{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 15.0, "eval_steps": 500, "global_step": 55305, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0027122321670735015, "grad_norm": 29.234992380530397, "learning_rate": 1.6271921894774907e-08, "loss": 4.7852, "step": 10 }, { "epoch": 0.005424464334147003, "grad_norm": 29.866194958292315, "learning_rate": 3.435183511119147e-08, "loss": 4.7636, "step": 20 }, { "epoch": 0.008136696501220505, "grad_norm": 25.187897217737806, "learning_rate": 5.243174832760803e-08, "loss": 4.7024, "step": 30 }, { "epoch": 0.010848928668294006, "grad_norm": 26.920314306519167, "learning_rate": 7.05116615440246e-08, "loss": 4.7061, "step": 40 }, { "epoch": 0.013561160835367507, "grad_norm": 25.22299668122294, "learning_rate": 8.859157476044116e-08, "loss": 4.7814, "step": 50 }, { "epoch": 0.01627339300244101, "grad_norm": 27.941637516338176, "learning_rate": 1.0667148797685772e-07, "loss": 4.6904, "step": 60 }, { "epoch": 0.01898562516951451, "grad_norm": 31.159265900997948, "learning_rate": 1.2475140119327429e-07, "loss": 4.7183, "step": 70 }, { "epoch": 0.021697857336588012, "grad_norm": 27.74006564083179, "learning_rate": 1.4283131440969085e-07, "loss": 4.7586, "step": 80 }, { "epoch": 0.024410089503661515, "grad_norm": 25.214266536899295, "learning_rate": 1.6091122762610739e-07, "loss": 4.6768, "step": 90 }, { "epoch": 0.027122321670735014, "grad_norm": 22.99808688479669, "learning_rate": 1.7899114084252395e-07, "loss": 4.6311, "step": 100 }, { "epoch": 0.029834553837808517, "grad_norm": 22.959881899963364, "learning_rate": 1.970710540589405e-07, "loss": 4.623, "step": 110 }, { "epoch": 0.03254678600488202, "grad_norm": 17.830599823903277, "learning_rate": 2.1515096727535708e-07, "loss": 4.6255, "step": 120 }, { "epoch": 0.03525901817195552, "grad_norm": 14.589545506224427, "learning_rate": 2.3323088049177364e-07, "loss": 4.5298, "step": 130 }, { "epoch": 0.03797125033902902, "grad_norm": 17.01493096340876, "learning_rate": 2.5131079370819023e-07, "loss": 4.5051, "step": 140 }, { "epoch": 0.04068348250610252, "grad_norm": 15.899843091793233, "learning_rate": 2.693907069246068e-07, "loss": 4.4702, "step": 150 }, { "epoch": 0.043395714673176024, "grad_norm": 17.225752395503978, "learning_rate": 2.8747062014102335e-07, "loss": 4.4415, "step": 160 }, { "epoch": 0.04610794684024953, "grad_norm": 14.019499701501443, "learning_rate": 3.055505333574399e-07, "loss": 4.3882, "step": 170 }, { "epoch": 0.04882017900732303, "grad_norm": 14.543032709713735, "learning_rate": 3.236304465738565e-07, "loss": 4.4889, "step": 180 }, { "epoch": 0.051532411174396525, "grad_norm": 13.233793937488173, "learning_rate": 3.4171035979027304e-07, "loss": 4.3834, "step": 190 }, { "epoch": 0.05424464334147003, "grad_norm": 13.092943926448488, "learning_rate": 3.597902730066896e-07, "loss": 4.4789, "step": 200 }, { "epoch": 0.05695687550854353, "grad_norm": 15.04393829749576, "learning_rate": 3.7787018622310617e-07, "loss": 4.3622, "step": 210 }, { "epoch": 0.059669107675617034, "grad_norm": 13.839493050721675, "learning_rate": 3.9595009943952273e-07, "loss": 4.3977, "step": 220 }, { "epoch": 0.062381339842690536, "grad_norm": 13.658557593061996, "learning_rate": 4.140300126559393e-07, "loss": 4.4393, "step": 230 }, { "epoch": 0.06509357200976404, "grad_norm": 13.729298801660116, "learning_rate": 4.3210992587235586e-07, "loss": 4.3637, "step": 240 }, { "epoch": 0.06780580417683754, "grad_norm": 15.41945160750915, "learning_rate": 4.5018983908877237e-07, "loss": 4.389, "step": 250 }, { "epoch": 0.07051803634391104, "grad_norm": 14.318729034682729, "learning_rate": 4.68269752305189e-07, "loss": 4.4047, "step": 260 }, { "epoch": 0.07323026851098453, "grad_norm": 13.819336802666294, "learning_rate": 4.863496655216055e-07, "loss": 4.3047, "step": 270 }, { "epoch": 0.07594250067805804, "grad_norm": 13.021006918000907, "learning_rate": 5.044295787380222e-07, "loss": 4.401, "step": 280 }, { "epoch": 0.07865473284513154, "grad_norm": 13.431808217026937, "learning_rate": 5.225094919544386e-07, "loss": 4.3866, "step": 290 }, { "epoch": 0.08136696501220504, "grad_norm": 14.036901008874946, "learning_rate": 5.405894051708553e-07, "loss": 4.313, "step": 300 }, { "epoch": 0.08407919717927854, "grad_norm": 13.374050817534993, "learning_rate": 5.586693183872717e-07, "loss": 4.2673, "step": 310 }, { "epoch": 0.08679142934635205, "grad_norm": 13.330133965995783, "learning_rate": 5.767492316036884e-07, "loss": 4.3291, "step": 320 }, { "epoch": 0.08950366151342555, "grad_norm": 12.82003843416354, "learning_rate": 5.948291448201049e-07, "loss": 4.4483, "step": 330 }, { "epoch": 0.09221589368049905, "grad_norm": 14.393543197666967, "learning_rate": 6.129090580365215e-07, "loss": 4.3692, "step": 340 }, { "epoch": 0.09492812584757256, "grad_norm": 13.822599092866522, "learning_rate": 6.309889712529381e-07, "loss": 4.3624, "step": 350 }, { "epoch": 0.09764035801464606, "grad_norm": 12.897031913470373, "learning_rate": 6.490688844693547e-07, "loss": 4.2708, "step": 360 }, { "epoch": 0.10035259018171956, "grad_norm": 16.303157251933566, "learning_rate": 6.671487976857711e-07, "loss": 4.3523, "step": 370 }, { "epoch": 0.10306482234879305, "grad_norm": 13.526632404178025, "learning_rate": 6.852287109021877e-07, "loss": 4.2914, "step": 380 }, { "epoch": 0.10577705451586655, "grad_norm": 14.449504698019009, "learning_rate": 7.033086241186044e-07, "loss": 4.2882, "step": 390 }, { "epoch": 0.10848928668294006, "grad_norm": 13.251351828889392, "learning_rate": 7.213885373350209e-07, "loss": 4.249, "step": 400 }, { "epoch": 0.11120151885001356, "grad_norm": 13.320854541661046, "learning_rate": 7.394684505514374e-07, "loss": 4.3215, "step": 410 }, { "epoch": 0.11391375101708706, "grad_norm": 13.600090892167572, "learning_rate": 7.575483637678539e-07, "loss": 4.1808, "step": 420 }, { "epoch": 0.11662598318416056, "grad_norm": 15.487971803294583, "learning_rate": 7.756282769842706e-07, "loss": 4.2437, "step": 430 }, { "epoch": 0.11933821535123407, "grad_norm": 13.214473772393566, "learning_rate": 7.937081902006872e-07, "loss": 4.2576, "step": 440 }, { "epoch": 0.12205044751830757, "grad_norm": 15.78029461717227, "learning_rate": 8.117881034171036e-07, "loss": 4.2355, "step": 450 }, { "epoch": 0.12476267968538107, "grad_norm": 13.023946815505676, "learning_rate": 8.298680166335202e-07, "loss": 4.2677, "step": 460 }, { "epoch": 0.12747491185245458, "grad_norm": 13.313093112441901, "learning_rate": 8.479479298499369e-07, "loss": 4.2108, "step": 470 }, { "epoch": 0.13018714401952808, "grad_norm": 13.453086281279976, "learning_rate": 8.660278430663533e-07, "loss": 4.2319, "step": 480 }, { "epoch": 0.13289937618660158, "grad_norm": 12.62767069120399, "learning_rate": 8.841077562827699e-07, "loss": 4.2966, "step": 490 }, { "epoch": 0.13561160835367508, "grad_norm": 13.71579353256742, "learning_rate": 9.021876694991864e-07, "loss": 4.15, "step": 500 }, { "epoch": 0.1383238405207486, "grad_norm": 13.091962225921964, "learning_rate": 9.202675827156031e-07, "loss": 4.2308, "step": 510 }, { "epoch": 0.1410360726878221, "grad_norm": 12.931211360154784, "learning_rate": 9.383474959320196e-07, "loss": 4.1481, "step": 520 }, { "epoch": 0.1437483048548956, "grad_norm": 12.67785399788353, "learning_rate": 9.564274091484362e-07, "loss": 4.1907, "step": 530 }, { "epoch": 0.14646053702196907, "grad_norm": 16.23599075359464, "learning_rate": 9.745073223648527e-07, "loss": 4.1921, "step": 540 }, { "epoch": 0.14917276918904257, "grad_norm": 14.535038960183437, "learning_rate": 9.925872355812694e-07, "loss": 4.1715, "step": 550 }, { "epoch": 0.15188500135611607, "grad_norm": 13.795996781461202, "learning_rate": 1.0106671487976858e-06, "loss": 4.1325, "step": 560 }, { "epoch": 0.15459723352318958, "grad_norm": 13.88765872728385, "learning_rate": 1.0287470620141025e-06, "loss": 4.168, "step": 570 }, { "epoch": 0.15730946569026308, "grad_norm": 14.868807422333328, "learning_rate": 1.046826975230519e-06, "loss": 4.1293, "step": 580 }, { "epoch": 0.16002169785733658, "grad_norm": 14.809291535345027, "learning_rate": 1.0649068884469356e-06, "loss": 4.1411, "step": 590 }, { "epoch": 0.16273393002441008, "grad_norm": 15.386203652584644, "learning_rate": 1.082986801663352e-06, "loss": 4.1928, "step": 600 }, { "epoch": 0.1654461621914836, "grad_norm": 13.898293684769802, "learning_rate": 1.1010667148797687e-06, "loss": 4.1443, "step": 610 }, { "epoch": 0.1681583943585571, "grad_norm": 16.748701318186207, "learning_rate": 1.1191466280961852e-06, "loss": 4.1963, "step": 620 }, { "epoch": 0.1708706265256306, "grad_norm": 15.152537496972242, "learning_rate": 1.1372265413126019e-06, "loss": 4.1719, "step": 630 }, { "epoch": 0.1735828586927041, "grad_norm": 12.934599896208905, "learning_rate": 1.1553064545290183e-06, "loss": 4.1117, "step": 640 }, { "epoch": 0.1762950908597776, "grad_norm": 14.038074090319338, "learning_rate": 1.173386367745435e-06, "loss": 4.1551, "step": 650 }, { "epoch": 0.1790073230268511, "grad_norm": 14.084906227563744, "learning_rate": 1.1914662809618515e-06, "loss": 4.2041, "step": 660 }, { "epoch": 0.1817195551939246, "grad_norm": 15.492186224693576, "learning_rate": 1.2095461941782681e-06, "loss": 4.0362, "step": 670 }, { "epoch": 0.1844317873609981, "grad_norm": 14.469353240063771, "learning_rate": 1.2276261073946846e-06, "loss": 4.1344, "step": 680 }, { "epoch": 0.1871440195280716, "grad_norm": 15.407270925472952, "learning_rate": 1.2457060206111013e-06, "loss": 4.0513, "step": 690 }, { "epoch": 0.1898562516951451, "grad_norm": 13.185916674059047, "learning_rate": 1.2637859338275177e-06, "loss": 4.052, "step": 700 }, { "epoch": 0.19256848386221861, "grad_norm": 12.632001817697322, "learning_rate": 1.2818658470439344e-06, "loss": 4.1918, "step": 710 }, { "epoch": 0.19528071602929212, "grad_norm": 15.241870167568408, "learning_rate": 1.2999457602603508e-06, "loss": 4.0931, "step": 720 }, { "epoch": 0.19799294819636562, "grad_norm": 12.508249202808527, "learning_rate": 1.3180256734767675e-06, "loss": 4.1227, "step": 730 }, { "epoch": 0.20070518036343912, "grad_norm": 13.369115364877795, "learning_rate": 1.3361055866931842e-06, "loss": 4.1315, "step": 740 }, { "epoch": 0.20341741253051263, "grad_norm": 15.072368573519993, "learning_rate": 1.3541854999096004e-06, "loss": 4.041, "step": 750 }, { "epoch": 0.2061296446975861, "grad_norm": 13.457265874853475, "learning_rate": 1.372265413126017e-06, "loss": 4.1177, "step": 760 }, { "epoch": 0.2088418768646596, "grad_norm": 14.679618961841538, "learning_rate": 1.3903453263424338e-06, "loss": 4.1193, "step": 770 }, { "epoch": 0.2115541090317331, "grad_norm": 13.274209924963621, "learning_rate": 1.4084252395588502e-06, "loss": 4.0137, "step": 780 }, { "epoch": 0.2142663411988066, "grad_norm": 14.031683006872955, "learning_rate": 1.4265051527752669e-06, "loss": 4.09, "step": 790 }, { "epoch": 0.2169785733658801, "grad_norm": 13.388005569860587, "learning_rate": 1.4445850659916833e-06, "loss": 4.1734, "step": 800 }, { "epoch": 0.21969080553295361, "grad_norm": 13.798572055117402, "learning_rate": 1.4626649792081e-06, "loss": 4.0859, "step": 810 }, { "epoch": 0.22240303770002712, "grad_norm": 13.918412736439748, "learning_rate": 1.4807448924245167e-06, "loss": 4.1053, "step": 820 }, { "epoch": 0.22511526986710062, "grad_norm": 12.708501257413488, "learning_rate": 1.498824805640933e-06, "loss": 4.0999, "step": 830 }, { "epoch": 0.22782750203417412, "grad_norm": 14.112457574669156, "learning_rate": 1.5169047188573496e-06, "loss": 4.1387, "step": 840 }, { "epoch": 0.23053973420124763, "grad_norm": 14.10021218365695, "learning_rate": 1.5349846320737663e-06, "loss": 4.0353, "step": 850 }, { "epoch": 0.23325196636832113, "grad_norm": 13.033965645128879, "learning_rate": 1.5530645452901827e-06, "loss": 4.0238, "step": 860 }, { "epoch": 0.23596419853539463, "grad_norm": 15.887602244962897, "learning_rate": 1.5711444585065994e-06, "loss": 3.9962, "step": 870 }, { "epoch": 0.23867643070246813, "grad_norm": 13.274875212570835, "learning_rate": 1.5892243717230158e-06, "loss": 4.1162, "step": 880 }, { "epoch": 0.24138866286954164, "grad_norm": 12.706368939055737, "learning_rate": 1.6073042849394325e-06, "loss": 4.1144, "step": 890 }, { "epoch": 0.24410089503661514, "grad_norm": 13.05760405079636, "learning_rate": 1.625384198155849e-06, "loss": 4.0079, "step": 900 }, { "epoch": 0.24681312720368864, "grad_norm": 13.895289886060388, "learning_rate": 1.6434641113722654e-06, "loss": 4.0209, "step": 910 }, { "epoch": 0.24952535937076215, "grad_norm": 13.300425087616022, "learning_rate": 1.661544024588682e-06, "loss": 4.0199, "step": 920 }, { "epoch": 0.25223759153783565, "grad_norm": 12.686913322550453, "learning_rate": 1.6796239378050988e-06, "loss": 4.0526, "step": 930 }, { "epoch": 0.25494982370490915, "grad_norm": 15.024789366966468, "learning_rate": 1.6977038510215152e-06, "loss": 4.0259, "step": 940 }, { "epoch": 0.25766205587198265, "grad_norm": 14.39229621129743, "learning_rate": 1.715783764237932e-06, "loss": 4.1294, "step": 950 }, { "epoch": 0.26037428803905616, "grad_norm": 12.549862389809645, "learning_rate": 1.7338636774543484e-06, "loss": 3.966, "step": 960 }, { "epoch": 0.26308652020612966, "grad_norm": 12.923710868476228, "learning_rate": 1.7519435906707648e-06, "loss": 3.9575, "step": 970 }, { "epoch": 0.26579875237320316, "grad_norm": 12.59765997142362, "learning_rate": 1.7700235038871815e-06, "loss": 4.0288, "step": 980 }, { "epoch": 0.26851098454027666, "grad_norm": 12.404735317737702, "learning_rate": 1.788103417103598e-06, "loss": 3.9576, "step": 990 }, { "epoch": 0.27122321670735017, "grad_norm": 12.756141555120886, "learning_rate": 1.8061833303200146e-06, "loss": 4.0877, "step": 1000 }, { "epoch": 0.27393544887442367, "grad_norm": 14.151332192232289, "learning_rate": 1.8242632435364313e-06, "loss": 3.9144, "step": 1010 }, { "epoch": 0.2766476810414972, "grad_norm": 14.414266018992917, "learning_rate": 1.8423431567528477e-06, "loss": 4.081, "step": 1020 }, { "epoch": 0.2793599132085707, "grad_norm": 14.416371100010593, "learning_rate": 1.8604230699692644e-06, "loss": 4.0474, "step": 1030 }, { "epoch": 0.2820721453756442, "grad_norm": 12.577192712456604, "learning_rate": 1.8785029831856806e-06, "loss": 4.0247, "step": 1040 }, { "epoch": 0.2847843775427177, "grad_norm": 11.682532712466617, "learning_rate": 1.8965828964020973e-06, "loss": 4.0606, "step": 1050 }, { "epoch": 0.2874966097097912, "grad_norm": 14.232651462218518, "learning_rate": 1.914662809618514e-06, "loss": 3.9761, "step": 1060 }, { "epoch": 0.2902088418768647, "grad_norm": 11.834767113973433, "learning_rate": 1.9327427228349304e-06, "loss": 3.9883, "step": 1070 }, { "epoch": 0.29292107404393813, "grad_norm": 13.078120264616075, "learning_rate": 1.950822636051347e-06, "loss": 3.9621, "step": 1080 }, { "epoch": 0.29563330621101164, "grad_norm": 15.046678338541682, "learning_rate": 1.9689025492677638e-06, "loss": 4.1448, "step": 1090 }, { "epoch": 0.29834553837808514, "grad_norm": 12.050710240730098, "learning_rate": 1.98698246248418e-06, "loss": 4.0172, "step": 1100 }, { "epoch": 0.30105777054515864, "grad_norm": 12.820130911132795, "learning_rate": 2.0050623757005967e-06, "loss": 3.9178, "step": 1110 }, { "epoch": 0.30377000271223215, "grad_norm": 13.949466177109464, "learning_rate": 2.0231422889170134e-06, "loss": 4.0867, "step": 1120 }, { "epoch": 0.30648223487930565, "grad_norm": 14.675431806260724, "learning_rate": 2.04122220213343e-06, "loss": 4.0163, "step": 1130 }, { "epoch": 0.30919446704637915, "grad_norm": 14.704793384687548, "learning_rate": 2.0593021153498467e-06, "loss": 3.9384, "step": 1140 }, { "epoch": 0.31190669921345265, "grad_norm": 13.547312395576375, "learning_rate": 2.077382028566263e-06, "loss": 3.8928, "step": 1150 }, { "epoch": 0.31461893138052616, "grad_norm": 14.097027846106963, "learning_rate": 2.0954619417826796e-06, "loss": 3.9725, "step": 1160 }, { "epoch": 0.31733116354759966, "grad_norm": 12.400061630505512, "learning_rate": 2.1135418549990963e-06, "loss": 4.0311, "step": 1170 }, { "epoch": 0.32004339571467316, "grad_norm": 13.557655261844179, "learning_rate": 2.1316217682155125e-06, "loss": 4.001, "step": 1180 }, { "epoch": 0.32275562788174667, "grad_norm": 16.381196144332066, "learning_rate": 2.149701681431929e-06, "loss": 4.1063, "step": 1190 }, { "epoch": 0.32546786004882017, "grad_norm": 12.995049821214893, "learning_rate": 2.167781594648346e-06, "loss": 4.0115, "step": 1200 }, { "epoch": 0.32818009221589367, "grad_norm": 11.64262748130277, "learning_rate": 2.1858615078647625e-06, "loss": 3.9593, "step": 1210 }, { "epoch": 0.3308923243829672, "grad_norm": 13.10674137030383, "learning_rate": 2.203941421081179e-06, "loss": 4.0291, "step": 1220 }, { "epoch": 0.3336045565500407, "grad_norm": 16.2427684001199, "learning_rate": 2.2220213342975955e-06, "loss": 3.9953, "step": 1230 }, { "epoch": 0.3363167887171142, "grad_norm": 12.863707313605202, "learning_rate": 2.240101247514012e-06, "loss": 4.0433, "step": 1240 }, { "epoch": 0.3390290208841877, "grad_norm": 11.951841505758546, "learning_rate": 2.258181160730429e-06, "loss": 4.0223, "step": 1250 }, { "epoch": 0.3417412530512612, "grad_norm": 12.371971600807923, "learning_rate": 2.276261073946845e-06, "loss": 3.9531, "step": 1260 }, { "epoch": 0.3444534852183347, "grad_norm": 13.092330733865937, "learning_rate": 2.2943409871632617e-06, "loss": 3.9822, "step": 1270 }, { "epoch": 0.3471657173854082, "grad_norm": 13.361752426020843, "learning_rate": 2.3124209003796784e-06, "loss": 4.0502, "step": 1280 }, { "epoch": 0.3498779495524817, "grad_norm": 14.224226616715113, "learning_rate": 2.330500813596095e-06, "loss": 4.069, "step": 1290 }, { "epoch": 0.3525901817195552, "grad_norm": 13.408899892012645, "learning_rate": 2.3485807268125117e-06, "loss": 3.9982, "step": 1300 }, { "epoch": 0.3553024138866287, "grad_norm": 13.251802382435033, "learning_rate": 2.366660640028928e-06, "loss": 4.0106, "step": 1310 }, { "epoch": 0.3580146460537022, "grad_norm": 14.138621170144297, "learning_rate": 2.3847405532453446e-06, "loss": 4.0033, "step": 1320 }, { "epoch": 0.3607268782207757, "grad_norm": 11.346171575106473, "learning_rate": 2.4028204664617613e-06, "loss": 3.9854, "step": 1330 }, { "epoch": 0.3634391103878492, "grad_norm": 13.887297823316333, "learning_rate": 2.4209003796781775e-06, "loss": 3.9235, "step": 1340 }, { "epoch": 0.3661513425549227, "grad_norm": 12.08655132632429, "learning_rate": 2.438980292894594e-06, "loss": 3.8951, "step": 1350 }, { "epoch": 0.3688635747219962, "grad_norm": 12.780525375344373, "learning_rate": 2.457060206111011e-06, "loss": 4.0178, "step": 1360 }, { "epoch": 0.3715758068890697, "grad_norm": 12.877217818566415, "learning_rate": 2.4751401193274275e-06, "loss": 4.0165, "step": 1370 }, { "epoch": 0.3742880390561432, "grad_norm": 12.397682558989823, "learning_rate": 2.4932200325438442e-06, "loss": 3.9743, "step": 1380 }, { "epoch": 0.3770002712232167, "grad_norm": 12.509386828153037, "learning_rate": 2.5112999457602605e-06, "loss": 3.8757, "step": 1390 }, { "epoch": 0.3797125033902902, "grad_norm": 14.658910172635418, "learning_rate": 2.529379858976677e-06, "loss": 3.9652, "step": 1400 }, { "epoch": 0.3824247355573637, "grad_norm": 11.146789082397893, "learning_rate": 2.547459772193094e-06, "loss": 3.9808, "step": 1410 }, { "epoch": 0.38513696772443723, "grad_norm": 12.78197069441951, "learning_rate": 2.5655396854095105e-06, "loss": 3.9607, "step": 1420 }, { "epoch": 0.38784919989151073, "grad_norm": 12.648853089940642, "learning_rate": 2.5836195986259267e-06, "loss": 3.9787, "step": 1430 }, { "epoch": 0.39056143205858423, "grad_norm": 13.173712406733536, "learning_rate": 2.6016995118423434e-06, "loss": 3.9847, "step": 1440 }, { "epoch": 0.39327366422565774, "grad_norm": 13.945896159719027, "learning_rate": 2.61977942505876e-06, "loss": 4.0322, "step": 1450 }, { "epoch": 0.39598589639273124, "grad_norm": 13.3284754510414, "learning_rate": 2.6378593382751767e-06, "loss": 3.8631, "step": 1460 }, { "epoch": 0.39869812855980474, "grad_norm": 11.792033683693688, "learning_rate": 2.6559392514915934e-06, "loss": 3.963, "step": 1470 }, { "epoch": 0.40141036072687825, "grad_norm": 12.847399025820105, "learning_rate": 2.6740191647080092e-06, "loss": 4.0428, "step": 1480 }, { "epoch": 0.40412259289395175, "grad_norm": 11.748127508351029, "learning_rate": 2.692099077924426e-06, "loss": 4.0881, "step": 1490 }, { "epoch": 0.40683482506102525, "grad_norm": 13.183615268569895, "learning_rate": 2.7101789911408426e-06, "loss": 4.0291, "step": 1500 }, { "epoch": 0.4095470572280987, "grad_norm": 14.21478282441351, "learning_rate": 2.7282589043572592e-06, "loss": 4.0158, "step": 1510 }, { "epoch": 0.4122592893951722, "grad_norm": 20.349995994627943, "learning_rate": 2.746338817573676e-06, "loss": 3.9919, "step": 1520 }, { "epoch": 0.4149715215622457, "grad_norm": 13.324071288858335, "learning_rate": 2.7644187307900926e-06, "loss": 3.9645, "step": 1530 }, { "epoch": 0.4176837537293192, "grad_norm": 11.845122211343067, "learning_rate": 2.782498644006509e-06, "loss": 3.9864, "step": 1540 }, { "epoch": 0.4203959858963927, "grad_norm": 12.493461149182778, "learning_rate": 2.8005785572229255e-06, "loss": 3.9775, "step": 1550 }, { "epoch": 0.4231082180634662, "grad_norm": 12.562054967401073, "learning_rate": 2.818658470439342e-06, "loss": 3.9745, "step": 1560 }, { "epoch": 0.4258204502305397, "grad_norm": 12.552296812224485, "learning_rate": 2.836738383655759e-06, "loss": 3.9277, "step": 1570 }, { "epoch": 0.4285326823976132, "grad_norm": 13.636268218007535, "learning_rate": 2.8548182968721755e-06, "loss": 3.9175, "step": 1580 }, { "epoch": 0.4312449145646867, "grad_norm": 14.320154587158468, "learning_rate": 2.8728982100885917e-06, "loss": 3.893, "step": 1590 }, { "epoch": 0.4339571467317602, "grad_norm": 15.711345322048398, "learning_rate": 2.8909781233050084e-06, "loss": 3.9615, "step": 1600 }, { "epoch": 0.4366693788988337, "grad_norm": 12.842904636494813, "learning_rate": 2.909058036521425e-06, "loss": 3.9274, "step": 1610 }, { "epoch": 0.43938161106590723, "grad_norm": 14.182851966705977, "learning_rate": 2.9271379497378417e-06, "loss": 3.9732, "step": 1620 }, { "epoch": 0.44209384323298073, "grad_norm": 15.749079506985254, "learning_rate": 2.9452178629542584e-06, "loss": 3.829, "step": 1630 }, { "epoch": 0.44480607540005423, "grad_norm": 11.867518070425094, "learning_rate": 2.9632977761706742e-06, "loss": 3.9236, "step": 1640 }, { "epoch": 0.44751830756712774, "grad_norm": 12.556643690081478, "learning_rate": 2.981377689387091e-06, "loss": 3.8797, "step": 1650 }, { "epoch": 0.45023053973420124, "grad_norm": 14.5858069410831, "learning_rate": 2.9994576026035076e-06, "loss": 3.9235, "step": 1660 }, { "epoch": 0.45294277190127474, "grad_norm": 11.612038205382161, "learning_rate": 3.0175375158199242e-06, "loss": 3.8556, "step": 1670 }, { "epoch": 0.45565500406834825, "grad_norm": 13.506638821420506, "learning_rate": 3.035617429036341e-06, "loss": 3.9297, "step": 1680 }, { "epoch": 0.45836723623542175, "grad_norm": 11.422459291871927, "learning_rate": 3.0536973422527576e-06, "loss": 3.882, "step": 1690 }, { "epoch": 0.46107946840249525, "grad_norm": 18.450971951247272, "learning_rate": 3.071777255469174e-06, "loss": 3.9306, "step": 1700 }, { "epoch": 0.46379170056956875, "grad_norm": 12.26569395318839, "learning_rate": 3.0898571686855905e-06, "loss": 3.9972, "step": 1710 }, { "epoch": 0.46650393273664226, "grad_norm": 13.087078752709406, "learning_rate": 3.107937081902007e-06, "loss": 3.933, "step": 1720 }, { "epoch": 0.46921616490371576, "grad_norm": 14.747112551810982, "learning_rate": 3.126016995118424e-06, "loss": 3.9455, "step": 1730 }, { "epoch": 0.47192839707078926, "grad_norm": 11.53040547011405, "learning_rate": 3.1440969083348405e-06, "loss": 3.9295, "step": 1740 }, { "epoch": 0.47464062923786277, "grad_norm": 16.25689913464232, "learning_rate": 3.1621768215512567e-06, "loss": 3.8138, "step": 1750 }, { "epoch": 0.47735286140493627, "grad_norm": 13.387272818141222, "learning_rate": 3.1802567347676734e-06, "loss": 3.946, "step": 1760 }, { "epoch": 0.48006509357200977, "grad_norm": 12.856875515376494, "learning_rate": 3.19833664798409e-06, "loss": 3.9766, "step": 1770 }, { "epoch": 0.4827773257390833, "grad_norm": 13.680281872251493, "learning_rate": 3.2164165612005067e-06, "loss": 3.8251, "step": 1780 }, { "epoch": 0.4854895579061568, "grad_norm": 12.694040586201918, "learning_rate": 3.2344964744169234e-06, "loss": 4.0124, "step": 1790 }, { "epoch": 0.4882017900732303, "grad_norm": 11.104136755199528, "learning_rate": 3.2525763876333392e-06, "loss": 3.8558, "step": 1800 }, { "epoch": 0.4909140222403038, "grad_norm": 13.787555014201764, "learning_rate": 3.270656300849756e-06, "loss": 3.8592, "step": 1810 }, { "epoch": 0.4936262544073773, "grad_norm": 12.120847587492477, "learning_rate": 3.2887362140661726e-06, "loss": 3.9008, "step": 1820 }, { "epoch": 0.4963384865744508, "grad_norm": 11.19553156977807, "learning_rate": 3.3068161272825892e-06, "loss": 3.9435, "step": 1830 }, { "epoch": 0.4990507187415243, "grad_norm": 13.676511994238854, "learning_rate": 3.324896040499006e-06, "loss": 3.9437, "step": 1840 }, { "epoch": 0.5017629509085978, "grad_norm": 14.621597105589913, "learning_rate": 3.342975953715422e-06, "loss": 4.0213, "step": 1850 }, { "epoch": 0.5044751830756713, "grad_norm": 12.698285064351117, "learning_rate": 3.361055866931839e-06, "loss": 4.0509, "step": 1860 }, { "epoch": 0.5071874152427448, "grad_norm": 11.752710436417889, "learning_rate": 3.3791357801482555e-06, "loss": 3.9186, "step": 1870 }, { "epoch": 0.5098996474098183, "grad_norm": 11.510937506229855, "learning_rate": 3.397215693364672e-06, "loss": 3.8712, "step": 1880 }, { "epoch": 0.5126118795768918, "grad_norm": 15.168590045397885, "learning_rate": 3.415295606581089e-06, "loss": 3.912, "step": 1890 }, { "epoch": 0.5153241117439653, "grad_norm": 11.356032008716202, "learning_rate": 3.4333755197975055e-06, "loss": 3.8964, "step": 1900 }, { "epoch": 0.5180363439110388, "grad_norm": 12.786307922988563, "learning_rate": 3.4514554330139217e-06, "loss": 3.9336, "step": 1910 }, { "epoch": 0.5207485760781123, "grad_norm": 12.694810325884477, "learning_rate": 3.4695353462303384e-06, "loss": 3.9115, "step": 1920 }, { "epoch": 0.5234608082451858, "grad_norm": 10.750121948096671, "learning_rate": 3.487615259446755e-06, "loss": 3.8799, "step": 1930 }, { "epoch": 0.5261730404122593, "grad_norm": 12.430213857359213, "learning_rate": 3.5056951726631718e-06, "loss": 3.7581, "step": 1940 }, { "epoch": 0.5288852725793328, "grad_norm": 12.536667314342091, "learning_rate": 3.5237750858795884e-06, "loss": 3.9481, "step": 1950 }, { "epoch": 0.5315975047464063, "grad_norm": 11.210684428052708, "learning_rate": 3.5418549990960042e-06, "loss": 3.9843, "step": 1960 }, { "epoch": 0.5343097369134798, "grad_norm": 12.004766183049412, "learning_rate": 3.559934912312421e-06, "loss": 3.9456, "step": 1970 }, { "epoch": 0.5370219690805533, "grad_norm": 12.057151080102477, "learning_rate": 3.5780148255288376e-06, "loss": 3.8938, "step": 1980 }, { "epoch": 0.5397342012476268, "grad_norm": 11.502764433040888, "learning_rate": 3.5960947387452543e-06, "loss": 3.9093, "step": 1990 }, { "epoch": 0.5424464334147003, "grad_norm": 12.157595118461936, "learning_rate": 3.614174651961671e-06, "loss": 3.9034, "step": 2000 }, { "epoch": 0.5451586655817738, "grad_norm": 12.590747815130253, "learning_rate": 3.632254565178087e-06, "loss": 3.9476, "step": 2010 }, { "epoch": 0.5478708977488473, "grad_norm": 11.012151876670888, "learning_rate": 3.650334478394504e-06, "loss": 3.9158, "step": 2020 }, { "epoch": 0.5505831299159208, "grad_norm": 13.47477140838514, "learning_rate": 3.6684143916109205e-06, "loss": 3.9717, "step": 2030 }, { "epoch": 0.5532953620829943, "grad_norm": 12.943102522028964, "learning_rate": 3.686494304827337e-06, "loss": 3.9473, "step": 2040 }, { "epoch": 0.5560075942500678, "grad_norm": 11.708064749316591, "learning_rate": 3.704574218043754e-06, "loss": 3.932, "step": 2050 }, { "epoch": 0.5587198264171414, "grad_norm": 13.080053472108048, "learning_rate": 3.7226541312601705e-06, "loss": 3.89, "step": 2060 }, { "epoch": 0.5614320585842149, "grad_norm": 10.539277134395068, "learning_rate": 3.7407340444765868e-06, "loss": 3.9671, "step": 2070 }, { "epoch": 0.5641442907512884, "grad_norm": 12.835475463452825, "learning_rate": 3.7588139576930034e-06, "loss": 3.9441, "step": 2080 }, { "epoch": 0.5668565229183619, "grad_norm": 12.696107230122212, "learning_rate": 3.77689387090942e-06, "loss": 4.0242, "step": 2090 }, { "epoch": 0.5695687550854354, "grad_norm": 12.147406397603163, "learning_rate": 3.7949737841258368e-06, "loss": 3.9726, "step": 2100 }, { "epoch": 0.5722809872525089, "grad_norm": 12.267298820304172, "learning_rate": 3.8130536973422534e-06, "loss": 3.9066, "step": 2110 }, { "epoch": 0.5749932194195824, "grad_norm": 13.043429264921691, "learning_rate": 3.831133610558669e-06, "loss": 3.9515, "step": 2120 }, { "epoch": 0.5777054515866559, "grad_norm": 12.77440963209346, "learning_rate": 3.849213523775086e-06, "loss": 3.7644, "step": 2130 }, { "epoch": 0.5804176837537294, "grad_norm": 12.816194641139372, "learning_rate": 3.867293436991503e-06, "loss": 3.9741, "step": 2140 }, { "epoch": 0.5831299159208028, "grad_norm": 13.75954713044274, "learning_rate": 3.88537335020792e-06, "loss": 3.9633, "step": 2150 }, { "epoch": 0.5858421480878763, "grad_norm": 11.448752429757427, "learning_rate": 3.903453263424336e-06, "loss": 4.0077, "step": 2160 }, { "epoch": 0.5885543802549498, "grad_norm": 13.108484092207295, "learning_rate": 3.921533176640752e-06, "loss": 3.8084, "step": 2170 }, { "epoch": 0.5912666124220233, "grad_norm": 14.16551190951394, "learning_rate": 3.9396130898571684e-06, "loss": 3.9106, "step": 2180 }, { "epoch": 0.5939788445890968, "grad_norm": 11.028342646325182, "learning_rate": 3.9576930030735855e-06, "loss": 3.8935, "step": 2190 }, { "epoch": 0.5966910767561703, "grad_norm": 10.57445177317351, "learning_rate": 3.975772916290002e-06, "loss": 3.9363, "step": 2200 }, { "epoch": 0.5994033089232438, "grad_norm": 12.394557139520593, "learning_rate": 3.993852829506419e-06, "loss": 3.9073, "step": 2210 }, { "epoch": 0.6021155410903173, "grad_norm": 14.393110788375015, "learning_rate": 4.011932742722835e-06, "loss": 3.8491, "step": 2220 }, { "epoch": 0.6048277732573908, "grad_norm": 12.525193679869188, "learning_rate": 4.030012655939251e-06, "loss": 3.778, "step": 2230 }, { "epoch": 0.6075400054244643, "grad_norm": 13.558236027676076, "learning_rate": 4.0480925691556684e-06, "loss": 3.8103, "step": 2240 }, { "epoch": 0.6102522375915378, "grad_norm": 12.69350829302961, "learning_rate": 4.066172482372085e-06, "loss": 3.9625, "step": 2250 }, { "epoch": 0.6129644697586113, "grad_norm": 10.916617842225943, "learning_rate": 4.084252395588502e-06, "loss": 3.9342, "step": 2260 }, { "epoch": 0.6156767019256848, "grad_norm": 13.125949099970542, "learning_rate": 4.102332308804918e-06, "loss": 3.9154, "step": 2270 }, { "epoch": 0.6183889340927583, "grad_norm": 17.12327770014318, "learning_rate": 4.120412222021334e-06, "loss": 3.8543, "step": 2280 }, { "epoch": 0.6211011662598318, "grad_norm": 11.380052430887456, "learning_rate": 4.138492135237751e-06, "loss": 3.9054, "step": 2290 }, { "epoch": 0.6238133984269053, "grad_norm": 12.325855023164907, "learning_rate": 4.156572048454168e-06, "loss": 3.902, "step": 2300 }, { "epoch": 0.6265256305939788, "grad_norm": 12.11239448952803, "learning_rate": 4.174651961670585e-06, "loss": 3.8745, "step": 2310 }, { "epoch": 0.6292378627610523, "grad_norm": 12.571574865016068, "learning_rate": 4.192731874887001e-06, "loss": 3.9762, "step": 2320 }, { "epoch": 0.6319500949281258, "grad_norm": 11.027029318313433, "learning_rate": 4.210811788103417e-06, "loss": 3.9028, "step": 2330 }, { "epoch": 0.6346623270951993, "grad_norm": 12.763903274669145, "learning_rate": 4.2288917013198334e-06, "loss": 3.8871, "step": 2340 }, { "epoch": 0.6373745592622728, "grad_norm": 12.852798170990782, "learning_rate": 4.2469716145362505e-06, "loss": 3.8295, "step": 2350 }, { "epoch": 0.6400867914293463, "grad_norm": 12.443733689597178, "learning_rate": 4.265051527752667e-06, "loss": 3.926, "step": 2360 }, { "epoch": 0.6427990235964198, "grad_norm": 12.440434827195787, "learning_rate": 4.283131440969084e-06, "loss": 3.8863, "step": 2370 }, { "epoch": 0.6455112557634933, "grad_norm": 15.754177197924054, "learning_rate": 4.3012113541855e-06, "loss": 3.8977, "step": 2380 }, { "epoch": 0.6482234879305668, "grad_norm": 13.339165824748694, "learning_rate": 4.319291267401916e-06, "loss": 3.7911, "step": 2390 }, { "epoch": 0.6509357200976403, "grad_norm": 12.10310396716957, "learning_rate": 4.3373711806183334e-06, "loss": 3.8586, "step": 2400 }, { "epoch": 0.6536479522647138, "grad_norm": 12.462575484159938, "learning_rate": 4.35545109383475e-06, "loss": 3.7856, "step": 2410 }, { "epoch": 0.6563601844317873, "grad_norm": 11.814285209162506, "learning_rate": 4.373531007051167e-06, "loss": 3.9801, "step": 2420 }, { "epoch": 0.6590724165988608, "grad_norm": 12.783724706074985, "learning_rate": 4.391610920267583e-06, "loss": 3.9122, "step": 2430 }, { "epoch": 0.6617846487659343, "grad_norm": 13.507898727562706, "learning_rate": 4.409690833483999e-06, "loss": 3.8873, "step": 2440 }, { "epoch": 0.6644968809330078, "grad_norm": 11.31263561649439, "learning_rate": 4.427770746700416e-06, "loss": 3.8727, "step": 2450 }, { "epoch": 0.6672091131000814, "grad_norm": 11.259426006898144, "learning_rate": 4.445850659916833e-06, "loss": 3.9149, "step": 2460 }, { "epoch": 0.6699213452671549, "grad_norm": 12.575255056326984, "learning_rate": 4.46393057313325e-06, "loss": 3.9339, "step": 2470 }, { "epoch": 0.6726335774342284, "grad_norm": 13.182076976452553, "learning_rate": 4.482010486349666e-06, "loss": 3.8904, "step": 2480 }, { "epoch": 0.6753458096013019, "grad_norm": 10.67245439103985, "learning_rate": 4.500090399566082e-06, "loss": 3.9051, "step": 2490 }, { "epoch": 0.6780580417683754, "grad_norm": 14.765214373321312, "learning_rate": 4.5181703127824984e-06, "loss": 4.032, "step": 2500 }, { "epoch": 0.6807702739354489, "grad_norm": 12.6070347112421, "learning_rate": 4.5362502259989155e-06, "loss": 3.8311, "step": 2510 }, { "epoch": 0.6834825061025224, "grad_norm": 11.54446816115957, "learning_rate": 4.554330139215332e-06, "loss": 3.9221, "step": 2520 }, { "epoch": 0.6861947382695959, "grad_norm": 14.429869794012275, "learning_rate": 4.572410052431749e-06, "loss": 3.8903, "step": 2530 }, { "epoch": 0.6889069704366694, "grad_norm": 12.55015047469403, "learning_rate": 4.590489965648165e-06, "loss": 3.9057, "step": 2540 }, { "epoch": 0.6916192026037429, "grad_norm": 10.94425995502571, "learning_rate": 4.608569878864581e-06, "loss": 3.9798, "step": 2550 }, { "epoch": 0.6943314347708164, "grad_norm": 11.751749593468364, "learning_rate": 4.6266497920809985e-06, "loss": 3.9801, "step": 2560 }, { "epoch": 0.6970436669378899, "grad_norm": 12.601071532078807, "learning_rate": 4.644729705297415e-06, "loss": 3.8737, "step": 2570 }, { "epoch": 0.6997558991049634, "grad_norm": 12.46546014119485, "learning_rate": 4.662809618513832e-06, "loss": 3.8675, "step": 2580 }, { "epoch": 0.7024681312720369, "grad_norm": 13.724998288731115, "learning_rate": 4.680889531730248e-06, "loss": 3.894, "step": 2590 }, { "epoch": 0.7051803634391104, "grad_norm": 11.388614340136172, "learning_rate": 4.698969444946664e-06, "loss": 3.8888, "step": 2600 }, { "epoch": 0.7078925956061839, "grad_norm": 11.634522374855255, "learning_rate": 4.717049358163081e-06, "loss": 3.8467, "step": 2610 }, { "epoch": 0.7106048277732574, "grad_norm": 12.428733042492535, "learning_rate": 4.735129271379498e-06, "loss": 3.881, "step": 2620 }, { "epoch": 0.7133170599403309, "grad_norm": 12.437401378634034, "learning_rate": 4.753209184595915e-06, "loss": 3.8828, "step": 2630 }, { "epoch": 0.7160292921074044, "grad_norm": 11.08406488779486, "learning_rate": 4.771289097812331e-06, "loss": 3.8228, "step": 2640 }, { "epoch": 0.7187415242744779, "grad_norm": 13.369279562807007, "learning_rate": 4.789369011028747e-06, "loss": 3.8438, "step": 2650 }, { "epoch": 0.7214537564415514, "grad_norm": 11.28630476161874, "learning_rate": 4.8074489242451635e-06, "loss": 3.9913, "step": 2660 }, { "epoch": 0.7241659886086249, "grad_norm": 14.043871962080951, "learning_rate": 4.8255288374615805e-06, "loss": 3.8245, "step": 2670 }, { "epoch": 0.7268782207756984, "grad_norm": 12.726266714860555, "learning_rate": 4.843608750677997e-06, "loss": 3.8089, "step": 2680 }, { "epoch": 0.7295904529427719, "grad_norm": 14.922017278931872, "learning_rate": 4.861688663894414e-06, "loss": 3.8567, "step": 2690 }, { "epoch": 0.7323026851098454, "grad_norm": 14.019596468106354, "learning_rate": 4.87976857711083e-06, "loss": 3.9095, "step": 2700 }, { "epoch": 0.7350149172769189, "grad_norm": 13.273138155909116, "learning_rate": 4.897848490327246e-06, "loss": 3.8623, "step": 2710 }, { "epoch": 0.7377271494439924, "grad_norm": 12.980035332166532, "learning_rate": 4.9159284035436635e-06, "loss": 4.0381, "step": 2720 }, { "epoch": 0.7404393816110659, "grad_norm": 11.32891111419917, "learning_rate": 4.93400831676008e-06, "loss": 3.8853, "step": 2730 }, { "epoch": 0.7431516137781394, "grad_norm": 11.752654640986231, "learning_rate": 4.952088229976497e-06, "loss": 3.8636, "step": 2740 }, { "epoch": 0.7458638459452129, "grad_norm": 11.170920568261982, "learning_rate": 4.970168143192913e-06, "loss": 3.8466, "step": 2750 }, { "epoch": 0.7485760781122864, "grad_norm": 12.399500034098923, "learning_rate": 4.988248056409329e-06, "loss": 3.9353, "step": 2760 }, { "epoch": 0.7512883102793599, "grad_norm": 13.40944991262581, "learning_rate": 5.006327969625746e-06, "loss": 3.848, "step": 2770 }, { "epoch": 0.7540005424464334, "grad_norm": 10.32788977620122, "learning_rate": 5.024407882842163e-06, "loss": 3.8748, "step": 2780 }, { "epoch": 0.7567127746135069, "grad_norm": 14.210508272202755, "learning_rate": 5.04248779605858e-06, "loss": 3.8164, "step": 2790 }, { "epoch": 0.7594250067805804, "grad_norm": 10.832267340095777, "learning_rate": 5.060567709274996e-06, "loss": 3.902, "step": 2800 }, { "epoch": 0.762137238947654, "grad_norm": 11.316890120967834, "learning_rate": 5.078647622491413e-06, "loss": 3.9181, "step": 2810 }, { "epoch": 0.7648494711147275, "grad_norm": 12.764129017595696, "learning_rate": 5.096727535707829e-06, "loss": 3.8124, "step": 2820 }, { "epoch": 0.767561703281801, "grad_norm": 12.714671675515273, "learning_rate": 5.114807448924246e-06, "loss": 3.7939, "step": 2830 }, { "epoch": 0.7702739354488745, "grad_norm": 11.197122688118688, "learning_rate": 5.132887362140662e-06, "loss": 3.9953, "step": 2840 }, { "epoch": 0.772986167615948, "grad_norm": 10.78788338848926, "learning_rate": 5.150967275357078e-06, "loss": 3.9788, "step": 2850 }, { "epoch": 0.7756983997830215, "grad_norm": 11.082764080142768, "learning_rate": 5.169047188573495e-06, "loss": 3.9037, "step": 2860 }, { "epoch": 0.778410631950095, "grad_norm": 10.986758970983113, "learning_rate": 5.187127101789911e-06, "loss": 3.8449, "step": 2870 }, { "epoch": 0.7811228641171685, "grad_norm": 14.501249928664988, "learning_rate": 5.2052070150063285e-06, "loss": 3.8716, "step": 2880 }, { "epoch": 0.783835096284242, "grad_norm": 12.026704120305002, "learning_rate": 5.223286928222745e-06, "loss": 3.9395, "step": 2890 }, { "epoch": 0.7865473284513155, "grad_norm": 11.435784912398447, "learning_rate": 5.241366841439162e-06, "loss": 3.8216, "step": 2900 }, { "epoch": 0.789259560618389, "grad_norm": 12.177914752453622, "learning_rate": 5.259446754655578e-06, "loss": 3.9007, "step": 2910 }, { "epoch": 0.7919717927854625, "grad_norm": 12.229784399246146, "learning_rate": 5.277526667871995e-06, "loss": 3.9096, "step": 2920 }, { "epoch": 0.794684024952536, "grad_norm": 10.289165636940625, "learning_rate": 5.295606581088411e-06, "loss": 3.8762, "step": 2930 }, { "epoch": 0.7973962571196095, "grad_norm": 12.536165730753236, "learning_rate": 5.3136864943048285e-06, "loss": 3.8316, "step": 2940 }, { "epoch": 0.800108489286683, "grad_norm": 13.825964188634316, "learning_rate": 5.331766407521244e-06, "loss": 3.8703, "step": 2950 }, { "epoch": 0.8028207214537565, "grad_norm": 11.457089244475652, "learning_rate": 5.34984632073766e-06, "loss": 3.7506, "step": 2960 }, { "epoch": 0.80553295362083, "grad_norm": 13.143298424584861, "learning_rate": 5.367926233954077e-06, "loss": 3.9302, "step": 2970 }, { "epoch": 0.8082451857879035, "grad_norm": 10.11749734350807, "learning_rate": 5.3860061471704935e-06, "loss": 3.8539, "step": 2980 }, { "epoch": 0.810957417954977, "grad_norm": 12.337978849844081, "learning_rate": 5.4040860603869106e-06, "loss": 3.8341, "step": 2990 }, { "epoch": 0.8136696501220505, "grad_norm": 13.071513292215652, "learning_rate": 5.422165973603327e-06, "loss": 3.7991, "step": 3000 }, { "epoch": 0.8163818822891239, "grad_norm": 12.391312241259424, "learning_rate": 5.440245886819744e-06, "loss": 3.8917, "step": 3010 }, { "epoch": 0.8190941144561974, "grad_norm": 10.72789704749546, "learning_rate": 5.45832580003616e-06, "loss": 3.9737, "step": 3020 }, { "epoch": 0.8218063466232709, "grad_norm": 10.89618062665714, "learning_rate": 5.476405713252577e-06, "loss": 3.8685, "step": 3030 }, { "epoch": 0.8245185787903444, "grad_norm": 10.629231197455578, "learning_rate": 5.4944856264689935e-06, "loss": 3.875, "step": 3040 }, { "epoch": 0.8272308109574179, "grad_norm": 10.17436809183992, "learning_rate": 5.51256553968541e-06, "loss": 3.9485, "step": 3050 }, { "epoch": 0.8299430431244914, "grad_norm": 11.542915846354557, "learning_rate": 5.530645452901826e-06, "loss": 3.8443, "step": 3060 }, { "epoch": 0.8326552752915649, "grad_norm": 11.01712703720843, "learning_rate": 5.548725366118243e-06, "loss": 3.9139, "step": 3070 }, { "epoch": 0.8353675074586384, "grad_norm": 12.267456973800702, "learning_rate": 5.566805279334659e-06, "loss": 3.7774, "step": 3080 }, { "epoch": 0.8380797396257119, "grad_norm": 11.56476700186551, "learning_rate": 5.584885192551076e-06, "loss": 3.8255, "step": 3090 }, { "epoch": 0.8407919717927854, "grad_norm": 11.744806631405055, "learning_rate": 5.602965105767493e-06, "loss": 3.8808, "step": 3100 }, { "epoch": 0.8435042039598589, "grad_norm": 11.603243141836423, "learning_rate": 5.62104501898391e-06, "loss": 3.8765, "step": 3110 }, { "epoch": 0.8462164361269324, "grad_norm": 11.462541278391337, "learning_rate": 5.639124932200326e-06, "loss": 3.7862, "step": 3120 }, { "epoch": 0.8489286682940059, "grad_norm": 12.319473647844573, "learning_rate": 5.657204845416743e-06, "loss": 3.8062, "step": 3130 }, { "epoch": 0.8516409004610794, "grad_norm": 12.770469608351746, "learning_rate": 5.675284758633159e-06, "loss": 3.8703, "step": 3140 }, { "epoch": 0.8543531326281529, "grad_norm": 12.000919744558267, "learning_rate": 5.6933646718495764e-06, "loss": 3.7807, "step": 3150 }, { "epoch": 0.8570653647952264, "grad_norm": 11.335911379945044, "learning_rate": 5.711444585065992e-06, "loss": 3.9204, "step": 3160 }, { "epoch": 0.8597775969622999, "grad_norm": 10.474273483224678, "learning_rate": 5.729524498282408e-06, "loss": 3.9837, "step": 3170 }, { "epoch": 0.8624898291293734, "grad_norm": 11.184808110887175, "learning_rate": 5.747604411498825e-06, "loss": 3.7977, "step": 3180 }, { "epoch": 0.865202061296447, "grad_norm": 11.695899039815135, "learning_rate": 5.765684324715241e-06, "loss": 3.9197, "step": 3190 }, { "epoch": 0.8679142934635204, "grad_norm": 10.971156100337545, "learning_rate": 5.7837642379316585e-06, "loss": 3.8937, "step": 3200 }, { "epoch": 0.870626525630594, "grad_norm": 11.193448922094877, "learning_rate": 5.801844151148075e-06, "loss": 3.8687, "step": 3210 }, { "epoch": 0.8733387577976675, "grad_norm": 11.708241605508176, "learning_rate": 5.819924064364492e-06, "loss": 3.9562, "step": 3220 }, { "epoch": 0.876050989964741, "grad_norm": 12.456853643890469, "learning_rate": 5.838003977580908e-06, "loss": 3.8762, "step": 3230 }, { "epoch": 0.8787632221318145, "grad_norm": 11.91242375730322, "learning_rate": 5.856083890797325e-06, "loss": 3.8876, "step": 3240 }, { "epoch": 0.881475454298888, "grad_norm": 10.92503112545494, "learning_rate": 5.874163804013741e-06, "loss": 3.8262, "step": 3250 }, { "epoch": 0.8841876864659615, "grad_norm": 12.526546246818437, "learning_rate": 5.8922437172301585e-06, "loss": 3.8972, "step": 3260 }, { "epoch": 0.886899918633035, "grad_norm": 12.382156266473576, "learning_rate": 5.910323630446574e-06, "loss": 3.8219, "step": 3270 }, { "epoch": 0.8896121508001085, "grad_norm": 11.202146992189107, "learning_rate": 5.92840354366299e-06, "loss": 3.9858, "step": 3280 }, { "epoch": 0.892324382967182, "grad_norm": 10.459767335609188, "learning_rate": 5.946483456879407e-06, "loss": 3.8749, "step": 3290 }, { "epoch": 0.8950366151342555, "grad_norm": 11.558975845588217, "learning_rate": 5.9645633700958235e-06, "loss": 3.7737, "step": 3300 }, { "epoch": 0.897748847301329, "grad_norm": 13.113640413270446, "learning_rate": 5.982643283312241e-06, "loss": 3.8556, "step": 3310 }, { "epoch": 0.9004610794684025, "grad_norm": 10.642108628031046, "learning_rate": 6.000723196528657e-06, "loss": 3.9248, "step": 3320 }, { "epoch": 0.903173311635476, "grad_norm": 11.871718759090701, "learning_rate": 6.018803109745074e-06, "loss": 3.8266, "step": 3330 }, { "epoch": 0.9058855438025495, "grad_norm": 11.87982620524846, "learning_rate": 6.03688302296149e-06, "loss": 3.8662, "step": 3340 }, { "epoch": 0.908597775969623, "grad_norm": 12.005118352031447, "learning_rate": 6.054962936177907e-06, "loss": 3.9724, "step": 3350 }, { "epoch": 0.9113100081366965, "grad_norm": 13.62853207710492, "learning_rate": 6.0730428493943235e-06, "loss": 3.7784, "step": 3360 }, { "epoch": 0.91402224030377, "grad_norm": 11.31306850410967, "learning_rate": 6.09112276261074e-06, "loss": 3.7916, "step": 3370 }, { "epoch": 0.9167344724708435, "grad_norm": 10.928478414709017, "learning_rate": 6.109202675827156e-06, "loss": 3.8522, "step": 3380 }, { "epoch": 0.919446704637917, "grad_norm": 10.81930767493218, "learning_rate": 6.127282589043573e-06, "loss": 3.9097, "step": 3390 }, { "epoch": 0.9221589368049905, "grad_norm": 12.704669345271268, "learning_rate": 6.145362502259989e-06, "loss": 3.8771, "step": 3400 }, { "epoch": 0.924871168972064, "grad_norm": 10.312159531348646, "learning_rate": 6.1634424154764064e-06, "loss": 3.9024, "step": 3410 }, { "epoch": 0.9275834011391375, "grad_norm": 11.512243615261779, "learning_rate": 6.181522328692823e-06, "loss": 3.9105, "step": 3420 }, { "epoch": 0.930295633306211, "grad_norm": 11.598847817917097, "learning_rate": 6.19960224190924e-06, "loss": 3.8722, "step": 3430 }, { "epoch": 0.9330078654732845, "grad_norm": 10.933903775197805, "learning_rate": 6.217682155125656e-06, "loss": 3.7957, "step": 3440 }, { "epoch": 0.935720097640358, "grad_norm": 10.740220019013375, "learning_rate": 6.235762068342073e-06, "loss": 3.8105, "step": 3450 }, { "epoch": 0.9384323298074315, "grad_norm": 11.427375754054378, "learning_rate": 6.253841981558489e-06, "loss": 3.8883, "step": 3460 }, { "epoch": 0.941144561974505, "grad_norm": 11.249631394136593, "learning_rate": 6.2719218947749064e-06, "loss": 3.9011, "step": 3470 }, { "epoch": 0.9438567941415785, "grad_norm": 11.847867728459605, "learning_rate": 6.290001807991322e-06, "loss": 3.8657, "step": 3480 }, { "epoch": 0.946569026308652, "grad_norm": 11.472411399025512, "learning_rate": 6.308081721207738e-06, "loss": 3.8902, "step": 3490 }, { "epoch": 0.9492812584757255, "grad_norm": 10.559351744762203, "learning_rate": 6.326161634424155e-06, "loss": 3.8817, "step": 3500 }, { "epoch": 0.951993490642799, "grad_norm": 12.253100895942652, "learning_rate": 6.3442415476405714e-06, "loss": 3.8222, "step": 3510 }, { "epoch": 0.9547057228098725, "grad_norm": 11.416361465259525, "learning_rate": 6.3623214608569885e-06, "loss": 3.7998, "step": 3520 }, { "epoch": 0.957417954976946, "grad_norm": 10.968597823723709, "learning_rate": 6.380401374073405e-06, "loss": 3.8251, "step": 3530 }, { "epoch": 0.9601301871440195, "grad_norm": 10.762830561698085, "learning_rate": 6.398481287289822e-06, "loss": 3.8313, "step": 3540 }, { "epoch": 0.962842419311093, "grad_norm": 12.01246699003259, "learning_rate": 6.416561200506238e-06, "loss": 3.8562, "step": 3550 }, { "epoch": 0.9655546514781665, "grad_norm": 10.438081662392598, "learning_rate": 6.434641113722655e-06, "loss": 3.9064, "step": 3560 }, { "epoch": 0.96826688364524, "grad_norm": 10.909397627530705, "learning_rate": 6.4527210269390714e-06, "loss": 3.9339, "step": 3570 }, { "epoch": 0.9709791158123136, "grad_norm": 11.55618457077656, "learning_rate": 6.4708009401554885e-06, "loss": 3.8522, "step": 3580 }, { "epoch": 0.9736913479793871, "grad_norm": 10.915341204142443, "learning_rate": 6.488880853371904e-06, "loss": 3.8318, "step": 3590 }, { "epoch": 0.9764035801464606, "grad_norm": 12.318572282379572, "learning_rate": 6.50696076658832e-06, "loss": 3.9451, "step": 3600 }, { "epoch": 0.9791158123135341, "grad_norm": 11.061169882076046, "learning_rate": 6.525040679804737e-06, "loss": 3.8756, "step": 3610 }, { "epoch": 0.9818280444806076, "grad_norm": 11.317835219898091, "learning_rate": 6.5431205930211535e-06, "loss": 3.9096, "step": 3620 }, { "epoch": 0.9845402766476811, "grad_norm": 10.546731094178764, "learning_rate": 6.561200506237571e-06, "loss": 3.8319, "step": 3630 }, { "epoch": 0.9872525088147546, "grad_norm": 14.34985524826246, "learning_rate": 6.579280419453987e-06, "loss": 3.8596, "step": 3640 }, { "epoch": 0.9899647409818281, "grad_norm": 11.2472263657756, "learning_rate": 6.597360332670404e-06, "loss": 3.8284, "step": 3650 }, { "epoch": 0.9926769731489016, "grad_norm": 10.87930698706768, "learning_rate": 6.61544024588682e-06, "loss": 3.9277, "step": 3660 }, { "epoch": 0.9953892053159751, "grad_norm": 11.246422738680595, "learning_rate": 6.633520159103237e-06, "loss": 3.7726, "step": 3670 }, { "epoch": 0.9981014374830486, "grad_norm": 10.908648330010463, "learning_rate": 6.6516000723196535e-06, "loss": 3.8719, "step": 3680 }, { "epoch": 1.000813669650122, "grad_norm": 9.798123764684597, "learning_rate": 6.66967998553607e-06, "loss": 3.8518, "step": 3690 }, { "epoch": 1.0035259018171956, "grad_norm": 11.330294570755793, "learning_rate": 6.687759898752486e-06, "loss": 3.6606, "step": 3700 }, { "epoch": 1.006238133984269, "grad_norm": 11.35513661127942, "learning_rate": 6.705839811968903e-06, "loss": 3.7916, "step": 3710 }, { "epoch": 1.0089503661513426, "grad_norm": 11.858706161335341, "learning_rate": 6.723919725185319e-06, "loss": 3.7427, "step": 3720 }, { "epoch": 1.011662598318416, "grad_norm": 12.964004975624764, "learning_rate": 6.7419996384017365e-06, "loss": 3.6508, "step": 3730 }, { "epoch": 1.0143748304854896, "grad_norm": 11.00466366037197, "learning_rate": 6.760079551618153e-06, "loss": 3.7436, "step": 3740 }, { "epoch": 1.017087062652563, "grad_norm": 10.696194728411998, "learning_rate": 6.77815946483457e-06, "loss": 3.7012, "step": 3750 }, { "epoch": 1.0197992948196366, "grad_norm": 10.967201330748885, "learning_rate": 6.796239378050986e-06, "loss": 3.7913, "step": 3760 }, { "epoch": 1.02251152698671, "grad_norm": 10.966549942920784, "learning_rate": 6.814319291267403e-06, "loss": 3.7498, "step": 3770 }, { "epoch": 1.0252237591537836, "grad_norm": 12.214356201996573, "learning_rate": 6.832399204483819e-06, "loss": 3.6254, "step": 3780 }, { "epoch": 1.027935991320857, "grad_norm": 12.026852005818608, "learning_rate": 6.8504791177002365e-06, "loss": 3.7462, "step": 3790 }, { "epoch": 1.0306482234879306, "grad_norm": 12.32268658559643, "learning_rate": 6.868559030916652e-06, "loss": 3.6595, "step": 3800 }, { "epoch": 1.033360455655004, "grad_norm": 13.494583249916213, "learning_rate": 6.886638944133068e-06, "loss": 3.7049, "step": 3810 }, { "epoch": 1.0360726878220776, "grad_norm": 10.765966458838351, "learning_rate": 6.904718857349485e-06, "loss": 3.739, "step": 3820 }, { "epoch": 1.038784919989151, "grad_norm": 12.0308046264097, "learning_rate": 6.9227987705659015e-06, "loss": 3.7772, "step": 3830 }, { "epoch": 1.0414971521562246, "grad_norm": 13.240674278595574, "learning_rate": 6.9408786837823185e-06, "loss": 3.8111, "step": 3840 }, { "epoch": 1.044209384323298, "grad_norm": 11.074818586991242, "learning_rate": 6.958958596998735e-06, "loss": 3.9116, "step": 3850 }, { "epoch": 1.0469216164903716, "grad_norm": 11.203644189602882, "learning_rate": 6.977038510215152e-06, "loss": 3.7272, "step": 3860 }, { "epoch": 1.049633848657445, "grad_norm": 13.05323335208992, "learning_rate": 6.995118423431568e-06, "loss": 3.7541, "step": 3870 }, { "epoch": 1.0523460808245186, "grad_norm": 11.880778298041832, "learning_rate": 7.013198336647985e-06, "loss": 3.6316, "step": 3880 }, { "epoch": 1.055058312991592, "grad_norm": 12.583684667366613, "learning_rate": 7.0312782498644015e-06, "loss": 3.7102, "step": 3890 }, { "epoch": 1.0577705451586656, "grad_norm": 10.98180200999189, "learning_rate": 7.0493581630808186e-06, "loss": 3.7546, "step": 3900 }, { "epoch": 1.060482777325739, "grad_norm": 10.183081367189304, "learning_rate": 7.067438076297234e-06, "loss": 3.7373, "step": 3910 }, { "epoch": 1.0631950094928126, "grad_norm": 10.155447557489635, "learning_rate": 7.08551798951365e-06, "loss": 3.8109, "step": 3920 }, { "epoch": 1.065907241659886, "grad_norm": 10.758051674108895, "learning_rate": 7.103597902730067e-06, "loss": 3.8145, "step": 3930 }, { "epoch": 1.0686194738269597, "grad_norm": 11.227698839445354, "learning_rate": 7.1216778159464835e-06, "loss": 3.7663, "step": 3940 }, { "epoch": 1.071331705994033, "grad_norm": 11.043419776815735, "learning_rate": 7.139757729162901e-06, "loss": 3.7825, "step": 3950 }, { "epoch": 1.0740439381611067, "grad_norm": 10.93761496981384, "learning_rate": 7.157837642379317e-06, "loss": 3.7882, "step": 3960 }, { "epoch": 1.07675617032818, "grad_norm": 11.113324851051482, "learning_rate": 7.175917555595734e-06, "loss": 3.7719, "step": 3970 }, { "epoch": 1.0794684024952537, "grad_norm": 10.969897510278006, "learning_rate": 7.19399746881215e-06, "loss": 3.7017, "step": 3980 }, { "epoch": 1.082180634662327, "grad_norm": 11.503079554957422, "learning_rate": 7.212077382028567e-06, "loss": 3.7117, "step": 3990 }, { "epoch": 1.0848928668294007, "grad_norm": 11.725262877001525, "learning_rate": 7.2301572952449836e-06, "loss": 3.7385, "step": 4000 }, { "epoch": 1.087605098996474, "grad_norm": 12.125693138614068, "learning_rate": 7.2482372084614e-06, "loss": 3.7286, "step": 4010 }, { "epoch": 1.0903173311635477, "grad_norm": 10.052331041724933, "learning_rate": 7.266317121677816e-06, "loss": 3.7659, "step": 4020 }, { "epoch": 1.093029563330621, "grad_norm": 11.474839894904266, "learning_rate": 7.284397034894233e-06, "loss": 3.6826, "step": 4030 }, { "epoch": 1.0957417954976947, "grad_norm": 11.563756904147676, "learning_rate": 7.302476948110649e-06, "loss": 3.7938, "step": 4040 }, { "epoch": 1.098454027664768, "grad_norm": 12.039911432388173, "learning_rate": 7.3205568613270665e-06, "loss": 3.7975, "step": 4050 }, { "epoch": 1.1011662598318417, "grad_norm": 12.0399651566778, "learning_rate": 7.338636774543483e-06, "loss": 3.6104, "step": 4060 }, { "epoch": 1.103878491998915, "grad_norm": 11.07451239652187, "learning_rate": 7.3567166877599e-06, "loss": 3.823, "step": 4070 }, { "epoch": 1.1065907241659887, "grad_norm": 11.209294602366404, "learning_rate": 7.374796600976316e-06, "loss": 3.7649, "step": 4080 }, { "epoch": 1.109302956333062, "grad_norm": 11.231914002199149, "learning_rate": 7.392876514192733e-06, "loss": 3.8025, "step": 4090 }, { "epoch": 1.1120151885001357, "grad_norm": 10.904406615279868, "learning_rate": 7.410956427409149e-06, "loss": 3.7837, "step": 4100 }, { "epoch": 1.114727420667209, "grad_norm": 10.945707078226178, "learning_rate": 7.4290363406255665e-06, "loss": 3.7664, "step": 4110 }, { "epoch": 1.1174396528342827, "grad_norm": 10.33029858181482, "learning_rate": 7.447116253841982e-06, "loss": 3.6613, "step": 4120 }, { "epoch": 1.120151885001356, "grad_norm": 11.428191585537133, "learning_rate": 7.465196167058398e-06, "loss": 3.6935, "step": 4130 }, { "epoch": 1.1228641171684297, "grad_norm": 10.691476956745225, "learning_rate": 7.483276080274815e-06, "loss": 3.7193, "step": 4140 }, { "epoch": 1.125576349335503, "grad_norm": 9.638343698948212, "learning_rate": 7.5013559934912315e-06, "loss": 3.7826, "step": 4150 }, { "epoch": 1.1282885815025767, "grad_norm": 12.774336021276158, "learning_rate": 7.5194359067076486e-06, "loss": 3.7173, "step": 4160 }, { "epoch": 1.13100081366965, "grad_norm": 11.394459654564558, "learning_rate": 7.537515819924065e-06, "loss": 3.6297, "step": 4170 }, { "epoch": 1.1337130458367237, "grad_norm": 10.711590416823018, "learning_rate": 7.555595733140482e-06, "loss": 3.6935, "step": 4180 }, { "epoch": 1.1364252780037971, "grad_norm": 11.126418287985162, "learning_rate": 7.573675646356898e-06, "loss": 3.7648, "step": 4190 }, { "epoch": 1.1391375101708707, "grad_norm": 9.69306301899103, "learning_rate": 7.591755559573315e-06, "loss": 3.8762, "step": 4200 }, { "epoch": 1.1418497423379441, "grad_norm": 10.555927433042044, "learning_rate": 7.6098354727897315e-06, "loss": 3.7947, "step": 4210 }, { "epoch": 1.1445619745050175, "grad_norm": 10.911173498648287, "learning_rate": 7.627915386006149e-06, "loss": 3.7808, "step": 4220 }, { "epoch": 1.1472742066720911, "grad_norm": 11.080840880961098, "learning_rate": 7.645995299222563e-06, "loss": 3.6991, "step": 4230 }, { "epoch": 1.1499864388391647, "grad_norm": 10.236508861257624, "learning_rate": 7.66407521243898e-06, "loss": 3.748, "step": 4240 }, { "epoch": 1.1526986710062381, "grad_norm": 10.670381085500223, "learning_rate": 7.682155125655397e-06, "loss": 3.7719, "step": 4250 }, { "epoch": 1.1554109031733115, "grad_norm": 11.086662766429049, "learning_rate": 7.700235038871814e-06, "loss": 3.7394, "step": 4260 }, { "epoch": 1.1581231353403851, "grad_norm": 12.527615349075148, "learning_rate": 7.71831495208823e-06, "loss": 3.7827, "step": 4270 }, { "epoch": 1.1608353675074587, "grad_norm": 10.52216219482226, "learning_rate": 7.736394865304647e-06, "loss": 3.7489, "step": 4280 }, { "epoch": 1.1635475996745321, "grad_norm": 11.423814559294847, "learning_rate": 7.754474778521064e-06, "loss": 3.7115, "step": 4290 }, { "epoch": 1.1662598318416055, "grad_norm": 11.28980024839117, "learning_rate": 7.772554691737481e-06, "loss": 3.7329, "step": 4300 }, { "epoch": 1.1689720640086791, "grad_norm": 11.230853392850149, "learning_rate": 7.790634604953896e-06, "loss": 3.6796, "step": 4310 }, { "epoch": 1.1716842961757528, "grad_norm": 11.317693765701632, "learning_rate": 7.808714518170314e-06, "loss": 3.7671, "step": 4320 }, { "epoch": 1.1743965283428262, "grad_norm": 12.122308825793272, "learning_rate": 7.826794431386729e-06, "loss": 3.7634, "step": 4330 }, { "epoch": 1.1771087605098995, "grad_norm": 10.398634452836472, "learning_rate": 7.844874344603146e-06, "loss": 3.7346, "step": 4340 }, { "epoch": 1.1798209926769732, "grad_norm": 11.68370672906272, "learning_rate": 7.862954257819563e-06, "loss": 3.7077, "step": 4350 }, { "epoch": 1.1825332248440468, "grad_norm": 11.100382090060014, "learning_rate": 7.88103417103598e-06, "loss": 3.6966, "step": 4360 }, { "epoch": 1.1852454570111202, "grad_norm": 10.36251805400384, "learning_rate": 7.899114084252396e-06, "loss": 3.681, "step": 4370 }, { "epoch": 1.1879576891781936, "grad_norm": 9.702750216871296, "learning_rate": 7.917193997468813e-06, "loss": 3.7537, "step": 4380 }, { "epoch": 1.1906699213452672, "grad_norm": 12.08067109278004, "learning_rate": 7.93527391068523e-06, "loss": 3.734, "step": 4390 }, { "epoch": 1.1933821535123406, "grad_norm": 10.264143043942823, "learning_rate": 7.953353823901647e-06, "loss": 3.8635, "step": 4400 }, { "epoch": 1.1960943856794142, "grad_norm": 10.794196286694742, "learning_rate": 7.971433737118062e-06, "loss": 3.6818, "step": 4410 }, { "epoch": 1.1988066178464876, "grad_norm": 10.958036287615686, "learning_rate": 7.98951365033448e-06, "loss": 3.791, "step": 4420 }, { "epoch": 1.2015188500135612, "grad_norm": 10.043069479619303, "learning_rate": 8.007593563550897e-06, "loss": 3.7815, "step": 4430 }, { "epoch": 1.2042310821806346, "grad_norm": 11.545481510042931, "learning_rate": 8.025673476767312e-06, "loss": 3.717, "step": 4440 }, { "epoch": 1.2069433143477082, "grad_norm": 11.561008218532047, "learning_rate": 8.043753389983729e-06, "loss": 3.8105, "step": 4450 }, { "epoch": 1.2096555465147816, "grad_norm": 11.806617258261014, "learning_rate": 8.061833303200144e-06, "loss": 3.7817, "step": 4460 }, { "epoch": 1.2123677786818552, "grad_norm": 10.023662231601683, "learning_rate": 8.079913216416561e-06, "loss": 3.7222, "step": 4470 }, { "epoch": 1.2150800108489286, "grad_norm": 10.994504467523782, "learning_rate": 8.097993129632979e-06, "loss": 3.8084, "step": 4480 }, { "epoch": 1.2177922430160022, "grad_norm": 11.24278484278038, "learning_rate": 8.116073042849396e-06, "loss": 3.6735, "step": 4490 }, { "epoch": 1.2205044751830756, "grad_norm": 10.921899922652464, "learning_rate": 8.134152956065811e-06, "loss": 3.8142, "step": 4500 }, { "epoch": 1.2232167073501492, "grad_norm": 12.058901653987665, "learning_rate": 8.152232869282228e-06, "loss": 3.6977, "step": 4510 }, { "epoch": 1.2259289395172226, "grad_norm": 11.454973422644745, "learning_rate": 8.170312782498645e-06, "loss": 3.8189, "step": 4520 }, { "epoch": 1.2286411716842962, "grad_norm": 10.156967381247352, "learning_rate": 8.188392695715062e-06, "loss": 3.7352, "step": 4530 }, { "epoch": 1.2313534038513696, "grad_norm": 11.420540867782037, "learning_rate": 8.206472608931478e-06, "loss": 3.7095, "step": 4540 }, { "epoch": 1.2340656360184432, "grad_norm": 11.729311667818148, "learning_rate": 8.224552522147893e-06, "loss": 3.7824, "step": 4550 }, { "epoch": 1.2367778681855166, "grad_norm": 10.397293114958282, "learning_rate": 8.24263243536431e-06, "loss": 3.741, "step": 4560 }, { "epoch": 1.2394901003525902, "grad_norm": 11.307340425981543, "learning_rate": 8.260712348580727e-06, "loss": 3.6208, "step": 4570 }, { "epoch": 1.2422023325196636, "grad_norm": 11.22767113619351, "learning_rate": 8.278792261797144e-06, "loss": 3.7017, "step": 4580 }, { "epoch": 1.2449145646867372, "grad_norm": 11.875632624668798, "learning_rate": 8.29687217501356e-06, "loss": 3.7884, "step": 4590 }, { "epoch": 1.2476267968538106, "grad_norm": 11.083336896058244, "learning_rate": 8.314952088229977e-06, "loss": 3.756, "step": 4600 }, { "epoch": 1.2503390290208842, "grad_norm": 13.834981813802047, "learning_rate": 8.333032001446394e-06, "loss": 3.7597, "step": 4610 }, { "epoch": 1.2530512611879576, "grad_norm": 10.403441255662347, "learning_rate": 8.351111914662811e-06, "loss": 3.7848, "step": 4620 }, { "epoch": 1.2557634933550312, "grad_norm": 11.024432327844368, "learning_rate": 8.369191827879227e-06, "loss": 3.7427, "step": 4630 }, { "epoch": 1.2584757255221046, "grad_norm": 10.859328110134767, "learning_rate": 8.387271741095644e-06, "loss": 3.6839, "step": 4640 }, { "epoch": 1.2611879576891782, "grad_norm": 11.274965444153624, "learning_rate": 8.405351654312059e-06, "loss": 3.7753, "step": 4650 }, { "epoch": 1.2639001898562516, "grad_norm": 10.43181347434775, "learning_rate": 8.423431567528476e-06, "loss": 3.7752, "step": 4660 }, { "epoch": 1.2666124220233252, "grad_norm": 10.920323141973757, "learning_rate": 8.441511480744893e-06, "loss": 3.6464, "step": 4670 }, { "epoch": 1.2693246541903986, "grad_norm": 14.082616841905221, "learning_rate": 8.45959139396131e-06, "loss": 3.6485, "step": 4680 }, { "epoch": 1.2720368863574723, "grad_norm": 12.77629694047138, "learning_rate": 8.477671307177726e-06, "loss": 3.8182, "step": 4690 }, { "epoch": 1.2747491185245456, "grad_norm": 11.164158775254085, "learning_rate": 8.495751220394143e-06, "loss": 3.6392, "step": 4700 }, { "epoch": 1.2774613506916193, "grad_norm": 10.431053863589845, "learning_rate": 8.51383113361056e-06, "loss": 3.679, "step": 4710 }, { "epoch": 1.2801735828586926, "grad_norm": 9.987456529592675, "learning_rate": 8.531911046826977e-06, "loss": 3.811, "step": 4720 }, { "epoch": 1.2828858150257663, "grad_norm": 10.832113140707932, "learning_rate": 8.549990960043392e-06, "loss": 3.7681, "step": 4730 }, { "epoch": 1.2855980471928397, "grad_norm": 9.347785049398839, "learning_rate": 8.56807087325981e-06, "loss": 3.7883, "step": 4740 }, { "epoch": 1.2883102793599133, "grad_norm": 10.928563504055852, "learning_rate": 8.586150786476227e-06, "loss": 3.6248, "step": 4750 }, { "epoch": 1.2910225115269867, "grad_norm": 10.978384803299235, "learning_rate": 8.604230699692642e-06, "loss": 3.7827, "step": 4760 }, { "epoch": 1.2937347436940603, "grad_norm": 11.016143483188108, "learning_rate": 8.622310612909059e-06, "loss": 3.6963, "step": 4770 }, { "epoch": 1.2964469758611337, "grad_norm": 12.797028110250281, "learning_rate": 8.640390526125474e-06, "loss": 3.726, "step": 4780 }, { "epoch": 1.2991592080282073, "grad_norm": 10.190821096790112, "learning_rate": 8.658470439341892e-06, "loss": 3.8058, "step": 4790 }, { "epoch": 1.3018714401952807, "grad_norm": 11.054924311833146, "learning_rate": 8.676550352558309e-06, "loss": 3.7374, "step": 4800 }, { "epoch": 1.3045836723623543, "grad_norm": 10.593974878811935, "learning_rate": 8.694630265774726e-06, "loss": 3.7679, "step": 4810 }, { "epoch": 1.3072959045294277, "grad_norm": 11.278550835542873, "learning_rate": 8.712710178991141e-06, "loss": 3.6704, "step": 4820 }, { "epoch": 1.3100081366965013, "grad_norm": 10.40323790206987, "learning_rate": 8.730790092207558e-06, "loss": 3.8059, "step": 4830 }, { "epoch": 1.3127203688635747, "grad_norm": 10.322732494963619, "learning_rate": 8.748870005423975e-06, "loss": 3.8964, "step": 4840 }, { "epoch": 1.3154326010306483, "grad_norm": 11.968428637377265, "learning_rate": 8.766949918640392e-06, "loss": 3.7392, "step": 4850 }, { "epoch": 1.3181448331977217, "grad_norm": 9.600544747153286, "learning_rate": 8.785029831856808e-06, "loss": 3.7997, "step": 4860 }, { "epoch": 1.3208570653647953, "grad_norm": 10.73819038296125, "learning_rate": 8.803109745073223e-06, "loss": 3.7462, "step": 4870 }, { "epoch": 1.3235692975318687, "grad_norm": 10.018281655193094, "learning_rate": 8.82118965828964e-06, "loss": 3.7517, "step": 4880 }, { "epoch": 1.3262815296989423, "grad_norm": 11.136942601564886, "learning_rate": 8.839269571506057e-06, "loss": 3.7243, "step": 4890 }, { "epoch": 1.3289937618660157, "grad_norm": 10.697421528457879, "learning_rate": 8.857349484722474e-06, "loss": 3.6908, "step": 4900 }, { "epoch": 1.3317059940330893, "grad_norm": 11.039438149635536, "learning_rate": 8.87542939793889e-06, "loss": 3.6751, "step": 4910 }, { "epoch": 1.3344182262001627, "grad_norm": 11.498279703063547, "learning_rate": 8.893509311155307e-06, "loss": 3.7261, "step": 4920 }, { "epoch": 1.3371304583672363, "grad_norm": 10.01866447920304, "learning_rate": 8.911589224371724e-06, "loss": 3.8447, "step": 4930 }, { "epoch": 1.3398426905343097, "grad_norm": 11.332561099598884, "learning_rate": 8.929669137588141e-06, "loss": 3.7987, "step": 4940 }, { "epoch": 1.342554922701383, "grad_norm": 11.172421894471505, "learning_rate": 8.947749050804557e-06, "loss": 3.8074, "step": 4950 }, { "epoch": 1.3452671548684567, "grad_norm": 9.545014889898086, "learning_rate": 8.965828964020974e-06, "loss": 3.7121, "step": 4960 }, { "epoch": 1.3479793870355303, "grad_norm": 11.397310641956896, "learning_rate": 8.983908877237389e-06, "loss": 3.651, "step": 4970 }, { "epoch": 1.3506916192026037, "grad_norm": 11.405839813807962, "learning_rate": 9.001988790453806e-06, "loss": 3.695, "step": 4980 }, { "epoch": 1.3534038513696771, "grad_norm": 11.216439943763065, "learning_rate": 9.020068703670223e-06, "loss": 3.6464, "step": 4990 }, { "epoch": 1.3561160835367507, "grad_norm": 10.235823270448314, "learning_rate": 9.03814861688664e-06, "loss": 3.8656, "step": 5000 }, { "epoch": 1.3588283157038243, "grad_norm": 10.015377673233223, "learning_rate": 9.056228530103056e-06, "loss": 3.7656, "step": 5010 }, { "epoch": 1.3615405478708977, "grad_norm": 10.797417740099464, "learning_rate": 9.074308443319473e-06, "loss": 3.7314, "step": 5020 }, { "epoch": 1.3642527800379711, "grad_norm": 11.46850025578586, "learning_rate": 9.09238835653589e-06, "loss": 3.7048, "step": 5030 }, { "epoch": 1.3669650122050447, "grad_norm": 10.798281435294742, "learning_rate": 9.110468269752307e-06, "loss": 3.749, "step": 5040 }, { "epoch": 1.3696772443721184, "grad_norm": 10.5593729478773, "learning_rate": 9.128548182968722e-06, "loss": 3.7367, "step": 5050 }, { "epoch": 1.3723894765391917, "grad_norm": 10.775307605765093, "learning_rate": 9.14662809618514e-06, "loss": 3.8255, "step": 5060 }, { "epoch": 1.3751017087062651, "grad_norm": 11.671836329127208, "learning_rate": 9.164708009401557e-06, "loss": 3.6875, "step": 5070 }, { "epoch": 1.3778139408733387, "grad_norm": 11.690336656054072, "learning_rate": 9.182787922617972e-06, "loss": 3.7413, "step": 5080 }, { "epoch": 1.3805261730404124, "grad_norm": 9.778851302235964, "learning_rate": 9.200867835834389e-06, "loss": 3.81, "step": 5090 }, { "epoch": 1.3832384052074858, "grad_norm": 10.732182224594617, "learning_rate": 9.218947749050804e-06, "loss": 3.8125, "step": 5100 }, { "epoch": 1.3859506373745591, "grad_norm": 10.275728852498137, "learning_rate": 9.237027662267222e-06, "loss": 3.7776, "step": 5110 }, { "epoch": 1.3886628695416328, "grad_norm": 10.742071442514563, "learning_rate": 9.255107575483639e-06, "loss": 3.6496, "step": 5120 }, { "epoch": 1.3913751017087064, "grad_norm": 11.025431582893285, "learning_rate": 9.273187488700056e-06, "loss": 3.7528, "step": 5130 }, { "epoch": 1.3940873338757798, "grad_norm": 10.40511446840101, "learning_rate": 9.291267401916471e-06, "loss": 3.7953, "step": 5140 }, { "epoch": 1.3967995660428532, "grad_norm": 9.918132860891761, "learning_rate": 9.309347315132888e-06, "loss": 3.7702, "step": 5150 }, { "epoch": 1.3995117982099268, "grad_norm": 11.48770586169003, "learning_rate": 9.327427228349305e-06, "loss": 3.6879, "step": 5160 }, { "epoch": 1.4022240303770004, "grad_norm": 9.923345806038597, "learning_rate": 9.345507141565722e-06, "loss": 3.7578, "step": 5170 }, { "epoch": 1.4049362625440738, "grad_norm": 11.196811053562133, "learning_rate": 9.363587054782138e-06, "loss": 3.7022, "step": 5180 }, { "epoch": 1.4076484947111472, "grad_norm": 10.718213954452898, "learning_rate": 9.381666967998553e-06, "loss": 3.6743, "step": 5190 }, { "epoch": 1.4103607268782208, "grad_norm": 11.427111758052625, "learning_rate": 9.39974688121497e-06, "loss": 3.6871, "step": 5200 }, { "epoch": 1.4130729590452944, "grad_norm": 12.390442213589965, "learning_rate": 9.417826794431387e-06, "loss": 3.6957, "step": 5210 }, { "epoch": 1.4157851912123678, "grad_norm": 11.425472376866987, "learning_rate": 9.435906707647804e-06, "loss": 3.7867, "step": 5220 }, { "epoch": 1.4184974233794412, "grad_norm": 10.916452732081893, "learning_rate": 9.45398662086422e-06, "loss": 3.7807, "step": 5230 }, { "epoch": 1.4212096555465148, "grad_norm": 11.388042059321435, "learning_rate": 9.472066534080637e-06, "loss": 3.6485, "step": 5240 }, { "epoch": 1.4239218877135884, "grad_norm": 11.596946393507668, "learning_rate": 9.490146447297054e-06, "loss": 3.7187, "step": 5250 }, { "epoch": 1.4266341198806618, "grad_norm": 9.223057871291557, "learning_rate": 9.508226360513471e-06, "loss": 3.7781, "step": 5260 }, { "epoch": 1.4293463520477352, "grad_norm": 11.12466046604056, "learning_rate": 9.526306273729887e-06, "loss": 3.7161, "step": 5270 }, { "epoch": 1.4320585842148088, "grad_norm": 9.916920550472875, "learning_rate": 9.544386186946304e-06, "loss": 3.7749, "step": 5280 }, { "epoch": 1.4347708163818824, "grad_norm": 9.837873500255544, "learning_rate": 9.562466100162719e-06, "loss": 3.6448, "step": 5290 }, { "epoch": 1.4374830485489558, "grad_norm": 10.268763445450904, "learning_rate": 9.580546013379136e-06, "loss": 3.6149, "step": 5300 }, { "epoch": 1.4401952807160292, "grad_norm": 12.885307923761378, "learning_rate": 9.598625926595553e-06, "loss": 3.6487, "step": 5310 }, { "epoch": 1.4429075128831028, "grad_norm": 10.488270217746019, "learning_rate": 9.61670583981197e-06, "loss": 3.751, "step": 5320 }, { "epoch": 1.4456197450501764, "grad_norm": 10.275870701182678, "learning_rate": 9.634785753028386e-06, "loss": 3.6893, "step": 5330 }, { "epoch": 1.4483319772172498, "grad_norm": 12.780006886503534, "learning_rate": 9.652865666244803e-06, "loss": 3.7971, "step": 5340 }, { "epoch": 1.4510442093843232, "grad_norm": 11.231231777754825, "learning_rate": 9.67094557946122e-06, "loss": 3.7553, "step": 5350 }, { "epoch": 1.4537564415513968, "grad_norm": 10.457778794923447, "learning_rate": 9.689025492677637e-06, "loss": 3.7168, "step": 5360 }, { "epoch": 1.4564686737184702, "grad_norm": 10.495928968396719, "learning_rate": 9.707105405894052e-06, "loss": 3.6642, "step": 5370 }, { "epoch": 1.4591809058855438, "grad_norm": 10.20057921946344, "learning_rate": 9.72518531911047e-06, "loss": 3.7841, "step": 5380 }, { "epoch": 1.4618931380526172, "grad_norm": 10.448318613403961, "learning_rate": 9.743265232326887e-06, "loss": 3.7883, "step": 5390 }, { "epoch": 1.4646053702196908, "grad_norm": 11.337790483537821, "learning_rate": 9.761345145543302e-06, "loss": 3.751, "step": 5400 }, { "epoch": 1.4673176023867642, "grad_norm": 10.195392779391037, "learning_rate": 9.779425058759719e-06, "loss": 3.8022, "step": 5410 }, { "epoch": 1.4700298345538378, "grad_norm": 11.277573516594822, "learning_rate": 9.797504971976134e-06, "loss": 3.7462, "step": 5420 }, { "epoch": 1.4727420667209112, "grad_norm": 10.287536160824914, "learning_rate": 9.815584885192552e-06, "loss": 3.6779, "step": 5430 }, { "epoch": 1.4754542988879848, "grad_norm": 9.90033419633516, "learning_rate": 9.833664798408969e-06, "loss": 3.7228, "step": 5440 }, { "epoch": 1.4781665310550582, "grad_norm": 10.55380925100051, "learning_rate": 9.851744711625386e-06, "loss": 3.6731, "step": 5450 }, { "epoch": 1.4808787632221319, "grad_norm": 10.015395725797951, "learning_rate": 9.869824624841801e-06, "loss": 3.6495, "step": 5460 }, { "epoch": 1.4835909953892052, "grad_norm": 9.691136738128764, "learning_rate": 9.887904538058218e-06, "loss": 3.6997, "step": 5470 }, { "epoch": 1.4863032275562789, "grad_norm": 10.143369947026784, "learning_rate": 9.905984451274635e-06, "loss": 3.8288, "step": 5480 }, { "epoch": 1.4890154597233523, "grad_norm": 11.85029494040246, "learning_rate": 9.924064364491052e-06, "loss": 3.6763, "step": 5490 }, { "epoch": 1.4917276918904259, "grad_norm": 11.195175188668333, "learning_rate": 9.942144277707468e-06, "loss": 3.7057, "step": 5500 }, { "epoch": 1.4944399240574993, "grad_norm": 10.723341607013733, "learning_rate": 9.960224190923883e-06, "loss": 3.7514, "step": 5510 }, { "epoch": 1.4971521562245729, "grad_norm": 11.001701390872864, "learning_rate": 9.9783041041403e-06, "loss": 3.8733, "step": 5520 }, { "epoch": 1.4998643883916463, "grad_norm": 11.004843759879867, "learning_rate": 9.996384017356717e-06, "loss": 3.6338, "step": 5530 }, { "epoch": 1.5025766205587199, "grad_norm": 11.428778441279238, "learning_rate": 9.999999362596224e-06, "loss": 3.7917, "step": 5540 }, { "epoch": 1.5052888527257933, "grad_norm": 10.16506414765493, "learning_rate": 9.999996773143663e-06, "loss": 3.8566, "step": 5550 }, { "epoch": 1.5080010848928669, "grad_norm": 9.589080099708125, "learning_rate": 9.999992191805612e-06, "loss": 3.6815, "step": 5560 }, { "epoch": 1.5107133170599405, "grad_norm": 11.397413179516727, "learning_rate": 9.999985618583894e-06, "loss": 3.7327, "step": 5570 }, { "epoch": 1.5134255492270139, "grad_norm": 10.32888185076621, "learning_rate": 9.99997705348113e-06, "loss": 3.8466, "step": 5580 }, { "epoch": 1.5161377813940873, "grad_norm": 10.416422678421183, "learning_rate": 9.99996649650073e-06, "loss": 3.7171, "step": 5590 }, { "epoch": 1.5188500135611607, "grad_norm": 11.592571679196201, "learning_rate": 9.999953947646903e-06, "loss": 3.7434, "step": 5600 }, { "epoch": 1.5215622457282343, "grad_norm": 12.002857382952216, "learning_rate": 9.999939406924645e-06, "loss": 3.7716, "step": 5610 }, { "epoch": 1.524274477895308, "grad_norm": 10.557364262138561, "learning_rate": 9.999922874339749e-06, "loss": 3.8011, "step": 5620 }, { "epoch": 1.5269867100623813, "grad_norm": 10.181956125976306, "learning_rate": 9.999904349898804e-06, "loss": 3.8043, "step": 5630 }, { "epoch": 1.5296989422294547, "grad_norm": 10.650033309930544, "learning_rate": 9.999883833609185e-06, "loss": 3.8134, "step": 5640 }, { "epoch": 1.5324111743965283, "grad_norm": 11.051739634270133, "learning_rate": 9.99986132547907e-06, "loss": 3.7503, "step": 5650 }, { "epoch": 1.535123406563602, "grad_norm": 11.329242745796922, "learning_rate": 9.999836825517423e-06, "loss": 3.6408, "step": 5660 }, { "epoch": 1.5378356387306753, "grad_norm": 9.4872171553482, "learning_rate": 9.999810333734006e-06, "loss": 3.8054, "step": 5670 }, { "epoch": 1.5405478708977487, "grad_norm": 9.277167302329932, "learning_rate": 9.999781850139371e-06, "loss": 3.8565, "step": 5680 }, { "epoch": 1.5432601030648223, "grad_norm": 10.146401369972558, "learning_rate": 9.999751374744867e-06, "loss": 3.6287, "step": 5690 }, { "epoch": 1.545972335231896, "grad_norm": 11.495012836927643, "learning_rate": 9.99971890756263e-06, "loss": 3.6998, "step": 5700 }, { "epoch": 1.5486845673989693, "grad_norm": 10.12645286455962, "learning_rate": 9.9996844486056e-06, "loss": 3.7559, "step": 5710 }, { "epoch": 1.5513967995660427, "grad_norm": 11.402607819939783, "learning_rate": 9.999647997887503e-06, "loss": 3.8399, "step": 5720 }, { "epoch": 1.5541090317331163, "grad_norm": 10.169855108864976, "learning_rate": 9.99960955542286e-06, "loss": 3.7977, "step": 5730 }, { "epoch": 1.55682126390019, "grad_norm": 11.2406515415903, "learning_rate": 9.999569121226983e-06, "loss": 3.7999, "step": 5740 }, { "epoch": 1.5595334960672633, "grad_norm": 10.708036067230426, "learning_rate": 9.999526695315984e-06, "loss": 3.7296, "step": 5750 }, { "epoch": 1.5622457282343367, "grad_norm": 9.124294738113532, "learning_rate": 9.999482277706762e-06, "loss": 3.7592, "step": 5760 }, { "epoch": 1.5649579604014103, "grad_norm": 10.123659269342975, "learning_rate": 9.999435868417014e-06, "loss": 3.7381, "step": 5770 }, { "epoch": 1.567670192568484, "grad_norm": 10.553976767052996, "learning_rate": 9.999387467465224e-06, "loss": 3.8008, "step": 5780 }, { "epoch": 1.5703824247355573, "grad_norm": 10.084294798895911, "learning_rate": 9.99933707487068e-06, "loss": 3.7897, "step": 5790 }, { "epoch": 1.5730946569026307, "grad_norm": 10.35290236792001, "learning_rate": 9.999284690653452e-06, "loss": 3.794, "step": 5800 }, { "epoch": 1.5758068890697043, "grad_norm": 10.96880379565306, "learning_rate": 9.999230314834414e-06, "loss": 3.7597, "step": 5810 }, { "epoch": 1.578519121236778, "grad_norm": 11.677382528498402, "learning_rate": 9.999173947435224e-06, "loss": 3.7134, "step": 5820 }, { "epoch": 1.5812313534038513, "grad_norm": 11.600581452330355, "learning_rate": 9.999115588478337e-06, "loss": 3.7493, "step": 5830 }, { "epoch": 1.5839435855709247, "grad_norm": 10.204094874208836, "learning_rate": 9.999055237987002e-06, "loss": 3.6992, "step": 5840 }, { "epoch": 1.5866558177379984, "grad_norm": 10.41715993080021, "learning_rate": 9.998992895985267e-06, "loss": 3.8152, "step": 5850 }, { "epoch": 1.589368049905072, "grad_norm": 10.244422217173314, "learning_rate": 9.99892856249796e-06, "loss": 3.8009, "step": 5860 }, { "epoch": 1.5920802820721454, "grad_norm": 9.934463449614945, "learning_rate": 9.998862237550715e-06, "loss": 3.6712, "step": 5870 }, { "epoch": 1.5947925142392188, "grad_norm": 11.719419572640572, "learning_rate": 9.998793921169954e-06, "loss": 3.7271, "step": 5880 }, { "epoch": 1.5975047464062924, "grad_norm": 10.699268653575228, "learning_rate": 9.998723613382887e-06, "loss": 3.6686, "step": 5890 }, { "epoch": 1.600216978573366, "grad_norm": 10.227548595592754, "learning_rate": 9.998651314217531e-06, "loss": 3.7056, "step": 5900 }, { "epoch": 1.6029292107404394, "grad_norm": 9.167263409623525, "learning_rate": 9.998577023702685e-06, "loss": 3.7995, "step": 5910 }, { "epoch": 1.6056414429075128, "grad_norm": 11.323060155254497, "learning_rate": 9.998500741867943e-06, "loss": 3.8251, "step": 5920 }, { "epoch": 1.6083536750745864, "grad_norm": 9.198570393581404, "learning_rate": 9.998422468743695e-06, "loss": 3.7772, "step": 5930 }, { "epoch": 1.61106590724166, "grad_norm": 9.541426758140537, "learning_rate": 9.998342204361126e-06, "loss": 3.6937, "step": 5940 }, { "epoch": 1.6137781394087334, "grad_norm": 10.410569897120702, "learning_rate": 9.998259948752207e-06, "loss": 3.8243, "step": 5950 }, { "epoch": 1.6164903715758068, "grad_norm": 10.025120207156514, "learning_rate": 9.998175701949709e-06, "loss": 3.6639, "step": 5960 }, { "epoch": 1.6192026037428804, "grad_norm": 10.995608726309161, "learning_rate": 9.998089463987194e-06, "loss": 3.6504, "step": 5970 }, { "epoch": 1.621914835909954, "grad_norm": 9.150147692119972, "learning_rate": 9.998001234899017e-06, "loss": 3.7061, "step": 5980 }, { "epoch": 1.6246270680770274, "grad_norm": 10.314503472147571, "learning_rate": 9.997911014720328e-06, "loss": 3.8121, "step": 5990 }, { "epoch": 1.6273393002441008, "grad_norm": 10.504396521395732, "learning_rate": 9.997818803487067e-06, "loss": 3.7279, "step": 6000 }, { "epoch": 1.6300515324111744, "grad_norm": 9.732631436939013, "learning_rate": 9.997724601235968e-06, "loss": 3.646, "step": 6010 }, { "epoch": 1.632763764578248, "grad_norm": 12.155345668425214, "learning_rate": 9.997628408004562e-06, "loss": 3.8227, "step": 6020 }, { "epoch": 1.6354759967453214, "grad_norm": 9.66229147223548, "learning_rate": 9.997530223831166e-06, "loss": 3.6983, "step": 6030 }, { "epoch": 1.6381882289123948, "grad_norm": 10.322200756176509, "learning_rate": 9.997430048754899e-06, "loss": 3.7005, "step": 6040 }, { "epoch": 1.6409004610794684, "grad_norm": 9.814392638986387, "learning_rate": 9.997327882815667e-06, "loss": 3.7785, "step": 6050 }, { "epoch": 1.643612693246542, "grad_norm": 9.534793575212174, "learning_rate": 9.997223726054167e-06, "loss": 3.7216, "step": 6060 }, { "epoch": 1.6463249254136154, "grad_norm": 10.238078539556966, "learning_rate": 9.997117578511896e-06, "loss": 3.7861, "step": 6070 }, { "epoch": 1.6490371575806888, "grad_norm": 10.385199917968801, "learning_rate": 9.997009440231143e-06, "loss": 3.7889, "step": 6080 }, { "epoch": 1.6517493897477624, "grad_norm": 10.6728058789742, "learning_rate": 9.996899311254982e-06, "loss": 3.6949, "step": 6090 }, { "epoch": 1.654461621914836, "grad_norm": 10.19404406390347, "learning_rate": 9.996787191627291e-06, "loss": 3.8023, "step": 6100 }, { "epoch": 1.6571738540819094, "grad_norm": 9.628312727684255, "learning_rate": 9.996673081392734e-06, "loss": 3.7279, "step": 6110 }, { "epoch": 1.6598860862489828, "grad_norm": 9.967511896175251, "learning_rate": 9.99655698059677e-06, "loss": 3.7597, "step": 6120 }, { "epoch": 1.6625983184160564, "grad_norm": 9.838464879802347, "learning_rate": 9.99643888928565e-06, "loss": 3.7054, "step": 6130 }, { "epoch": 1.66531055058313, "grad_norm": 11.615271669474192, "learning_rate": 9.99631880750642e-06, "loss": 3.8393, "step": 6140 }, { "epoch": 1.6680227827502034, "grad_norm": 9.852816549409146, "learning_rate": 9.996196735306918e-06, "loss": 3.8173, "step": 6150 }, { "epoch": 1.6707350149172768, "grad_norm": 11.138787592683917, "learning_rate": 9.996072672735775e-06, "loss": 3.7356, "step": 6160 }, { "epoch": 1.6734472470843504, "grad_norm": 11.020853654220547, "learning_rate": 9.995946619842413e-06, "loss": 3.7638, "step": 6170 }, { "epoch": 1.676159479251424, "grad_norm": 10.798065289943311, "learning_rate": 9.99581857667705e-06, "loss": 3.8113, "step": 6180 }, { "epoch": 1.6788717114184974, "grad_norm": 9.453782960449443, "learning_rate": 9.995688543290695e-06, "loss": 3.9342, "step": 6190 }, { "epoch": 1.6815839435855708, "grad_norm": 9.625308517197212, "learning_rate": 9.995556519735152e-06, "loss": 3.7719, "step": 6200 }, { "epoch": 1.6842961757526445, "grad_norm": 11.794932227604816, "learning_rate": 9.995422506063013e-06, "loss": 3.729, "step": 6210 }, { "epoch": 1.687008407919718, "grad_norm": 10.08136630839966, "learning_rate": 9.995286502327667e-06, "loss": 3.7882, "step": 6220 }, { "epoch": 1.6897206400867915, "grad_norm": 10.196419581026273, "learning_rate": 9.9951485085833e-06, "loss": 3.7455, "step": 6230 }, { "epoch": 1.6924328722538649, "grad_norm": 9.42579636199513, "learning_rate": 9.995008524884878e-06, "loss": 3.8194, "step": 6240 }, { "epoch": 1.6951451044209385, "grad_norm": 9.400696141258749, "learning_rate": 9.99486655128817e-06, "loss": 3.8003, "step": 6250 }, { "epoch": 1.697857336588012, "grad_norm": 8.997829490965419, "learning_rate": 9.994722587849737e-06, "loss": 3.736, "step": 6260 }, { "epoch": 1.7005695687550855, "grad_norm": 10.08758978048927, "learning_rate": 9.994576634626929e-06, "loss": 3.7859, "step": 6270 }, { "epoch": 1.7032818009221589, "grad_norm": 11.493111797069837, "learning_rate": 9.994428691677891e-06, "loss": 3.7204, "step": 6280 }, { "epoch": 1.7059940330892325, "grad_norm": 10.028101523583999, "learning_rate": 9.99427875906156e-06, "loss": 3.8396, "step": 6290 }, { "epoch": 1.708706265256306, "grad_norm": 9.175178605406396, "learning_rate": 9.994126836837665e-06, "loss": 3.724, "step": 6300 }, { "epoch": 1.7114184974233795, "grad_norm": 10.770585444402144, "learning_rate": 9.993972925066731e-06, "loss": 3.7623, "step": 6310 }, { "epoch": 1.7141307295904529, "grad_norm": 11.163673961438784, "learning_rate": 9.993817023810068e-06, "loss": 3.7362, "step": 6320 }, { "epoch": 1.7168429617575265, "grad_norm": 9.353779079726237, "learning_rate": 9.99365913312979e-06, "loss": 3.7036, "step": 6330 }, { "epoch": 1.7195551939246, "grad_norm": 9.75659946145151, "learning_rate": 9.993499253088792e-06, "loss": 3.704, "step": 6340 }, { "epoch": 1.7222674260916735, "grad_norm": 10.076598795164875, "learning_rate": 9.993337383750768e-06, "loss": 3.7533, "step": 6350 }, { "epoch": 1.7249796582587469, "grad_norm": 12.192112733919261, "learning_rate": 9.993173525180204e-06, "loss": 3.6555, "step": 6360 }, { "epoch": 1.7276918904258205, "grad_norm": 9.52275313404009, "learning_rate": 9.993007677442377e-06, "loss": 3.7958, "step": 6370 }, { "epoch": 1.7304041225928941, "grad_norm": 10.16726140114274, "learning_rate": 9.992839840603356e-06, "loss": 3.7523, "step": 6380 }, { "epoch": 1.7331163547599675, "grad_norm": 9.340669684763402, "learning_rate": 9.992670014730004e-06, "loss": 3.7159, "step": 6390 }, { "epoch": 1.735828586927041, "grad_norm": 9.89002938644933, "learning_rate": 9.992498199889978e-06, "loss": 3.682, "step": 6400 }, { "epoch": 1.7385408190941145, "grad_norm": 10.419510438668452, "learning_rate": 9.99232439615172e-06, "loss": 3.7422, "step": 6410 }, { "epoch": 1.7412530512611881, "grad_norm": 9.823726080425285, "learning_rate": 9.992148603584475e-06, "loss": 3.6754, "step": 6420 }, { "epoch": 1.7439652834282615, "grad_norm": 10.779815591733746, "learning_rate": 9.991970822258272e-06, "loss": 3.7498, "step": 6430 }, { "epoch": 1.746677515595335, "grad_norm": 10.28750065549012, "learning_rate": 9.991791052243937e-06, "loss": 3.7302, "step": 6440 }, { "epoch": 1.7493897477624083, "grad_norm": 9.941182419005072, "learning_rate": 9.991609293613082e-06, "loss": 3.777, "step": 6450 }, { "epoch": 1.752101979929482, "grad_norm": 9.80422881826482, "learning_rate": 9.991425546438121e-06, "loss": 3.76, "step": 6460 }, { "epoch": 1.7548142120965555, "grad_norm": 9.970765566898464, "learning_rate": 9.99123981079225e-06, "loss": 3.7093, "step": 6470 }, { "epoch": 1.757526444263629, "grad_norm": 10.917666442780964, "learning_rate": 9.991052086749464e-06, "loss": 3.7674, "step": 6480 }, { "epoch": 1.7602386764307023, "grad_norm": 10.29539986878929, "learning_rate": 9.990862374384548e-06, "loss": 3.7723, "step": 6490 }, { "epoch": 1.762950908597776, "grad_norm": 10.57822684509711, "learning_rate": 9.99067067377308e-06, "loss": 3.8104, "step": 6500 }, { "epoch": 1.7656631407648495, "grad_norm": 9.52147839982269, "learning_rate": 9.990476984991429e-06, "loss": 3.7898, "step": 6510 }, { "epoch": 1.768375372931923, "grad_norm": 9.9705797741706, "learning_rate": 9.990281308116754e-06, "loss": 3.7438, "step": 6520 }, { "epoch": 1.7710876050989963, "grad_norm": 9.204688200950647, "learning_rate": 9.99008364322701e-06, "loss": 3.7937, "step": 6530 }, { "epoch": 1.77379983726607, "grad_norm": 11.026303456959635, "learning_rate": 9.989883990400942e-06, "loss": 3.7481, "step": 6540 }, { "epoch": 1.7765120694331435, "grad_norm": 10.402718724667409, "learning_rate": 9.989682349718085e-06, "loss": 3.7616, "step": 6550 }, { "epoch": 1.779224301600217, "grad_norm": 8.833065538242865, "learning_rate": 9.989478721258773e-06, "loss": 3.7361, "step": 6560 }, { "epoch": 1.7819365337672903, "grad_norm": 9.840961507742868, "learning_rate": 9.989273105104123e-06, "loss": 3.7683, "step": 6570 }, { "epoch": 1.784648765934364, "grad_norm": 9.404249471187812, "learning_rate": 9.989065501336048e-06, "loss": 3.6158, "step": 6580 }, { "epoch": 1.7873609981014376, "grad_norm": 10.561672120928208, "learning_rate": 9.988855910037256e-06, "loss": 3.7864, "step": 6590 }, { "epoch": 1.790073230268511, "grad_norm": 10.681453705990164, "learning_rate": 9.98864433129124e-06, "loss": 3.7925, "step": 6600 }, { "epoch": 1.7927854624355843, "grad_norm": 9.568301135003823, "learning_rate": 9.988430765182288e-06, "loss": 3.7372, "step": 6610 }, { "epoch": 1.795497694602658, "grad_norm": 10.280514023215494, "learning_rate": 9.98821521179548e-06, "loss": 3.8058, "step": 6620 }, { "epoch": 1.7982099267697316, "grad_norm": 10.031128959973039, "learning_rate": 9.987997671216692e-06, "loss": 3.772, "step": 6630 }, { "epoch": 1.800922158936805, "grad_norm": 10.251796302736118, "learning_rate": 9.987778143532583e-06, "loss": 3.7668, "step": 6640 }, { "epoch": 1.8036343911038784, "grad_norm": 9.084877041282393, "learning_rate": 9.987556628830606e-06, "loss": 3.7928, "step": 6650 }, { "epoch": 1.806346623270952, "grad_norm": 9.90395956624837, "learning_rate": 9.987333127199013e-06, "loss": 3.752, "step": 6660 }, { "epoch": 1.8090588554380256, "grad_norm": 9.034105444918191, "learning_rate": 9.987107638726838e-06, "loss": 3.728, "step": 6670 }, { "epoch": 1.811771087605099, "grad_norm": 10.279755257388318, "learning_rate": 9.986880163503913e-06, "loss": 3.7947, "step": 6680 }, { "epoch": 1.8144833197721724, "grad_norm": 10.341374443627032, "learning_rate": 9.986650701620857e-06, "loss": 3.6991, "step": 6690 }, { "epoch": 1.817195551939246, "grad_norm": 10.80918946652945, "learning_rate": 9.986419253169081e-06, "loss": 3.6982, "step": 6700 }, { "epoch": 1.8199077841063196, "grad_norm": 11.671734329361655, "learning_rate": 9.986185818240795e-06, "loss": 3.7018, "step": 6710 }, { "epoch": 1.822620016273393, "grad_norm": 10.309645644664105, "learning_rate": 9.985950396928987e-06, "loss": 3.7662, "step": 6720 }, { "epoch": 1.8253322484404664, "grad_norm": 10.936204426654431, "learning_rate": 9.985712989327448e-06, "loss": 3.7654, "step": 6730 }, { "epoch": 1.82804448060754, "grad_norm": 9.344315474936113, "learning_rate": 9.985473595530753e-06, "loss": 3.7346, "step": 6740 }, { "epoch": 1.8307567127746136, "grad_norm": 10.177197571795503, "learning_rate": 9.985232215634273e-06, "loss": 3.8722, "step": 6750 }, { "epoch": 1.833468944941687, "grad_norm": 10.654351367246818, "learning_rate": 9.984988849734168e-06, "loss": 3.6369, "step": 6760 }, { "epoch": 1.8361811771087604, "grad_norm": 9.380257553265057, "learning_rate": 9.98474349792739e-06, "loss": 3.8065, "step": 6770 }, { "epoch": 1.838893409275834, "grad_norm": 9.55367148986276, "learning_rate": 9.984496160311681e-06, "loss": 3.7499, "step": 6780 }, { "epoch": 1.8416056414429076, "grad_norm": 10.04911757098543, "learning_rate": 9.984246836985574e-06, "loss": 3.6812, "step": 6790 }, { "epoch": 1.844317873609981, "grad_norm": 9.657265016892563, "learning_rate": 9.983995528048394e-06, "loss": 3.6655, "step": 6800 }, { "epoch": 1.8470301057770544, "grad_norm": 9.960130135994008, "learning_rate": 9.983742233600256e-06, "loss": 3.7426, "step": 6810 }, { "epoch": 1.849742337944128, "grad_norm": 10.826252024105024, "learning_rate": 9.98348695374207e-06, "loss": 3.698, "step": 6820 }, { "epoch": 1.8524545701112016, "grad_norm": 9.592366001716606, "learning_rate": 9.983229688575531e-06, "loss": 3.7827, "step": 6830 }, { "epoch": 1.855166802278275, "grad_norm": 10.107628600825535, "learning_rate": 9.982970438203127e-06, "loss": 3.7304, "step": 6840 }, { "epoch": 1.8578790344453484, "grad_norm": 9.00923240453071, "learning_rate": 9.98270920272814e-06, "loss": 3.635, "step": 6850 }, { "epoch": 1.860591266612422, "grad_norm": 9.156380772288612, "learning_rate": 9.982445982254638e-06, "loss": 3.6395, "step": 6860 }, { "epoch": 1.8633034987794956, "grad_norm": 10.143291348573653, "learning_rate": 9.982180776887485e-06, "loss": 3.6727, "step": 6870 }, { "epoch": 1.866015730946569, "grad_norm": 9.903918316255401, "learning_rate": 9.981913586732329e-06, "loss": 3.8011, "step": 6880 }, { "epoch": 1.8687279631136424, "grad_norm": 9.759692692933308, "learning_rate": 9.981644411895615e-06, "loss": 3.6857, "step": 6890 }, { "epoch": 1.871440195280716, "grad_norm": 9.4353685880923, "learning_rate": 9.981373252484574e-06, "loss": 3.8006, "step": 6900 }, { "epoch": 1.8741524274477896, "grad_norm": 10.363214006193564, "learning_rate": 9.981100108607234e-06, "loss": 3.7623, "step": 6910 }, { "epoch": 1.876864659614863, "grad_norm": 10.096136727778262, "learning_rate": 9.980824980372406e-06, "loss": 3.7523, "step": 6920 }, { "epoch": 1.8795768917819364, "grad_norm": 9.897387589250945, "learning_rate": 9.980547867889694e-06, "loss": 3.7349, "step": 6930 }, { "epoch": 1.88228912394901, "grad_norm": 9.293746813665958, "learning_rate": 9.980268771269495e-06, "loss": 3.7216, "step": 6940 }, { "epoch": 1.8850013561160837, "grad_norm": 9.672288678636805, "learning_rate": 9.979987690622996e-06, "loss": 3.7018, "step": 6950 }, { "epoch": 1.887713588283157, "grad_norm": 9.364746743259836, "learning_rate": 9.979704626062171e-06, "loss": 3.7395, "step": 6960 }, { "epoch": 1.8904258204502304, "grad_norm": 9.921462208502103, "learning_rate": 9.979419577699787e-06, "loss": 3.6519, "step": 6970 }, { "epoch": 1.893138052617304, "grad_norm": 10.110560820942222, "learning_rate": 9.979132545649402e-06, "loss": 3.7194, "step": 6980 }, { "epoch": 1.8958502847843777, "grad_norm": 9.053651228980396, "learning_rate": 9.978843530025362e-06, "loss": 3.7142, "step": 6990 }, { "epoch": 1.898562516951451, "grad_norm": 8.912442245106822, "learning_rate": 9.978552530942804e-06, "loss": 3.7526, "step": 7000 }, { "epoch": 1.9012747491185245, "grad_norm": 10.283410421735962, "learning_rate": 9.978259548517656e-06, "loss": 3.6921, "step": 7010 }, { "epoch": 1.903986981285598, "grad_norm": 10.815770987762495, "learning_rate": 9.977964582866635e-06, "loss": 3.7625, "step": 7020 }, { "epoch": 1.9066992134526717, "grad_norm": 10.05267187707207, "learning_rate": 9.97766763410725e-06, "loss": 3.6881, "step": 7030 }, { "epoch": 1.909411445619745, "grad_norm": 9.637704433469391, "learning_rate": 9.977368702357797e-06, "loss": 3.6951, "step": 7040 }, { "epoch": 1.9121236777868185, "grad_norm": 9.155923819668526, "learning_rate": 9.977067787737366e-06, "loss": 3.744, "step": 7050 }, { "epoch": 1.914835909953892, "grad_norm": 9.125851195933297, "learning_rate": 9.97676489036583e-06, "loss": 3.7607, "step": 7060 }, { "epoch": 1.9175481421209657, "grad_norm": 10.044633624206845, "learning_rate": 9.976460010363862e-06, "loss": 3.8425, "step": 7070 }, { "epoch": 1.920260374288039, "grad_norm": 9.313160239241439, "learning_rate": 9.976153147852915e-06, "loss": 3.7025, "step": 7080 }, { "epoch": 1.9229726064551125, "grad_norm": 8.948637970757336, "learning_rate": 9.975844302955239e-06, "loss": 3.679, "step": 7090 }, { "epoch": 1.925684838622186, "grad_norm": 9.004606828143215, "learning_rate": 9.97553347579387e-06, "loss": 3.8257, "step": 7100 }, { "epoch": 1.9283970707892597, "grad_norm": 10.434827694230572, "learning_rate": 9.975220666492634e-06, "loss": 3.6705, "step": 7110 }, { "epoch": 1.931109302956333, "grad_norm": 9.181511915499941, "learning_rate": 9.974905875176146e-06, "loss": 3.6639, "step": 7120 }, { "epoch": 1.9338215351234065, "grad_norm": 9.736321334703938, "learning_rate": 9.974589101969813e-06, "loss": 3.7529, "step": 7130 }, { "epoch": 1.93653376729048, "grad_norm": 9.165765734407383, "learning_rate": 9.974270346999832e-06, "loss": 3.7495, "step": 7140 }, { "epoch": 1.9392459994575537, "grad_norm": 8.743402991250289, "learning_rate": 9.973949610393186e-06, "loss": 3.761, "step": 7150 }, { "epoch": 1.941958231624627, "grad_norm": 9.459147991530394, "learning_rate": 9.973626892277647e-06, "loss": 3.6822, "step": 7160 }, { "epoch": 1.9446704637917005, "grad_norm": 9.662707897781592, "learning_rate": 9.973302192781782e-06, "loss": 3.7054, "step": 7170 }, { "epoch": 1.9473826959587741, "grad_norm": 9.065622328664764, "learning_rate": 9.972975512034943e-06, "loss": 3.756, "step": 7180 }, { "epoch": 1.9500949281258477, "grad_norm": 11.5671610444023, "learning_rate": 9.972646850167272e-06, "loss": 3.7231, "step": 7190 }, { "epoch": 1.9528071602929211, "grad_norm": 9.806320712745649, "learning_rate": 9.9723162073097e-06, "loss": 3.7521, "step": 7200 }, { "epoch": 1.9555193924599945, "grad_norm": 10.586203021837838, "learning_rate": 9.971983583593948e-06, "loss": 3.6466, "step": 7210 }, { "epoch": 1.9582316246270681, "grad_norm": 10.421458695268827, "learning_rate": 9.971648979152526e-06, "loss": 3.809, "step": 7220 }, { "epoch": 1.9609438567941417, "grad_norm": 11.412307835693195, "learning_rate": 9.97131239411873e-06, "loss": 3.7136, "step": 7230 }, { "epoch": 1.9636560889612151, "grad_norm": 11.692630279729093, "learning_rate": 9.970973828626653e-06, "loss": 3.6402, "step": 7240 }, { "epoch": 1.9663683211282885, "grad_norm": 9.689949800436377, "learning_rate": 9.970633282811171e-06, "loss": 3.7881, "step": 7250 }, { "epoch": 1.9690805532953621, "grad_norm": 9.380294331322336, "learning_rate": 9.970290756807946e-06, "loss": 3.7759, "step": 7260 }, { "epoch": 1.9717927854624357, "grad_norm": 10.10033136418225, "learning_rate": 9.969946250753433e-06, "loss": 3.6919, "step": 7270 }, { "epoch": 1.9745050176295091, "grad_norm": 9.815525544394267, "learning_rate": 9.96959976478488e-06, "loss": 3.7105, "step": 7280 }, { "epoch": 1.9772172497965825, "grad_norm": 8.929409449904606, "learning_rate": 9.969251299040314e-06, "loss": 3.7882, "step": 7290 }, { "epoch": 1.979929481963656, "grad_norm": 9.686483380280356, "learning_rate": 9.968900853658561e-06, "loss": 3.693, "step": 7300 }, { "epoch": 1.9826417141307295, "grad_norm": 10.160481721597176, "learning_rate": 9.968548428779224e-06, "loss": 3.758, "step": 7310 }, { "epoch": 1.9853539462978032, "grad_norm": 9.139557082449334, "learning_rate": 9.968194024542706e-06, "loss": 3.7131, "step": 7320 }, { "epoch": 1.9880661784648765, "grad_norm": 10.108900567210624, "learning_rate": 9.967837641090194e-06, "loss": 3.8425, "step": 7330 }, { "epoch": 1.99077841063195, "grad_norm": 10.283137034109744, "learning_rate": 9.96747927856366e-06, "loss": 3.7244, "step": 7340 }, { "epoch": 1.9934906427990235, "grad_norm": 10.14101489594113, "learning_rate": 9.96711893710587e-06, "loss": 3.6232, "step": 7350 }, { "epoch": 1.9962028749660972, "grad_norm": 8.654194871763018, "learning_rate": 9.96675661686037e-06, "loss": 3.8081, "step": 7360 }, { "epoch": 1.9989151071331706, "grad_norm": 10.13330549277211, "learning_rate": 9.966392317971511e-06, "loss": 3.7067, "step": 7370 }, { "epoch": 2.001627339300244, "grad_norm": 9.13542667200247, "learning_rate": 9.966026040584412e-06, "loss": 3.5784, "step": 7380 }, { "epoch": 2.004339571467318, "grad_norm": 11.987169890687785, "learning_rate": 9.965657784844994e-06, "loss": 3.2708, "step": 7390 }, { "epoch": 2.007051803634391, "grad_norm": 10.318922438153235, "learning_rate": 9.965287550899962e-06, "loss": 3.2915, "step": 7400 }, { "epoch": 2.0097640358014646, "grad_norm": 11.027342509546054, "learning_rate": 9.964915338896806e-06, "loss": 3.3603, "step": 7410 }, { "epoch": 2.012476267968538, "grad_norm": 11.246300781941542, "learning_rate": 9.964541148983808e-06, "loss": 3.2969, "step": 7420 }, { "epoch": 2.015188500135612, "grad_norm": 10.722984105340931, "learning_rate": 9.964164981310038e-06, "loss": 3.4482, "step": 7430 }, { "epoch": 2.017900732302685, "grad_norm": 9.499491638224583, "learning_rate": 9.96378683602535e-06, "loss": 3.3673, "step": 7440 }, { "epoch": 2.0206129644697586, "grad_norm": 10.749367783266761, "learning_rate": 9.963406713280394e-06, "loss": 3.3571, "step": 7450 }, { "epoch": 2.023325196636832, "grad_norm": 9.180608795405549, "learning_rate": 9.963024613226597e-06, "loss": 3.3401, "step": 7460 }, { "epoch": 2.026037428803906, "grad_norm": 9.814574479620168, "learning_rate": 9.962640536016179e-06, "loss": 3.3948, "step": 7470 }, { "epoch": 2.028749660970979, "grad_norm": 10.83048616477613, "learning_rate": 9.962254481802151e-06, "loss": 3.4034, "step": 7480 }, { "epoch": 2.0314618931380526, "grad_norm": 10.113814962901321, "learning_rate": 9.961866450738307e-06, "loss": 3.4076, "step": 7490 }, { "epoch": 2.034174125305126, "grad_norm": 10.800697232470654, "learning_rate": 9.96147644297923e-06, "loss": 3.4069, "step": 7500 }, { "epoch": 2.0368863574722, "grad_norm": 10.886715988286236, "learning_rate": 9.961084458680287e-06, "loss": 3.47, "step": 7510 }, { "epoch": 2.039598589639273, "grad_norm": 9.703912531259055, "learning_rate": 9.960690497997641e-06, "loss": 3.4172, "step": 7520 }, { "epoch": 2.0423108218063466, "grad_norm": 9.953619956934233, "learning_rate": 9.960294561088233e-06, "loss": 3.4575, "step": 7530 }, { "epoch": 2.04502305397342, "grad_norm": 10.557291297114917, "learning_rate": 9.959896648109797e-06, "loss": 3.3437, "step": 7540 }, { "epoch": 2.047735286140494, "grad_norm": 10.506241775870533, "learning_rate": 9.959496759220853e-06, "loss": 3.4301, "step": 7550 }, { "epoch": 2.050447518307567, "grad_norm": 10.188685575284435, "learning_rate": 9.959094894580704e-06, "loss": 3.369, "step": 7560 }, { "epoch": 2.0531597504746406, "grad_norm": 10.063341579549862, "learning_rate": 9.95869105434945e-06, "loss": 3.3077, "step": 7570 }, { "epoch": 2.055871982641714, "grad_norm": 10.380240571915156, "learning_rate": 9.958285238687965e-06, "loss": 3.5214, "step": 7580 }, { "epoch": 2.058584214808788, "grad_norm": 10.986519506746532, "learning_rate": 9.957877447757922e-06, "loss": 3.3804, "step": 7590 }, { "epoch": 2.0612964469758612, "grad_norm": 11.721394918450745, "learning_rate": 9.957467681721774e-06, "loss": 3.4743, "step": 7600 }, { "epoch": 2.0640086791429346, "grad_norm": 10.661363105968636, "learning_rate": 9.957055940742764e-06, "loss": 3.3823, "step": 7610 }, { "epoch": 2.066720911310008, "grad_norm": 9.509047638870625, "learning_rate": 9.956642224984916e-06, "loss": 3.4391, "step": 7620 }, { "epoch": 2.069433143477082, "grad_norm": 10.323606367670724, "learning_rate": 9.95622653461305e-06, "loss": 3.3343, "step": 7630 }, { "epoch": 2.0721453756441552, "grad_norm": 9.768452968799643, "learning_rate": 9.955808869792762e-06, "loss": 3.4563, "step": 7640 }, { "epoch": 2.0748576078112286, "grad_norm": 9.880019840915178, "learning_rate": 9.955389230690446e-06, "loss": 3.4013, "step": 7650 }, { "epoch": 2.077569839978302, "grad_norm": 10.540100869572727, "learning_rate": 9.954967617473273e-06, "loss": 3.4976, "step": 7660 }, { "epoch": 2.0802820721453754, "grad_norm": 9.4847105034066, "learning_rate": 9.954544030309205e-06, "loss": 3.5477, "step": 7670 }, { "epoch": 2.0829943043124493, "grad_norm": 11.001700685440756, "learning_rate": 9.954118469366992e-06, "loss": 3.3587, "step": 7680 }, { "epoch": 2.0857065364795226, "grad_norm": 9.47657791488699, "learning_rate": 9.953690934816164e-06, "loss": 3.5451, "step": 7690 }, { "epoch": 2.088418768646596, "grad_norm": 10.760994951223797, "learning_rate": 9.953261426827043e-06, "loss": 3.3382, "step": 7700 }, { "epoch": 2.09113100081367, "grad_norm": 10.496101690857975, "learning_rate": 9.952829945570736e-06, "loss": 3.4342, "step": 7710 }, { "epoch": 2.0938432329807433, "grad_norm": 9.513293589262233, "learning_rate": 9.952396491219133e-06, "loss": 3.5564, "step": 7720 }, { "epoch": 2.0965554651478167, "grad_norm": 10.15331716094316, "learning_rate": 9.951961063944915e-06, "loss": 3.4204, "step": 7730 }, { "epoch": 2.09926769731489, "grad_norm": 9.98768804256586, "learning_rate": 9.951523663921545e-06, "loss": 3.3224, "step": 7740 }, { "epoch": 2.1019799294819634, "grad_norm": 12.440853798508165, "learning_rate": 9.951084291323274e-06, "loss": 3.4368, "step": 7750 }, { "epoch": 2.1046921616490373, "grad_norm": 9.871288951137494, "learning_rate": 9.950642946325135e-06, "loss": 3.4618, "step": 7760 }, { "epoch": 2.1074043938161107, "grad_norm": 11.302215437546193, "learning_rate": 9.950199629102955e-06, "loss": 3.3541, "step": 7770 }, { "epoch": 2.110116625983184, "grad_norm": 10.980548139712099, "learning_rate": 9.949754339833339e-06, "loss": 3.4236, "step": 7780 }, { "epoch": 2.1128288581502574, "grad_norm": 9.661729290925104, "learning_rate": 9.949307078693678e-06, "loss": 3.4527, "step": 7790 }, { "epoch": 2.1155410903173313, "grad_norm": 10.316268343774704, "learning_rate": 9.948857845862154e-06, "loss": 3.3884, "step": 7800 }, { "epoch": 2.1182533224844047, "grad_norm": 9.509785394186839, "learning_rate": 9.948406641517732e-06, "loss": 3.3515, "step": 7810 }, { "epoch": 2.120965554651478, "grad_norm": 11.761780735871874, "learning_rate": 9.947953465840157e-06, "loss": 3.4155, "step": 7820 }, { "epoch": 2.1236777868185515, "grad_norm": 10.252335274890898, "learning_rate": 9.947498319009967e-06, "loss": 3.3435, "step": 7830 }, { "epoch": 2.1263900189856253, "grad_norm": 9.361790437899543, "learning_rate": 9.947041201208482e-06, "loss": 3.3728, "step": 7840 }, { "epoch": 2.1291022511526987, "grad_norm": 10.770139464212892, "learning_rate": 9.946582112617806e-06, "loss": 3.3304, "step": 7850 }, { "epoch": 2.131814483319772, "grad_norm": 10.004778034247783, "learning_rate": 9.946121053420832e-06, "loss": 3.3291, "step": 7860 }, { "epoch": 2.1345267154868455, "grad_norm": 9.601917145354768, "learning_rate": 9.945658023801235e-06, "loss": 3.4182, "step": 7870 }, { "epoch": 2.1372389476539193, "grad_norm": 9.650395123503687, "learning_rate": 9.945193023943472e-06, "loss": 3.3966, "step": 7880 }, { "epoch": 2.1399511798209927, "grad_norm": 9.769049241010425, "learning_rate": 9.944726054032793e-06, "loss": 3.3201, "step": 7890 }, { "epoch": 2.142663411988066, "grad_norm": 9.742810091366463, "learning_rate": 9.944257114255225e-06, "loss": 3.5166, "step": 7900 }, { "epoch": 2.1453756441551395, "grad_norm": 9.820705601716355, "learning_rate": 9.943786204797586e-06, "loss": 3.448, "step": 7910 }, { "epoch": 2.1480878763222133, "grad_norm": 11.506005145085979, "learning_rate": 9.943313325847472e-06, "loss": 3.2802, "step": 7920 }, { "epoch": 2.1508001084892867, "grad_norm": 9.42975674711527, "learning_rate": 9.94283847759327e-06, "loss": 3.4306, "step": 7930 }, { "epoch": 2.15351234065636, "grad_norm": 10.668705622558932, "learning_rate": 9.942361660224149e-06, "loss": 3.4768, "step": 7940 }, { "epoch": 2.1562245728234335, "grad_norm": 10.412627255394328, "learning_rate": 9.941882873930061e-06, "loss": 3.5024, "step": 7950 }, { "epoch": 2.1589368049905073, "grad_norm": 10.493721436691686, "learning_rate": 9.941402118901743e-06, "loss": 3.3715, "step": 7960 }, { "epoch": 2.1616490371575807, "grad_norm": 10.403463168674396, "learning_rate": 9.940919395330719e-06, "loss": 3.4694, "step": 7970 }, { "epoch": 2.164361269324654, "grad_norm": 11.585213754224668, "learning_rate": 9.940434703409292e-06, "loss": 3.4207, "step": 7980 }, { "epoch": 2.1670735014917275, "grad_norm": 11.30016054268739, "learning_rate": 9.939948043330556e-06, "loss": 3.4619, "step": 7990 }, { "epoch": 2.1697857336588013, "grad_norm": 10.150493977420043, "learning_rate": 9.939459415288384e-06, "loss": 3.4548, "step": 8000 }, { "epoch": 2.1724979658258747, "grad_norm": 10.574184003122834, "learning_rate": 9.938968819477432e-06, "loss": 3.494, "step": 8010 }, { "epoch": 2.175210197992948, "grad_norm": 10.649352948068646, "learning_rate": 9.938476256093147e-06, "loss": 3.453, "step": 8020 }, { "epoch": 2.1779224301600215, "grad_norm": 10.03330657084772, "learning_rate": 9.937981725331748e-06, "loss": 3.4133, "step": 8030 }, { "epoch": 2.1806346623270954, "grad_norm": 10.425920818878048, "learning_rate": 9.937485227390252e-06, "loss": 3.3523, "step": 8040 }, { "epoch": 2.1833468944941687, "grad_norm": 12.037097218012258, "learning_rate": 9.93698676246645e-06, "loss": 3.306, "step": 8050 }, { "epoch": 2.186059126661242, "grad_norm": 9.548559665936004, "learning_rate": 9.936486330758916e-06, "loss": 3.4133, "step": 8060 }, { "epoch": 2.1887713588283155, "grad_norm": 10.178457453023222, "learning_rate": 9.935983932467018e-06, "loss": 3.4542, "step": 8070 }, { "epoch": 2.1914835909953894, "grad_norm": 10.706443093076846, "learning_rate": 9.935479567790892e-06, "loss": 3.4314, "step": 8080 }, { "epoch": 2.1941958231624628, "grad_norm": 10.333595603493151, "learning_rate": 9.934973236931468e-06, "loss": 3.3591, "step": 8090 }, { "epoch": 2.196908055329536, "grad_norm": 10.350253024832542, "learning_rate": 9.934464940090461e-06, "loss": 3.4189, "step": 8100 }, { "epoch": 2.1996202874966095, "grad_norm": 11.974242345676577, "learning_rate": 9.93395467747036e-06, "loss": 3.3789, "step": 8110 }, { "epoch": 2.2023325196636834, "grad_norm": 10.115429695187085, "learning_rate": 9.933442449274446e-06, "loss": 3.4333, "step": 8120 }, { "epoch": 2.2050447518307568, "grad_norm": 10.340942805595137, "learning_rate": 9.932928255706776e-06, "loss": 3.462, "step": 8130 }, { "epoch": 2.20775698399783, "grad_norm": 10.686434416794844, "learning_rate": 9.932412096972192e-06, "loss": 3.3879, "step": 8140 }, { "epoch": 2.2104692161649035, "grad_norm": 10.693254572413275, "learning_rate": 9.931893973276323e-06, "loss": 3.4147, "step": 8150 }, { "epoch": 2.2131814483319774, "grad_norm": 9.678665161180483, "learning_rate": 9.931373884825578e-06, "loss": 3.3804, "step": 8160 }, { "epoch": 2.2158936804990508, "grad_norm": 11.231137619926697, "learning_rate": 9.930851831827145e-06, "loss": 3.3836, "step": 8170 }, { "epoch": 2.218605912666124, "grad_norm": 10.506057988837458, "learning_rate": 9.930327814489002e-06, "loss": 3.3641, "step": 8180 }, { "epoch": 2.2213181448331976, "grad_norm": 11.218178370313979, "learning_rate": 9.929801833019902e-06, "loss": 3.3327, "step": 8190 }, { "epoch": 2.2240303770002714, "grad_norm": 10.449090161412462, "learning_rate": 9.929273887629388e-06, "loss": 3.4698, "step": 8200 }, { "epoch": 2.226742609167345, "grad_norm": 9.444793899239595, "learning_rate": 9.92874397852778e-06, "loss": 3.4493, "step": 8210 }, { "epoch": 2.229454841334418, "grad_norm": 10.579120091775607, "learning_rate": 9.92821210592618e-06, "loss": 3.367, "step": 8220 }, { "epoch": 2.2321670735014916, "grad_norm": 9.969524764813528, "learning_rate": 9.927678270036473e-06, "loss": 3.3247, "step": 8230 }, { "epoch": 2.2348793056685654, "grad_norm": 11.008050913184828, "learning_rate": 9.927142471071334e-06, "loss": 3.317, "step": 8240 }, { "epoch": 2.237591537835639, "grad_norm": 9.93650891666251, "learning_rate": 9.926604709244206e-06, "loss": 3.4391, "step": 8250 }, { "epoch": 2.240303770002712, "grad_norm": 9.977820835994976, "learning_rate": 9.926064984769323e-06, "loss": 3.3937, "step": 8260 }, { "epoch": 2.2430160021697856, "grad_norm": 10.796230104469485, "learning_rate": 9.9255232978617e-06, "loss": 3.2945, "step": 8270 }, { "epoch": 2.2457282343368594, "grad_norm": 9.381037885280042, "learning_rate": 9.924979648737134e-06, "loss": 3.4622, "step": 8280 }, { "epoch": 2.248440466503933, "grad_norm": 10.47859135993876, "learning_rate": 9.9244340376122e-06, "loss": 3.4043, "step": 8290 }, { "epoch": 2.251152698671006, "grad_norm": 10.165001204182703, "learning_rate": 9.92388646470426e-06, "loss": 3.4152, "step": 8300 }, { "epoch": 2.2538649308380796, "grad_norm": 10.622939597171317, "learning_rate": 9.923336930231451e-06, "loss": 3.4843, "step": 8310 }, { "epoch": 2.2565771630051534, "grad_norm": 10.43487164415416, "learning_rate": 9.922785434412697e-06, "loss": 3.3813, "step": 8320 }, { "epoch": 2.259289395172227, "grad_norm": 11.05748957547795, "learning_rate": 9.922231977467701e-06, "loss": 3.4217, "step": 8330 }, { "epoch": 2.2620016273393, "grad_norm": 11.20845621992615, "learning_rate": 9.921676559616948e-06, "loss": 3.4393, "step": 8340 }, { "epoch": 2.2647138595063736, "grad_norm": 10.107640834419353, "learning_rate": 9.921119181081705e-06, "loss": 3.3688, "step": 8350 }, { "epoch": 2.2674260916734474, "grad_norm": 10.631971799298334, "learning_rate": 9.920559842084017e-06, "loss": 3.4804, "step": 8360 }, { "epoch": 2.270138323840521, "grad_norm": 10.856638949996453, "learning_rate": 9.919998542846714e-06, "loss": 3.3261, "step": 8370 }, { "epoch": 2.2728505560075942, "grad_norm": 10.27941104766489, "learning_rate": 9.919435283593405e-06, "loss": 3.3732, "step": 8380 }, { "epoch": 2.2755627881746676, "grad_norm": 11.250216495780256, "learning_rate": 9.918870064548475e-06, "loss": 3.4624, "step": 8390 }, { "epoch": 2.2782750203417415, "grad_norm": 9.931253841337675, "learning_rate": 9.9183028859371e-06, "loss": 3.4906, "step": 8400 }, { "epoch": 2.280987252508815, "grad_norm": 10.231436125331088, "learning_rate": 9.917733747985228e-06, "loss": 3.5123, "step": 8410 }, { "epoch": 2.2836994846758882, "grad_norm": 9.887377526254799, "learning_rate": 9.917162650919594e-06, "loss": 3.4192, "step": 8420 }, { "epoch": 2.2864117168429616, "grad_norm": 11.524920874305172, "learning_rate": 9.916589594967705e-06, "loss": 3.381, "step": 8430 }, { "epoch": 2.289123949010035, "grad_norm": 11.33647845048549, "learning_rate": 9.916014580357857e-06, "loss": 3.294, "step": 8440 }, { "epoch": 2.291836181177109, "grad_norm": 9.155967118486437, "learning_rate": 9.915437607319124e-06, "loss": 3.4515, "step": 8450 }, { "epoch": 2.2945484133441822, "grad_norm": 9.753822713735172, "learning_rate": 9.914858676081355e-06, "loss": 3.4496, "step": 8460 }, { "epoch": 2.2972606455112556, "grad_norm": 10.14169515864851, "learning_rate": 9.914277786875188e-06, "loss": 3.4472, "step": 8470 }, { "epoch": 2.2999728776783295, "grad_norm": 10.834426942056453, "learning_rate": 9.913694939932033e-06, "loss": 3.4363, "step": 8480 }, { "epoch": 2.302685109845403, "grad_norm": 11.523503092905274, "learning_rate": 9.913110135484083e-06, "loss": 3.3922, "step": 8490 }, { "epoch": 2.3053973420124763, "grad_norm": 10.408846597330367, "learning_rate": 9.912523373764311e-06, "loss": 3.4427, "step": 8500 }, { "epoch": 2.3081095741795497, "grad_norm": 10.309426855650706, "learning_rate": 9.91193465500647e-06, "loss": 3.4471, "step": 8510 }, { "epoch": 2.310821806346623, "grad_norm": 10.451228766137904, "learning_rate": 9.911343979445092e-06, "loss": 3.4347, "step": 8520 }, { "epoch": 2.313534038513697, "grad_norm": 10.680878649546933, "learning_rate": 9.910751347315488e-06, "loss": 3.4618, "step": 8530 }, { "epoch": 2.3162462706807703, "grad_norm": 9.439928534759309, "learning_rate": 9.910156758853751e-06, "loss": 3.3517, "step": 8540 }, { "epoch": 2.3189585028478437, "grad_norm": 10.951665433002338, "learning_rate": 9.909560214296752e-06, "loss": 3.5069, "step": 8550 }, { "epoch": 2.3216707350149175, "grad_norm": 9.701229061351214, "learning_rate": 9.908961713882137e-06, "loss": 3.4078, "step": 8560 }, { "epoch": 2.324382967181991, "grad_norm": 11.582348306237218, "learning_rate": 9.90836125784834e-06, "loss": 3.4032, "step": 8570 }, { "epoch": 2.3270951993490643, "grad_norm": 10.629049379394075, "learning_rate": 9.907758846434565e-06, "loss": 3.3816, "step": 8580 }, { "epoch": 2.3298074315161377, "grad_norm": 9.83628441171694, "learning_rate": 9.9071544798808e-06, "loss": 3.5596, "step": 8590 }, { "epoch": 2.332519663683211, "grad_norm": 9.777185196318918, "learning_rate": 9.906548158427813e-06, "loss": 3.366, "step": 8600 }, { "epoch": 2.335231895850285, "grad_norm": 11.107416189408939, "learning_rate": 9.905939882317148e-06, "loss": 3.3993, "step": 8610 }, { "epoch": 2.3379441280173583, "grad_norm": 9.705646699984285, "learning_rate": 9.905329651791127e-06, "loss": 3.4172, "step": 8620 }, { "epoch": 2.3406563601844317, "grad_norm": 10.205055324730134, "learning_rate": 9.904717467092853e-06, "loss": 3.5065, "step": 8630 }, { "epoch": 2.3433685923515055, "grad_norm": 9.68564547724862, "learning_rate": 9.904103328466206e-06, "loss": 3.4716, "step": 8640 }, { "epoch": 2.346080824518579, "grad_norm": 10.194887233876539, "learning_rate": 9.903487236155847e-06, "loss": 3.4618, "step": 8650 }, { "epoch": 2.3487930566856523, "grad_norm": 10.115750850618353, "learning_rate": 9.90286919040721e-06, "loss": 3.444, "step": 8660 }, { "epoch": 2.3515052888527257, "grad_norm": 10.224614070347478, "learning_rate": 9.902249191466511e-06, "loss": 3.3939, "step": 8670 }, { "epoch": 2.354217521019799, "grad_norm": 9.788284488236474, "learning_rate": 9.901627239580746e-06, "loss": 3.5049, "step": 8680 }, { "epoch": 2.356929753186873, "grad_norm": 10.096196183489017, "learning_rate": 9.901003334997684e-06, "loss": 3.3888, "step": 8690 }, { "epoch": 2.3596419853539463, "grad_norm": 10.612603315646641, "learning_rate": 9.900377477965876e-06, "loss": 3.3866, "step": 8700 }, { "epoch": 2.3623542175210197, "grad_norm": 9.202797220205904, "learning_rate": 9.899749668734648e-06, "loss": 3.4565, "step": 8710 }, { "epoch": 2.3650664496880935, "grad_norm": 9.313031183995781, "learning_rate": 9.899119907554107e-06, "loss": 3.479, "step": 8720 }, { "epoch": 2.367778681855167, "grad_norm": 10.641672316127686, "learning_rate": 9.898488194675133e-06, "loss": 3.4291, "step": 8730 }, { "epoch": 2.3704909140222403, "grad_norm": 9.912674939637327, "learning_rate": 9.897854530349388e-06, "loss": 3.4655, "step": 8740 }, { "epoch": 2.3732031461893137, "grad_norm": 9.731698712172914, "learning_rate": 9.89721891482931e-06, "loss": 3.4508, "step": 8750 }, { "epoch": 2.375915378356387, "grad_norm": 10.809335628710361, "learning_rate": 9.896581348368111e-06, "loss": 3.3604, "step": 8760 }, { "epoch": 2.378627610523461, "grad_norm": 9.324316198291728, "learning_rate": 9.895941831219786e-06, "loss": 3.3731, "step": 8770 }, { "epoch": 2.3813398426905343, "grad_norm": 10.322126033479222, "learning_rate": 9.895300363639103e-06, "loss": 3.4084, "step": 8780 }, { "epoch": 2.3840520748576077, "grad_norm": 10.354706349708307, "learning_rate": 9.894656945881607e-06, "loss": 3.436, "step": 8790 }, { "epoch": 2.386764307024681, "grad_norm": 9.772379662244695, "learning_rate": 9.894011578203623e-06, "loss": 3.4193, "step": 8800 }, { "epoch": 2.389476539191755, "grad_norm": 9.079369070127115, "learning_rate": 9.89336426086225e-06, "loss": 3.5137, "step": 8810 }, { "epoch": 2.3921887713588283, "grad_norm": 10.336238846153842, "learning_rate": 9.892714994115366e-06, "loss": 3.347, "step": 8820 }, { "epoch": 2.3949010035259017, "grad_norm": 10.515110798940356, "learning_rate": 9.89206377822162e-06, "loss": 3.4809, "step": 8830 }, { "epoch": 2.397613235692975, "grad_norm": 11.429919698411393, "learning_rate": 9.891410613440446e-06, "loss": 3.3854, "step": 8840 }, { "epoch": 2.400325467860049, "grad_norm": 9.867262995217226, "learning_rate": 9.89075550003205e-06, "loss": 3.4903, "step": 8850 }, { "epoch": 2.4030377000271224, "grad_norm": 11.440311224937888, "learning_rate": 9.89009843825741e-06, "loss": 3.3709, "step": 8860 }, { "epoch": 2.4057499321941958, "grad_norm": 10.39444792999328, "learning_rate": 9.88943942837829e-06, "loss": 3.5097, "step": 8870 }, { "epoch": 2.408462164361269, "grad_norm": 11.536941795200203, "learning_rate": 9.88877847065722e-06, "loss": 3.446, "step": 8880 }, { "epoch": 2.411174396528343, "grad_norm": 10.34113549683298, "learning_rate": 9.888115565357513e-06, "loss": 3.4652, "step": 8890 }, { "epoch": 2.4138866286954164, "grad_norm": 9.987173723069171, "learning_rate": 9.887450712743256e-06, "loss": 3.4245, "step": 8900 }, { "epoch": 2.4165988608624898, "grad_norm": 10.129800263713726, "learning_rate": 9.88678391307931e-06, "loss": 3.4724, "step": 8910 }, { "epoch": 2.419311093029563, "grad_norm": 10.063033974817246, "learning_rate": 9.886115166631313e-06, "loss": 3.4764, "step": 8920 }, { "epoch": 2.422023325196637, "grad_norm": 12.723937017484182, "learning_rate": 9.885444473665678e-06, "loss": 3.4689, "step": 8930 }, { "epoch": 2.4247355573637104, "grad_norm": 9.639615342117049, "learning_rate": 9.884771834449597e-06, "loss": 3.4627, "step": 8940 }, { "epoch": 2.4274477895307838, "grad_norm": 10.900361055846378, "learning_rate": 9.884097249251028e-06, "loss": 3.5031, "step": 8950 }, { "epoch": 2.430160021697857, "grad_norm": 10.070434056122316, "learning_rate": 9.883420718338716e-06, "loss": 3.4232, "step": 8960 }, { "epoch": 2.432872253864931, "grad_norm": 10.323380389204111, "learning_rate": 9.882742241982172e-06, "loss": 3.3928, "step": 8970 }, { "epoch": 2.4355844860320044, "grad_norm": 10.667504699425857, "learning_rate": 9.882061820451691e-06, "loss": 3.4807, "step": 8980 }, { "epoch": 2.438296718199078, "grad_norm": 10.47655625819323, "learning_rate": 9.88137945401833e-06, "loss": 3.3687, "step": 8990 }, { "epoch": 2.441008950366151, "grad_norm": 9.919591531423523, "learning_rate": 9.880695142953935e-06, "loss": 3.3991, "step": 9000 }, { "epoch": 2.443721182533225, "grad_norm": 10.572090816412008, "learning_rate": 9.880008887531118e-06, "loss": 3.445, "step": 9010 }, { "epoch": 2.4464334147002984, "grad_norm": 10.120998900538801, "learning_rate": 9.879320688023266e-06, "loss": 3.3804, "step": 9020 }, { "epoch": 2.449145646867372, "grad_norm": 9.737091620746702, "learning_rate": 9.878630544704541e-06, "loss": 3.5311, "step": 9030 }, { "epoch": 2.451857879034445, "grad_norm": 10.7498809327433, "learning_rate": 9.877938457849885e-06, "loss": 3.499, "step": 9040 }, { "epoch": 2.454570111201519, "grad_norm": 9.15679589863431, "learning_rate": 9.877244427735007e-06, "loss": 3.4104, "step": 9050 }, { "epoch": 2.4572823433685924, "grad_norm": 10.281886526115436, "learning_rate": 9.876548454636392e-06, "loss": 3.4891, "step": 9060 }, { "epoch": 2.459994575535666, "grad_norm": 10.904395482992655, "learning_rate": 9.875850538831302e-06, "loss": 3.398, "step": 9070 }, { "epoch": 2.462706807702739, "grad_norm": 11.589056262486134, "learning_rate": 9.87515068059777e-06, "loss": 3.4085, "step": 9080 }, { "epoch": 2.465419039869813, "grad_norm": 10.152837656691984, "learning_rate": 9.874448880214602e-06, "loss": 3.4673, "step": 9090 }, { "epoch": 2.4681312720368864, "grad_norm": 10.234149645527552, "learning_rate": 9.873745137961382e-06, "loss": 3.487, "step": 9100 }, { "epoch": 2.47084350420396, "grad_norm": 10.36410241437007, "learning_rate": 9.873039454118465e-06, "loss": 3.4707, "step": 9110 }, { "epoch": 2.473555736371033, "grad_norm": 9.927163649448115, "learning_rate": 9.872331828966976e-06, "loss": 3.3678, "step": 9120 }, { "epoch": 2.476267968538107, "grad_norm": 9.974914527558592, "learning_rate": 9.871622262788822e-06, "loss": 3.427, "step": 9130 }, { "epoch": 2.4789802007051804, "grad_norm": 9.427346485347568, "learning_rate": 9.870910755866674e-06, "loss": 3.4945, "step": 9140 }, { "epoch": 2.481692432872254, "grad_norm": 10.374465590426873, "learning_rate": 9.870197308483981e-06, "loss": 3.3876, "step": 9150 }, { "epoch": 2.484404665039327, "grad_norm": 10.553555213124271, "learning_rate": 9.869481920924965e-06, "loss": 3.4331, "step": 9160 }, { "epoch": 2.487116897206401, "grad_norm": 9.790514652813167, "learning_rate": 9.86876459347462e-06, "loss": 3.4909, "step": 9170 }, { "epoch": 2.4898291293734744, "grad_norm": 9.679464371457316, "learning_rate": 9.868045326418714e-06, "loss": 3.5031, "step": 9180 }, { "epoch": 2.492541361540548, "grad_norm": 10.764660953853978, "learning_rate": 9.867324120043785e-06, "loss": 3.3019, "step": 9190 }, { "epoch": 2.4952535937076212, "grad_norm": 10.733163535492205, "learning_rate": 9.866600974637146e-06, "loss": 3.4708, "step": 9200 }, { "epoch": 2.497965825874695, "grad_norm": 9.477901869418446, "learning_rate": 9.865875890486882e-06, "loss": 3.4545, "step": 9210 }, { "epoch": 2.5006780580417685, "grad_norm": 12.22909141660565, "learning_rate": 9.86514886788185e-06, "loss": 3.4638, "step": 9220 }, { "epoch": 2.503390290208842, "grad_norm": 9.658919250527761, "learning_rate": 9.86441990711168e-06, "loss": 3.5646, "step": 9230 }, { "epoch": 2.5061025223759152, "grad_norm": 8.843357429189407, "learning_rate": 9.86368900846677e-06, "loss": 3.5219, "step": 9240 }, { "epoch": 2.508814754542989, "grad_norm": 11.191139204933263, "learning_rate": 9.862956172238298e-06, "loss": 3.4463, "step": 9250 }, { "epoch": 2.5115269867100625, "grad_norm": 11.326375049923682, "learning_rate": 9.862221398718208e-06, "loss": 3.3362, "step": 9260 }, { "epoch": 2.514239218877136, "grad_norm": 10.157993770748513, "learning_rate": 9.861484688199216e-06, "loss": 3.5665, "step": 9270 }, { "epoch": 2.5169514510442093, "grad_norm": 10.0699233906355, "learning_rate": 9.860746040974813e-06, "loss": 3.5962, "step": 9280 }, { "epoch": 2.5196636832112826, "grad_norm": 9.572715560661596, "learning_rate": 9.860005457339257e-06, "loss": 3.3563, "step": 9290 }, { "epoch": 2.5223759153783565, "grad_norm": 11.168444642722449, "learning_rate": 9.85926293758758e-06, "loss": 3.3924, "step": 9300 }, { "epoch": 2.52508814754543, "grad_norm": 11.44666835730229, "learning_rate": 9.858518482015586e-06, "loss": 3.4901, "step": 9310 }, { "epoch": 2.5278003797125033, "grad_norm": 10.53524238752234, "learning_rate": 9.857772090919848e-06, "loss": 3.3839, "step": 9320 }, { "epoch": 2.530512611879577, "grad_norm": 9.055677708784701, "learning_rate": 9.857023764597712e-06, "loss": 3.5096, "step": 9330 }, { "epoch": 2.5332248440466505, "grad_norm": 10.042669401145451, "learning_rate": 9.856273503347296e-06, "loss": 3.4517, "step": 9340 }, { "epoch": 2.535937076213724, "grad_norm": 10.519873530169068, "learning_rate": 9.855521307467484e-06, "loss": 3.4222, "step": 9350 }, { "epoch": 2.5386493083807973, "grad_norm": 9.34600946512369, "learning_rate": 9.854767177257937e-06, "loss": 3.3241, "step": 9360 }, { "epoch": 2.5413615405478707, "grad_norm": 9.914083619241351, "learning_rate": 9.85401111301908e-06, "loss": 3.4074, "step": 9370 }, { "epoch": 2.5440737727149445, "grad_norm": 10.729176930239051, "learning_rate": 9.853253115052115e-06, "loss": 3.5059, "step": 9380 }, { "epoch": 2.546786004882018, "grad_norm": 9.666410410246591, "learning_rate": 9.852493183659008e-06, "loss": 3.3972, "step": 9390 }, { "epoch": 2.5494982370490913, "grad_norm": 9.605317102955839, "learning_rate": 9.851731319142503e-06, "loss": 3.5404, "step": 9400 }, { "epoch": 2.552210469216165, "grad_norm": 10.178824313856426, "learning_rate": 9.850967521806105e-06, "loss": 3.3156, "step": 9410 }, { "epoch": 2.5549227013832385, "grad_norm": 10.650568907605978, "learning_rate": 9.850201791954095e-06, "loss": 3.3868, "step": 9420 }, { "epoch": 2.557634933550312, "grad_norm": 9.547582739434002, "learning_rate": 9.849434129891524e-06, "loss": 3.3696, "step": 9430 }, { "epoch": 2.5603471657173853, "grad_norm": 9.523290621093352, "learning_rate": 9.84866453592421e-06, "loss": 3.4654, "step": 9440 }, { "epoch": 2.5630593978844587, "grad_norm": 10.043505240496524, "learning_rate": 9.84789301035874e-06, "loss": 3.355, "step": 9450 }, { "epoch": 2.5657716300515325, "grad_norm": 10.696589716344223, "learning_rate": 9.847119553502476e-06, "loss": 3.4762, "step": 9460 }, { "epoch": 2.568483862218606, "grad_norm": 10.466147721754053, "learning_rate": 9.846344165663544e-06, "loss": 3.4037, "step": 9470 }, { "epoch": 2.5711960943856793, "grad_norm": 9.82190175611993, "learning_rate": 9.84556684715084e-06, "loss": 3.4809, "step": 9480 }, { "epoch": 2.573908326552753, "grad_norm": 9.501834275138668, "learning_rate": 9.84478759827403e-06, "loss": 3.4747, "step": 9490 }, { "epoch": 2.5766205587198265, "grad_norm": 9.49559855680466, "learning_rate": 9.844006419343553e-06, "loss": 3.5479, "step": 9500 }, { "epoch": 2.5793327908869, "grad_norm": 10.376933277078258, "learning_rate": 9.843223310670606e-06, "loss": 3.3692, "step": 9510 }, { "epoch": 2.5820450230539733, "grad_norm": 11.12874706289573, "learning_rate": 9.842438272567169e-06, "loss": 3.3543, "step": 9520 }, { "epoch": 2.5847572552210467, "grad_norm": 8.884671659537398, "learning_rate": 9.841651305345979e-06, "loss": 3.3372, "step": 9530 }, { "epoch": 2.5874694873881205, "grad_norm": 8.66984434283123, "learning_rate": 9.840862409320546e-06, "loss": 3.5272, "step": 9540 }, { "epoch": 2.590181719555194, "grad_norm": 9.841967874577385, "learning_rate": 9.840071584805152e-06, "loss": 3.4839, "step": 9550 }, { "epoch": 2.5928939517222673, "grad_norm": 10.516135307275531, "learning_rate": 9.839278832114838e-06, "loss": 3.4341, "step": 9560 }, { "epoch": 2.595606183889341, "grad_norm": 9.703200776491407, "learning_rate": 9.838484151565425e-06, "loss": 3.4292, "step": 9570 }, { "epoch": 2.5983184160564146, "grad_norm": 10.635464639767122, "learning_rate": 9.837687543473488e-06, "loss": 3.4797, "step": 9580 }, { "epoch": 2.601030648223488, "grad_norm": 10.732466511693207, "learning_rate": 9.836889008156386e-06, "loss": 3.4823, "step": 9590 }, { "epoch": 2.6037428803905613, "grad_norm": 10.79559944696922, "learning_rate": 9.836088545932233e-06, "loss": 3.5804, "step": 9600 }, { "epoch": 2.6064551125576347, "grad_norm": 10.779036482937709, "learning_rate": 9.835286157119911e-06, "loss": 3.4549, "step": 9610 }, { "epoch": 2.6091673447247086, "grad_norm": 9.982597852674857, "learning_rate": 9.834481842039083e-06, "loss": 3.4232, "step": 9620 }, { "epoch": 2.611879576891782, "grad_norm": 9.742520408240042, "learning_rate": 9.833675601010165e-06, "loss": 3.5375, "step": 9630 }, { "epoch": 2.6145918090588554, "grad_norm": 10.983802347840083, "learning_rate": 9.832867434354343e-06, "loss": 3.4363, "step": 9640 }, { "epoch": 2.617304041225929, "grad_norm": 10.407298392783357, "learning_rate": 9.832057342393575e-06, "loss": 3.2944, "step": 9650 }, { "epoch": 2.6200162733930026, "grad_norm": 9.886116108518205, "learning_rate": 9.831245325450582e-06, "loss": 3.5574, "step": 9660 }, { "epoch": 2.622728505560076, "grad_norm": 11.02018328243985, "learning_rate": 9.830431383848853e-06, "loss": 3.499, "step": 9670 }, { "epoch": 2.6254407377271494, "grad_norm": 9.68082394696806, "learning_rate": 9.829615517912647e-06, "loss": 3.4858, "step": 9680 }, { "epoch": 2.6281529698942228, "grad_norm": 10.435160016099827, "learning_rate": 9.828797727966983e-06, "loss": 3.4285, "step": 9690 }, { "epoch": 2.6308652020612966, "grad_norm": 10.312206940350645, "learning_rate": 9.827978014337652e-06, "loss": 3.4532, "step": 9700 }, { "epoch": 2.63357743422837, "grad_norm": 10.66414908976704, "learning_rate": 9.827156377351207e-06, "loss": 3.4815, "step": 9710 }, { "epoch": 2.6362896663954434, "grad_norm": 10.200752630745937, "learning_rate": 9.826332817334973e-06, "loss": 3.4386, "step": 9720 }, { "epoch": 2.639001898562517, "grad_norm": 9.872833548878132, "learning_rate": 9.825507334617035e-06, "loss": 3.4718, "step": 9730 }, { "epoch": 2.6417141307295906, "grad_norm": 10.71269223158453, "learning_rate": 9.824679929526247e-06, "loss": 3.3538, "step": 9740 }, { "epoch": 2.644426362896664, "grad_norm": 9.878350940212437, "learning_rate": 9.82385060239223e-06, "loss": 3.4463, "step": 9750 }, { "epoch": 2.6471385950637374, "grad_norm": 9.765326103470969, "learning_rate": 9.823019353545366e-06, "loss": 3.5094, "step": 9760 }, { "epoch": 2.649850827230811, "grad_norm": 10.376491593717088, "learning_rate": 9.82218618331681e-06, "loss": 3.4346, "step": 9770 }, { "epoch": 2.6525630593978846, "grad_norm": 9.727037075217455, "learning_rate": 9.821351092038476e-06, "loss": 3.5388, "step": 9780 }, { "epoch": 2.655275291564958, "grad_norm": 11.270626479360988, "learning_rate": 9.820514080043045e-06, "loss": 3.4809, "step": 9790 }, { "epoch": 2.6579875237320314, "grad_norm": 11.318239784551778, "learning_rate": 9.819675147663964e-06, "loss": 3.5167, "step": 9800 }, { "epoch": 2.6606997558991052, "grad_norm": 10.677817758779256, "learning_rate": 9.818834295235442e-06, "loss": 3.5378, "step": 9810 }, { "epoch": 2.6634119880661786, "grad_norm": 10.20346665858874, "learning_rate": 9.817991523092461e-06, "loss": 3.4014, "step": 9820 }, { "epoch": 2.666124220233252, "grad_norm": 9.601737099383941, "learning_rate": 9.81714683157076e-06, "loss": 3.4522, "step": 9830 }, { "epoch": 2.6688364524003254, "grad_norm": 9.358368423844398, "learning_rate": 9.816300221006843e-06, "loss": 3.3823, "step": 9840 }, { "epoch": 2.671548684567399, "grad_norm": 9.482501546710456, "learning_rate": 9.815451691737981e-06, "loss": 3.475, "step": 9850 }, { "epoch": 2.6742609167344726, "grad_norm": 10.43028677479889, "learning_rate": 9.814601244102212e-06, "loss": 3.4981, "step": 9860 }, { "epoch": 2.676973148901546, "grad_norm": 10.973002558434368, "learning_rate": 9.813748878438332e-06, "loss": 3.3744, "step": 9870 }, { "epoch": 2.6796853810686194, "grad_norm": 10.839789664214694, "learning_rate": 9.812894595085905e-06, "loss": 3.4247, "step": 9880 }, { "epoch": 2.682397613235693, "grad_norm": 9.736989674971193, "learning_rate": 9.81203839438526e-06, "loss": 3.4475, "step": 9890 }, { "epoch": 2.685109845402766, "grad_norm": 10.29311568732616, "learning_rate": 9.811180276677483e-06, "loss": 3.3607, "step": 9900 }, { "epoch": 2.68782207756984, "grad_norm": 9.889822618276874, "learning_rate": 9.810320242304435e-06, "loss": 3.5245, "step": 9910 }, { "epoch": 2.6905343097369134, "grad_norm": 9.723963701584994, "learning_rate": 9.809458291608729e-06, "loss": 3.4921, "step": 9920 }, { "epoch": 2.693246541903987, "grad_norm": 10.904796793698711, "learning_rate": 9.808594424933751e-06, "loss": 3.3961, "step": 9930 }, { "epoch": 2.6959587740710607, "grad_norm": 10.779815295660054, "learning_rate": 9.807728642623642e-06, "loss": 3.4857, "step": 9940 }, { "epoch": 2.698671006238134, "grad_norm": 9.345150475717094, "learning_rate": 9.80686094502331e-06, "loss": 3.5051, "step": 9950 }, { "epoch": 2.7013832384052074, "grad_norm": 10.029142763361438, "learning_rate": 9.80599133247843e-06, "loss": 3.5259, "step": 9960 }, { "epoch": 2.704095470572281, "grad_norm": 10.361691538964259, "learning_rate": 9.805119805335435e-06, "loss": 3.4756, "step": 9970 }, { "epoch": 2.7068077027393542, "grad_norm": 10.3273266017323, "learning_rate": 9.804246363941518e-06, "loss": 3.3992, "step": 9980 }, { "epoch": 2.709519934906428, "grad_norm": 9.707521645933115, "learning_rate": 9.803371008644643e-06, "loss": 3.4771, "step": 9990 }, { "epoch": 2.7122321670735015, "grad_norm": 10.765684417060063, "learning_rate": 9.802493739793528e-06, "loss": 3.4212, "step": 10000 }, { "epoch": 2.714944399240575, "grad_norm": 9.595106682572101, "learning_rate": 9.801614557737658e-06, "loss": 3.3877, "step": 10010 }, { "epoch": 2.7176566314076487, "grad_norm": 10.39370602613928, "learning_rate": 9.80073346282728e-06, "loss": 3.5319, "step": 10020 }, { "epoch": 2.720368863574722, "grad_norm": 10.797526031411486, "learning_rate": 9.799850455413402e-06, "loss": 3.4351, "step": 10030 }, { "epoch": 2.7230810957417955, "grad_norm": 10.167004944938801, "learning_rate": 9.798965535847796e-06, "loss": 3.4462, "step": 10040 }, { "epoch": 2.725793327908869, "grad_norm": 9.70909155514413, "learning_rate": 9.79807870448299e-06, "loss": 3.3708, "step": 10050 }, { "epoch": 2.7285055600759422, "grad_norm": 9.337194082189718, "learning_rate": 9.79718996167228e-06, "loss": 3.5764, "step": 10060 }, { "epoch": 2.731217792243016, "grad_norm": 10.073824720870501, "learning_rate": 9.796299307769722e-06, "loss": 3.4469, "step": 10070 }, { "epoch": 2.7339300244100895, "grad_norm": 10.292369651758396, "learning_rate": 9.795406743130129e-06, "loss": 3.5857, "step": 10080 }, { "epoch": 2.736642256577163, "grad_norm": 9.594697916947514, "learning_rate": 9.79451226810908e-06, "loss": 3.4979, "step": 10090 }, { "epoch": 2.7393544887442367, "grad_norm": 10.408669430007736, "learning_rate": 9.793615883062915e-06, "loss": 3.4373, "step": 10100 }, { "epoch": 2.74206672091131, "grad_norm": 10.570702061402606, "learning_rate": 9.792717588348733e-06, "loss": 3.5146, "step": 10110 }, { "epoch": 2.7447789530783835, "grad_norm": 9.851442070221347, "learning_rate": 9.791817384324393e-06, "loss": 3.4735, "step": 10120 }, { "epoch": 2.747491185245457, "grad_norm": 9.701173439484617, "learning_rate": 9.790915271348518e-06, "loss": 3.4242, "step": 10130 }, { "epoch": 2.7502034174125303, "grad_norm": 10.933188285928338, "learning_rate": 9.790011249780487e-06, "loss": 3.4548, "step": 10140 }, { "epoch": 2.752915649579604, "grad_norm": 9.354975823632824, "learning_rate": 9.789105319980443e-06, "loss": 3.3603, "step": 10150 }, { "epoch": 2.7556278817466775, "grad_norm": 10.335973419049127, "learning_rate": 9.788197482309287e-06, "loss": 3.3247, "step": 10160 }, { "epoch": 2.758340113913751, "grad_norm": 9.447465318507689, "learning_rate": 9.787287737128684e-06, "loss": 3.3603, "step": 10170 }, { "epoch": 2.7610523460808247, "grad_norm": 10.125881649021302, "learning_rate": 9.786376084801052e-06, "loss": 3.4048, "step": 10180 }, { "epoch": 2.763764578247898, "grad_norm": 11.532384745202627, "learning_rate": 9.785462525689575e-06, "loss": 3.3281, "step": 10190 }, { "epoch": 2.7664768104149715, "grad_norm": 10.407401857220938, "learning_rate": 9.784547060158192e-06, "loss": 3.4684, "step": 10200 }, { "epoch": 2.769189042582045, "grad_norm": 10.163941080974125, "learning_rate": 9.783629688571606e-06, "loss": 3.419, "step": 10210 }, { "epoch": 2.7719012747491183, "grad_norm": 10.220165582727764, "learning_rate": 9.782710411295278e-06, "loss": 3.3267, "step": 10220 }, { "epoch": 2.774613506916192, "grad_norm": 9.532744982576778, "learning_rate": 9.781789228695423e-06, "loss": 3.4618, "step": 10230 }, { "epoch": 2.7773257390832655, "grad_norm": 10.135427064139503, "learning_rate": 9.780866141139023e-06, "loss": 3.5012, "step": 10240 }, { "epoch": 2.780037971250339, "grad_norm": 10.305243225400707, "learning_rate": 9.779941148993815e-06, "loss": 3.4363, "step": 10250 }, { "epoch": 2.7827502034174127, "grad_norm": 9.872443799439445, "learning_rate": 9.779014252628292e-06, "loss": 3.3905, "step": 10260 }, { "epoch": 2.785462435584486, "grad_norm": 11.760761201025618, "learning_rate": 9.778085452411711e-06, "loss": 3.4384, "step": 10270 }, { "epoch": 2.7881746677515595, "grad_norm": 10.260691829442932, "learning_rate": 9.777154748714087e-06, "loss": 3.4333, "step": 10280 }, { "epoch": 2.790886899918633, "grad_norm": 9.771542256865438, "learning_rate": 9.776222141906184e-06, "loss": 3.4427, "step": 10290 }, { "epoch": 2.7935991320857063, "grad_norm": 10.543929575600284, "learning_rate": 9.77528763235954e-06, "loss": 3.4919, "step": 10300 }, { "epoch": 2.79631136425278, "grad_norm": 10.468881560162076, "learning_rate": 9.774351220446437e-06, "loss": 3.4312, "step": 10310 }, { "epoch": 2.7990235964198535, "grad_norm": 9.655069832906426, "learning_rate": 9.773412906539924e-06, "loss": 3.4487, "step": 10320 }, { "epoch": 2.801735828586927, "grad_norm": 9.200160662802745, "learning_rate": 9.772472691013798e-06, "loss": 3.408, "step": 10330 }, { "epoch": 2.8044480607540008, "grad_norm": 9.522083973033496, "learning_rate": 9.771530574242627e-06, "loss": 3.5324, "step": 10340 }, { "epoch": 2.807160292921074, "grad_norm": 9.001621783605488, "learning_rate": 9.770586556601724e-06, "loss": 3.5371, "step": 10350 }, { "epoch": 2.8098725250881476, "grad_norm": 10.197473733572782, "learning_rate": 9.769640638467166e-06, "loss": 3.4709, "step": 10360 }, { "epoch": 2.812584757255221, "grad_norm": 9.96815061930099, "learning_rate": 9.768692820215784e-06, "loss": 3.4596, "step": 10370 }, { "epoch": 2.8152969894222943, "grad_norm": 10.48209536507704, "learning_rate": 9.76774310222517e-06, "loss": 3.4741, "step": 10380 }, { "epoch": 2.818009221589368, "grad_norm": 10.153662892226977, "learning_rate": 9.766791484873668e-06, "loss": 3.5534, "step": 10390 }, { "epoch": 2.8207214537564416, "grad_norm": 10.837642157920115, "learning_rate": 9.765837968540381e-06, "loss": 3.4165, "step": 10400 }, { "epoch": 2.823433685923515, "grad_norm": 10.155836662075059, "learning_rate": 9.764882553605169e-06, "loss": 3.4788, "step": 10410 }, { "epoch": 2.826145918090589, "grad_norm": 9.17904404839413, "learning_rate": 9.763925240448649e-06, "loss": 3.4057, "step": 10420 }, { "epoch": 2.828858150257662, "grad_norm": 11.112721963467484, "learning_rate": 9.76296602945219e-06, "loss": 3.4534, "step": 10430 }, { "epoch": 2.8315703824247356, "grad_norm": 9.278575826034734, "learning_rate": 9.762004920997919e-06, "loss": 3.3602, "step": 10440 }, { "epoch": 2.834282614591809, "grad_norm": 10.980763256954104, "learning_rate": 9.761041915468726e-06, "loss": 3.5063, "step": 10450 }, { "epoch": 2.8369948467588824, "grad_norm": 9.084004430811538, "learning_rate": 9.760077013248243e-06, "loss": 3.586, "step": 10460 }, { "epoch": 2.839707078925956, "grad_norm": 10.505079448171504, "learning_rate": 9.75911021472087e-06, "loss": 3.4666, "step": 10470 }, { "epoch": 2.8424193110930296, "grad_norm": 10.967348240725993, "learning_rate": 9.758141520271757e-06, "loss": 3.4422, "step": 10480 }, { "epoch": 2.845131543260103, "grad_norm": 9.813319132188182, "learning_rate": 9.75717093028681e-06, "loss": 3.4942, "step": 10490 }, { "epoch": 2.847843775427177, "grad_norm": 11.432622360413173, "learning_rate": 9.756198445152688e-06, "loss": 3.5301, "step": 10500 }, { "epoch": 2.85055600759425, "grad_norm": 10.38183821667311, "learning_rate": 9.755224065256809e-06, "loss": 3.4472, "step": 10510 }, { "epoch": 2.8532682397613236, "grad_norm": 9.845903860296463, "learning_rate": 9.754247790987343e-06, "loss": 3.4704, "step": 10520 }, { "epoch": 2.855980471928397, "grad_norm": 9.658687707017815, "learning_rate": 9.753269622733218e-06, "loss": 3.4513, "step": 10530 }, { "epoch": 2.8586927040954704, "grad_norm": 9.717123151739488, "learning_rate": 9.75228956088411e-06, "loss": 3.3758, "step": 10540 }, { "epoch": 2.861404936262544, "grad_norm": 10.045716334812301, "learning_rate": 9.751307605830456e-06, "loss": 3.5792, "step": 10550 }, { "epoch": 2.8641171684296176, "grad_norm": 10.355273468490328, "learning_rate": 9.750323757963445e-06, "loss": 3.4387, "step": 10560 }, { "epoch": 2.866829400596691, "grad_norm": 10.439707615000527, "learning_rate": 9.749338017675017e-06, "loss": 3.4321, "step": 10570 }, { "epoch": 2.869541632763765, "grad_norm": 10.03865920022432, "learning_rate": 9.748350385357873e-06, "loss": 3.4648, "step": 10580 }, { "epoch": 2.8722538649308382, "grad_norm": 10.700178562367695, "learning_rate": 9.74736086140546e-06, "loss": 3.3914, "step": 10590 }, { "epoch": 2.8749660970979116, "grad_norm": 10.041539289990034, "learning_rate": 9.74636944621198e-06, "loss": 3.5404, "step": 10600 }, { "epoch": 2.877678329264985, "grad_norm": 10.231489379195137, "learning_rate": 9.745376140172395e-06, "loss": 3.4847, "step": 10610 }, { "epoch": 2.8803905614320584, "grad_norm": 10.30117713169905, "learning_rate": 9.744380943682415e-06, "loss": 3.4476, "step": 10620 }, { "epoch": 2.8831027935991322, "grad_norm": 9.100236896649143, "learning_rate": 9.743383857138502e-06, "loss": 3.4943, "step": 10630 }, { "epoch": 2.8858150257662056, "grad_norm": 9.684192343601577, "learning_rate": 9.742384880937872e-06, "loss": 3.4406, "step": 10640 }, { "epoch": 2.888527257933279, "grad_norm": 10.48222476267502, "learning_rate": 9.741384015478496e-06, "loss": 3.4388, "step": 10650 }, { "epoch": 2.891239490100353, "grad_norm": 9.133233460132743, "learning_rate": 9.740381261159098e-06, "loss": 3.4833, "step": 10660 }, { "epoch": 2.8939517222674263, "grad_norm": 10.122289122811777, "learning_rate": 9.739376618379149e-06, "loss": 3.3913, "step": 10670 }, { "epoch": 2.8966639544344996, "grad_norm": 9.262411875263421, "learning_rate": 9.738370087538876e-06, "loss": 3.5809, "step": 10680 }, { "epoch": 2.899376186601573, "grad_norm": 9.652520814781607, "learning_rate": 9.73736166903926e-06, "loss": 3.3401, "step": 10690 }, { "epoch": 2.9020884187686464, "grad_norm": 10.537204227109545, "learning_rate": 9.736351363282033e-06, "loss": 3.4757, "step": 10700 }, { "epoch": 2.9048006509357203, "grad_norm": 9.583417772800727, "learning_rate": 9.735339170669675e-06, "loss": 3.3727, "step": 10710 }, { "epoch": 2.9075128831027937, "grad_norm": 9.505107873712177, "learning_rate": 9.734325091605423e-06, "loss": 3.4392, "step": 10720 }, { "epoch": 2.910225115269867, "grad_norm": 10.163665034682968, "learning_rate": 9.733309126493263e-06, "loss": 3.498, "step": 10730 }, { "epoch": 2.9129373474369404, "grad_norm": 9.896925979982651, "learning_rate": 9.73229127573793e-06, "loss": 3.434, "step": 10740 }, { "epoch": 2.915649579604014, "grad_norm": 10.303616721585643, "learning_rate": 9.731271539744915e-06, "loss": 3.365, "step": 10750 }, { "epoch": 2.9183618117710877, "grad_norm": 11.372947099010862, "learning_rate": 9.730249918920458e-06, "loss": 3.5684, "step": 10760 }, { "epoch": 2.921074043938161, "grad_norm": 11.199267819663378, "learning_rate": 9.729226413671548e-06, "loss": 3.5233, "step": 10770 }, { "epoch": 2.9237862761052344, "grad_norm": 10.543527401415657, "learning_rate": 9.728201024405928e-06, "loss": 3.4853, "step": 10780 }, { "epoch": 2.9264985082723083, "grad_norm": 9.510868691942445, "learning_rate": 9.727173751532087e-06, "loss": 3.5089, "step": 10790 }, { "epoch": 2.9292107404393817, "grad_norm": 9.65384893038202, "learning_rate": 9.72614459545927e-06, "loss": 3.4062, "step": 10800 }, { "epoch": 2.931922972606455, "grad_norm": 10.030477554664966, "learning_rate": 9.725113556597469e-06, "loss": 3.4302, "step": 10810 }, { "epoch": 2.9346352047735285, "grad_norm": 8.980326013472855, "learning_rate": 9.724080635357428e-06, "loss": 3.4935, "step": 10820 }, { "epoch": 2.937347436940602, "grad_norm": 9.858901064941199, "learning_rate": 9.723045832150636e-06, "loss": 3.5249, "step": 10830 }, { "epoch": 2.9400596691076757, "grad_norm": 9.047438395980413, "learning_rate": 9.722009147389336e-06, "loss": 3.4705, "step": 10840 }, { "epoch": 2.942771901274749, "grad_norm": 9.871595556421886, "learning_rate": 9.720970581486521e-06, "loss": 3.4747, "step": 10850 }, { "epoch": 2.9454841334418225, "grad_norm": 9.314541912209856, "learning_rate": 9.719930134855931e-06, "loss": 3.5198, "step": 10860 }, { "epoch": 2.9481963656088963, "grad_norm": 8.995470476443431, "learning_rate": 9.718887807912059e-06, "loss": 3.5844, "step": 10870 }, { "epoch": 2.9509085977759697, "grad_norm": 10.670892414729602, "learning_rate": 9.71784360107014e-06, "loss": 3.4395, "step": 10880 }, { "epoch": 2.953620829943043, "grad_norm": 10.498430818019058, "learning_rate": 9.716797514746167e-06, "loss": 3.4655, "step": 10890 }, { "epoch": 2.9563330621101165, "grad_norm": 9.135193308003585, "learning_rate": 9.715749549356874e-06, "loss": 3.4942, "step": 10900 }, { "epoch": 2.95904529427719, "grad_norm": 9.608886917821126, "learning_rate": 9.714699705319749e-06, "loss": 3.4354, "step": 10910 }, { "epoch": 2.9617575264442637, "grad_norm": 10.329425595890054, "learning_rate": 9.713647983053023e-06, "loss": 3.5074, "step": 10920 }, { "epoch": 2.964469758611337, "grad_norm": 10.240508803136011, "learning_rate": 9.712594382975682e-06, "loss": 3.5447, "step": 10930 }, { "epoch": 2.9671819907784105, "grad_norm": 10.242016566666063, "learning_rate": 9.711538905507452e-06, "loss": 3.5467, "step": 10940 }, { "epoch": 2.9698942229454843, "grad_norm": 9.271040663266579, "learning_rate": 9.710481551068815e-06, "loss": 3.5529, "step": 10950 }, { "epoch": 2.9726064551125577, "grad_norm": 10.564236283592845, "learning_rate": 9.709422320080997e-06, "loss": 3.5076, "step": 10960 }, { "epoch": 2.975318687279631, "grad_norm": 10.114823465630042, "learning_rate": 9.708361212965969e-06, "loss": 3.4457, "step": 10970 }, { "epoch": 2.9780309194467045, "grad_norm": 9.297118717046533, "learning_rate": 9.707298230146454e-06, "loss": 3.5158, "step": 10980 }, { "epoch": 2.980743151613778, "grad_norm": 8.942011002942428, "learning_rate": 9.70623337204592e-06, "loss": 3.5063, "step": 10990 }, { "epoch": 2.9834553837808517, "grad_norm": 10.819061103008606, "learning_rate": 9.705166639088581e-06, "loss": 3.47, "step": 11000 }, { "epoch": 2.986167615947925, "grad_norm": 9.547334364575718, "learning_rate": 9.704098031699402e-06, "loss": 3.4659, "step": 11010 }, { "epoch": 2.9888798481149985, "grad_norm": 10.270463384387668, "learning_rate": 9.703027550304091e-06, "loss": 3.4958, "step": 11020 }, { "epoch": 2.9915920802820724, "grad_norm": 9.88297412488453, "learning_rate": 9.701955195329101e-06, "loss": 3.4757, "step": 11030 }, { "epoch": 2.9943043124491457, "grad_norm": 9.418637538315384, "learning_rate": 9.700880967201636e-06, "loss": 3.3999, "step": 11040 }, { "epoch": 2.997016544616219, "grad_norm": 10.117260496693351, "learning_rate": 9.699804866349646e-06, "loss": 3.3404, "step": 11050 }, { "epoch": 2.9997287767832925, "grad_norm": 10.002891668161308, "learning_rate": 9.698726893201822e-06, "loss": 3.3967, "step": 11060 }, { "epoch": 3.0024410089503664, "grad_norm": 9.602497816514799, "learning_rate": 9.697647048187606e-06, "loss": 3.1445, "step": 11070 }, { "epoch": 3.0051532411174398, "grad_norm": 11.133756824875432, "learning_rate": 9.696565331737181e-06, "loss": 3.0066, "step": 11080 }, { "epoch": 3.007865473284513, "grad_norm": 12.122085131521601, "learning_rate": 9.695481744281485e-06, "loss": 2.9699, "step": 11090 }, { "epoch": 3.0105777054515865, "grad_norm": 10.850568945020859, "learning_rate": 9.694396286252186e-06, "loss": 3.0162, "step": 11100 }, { "epoch": 3.0132899376186604, "grad_norm": 10.847859615835201, "learning_rate": 9.693308958081714e-06, "loss": 3.0513, "step": 11110 }, { "epoch": 3.0160021697857338, "grad_norm": 10.878414491715628, "learning_rate": 9.692219760203228e-06, "loss": 2.9718, "step": 11120 }, { "epoch": 3.018714401952807, "grad_norm": 11.310330081700503, "learning_rate": 9.691128693050646e-06, "loss": 3.0113, "step": 11130 }, { "epoch": 3.0214266341198805, "grad_norm": 12.171596767997062, "learning_rate": 9.690035757058622e-06, "loss": 2.957, "step": 11140 }, { "epoch": 3.024138866286954, "grad_norm": 10.7591225283794, "learning_rate": 9.688940952662556e-06, "loss": 2.9964, "step": 11150 }, { "epoch": 3.0268510984540278, "grad_norm": 11.062685774960048, "learning_rate": 9.687844280298596e-06, "loss": 3.1326, "step": 11160 }, { "epoch": 3.029563330621101, "grad_norm": 11.136032426173223, "learning_rate": 9.686745740403628e-06, "loss": 2.9049, "step": 11170 }, { "epoch": 3.0322755627881746, "grad_norm": 10.744935950983592, "learning_rate": 9.68564533341529e-06, "loss": 3.0062, "step": 11180 }, { "epoch": 3.034987794955248, "grad_norm": 11.669716038390924, "learning_rate": 9.684543059771955e-06, "loss": 3.0747, "step": 11190 }, { "epoch": 3.037700027122322, "grad_norm": 10.796222809485593, "learning_rate": 9.683438919912746e-06, "loss": 2.9779, "step": 11200 }, { "epoch": 3.040412259289395, "grad_norm": 11.767920662764805, "learning_rate": 9.682332914277527e-06, "loss": 3.0319, "step": 11210 }, { "epoch": 3.0431244914564686, "grad_norm": 10.733104983039201, "learning_rate": 9.681225043306906e-06, "loss": 3.0335, "step": 11220 }, { "epoch": 3.045836723623542, "grad_norm": 10.724211258430056, "learning_rate": 9.680115307442231e-06, "loss": 2.9583, "step": 11230 }, { "epoch": 3.048548955790616, "grad_norm": 10.902764684640736, "learning_rate": 9.679003707125598e-06, "loss": 2.9984, "step": 11240 }, { "epoch": 3.051261187957689, "grad_norm": 10.992669235278502, "learning_rate": 9.677890242799843e-06, "loss": 3.0368, "step": 11250 }, { "epoch": 3.0539734201247626, "grad_norm": 11.654270531365109, "learning_rate": 9.676774914908546e-06, "loss": 3.0546, "step": 11260 }, { "epoch": 3.056685652291836, "grad_norm": 10.722033045956922, "learning_rate": 9.675657723896027e-06, "loss": 2.9775, "step": 11270 }, { "epoch": 3.05939788445891, "grad_norm": 11.870774168954423, "learning_rate": 9.674538670207351e-06, "loss": 2.9951, "step": 11280 }, { "epoch": 3.062110116625983, "grad_norm": 10.381884632384942, "learning_rate": 9.673417754288323e-06, "loss": 3.0393, "step": 11290 }, { "epoch": 3.0648223487930566, "grad_norm": 11.895167373446938, "learning_rate": 9.672294976585489e-06, "loss": 3.0138, "step": 11300 }, { "epoch": 3.06753458096013, "grad_norm": 10.47313905756031, "learning_rate": 9.671170337546139e-06, "loss": 2.9943, "step": 11310 }, { "epoch": 3.070246813127204, "grad_norm": 11.885777714354866, "learning_rate": 9.670043837618306e-06, "loss": 3.0499, "step": 11320 }, { "epoch": 3.072959045294277, "grad_norm": 11.959766756941377, "learning_rate": 9.668915477250757e-06, "loss": 3.1811, "step": 11330 }, { "epoch": 3.0756712774613506, "grad_norm": 11.606140793308994, "learning_rate": 9.667785256893012e-06, "loss": 2.9988, "step": 11340 }, { "epoch": 3.078383509628424, "grad_norm": 11.124924630878885, "learning_rate": 9.66665317699532e-06, "loss": 2.9013, "step": 11350 }, { "epoch": 3.081095741795498, "grad_norm": 11.645506444948666, "learning_rate": 9.665519238008677e-06, "loss": 2.9497, "step": 11360 }, { "epoch": 3.0838079739625712, "grad_norm": 12.11427239706539, "learning_rate": 9.66438344038482e-06, "loss": 2.8624, "step": 11370 }, { "epoch": 3.0865202061296446, "grad_norm": 11.815344069463123, "learning_rate": 9.663245784576225e-06, "loss": 3.0231, "step": 11380 }, { "epoch": 3.089232438296718, "grad_norm": 13.02695277112266, "learning_rate": 9.662106271036106e-06, "loss": 2.9903, "step": 11390 }, { "epoch": 3.091944670463792, "grad_norm": 12.048683510592387, "learning_rate": 9.660964900218421e-06, "loss": 2.972, "step": 11400 }, { "epoch": 3.0946569026308652, "grad_norm": 12.145728373842505, "learning_rate": 9.659821672577866e-06, "loss": 2.9262, "step": 11410 }, { "epoch": 3.0973691347979386, "grad_norm": 11.380338089055892, "learning_rate": 9.65867658856988e-06, "loss": 2.9512, "step": 11420 }, { "epoch": 3.100081366965012, "grad_norm": 11.244537740889191, "learning_rate": 9.657529648650634e-06, "loss": 3.0546, "step": 11430 }, { "epoch": 3.102793599132086, "grad_norm": 11.187127208679843, "learning_rate": 9.656380853277045e-06, "loss": 3.0719, "step": 11440 }, { "epoch": 3.1055058312991592, "grad_norm": 10.635148193666314, "learning_rate": 9.655230202906767e-06, "loss": 2.9673, "step": 11450 }, { "epoch": 3.1082180634662326, "grad_norm": 11.608171994073471, "learning_rate": 9.654077697998191e-06, "loss": 3.0275, "step": 11460 }, { "epoch": 3.110930295633306, "grad_norm": 12.737646636820626, "learning_rate": 9.652923339010455e-06, "loss": 3.139, "step": 11470 }, { "epoch": 3.11364252780038, "grad_norm": 12.70135878093514, "learning_rate": 9.65176712640342e-06, "loss": 3.0449, "step": 11480 }, { "epoch": 3.1163547599674533, "grad_norm": 11.604820089360977, "learning_rate": 9.650609060637703e-06, "loss": 3.0905, "step": 11490 }, { "epoch": 3.1190669921345267, "grad_norm": 12.725731599018527, "learning_rate": 9.649449142174646e-06, "loss": 3.0002, "step": 11500 }, { "epoch": 3.1217792243016, "grad_norm": 12.218445476451668, "learning_rate": 9.64828737147634e-06, "loss": 3.0477, "step": 11510 }, { "epoch": 3.124491456468674, "grad_norm": 12.550128940911769, "learning_rate": 9.647123749005603e-06, "loss": 2.9345, "step": 11520 }, { "epoch": 3.1272036886357473, "grad_norm": 11.157604566908777, "learning_rate": 9.645958275225998e-06, "loss": 3.009, "step": 11530 }, { "epoch": 3.1299159208028207, "grad_norm": 11.751808921611053, "learning_rate": 9.644790950601821e-06, "loss": 2.9812, "step": 11540 }, { "epoch": 3.132628152969894, "grad_norm": 11.5578378325564, "learning_rate": 9.643621775598111e-06, "loss": 3.0836, "step": 11550 }, { "epoch": 3.135340385136968, "grad_norm": 11.704090569471386, "learning_rate": 9.642450750680638e-06, "loss": 2.9653, "step": 11560 }, { "epoch": 3.1380526173040413, "grad_norm": 12.611324207791647, "learning_rate": 9.641277876315915e-06, "loss": 2.9363, "step": 11570 }, { "epoch": 3.1407648494711147, "grad_norm": 11.344792077724774, "learning_rate": 9.640103152971187e-06, "loss": 2.9808, "step": 11580 }, { "epoch": 3.143477081638188, "grad_norm": 11.716110278081063, "learning_rate": 9.638926581114436e-06, "loss": 3.0076, "step": 11590 }, { "epoch": 3.146189313805262, "grad_norm": 12.658178369656964, "learning_rate": 9.637748161214382e-06, "loss": 3.0471, "step": 11600 }, { "epoch": 3.1489015459723353, "grad_norm": 11.699848797621163, "learning_rate": 9.636567893740483e-06, "loss": 3.0163, "step": 11610 }, { "epoch": 3.1516137781394087, "grad_norm": 11.043562410113532, "learning_rate": 9.635385779162926e-06, "loss": 3.1311, "step": 11620 }, { "epoch": 3.154326010306482, "grad_norm": 10.771208730172196, "learning_rate": 9.634201817952645e-06, "loss": 3.0625, "step": 11630 }, { "epoch": 3.157038242473556, "grad_norm": 10.46349832256357, "learning_rate": 9.6330160105813e-06, "loss": 2.9773, "step": 11640 }, { "epoch": 3.1597504746406293, "grad_norm": 11.847377921966348, "learning_rate": 9.631828357521289e-06, "loss": 3.1022, "step": 11650 }, { "epoch": 3.1624627068077027, "grad_norm": 11.438522853545896, "learning_rate": 9.630638859245747e-06, "loss": 3.0989, "step": 11660 }, { "epoch": 3.165174938974776, "grad_norm": 10.898501721458944, "learning_rate": 9.629447516228544e-06, "loss": 3.1104, "step": 11670 }, { "epoch": 3.16788717114185, "grad_norm": 11.195613224660148, "learning_rate": 9.628254328944283e-06, "loss": 3.0713, "step": 11680 }, { "epoch": 3.1705994033089233, "grad_norm": 10.218310292752651, "learning_rate": 9.627059297868303e-06, "loss": 3.0364, "step": 11690 }, { "epoch": 3.1733116354759967, "grad_norm": 10.62541870149064, "learning_rate": 9.625862423476678e-06, "loss": 3.0106, "step": 11700 }, { "epoch": 3.17602386764307, "grad_norm": 11.961629162333788, "learning_rate": 9.624663706246214e-06, "loss": 2.959, "step": 11710 }, { "epoch": 3.178736099810144, "grad_norm": 11.020752406873186, "learning_rate": 9.623463146654455e-06, "loss": 3.1408, "step": 11720 }, { "epoch": 3.1814483319772173, "grad_norm": 11.00542069689833, "learning_rate": 9.622260745179675e-06, "loss": 3.0526, "step": 11730 }, { "epoch": 3.1841605641442907, "grad_norm": 11.288837467698157, "learning_rate": 9.621056502300885e-06, "loss": 3.0242, "step": 11740 }, { "epoch": 3.186872796311364, "grad_norm": 11.76000841826907, "learning_rate": 9.619850418497826e-06, "loss": 3.0675, "step": 11750 }, { "epoch": 3.189585028478438, "grad_norm": 11.413820669614887, "learning_rate": 9.618642494250977e-06, "loss": 2.9928, "step": 11760 }, { "epoch": 3.1922972606455113, "grad_norm": 11.868711755452315, "learning_rate": 9.617432730041546e-06, "loss": 2.9846, "step": 11770 }, { "epoch": 3.1950094928125847, "grad_norm": 10.371658195986083, "learning_rate": 9.616221126351476e-06, "loss": 2.9757, "step": 11780 }, { "epoch": 3.197721724979658, "grad_norm": 10.65072886879809, "learning_rate": 9.615007683663444e-06, "loss": 3.0225, "step": 11790 }, { "epoch": 3.200433957146732, "grad_norm": 11.227291949330462, "learning_rate": 9.613792402460855e-06, "loss": 3.0903, "step": 11800 }, { "epoch": 3.2031461893138053, "grad_norm": 11.083897885295846, "learning_rate": 9.612575283227851e-06, "loss": 3.0224, "step": 11810 }, { "epoch": 3.2058584214808787, "grad_norm": 12.127338157900144, "learning_rate": 9.611356326449305e-06, "loss": 2.9848, "step": 11820 }, { "epoch": 3.208570653647952, "grad_norm": 10.528622098775031, "learning_rate": 9.610135532610823e-06, "loss": 3.0248, "step": 11830 }, { "epoch": 3.211282885815026, "grad_norm": 13.169729130322565, "learning_rate": 9.60891290219874e-06, "loss": 2.9309, "step": 11840 }, { "epoch": 3.2139951179820994, "grad_norm": 10.674143072053615, "learning_rate": 9.607688435700123e-06, "loss": 3.0415, "step": 11850 }, { "epoch": 3.2167073501491728, "grad_norm": 11.319966667951086, "learning_rate": 9.606462133602774e-06, "loss": 3.1127, "step": 11860 }, { "epoch": 3.219419582316246, "grad_norm": 11.611970674164807, "learning_rate": 9.605233996395225e-06, "loss": 3.0048, "step": 11870 }, { "epoch": 3.22213181448332, "grad_norm": 12.600584657292098, "learning_rate": 9.604004024566735e-06, "loss": 2.9964, "step": 11880 }, { "epoch": 3.2248440466503934, "grad_norm": 11.03257806440029, "learning_rate": 9.602772218607297e-06, "loss": 3.0253, "step": 11890 }, { "epoch": 3.2275562788174668, "grad_norm": 11.063522941885877, "learning_rate": 9.601538579007638e-06, "loss": 3.0413, "step": 11900 }, { "epoch": 3.23026851098454, "grad_norm": 12.409979773981474, "learning_rate": 9.600303106259208e-06, "loss": 3.0619, "step": 11910 }, { "epoch": 3.2329807431516135, "grad_norm": 11.825249483216535, "learning_rate": 9.599065800854195e-06, "loss": 3.0558, "step": 11920 }, { "epoch": 3.2356929753186874, "grad_norm": 11.24676891869841, "learning_rate": 9.597826663285512e-06, "loss": 3.1371, "step": 11930 }, { "epoch": 3.2384052074857608, "grad_norm": 12.556057552797714, "learning_rate": 9.5965856940468e-06, "loss": 3.053, "step": 11940 }, { "epoch": 3.241117439652834, "grad_norm": 10.663053523094515, "learning_rate": 9.59534289363244e-06, "loss": 2.9889, "step": 11950 }, { "epoch": 3.243829671819908, "grad_norm": 10.53764608254658, "learning_rate": 9.59409826253753e-06, "loss": 3.0049, "step": 11960 }, { "epoch": 3.2465419039869814, "grad_norm": 11.343314980679333, "learning_rate": 9.592851801257905e-06, "loss": 3.0353, "step": 11970 }, { "epoch": 3.249254136154055, "grad_norm": 12.09280635432137, "learning_rate": 9.591603510290127e-06, "loss": 3.236, "step": 11980 }, { "epoch": 3.251966368321128, "grad_norm": 10.995301952151857, "learning_rate": 9.590353390131483e-06, "loss": 3.0361, "step": 11990 }, { "epoch": 3.2546786004882016, "grad_norm": 12.816476563167843, "learning_rate": 9.58910144128e-06, "loss": 2.9848, "step": 12000 }, { "epoch": 3.2573908326552754, "grad_norm": 11.813047150431085, "learning_rate": 9.58784766423442e-06, "loss": 3.0068, "step": 12010 }, { "epoch": 3.260103064822349, "grad_norm": 12.144950963303746, "learning_rate": 9.586592059494222e-06, "loss": 3.01, "step": 12020 }, { "epoch": 3.262815296989422, "grad_norm": 12.0712513386369, "learning_rate": 9.58533462755961e-06, "loss": 3.0827, "step": 12030 }, { "epoch": 3.265527529156496, "grad_norm": 11.762561899653539, "learning_rate": 9.584075368931516e-06, "loss": 3.0638, "step": 12040 }, { "epoch": 3.2682397613235694, "grad_norm": 11.78656631477849, "learning_rate": 9.5828142841116e-06, "loss": 3.0655, "step": 12050 }, { "epoch": 3.270951993490643, "grad_norm": 11.325018436080722, "learning_rate": 9.581551373602251e-06, "loss": 3.1435, "step": 12060 }, { "epoch": 3.273664225657716, "grad_norm": 11.782251153550625, "learning_rate": 9.580286637906581e-06, "loss": 3.1106, "step": 12070 }, { "epoch": 3.2763764578247896, "grad_norm": 11.800262131425331, "learning_rate": 9.579020077528437e-06, "loss": 3.081, "step": 12080 }, { "epoch": 3.2790886899918634, "grad_norm": 12.265683099525484, "learning_rate": 9.577751692972383e-06, "loss": 3.0124, "step": 12090 }, { "epoch": 3.281800922158937, "grad_norm": 10.94125605258509, "learning_rate": 9.576481484743717e-06, "loss": 2.9408, "step": 12100 }, { "epoch": 3.28451315432601, "grad_norm": 11.67009905019906, "learning_rate": 9.575209453348461e-06, "loss": 3.0794, "step": 12110 }, { "epoch": 3.287225386493084, "grad_norm": 11.537374740181109, "learning_rate": 9.573935599293363e-06, "loss": 2.9288, "step": 12120 }, { "epoch": 3.2899376186601574, "grad_norm": 11.501672041036317, "learning_rate": 9.572659923085898e-06, "loss": 3.0571, "step": 12130 }, { "epoch": 3.292649850827231, "grad_norm": 11.818615548547962, "learning_rate": 9.571382425234269e-06, "loss": 2.9921, "step": 12140 }, { "epoch": 3.295362082994304, "grad_norm": 10.367421731329298, "learning_rate": 9.570103106247396e-06, "loss": 3.0119, "step": 12150 }, { "epoch": 3.2980743151613776, "grad_norm": 10.737978092129882, "learning_rate": 9.568821966634938e-06, "loss": 3.0546, "step": 12160 }, { "epoch": 3.3007865473284514, "grad_norm": 10.947040730417713, "learning_rate": 9.567539006907265e-06, "loss": 3.1028, "step": 12170 }, { "epoch": 3.303498779495525, "grad_norm": 11.179210207337603, "learning_rate": 9.566254227575484e-06, "loss": 3.1697, "step": 12180 }, { "epoch": 3.3062110116625982, "grad_norm": 12.356103665727655, "learning_rate": 9.564967629151422e-06, "loss": 2.9146, "step": 12190 }, { "epoch": 3.308923243829672, "grad_norm": 12.173731778950735, "learning_rate": 9.563679212147627e-06, "loss": 3.0194, "step": 12200 }, { "epoch": 3.3116354759967455, "grad_norm": 10.707767375910507, "learning_rate": 9.56238897707738e-06, "loss": 3.1382, "step": 12210 }, { "epoch": 3.314347708163819, "grad_norm": 11.517534516000755, "learning_rate": 9.561096924454676e-06, "loss": 3.1362, "step": 12220 }, { "epoch": 3.3170599403308922, "grad_norm": 11.632919919367666, "learning_rate": 9.559803054794244e-06, "loss": 3.0779, "step": 12230 }, { "epoch": 3.3197721724979656, "grad_norm": 12.692622232028217, "learning_rate": 9.55850736861153e-06, "loss": 3.0796, "step": 12240 }, { "epoch": 3.3224844046650395, "grad_norm": 12.576594127040224, "learning_rate": 9.557209866422707e-06, "loss": 3.102, "step": 12250 }, { "epoch": 3.325196636832113, "grad_norm": 10.455021292518971, "learning_rate": 9.55591054874467e-06, "loss": 3.1162, "step": 12260 }, { "epoch": 3.3279088689991863, "grad_norm": 12.139191140675484, "learning_rate": 9.554609416095038e-06, "loss": 3.0383, "step": 12270 }, { "epoch": 3.3306211011662596, "grad_norm": 10.998773598362314, "learning_rate": 9.553306468992153e-06, "loss": 3.0624, "step": 12280 }, { "epoch": 3.3333333333333335, "grad_norm": 11.724481103058018, "learning_rate": 9.55200170795508e-06, "loss": 3.1375, "step": 12290 }, { "epoch": 3.336045565500407, "grad_norm": 13.727922667231674, "learning_rate": 9.550695133503605e-06, "loss": 3.0781, "step": 12300 }, { "epoch": 3.3387577976674803, "grad_norm": 11.816398571721821, "learning_rate": 9.549386746158237e-06, "loss": 3.0802, "step": 12310 }, { "epoch": 3.3414700298345537, "grad_norm": 11.51977484115356, "learning_rate": 9.548076546440211e-06, "loss": 3.0998, "step": 12320 }, { "epoch": 3.3441822620016275, "grad_norm": 10.888229966147986, "learning_rate": 9.546764534871478e-06, "loss": 3.0659, "step": 12330 }, { "epoch": 3.346894494168701, "grad_norm": 11.504669634855372, "learning_rate": 9.545450711974715e-06, "loss": 3.0913, "step": 12340 }, { "epoch": 3.3496067263357743, "grad_norm": 11.346017715030149, "learning_rate": 9.54413507827332e-06, "loss": 3.019, "step": 12350 }, { "epoch": 3.3523189585028477, "grad_norm": 12.686482230884591, "learning_rate": 9.54281763429141e-06, "loss": 3.0777, "step": 12360 }, { "epoch": 3.3550311906699215, "grad_norm": 9.966581547424655, "learning_rate": 9.541498380553827e-06, "loss": 3.0829, "step": 12370 }, { "epoch": 3.357743422836995, "grad_norm": 11.702195056612089, "learning_rate": 9.540177317586128e-06, "loss": 3.0625, "step": 12380 }, { "epoch": 3.3604556550040683, "grad_norm": 12.628055352577206, "learning_rate": 9.538854445914598e-06, "loss": 3.0125, "step": 12390 }, { "epoch": 3.3631678871711417, "grad_norm": 11.67240553852873, "learning_rate": 9.537529766066237e-06, "loss": 2.9867, "step": 12400 }, { "epoch": 3.3658801193382155, "grad_norm": 12.116750164076146, "learning_rate": 9.536203278568772e-06, "loss": 2.9561, "step": 12410 }, { "epoch": 3.368592351505289, "grad_norm": 12.51150706543475, "learning_rate": 9.534874983950639e-06, "loss": 2.9965, "step": 12420 }, { "epoch": 3.3713045836723623, "grad_norm": 11.957363908552859, "learning_rate": 9.533544882741007e-06, "loss": 3.0006, "step": 12430 }, { "epoch": 3.3740168158394357, "grad_norm": 11.04123635528423, "learning_rate": 9.53221297546975e-06, "loss": 3.0915, "step": 12440 }, { "epoch": 3.3767290480065095, "grad_norm": 11.705071688833986, "learning_rate": 9.530879262667478e-06, "loss": 3.0675, "step": 12450 }, { "epoch": 3.379441280173583, "grad_norm": 10.54467149048281, "learning_rate": 9.529543744865507e-06, "loss": 2.9501, "step": 12460 }, { "epoch": 3.3821535123406563, "grad_norm": 11.26951651085401, "learning_rate": 9.528206422595879e-06, "loss": 3.0221, "step": 12470 }, { "epoch": 3.3848657445077297, "grad_norm": 10.479226498982563, "learning_rate": 9.526867296391353e-06, "loss": 3.1162, "step": 12480 }, { "epoch": 3.3875779766748035, "grad_norm": 10.927071877949368, "learning_rate": 9.525526366785405e-06, "loss": 3.1229, "step": 12490 }, { "epoch": 3.390290208841877, "grad_norm": 10.282877008781082, "learning_rate": 9.524183634312233e-06, "loss": 3.1634, "step": 12500 }, { "epoch": 3.3930024410089503, "grad_norm": 11.175382751461113, "learning_rate": 9.52283909950675e-06, "loss": 3.1243, "step": 12510 }, { "epoch": 3.3957146731760237, "grad_norm": 11.285451023799476, "learning_rate": 9.521492762904588e-06, "loss": 3.0228, "step": 12520 }, { "epoch": 3.3984269053430975, "grad_norm": 12.561194135038823, "learning_rate": 9.520144625042098e-06, "loss": 3.008, "step": 12530 }, { "epoch": 3.401139137510171, "grad_norm": 12.488778330073332, "learning_rate": 9.518794686456347e-06, "loss": 2.9811, "step": 12540 }, { "epoch": 3.4038513696772443, "grad_norm": 11.112821176949089, "learning_rate": 9.51744294768512e-06, "loss": 2.979, "step": 12550 }, { "epoch": 3.4065636018443177, "grad_norm": 11.23871457570514, "learning_rate": 9.516089409266918e-06, "loss": 3.0413, "step": 12560 }, { "epoch": 3.4092758340113916, "grad_norm": 11.444536627220538, "learning_rate": 9.514734071740963e-06, "loss": 2.9698, "step": 12570 }, { "epoch": 3.411988066178465, "grad_norm": 11.38950692477146, "learning_rate": 9.513376935647188e-06, "loss": 3.067, "step": 12580 }, { "epoch": 3.4147002983455383, "grad_norm": 10.891673972521964, "learning_rate": 9.512018001526246e-06, "loss": 3.1777, "step": 12590 }, { "epoch": 3.4174125305126117, "grad_norm": 11.380169379896985, "learning_rate": 9.510657269919508e-06, "loss": 2.9988, "step": 12600 }, { "epoch": 3.4201247626796856, "grad_norm": 13.058064754836357, "learning_rate": 9.509294741369054e-06, "loss": 3.0014, "step": 12610 }, { "epoch": 3.422836994846759, "grad_norm": 10.685841736468568, "learning_rate": 9.507930416417688e-06, "loss": 3.1797, "step": 12620 }, { "epoch": 3.4255492270138324, "grad_norm": 13.435665153672996, "learning_rate": 9.506564295608925e-06, "loss": 2.9228, "step": 12630 }, { "epoch": 3.4282614591809057, "grad_norm": 11.195758114445422, "learning_rate": 9.505196379486995e-06, "loss": 3.0727, "step": 12640 }, { "epoch": 3.4309736913479796, "grad_norm": 12.767427132750939, "learning_rate": 9.50382666859685e-06, "loss": 3.0518, "step": 12650 }, { "epoch": 3.433685923515053, "grad_norm": 11.335558047815253, "learning_rate": 9.502455163484146e-06, "loss": 3.0444, "step": 12660 }, { "epoch": 3.4363981556821264, "grad_norm": 10.93319733678939, "learning_rate": 9.501081864695262e-06, "loss": 3.0413, "step": 12670 }, { "epoch": 3.4391103878491998, "grad_norm": 13.638530047263227, "learning_rate": 9.49970677277729e-06, "loss": 3.0616, "step": 12680 }, { "epoch": 3.441822620016273, "grad_norm": 11.097101989544203, "learning_rate": 9.498329888278035e-06, "loss": 3.1437, "step": 12690 }, { "epoch": 3.444534852183347, "grad_norm": 11.608634588880639, "learning_rate": 9.496951211746018e-06, "loss": 3.0617, "step": 12700 }, { "epoch": 3.4472470843504204, "grad_norm": 15.580803993992447, "learning_rate": 9.495570743730466e-06, "loss": 3.0856, "step": 12710 }, { "epoch": 3.4499593165174938, "grad_norm": 13.156750513953199, "learning_rate": 9.494188484781335e-06, "loss": 2.9589, "step": 12720 }, { "epoch": 3.4526715486845676, "grad_norm": 10.674706558210048, "learning_rate": 9.492804435449278e-06, "loss": 3.152, "step": 12730 }, { "epoch": 3.455383780851641, "grad_norm": 11.636309047365422, "learning_rate": 9.491418596285675e-06, "loss": 2.9783, "step": 12740 }, { "epoch": 3.4580960130187144, "grad_norm": 11.241072911633, "learning_rate": 9.49003096784261e-06, "loss": 3.0083, "step": 12750 }, { "epoch": 3.460808245185788, "grad_norm": 10.541905638781358, "learning_rate": 9.488641550672883e-06, "loss": 3.0768, "step": 12760 }, { "epoch": 3.463520477352861, "grad_norm": 12.244600826903524, "learning_rate": 9.487250345330006e-06, "loss": 3.0953, "step": 12770 }, { "epoch": 3.466232709519935, "grad_norm": 12.126150209222072, "learning_rate": 9.485857352368206e-06, "loss": 3.1033, "step": 12780 }, { "epoch": 3.4689449416870084, "grad_norm": 11.417231229561393, "learning_rate": 9.484462572342417e-06, "loss": 3.0922, "step": 12790 }, { "epoch": 3.471657173854082, "grad_norm": 11.11680879842505, "learning_rate": 9.483066005808287e-06, "loss": 3.1248, "step": 12800 }, { "epoch": 3.4743694060211556, "grad_norm": 10.739969294554044, "learning_rate": 9.481667653322181e-06, "loss": 3.1643, "step": 12810 }, { "epoch": 3.477081638188229, "grad_norm": 11.083898485175151, "learning_rate": 9.480267515441166e-06, "loss": 3.0527, "step": 12820 }, { "epoch": 3.4797938703553024, "grad_norm": 11.725300599062622, "learning_rate": 9.47886559272303e-06, "loss": 3.0448, "step": 12830 }, { "epoch": 3.482506102522376, "grad_norm": 11.413650283769963, "learning_rate": 9.477461885726263e-06, "loss": 3.1189, "step": 12840 }, { "epoch": 3.485218334689449, "grad_norm": 12.64323112294165, "learning_rate": 9.476056395010073e-06, "loss": 3.0413, "step": 12850 }, { "epoch": 3.487930566856523, "grad_norm": 11.446588381704128, "learning_rate": 9.474649121134375e-06, "loss": 3.13, "step": 12860 }, { "epoch": 3.4906427990235964, "grad_norm": 10.653280420155198, "learning_rate": 9.473240064659792e-06, "loss": 3.1238, "step": 12870 }, { "epoch": 3.49335503119067, "grad_norm": 11.221633359233289, "learning_rate": 9.471829226147664e-06, "loss": 3.0941, "step": 12880 }, { "epoch": 3.4960672633577436, "grad_norm": 11.183388773803038, "learning_rate": 9.470416606160037e-06, "loss": 3.1341, "step": 12890 }, { "epoch": 3.498779495524817, "grad_norm": 10.241966529909218, "learning_rate": 9.469002205259664e-06, "loss": 3.0118, "step": 12900 }, { "epoch": 3.5014917276918904, "grad_norm": 11.917118419641191, "learning_rate": 9.467586024010014e-06, "loss": 3.1118, "step": 12910 }, { "epoch": 3.504203959858964, "grad_norm": 10.455706944713013, "learning_rate": 9.466168062975258e-06, "loss": 3.077, "step": 12920 }, { "epoch": 3.506916192026037, "grad_norm": 10.55165444251888, "learning_rate": 9.46474832272028e-06, "loss": 3.1157, "step": 12930 }, { "epoch": 3.509628424193111, "grad_norm": 10.541126093032, "learning_rate": 9.463326803810676e-06, "loss": 3.1132, "step": 12940 }, { "epoch": 3.5123406563601844, "grad_norm": 11.78301200684155, "learning_rate": 9.461903506812745e-06, "loss": 3.0078, "step": 12950 }, { "epoch": 3.515052888527258, "grad_norm": 12.070016229488912, "learning_rate": 9.460478432293494e-06, "loss": 3.0795, "step": 12960 }, { "epoch": 3.5177651206943317, "grad_norm": 13.610744104458723, "learning_rate": 9.459051580820641e-06, "loss": 2.9826, "step": 12970 }, { "epoch": 3.520477352861405, "grad_norm": 10.907197661798278, "learning_rate": 9.457622952962614e-06, "loss": 3.0856, "step": 12980 }, { "epoch": 3.5231895850284785, "grad_norm": 10.418517477951424, "learning_rate": 9.456192549288544e-06, "loss": 3.0612, "step": 12990 }, { "epoch": 3.525901817195552, "grad_norm": 10.991844301529525, "learning_rate": 9.454760370368272e-06, "loss": 3.0101, "step": 13000 }, { "epoch": 3.5286140493626252, "grad_norm": 11.267306539451093, "learning_rate": 9.453326416772346e-06, "loss": 3.0259, "step": 13010 }, { "epoch": 3.531326281529699, "grad_norm": 11.840400002691453, "learning_rate": 9.451890689072019e-06, "loss": 2.9954, "step": 13020 }, { "epoch": 3.5340385136967725, "grad_norm": 11.328440236752005, "learning_rate": 9.450453187839255e-06, "loss": 3.1014, "step": 13030 }, { "epoch": 3.536750745863846, "grad_norm": 11.927378290222375, "learning_rate": 9.449013913646718e-06, "loss": 3.0909, "step": 13040 }, { "epoch": 3.5394629780309197, "grad_norm": 12.16395113251231, "learning_rate": 9.447572867067786e-06, "loss": 3.0752, "step": 13050 }, { "epoch": 3.542175210197993, "grad_norm": 11.059035000512074, "learning_rate": 9.44613004867654e-06, "loss": 3.1445, "step": 13060 }, { "epoch": 3.5448874423650665, "grad_norm": 9.841210378678433, "learning_rate": 9.444685459047763e-06, "loss": 3.19, "step": 13070 }, { "epoch": 3.54759967453214, "grad_norm": 11.559451736113772, "learning_rate": 9.443239098756948e-06, "loss": 3.1493, "step": 13080 }, { "epoch": 3.5503119066992133, "grad_norm": 12.786794426341041, "learning_rate": 9.441790968380293e-06, "loss": 3.0153, "step": 13090 }, { "epoch": 3.553024138866287, "grad_norm": 13.13291310936245, "learning_rate": 9.440341068494698e-06, "loss": 3.0522, "step": 13100 }, { "epoch": 3.5557363710333605, "grad_norm": 10.992759381592839, "learning_rate": 9.438889399677772e-06, "loss": 3.2156, "step": 13110 }, { "epoch": 3.558448603200434, "grad_norm": 10.24631616768907, "learning_rate": 9.437435962507827e-06, "loss": 3.14, "step": 13120 }, { "epoch": 3.5611608353675077, "grad_norm": 11.15124699787642, "learning_rate": 9.43598075756388e-06, "loss": 3.1862, "step": 13130 }, { "epoch": 3.563873067534581, "grad_norm": 10.190808897955321, "learning_rate": 9.434523785425651e-06, "loss": 3.0063, "step": 13140 }, { "epoch": 3.5665852997016545, "grad_norm": 12.348055235015147, "learning_rate": 9.433065046673564e-06, "loss": 3.1307, "step": 13150 }, { "epoch": 3.569297531868728, "grad_norm": 11.674779103195297, "learning_rate": 9.431604541888748e-06, "loss": 2.9996, "step": 13160 }, { "epoch": 3.5720097640358013, "grad_norm": 11.326903526665856, "learning_rate": 9.430142271653036e-06, "loss": 3.1447, "step": 13170 }, { "epoch": 3.574721996202875, "grad_norm": 10.442054990930492, "learning_rate": 9.428678236548963e-06, "loss": 3.143, "step": 13180 }, { "epoch": 3.5774342283699485, "grad_norm": 10.63074815563649, "learning_rate": 9.427212437159765e-06, "loss": 3.0562, "step": 13190 }, { "epoch": 3.580146460537022, "grad_norm": 11.179977679096751, "learning_rate": 9.425744874069387e-06, "loss": 3.0474, "step": 13200 }, { "epoch": 3.5828586927040957, "grad_norm": 10.852824805724302, "learning_rate": 9.424275547862471e-06, "loss": 2.9213, "step": 13210 }, { "epoch": 3.585570924871169, "grad_norm": 11.2373223992278, "learning_rate": 9.422804459124365e-06, "loss": 3.2096, "step": 13220 }, { "epoch": 3.5882831570382425, "grad_norm": 12.899622022545087, "learning_rate": 9.421331608441114e-06, "loss": 3.0292, "step": 13230 }, { "epoch": 3.590995389205316, "grad_norm": 10.869079048944702, "learning_rate": 9.41985699639947e-06, "loss": 3.0609, "step": 13240 }, { "epoch": 3.5937076213723893, "grad_norm": 11.449595064686077, "learning_rate": 9.418380623586885e-06, "loss": 3.1585, "step": 13250 }, { "epoch": 3.596419853539463, "grad_norm": 11.977241503252927, "learning_rate": 9.416902490591514e-06, "loss": 3.0561, "step": 13260 }, { "epoch": 3.5991320857065365, "grad_norm": 11.24385060141734, "learning_rate": 9.41542259800221e-06, "loss": 3.066, "step": 13270 }, { "epoch": 3.60184431787361, "grad_norm": 11.769669589972121, "learning_rate": 9.413940946408529e-06, "loss": 3.0835, "step": 13280 }, { "epoch": 3.6045565500406838, "grad_norm": 11.949888190935273, "learning_rate": 9.412457536400728e-06, "loss": 3.0078, "step": 13290 }, { "epoch": 3.6072687822077567, "grad_norm": 13.402866336681539, "learning_rate": 9.410972368569763e-06, "loss": 3.1584, "step": 13300 }, { "epoch": 3.6099810143748305, "grad_norm": 11.489118095314213, "learning_rate": 9.409485443507294e-06, "loss": 3.105, "step": 13310 }, { "epoch": 3.612693246541904, "grad_norm": 11.750301552653873, "learning_rate": 9.407996761805674e-06, "loss": 3.0222, "step": 13320 }, { "epoch": 3.6154054787089773, "grad_norm": 10.830485889895906, "learning_rate": 9.406506324057962e-06, "loss": 3.033, "step": 13330 }, { "epoch": 3.618117710876051, "grad_norm": 11.69576012798874, "learning_rate": 9.405014130857917e-06, "loss": 3.0757, "step": 13340 }, { "epoch": 3.6208299430431246, "grad_norm": 11.07077852370341, "learning_rate": 9.403520182799992e-06, "loss": 3.0479, "step": 13350 }, { "epoch": 3.623542175210198, "grad_norm": 11.079870195741872, "learning_rate": 9.40202448047934e-06, "loss": 3.1476, "step": 13360 }, { "epoch": 3.6262544073772713, "grad_norm": 11.68556374092015, "learning_rate": 9.400527024491822e-06, "loss": 3.033, "step": 13370 }, { "epoch": 3.6289666395443447, "grad_norm": 11.137323402621524, "learning_rate": 9.399027815433985e-06, "loss": 2.971, "step": 13380 }, { "epoch": 3.6316788717114186, "grad_norm": 10.132071321964336, "learning_rate": 9.397526853903081e-06, "loss": 3.141, "step": 13390 }, { "epoch": 3.634391103878492, "grad_norm": 11.65973377440923, "learning_rate": 9.39602414049706e-06, "loss": 3.1832, "step": 13400 }, { "epoch": 3.6371033360455653, "grad_norm": 13.332429701208943, "learning_rate": 9.394519675814568e-06, "loss": 3.0482, "step": 13410 }, { "epoch": 3.639815568212639, "grad_norm": 12.302400919554815, "learning_rate": 9.393013460454949e-06, "loss": 3.0716, "step": 13420 }, { "epoch": 3.6425278003797126, "grad_norm": 12.346267525187685, "learning_rate": 9.391505495018247e-06, "loss": 3.1222, "step": 13430 }, { "epoch": 3.645240032546786, "grad_norm": 10.531927771419673, "learning_rate": 9.389995780105201e-06, "loss": 3.1402, "step": 13440 }, { "epoch": 3.6479522647138594, "grad_norm": 11.468922929148496, "learning_rate": 9.388484316317247e-06, "loss": 3.0308, "step": 13450 }, { "epoch": 3.6506644968809328, "grad_norm": 11.764418257994782, "learning_rate": 9.386971104256517e-06, "loss": 3.1379, "step": 13460 }, { "epoch": 3.6533767290480066, "grad_norm": 11.77327381980495, "learning_rate": 9.385456144525842e-06, "loss": 3.0996, "step": 13470 }, { "epoch": 3.65608896121508, "grad_norm": 10.86167263183151, "learning_rate": 9.383939437728745e-06, "loss": 3.0005, "step": 13480 }, { "epoch": 3.6588011933821534, "grad_norm": 12.057149401977659, "learning_rate": 9.38242098446945e-06, "loss": 3.1157, "step": 13490 }, { "epoch": 3.661513425549227, "grad_norm": 10.941692505183202, "learning_rate": 9.380900785352874e-06, "loss": 3.091, "step": 13500 }, { "epoch": 3.6642256577163006, "grad_norm": 12.948395522506699, "learning_rate": 9.379378840984629e-06, "loss": 3.0918, "step": 13510 }, { "epoch": 3.666937889883374, "grad_norm": 11.67336917811239, "learning_rate": 9.377855151971025e-06, "loss": 3.0864, "step": 13520 }, { "epoch": 3.6696501220504474, "grad_norm": 10.699322044922129, "learning_rate": 9.376329718919065e-06, "loss": 3.0878, "step": 13530 }, { "epoch": 3.6723623542175208, "grad_norm": 11.724440961578296, "learning_rate": 9.374802542436443e-06, "loss": 3.0255, "step": 13540 }, { "epoch": 3.6750745863845946, "grad_norm": 10.834424807697852, "learning_rate": 9.373273623131556e-06, "loss": 3.0931, "step": 13550 }, { "epoch": 3.677786818551668, "grad_norm": 10.257236462134204, "learning_rate": 9.37174296161349e-06, "loss": 2.9836, "step": 13560 }, { "epoch": 3.6804990507187414, "grad_norm": 10.982043876774572, "learning_rate": 9.370210558492024e-06, "loss": 3.1264, "step": 13570 }, { "epoch": 3.6832112828858152, "grad_norm": 12.534591487077344, "learning_rate": 9.368676414377634e-06, "loss": 3.1147, "step": 13580 }, { "epoch": 3.6859235150528886, "grad_norm": 11.071134105151831, "learning_rate": 9.367140529881489e-06, "loss": 3.1108, "step": 13590 }, { "epoch": 3.688635747219962, "grad_norm": 10.538686405994069, "learning_rate": 9.36560290561545e-06, "loss": 3.1119, "step": 13600 }, { "epoch": 3.6913479793870354, "grad_norm": 11.10612092545836, "learning_rate": 9.36406354219207e-06, "loss": 3.2122, "step": 13610 }, { "epoch": 3.694060211554109, "grad_norm": 11.361511110771488, "learning_rate": 9.362522440224599e-06, "loss": 3.1751, "step": 13620 }, { "epoch": 3.6967724437211826, "grad_norm": 11.305897591455496, "learning_rate": 9.360979600326976e-06, "loss": 3.1218, "step": 13630 }, { "epoch": 3.699484675888256, "grad_norm": 11.395894028920347, "learning_rate": 9.359435023113835e-06, "loss": 3.1221, "step": 13640 }, { "epoch": 3.7021969080553294, "grad_norm": 10.710046349129446, "learning_rate": 9.357888709200498e-06, "loss": 3.0446, "step": 13650 }, { "epoch": 3.7049091402224033, "grad_norm": 11.232567497865213, "learning_rate": 9.356340659202983e-06, "loss": 3.0853, "step": 13660 }, { "epoch": 3.7076213723894766, "grad_norm": 9.827060315239661, "learning_rate": 9.354790873737997e-06, "loss": 3.2969, "step": 13670 }, { "epoch": 3.71033360455655, "grad_norm": 11.48195015137984, "learning_rate": 9.353239353422941e-06, "loss": 3.0711, "step": 13680 }, { "epoch": 3.7130458367236234, "grad_norm": 10.970359141072299, "learning_rate": 9.351686098875905e-06, "loss": 3.0503, "step": 13690 }, { "epoch": 3.715758068890697, "grad_norm": 11.515384953708654, "learning_rate": 9.35013111071567e-06, "loss": 3.1191, "step": 13700 }, { "epoch": 3.7184703010577707, "grad_norm": 11.562885593058141, "learning_rate": 9.348574389561709e-06, "loss": 3.0748, "step": 13710 }, { "epoch": 3.721182533224844, "grad_norm": 14.018213496695196, "learning_rate": 9.347015936034182e-06, "loss": 3.0414, "step": 13720 }, { "epoch": 3.7238947653919174, "grad_norm": 10.484610446818694, "learning_rate": 9.345455750753945e-06, "loss": 3.1598, "step": 13730 }, { "epoch": 3.7266069975589913, "grad_norm": 12.373034304724982, "learning_rate": 9.343893834342537e-06, "loss": 3.1287, "step": 13740 }, { "epoch": 3.7293192297260647, "grad_norm": 13.152696526551193, "learning_rate": 9.342330187422193e-06, "loss": 3.0735, "step": 13750 }, { "epoch": 3.732031461893138, "grad_norm": 10.819737046954375, "learning_rate": 9.340764810615832e-06, "loss": 3.1236, "step": 13760 }, { "epoch": 3.7347436940602114, "grad_norm": 10.936192636707878, "learning_rate": 9.339197704547068e-06, "loss": 3.1593, "step": 13770 }, { "epoch": 3.737455926227285, "grad_norm": 12.087207090401865, "learning_rate": 9.337628869840199e-06, "loss": 3.0706, "step": 13780 }, { "epoch": 3.7401681583943587, "grad_norm": 10.978700227256224, "learning_rate": 9.33605830712021e-06, "loss": 3.1317, "step": 13790 }, { "epoch": 3.742880390561432, "grad_norm": 11.836475928365878, "learning_rate": 9.334486017012782e-06, "loss": 3.158, "step": 13800 }, { "epoch": 3.7455926227285055, "grad_norm": 11.239552869395771, "learning_rate": 9.33291200014428e-06, "loss": 3.1066, "step": 13810 }, { "epoch": 3.7483048548955793, "grad_norm": 11.247509013847699, "learning_rate": 9.331336257141754e-06, "loss": 3.0643, "step": 13820 }, { "epoch": 3.7510170870626527, "grad_norm": 10.546106947809713, "learning_rate": 9.329758788632945e-06, "loss": 3.0801, "step": 13830 }, { "epoch": 3.753729319229726, "grad_norm": 10.891150384165524, "learning_rate": 9.32817959524628e-06, "loss": 3.1608, "step": 13840 }, { "epoch": 3.7564415513967995, "grad_norm": 10.82185278266411, "learning_rate": 9.326598677610877e-06, "loss": 3.1182, "step": 13850 }, { "epoch": 3.759153783563873, "grad_norm": 13.383382107747348, "learning_rate": 9.325016036356533e-06, "loss": 2.9674, "step": 13860 }, { "epoch": 3.7618660157309467, "grad_norm": 11.30402698828102, "learning_rate": 9.323431672113741e-06, "loss": 3.1005, "step": 13870 }, { "epoch": 3.76457824789802, "grad_norm": 11.763451378565403, "learning_rate": 9.321845585513675e-06, "loss": 3.1188, "step": 13880 }, { "epoch": 3.7672904800650935, "grad_norm": 14.165047305021604, "learning_rate": 9.320257777188193e-06, "loss": 3.1515, "step": 13890 }, { "epoch": 3.7700027122321673, "grad_norm": 11.620179176282544, "learning_rate": 9.318668247769846e-06, "loss": 2.9781, "step": 13900 }, { "epoch": 3.7727149443992407, "grad_norm": 10.39104388829201, "learning_rate": 9.317076997891863e-06, "loss": 2.9986, "step": 13910 }, { "epoch": 3.775427176566314, "grad_norm": 11.124109012965054, "learning_rate": 9.315484028188163e-06, "loss": 3.1429, "step": 13920 }, { "epoch": 3.7781394087333875, "grad_norm": 11.49419467072653, "learning_rate": 9.313889339293349e-06, "loss": 3.0979, "step": 13930 }, { "epoch": 3.780851640900461, "grad_norm": 11.876980781889488, "learning_rate": 9.312292931842711e-06, "loss": 3.1554, "step": 13940 }, { "epoch": 3.7835638730675347, "grad_norm": 11.273230293282468, "learning_rate": 9.310694806472218e-06, "loss": 3.122, "step": 13950 }, { "epoch": 3.786276105234608, "grad_norm": 11.268057694486075, "learning_rate": 9.30909496381853e-06, "loss": 3.062, "step": 13960 }, { "epoch": 3.7889883374016815, "grad_norm": 10.671454202234983, "learning_rate": 9.307493404518986e-06, "loss": 3.1292, "step": 13970 }, { "epoch": 3.7917005695687553, "grad_norm": 11.035798307628033, "learning_rate": 9.305890129211613e-06, "loss": 3.0861, "step": 13980 }, { "epoch": 3.7944128017358287, "grad_norm": 11.527961927190296, "learning_rate": 9.304285138535119e-06, "loss": 3.1257, "step": 13990 }, { "epoch": 3.797125033902902, "grad_norm": 10.648035099157472, "learning_rate": 9.302678433128893e-06, "loss": 3.1402, "step": 14000 }, { "epoch": 3.7998372660699755, "grad_norm": 11.163179766071217, "learning_rate": 9.301070013633013e-06, "loss": 3.139, "step": 14010 }, { "epoch": 3.802549498237049, "grad_norm": 13.673256654184025, "learning_rate": 9.299459880688239e-06, "loss": 3.1242, "step": 14020 }, { "epoch": 3.8052617304041227, "grad_norm": 12.032863293656641, "learning_rate": 9.297848034936007e-06, "loss": 3.2012, "step": 14030 }, { "epoch": 3.807973962571196, "grad_norm": 11.47565047764239, "learning_rate": 9.296234477018441e-06, "loss": 3.1589, "step": 14040 }, { "epoch": 3.8106861947382695, "grad_norm": 11.920929479237344, "learning_rate": 9.294619207578349e-06, "loss": 2.9626, "step": 14050 }, { "epoch": 3.8133984269053434, "grad_norm": 10.59638081936334, "learning_rate": 9.293002227259212e-06, "loss": 3.1371, "step": 14060 }, { "epoch": 3.8161106590724168, "grad_norm": 11.397039155984338, "learning_rate": 9.291383536705204e-06, "loss": 3.1344, "step": 14070 }, { "epoch": 3.81882289123949, "grad_norm": 11.709289081474438, "learning_rate": 9.289763136561172e-06, "loss": 3.1654, "step": 14080 }, { "epoch": 3.8215351234065635, "grad_norm": 10.595151894856611, "learning_rate": 9.288141027472647e-06, "loss": 3.0088, "step": 14090 }, { "epoch": 3.824247355573637, "grad_norm": 13.489420473225257, "learning_rate": 9.286517210085841e-06, "loss": 3.0015, "step": 14100 }, { "epoch": 3.8269595877407108, "grad_norm": 12.567459450807132, "learning_rate": 9.284891685047646e-06, "loss": 3.0481, "step": 14110 }, { "epoch": 3.829671819907784, "grad_norm": 10.80263452962618, "learning_rate": 9.283264453005634e-06, "loss": 3.0658, "step": 14120 }, { "epoch": 3.8323840520748575, "grad_norm": 11.085986164329174, "learning_rate": 9.281635514608055e-06, "loss": 3.0149, "step": 14130 }, { "epoch": 3.8350962842419314, "grad_norm": 11.281126347080729, "learning_rate": 9.280004870503848e-06, "loss": 3.0944, "step": 14140 }, { "epoch": 3.8378085164090043, "grad_norm": 11.103272750046086, "learning_rate": 9.278372521342617e-06, "loss": 3.0042, "step": 14150 }, { "epoch": 3.840520748576078, "grad_norm": 10.578807232532952, "learning_rate": 9.276738467774658e-06, "loss": 2.9729, "step": 14160 }, { "epoch": 3.8432329807431516, "grad_norm": 14.370239764064369, "learning_rate": 9.275102710450939e-06, "loss": 3.064, "step": 14170 }, { "epoch": 3.845945212910225, "grad_norm": 10.781953857777397, "learning_rate": 9.273465250023109e-06, "loss": 3.1442, "step": 14180 }, { "epoch": 3.848657445077299, "grad_norm": 11.109708052835975, "learning_rate": 9.271826087143494e-06, "loss": 3.1075, "step": 14190 }, { "epoch": 3.851369677244372, "grad_norm": 11.905073798960565, "learning_rate": 9.2701852224651e-06, "loss": 2.9884, "step": 14200 }, { "epoch": 3.8540819094114456, "grad_norm": 12.59975729526938, "learning_rate": 9.268542656641612e-06, "loss": 3.1013, "step": 14210 }, { "epoch": 3.856794141578519, "grad_norm": 11.734177077150688, "learning_rate": 9.26689839032739e-06, "loss": 3.1, "step": 14220 }, { "epoch": 3.8595063737455924, "grad_norm": 12.149783762141059, "learning_rate": 9.265252424177472e-06, "loss": 3.1679, "step": 14230 }, { "epoch": 3.862218605912666, "grad_norm": 10.59456357977352, "learning_rate": 9.263604758847574e-06, "loss": 3.1197, "step": 14240 }, { "epoch": 3.8649308380797396, "grad_norm": 12.228185490828627, "learning_rate": 9.261955394994088e-06, "loss": 3.1155, "step": 14250 }, { "epoch": 3.867643070246813, "grad_norm": 11.733890006398951, "learning_rate": 9.260304333274084e-06, "loss": 3.2083, "step": 14260 }, { "epoch": 3.870355302413887, "grad_norm": 10.691069837569062, "learning_rate": 9.258651574345307e-06, "loss": 3.1179, "step": 14270 }, { "epoch": 3.87306753458096, "grad_norm": 11.294772042356737, "learning_rate": 9.25699711886618e-06, "loss": 3.1449, "step": 14280 }, { "epoch": 3.8757797667480336, "grad_norm": 10.527813276856406, "learning_rate": 9.255340967495798e-06, "loss": 3.1127, "step": 14290 }, { "epoch": 3.878491998915107, "grad_norm": 10.780616253744885, "learning_rate": 9.253683120893937e-06, "loss": 3.137, "step": 14300 }, { "epoch": 3.8812042310821804, "grad_norm": 11.806755337417517, "learning_rate": 9.252023579721045e-06, "loss": 3.0849, "step": 14310 }, { "epoch": 3.883916463249254, "grad_norm": 11.718557245946222, "learning_rate": 9.250362344638245e-06, "loss": 3.0732, "step": 14320 }, { "epoch": 3.8866286954163276, "grad_norm": 11.67072585376973, "learning_rate": 9.248699416307335e-06, "loss": 3.0635, "step": 14330 }, { "epoch": 3.889340927583401, "grad_norm": 11.712088880735095, "learning_rate": 9.247034795390786e-06, "loss": 3.121, "step": 14340 }, { "epoch": 3.892053159750475, "grad_norm": 12.06063612114484, "learning_rate": 9.24536848255175e-06, "loss": 3.1005, "step": 14350 }, { "epoch": 3.8947653919175482, "grad_norm": 11.542162686480124, "learning_rate": 9.243700478454046e-06, "loss": 3.1742, "step": 14360 }, { "epoch": 3.8974776240846216, "grad_norm": 10.864474233832576, "learning_rate": 9.242030783762169e-06, "loss": 3.1556, "step": 14370 }, { "epoch": 3.900189856251695, "grad_norm": 11.077629572810341, "learning_rate": 9.240359399141286e-06, "loss": 3.1618, "step": 14380 }, { "epoch": 3.9029020884187684, "grad_norm": 11.796321198686883, "learning_rate": 9.238686325257242e-06, "loss": 3.1019, "step": 14390 }, { "epoch": 3.9056143205858422, "grad_norm": 11.58337274876981, "learning_rate": 9.23701156277655e-06, "loss": 3.1949, "step": 14400 }, { "epoch": 3.9083265527529156, "grad_norm": 11.181715172210398, "learning_rate": 9.235335112366397e-06, "loss": 3.1093, "step": 14410 }, { "epoch": 3.911038784919989, "grad_norm": 11.219765034410806, "learning_rate": 9.233656974694643e-06, "loss": 3.0975, "step": 14420 }, { "epoch": 3.913751017087063, "grad_norm": 13.766224939343175, "learning_rate": 9.231977150429822e-06, "loss": 3.1234, "step": 14430 }, { "epoch": 3.9164632492541362, "grad_norm": 11.663475048043502, "learning_rate": 9.230295640241136e-06, "loss": 3.0903, "step": 14440 }, { "epoch": 3.9191754814212096, "grad_norm": 12.577059727821103, "learning_rate": 9.22861244479846e-06, "loss": 3.0931, "step": 14450 }, { "epoch": 3.921887713588283, "grad_norm": 11.685570695710865, "learning_rate": 9.226927564772343e-06, "loss": 3.1466, "step": 14460 }, { "epoch": 3.9245999457553564, "grad_norm": 10.712863533223342, "learning_rate": 9.225241000834002e-06, "loss": 3.1146, "step": 14470 }, { "epoch": 3.9273121779224303, "grad_norm": 11.295662448117314, "learning_rate": 9.223552753655325e-06, "loss": 3.139, "step": 14480 }, { "epoch": 3.9300244100895037, "grad_norm": 10.583127399572737, "learning_rate": 9.221862823908873e-06, "loss": 3.0921, "step": 14490 }, { "epoch": 3.932736642256577, "grad_norm": 11.827416897146506, "learning_rate": 9.220171212267877e-06, "loss": 2.9822, "step": 14500 }, { "epoch": 3.935448874423651, "grad_norm": 10.442053738732062, "learning_rate": 9.218477919406233e-06, "loss": 3.0618, "step": 14510 }, { "epoch": 3.9381611065907243, "grad_norm": 12.323874181591778, "learning_rate": 9.216782945998512e-06, "loss": 3.1671, "step": 14520 }, { "epoch": 3.9408733387577977, "grad_norm": 13.677133204208735, "learning_rate": 9.215086292719954e-06, "loss": 3.1203, "step": 14530 }, { "epoch": 3.943585570924871, "grad_norm": 11.75538095719964, "learning_rate": 9.213387960246466e-06, "loss": 3.146, "step": 14540 }, { "epoch": 3.9462978030919444, "grad_norm": 11.072791322340722, "learning_rate": 9.211687949254626e-06, "loss": 3.0323, "step": 14550 }, { "epoch": 3.9490100352590183, "grad_norm": 12.390176054556084, "learning_rate": 9.209986260421682e-06, "loss": 3.1621, "step": 14560 }, { "epoch": 3.9517222674260917, "grad_norm": 11.391890185808995, "learning_rate": 9.208282894425544e-06, "loss": 3.0798, "step": 14570 }, { "epoch": 3.954434499593165, "grad_norm": 11.015530560013493, "learning_rate": 9.206577851944796e-06, "loss": 3.1855, "step": 14580 }, { "epoch": 3.957146731760239, "grad_norm": 11.949594612189767, "learning_rate": 9.20487113365869e-06, "loss": 3.0159, "step": 14590 }, { "epoch": 3.9598589639273123, "grad_norm": 12.106290004042883, "learning_rate": 9.203162740247142e-06, "loss": 3.1378, "step": 14600 }, { "epoch": 3.9625711960943857, "grad_norm": 11.390933232024087, "learning_rate": 9.201452672390739e-06, "loss": 3.0995, "step": 14610 }, { "epoch": 3.965283428261459, "grad_norm": 10.772811906562197, "learning_rate": 9.199740930770732e-06, "loss": 3.1977, "step": 14620 }, { "epoch": 3.9679956604285325, "grad_norm": 12.001384851067803, "learning_rate": 9.19802751606904e-06, "loss": 3.1282, "step": 14630 }, { "epoch": 3.9707078925956063, "grad_norm": 11.889693282547054, "learning_rate": 9.196312428968249e-06, "loss": 3.1105, "step": 14640 }, { "epoch": 3.9734201247626797, "grad_norm": 11.05363096537872, "learning_rate": 9.19459567015161e-06, "loss": 3.0982, "step": 14650 }, { "epoch": 3.976132356929753, "grad_norm": 11.13713371228251, "learning_rate": 9.192877240303042e-06, "loss": 3.088, "step": 14660 }, { "epoch": 3.978844589096827, "grad_norm": 13.468601772795934, "learning_rate": 9.19115714010713e-06, "loss": 3.1534, "step": 14670 }, { "epoch": 3.9815568212639003, "grad_norm": 11.193621608470696, "learning_rate": 9.18943537024912e-06, "loss": 3.0187, "step": 14680 }, { "epoch": 3.9842690534309737, "grad_norm": 12.043125446818767, "learning_rate": 9.187711931414927e-06, "loss": 3.1737, "step": 14690 }, { "epoch": 3.986981285598047, "grad_norm": 11.283478073072096, "learning_rate": 9.18598682429113e-06, "loss": 2.9636, "step": 14700 }, { "epoch": 3.9896935177651205, "grad_norm": 11.34035323099169, "learning_rate": 9.184260049564975e-06, "loss": 3.0635, "step": 14710 }, { "epoch": 3.9924057499321943, "grad_norm": 11.485839074400388, "learning_rate": 9.182531607924366e-06, "loss": 3.0536, "step": 14720 }, { "epoch": 3.9951179820992677, "grad_norm": 10.489154569525134, "learning_rate": 9.180801500057879e-06, "loss": 3.1876, "step": 14730 }, { "epoch": 3.997830214266341, "grad_norm": 10.766255777735454, "learning_rate": 9.179069726654746e-06, "loss": 3.0195, "step": 14740 }, { "epoch": 4.000542446433415, "grad_norm": 11.791653034419024, "learning_rate": 9.177336288404867e-06, "loss": 2.9558, "step": 14750 }, { "epoch": 4.003254678600488, "grad_norm": 13.021578920848572, "learning_rate": 9.175601185998808e-06, "loss": 2.6632, "step": 14760 }, { "epoch": 4.005966910767562, "grad_norm": 14.030468553788818, "learning_rate": 9.17386442012779e-06, "loss": 2.5821, "step": 14770 }, { "epoch": 4.008679142934636, "grad_norm": 13.978557350335365, "learning_rate": 9.172125991483705e-06, "loss": 2.5084, "step": 14780 }, { "epoch": 4.0113913751017085, "grad_norm": 11.820294699546888, "learning_rate": 9.1703859007591e-06, "loss": 2.6642, "step": 14790 }, { "epoch": 4.014103607268782, "grad_norm": 11.324424242104367, "learning_rate": 9.168644148647192e-06, "loss": 2.6694, "step": 14800 }, { "epoch": 4.016815839435855, "grad_norm": 12.622796184520702, "learning_rate": 9.16690073584185e-06, "loss": 2.5403, "step": 14810 }, { "epoch": 4.019528071602929, "grad_norm": 13.592309232962865, "learning_rate": 9.165155663037616e-06, "loss": 2.5423, "step": 14820 }, { "epoch": 4.022240303770003, "grad_norm": 13.465311046094893, "learning_rate": 9.163408930929685e-06, "loss": 2.4562, "step": 14830 }, { "epoch": 4.024952535937076, "grad_norm": 12.741747504435354, "learning_rate": 9.161660540213915e-06, "loss": 2.6159, "step": 14840 }, { "epoch": 4.02766476810415, "grad_norm": 12.378304046864471, "learning_rate": 9.159910491586824e-06, "loss": 2.7492, "step": 14850 }, { "epoch": 4.030377000271224, "grad_norm": 13.359000927373536, "learning_rate": 9.158158785745594e-06, "loss": 2.5765, "step": 14860 }, { "epoch": 4.0330892324382965, "grad_norm": 13.366476830893303, "learning_rate": 9.156405423388066e-06, "loss": 2.4966, "step": 14870 }, { "epoch": 4.03580146460537, "grad_norm": 13.209594952427974, "learning_rate": 9.154650405212738e-06, "loss": 2.5518, "step": 14880 }, { "epoch": 4.038513696772443, "grad_norm": 13.262652887998318, "learning_rate": 9.152893731918768e-06, "loss": 2.5406, "step": 14890 }, { "epoch": 4.041225928939517, "grad_norm": 13.97891694100065, "learning_rate": 9.151135404205978e-06, "loss": 2.5736, "step": 14900 }, { "epoch": 4.043938161106591, "grad_norm": 13.559236608197052, "learning_rate": 9.149375422774843e-06, "loss": 2.5844, "step": 14910 }, { "epoch": 4.046650393273664, "grad_norm": 12.614428311267254, "learning_rate": 9.147613788326502e-06, "loss": 2.555, "step": 14920 }, { "epoch": 4.049362625440738, "grad_norm": 15.26152080888392, "learning_rate": 9.14585050156275e-06, "loss": 2.5097, "step": 14930 }, { "epoch": 4.052074857607812, "grad_norm": 12.581870683564429, "learning_rate": 9.14408556318604e-06, "loss": 2.6004, "step": 14940 }, { "epoch": 4.054787089774885, "grad_norm": 17.252391722995103, "learning_rate": 9.142318973899482e-06, "loss": 2.5733, "step": 14950 }, { "epoch": 4.057499321941958, "grad_norm": 13.334359041721635, "learning_rate": 9.140550734406849e-06, "loss": 2.6107, "step": 14960 }, { "epoch": 4.060211554109031, "grad_norm": 13.371198513122529, "learning_rate": 9.138780845412565e-06, "loss": 2.4905, "step": 14970 }, { "epoch": 4.062923786276105, "grad_norm": 14.648978584754259, "learning_rate": 9.137009307621713e-06, "loss": 2.5355, "step": 14980 }, { "epoch": 4.065636018443179, "grad_norm": 13.342796075229867, "learning_rate": 9.135236121740034e-06, "loss": 2.6818, "step": 14990 }, { "epoch": 4.068348250610252, "grad_norm": 13.142195188607923, "learning_rate": 9.133461288473928e-06, "loss": 2.5924, "step": 15000 }, { "epoch": 4.071060482777326, "grad_norm": 13.024515313516266, "learning_rate": 9.131684808530443e-06, "loss": 2.5753, "step": 15010 }, { "epoch": 4.0737727149444, "grad_norm": 12.989503391128117, "learning_rate": 9.129906682617293e-06, "loss": 2.4545, "step": 15020 }, { "epoch": 4.076484947111473, "grad_norm": 15.402323920383967, "learning_rate": 9.128126911442841e-06, "loss": 2.6466, "step": 15030 }, { "epoch": 4.079197179278546, "grad_norm": 13.148650051070684, "learning_rate": 9.12634549571611e-06, "loss": 2.5261, "step": 15040 }, { "epoch": 4.081909411445619, "grad_norm": 12.604008715286968, "learning_rate": 9.124562436146772e-06, "loss": 2.5432, "step": 15050 }, { "epoch": 4.084621643612693, "grad_norm": 14.823505201910177, "learning_rate": 9.122777733445157e-06, "loss": 2.6576, "step": 15060 }, { "epoch": 4.087333875779767, "grad_norm": 12.574935202320557, "learning_rate": 9.120991388322256e-06, "loss": 2.4756, "step": 15070 }, { "epoch": 4.09004610794684, "grad_norm": 13.475980603871731, "learning_rate": 9.119203401489703e-06, "loss": 2.619, "step": 15080 }, { "epoch": 4.092758340113914, "grad_norm": 13.108711053816977, "learning_rate": 9.117413773659793e-06, "loss": 2.5617, "step": 15090 }, { "epoch": 4.095470572280988, "grad_norm": 13.926679465044385, "learning_rate": 9.115622505545473e-06, "loss": 2.6049, "step": 15100 }, { "epoch": 4.098182804448061, "grad_norm": 13.465642515422633, "learning_rate": 9.113829597860347e-06, "loss": 2.6417, "step": 15110 }, { "epoch": 4.100895036615134, "grad_norm": 12.362602348619026, "learning_rate": 9.112035051318664e-06, "loss": 2.5901, "step": 15120 }, { "epoch": 4.103607268782207, "grad_norm": 14.523572415727088, "learning_rate": 9.110238866635333e-06, "loss": 2.6021, "step": 15130 }, { "epoch": 4.106319500949281, "grad_norm": 14.720937563149164, "learning_rate": 9.108441044525913e-06, "loss": 2.6064, "step": 15140 }, { "epoch": 4.109031733116355, "grad_norm": 13.4732190148333, "learning_rate": 9.106641585706616e-06, "loss": 2.6158, "step": 15150 }, { "epoch": 4.111743965283428, "grad_norm": 13.929944043794519, "learning_rate": 9.104840490894305e-06, "loss": 2.5715, "step": 15160 }, { "epoch": 4.114456197450502, "grad_norm": 14.926381865524721, "learning_rate": 9.103037760806494e-06, "loss": 2.6082, "step": 15170 }, { "epoch": 4.117168429617576, "grad_norm": 12.494519084751746, "learning_rate": 9.101233396161354e-06, "loss": 2.4996, "step": 15180 }, { "epoch": 4.119880661784649, "grad_norm": 14.7639441498568, "learning_rate": 9.0994273976777e-06, "loss": 2.5606, "step": 15190 }, { "epoch": 4.1225928939517225, "grad_norm": 15.20309485943419, "learning_rate": 9.097619766075001e-06, "loss": 2.6388, "step": 15200 }, { "epoch": 4.125305126118795, "grad_norm": 15.06356143724503, "learning_rate": 9.095810502073376e-06, "loss": 2.5724, "step": 15210 }, { "epoch": 4.128017358285869, "grad_norm": 12.017386419697296, "learning_rate": 9.093999606393596e-06, "loss": 2.7031, "step": 15220 }, { "epoch": 4.130729590452943, "grad_norm": 15.325802146930648, "learning_rate": 9.092187079757082e-06, "loss": 2.6069, "step": 15230 }, { "epoch": 4.133441822620016, "grad_norm": 14.511825583057604, "learning_rate": 9.0903729228859e-06, "loss": 2.6874, "step": 15240 }, { "epoch": 4.13615405478709, "grad_norm": 12.78593802223587, "learning_rate": 9.088557136502773e-06, "loss": 2.5893, "step": 15250 }, { "epoch": 4.138866286954164, "grad_norm": 12.932900141902056, "learning_rate": 9.086739721331066e-06, "loss": 2.6352, "step": 15260 }, { "epoch": 4.141578519121237, "grad_norm": 13.43703757936368, "learning_rate": 9.084920678094799e-06, "loss": 2.6155, "step": 15270 }, { "epoch": 4.1442907512883105, "grad_norm": 13.946718810978899, "learning_rate": 9.083100007518635e-06, "loss": 2.6006, "step": 15280 }, { "epoch": 4.147002983455383, "grad_norm": 13.926842369823843, "learning_rate": 9.081277710327887e-06, "loss": 2.6585, "step": 15290 }, { "epoch": 4.149715215622457, "grad_norm": 13.92318223654076, "learning_rate": 9.079453787248521e-06, "loss": 2.6121, "step": 15300 }, { "epoch": 4.152427447789531, "grad_norm": 13.349085482197074, "learning_rate": 9.077628239007145e-06, "loss": 2.5206, "step": 15310 }, { "epoch": 4.155139679956604, "grad_norm": 14.488436067391048, "learning_rate": 9.075801066331014e-06, "loss": 2.5493, "step": 15320 }, { "epoch": 4.157851912123678, "grad_norm": 13.616074024822963, "learning_rate": 9.073972269948035e-06, "loss": 2.62, "step": 15330 }, { "epoch": 4.160564144290751, "grad_norm": 13.824324107783253, "learning_rate": 9.072141850586758e-06, "loss": 2.5855, "step": 15340 }, { "epoch": 4.163276376457825, "grad_norm": 14.457577723118552, "learning_rate": 9.070309808976378e-06, "loss": 2.5749, "step": 15350 }, { "epoch": 4.1659886086248985, "grad_norm": 13.942483784803677, "learning_rate": 9.068476145846745e-06, "loss": 2.6291, "step": 15360 }, { "epoch": 4.1687008407919715, "grad_norm": 13.518334479120131, "learning_rate": 9.066640861928342e-06, "loss": 2.6417, "step": 15370 }, { "epoch": 4.171413072959045, "grad_norm": 12.892059004340734, "learning_rate": 9.06480395795231e-06, "loss": 2.6525, "step": 15380 }, { "epoch": 4.174125305126119, "grad_norm": 15.368378941379772, "learning_rate": 9.062965434650426e-06, "loss": 2.5422, "step": 15390 }, { "epoch": 4.176837537293192, "grad_norm": 12.370799540454232, "learning_rate": 9.061125292755119e-06, "loss": 2.6059, "step": 15400 }, { "epoch": 4.179549769460266, "grad_norm": 13.11913524128977, "learning_rate": 9.059283532999458e-06, "loss": 2.5189, "step": 15410 }, { "epoch": 4.18226200162734, "grad_norm": 13.515553840584948, "learning_rate": 9.05744015611716e-06, "loss": 2.6525, "step": 15420 }, { "epoch": 4.184974233794413, "grad_norm": 12.787776212063894, "learning_rate": 9.055595162842584e-06, "loss": 2.5363, "step": 15430 }, { "epoch": 4.1876864659614865, "grad_norm": 14.700504107936979, "learning_rate": 9.05374855391073e-06, "loss": 2.6744, "step": 15440 }, { "epoch": 4.1903986981285595, "grad_norm": 14.782667142016232, "learning_rate": 9.051900330057252e-06, "loss": 2.638, "step": 15450 }, { "epoch": 4.193110930295633, "grad_norm": 13.804141824817702, "learning_rate": 9.050050492018437e-06, "loss": 2.7579, "step": 15460 }, { "epoch": 4.195823162462707, "grad_norm": 14.772817248374214, "learning_rate": 9.048199040531216e-06, "loss": 2.5316, "step": 15470 }, { "epoch": 4.19853539462978, "grad_norm": 13.74855423168096, "learning_rate": 9.046345976333169e-06, "loss": 2.518, "step": 15480 }, { "epoch": 4.201247626796854, "grad_norm": 13.919543577035164, "learning_rate": 9.044491300162511e-06, "loss": 2.6227, "step": 15490 }, { "epoch": 4.203959858963927, "grad_norm": 13.550222402973677, "learning_rate": 9.04263501275811e-06, "loss": 2.6737, "step": 15500 }, { "epoch": 4.206672091131001, "grad_norm": 13.99607747317417, "learning_rate": 9.040777114859462e-06, "loss": 2.5077, "step": 15510 }, { "epoch": 4.2093843232980745, "grad_norm": 14.159521303222386, "learning_rate": 9.038917607206716e-06, "loss": 2.588, "step": 15520 }, { "epoch": 4.2120965554651475, "grad_norm": 13.025458571273742, "learning_rate": 9.037056490540652e-06, "loss": 2.6387, "step": 15530 }, { "epoch": 4.214808787632221, "grad_norm": 14.558071393596137, "learning_rate": 9.035193765602703e-06, "loss": 2.631, "step": 15540 }, { "epoch": 4.217521019799295, "grad_norm": 13.877918361018137, "learning_rate": 9.033329433134934e-06, "loss": 2.5213, "step": 15550 }, { "epoch": 4.220233251966368, "grad_norm": 13.242708924199905, "learning_rate": 9.031463493880053e-06, "loss": 2.6542, "step": 15560 }, { "epoch": 4.222945484133442, "grad_norm": 14.887205750464286, "learning_rate": 9.029595948581407e-06, "loss": 2.671, "step": 15570 }, { "epoch": 4.225657716300515, "grad_norm": 12.82271852502033, "learning_rate": 9.027726797982983e-06, "loss": 2.6161, "step": 15580 }, { "epoch": 4.228369948467589, "grad_norm": 14.900082939399109, "learning_rate": 9.025856042829412e-06, "loss": 2.5922, "step": 15590 }, { "epoch": 4.231082180634663, "grad_norm": 14.822946063579659, "learning_rate": 9.023983683865957e-06, "loss": 2.62, "step": 15600 }, { "epoch": 4.2337944128017355, "grad_norm": 14.317434135290721, "learning_rate": 9.022109721838524e-06, "loss": 2.5264, "step": 15610 }, { "epoch": 4.236506644968809, "grad_norm": 12.92054496102857, "learning_rate": 9.020234157493659e-06, "loss": 2.5853, "step": 15620 }, { "epoch": 4.239218877135883, "grad_norm": 16.55558230262288, "learning_rate": 9.018356991578542e-06, "loss": 2.601, "step": 15630 }, { "epoch": 4.241931109302956, "grad_norm": 12.84601541541041, "learning_rate": 9.016478224840995e-06, "loss": 2.6337, "step": 15640 }, { "epoch": 4.24464334147003, "grad_norm": 13.613048667622737, "learning_rate": 9.014597858029475e-06, "loss": 2.6876, "step": 15650 }, { "epoch": 4.247355573637103, "grad_norm": 13.139346017266886, "learning_rate": 9.012715891893078e-06, "loss": 2.7854, "step": 15660 }, { "epoch": 4.250067805804177, "grad_norm": 13.182393501632387, "learning_rate": 9.010832327181537e-06, "loss": 2.4727, "step": 15670 }, { "epoch": 4.252780037971251, "grad_norm": 14.390312838672445, "learning_rate": 9.00894716464522e-06, "loss": 2.5698, "step": 15680 }, { "epoch": 4.2554922701383235, "grad_norm": 13.674777619089005, "learning_rate": 9.007060405035136e-06, "loss": 2.7055, "step": 15690 }, { "epoch": 4.258204502305397, "grad_norm": 14.354110252201265, "learning_rate": 9.005172049102927e-06, "loss": 2.5953, "step": 15700 }, { "epoch": 4.260916734472471, "grad_norm": 13.2353924966463, "learning_rate": 9.003282097600867e-06, "loss": 2.6524, "step": 15710 }, { "epoch": 4.263628966639544, "grad_norm": 14.209459454733931, "learning_rate": 9.001390551281874e-06, "loss": 2.6221, "step": 15720 }, { "epoch": 4.266341198806618, "grad_norm": 14.07917264144613, "learning_rate": 8.999497410899495e-06, "loss": 2.641, "step": 15730 }, { "epoch": 4.269053430973691, "grad_norm": 15.138611540112821, "learning_rate": 8.997602677207916e-06, "loss": 2.6183, "step": 15740 }, { "epoch": 4.271765663140765, "grad_norm": 14.831749297384851, "learning_rate": 8.995706350961955e-06, "loss": 2.7714, "step": 15750 }, { "epoch": 4.274477895307839, "grad_norm": 14.855976313628299, "learning_rate": 8.993808432917066e-06, "loss": 2.5832, "step": 15760 }, { "epoch": 4.277190127474912, "grad_norm": 14.191222375308095, "learning_rate": 8.991908923829335e-06, "loss": 2.5567, "step": 15770 }, { "epoch": 4.279902359641985, "grad_norm": 12.625194603801463, "learning_rate": 8.990007824455487e-06, "loss": 2.7342, "step": 15780 }, { "epoch": 4.282614591809059, "grad_norm": 13.146027564838628, "learning_rate": 8.988105135552873e-06, "loss": 2.573, "step": 15790 }, { "epoch": 4.285326823976132, "grad_norm": 13.923101008401474, "learning_rate": 8.986200857879484e-06, "loss": 2.6401, "step": 15800 }, { "epoch": 4.288039056143206, "grad_norm": 12.891036564142698, "learning_rate": 8.984294992193939e-06, "loss": 2.5339, "step": 15810 }, { "epoch": 4.290751288310279, "grad_norm": 12.880374906744567, "learning_rate": 8.982387539255493e-06, "loss": 2.7033, "step": 15820 }, { "epoch": 4.293463520477353, "grad_norm": 14.586559153981183, "learning_rate": 8.98047849982403e-06, "loss": 2.5315, "step": 15830 }, { "epoch": 4.296175752644427, "grad_norm": 13.13677005385445, "learning_rate": 8.97856787466007e-06, "loss": 2.5463, "step": 15840 }, { "epoch": 4.2988879848115, "grad_norm": 12.717981883580586, "learning_rate": 8.976655664524764e-06, "loss": 2.5159, "step": 15850 }, { "epoch": 4.301600216978573, "grad_norm": 13.275684655859314, "learning_rate": 8.974741870179892e-06, "loss": 2.6638, "step": 15860 }, { "epoch": 4.304312449145647, "grad_norm": 14.994957235831588, "learning_rate": 8.972826492387865e-06, "loss": 2.6092, "step": 15870 }, { "epoch": 4.30702468131272, "grad_norm": 13.02999386660378, "learning_rate": 8.970909531911728e-06, "loss": 2.5584, "step": 15880 }, { "epoch": 4.309736913479794, "grad_norm": 13.466162627165081, "learning_rate": 8.968990989515152e-06, "loss": 2.675, "step": 15890 }, { "epoch": 4.312449145646867, "grad_norm": 13.290965745325941, "learning_rate": 8.967070865962445e-06, "loss": 2.6904, "step": 15900 }, { "epoch": 4.315161377813941, "grad_norm": 12.61026103269128, "learning_rate": 8.965149162018538e-06, "loss": 2.6704, "step": 15910 }, { "epoch": 4.317873609981015, "grad_norm": 13.628078182839452, "learning_rate": 8.963225878448994e-06, "loss": 2.6904, "step": 15920 }, { "epoch": 4.320585842148088, "grad_norm": 13.516725424733131, "learning_rate": 8.961301016020008e-06, "loss": 2.6381, "step": 15930 }, { "epoch": 4.323298074315161, "grad_norm": 15.706248374376287, "learning_rate": 8.959374575498397e-06, "loss": 2.5781, "step": 15940 }, { "epoch": 4.326010306482235, "grad_norm": 13.395195170979687, "learning_rate": 8.957446557651618e-06, "loss": 2.668, "step": 15950 }, { "epoch": 4.328722538649308, "grad_norm": 13.084083693170495, "learning_rate": 8.955516963247743e-06, "loss": 2.636, "step": 15960 }, { "epoch": 4.331434770816382, "grad_norm": 13.68263181426415, "learning_rate": 8.953585793055482e-06, "loss": 2.5969, "step": 15970 }, { "epoch": 4.334147002983455, "grad_norm": 14.574534385523755, "learning_rate": 8.95165304784417e-06, "loss": 2.6156, "step": 15980 }, { "epoch": 4.336859235150529, "grad_norm": 13.550530320440801, "learning_rate": 8.949718728383768e-06, "loss": 2.6558, "step": 15990 }, { "epoch": 4.339571467317603, "grad_norm": 14.785062676513023, "learning_rate": 8.947782835444864e-06, "loss": 2.6067, "step": 16000 }, { "epoch": 4.342283699484676, "grad_norm": 13.289616566937157, "learning_rate": 8.945845369798676e-06, "loss": 2.7438, "step": 16010 }, { "epoch": 4.3449959316517495, "grad_norm": 13.509344597273266, "learning_rate": 8.943906332217044e-06, "loss": 2.6654, "step": 16020 }, { "epoch": 4.347708163818823, "grad_norm": 14.531902498175196, "learning_rate": 8.94196572347244e-06, "loss": 2.6484, "step": 16030 }, { "epoch": 4.350420395985896, "grad_norm": 13.561878390021475, "learning_rate": 8.940023544337957e-06, "loss": 2.6469, "step": 16040 }, { "epoch": 4.35313262815297, "grad_norm": 12.0928041491658, "learning_rate": 8.938079795587315e-06, "loss": 2.7179, "step": 16050 }, { "epoch": 4.355844860320043, "grad_norm": 13.090152424283533, "learning_rate": 8.936134477994858e-06, "loss": 2.6798, "step": 16060 }, { "epoch": 4.358557092487117, "grad_norm": 12.672225822965165, "learning_rate": 8.934187592335559e-06, "loss": 2.5186, "step": 16070 }, { "epoch": 4.361269324654191, "grad_norm": 14.932096159615002, "learning_rate": 8.93223913938501e-06, "loss": 2.6887, "step": 16080 }, { "epoch": 4.363981556821264, "grad_norm": 12.969521923616968, "learning_rate": 8.930289119919435e-06, "loss": 2.6761, "step": 16090 }, { "epoch": 4.3666937889883375, "grad_norm": 13.714333002485027, "learning_rate": 8.928337534715674e-06, "loss": 2.5942, "step": 16100 }, { "epoch": 4.369406021155411, "grad_norm": 14.361336040787387, "learning_rate": 8.926384384551197e-06, "loss": 2.5778, "step": 16110 }, { "epoch": 4.372118253322484, "grad_norm": 14.88885215463092, "learning_rate": 8.92442967020409e-06, "loss": 2.6613, "step": 16120 }, { "epoch": 4.374830485489558, "grad_norm": 12.622621460611468, "learning_rate": 8.922473392453073e-06, "loss": 2.6271, "step": 16130 }, { "epoch": 4.377542717656631, "grad_norm": 13.642482477525412, "learning_rate": 8.92051555207748e-06, "loss": 2.592, "step": 16140 }, { "epoch": 4.380254949823705, "grad_norm": 12.40302498723259, "learning_rate": 8.91855614985727e-06, "loss": 2.6552, "step": 16150 }, { "epoch": 4.382967181990779, "grad_norm": 13.979898228053742, "learning_rate": 8.916595186573023e-06, "loss": 2.7992, "step": 16160 }, { "epoch": 4.385679414157852, "grad_norm": 12.600927279785724, "learning_rate": 8.914632663005946e-06, "loss": 2.6115, "step": 16170 }, { "epoch": 4.3883916463249255, "grad_norm": 14.486604756801555, "learning_rate": 8.912668579937861e-06, "loss": 2.4797, "step": 16180 }, { "epoch": 4.3911038784919985, "grad_norm": 14.215809510740819, "learning_rate": 8.910702938151215e-06, "loss": 2.6909, "step": 16190 }, { "epoch": 4.393816110659072, "grad_norm": 13.582628693500897, "learning_rate": 8.908735738429077e-06, "loss": 2.5996, "step": 16200 }, { "epoch": 4.396528342826146, "grad_norm": 14.034980933615813, "learning_rate": 8.906766981555131e-06, "loss": 2.632, "step": 16210 }, { "epoch": 4.399240574993219, "grad_norm": 13.402313920432963, "learning_rate": 8.904796668313689e-06, "loss": 2.6641, "step": 16220 }, { "epoch": 4.401952807160293, "grad_norm": 13.534211435280158, "learning_rate": 8.902824799489677e-06, "loss": 2.6379, "step": 16230 }, { "epoch": 4.404665039327367, "grad_norm": 12.90766444782416, "learning_rate": 8.900851375868645e-06, "loss": 2.725, "step": 16240 }, { "epoch": 4.40737727149444, "grad_norm": 13.609806617038666, "learning_rate": 8.898876398236758e-06, "loss": 2.657, "step": 16250 }, { "epoch": 4.4100895036615135, "grad_norm": 13.105053912308625, "learning_rate": 8.896899867380801e-06, "loss": 2.6315, "step": 16260 }, { "epoch": 4.412801735828587, "grad_norm": 13.846033831545851, "learning_rate": 8.894921784088183e-06, "loss": 2.6399, "step": 16270 }, { "epoch": 4.41551396799566, "grad_norm": 14.566138438386314, "learning_rate": 8.892942149146928e-06, "loss": 2.6509, "step": 16280 }, { "epoch": 4.418226200162734, "grad_norm": 12.356194600484441, "learning_rate": 8.890960963345672e-06, "loss": 2.6084, "step": 16290 }, { "epoch": 4.420938432329807, "grad_norm": 14.330084388836152, "learning_rate": 8.888978227473682e-06, "loss": 2.6892, "step": 16300 }, { "epoch": 4.423650664496881, "grad_norm": 12.818114401458178, "learning_rate": 8.886993942320827e-06, "loss": 2.7558, "step": 16310 }, { "epoch": 4.426362896663955, "grad_norm": 12.930493551033186, "learning_rate": 8.88500810867761e-06, "loss": 2.506, "step": 16320 }, { "epoch": 4.429075128831028, "grad_norm": 13.691427451129636, "learning_rate": 8.883020727335133e-06, "loss": 2.6445, "step": 16330 }, { "epoch": 4.4317873609981016, "grad_norm": 13.77945611590475, "learning_rate": 8.881031799085131e-06, "loss": 2.6177, "step": 16340 }, { "epoch": 4.4344995931651745, "grad_norm": 13.664259129686581, "learning_rate": 8.879041324719943e-06, "loss": 2.6798, "step": 16350 }, { "epoch": 4.437211825332248, "grad_norm": 13.221516156988962, "learning_rate": 8.877049305032535e-06, "loss": 2.748, "step": 16360 }, { "epoch": 4.439924057499322, "grad_norm": 13.314653114132174, "learning_rate": 8.875055740816475e-06, "loss": 2.6256, "step": 16370 }, { "epoch": 4.442636289666395, "grad_norm": 15.000398515946166, "learning_rate": 8.873060632865961e-06, "loss": 2.6755, "step": 16380 }, { "epoch": 4.445348521833469, "grad_norm": 14.960608291862874, "learning_rate": 8.871063981975793e-06, "loss": 2.7023, "step": 16390 }, { "epoch": 4.448060754000543, "grad_norm": 12.521152513611598, "learning_rate": 8.869065788941395e-06, "loss": 2.6444, "step": 16400 }, { "epoch": 4.450772986167616, "grad_norm": 13.769265232234918, "learning_rate": 8.867066054558802e-06, "loss": 2.6037, "step": 16410 }, { "epoch": 4.45348521833469, "grad_norm": 13.762770566078917, "learning_rate": 8.86506477962466e-06, "loss": 2.5903, "step": 16420 }, { "epoch": 4.456197450501763, "grad_norm": 15.063551179626417, "learning_rate": 8.863061964936235e-06, "loss": 2.5806, "step": 16430 }, { "epoch": 4.458909682668836, "grad_norm": 14.87236134467505, "learning_rate": 8.861057611291399e-06, "loss": 2.6828, "step": 16440 }, { "epoch": 4.46162191483591, "grad_norm": 13.279427690522997, "learning_rate": 8.859051719488644e-06, "loss": 2.7982, "step": 16450 }, { "epoch": 4.464334147002983, "grad_norm": 13.988040704554429, "learning_rate": 8.857044290327072e-06, "loss": 2.5935, "step": 16460 }, { "epoch": 4.467046379170057, "grad_norm": 13.02151378423945, "learning_rate": 8.855035324606398e-06, "loss": 2.6445, "step": 16470 }, { "epoch": 4.469758611337131, "grad_norm": 14.063476520488845, "learning_rate": 8.853024823126945e-06, "loss": 2.6554, "step": 16480 }, { "epoch": 4.472470843504204, "grad_norm": 13.481504351620433, "learning_rate": 8.851012786689654e-06, "loss": 2.6729, "step": 16490 }, { "epoch": 4.475183075671278, "grad_norm": 14.293845578422342, "learning_rate": 8.848999216096074e-06, "loss": 2.666, "step": 16500 }, { "epoch": 4.4778953078383505, "grad_norm": 14.769142042817027, "learning_rate": 8.846984112148366e-06, "loss": 2.5935, "step": 16510 }, { "epoch": 4.480607540005424, "grad_norm": 13.72119596253004, "learning_rate": 8.844967475649304e-06, "loss": 2.7664, "step": 16520 }, { "epoch": 4.483319772172498, "grad_norm": 14.361343255364984, "learning_rate": 8.842949307402267e-06, "loss": 2.5945, "step": 16530 }, { "epoch": 4.486032004339571, "grad_norm": 12.757185897845082, "learning_rate": 8.840929608211247e-06, "loss": 2.6665, "step": 16540 }, { "epoch": 4.488744236506645, "grad_norm": 14.791775634839109, "learning_rate": 8.838908378880849e-06, "loss": 2.6152, "step": 16550 }, { "epoch": 4.491456468673719, "grad_norm": 12.815323768182463, "learning_rate": 8.836885620216285e-06, "loss": 2.5712, "step": 16560 }, { "epoch": 4.494168700840792, "grad_norm": 12.758593780928523, "learning_rate": 8.834861333023374e-06, "loss": 2.5467, "step": 16570 }, { "epoch": 4.496880933007866, "grad_norm": 12.914558073673346, "learning_rate": 8.832835518108546e-06, "loss": 2.5691, "step": 16580 }, { "epoch": 4.499593165174939, "grad_norm": 16.575631714724977, "learning_rate": 8.830808176278844e-06, "loss": 2.6878, "step": 16590 }, { "epoch": 4.502305397342012, "grad_norm": 13.747660804780335, "learning_rate": 8.828779308341912e-06, "loss": 2.5782, "step": 16600 }, { "epoch": 4.505017629509086, "grad_norm": 12.038553930062681, "learning_rate": 8.826748915106005e-06, "loss": 2.7774, "step": 16610 }, { "epoch": 4.507729861676159, "grad_norm": 15.634107563499644, "learning_rate": 8.824716997379987e-06, "loss": 2.6646, "step": 16620 }, { "epoch": 4.510442093843233, "grad_norm": 13.65338693671866, "learning_rate": 8.822683555973326e-06, "loss": 2.6841, "step": 16630 }, { "epoch": 4.513154326010307, "grad_norm": 13.649298905621068, "learning_rate": 8.8206485916961e-06, "loss": 2.6347, "step": 16640 }, { "epoch": 4.51586655817738, "grad_norm": 13.92802726327037, "learning_rate": 8.818612105358994e-06, "loss": 2.6878, "step": 16650 }, { "epoch": 4.518578790344454, "grad_norm": 12.511882226168312, "learning_rate": 8.816574097773296e-06, "loss": 2.7857, "step": 16660 }, { "epoch": 4.521291022511527, "grad_norm": 14.501547451150122, "learning_rate": 8.814534569750903e-06, "loss": 2.6277, "step": 16670 }, { "epoch": 4.5240032546786, "grad_norm": 13.369700157500223, "learning_rate": 8.812493522104317e-06, "loss": 2.8704, "step": 16680 }, { "epoch": 4.526715486845674, "grad_norm": 14.948518241452929, "learning_rate": 8.810450955646645e-06, "loss": 2.7093, "step": 16690 }, { "epoch": 4.529427719012747, "grad_norm": 14.834467546048304, "learning_rate": 8.808406871191597e-06, "loss": 2.7031, "step": 16700 }, { "epoch": 4.532139951179821, "grad_norm": 13.004469212843171, "learning_rate": 8.806361269553495e-06, "loss": 2.8052, "step": 16710 }, { "epoch": 4.534852183346895, "grad_norm": 13.990917747616392, "learning_rate": 8.804314151547257e-06, "loss": 2.6276, "step": 16720 }, { "epoch": 4.537564415513968, "grad_norm": 12.624982748221838, "learning_rate": 8.802265517988408e-06, "loss": 2.6506, "step": 16730 }, { "epoch": 4.540276647681042, "grad_norm": 13.69315157152237, "learning_rate": 8.800215369693076e-06, "loss": 2.7208, "step": 16740 }, { "epoch": 4.542988879848115, "grad_norm": 13.07428810912908, "learning_rate": 8.798163707478e-06, "loss": 2.7658, "step": 16750 }, { "epoch": 4.5457011120151884, "grad_norm": 13.906130974410114, "learning_rate": 8.796110532160509e-06, "loss": 2.4802, "step": 16760 }, { "epoch": 4.548413344182262, "grad_norm": 13.26812194282171, "learning_rate": 8.794055844558545e-06, "loss": 2.6887, "step": 16770 }, { "epoch": 4.551125576349335, "grad_norm": 14.182100247567707, "learning_rate": 8.791999645490648e-06, "loss": 2.5679, "step": 16780 }, { "epoch": 4.553837808516409, "grad_norm": 15.05601149252244, "learning_rate": 8.789941935775961e-06, "loss": 2.7244, "step": 16790 }, { "epoch": 4.556550040683483, "grad_norm": 14.30113707134802, "learning_rate": 8.787882716234228e-06, "loss": 2.7499, "step": 16800 }, { "epoch": 4.559262272850556, "grad_norm": 14.03326336696086, "learning_rate": 8.7858219876858e-06, "loss": 2.7304, "step": 16810 }, { "epoch": 4.56197450501763, "grad_norm": 13.348964613687567, "learning_rate": 8.783759750951617e-06, "loss": 2.683, "step": 16820 }, { "epoch": 4.564686737184703, "grad_norm": 13.416672363362938, "learning_rate": 8.781696006853233e-06, "loss": 2.7055, "step": 16830 }, { "epoch": 4.5673989693517765, "grad_norm": 13.124910652201656, "learning_rate": 8.779630756212796e-06, "loss": 2.6736, "step": 16840 }, { "epoch": 4.57011120151885, "grad_norm": 14.619135741138255, "learning_rate": 8.777563999853055e-06, "loss": 2.6425, "step": 16850 }, { "epoch": 4.572823433685923, "grad_norm": 12.937442857798516, "learning_rate": 8.775495738597359e-06, "loss": 2.7291, "step": 16860 }, { "epoch": 4.575535665852997, "grad_norm": 13.848346213856477, "learning_rate": 8.773425973269655e-06, "loss": 2.6572, "step": 16870 }, { "epoch": 4.57824789802007, "grad_norm": 16.614145524044357, "learning_rate": 8.771354704694491e-06, "loss": 2.5669, "step": 16880 }, { "epoch": 4.580960130187144, "grad_norm": 13.448171312429697, "learning_rate": 8.769281933697015e-06, "loss": 2.5318, "step": 16890 }, { "epoch": 4.583672362354218, "grad_norm": 13.563786477253, "learning_rate": 8.767207661102972e-06, "loss": 2.6312, "step": 16900 }, { "epoch": 4.586384594521291, "grad_norm": 14.236375915524155, "learning_rate": 8.765131887738703e-06, "loss": 2.6776, "step": 16910 }, { "epoch": 4.5890968266883645, "grad_norm": 13.331947529142106, "learning_rate": 8.763054614431152e-06, "loss": 2.7076, "step": 16920 }, { "epoch": 4.591809058855438, "grad_norm": 14.587146086950883, "learning_rate": 8.760975842007854e-06, "loss": 2.6977, "step": 16930 }, { "epoch": 4.594521291022511, "grad_norm": 13.507299927658826, "learning_rate": 8.75889557129695e-06, "loss": 2.8226, "step": 16940 }, { "epoch": 4.597233523189585, "grad_norm": 15.280404632968319, "learning_rate": 8.756813803127168e-06, "loss": 2.5207, "step": 16950 }, { "epoch": 4.599945755356659, "grad_norm": 13.383344809426688, "learning_rate": 8.754730538327839e-06, "loss": 2.762, "step": 16960 }, { "epoch": 4.602657987523732, "grad_norm": 13.842333844759283, "learning_rate": 8.752645777728887e-06, "loss": 2.623, "step": 16970 }, { "epoch": 4.605370219690806, "grad_norm": 13.648147036223218, "learning_rate": 8.750559522160838e-06, "loss": 2.8259, "step": 16980 }, { "epoch": 4.608082451857879, "grad_norm": 16.50688542886367, "learning_rate": 8.748471772454805e-06, "loss": 2.6872, "step": 16990 }, { "epoch": 4.6107946840249525, "grad_norm": 12.635942276478028, "learning_rate": 8.746382529442501e-06, "loss": 2.8033, "step": 17000 }, { "epoch": 4.613506916192026, "grad_norm": 14.80647107310184, "learning_rate": 8.744291793956232e-06, "loss": 2.766, "step": 17010 }, { "epoch": 4.616219148359099, "grad_norm": 13.856151431331572, "learning_rate": 8.742199566828903e-06, "loss": 2.712, "step": 17020 }, { "epoch": 4.618931380526173, "grad_norm": 13.502921758196702, "learning_rate": 8.740105848894007e-06, "loss": 2.6264, "step": 17030 }, { "epoch": 4.621643612693246, "grad_norm": 12.780800939248818, "learning_rate": 8.738010640985634e-06, "loss": 2.7189, "step": 17040 }, { "epoch": 4.62435584486032, "grad_norm": 13.598874608481632, "learning_rate": 8.73591394393847e-06, "loss": 2.7199, "step": 17050 }, { "epoch": 4.627068077027394, "grad_norm": 13.176293915263358, "learning_rate": 8.733815758587787e-06, "loss": 2.7847, "step": 17060 }, { "epoch": 4.629780309194467, "grad_norm": 14.431965047089138, "learning_rate": 8.731716085769459e-06, "loss": 2.5856, "step": 17070 }, { "epoch": 4.6324925413615405, "grad_norm": 14.401075891739607, "learning_rate": 8.729614926319946e-06, "loss": 2.7141, "step": 17080 }, { "epoch": 4.635204773528614, "grad_norm": 12.873389368222897, "learning_rate": 8.727512281076303e-06, "loss": 2.6977, "step": 17090 }, { "epoch": 4.637917005695687, "grad_norm": 13.892034029246943, "learning_rate": 8.725408150876176e-06, "loss": 2.4522, "step": 17100 }, { "epoch": 4.640629237862761, "grad_norm": 15.272352501957682, "learning_rate": 8.7233025365578e-06, "loss": 2.5571, "step": 17110 }, { "epoch": 4.643341470029835, "grad_norm": 13.37249672669737, "learning_rate": 8.72119543896001e-06, "loss": 2.7011, "step": 17120 }, { "epoch": 4.646053702196908, "grad_norm": 17.729283728387045, "learning_rate": 8.719086858922223e-06, "loss": 2.7431, "step": 17130 }, { "epoch": 4.648765934363982, "grad_norm": 13.420759017204066, "learning_rate": 8.716976797284447e-06, "loss": 2.6507, "step": 17140 }, { "epoch": 4.651478166531055, "grad_norm": 15.296008772410596, "learning_rate": 8.714865254887285e-06, "loss": 2.7586, "step": 17150 }, { "epoch": 4.654190398698129, "grad_norm": 13.661830174148838, "learning_rate": 8.712752232571931e-06, "loss": 2.5765, "step": 17160 }, { "epoch": 4.656902630865202, "grad_norm": 14.786596819795568, "learning_rate": 8.710637731180159e-06, "loss": 2.6805, "step": 17170 }, { "epoch": 4.659614863032275, "grad_norm": 12.78031645498752, "learning_rate": 8.708521751554344e-06, "loss": 2.647, "step": 17180 }, { "epoch": 4.662327095199349, "grad_norm": 14.71983597355939, "learning_rate": 8.706404294537438e-06, "loss": 2.6741, "step": 17190 }, { "epoch": 4.665039327366422, "grad_norm": 13.750960122783692, "learning_rate": 8.704285360972995e-06, "loss": 2.7363, "step": 17200 }, { "epoch": 4.667751559533496, "grad_norm": 12.616544312504637, "learning_rate": 8.702164951705145e-06, "loss": 2.6446, "step": 17210 }, { "epoch": 4.67046379170057, "grad_norm": 13.548943595491918, "learning_rate": 8.700043067578613e-06, "loss": 2.7444, "step": 17220 }, { "epoch": 4.673176023867643, "grad_norm": 13.64269084820044, "learning_rate": 8.697919709438712e-06, "loss": 2.6546, "step": 17230 }, { "epoch": 4.675888256034717, "grad_norm": 13.324999377112906, "learning_rate": 8.695794878131336e-06, "loss": 2.5761, "step": 17240 }, { "epoch": 4.67860048820179, "grad_norm": 13.023916442728208, "learning_rate": 8.693668574502972e-06, "loss": 2.7588, "step": 17250 }, { "epoch": 4.681312720368863, "grad_norm": 13.384906301111474, "learning_rate": 8.691540799400689e-06, "loss": 2.6111, "step": 17260 }, { "epoch": 4.684024952535937, "grad_norm": 15.040826319098624, "learning_rate": 8.689411553672144e-06, "loss": 2.6325, "step": 17270 }, { "epoch": 4.686737184703011, "grad_norm": 13.242277821318137, "learning_rate": 8.687280838165584e-06, "loss": 2.6772, "step": 17280 }, { "epoch": 4.689449416870084, "grad_norm": 12.915372761555712, "learning_rate": 8.685148653729836e-06, "loss": 2.667, "step": 17290 }, { "epoch": 4.692161649037158, "grad_norm": 14.674193762147555, "learning_rate": 8.683015001214312e-06, "loss": 2.6123, "step": 17300 }, { "epoch": 4.694873881204231, "grad_norm": 12.992309157291842, "learning_rate": 8.680879881469014e-06, "loss": 2.7348, "step": 17310 }, { "epoch": 4.697586113371305, "grad_norm": 16.097306400388455, "learning_rate": 8.678743295344524e-06, "loss": 2.6684, "step": 17320 }, { "epoch": 4.700298345538378, "grad_norm": 14.728805203609047, "learning_rate": 8.676605243692009e-06, "loss": 2.6157, "step": 17330 }, { "epoch": 4.703010577705451, "grad_norm": 14.163753969374723, "learning_rate": 8.674465727363222e-06, "loss": 2.6403, "step": 17340 }, { "epoch": 4.705722809872525, "grad_norm": 13.309833651167596, "learning_rate": 8.672324747210495e-06, "loss": 2.7072, "step": 17350 }, { "epoch": 4.708435042039598, "grad_norm": 13.060290943959346, "learning_rate": 8.670182304086747e-06, "loss": 2.7775, "step": 17360 }, { "epoch": 4.711147274206672, "grad_norm": 13.762506411034133, "learning_rate": 8.668038398845482e-06, "loss": 2.6988, "step": 17370 }, { "epoch": 4.713859506373746, "grad_norm": 13.507015676109608, "learning_rate": 8.665893032340779e-06, "loss": 2.5561, "step": 17380 }, { "epoch": 4.716571738540819, "grad_norm": 12.587849433913325, "learning_rate": 8.663746205427306e-06, "loss": 2.6729, "step": 17390 }, { "epoch": 4.719283970707893, "grad_norm": 13.66605582853189, "learning_rate": 8.661597918960309e-06, "loss": 2.537, "step": 17400 }, { "epoch": 4.7219962028749665, "grad_norm": 13.681663142424096, "learning_rate": 8.659448173795618e-06, "loss": 2.6287, "step": 17410 }, { "epoch": 4.724708435042039, "grad_norm": 12.464961278156949, "learning_rate": 8.657296970789641e-06, "loss": 2.8029, "step": 17420 }, { "epoch": 4.727420667209113, "grad_norm": 14.051788096571483, "learning_rate": 8.65514431079937e-06, "loss": 2.6177, "step": 17430 }, { "epoch": 4.730132899376187, "grad_norm": 15.343296406652277, "learning_rate": 8.652990194682375e-06, "loss": 2.6951, "step": 17440 }, { "epoch": 4.73284513154326, "grad_norm": 13.833822348182682, "learning_rate": 8.650834623296809e-06, "loss": 2.686, "step": 17450 }, { "epoch": 4.735557363710334, "grad_norm": 14.918469257695985, "learning_rate": 8.648677597501401e-06, "loss": 2.5552, "step": 17460 }, { "epoch": 4.738269595877407, "grad_norm": 16.37428204917812, "learning_rate": 8.646519118155461e-06, "loss": 2.6486, "step": 17470 }, { "epoch": 4.740981828044481, "grad_norm": 13.283314441836296, "learning_rate": 8.644359186118877e-06, "loss": 2.7302, "step": 17480 }, { "epoch": 4.7436940602115545, "grad_norm": 12.542523166823456, "learning_rate": 8.642197802252123e-06, "loss": 2.6603, "step": 17490 }, { "epoch": 4.746406292378627, "grad_norm": 13.9280023507465, "learning_rate": 8.64003496741624e-06, "loss": 2.6105, "step": 17500 }, { "epoch": 4.749118524545701, "grad_norm": 13.5537149640925, "learning_rate": 8.637870682472854e-06, "loss": 2.6719, "step": 17510 }, { "epoch": 4.751830756712774, "grad_norm": 14.449328692766871, "learning_rate": 8.635704948284164e-06, "loss": 2.8152, "step": 17520 }, { "epoch": 4.754542988879848, "grad_norm": 13.518051080849476, "learning_rate": 8.633537765712955e-06, "loss": 2.6471, "step": 17530 }, { "epoch": 4.757255221046922, "grad_norm": 12.902193344043402, "learning_rate": 8.63136913562258e-06, "loss": 2.69, "step": 17540 }, { "epoch": 4.759967453213995, "grad_norm": 12.421131440179838, "learning_rate": 8.629199058876973e-06, "loss": 2.6881, "step": 17550 }, { "epoch": 4.762679685381069, "grad_norm": 13.293415429340714, "learning_rate": 8.627027536340645e-06, "loss": 2.6549, "step": 17560 }, { "epoch": 4.7653919175481425, "grad_norm": 13.409316292409748, "learning_rate": 8.624854568878678e-06, "loss": 2.7382, "step": 17570 }, { "epoch": 4.7681041497152155, "grad_norm": 13.434301032143962, "learning_rate": 8.622680157356736e-06, "loss": 2.5886, "step": 17580 }, { "epoch": 4.770816381882289, "grad_norm": 13.743503757827945, "learning_rate": 8.620504302641051e-06, "loss": 2.5976, "step": 17590 }, { "epoch": 4.773528614049362, "grad_norm": 16.10738876117641, "learning_rate": 8.61832700559844e-06, "loss": 2.7197, "step": 17600 }, { "epoch": 4.776240846216436, "grad_norm": 13.93949305730481, "learning_rate": 8.616148267096287e-06, "loss": 2.6287, "step": 17610 }, { "epoch": 4.77895307838351, "grad_norm": 14.239018262888717, "learning_rate": 8.613968088002549e-06, "loss": 2.5176, "step": 17620 }, { "epoch": 4.781665310550583, "grad_norm": 13.207493438862645, "learning_rate": 8.611786469185764e-06, "loss": 2.8071, "step": 17630 }, { "epoch": 4.784377542717657, "grad_norm": 13.776299126705592, "learning_rate": 8.609603411515036e-06, "loss": 2.6058, "step": 17640 }, { "epoch": 4.7870897748847305, "grad_norm": 14.206277458556515, "learning_rate": 8.607418915860049e-06, "loss": 2.5653, "step": 17650 }, { "epoch": 4.7898020070518035, "grad_norm": 14.461722246572352, "learning_rate": 8.605232983091054e-06, "loss": 2.6426, "step": 17660 }, { "epoch": 4.792514239218877, "grad_norm": 13.31260874274617, "learning_rate": 8.60304561407888e-06, "loss": 2.7115, "step": 17670 }, { "epoch": 4.79522647138595, "grad_norm": 13.283806407439828, "learning_rate": 8.600856809694922e-06, "loss": 2.6532, "step": 17680 }, { "epoch": 4.797938703553024, "grad_norm": 15.634737222900277, "learning_rate": 8.598666570811151e-06, "loss": 2.5342, "step": 17690 }, { "epoch": 4.800650935720098, "grad_norm": 14.967293893335187, "learning_rate": 8.596474898300111e-06, "loss": 2.6942, "step": 17700 }, { "epoch": 4.803363167887171, "grad_norm": 13.983527101536005, "learning_rate": 8.594281793034913e-06, "loss": 2.6254, "step": 17710 }, { "epoch": 4.806075400054245, "grad_norm": 16.4774653935633, "learning_rate": 8.59208725588924e-06, "loss": 2.6407, "step": 17720 }, { "epoch": 4.808787632221318, "grad_norm": 15.303523441152374, "learning_rate": 8.589891287737346e-06, "loss": 2.721, "step": 17730 }, { "epoch": 4.8114998643883915, "grad_norm": 16.52018305961032, "learning_rate": 8.587693889454055e-06, "loss": 2.571, "step": 17740 }, { "epoch": 4.814212096555465, "grad_norm": 16.503516953600865, "learning_rate": 8.58549506191476e-06, "loss": 2.7775, "step": 17750 }, { "epoch": 4.816924328722538, "grad_norm": 12.697804033415387, "learning_rate": 8.583294805995427e-06, "loss": 2.6792, "step": 17760 }, { "epoch": 4.819636560889612, "grad_norm": 14.57950868535272, "learning_rate": 8.581093122572585e-06, "loss": 2.7099, "step": 17770 }, { "epoch": 4.822348793056686, "grad_norm": 14.43296807385412, "learning_rate": 8.578890012523336e-06, "loss": 2.7108, "step": 17780 }, { "epoch": 4.825061025223759, "grad_norm": 14.510133506049852, "learning_rate": 8.57668547672535e-06, "loss": 2.6902, "step": 17790 }, { "epoch": 4.827773257390833, "grad_norm": 13.519612777521239, "learning_rate": 8.574479516056864e-06, "loss": 2.6327, "step": 17800 }, { "epoch": 4.830485489557907, "grad_norm": 13.785407940268515, "learning_rate": 8.57227213139668e-06, "loss": 2.6338, "step": 17810 }, { "epoch": 4.8331977217249795, "grad_norm": 13.512749658868994, "learning_rate": 8.570063323624174e-06, "loss": 2.7323, "step": 17820 }, { "epoch": 4.835909953892053, "grad_norm": 12.564378512940648, "learning_rate": 8.567853093619282e-06, "loss": 2.5658, "step": 17830 }, { "epoch": 4.838622186059126, "grad_norm": 13.847009370060354, "learning_rate": 8.56564144226251e-06, "loss": 2.6648, "step": 17840 }, { "epoch": 4.8413344182262, "grad_norm": 12.048426684577219, "learning_rate": 8.563428370434931e-06, "loss": 2.7555, "step": 17850 }, { "epoch": 4.844046650393274, "grad_norm": 14.395991051356454, "learning_rate": 8.561213879018184e-06, "loss": 2.7315, "step": 17860 }, { "epoch": 4.846758882560347, "grad_norm": 12.68618980469867, "learning_rate": 8.55899796889447e-06, "loss": 2.6814, "step": 17870 }, { "epoch": 4.849471114727421, "grad_norm": 14.276013385082884, "learning_rate": 8.556780640946558e-06, "loss": 2.5803, "step": 17880 }, { "epoch": 4.852183346894494, "grad_norm": 14.56560993536756, "learning_rate": 8.55456189605778e-06, "loss": 2.6393, "step": 17890 }, { "epoch": 4.8548955790615675, "grad_norm": 13.016518533191128, "learning_rate": 8.552341735112036e-06, "loss": 2.6063, "step": 17900 }, { "epoch": 4.857607811228641, "grad_norm": 12.930254942724071, "learning_rate": 8.550120158993786e-06, "loss": 2.7387, "step": 17910 }, { "epoch": 4.860320043395714, "grad_norm": 14.736487385808942, "learning_rate": 8.547897168588056e-06, "loss": 2.539, "step": 17920 }, { "epoch": 4.863032275562788, "grad_norm": 13.257439450789638, "learning_rate": 8.545672764780438e-06, "loss": 2.6789, "step": 17930 }, { "epoch": 4.865744507729862, "grad_norm": 12.630713596181605, "learning_rate": 8.543446948457077e-06, "loss": 2.7463, "step": 17940 }, { "epoch": 4.868456739896935, "grad_norm": 14.374486654719194, "learning_rate": 8.541219720504696e-06, "loss": 2.6485, "step": 17950 }, { "epoch": 4.871168972064009, "grad_norm": 12.911864323438182, "learning_rate": 8.538991081810567e-06, "loss": 2.622, "step": 17960 }, { "epoch": 4.873881204231083, "grad_norm": 13.174263164457269, "learning_rate": 8.53676103326253e-06, "loss": 2.6635, "step": 17970 }, { "epoch": 4.876593436398156, "grad_norm": 13.738587437247453, "learning_rate": 8.534529575748988e-06, "loss": 2.8014, "step": 17980 }, { "epoch": 4.879305668565229, "grad_norm": 13.406240974608775, "learning_rate": 8.5322967101589e-06, "loss": 2.6963, "step": 17990 }, { "epoch": 4.882017900732302, "grad_norm": 13.272601107859362, "learning_rate": 8.530062437381792e-06, "loss": 2.6376, "step": 18000 }, { "epoch": 4.884730132899376, "grad_norm": 13.946829548076384, "learning_rate": 8.527826758307746e-06, "loss": 2.6583, "step": 18010 }, { "epoch": 4.88744236506645, "grad_norm": 13.446239236173966, "learning_rate": 8.525589673827408e-06, "loss": 2.6853, "step": 18020 }, { "epoch": 4.890154597233523, "grad_norm": 13.525715665728194, "learning_rate": 8.523351184831979e-06, "loss": 2.7209, "step": 18030 }, { "epoch": 4.892866829400597, "grad_norm": 12.188314342868711, "learning_rate": 8.521111292213221e-06, "loss": 2.7904, "step": 18040 }, { "epoch": 4.89557906156767, "grad_norm": 13.759196430332569, "learning_rate": 8.518869996863463e-06, "loss": 2.8072, "step": 18050 }, { "epoch": 4.898291293734744, "grad_norm": 14.27597876802336, "learning_rate": 8.51662729967558e-06, "loss": 2.6825, "step": 18060 }, { "epoch": 4.901003525901817, "grad_norm": 12.31733830459825, "learning_rate": 8.514383201543016e-06, "loss": 2.6462, "step": 18070 }, { "epoch": 4.90371575806889, "grad_norm": 13.761198205117674, "learning_rate": 8.512137703359765e-06, "loss": 2.7548, "step": 18080 }, { "epoch": 4.906427990235964, "grad_norm": 12.874011439001412, "learning_rate": 8.509890806020385e-06, "loss": 2.8556, "step": 18090 }, { "epoch": 4.909140222403038, "grad_norm": 13.43184637570081, "learning_rate": 8.507642510419988e-06, "loss": 2.6251, "step": 18100 }, { "epoch": 4.911852454570111, "grad_norm": 14.623307944944697, "learning_rate": 8.505392817454245e-06, "loss": 2.6714, "step": 18110 }, { "epoch": 4.914564686737185, "grad_norm": 13.80483850356093, "learning_rate": 8.503141728019384e-06, "loss": 2.7318, "step": 18120 }, { "epoch": 4.917276918904259, "grad_norm": 13.077649563550708, "learning_rate": 8.500889243012184e-06, "loss": 2.6435, "step": 18130 }, { "epoch": 4.919989151071332, "grad_norm": 13.048012288804317, "learning_rate": 8.498635363329988e-06, "loss": 2.73, "step": 18140 }, { "epoch": 4.9227013832384054, "grad_norm": 13.071135812252063, "learning_rate": 8.496380089870688e-06, "loss": 2.7663, "step": 18150 }, { "epoch": 4.925413615405478, "grad_norm": 13.632509104738997, "learning_rate": 8.494123423532737e-06, "loss": 2.6723, "step": 18160 }, { "epoch": 4.928125847572552, "grad_norm": 13.464782887768992, "learning_rate": 8.491865365215135e-06, "loss": 2.6968, "step": 18170 }, { "epoch": 4.930838079739626, "grad_norm": 13.846018216966286, "learning_rate": 8.489605915817446e-06, "loss": 2.7508, "step": 18180 }, { "epoch": 4.933550311906699, "grad_norm": 15.732123760987147, "learning_rate": 8.487345076239781e-06, "loss": 2.7354, "step": 18190 }, { "epoch": 4.936262544073773, "grad_norm": 12.279354452978303, "learning_rate": 8.485082847382808e-06, "loss": 2.7636, "step": 18200 }, { "epoch": 4.938974776240846, "grad_norm": 13.245077911529165, "learning_rate": 8.482819230147747e-06, "loss": 2.6647, "step": 18210 }, { "epoch": 4.94168700840792, "grad_norm": 12.74968815423554, "learning_rate": 8.480554225436373e-06, "loss": 2.7732, "step": 18220 }, { "epoch": 4.9443992405749935, "grad_norm": 14.449317896941091, "learning_rate": 8.47828783415101e-06, "loss": 2.6957, "step": 18230 }, { "epoch": 4.947111472742066, "grad_norm": 14.189706589354747, "learning_rate": 8.47602005719454e-06, "loss": 2.7549, "step": 18240 }, { "epoch": 4.94982370490914, "grad_norm": 14.164617058321637, "learning_rate": 8.473750895470394e-06, "loss": 2.74, "step": 18250 }, { "epoch": 4.952535937076214, "grad_norm": 13.584707891376341, "learning_rate": 8.471480349882553e-06, "loss": 2.6417, "step": 18260 }, { "epoch": 4.955248169243287, "grad_norm": 13.201522344005394, "learning_rate": 8.46920842133555e-06, "loss": 2.6199, "step": 18270 }, { "epoch": 4.957960401410361, "grad_norm": 11.722929260624122, "learning_rate": 8.466935110734475e-06, "loss": 2.7681, "step": 18280 }, { "epoch": 4.960672633577435, "grad_norm": 15.691329916544706, "learning_rate": 8.464660418984957e-06, "loss": 2.7449, "step": 18290 }, { "epoch": 4.963384865744508, "grad_norm": 13.26978648852302, "learning_rate": 8.462384346993186e-06, "loss": 2.7155, "step": 18300 }, { "epoch": 4.9660970979115815, "grad_norm": 13.70533236998019, "learning_rate": 8.460106895665895e-06, "loss": 2.7113, "step": 18310 }, { "epoch": 4.968809330078654, "grad_norm": 14.311974867481968, "learning_rate": 8.45782806591037e-06, "loss": 2.7682, "step": 18320 }, { "epoch": 4.971521562245728, "grad_norm": 13.640778562065156, "learning_rate": 8.455547858634445e-06, "loss": 2.6703, "step": 18330 }, { "epoch": 4.974233794412802, "grad_norm": 13.511911830160628, "learning_rate": 8.453266274746505e-06, "loss": 2.6232, "step": 18340 }, { "epoch": 4.976946026579875, "grad_norm": 15.389659241401079, "learning_rate": 8.450983315155479e-06, "loss": 2.7914, "step": 18350 }, { "epoch": 4.979658258746949, "grad_norm": 13.727019026037636, "learning_rate": 8.448698980770846e-06, "loss": 2.5677, "step": 18360 }, { "epoch": 4.982370490914022, "grad_norm": 13.345322529355519, "learning_rate": 8.446413272502635e-06, "loss": 2.6392, "step": 18370 }, { "epoch": 4.985082723081096, "grad_norm": 12.90571855352788, "learning_rate": 8.444126191261419e-06, "loss": 2.6597, "step": 18380 }, { "epoch": 4.9877949552481695, "grad_norm": 12.872281294242681, "learning_rate": 8.44183773795832e-06, "loss": 2.7495, "step": 18390 }, { "epoch": 4.9905071874152425, "grad_norm": 13.445899379913042, "learning_rate": 8.439547913505006e-06, "loss": 2.6874, "step": 18400 }, { "epoch": 4.993219419582316, "grad_norm": 11.596646410090548, "learning_rate": 8.437256718813692e-06, "loss": 2.7577, "step": 18410 }, { "epoch": 4.99593165174939, "grad_norm": 13.1580307915758, "learning_rate": 8.434964154797136e-06, "loss": 2.7104, "step": 18420 }, { "epoch": 4.998643883916463, "grad_norm": 13.71174751457682, "learning_rate": 8.432670222368643e-06, "loss": 2.7645, "step": 18430 }, { "epoch": 5.001356116083537, "grad_norm": 13.528916724930992, "learning_rate": 8.430374922442068e-06, "loss": 2.4203, "step": 18440 }, { "epoch": 5.00406834825061, "grad_norm": 15.882800252470672, "learning_rate": 8.428078255931804e-06, "loss": 2.2077, "step": 18450 }, { "epoch": 5.006780580417684, "grad_norm": 14.881526536012991, "learning_rate": 8.425780223752791e-06, "loss": 2.1805, "step": 18460 }, { "epoch": 5.0094928125847575, "grad_norm": 15.324594446936311, "learning_rate": 8.423480826820512e-06, "loss": 2.1409, "step": 18470 }, { "epoch": 5.0122050447518305, "grad_norm": 13.751732274202018, "learning_rate": 8.421180066050996e-06, "loss": 2.0451, "step": 18480 }, { "epoch": 5.014917276918904, "grad_norm": 15.8807491730725, "learning_rate": 8.418877942360812e-06, "loss": 2.0094, "step": 18490 }, { "epoch": 5.017629509085978, "grad_norm": 14.03017429267143, "learning_rate": 8.416574456667077e-06, "loss": 2.1888, "step": 18500 }, { "epoch": 5.020341741253051, "grad_norm": 14.720531804576803, "learning_rate": 8.414269609887446e-06, "loss": 2.1741, "step": 18510 }, { "epoch": 5.023053973420125, "grad_norm": 14.192094382907161, "learning_rate": 8.41196340294012e-06, "loss": 2.2388, "step": 18520 }, { "epoch": 5.025766205587198, "grad_norm": 17.269394607227035, "learning_rate": 8.409655836743833e-06, "loss": 1.9993, "step": 18530 }, { "epoch": 5.028478437754272, "grad_norm": 15.576385970481855, "learning_rate": 8.407346912217875e-06, "loss": 2.0905, "step": 18540 }, { "epoch": 5.031190669921346, "grad_norm": 16.35563370196204, "learning_rate": 8.405036630282063e-06, "loss": 2.1836, "step": 18550 }, { "epoch": 5.0339029020884185, "grad_norm": 15.958762378535535, "learning_rate": 8.402724991856766e-06, "loss": 2.1124, "step": 18560 }, { "epoch": 5.036615134255492, "grad_norm": 15.153717214439391, "learning_rate": 8.400411997862885e-06, "loss": 2.1467, "step": 18570 }, { "epoch": 5.039327366422566, "grad_norm": 15.489932171515386, "learning_rate": 8.398097649221865e-06, "loss": 2.222, "step": 18580 }, { "epoch": 5.042039598589639, "grad_norm": 15.913749277526303, "learning_rate": 8.395781946855692e-06, "loss": 2.1089, "step": 18590 }, { "epoch": 5.044751830756713, "grad_norm": 14.332153946383157, "learning_rate": 8.39346489168689e-06, "loss": 2.1586, "step": 18600 }, { "epoch": 5.047464062923786, "grad_norm": 18.27921389351497, "learning_rate": 8.391146484638516e-06, "loss": 2.1288, "step": 18610 }, { "epoch": 5.05017629509086, "grad_norm": 17.48516295723462, "learning_rate": 8.388826726634176e-06, "loss": 2.184, "step": 18620 }, { "epoch": 5.052888527257934, "grad_norm": 13.8994477501439, "learning_rate": 8.386505618598006e-06, "loss": 2.0204, "step": 18630 }, { "epoch": 5.0556007594250065, "grad_norm": 16.402772731892462, "learning_rate": 8.384183161454687e-06, "loss": 2.156, "step": 18640 }, { "epoch": 5.05831299159208, "grad_norm": 14.81839411438416, "learning_rate": 8.381859356129427e-06, "loss": 2.1264, "step": 18650 }, { "epoch": 5.061025223759154, "grad_norm": 15.93768545914976, "learning_rate": 8.379534203547983e-06, "loss": 2.111, "step": 18660 }, { "epoch": 5.063737455926227, "grad_norm": 14.54847038089006, "learning_rate": 8.37720770463664e-06, "loss": 2.0215, "step": 18670 }, { "epoch": 5.066449688093301, "grad_norm": 16.655828865292502, "learning_rate": 8.374879860322226e-06, "loss": 2.1104, "step": 18680 }, { "epoch": 5.069161920260374, "grad_norm": 14.766278492715498, "learning_rate": 8.372550671532096e-06, "loss": 2.0753, "step": 18690 }, { "epoch": 5.071874152427448, "grad_norm": 15.990563578737197, "learning_rate": 8.370220139194152e-06, "loss": 2.1312, "step": 18700 }, { "epoch": 5.074586384594522, "grad_norm": 15.620672800341469, "learning_rate": 8.36788826423682e-06, "loss": 2.0553, "step": 18710 }, { "epoch": 5.0772986167615946, "grad_norm": 15.103175647661342, "learning_rate": 8.365555047589071e-06, "loss": 2.0157, "step": 18720 }, { "epoch": 5.080010848928668, "grad_norm": 17.301834926261435, "learning_rate": 8.363220490180402e-06, "loss": 2.1604, "step": 18730 }, { "epoch": 5.082723081095742, "grad_norm": 16.31321472571673, "learning_rate": 8.360884592940848e-06, "loss": 2.119, "step": 18740 }, { "epoch": 5.085435313262815, "grad_norm": 17.228880788835312, "learning_rate": 8.35854735680098e-06, "loss": 1.9593, "step": 18750 }, { "epoch": 5.088147545429889, "grad_norm": 15.413108747715361, "learning_rate": 8.3562087826919e-06, "loss": 2.1539, "step": 18760 }, { "epoch": 5.090859777596962, "grad_norm": 15.099361694430046, "learning_rate": 8.35386887154524e-06, "loss": 2.0211, "step": 18770 }, { "epoch": 5.093572009764036, "grad_norm": 15.243410558244175, "learning_rate": 8.351527624293167e-06, "loss": 2.1848, "step": 18780 }, { "epoch": 5.09628424193111, "grad_norm": 14.497475413915657, "learning_rate": 8.349185041868385e-06, "loss": 2.222, "step": 18790 }, { "epoch": 5.098996474098183, "grad_norm": 17.235425723928937, "learning_rate": 8.346841125204124e-06, "loss": 2.121, "step": 18800 }, { "epoch": 5.101708706265256, "grad_norm": 15.594304909805977, "learning_rate": 8.344495875234148e-06, "loss": 2.0789, "step": 18810 }, { "epoch": 5.10442093843233, "grad_norm": 16.90754729962433, "learning_rate": 8.34214929289275e-06, "loss": 2.1604, "step": 18820 }, { "epoch": 5.107133170599403, "grad_norm": 16.298054872137474, "learning_rate": 8.339801379114756e-06, "loss": 2.2375, "step": 18830 }, { "epoch": 5.109845402766477, "grad_norm": 15.650258967435393, "learning_rate": 8.337452134835522e-06, "loss": 2.0874, "step": 18840 }, { "epoch": 5.11255763493355, "grad_norm": 17.869238464845463, "learning_rate": 8.335101560990933e-06, "loss": 2.071, "step": 18850 }, { "epoch": 5.115269867100624, "grad_norm": 15.827732911278884, "learning_rate": 8.332749658517404e-06, "loss": 2.0662, "step": 18860 }, { "epoch": 5.117982099267698, "grad_norm": 15.697648628154491, "learning_rate": 8.330396428351881e-06, "loss": 2.1123, "step": 18870 }, { "epoch": 5.120694331434771, "grad_norm": 15.46455671937288, "learning_rate": 8.328041871431837e-06, "loss": 2.1915, "step": 18880 }, { "epoch": 5.123406563601844, "grad_norm": 16.586311625169717, "learning_rate": 8.325685988695274e-06, "loss": 2.0412, "step": 18890 }, { "epoch": 5.126118795768917, "grad_norm": 14.353290075770762, "learning_rate": 8.323328781080724e-06, "loss": 2.1936, "step": 18900 }, { "epoch": 5.128831027935991, "grad_norm": 16.539378814766206, "learning_rate": 8.320970249527241e-06, "loss": 2.1074, "step": 18910 }, { "epoch": 5.131543260103065, "grad_norm": 14.899286156203662, "learning_rate": 8.318610394974415e-06, "loss": 2.1114, "step": 18920 }, { "epoch": 5.134255492270138, "grad_norm": 16.006687411783222, "learning_rate": 8.316249218362357e-06, "loss": 2.0416, "step": 18930 }, { "epoch": 5.136967724437212, "grad_norm": 16.53398059649155, "learning_rate": 8.313886720631705e-06, "loss": 2.195, "step": 18940 }, { "epoch": 5.139679956604286, "grad_norm": 16.819249064247007, "learning_rate": 8.311522902723627e-06, "loss": 2.1534, "step": 18950 }, { "epoch": 5.142392188771359, "grad_norm": 16.931415528933698, "learning_rate": 8.309157765579814e-06, "loss": 2.2452, "step": 18960 }, { "epoch": 5.1451044209384325, "grad_norm": 15.088776919758375, "learning_rate": 8.306791310142481e-06, "loss": 2.0879, "step": 18970 }, { "epoch": 5.147816653105505, "grad_norm": 17.157444875230258, "learning_rate": 8.304423537354372e-06, "loss": 2.1803, "step": 18980 }, { "epoch": 5.150528885272579, "grad_norm": 17.0877718144278, "learning_rate": 8.302054448158751e-06, "loss": 2.2011, "step": 18990 }, { "epoch": 5.153241117439653, "grad_norm": 15.926642289181856, "learning_rate": 8.299684043499415e-06, "loss": 2.2798, "step": 19000 }, { "epoch": 5.155953349606726, "grad_norm": 15.621432176763534, "learning_rate": 8.297312324320676e-06, "loss": 2.2403, "step": 19010 }, { "epoch": 5.1586655817738, "grad_norm": 15.82057552609413, "learning_rate": 8.294939291567374e-06, "loss": 2.1312, "step": 19020 }, { "epoch": 5.161377813940874, "grad_norm": 17.6176195390361, "learning_rate": 8.292564946184873e-06, "loss": 2.0495, "step": 19030 }, { "epoch": 5.164090046107947, "grad_norm": 15.742089984194145, "learning_rate": 8.290189289119054e-06, "loss": 2.1955, "step": 19040 }, { "epoch": 5.1668022782750205, "grad_norm": 16.263611332994678, "learning_rate": 8.28781232131633e-06, "loss": 2.0421, "step": 19050 }, { "epoch": 5.169514510442093, "grad_norm": 18.107213895605938, "learning_rate": 8.285434043723627e-06, "loss": 2.0407, "step": 19060 }, { "epoch": 5.172226742609167, "grad_norm": 15.82147807940213, "learning_rate": 8.2830544572884e-06, "loss": 2.1814, "step": 19070 }, { "epoch": 5.174938974776241, "grad_norm": 17.472000994294465, "learning_rate": 8.280673562958622e-06, "loss": 2.1557, "step": 19080 }, { "epoch": 5.177651206943314, "grad_norm": 15.206101368031495, "learning_rate": 8.278291361682786e-06, "loss": 2.1924, "step": 19090 }, { "epoch": 5.180363439110388, "grad_norm": 14.96563962085415, "learning_rate": 8.275907854409909e-06, "loss": 2.1087, "step": 19100 }, { "epoch": 5.183075671277462, "grad_norm": 14.474236733973989, "learning_rate": 8.273523042089522e-06, "loss": 2.2117, "step": 19110 }, { "epoch": 5.185787903444535, "grad_norm": 17.013850739784434, "learning_rate": 8.271136925671686e-06, "loss": 2.2211, "step": 19120 }, { "epoch": 5.1885001356116085, "grad_norm": 14.643320926516239, "learning_rate": 8.26874950610697e-06, "loss": 2.2176, "step": 19130 }, { "epoch": 5.1912123677786814, "grad_norm": 15.702832605665217, "learning_rate": 8.266360784346475e-06, "loss": 2.136, "step": 19140 }, { "epoch": 5.193924599945755, "grad_norm": 16.79055911284228, "learning_rate": 8.263970761341806e-06, "loss": 2.1882, "step": 19150 }, { "epoch": 5.196636832112829, "grad_norm": 15.376244499855357, "learning_rate": 8.2615794380451e-06, "loss": 2.111, "step": 19160 }, { "epoch": 5.199349064279902, "grad_norm": 16.175662106030565, "learning_rate": 8.259186815409001e-06, "loss": 2.113, "step": 19170 }, { "epoch": 5.202061296446976, "grad_norm": 16.7213624769967, "learning_rate": 8.25679289438668e-06, "loss": 2.2056, "step": 19180 }, { "epoch": 5.20477352861405, "grad_norm": 15.705221200978485, "learning_rate": 8.25439767593182e-06, "loss": 2.1408, "step": 19190 }, { "epoch": 5.207485760781123, "grad_norm": 15.41239046746615, "learning_rate": 8.25200116099862e-06, "loss": 2.3146, "step": 19200 }, { "epoch": 5.2101979929481965, "grad_norm": 14.235341897607679, "learning_rate": 8.249603350541796e-06, "loss": 2.1858, "step": 19210 }, { "epoch": 5.2129102251152695, "grad_norm": 15.403067803423001, "learning_rate": 8.247204245516588e-06, "loss": 2.1267, "step": 19220 }, { "epoch": 5.215622457282343, "grad_norm": 16.091946051459516, "learning_rate": 8.244803846878739e-06, "loss": 2.367, "step": 19230 }, { "epoch": 5.218334689449417, "grad_norm": 15.33881704124221, "learning_rate": 8.242402155584514e-06, "loss": 2.2882, "step": 19240 }, { "epoch": 5.22104692161649, "grad_norm": 15.483121940854678, "learning_rate": 8.239999172590696e-06, "loss": 2.0526, "step": 19250 }, { "epoch": 5.223759153783564, "grad_norm": 16.881872219985933, "learning_rate": 8.237594898854575e-06, "loss": 2.1877, "step": 19260 }, { "epoch": 5.226471385950638, "grad_norm": 17.789404034964644, "learning_rate": 8.235189335333963e-06, "loss": 2.1941, "step": 19270 }, { "epoch": 5.229183618117711, "grad_norm": 19.577464816120525, "learning_rate": 8.232782482987178e-06, "loss": 2.1148, "step": 19280 }, { "epoch": 5.2318958502847845, "grad_norm": 14.386322261392163, "learning_rate": 8.230374342773057e-06, "loss": 2.1425, "step": 19290 }, { "epoch": 5.2346080824518575, "grad_norm": 15.339737362684938, "learning_rate": 8.22796491565095e-06, "loss": 2.1741, "step": 19300 }, { "epoch": 5.237320314618931, "grad_norm": 14.720374847143619, "learning_rate": 8.225554202580717e-06, "loss": 2.118, "step": 19310 }, { "epoch": 5.240032546786005, "grad_norm": 17.96692682906325, "learning_rate": 8.223142204522731e-06, "loss": 2.0662, "step": 19320 }, { "epoch": 5.242744778953078, "grad_norm": 16.536957243132015, "learning_rate": 8.22072892243788e-06, "loss": 2.0971, "step": 19330 }, { "epoch": 5.245457011120152, "grad_norm": 17.098741804759115, "learning_rate": 8.218314357287558e-06, "loss": 2.1513, "step": 19340 }, { "epoch": 5.248169243287226, "grad_norm": 18.325772977820126, "learning_rate": 8.215898510033672e-06, "loss": 2.1311, "step": 19350 }, { "epoch": 5.250881475454299, "grad_norm": 17.860826404194306, "learning_rate": 8.213481381638646e-06, "loss": 2.1763, "step": 19360 }, { "epoch": 5.253593707621373, "grad_norm": 15.453975957997534, "learning_rate": 8.211062973065405e-06, "loss": 2.2528, "step": 19370 }, { "epoch": 5.2563059397884455, "grad_norm": 15.606334131524694, "learning_rate": 8.208643285277389e-06, "loss": 2.1831, "step": 19380 }, { "epoch": 5.259018171955519, "grad_norm": 17.584037624410424, "learning_rate": 8.206222319238547e-06, "loss": 2.1967, "step": 19390 }, { "epoch": 5.261730404122593, "grad_norm": 17.275141156269335, "learning_rate": 8.203800075913338e-06, "loss": 2.2156, "step": 19400 }, { "epoch": 5.264442636289666, "grad_norm": 15.9026788364695, "learning_rate": 8.201376556266728e-06, "loss": 2.1721, "step": 19410 }, { "epoch": 5.26715486845674, "grad_norm": 16.639154319720756, "learning_rate": 8.198951761264192e-06, "loss": 2.1666, "step": 19420 }, { "epoch": 5.269867100623814, "grad_norm": 16.100031573788947, "learning_rate": 8.196525691871714e-06, "loss": 2.1591, "step": 19430 }, { "epoch": 5.272579332790887, "grad_norm": 16.928726429704128, "learning_rate": 8.194098349055785e-06, "loss": 2.1765, "step": 19440 }, { "epoch": 5.275291564957961, "grad_norm": 14.62948775634189, "learning_rate": 8.191669733783402e-06, "loss": 2.2231, "step": 19450 }, { "epoch": 5.2780037971250335, "grad_norm": 14.433376856025946, "learning_rate": 8.189239847022074e-06, "loss": 2.2221, "step": 19460 }, { "epoch": 5.280716029292107, "grad_norm": 17.008600645100138, "learning_rate": 8.186808689739811e-06, "loss": 2.2721, "step": 19470 }, { "epoch": 5.283428261459181, "grad_norm": 16.014341151412815, "learning_rate": 8.184376262905129e-06, "loss": 2.2613, "step": 19480 }, { "epoch": 5.286140493626254, "grad_norm": 16.937519607323864, "learning_rate": 8.181942567487054e-06, "loss": 2.196, "step": 19490 }, { "epoch": 5.288852725793328, "grad_norm": 17.849985423501238, "learning_rate": 8.179507604455113e-06, "loss": 2.1887, "step": 19500 }, { "epoch": 5.291564957960402, "grad_norm": 16.172460028386002, "learning_rate": 8.177071374779342e-06, "loss": 2.0393, "step": 19510 }, { "epoch": 5.294277190127475, "grad_norm": 15.345632052762186, "learning_rate": 8.174633879430282e-06, "loss": 2.1762, "step": 19520 }, { "epoch": 5.296989422294549, "grad_norm": 15.210315222031847, "learning_rate": 8.172195119378971e-06, "loss": 2.173, "step": 19530 }, { "epoch": 5.299701654461622, "grad_norm": 16.197714070190685, "learning_rate": 8.169755095596958e-06, "loss": 2.1663, "step": 19540 }, { "epoch": 5.302413886628695, "grad_norm": 17.558159198165516, "learning_rate": 8.167313809056294e-06, "loss": 2.156, "step": 19550 }, { "epoch": 5.305126118795769, "grad_norm": 16.539634398526527, "learning_rate": 8.164871260729532e-06, "loss": 2.1746, "step": 19560 }, { "epoch": 5.307838350962842, "grad_norm": 14.771010384656496, "learning_rate": 8.162427451589726e-06, "loss": 2.2561, "step": 19570 }, { "epoch": 5.310550583129916, "grad_norm": 16.506577986155484, "learning_rate": 8.159982382610438e-06, "loss": 2.1109, "step": 19580 }, { "epoch": 5.313262815296989, "grad_norm": 15.998429865020467, "learning_rate": 8.157536054765725e-06, "loss": 2.0621, "step": 19590 }, { "epoch": 5.315975047464063, "grad_norm": 16.033833777353678, "learning_rate": 8.155088469030147e-06, "loss": 2.1976, "step": 19600 }, { "epoch": 5.318687279631137, "grad_norm": 16.17332721515477, "learning_rate": 8.152639626378773e-06, "loss": 2.1926, "step": 19610 }, { "epoch": 5.32139951179821, "grad_norm": 15.887866032945395, "learning_rate": 8.15018952778716e-06, "loss": 2.1589, "step": 19620 }, { "epoch": 5.324111743965283, "grad_norm": 16.177921426137132, "learning_rate": 8.147738174231376e-06, "loss": 2.1663, "step": 19630 }, { "epoch": 5.326823976132357, "grad_norm": 16.85837403901552, "learning_rate": 8.145285566687983e-06, "loss": 2.0816, "step": 19640 }, { "epoch": 5.32953620829943, "grad_norm": 15.58701343300539, "learning_rate": 8.142831706134045e-06, "loss": 2.2063, "step": 19650 }, { "epoch": 5.332248440466504, "grad_norm": 16.506497487965806, "learning_rate": 8.140376593547122e-06, "loss": 2.2024, "step": 19660 }, { "epoch": 5.334960672633578, "grad_norm": 16.78414763443575, "learning_rate": 8.13792022990528e-06, "loss": 2.1163, "step": 19670 }, { "epoch": 5.337672904800651, "grad_norm": 15.821506977280531, "learning_rate": 8.135462616187073e-06, "loss": 2.3137, "step": 19680 }, { "epoch": 5.340385136967725, "grad_norm": 18.908441926612525, "learning_rate": 8.133003753371564e-06, "loss": 2.2016, "step": 19690 }, { "epoch": 5.343097369134798, "grad_norm": 15.073716492067412, "learning_rate": 8.130543642438304e-06, "loss": 2.1738, "step": 19700 }, { "epoch": 5.345809601301871, "grad_norm": 14.457968672557586, "learning_rate": 8.12808228436735e-06, "loss": 2.1712, "step": 19710 }, { "epoch": 5.348521833468945, "grad_norm": 15.633393109635263, "learning_rate": 8.125619680139245e-06, "loss": 2.2443, "step": 19720 }, { "epoch": 5.351234065636018, "grad_norm": 15.320690960323384, "learning_rate": 8.12315583073504e-06, "loss": 2.4915, "step": 19730 }, { "epoch": 5.353946297803092, "grad_norm": 15.753037744911492, "learning_rate": 8.120690737136276e-06, "loss": 2.1147, "step": 19740 }, { "epoch": 5.356658529970165, "grad_norm": 16.023491086924558, "learning_rate": 8.11822440032499e-06, "loss": 2.2736, "step": 19750 }, { "epoch": 5.359370762137239, "grad_norm": 18.506019969409163, "learning_rate": 8.115756821283714e-06, "loss": 2.1451, "step": 19760 }, { "epoch": 5.362082994304313, "grad_norm": 15.912759577429394, "learning_rate": 8.113288000995473e-06, "loss": 2.2136, "step": 19770 }, { "epoch": 5.364795226471386, "grad_norm": 16.34065190237075, "learning_rate": 8.110817940443795e-06, "loss": 2.1741, "step": 19780 }, { "epoch": 5.3675074586384595, "grad_norm": 17.561562453493714, "learning_rate": 8.108346640612694e-06, "loss": 2.1868, "step": 19790 }, { "epoch": 5.370219690805533, "grad_norm": 14.538121711252645, "learning_rate": 8.105874102486679e-06, "loss": 2.1306, "step": 19800 }, { "epoch": 5.372931922972606, "grad_norm": 15.00929700433839, "learning_rate": 8.103400327050752e-06, "loss": 2.3086, "step": 19810 }, { "epoch": 5.37564415513968, "grad_norm": 14.92701458368847, "learning_rate": 8.10092531529041e-06, "loss": 2.2053, "step": 19820 }, { "epoch": 5.378356387306754, "grad_norm": 17.671643955631396, "learning_rate": 8.098449068191645e-06, "loss": 2.1875, "step": 19830 }, { "epoch": 5.381068619473827, "grad_norm": 17.3725926531094, "learning_rate": 8.095971586740935e-06, "loss": 2.1776, "step": 19840 }, { "epoch": 5.383780851640901, "grad_norm": 15.264437194607344, "learning_rate": 8.093492871925248e-06, "loss": 2.2902, "step": 19850 }, { "epoch": 5.386493083807974, "grad_norm": 17.898650835862508, "learning_rate": 8.091012924732056e-06, "loss": 2.2225, "step": 19860 }, { "epoch": 5.3892053159750475, "grad_norm": 15.15384870042739, "learning_rate": 8.088531746149308e-06, "loss": 2.3311, "step": 19870 }, { "epoch": 5.391917548142121, "grad_norm": 16.142306182162923, "learning_rate": 8.086049337165453e-06, "loss": 2.1499, "step": 19880 }, { "epoch": 5.394629780309194, "grad_norm": 16.37614800054793, "learning_rate": 8.083565698769424e-06, "loss": 2.0142, "step": 19890 }, { "epoch": 5.397342012476268, "grad_norm": 16.31718742486681, "learning_rate": 8.081080831950648e-06, "loss": 2.1053, "step": 19900 }, { "epoch": 5.400054244643341, "grad_norm": 15.268637654752341, "learning_rate": 8.078594737699038e-06, "loss": 2.1813, "step": 19910 }, { "epoch": 5.402766476810415, "grad_norm": 15.35018176625123, "learning_rate": 8.076107417005e-06, "loss": 2.196, "step": 19920 }, { "epoch": 5.405478708977489, "grad_norm": 17.029895203252206, "learning_rate": 8.073618870859424e-06, "loss": 2.2261, "step": 19930 }, { "epoch": 5.408190941144562, "grad_norm": 16.321241569055733, "learning_rate": 8.071129100253692e-06, "loss": 2.2161, "step": 19940 }, { "epoch": 5.4109031733116355, "grad_norm": 16.08057813381641, "learning_rate": 8.068638106179671e-06, "loss": 2.1294, "step": 19950 }, { "epoch": 5.413615405478709, "grad_norm": 15.68815229780837, "learning_rate": 8.066145889629718e-06, "loss": 2.0948, "step": 19960 }, { "epoch": 5.416327637645782, "grad_norm": 17.21120048010628, "learning_rate": 8.063652451596674e-06, "loss": 2.1721, "step": 19970 }, { "epoch": 5.419039869812856, "grad_norm": 16.45555992158113, "learning_rate": 8.061157793073868e-06, "loss": 2.1365, "step": 19980 }, { "epoch": 5.421752101979929, "grad_norm": 16.29802154380825, "learning_rate": 8.058661915055117e-06, "loss": 2.1298, "step": 19990 }, { "epoch": 5.424464334147003, "grad_norm": 15.748922395194874, "learning_rate": 8.056164818534723e-06, "loss": 2.2644, "step": 20000 }, { "epoch": 5.427176566314077, "grad_norm": 16.839266453318217, "learning_rate": 8.05366650450747e-06, "loss": 2.1847, "step": 20010 }, { "epoch": 5.42988879848115, "grad_norm": 16.173518686513816, "learning_rate": 8.051166973968631e-06, "loss": 2.2309, "step": 20020 }, { "epoch": 5.4326010306482235, "grad_norm": 15.43729268465214, "learning_rate": 8.048666227913964e-06, "loss": 2.2485, "step": 20030 }, { "epoch": 5.435313262815297, "grad_norm": 16.071225314987327, "learning_rate": 8.046164267339708e-06, "loss": 2.1846, "step": 20040 }, { "epoch": 5.43802549498237, "grad_norm": 16.848516825434746, "learning_rate": 8.043661093242587e-06, "loss": 2.1073, "step": 20050 }, { "epoch": 5.440737727149444, "grad_norm": 16.701194125857384, "learning_rate": 8.041156706619811e-06, "loss": 2.1304, "step": 20060 }, { "epoch": 5.443449959316517, "grad_norm": 16.541505741609683, "learning_rate": 8.038651108469066e-06, "loss": 2.2445, "step": 20070 }, { "epoch": 5.446162191483591, "grad_norm": 15.668967425353303, "learning_rate": 8.036144299788533e-06, "loss": 2.1742, "step": 20080 }, { "epoch": 5.448874423650665, "grad_norm": 16.60262500682494, "learning_rate": 8.03363628157686e-06, "loss": 2.0568, "step": 20090 }, { "epoch": 5.451586655817738, "grad_norm": 18.972475872985193, "learning_rate": 8.031127054833192e-06, "loss": 2.2315, "step": 20100 }, { "epoch": 5.4542988879848116, "grad_norm": 16.738272676346988, "learning_rate": 8.028616620557142e-06, "loss": 2.2986, "step": 20110 }, { "epoch": 5.457011120151885, "grad_norm": 16.25147505222318, "learning_rate": 8.026104979748814e-06, "loss": 2.0782, "step": 20120 }, { "epoch": 5.459723352318958, "grad_norm": 16.686614722182252, "learning_rate": 8.023592133408786e-06, "loss": 2.3051, "step": 20130 }, { "epoch": 5.462435584486032, "grad_norm": 19.5372555371867, "learning_rate": 8.021078082538123e-06, "loss": 2.2422, "step": 20140 }, { "epoch": 5.465147816653105, "grad_norm": 16.834205332272205, "learning_rate": 8.01856282813836e-06, "loss": 2.1325, "step": 20150 }, { "epoch": 5.467860048820179, "grad_norm": 15.681691445974009, "learning_rate": 8.016046371211525e-06, "loss": 2.2561, "step": 20160 }, { "epoch": 5.470572280987253, "grad_norm": 15.465386003940761, "learning_rate": 8.01352871276011e-06, "loss": 2.2946, "step": 20170 }, { "epoch": 5.473284513154326, "grad_norm": 14.403694800766393, "learning_rate": 8.0110098537871e-06, "loss": 2.1094, "step": 20180 }, { "epoch": 5.4759967453214, "grad_norm": 17.011955644904052, "learning_rate": 8.008489795295946e-06, "loss": 2.0368, "step": 20190 }, { "epoch": 5.478708977488473, "grad_norm": 16.28785224472087, "learning_rate": 8.005968538290585e-06, "loss": 2.1833, "step": 20200 }, { "epoch": 5.481421209655546, "grad_norm": 15.435394287767373, "learning_rate": 8.003446083775426e-06, "loss": 2.1275, "step": 20210 }, { "epoch": 5.48413344182262, "grad_norm": 17.443852723781134, "learning_rate": 8.00092243275536e-06, "loss": 2.2551, "step": 20220 }, { "epoch": 5.486845673989693, "grad_norm": 17.098436567101654, "learning_rate": 7.998397586235753e-06, "loss": 2.2016, "step": 20230 }, { "epoch": 5.489557906156767, "grad_norm": 19.185876161375386, "learning_rate": 7.995871545222446e-06, "loss": 2.1891, "step": 20240 }, { "epoch": 5.492270138323841, "grad_norm": 15.274844789877553, "learning_rate": 7.993344310721755e-06, "loss": 2.18, "step": 20250 }, { "epoch": 5.494982370490914, "grad_norm": 17.99177320180988, "learning_rate": 7.990815883740476e-06, "loss": 2.2182, "step": 20260 }, { "epoch": 5.497694602657988, "grad_norm": 16.224479881361297, "learning_rate": 7.988286265285873e-06, "loss": 2.1596, "step": 20270 }, { "epoch": 5.5004068348250605, "grad_norm": 17.484596470341252, "learning_rate": 7.985755456365693e-06, "loss": 2.2076, "step": 20280 }, { "epoch": 5.503119066992134, "grad_norm": 16.322162093991402, "learning_rate": 7.98322345798815e-06, "loss": 2.1857, "step": 20290 }, { "epoch": 5.505831299159208, "grad_norm": 16.045975593705833, "learning_rate": 7.980690271161937e-06, "loss": 2.1767, "step": 20300 }, { "epoch": 5.508543531326281, "grad_norm": 16.376096926327172, "learning_rate": 7.978155896896216e-06, "loss": 2.2625, "step": 20310 }, { "epoch": 5.511255763493355, "grad_norm": 16.281152546218514, "learning_rate": 7.975620336200624e-06, "loss": 2.2028, "step": 20320 }, { "epoch": 5.513967995660429, "grad_norm": 15.718556212880236, "learning_rate": 7.973083590085272e-06, "loss": 2.0961, "step": 20330 }, { "epoch": 5.516680227827502, "grad_norm": 16.26791814510362, "learning_rate": 7.970545659560747e-06, "loss": 2.2257, "step": 20340 }, { "epoch": 5.519392459994576, "grad_norm": 15.247843628074397, "learning_rate": 7.968006545638095e-06, "loss": 2.2515, "step": 20350 }, { "epoch": 5.5221046921616495, "grad_norm": 16.06062766466766, "learning_rate": 7.965466249328846e-06, "loss": 2.2617, "step": 20360 }, { "epoch": 5.524816924328722, "grad_norm": 16.309323620413377, "learning_rate": 7.962924771644996e-06, "loss": 2.2472, "step": 20370 }, { "epoch": 5.527529156495796, "grad_norm": 16.18920051902327, "learning_rate": 7.96038211359901e-06, "loss": 2.2227, "step": 20380 }, { "epoch": 5.530241388662869, "grad_norm": 17.18374244865039, "learning_rate": 7.957838276203828e-06, "loss": 2.1012, "step": 20390 }, { "epoch": 5.532953620829943, "grad_norm": 15.251418042328403, "learning_rate": 7.955293260472859e-06, "loss": 2.1347, "step": 20400 }, { "epoch": 5.535665852997017, "grad_norm": 16.50704812489312, "learning_rate": 7.952747067419973e-06, "loss": 2.2308, "step": 20410 }, { "epoch": 5.53837808516409, "grad_norm": 16.46715614320502, "learning_rate": 7.950199698059523e-06, "loss": 2.1622, "step": 20420 }, { "epoch": 5.541090317331164, "grad_norm": 15.794578039681037, "learning_rate": 7.947651153406318e-06, "loss": 2.1226, "step": 20430 }, { "epoch": 5.543802549498237, "grad_norm": 17.467342384848383, "learning_rate": 7.945101434475642e-06, "loss": 2.1772, "step": 20440 }, { "epoch": 5.54651478166531, "grad_norm": 16.274518340853646, "learning_rate": 7.942550542283245e-06, "loss": 2.2428, "step": 20450 }, { "epoch": 5.549227013832384, "grad_norm": 15.823926538123283, "learning_rate": 7.939998477845348e-06, "loss": 2.2596, "step": 20460 }, { "epoch": 5.551939245999457, "grad_norm": 15.043205286778049, "learning_rate": 7.937445242178632e-06, "loss": 2.3016, "step": 20470 }, { "epoch": 5.554651478166531, "grad_norm": 17.778149694130107, "learning_rate": 7.934890836300249e-06, "loss": 2.3002, "step": 20480 }, { "epoch": 5.557363710333605, "grad_norm": 15.713134653555121, "learning_rate": 7.932335261227816e-06, "loss": 2.193, "step": 20490 }, { "epoch": 5.560075942500678, "grad_norm": 16.497242032379862, "learning_rate": 7.929778517979419e-06, "loss": 2.1372, "step": 20500 }, { "epoch": 5.562788174667752, "grad_norm": 19.307545455473974, "learning_rate": 7.927220607573603e-06, "loss": 2.285, "step": 20510 }, { "epoch": 5.5655004068348255, "grad_norm": 15.60329117716479, "learning_rate": 7.924661531029384e-06, "loss": 2.294, "step": 20520 }, { "epoch": 5.5682126390018984, "grad_norm": 16.788791192298408, "learning_rate": 7.922101289366237e-06, "loss": 2.2141, "step": 20530 }, { "epoch": 5.570924871168972, "grad_norm": 15.352328263421315, "learning_rate": 7.91953988360411e-06, "loss": 2.242, "step": 20540 }, { "epoch": 5.573637103336045, "grad_norm": 16.6890501178051, "learning_rate": 7.916977314763404e-06, "loss": 2.1446, "step": 20550 }, { "epoch": 5.576349335503119, "grad_norm": 17.25047965355171, "learning_rate": 7.91441358386499e-06, "loss": 2.1628, "step": 20560 }, { "epoch": 5.579061567670193, "grad_norm": 17.366685903090445, "learning_rate": 7.911848691930202e-06, "loss": 2.2199, "step": 20570 }, { "epoch": 5.581773799837266, "grad_norm": 15.149427901344525, "learning_rate": 7.909282639980832e-06, "loss": 2.2123, "step": 20580 }, { "epoch": 5.58448603200434, "grad_norm": 14.965127335573252, "learning_rate": 7.906715429039138e-06, "loss": 2.1789, "step": 20590 }, { "epoch": 5.587198264171413, "grad_norm": 15.903103284139386, "learning_rate": 7.90414706012784e-06, "loss": 2.256, "step": 20600 }, { "epoch": 5.5899104963384865, "grad_norm": 17.00594046682566, "learning_rate": 7.901577534270117e-06, "loss": 2.109, "step": 20610 }, { "epoch": 5.59262272850556, "grad_norm": 17.160088640887107, "learning_rate": 7.89900685248961e-06, "loss": 2.1318, "step": 20620 }, { "epoch": 5.595334960672633, "grad_norm": 16.072898648037537, "learning_rate": 7.896435015810419e-06, "loss": 2.3814, "step": 20630 }, { "epoch": 5.598047192839707, "grad_norm": 17.48664222225543, "learning_rate": 7.893862025257108e-06, "loss": 2.0463, "step": 20640 }, { "epoch": 5.600759425006781, "grad_norm": 15.560931152629117, "learning_rate": 7.891287881854697e-06, "loss": 2.2711, "step": 20650 }, { "epoch": 5.603471657173854, "grad_norm": 17.372199777560006, "learning_rate": 7.888712586628665e-06, "loss": 2.1907, "step": 20660 }, { "epoch": 5.606183889340928, "grad_norm": 15.220713295710134, "learning_rate": 7.886136140604954e-06, "loss": 2.2552, "step": 20670 }, { "epoch": 5.6088961215080015, "grad_norm": 16.47440609665633, "learning_rate": 7.883558544809962e-06, "loss": 2.1782, "step": 20680 }, { "epoch": 5.6116083536750745, "grad_norm": 15.320514073278133, "learning_rate": 7.88097980027054e-06, "loss": 2.4577, "step": 20690 }, { "epoch": 5.614320585842148, "grad_norm": 14.773234536914938, "learning_rate": 7.878399908014006e-06, "loss": 2.2003, "step": 20700 }, { "epoch": 5.617032818009221, "grad_norm": 16.579056260618565, "learning_rate": 7.875818869068131e-06, "loss": 2.194, "step": 20710 }, { "epoch": 5.619745050176295, "grad_norm": 15.930634277660664, "learning_rate": 7.873236684461139e-06, "loss": 2.2953, "step": 20720 }, { "epoch": 5.622457282343369, "grad_norm": 16.095963516024785, "learning_rate": 7.870653355221715e-06, "loss": 2.2462, "step": 20730 }, { "epoch": 5.625169514510442, "grad_norm": 16.863023198703257, "learning_rate": 7.868068882379003e-06, "loss": 2.1714, "step": 20740 }, { "epoch": 5.627881746677516, "grad_norm": 19.618947388995906, "learning_rate": 7.865483266962592e-06, "loss": 2.2849, "step": 20750 }, { "epoch": 5.630593978844589, "grad_norm": 14.410354665255712, "learning_rate": 7.862896510002539e-06, "loss": 2.2205, "step": 20760 }, { "epoch": 5.6333062110116625, "grad_norm": 18.41453215210802, "learning_rate": 7.860308612529342e-06, "loss": 2.2058, "step": 20770 }, { "epoch": 5.636018443178736, "grad_norm": 16.84253678597235, "learning_rate": 7.857719575573966e-06, "loss": 2.091, "step": 20780 }, { "epoch": 5.638730675345809, "grad_norm": 17.553880752092606, "learning_rate": 7.855129400167825e-06, "loss": 2.1488, "step": 20790 }, { "epoch": 5.641442907512883, "grad_norm": 16.987740147240718, "learning_rate": 7.852538087342785e-06, "loss": 2.2363, "step": 20800 }, { "epoch": 5.644155139679957, "grad_norm": 16.48127077636474, "learning_rate": 7.849945638131164e-06, "loss": 2.2921, "step": 20810 }, { "epoch": 5.64686737184703, "grad_norm": 17.829900137528078, "learning_rate": 7.847352053565738e-06, "loss": 2.2142, "step": 20820 }, { "epoch": 5.649579604014104, "grad_norm": 16.47456024803584, "learning_rate": 7.844757334679732e-06, "loss": 2.1684, "step": 20830 }, { "epoch": 5.652291836181178, "grad_norm": 16.647366900902494, "learning_rate": 7.842161482506822e-06, "loss": 2.2279, "step": 20840 }, { "epoch": 5.6550040683482505, "grad_norm": 17.156029381768725, "learning_rate": 7.839564498081138e-06, "loss": 2.1315, "step": 20850 }, { "epoch": 5.657716300515324, "grad_norm": 16.89595907867913, "learning_rate": 7.83696638243726e-06, "loss": 2.1986, "step": 20860 }, { "epoch": 5.660428532682397, "grad_norm": 16.3049506200326, "learning_rate": 7.834367136610216e-06, "loss": 2.2747, "step": 20870 }, { "epoch": 5.663140764849471, "grad_norm": 16.381357363237708, "learning_rate": 7.83176676163549e-06, "loss": 2.2293, "step": 20880 }, { "epoch": 5.665852997016545, "grad_norm": 16.590467934128434, "learning_rate": 7.82916525854901e-06, "loss": 2.2507, "step": 20890 }, { "epoch": 5.668565229183618, "grad_norm": 15.211384346084758, "learning_rate": 7.826562628387157e-06, "loss": 2.1634, "step": 20900 }, { "epoch": 5.671277461350692, "grad_norm": 19.258568783085067, "learning_rate": 7.823958872186758e-06, "loss": 2.1546, "step": 20910 }, { "epoch": 5.673989693517765, "grad_norm": 16.254900982391685, "learning_rate": 7.821353990985096e-06, "loss": 2.3527, "step": 20920 }, { "epoch": 5.676701925684839, "grad_norm": 15.574181696935819, "learning_rate": 7.81874798581989e-06, "loss": 2.2165, "step": 20930 }, { "epoch": 5.679414157851912, "grad_norm": 17.6663545636317, "learning_rate": 7.816140857729317e-06, "loss": 2.155, "step": 20940 }, { "epoch": 5.682126390018985, "grad_norm": 14.859021887499562, "learning_rate": 7.813532607751998e-06, "loss": 2.1801, "step": 20950 }, { "epoch": 5.684838622186059, "grad_norm": 15.600672666307071, "learning_rate": 7.810923236926999e-06, "loss": 2.1809, "step": 20960 }, { "epoch": 5.687550854353133, "grad_norm": 15.31909261688448, "learning_rate": 7.808312746293835e-06, "loss": 2.2755, "step": 20970 }, { "epoch": 5.690263086520206, "grad_norm": 17.339494236967898, "learning_rate": 7.805701136892467e-06, "loss": 2.079, "step": 20980 }, { "epoch": 5.69297531868728, "grad_norm": 17.022298858428062, "learning_rate": 7.803088409763302e-06, "loss": 2.2895, "step": 20990 }, { "epoch": 5.695687550854354, "grad_norm": 15.150637923918989, "learning_rate": 7.800474565947188e-06, "loss": 2.3131, "step": 21000 }, { "epoch": 5.698399783021427, "grad_norm": 16.206505075141386, "learning_rate": 7.797859606485422e-06, "loss": 2.2718, "step": 21010 }, { "epoch": 5.7011120151885, "grad_norm": 16.730795700577605, "learning_rate": 7.79524353241975e-06, "loss": 2.1914, "step": 21020 }, { "epoch": 5.703824247355573, "grad_norm": 16.657075246898497, "learning_rate": 7.792626344792347e-06, "loss": 2.2014, "step": 21030 }, { "epoch": 5.706536479522647, "grad_norm": 15.8271096261199, "learning_rate": 7.79000804464585e-06, "loss": 2.2114, "step": 21040 }, { "epoch": 5.709248711689721, "grad_norm": 15.23839746952993, "learning_rate": 7.787388633023327e-06, "loss": 2.2844, "step": 21050 }, { "epoch": 5.711960943856794, "grad_norm": 16.221319645921223, "learning_rate": 7.784768110968292e-06, "loss": 2.2162, "step": 21060 }, { "epoch": 5.714673176023868, "grad_norm": 14.867241542432692, "learning_rate": 7.782146479524701e-06, "loss": 2.2396, "step": 21070 }, { "epoch": 5.717385408190941, "grad_norm": 15.540602544093382, "learning_rate": 7.779523739736955e-06, "loss": 2.2228, "step": 21080 }, { "epoch": 5.720097640358015, "grad_norm": 16.553336969764153, "learning_rate": 7.776899892649892e-06, "loss": 2.3026, "step": 21090 }, { "epoch": 5.722809872525088, "grad_norm": 16.995587156836308, "learning_rate": 7.774274939308794e-06, "loss": 2.234, "step": 21100 }, { "epoch": 5.725522104692161, "grad_norm": 15.779949068198244, "learning_rate": 7.771648880759383e-06, "loss": 2.1959, "step": 21110 }, { "epoch": 5.728234336859235, "grad_norm": 15.516799610523027, "learning_rate": 7.769021718047823e-06, "loss": 2.1844, "step": 21120 }, { "epoch": 5.730946569026308, "grad_norm": 16.89325929968585, "learning_rate": 7.766393452220712e-06, "loss": 2.2341, "step": 21130 }, { "epoch": 5.733658801193382, "grad_norm": 15.597733600052484, "learning_rate": 7.763764084325096e-06, "loss": 2.232, "step": 21140 }, { "epoch": 5.736371033360456, "grad_norm": 14.46478129800222, "learning_rate": 7.761133615408452e-06, "loss": 2.0507, "step": 21150 }, { "epoch": 5.739083265527529, "grad_norm": 17.790009777607498, "learning_rate": 7.758502046518701e-06, "loss": 2.2025, "step": 21160 }, { "epoch": 5.741795497694603, "grad_norm": 16.076247369999415, "learning_rate": 7.755869378704202e-06, "loss": 2.0367, "step": 21170 }, { "epoch": 5.7445077298616765, "grad_norm": 16.621092119754778, "learning_rate": 7.753235613013747e-06, "loss": 2.2128, "step": 21180 }, { "epoch": 5.747219962028749, "grad_norm": 17.892248870670315, "learning_rate": 7.75060075049657e-06, "loss": 2.2711, "step": 21190 }, { "epoch": 5.749932194195823, "grad_norm": 15.15622747452578, "learning_rate": 7.747964792202342e-06, "loss": 2.2847, "step": 21200 }, { "epoch": 5.752644426362897, "grad_norm": 15.481037723858577, "learning_rate": 7.745327739181166e-06, "loss": 2.2363, "step": 21210 }, { "epoch": 5.75535665852997, "grad_norm": 16.232683616828936, "learning_rate": 7.742689592483588e-06, "loss": 2.32, "step": 21220 }, { "epoch": 5.758068890697044, "grad_norm": 16.056305853000495, "learning_rate": 7.740050353160583e-06, "loss": 2.2132, "step": 21230 }, { "epoch": 5.760781122864117, "grad_norm": 17.255695418193763, "learning_rate": 7.737410022263567e-06, "loss": 2.2647, "step": 21240 }, { "epoch": 5.763493355031191, "grad_norm": 16.311501958710465, "learning_rate": 7.734768600844383e-06, "loss": 2.3291, "step": 21250 }, { "epoch": 5.7662055871982645, "grad_norm": 15.992758838300785, "learning_rate": 7.732126089955319e-06, "loss": 2.3509, "step": 21260 }, { "epoch": 5.768917819365337, "grad_norm": 15.445491517071932, "learning_rate": 7.729482490649087e-06, "loss": 2.1573, "step": 21270 }, { "epoch": 5.771630051532411, "grad_norm": 15.203702387886294, "learning_rate": 7.726837803978842e-06, "loss": 2.196, "step": 21280 }, { "epoch": 5.774342283699484, "grad_norm": 17.899746581214913, "learning_rate": 7.724192030998163e-06, "loss": 2.2874, "step": 21290 }, { "epoch": 5.777054515866558, "grad_norm": 18.246599541602034, "learning_rate": 7.721545172761067e-06, "loss": 2.2761, "step": 21300 }, { "epoch": 5.779766748033632, "grad_norm": 15.228111015345563, "learning_rate": 7.718897230322004e-06, "loss": 2.2538, "step": 21310 }, { "epoch": 5.782478980200705, "grad_norm": 15.96216006455379, "learning_rate": 7.716248204735853e-06, "loss": 2.3167, "step": 21320 }, { "epoch": 5.785191212367779, "grad_norm": 16.701993419645888, "learning_rate": 7.713598097057926e-06, "loss": 2.3746, "step": 21330 }, { "epoch": 5.7879034445348525, "grad_norm": 16.13960051254094, "learning_rate": 7.710946908343967e-06, "loss": 2.2858, "step": 21340 }, { "epoch": 5.7906156767019255, "grad_norm": 16.196899939358058, "learning_rate": 7.708294639650146e-06, "loss": 2.2811, "step": 21350 }, { "epoch": 5.793327908868999, "grad_norm": 15.898577615810082, "learning_rate": 7.705641292033072e-06, "loss": 2.1774, "step": 21360 }, { "epoch": 5.796040141036073, "grad_norm": 17.918580795619977, "learning_rate": 7.702986866549775e-06, "loss": 2.2528, "step": 21370 }, { "epoch": 5.798752373203146, "grad_norm": 16.290178639512277, "learning_rate": 7.700331364257718e-06, "loss": 2.2011, "step": 21380 }, { "epoch": 5.80146460537022, "grad_norm": 16.92431352283922, "learning_rate": 7.697674786214796e-06, "loss": 2.3081, "step": 21390 }, { "epoch": 5.804176837537293, "grad_norm": 15.715570591036402, "learning_rate": 7.695017133479326e-06, "loss": 2.2638, "step": 21400 }, { "epoch": 5.806889069704367, "grad_norm": 16.575734726490243, "learning_rate": 7.692358407110057e-06, "loss": 2.3996, "step": 21410 }, { "epoch": 5.8096013018714405, "grad_norm": 15.453997016325964, "learning_rate": 7.689698608166168e-06, "loss": 2.2781, "step": 21420 }, { "epoch": 5.8123135340385135, "grad_norm": 16.325368593406388, "learning_rate": 7.68703773770726e-06, "loss": 2.18, "step": 21430 }, { "epoch": 5.815025766205587, "grad_norm": 15.710262662758929, "learning_rate": 7.684375796793365e-06, "loss": 2.2985, "step": 21440 }, { "epoch": 5.81773799837266, "grad_norm": 18.825235446790266, "learning_rate": 7.68171278648494e-06, "loss": 2.3221, "step": 21450 }, { "epoch": 5.820450230539734, "grad_norm": 17.294384459110432, "learning_rate": 7.67904870784287e-06, "loss": 2.3236, "step": 21460 }, { "epoch": 5.823162462706808, "grad_norm": 15.984390081490194, "learning_rate": 7.676383561928457e-06, "loss": 2.1469, "step": 21470 }, { "epoch": 5.825874694873881, "grad_norm": 17.516167374301663, "learning_rate": 7.673717349803442e-06, "loss": 2.1113, "step": 21480 }, { "epoch": 5.828586927040955, "grad_norm": 15.469557116208575, "learning_rate": 7.671050072529978e-06, "loss": 2.1559, "step": 21490 }, { "epoch": 5.8312991592080285, "grad_norm": 16.621409651498904, "learning_rate": 7.668381731170653e-06, "loss": 2.3777, "step": 21500 }, { "epoch": 5.8340113913751015, "grad_norm": 16.881377307585147, "learning_rate": 7.665712326788468e-06, "loss": 2.2218, "step": 21510 }, { "epoch": 5.836723623542175, "grad_norm": 17.56072687164579, "learning_rate": 7.66304186044686e-06, "loss": 2.2425, "step": 21520 }, { "epoch": 5.839435855709249, "grad_norm": 17.10183767336952, "learning_rate": 7.660370333209678e-06, "loss": 2.343, "step": 21530 }, { "epoch": 5.842148087876322, "grad_norm": 16.749208811797804, "learning_rate": 7.6576977461412e-06, "loss": 2.1415, "step": 21540 }, { "epoch": 5.844860320043396, "grad_norm": 15.417283596686653, "learning_rate": 7.65502410030612e-06, "loss": 2.2178, "step": 21550 }, { "epoch": 5.847572552210469, "grad_norm": 18.119439642306293, "learning_rate": 7.652349396769562e-06, "loss": 2.3364, "step": 21560 }, { "epoch": 5.850284784377543, "grad_norm": 17.265533745766028, "learning_rate": 7.649673636597066e-06, "loss": 2.3878, "step": 21570 }, { "epoch": 5.852997016544617, "grad_norm": 15.46194225284409, "learning_rate": 7.646996820854595e-06, "loss": 2.242, "step": 21580 }, { "epoch": 5.8557092487116895, "grad_norm": 15.152205595702073, "learning_rate": 7.64431895060853e-06, "loss": 2.1811, "step": 21590 }, { "epoch": 5.858421480878763, "grad_norm": 17.167819525474513, "learning_rate": 7.641640026925676e-06, "loss": 2.2004, "step": 21600 }, { "epoch": 5.861133713045836, "grad_norm": 16.425706096791952, "learning_rate": 7.638960050873256e-06, "loss": 2.2783, "step": 21610 }, { "epoch": 5.86384594521291, "grad_norm": 15.76413864312519, "learning_rate": 7.636279023518908e-06, "loss": 2.2027, "step": 21620 }, { "epoch": 5.866558177379984, "grad_norm": 15.35266612637476, "learning_rate": 7.633596945930696e-06, "loss": 2.1782, "step": 21630 }, { "epoch": 5.869270409547057, "grad_norm": 16.335293912449824, "learning_rate": 7.630913819177099e-06, "loss": 2.2299, "step": 21640 }, { "epoch": 5.871982641714131, "grad_norm": 16.452300923952304, "learning_rate": 7.628229644327011e-06, "loss": 2.1842, "step": 21650 }, { "epoch": 5.874694873881205, "grad_norm": 14.551319611679013, "learning_rate": 7.62554442244975e-06, "loss": 2.1442, "step": 21660 }, { "epoch": 5.8774071060482775, "grad_norm": 15.556059635217226, "learning_rate": 7.622858154615045e-06, "loss": 2.2424, "step": 21670 }, { "epoch": 5.880119338215351, "grad_norm": 16.342891155678885, "learning_rate": 7.6201708418930445e-06, "loss": 2.3946, "step": 21680 }, { "epoch": 5.882831570382425, "grad_norm": 16.555078961491507, "learning_rate": 7.617482485354315e-06, "loss": 2.2454, "step": 21690 }, { "epoch": 5.885543802549498, "grad_norm": 16.454843980614992, "learning_rate": 7.614793086069836e-06, "loss": 2.3455, "step": 21700 }, { "epoch": 5.888256034716572, "grad_norm": 17.82552803330702, "learning_rate": 7.612102645111002e-06, "loss": 2.2985, "step": 21710 }, { "epoch": 5.890968266883645, "grad_norm": 16.75212087928138, "learning_rate": 7.609411163549626e-06, "loss": 2.1844, "step": 21720 }, { "epoch": 5.893680499050719, "grad_norm": 18.04998123900337, "learning_rate": 7.606718642457931e-06, "loss": 2.2402, "step": 21730 }, { "epoch": 5.896392731217793, "grad_norm": 15.653328282086651, "learning_rate": 7.604025082908557e-06, "loss": 2.3354, "step": 21740 }, { "epoch": 5.899104963384866, "grad_norm": 16.949546110926825, "learning_rate": 7.601330485974557e-06, "loss": 2.2804, "step": 21750 }, { "epoch": 5.901817195551939, "grad_norm": 16.14690296508219, "learning_rate": 7.5986348527294e-06, "loss": 2.2801, "step": 21760 }, { "epoch": 5.904529427719012, "grad_norm": 15.093214332438354, "learning_rate": 7.595938184246962e-06, "loss": 2.3553, "step": 21770 }, { "epoch": 5.907241659886086, "grad_norm": 18.87580048723903, "learning_rate": 7.593240481601536e-06, "loss": 2.4192, "step": 21780 }, { "epoch": 5.90995389205316, "grad_norm": 15.346272865520701, "learning_rate": 7.590541745867826e-06, "loss": 2.212, "step": 21790 }, { "epoch": 5.912666124220233, "grad_norm": 17.544775674708603, "learning_rate": 7.5878419781209445e-06, "loss": 2.2633, "step": 21800 }, { "epoch": 5.915378356387307, "grad_norm": 16.82044577322912, "learning_rate": 7.585141179436422e-06, "loss": 2.2589, "step": 21810 }, { "epoch": 5.918090588554381, "grad_norm": 16.392981091830706, "learning_rate": 7.582439350890194e-06, "loss": 2.3418, "step": 21820 }, { "epoch": 5.920802820721454, "grad_norm": 14.823138983367016, "learning_rate": 7.579736493558606e-06, "loss": 2.2867, "step": 21830 }, { "epoch": 5.923515052888527, "grad_norm": 14.60651084765073, "learning_rate": 7.577032608518417e-06, "loss": 2.1632, "step": 21840 }, { "epoch": 5.926227285055601, "grad_norm": 16.61322002853419, "learning_rate": 7.574327696846794e-06, "loss": 2.319, "step": 21850 }, { "epoch": 5.928939517222674, "grad_norm": 16.232869747780352, "learning_rate": 7.57162175962131e-06, "loss": 2.3681, "step": 21860 }, { "epoch": 5.931651749389748, "grad_norm": 15.308723563254361, "learning_rate": 7.568914797919952e-06, "loss": 2.3693, "step": 21870 }, { "epoch": 5.934363981556821, "grad_norm": 16.026355592133868, "learning_rate": 7.566206812821111e-06, "loss": 2.1531, "step": 21880 }, { "epoch": 5.937076213723895, "grad_norm": 17.567619580050525, "learning_rate": 7.563497805403587e-06, "loss": 2.2093, "step": 21890 }, { "epoch": 5.939788445890969, "grad_norm": 15.939542815111928, "learning_rate": 7.560787776746588e-06, "loss": 2.241, "step": 21900 }, { "epoch": 5.942500678058042, "grad_norm": 14.93670995410711, "learning_rate": 7.558076727929724e-06, "loss": 2.3657, "step": 21910 }, { "epoch": 5.945212910225115, "grad_norm": 17.727799135388246, "learning_rate": 7.555364660033021e-06, "loss": 2.2284, "step": 21920 }, { "epoch": 5.947925142392188, "grad_norm": 16.803696785289702, "learning_rate": 7.552651574136902e-06, "loss": 2.1376, "step": 21930 }, { "epoch": 5.950637374559262, "grad_norm": 16.39985099907759, "learning_rate": 7.549937471322202e-06, "loss": 2.2114, "step": 21940 }, { "epoch": 5.953349606726336, "grad_norm": 18.103511271314733, "learning_rate": 7.547222352670155e-06, "loss": 2.1693, "step": 21950 }, { "epoch": 5.956061838893409, "grad_norm": 16.122905001664535, "learning_rate": 7.544506219262404e-06, "loss": 2.1976, "step": 21960 }, { "epoch": 5.958774071060483, "grad_norm": 16.090418570733103, "learning_rate": 7.541789072180996e-06, "loss": 2.3143, "step": 21970 }, { "epoch": 5.961486303227556, "grad_norm": 16.177647689493433, "learning_rate": 7.53907091250838e-06, "loss": 2.0788, "step": 21980 }, { "epoch": 5.96419853539463, "grad_norm": 16.892098887521936, "learning_rate": 7.536351741327406e-06, "loss": 2.2579, "step": 21990 }, { "epoch": 5.9669107675617035, "grad_norm": 17.194105285137898, "learning_rate": 7.5336315597213375e-06, "loss": 2.266, "step": 22000 }, { "epoch": 5.969622999728776, "grad_norm": 18.023950943184147, "learning_rate": 7.530910368773826e-06, "loss": 2.3401, "step": 22010 }, { "epoch": 5.97233523189585, "grad_norm": 17.705156361217487, "learning_rate": 7.5281881695689376e-06, "loss": 2.2635, "step": 22020 }, { "epoch": 5.975047464062924, "grad_norm": 17.5098642094785, "learning_rate": 7.5254649631911316e-06, "loss": 2.1892, "step": 22030 }, { "epoch": 5.977759696229997, "grad_norm": 16.471017249072663, "learning_rate": 7.5227407507252734e-06, "loss": 2.3034, "step": 22040 }, { "epoch": 5.980471928397071, "grad_norm": 15.760443296722736, "learning_rate": 7.520015533256626e-06, "loss": 2.2163, "step": 22050 }, { "epoch": 5.983184160564145, "grad_norm": 14.921493096299022, "learning_rate": 7.5172893118708565e-06, "loss": 2.2855, "step": 22060 }, { "epoch": 5.985896392731218, "grad_norm": 15.409266663157927, "learning_rate": 7.514562087654027e-06, "loss": 2.4176, "step": 22070 }, { "epoch": 5.9886086248982915, "grad_norm": 15.568258098435807, "learning_rate": 7.511833861692606e-06, "loss": 2.2502, "step": 22080 }, { "epoch": 5.991320857065364, "grad_norm": 16.327735634939494, "learning_rate": 7.509104635073452e-06, "loss": 2.333, "step": 22090 }, { "epoch": 5.994033089232438, "grad_norm": 15.294054967042953, "learning_rate": 7.506374408883832e-06, "loss": 2.2476, "step": 22100 }, { "epoch": 5.996745321399512, "grad_norm": 16.76109998660596, "learning_rate": 7.503643184211402e-06, "loss": 2.2234, "step": 22110 }, { "epoch": 5.999457553566585, "grad_norm": 15.88273960876096, "learning_rate": 7.500910962144222e-06, "loss": 2.2888, "step": 22120 }, { "epoch": 6.002169785733659, "grad_norm": 16.616101446052465, "learning_rate": 7.498177743770747e-06, "loss": 1.8312, "step": 22130 }, { "epoch": 6.004882017900733, "grad_norm": 18.755387683242567, "learning_rate": 7.49544353017983e-06, "loss": 1.6425, "step": 22140 }, { "epoch": 6.007594250067806, "grad_norm": 15.511278845285528, "learning_rate": 7.49270832246072e-06, "loss": 1.7212, "step": 22150 }, { "epoch": 6.0103064822348795, "grad_norm": 16.4484142507517, "learning_rate": 7.48997212170306e-06, "loss": 1.5719, "step": 22160 }, { "epoch": 6.0130187144019525, "grad_norm": 17.31633350644476, "learning_rate": 7.487234928996894e-06, "loss": 1.695, "step": 22170 }, { "epoch": 6.015730946569026, "grad_norm": 17.335102631195795, "learning_rate": 7.484496745432652e-06, "loss": 1.6239, "step": 22180 }, { "epoch": 6.0184431787361, "grad_norm": 17.878597454745332, "learning_rate": 7.4817575721011695e-06, "loss": 1.7511, "step": 22190 }, { "epoch": 6.021155410903173, "grad_norm": 16.62697604914109, "learning_rate": 7.479017410093668e-06, "loss": 1.6458, "step": 22200 }, { "epoch": 6.023867643070247, "grad_norm": 16.59368206997119, "learning_rate": 7.476276260501767e-06, "loss": 1.7097, "step": 22210 }, { "epoch": 6.026579875237321, "grad_norm": 16.458720983147117, "learning_rate": 7.47353412441748e-06, "loss": 1.6421, "step": 22220 }, { "epoch": 6.029292107404394, "grad_norm": 20.715845008650593, "learning_rate": 7.47079100293321e-06, "loss": 1.7546, "step": 22230 }, { "epoch": 6.0320043395714675, "grad_norm": 16.46699881772929, "learning_rate": 7.468046897141754e-06, "loss": 1.7129, "step": 22240 }, { "epoch": 6.0347165717385405, "grad_norm": 17.812317487057676, "learning_rate": 7.465301808136303e-06, "loss": 1.7448, "step": 22250 }, { "epoch": 6.037428803905614, "grad_norm": 18.62017237441634, "learning_rate": 7.462555737010437e-06, "loss": 1.7074, "step": 22260 }, { "epoch": 6.040141036072688, "grad_norm": 17.930528833502752, "learning_rate": 7.459808684858132e-06, "loss": 1.7486, "step": 22270 }, { "epoch": 6.042853268239761, "grad_norm": 18.935721642357258, "learning_rate": 7.457060652773747e-06, "loss": 1.6847, "step": 22280 }, { "epoch": 6.045565500406835, "grad_norm": 16.891702340314747, "learning_rate": 7.454311641852038e-06, "loss": 1.6471, "step": 22290 }, { "epoch": 6.048277732573908, "grad_norm": 15.834611071239618, "learning_rate": 7.451561653188148e-06, "loss": 1.7751, "step": 22300 }, { "epoch": 6.050989964740982, "grad_norm": 16.489072008422927, "learning_rate": 7.448810687877609e-06, "loss": 1.5871, "step": 22310 }, { "epoch": 6.0537021969080556, "grad_norm": 20.1458615127966, "learning_rate": 7.446058747016346e-06, "loss": 1.6824, "step": 22320 }, { "epoch": 6.0564144290751285, "grad_norm": 17.821882920330673, "learning_rate": 7.443305831700669e-06, "loss": 1.602, "step": 22330 }, { "epoch": 6.059126661242202, "grad_norm": 20.216751177502044, "learning_rate": 7.440551943027276e-06, "loss": 1.7025, "step": 22340 }, { "epoch": 6.061838893409276, "grad_norm": 16.196464345104516, "learning_rate": 7.437797082093256e-06, "loss": 1.5733, "step": 22350 }, { "epoch": 6.064551125576349, "grad_norm": 19.34974585373455, "learning_rate": 7.435041249996081e-06, "loss": 1.746, "step": 22360 }, { "epoch": 6.067263357743423, "grad_norm": 17.401436782951308, "learning_rate": 7.432284447833613e-06, "loss": 1.6677, "step": 22370 }, { "epoch": 6.069975589910496, "grad_norm": 16.453692075041488, "learning_rate": 7.4295266767041e-06, "loss": 1.6565, "step": 22380 }, { "epoch": 6.07268782207757, "grad_norm": 17.392497413526183, "learning_rate": 7.426767937706174e-06, "loss": 1.7869, "step": 22390 }, { "epoch": 6.075400054244644, "grad_norm": 16.338520902803566, "learning_rate": 7.424008231938856e-06, "loss": 1.6985, "step": 22400 }, { "epoch": 6.0781122864117165, "grad_norm": 16.856977353031315, "learning_rate": 7.42124756050155e-06, "loss": 1.6792, "step": 22410 }, { "epoch": 6.08082451857879, "grad_norm": 18.5217818942569, "learning_rate": 7.418485924494044e-06, "loss": 1.7631, "step": 22420 }, { "epoch": 6.083536750745864, "grad_norm": 18.477022329179206, "learning_rate": 7.415723325016512e-06, "loss": 1.6713, "step": 22430 }, { "epoch": 6.086248982912937, "grad_norm": 17.651515726786794, "learning_rate": 7.41295976316951e-06, "loss": 1.7352, "step": 22440 }, { "epoch": 6.088961215080011, "grad_norm": 17.272128456555365, "learning_rate": 7.41019524005398e-06, "loss": 1.6568, "step": 22450 }, { "epoch": 6.091673447247084, "grad_norm": 17.55461993267076, "learning_rate": 7.407429756771245e-06, "loss": 1.6255, "step": 22460 }, { "epoch": 6.094385679414158, "grad_norm": 18.651322164069107, "learning_rate": 7.404663314423011e-06, "loss": 1.6428, "step": 22470 }, { "epoch": 6.097097911581232, "grad_norm": 16.76581418202974, "learning_rate": 7.401895914111363e-06, "loss": 1.794, "step": 22480 }, { "epoch": 6.0998101437483045, "grad_norm": 19.873951175700622, "learning_rate": 7.3991275569387755e-06, "loss": 1.7086, "step": 22490 }, { "epoch": 6.102522375915378, "grad_norm": 17.444018380270833, "learning_rate": 7.396358244008096e-06, "loss": 1.8316, "step": 22500 }, { "epoch": 6.105234608082452, "grad_norm": 17.340492747241782, "learning_rate": 7.393587976422558e-06, "loss": 1.7069, "step": 22510 }, { "epoch": 6.107946840249525, "grad_norm": 19.171894307359107, "learning_rate": 7.390816755285772e-06, "loss": 1.6441, "step": 22520 }, { "epoch": 6.110659072416599, "grad_norm": 17.79505846807756, "learning_rate": 7.388044581701731e-06, "loss": 1.6921, "step": 22530 }, { "epoch": 6.113371304583672, "grad_norm": 17.590712687674262, "learning_rate": 7.385271456774805e-06, "loss": 1.6596, "step": 22540 }, { "epoch": 6.116083536750746, "grad_norm": 16.818563876377, "learning_rate": 7.382497381609745e-06, "loss": 1.6551, "step": 22550 }, { "epoch": 6.11879576891782, "grad_norm": 16.662178215249856, "learning_rate": 7.379722357311676e-06, "loss": 1.7056, "step": 22560 }, { "epoch": 6.121508001084893, "grad_norm": 20.318937167127203, "learning_rate": 7.3769463849861125e-06, "loss": 1.6405, "step": 22570 }, { "epoch": 6.124220233251966, "grad_norm": 18.721228025953298, "learning_rate": 7.374169465738931e-06, "loss": 1.642, "step": 22580 }, { "epoch": 6.12693246541904, "grad_norm": 18.29651825749388, "learning_rate": 7.371391600676401e-06, "loss": 1.6918, "step": 22590 }, { "epoch": 6.129644697586113, "grad_norm": 18.797641087227397, "learning_rate": 7.368612790905154e-06, "loss": 1.6477, "step": 22600 }, { "epoch": 6.132356929753187, "grad_norm": 17.694302819368218, "learning_rate": 7.365833037532209e-06, "loss": 1.622, "step": 22610 }, { "epoch": 6.13506916192026, "grad_norm": 17.281502974897112, "learning_rate": 7.3630523416649536e-06, "loss": 1.7069, "step": 22620 }, { "epoch": 6.137781394087334, "grad_norm": 18.984671041703354, "learning_rate": 7.360270704411158e-06, "loss": 1.7055, "step": 22630 }, { "epoch": 6.140493626254408, "grad_norm": 17.494221846147255, "learning_rate": 7.357488126878961e-06, "loss": 1.6456, "step": 22640 }, { "epoch": 6.143205858421481, "grad_norm": 18.32056896368272, "learning_rate": 7.3547046101768805e-06, "loss": 1.6866, "step": 22650 }, { "epoch": 6.145918090588554, "grad_norm": 17.73926995987851, "learning_rate": 7.351920155413803e-06, "loss": 1.7601, "step": 22660 }, { "epoch": 6.148630322755628, "grad_norm": 17.837604726532486, "learning_rate": 7.349134763698997e-06, "loss": 1.6729, "step": 22670 }, { "epoch": 6.151342554922701, "grad_norm": 16.301834479811454, "learning_rate": 7.346348436142093e-06, "loss": 1.8028, "step": 22680 }, { "epoch": 6.154054787089775, "grad_norm": 16.802518861907085, "learning_rate": 7.343561173853106e-06, "loss": 1.5696, "step": 22690 }, { "epoch": 6.156767019256848, "grad_norm": 18.336412515698367, "learning_rate": 7.340772977942417e-06, "loss": 1.5878, "step": 22700 }, { "epoch": 6.159479251423922, "grad_norm": 16.996984368858854, "learning_rate": 7.337983849520781e-06, "loss": 1.6661, "step": 22710 }, { "epoch": 6.162191483590996, "grad_norm": 17.789725287860495, "learning_rate": 7.335193789699322e-06, "loss": 1.7365, "step": 22720 }, { "epoch": 6.164903715758069, "grad_norm": 15.820406171484162, "learning_rate": 7.3324027995895374e-06, "loss": 1.6825, "step": 22730 }, { "epoch": 6.1676159479251425, "grad_norm": 18.658106022416256, "learning_rate": 7.329610880303293e-06, "loss": 1.8406, "step": 22740 }, { "epoch": 6.170328180092216, "grad_norm": 19.575781179778513, "learning_rate": 7.3268180329528274e-06, "loss": 1.7277, "step": 22750 }, { "epoch": 6.173040412259289, "grad_norm": 18.877553720560265, "learning_rate": 7.3240242586507485e-06, "loss": 1.6304, "step": 22760 }, { "epoch": 6.175752644426363, "grad_norm": 18.546555486722216, "learning_rate": 7.321229558510033e-06, "loss": 1.6942, "step": 22770 }, { "epoch": 6.178464876593436, "grad_norm": 17.748525434279035, "learning_rate": 7.318433933644022e-06, "loss": 1.6939, "step": 22780 }, { "epoch": 6.18117710876051, "grad_norm": 18.117996082712803, "learning_rate": 7.3156373851664355e-06, "loss": 1.5702, "step": 22790 }, { "epoch": 6.183889340927584, "grad_norm": 16.879796139946222, "learning_rate": 7.31283991419135e-06, "loss": 1.754, "step": 22800 }, { "epoch": 6.186601573094657, "grad_norm": 17.62809883749084, "learning_rate": 7.310041521833217e-06, "loss": 1.7601, "step": 22810 }, { "epoch": 6.1893138052617305, "grad_norm": 17.34195794028807, "learning_rate": 7.307242209206851e-06, "loss": 1.7014, "step": 22820 }, { "epoch": 6.192026037428804, "grad_norm": 18.49222784904851, "learning_rate": 7.304441977427437e-06, "loss": 1.6635, "step": 22830 }, { "epoch": 6.194738269595877, "grad_norm": 16.41125395456463, "learning_rate": 7.301640827610523e-06, "loss": 1.5766, "step": 22840 }, { "epoch": 6.197450501762951, "grad_norm": 17.59938687969622, "learning_rate": 7.298838760872023e-06, "loss": 1.6813, "step": 22850 }, { "epoch": 6.200162733930024, "grad_norm": 18.304590876294597, "learning_rate": 7.296035778328218e-06, "loss": 1.6827, "step": 22860 }, { "epoch": 6.202874966097098, "grad_norm": 18.689831529887332, "learning_rate": 7.293231881095752e-06, "loss": 1.7848, "step": 22870 }, { "epoch": 6.205587198264172, "grad_norm": 18.811746808978903, "learning_rate": 7.2904270702916324e-06, "loss": 1.783, "step": 22880 }, { "epoch": 6.208299430431245, "grad_norm": 18.095495069433596, "learning_rate": 7.287621347033237e-06, "loss": 1.6388, "step": 22890 }, { "epoch": 6.2110116625983185, "grad_norm": 16.781366655642476, "learning_rate": 7.284814712438297e-06, "loss": 1.7031, "step": 22900 }, { "epoch": 6.213723894765392, "grad_norm": 17.974846942075587, "learning_rate": 7.282007167624917e-06, "loss": 1.7805, "step": 22910 }, { "epoch": 6.216436126932465, "grad_norm": 17.85928478830078, "learning_rate": 7.279198713711556e-06, "loss": 1.7213, "step": 22920 }, { "epoch": 6.219148359099539, "grad_norm": 18.77003925584053, "learning_rate": 7.276389351817039e-06, "loss": 1.7407, "step": 22930 }, { "epoch": 6.221860591266612, "grad_norm": 18.876008045829348, "learning_rate": 7.2735790830605524e-06, "loss": 1.7501, "step": 22940 }, { "epoch": 6.224572823433686, "grad_norm": 16.343950818658872, "learning_rate": 7.270767908561645e-06, "loss": 1.7219, "step": 22950 }, { "epoch": 6.22728505560076, "grad_norm": 17.883800786884102, "learning_rate": 7.267955829440223e-06, "loss": 1.6103, "step": 22960 }, { "epoch": 6.229997287767833, "grad_norm": 18.235717095537716, "learning_rate": 7.265142846816557e-06, "loss": 1.7931, "step": 22970 }, { "epoch": 6.2327095199349065, "grad_norm": 17.469522765813295, "learning_rate": 7.262328961811273e-06, "loss": 1.6761, "step": 22980 }, { "epoch": 6.23542175210198, "grad_norm": 18.486270165617995, "learning_rate": 7.259514175545361e-06, "loss": 1.7997, "step": 22990 }, { "epoch": 6.238133984269053, "grad_norm": 17.89281180295903, "learning_rate": 7.256698489140167e-06, "loss": 1.7602, "step": 23000 }, { "epoch": 6.240846216436127, "grad_norm": 17.17301724566747, "learning_rate": 7.253881903717399e-06, "loss": 1.8927, "step": 23010 }, { "epoch": 6.2435584486032, "grad_norm": 18.66758080671406, "learning_rate": 7.2510644203991186e-06, "loss": 1.7714, "step": 23020 }, { "epoch": 6.246270680770274, "grad_norm": 17.76503879459526, "learning_rate": 7.248246040307748e-06, "loss": 1.7933, "step": 23030 }, { "epoch": 6.248982912937348, "grad_norm": 19.210538793735893, "learning_rate": 7.2454267645660656e-06, "loss": 1.5631, "step": 23040 }, { "epoch": 6.251695145104421, "grad_norm": 19.50143979985811, "learning_rate": 7.242606594297208e-06, "loss": 1.8719, "step": 23050 }, { "epoch": 6.2544073772714945, "grad_norm": 17.367993624894076, "learning_rate": 7.239785530624666e-06, "loss": 1.6281, "step": 23060 }, { "epoch": 6.257119609438568, "grad_norm": 19.079419932567422, "learning_rate": 7.236963574672288e-06, "loss": 1.7891, "step": 23070 }, { "epoch": 6.259831841605641, "grad_norm": 18.538562820209265, "learning_rate": 7.234140727564277e-06, "loss": 1.7381, "step": 23080 }, { "epoch": 6.262544073772715, "grad_norm": 16.494069282813776, "learning_rate": 7.231316990425193e-06, "loss": 1.7277, "step": 23090 }, { "epoch": 6.265256305939788, "grad_norm": 18.771778352585695, "learning_rate": 7.228492364379946e-06, "loss": 1.8747, "step": 23100 }, { "epoch": 6.267968538106862, "grad_norm": 17.50173704897362, "learning_rate": 7.225666850553808e-06, "loss": 1.6771, "step": 23110 }, { "epoch": 6.270680770273936, "grad_norm": 19.459731132649534, "learning_rate": 7.222840450072392e-06, "loss": 1.7083, "step": 23120 }, { "epoch": 6.273393002441009, "grad_norm": 18.3393343644881, "learning_rate": 7.22001316406168e-06, "loss": 1.6754, "step": 23130 }, { "epoch": 6.276105234608083, "grad_norm": 18.036450713566907, "learning_rate": 7.217184993647993e-06, "loss": 1.6886, "step": 23140 }, { "epoch": 6.2788174667751555, "grad_norm": 19.25657105949883, "learning_rate": 7.214355939958014e-06, "loss": 1.6566, "step": 23150 }, { "epoch": 6.281529698942229, "grad_norm": 18.740428181695144, "learning_rate": 7.21152600411877e-06, "loss": 1.7825, "step": 23160 }, { "epoch": 6.284241931109303, "grad_norm": 19.742749965187244, "learning_rate": 7.2086951872576485e-06, "loss": 1.8867, "step": 23170 }, { "epoch": 6.286954163276376, "grad_norm": 18.26058475631444, "learning_rate": 7.205863490502377e-06, "loss": 1.8143, "step": 23180 }, { "epoch": 6.28966639544345, "grad_norm": 18.169916397797255, "learning_rate": 7.203030914981043e-06, "loss": 1.621, "step": 23190 }, { "epoch": 6.292378627610524, "grad_norm": 17.90603422815768, "learning_rate": 7.2001974618220795e-06, "loss": 1.6517, "step": 23200 }, { "epoch": 6.295090859777597, "grad_norm": 20.63809004190008, "learning_rate": 7.197363132154271e-06, "loss": 1.7131, "step": 23210 }, { "epoch": 6.297803091944671, "grad_norm": 18.285487915941886, "learning_rate": 7.1945279271067495e-06, "loss": 1.8278, "step": 23220 }, { "epoch": 6.300515324111744, "grad_norm": 16.254990721132934, "learning_rate": 7.191691847808996e-06, "loss": 1.7449, "step": 23230 }, { "epoch": 6.303227556278817, "grad_norm": 18.8971447240364, "learning_rate": 7.18885489539084e-06, "loss": 1.7244, "step": 23240 }, { "epoch": 6.305939788445891, "grad_norm": 18.888211153631396, "learning_rate": 7.186017070982461e-06, "loss": 1.7233, "step": 23250 }, { "epoch": 6.308652020612964, "grad_norm": 18.069393830433686, "learning_rate": 7.183178375714382e-06, "loss": 1.7454, "step": 23260 }, { "epoch": 6.311364252780038, "grad_norm": 18.15419727877763, "learning_rate": 7.1803388107174755e-06, "loss": 1.7931, "step": 23270 }, { "epoch": 6.314076484947112, "grad_norm": 17.49572754784101, "learning_rate": 7.17749837712296e-06, "loss": 1.7215, "step": 23280 }, { "epoch": 6.316788717114185, "grad_norm": 17.399388774387678, "learning_rate": 7.174657076062402e-06, "loss": 1.6278, "step": 23290 }, { "epoch": 6.319500949281259, "grad_norm": 19.66064194421676, "learning_rate": 7.171814908667707e-06, "loss": 1.7784, "step": 23300 }, { "epoch": 6.3222131814483316, "grad_norm": 19.840948054735474, "learning_rate": 7.168971876071133e-06, "loss": 1.7366, "step": 23310 }, { "epoch": 6.324925413615405, "grad_norm": 20.317832370651207, "learning_rate": 7.16612797940528e-06, "loss": 1.6951, "step": 23320 }, { "epoch": 6.327637645782479, "grad_norm": 19.602975025704986, "learning_rate": 7.163283219803092e-06, "loss": 1.8521, "step": 23330 }, { "epoch": 6.330349877949552, "grad_norm": 17.60043757833827, "learning_rate": 7.160437598397856e-06, "loss": 1.7074, "step": 23340 }, { "epoch": 6.333062110116626, "grad_norm": 20.103254149263357, "learning_rate": 7.1575911163232035e-06, "loss": 1.6366, "step": 23350 }, { "epoch": 6.3357743422837, "grad_norm": 19.183213569152244, "learning_rate": 7.154743774713109e-06, "loss": 1.589, "step": 23360 }, { "epoch": 6.338486574450773, "grad_norm": 19.29163751562287, "learning_rate": 7.151895574701887e-06, "loss": 1.7159, "step": 23370 }, { "epoch": 6.341198806617847, "grad_norm": 17.275047002005092, "learning_rate": 7.149046517424197e-06, "loss": 1.8015, "step": 23380 }, { "epoch": 6.34391103878492, "grad_norm": 17.060405731422513, "learning_rate": 7.146196604015041e-06, "loss": 1.7481, "step": 23390 }, { "epoch": 6.346623270951993, "grad_norm": 19.681082291833842, "learning_rate": 7.143345835609757e-06, "loss": 1.8159, "step": 23400 }, { "epoch": 6.349335503119067, "grad_norm": 17.389095430658553, "learning_rate": 7.140494213344029e-06, "loss": 1.7644, "step": 23410 }, { "epoch": 6.35204773528614, "grad_norm": 19.02030141543921, "learning_rate": 7.137641738353876e-06, "loss": 1.6872, "step": 23420 }, { "epoch": 6.354759967453214, "grad_norm": 17.870282121130238, "learning_rate": 7.1347884117756605e-06, "loss": 1.5514, "step": 23430 }, { "epoch": 6.357472199620288, "grad_norm": 18.50468627600644, "learning_rate": 7.131934234746084e-06, "loss": 1.6661, "step": 23440 }, { "epoch": 6.360184431787361, "grad_norm": 15.610527068002124, "learning_rate": 7.1290792084021865e-06, "loss": 1.7286, "step": 23450 }, { "epoch": 6.362896663954435, "grad_norm": 18.195777140090858, "learning_rate": 7.126223333881343e-06, "loss": 1.6979, "step": 23460 }, { "epoch": 6.365608896121508, "grad_norm": 16.864412521252337, "learning_rate": 7.123366612321274e-06, "loss": 1.6841, "step": 23470 }, { "epoch": 6.368321128288581, "grad_norm": 18.995916718652495, "learning_rate": 7.120509044860028e-06, "loss": 1.7776, "step": 23480 }, { "epoch": 6.371033360455655, "grad_norm": 18.14890108319271, "learning_rate": 7.1176506326359965e-06, "loss": 1.7822, "step": 23490 }, { "epoch": 6.373745592622728, "grad_norm": 17.66485263543841, "learning_rate": 7.114791376787906e-06, "loss": 1.8312, "step": 23500 }, { "epoch": 6.376457824789802, "grad_norm": 17.607117987502374, "learning_rate": 7.111931278454823e-06, "loss": 1.6408, "step": 23510 }, { "epoch": 6.379170056956876, "grad_norm": 19.5011219867307, "learning_rate": 7.1090703387761405e-06, "loss": 1.7515, "step": 23520 }, { "epoch": 6.381882289123949, "grad_norm": 17.28025276374495, "learning_rate": 7.106208558891596e-06, "loss": 1.7036, "step": 23530 }, { "epoch": 6.384594521291023, "grad_norm": 17.07356107004101, "learning_rate": 7.103345939941255e-06, "loss": 1.8327, "step": 23540 }, { "epoch": 6.387306753458096, "grad_norm": 19.076062890730157, "learning_rate": 7.1004824830655215e-06, "loss": 1.7557, "step": 23550 }, { "epoch": 6.3900189856251695, "grad_norm": 19.848649890159358, "learning_rate": 7.097618189405131e-06, "loss": 1.7369, "step": 23560 }, { "epoch": 6.392731217792243, "grad_norm": 19.640980554878393, "learning_rate": 7.094753060101156e-06, "loss": 1.787, "step": 23570 }, { "epoch": 6.395443449959316, "grad_norm": 18.51771342375885, "learning_rate": 7.091887096294996e-06, "loss": 1.8434, "step": 23580 }, { "epoch": 6.39815568212639, "grad_norm": 19.67286343884004, "learning_rate": 7.089020299128388e-06, "loss": 1.7315, "step": 23590 }, { "epoch": 6.400867914293464, "grad_norm": 20.181983401389605, "learning_rate": 7.086152669743397e-06, "loss": 1.7179, "step": 23600 }, { "epoch": 6.403580146460537, "grad_norm": 18.27179153869595, "learning_rate": 7.083284209282424e-06, "loss": 1.8266, "step": 23610 }, { "epoch": 6.406292378627611, "grad_norm": 18.054743123022924, "learning_rate": 7.080414918888195e-06, "loss": 1.8049, "step": 23620 }, { "epoch": 6.409004610794684, "grad_norm": 17.734603289563328, "learning_rate": 7.077544799703775e-06, "loss": 1.6744, "step": 23630 }, { "epoch": 6.4117168429617575, "grad_norm": 19.227303871361293, "learning_rate": 7.074673852872551e-06, "loss": 1.8488, "step": 23640 }, { "epoch": 6.414429075128831, "grad_norm": 17.211297769682368, "learning_rate": 7.071802079538247e-06, "loss": 1.6786, "step": 23650 }, { "epoch": 6.417141307295904, "grad_norm": 19.371722176409033, "learning_rate": 7.068929480844908e-06, "loss": 1.7065, "step": 23660 }, { "epoch": 6.419853539462978, "grad_norm": 17.82374490327476, "learning_rate": 7.066056057936915e-06, "loss": 1.7193, "step": 23670 }, { "epoch": 6.422565771630052, "grad_norm": 17.353680636546468, "learning_rate": 7.063181811958973e-06, "loss": 1.622, "step": 23680 }, { "epoch": 6.425278003797125, "grad_norm": 19.657162789610556, "learning_rate": 7.060306744056119e-06, "loss": 1.8634, "step": 23690 }, { "epoch": 6.427990235964199, "grad_norm": 18.608674634633577, "learning_rate": 7.057430855373711e-06, "loss": 1.6629, "step": 23700 }, { "epoch": 6.430702468131272, "grad_norm": 18.50718240209421, "learning_rate": 7.054554147057443e-06, "loss": 1.7798, "step": 23710 }, { "epoch": 6.4334147002983455, "grad_norm": 16.951014248120654, "learning_rate": 7.051676620253326e-06, "loss": 1.7986, "step": 23720 }, { "epoch": 6.436126932465419, "grad_norm": 19.53065298056961, "learning_rate": 7.048798276107705e-06, "loss": 1.8192, "step": 23730 }, { "epoch": 6.438839164632492, "grad_norm": 17.69735673147533, "learning_rate": 7.045919115767244e-06, "loss": 1.6187, "step": 23740 }, { "epoch": 6.441551396799566, "grad_norm": 17.89123266881945, "learning_rate": 7.043039140378936e-06, "loss": 1.7404, "step": 23750 }, { "epoch": 6.44426362896664, "grad_norm": 19.117606574688903, "learning_rate": 7.040158351090099e-06, "loss": 1.6373, "step": 23760 }, { "epoch": 6.446975861133713, "grad_norm": 19.79197279702475, "learning_rate": 7.0372767490483755e-06, "loss": 1.7322, "step": 23770 }, { "epoch": 6.449688093300787, "grad_norm": 17.101212540371087, "learning_rate": 7.034394335401726e-06, "loss": 1.7681, "step": 23780 }, { "epoch": 6.45240032546786, "grad_norm": 17.371203852825197, "learning_rate": 7.031511111298445e-06, "loss": 1.7656, "step": 23790 }, { "epoch": 6.4551125576349335, "grad_norm": 19.02647839261626, "learning_rate": 7.028627077887138e-06, "loss": 1.6661, "step": 23800 }, { "epoch": 6.457824789802007, "grad_norm": 18.264471187089764, "learning_rate": 7.025742236316742e-06, "loss": 1.7873, "step": 23810 }, { "epoch": 6.46053702196908, "grad_norm": 17.91345244003105, "learning_rate": 7.022856587736511e-06, "loss": 1.6246, "step": 23820 }, { "epoch": 6.463249254136154, "grad_norm": 18.552234679635383, "learning_rate": 7.019970133296023e-06, "loss": 1.8426, "step": 23830 }, { "epoch": 6.465961486303227, "grad_norm": 16.64594340164597, "learning_rate": 7.017082874145176e-06, "loss": 1.8004, "step": 23840 }, { "epoch": 6.468673718470301, "grad_norm": 19.75058217940333, "learning_rate": 7.014194811434187e-06, "loss": 1.8368, "step": 23850 }, { "epoch": 6.471385950637375, "grad_norm": 18.552489308841587, "learning_rate": 7.011305946313598e-06, "loss": 1.6526, "step": 23860 }, { "epoch": 6.474098182804448, "grad_norm": 21.342404198834874, "learning_rate": 7.008416279934264e-06, "loss": 1.8486, "step": 23870 }, { "epoch": 6.4768104149715215, "grad_norm": 17.896113726470812, "learning_rate": 7.005525813447364e-06, "loss": 1.6862, "step": 23880 }, { "epoch": 6.479522647138595, "grad_norm": 17.691089194299327, "learning_rate": 7.002634548004397e-06, "loss": 1.7534, "step": 23890 }, { "epoch": 6.482234879305668, "grad_norm": 19.03985957921553, "learning_rate": 6.9997424847571725e-06, "loss": 1.8649, "step": 23900 }, { "epoch": 6.484947111472742, "grad_norm": 17.883552690320673, "learning_rate": 6.996849624857828e-06, "loss": 1.6388, "step": 23910 }, { "epoch": 6.487659343639816, "grad_norm": 19.276688498708562, "learning_rate": 6.99395596945881e-06, "loss": 1.7023, "step": 23920 }, { "epoch": 6.490371575806889, "grad_norm": 18.943516037551863, "learning_rate": 6.991061519712886e-06, "loss": 1.8628, "step": 23930 }, { "epoch": 6.493083807973963, "grad_norm": 20.442311509563666, "learning_rate": 6.988166276773141e-06, "loss": 1.7607, "step": 23940 }, { "epoch": 6.495796040141036, "grad_norm": 19.906533234911134, "learning_rate": 6.985270241792972e-06, "loss": 1.7677, "step": 23950 }, { "epoch": 6.49850827230811, "grad_norm": 18.882876774011486, "learning_rate": 6.982373415926095e-06, "loss": 1.8159, "step": 23960 }, { "epoch": 6.501220504475183, "grad_norm": 17.69953132280601, "learning_rate": 6.97947580032654e-06, "loss": 1.7107, "step": 23970 }, { "epoch": 6.503932736642256, "grad_norm": 17.58922394337728, "learning_rate": 6.976577396148649e-06, "loss": 1.8326, "step": 23980 }, { "epoch": 6.50664496880933, "grad_norm": 17.88730879902384, "learning_rate": 6.973678204547084e-06, "loss": 1.828, "step": 23990 }, { "epoch": 6.509357200976403, "grad_norm": 16.945621931483302, "learning_rate": 6.9707782266768156e-06, "loss": 1.7078, "step": 24000 }, { "epoch": 6.512069433143477, "grad_norm": 20.17536325229849, "learning_rate": 6.96787746369313e-06, "loss": 1.7524, "step": 24010 }, { "epoch": 6.514781665310551, "grad_norm": 17.466502756397272, "learning_rate": 6.964975916751625e-06, "loss": 1.8392, "step": 24020 }, { "epoch": 6.517493897477624, "grad_norm": 16.701864676153953, "learning_rate": 6.962073587008211e-06, "loss": 1.7593, "step": 24030 }, { "epoch": 6.520206129644698, "grad_norm": 19.056741989252895, "learning_rate": 6.95917047561911e-06, "loss": 1.8133, "step": 24040 }, { "epoch": 6.522918361811771, "grad_norm": 17.15266152541875, "learning_rate": 6.956266583740858e-06, "loss": 1.6743, "step": 24050 }, { "epoch": 6.525630593978844, "grad_norm": 20.980874428682917, "learning_rate": 6.953361912530296e-06, "loss": 1.772, "step": 24060 }, { "epoch": 6.528342826145918, "grad_norm": 18.928897138800377, "learning_rate": 6.950456463144583e-06, "loss": 1.7249, "step": 24070 }, { "epoch": 6.531055058312992, "grad_norm": 17.62909241657858, "learning_rate": 6.947550236741183e-06, "loss": 1.7784, "step": 24080 }, { "epoch": 6.533767290480065, "grad_norm": 18.198535631493414, "learning_rate": 6.94464323447787e-06, "loss": 1.7729, "step": 24090 }, { "epoch": 6.536479522647139, "grad_norm": 18.327367272068717, "learning_rate": 6.941735457512729e-06, "loss": 2.0151, "step": 24100 }, { "epoch": 6.539191754814212, "grad_norm": 16.710448610781455, "learning_rate": 6.938826907004153e-06, "loss": 1.8712, "step": 24110 }, { "epoch": 6.541903986981286, "grad_norm": 16.02815382494221, "learning_rate": 6.935917584110837e-06, "loss": 1.8244, "step": 24120 }, { "epoch": 6.5446162191483594, "grad_norm": 19.296967682634236, "learning_rate": 6.9330074899917976e-06, "loss": 1.8566, "step": 24130 }, { "epoch": 6.547328451315432, "grad_norm": 19.739002604677058, "learning_rate": 6.9300966258063464e-06, "loss": 1.7097, "step": 24140 }, { "epoch": 6.550040683482506, "grad_norm": 17.675856732951473, "learning_rate": 6.9271849927141065e-06, "loss": 1.8221, "step": 24150 }, { "epoch": 6.552752915649579, "grad_norm": 18.609550241539676, "learning_rate": 6.924272591875004e-06, "loss": 1.8237, "step": 24160 }, { "epoch": 6.555465147816653, "grad_norm": 17.543352624652655, "learning_rate": 6.921359424449279e-06, "loss": 1.7677, "step": 24170 }, { "epoch": 6.558177379983727, "grad_norm": 18.018675774509546, "learning_rate": 6.918445491597465e-06, "loss": 1.7801, "step": 24180 }, { "epoch": 6.5608896121508, "grad_norm": 17.934281741701287, "learning_rate": 6.915530794480411e-06, "loss": 1.8444, "step": 24190 }, { "epoch": 6.563601844317874, "grad_norm": 19.627385351522094, "learning_rate": 6.912615334259264e-06, "loss": 1.7471, "step": 24200 }, { "epoch": 6.5663140764849475, "grad_norm": 18.790558170376816, "learning_rate": 6.90969911209548e-06, "loss": 1.6841, "step": 24210 }, { "epoch": 6.56902630865202, "grad_norm": 18.535514013462084, "learning_rate": 6.9067821291508154e-06, "loss": 1.6881, "step": 24220 }, { "epoch": 6.571738540819094, "grad_norm": 17.018640028658282, "learning_rate": 6.903864386587328e-06, "loss": 1.6415, "step": 24230 }, { "epoch": 6.574450772986168, "grad_norm": 18.323216108735984, "learning_rate": 6.900945885567381e-06, "loss": 1.7173, "step": 24240 }, { "epoch": 6.577163005153241, "grad_norm": 18.25631985637151, "learning_rate": 6.89802662725364e-06, "loss": 1.7846, "step": 24250 }, { "epoch": 6.579875237320315, "grad_norm": 19.196364418063272, "learning_rate": 6.895106612809072e-06, "loss": 1.6918, "step": 24260 }, { "epoch": 6.582587469487388, "grad_norm": 18.97723809702243, "learning_rate": 6.892185843396944e-06, "loss": 1.5827, "step": 24270 }, { "epoch": 6.585299701654462, "grad_norm": 18.62667558702463, "learning_rate": 6.889264320180822e-06, "loss": 1.7871, "step": 24280 }, { "epoch": 6.5880119338215355, "grad_norm": 16.283152873410064, "learning_rate": 6.886342044324579e-06, "loss": 1.5436, "step": 24290 }, { "epoch": 6.590724165988608, "grad_norm": 19.208174241864512, "learning_rate": 6.88341901699238e-06, "loss": 1.8532, "step": 24300 }, { "epoch": 6.593436398155682, "grad_norm": 20.903094811402454, "learning_rate": 6.8804952393486945e-06, "loss": 1.6492, "step": 24310 }, { "epoch": 6.596148630322755, "grad_norm": 18.966605259317785, "learning_rate": 6.877570712558289e-06, "loss": 1.8392, "step": 24320 }, { "epoch": 6.598860862489829, "grad_norm": 18.40231548457875, "learning_rate": 6.874645437786229e-06, "loss": 1.7605, "step": 24330 }, { "epoch": 6.601573094656903, "grad_norm": 19.389044484815166, "learning_rate": 6.871719416197877e-06, "loss": 1.7933, "step": 24340 }, { "epoch": 6.604285326823976, "grad_norm": 19.0563220282301, "learning_rate": 6.868792648958896e-06, "loss": 1.7589, "step": 24350 }, { "epoch": 6.60699755899105, "grad_norm": 19.237546016766682, "learning_rate": 6.86586513723524e-06, "loss": 1.8746, "step": 24360 }, { "epoch": 6.6097097911581235, "grad_norm": 19.562058147882865, "learning_rate": 6.8629368821931664e-06, "loss": 1.6939, "step": 24370 }, { "epoch": 6.6124220233251965, "grad_norm": 16.64757716210416, "learning_rate": 6.860007884999225e-06, "loss": 1.7294, "step": 24380 }, { "epoch": 6.61513425549227, "grad_norm": 17.009649377763978, "learning_rate": 6.857078146820262e-06, "loss": 1.714, "step": 24390 }, { "epoch": 6.617846487659344, "grad_norm": 18.4864687003004, "learning_rate": 6.854147668823417e-06, "loss": 1.7883, "step": 24400 }, { "epoch": 6.620558719826417, "grad_norm": 18.756949342722887, "learning_rate": 6.851216452176128e-06, "loss": 1.8291, "step": 24410 }, { "epoch": 6.623270951993491, "grad_norm": 20.00581115005652, "learning_rate": 6.848284498046125e-06, "loss": 1.7123, "step": 24420 }, { "epoch": 6.625983184160564, "grad_norm": 16.59018450483297, "learning_rate": 6.845351807601432e-06, "loss": 1.7423, "step": 24430 }, { "epoch": 6.628695416327638, "grad_norm": 20.73033204607409, "learning_rate": 6.842418382010366e-06, "loss": 1.8468, "step": 24440 }, { "epoch": 6.6314076484947115, "grad_norm": 16.260217347137157, "learning_rate": 6.839484222441539e-06, "loss": 1.6984, "step": 24450 }, { "epoch": 6.6341198806617845, "grad_norm": 19.745335933762334, "learning_rate": 6.8365493300638516e-06, "loss": 1.8025, "step": 24460 }, { "epoch": 6.636832112828858, "grad_norm": 17.276436255448356, "learning_rate": 6.8336137060465e-06, "loss": 1.7573, "step": 24470 }, { "epoch": 6.639544344995931, "grad_norm": 17.056067887900436, "learning_rate": 6.83067735155897e-06, "loss": 1.7306, "step": 24480 }, { "epoch": 6.642256577163005, "grad_norm": 18.37139595597807, "learning_rate": 6.827740267771037e-06, "loss": 1.6526, "step": 24490 }, { "epoch": 6.644968809330079, "grad_norm": 20.37338255756979, "learning_rate": 6.82480245585277e-06, "loss": 1.7599, "step": 24500 }, { "epoch": 6.647681041497152, "grad_norm": 20.431765357003872, "learning_rate": 6.821863916974528e-06, "loss": 1.8992, "step": 24510 }, { "epoch": 6.650393273664226, "grad_norm": 17.683994487335923, "learning_rate": 6.818924652306956e-06, "loss": 1.7524, "step": 24520 }, { "epoch": 6.653105505831299, "grad_norm": 21.448847709351156, "learning_rate": 6.815984663020993e-06, "loss": 1.8032, "step": 24530 }, { "epoch": 6.6558177379983725, "grad_norm": 16.361234601312177, "learning_rate": 6.813043950287862e-06, "loss": 1.8349, "step": 24540 }, { "epoch": 6.658529970165446, "grad_norm": 18.19422742445778, "learning_rate": 6.810102515279078e-06, "loss": 1.7322, "step": 24550 }, { "epoch": 6.661242202332519, "grad_norm": 20.286299967203927, "learning_rate": 6.80716035916644e-06, "loss": 1.7644, "step": 24560 }, { "epoch": 6.663954434499593, "grad_norm": 18.54171847478258, "learning_rate": 6.804217483122039e-06, "loss": 1.8279, "step": 24570 }, { "epoch": 6.666666666666667, "grad_norm": 18.883344332605677, "learning_rate": 6.801273888318247e-06, "loss": 1.7226, "step": 24580 }, { "epoch": 6.66937889883374, "grad_norm": 17.804653986787965, "learning_rate": 6.79832957592773e-06, "loss": 1.7471, "step": 24590 }, { "epoch": 6.672091131000814, "grad_norm": 16.761945753834773, "learning_rate": 6.7953845471234315e-06, "loss": 1.7942, "step": 24600 }, { "epoch": 6.674803363167888, "grad_norm": 17.86316634869927, "learning_rate": 6.792438803078585e-06, "loss": 1.9202, "step": 24610 }, { "epoch": 6.6775155953349605, "grad_norm": 19.060367748417196, "learning_rate": 6.789492344966709e-06, "loss": 1.6676, "step": 24620 }, { "epoch": 6.680227827502034, "grad_norm": 18.24495752692355, "learning_rate": 6.786545173961605e-06, "loss": 1.9282, "step": 24630 }, { "epoch": 6.682940059669107, "grad_norm": 17.385004772569868, "learning_rate": 6.783597291237361e-06, "loss": 1.927, "step": 24640 }, { "epoch": 6.685652291836181, "grad_norm": 17.921064243308532, "learning_rate": 6.780648697968344e-06, "loss": 1.7394, "step": 24650 }, { "epoch": 6.688364524003255, "grad_norm": 18.514549426783713, "learning_rate": 6.777699395329208e-06, "loss": 1.6856, "step": 24660 }, { "epoch": 6.691076756170328, "grad_norm": 18.298417137293356, "learning_rate": 6.77474938449489e-06, "loss": 1.7488, "step": 24670 }, { "epoch": 6.693788988337402, "grad_norm": 18.22855939100513, "learning_rate": 6.771798666640604e-06, "loss": 1.8, "step": 24680 }, { "epoch": 6.696501220504475, "grad_norm": 18.156392440173274, "learning_rate": 6.768847242941853e-06, "loss": 1.7424, "step": 24690 }, { "epoch": 6.6992134526715486, "grad_norm": 17.146971055913635, "learning_rate": 6.765895114574414e-06, "loss": 1.7705, "step": 24700 }, { "epoch": 6.701925684838622, "grad_norm": 21.800106733513093, "learning_rate": 6.762942282714351e-06, "loss": 1.7638, "step": 24710 }, { "epoch": 6.704637917005695, "grad_norm": 19.567296914648452, "learning_rate": 6.759988748538003e-06, "loss": 1.7913, "step": 24720 }, { "epoch": 6.707350149172769, "grad_norm": 20.888876852866105, "learning_rate": 6.757034513221994e-06, "loss": 1.7591, "step": 24730 }, { "epoch": 6.710062381339843, "grad_norm": 20.910732887342736, "learning_rate": 6.75407957794322e-06, "loss": 1.73, "step": 24740 }, { "epoch": 6.712774613506916, "grad_norm": 19.065339502597595, "learning_rate": 6.751123943878864e-06, "loss": 1.8143, "step": 24750 }, { "epoch": 6.71548684567399, "grad_norm": 18.362794054377375, "learning_rate": 6.748167612206382e-06, "loss": 1.9543, "step": 24760 }, { "epoch": 6.718199077841064, "grad_norm": 21.732415273439486, "learning_rate": 6.745210584103512e-06, "loss": 1.721, "step": 24770 }, { "epoch": 6.720911310008137, "grad_norm": 18.22828117451675, "learning_rate": 6.742252860748264e-06, "loss": 1.6068, "step": 24780 }, { "epoch": 6.72362354217521, "grad_norm": 18.124639118525742, "learning_rate": 6.73929444331893e-06, "loss": 1.7432, "step": 24790 }, { "epoch": 6.726335774342283, "grad_norm": 18.484064588989693, "learning_rate": 6.736335332994075e-06, "loss": 1.7174, "step": 24800 }, { "epoch": 6.729048006509357, "grad_norm": 21.903479158710773, "learning_rate": 6.733375530952542e-06, "loss": 1.9092, "step": 24810 }, { "epoch": 6.731760238676431, "grad_norm": 18.12345327430934, "learning_rate": 6.730415038373449e-06, "loss": 1.8142, "step": 24820 }, { "epoch": 6.734472470843504, "grad_norm": 20.472714520471367, "learning_rate": 6.7274538564361905e-06, "loss": 1.8162, "step": 24830 }, { "epoch": 6.737184703010578, "grad_norm": 18.630349146946717, "learning_rate": 6.724491986320433e-06, "loss": 1.8356, "step": 24840 }, { "epoch": 6.739896935177651, "grad_norm": 18.28039337055057, "learning_rate": 6.721529429206119e-06, "loss": 1.7367, "step": 24850 }, { "epoch": 6.742609167344725, "grad_norm": 20.05733744832361, "learning_rate": 6.718566186273465e-06, "loss": 1.8125, "step": 24860 }, { "epoch": 6.745321399511798, "grad_norm": 18.686878294736886, "learning_rate": 6.715602258702958e-06, "loss": 1.7663, "step": 24870 }, { "epoch": 6.748033631678871, "grad_norm": 20.677756064066383, "learning_rate": 6.7126376476753595e-06, "loss": 1.6436, "step": 24880 }, { "epoch": 6.750745863845945, "grad_norm": 19.349078825051294, "learning_rate": 6.709672354371706e-06, "loss": 1.6834, "step": 24890 }, { "epoch": 6.753458096013019, "grad_norm": 18.449526197501573, "learning_rate": 6.706706379973301e-06, "loss": 1.8985, "step": 24900 }, { "epoch": 6.756170328180092, "grad_norm": 17.899848208625517, "learning_rate": 6.703739725661723e-06, "loss": 1.7815, "step": 24910 }, { "epoch": 6.758882560347166, "grad_norm": 18.833634325681285, "learning_rate": 6.700772392618819e-06, "loss": 1.8017, "step": 24920 }, { "epoch": 6.76159479251424, "grad_norm": 18.62634806040012, "learning_rate": 6.697804382026706e-06, "loss": 1.8433, "step": 24930 }, { "epoch": 6.764307024681313, "grad_norm": 18.72244415036686, "learning_rate": 6.694835695067774e-06, "loss": 1.719, "step": 24940 }, { "epoch": 6.7670192568483865, "grad_norm": 16.907101806743896, "learning_rate": 6.691866332924681e-06, "loss": 1.8082, "step": 24950 }, { "epoch": 6.769731489015459, "grad_norm": 18.69070502734929, "learning_rate": 6.688896296780354e-06, "loss": 1.7137, "step": 24960 }, { "epoch": 6.772443721182533, "grad_norm": 17.610732794988156, "learning_rate": 6.685925587817985e-06, "loss": 1.8707, "step": 24970 }, { "epoch": 6.775155953349607, "grad_norm": 19.745345897841897, "learning_rate": 6.682954207221041e-06, "loss": 1.8483, "step": 24980 }, { "epoch": 6.77786818551668, "grad_norm": 17.875590373150292, "learning_rate": 6.679982156173249e-06, "loss": 1.7297, "step": 24990 }, { "epoch": 6.780580417683754, "grad_norm": 19.995389050384862, "learning_rate": 6.67700943585861e-06, "loss": 1.7346, "step": 25000 }, { "epoch": 6.783292649850827, "grad_norm": 20.321368867144106, "learning_rate": 6.674036047461387e-06, "loss": 1.7963, "step": 25010 }, { "epoch": 6.786004882017901, "grad_norm": 18.660316741324717, "learning_rate": 6.6710619921661105e-06, "loss": 1.7161, "step": 25020 }, { "epoch": 6.7887171141849745, "grad_norm": 20.534901226648255, "learning_rate": 6.6680872711575785e-06, "loss": 1.7918, "step": 25030 }, { "epoch": 6.791429346352047, "grad_norm": 19.152202550153948, "learning_rate": 6.6651118856208495e-06, "loss": 1.7032, "step": 25040 }, { "epoch": 6.794141578519121, "grad_norm": 16.186646675047754, "learning_rate": 6.662135836741252e-06, "loss": 1.7722, "step": 25050 }, { "epoch": 6.796853810686195, "grad_norm": 18.201278174838347, "learning_rate": 6.659159125704375e-06, "loss": 1.8457, "step": 25060 }, { "epoch": 6.799566042853268, "grad_norm": 18.647437094733707, "learning_rate": 6.656181753696076e-06, "loss": 1.6664, "step": 25070 }, { "epoch": 6.802278275020342, "grad_norm": 18.966031597712472, "learning_rate": 6.653203721902468e-06, "loss": 1.7915, "step": 25080 }, { "epoch": 6.804990507187416, "grad_norm": 18.430563804619474, "learning_rate": 6.650225031509936e-06, "loss": 1.8832, "step": 25090 }, { "epoch": 6.807702739354489, "grad_norm": 19.009294550585054, "learning_rate": 6.647245683705118e-06, "loss": 1.7816, "step": 25100 }, { "epoch": 6.8104149715215625, "grad_norm": 18.867077109009866, "learning_rate": 6.644265679674922e-06, "loss": 1.808, "step": 25110 }, { "epoch": 6.8131272036886354, "grad_norm": 17.94403973459315, "learning_rate": 6.6412850206065126e-06, "loss": 1.7315, "step": 25120 }, { "epoch": 6.815839435855709, "grad_norm": 17.517706732855306, "learning_rate": 6.638303707687319e-06, "loss": 1.8338, "step": 25130 }, { "epoch": 6.818551668022783, "grad_norm": 18.9361462926492, "learning_rate": 6.6353217421050254e-06, "loss": 1.8125, "step": 25140 }, { "epoch": 6.821263900189856, "grad_norm": 17.959046603991684, "learning_rate": 6.632339125047583e-06, "loss": 1.759, "step": 25150 }, { "epoch": 6.82397613235693, "grad_norm": 18.110943654656616, "learning_rate": 6.629355857703196e-06, "loss": 1.7466, "step": 25160 }, { "epoch": 6.826688364524003, "grad_norm": 17.75707209175116, "learning_rate": 6.6263719412603326e-06, "loss": 1.7866, "step": 25170 }, { "epoch": 6.829400596691077, "grad_norm": 18.715266158722983, "learning_rate": 6.623387376907716e-06, "loss": 1.9688, "step": 25180 }, { "epoch": 6.8321128288581505, "grad_norm": 20.73124098549884, "learning_rate": 6.6204021658343294e-06, "loss": 1.7348, "step": 25190 }, { "epoch": 6.8348250610252235, "grad_norm": 17.44084241057477, "learning_rate": 6.617416309229414e-06, "loss": 1.6478, "step": 25200 }, { "epoch": 6.837537293192297, "grad_norm": 19.189628383030016, "learning_rate": 6.614429808282468e-06, "loss": 1.7666, "step": 25210 }, { "epoch": 6.840249525359371, "grad_norm": 16.461893354694336, "learning_rate": 6.611442664183243e-06, "loss": 1.8507, "step": 25220 }, { "epoch": 6.842961757526444, "grad_norm": 18.81104370454292, "learning_rate": 6.608454878121753e-06, "loss": 1.9244, "step": 25230 }, { "epoch": 6.845673989693518, "grad_norm": 19.498326323197023, "learning_rate": 6.6054664512882615e-06, "loss": 1.8801, "step": 25240 }, { "epoch": 6.848386221860592, "grad_norm": 17.164785612830155, "learning_rate": 6.602477384873292e-06, "loss": 1.8861, "step": 25250 }, { "epoch": 6.851098454027665, "grad_norm": 17.234821774701956, "learning_rate": 6.599487680067618e-06, "loss": 1.7018, "step": 25260 }, { "epoch": 6.8538106861947385, "grad_norm": 21.68009214526514, "learning_rate": 6.596497338062275e-06, "loss": 1.7073, "step": 25270 }, { "epoch": 6.8565229183618115, "grad_norm": 17.00329907451891, "learning_rate": 6.5935063600485424e-06, "loss": 1.7468, "step": 25280 }, { "epoch": 6.859235150528885, "grad_norm": 17.702778354407187, "learning_rate": 6.5905147472179645e-06, "loss": 1.7134, "step": 25290 }, { "epoch": 6.861947382695959, "grad_norm": 18.540844130783253, "learning_rate": 6.5875225007623245e-06, "loss": 1.7722, "step": 25300 }, { "epoch": 6.864659614863032, "grad_norm": 18.65291697512774, "learning_rate": 6.58452962187367e-06, "loss": 1.7857, "step": 25310 }, { "epoch": 6.867371847030106, "grad_norm": 17.52871638887818, "learning_rate": 6.581536111744295e-06, "loss": 1.8223, "step": 25320 }, { "epoch": 6.870084079197179, "grad_norm": 17.540144896602598, "learning_rate": 6.5785419715667486e-06, "loss": 1.7955, "step": 25330 }, { "epoch": 6.872796311364253, "grad_norm": 17.47306761743786, "learning_rate": 6.5755472025338255e-06, "loss": 1.8072, "step": 25340 }, { "epoch": 6.875508543531327, "grad_norm": 19.010529415223164, "learning_rate": 6.572551805838575e-06, "loss": 1.714, "step": 25350 }, { "epoch": 6.8782207756983995, "grad_norm": 17.73666406459359, "learning_rate": 6.5695557826742955e-06, "loss": 1.6819, "step": 25360 }, { "epoch": 6.880933007865473, "grad_norm": 18.336454666591898, "learning_rate": 6.5665591342345326e-06, "loss": 1.8778, "step": 25370 }, { "epoch": 6.883645240032546, "grad_norm": 18.103752295885695, "learning_rate": 6.5635618617130855e-06, "loss": 1.8796, "step": 25380 }, { "epoch": 6.88635747219962, "grad_norm": 18.047198434535282, "learning_rate": 6.560563966304001e-06, "loss": 1.7504, "step": 25390 }, { "epoch": 6.889069704366694, "grad_norm": 18.31100654194271, "learning_rate": 6.557565449201569e-06, "loss": 1.9053, "step": 25400 }, { "epoch": 6.891781936533767, "grad_norm": 18.671000567703224, "learning_rate": 6.554566311600335e-06, "loss": 1.7409, "step": 25410 }, { "epoch": 6.894494168700841, "grad_norm": 18.99997044004748, "learning_rate": 6.551566554695083e-06, "loss": 1.8539, "step": 25420 }, { "epoch": 6.897206400867915, "grad_norm": 16.439815717724702, "learning_rate": 6.5485661796808515e-06, "loss": 1.884, "step": 25430 }, { "epoch": 6.8999186330349875, "grad_norm": 20.504107881022158, "learning_rate": 6.5455651877529205e-06, "loss": 1.6765, "step": 25440 }, { "epoch": 6.902630865202061, "grad_norm": 20.021397143572283, "learning_rate": 6.542563580106819e-06, "loss": 1.7104, "step": 25450 }, { "epoch": 6.905343097369135, "grad_norm": 18.68207492058619, "learning_rate": 6.539561357938316e-06, "loss": 1.8191, "step": 25460 }, { "epoch": 6.908055329536208, "grad_norm": 18.15507118659667, "learning_rate": 6.536558522443434e-06, "loss": 1.8505, "step": 25470 }, { "epoch": 6.910767561703282, "grad_norm": 19.659309575315937, "learning_rate": 6.533555074818429e-06, "loss": 1.8413, "step": 25480 }, { "epoch": 6.913479793870355, "grad_norm": 20.23708885037468, "learning_rate": 6.5305510162598095e-06, "loss": 1.645, "step": 25490 }, { "epoch": 6.916192026037429, "grad_norm": 19.143923532277686, "learning_rate": 6.527546347964325e-06, "loss": 1.7496, "step": 25500 }, { "epoch": 6.918904258204503, "grad_norm": 17.85836129124824, "learning_rate": 6.524541071128967e-06, "loss": 1.8155, "step": 25510 }, { "epoch": 6.921616490371576, "grad_norm": 20.93290196376465, "learning_rate": 6.521535186950968e-06, "loss": 1.733, "step": 25520 }, { "epoch": 6.924328722538649, "grad_norm": 18.363453113033334, "learning_rate": 6.518528696627806e-06, "loss": 1.8783, "step": 25530 }, { "epoch": 6.927040954705722, "grad_norm": 19.50262463931342, "learning_rate": 6.515521601357198e-06, "loss": 1.735, "step": 25540 }, { "epoch": 6.929753186872796, "grad_norm": 17.27315770549501, "learning_rate": 6.512513902337102e-06, "loss": 1.7775, "step": 25550 }, { "epoch": 6.93246541903987, "grad_norm": 17.782978791159643, "learning_rate": 6.509505600765718e-06, "loss": 1.8029, "step": 25560 }, { "epoch": 6.935177651206943, "grad_norm": 17.91723811320076, "learning_rate": 6.506496697841485e-06, "loss": 1.6835, "step": 25570 }, { "epoch": 6.937889883374017, "grad_norm": 18.774769497720587, "learning_rate": 6.503487194763083e-06, "loss": 1.9072, "step": 25580 }, { "epoch": 6.940602115541091, "grad_norm": 21.193259396341244, "learning_rate": 6.500477092729428e-06, "loss": 1.8589, "step": 25590 }, { "epoch": 6.943314347708164, "grad_norm": 21.726238163981865, "learning_rate": 6.497466392939677e-06, "loss": 1.8527, "step": 25600 }, { "epoch": 6.946026579875237, "grad_norm": 19.263457712350792, "learning_rate": 6.494455096593226e-06, "loss": 1.7572, "step": 25610 }, { "epoch": 6.948738812042311, "grad_norm": 18.144998215285806, "learning_rate": 6.491443204889704e-06, "loss": 1.6443, "step": 25620 }, { "epoch": 6.951451044209384, "grad_norm": 20.93095350104292, "learning_rate": 6.488430719028984e-06, "loss": 1.8607, "step": 25630 }, { "epoch": 6.954163276376458, "grad_norm": 18.73720702772335, "learning_rate": 6.48541764021117e-06, "loss": 1.8286, "step": 25640 }, { "epoch": 6.956875508543531, "grad_norm": 18.253924564829457, "learning_rate": 6.482403969636606e-06, "loss": 1.7973, "step": 25650 }, { "epoch": 6.959587740710605, "grad_norm": 19.41658261907089, "learning_rate": 6.479389708505869e-06, "loss": 1.6878, "step": 25660 }, { "epoch": 6.962299972877679, "grad_norm": 18.240355946605295, "learning_rate": 6.476374858019773e-06, "loss": 1.7645, "step": 25670 }, { "epoch": 6.965012205044752, "grad_norm": 18.587422937867196, "learning_rate": 6.473359419379364e-06, "loss": 1.9127, "step": 25680 }, { "epoch": 6.967724437211825, "grad_norm": 17.67455011449157, "learning_rate": 6.470343393785927e-06, "loss": 1.868, "step": 25690 }, { "epoch": 6.970436669378898, "grad_norm": 16.690452383834156, "learning_rate": 6.467326782440978e-06, "loss": 1.7404, "step": 25700 }, { "epoch": 6.973148901545972, "grad_norm": 17.661478923675592, "learning_rate": 6.464309586546267e-06, "loss": 1.8536, "step": 25710 }, { "epoch": 6.975861133713046, "grad_norm": 17.903276286272202, "learning_rate": 6.461291807303773e-06, "loss": 1.7221, "step": 25720 }, { "epoch": 6.978573365880119, "grad_norm": 19.8105060389201, "learning_rate": 6.458273445915717e-06, "loss": 1.8684, "step": 25730 }, { "epoch": 6.981285598047193, "grad_norm": 19.761342909734005, "learning_rate": 6.4552545035845405e-06, "loss": 1.7837, "step": 25740 }, { "epoch": 6.983997830214267, "grad_norm": 17.76802771671153, "learning_rate": 6.452234981512923e-06, "loss": 1.7557, "step": 25750 }, { "epoch": 6.98671006238134, "grad_norm": 18.136399769085454, "learning_rate": 6.449214880903775e-06, "loss": 1.6707, "step": 25760 }, { "epoch": 6.9894222945484135, "grad_norm": 17.131650775829744, "learning_rate": 6.4461942029602365e-06, "loss": 1.8032, "step": 25770 }, { "epoch": 6.992134526715487, "grad_norm": 18.920080706935842, "learning_rate": 6.4431729488856756e-06, "loss": 1.861, "step": 25780 }, { "epoch": 6.99484675888256, "grad_norm": 19.250809004611344, "learning_rate": 6.440151119883693e-06, "loss": 1.8356, "step": 25790 }, { "epoch": 6.997558991049634, "grad_norm": 17.323948755355985, "learning_rate": 6.437128717158114e-06, "loss": 1.9415, "step": 25800 }, { "epoch": 7.000271223216707, "grad_norm": 16.72584167311582, "learning_rate": 6.434105741912999e-06, "loss": 1.7613, "step": 25810 }, { "epoch": 7.002983455383781, "grad_norm": 18.99371457168076, "learning_rate": 6.431082195352632e-06, "loss": 1.2271, "step": 25820 }, { "epoch": 7.005695687550855, "grad_norm": 19.381839173291095, "learning_rate": 6.428058078681525e-06, "loss": 1.2624, "step": 25830 }, { "epoch": 7.008407919717928, "grad_norm": 16.872504428328725, "learning_rate": 6.425033393104416e-06, "loss": 1.3771, "step": 25840 }, { "epoch": 7.0111201518850015, "grad_norm": 21.419502097520308, "learning_rate": 6.422008139826274e-06, "loss": 1.2896, "step": 25850 }, { "epoch": 7.013832384052074, "grad_norm": 18.912606131288392, "learning_rate": 6.418982320052289e-06, "loss": 1.2524, "step": 25860 }, { "epoch": 7.016544616219148, "grad_norm": 23.095533706392736, "learning_rate": 6.415955934987881e-06, "loss": 1.3471, "step": 25870 }, { "epoch": 7.019256848386222, "grad_norm": 19.827580211217498, "learning_rate": 6.412928985838693e-06, "loss": 1.2332, "step": 25880 }, { "epoch": 7.021969080553295, "grad_norm": 16.07060894497139, "learning_rate": 6.409901473810593e-06, "loss": 1.2241, "step": 25890 }, { "epoch": 7.024681312720369, "grad_norm": 21.014044881281926, "learning_rate": 6.40687340010967e-06, "loss": 1.2491, "step": 25900 }, { "epoch": 7.027393544887443, "grad_norm": 15.691763747445954, "learning_rate": 6.4038447659422454e-06, "loss": 1.2909, "step": 25910 }, { "epoch": 7.030105777054516, "grad_norm": 18.910119495230692, "learning_rate": 6.4008155725148545e-06, "loss": 1.2439, "step": 25920 }, { "epoch": 7.0328180092215895, "grad_norm": 18.910739238459342, "learning_rate": 6.397785821034262e-06, "loss": 1.2426, "step": 25930 }, { "epoch": 7.0355302413886625, "grad_norm": 19.914322672710913, "learning_rate": 6.394755512707449e-06, "loss": 1.1381, "step": 25940 }, { "epoch": 7.038242473555736, "grad_norm": 17.103081098818908, "learning_rate": 6.391724648741625e-06, "loss": 1.1873, "step": 25950 }, { "epoch": 7.04095470572281, "grad_norm": 16.500157180228896, "learning_rate": 6.388693230344216e-06, "loss": 1.3058, "step": 25960 }, { "epoch": 7.043666937889883, "grad_norm": 19.054683623989774, "learning_rate": 6.38566125872287e-06, "loss": 1.3466, "step": 25970 }, { "epoch": 7.046379170056957, "grad_norm": 18.220528961254576, "learning_rate": 6.382628735085457e-06, "loss": 1.289, "step": 25980 }, { "epoch": 7.049091402224031, "grad_norm": 19.12425850722809, "learning_rate": 6.379595660640065e-06, "loss": 1.241, "step": 25990 }, { "epoch": 7.051803634391104, "grad_norm": 23.132081115965224, "learning_rate": 6.376562036595003e-06, "loss": 1.2874, "step": 26000 }, { "epoch": 7.0545158665581775, "grad_norm": 21.37238041069356, "learning_rate": 6.373527864158799e-06, "loss": 1.288, "step": 26010 }, { "epoch": 7.0572280987252505, "grad_norm": 18.278122066437497, "learning_rate": 6.370493144540194e-06, "loss": 1.2786, "step": 26020 }, { "epoch": 7.059940330892324, "grad_norm": 18.30404990642974, "learning_rate": 6.367457878948158e-06, "loss": 1.2007, "step": 26030 }, { "epoch": 7.062652563059398, "grad_norm": 20.10245002364024, "learning_rate": 6.364422068591866e-06, "loss": 1.209, "step": 26040 }, { "epoch": 7.065364795226471, "grad_norm": 18.325569100814718, "learning_rate": 6.36138571468072e-06, "loss": 1.2086, "step": 26050 }, { "epoch": 7.068077027393545, "grad_norm": 21.067327895693772, "learning_rate": 6.3583488184243335e-06, "loss": 1.2197, "step": 26060 }, { "epoch": 7.070789259560619, "grad_norm": 17.38369916079102, "learning_rate": 6.3553113810325375e-06, "loss": 1.2462, "step": 26070 }, { "epoch": 7.073501491727692, "grad_norm": 19.440264995171354, "learning_rate": 6.352273403715378e-06, "loss": 1.2447, "step": 26080 }, { "epoch": 7.0762137238947656, "grad_norm": 21.66902927217352, "learning_rate": 6.349234887683117e-06, "loss": 1.2932, "step": 26090 }, { "epoch": 7.0789259560618385, "grad_norm": 19.974070358694057, "learning_rate": 6.346195834146228e-06, "loss": 1.4278, "step": 26100 }, { "epoch": 7.081638188228912, "grad_norm": 19.321419607667757, "learning_rate": 6.343156244315405e-06, "loss": 1.256, "step": 26110 }, { "epoch": 7.084350420395986, "grad_norm": 19.84206284643965, "learning_rate": 6.340116119401547e-06, "loss": 1.2744, "step": 26120 }, { "epoch": 7.087062652563059, "grad_norm": 20.184432241219408, "learning_rate": 6.337075460615776e-06, "loss": 1.2679, "step": 26130 }, { "epoch": 7.089774884730133, "grad_norm": 20.203539963015597, "learning_rate": 6.334034269169419e-06, "loss": 1.2781, "step": 26140 }, { "epoch": 7.092487116897207, "grad_norm": 20.97392405801047, "learning_rate": 6.3309925462740184e-06, "loss": 1.2857, "step": 26150 }, { "epoch": 7.09519934906428, "grad_norm": 18.36954016756257, "learning_rate": 6.327950293141326e-06, "loss": 1.3749, "step": 26160 }, { "epoch": 7.097911581231354, "grad_norm": 17.689882244690825, "learning_rate": 6.32490751098331e-06, "loss": 1.2322, "step": 26170 }, { "epoch": 7.1006238133984265, "grad_norm": 17.570749153171494, "learning_rate": 6.321864201012143e-06, "loss": 1.1896, "step": 26180 }, { "epoch": 7.1033360455655, "grad_norm": 20.088054043307192, "learning_rate": 6.31882036444021e-06, "loss": 1.3002, "step": 26190 }, { "epoch": 7.106048277732574, "grad_norm": 19.27329236174834, "learning_rate": 6.31577600248011e-06, "loss": 1.2416, "step": 26200 }, { "epoch": 7.108760509899647, "grad_norm": 18.152760287150706, "learning_rate": 6.312731116344646e-06, "loss": 1.1721, "step": 26210 }, { "epoch": 7.111472742066721, "grad_norm": 18.14390866819099, "learning_rate": 6.309685707246831e-06, "loss": 1.2163, "step": 26220 }, { "epoch": 7.114184974233795, "grad_norm": 19.672284746533325, "learning_rate": 6.306639776399888e-06, "loss": 1.2775, "step": 26230 }, { "epoch": 7.116897206400868, "grad_norm": 19.50393897087949, "learning_rate": 6.303593325017247e-06, "loss": 1.2926, "step": 26240 }, { "epoch": 7.119609438567942, "grad_norm": 20.40446543940182, "learning_rate": 6.3005463543125444e-06, "loss": 1.2522, "step": 26250 }, { "epoch": 7.1223216707350145, "grad_norm": 16.585140678016952, "learning_rate": 6.297498865499625e-06, "loss": 1.2472, "step": 26260 }, { "epoch": 7.125033902902088, "grad_norm": 17.975272228272733, "learning_rate": 6.294450859792539e-06, "loss": 1.2866, "step": 26270 }, { "epoch": 7.127746135069162, "grad_norm": 18.07707324498092, "learning_rate": 6.2914023384055435e-06, "loss": 1.301, "step": 26280 }, { "epoch": 7.130458367236235, "grad_norm": 18.285264304640734, "learning_rate": 6.2883533025531e-06, "loss": 1.3207, "step": 26290 }, { "epoch": 7.133170599403309, "grad_norm": 17.14955673459912, "learning_rate": 6.285303753449874e-06, "loss": 1.2979, "step": 26300 }, { "epoch": 7.135882831570383, "grad_norm": 17.9733103691796, "learning_rate": 6.282253692310739e-06, "loss": 1.2678, "step": 26310 }, { "epoch": 7.138595063737456, "grad_norm": 19.142715031436346, "learning_rate": 6.2792031203507685e-06, "loss": 1.2992, "step": 26320 }, { "epoch": 7.14130729590453, "grad_norm": 18.032132937519833, "learning_rate": 6.276152038785244e-06, "loss": 1.309, "step": 26330 }, { "epoch": 7.144019528071603, "grad_norm": 19.854157652660945, "learning_rate": 6.273100448829644e-06, "loss": 1.2777, "step": 26340 }, { "epoch": 7.146731760238676, "grad_norm": 20.066650248624356, "learning_rate": 6.270048351699656e-06, "loss": 1.3091, "step": 26350 }, { "epoch": 7.14944399240575, "grad_norm": 20.530494837507643, "learning_rate": 6.266995748611162e-06, "loss": 1.2493, "step": 26360 }, { "epoch": 7.152156224572823, "grad_norm": 19.84878019021024, "learning_rate": 6.263942640780254e-06, "loss": 1.2462, "step": 26370 }, { "epoch": 7.154868456739897, "grad_norm": 19.109899170732245, "learning_rate": 6.260889029423219e-06, "loss": 1.3214, "step": 26380 }, { "epoch": 7.157580688906971, "grad_norm": 21.194976570258245, "learning_rate": 6.257834915756549e-06, "loss": 1.3093, "step": 26390 }, { "epoch": 7.160292921074044, "grad_norm": 19.81204955844538, "learning_rate": 6.25478030099693e-06, "loss": 1.3773, "step": 26400 }, { "epoch": 7.163005153241118, "grad_norm": 24.648058032111944, "learning_rate": 6.251725186361254e-06, "loss": 1.2928, "step": 26410 }, { "epoch": 7.165717385408191, "grad_norm": 20.16017900508635, "learning_rate": 6.248669573066608e-06, "loss": 1.4332, "step": 26420 }, { "epoch": 7.168429617575264, "grad_norm": 20.694339629281234, "learning_rate": 6.245613462330281e-06, "loss": 1.2631, "step": 26430 }, { "epoch": 7.171141849742338, "grad_norm": 21.391749353751347, "learning_rate": 6.242556855369757e-06, "loss": 1.1591, "step": 26440 }, { "epoch": 7.173854081909411, "grad_norm": 19.749906181605976, "learning_rate": 6.239499753402719e-06, "loss": 1.3267, "step": 26450 }, { "epoch": 7.176566314076485, "grad_norm": 20.780174441563084, "learning_rate": 6.236442157647048e-06, "loss": 1.2785, "step": 26460 }, { "epoch": 7.179278546243559, "grad_norm": 19.022987112223134, "learning_rate": 6.2333840693208205e-06, "loss": 1.2847, "step": 26470 }, { "epoch": 7.181990778410632, "grad_norm": 18.282661246503643, "learning_rate": 6.230325489642308e-06, "loss": 1.2705, "step": 26480 }, { "epoch": 7.184703010577706, "grad_norm": 20.281590191524817, "learning_rate": 6.227266419829981e-06, "loss": 1.3869, "step": 26490 }, { "epoch": 7.187415242744779, "grad_norm": 18.530834396128277, "learning_rate": 6.224206861102505e-06, "loss": 1.3108, "step": 26500 }, { "epoch": 7.1901274749118524, "grad_norm": 19.287084052526083, "learning_rate": 6.221146814678737e-06, "loss": 1.2382, "step": 26510 }, { "epoch": 7.192839707078926, "grad_norm": 18.542800179648967, "learning_rate": 6.21808628177773e-06, "loss": 1.2883, "step": 26520 }, { "epoch": 7.195551939245999, "grad_norm": 19.98633856293869, "learning_rate": 6.2150252636187335e-06, "loss": 1.2602, "step": 26530 }, { "epoch": 7.198264171413073, "grad_norm": 20.5202755917175, "learning_rate": 6.211963761421184e-06, "loss": 1.3168, "step": 26540 }, { "epoch": 7.200976403580146, "grad_norm": 20.10151934818811, "learning_rate": 6.208901776404716e-06, "loss": 1.2972, "step": 26550 }, { "epoch": 7.20368863574722, "grad_norm": 18.68961907209884, "learning_rate": 6.205839309789157e-06, "loss": 1.3222, "step": 26560 }, { "epoch": 7.206400867914294, "grad_norm": 19.85993595920866, "learning_rate": 6.202776362794523e-06, "loss": 1.2482, "step": 26570 }, { "epoch": 7.209113100081367, "grad_norm": 18.76168521745466, "learning_rate": 6.199712936641022e-06, "loss": 1.3385, "step": 26580 }, { "epoch": 7.2118253322484405, "grad_norm": 21.1538846710735, "learning_rate": 6.196649032549056e-06, "loss": 1.4177, "step": 26590 }, { "epoch": 7.214537564415514, "grad_norm": 19.409361174982216, "learning_rate": 6.1935846517392105e-06, "loss": 1.2634, "step": 26600 }, { "epoch": 7.217249796582587, "grad_norm": 18.818440524414772, "learning_rate": 6.190519795432271e-06, "loss": 1.2318, "step": 26610 }, { "epoch": 7.219962028749661, "grad_norm": 21.197694052733677, "learning_rate": 6.187454464849201e-06, "loss": 1.3459, "step": 26620 }, { "epoch": 7.222674260916735, "grad_norm": 19.870418197157026, "learning_rate": 6.184388661211165e-06, "loss": 1.3257, "step": 26630 }, { "epoch": 7.225386493083808, "grad_norm": 17.762612144625933, "learning_rate": 6.1813223857395044e-06, "loss": 1.2969, "step": 26640 }, { "epoch": 7.228098725250882, "grad_norm": 22.962118308322346, "learning_rate": 6.178255639655757e-06, "loss": 1.403, "step": 26650 }, { "epoch": 7.230810957417955, "grad_norm": 20.794158141568925, "learning_rate": 6.175188424181644e-06, "loss": 1.3054, "step": 26660 }, { "epoch": 7.2335231895850285, "grad_norm": 17.3119354737425, "learning_rate": 6.172120740539075e-06, "loss": 1.1923, "step": 26670 }, { "epoch": 7.236235421752102, "grad_norm": 18.80509456581159, "learning_rate": 6.169052589950144e-06, "loss": 1.4897, "step": 26680 }, { "epoch": 7.238947653919175, "grad_norm": 19.734748448750178, "learning_rate": 6.165983973637135e-06, "loss": 1.4215, "step": 26690 }, { "epoch": 7.241659886086249, "grad_norm": 23.460791226490226, "learning_rate": 6.162914892822515e-06, "loss": 1.312, "step": 26700 }, { "epoch": 7.244372118253322, "grad_norm": 19.96098827568085, "learning_rate": 6.159845348728936e-06, "loss": 1.3997, "step": 26710 }, { "epoch": 7.247084350420396, "grad_norm": 20.96920370576328, "learning_rate": 6.156775342579233e-06, "loss": 1.3658, "step": 26720 }, { "epoch": 7.24979658258747, "grad_norm": 18.34227476301692, "learning_rate": 6.153704875596429e-06, "loss": 1.2053, "step": 26730 }, { "epoch": 7.252508814754543, "grad_norm": 19.1451188760118, "learning_rate": 6.150633949003728e-06, "loss": 1.3709, "step": 26740 }, { "epoch": 7.2552210469216165, "grad_norm": 20.822246843347664, "learning_rate": 6.147562564024518e-06, "loss": 1.2547, "step": 26750 }, { "epoch": 7.25793327908869, "grad_norm": 20.278867905059904, "learning_rate": 6.144490721882367e-06, "loss": 1.3503, "step": 26760 }, { "epoch": 7.260645511255763, "grad_norm": 19.644254586350076, "learning_rate": 6.141418423801031e-06, "loss": 1.313, "step": 26770 }, { "epoch": 7.263357743422837, "grad_norm": 22.10435334621629, "learning_rate": 6.138345671004442e-06, "loss": 1.2883, "step": 26780 }, { "epoch": 7.266069975589911, "grad_norm": 17.896550385801312, "learning_rate": 6.135272464716715e-06, "loss": 1.3607, "step": 26790 }, { "epoch": 7.268782207756984, "grad_norm": 24.505257330120394, "learning_rate": 6.132198806162146e-06, "loss": 1.249, "step": 26800 }, { "epoch": 7.271494439924058, "grad_norm": 18.712317056756547, "learning_rate": 6.129124696565211e-06, "loss": 1.1808, "step": 26810 }, { "epoch": 7.274206672091131, "grad_norm": 17.09704330207528, "learning_rate": 6.126050137150564e-06, "loss": 1.32, "step": 26820 }, { "epoch": 7.2769189042582045, "grad_norm": 20.478810252275657, "learning_rate": 6.122975129143045e-06, "loss": 1.3183, "step": 26830 }, { "epoch": 7.279631136425278, "grad_norm": 17.383556347867167, "learning_rate": 6.119899673767662e-06, "loss": 1.3844, "step": 26840 }, { "epoch": 7.282343368592351, "grad_norm": 20.036170679785485, "learning_rate": 6.1168237722496095e-06, "loss": 1.4221, "step": 26850 }, { "epoch": 7.285055600759425, "grad_norm": 19.106297392169953, "learning_rate": 6.1137474258142545e-06, "loss": 1.303, "step": 26860 }, { "epoch": 7.287767832926498, "grad_norm": 18.732905071053008, "learning_rate": 6.110670635687148e-06, "loss": 1.2005, "step": 26870 }, { "epoch": 7.290480065093572, "grad_norm": 19.35005951094739, "learning_rate": 6.10759340309401e-06, "loss": 1.2719, "step": 26880 }, { "epoch": 7.293192297260646, "grad_norm": 19.8724133678053, "learning_rate": 6.104515729260742e-06, "loss": 1.386, "step": 26890 }, { "epoch": 7.295904529427719, "grad_norm": 18.53281431159985, "learning_rate": 6.1014376154134185e-06, "loss": 1.3449, "step": 26900 }, { "epoch": 7.298616761594793, "grad_norm": 20.642314926752885, "learning_rate": 6.098359062778291e-06, "loss": 1.3271, "step": 26910 }, { "epoch": 7.301328993761866, "grad_norm": 19.23108987058909, "learning_rate": 6.095280072581784e-06, "loss": 1.3563, "step": 26920 }, { "epoch": 7.304041225928939, "grad_norm": 17.482927776001844, "learning_rate": 6.092200646050499e-06, "loss": 1.2822, "step": 26930 }, { "epoch": 7.306753458096013, "grad_norm": 23.116490679253193, "learning_rate": 6.08912078441121e-06, "loss": 1.3281, "step": 26940 }, { "epoch": 7.309465690263086, "grad_norm": 18.39412647558001, "learning_rate": 6.086040488890863e-06, "loss": 1.2215, "step": 26950 }, { "epoch": 7.31217792243016, "grad_norm": 19.061486315528846, "learning_rate": 6.082959760716578e-06, "loss": 1.3578, "step": 26960 }, { "epoch": 7.314890154597234, "grad_norm": 20.20336613108401, "learning_rate": 6.0798786011156494e-06, "loss": 1.4687, "step": 26970 }, { "epoch": 7.317602386764307, "grad_norm": 16.469742096286517, "learning_rate": 6.076797011315539e-06, "loss": 1.3166, "step": 26980 }, { "epoch": 7.320314618931381, "grad_norm": 20.007329912404582, "learning_rate": 6.073714992543881e-06, "loss": 1.3014, "step": 26990 }, { "epoch": 7.323026851098454, "grad_norm": 25.151754876961196, "learning_rate": 6.070632546028486e-06, "loss": 1.3236, "step": 27000 }, { "epoch": 7.325739083265527, "grad_norm": 19.463518737963867, "learning_rate": 6.0675496729973285e-06, "loss": 1.4558, "step": 27010 }, { "epoch": 7.328451315432601, "grad_norm": 19.83456393471526, "learning_rate": 6.064466374678556e-06, "loss": 1.2785, "step": 27020 }, { "epoch": 7.331163547599674, "grad_norm": 17.544299986970454, "learning_rate": 6.061382652300485e-06, "loss": 1.2809, "step": 27030 }, { "epoch": 7.333875779766748, "grad_norm": 18.876674526080095, "learning_rate": 6.058298507091598e-06, "loss": 1.3693, "step": 27040 }, { "epoch": 7.336588011933822, "grad_norm": 21.249973601986405, "learning_rate": 6.055213940280552e-06, "loss": 1.2735, "step": 27050 }, { "epoch": 7.339300244100895, "grad_norm": 17.395126518585467, "learning_rate": 6.0521289530961665e-06, "loss": 1.2276, "step": 27060 }, { "epoch": 7.342012476267969, "grad_norm": 17.671400243836807, "learning_rate": 6.049043546767432e-06, "loss": 1.326, "step": 27070 }, { "epoch": 7.344724708435042, "grad_norm": 19.065498331414673, "learning_rate": 6.0459577225235044e-06, "loss": 1.2863, "step": 27080 }, { "epoch": 7.347436940602115, "grad_norm": 23.606022396843162, "learning_rate": 6.042871481593706e-06, "loss": 1.3517, "step": 27090 }, { "epoch": 7.350149172769189, "grad_norm": 18.06226715501982, "learning_rate": 6.039784825207523e-06, "loss": 1.3599, "step": 27100 }, { "epoch": 7.352861404936262, "grad_norm": 21.48859830218694, "learning_rate": 6.036697754594614e-06, "loss": 1.4139, "step": 27110 }, { "epoch": 7.355573637103336, "grad_norm": 20.43355463377456, "learning_rate": 6.033610270984792e-06, "loss": 1.2971, "step": 27120 }, { "epoch": 7.35828586927041, "grad_norm": 19.484300752485687, "learning_rate": 6.030522375608046e-06, "loss": 1.3427, "step": 27130 }, { "epoch": 7.360998101437483, "grad_norm": 18.82139855080072, "learning_rate": 6.027434069694521e-06, "loss": 1.305, "step": 27140 }, { "epoch": 7.363710333604557, "grad_norm": 19.84693439096365, "learning_rate": 6.024345354474529e-06, "loss": 1.2422, "step": 27150 }, { "epoch": 7.3664225657716305, "grad_norm": 17.165391937773624, "learning_rate": 6.021256231178544e-06, "loss": 1.4084, "step": 27160 }, { "epoch": 7.369134797938703, "grad_norm": 18.313255896650677, "learning_rate": 6.018166701037202e-06, "loss": 1.3096, "step": 27170 }, { "epoch": 7.371847030105777, "grad_norm": 21.18561689034349, "learning_rate": 6.015076765281301e-06, "loss": 1.2264, "step": 27180 }, { "epoch": 7.37455926227285, "grad_norm": 19.66418770486974, "learning_rate": 6.011986425141806e-06, "loss": 1.2708, "step": 27190 }, { "epoch": 7.377271494439924, "grad_norm": 20.58574754458175, "learning_rate": 6.008895681849833e-06, "loss": 1.2743, "step": 27200 }, { "epoch": 7.379983726606998, "grad_norm": 20.848186048266605, "learning_rate": 6.0058045366366695e-06, "loss": 1.3171, "step": 27210 }, { "epoch": 7.382695958774071, "grad_norm": 20.896160496670685, "learning_rate": 6.0027129907337525e-06, "loss": 1.4193, "step": 27220 }, { "epoch": 7.385408190941145, "grad_norm": 20.37225539642399, "learning_rate": 5.999621045372688e-06, "loss": 1.4611, "step": 27230 }, { "epoch": 7.388120423108218, "grad_norm": 18.20089327597698, "learning_rate": 5.9965287017852325e-06, "loss": 1.3879, "step": 27240 }, { "epoch": 7.390832655275291, "grad_norm": 18.591999319282127, "learning_rate": 5.99343596120331e-06, "loss": 1.3477, "step": 27250 }, { "epoch": 7.393544887442365, "grad_norm": 18.02406429098716, "learning_rate": 5.990342824858996e-06, "loss": 1.2851, "step": 27260 }, { "epoch": 7.396257119609438, "grad_norm": 21.90306079579659, "learning_rate": 5.987249293984528e-06, "loss": 1.3513, "step": 27270 }, { "epoch": 7.398969351776512, "grad_norm": 19.82812025087207, "learning_rate": 5.984155369812296e-06, "loss": 1.4446, "step": 27280 }, { "epoch": 7.401681583943586, "grad_norm": 21.443856561324107, "learning_rate": 5.981061053574851e-06, "loss": 1.2702, "step": 27290 }, { "epoch": 7.404393816110659, "grad_norm": 20.787717907750594, "learning_rate": 5.9779663465048984e-06, "loss": 1.3151, "step": 27300 }, { "epoch": 7.407106048277733, "grad_norm": 21.682387989404415, "learning_rate": 5.974871249835299e-06, "loss": 1.337, "step": 27310 }, { "epoch": 7.4098182804448065, "grad_norm": 20.762116777083403, "learning_rate": 5.971775764799068e-06, "loss": 1.3578, "step": 27320 }, { "epoch": 7.4125305126118795, "grad_norm": 21.937815094335374, "learning_rate": 5.968679892629381e-06, "loss": 1.3126, "step": 27330 }, { "epoch": 7.415242744778953, "grad_norm": 19.48508606382215, "learning_rate": 5.965583634559558e-06, "loss": 1.2864, "step": 27340 }, { "epoch": 7.417954976946026, "grad_norm": 19.106628502428997, "learning_rate": 5.962486991823082e-06, "loss": 1.3584, "step": 27350 }, { "epoch": 7.4206672091131, "grad_norm": 21.34275539279353, "learning_rate": 5.9593899656535825e-06, "loss": 1.3058, "step": 27360 }, { "epoch": 7.423379441280174, "grad_norm": 17.55855046931082, "learning_rate": 5.956292557284846e-06, "loss": 1.2966, "step": 27370 }, { "epoch": 7.426091673447247, "grad_norm": 20.289039010320106, "learning_rate": 5.95319476795081e-06, "loss": 1.374, "step": 27380 }, { "epoch": 7.428803905614321, "grad_norm": 21.495289778126402, "learning_rate": 5.950096598885564e-06, "loss": 1.3342, "step": 27390 }, { "epoch": 7.431516137781394, "grad_norm": 21.72284191214462, "learning_rate": 5.946998051323346e-06, "loss": 1.339, "step": 27400 }, { "epoch": 7.4342283699484675, "grad_norm": 19.580535496342673, "learning_rate": 5.943899126498551e-06, "loss": 1.3165, "step": 27410 }, { "epoch": 7.436940602115541, "grad_norm": 20.868268266736, "learning_rate": 5.940799825645717e-06, "loss": 1.2628, "step": 27420 }, { "epoch": 7.439652834282614, "grad_norm": 20.362514198052555, "learning_rate": 5.9377001499995365e-06, "loss": 1.3853, "step": 27430 }, { "epoch": 7.442365066449688, "grad_norm": 19.366237179104665, "learning_rate": 5.9346001007948505e-06, "loss": 1.3411, "step": 27440 }, { "epoch": 7.445077298616762, "grad_norm": 19.137609459634326, "learning_rate": 5.9314996792666494e-06, "loss": 1.3178, "step": 27450 }, { "epoch": 7.447789530783835, "grad_norm": 20.415701721235042, "learning_rate": 5.928398886650068e-06, "loss": 1.2736, "step": 27460 }, { "epoch": 7.450501762950909, "grad_norm": 18.87839739492888, "learning_rate": 5.925297724180396e-06, "loss": 1.386, "step": 27470 }, { "epoch": 7.4532139951179825, "grad_norm": 19.871609753865066, "learning_rate": 5.922196193093062e-06, "loss": 1.337, "step": 27480 }, { "epoch": 7.4559262272850555, "grad_norm": 18.24789693546917, "learning_rate": 5.919094294623646e-06, "loss": 1.2747, "step": 27490 }, { "epoch": 7.458638459452129, "grad_norm": 23.183846832733007, "learning_rate": 5.915992030007877e-06, "loss": 1.2387, "step": 27500 }, { "epoch": 7.461350691619202, "grad_norm": 19.558635522797218, "learning_rate": 5.9128894004816274e-06, "loss": 1.3955, "step": 27510 }, { "epoch": 7.464062923786276, "grad_norm": 19.67450450302727, "learning_rate": 5.909786407280911e-06, "loss": 1.3393, "step": 27520 }, { "epoch": 7.46677515595335, "grad_norm": 18.579460583359932, "learning_rate": 5.906683051641894e-06, "loss": 1.3587, "step": 27530 }, { "epoch": 7.469487388120423, "grad_norm": 21.824216613322854, "learning_rate": 5.9035793348008784e-06, "loss": 1.3909, "step": 27540 }, { "epoch": 7.472199620287497, "grad_norm": 20.940686903657305, "learning_rate": 5.9004752579943184e-06, "loss": 1.3002, "step": 27550 }, { "epoch": 7.47491185245457, "grad_norm": 19.94646412008084, "learning_rate": 5.8973708224588055e-06, "loss": 1.2639, "step": 27560 }, { "epoch": 7.4776240846216435, "grad_norm": 17.951944780360716, "learning_rate": 5.89426602943108e-06, "loss": 1.2745, "step": 27570 }, { "epoch": 7.480336316788717, "grad_norm": 18.383380098980332, "learning_rate": 5.891160880148016e-06, "loss": 1.3777, "step": 27580 }, { "epoch": 7.48304854895579, "grad_norm": 19.563974228304097, "learning_rate": 5.88805537584664e-06, "loss": 1.2714, "step": 27590 }, { "epoch": 7.485760781122864, "grad_norm": 17.963794974335613, "learning_rate": 5.88494951776411e-06, "loss": 1.3567, "step": 27600 }, { "epoch": 7.488473013289938, "grad_norm": 21.840427419961998, "learning_rate": 5.881843307137733e-06, "loss": 1.3604, "step": 27610 }, { "epoch": 7.491185245457011, "grad_norm": 18.70686199657987, "learning_rate": 5.87873674520495e-06, "loss": 1.3272, "step": 27620 }, { "epoch": 7.493897477624085, "grad_norm": 20.472388348635608, "learning_rate": 5.875629833203348e-06, "loss": 1.3561, "step": 27630 }, { "epoch": 7.496609709791159, "grad_norm": 23.499706552369922, "learning_rate": 5.872522572370648e-06, "loss": 1.3946, "step": 27640 }, { "epoch": 7.4993219419582315, "grad_norm": 21.080679994408083, "learning_rate": 5.869414963944715e-06, "loss": 1.2879, "step": 27650 }, { "epoch": 7.502034174125305, "grad_norm": 19.447038524908386, "learning_rate": 5.866307009163546e-06, "loss": 1.4004, "step": 27660 }, { "epoch": 7.504746406292378, "grad_norm": 19.038938742101518, "learning_rate": 5.8631987092652834e-06, "loss": 1.3163, "step": 27670 }, { "epoch": 7.507458638459452, "grad_norm": 18.342395797798883, "learning_rate": 5.8600900654882e-06, "loss": 1.2587, "step": 27680 }, { "epoch": 7.510170870626526, "grad_norm": 17.744361328707843, "learning_rate": 5.856981079070712e-06, "loss": 1.3208, "step": 27690 }, { "epoch": 7.512883102793599, "grad_norm": 20.5471852879858, "learning_rate": 5.85387175125137e-06, "loss": 1.2694, "step": 27700 }, { "epoch": 7.515595334960673, "grad_norm": 19.538399227998983, "learning_rate": 5.8507620832688565e-06, "loss": 1.3993, "step": 27710 }, { "epoch": 7.518307567127746, "grad_norm": 19.206367108234353, "learning_rate": 5.847652076361995e-06, "loss": 1.3091, "step": 27720 }, { "epoch": 7.52101979929482, "grad_norm": 19.019797839717246, "learning_rate": 5.84454173176974e-06, "loss": 1.3994, "step": 27730 }, { "epoch": 7.523732031461893, "grad_norm": 18.530409076737847, "learning_rate": 5.8414310507311845e-06, "loss": 1.339, "step": 27740 }, { "epoch": 7.526444263628966, "grad_norm": 20.66031436034748, "learning_rate": 5.838320034485551e-06, "loss": 1.292, "step": 27750 }, { "epoch": 7.52915649579604, "grad_norm": 20.188559910815123, "learning_rate": 5.835208684272201e-06, "loss": 1.2687, "step": 27760 }, { "epoch": 7.531868727963114, "grad_norm": 19.18315088066107, "learning_rate": 5.832097001330623e-06, "loss": 1.2709, "step": 27770 }, { "epoch": 7.534580960130187, "grad_norm": 17.35646954300402, "learning_rate": 5.828984986900444e-06, "loss": 1.3315, "step": 27780 }, { "epoch": 7.537293192297261, "grad_norm": 20.5303812793052, "learning_rate": 5.825872642221417e-06, "loss": 1.2301, "step": 27790 }, { "epoch": 7.540005424464335, "grad_norm": 21.288244576092595, "learning_rate": 5.8227599685334326e-06, "loss": 1.4412, "step": 27800 }, { "epoch": 7.542717656631408, "grad_norm": 19.886718313308656, "learning_rate": 5.8196469670765065e-06, "loss": 1.3178, "step": 27810 }, { "epoch": 7.545429888798481, "grad_norm": 23.144132232848047, "learning_rate": 5.816533639090789e-06, "loss": 1.2977, "step": 27820 }, { "epoch": 7.548142120965554, "grad_norm": 17.62257789119255, "learning_rate": 5.813419985816561e-06, "loss": 1.2421, "step": 27830 }, { "epoch": 7.550854353132628, "grad_norm": 20.440443651526095, "learning_rate": 5.810306008494231e-06, "loss": 1.4205, "step": 27840 }, { "epoch": 7.553566585299702, "grad_norm": 19.259465966378574, "learning_rate": 5.807191708364335e-06, "loss": 1.2918, "step": 27850 }, { "epoch": 7.556278817466775, "grad_norm": 21.894764643463933, "learning_rate": 5.804077086667543e-06, "loss": 1.362, "step": 27860 }, { "epoch": 7.558991049633849, "grad_norm": 19.348373540757763, "learning_rate": 5.800962144644645e-06, "loss": 1.2647, "step": 27870 }, { "epoch": 7.561703281800922, "grad_norm": 19.67943098655043, "learning_rate": 5.797846883536566e-06, "loss": 1.3221, "step": 27880 }, { "epoch": 7.564415513967996, "grad_norm": 22.44730975339954, "learning_rate": 5.794731304584358e-06, "loss": 1.3394, "step": 27890 }, { "epoch": 7.567127746135069, "grad_norm": 18.78320221463498, "learning_rate": 5.791615409029194e-06, "loss": 1.246, "step": 27900 }, { "epoch": 7.569839978302142, "grad_norm": 18.640165500080858, "learning_rate": 5.788499198112377e-06, "loss": 1.3035, "step": 27910 }, { "epoch": 7.572552210469216, "grad_norm": 19.61641633306692, "learning_rate": 5.785382673075336e-06, "loss": 1.3473, "step": 27920 }, { "epoch": 7.575264442636289, "grad_norm": 18.574220248699458, "learning_rate": 5.782265835159622e-06, "loss": 1.3576, "step": 27930 }, { "epoch": 7.577976674803363, "grad_norm": 21.51726843536977, "learning_rate": 5.779148685606912e-06, "loss": 1.3261, "step": 27940 }, { "epoch": 7.580688906970437, "grad_norm": 19.273676312428737, "learning_rate": 5.77603122565901e-06, "loss": 1.326, "step": 27950 }, { "epoch": 7.58340113913751, "grad_norm": 21.915502878559078, "learning_rate": 5.772913456557842e-06, "loss": 1.3479, "step": 27960 }, { "epoch": 7.586113371304584, "grad_norm": 19.693812088603146, "learning_rate": 5.769795379545454e-06, "loss": 1.3745, "step": 27970 }, { "epoch": 7.5888256034716575, "grad_norm": 20.619193087438248, "learning_rate": 5.76667699586402e-06, "loss": 1.4316, "step": 27980 }, { "epoch": 7.59153783563873, "grad_norm": 17.895276097201105, "learning_rate": 5.763558306755832e-06, "loss": 1.3154, "step": 27990 }, { "epoch": 7.594250067805804, "grad_norm": 19.156050740785478, "learning_rate": 5.760439313463304e-06, "loss": 1.3143, "step": 28000 }, { "epoch": 7.596962299972878, "grad_norm": 20.120199321869904, "learning_rate": 5.757320017228973e-06, "loss": 1.3449, "step": 28010 }, { "epoch": 7.599674532139951, "grad_norm": 22.29062463054425, "learning_rate": 5.754200419295499e-06, "loss": 1.3296, "step": 28020 }, { "epoch": 7.602386764307025, "grad_norm": 20.49270684497802, "learning_rate": 5.751080520905654e-06, "loss": 1.3511, "step": 28030 }, { "epoch": 7.605098996474098, "grad_norm": 20.023778324851584, "learning_rate": 5.74796032330234e-06, "loss": 1.3294, "step": 28040 }, { "epoch": 7.607811228641172, "grad_norm": 22.86633636812581, "learning_rate": 5.744839827728569e-06, "loss": 1.3151, "step": 28050 }, { "epoch": 7.6105234608082455, "grad_norm": 19.496060900351697, "learning_rate": 5.741719035427477e-06, "loss": 1.289, "step": 28060 }, { "epoch": 7.613235692975318, "grad_norm": 17.22571078356128, "learning_rate": 5.7385979476423185e-06, "loss": 1.2898, "step": 28070 }, { "epoch": 7.615947925142392, "grad_norm": 19.685749241588596, "learning_rate": 5.735476565616462e-06, "loss": 1.4507, "step": 28080 }, { "epoch": 7.618660157309465, "grad_norm": 20.281340755621795, "learning_rate": 5.732354890593398e-06, "loss": 1.2776, "step": 28090 }, { "epoch": 7.621372389476539, "grad_norm": 20.181591400857698, "learning_rate": 5.729232923816729e-06, "loss": 1.393, "step": 28100 }, { "epoch": 7.624084621643613, "grad_norm": 19.799135250570814, "learning_rate": 5.726110666530176e-06, "loss": 1.2333, "step": 28110 }, { "epoch": 7.626796853810686, "grad_norm": 17.57832783245684, "learning_rate": 5.722988119977575e-06, "loss": 1.2666, "step": 28120 }, { "epoch": 7.62950908597776, "grad_norm": 19.653463201753553, "learning_rate": 5.7198652854028805e-06, "loss": 1.3468, "step": 28130 }, { "epoch": 7.6322213181448335, "grad_norm": 19.50686909262883, "learning_rate": 5.716742164050158e-06, "loss": 1.505, "step": 28140 }, { "epoch": 7.6349335503119065, "grad_norm": 22.082110199422736, "learning_rate": 5.713618757163586e-06, "loss": 1.3914, "step": 28150 }, { "epoch": 7.63764578247898, "grad_norm": 18.52788150738289, "learning_rate": 5.7104950659874625e-06, "loss": 1.3742, "step": 28160 }, { "epoch": 7.640358014646054, "grad_norm": 22.191587289002616, "learning_rate": 5.707371091766193e-06, "loss": 1.3242, "step": 28170 }, { "epoch": 7.643070246813127, "grad_norm": 19.82496544150196, "learning_rate": 5.704246835744298e-06, "loss": 1.3101, "step": 28180 }, { "epoch": 7.645782478980201, "grad_norm": 20.077769927835487, "learning_rate": 5.7011222991664115e-06, "loss": 1.2611, "step": 28190 }, { "epoch": 7.648494711147274, "grad_norm": 17.398742365517084, "learning_rate": 5.6979974832772786e-06, "loss": 1.2093, "step": 28200 }, { "epoch": 7.651206943314348, "grad_norm": 20.18074487007673, "learning_rate": 5.694872389321754e-06, "loss": 1.3079, "step": 28210 }, { "epoch": 7.6539191754814215, "grad_norm": 19.148488781803245, "learning_rate": 5.691747018544805e-06, "loss": 1.4257, "step": 28220 }, { "epoch": 7.6566314076484945, "grad_norm": 19.821430823843325, "learning_rate": 5.688621372191508e-06, "loss": 1.4147, "step": 28230 }, { "epoch": 7.659343639815568, "grad_norm": 21.269455338045834, "learning_rate": 5.6854954515070514e-06, "loss": 1.2781, "step": 28240 }, { "epoch": 7.662055871982641, "grad_norm": 22.974313140238714, "learning_rate": 5.682369257736726e-06, "loss": 1.4317, "step": 28250 }, { "epoch": 7.664768104149715, "grad_norm": 20.53944105441944, "learning_rate": 5.6792427921259445e-06, "loss": 1.381, "step": 28260 }, { "epoch": 7.667480336316789, "grad_norm": 21.73942612795251, "learning_rate": 5.676116055920213e-06, "loss": 1.3512, "step": 28270 }, { "epoch": 7.670192568483862, "grad_norm": 19.65484441016435, "learning_rate": 5.672989050365157e-06, "loss": 1.3481, "step": 28280 }, { "epoch": 7.672904800650936, "grad_norm": 18.704877039537198, "learning_rate": 5.669861776706503e-06, "loss": 1.3817, "step": 28290 }, { "epoch": 7.6756170328180096, "grad_norm": 20.42837513574169, "learning_rate": 5.6667342361900865e-06, "loss": 1.2944, "step": 28300 }, { "epoch": 7.6783292649850825, "grad_norm": 19.069497000201874, "learning_rate": 5.663606430061848e-06, "loss": 1.3534, "step": 28310 }, { "epoch": 7.681041497152156, "grad_norm": 21.53839683916989, "learning_rate": 5.660478359567833e-06, "loss": 1.35, "step": 28320 }, { "epoch": 7.68375372931923, "grad_norm": 18.50692384019326, "learning_rate": 5.657350025954197e-06, "loss": 1.4171, "step": 28330 }, { "epoch": 7.686465961486303, "grad_norm": 20.412315174402615, "learning_rate": 5.654221430467197e-06, "loss": 1.3927, "step": 28340 }, { "epoch": 7.689178193653377, "grad_norm": 18.47625501043882, "learning_rate": 5.6510925743531945e-06, "loss": 1.4623, "step": 28350 }, { "epoch": 7.69189042582045, "grad_norm": 21.051878338941204, "learning_rate": 5.647963458858653e-06, "loss": 1.2609, "step": 28360 }, { "epoch": 7.694602657987524, "grad_norm": 20.81521163073713, "learning_rate": 5.644834085230142e-06, "loss": 1.3748, "step": 28370 }, { "epoch": 7.697314890154598, "grad_norm": 19.71341419745233, "learning_rate": 5.641704454714334e-06, "loss": 1.2155, "step": 28380 }, { "epoch": 7.7000271223216705, "grad_norm": 18.952689299862396, "learning_rate": 5.638574568558002e-06, "loss": 1.4045, "step": 28390 }, { "epoch": 7.702739354488744, "grad_norm": 17.575609017350114, "learning_rate": 5.635444428008023e-06, "loss": 1.4267, "step": 28400 }, { "epoch": 7.705451586655817, "grad_norm": 19.704294829685676, "learning_rate": 5.632314034311374e-06, "loss": 1.3152, "step": 28410 }, { "epoch": 7.708163818822891, "grad_norm": 17.945003990656783, "learning_rate": 5.629183388715133e-06, "loss": 1.3219, "step": 28420 }, { "epoch": 7.710876050989965, "grad_norm": 20.234346063469566, "learning_rate": 5.626052492466475e-06, "loss": 1.3585, "step": 28430 }, { "epoch": 7.713588283157038, "grad_norm": 18.592883187585837, "learning_rate": 5.622921346812682e-06, "loss": 1.4761, "step": 28440 }, { "epoch": 7.716300515324112, "grad_norm": 20.018167229297422, "learning_rate": 5.61978995300113e-06, "loss": 1.3526, "step": 28450 }, { "epoch": 7.719012747491186, "grad_norm": 22.460512603986547, "learning_rate": 5.616658312279298e-06, "loss": 1.3764, "step": 28460 }, { "epoch": 7.7217249796582585, "grad_norm": 19.020754934312397, "learning_rate": 5.613526425894755e-06, "loss": 1.3686, "step": 28470 }, { "epoch": 7.724437211825332, "grad_norm": 18.35441253569414, "learning_rate": 5.610394295095179e-06, "loss": 1.2256, "step": 28480 }, { "epoch": 7.727149443992406, "grad_norm": 20.93623448896858, "learning_rate": 5.607261921128339e-06, "loss": 1.3126, "step": 28490 }, { "epoch": 7.729861676159479, "grad_norm": 18.262764227744753, "learning_rate": 5.604129305242098e-06, "loss": 1.2661, "step": 28500 }, { "epoch": 7.732573908326553, "grad_norm": 18.358837830186697, "learning_rate": 5.600996448684425e-06, "loss": 1.3604, "step": 28510 }, { "epoch": 7.735286140493626, "grad_norm": 21.54562443256278, "learning_rate": 5.597863352703376e-06, "loss": 1.3231, "step": 28520 }, { "epoch": 7.7379983726607, "grad_norm": 22.165148829083662, "learning_rate": 5.594730018547104e-06, "loss": 1.3474, "step": 28530 }, { "epoch": 7.740710604827774, "grad_norm": 20.01068549098803, "learning_rate": 5.591596447463862e-06, "loss": 1.334, "step": 28540 }, { "epoch": 7.743422836994847, "grad_norm": 20.879258538491573, "learning_rate": 5.58846264070199e-06, "loss": 1.4987, "step": 28550 }, { "epoch": 7.74613506916192, "grad_norm": 18.908067468778654, "learning_rate": 5.58532859950993e-06, "loss": 1.3538, "step": 28560 }, { "epoch": 7.748847301328993, "grad_norm": 21.78836385147034, "learning_rate": 5.582194325136207e-06, "loss": 1.3656, "step": 28570 }, { "epoch": 7.751559533496067, "grad_norm": 21.821886259546087, "learning_rate": 5.579059818829452e-06, "loss": 1.4112, "step": 28580 }, { "epoch": 7.754271765663141, "grad_norm": 21.83161892886313, "learning_rate": 5.575925081838377e-06, "loss": 1.265, "step": 28590 }, { "epoch": 7.756983997830214, "grad_norm": 25.17721804030395, "learning_rate": 5.572790115411791e-06, "loss": 1.2892, "step": 28600 }, { "epoch": 7.759696229997288, "grad_norm": 19.859814363532216, "learning_rate": 5.569654920798593e-06, "loss": 1.2003, "step": 28610 }, { "epoch": 7.762408462164362, "grad_norm": 18.921344644649082, "learning_rate": 5.566519499247775e-06, "loss": 1.4463, "step": 28620 }, { "epoch": 7.765120694331435, "grad_norm": 20.047350177238695, "learning_rate": 5.563383852008416e-06, "loss": 1.442, "step": 28630 }, { "epoch": 7.767832926498508, "grad_norm": 23.48921024884581, "learning_rate": 5.5602479803296895e-06, "loss": 1.5002, "step": 28640 }, { "epoch": 7.770545158665582, "grad_norm": 20.849364848358974, "learning_rate": 5.557111885460853e-06, "loss": 1.3561, "step": 28650 }, { "epoch": 7.773257390832655, "grad_norm": 21.163029953701457, "learning_rate": 5.553975568651259e-06, "loss": 1.4019, "step": 28660 }, { "epoch": 7.775969622999729, "grad_norm": 18.904209529756365, "learning_rate": 5.550839031150342e-06, "loss": 1.4375, "step": 28670 }, { "epoch": 7.778681855166802, "grad_norm": 20.482025221817732, "learning_rate": 5.547702274207628e-06, "loss": 1.3749, "step": 28680 }, { "epoch": 7.781394087333876, "grad_norm": 21.198442416161484, "learning_rate": 5.54456529907273e-06, "loss": 1.4116, "step": 28690 }, { "epoch": 7.78410631950095, "grad_norm": 21.012071131331155, "learning_rate": 5.54142810699535e-06, "loss": 1.2781, "step": 28700 }, { "epoch": 7.786818551668023, "grad_norm": 19.06061116594558, "learning_rate": 5.538290699225273e-06, "loss": 1.2889, "step": 28710 }, { "epoch": 7.7895307838350965, "grad_norm": 19.485634978072135, "learning_rate": 5.53515307701237e-06, "loss": 1.3676, "step": 28720 }, { "epoch": 7.792243016002169, "grad_norm": 18.732810683983367, "learning_rate": 5.5320152416065995e-06, "loss": 1.4404, "step": 28730 }, { "epoch": 7.794955248169243, "grad_norm": 20.064078733607612, "learning_rate": 5.528877194258004e-06, "loss": 1.2382, "step": 28740 }, { "epoch": 7.797667480336317, "grad_norm": 19.240660488163506, "learning_rate": 5.52573893621671e-06, "loss": 1.3311, "step": 28750 }, { "epoch": 7.80037971250339, "grad_norm": 20.813921861948057, "learning_rate": 5.522600468732931e-06, "loss": 1.2876, "step": 28760 }, { "epoch": 7.803091944670464, "grad_norm": 18.470003173673295, "learning_rate": 5.519461793056958e-06, "loss": 1.5009, "step": 28770 }, { "epoch": 7.805804176837537, "grad_norm": 19.471515449991625, "learning_rate": 5.51632291043917e-06, "loss": 1.4127, "step": 28780 }, { "epoch": 7.808516409004611, "grad_norm": 21.07388172187542, "learning_rate": 5.513183822130027e-06, "loss": 1.2991, "step": 28790 }, { "epoch": 7.8112286411716845, "grad_norm": 21.718428855795132, "learning_rate": 5.510044529380071e-06, "loss": 1.4229, "step": 28800 }, { "epoch": 7.813940873338757, "grad_norm": 19.320745970011636, "learning_rate": 5.506905033439922e-06, "loss": 1.2718, "step": 28810 }, { "epoch": 7.816653105505831, "grad_norm": 20.63753020958724, "learning_rate": 5.503765335560287e-06, "loss": 1.3447, "step": 28820 }, { "epoch": 7.819365337672905, "grad_norm": 22.708411642864146, "learning_rate": 5.500625436991951e-06, "loss": 1.2639, "step": 28830 }, { "epoch": 7.822077569839978, "grad_norm": 19.269487277285396, "learning_rate": 5.497485338985777e-06, "loss": 1.4656, "step": 28840 }, { "epoch": 7.824789802007052, "grad_norm": 21.91797026290559, "learning_rate": 5.494345042792709e-06, "loss": 1.3551, "step": 28850 }, { "epoch": 7.827502034174126, "grad_norm": 19.658996150647507, "learning_rate": 5.491204549663771e-06, "loss": 1.4264, "step": 28860 }, { "epoch": 7.830214266341199, "grad_norm": 22.13540687345259, "learning_rate": 5.488063860850063e-06, "loss": 1.3076, "step": 28870 }, { "epoch": 7.8329264985082725, "grad_norm": 19.46366903783692, "learning_rate": 5.484922977602766e-06, "loss": 1.339, "step": 28880 }, { "epoch": 7.835638730675345, "grad_norm": 21.362983389330825, "learning_rate": 5.481781901173134e-06, "loss": 1.3308, "step": 28890 }, { "epoch": 7.838350962842419, "grad_norm": 19.24289686397816, "learning_rate": 5.478640632812504e-06, "loss": 1.2946, "step": 28900 }, { "epoch": 7.841063195009493, "grad_norm": 20.869962980890314, "learning_rate": 5.475499173772283e-06, "loss": 1.3931, "step": 28910 }, { "epoch": 7.843775427176566, "grad_norm": 18.355356846007187, "learning_rate": 5.472357525303961e-06, "loss": 1.4218, "step": 28920 }, { "epoch": 7.84648765934364, "grad_norm": 18.11943317159349, "learning_rate": 5.469215688659095e-06, "loss": 1.3926, "step": 28930 }, { "epoch": 7.849199891510713, "grad_norm": 19.333506315354082, "learning_rate": 5.466073665089325e-06, "loss": 1.3529, "step": 28940 }, { "epoch": 7.851912123677787, "grad_norm": 21.17043050019844, "learning_rate": 5.4629314558463605e-06, "loss": 1.4094, "step": 28950 }, { "epoch": 7.8546243558448605, "grad_norm": 20.158914888907944, "learning_rate": 5.4597890621819886e-06, "loss": 1.3524, "step": 28960 }, { "epoch": 7.8573365880119335, "grad_norm": 21.186219002047437, "learning_rate": 5.456646485348064e-06, "loss": 1.3814, "step": 28970 }, { "epoch": 7.860048820179007, "grad_norm": 21.063480525030332, "learning_rate": 5.453503726596522e-06, "loss": 1.3982, "step": 28980 }, { "epoch": 7.862761052346081, "grad_norm": 19.598782874304874, "learning_rate": 5.450360787179364e-06, "loss": 1.3986, "step": 28990 }, { "epoch": 7.865473284513154, "grad_norm": 19.942390363918424, "learning_rate": 5.447217668348667e-06, "loss": 1.4209, "step": 29000 }, { "epoch": 7.868185516680228, "grad_norm": 23.707602950543933, "learning_rate": 5.444074371356577e-06, "loss": 1.3151, "step": 29010 }, { "epoch": 7.870897748847302, "grad_norm": 18.99234950878351, "learning_rate": 5.4409308974553145e-06, "loss": 1.3504, "step": 29020 }, { "epoch": 7.873609981014375, "grad_norm": 18.02599394947115, "learning_rate": 5.437787247897167e-06, "loss": 1.3667, "step": 29030 }, { "epoch": 7.8763222131814485, "grad_norm": 21.57928404887783, "learning_rate": 5.4346434239344934e-06, "loss": 1.3145, "step": 29040 }, { "epoch": 7.8790344453485215, "grad_norm": 20.885789767013737, "learning_rate": 5.431499426819722e-06, "loss": 1.4121, "step": 29050 }, { "epoch": 7.881746677515595, "grad_norm": 20.54630188308527, "learning_rate": 5.42835525780535e-06, "loss": 1.3642, "step": 29060 }, { "epoch": 7.884458909682669, "grad_norm": 21.86138630685894, "learning_rate": 5.425210918143944e-06, "loss": 1.3099, "step": 29070 }, { "epoch": 7.887171141849742, "grad_norm": 22.426448154640852, "learning_rate": 5.422066409088137e-06, "loss": 1.3005, "step": 29080 }, { "epoch": 7.889883374016816, "grad_norm": 19.763506344659856, "learning_rate": 5.41892173189063e-06, "loss": 1.3317, "step": 29090 }, { "epoch": 7.892595606183889, "grad_norm": 19.95226001216523, "learning_rate": 5.415776887804191e-06, "loss": 1.3332, "step": 29100 }, { "epoch": 7.895307838350963, "grad_norm": 20.42064763008763, "learning_rate": 5.412631878081656e-06, "loss": 1.4804, "step": 29110 }, { "epoch": 7.898020070518037, "grad_norm": 21.0211426308731, "learning_rate": 5.409486703975924e-06, "loss": 1.2641, "step": 29120 }, { "epoch": 7.9007323026851095, "grad_norm": 20.82605202934838, "learning_rate": 5.406341366739961e-06, "loss": 1.4054, "step": 29130 }, { "epoch": 7.903444534852183, "grad_norm": 18.7901279749451, "learning_rate": 5.4031958676268016e-06, "loss": 1.3319, "step": 29140 }, { "epoch": 7.906156767019257, "grad_norm": 20.86550720987806, "learning_rate": 5.400050207889537e-06, "loss": 1.3691, "step": 29150 }, { "epoch": 7.90886899918633, "grad_norm": 20.10186305531796, "learning_rate": 5.396904388781329e-06, "loss": 1.3229, "step": 29160 }, { "epoch": 7.911581231353404, "grad_norm": 18.620017292863825, "learning_rate": 5.3937584115553994e-06, "loss": 1.3878, "step": 29170 }, { "epoch": 7.914293463520478, "grad_norm": 18.94926372207524, "learning_rate": 5.390612277465036e-06, "loss": 1.371, "step": 29180 }, { "epoch": 7.917005695687551, "grad_norm": 23.194791531592283, "learning_rate": 5.387465987763586e-06, "loss": 1.2863, "step": 29190 }, { "epoch": 7.919717927854625, "grad_norm": 19.688532739928487, "learning_rate": 5.384319543704462e-06, "loss": 1.188, "step": 29200 }, { "epoch": 7.9224301600216975, "grad_norm": 21.771527106301573, "learning_rate": 5.381172946541133e-06, "loss": 1.3792, "step": 29210 }, { "epoch": 7.925142392188771, "grad_norm": 21.069704597840307, "learning_rate": 5.378026197527135e-06, "loss": 1.2439, "step": 29220 }, { "epoch": 7.927854624355845, "grad_norm": 20.060866709069806, "learning_rate": 5.37487929791606e-06, "loss": 1.4154, "step": 29230 }, { "epoch": 7.930566856522918, "grad_norm": 20.897841483356775, "learning_rate": 5.371732248961562e-06, "loss": 1.3701, "step": 29240 }, { "epoch": 7.933279088689992, "grad_norm": 18.70376235731793, "learning_rate": 5.368585051917352e-06, "loss": 1.3842, "step": 29250 }, { "epoch": 7.935991320857065, "grad_norm": 19.33457472461042, "learning_rate": 5.365437708037205e-06, "loss": 1.3254, "step": 29260 }, { "epoch": 7.938703553024139, "grad_norm": 21.409364951958917, "learning_rate": 5.36229021857495e-06, "loss": 1.4187, "step": 29270 }, { "epoch": 7.941415785191213, "grad_norm": 19.02534708283433, "learning_rate": 5.359142584784478e-06, "loss": 1.3289, "step": 29280 }, { "epoch": 7.9441280173582856, "grad_norm": 24.29815916477905, "learning_rate": 5.355994807919731e-06, "loss": 1.422, "step": 29290 }, { "epoch": 7.946840249525359, "grad_norm": 20.639841122578257, "learning_rate": 5.352846889234715e-06, "loss": 1.3513, "step": 29300 }, { "epoch": 7.949552481692433, "grad_norm": 19.84382323175198, "learning_rate": 5.349698829983487e-06, "loss": 1.4096, "step": 29310 }, { "epoch": 7.952264713859506, "grad_norm": 21.287076817894622, "learning_rate": 5.346550631420165e-06, "loss": 1.3995, "step": 29320 }, { "epoch": 7.95497694602658, "grad_norm": 18.370410522640938, "learning_rate": 5.343402294798918e-06, "loss": 1.3524, "step": 29330 }, { "epoch": 7.957689178193654, "grad_norm": 21.75759179587464, "learning_rate": 5.340253821373974e-06, "loss": 1.4647, "step": 29340 }, { "epoch": 7.960401410360727, "grad_norm": 19.876249048186654, "learning_rate": 5.3371052123996115e-06, "loss": 1.4128, "step": 29350 }, { "epoch": 7.963113642527801, "grad_norm": 19.922514129172736, "learning_rate": 5.333956469130168e-06, "loss": 1.3358, "step": 29360 }, { "epoch": 7.965825874694874, "grad_norm": 19.2895080068043, "learning_rate": 5.330807592820027e-06, "loss": 1.2745, "step": 29370 }, { "epoch": 7.968538106861947, "grad_norm": 20.73734813638228, "learning_rate": 5.327658584723634e-06, "loss": 1.3996, "step": 29380 }, { "epoch": 7.971250339029021, "grad_norm": 20.051703646873644, "learning_rate": 5.3245094460954794e-06, "loss": 1.3672, "step": 29390 }, { "epoch": 7.973962571196094, "grad_norm": 18.998540935707986, "learning_rate": 5.321360178190111e-06, "loss": 1.4533, "step": 29400 }, { "epoch": 7.976674803363168, "grad_norm": 19.28243643818633, "learning_rate": 5.318210782262124e-06, "loss": 1.3437, "step": 29410 }, { "epoch": 7.979387035530241, "grad_norm": 20.672794778728925, "learning_rate": 5.315061259566169e-06, "loss": 1.3597, "step": 29420 }, { "epoch": 7.982099267697315, "grad_norm": 18.3571562972601, "learning_rate": 5.31191161135694e-06, "loss": 1.4371, "step": 29430 }, { "epoch": 7.984811499864389, "grad_norm": 19.605415085318302, "learning_rate": 5.3087618388891895e-06, "loss": 1.3143, "step": 29440 }, { "epoch": 7.987523732031462, "grad_norm": 20.443593257909857, "learning_rate": 5.305611943417714e-06, "loss": 1.4639, "step": 29450 }, { "epoch": 7.990235964198535, "grad_norm": 19.26706316701953, "learning_rate": 5.302461926197362e-06, "loss": 1.4105, "step": 29460 }, { "epoch": 7.992948196365609, "grad_norm": 19.260462878242816, "learning_rate": 5.299311788483026e-06, "loss": 1.2519, "step": 29470 }, { "epoch": 7.995660428532682, "grad_norm": 21.816372952794534, "learning_rate": 5.296161531529652e-06, "loss": 1.3648, "step": 29480 }, { "epoch": 7.998372660699756, "grad_norm": 18.48308590758304, "learning_rate": 5.29301115659223e-06, "loss": 1.3173, "step": 29490 }, { "epoch": 8.00108489286683, "grad_norm": 14.769970197743634, "learning_rate": 5.289860664925799e-06, "loss": 1.1754, "step": 29500 }, { "epoch": 8.003797125033904, "grad_norm": 19.521671590292158, "learning_rate": 5.286710057785443e-06, "loss": 0.9094, "step": 29510 }, { "epoch": 8.006509357200976, "grad_norm": 19.186984439579074, "learning_rate": 5.283559336426292e-06, "loss": 1.0123, "step": 29520 }, { "epoch": 8.00922158936805, "grad_norm": 19.89730360086172, "learning_rate": 5.280408502103523e-06, "loss": 1.0072, "step": 29530 }, { "epoch": 8.011933821535123, "grad_norm": 22.93725702344847, "learning_rate": 5.277257556072356e-06, "loss": 0.9207, "step": 29540 }, { "epoch": 8.014646053702197, "grad_norm": 17.36621290262972, "learning_rate": 5.274106499588056e-06, "loss": 0.8344, "step": 29550 }, { "epoch": 8.017358285869271, "grad_norm": 20.014135347215184, "learning_rate": 5.270955333905935e-06, "loss": 0.9088, "step": 29560 }, { "epoch": 8.020070518036343, "grad_norm": 18.44753497217023, "learning_rate": 5.267804060281344e-06, "loss": 0.9557, "step": 29570 }, { "epoch": 8.022782750203417, "grad_norm": 20.63210320959081, "learning_rate": 5.264652679969681e-06, "loss": 0.8918, "step": 29580 }, { "epoch": 8.02549498237049, "grad_norm": 15.783811312876017, "learning_rate": 5.2615011942263825e-06, "loss": 0.8829, "step": 29590 }, { "epoch": 8.028207214537565, "grad_norm": 17.933263602222436, "learning_rate": 5.258349604306932e-06, "loss": 0.936, "step": 29600 }, { "epoch": 8.030919446704639, "grad_norm": 21.468444995518364, "learning_rate": 5.255197911466846e-06, "loss": 0.8936, "step": 29610 }, { "epoch": 8.03363167887171, "grad_norm": 16.13398786974443, "learning_rate": 5.252046116961692e-06, "loss": 0.9552, "step": 29620 }, { "epoch": 8.036343911038784, "grad_norm": 19.571416224306738, "learning_rate": 5.248894222047073e-06, "loss": 0.8642, "step": 29630 }, { "epoch": 8.039056143205858, "grad_norm": 21.407093193025496, "learning_rate": 5.245742227978632e-06, "loss": 0.9517, "step": 29640 }, { "epoch": 8.041768375372932, "grad_norm": 18.50926606413127, "learning_rate": 5.242590136012053e-06, "loss": 0.9216, "step": 29650 }, { "epoch": 8.044480607540006, "grad_norm": 17.951170192493873, "learning_rate": 5.239437947403056e-06, "loss": 0.9409, "step": 29660 }, { "epoch": 8.04719283970708, "grad_norm": 20.392030704819796, "learning_rate": 5.236285663407404e-06, "loss": 0.9302, "step": 29670 }, { "epoch": 8.049905071874152, "grad_norm": 17.07406441734328, "learning_rate": 5.2331332852808925e-06, "loss": 0.9346, "step": 29680 }, { "epoch": 8.052617304041226, "grad_norm": 22.880288697947282, "learning_rate": 5.22998081427936e-06, "loss": 0.9035, "step": 29690 }, { "epoch": 8.0553295362083, "grad_norm": 20.81811560167967, "learning_rate": 5.22682825165868e-06, "loss": 0.954, "step": 29700 }, { "epoch": 8.058041768375373, "grad_norm": 20.685194481545764, "learning_rate": 5.22367559867476e-06, "loss": 0.9573, "step": 29710 }, { "epoch": 8.060754000542447, "grad_norm": 16.837833549727367, "learning_rate": 5.220522856583547e-06, "loss": 0.8526, "step": 29720 }, { "epoch": 8.06346623270952, "grad_norm": 18.990916961665608, "learning_rate": 5.217370026641021e-06, "loss": 0.9204, "step": 29730 }, { "epoch": 8.066178464876593, "grad_norm": 17.907876950278155, "learning_rate": 5.2142171101032e-06, "loss": 0.849, "step": 29740 }, { "epoch": 8.068890697043667, "grad_norm": 18.93157173595001, "learning_rate": 5.211064108226131e-06, "loss": 0.7994, "step": 29750 }, { "epoch": 8.07160292921074, "grad_norm": 22.640444670315112, "learning_rate": 5.207911022265903e-06, "loss": 1.0213, "step": 29760 }, { "epoch": 8.074315161377815, "grad_norm": 21.043732713193318, "learning_rate": 5.204757853478631e-06, "loss": 0.9439, "step": 29770 }, { "epoch": 8.077027393544887, "grad_norm": 18.676401727361164, "learning_rate": 5.201604603120468e-06, "loss": 0.9802, "step": 29780 }, { "epoch": 8.07973962571196, "grad_norm": 20.342207151394025, "learning_rate": 5.198451272447595e-06, "loss": 0.9316, "step": 29790 }, { "epoch": 8.082451857879034, "grad_norm": 20.930711922718917, "learning_rate": 5.19529786271623e-06, "loss": 1.0086, "step": 29800 }, { "epoch": 8.085164090046108, "grad_norm": 18.762544708791165, "learning_rate": 5.192144375182617e-06, "loss": 0.949, "step": 29810 }, { "epoch": 8.087876322213182, "grad_norm": 19.42472045455032, "learning_rate": 5.188990811103038e-06, "loss": 0.9309, "step": 29820 }, { "epoch": 8.090588554380256, "grad_norm": 22.35265174403685, "learning_rate": 5.185837171733799e-06, "loss": 0.9169, "step": 29830 }, { "epoch": 8.093300786547328, "grad_norm": 19.349703324303132, "learning_rate": 5.182683458331238e-06, "loss": 0.9339, "step": 29840 }, { "epoch": 8.096013018714402, "grad_norm": 17.894276682143982, "learning_rate": 5.179529672151725e-06, "loss": 0.8963, "step": 29850 }, { "epoch": 8.098725250881476, "grad_norm": 20.535292371278302, "learning_rate": 5.1763758144516564e-06, "loss": 0.9713, "step": 29860 }, { "epoch": 8.10143748304855, "grad_norm": 18.2593744105866, "learning_rate": 5.173221886487456e-06, "loss": 0.8698, "step": 29870 }, { "epoch": 8.104149715215623, "grad_norm": 20.124717382031044, "learning_rate": 5.17006788951558e-06, "loss": 0.8579, "step": 29880 }, { "epoch": 8.106861947382695, "grad_norm": 18.435380283580276, "learning_rate": 5.166913824792507e-06, "loss": 0.9097, "step": 29890 }, { "epoch": 8.10957417954977, "grad_norm": 22.239718019163472, "learning_rate": 5.163759693574747e-06, "loss": 0.9103, "step": 29900 }, { "epoch": 8.112286411716843, "grad_norm": 20.174926715293157, "learning_rate": 5.160605497118831e-06, "loss": 0.9586, "step": 29910 }, { "epoch": 8.114998643883917, "grad_norm": 20.47346153092586, "learning_rate": 5.157451236681324e-06, "loss": 0.9809, "step": 29920 }, { "epoch": 8.11771087605099, "grad_norm": 19.83582233339297, "learning_rate": 5.1542969135188094e-06, "loss": 0.9884, "step": 29930 }, { "epoch": 8.120423108218063, "grad_norm": 23.53158953055386, "learning_rate": 5.151142528887898e-06, "loss": 0.9079, "step": 29940 }, { "epoch": 8.123135340385137, "grad_norm": 16.64547048899188, "learning_rate": 5.147988084045226e-06, "loss": 0.9436, "step": 29950 }, { "epoch": 8.12584757255221, "grad_norm": 15.139061281963649, "learning_rate": 5.144833580247454e-06, "loss": 0.9064, "step": 29960 }, { "epoch": 8.128559804719284, "grad_norm": 20.047239446983852, "learning_rate": 5.14167901875126e-06, "loss": 0.9175, "step": 29970 }, { "epoch": 8.131272036886358, "grad_norm": 19.4067488245791, "learning_rate": 5.138524400813356e-06, "loss": 0.8343, "step": 29980 }, { "epoch": 8.133984269053432, "grad_norm": 20.364430272819856, "learning_rate": 5.1353697276904665e-06, "loss": 0.9622, "step": 29990 }, { "epoch": 8.136696501220504, "grad_norm": 17.984196812823296, "learning_rate": 5.132215000639342e-06, "loss": 0.969, "step": 30000 }, { "epoch": 8.139408733387578, "grad_norm": 19.224917822501595, "learning_rate": 5.129060220916756e-06, "loss": 0.9471, "step": 30010 }, { "epoch": 8.142120965554652, "grad_norm": 19.199129803656046, "learning_rate": 5.125905389779502e-06, "loss": 0.9251, "step": 30020 }, { "epoch": 8.144833197721725, "grad_norm": 19.6753977519566, "learning_rate": 5.12275050848439e-06, "loss": 0.9845, "step": 30030 }, { "epoch": 8.1475454298888, "grad_norm": 21.551685078122073, "learning_rate": 5.119595578288257e-06, "loss": 0.966, "step": 30040 }, { "epoch": 8.150257662055871, "grad_norm": 19.410778303327636, "learning_rate": 5.116440600447951e-06, "loss": 0.9199, "step": 30050 }, { "epoch": 8.152969894222945, "grad_norm": 20.20490070812411, "learning_rate": 5.113285576220348e-06, "loss": 0.8317, "step": 30060 }, { "epoch": 8.155682126390019, "grad_norm": 19.0700394609665, "learning_rate": 5.110130506862336e-06, "loss": 0.8703, "step": 30070 }, { "epoch": 8.158394358557093, "grad_norm": 19.53588235822137, "learning_rate": 5.1069753936308265e-06, "loss": 0.9283, "step": 30080 }, { "epoch": 8.161106590724167, "grad_norm": 19.61330853528208, "learning_rate": 5.1038202377827395e-06, "loss": 0.8702, "step": 30090 }, { "epoch": 8.163818822891239, "grad_norm": 20.901462217669426, "learning_rate": 5.100665040575024e-06, "loss": 0.9485, "step": 30100 }, { "epoch": 8.166531055058313, "grad_norm": 19.26372858441344, "learning_rate": 5.097509803264633e-06, "loss": 0.9451, "step": 30110 }, { "epoch": 8.169243287225386, "grad_norm": 19.56793777589489, "learning_rate": 5.0943545271085445e-06, "loss": 0.9234, "step": 30120 }, { "epoch": 8.17195551939246, "grad_norm": 20.814985946889745, "learning_rate": 5.09119921336375e-06, "loss": 0.917, "step": 30130 }, { "epoch": 8.174667751559534, "grad_norm": 19.109041436978732, "learning_rate": 5.088043863287255e-06, "loss": 1.0319, "step": 30140 }, { "epoch": 8.177379983726606, "grad_norm": 19.081500373639738, "learning_rate": 5.084888478136078e-06, "loss": 0.8522, "step": 30150 }, { "epoch": 8.18009221589368, "grad_norm": 19.221086556981362, "learning_rate": 5.081733059167253e-06, "loss": 0.9183, "step": 30160 }, { "epoch": 8.182804448060754, "grad_norm": 19.642558316874844, "learning_rate": 5.078577607637828e-06, "loss": 1.0013, "step": 30170 }, { "epoch": 8.185516680227828, "grad_norm": 20.26809870122163, "learning_rate": 5.075422124804863e-06, "loss": 0.8856, "step": 30180 }, { "epoch": 8.188228912394901, "grad_norm": 18.148585638633392, "learning_rate": 5.0722666119254324e-06, "loss": 0.9252, "step": 30190 }, { "epoch": 8.190941144561975, "grad_norm": 19.246559221094067, "learning_rate": 5.0691110702566205e-06, "loss": 0.8799, "step": 30200 }, { "epoch": 8.193653376729047, "grad_norm": 18.712317407734467, "learning_rate": 5.065955501055522e-06, "loss": 0.9418, "step": 30210 }, { "epoch": 8.196365608896121, "grad_norm": 18.51087888500703, "learning_rate": 5.062799905579247e-06, "loss": 0.9332, "step": 30220 }, { "epoch": 8.199077841063195, "grad_norm": 21.10935138382475, "learning_rate": 5.05964428508491e-06, "loss": 0.9711, "step": 30230 }, { "epoch": 8.201790073230269, "grad_norm": 18.486871579274467, "learning_rate": 5.056488640829642e-06, "loss": 0.7994, "step": 30240 }, { "epoch": 8.204502305397343, "grad_norm": 19.028308713545805, "learning_rate": 5.053332974070576e-06, "loss": 0.9246, "step": 30250 }, { "epoch": 8.207214537564415, "grad_norm": 18.866007899197115, "learning_rate": 5.050177286064864e-06, "loss": 0.9645, "step": 30260 }, { "epoch": 8.209926769731489, "grad_norm": 19.371780018311767, "learning_rate": 5.047021578069655e-06, "loss": 0.9371, "step": 30270 }, { "epoch": 8.212639001898562, "grad_norm": 17.62379419022932, "learning_rate": 5.043865851342117e-06, "loss": 0.9153, "step": 30280 }, { "epoch": 8.215351234065636, "grad_norm": 20.381545385314915, "learning_rate": 5.040710107139415e-06, "loss": 0.984, "step": 30290 }, { "epoch": 8.21806346623271, "grad_norm": 23.737022439865843, "learning_rate": 5.0375543467187295e-06, "loss": 0.9689, "step": 30300 }, { "epoch": 8.220775698399782, "grad_norm": 18.946075025870435, "learning_rate": 5.03439857133724e-06, "loss": 0.8545, "step": 30310 }, { "epoch": 8.223487930566856, "grad_norm": 19.415801873855273, "learning_rate": 5.03124278225214e-06, "loss": 0.9495, "step": 30320 }, { "epoch": 8.22620016273393, "grad_norm": 20.426436603684074, "learning_rate": 5.028086980720623e-06, "loss": 0.9613, "step": 30330 }, { "epoch": 8.228912394901004, "grad_norm": 20.9155244908581, "learning_rate": 5.024931167999891e-06, "loss": 0.9753, "step": 30340 }, { "epoch": 8.231624627068078, "grad_norm": 21.736065283864974, "learning_rate": 5.021775345347144e-06, "loss": 0.9724, "step": 30350 }, { "epoch": 8.234336859235151, "grad_norm": 21.427795074274396, "learning_rate": 5.0186195140195935e-06, "loss": 1.0235, "step": 30360 }, { "epoch": 8.237049091402223, "grad_norm": 19.672478809128545, "learning_rate": 5.01546367527445e-06, "loss": 0.9184, "step": 30370 }, { "epoch": 8.239761323569297, "grad_norm": 21.105086046997734, "learning_rate": 5.012307830368928e-06, "loss": 1.0353, "step": 30380 }, { "epoch": 8.242473555736371, "grad_norm": 21.08856809417945, "learning_rate": 5.0091519805602455e-06, "loss": 0.9342, "step": 30390 }, { "epoch": 8.245185787903445, "grad_norm": 22.640682260685427, "learning_rate": 5.005996127105621e-06, "loss": 0.895, "step": 30400 }, { "epoch": 8.247898020070519, "grad_norm": 22.712955731999816, "learning_rate": 5.002840271262275e-06, "loss": 0.9855, "step": 30410 }, { "epoch": 8.25061025223759, "grad_norm": 19.549877460268316, "learning_rate": 4.99968441428743e-06, "loss": 0.9217, "step": 30420 }, { "epoch": 8.253322484404665, "grad_norm": 22.27417459916291, "learning_rate": 4.996528557438306e-06, "loss": 0.9097, "step": 30430 }, { "epoch": 8.256034716571738, "grad_norm": 20.99674013465709, "learning_rate": 4.9933727019721265e-06, "loss": 1.0117, "step": 30440 }, { "epoch": 8.258746948738812, "grad_norm": 19.087102968019867, "learning_rate": 4.990216849146113e-06, "loss": 0.9034, "step": 30450 }, { "epoch": 8.261459180905886, "grad_norm": 19.567288404434276, "learning_rate": 4.987061000217486e-06, "loss": 0.8458, "step": 30460 }, { "epoch": 8.264171413072958, "grad_norm": 21.2479160960862, "learning_rate": 4.983905156443464e-06, "loss": 0.9898, "step": 30470 }, { "epoch": 8.266883645240032, "grad_norm": 21.091318595815334, "learning_rate": 4.9807493190812596e-06, "loss": 0.8533, "step": 30480 }, { "epoch": 8.269595877407106, "grad_norm": 17.30397052113536, "learning_rate": 4.9775934893880935e-06, "loss": 0.9885, "step": 30490 }, { "epoch": 8.27230810957418, "grad_norm": 20.34107697329088, "learning_rate": 4.9744376686211725e-06, "loss": 0.9222, "step": 30500 }, { "epoch": 8.275020341741254, "grad_norm": 21.057544276737012, "learning_rate": 4.971281858037707e-06, "loss": 0.94, "step": 30510 }, { "epoch": 8.277732573908327, "grad_norm": 20.310164982121382, "learning_rate": 4.968126058894896e-06, "loss": 0.9033, "step": 30520 }, { "epoch": 8.2804448060754, "grad_norm": 21.229335089933727, "learning_rate": 4.964970272449944e-06, "loss": 0.9389, "step": 30530 }, { "epoch": 8.283157038242473, "grad_norm": 21.72926823984425, "learning_rate": 4.961814499960042e-06, "loss": 0.9662, "step": 30540 }, { "epoch": 8.285869270409547, "grad_norm": 20.668452677055505, "learning_rate": 4.958658742682378e-06, "loss": 0.9509, "step": 30550 }, { "epoch": 8.288581502576621, "grad_norm": 19.789431401038556, "learning_rate": 4.955503001874134e-06, "loss": 1.0256, "step": 30560 }, { "epoch": 8.291293734743695, "grad_norm": 20.463805513626014, "learning_rate": 4.952347278792487e-06, "loss": 0.9557, "step": 30570 }, { "epoch": 8.294005966910767, "grad_norm": 20.24303920388717, "learning_rate": 4.949191574694607e-06, "loss": 1.0556, "step": 30580 }, { "epoch": 8.29671819907784, "grad_norm": 20.33505762890305, "learning_rate": 4.946035890837651e-06, "loss": 0.9623, "step": 30590 }, { "epoch": 8.299430431244915, "grad_norm": 19.793967969436405, "learning_rate": 4.942880228478774e-06, "loss": 0.9681, "step": 30600 }, { "epoch": 8.302142663411988, "grad_norm": 19.82607306217441, "learning_rate": 4.939724588875121e-06, "loss": 1.1242, "step": 30610 }, { "epoch": 8.304854895579062, "grad_norm": 19.814476428699635, "learning_rate": 4.936568973283827e-06, "loss": 0.9521, "step": 30620 }, { "epoch": 8.307567127746134, "grad_norm": 20.17352221747112, "learning_rate": 4.933413382962018e-06, "loss": 0.9709, "step": 30630 }, { "epoch": 8.310279359913208, "grad_norm": 18.709597587785296, "learning_rate": 4.9302578191668075e-06, "loss": 0.9362, "step": 30640 }, { "epoch": 8.312991592080282, "grad_norm": 19.514419098866377, "learning_rate": 4.927102283155304e-06, "loss": 0.9161, "step": 30650 }, { "epoch": 8.315703824247356, "grad_norm": 19.682787943850663, "learning_rate": 4.923946776184601e-06, "loss": 1.001, "step": 30660 }, { "epoch": 8.31841605641443, "grad_norm": 20.2467246859726, "learning_rate": 4.920791299511781e-06, "loss": 0.9611, "step": 30670 }, { "epoch": 8.321128288581502, "grad_norm": 18.60222923734696, "learning_rate": 4.917635854393909e-06, "loss": 0.9121, "step": 30680 }, { "epoch": 8.323840520748575, "grad_norm": 18.537092585858744, "learning_rate": 4.914480442088051e-06, "loss": 0.9105, "step": 30690 }, { "epoch": 8.32655275291565, "grad_norm": 19.97086149043305, "learning_rate": 4.9113250638512485e-06, "loss": 0.9821, "step": 30700 }, { "epoch": 8.329264985082723, "grad_norm": 20.850078463054267, "learning_rate": 4.908169720940532e-06, "loss": 0.9799, "step": 30710 }, { "epoch": 8.331977217249797, "grad_norm": 18.962166598323595, "learning_rate": 4.905014414612918e-06, "loss": 0.8966, "step": 30720 }, { "epoch": 8.33468944941687, "grad_norm": 19.976561614451942, "learning_rate": 4.901859146125412e-06, "loss": 0.9628, "step": 30730 }, { "epoch": 8.337401681583943, "grad_norm": 19.047300992547996, "learning_rate": 4.898703916734999e-06, "loss": 0.8767, "step": 30740 }, { "epoch": 8.340113913751017, "grad_norm": 18.719017985066774, "learning_rate": 4.895548727698652e-06, "loss": 0.8879, "step": 30750 }, { "epoch": 8.34282614591809, "grad_norm": 20.002179094277942, "learning_rate": 4.892393580273324e-06, "loss": 0.9016, "step": 30760 }, { "epoch": 8.345538378085164, "grad_norm": 18.651412010629414, "learning_rate": 4.88923847571596e-06, "loss": 0.9682, "step": 30770 }, { "epoch": 8.348250610252238, "grad_norm": 20.329220238153365, "learning_rate": 4.886083415283479e-06, "loss": 0.9369, "step": 30780 }, { "epoch": 8.35096284241931, "grad_norm": 17.305759729127537, "learning_rate": 4.882928400232786e-06, "loss": 0.899, "step": 30790 }, { "epoch": 8.353675074586384, "grad_norm": 20.20237002058938, "learning_rate": 4.879773431820765e-06, "loss": 0.9173, "step": 30800 }, { "epoch": 8.356387306753458, "grad_norm": 19.796088415594266, "learning_rate": 4.876618511304287e-06, "loss": 0.9812, "step": 30810 }, { "epoch": 8.359099538920532, "grad_norm": 21.36276147014092, "learning_rate": 4.873463639940202e-06, "loss": 1.0248, "step": 30820 }, { "epoch": 8.361811771087606, "grad_norm": 19.736840585311743, "learning_rate": 4.870308818985337e-06, "loss": 0.7886, "step": 30830 }, { "epoch": 8.36452400325468, "grad_norm": 20.540381300802736, "learning_rate": 4.867154049696499e-06, "loss": 0.9221, "step": 30840 }, { "epoch": 8.367236235421752, "grad_norm": 20.319064648584366, "learning_rate": 4.863999333330482e-06, "loss": 0.9517, "step": 30850 }, { "epoch": 8.369948467588825, "grad_norm": 18.602434634896095, "learning_rate": 4.860844671144051e-06, "loss": 0.9708, "step": 30860 }, { "epoch": 8.3726606997559, "grad_norm": 20.26084403961472, "learning_rate": 4.85769006439395e-06, "loss": 0.9569, "step": 30870 }, { "epoch": 8.375372931922973, "grad_norm": 19.91536897156079, "learning_rate": 4.8545355143369045e-06, "loss": 0.9587, "step": 30880 }, { "epoch": 8.378085164090047, "grad_norm": 18.88542910790931, "learning_rate": 4.851381022229619e-06, "loss": 0.8743, "step": 30890 }, { "epoch": 8.380797396257119, "grad_norm": 17.17908392193112, "learning_rate": 4.848226589328768e-06, "loss": 0.9745, "step": 30900 }, { "epoch": 8.383509628424193, "grad_norm": 24.30139186819621, "learning_rate": 4.845072216891006e-06, "loss": 1.0212, "step": 30910 }, { "epoch": 8.386221860591267, "grad_norm": 20.298294531881403, "learning_rate": 4.841917906172963e-06, "loss": 0.9191, "step": 30920 }, { "epoch": 8.38893409275834, "grad_norm": 19.764416920151067, "learning_rate": 4.838763658431249e-06, "loss": 0.9256, "step": 30930 }, { "epoch": 8.391646324925414, "grad_norm": 19.24412936471786, "learning_rate": 4.835609474922441e-06, "loss": 0.9204, "step": 30940 }, { "epoch": 8.394358557092486, "grad_norm": 21.383613533833998, "learning_rate": 4.8324553569030965e-06, "loss": 0.9154, "step": 30950 }, { "epoch": 8.39707078925956, "grad_norm": 21.438401462638623, "learning_rate": 4.829301305629742e-06, "loss": 0.8423, "step": 30960 }, { "epoch": 8.399783021426634, "grad_norm": 21.05122754475836, "learning_rate": 4.826147322358884e-06, "loss": 0.8759, "step": 30970 }, { "epoch": 8.402495253593708, "grad_norm": 20.694980582904723, "learning_rate": 4.8229934083469956e-06, "loss": 0.984, "step": 30980 }, { "epoch": 8.405207485760782, "grad_norm": 22.55649359224389, "learning_rate": 4.819839564850524e-06, "loss": 0.8741, "step": 30990 }, { "epoch": 8.407919717927854, "grad_norm": 21.83518460466324, "learning_rate": 4.816685793125891e-06, "loss": 1.0015, "step": 31000 }, { "epoch": 8.410631950094928, "grad_norm": 22.63724164038145, "learning_rate": 4.813532094429486e-06, "loss": 1.002, "step": 31010 }, { "epoch": 8.413344182262001, "grad_norm": 22.33403037845242, "learning_rate": 4.810378470017673e-06, "loss": 0.9625, "step": 31020 }, { "epoch": 8.416056414429075, "grad_norm": 21.663859250299257, "learning_rate": 4.807224921146784e-06, "loss": 0.9369, "step": 31030 }, { "epoch": 8.418768646596149, "grad_norm": 18.109656213601703, "learning_rate": 4.804071449073119e-06, "loss": 0.9987, "step": 31040 }, { "epoch": 8.421480878763223, "grad_norm": 19.95573921871464, "learning_rate": 4.800918055052953e-06, "loss": 1.0049, "step": 31050 }, { "epoch": 8.424193110930295, "grad_norm": 19.264252937787248, "learning_rate": 4.797764740342526e-06, "loss": 1.0135, "step": 31060 }, { "epoch": 8.426905343097369, "grad_norm": 17.327729880160152, "learning_rate": 4.794611506198047e-06, "loss": 1.0564, "step": 31070 }, { "epoch": 8.429617575264443, "grad_norm": 21.529693065638785, "learning_rate": 4.791458353875691e-06, "loss": 0.9496, "step": 31080 }, { "epoch": 8.432329807431516, "grad_norm": 19.698623788762134, "learning_rate": 4.7883052846316076e-06, "loss": 0.9628, "step": 31090 }, { "epoch": 8.43504203959859, "grad_norm": 21.01559422233646, "learning_rate": 4.785152299721904e-06, "loss": 1.0044, "step": 31100 }, { "epoch": 8.437754271765662, "grad_norm": 18.628102319598263, "learning_rate": 4.781999400402658e-06, "loss": 0.9664, "step": 31110 }, { "epoch": 8.440466503932736, "grad_norm": 22.46684918048349, "learning_rate": 4.778846587929915e-06, "loss": 0.9246, "step": 31120 }, { "epoch": 8.44317873609981, "grad_norm": 20.909380292360552, "learning_rate": 4.775693863559683e-06, "loss": 0.9576, "step": 31130 }, { "epoch": 8.445890968266884, "grad_norm": 20.004971507683003, "learning_rate": 4.772541228547937e-06, "loss": 0.9493, "step": 31140 }, { "epoch": 8.448603200433958, "grad_norm": 19.021387808989292, "learning_rate": 4.769388684150616e-06, "loss": 1.0108, "step": 31150 }, { "epoch": 8.45131543260103, "grad_norm": 19.209890611461365, "learning_rate": 4.766236231623617e-06, "loss": 1.0086, "step": 31160 }, { "epoch": 8.454027664768104, "grad_norm": 19.724402030751246, "learning_rate": 4.763083872222814e-06, "loss": 0.9974, "step": 31170 }, { "epoch": 8.456739896935177, "grad_norm": 20.119769918992734, "learning_rate": 4.759931607204029e-06, "loss": 0.9749, "step": 31180 }, { "epoch": 8.459452129102251, "grad_norm": 20.207207850244913, "learning_rate": 4.756779437823057e-06, "loss": 1.0276, "step": 31190 }, { "epoch": 8.462164361269325, "grad_norm": 20.650575994695057, "learning_rate": 4.753627365335646e-06, "loss": 0.9591, "step": 31200 }, { "epoch": 8.464876593436399, "grad_norm": 20.503611961341658, "learning_rate": 4.750475390997517e-06, "loss": 1.0079, "step": 31210 }, { "epoch": 8.467588825603471, "grad_norm": 23.209048400163038, "learning_rate": 4.747323516064342e-06, "loss": 1.0071, "step": 31220 }, { "epoch": 8.470301057770545, "grad_norm": 21.228842724046633, "learning_rate": 4.744171741791756e-06, "loss": 0.9759, "step": 31230 }, { "epoch": 8.473013289937619, "grad_norm": 18.151059323514353, "learning_rate": 4.741020069435352e-06, "loss": 1.0185, "step": 31240 }, { "epoch": 8.475725522104693, "grad_norm": 18.498540458796462, "learning_rate": 4.73786850025069e-06, "loss": 1.0069, "step": 31250 }, { "epoch": 8.478437754271766, "grad_norm": 23.52727529691736, "learning_rate": 4.734717035493282e-06, "loss": 0.9128, "step": 31260 }, { "epoch": 8.481149986438838, "grad_norm": 20.194430194653812, "learning_rate": 4.7315656764186e-06, "loss": 0.978, "step": 31270 }, { "epoch": 8.483862218605912, "grad_norm": 19.90253528363005, "learning_rate": 4.7284144242820714e-06, "loss": 1.0342, "step": 31280 }, { "epoch": 8.486574450772986, "grad_norm": 23.69346267912517, "learning_rate": 4.7252632803390886e-06, "loss": 0.982, "step": 31290 }, { "epoch": 8.48928668294006, "grad_norm": 16.151539712142984, "learning_rate": 4.722112245844994e-06, "loss": 0.9112, "step": 31300 }, { "epoch": 8.491998915107134, "grad_norm": 18.70650716912795, "learning_rate": 4.718961322055086e-06, "loss": 1.0633, "step": 31310 }, { "epoch": 8.494711147274206, "grad_norm": 17.681325043372595, "learning_rate": 4.715810510224623e-06, "loss": 0.91, "step": 31320 }, { "epoch": 8.49742337944128, "grad_norm": 19.349078951559974, "learning_rate": 4.712659811608819e-06, "loss": 0.9815, "step": 31330 }, { "epoch": 8.500135611608354, "grad_norm": 18.034029638584514, "learning_rate": 4.709509227462839e-06, "loss": 0.9555, "step": 31340 }, { "epoch": 8.502847843775427, "grad_norm": 17.155474719427158, "learning_rate": 4.706358759041806e-06, "loss": 0.9639, "step": 31350 }, { "epoch": 8.505560075942501, "grad_norm": 20.72994446520045, "learning_rate": 4.70320840760079e-06, "loss": 1.0771, "step": 31360 }, { "epoch": 8.508272308109575, "grad_norm": 21.67887539107142, "learning_rate": 4.700058174394827e-06, "loss": 0.9198, "step": 31370 }, { "epoch": 8.510984540276647, "grad_norm": 20.302939314983554, "learning_rate": 4.696908060678893e-06, "loss": 0.9049, "step": 31380 }, { "epoch": 8.513696772443721, "grad_norm": 21.714671045452924, "learning_rate": 4.693758067707927e-06, "loss": 0.9371, "step": 31390 }, { "epoch": 8.516409004610795, "grad_norm": 20.14825667321153, "learning_rate": 4.690608196736809e-06, "loss": 0.9913, "step": 31400 }, { "epoch": 8.519121236777869, "grad_norm": 18.987446827646068, "learning_rate": 4.687458449020381e-06, "loss": 0.9356, "step": 31410 }, { "epoch": 8.521833468944942, "grad_norm": 21.04280521198988, "learning_rate": 4.68430882581343e-06, "loss": 1.0079, "step": 31420 }, { "epoch": 8.524545701112014, "grad_norm": 22.523836500691484, "learning_rate": 4.681159328370692e-06, "loss": 1.0187, "step": 31430 }, { "epoch": 8.527257933279088, "grad_norm": 19.108889210065502, "learning_rate": 4.678009957946859e-06, "loss": 0.9127, "step": 31440 }, { "epoch": 8.529970165446162, "grad_norm": 20.589579858371863, "learning_rate": 4.6748607157965665e-06, "loss": 0.9449, "step": 31450 }, { "epoch": 8.532682397613236, "grad_norm": 20.25408613625378, "learning_rate": 4.671711603174404e-06, "loss": 0.9096, "step": 31460 }, { "epoch": 8.53539462978031, "grad_norm": 20.059205320901828, "learning_rate": 4.668562621334904e-06, "loss": 0.9856, "step": 31470 }, { "epoch": 8.538106861947382, "grad_norm": 21.506581013149273, "learning_rate": 4.665413771532548e-06, "loss": 1.0867, "step": 31480 }, { "epoch": 8.540819094114456, "grad_norm": 21.48927800076611, "learning_rate": 4.6622650550217715e-06, "loss": 0.9598, "step": 31490 }, { "epoch": 8.54353132628153, "grad_norm": 18.09813648401641, "learning_rate": 4.659116473056948e-06, "loss": 0.9184, "step": 31500 }, { "epoch": 8.546243558448603, "grad_norm": 21.555858524005384, "learning_rate": 4.655968026892403e-06, "loss": 0.9007, "step": 31510 }, { "epoch": 8.548955790615677, "grad_norm": 17.75125761830926, "learning_rate": 4.6528197177824035e-06, "loss": 0.9802, "step": 31520 }, { "epoch": 8.55166802278275, "grad_norm": 21.686726597251656, "learning_rate": 4.6496715469811696e-06, "loss": 1.0728, "step": 31530 }, { "epoch": 8.554380254949823, "grad_norm": 20.03402364631323, "learning_rate": 4.646523515742857e-06, "loss": 1.0229, "step": 31540 }, { "epoch": 8.557092487116897, "grad_norm": 19.99227086149181, "learning_rate": 4.643375625321571e-06, "loss": 1.0336, "step": 31550 }, { "epoch": 8.55980471928397, "grad_norm": 19.74230876412437, "learning_rate": 4.64022787697136e-06, "loss": 0.989, "step": 31560 }, { "epoch": 8.562516951451045, "grad_norm": 19.106416304658094, "learning_rate": 4.637080271946216e-06, "loss": 1.027, "step": 31570 }, { "epoch": 8.565229183618118, "grad_norm": 22.493029243024242, "learning_rate": 4.633932811500075e-06, "loss": 0.9953, "step": 31580 }, { "epoch": 8.56794141578519, "grad_norm": 18.26351295411168, "learning_rate": 4.630785496886812e-06, "loss": 1.0109, "step": 31590 }, { "epoch": 8.570653647952264, "grad_norm": 21.366003092199993, "learning_rate": 4.627638329360243e-06, "loss": 0.8896, "step": 31600 }, { "epoch": 8.573365880119338, "grad_norm": 22.84030903705365, "learning_rate": 4.624491310174135e-06, "loss": 0.9838, "step": 31610 }, { "epoch": 8.576078112286412, "grad_norm": 23.12609414906719, "learning_rate": 4.621344440582185e-06, "loss": 1.0118, "step": 31620 }, { "epoch": 8.578790344453486, "grad_norm": 16.348535398274176, "learning_rate": 4.618197721838035e-06, "loss": 0.8672, "step": 31630 }, { "epoch": 8.581502576620558, "grad_norm": 19.73464700397891, "learning_rate": 4.615051155195266e-06, "loss": 0.9411, "step": 31640 }, { "epoch": 8.584214808787632, "grad_norm": 18.873500601489365, "learning_rate": 4.6119047419074e-06, "loss": 0.9679, "step": 31650 }, { "epoch": 8.586927040954706, "grad_norm": 20.216987831297136, "learning_rate": 4.608758483227898e-06, "loss": 0.9942, "step": 31660 }, { "epoch": 8.58963927312178, "grad_norm": 19.54485750748144, "learning_rate": 4.605612380410156e-06, "loss": 1.0285, "step": 31670 }, { "epoch": 8.592351505288853, "grad_norm": 19.359950503972694, "learning_rate": 4.602466434707509e-06, "loss": 0.9992, "step": 31680 }, { "epoch": 8.595063737455927, "grad_norm": 20.905788827044635, "learning_rate": 4.599320647373233e-06, "loss": 1.0024, "step": 31690 }, { "epoch": 8.597775969623, "grad_norm": 21.492848926833464, "learning_rate": 4.59617501966054e-06, "loss": 0.9028, "step": 31700 }, { "epoch": 8.600488201790073, "grad_norm": 17.985947848219414, "learning_rate": 4.593029552822574e-06, "loss": 0.9624, "step": 31710 }, { "epoch": 8.603200433957147, "grad_norm": 19.52696200571577, "learning_rate": 4.589884248112416e-06, "loss": 0.9469, "step": 31720 }, { "epoch": 8.60591266612422, "grad_norm": 21.614569088027185, "learning_rate": 4.586739106783089e-06, "loss": 0.916, "step": 31730 }, { "epoch": 8.608624898291295, "grad_norm": 20.72559696562766, "learning_rate": 4.5835941300875435e-06, "loss": 1.0353, "step": 31740 }, { "epoch": 8.611337130458367, "grad_norm": 21.809528343040505, "learning_rate": 4.580449319278666e-06, "loss": 0.9792, "step": 31750 }, { "epoch": 8.61404936262544, "grad_norm": 21.038435483729646, "learning_rate": 4.577304675609279e-06, "loss": 0.9483, "step": 31760 }, { "epoch": 8.616761594792514, "grad_norm": 19.813163223059615, "learning_rate": 4.574160200332138e-06, "loss": 0.966, "step": 31770 }, { "epoch": 8.619473826959588, "grad_norm": 18.938686845156507, "learning_rate": 4.5710158946999315e-06, "loss": 1.0335, "step": 31780 }, { "epoch": 8.622186059126662, "grad_norm": 20.394426358008158, "learning_rate": 4.5678717599652776e-06, "loss": 0.8969, "step": 31790 }, { "epoch": 8.624898291293734, "grad_norm": 22.010609386355526, "learning_rate": 4.5647277973807265e-06, "loss": 1.0308, "step": 31800 }, { "epoch": 8.627610523460808, "grad_norm": 19.642730760137095, "learning_rate": 4.561584008198766e-06, "loss": 1.028, "step": 31810 }, { "epoch": 8.630322755627882, "grad_norm": 17.535994070033396, "learning_rate": 4.5584403936718095e-06, "loss": 1.0054, "step": 31820 }, { "epoch": 8.633034987794955, "grad_norm": 20.64440563236174, "learning_rate": 4.555296955052201e-06, "loss": 0.993, "step": 31830 }, { "epoch": 8.63574721996203, "grad_norm": 15.009299313517174, "learning_rate": 4.552153693592214e-06, "loss": 0.9544, "step": 31840 }, { "epoch": 8.638459452129101, "grad_norm": 20.23981198757384, "learning_rate": 4.549010610544055e-06, "loss": 0.9099, "step": 31850 }, { "epoch": 8.641171684296175, "grad_norm": 21.26854454127017, "learning_rate": 4.545867707159858e-06, "loss": 0.9618, "step": 31860 }, { "epoch": 8.643883916463249, "grad_norm": 20.637635765560173, "learning_rate": 4.54272498469168e-06, "loss": 0.9785, "step": 31870 }, { "epoch": 8.646596148630323, "grad_norm": 22.877869273270893, "learning_rate": 4.539582444391513e-06, "loss": 1.1687, "step": 31880 }, { "epoch": 8.649308380797397, "grad_norm": 19.565493534385592, "learning_rate": 4.5364400875112764e-06, "loss": 0.9224, "step": 31890 }, { "epoch": 8.65202061296447, "grad_norm": 22.230469919125763, "learning_rate": 4.533297915302811e-06, "loss": 0.9757, "step": 31900 }, { "epoch": 8.654732845131543, "grad_norm": 21.49865325330261, "learning_rate": 4.5301559290178884e-06, "loss": 0.9316, "step": 31910 }, { "epoch": 8.657445077298616, "grad_norm": 20.355660882484027, "learning_rate": 4.527014129908201e-06, "loss": 0.881, "step": 31920 }, { "epoch": 8.66015730946569, "grad_norm": 20.71924150299971, "learning_rate": 4.523872519225378e-06, "loss": 1.013, "step": 31930 }, { "epoch": 8.662869541632764, "grad_norm": 19.86596376318049, "learning_rate": 4.520731098220958e-06, "loss": 0.9007, "step": 31940 }, { "epoch": 8.665581773799838, "grad_norm": 20.860077167508674, "learning_rate": 4.517589868146418e-06, "loss": 1.0165, "step": 31950 }, { "epoch": 8.66829400596691, "grad_norm": 19.72845792537069, "learning_rate": 4.5144488302531475e-06, "loss": 1.001, "step": 31960 }, { "epoch": 8.671006238133984, "grad_norm": 21.804836887170357, "learning_rate": 4.5113079857924704e-06, "loss": 0.9608, "step": 31970 }, { "epoch": 8.673718470301058, "grad_norm": 18.348853783623337, "learning_rate": 4.508167336015626e-06, "loss": 0.9669, "step": 31980 }, { "epoch": 8.676430702468132, "grad_norm": 20.59325905038565, "learning_rate": 4.5050268821737755e-06, "loss": 0.9369, "step": 31990 }, { "epoch": 8.679142934635205, "grad_norm": 19.754328670444064, "learning_rate": 4.501886625518007e-06, "loss": 0.8198, "step": 32000 }, { "epoch": 8.68185516680228, "grad_norm": 19.86104298520681, "learning_rate": 4.498746567299327e-06, "loss": 0.9274, "step": 32010 }, { "epoch": 8.684567398969351, "grad_norm": 21.417218583751247, "learning_rate": 4.495606708768665e-06, "loss": 1.0547, "step": 32020 }, { "epoch": 8.687279631136425, "grad_norm": 18.89288244837402, "learning_rate": 4.492467051176869e-06, "loss": 1.0288, "step": 32030 }, { "epoch": 8.689991863303499, "grad_norm": 22.93685065523436, "learning_rate": 4.489327595774704e-06, "loss": 0.826, "step": 32040 }, { "epoch": 8.692704095470573, "grad_norm": 20.293557579139428, "learning_rate": 4.486188343812862e-06, "loss": 1.0152, "step": 32050 }, { "epoch": 8.695416327637647, "grad_norm": 23.037707980190923, "learning_rate": 4.483049296541949e-06, "loss": 0.924, "step": 32060 }, { "epoch": 8.698128559804719, "grad_norm": 18.321365823852375, "learning_rate": 4.4799104552124905e-06, "loss": 0.9976, "step": 32070 }, { "epoch": 8.700840791971792, "grad_norm": 17.78592058938146, "learning_rate": 4.476771821074928e-06, "loss": 0.9811, "step": 32080 }, { "epoch": 8.703553024138866, "grad_norm": 22.09556919669915, "learning_rate": 4.473633395379625e-06, "loss": 1.0178, "step": 32090 }, { "epoch": 8.70626525630594, "grad_norm": 22.68136688195156, "learning_rate": 4.470495179376858e-06, "loss": 0.8834, "step": 32100 }, { "epoch": 8.708977488473014, "grad_norm": 21.52107096801186, "learning_rate": 4.46735717431682e-06, "loss": 0.9895, "step": 32110 }, { "epoch": 8.711689720640086, "grad_norm": 21.28864970825043, "learning_rate": 4.4642193814496215e-06, "loss": 0.9815, "step": 32120 }, { "epoch": 8.71440195280716, "grad_norm": 17.329631095183952, "learning_rate": 4.461081802025289e-06, "loss": 0.9117, "step": 32130 }, { "epoch": 8.717114184974234, "grad_norm": 22.511774691053944, "learning_rate": 4.457944437293765e-06, "loss": 1.0017, "step": 32140 }, { "epoch": 8.719826417141308, "grad_norm": 18.91447246688527, "learning_rate": 4.454807288504901e-06, "loss": 0.8888, "step": 32150 }, { "epoch": 8.722538649308381, "grad_norm": 21.401691246992687, "learning_rate": 4.451670356908466e-06, "loss": 0.9212, "step": 32160 }, { "epoch": 8.725250881475453, "grad_norm": 22.384691009168534, "learning_rate": 4.448533643754147e-06, "loss": 0.946, "step": 32170 }, { "epoch": 8.727963113642527, "grad_norm": 17.55845111102375, "learning_rate": 4.445397150291534e-06, "loss": 1.0018, "step": 32180 }, { "epoch": 8.730675345809601, "grad_norm": 19.65008465201094, "learning_rate": 4.44226087777014e-06, "loss": 1.0537, "step": 32190 }, { "epoch": 8.733387577976675, "grad_norm": 19.911687438029677, "learning_rate": 4.43912482743938e-06, "loss": 0.9768, "step": 32200 }, { "epoch": 8.736099810143749, "grad_norm": 21.715401382220954, "learning_rate": 4.435989000548589e-06, "loss": 1.0218, "step": 32210 }, { "epoch": 8.738812042310823, "grad_norm": 21.50716142935153, "learning_rate": 4.432853398347008e-06, "loss": 1.0162, "step": 32220 }, { "epoch": 8.741524274477895, "grad_norm": 19.83850957896116, "learning_rate": 4.4297180220837915e-06, "loss": 0.9619, "step": 32230 }, { "epoch": 8.744236506644969, "grad_norm": 20.603522343538053, "learning_rate": 4.426582873007999e-06, "loss": 0.9422, "step": 32240 }, { "epoch": 8.746948738812042, "grad_norm": 22.033002696807664, "learning_rate": 4.423447952368604e-06, "loss": 0.9933, "step": 32250 }, { "epoch": 8.749660970979116, "grad_norm": 17.275684855955525, "learning_rate": 4.420313261414493e-06, "loss": 0.9218, "step": 32260 }, { "epoch": 8.75237320314619, "grad_norm": 20.119811904790666, "learning_rate": 4.417178801394449e-06, "loss": 0.9431, "step": 32270 }, { "epoch": 8.755085435313262, "grad_norm": 21.53006522664635, "learning_rate": 4.414044573557172e-06, "loss": 1.0703, "step": 32280 }, { "epoch": 8.757797667480336, "grad_norm": 17.221165699024528, "learning_rate": 4.41091057915127e-06, "loss": 1.025, "step": 32290 }, { "epoch": 8.76050989964741, "grad_norm": 19.621796047912433, "learning_rate": 4.407776819425253e-06, "loss": 1.0539, "step": 32300 }, { "epoch": 8.763222131814484, "grad_norm": 21.311706170199198, "learning_rate": 4.404643295627541e-06, "loss": 1.0101, "step": 32310 }, { "epoch": 8.765934363981557, "grad_norm": 19.178548111565, "learning_rate": 4.401510009006455e-06, "loss": 0.9211, "step": 32320 }, { "epoch": 8.76864659614863, "grad_norm": 21.192792435891157, "learning_rate": 4.3983769608102314e-06, "loss": 0.9702, "step": 32330 }, { "epoch": 8.771358828315703, "grad_norm": 22.61188675864982, "learning_rate": 4.3952441522870035e-06, "loss": 1.1068, "step": 32340 }, { "epoch": 8.774071060482777, "grad_norm": 20.789313514469104, "learning_rate": 4.392111584684811e-06, "loss": 0.9399, "step": 32350 }, { "epoch": 8.776783292649851, "grad_norm": 18.919229716324647, "learning_rate": 4.388979259251595e-06, "loss": 0.9624, "step": 32360 }, { "epoch": 8.779495524816925, "grad_norm": 20.443486329158375, "learning_rate": 4.385847177235208e-06, "loss": 1.0175, "step": 32370 }, { "epoch": 8.782207756983997, "grad_norm": 23.642307080334017, "learning_rate": 4.382715339883397e-06, "loss": 0.9787, "step": 32380 }, { "epoch": 8.78491998915107, "grad_norm": 17.234646862962006, "learning_rate": 4.3795837484438185e-06, "loss": 1.0927, "step": 32390 }, { "epoch": 8.787632221318145, "grad_norm": 18.715876240013618, "learning_rate": 4.376452404164021e-06, "loss": 0.953, "step": 32400 }, { "epoch": 8.790344453485218, "grad_norm": 20.652149161314842, "learning_rate": 4.37332130829147e-06, "loss": 0.9261, "step": 32410 }, { "epoch": 8.793056685652292, "grad_norm": 20.354247593247074, "learning_rate": 4.370190462073518e-06, "loss": 0.996, "step": 32420 }, { "epoch": 8.795768917819366, "grad_norm": 19.300639866738702, "learning_rate": 4.367059866757423e-06, "loss": 1.0345, "step": 32430 }, { "epoch": 8.798481149986438, "grad_norm": 18.5177509765031, "learning_rate": 4.363929523590343e-06, "loss": 0.9946, "step": 32440 }, { "epoch": 8.801193382153512, "grad_norm": 17.24813226000699, "learning_rate": 4.360799433819339e-06, "loss": 1.1193, "step": 32450 }, { "epoch": 8.803905614320586, "grad_norm": 22.465048184662436, "learning_rate": 4.357669598691366e-06, "loss": 1.0676, "step": 32460 }, { "epoch": 8.80661784648766, "grad_norm": 22.114655271679354, "learning_rate": 4.3545400194532795e-06, "loss": 0.9713, "step": 32470 }, { "epoch": 8.809330078654733, "grad_norm": 24.385614046740695, "learning_rate": 4.351410697351831e-06, "loss": 0.9151, "step": 32480 }, { "epoch": 8.812042310821806, "grad_norm": 18.718044585008535, "learning_rate": 4.348281633633676e-06, "loss": 0.8927, "step": 32490 }, { "epoch": 8.81475454298888, "grad_norm": 20.56640663434684, "learning_rate": 4.345152829545359e-06, "loss": 0.9562, "step": 32500 }, { "epoch": 8.817466775155953, "grad_norm": 23.01087413281028, "learning_rate": 4.342024286333326e-06, "loss": 1.003, "step": 32510 }, { "epoch": 8.820179007323027, "grad_norm": 23.048887608834693, "learning_rate": 4.3388960052439166e-06, "loss": 0.9503, "step": 32520 }, { "epoch": 8.822891239490101, "grad_norm": 19.682501803024948, "learning_rate": 4.33576798752337e-06, "loss": 0.9385, "step": 32530 }, { "epoch": 8.825603471657175, "grad_norm": 19.1490900471492, "learning_rate": 4.332640234417815e-06, "loss": 0.9017, "step": 32540 }, { "epoch": 8.828315703824247, "grad_norm": 23.113495004350064, "learning_rate": 4.329512747173279e-06, "loss": 1.0281, "step": 32550 }, { "epoch": 8.83102793599132, "grad_norm": 20.784653802140443, "learning_rate": 4.32638552703568e-06, "loss": 0.9936, "step": 32560 }, { "epoch": 8.833740168158394, "grad_norm": 20.69542718142278, "learning_rate": 4.3232585752508335e-06, "loss": 0.9678, "step": 32570 }, { "epoch": 8.836452400325468, "grad_norm": 19.87628084317382, "learning_rate": 4.320131893064447e-06, "loss": 0.9412, "step": 32580 }, { "epoch": 8.839164632492542, "grad_norm": 20.224447780758954, "learning_rate": 4.317005481722119e-06, "loss": 1.0561, "step": 32590 }, { "epoch": 8.841876864659614, "grad_norm": 20.8000193710058, "learning_rate": 4.3138793424693385e-06, "loss": 0.9965, "step": 32600 }, { "epoch": 8.844589096826688, "grad_norm": 20.172445274905158, "learning_rate": 4.3107534765514935e-06, "loss": 1.0278, "step": 32610 }, { "epoch": 8.847301328993762, "grad_norm": 21.61131587503655, "learning_rate": 4.307627885213854e-06, "loss": 1.0512, "step": 32620 }, { "epoch": 8.850013561160836, "grad_norm": 20.63505031941217, "learning_rate": 4.304502569701588e-06, "loss": 0.9709, "step": 32630 }, { "epoch": 8.85272579332791, "grad_norm": 22.440972829297245, "learning_rate": 4.301377531259746e-06, "loss": 0.9834, "step": 32640 }, { "epoch": 8.855438025494982, "grad_norm": 19.699688916868038, "learning_rate": 4.298252771133278e-06, "loss": 0.9735, "step": 32650 }, { "epoch": 8.858150257662055, "grad_norm": 17.983957092328126, "learning_rate": 4.295128290567015e-06, "loss": 1.0048, "step": 32660 }, { "epoch": 8.86086248982913, "grad_norm": 19.81484322653689, "learning_rate": 4.29200409080568e-06, "loss": 0.9476, "step": 32670 }, { "epoch": 8.863574721996203, "grad_norm": 23.01664399440499, "learning_rate": 4.28888017309388e-06, "loss": 0.9876, "step": 32680 }, { "epoch": 8.866286954163277, "grad_norm": 20.92605575027385, "learning_rate": 4.285756538676117e-06, "loss": 1.0276, "step": 32690 }, { "epoch": 8.868999186330349, "grad_norm": 23.719293981362153, "learning_rate": 4.282633188796776e-06, "loss": 0.9431, "step": 32700 }, { "epoch": 8.871711418497423, "grad_norm": 21.160750309754047, "learning_rate": 4.279510124700128e-06, "loss": 0.912, "step": 32710 }, { "epoch": 8.874423650664497, "grad_norm": 17.92201932201363, "learning_rate": 4.276387347630329e-06, "loss": 1.0658, "step": 32720 }, { "epoch": 8.87713588283157, "grad_norm": 22.606488438613102, "learning_rate": 4.273264858831428e-06, "loss": 0.9995, "step": 32730 }, { "epoch": 8.879848114998644, "grad_norm": 20.757760994457918, "learning_rate": 4.2701426595473474e-06, "loss": 1.0252, "step": 32740 }, { "epoch": 8.882560347165718, "grad_norm": 18.884158631454092, "learning_rate": 4.2670207510219055e-06, "loss": 0.9663, "step": 32750 }, { "epoch": 8.88527257933279, "grad_norm": 20.772389393515304, "learning_rate": 4.263899134498796e-06, "loss": 0.9495, "step": 32760 }, { "epoch": 8.887984811499864, "grad_norm": 18.663706944715138, "learning_rate": 4.260777811221605e-06, "loss": 0.9531, "step": 32770 }, { "epoch": 8.890697043666938, "grad_norm": 22.69282667834446, "learning_rate": 4.257656782433794e-06, "loss": 1.0593, "step": 32780 }, { "epoch": 8.893409275834012, "grad_norm": 21.556243230728505, "learning_rate": 4.254536049378712e-06, "loss": 0.9721, "step": 32790 }, { "epoch": 8.896121508001086, "grad_norm": 22.629420030777368, "learning_rate": 4.251415613299584e-06, "loss": 1.0194, "step": 32800 }, { "epoch": 8.898833740168158, "grad_norm": 21.052507641128123, "learning_rate": 4.248295475439525e-06, "loss": 0.9947, "step": 32810 }, { "epoch": 8.901545972335231, "grad_norm": 20.616078333240104, "learning_rate": 4.2451756370415275e-06, "loss": 0.9707, "step": 32820 }, { "epoch": 8.904258204502305, "grad_norm": 18.683379571203027, "learning_rate": 4.242056099348464e-06, "loss": 0.9504, "step": 32830 }, { "epoch": 8.90697043666938, "grad_norm": 20.28307004532603, "learning_rate": 4.238936863603084e-06, "loss": 1.0605, "step": 32840 }, { "epoch": 8.909682668836453, "grad_norm": 23.80589327790313, "learning_rate": 4.235817931048026e-06, "loss": 1.0337, "step": 32850 }, { "epoch": 8.912394901003527, "grad_norm": 20.268975691212574, "learning_rate": 4.2326993029258e-06, "loss": 0.9647, "step": 32860 }, { "epoch": 8.915107133170599, "grad_norm": 19.380807816914903, "learning_rate": 4.229580980478795e-06, "loss": 0.8597, "step": 32870 }, { "epoch": 8.917819365337673, "grad_norm": 21.851441112562032, "learning_rate": 4.226462964949279e-06, "loss": 0.9943, "step": 32880 }, { "epoch": 8.920531597504747, "grad_norm": 21.213991811904503, "learning_rate": 4.223345257579405e-06, "loss": 0.9131, "step": 32890 }, { "epoch": 8.92324382967182, "grad_norm": 20.12547228750864, "learning_rate": 4.220227859611193e-06, "loss": 1.0575, "step": 32900 }, { "epoch": 8.925956061838894, "grad_norm": 20.253035423947903, "learning_rate": 4.217110772286543e-06, "loss": 0.973, "step": 32910 }, { "epoch": 8.928668294005966, "grad_norm": 20.135429002057734, "learning_rate": 4.2139939968472325e-06, "loss": 1.0085, "step": 32920 }, { "epoch": 8.93138052617304, "grad_norm": 18.041383360436093, "learning_rate": 4.2108775345349155e-06, "loss": 1.0066, "step": 32930 }, { "epoch": 8.934092758340114, "grad_norm": 19.31826470623686, "learning_rate": 4.207761386591118e-06, "loss": 0.949, "step": 32940 }, { "epoch": 8.936804990507188, "grad_norm": 21.736124636049265, "learning_rate": 4.204645554257246e-06, "loss": 0.9149, "step": 32950 }, { "epoch": 8.939517222674262, "grad_norm": 20.511771207731126, "learning_rate": 4.201530038774572e-06, "loss": 1.0361, "step": 32960 }, { "epoch": 8.942229454841334, "grad_norm": 18.494879445082752, "learning_rate": 4.198414841384252e-06, "loss": 0.903, "step": 32970 }, { "epoch": 8.944941687008408, "grad_norm": 22.41016346205478, "learning_rate": 4.1952999633273065e-06, "loss": 1.0547, "step": 32980 }, { "epoch": 8.947653919175481, "grad_norm": 22.05606516901471, "learning_rate": 4.192185405844633e-06, "loss": 0.9311, "step": 32990 }, { "epoch": 8.950366151342555, "grad_norm": 21.51914153056629, "learning_rate": 4.189071170177001e-06, "loss": 1.021, "step": 33000 }, { "epoch": 8.953078383509629, "grad_norm": 17.06719011891294, "learning_rate": 4.185957257565051e-06, "loss": 0.9137, "step": 33010 }, { "epoch": 8.955790615676701, "grad_norm": 21.4204183557601, "learning_rate": 4.182843669249298e-06, "loss": 0.9634, "step": 33020 }, { "epoch": 8.958502847843775, "grad_norm": 20.276037569464446, "learning_rate": 4.1797304064701214e-06, "loss": 0.9825, "step": 33030 }, { "epoch": 8.961215080010849, "grad_norm": 21.419800936716207, "learning_rate": 4.176617470467774e-06, "loss": 1.0023, "step": 33040 }, { "epoch": 8.963927312177923, "grad_norm": 20.12649586075173, "learning_rate": 4.1735048624823836e-06, "loss": 0.9524, "step": 33050 }, { "epoch": 8.966639544344996, "grad_norm": 22.696627428327954, "learning_rate": 4.17039258375394e-06, "loss": 1.0078, "step": 33060 }, { "epoch": 8.96935177651207, "grad_norm": 20.608700246758307, "learning_rate": 4.167280635522305e-06, "loss": 0.9892, "step": 33070 }, { "epoch": 8.972064008679142, "grad_norm": 21.509993368283677, "learning_rate": 4.164169019027205e-06, "loss": 1.0718, "step": 33080 }, { "epoch": 8.974776240846216, "grad_norm": 21.96128024874989, "learning_rate": 4.161057735508245e-06, "loss": 0.9762, "step": 33090 }, { "epoch": 8.97748847301329, "grad_norm": 18.95661097182884, "learning_rate": 4.1579467862048834e-06, "loss": 0.9915, "step": 33100 }, { "epoch": 8.980200705180364, "grad_norm": 18.860294803813467, "learning_rate": 4.154836172356454e-06, "loss": 1.1222, "step": 33110 }, { "epoch": 8.982912937347438, "grad_norm": 20.304129885590616, "learning_rate": 4.151725895202155e-06, "loss": 0.9591, "step": 33120 }, { "epoch": 8.98562516951451, "grad_norm": 20.491423121423473, "learning_rate": 4.1486159559810504e-06, "loss": 1.0451, "step": 33130 }, { "epoch": 8.988337401681584, "grad_norm": 22.57247208899392, "learning_rate": 4.14550635593207e-06, "loss": 1.0613, "step": 33140 }, { "epoch": 8.991049633848657, "grad_norm": 21.180911690077103, "learning_rate": 4.142397096294008e-06, "loss": 1.1075, "step": 33150 }, { "epoch": 8.993761866015731, "grad_norm": 20.341539781531253, "learning_rate": 4.13928817830552e-06, "loss": 1.0439, "step": 33160 }, { "epoch": 8.996474098182805, "grad_norm": 18.622682881659752, "learning_rate": 4.136179603205134e-06, "loss": 0.9349, "step": 33170 }, { "epoch": 8.999186330349877, "grad_norm": 22.43128443099554, "learning_rate": 4.133071372231232e-06, "loss": 0.9717, "step": 33180 }, { "epoch": 9.001898562516951, "grad_norm": 15.572227014051018, "learning_rate": 4.129963486622064e-06, "loss": 0.7328, "step": 33190 }, { "epoch": 9.004610794684025, "grad_norm": 17.80998977958041, "learning_rate": 4.12685594761574e-06, "loss": 0.6472, "step": 33200 }, { "epoch": 9.007323026851099, "grad_norm": 15.562237506916865, "learning_rate": 4.123748756450236e-06, "loss": 0.6193, "step": 33210 }, { "epoch": 9.010035259018172, "grad_norm": 18.72847828960028, "learning_rate": 4.120641914363385e-06, "loss": 0.6106, "step": 33220 }, { "epoch": 9.012747491185246, "grad_norm": 17.404447993031116, "learning_rate": 4.117535422592881e-06, "loss": 0.6299, "step": 33230 }, { "epoch": 9.015459723352318, "grad_norm": 19.658703804164446, "learning_rate": 4.114429282376282e-06, "loss": 0.6632, "step": 33240 }, { "epoch": 9.018171955519392, "grad_norm": 18.82239685719204, "learning_rate": 4.1113234949510035e-06, "loss": 0.6585, "step": 33250 }, { "epoch": 9.020884187686466, "grad_norm": 20.473427417277993, "learning_rate": 4.108218061554321e-06, "loss": 0.6456, "step": 33260 }, { "epoch": 9.02359641985354, "grad_norm": 18.74869507152634, "learning_rate": 4.1051129834233675e-06, "loss": 0.6515, "step": 33270 }, { "epoch": 9.026308652020614, "grad_norm": 20.841426129347376, "learning_rate": 4.102008261795139e-06, "loss": 0.6335, "step": 33280 }, { "epoch": 9.029020884187686, "grad_norm": 17.626026423288813, "learning_rate": 4.098903897906483e-06, "loss": 0.6491, "step": 33290 }, { "epoch": 9.03173311635476, "grad_norm": 19.088469209876333, "learning_rate": 4.09579989299411e-06, "loss": 0.7083, "step": 33300 }, { "epoch": 9.034445348521833, "grad_norm": 19.76990848054798, "learning_rate": 4.092696248294581e-06, "loss": 0.6611, "step": 33310 }, { "epoch": 9.037157580688907, "grad_norm": 19.150839204360185, "learning_rate": 4.089592965044323e-06, "loss": 0.6252, "step": 33320 }, { "epoch": 9.039869812855981, "grad_norm": 18.214212686363705, "learning_rate": 4.086490044479612e-06, "loss": 0.6157, "step": 33330 }, { "epoch": 9.042582045023053, "grad_norm": 18.078844261345623, "learning_rate": 4.083387487836582e-06, "loss": 0.6767, "step": 33340 }, { "epoch": 9.045294277190127, "grad_norm": 14.36443030059754, "learning_rate": 4.080285296351218e-06, "loss": 0.6252, "step": 33350 }, { "epoch": 9.0480065093572, "grad_norm": 15.794499563918684, "learning_rate": 4.077183471259367e-06, "loss": 0.6262, "step": 33360 }, { "epoch": 9.050718741524275, "grad_norm": 19.308916237405835, "learning_rate": 4.074082013796724e-06, "loss": 0.6761, "step": 33370 }, { "epoch": 9.053430973691349, "grad_norm": 14.287729382383203, "learning_rate": 4.070980925198838e-06, "loss": 0.6027, "step": 33380 }, { "epoch": 9.056143205858422, "grad_norm": 18.176456846496425, "learning_rate": 4.067880206701114e-06, "loss": 0.6415, "step": 33390 }, { "epoch": 9.058855438025494, "grad_norm": 20.706921844204793, "learning_rate": 4.064779859538809e-06, "loss": 0.6544, "step": 33400 }, { "epoch": 9.061567670192568, "grad_norm": 17.310968087630012, "learning_rate": 4.061679884947031e-06, "loss": 0.6743, "step": 33410 }, { "epoch": 9.064279902359642, "grad_norm": 21.73186218810043, "learning_rate": 4.058580284160739e-06, "loss": 0.6316, "step": 33420 }, { "epoch": 9.066992134526716, "grad_norm": 17.489000082393822, "learning_rate": 4.0554810584147404e-06, "loss": 0.5903, "step": 33430 }, { "epoch": 9.06970436669379, "grad_norm": 14.230334977680926, "learning_rate": 4.0523822089437025e-06, "loss": 0.6181, "step": 33440 }, { "epoch": 9.072416598860862, "grad_norm": 15.69441995641486, "learning_rate": 4.049283736982133e-06, "loss": 0.6657, "step": 33450 }, { "epoch": 9.075128831027936, "grad_norm": 18.66604728909722, "learning_rate": 4.046185643764395e-06, "loss": 0.6677, "step": 33460 }, { "epoch": 9.07784106319501, "grad_norm": 18.731044414712684, "learning_rate": 4.043087930524695e-06, "loss": 0.6795, "step": 33470 }, { "epoch": 9.080553295362083, "grad_norm": 22.383611025645195, "learning_rate": 4.039990598497097e-06, "loss": 0.7026, "step": 33480 }, { "epoch": 9.083265527529157, "grad_norm": 14.610596367238223, "learning_rate": 4.0368936489155055e-06, "loss": 0.6141, "step": 33490 }, { "epoch": 9.08597775969623, "grad_norm": 17.674306716755307, "learning_rate": 4.033797083013673e-06, "loss": 0.6678, "step": 33500 }, { "epoch": 9.088689991863303, "grad_norm": 18.128185321092737, "learning_rate": 4.030700902025202e-06, "loss": 0.6186, "step": 33510 }, { "epoch": 9.091402224030377, "grad_norm": 21.217963659708687, "learning_rate": 4.0276051071835455e-06, "loss": 0.6587, "step": 33520 }, { "epoch": 9.09411445619745, "grad_norm": 21.94525932204732, "learning_rate": 4.024509699721994e-06, "loss": 0.6695, "step": 33530 }, { "epoch": 9.096826688364525, "grad_norm": 20.752645088180667, "learning_rate": 4.021414680873688e-06, "loss": 0.6406, "step": 33540 }, { "epoch": 9.099538920531597, "grad_norm": 24.08416955587967, "learning_rate": 4.0183200518716114e-06, "loss": 0.6687, "step": 33550 }, { "epoch": 9.10225115269867, "grad_norm": 17.138689566978876, "learning_rate": 4.015225813948599e-06, "loss": 0.6139, "step": 33560 }, { "epoch": 9.104963384865744, "grad_norm": 20.598241427482193, "learning_rate": 4.012131968337321e-06, "loss": 0.6375, "step": 33570 }, { "epoch": 9.107675617032818, "grad_norm": 17.125384510984432, "learning_rate": 4.009038516270297e-06, "loss": 0.6739, "step": 33580 }, { "epoch": 9.110387849199892, "grad_norm": 16.65429401663914, "learning_rate": 4.005945458979886e-06, "loss": 0.623, "step": 33590 }, { "epoch": 9.113100081366966, "grad_norm": 19.78634067760296, "learning_rate": 4.0028527976982965e-06, "loss": 0.7113, "step": 33600 }, { "epoch": 9.115812313534038, "grad_norm": 22.269781465122737, "learning_rate": 3.999760533657573e-06, "loss": 0.6049, "step": 33610 }, { "epoch": 9.118524545701112, "grad_norm": 21.006813095723743, "learning_rate": 3.9966686680896005e-06, "loss": 0.6498, "step": 33620 }, { "epoch": 9.121236777868186, "grad_norm": 18.423199163599055, "learning_rate": 3.993577202226111e-06, "loss": 0.5838, "step": 33630 }, { "epoch": 9.12394901003526, "grad_norm": 18.001904517347505, "learning_rate": 3.990486137298673e-06, "loss": 0.6155, "step": 33640 }, { "epoch": 9.126661242202333, "grad_norm": 17.814590702767024, "learning_rate": 3.9873954745387e-06, "loss": 0.5794, "step": 33650 }, { "epoch": 9.129373474369405, "grad_norm": 18.290755710793235, "learning_rate": 3.984305215177439e-06, "loss": 0.6279, "step": 33660 }, { "epoch": 9.132085706536479, "grad_norm": 17.987780604926815, "learning_rate": 3.981215360445978e-06, "loss": 0.6434, "step": 33670 }, { "epoch": 9.134797938703553, "grad_norm": 18.40070604071608, "learning_rate": 3.9781259115752495e-06, "loss": 0.6809, "step": 33680 }, { "epoch": 9.137510170870627, "grad_norm": 21.67871441049607, "learning_rate": 3.975036869796017e-06, "loss": 0.6681, "step": 33690 }, { "epoch": 9.1402224030377, "grad_norm": 20.921290918960885, "learning_rate": 3.971948236338886e-06, "loss": 0.6258, "step": 33700 }, { "epoch": 9.142934635204773, "grad_norm": 19.138208636547454, "learning_rate": 3.968860012434295e-06, "loss": 0.6519, "step": 33710 }, { "epoch": 9.145646867371847, "grad_norm": 18.149690259409, "learning_rate": 3.965772199312529e-06, "loss": 0.6609, "step": 33720 }, { "epoch": 9.14835909953892, "grad_norm": 19.93910443188718, "learning_rate": 3.962684798203697e-06, "loss": 0.686, "step": 33730 }, { "epoch": 9.151071331705994, "grad_norm": 17.977501861538848, "learning_rate": 3.959597810337752e-06, "loss": 0.7088, "step": 33740 }, { "epoch": 9.153783563873068, "grad_norm": 21.15123811908339, "learning_rate": 3.956511236944479e-06, "loss": 0.6402, "step": 33750 }, { "epoch": 9.156495796040142, "grad_norm": 21.429399690383107, "learning_rate": 3.9534250792535e-06, "loss": 0.6498, "step": 33760 }, { "epoch": 9.159208028207214, "grad_norm": 17.875739445161603, "learning_rate": 3.950339338494271e-06, "loss": 0.6841, "step": 33770 }, { "epoch": 9.161920260374288, "grad_norm": 15.702458366287019, "learning_rate": 3.94725401589608e-06, "loss": 0.5689, "step": 33780 }, { "epoch": 9.164632492541362, "grad_norm": 18.977506979710903, "learning_rate": 3.9441691126880475e-06, "loss": 0.6656, "step": 33790 }, { "epoch": 9.167344724708435, "grad_norm": 16.541167891931256, "learning_rate": 3.941084630099134e-06, "loss": 0.6423, "step": 33800 }, { "epoch": 9.17005695687551, "grad_norm": 18.670544186957233, "learning_rate": 3.938000569358124e-06, "loss": 0.596, "step": 33810 }, { "epoch": 9.172769189042581, "grad_norm": 17.789394670905818, "learning_rate": 3.9349169316936394e-06, "loss": 0.687, "step": 33820 }, { "epoch": 9.175481421209655, "grad_norm": 21.499909261291393, "learning_rate": 3.931833718334128e-06, "loss": 0.6642, "step": 33830 }, { "epoch": 9.178193653376729, "grad_norm": 18.77480207574815, "learning_rate": 3.928750930507878e-06, "loss": 0.6373, "step": 33840 }, { "epoch": 9.180905885543803, "grad_norm": 19.04431123033377, "learning_rate": 3.9256685694429976e-06, "loss": 0.6079, "step": 33850 }, { "epoch": 9.183618117710877, "grad_norm": 24.655162434540266, "learning_rate": 3.922586636367432e-06, "loss": 0.7195, "step": 33860 }, { "epoch": 9.186330349877949, "grad_norm": 17.81714744198897, "learning_rate": 3.919505132508951e-06, "loss": 0.6179, "step": 33870 }, { "epoch": 9.189042582045023, "grad_norm": 17.96058276783991, "learning_rate": 3.916424059095158e-06, "loss": 0.6666, "step": 33880 }, { "epoch": 9.191754814212096, "grad_norm": 17.025787918389984, "learning_rate": 3.9133434173534835e-06, "loss": 0.5947, "step": 33890 }, { "epoch": 9.19446704637917, "grad_norm": 19.45692282452676, "learning_rate": 3.910263208511184e-06, "loss": 0.6597, "step": 33900 }, { "epoch": 9.197179278546244, "grad_norm": 20.291756103182045, "learning_rate": 3.9071834337953444e-06, "loss": 0.6449, "step": 33910 }, { "epoch": 9.199891510713318, "grad_norm": 22.086815045733204, "learning_rate": 3.90410409443288e-06, "loss": 0.6275, "step": 33920 }, { "epoch": 9.20260374288039, "grad_norm": 20.144929453959296, "learning_rate": 3.901025191650527e-06, "loss": 0.6142, "step": 33930 }, { "epoch": 9.205315975047464, "grad_norm": 18.6882568118745, "learning_rate": 3.897946726674852e-06, "loss": 0.7329, "step": 33940 }, { "epoch": 9.208028207214538, "grad_norm": 22.05474627665476, "learning_rate": 3.894868700732244e-06, "loss": 0.6368, "step": 33950 }, { "epoch": 9.210740439381611, "grad_norm": 17.998296298257536, "learning_rate": 3.891791115048922e-06, "loss": 0.5917, "step": 33960 }, { "epoch": 9.213452671548685, "grad_norm": 16.43005713391846, "learning_rate": 3.888713970850924e-06, "loss": 0.5908, "step": 33970 }, { "epoch": 9.216164903715757, "grad_norm": 19.921217756662177, "learning_rate": 3.885637269364117e-06, "loss": 0.6703, "step": 33980 }, { "epoch": 9.218877135882831, "grad_norm": 19.90174098795677, "learning_rate": 3.882561011814183e-06, "loss": 0.7057, "step": 33990 }, { "epoch": 9.221589368049905, "grad_norm": 20.068140295087026, "learning_rate": 3.879485199426641e-06, "loss": 0.6828, "step": 34000 }, { "epoch": 9.224301600216979, "grad_norm": 17.084202953252245, "learning_rate": 3.876409833426821e-06, "loss": 0.6639, "step": 34010 }, { "epoch": 9.227013832384053, "grad_norm": 19.31614241235333, "learning_rate": 3.87333491503988e-06, "loss": 0.6778, "step": 34020 }, { "epoch": 9.229726064551125, "grad_norm": 18.357059769027874, "learning_rate": 3.870260445490794e-06, "loss": 0.647, "step": 34030 }, { "epoch": 9.232438296718199, "grad_norm": 17.651504827302368, "learning_rate": 3.867186426004367e-06, "loss": 0.6708, "step": 34040 }, { "epoch": 9.235150528885272, "grad_norm": 18.572927528459417, "learning_rate": 3.8641128578052144e-06, "loss": 0.6308, "step": 34050 }, { "epoch": 9.237862761052346, "grad_norm": 19.512928178595597, "learning_rate": 3.861039742117777e-06, "loss": 0.6682, "step": 34060 }, { "epoch": 9.24057499321942, "grad_norm": 18.259936253234233, "learning_rate": 3.857967080166314e-06, "loss": 0.6729, "step": 34070 }, { "epoch": 9.243287225386494, "grad_norm": 19.229565777583176, "learning_rate": 3.854894873174906e-06, "loss": 0.7022, "step": 34080 }, { "epoch": 9.245999457553566, "grad_norm": 17.589764499791627, "learning_rate": 3.8518231223674504e-06, "loss": 0.6953, "step": 34090 }, { "epoch": 9.24871168972064, "grad_norm": 20.523121064888457, "learning_rate": 3.848751828967663e-06, "loss": 0.654, "step": 34100 }, { "epoch": 9.251423921887714, "grad_norm": 19.6626731018816, "learning_rate": 3.8456809941990745e-06, "loss": 0.6019, "step": 34110 }, { "epoch": 9.254136154054788, "grad_norm": 17.605240951461393, "learning_rate": 3.842610619285041e-06, "loss": 0.651, "step": 34120 }, { "epoch": 9.256848386221861, "grad_norm": 19.806915717182623, "learning_rate": 3.8395407054487276e-06, "loss": 0.6313, "step": 34130 }, { "epoch": 9.259560618388933, "grad_norm": 21.75351012928336, "learning_rate": 3.836471253913119e-06, "loss": 0.6408, "step": 34140 }, { "epoch": 9.262272850556007, "grad_norm": 18.065586959837116, "learning_rate": 3.833402265901013e-06, "loss": 0.6207, "step": 34150 }, { "epoch": 9.264985082723081, "grad_norm": 21.755739112625275, "learning_rate": 3.830333742635029e-06, "loss": 0.6018, "step": 34160 }, { "epoch": 9.267697314890155, "grad_norm": 18.099655240044996, "learning_rate": 3.827265685337595e-06, "loss": 0.6086, "step": 34170 }, { "epoch": 9.270409547057229, "grad_norm": 18.812571616560017, "learning_rate": 3.824198095230955e-06, "loss": 0.7406, "step": 34180 }, { "epoch": 9.2731217792243, "grad_norm": 17.715266703391034, "learning_rate": 3.821130973537167e-06, "loss": 0.6403, "step": 34190 }, { "epoch": 9.275834011391375, "grad_norm": 17.42581952292213, "learning_rate": 3.818064321478106e-06, "loss": 0.692, "step": 34200 }, { "epoch": 9.278546243558448, "grad_norm": 19.98376826326203, "learning_rate": 3.814998140275455e-06, "loss": 0.6803, "step": 34210 }, { "epoch": 9.281258475725522, "grad_norm": 19.199573361883388, "learning_rate": 3.8119324311507095e-06, "loss": 0.6367, "step": 34220 }, { "epoch": 9.283970707892596, "grad_norm": 18.37622205031883, "learning_rate": 3.8088671953251787e-06, "loss": 0.6491, "step": 34230 }, { "epoch": 9.28668294005967, "grad_norm": 21.44459657347252, "learning_rate": 3.8058024340199864e-06, "loss": 0.6903, "step": 34240 }, { "epoch": 9.289395172226742, "grad_norm": 14.448590752168863, "learning_rate": 3.802738148456061e-06, "loss": 0.6005, "step": 34250 }, { "epoch": 9.292107404393816, "grad_norm": 15.729326184284131, "learning_rate": 3.7996743398541463e-06, "loss": 0.7229, "step": 34260 }, { "epoch": 9.29481963656089, "grad_norm": 21.36332866022379, "learning_rate": 3.796611009434792e-06, "loss": 0.7032, "step": 34270 }, { "epoch": 9.297531868727964, "grad_norm": 21.22385447191357, "learning_rate": 3.7935481584183625e-06, "loss": 0.7059, "step": 34280 }, { "epoch": 9.300244100895037, "grad_norm": 22.09619576611952, "learning_rate": 3.790485788025027e-06, "loss": 0.7051, "step": 34290 }, { "epoch": 9.30295633306211, "grad_norm": 20.686754664182608, "learning_rate": 3.787423899474765e-06, "loss": 0.6571, "step": 34300 }, { "epoch": 9.305668565229183, "grad_norm": 19.029106185446985, "learning_rate": 3.7843624939873604e-06, "loss": 0.7075, "step": 34310 }, { "epoch": 9.308380797396257, "grad_norm": 20.938781721033916, "learning_rate": 3.781301572782411e-06, "loss": 0.6716, "step": 34320 }, { "epoch": 9.311093029563331, "grad_norm": 22.079967164824918, "learning_rate": 3.7782411370793193e-06, "loss": 0.698, "step": 34330 }, { "epoch": 9.313805261730405, "grad_norm": 17.22307446021487, "learning_rate": 3.775181188097292e-06, "loss": 0.6284, "step": 34340 }, { "epoch": 9.316517493897477, "grad_norm": 20.291023218998735, "learning_rate": 3.7721217270553413e-06, "loss": 0.7078, "step": 34350 }, { "epoch": 9.31922972606455, "grad_norm": 17.572390673249185, "learning_rate": 3.7690627551722917e-06, "loss": 0.6112, "step": 34360 }, { "epoch": 9.321941958231625, "grad_norm": 15.072224891162108, "learning_rate": 3.7660042736667667e-06, "loss": 0.6668, "step": 34370 }, { "epoch": 9.324654190398698, "grad_norm": 19.385783972817748, "learning_rate": 3.762946283757194e-06, "loss": 0.655, "step": 34380 }, { "epoch": 9.327366422565772, "grad_norm": 15.387313313970337, "learning_rate": 3.759888786661808e-06, "loss": 0.6318, "step": 34390 }, { "epoch": 9.330078654732844, "grad_norm": 21.589140735884577, "learning_rate": 3.7568317835986494e-06, "loss": 0.6486, "step": 34400 }, { "epoch": 9.332790886899918, "grad_norm": 19.70057630858124, "learning_rate": 3.753775275785557e-06, "loss": 0.6844, "step": 34410 }, { "epoch": 9.335503119066992, "grad_norm": 15.962351398088265, "learning_rate": 3.7507192644401746e-06, "loss": 0.6797, "step": 34420 }, { "epoch": 9.338215351234066, "grad_norm": 29.112982736457315, "learning_rate": 3.747663750779945e-06, "loss": 0.68, "step": 34430 }, { "epoch": 9.34092758340114, "grad_norm": 17.27346296343708, "learning_rate": 3.7446087360221195e-06, "loss": 0.658, "step": 34440 }, { "epoch": 9.343639815568213, "grad_norm": 18.680267157623017, "learning_rate": 3.7415542213837463e-06, "loss": 0.6894, "step": 34450 }, { "epoch": 9.346352047735285, "grad_norm": 21.25235046989756, "learning_rate": 3.7385002080816745e-06, "loss": 0.6525, "step": 34460 }, { "epoch": 9.34906427990236, "grad_norm": 15.839837173535981, "learning_rate": 3.735446697332551e-06, "loss": 0.6348, "step": 34470 }, { "epoch": 9.351776512069433, "grad_norm": 18.60375248142456, "learning_rate": 3.73239369035283e-06, "loss": 0.6775, "step": 34480 }, { "epoch": 9.354488744236507, "grad_norm": 17.073549713179766, "learning_rate": 3.729341188358758e-06, "loss": 0.5761, "step": 34490 }, { "epoch": 9.35720097640358, "grad_norm": 19.778925992036193, "learning_rate": 3.7262891925663813e-06, "loss": 0.6867, "step": 34500 }, { "epoch": 9.359913208570653, "grad_norm": 18.831563714401685, "learning_rate": 3.723237704191547e-06, "loss": 0.6683, "step": 34510 }, { "epoch": 9.362625440737727, "grad_norm": 19.22893675213191, "learning_rate": 3.7201867244499012e-06, "loss": 0.6562, "step": 34520 }, { "epoch": 9.3653376729048, "grad_norm": 18.165361665609908, "learning_rate": 3.7171362545568835e-06, "loss": 0.6233, "step": 34530 }, { "epoch": 9.368049905071874, "grad_norm": 18.34115199605992, "learning_rate": 3.7140862957277314e-06, "loss": 0.6148, "step": 34540 }, { "epoch": 9.370762137238948, "grad_norm": 19.16591763585986, "learning_rate": 3.7110368491774783e-06, "loss": 0.5643, "step": 34550 }, { "epoch": 9.37347436940602, "grad_norm": 20.047368503226764, "learning_rate": 3.7079879161209577e-06, "loss": 0.6631, "step": 34560 }, { "epoch": 9.376186601573094, "grad_norm": 18.984291741050193, "learning_rate": 3.704939497772793e-06, "loss": 0.6573, "step": 34570 }, { "epoch": 9.378898833740168, "grad_norm": 20.61051889127465, "learning_rate": 3.701891595347408e-06, "loss": 0.7061, "step": 34580 }, { "epoch": 9.381611065907242, "grad_norm": 19.81861290743322, "learning_rate": 3.698844210059013e-06, "loss": 0.6919, "step": 34590 }, { "epoch": 9.384323298074316, "grad_norm": 17.29735726518582, "learning_rate": 3.695797343121622e-06, "loss": 0.6202, "step": 34600 }, { "epoch": 9.38703553024139, "grad_norm": 18.799768359553298, "learning_rate": 3.692750995749037e-06, "loss": 0.6367, "step": 34610 }, { "epoch": 9.389747762408462, "grad_norm": 19.051076059343146, "learning_rate": 3.689705169154851e-06, "loss": 0.667, "step": 34620 }, { "epoch": 9.392459994575535, "grad_norm": 21.67165532448825, "learning_rate": 3.6866598645524555e-06, "loss": 0.6808, "step": 34630 }, { "epoch": 9.39517222674261, "grad_norm": 18.952517937370924, "learning_rate": 3.683615083155029e-06, "loss": 0.6557, "step": 34640 }, { "epoch": 9.397884458909683, "grad_norm": 20.024539293538712, "learning_rate": 3.6805708261755457e-06, "loss": 0.6621, "step": 34650 }, { "epoch": 9.400596691076757, "grad_norm": 18.65481492696159, "learning_rate": 3.677527094826766e-06, "loss": 0.6502, "step": 34660 }, { "epoch": 9.403308923243829, "grad_norm": 18.694229101099403, "learning_rate": 3.674483890321243e-06, "loss": 0.6706, "step": 34670 }, { "epoch": 9.406021155410903, "grad_norm": 18.305304670233582, "learning_rate": 3.671441213871325e-06, "loss": 0.5976, "step": 34680 }, { "epoch": 9.408733387577977, "grad_norm": 19.491004879928788, "learning_rate": 3.6683990666891406e-06, "loss": 0.6734, "step": 34690 }, { "epoch": 9.41144561974505, "grad_norm": 18.20109447476541, "learning_rate": 3.6653574499866153e-06, "loss": 0.6755, "step": 34700 }, { "epoch": 9.414157851912124, "grad_norm": 18.615781547713485, "learning_rate": 3.6623163649754573e-06, "loss": 0.6676, "step": 34710 }, { "epoch": 9.416870084079196, "grad_norm": 20.443768661679833, "learning_rate": 3.65927581286717e-06, "loss": 0.6963, "step": 34720 }, { "epoch": 9.41958231624627, "grad_norm": 18.431961613093282, "learning_rate": 3.656235794873039e-06, "loss": 0.7155, "step": 34730 }, { "epoch": 9.422294548413344, "grad_norm": 19.47101646495912, "learning_rate": 3.653196312204136e-06, "loss": 0.6242, "step": 34740 }, { "epoch": 9.425006780580418, "grad_norm": 21.508019429659964, "learning_rate": 3.650157366071324e-06, "loss": 0.6997, "step": 34750 }, { "epoch": 9.427719012747492, "grad_norm": 19.748465729972224, "learning_rate": 3.6471189576852516e-06, "loss": 0.6846, "step": 34760 }, { "epoch": 9.430431244914566, "grad_norm": 16.53182915553078, "learning_rate": 3.6440810882563515e-06, "loss": 0.6722, "step": 34770 }, { "epoch": 9.433143477081638, "grad_norm": 20.442521299538285, "learning_rate": 3.641043758994841e-06, "loss": 0.6655, "step": 34780 }, { "epoch": 9.435855709248711, "grad_norm": 17.606951348614373, "learning_rate": 3.6380069711107213e-06, "loss": 0.7136, "step": 34790 }, { "epoch": 9.438567941415785, "grad_norm": 15.517758800120859, "learning_rate": 3.6349707258137847e-06, "loss": 0.5831, "step": 34800 }, { "epoch": 9.441280173582859, "grad_norm": 20.739459895090416, "learning_rate": 3.6319350243136e-06, "loss": 0.6681, "step": 34810 }, { "epoch": 9.443992405749933, "grad_norm": 18.053173856695913, "learning_rate": 3.628899867819522e-06, "loss": 0.729, "step": 34820 }, { "epoch": 9.446704637917005, "grad_norm": 21.213625126572317, "learning_rate": 3.625865257540686e-06, "loss": 0.6485, "step": 34830 }, { "epoch": 9.449416870084079, "grad_norm": 22.097370024778712, "learning_rate": 3.622831194686017e-06, "loss": 0.7364, "step": 34840 }, { "epoch": 9.452129102251153, "grad_norm": 20.835898272391148, "learning_rate": 3.619797680464215e-06, "loss": 0.6167, "step": 34850 }, { "epoch": 9.454841334418226, "grad_norm": 20.49853880594271, "learning_rate": 3.616764716083762e-06, "loss": 0.6652, "step": 34860 }, { "epoch": 9.4575535665853, "grad_norm": 16.93925236365519, "learning_rate": 3.6137323027529215e-06, "loss": 0.6937, "step": 34870 }, { "epoch": 9.460265798752372, "grad_norm": 23.670206890687393, "learning_rate": 3.6107004416797404e-06, "loss": 0.7062, "step": 34880 }, { "epoch": 9.462978030919446, "grad_norm": 19.068295627680325, "learning_rate": 3.607669134072044e-06, "loss": 0.7249, "step": 34890 }, { "epoch": 9.46569026308652, "grad_norm": 20.781059272716856, "learning_rate": 3.6046383811374357e-06, "loss": 0.6524, "step": 34900 }, { "epoch": 9.468402495253594, "grad_norm": 19.52735972714015, "learning_rate": 3.601608184083296e-06, "loss": 0.623, "step": 34910 }, { "epoch": 9.471114727420668, "grad_norm": 17.566112446558506, "learning_rate": 3.598578544116793e-06, "loss": 0.5817, "step": 34920 }, { "epoch": 9.473826959587742, "grad_norm": 19.195716588970317, "learning_rate": 3.595549462444862e-06, "loss": 0.6773, "step": 34930 }, { "epoch": 9.476539191754814, "grad_norm": 20.932809100373987, "learning_rate": 3.592520940274221e-06, "loss": 0.6436, "step": 34940 }, { "epoch": 9.479251423921887, "grad_norm": 18.86666300031253, "learning_rate": 3.5894929788113642e-06, "loss": 0.6661, "step": 34950 }, { "epoch": 9.481963656088961, "grad_norm": 17.89479807246198, "learning_rate": 3.5864655792625663e-06, "loss": 0.68, "step": 34960 }, { "epoch": 9.484675888256035, "grad_norm": 16.636882052676008, "learning_rate": 3.5834387428338726e-06, "loss": 0.6662, "step": 34970 }, { "epoch": 9.487388120423109, "grad_norm": 20.85890930914326, "learning_rate": 3.5804124707311057e-06, "loss": 0.7003, "step": 34980 }, { "epoch": 9.490100352590181, "grad_norm": 20.29162645589901, "learning_rate": 3.577386764159862e-06, "loss": 0.7293, "step": 34990 }, { "epoch": 9.492812584757255, "grad_norm": 20.57104494276753, "learning_rate": 3.5743616243255187e-06, "loss": 0.5645, "step": 35000 }, { "epoch": 9.495524816924329, "grad_norm": 20.59504676438792, "learning_rate": 3.5713370524332202e-06, "loss": 0.8047, "step": 35010 }, { "epoch": 9.498237049091403, "grad_norm": 17.782276817387753, "learning_rate": 3.5683130496878894e-06, "loss": 0.6669, "step": 35020 }, { "epoch": 9.500949281258476, "grad_norm": 21.97116262889878, "learning_rate": 3.5652896172942176e-06, "loss": 0.6759, "step": 35030 }, { "epoch": 9.503661513425548, "grad_norm": 19.425328800003175, "learning_rate": 3.5622667564566747e-06, "loss": 0.7428, "step": 35040 }, { "epoch": 9.506373745592622, "grad_norm": 18.247578955005203, "learning_rate": 3.5592444683795e-06, "loss": 0.6367, "step": 35050 }, { "epoch": 9.509085977759696, "grad_norm": 16.52976061329653, "learning_rate": 3.5562227542667016e-06, "loss": 0.6063, "step": 35060 }, { "epoch": 9.51179820992677, "grad_norm": 18.919780550214323, "learning_rate": 3.5532016153220645e-06, "loss": 0.629, "step": 35070 }, { "epoch": 9.514510442093844, "grad_norm": 20.389136488482958, "learning_rate": 3.550181052749141e-06, "loss": 0.6337, "step": 35080 }, { "epoch": 9.517222674260918, "grad_norm": 23.13374566169376, "learning_rate": 3.5471610677512564e-06, "loss": 0.6716, "step": 35090 }, { "epoch": 9.51993490642799, "grad_norm": 23.40036177542879, "learning_rate": 3.544141661531503e-06, "loss": 0.6169, "step": 35100 }, { "epoch": 9.522647138595064, "grad_norm": 19.80788906748349, "learning_rate": 3.5411228352927424e-06, "loss": 0.5879, "step": 35110 }, { "epoch": 9.525359370762137, "grad_norm": 18.792268466944982, "learning_rate": 3.53810459023761e-06, "loss": 0.6653, "step": 35120 }, { "epoch": 9.528071602929211, "grad_norm": 18.160408433509588, "learning_rate": 3.5350869275685035e-06, "loss": 0.6812, "step": 35130 }, { "epoch": 9.530783835096285, "grad_norm": 22.741240176903755, "learning_rate": 3.5320698484875936e-06, "loss": 0.7162, "step": 35140 }, { "epoch": 9.533496067263357, "grad_norm": 16.23128541548574, "learning_rate": 3.529053354196813e-06, "loss": 0.6706, "step": 35150 }, { "epoch": 9.536208299430431, "grad_norm": 18.962845778041522, "learning_rate": 3.5260374458978693e-06, "loss": 0.7024, "step": 35160 }, { "epoch": 9.538920531597505, "grad_norm": 17.902124540410124, "learning_rate": 3.5230221247922293e-06, "loss": 0.7193, "step": 35170 }, { "epoch": 9.541632763764579, "grad_norm": 19.54498932947124, "learning_rate": 3.5200073920811274e-06, "loss": 0.6269, "step": 35180 }, { "epoch": 9.544344995931652, "grad_norm": 19.907127199012322, "learning_rate": 3.5169932489655668e-06, "loss": 0.6919, "step": 35190 }, { "epoch": 9.547057228098724, "grad_norm": 15.383817201823218, "learning_rate": 3.5139796966463134e-06, "loss": 0.652, "step": 35200 }, { "epoch": 9.549769460265798, "grad_norm": 20.127308792643262, "learning_rate": 3.5109667363239e-06, "loss": 0.6101, "step": 35210 }, { "epoch": 9.552481692432872, "grad_norm": 19.016030715205737, "learning_rate": 3.5079543691986183e-06, "loss": 0.6072, "step": 35220 }, { "epoch": 9.555193924599946, "grad_norm": 20.339902335314513, "learning_rate": 3.5049425964705285e-06, "loss": 0.6205, "step": 35230 }, { "epoch": 9.55790615676702, "grad_norm": 22.4114325278595, "learning_rate": 3.5019314193394544e-06, "loss": 0.6839, "step": 35240 }, { "epoch": 9.560618388934092, "grad_norm": 17.084590549729747, "learning_rate": 3.4989208390049784e-06, "loss": 0.6463, "step": 35250 }, { "epoch": 9.563330621101166, "grad_norm": 19.212718272076945, "learning_rate": 3.4959108566664496e-06, "loss": 0.6071, "step": 35260 }, { "epoch": 9.56604285326824, "grad_norm": 17.972429928846125, "learning_rate": 3.492901473522974e-06, "loss": 0.5842, "step": 35270 }, { "epoch": 9.568755085435313, "grad_norm": 19.034255250801724, "learning_rate": 3.4898926907734243e-06, "loss": 0.644, "step": 35280 }, { "epoch": 9.571467317602387, "grad_norm": 21.99648209057993, "learning_rate": 3.4868845096164317e-06, "loss": 0.6592, "step": 35290 }, { "epoch": 9.574179549769461, "grad_norm": 18.165479683180962, "learning_rate": 3.483876931250386e-06, "loss": 0.629, "step": 35300 }, { "epoch": 9.576891781936533, "grad_norm": 20.28824575955119, "learning_rate": 3.4808699568734365e-06, "loss": 0.7318, "step": 35310 }, { "epoch": 9.579604014103607, "grad_norm": 19.417117543522114, "learning_rate": 3.477863587683497e-06, "loss": 0.698, "step": 35320 }, { "epoch": 9.58231624627068, "grad_norm": 17.278604020770615, "learning_rate": 3.4748578248782367e-06, "loss": 0.6626, "step": 35330 }, { "epoch": 9.585028478437755, "grad_norm": 19.43727316405258, "learning_rate": 3.4718526696550825e-06, "loss": 0.7043, "step": 35340 }, { "epoch": 9.587740710604828, "grad_norm": 19.31122490326134, "learning_rate": 3.468848123211218e-06, "loss": 0.6213, "step": 35350 }, { "epoch": 9.5904529427719, "grad_norm": 20.118719255592282, "learning_rate": 3.4658441867435906e-06, "loss": 0.6726, "step": 35360 }, { "epoch": 9.593165174938974, "grad_norm": 17.130354836190882, "learning_rate": 3.4628408614488974e-06, "loss": 0.6388, "step": 35370 }, { "epoch": 9.595877407106048, "grad_norm": 21.96980239597461, "learning_rate": 3.459838148523598e-06, "loss": 0.6753, "step": 35380 }, { "epoch": 9.598589639273122, "grad_norm": 19.661929559884292, "learning_rate": 3.4568360491639008e-06, "loss": 0.5516, "step": 35390 }, { "epoch": 9.601301871440196, "grad_norm": 18.6468146688527, "learning_rate": 3.4538345645657794e-06, "loss": 0.6595, "step": 35400 }, { "epoch": 9.60401410360727, "grad_norm": 17.551765613409703, "learning_rate": 3.4508336959249557e-06, "loss": 0.6747, "step": 35410 }, { "epoch": 9.606726335774342, "grad_norm": 18.109682439151882, "learning_rate": 3.4478334444369056e-06, "loss": 0.6404, "step": 35420 }, { "epoch": 9.609438567941416, "grad_norm": 20.34155623796744, "learning_rate": 3.4448338112968617e-06, "loss": 0.7176, "step": 35430 }, { "epoch": 9.61215080010849, "grad_norm": 12.09195742246439, "learning_rate": 3.441834797699811e-06, "loss": 0.551, "step": 35440 }, { "epoch": 9.614863032275563, "grad_norm": 18.605529148040223, "learning_rate": 3.438836404840494e-06, "loss": 0.6203, "step": 35450 }, { "epoch": 9.617575264442637, "grad_norm": 17.277647459549556, "learning_rate": 3.435838633913402e-06, "loss": 0.5909, "step": 35460 }, { "epoch": 9.62028749660971, "grad_norm": 22.65741293882537, "learning_rate": 3.4328414861127756e-06, "loss": 0.6221, "step": 35470 }, { "epoch": 9.622999728776783, "grad_norm": 17.452930730805505, "learning_rate": 3.4298449626326146e-06, "loss": 0.691, "step": 35480 }, { "epoch": 9.625711960943857, "grad_norm": 18.89177617828101, "learning_rate": 3.426849064666666e-06, "loss": 0.6433, "step": 35490 }, { "epoch": 9.62842419311093, "grad_norm": 24.90424232713079, "learning_rate": 3.4238537934084244e-06, "loss": 0.7274, "step": 35500 }, { "epoch": 9.631136425278005, "grad_norm": 23.436950282081682, "learning_rate": 3.420859150051139e-06, "loss": 0.7428, "step": 35510 }, { "epoch": 9.633848657445077, "grad_norm": 18.616055895719917, "learning_rate": 3.417865135787811e-06, "loss": 0.6715, "step": 35520 }, { "epoch": 9.63656088961215, "grad_norm": 21.763092773022876, "learning_rate": 3.4148717518111863e-06, "loss": 0.6747, "step": 35530 }, { "epoch": 9.639273121779224, "grad_norm": 17.94843656801715, "learning_rate": 3.41187899931376e-06, "loss": 0.6339, "step": 35540 }, { "epoch": 9.641985353946298, "grad_norm": 20.027865951046053, "learning_rate": 3.4088868794877762e-06, "loss": 0.7263, "step": 35550 }, { "epoch": 9.644697586113372, "grad_norm": 17.72686186384433, "learning_rate": 3.4058953935252316e-06, "loss": 0.7039, "step": 35560 }, { "epoch": 9.647409818280444, "grad_norm": 21.046927911721927, "learning_rate": 3.402904542617862e-06, "loss": 0.6662, "step": 35570 }, { "epoch": 9.650122050447518, "grad_norm": 20.282650209657785, "learning_rate": 3.3999143279571585e-06, "loss": 0.6752, "step": 35580 }, { "epoch": 9.652834282614592, "grad_norm": 18.786984720196195, "learning_rate": 3.3969247507343507e-06, "loss": 0.7438, "step": 35590 }, { "epoch": 9.655546514781665, "grad_norm": 21.893429619150524, "learning_rate": 3.393935812140423e-06, "loss": 0.676, "step": 35600 }, { "epoch": 9.65825874694874, "grad_norm": 20.021438782287213, "learning_rate": 3.3909475133661e-06, "loss": 0.7361, "step": 35610 }, { "epoch": 9.660970979115813, "grad_norm": 19.692744188428996, "learning_rate": 3.3879598556018484e-06, "loss": 0.639, "step": 35620 }, { "epoch": 9.663683211282885, "grad_norm": 17.348842489987057, "learning_rate": 3.384972840037887e-06, "loss": 0.6261, "step": 35630 }, { "epoch": 9.666395443449959, "grad_norm": 18.10266174479516, "learning_rate": 3.381986467864173e-06, "loss": 0.6956, "step": 35640 }, { "epoch": 9.669107675617033, "grad_norm": 19.060684821390126, "learning_rate": 3.3790007402704126e-06, "loss": 0.7127, "step": 35650 }, { "epoch": 9.671819907784107, "grad_norm": 17.746720193614156, "learning_rate": 3.3760156584460503e-06, "loss": 0.6421, "step": 35660 }, { "epoch": 9.67453213995118, "grad_norm": 20.870014437318705, "learning_rate": 3.373031223580273e-06, "loss": 0.6486, "step": 35670 }, { "epoch": 9.677244372118253, "grad_norm": 18.341082850867167, "learning_rate": 3.370047436862015e-06, "loss": 0.5981, "step": 35680 }, { "epoch": 9.679956604285326, "grad_norm": 20.204784005221487, "learning_rate": 3.367064299479948e-06, "loss": 0.6923, "step": 35690 }, { "epoch": 9.6826688364524, "grad_norm": 18.353017901929142, "learning_rate": 3.3640818126224873e-06, "loss": 0.6469, "step": 35700 }, { "epoch": 9.685381068619474, "grad_norm": 20.223181296332992, "learning_rate": 3.361099977477785e-06, "loss": 0.6464, "step": 35710 }, { "epoch": 9.688093300786548, "grad_norm": 16.149346908230164, "learning_rate": 3.3581187952337423e-06, "loss": 0.668, "step": 35720 }, { "epoch": 9.69080553295362, "grad_norm": 18.870389695025704, "learning_rate": 3.355138267077991e-06, "loss": 0.6532, "step": 35730 }, { "epoch": 9.693517765120694, "grad_norm": 20.358806390157138, "learning_rate": 3.3521583941979053e-06, "loss": 0.6914, "step": 35740 }, { "epoch": 9.696229997287768, "grad_norm": 17.87191219834831, "learning_rate": 3.3491791777806004e-06, "loss": 0.6617, "step": 35750 }, { "epoch": 9.698942229454842, "grad_norm": 18.99619745383539, "learning_rate": 3.3462006190129283e-06, "loss": 0.6267, "step": 35760 }, { "epoch": 9.701654461621915, "grad_norm": 19.176945624638762, "learning_rate": 3.3432227190814816e-06, "loss": 0.7096, "step": 35770 }, { "epoch": 9.704366693788987, "grad_norm": 18.544511093159812, "learning_rate": 3.3402454791725856e-06, "loss": 0.6447, "step": 35780 }, { "epoch": 9.707078925956061, "grad_norm": 18.056301123844758, "learning_rate": 3.3372689004723035e-06, "loss": 0.6437, "step": 35790 }, { "epoch": 9.709791158123135, "grad_norm": 18.228820015731028, "learning_rate": 3.334292984166442e-06, "loss": 0.6644, "step": 35800 }, { "epoch": 9.712503390290209, "grad_norm": 21.572936459106423, "learning_rate": 3.331317731440535e-06, "loss": 0.7148, "step": 35810 }, { "epoch": 9.715215622457283, "grad_norm": 18.550099805491154, "learning_rate": 3.3283431434798576e-06, "loss": 0.62, "step": 35820 }, { "epoch": 9.717927854624357, "grad_norm": 17.4403791131388, "learning_rate": 3.3253692214694155e-06, "loss": 0.6956, "step": 35830 }, { "epoch": 9.720640086791429, "grad_norm": 18.98445781747837, "learning_rate": 3.3223959665939566e-06, "loss": 0.6899, "step": 35840 }, { "epoch": 9.723352318958502, "grad_norm": 21.938473791688892, "learning_rate": 3.3194233800379547e-06, "loss": 0.6569, "step": 35850 }, { "epoch": 9.726064551125576, "grad_norm": 20.452648820778627, "learning_rate": 3.3164514629856226e-06, "loss": 0.6832, "step": 35860 }, { "epoch": 9.72877678329265, "grad_norm": 20.454257450973124, "learning_rate": 3.3134802166209013e-06, "loss": 0.6425, "step": 35870 }, { "epoch": 9.731489015459724, "grad_norm": 21.494936651337042, "learning_rate": 3.310509642127472e-06, "loss": 0.6096, "step": 35880 }, { "epoch": 9.734201247626796, "grad_norm": 18.846458051291062, "learning_rate": 3.3075397406887443e-06, "loss": 0.7359, "step": 35890 }, { "epoch": 9.73691347979387, "grad_norm": 19.927792079953882, "learning_rate": 3.3045705134878594e-06, "loss": 0.6458, "step": 35900 }, { "epoch": 9.739625711960944, "grad_norm": 19.611028280628666, "learning_rate": 3.301601961707686e-06, "loss": 0.6697, "step": 35910 }, { "epoch": 9.742337944128018, "grad_norm": 25.39528721150291, "learning_rate": 3.298634086530834e-06, "loss": 0.6376, "step": 35920 }, { "epoch": 9.745050176295091, "grad_norm": 22.0466405656409, "learning_rate": 3.2956668891396348e-06, "loss": 0.6325, "step": 35930 }, { "epoch": 9.747762408462165, "grad_norm": 14.742606137397624, "learning_rate": 3.292700370716151e-06, "loss": 0.7031, "step": 35940 }, { "epoch": 9.750474640629237, "grad_norm": 20.91462551665028, "learning_rate": 3.2897345324421774e-06, "loss": 0.7004, "step": 35950 }, { "epoch": 9.753186872796311, "grad_norm": 19.30815273649762, "learning_rate": 3.2867693754992404e-06, "loss": 0.6062, "step": 35960 }, { "epoch": 9.755899104963385, "grad_norm": 18.286278650318497, "learning_rate": 3.2838049010685873e-06, "loss": 0.6306, "step": 35970 }, { "epoch": 9.758611337130459, "grad_norm": 17.33109276104339, "learning_rate": 3.2808411103311995e-06, "loss": 0.6535, "step": 35980 }, { "epoch": 9.761323569297533, "grad_norm": 19.15196231268548, "learning_rate": 3.2778780044677805e-06, "loss": 0.6694, "step": 35990 }, { "epoch": 9.764035801464605, "grad_norm": 18.61662021841624, "learning_rate": 3.2749155846587705e-06, "loss": 0.6411, "step": 36000 }, { "epoch": 9.766748033631679, "grad_norm": 23.104449088130846, "learning_rate": 3.271953852084326e-06, "loss": 0.6952, "step": 36010 }, { "epoch": 9.769460265798752, "grad_norm": 18.42428042142012, "learning_rate": 3.268992807924336e-06, "loss": 0.6771, "step": 36020 }, { "epoch": 9.772172497965826, "grad_norm": 21.610524638452638, "learning_rate": 3.266032453358412e-06, "loss": 0.6967, "step": 36030 }, { "epoch": 9.7748847301329, "grad_norm": 22.110981145876895, "learning_rate": 3.2630727895658947e-06, "loss": 0.6809, "step": 36040 }, { "epoch": 9.777596962299972, "grad_norm": 17.3379913347575, "learning_rate": 3.2601138177258462e-06, "loss": 0.7113, "step": 36050 }, { "epoch": 9.780309194467046, "grad_norm": 20.689863409831354, "learning_rate": 3.2571555390170517e-06, "loss": 0.6664, "step": 36060 }, { "epoch": 9.78302142663412, "grad_norm": 24.638091371208244, "learning_rate": 3.2541979546180246e-06, "loss": 0.6557, "step": 36070 }, { "epoch": 9.785733658801194, "grad_norm": 23.908273525635067, "learning_rate": 3.2512410657069986e-06, "loss": 0.6921, "step": 36080 }, { "epoch": 9.788445890968267, "grad_norm": 18.392785436297082, "learning_rate": 3.248284873461933e-06, "loss": 0.6962, "step": 36090 }, { "epoch": 9.79115812313534, "grad_norm": 18.530572860610498, "learning_rate": 3.2453293790605066e-06, "loss": 0.7425, "step": 36100 }, { "epoch": 9.793870355302413, "grad_norm": 21.566437731408897, "learning_rate": 3.2423745836801203e-06, "loss": 0.6681, "step": 36110 }, { "epoch": 9.796582587469487, "grad_norm": 18.826379465488436, "learning_rate": 3.2394204884978996e-06, "loss": 0.7029, "step": 36120 }, { "epoch": 9.799294819636561, "grad_norm": 17.060984670266023, "learning_rate": 3.2364670946906886e-06, "loss": 0.6417, "step": 36130 }, { "epoch": 9.802007051803635, "grad_norm": 20.64351705869162, "learning_rate": 3.2335144034350517e-06, "loss": 0.6725, "step": 36140 }, { "epoch": 9.804719283970709, "grad_norm": 21.453446269467435, "learning_rate": 3.2305624159072737e-06, "loss": 0.7009, "step": 36150 }, { "epoch": 9.80743151613778, "grad_norm": 17.11267023165223, "learning_rate": 3.2276111332833624e-06, "loss": 0.5895, "step": 36160 }, { "epoch": 9.810143748304855, "grad_norm": 19.759389792387186, "learning_rate": 3.224660556739041e-06, "loss": 0.6484, "step": 36170 }, { "epoch": 9.812855980471928, "grad_norm": 20.17349662249945, "learning_rate": 3.2217106874497494e-06, "loss": 0.6817, "step": 36180 }, { "epoch": 9.815568212639002, "grad_norm": 18.70237447347419, "learning_rate": 3.2187615265906517e-06, "loss": 0.6787, "step": 36190 }, { "epoch": 9.818280444806076, "grad_norm": 20.517725848630317, "learning_rate": 3.2158130753366256e-06, "loss": 0.6751, "step": 36200 }, { "epoch": 9.820992676973148, "grad_norm": 22.241261359006607, "learning_rate": 3.2128653348622694e-06, "loss": 0.652, "step": 36210 }, { "epoch": 9.823704909140222, "grad_norm": 17.710544371127774, "learning_rate": 3.2099183063418948e-06, "loss": 0.6622, "step": 36220 }, { "epoch": 9.826417141307296, "grad_norm": 19.671318899591387, "learning_rate": 3.206971990949528e-06, "loss": 0.6361, "step": 36230 }, { "epoch": 9.82912937347437, "grad_norm": 21.827321552372357, "learning_rate": 3.20402638985892e-06, "loss": 0.7553, "step": 36240 }, { "epoch": 9.831841605641443, "grad_norm": 18.80041841971499, "learning_rate": 3.2010815042435274e-06, "loss": 0.6232, "step": 36250 }, { "epoch": 9.834553837808517, "grad_norm": 20.118449207190626, "learning_rate": 3.1981373352765282e-06, "loss": 0.6677, "step": 36260 }, { "epoch": 9.83726606997559, "grad_norm": 15.989346638451307, "learning_rate": 3.19519388413081e-06, "loss": 0.6138, "step": 36270 }, { "epoch": 9.839978302142663, "grad_norm": 16.982570991165225, "learning_rate": 3.192251151978981e-06, "loss": 0.6064, "step": 36280 }, { "epoch": 9.842690534309737, "grad_norm": 18.226226624742758, "learning_rate": 3.1893091399933574e-06, "loss": 0.7066, "step": 36290 }, { "epoch": 9.845402766476811, "grad_norm": 17.23913970144218, "learning_rate": 3.186367849345969e-06, "loss": 0.6532, "step": 36300 }, { "epoch": 9.848114998643885, "grad_norm": 16.136980794808636, "learning_rate": 3.1834272812085586e-06, "loss": 0.6872, "step": 36310 }, { "epoch": 9.850827230810957, "grad_norm": 17.731377316435125, "learning_rate": 3.1804874367525846e-06, "loss": 0.7457, "step": 36320 }, { "epoch": 9.85353946297803, "grad_norm": 15.534457225310696, "learning_rate": 3.1775483171492136e-06, "loss": 0.6542, "step": 36330 }, { "epoch": 9.856251695145104, "grad_norm": 20.029654271878833, "learning_rate": 3.174609923569325e-06, "loss": 0.7224, "step": 36340 }, { "epoch": 9.858963927312178, "grad_norm": 21.105863173389643, "learning_rate": 3.1716722571835047e-06, "loss": 0.7298, "step": 36350 }, { "epoch": 9.861676159479252, "grad_norm": 19.597185482197855, "learning_rate": 3.1687353191620575e-06, "loss": 0.6366, "step": 36360 }, { "epoch": 9.864388391646324, "grad_norm": 24.346049765634003, "learning_rate": 3.1657991106749885e-06, "loss": 0.6817, "step": 36370 }, { "epoch": 9.867100623813398, "grad_norm": 18.585374950120883, "learning_rate": 3.162863632892021e-06, "loss": 0.6526, "step": 36380 }, { "epoch": 9.869812855980472, "grad_norm": 18.73320433751125, "learning_rate": 3.1599288869825777e-06, "loss": 0.6938, "step": 36390 }, { "epoch": 9.872525088147546, "grad_norm": 19.068810311437442, "learning_rate": 3.1569948741158007e-06, "loss": 0.6793, "step": 36400 }, { "epoch": 9.87523732031462, "grad_norm": 21.951161361850644, "learning_rate": 3.1540615954605303e-06, "loss": 0.6608, "step": 36410 }, { "epoch": 9.877949552481692, "grad_norm": 20.40469500194577, "learning_rate": 3.1511290521853204e-06, "loss": 0.7603, "step": 36420 }, { "epoch": 9.880661784648765, "grad_norm": 17.985968399068575, "learning_rate": 3.1481972454584264e-06, "loss": 0.745, "step": 36430 }, { "epoch": 9.88337401681584, "grad_norm": 18.03405910506731, "learning_rate": 3.145266176447817e-06, "loss": 0.6682, "step": 36440 }, { "epoch": 9.886086248982913, "grad_norm": 20.089475002270483, "learning_rate": 3.1423358463211637e-06, "loss": 0.7027, "step": 36450 }, { "epoch": 9.888798481149987, "grad_norm": 20.992069844834244, "learning_rate": 3.139406256245844e-06, "loss": 0.7075, "step": 36460 }, { "epoch": 9.89151071331706, "grad_norm": 23.505638212829826, "learning_rate": 3.1364774073889358e-06, "loss": 0.6919, "step": 36470 }, { "epoch": 9.894222945484133, "grad_norm": 17.11743887993556, "learning_rate": 3.1335493009172323e-06, "loss": 0.667, "step": 36480 }, { "epoch": 9.896935177651207, "grad_norm": 19.589795732299898, "learning_rate": 3.130621937997221e-06, "loss": 0.6874, "step": 36490 }, { "epoch": 9.89964740981828, "grad_norm": 22.48769616893248, "learning_rate": 3.1276953197950978e-06, "loss": 0.7104, "step": 36500 }, { "epoch": 9.902359641985354, "grad_norm": 18.32812083770545, "learning_rate": 3.1247694474767596e-06, "loss": 0.731, "step": 36510 }, { "epoch": 9.905071874152428, "grad_norm": 20.051663902881707, "learning_rate": 3.121844322207812e-06, "loss": 0.7683, "step": 36520 }, { "epoch": 9.9077841063195, "grad_norm": 22.478097278746535, "learning_rate": 3.118919945153556e-06, "loss": 0.7196, "step": 36530 }, { "epoch": 9.910496338486574, "grad_norm": 20.658348952961664, "learning_rate": 3.1159963174789963e-06, "loss": 0.7098, "step": 36540 }, { "epoch": 9.913208570653648, "grad_norm": 18.90092141314784, "learning_rate": 3.1130734403488396e-06, "loss": 0.6048, "step": 36550 }, { "epoch": 9.915920802820722, "grad_norm": 17.197249178849326, "learning_rate": 3.1101513149274974e-06, "loss": 0.7318, "step": 36560 }, { "epoch": 9.918633034987796, "grad_norm": 22.63098288857769, "learning_rate": 3.107229942379075e-06, "loss": 0.6641, "step": 36570 }, { "epoch": 9.921345267154868, "grad_norm": 19.13415895391823, "learning_rate": 3.1043093238673826e-06, "loss": 0.6144, "step": 36580 }, { "epoch": 9.924057499321941, "grad_norm": 20.261445370165315, "learning_rate": 3.1013894605559267e-06, "loss": 0.6752, "step": 36590 }, { "epoch": 9.926769731489015, "grad_norm": 20.894969771351327, "learning_rate": 3.0984703536079177e-06, "loss": 0.6613, "step": 36600 }, { "epoch": 9.92948196365609, "grad_norm": 15.649819083960885, "learning_rate": 3.095552004186261e-06, "loss": 0.6664, "step": 36610 }, { "epoch": 9.932194195823163, "grad_norm": 20.890505240785476, "learning_rate": 3.092634413453558e-06, "loss": 0.6058, "step": 36620 }, { "epoch": 9.934906427990235, "grad_norm": 16.29579203729413, "learning_rate": 3.0897175825721126e-06, "loss": 0.6216, "step": 36630 }, { "epoch": 9.937618660157309, "grad_norm": 18.491483674579836, "learning_rate": 3.0868015127039253e-06, "loss": 0.5903, "step": 36640 }, { "epoch": 9.940330892324383, "grad_norm": 20.335374311364035, "learning_rate": 3.083886205010692e-06, "loss": 0.6652, "step": 36650 }, { "epoch": 9.943043124491457, "grad_norm": 16.315336617203762, "learning_rate": 3.0809716606538043e-06, "loss": 0.677, "step": 36660 }, { "epoch": 9.94575535665853, "grad_norm": 15.754519364274445, "learning_rate": 3.0780578807943495e-06, "loss": 0.6531, "step": 36670 }, { "epoch": 9.948467588825604, "grad_norm": 19.804183237262844, "learning_rate": 3.075144866593115e-06, "loss": 0.6397, "step": 36680 }, { "epoch": 9.951179820992676, "grad_norm": 20.57597248368398, "learning_rate": 3.072232619210577e-06, "loss": 0.6358, "step": 36690 }, { "epoch": 9.95389205315975, "grad_norm": 13.82984750887314, "learning_rate": 3.06932113980691e-06, "loss": 0.6709, "step": 36700 }, { "epoch": 9.956604285326824, "grad_norm": 21.5477706155926, "learning_rate": 3.0664104295419796e-06, "loss": 0.6954, "step": 36710 }, { "epoch": 9.959316517493898, "grad_norm": 19.49029256783906, "learning_rate": 3.0635004895753496e-06, "loss": 0.5874, "step": 36720 }, { "epoch": 9.962028749660972, "grad_norm": 20.15046972858042, "learning_rate": 3.0605913210662737e-06, "loss": 0.6843, "step": 36730 }, { "epoch": 9.964740981828044, "grad_norm": 17.567031733419228, "learning_rate": 3.0576829251736966e-06, "loss": 0.6388, "step": 36740 }, { "epoch": 9.967453213995118, "grad_norm": 20.418616164354646, "learning_rate": 3.054775303056259e-06, "loss": 0.7625, "step": 36750 }, { "epoch": 9.970165446162191, "grad_norm": 21.609689065679444, "learning_rate": 3.051868455872291e-06, "loss": 0.757, "step": 36760 }, { "epoch": 9.972877678329265, "grad_norm": 21.17072095121045, "learning_rate": 3.0489623847798165e-06, "loss": 0.6517, "step": 36770 }, { "epoch": 9.975589910496339, "grad_norm": 20.62550809465741, "learning_rate": 3.0460570909365473e-06, "loss": 0.6223, "step": 36780 }, { "epoch": 9.978302142663413, "grad_norm": 19.25314698368681, "learning_rate": 3.043152575499885e-06, "loss": 0.6778, "step": 36790 }, { "epoch": 9.981014374830485, "grad_norm": 19.637455195804844, "learning_rate": 3.0402488396269255e-06, "loss": 0.681, "step": 36800 }, { "epoch": 9.983726606997559, "grad_norm": 19.46068762773329, "learning_rate": 3.0373458844744487e-06, "loss": 0.6053, "step": 36810 }, { "epoch": 9.986438839164633, "grad_norm": 19.148817062665866, "learning_rate": 3.0344437111989295e-06, "loss": 0.6623, "step": 36820 }, { "epoch": 9.989151071331706, "grad_norm": 19.277865748905974, "learning_rate": 3.0315423209565237e-06, "loss": 0.6969, "step": 36830 }, { "epoch": 9.99186330349878, "grad_norm": 22.122873226108904, "learning_rate": 3.0286417149030845e-06, "loss": 0.6848, "step": 36840 }, { "epoch": 9.994575535665852, "grad_norm": 22.51519264386723, "learning_rate": 3.025741894194144e-06, "loss": 0.7616, "step": 36850 }, { "epoch": 9.997287767832926, "grad_norm": 19.71081172584408, "learning_rate": 3.0228428599849268e-06, "loss": 0.6892, "step": 36860 }, { "epoch": 10.0, "grad_norm": 18.66931959650605, "learning_rate": 3.01994461343034e-06, "loss": 0.6452, "step": 36870 }, { "epoch": 10.002712232167074, "grad_norm": 14.004423001153569, "learning_rate": 3.017047155684983e-06, "loss": 0.3847, "step": 36880 }, { "epoch": 10.005424464334148, "grad_norm": 16.1095706408496, "learning_rate": 3.0141504879031357e-06, "loss": 0.4017, "step": 36890 }, { "epoch": 10.00813669650122, "grad_norm": 15.444090943798198, "learning_rate": 3.011254611238765e-06, "loss": 0.4225, "step": 36900 }, { "epoch": 10.010848928668294, "grad_norm": 16.374562070995438, "learning_rate": 3.0083595268455208e-06, "loss": 0.4399, "step": 36910 }, { "epoch": 10.013561160835367, "grad_norm": 15.374991286230868, "learning_rate": 3.0054652358767427e-06, "loss": 0.4145, "step": 36920 }, { "epoch": 10.016273393002441, "grad_norm": 15.897099297600663, "learning_rate": 3.0025717394854485e-06, "loss": 0.4207, "step": 36930 }, { "epoch": 10.018985625169515, "grad_norm": 17.55137077665052, "learning_rate": 2.999679038824341e-06, "loss": 0.3814, "step": 36940 }, { "epoch": 10.021697857336587, "grad_norm": 16.842222890989593, "learning_rate": 2.9967871350458074e-06, "loss": 0.4891, "step": 36950 }, { "epoch": 10.024410089503661, "grad_norm": 18.586149463586104, "learning_rate": 2.9938960293019183e-06, "loss": 0.3801, "step": 36960 }, { "epoch": 10.027122321670735, "grad_norm": 17.978574846110366, "learning_rate": 2.9910057227444234e-06, "loss": 0.4404, "step": 36970 }, { "epoch": 10.029834553837809, "grad_norm": 18.12824817979966, "learning_rate": 2.9881162165247557e-06, "loss": 0.4296, "step": 36980 }, { "epoch": 10.032546786004882, "grad_norm": 15.972235062534171, "learning_rate": 2.9852275117940254e-06, "loss": 0.4219, "step": 36990 }, { "epoch": 10.035259018171956, "grad_norm": 14.93756307351526, "learning_rate": 2.9823396097030333e-06, "loss": 0.4472, "step": 37000 }, { "epoch": 10.037971250339028, "grad_norm": 12.21629256800827, "learning_rate": 2.9794525114022502e-06, "loss": 0.3543, "step": 37010 }, { "epoch": 10.040683482506102, "grad_norm": 16.357246469518003, "learning_rate": 2.976566218041833e-06, "loss": 0.4343, "step": 37020 }, { "epoch": 10.043395714673176, "grad_norm": 13.530954784505713, "learning_rate": 2.973680730771612e-06, "loss": 0.4142, "step": 37030 }, { "epoch": 10.04610794684025, "grad_norm": 14.49460511160404, "learning_rate": 2.9707960507411037e-06, "loss": 0.3767, "step": 37040 }, { "epoch": 10.048820179007324, "grad_norm": 18.33129608545407, "learning_rate": 2.967912179099499e-06, "loss": 0.4051, "step": 37050 }, { "epoch": 10.051532411174396, "grad_norm": 14.268728393933142, "learning_rate": 2.9650291169956645e-06, "loss": 0.3613, "step": 37060 }, { "epoch": 10.05424464334147, "grad_norm": 15.421784346335624, "learning_rate": 2.9621468655781483e-06, "loss": 0.4149, "step": 37070 }, { "epoch": 10.056956875508543, "grad_norm": 13.460558270633861, "learning_rate": 2.9592654259951735e-06, "loss": 0.3999, "step": 37080 }, { "epoch": 10.059669107675617, "grad_norm": 16.875948275464932, "learning_rate": 2.9563847993946426e-06, "loss": 0.4006, "step": 37090 }, { "epoch": 10.062381339842691, "grad_norm": 15.7057971031292, "learning_rate": 2.9535049869241306e-06, "loss": 0.4042, "step": 37100 }, { "epoch": 10.065093572009763, "grad_norm": 17.538002746459735, "learning_rate": 2.9506259897308877e-06, "loss": 0.42, "step": 37110 }, { "epoch": 10.067805804176837, "grad_norm": 12.907397636367543, "learning_rate": 2.947747808961844e-06, "loss": 0.3888, "step": 37120 }, { "epoch": 10.07051803634391, "grad_norm": 20.060122486139733, "learning_rate": 2.9448704457635996e-06, "loss": 0.4208, "step": 37130 }, { "epoch": 10.073230268510985, "grad_norm": 16.87425527868761, "learning_rate": 2.9419939012824327e-06, "loss": 0.3926, "step": 37140 }, { "epoch": 10.075942500678059, "grad_norm": 16.526602593408814, "learning_rate": 2.9391181766642905e-06, "loss": 0.493, "step": 37150 }, { "epoch": 10.078654732845132, "grad_norm": 16.53244802599998, "learning_rate": 2.9362432730548006e-06, "loss": 0.442, "step": 37160 }, { "epoch": 10.081366965012204, "grad_norm": 15.939896531753595, "learning_rate": 2.9333691915992566e-06, "loss": 0.484, "step": 37170 }, { "epoch": 10.084079197179278, "grad_norm": 14.51679603600396, "learning_rate": 2.9304959334426286e-06, "loss": 0.5003, "step": 37180 }, { "epoch": 10.086791429346352, "grad_norm": 16.714481111511304, "learning_rate": 2.9276234997295566e-06, "loss": 0.4188, "step": 37190 }, { "epoch": 10.089503661513426, "grad_norm": 17.32316770498476, "learning_rate": 2.9247518916043538e-06, "loss": 0.4092, "step": 37200 }, { "epoch": 10.0922158936805, "grad_norm": 16.74425858135792, "learning_rate": 2.9218811102110046e-06, "loss": 0.4517, "step": 37210 }, { "epoch": 10.094928125847572, "grad_norm": 18.947848609034708, "learning_rate": 2.9190111566931624e-06, "loss": 0.475, "step": 37220 }, { "epoch": 10.097640358014646, "grad_norm": 18.691196977085948, "learning_rate": 2.916142032194151e-06, "loss": 0.4724, "step": 37230 }, { "epoch": 10.10035259018172, "grad_norm": 20.914029168161377, "learning_rate": 2.9132737378569653e-06, "loss": 0.4182, "step": 37240 }, { "epoch": 10.103064822348793, "grad_norm": 15.561837278879704, "learning_rate": 2.9104062748242717e-06, "loss": 0.4124, "step": 37250 }, { "epoch": 10.105777054515867, "grad_norm": 20.136936160600413, "learning_rate": 2.9075396442383974e-06, "loss": 0.4476, "step": 37260 }, { "epoch": 10.10848928668294, "grad_norm": 19.290942724330815, "learning_rate": 2.904673847241346e-06, "loss": 0.4329, "step": 37270 }, { "epoch": 10.111201518850013, "grad_norm": 17.606543539740144, "learning_rate": 2.901808884974785e-06, "loss": 0.4225, "step": 37280 }, { "epoch": 10.113913751017087, "grad_norm": 15.778129612838093, "learning_rate": 2.898944758580052e-06, "loss": 0.4052, "step": 37290 }, { "epoch": 10.11662598318416, "grad_norm": 17.209223396806607, "learning_rate": 2.8960814691981503e-06, "loss": 0.439, "step": 37300 }, { "epoch": 10.119338215351235, "grad_norm": 17.370203445119383, "learning_rate": 2.893219017969745e-06, "loss": 0.413, "step": 37310 }, { "epoch": 10.122050447518308, "grad_norm": 13.307574594693728, "learning_rate": 2.890357406035179e-06, "loss": 0.4318, "step": 37320 }, { "epoch": 10.12476267968538, "grad_norm": 20.975349520037344, "learning_rate": 2.8874966345344483e-06, "loss": 0.3926, "step": 37330 }, { "epoch": 10.127474911852454, "grad_norm": 19.30131794902573, "learning_rate": 2.8846367046072214e-06, "loss": 0.4498, "step": 37340 }, { "epoch": 10.130187144019528, "grad_norm": 16.962808160059602, "learning_rate": 2.881777617392829e-06, "loss": 0.4209, "step": 37350 }, { "epoch": 10.132899376186602, "grad_norm": 20.56077096142206, "learning_rate": 2.8789193740302677e-06, "loss": 0.4759, "step": 37360 }, { "epoch": 10.135611608353676, "grad_norm": 23.755854982052313, "learning_rate": 2.876061975658198e-06, "loss": 0.4364, "step": 37370 }, { "epoch": 10.138323840520748, "grad_norm": 15.069272435847516, "learning_rate": 2.873205423414939e-06, "loss": 0.4323, "step": 37380 }, { "epoch": 10.141036072687822, "grad_norm": 18.27587487353109, "learning_rate": 2.870349718438479e-06, "loss": 0.3814, "step": 37390 }, { "epoch": 10.143748304854896, "grad_norm": 20.848067867279685, "learning_rate": 2.8674948618664655e-06, "loss": 0.43, "step": 37400 }, { "epoch": 10.14646053702197, "grad_norm": 17.589033031914493, "learning_rate": 2.8646408548362094e-06, "loss": 0.4432, "step": 37410 }, { "epoch": 10.149172769189043, "grad_norm": 18.592488737287894, "learning_rate": 2.8617876984846828e-06, "loss": 0.4271, "step": 37420 }, { "epoch": 10.151885001356115, "grad_norm": 17.874135128204973, "learning_rate": 2.858935393948514e-06, "loss": 0.4211, "step": 37430 }, { "epoch": 10.154597233523189, "grad_norm": 16.37845464392175, "learning_rate": 2.8560839423640053e-06, "loss": 0.4724, "step": 37440 }, { "epoch": 10.157309465690263, "grad_norm": 15.194346334149564, "learning_rate": 2.853233344867103e-06, "loss": 0.4342, "step": 37450 }, { "epoch": 10.160021697857337, "grad_norm": 16.009696806504046, "learning_rate": 2.8503836025934227e-06, "loss": 0.4793, "step": 37460 }, { "epoch": 10.16273393002441, "grad_norm": 13.296979648188579, "learning_rate": 2.847534716678238e-06, "loss": 0.3826, "step": 37470 }, { "epoch": 10.165446162191484, "grad_norm": 16.104498775304297, "learning_rate": 2.8446866882564795e-06, "loss": 0.4079, "step": 37480 }, { "epoch": 10.168158394358557, "grad_norm": 17.55615882516133, "learning_rate": 2.84183951846274e-06, "loss": 0.3937, "step": 37490 }, { "epoch": 10.17087062652563, "grad_norm": 15.703089875616895, "learning_rate": 2.8389932084312636e-06, "loss": 0.3994, "step": 37500 }, { "epoch": 10.173582858692704, "grad_norm": 15.749832404896033, "learning_rate": 2.8361477592959575e-06, "loss": 0.459, "step": 37510 }, { "epoch": 10.176295090859778, "grad_norm": 15.736233486167773, "learning_rate": 2.8333031721903835e-06, "loss": 0.3918, "step": 37520 }, { "epoch": 10.179007323026852, "grad_norm": 15.832294556211947, "learning_rate": 2.830459448247762e-06, "loss": 0.4202, "step": 37530 }, { "epoch": 10.181719555193924, "grad_norm": 16.63818786323008, "learning_rate": 2.8276165886009693e-06, "loss": 0.4485, "step": 37540 }, { "epoch": 10.184431787360998, "grad_norm": 12.37866391984517, "learning_rate": 2.8247745943825307e-06, "loss": 0.3905, "step": 37550 }, { "epoch": 10.187144019528072, "grad_norm": 16.82840371578249, "learning_rate": 2.82193346672464e-06, "loss": 0.4444, "step": 37560 }, { "epoch": 10.189856251695145, "grad_norm": 16.750462507767693, "learning_rate": 2.8190932067591325e-06, "loss": 0.3952, "step": 37570 }, { "epoch": 10.19256848386222, "grad_norm": 18.7530808542219, "learning_rate": 2.816253815617507e-06, "loss": 0.4377, "step": 37580 }, { "epoch": 10.195280716029291, "grad_norm": 14.99897027337254, "learning_rate": 2.8134152944309067e-06, "loss": 0.4383, "step": 37590 }, { "epoch": 10.197992948196365, "grad_norm": 15.867951030965475, "learning_rate": 2.810577644330141e-06, "loss": 0.3928, "step": 37600 }, { "epoch": 10.200705180363439, "grad_norm": 17.72190368384724, "learning_rate": 2.807740866445664e-06, "loss": 0.4832, "step": 37610 }, { "epoch": 10.203417412530513, "grad_norm": 15.65104990307385, "learning_rate": 2.804904961907582e-06, "loss": 0.4448, "step": 37620 }, { "epoch": 10.206129644697587, "grad_norm": 16.085899636274174, "learning_rate": 2.8020699318456545e-06, "loss": 0.4424, "step": 37630 }, { "epoch": 10.20884187686466, "grad_norm": 21.6155133625052, "learning_rate": 2.7992357773892945e-06, "loss": 0.485, "step": 37640 }, { "epoch": 10.211554109031733, "grad_norm": 18.533863473905985, "learning_rate": 2.7964024996675677e-06, "loss": 0.4065, "step": 37650 }, { "epoch": 10.214266341198806, "grad_norm": 18.098752154584016, "learning_rate": 2.7935700998091823e-06, "loss": 0.4456, "step": 37660 }, { "epoch": 10.21697857336588, "grad_norm": 17.773364319002813, "learning_rate": 2.7907385789425035e-06, "loss": 0.4622, "step": 37670 }, { "epoch": 10.219690805532954, "grad_norm": 21.254629336873748, "learning_rate": 2.7879079381955505e-06, "loss": 0.4227, "step": 37680 }, { "epoch": 10.222403037700028, "grad_norm": 19.13535375045833, "learning_rate": 2.7850781786959804e-06, "loss": 0.4561, "step": 37690 }, { "epoch": 10.2251152698671, "grad_norm": 15.579769682396048, "learning_rate": 2.7822493015711092e-06, "loss": 0.4225, "step": 37700 }, { "epoch": 10.227827502034174, "grad_norm": 16.918951265270607, "learning_rate": 2.7794213079478925e-06, "loss": 0.4531, "step": 37710 }, { "epoch": 10.230539734201248, "grad_norm": 17.255992731240948, "learning_rate": 2.7765941989529444e-06, "loss": 0.4367, "step": 37720 }, { "epoch": 10.233251966368321, "grad_norm": 15.505110435036196, "learning_rate": 2.7737679757125203e-06, "loss": 0.4273, "step": 37730 }, { "epoch": 10.235964198535395, "grad_norm": 14.40770717995496, "learning_rate": 2.77094263935252e-06, "loss": 0.4724, "step": 37740 }, { "epoch": 10.238676430702467, "grad_norm": 16.498222913759946, "learning_rate": 2.768118190998496e-06, "loss": 0.4321, "step": 37750 }, { "epoch": 10.241388662869541, "grad_norm": 16.66037639411717, "learning_rate": 2.7652946317756447e-06, "loss": 0.4381, "step": 37760 }, { "epoch": 10.244100895036615, "grad_norm": 19.82963034593835, "learning_rate": 2.762471962808809e-06, "loss": 0.4361, "step": 37770 }, { "epoch": 10.246813127203689, "grad_norm": 19.11689879178498, "learning_rate": 2.759650185222473e-06, "loss": 0.4633, "step": 37780 }, { "epoch": 10.249525359370763, "grad_norm": 16.854261719484274, "learning_rate": 2.756829300140769e-06, "loss": 0.4523, "step": 37790 }, { "epoch": 10.252237591537835, "grad_norm": 24.344154711409203, "learning_rate": 2.7540093086874796e-06, "loss": 0.4349, "step": 37800 }, { "epoch": 10.254949823704909, "grad_norm": 17.583872451955987, "learning_rate": 2.7511902119860202e-06, "loss": 0.4247, "step": 37810 }, { "epoch": 10.257662055871982, "grad_norm": 16.093462834311627, "learning_rate": 2.7483720111594582e-06, "loss": 0.4325, "step": 37820 }, { "epoch": 10.260374288039056, "grad_norm": 17.499944345928327, "learning_rate": 2.7455547073304955e-06, "loss": 0.4633, "step": 37830 }, { "epoch": 10.26308652020613, "grad_norm": 15.761009315406286, "learning_rate": 2.7427383016214896e-06, "loss": 0.419, "step": 37840 }, { "epoch": 10.265798752373204, "grad_norm": 19.181425364848867, "learning_rate": 2.7399227951544272e-06, "loss": 0.4686, "step": 37850 }, { "epoch": 10.268510984540276, "grad_norm": 13.958264222725264, "learning_rate": 2.7371081890509448e-06, "loss": 0.453, "step": 37860 }, { "epoch": 10.27122321670735, "grad_norm": 18.496280637496525, "learning_rate": 2.7342944844323173e-06, "loss": 0.4342, "step": 37870 }, { "epoch": 10.273935448874424, "grad_norm": 14.345545382344868, "learning_rate": 2.7314816824194603e-06, "loss": 0.4345, "step": 37880 }, { "epoch": 10.276647681041498, "grad_norm": 15.119694208141533, "learning_rate": 2.7286697841329324e-06, "loss": 0.3978, "step": 37890 }, { "epoch": 10.279359913208571, "grad_norm": 15.558285651919185, "learning_rate": 2.7258587906929274e-06, "loss": 0.4463, "step": 37900 }, { "epoch": 10.282072145375643, "grad_norm": 17.443437973835863, "learning_rate": 2.7230487032192826e-06, "loss": 0.4686, "step": 37910 }, { "epoch": 10.284784377542717, "grad_norm": 21.938427324155512, "learning_rate": 2.7202395228314736e-06, "loss": 0.4512, "step": 37920 }, { "epoch": 10.287496609709791, "grad_norm": 15.454650536789357, "learning_rate": 2.717431250648614e-06, "loss": 0.4376, "step": 37930 }, { "epoch": 10.290208841876865, "grad_norm": 13.928085849925118, "learning_rate": 2.7146238877894577e-06, "loss": 0.3611, "step": 37940 }, { "epoch": 10.292921074043939, "grad_norm": 19.910574724242224, "learning_rate": 2.7118174353723886e-06, "loss": 0.4622, "step": 37950 }, { "epoch": 10.29563330621101, "grad_norm": 15.092660153390794, "learning_rate": 2.709011894515442e-06, "loss": 0.3903, "step": 37960 }, { "epoch": 10.298345538378085, "grad_norm": 18.926462295392966, "learning_rate": 2.7062072663362754e-06, "loss": 0.3678, "step": 37970 }, { "epoch": 10.301057770545158, "grad_norm": 18.33522960690953, "learning_rate": 2.7034035519521918e-06, "loss": 0.4261, "step": 37980 }, { "epoch": 10.303770002712232, "grad_norm": 17.630676743540167, "learning_rate": 2.700600752480127e-06, "loss": 0.4612, "step": 37990 }, { "epoch": 10.306482234879306, "grad_norm": 21.747327866660125, "learning_rate": 2.697798869036653e-06, "loss": 0.4443, "step": 38000 }, { "epoch": 10.30919446704638, "grad_norm": 22.310362526429188, "learning_rate": 2.694997902737979e-06, "loss": 0.4287, "step": 38010 }, { "epoch": 10.311906699213452, "grad_norm": 18.109016441696053, "learning_rate": 2.692197854699942e-06, "loss": 0.4877, "step": 38020 }, { "epoch": 10.314618931380526, "grad_norm": 16.004695643069578, "learning_rate": 2.6893987260380206e-06, "loss": 0.3969, "step": 38030 }, { "epoch": 10.3173311635476, "grad_norm": 13.881002454478516, "learning_rate": 2.686600517867323e-06, "loss": 0.4677, "step": 38040 }, { "epoch": 10.320043395714674, "grad_norm": 18.20144456579833, "learning_rate": 2.683803231302593e-06, "loss": 0.4054, "step": 38050 }, { "epoch": 10.322755627881747, "grad_norm": 16.918255794852268, "learning_rate": 2.681006867458208e-06, "loss": 0.4565, "step": 38060 }, { "epoch": 10.32546786004882, "grad_norm": 18.00780741276161, "learning_rate": 2.6782114274481695e-06, "loss": 0.414, "step": 38070 }, { "epoch": 10.328180092215893, "grad_norm": 14.714893016127348, "learning_rate": 2.675416912386126e-06, "loss": 0.4091, "step": 38080 }, { "epoch": 10.330892324382967, "grad_norm": 16.23183251043773, "learning_rate": 2.6726233233853433e-06, "loss": 0.4256, "step": 38090 }, { "epoch": 10.333604556550041, "grad_norm": 17.22474230411863, "learning_rate": 2.6698306615587254e-06, "loss": 0.4039, "step": 38100 }, { "epoch": 10.336316788717115, "grad_norm": 16.163646111083157, "learning_rate": 2.6670389280188056e-06, "loss": 0.4503, "step": 38110 }, { "epoch": 10.339029020884187, "grad_norm": 17.691665896625913, "learning_rate": 2.6642481238777472e-06, "loss": 0.4132, "step": 38120 }, { "epoch": 10.34174125305126, "grad_norm": 11.10499509951542, "learning_rate": 2.661458250247346e-06, "loss": 0.3935, "step": 38130 }, { "epoch": 10.344453485218335, "grad_norm": 16.171045473381135, "learning_rate": 2.658669308239021e-06, "loss": 0.4636, "step": 38140 }, { "epoch": 10.347165717385408, "grad_norm": 17.231332395319573, "learning_rate": 2.655881298963824e-06, "loss": 0.4079, "step": 38150 }, { "epoch": 10.349877949552482, "grad_norm": 17.190921201315163, "learning_rate": 2.653094223532435e-06, "loss": 0.469, "step": 38160 }, { "epoch": 10.352590181719556, "grad_norm": 15.00857835441651, "learning_rate": 2.6503080830551627e-06, "loss": 0.4214, "step": 38170 }, { "epoch": 10.355302413886628, "grad_norm": 13.181118050374268, "learning_rate": 2.6475228786419437e-06, "loss": 0.4404, "step": 38180 }, { "epoch": 10.358014646053702, "grad_norm": 19.311180805692725, "learning_rate": 2.644738611402334e-06, "loss": 0.3907, "step": 38190 }, { "epoch": 10.360726878220776, "grad_norm": 14.120824319771314, "learning_rate": 2.6419552824455318e-06, "loss": 0.4769, "step": 38200 }, { "epoch": 10.36343911038785, "grad_norm": 18.875420190321293, "learning_rate": 2.6391728928803463e-06, "loss": 0.5016, "step": 38210 }, { "epoch": 10.366151342554923, "grad_norm": 18.35759792163082, "learning_rate": 2.6363914438152182e-06, "loss": 0.4193, "step": 38220 }, { "epoch": 10.368863574721995, "grad_norm": 17.52754561380285, "learning_rate": 2.633610936358217e-06, "loss": 0.424, "step": 38230 }, { "epoch": 10.37157580688907, "grad_norm": 16.872687452081855, "learning_rate": 2.6308313716170316e-06, "loss": 0.4264, "step": 38240 }, { "epoch": 10.374288039056143, "grad_norm": 16.181314532509937, "learning_rate": 2.6280527506989807e-06, "loss": 0.4167, "step": 38250 }, { "epoch": 10.377000271223217, "grad_norm": 17.623879825680834, "learning_rate": 2.6252750747109995e-06, "loss": 0.4274, "step": 38260 }, { "epoch": 10.37971250339029, "grad_norm": 19.758000728859145, "learning_rate": 2.622498344759653e-06, "loss": 0.4311, "step": 38270 }, { "epoch": 10.382424735557363, "grad_norm": 19.314569387385053, "learning_rate": 2.619722561951129e-06, "loss": 0.3899, "step": 38280 }, { "epoch": 10.385136967724437, "grad_norm": 18.220155335047373, "learning_rate": 2.6169477273912347e-06, "loss": 0.4212, "step": 38290 }, { "epoch": 10.38784919989151, "grad_norm": 18.748671079820696, "learning_rate": 2.6141738421854046e-06, "loss": 0.3677, "step": 38300 }, { "epoch": 10.390561432058584, "grad_norm": 15.313856667259403, "learning_rate": 2.611400907438685e-06, "loss": 0.4486, "step": 38310 }, { "epoch": 10.393273664225658, "grad_norm": 19.33273893840344, "learning_rate": 2.6086289242557584e-06, "loss": 0.4219, "step": 38320 }, { "epoch": 10.395985896392732, "grad_norm": 26.3041563944588, "learning_rate": 2.6058578937409145e-06, "loss": 0.4232, "step": 38330 }, { "epoch": 10.398698128559804, "grad_norm": 19.745488273827426, "learning_rate": 2.603087816998071e-06, "loss": 0.4144, "step": 38340 }, { "epoch": 10.401410360726878, "grad_norm": 17.32360678542504, "learning_rate": 2.6003186951307633e-06, "loss": 0.4494, "step": 38350 }, { "epoch": 10.404122592893952, "grad_norm": 15.406770270262315, "learning_rate": 2.5975505292421476e-06, "loss": 0.4333, "step": 38360 }, { "epoch": 10.406834825061026, "grad_norm": 19.16289222911073, "learning_rate": 2.5947833204350003e-06, "loss": 0.4411, "step": 38370 }, { "epoch": 10.4095470572281, "grad_norm": 11.666429767377737, "learning_rate": 2.59201706981171e-06, "loss": 0.4035, "step": 38380 }, { "epoch": 10.412259289395172, "grad_norm": 19.790806695909545, "learning_rate": 2.589251778474291e-06, "loss": 0.4584, "step": 38390 }, { "epoch": 10.414971521562245, "grad_norm": 22.098832814085426, "learning_rate": 2.5864874475243727e-06, "loss": 0.467, "step": 38400 }, { "epoch": 10.41768375372932, "grad_norm": 16.285689635619335, "learning_rate": 2.5837240780632016e-06, "loss": 0.445, "step": 38410 }, { "epoch": 10.420395985896393, "grad_norm": 15.340606306737586, "learning_rate": 2.580961671191644e-06, "loss": 0.4097, "step": 38420 }, { "epoch": 10.423108218063467, "grad_norm": 17.04169041792144, "learning_rate": 2.578200228010174e-06, "loss": 0.4078, "step": 38430 }, { "epoch": 10.425820450230539, "grad_norm": 14.577107468217504, "learning_rate": 2.575439749618896e-06, "loss": 0.4566, "step": 38440 }, { "epoch": 10.428532682397613, "grad_norm": 14.279972270579725, "learning_rate": 2.5726802371175162e-06, "loss": 0.47, "step": 38450 }, { "epoch": 10.431244914564687, "grad_norm": 17.220755921803505, "learning_rate": 2.5699216916053636e-06, "loss": 0.3895, "step": 38460 }, { "epoch": 10.43395714673176, "grad_norm": 17.38654016138206, "learning_rate": 2.567164114181381e-06, "loss": 0.4461, "step": 38470 }, { "epoch": 10.436669378898834, "grad_norm": 13.733736869175305, "learning_rate": 2.5644075059441236e-06, "loss": 0.433, "step": 38480 }, { "epoch": 10.439381611065908, "grad_norm": 14.61719842412621, "learning_rate": 2.5616518679917646e-06, "loss": 0.4102, "step": 38490 }, { "epoch": 10.44209384323298, "grad_norm": 15.729636861351613, "learning_rate": 2.558897201422084e-06, "loss": 0.4607, "step": 38500 }, { "epoch": 10.444806075400054, "grad_norm": 16.016209900036053, "learning_rate": 2.556143507332479e-06, "loss": 0.4404, "step": 38510 }, { "epoch": 10.447518307567128, "grad_norm": 16.681383652202374, "learning_rate": 2.5533907868199616e-06, "loss": 0.4192, "step": 38520 }, { "epoch": 10.450230539734202, "grad_norm": 17.99037513538817, "learning_rate": 2.5506390409811504e-06, "loss": 0.461, "step": 38530 }, { "epoch": 10.452942771901276, "grad_norm": 13.570500291522286, "learning_rate": 2.5478882709122827e-06, "loss": 0.3998, "step": 38540 }, { "epoch": 10.455655004068348, "grad_norm": 18.12696255358241, "learning_rate": 2.5451384777091962e-06, "loss": 0.4891, "step": 38550 }, { "epoch": 10.458367236235421, "grad_norm": 19.325167471930154, "learning_rate": 2.5423896624673542e-06, "loss": 0.4215, "step": 38560 }, { "epoch": 10.461079468402495, "grad_norm": 18.334744567547176, "learning_rate": 2.539641826281817e-06, "loss": 0.4384, "step": 38570 }, { "epoch": 10.463791700569569, "grad_norm": 14.60545135738052, "learning_rate": 2.536894970247264e-06, "loss": 0.416, "step": 38580 }, { "epoch": 10.466503932736643, "grad_norm": 20.71856327556003, "learning_rate": 2.534149095457974e-06, "loss": 0.3973, "step": 38590 }, { "epoch": 10.469216164903715, "grad_norm": 18.414943303597457, "learning_rate": 2.531404203007848e-06, "loss": 0.422, "step": 38600 }, { "epoch": 10.471928397070789, "grad_norm": 20.67846649747711, "learning_rate": 2.5286602939903885e-06, "loss": 0.4331, "step": 38610 }, { "epoch": 10.474640629237863, "grad_norm": 21.64240254709942, "learning_rate": 2.5259173694987034e-06, "loss": 0.481, "step": 38620 }, { "epoch": 10.477352861404936, "grad_norm": 21.613801681910594, "learning_rate": 2.523175430625513e-06, "loss": 0.5041, "step": 38630 }, { "epoch": 10.48006509357201, "grad_norm": 20.618948100921966, "learning_rate": 2.5204344784631436e-06, "loss": 0.472, "step": 38640 }, { "epoch": 10.482777325739082, "grad_norm": 17.268402032036676, "learning_rate": 2.5176945141035293e-06, "loss": 0.493, "step": 38650 }, { "epoch": 10.485489557906156, "grad_norm": 13.11920807373632, "learning_rate": 2.514955538638211e-06, "loss": 0.3815, "step": 38660 }, { "epoch": 10.48820179007323, "grad_norm": 16.618205141773338, "learning_rate": 2.512217553158328e-06, "loss": 0.4135, "step": 38670 }, { "epoch": 10.490914022240304, "grad_norm": 16.743622146183466, "learning_rate": 2.509480558754642e-06, "loss": 0.4094, "step": 38680 }, { "epoch": 10.493626254407378, "grad_norm": 18.618653121507126, "learning_rate": 2.5067445565175006e-06, "loss": 0.4643, "step": 38690 }, { "epoch": 10.496338486574452, "grad_norm": 19.375718510851417, "learning_rate": 2.504009547536872e-06, "loss": 0.4358, "step": 38700 }, { "epoch": 10.499050718741524, "grad_norm": 16.845551158805538, "learning_rate": 2.501275532902314e-06, "loss": 0.4293, "step": 38710 }, { "epoch": 10.501762950908597, "grad_norm": 15.469267910351805, "learning_rate": 2.498542513703003e-06, "loss": 0.3961, "step": 38720 }, { "epoch": 10.504475183075671, "grad_norm": 16.774074734775027, "learning_rate": 2.495810491027711e-06, "loss": 0.4457, "step": 38730 }, { "epoch": 10.507187415242745, "grad_norm": 20.20422426522852, "learning_rate": 2.493079465964809e-06, "loss": 0.3961, "step": 38740 }, { "epoch": 10.509899647409819, "grad_norm": 18.774598346432185, "learning_rate": 2.4903494396022828e-06, "loss": 0.4605, "step": 38750 }, { "epoch": 10.512611879576891, "grad_norm": 16.164172986557457, "learning_rate": 2.4876204130277072e-06, "loss": 0.4596, "step": 38760 }, { "epoch": 10.515324111743965, "grad_norm": 18.42435837028434, "learning_rate": 2.4848923873282686e-06, "loss": 0.4849, "step": 38770 }, { "epoch": 10.518036343911039, "grad_norm": 14.185347333395729, "learning_rate": 2.4821653635907457e-06, "loss": 0.4323, "step": 38780 }, { "epoch": 10.520748576078113, "grad_norm": 20.707777935701095, "learning_rate": 2.4794393429015278e-06, "loss": 0.4342, "step": 38790 }, { "epoch": 10.523460808245186, "grad_norm": 16.50177194601137, "learning_rate": 2.4767143263466e-06, "loss": 0.4274, "step": 38800 }, { "epoch": 10.52617304041226, "grad_norm": 20.728284815804635, "learning_rate": 2.4739903150115434e-06, "loss": 0.4714, "step": 38810 }, { "epoch": 10.528885272579332, "grad_norm": 18.79177019931906, "learning_rate": 2.471267309981544e-06, "loss": 0.4047, "step": 38820 }, { "epoch": 10.531597504746406, "grad_norm": 18.07874191206222, "learning_rate": 2.4685453123413863e-06, "loss": 0.4936, "step": 38830 }, { "epoch": 10.53430973691348, "grad_norm": 16.399463605798555, "learning_rate": 2.4658243231754536e-06, "loss": 0.438, "step": 38840 }, { "epoch": 10.537021969080554, "grad_norm": 14.9053594322999, "learning_rate": 2.4631043435677225e-06, "loss": 0.428, "step": 38850 }, { "epoch": 10.539734201247628, "grad_norm": 18.270491295030393, "learning_rate": 2.460385374601772e-06, "loss": 0.4391, "step": 38860 }, { "epoch": 10.5424464334147, "grad_norm": 18.240437668596236, "learning_rate": 2.4576674173607827e-06, "loss": 0.4233, "step": 38870 }, { "epoch": 10.545158665581774, "grad_norm": 13.96026450954483, "learning_rate": 2.4549504729275224e-06, "loss": 0.4516, "step": 38880 }, { "epoch": 10.547870897748847, "grad_norm": 15.32371913837672, "learning_rate": 2.452234542384363e-06, "loss": 0.4394, "step": 38890 }, { "epoch": 10.550583129915921, "grad_norm": 16.217024564327495, "learning_rate": 2.449519626813265e-06, "loss": 0.4398, "step": 38900 }, { "epoch": 10.553295362082995, "grad_norm": 14.702345634490381, "learning_rate": 2.4468057272957963e-06, "loss": 0.4043, "step": 38910 }, { "epoch": 10.556007594250067, "grad_norm": 16.21287271980486, "learning_rate": 2.444092844913107e-06, "loss": 0.4431, "step": 38920 }, { "epoch": 10.558719826417141, "grad_norm": 15.25895668503815, "learning_rate": 2.4413809807459502e-06, "loss": 0.407, "step": 38930 }, { "epoch": 10.561432058584215, "grad_norm": 14.916541139173901, "learning_rate": 2.438670135874672e-06, "loss": 0.4536, "step": 38940 }, { "epoch": 10.564144290751289, "grad_norm": 13.865900456457357, "learning_rate": 2.435960311379211e-06, "loss": 0.4293, "step": 38950 }, { "epoch": 10.566856522918362, "grad_norm": 20.18769953213402, "learning_rate": 2.4332515083391007e-06, "loss": 0.4623, "step": 38960 }, { "epoch": 10.569568755085434, "grad_norm": 17.521007483657225, "learning_rate": 2.430543727833465e-06, "loss": 0.4325, "step": 38970 }, { "epoch": 10.572280987252508, "grad_norm": 16.641789491241067, "learning_rate": 2.427836970941023e-06, "loss": 0.4247, "step": 38980 }, { "epoch": 10.574993219419582, "grad_norm": 16.38889947550477, "learning_rate": 2.4251312387400856e-06, "loss": 0.5075, "step": 38990 }, { "epoch": 10.577705451586656, "grad_norm": 24.456069164210316, "learning_rate": 2.4224265323085555e-06, "loss": 0.4984, "step": 39000 }, { "epoch": 10.58041768375373, "grad_norm": 15.033215988923374, "learning_rate": 2.4197228527239288e-06, "loss": 0.3889, "step": 39010 }, { "epoch": 10.583129915920804, "grad_norm": 13.031457046019192, "learning_rate": 2.4170202010632833e-06, "loss": 0.4977, "step": 39020 }, { "epoch": 10.585842148087876, "grad_norm": 14.391263172021533, "learning_rate": 2.4143185784033026e-06, "loss": 0.3948, "step": 39030 }, { "epoch": 10.58855438025495, "grad_norm": 19.459515100231236, "learning_rate": 2.4116179858202458e-06, "loss": 0.4744, "step": 39040 }, { "epoch": 10.591266612422023, "grad_norm": 18.33261598072098, "learning_rate": 2.408918424389971e-06, "loss": 0.4377, "step": 39050 }, { "epoch": 10.593978844589097, "grad_norm": 15.631514405706223, "learning_rate": 2.406219895187921e-06, "loss": 0.4221, "step": 39060 }, { "epoch": 10.596691076756171, "grad_norm": 15.567066553803922, "learning_rate": 2.4035223992891288e-06, "loss": 0.4166, "step": 39070 }, { "epoch": 10.599403308923243, "grad_norm": 19.00715409634265, "learning_rate": 2.4008259377682186e-06, "loss": 0.4617, "step": 39080 }, { "epoch": 10.602115541090317, "grad_norm": 15.727498309582497, "learning_rate": 2.398130511699395e-06, "loss": 0.4196, "step": 39090 }, { "epoch": 10.60482777325739, "grad_norm": 17.07989577941781, "learning_rate": 2.3954361221564566e-06, "loss": 0.4078, "step": 39100 }, { "epoch": 10.607540005424465, "grad_norm": 17.00459666146143, "learning_rate": 2.3927427702127877e-06, "loss": 0.4345, "step": 39110 }, { "epoch": 10.610252237591538, "grad_norm": 19.208561332490124, "learning_rate": 2.3900504569413577e-06, "loss": 0.4028, "step": 39120 }, { "epoch": 10.61296446975861, "grad_norm": 21.017988729900928, "learning_rate": 2.3873591834147265e-06, "loss": 0.4497, "step": 39130 }, { "epoch": 10.615676701925684, "grad_norm": 20.285623758520153, "learning_rate": 2.3846689507050296e-06, "loss": 0.4676, "step": 39140 }, { "epoch": 10.618388934092758, "grad_norm": 19.50231213290663, "learning_rate": 2.381979759884003e-06, "loss": 0.4997, "step": 39150 }, { "epoch": 10.621101166259832, "grad_norm": 23.343657693256173, "learning_rate": 2.379291612022954e-06, "loss": 0.4963, "step": 39160 }, { "epoch": 10.623813398426906, "grad_norm": 16.711929327648914, "learning_rate": 2.3766045081927814e-06, "loss": 0.4556, "step": 39170 }, { "epoch": 10.626525630593978, "grad_norm": 18.64540924988773, "learning_rate": 2.3739184494639667e-06, "loss": 0.4194, "step": 39180 }, { "epoch": 10.629237862761052, "grad_norm": 19.71374079848367, "learning_rate": 2.3712334369065743e-06, "loss": 0.4307, "step": 39190 }, { "epoch": 10.631950094928126, "grad_norm": 16.480726586572903, "learning_rate": 2.3685494715902548e-06, "loss": 0.4258, "step": 39200 }, { "epoch": 10.6346623270952, "grad_norm": 18.553687944338993, "learning_rate": 2.365866554584235e-06, "loss": 0.4234, "step": 39210 }, { "epoch": 10.637374559262273, "grad_norm": 15.749776083450717, "learning_rate": 2.363184686957331e-06, "loss": 0.4714, "step": 39220 }, { "epoch": 10.640086791429347, "grad_norm": 22.97364588044247, "learning_rate": 2.3605038697779376e-06, "loss": 0.3922, "step": 39230 }, { "epoch": 10.64279902359642, "grad_norm": 18.73234179129017, "learning_rate": 2.357824104114031e-06, "loss": 0.4741, "step": 39240 }, { "epoch": 10.645511255763493, "grad_norm": 14.225025492588, "learning_rate": 2.3551453910331725e-06, "loss": 0.4135, "step": 39250 }, { "epoch": 10.648223487930567, "grad_norm": 16.139439005332623, "learning_rate": 2.352467731602494e-06, "loss": 0.4285, "step": 39260 }, { "epoch": 10.65093572009764, "grad_norm": 17.878714067980308, "learning_rate": 2.349791126888722e-06, "loss": 0.4467, "step": 39270 }, { "epoch": 10.653647952264715, "grad_norm": 17.266682018395787, "learning_rate": 2.34711557795815e-06, "loss": 0.439, "step": 39280 }, { "epoch": 10.656360184431787, "grad_norm": 14.504856037515466, "learning_rate": 2.3444410858766577e-06, "loss": 0.493, "step": 39290 }, { "epoch": 10.65907241659886, "grad_norm": 22.871092139991916, "learning_rate": 2.341767651709702e-06, "loss": 0.4524, "step": 39300 }, { "epoch": 10.661784648765934, "grad_norm": 18.089864637812354, "learning_rate": 2.3390952765223195e-06, "loss": 0.4473, "step": 39310 }, { "epoch": 10.664496880933008, "grad_norm": 16.56913615139851, "learning_rate": 2.3364239613791245e-06, "loss": 0.4223, "step": 39320 }, { "epoch": 10.667209113100082, "grad_norm": 20.67749994300025, "learning_rate": 2.3337537073443057e-06, "loss": 0.4298, "step": 39330 }, { "epoch": 10.669921345267156, "grad_norm": 17.0273688669222, "learning_rate": 2.331084515481633e-06, "loss": 0.427, "step": 39340 }, { "epoch": 10.672633577434228, "grad_norm": 21.984156280046776, "learning_rate": 2.3284163868544525e-06, "loss": 0.4285, "step": 39350 }, { "epoch": 10.675345809601302, "grad_norm": 14.87146799653432, "learning_rate": 2.3257493225256856e-06, "loss": 0.4435, "step": 39360 }, { "epoch": 10.678058041768375, "grad_norm": 17.62729450046788, "learning_rate": 2.323083323557833e-06, "loss": 0.5015, "step": 39370 }, { "epoch": 10.68077027393545, "grad_norm": 16.847747702782115, "learning_rate": 2.3204183910129623e-06, "loss": 0.4798, "step": 39380 }, { "epoch": 10.683482506102523, "grad_norm": 18.289613286251214, "learning_rate": 2.31775452595273e-06, "loss": 0.3737, "step": 39390 }, { "epoch": 10.686194738269595, "grad_norm": 14.837524279204418, "learning_rate": 2.315091729438354e-06, "loss": 0.455, "step": 39400 }, { "epoch": 10.688906970436669, "grad_norm": 20.47851299114392, "learning_rate": 2.3124300025306334e-06, "loss": 0.4889, "step": 39410 }, { "epoch": 10.691619202603743, "grad_norm": 17.518172760718084, "learning_rate": 2.3097693462899405e-06, "loss": 0.496, "step": 39420 }, { "epoch": 10.694331434770817, "grad_norm": 19.03773620982091, "learning_rate": 2.3071097617762206e-06, "loss": 0.4334, "step": 39430 }, { "epoch": 10.69704366693789, "grad_norm": 16.140160410332552, "learning_rate": 2.3044512500489933e-06, "loss": 0.4593, "step": 39440 }, { "epoch": 10.699755899104963, "grad_norm": 18.295091822076014, "learning_rate": 2.301793812167346e-06, "loss": 0.4104, "step": 39450 }, { "epoch": 10.702468131272036, "grad_norm": 17.00642197117806, "learning_rate": 2.2991374491899437e-06, "loss": 0.4632, "step": 39460 }, { "epoch": 10.70518036343911, "grad_norm": 17.880531983339413, "learning_rate": 2.296482162175021e-06, "loss": 0.4363, "step": 39470 }, { "epoch": 10.707892595606184, "grad_norm": 18.68172244035056, "learning_rate": 2.2938279521803847e-06, "loss": 0.4373, "step": 39480 }, { "epoch": 10.710604827773258, "grad_norm": 16.201740384331522, "learning_rate": 2.291174820263413e-06, "loss": 0.4226, "step": 39490 }, { "epoch": 10.71331705994033, "grad_norm": 20.557926155408026, "learning_rate": 2.288522767481049e-06, "loss": 0.4817, "step": 39500 }, { "epoch": 10.716029292107404, "grad_norm": 15.997139641984633, "learning_rate": 2.2858717948898174e-06, "loss": 0.4711, "step": 39510 }, { "epoch": 10.718741524274478, "grad_norm": 15.472191528235356, "learning_rate": 2.2832219035457996e-06, "loss": 0.4299, "step": 39520 }, { "epoch": 10.721453756441552, "grad_norm": 16.087027092628585, "learning_rate": 2.2805730945046544e-06, "loss": 0.4368, "step": 39530 }, { "epoch": 10.724165988608625, "grad_norm": 23.756966736391707, "learning_rate": 2.277925368821608e-06, "loss": 0.4627, "step": 39540 }, { "epoch": 10.7268782207757, "grad_norm": 15.847170123467457, "learning_rate": 2.2752787275514537e-06, "loss": 0.4169, "step": 39550 }, { "epoch": 10.729590452942771, "grad_norm": 20.369187000955893, "learning_rate": 2.2726331717485554e-06, "loss": 0.4375, "step": 39560 }, { "epoch": 10.732302685109845, "grad_norm": 19.704916035579867, "learning_rate": 2.2699887024668394e-06, "loss": 0.467, "step": 39570 }, { "epoch": 10.735014917276919, "grad_norm": 17.10086472278112, "learning_rate": 2.2673453207598033e-06, "loss": 0.3895, "step": 39580 }, { "epoch": 10.737727149443993, "grad_norm": 15.003692315559611, "learning_rate": 2.264703027680511e-06, "loss": 0.5147, "step": 39590 }, { "epoch": 10.740439381611067, "grad_norm": 14.285271750361115, "learning_rate": 2.2620618242815926e-06, "loss": 0.3987, "step": 39600 }, { "epoch": 10.743151613778139, "grad_norm": 18.815599889334976, "learning_rate": 2.259421711615245e-06, "loss": 0.4356, "step": 39610 }, { "epoch": 10.745863845945212, "grad_norm": 15.907279796735162, "learning_rate": 2.2567826907332246e-06, "loss": 0.3767, "step": 39620 }, { "epoch": 10.748576078112286, "grad_norm": 19.42449380633103, "learning_rate": 2.254144762686864e-06, "loss": 0.4508, "step": 39630 }, { "epoch": 10.75128831027936, "grad_norm": 16.015669642167136, "learning_rate": 2.2515079285270484e-06, "loss": 0.5022, "step": 39640 }, { "epoch": 10.754000542446434, "grad_norm": 20.34479962475831, "learning_rate": 2.248872189304235e-06, "loss": 0.4703, "step": 39650 }, { "epoch": 10.756712774613508, "grad_norm": 17.942582894093153, "learning_rate": 2.246237546068444e-06, "loss": 0.4007, "step": 39660 }, { "epoch": 10.75942500678058, "grad_norm": 15.67477658321903, "learning_rate": 2.243603999869255e-06, "loss": 0.4482, "step": 39670 }, { "epoch": 10.762137238947654, "grad_norm": 15.764949649829479, "learning_rate": 2.240971551755817e-06, "loss": 0.433, "step": 39680 }, { "epoch": 10.764849471114728, "grad_norm": 17.67167289666028, "learning_rate": 2.238340202776833e-06, "loss": 0.4445, "step": 39690 }, { "epoch": 10.767561703281801, "grad_norm": 19.20325629448629, "learning_rate": 2.235709953980575e-06, "loss": 0.4418, "step": 39700 }, { "epoch": 10.770273935448875, "grad_norm": 19.57228760713134, "learning_rate": 2.2330808064148745e-06, "loss": 0.4093, "step": 39710 }, { "epoch": 10.772986167615947, "grad_norm": 17.694744362559543, "learning_rate": 2.230452761127125e-06, "loss": 0.4521, "step": 39720 }, { "epoch": 10.775698399783021, "grad_norm": 16.169995385133085, "learning_rate": 2.227825819164281e-06, "loss": 0.4586, "step": 39730 }, { "epoch": 10.778410631950095, "grad_norm": 16.161486415999583, "learning_rate": 2.225199981572852e-06, "loss": 0.4359, "step": 39740 }, { "epoch": 10.781122864117169, "grad_norm": 16.69259965674746, "learning_rate": 2.222575249398919e-06, "loss": 0.4678, "step": 39750 }, { "epoch": 10.783835096284243, "grad_norm": 16.952005604454545, "learning_rate": 2.219951623688111e-06, "loss": 0.4398, "step": 39760 }, { "epoch": 10.786547328451315, "grad_norm": 18.96998875556029, "learning_rate": 2.217329105485624e-06, "loss": 0.3978, "step": 39770 }, { "epoch": 10.789259560618389, "grad_norm": 16.096183305373938, "learning_rate": 2.2147076958362047e-06, "loss": 0.3966, "step": 39780 }, { "epoch": 10.791971792785462, "grad_norm": 14.775770976031682, "learning_rate": 2.212087395784169e-06, "loss": 0.3948, "step": 39790 }, { "epoch": 10.794684024952536, "grad_norm": 16.678607111108146, "learning_rate": 2.209468206373385e-06, "loss": 0.4467, "step": 39800 }, { "epoch": 10.79739625711961, "grad_norm": 14.495201714420142, "learning_rate": 2.206850128647275e-06, "loss": 0.3741, "step": 39810 }, { "epoch": 10.800108489286682, "grad_norm": 15.184895516164934, "learning_rate": 2.2042331636488236e-06, "loss": 0.5051, "step": 39820 }, { "epoch": 10.802820721453756, "grad_norm": 17.755570148883475, "learning_rate": 2.2016173124205696e-06, "loss": 0.426, "step": 39830 }, { "epoch": 10.80553295362083, "grad_norm": 16.01129429804462, "learning_rate": 2.1990025760046117e-06, "loss": 0.4175, "step": 39840 }, { "epoch": 10.808245185787904, "grad_norm": 21.08466856145485, "learning_rate": 2.196388955442598e-06, "loss": 0.4156, "step": 39850 }, { "epoch": 10.810957417954977, "grad_norm": 18.7141060868103, "learning_rate": 2.1937764517757355e-06, "loss": 0.4319, "step": 39860 }, { "epoch": 10.813669650122051, "grad_norm": 18.844123390277193, "learning_rate": 2.1911650660447915e-06, "loss": 0.4725, "step": 39870 }, { "epoch": 10.816381882289123, "grad_norm": 16.17484997905716, "learning_rate": 2.188554799290078e-06, "loss": 0.4312, "step": 39880 }, { "epoch": 10.819094114456197, "grad_norm": 15.401855992370505, "learning_rate": 2.18594565255147e-06, "loss": 0.4379, "step": 39890 }, { "epoch": 10.821806346623271, "grad_norm": 18.582740843796927, "learning_rate": 2.1833376268683865e-06, "loss": 0.435, "step": 39900 }, { "epoch": 10.824518578790345, "grad_norm": 19.86934644178541, "learning_rate": 2.1807307232798134e-06, "loss": 0.424, "step": 39910 }, { "epoch": 10.827230810957419, "grad_norm": 14.73194666197724, "learning_rate": 2.178124942824277e-06, "loss": 0.4444, "step": 39920 }, { "epoch": 10.82994304312449, "grad_norm": 19.600322952216192, "learning_rate": 2.1755202865398622e-06, "loss": 0.4445, "step": 39930 }, { "epoch": 10.832655275291565, "grad_norm": 22.268631476940918, "learning_rate": 2.1729167554642057e-06, "loss": 0.4696, "step": 39940 }, { "epoch": 10.835367507458638, "grad_norm": 19.09604020966599, "learning_rate": 2.170314350634495e-06, "loss": 0.452, "step": 39950 }, { "epoch": 10.838079739625712, "grad_norm": 14.633276626819978, "learning_rate": 2.1677130730874713e-06, "loss": 0.4754, "step": 39960 }, { "epoch": 10.840791971792786, "grad_norm": 18.17250021906898, "learning_rate": 2.165112923859421e-06, "loss": 0.4826, "step": 39970 }, { "epoch": 10.843504203959858, "grad_norm": 19.447405756793565, "learning_rate": 2.1625139039861862e-06, "loss": 0.4885, "step": 39980 }, { "epoch": 10.846216436126932, "grad_norm": 17.17157894070352, "learning_rate": 2.1599160145031585e-06, "loss": 0.3868, "step": 39990 }, { "epoch": 10.848928668294006, "grad_norm": 21.828840403934812, "learning_rate": 2.157319256445277e-06, "loss": 0.514, "step": 40000 }, { "epoch": 10.85164090046108, "grad_norm": 19.693716891318772, "learning_rate": 2.154723630847033e-06, "loss": 0.4229, "step": 40010 }, { "epoch": 10.854353132628153, "grad_norm": 21.18812445759736, "learning_rate": 2.1521291387424614e-06, "loss": 0.3967, "step": 40020 }, { "epoch": 10.857065364795226, "grad_norm": 17.211887206692595, "learning_rate": 2.1495357811651545e-06, "loss": 0.4747, "step": 40030 }, { "epoch": 10.8597775969623, "grad_norm": 18.287203825530497, "learning_rate": 2.1469435591482425e-06, "loss": 0.4708, "step": 40040 }, { "epoch": 10.862489829129373, "grad_norm": 18.11152524515065, "learning_rate": 2.1443524737244097e-06, "loss": 0.4367, "step": 40050 }, { "epoch": 10.865202061296447, "grad_norm": 17.808437850278118, "learning_rate": 2.141762525925886e-06, "loss": 0.4411, "step": 40060 }, { "epoch": 10.867914293463521, "grad_norm": 16.689998861738186, "learning_rate": 2.139173716784448e-06, "loss": 0.4383, "step": 40070 }, { "epoch": 10.870626525630595, "grad_norm": 14.711304060973504, "learning_rate": 2.1365860473314197e-06, "loss": 0.477, "step": 40080 }, { "epoch": 10.873338757797667, "grad_norm": 20.26099942250379, "learning_rate": 2.1339995185976677e-06, "loss": 0.4895, "step": 40090 }, { "epoch": 10.87605098996474, "grad_norm": 16.087476712802882, "learning_rate": 2.1314141316136065e-06, "loss": 0.4796, "step": 40100 }, { "epoch": 10.878763222131814, "grad_norm": 19.127446614614275, "learning_rate": 2.1288298874091978e-06, "loss": 0.4068, "step": 40110 }, { "epoch": 10.881475454298888, "grad_norm": 18.45628071083782, "learning_rate": 2.1262467870139443e-06, "loss": 0.4186, "step": 40120 }, { "epoch": 10.884187686465962, "grad_norm": 16.758323387489384, "learning_rate": 2.1236648314568973e-06, "loss": 0.5057, "step": 40130 }, { "epoch": 10.886899918633034, "grad_norm": 18.27105972678019, "learning_rate": 2.121084021766644e-06, "loss": 0.4389, "step": 40140 }, { "epoch": 10.889612150800108, "grad_norm": 15.171489064076418, "learning_rate": 2.118504358971328e-06, "loss": 0.4524, "step": 40150 }, { "epoch": 10.892324382967182, "grad_norm": 14.055795877917193, "learning_rate": 2.115925844098622e-06, "loss": 0.4456, "step": 40160 }, { "epoch": 10.895036615134256, "grad_norm": 12.320389890845513, "learning_rate": 2.113348478175751e-06, "loss": 0.489, "step": 40170 }, { "epoch": 10.89774884730133, "grad_norm": 18.41111544547326, "learning_rate": 2.110772262229478e-06, "loss": 0.4752, "step": 40180 }, { "epoch": 10.900461079468403, "grad_norm": 18.245219946639804, "learning_rate": 2.1081971972861102e-06, "loss": 0.4556, "step": 40190 }, { "epoch": 10.903173311635475, "grad_norm": 14.951132386125702, "learning_rate": 2.1056232843714968e-06, "loss": 0.4627, "step": 40200 }, { "epoch": 10.90588554380255, "grad_norm": 17.524293719984577, "learning_rate": 2.103050524511022e-06, "loss": 0.4373, "step": 40210 }, { "epoch": 10.908597775969623, "grad_norm": 16.035222673275545, "learning_rate": 2.100478918729618e-06, "loss": 0.4718, "step": 40220 }, { "epoch": 10.911310008136697, "grad_norm": 17.49504182939066, "learning_rate": 2.0979084680517542e-06, "loss": 0.4472, "step": 40230 }, { "epoch": 10.91402224030377, "grad_norm": 15.54031853832311, "learning_rate": 2.095339173501439e-06, "loss": 0.4441, "step": 40240 }, { "epoch": 10.916734472470843, "grad_norm": 22.34169569388942, "learning_rate": 2.092771036102223e-06, "loss": 0.4527, "step": 40250 }, { "epoch": 10.919446704637917, "grad_norm": 20.084201174933124, "learning_rate": 2.0902040568771897e-06, "loss": 0.4567, "step": 40260 }, { "epoch": 10.92215893680499, "grad_norm": 18.28574740474085, "learning_rate": 2.0876382368489717e-06, "loss": 0.4951, "step": 40270 }, { "epoch": 10.924871168972064, "grad_norm": 17.570818606386304, "learning_rate": 2.0850735770397283e-06, "loss": 0.4439, "step": 40280 }, { "epoch": 10.927583401139138, "grad_norm": 18.824586512203442, "learning_rate": 2.082510078471165e-06, "loss": 0.421, "step": 40290 }, { "epoch": 10.93029563330621, "grad_norm": 18.245031302131817, "learning_rate": 2.0799477421645196e-06, "loss": 0.4656, "step": 40300 }, { "epoch": 10.933007865473284, "grad_norm": 18.4876927877877, "learning_rate": 2.07738656914057e-06, "loss": 0.4971, "step": 40310 }, { "epoch": 10.935720097640358, "grad_norm": 15.442490427507682, "learning_rate": 2.074826560419631e-06, "loss": 0.4338, "step": 40320 }, { "epoch": 10.938432329807432, "grad_norm": 18.391249433590826, "learning_rate": 2.0722677170215487e-06, "loss": 0.4711, "step": 40330 }, { "epoch": 10.941144561974506, "grad_norm": 16.30800761388552, "learning_rate": 2.06971003996571e-06, "loss": 0.464, "step": 40340 }, { "epoch": 10.943856794141578, "grad_norm": 19.46234146456396, "learning_rate": 2.0671535302710362e-06, "loss": 0.4739, "step": 40350 }, { "epoch": 10.946569026308651, "grad_norm": 17.802533059414355, "learning_rate": 2.064598188955981e-06, "loss": 0.4595, "step": 40360 }, { "epoch": 10.949281258475725, "grad_norm": 18.457313266893035, "learning_rate": 2.0620440170385394e-06, "loss": 0.4188, "step": 40370 }, { "epoch": 10.9519934906428, "grad_norm": 25.340755290892027, "learning_rate": 2.059491015536227e-06, "loss": 0.4979, "step": 40380 }, { "epoch": 10.954705722809873, "grad_norm": 18.02007054969556, "learning_rate": 2.0569391854661107e-06, "loss": 0.4488, "step": 40390 }, { "epoch": 10.957417954976947, "grad_norm": 18.671729635829283, "learning_rate": 2.0543885278447766e-06, "loss": 0.4119, "step": 40400 }, { "epoch": 10.960130187144019, "grad_norm": 16.538756604898378, "learning_rate": 2.05183904368835e-06, "loss": 0.4568, "step": 40410 }, { "epoch": 10.962842419311093, "grad_norm": 17.85243481719216, "learning_rate": 2.049290734012488e-06, "loss": 0.4609, "step": 40420 }, { "epoch": 10.965554651478167, "grad_norm": 16.79333625296192, "learning_rate": 2.0467435998323794e-06, "loss": 0.3795, "step": 40430 }, { "epoch": 10.96826688364524, "grad_norm": 18.986208513796537, "learning_rate": 2.0441976421627463e-06, "loss": 0.4435, "step": 40440 }, { "epoch": 10.970979115812314, "grad_norm": 15.623086203731377, "learning_rate": 2.041652862017838e-06, "loss": 0.3676, "step": 40450 }, { "epoch": 10.973691347979386, "grad_norm": 15.214538107229858, "learning_rate": 2.0391092604114376e-06, "loss": 0.481, "step": 40460 }, { "epoch": 10.97640358014646, "grad_norm": 15.399471981267421, "learning_rate": 2.036566838356859e-06, "loss": 0.4533, "step": 40470 }, { "epoch": 10.979115812313534, "grad_norm": 18.456069548890788, "learning_rate": 2.0340255968669463e-06, "loss": 0.4704, "step": 40480 }, { "epoch": 10.981828044480608, "grad_norm": 20.76127891135601, "learning_rate": 2.0314855369540738e-06, "loss": 0.4451, "step": 40490 }, { "epoch": 10.984540276647682, "grad_norm": 13.22233552114542, "learning_rate": 2.0289466596301376e-06, "loss": 0.4831, "step": 40500 }, { "epoch": 10.987252508814755, "grad_norm": 25.049712612896922, "learning_rate": 2.0264089659065777e-06, "loss": 0.4631, "step": 40510 }, { "epoch": 10.989964740981828, "grad_norm": 21.53609215685183, "learning_rate": 2.0238724567943473e-06, "loss": 0.4166, "step": 40520 }, { "epoch": 10.992676973148901, "grad_norm": 15.08050224487761, "learning_rate": 2.0213371333039374e-06, "loss": 0.4586, "step": 40530 }, { "epoch": 10.995389205315975, "grad_norm": 18.111621678914844, "learning_rate": 2.018802996445361e-06, "loss": 0.4576, "step": 40540 }, { "epoch": 10.998101437483049, "grad_norm": 16.760137337451248, "learning_rate": 2.016270047228163e-06, "loss": 0.4309, "step": 40550 }, { "epoch": 11.000813669650123, "grad_norm": 10.804615146645926, "learning_rate": 2.013738286661414e-06, "loss": 0.3769, "step": 40560 }, { "epoch": 11.003525901817195, "grad_norm": 12.3926162048683, "learning_rate": 2.0112077157537065e-06, "loss": 0.2534, "step": 40570 }, { "epoch": 11.006238133984269, "grad_norm": 15.198395898155278, "learning_rate": 2.0086783355131645e-06, "loss": 0.3, "step": 40580 }, { "epoch": 11.008950366151343, "grad_norm": 16.343706500535667, "learning_rate": 2.0061501469474357e-06, "loss": 0.3031, "step": 40590 }, { "epoch": 11.011662598318416, "grad_norm": 13.442564545173568, "learning_rate": 2.0036231510636935e-06, "loss": 0.2726, "step": 40600 }, { "epoch": 11.01437483048549, "grad_norm": 13.192460119162698, "learning_rate": 2.001097348868637e-06, "loss": 0.2923, "step": 40610 }, { "epoch": 11.017087062652562, "grad_norm": 11.379669744087632, "learning_rate": 1.9985727413684842e-06, "loss": 0.2683, "step": 40620 }, { "epoch": 11.019799294819636, "grad_norm": 15.122226407487267, "learning_rate": 1.9960493295689877e-06, "loss": 0.2504, "step": 40630 }, { "epoch": 11.02251152698671, "grad_norm": 22.50520845058223, "learning_rate": 1.9935271144754133e-06, "loss": 0.2966, "step": 40640 }, { "epoch": 11.025223759153784, "grad_norm": 11.947481514148837, "learning_rate": 1.991006097092555e-06, "loss": 0.2589, "step": 40650 }, { "epoch": 11.027935991320858, "grad_norm": 12.531839359680426, "learning_rate": 1.98848627842473e-06, "loss": 0.2355, "step": 40660 }, { "epoch": 11.03064822348793, "grad_norm": 12.697134991581175, "learning_rate": 1.985967659475776e-06, "loss": 0.2721, "step": 40670 }, { "epoch": 11.033360455655004, "grad_norm": 11.365502787350046, "learning_rate": 1.9834502412490564e-06, "loss": 0.2864, "step": 40680 }, { "epoch": 11.036072687822077, "grad_norm": 9.300827613275292, "learning_rate": 1.98093402474745e-06, "loss": 0.2455, "step": 40690 }, { "epoch": 11.038784919989151, "grad_norm": 12.4585941957385, "learning_rate": 1.978419010973362e-06, "loss": 0.2842, "step": 40700 }, { "epoch": 11.041497152156225, "grad_norm": 13.997326075815058, "learning_rate": 1.9759052009287172e-06, "loss": 0.3233, "step": 40710 }, { "epoch": 11.044209384323299, "grad_norm": 17.627580679180294, "learning_rate": 1.9733925956149603e-06, "loss": 0.3159, "step": 40720 }, { "epoch": 11.046921616490371, "grad_norm": 10.839812741767378, "learning_rate": 1.970881196033058e-06, "loss": 0.29, "step": 40730 }, { "epoch": 11.049633848657445, "grad_norm": 14.489867648882768, "learning_rate": 1.9683710031834902e-06, "loss": 0.3017, "step": 40740 }, { "epoch": 11.052346080824519, "grad_norm": 12.866910051895122, "learning_rate": 1.9658620180662674e-06, "loss": 0.279, "step": 40750 }, { "epoch": 11.055058312991592, "grad_norm": 12.557502560724243, "learning_rate": 1.963354241680908e-06, "loss": 0.2439, "step": 40760 }, { "epoch": 11.057770545158666, "grad_norm": 13.54109085915119, "learning_rate": 1.9608476750264556e-06, "loss": 0.2946, "step": 40770 }, { "epoch": 11.060482777325738, "grad_norm": 10.12397360661834, "learning_rate": 1.958342319101465e-06, "loss": 0.2654, "step": 40780 }, { "epoch": 11.063195009492812, "grad_norm": 14.156614864864693, "learning_rate": 1.955838174904019e-06, "loss": 0.2563, "step": 40790 }, { "epoch": 11.065907241659886, "grad_norm": 11.165495291737741, "learning_rate": 1.9533352434317116e-06, "loss": 0.2703, "step": 40800 }, { "epoch": 11.06861947382696, "grad_norm": 15.286271063643618, "learning_rate": 1.95083352568165e-06, "loss": 0.2873, "step": 40810 }, { "epoch": 11.071331705994034, "grad_norm": 16.89522519407749, "learning_rate": 1.9483330226504643e-06, "loss": 0.3028, "step": 40820 }, { "epoch": 11.074043938161106, "grad_norm": 14.73809417736403, "learning_rate": 1.945833735334298e-06, "loss": 0.3016, "step": 40830 }, { "epoch": 11.07675617032818, "grad_norm": 14.546717681220947, "learning_rate": 1.943335664728812e-06, "loss": 0.2756, "step": 40840 }, { "epoch": 11.079468402495253, "grad_norm": 11.486550936433243, "learning_rate": 1.9408388118291783e-06, "loss": 0.2802, "step": 40850 }, { "epoch": 11.082180634662327, "grad_norm": 13.945413127388518, "learning_rate": 1.9383431776300866e-06, "loss": 0.2861, "step": 40860 }, { "epoch": 11.084892866829401, "grad_norm": 12.557425997976454, "learning_rate": 1.935848763125745e-06, "loss": 0.2697, "step": 40870 }, { "epoch": 11.087605098996475, "grad_norm": 10.797326307072463, "learning_rate": 1.9333555693098684e-06, "loss": 0.2543, "step": 40880 }, { "epoch": 11.090317331163547, "grad_norm": 11.591664077863586, "learning_rate": 1.93086359717569e-06, "loss": 0.2573, "step": 40890 }, { "epoch": 11.09302956333062, "grad_norm": 13.813547119381596, "learning_rate": 1.928372847715952e-06, "loss": 0.2522, "step": 40900 }, { "epoch": 11.095741795497695, "grad_norm": 17.056997847301222, "learning_rate": 1.9258833219229182e-06, "loss": 0.2719, "step": 40910 }, { "epoch": 11.098454027664769, "grad_norm": 11.671091714645966, "learning_rate": 1.923395020788355e-06, "loss": 0.2637, "step": 40920 }, { "epoch": 11.101166259831842, "grad_norm": 11.01493415915117, "learning_rate": 1.920907945303546e-06, "loss": 0.2798, "step": 40930 }, { "epoch": 11.103878491998914, "grad_norm": 11.343530156072006, "learning_rate": 1.9184220964592865e-06, "loss": 0.2654, "step": 40940 }, { "epoch": 11.106590724165988, "grad_norm": 19.97465159015465, "learning_rate": 1.915937475245882e-06, "loss": 0.2862, "step": 40950 }, { "epoch": 11.109302956333062, "grad_norm": 13.476532335871255, "learning_rate": 1.913454082653151e-06, "loss": 0.2831, "step": 40960 }, { "epoch": 11.112015188500136, "grad_norm": 15.34830890181053, "learning_rate": 1.910971919670418e-06, "loss": 0.2629, "step": 40970 }, { "epoch": 11.11472742066721, "grad_norm": 13.052183769914462, "learning_rate": 1.9084909872865216e-06, "loss": 0.2875, "step": 40980 }, { "epoch": 11.117439652834282, "grad_norm": 11.574700445585759, "learning_rate": 1.9060112864898094e-06, "loss": 0.2942, "step": 40990 }, { "epoch": 11.120151885001356, "grad_norm": 12.696588446422629, "learning_rate": 1.903532818268138e-06, "loss": 0.2903, "step": 41000 }, { "epoch": 11.12286411716843, "grad_norm": 12.381750497270325, "learning_rate": 1.9010555836088746e-06, "loss": 0.2816, "step": 41010 }, { "epoch": 11.125576349335503, "grad_norm": 13.557485031502386, "learning_rate": 1.8985795834988885e-06, "loss": 0.2937, "step": 41020 }, { "epoch": 11.128288581502577, "grad_norm": 15.94493563852618, "learning_rate": 1.8961048189245685e-06, "loss": 0.2789, "step": 41030 }, { "epoch": 11.131000813669651, "grad_norm": 14.604742807977487, "learning_rate": 1.8936312908718002e-06, "loss": 0.2949, "step": 41040 }, { "epoch": 11.133713045836723, "grad_norm": 10.991368186565031, "learning_rate": 1.8911590003259822e-06, "loss": 0.2601, "step": 41050 }, { "epoch": 11.136425278003797, "grad_norm": 13.096580134426638, "learning_rate": 1.8886879482720199e-06, "loss": 0.272, "step": 41060 }, { "epoch": 11.13913751017087, "grad_norm": 14.539877759726105, "learning_rate": 1.8862181356943238e-06, "loss": 0.2541, "step": 41070 }, { "epoch": 11.141849742337945, "grad_norm": 13.593880071753537, "learning_rate": 1.8837495635768133e-06, "loss": 0.3076, "step": 41080 }, { "epoch": 11.144561974505018, "grad_norm": 14.736017113862637, "learning_rate": 1.881282232902908e-06, "loss": 0.308, "step": 41090 }, { "epoch": 11.14727420667209, "grad_norm": 11.46287697947656, "learning_rate": 1.8788161446555387e-06, "loss": 0.3213, "step": 41100 }, { "epoch": 11.149986438839164, "grad_norm": 15.141034864316868, "learning_rate": 1.8763512998171384e-06, "loss": 0.2839, "step": 41110 }, { "epoch": 11.152698671006238, "grad_norm": 14.807338240923228, "learning_rate": 1.8738876993696464e-06, "loss": 0.3113, "step": 41120 }, { "epoch": 11.155410903173312, "grad_norm": 12.115938372661265, "learning_rate": 1.8714253442945064e-06, "loss": 0.2803, "step": 41130 }, { "epoch": 11.158123135340386, "grad_norm": 14.219548671489818, "learning_rate": 1.8689642355726601e-06, "loss": 0.2549, "step": 41140 }, { "epoch": 11.160835367507458, "grad_norm": 9.697963308163647, "learning_rate": 1.8665043741845634e-06, "loss": 0.2679, "step": 41150 }, { "epoch": 11.163547599674532, "grad_norm": 15.137402795039979, "learning_rate": 1.8640457611101659e-06, "loss": 0.2864, "step": 41160 }, { "epoch": 11.166259831841606, "grad_norm": 11.94269482338191, "learning_rate": 1.861588397328924e-06, "loss": 0.2546, "step": 41170 }, { "epoch": 11.16897206400868, "grad_norm": 14.679822682674427, "learning_rate": 1.8591322838197956e-06, "loss": 0.2952, "step": 41180 }, { "epoch": 11.171684296175753, "grad_norm": 14.541736118630293, "learning_rate": 1.856677421561241e-06, "loss": 0.2732, "step": 41190 }, { "epoch": 11.174396528342825, "grad_norm": 15.110574086897158, "learning_rate": 1.854223811531224e-06, "loss": 0.2629, "step": 41200 }, { "epoch": 11.177108760509899, "grad_norm": 12.587024822385164, "learning_rate": 1.851771454707203e-06, "loss": 0.3055, "step": 41210 }, { "epoch": 11.179820992676973, "grad_norm": 13.370271401587244, "learning_rate": 1.8493203520661435e-06, "loss": 0.2824, "step": 41220 }, { "epoch": 11.182533224844047, "grad_norm": 13.181094299843286, "learning_rate": 1.8468705045845097e-06, "loss": 0.2639, "step": 41230 }, { "epoch": 11.18524545701112, "grad_norm": 14.10949980859087, "learning_rate": 1.8444219132382645e-06, "loss": 0.2962, "step": 41240 }, { "epoch": 11.187957689178194, "grad_norm": 12.939878960097499, "learning_rate": 1.8419745790028737e-06, "loss": 0.275, "step": 41250 }, { "epoch": 11.190669921345267, "grad_norm": 15.023330893304015, "learning_rate": 1.8395285028532944e-06, "loss": 0.2656, "step": 41260 }, { "epoch": 11.19338215351234, "grad_norm": 11.606179339019825, "learning_rate": 1.8370836857639945e-06, "loss": 0.2736, "step": 41270 }, { "epoch": 11.196094385679414, "grad_norm": 14.66690767834505, "learning_rate": 1.8346401287089294e-06, "loss": 0.3168, "step": 41280 }, { "epoch": 11.198806617846488, "grad_norm": 11.213771275077898, "learning_rate": 1.8321978326615575e-06, "loss": 0.2429, "step": 41290 }, { "epoch": 11.201518850013562, "grad_norm": 9.626057918912021, "learning_rate": 1.8297567985948355e-06, "loss": 0.2945, "step": 41300 }, { "epoch": 11.204231082180634, "grad_norm": 13.441711282471944, "learning_rate": 1.8273170274812152e-06, "loss": 0.2854, "step": 41310 }, { "epoch": 11.206943314347708, "grad_norm": 19.858948677669193, "learning_rate": 1.8248785202926483e-06, "loss": 0.3095, "step": 41320 }, { "epoch": 11.209655546514782, "grad_norm": 11.655781068310057, "learning_rate": 1.8224412780005768e-06, "loss": 0.2881, "step": 41330 }, { "epoch": 11.212367778681855, "grad_norm": 13.207212909431405, "learning_rate": 1.8200053015759456e-06, "loss": 0.243, "step": 41340 }, { "epoch": 11.21508001084893, "grad_norm": 16.476775996919507, "learning_rate": 1.8175705919891917e-06, "loss": 0.2451, "step": 41350 }, { "epoch": 11.217792243016001, "grad_norm": 15.096333665337257, "learning_rate": 1.815137150210249e-06, "loss": 0.3274, "step": 41360 }, { "epoch": 11.220504475183075, "grad_norm": 15.12566179565228, "learning_rate": 1.8127049772085458e-06, "loss": 0.3194, "step": 41370 }, { "epoch": 11.223216707350149, "grad_norm": 17.03592318654942, "learning_rate": 1.8102740739530018e-06, "loss": 0.2671, "step": 41380 }, { "epoch": 11.225928939517223, "grad_norm": 14.821543554374703, "learning_rate": 1.8078444414120388e-06, "loss": 0.2767, "step": 41390 }, { "epoch": 11.228641171684297, "grad_norm": 18.33740657646845, "learning_rate": 1.8054160805535631e-06, "loss": 0.2987, "step": 41400 }, { "epoch": 11.23135340385137, "grad_norm": 15.977011416803169, "learning_rate": 1.8029889923449807e-06, "loss": 0.2875, "step": 41410 }, { "epoch": 11.234065636018443, "grad_norm": 12.776975052706518, "learning_rate": 1.8005631777531873e-06, "loss": 0.2492, "step": 41420 }, { "epoch": 11.236777868185516, "grad_norm": 16.9273873543578, "learning_rate": 1.7981386377445742e-06, "loss": 0.2844, "step": 41430 }, { "epoch": 11.23949010035259, "grad_norm": 12.480675072662649, "learning_rate": 1.7957153732850236e-06, "loss": 0.2536, "step": 41440 }, { "epoch": 11.242202332519664, "grad_norm": 12.307007484022531, "learning_rate": 1.7932933853399066e-06, "loss": 0.2611, "step": 41450 }, { "epoch": 11.244914564686738, "grad_norm": 17.25577472998867, "learning_rate": 1.7908726748740896e-06, "loss": 0.2554, "step": 41460 }, { "epoch": 11.24762679685381, "grad_norm": 12.49623773735364, "learning_rate": 1.7884532428519285e-06, "loss": 0.2862, "step": 41470 }, { "epoch": 11.250339029020884, "grad_norm": 16.127474285045675, "learning_rate": 1.7860350902372708e-06, "loss": 0.2847, "step": 41480 }, { "epoch": 11.253051261187958, "grad_norm": 11.436230859518263, "learning_rate": 1.7836182179934558e-06, "loss": 0.2952, "step": 41490 }, { "epoch": 11.255763493355031, "grad_norm": 12.035340186596846, "learning_rate": 1.781202627083305e-06, "loss": 0.2834, "step": 41500 }, { "epoch": 11.258475725522105, "grad_norm": 9.96999248629861, "learning_rate": 1.778788318469142e-06, "loss": 0.2773, "step": 41510 }, { "epoch": 11.261187957689177, "grad_norm": 13.205204182055505, "learning_rate": 1.776375293112767e-06, "loss": 0.2883, "step": 41520 }, { "epoch": 11.263900189856251, "grad_norm": 13.172136122269553, "learning_rate": 1.7739635519754778e-06, "loss": 0.2802, "step": 41530 }, { "epoch": 11.266612422023325, "grad_norm": 11.956873929052016, "learning_rate": 1.771553096018056e-06, "loss": 0.3008, "step": 41540 }, { "epoch": 11.269324654190399, "grad_norm": 12.889475152379493, "learning_rate": 1.769143926200773e-06, "loss": 0.2467, "step": 41550 }, { "epoch": 11.272036886357473, "grad_norm": 11.871997536953566, "learning_rate": 1.7667360434833902e-06, "loss": 0.2916, "step": 41560 }, { "epoch": 11.274749118524547, "grad_norm": 17.383243668999327, "learning_rate": 1.7643294488251495e-06, "loss": 0.2634, "step": 41570 }, { "epoch": 11.277461350691619, "grad_norm": 13.993135020080201, "learning_rate": 1.761924143184785e-06, "loss": 0.2727, "step": 41580 }, { "epoch": 11.280173582858692, "grad_norm": 13.343210859572263, "learning_rate": 1.7595201275205158e-06, "loss": 0.2933, "step": 41590 }, { "epoch": 11.282885815025766, "grad_norm": 15.08527801175883, "learning_rate": 1.757117402790049e-06, "loss": 0.2696, "step": 41600 }, { "epoch": 11.28559804719284, "grad_norm": 12.305255602059425, "learning_rate": 1.7547159699505757e-06, "loss": 0.299, "step": 41610 }, { "epoch": 11.288310279359914, "grad_norm": 11.80040349261763, "learning_rate": 1.7523158299587679e-06, "loss": 0.2471, "step": 41620 }, { "epoch": 11.291022511526986, "grad_norm": 15.151663210731567, "learning_rate": 1.749916983770793e-06, "loss": 0.259, "step": 41630 }, { "epoch": 11.29373474369406, "grad_norm": 16.924071477967466, "learning_rate": 1.7475194323422934e-06, "loss": 0.2441, "step": 41640 }, { "epoch": 11.296446975861134, "grad_norm": 15.713769120359979, "learning_rate": 1.7451231766284004e-06, "loss": 0.2851, "step": 41650 }, { "epoch": 11.299159208028208, "grad_norm": 10.427194111108184, "learning_rate": 1.7427282175837273e-06, "loss": 0.2689, "step": 41660 }, { "epoch": 11.301871440195281, "grad_norm": 10.930615718400444, "learning_rate": 1.7403345561623725e-06, "loss": 0.2621, "step": 41670 }, { "epoch": 11.304583672362353, "grad_norm": 13.850265367748115, "learning_rate": 1.7379421933179181e-06, "loss": 0.2742, "step": 41680 }, { "epoch": 11.307295904529427, "grad_norm": 15.694486309651488, "learning_rate": 1.7355511300034228e-06, "loss": 0.3121, "step": 41690 }, { "epoch": 11.310008136696501, "grad_norm": 9.109042811155314, "learning_rate": 1.7331613671714354e-06, "loss": 0.2833, "step": 41700 }, { "epoch": 11.312720368863575, "grad_norm": 11.904198919127415, "learning_rate": 1.730772905773982e-06, "loss": 0.3029, "step": 41710 }, { "epoch": 11.315432601030649, "grad_norm": 15.17453657322871, "learning_rate": 1.728385746762572e-06, "loss": 0.2915, "step": 41720 }, { "epoch": 11.318144833197723, "grad_norm": 15.281442605557144, "learning_rate": 1.725999891088197e-06, "loss": 0.3227, "step": 41730 }, { "epoch": 11.320857065364795, "grad_norm": 15.096083931303017, "learning_rate": 1.7236153397013233e-06, "loss": 0.293, "step": 41740 }, { "epoch": 11.323569297531868, "grad_norm": 13.283493790578794, "learning_rate": 1.7212320935519089e-06, "loss": 0.3068, "step": 41750 }, { "epoch": 11.326281529698942, "grad_norm": 14.454883051547155, "learning_rate": 1.7188501535893792e-06, "loss": 0.2758, "step": 41760 }, { "epoch": 11.328993761866016, "grad_norm": 9.510933416375373, "learning_rate": 1.7164695207626498e-06, "loss": 0.2547, "step": 41770 }, { "epoch": 11.33170599403309, "grad_norm": 10.526309153131974, "learning_rate": 1.7140901960201046e-06, "loss": 0.2546, "step": 41780 }, { "epoch": 11.334418226200162, "grad_norm": 17.695642449235326, "learning_rate": 1.7117121803096183e-06, "loss": 0.3525, "step": 41790 }, { "epoch": 11.337130458367236, "grad_norm": 11.33819782035238, "learning_rate": 1.7093354745785384e-06, "loss": 0.2859, "step": 41800 }, { "epoch": 11.33984269053431, "grad_norm": 13.315154098222424, "learning_rate": 1.706960079773688e-06, "loss": 0.3068, "step": 41810 }, { "epoch": 11.342554922701384, "grad_norm": 14.04882877463476, "learning_rate": 1.704585996841371e-06, "loss": 0.2891, "step": 41820 }, { "epoch": 11.345267154868457, "grad_norm": 14.203003761946661, "learning_rate": 1.7022132267273689e-06, "loss": 0.265, "step": 41830 }, { "epoch": 11.34797938703553, "grad_norm": 14.766664328896736, "learning_rate": 1.6998417703769414e-06, "loss": 0.2765, "step": 41840 }, { "epoch": 11.350691619202603, "grad_norm": 14.843658483995865, "learning_rate": 1.6974716287348198e-06, "loss": 0.2355, "step": 41850 }, { "epoch": 11.353403851369677, "grad_norm": 13.435703110608408, "learning_rate": 1.6951028027452148e-06, "loss": 0.2657, "step": 41860 }, { "epoch": 11.356116083536751, "grad_norm": 13.656079362395602, "learning_rate": 1.6927352933518176e-06, "loss": 0.2789, "step": 41870 }, { "epoch": 11.358828315703825, "grad_norm": 16.22823847788136, "learning_rate": 1.6903691014977852e-06, "loss": 0.3116, "step": 41880 }, { "epoch": 11.361540547870899, "grad_norm": 15.588956181109866, "learning_rate": 1.688004228125759e-06, "loss": 0.2926, "step": 41890 }, { "epoch": 11.36425278003797, "grad_norm": 10.779196613174534, "learning_rate": 1.6856406741778453e-06, "loss": 0.2825, "step": 41900 }, { "epoch": 11.366965012205045, "grad_norm": 10.446017337505232, "learning_rate": 1.6832784405956376e-06, "loss": 0.3178, "step": 41910 }, { "epoch": 11.369677244372118, "grad_norm": 17.525270778851784, "learning_rate": 1.680917528320191e-06, "loss": 0.2963, "step": 41920 }, { "epoch": 11.372389476539192, "grad_norm": 16.49987714397211, "learning_rate": 1.6785579382920413e-06, "loss": 0.295, "step": 41930 }, { "epoch": 11.375101708706266, "grad_norm": 11.911252281049808, "learning_rate": 1.6761996714511957e-06, "loss": 0.3061, "step": 41940 }, { "epoch": 11.377813940873338, "grad_norm": 12.685900450834852, "learning_rate": 1.6738427287371344e-06, "loss": 0.3133, "step": 41950 }, { "epoch": 11.380526173040412, "grad_norm": 12.459476547401216, "learning_rate": 1.671487111088811e-06, "loss": 0.2886, "step": 41960 }, { "epoch": 11.383238405207486, "grad_norm": 13.630236999303635, "learning_rate": 1.6691328194446476e-06, "loss": 0.3018, "step": 41970 }, { "epoch": 11.38595063737456, "grad_norm": 10.465228761432874, "learning_rate": 1.6667798547425428e-06, "loss": 0.2842, "step": 41980 }, { "epoch": 11.388662869541633, "grad_norm": 14.013549177021511, "learning_rate": 1.6644282179198635e-06, "loss": 0.3484, "step": 41990 }, { "epoch": 11.391375101708705, "grad_norm": 12.41879271169404, "learning_rate": 1.6620779099134481e-06, "loss": 0.3284, "step": 42000 }, { "epoch": 11.39408733387578, "grad_norm": 15.327004436770496, "learning_rate": 1.6597289316596093e-06, "loss": 0.2753, "step": 42010 }, { "epoch": 11.396799566042853, "grad_norm": 14.387467702680016, "learning_rate": 1.6573812840941211e-06, "loss": 0.299, "step": 42020 }, { "epoch": 11.399511798209927, "grad_norm": 18.995209535921074, "learning_rate": 1.6550349681522398e-06, "loss": 0.3049, "step": 42030 }, { "epoch": 11.402224030377, "grad_norm": 10.328946297965325, "learning_rate": 1.6526899847686794e-06, "loss": 0.288, "step": 42040 }, { "epoch": 11.404936262544073, "grad_norm": 13.930322720766327, "learning_rate": 1.65034633487763e-06, "loss": 0.2594, "step": 42050 }, { "epoch": 11.407648494711147, "grad_norm": 20.862176594753898, "learning_rate": 1.6480040194127484e-06, "loss": 0.2897, "step": 42060 }, { "epoch": 11.41036072687822, "grad_norm": 14.42324214101958, "learning_rate": 1.6456630393071605e-06, "loss": 0.2672, "step": 42070 }, { "epoch": 11.413072959045294, "grad_norm": 26.98706954042835, "learning_rate": 1.6433233954934613e-06, "loss": 0.2963, "step": 42080 }, { "epoch": 11.415785191212368, "grad_norm": 15.672924209824075, "learning_rate": 1.6409850889037083e-06, "loss": 0.2877, "step": 42090 }, { "epoch": 11.418497423379442, "grad_norm": 13.411786267236424, "learning_rate": 1.6386481204694315e-06, "loss": 0.301, "step": 42100 }, { "epoch": 11.421209655546514, "grad_norm": 13.754244518674815, "learning_rate": 1.6363124911216265e-06, "loss": 0.3355, "step": 42110 }, { "epoch": 11.423921887713588, "grad_norm": 16.06889622197055, "learning_rate": 1.6339782017907558e-06, "loss": 0.2719, "step": 42120 }, { "epoch": 11.426634119880662, "grad_norm": 12.800563451447589, "learning_rate": 1.6316452534067472e-06, "loss": 0.2858, "step": 42130 }, { "epoch": 11.429346352047736, "grad_norm": 11.016107414868742, "learning_rate": 1.6293136468989918e-06, "loss": 0.2614, "step": 42140 }, { "epoch": 11.43205858421481, "grad_norm": 13.523156569250533, "learning_rate": 1.6269833831963533e-06, "loss": 0.2881, "step": 42150 }, { "epoch": 11.434770816381882, "grad_norm": 16.593160490609527, "learning_rate": 1.6246544632271526e-06, "loss": 0.2821, "step": 42160 }, { "epoch": 11.437483048548955, "grad_norm": 14.376848534121876, "learning_rate": 1.62232688791918e-06, "loss": 0.2733, "step": 42170 }, { "epoch": 11.44019528071603, "grad_norm": 11.663286649804546, "learning_rate": 1.620000658199688e-06, "loss": 0.251, "step": 42180 }, { "epoch": 11.442907512883103, "grad_norm": 12.486981572841483, "learning_rate": 1.6176757749953948e-06, "loss": 0.2715, "step": 42190 }, { "epoch": 11.445619745050177, "grad_norm": 12.78941796331161, "learning_rate": 1.6153522392324827e-06, "loss": 0.2921, "step": 42200 }, { "epoch": 11.44833197721725, "grad_norm": 17.100017298866863, "learning_rate": 1.6130300518365915e-06, "loss": 0.3365, "step": 42210 }, { "epoch": 11.451044209384323, "grad_norm": 13.68520424247513, "learning_rate": 1.6107092137328313e-06, "loss": 0.2867, "step": 42220 }, { "epoch": 11.453756441551397, "grad_norm": 13.789923701969583, "learning_rate": 1.6083897258457703e-06, "loss": 0.2526, "step": 42230 }, { "epoch": 11.45646867371847, "grad_norm": 15.58820593736075, "learning_rate": 1.60607158909944e-06, "loss": 0.2877, "step": 42240 }, { "epoch": 11.459180905885544, "grad_norm": 13.959960577917453, "learning_rate": 1.6037548044173356e-06, "loss": 0.2715, "step": 42250 }, { "epoch": 11.461893138052618, "grad_norm": 18.528722435472723, "learning_rate": 1.601439372722407e-06, "loss": 0.296, "step": 42260 }, { "epoch": 11.46460537021969, "grad_norm": 14.220524387907417, "learning_rate": 1.5991252949370756e-06, "loss": 0.2816, "step": 42270 }, { "epoch": 11.467317602386764, "grad_norm": 13.829468466093498, "learning_rate": 1.5968125719832128e-06, "loss": 0.3247, "step": 42280 }, { "epoch": 11.470029834553838, "grad_norm": 14.397749997608672, "learning_rate": 1.5945012047821573e-06, "loss": 0.2898, "step": 42290 }, { "epoch": 11.472742066720912, "grad_norm": 16.939608259172548, "learning_rate": 1.5921911942547047e-06, "loss": 0.3155, "step": 42300 }, { "epoch": 11.475454298887986, "grad_norm": 14.678564358133054, "learning_rate": 1.5898825413211116e-06, "loss": 0.2626, "step": 42310 }, { "epoch": 11.478166531055058, "grad_norm": 13.25811112571128, "learning_rate": 1.587575246901094e-06, "loss": 0.2457, "step": 42320 }, { "epoch": 11.480878763222131, "grad_norm": 15.913306149537618, "learning_rate": 1.585269311913823e-06, "loss": 0.3119, "step": 42330 }, { "epoch": 11.483590995389205, "grad_norm": 11.363036560167464, "learning_rate": 1.5829647372779323e-06, "loss": 0.279, "step": 42340 }, { "epoch": 11.486303227556279, "grad_norm": 11.74142436977856, "learning_rate": 1.5806615239115119e-06, "loss": 0.3059, "step": 42350 }, { "epoch": 11.489015459723353, "grad_norm": 12.124928750518245, "learning_rate": 1.5783596727321105e-06, "loss": 0.284, "step": 42360 }, { "epoch": 11.491727691890425, "grad_norm": 14.699509931537483, "learning_rate": 1.5760591846567347e-06, "loss": 0.2277, "step": 42370 }, { "epoch": 11.494439924057499, "grad_norm": 14.875358252209416, "learning_rate": 1.5737600606018421e-06, "loss": 0.3213, "step": 42380 }, { "epoch": 11.497152156224573, "grad_norm": 13.105727680219104, "learning_rate": 1.5714623014833586e-06, "loss": 0.2789, "step": 42390 }, { "epoch": 11.499864388391646, "grad_norm": 17.89615440056507, "learning_rate": 1.5691659082166538e-06, "loss": 0.2861, "step": 42400 }, { "epoch": 11.50257662055872, "grad_norm": 12.778990429196242, "learning_rate": 1.5668708817165601e-06, "loss": 0.2452, "step": 42410 }, { "epoch": 11.505288852725794, "grad_norm": 13.72803868155231, "learning_rate": 1.5645772228973655e-06, "loss": 0.2641, "step": 42420 }, { "epoch": 11.508001084892866, "grad_norm": 11.202388805237597, "learning_rate": 1.5622849326728106e-06, "loss": 0.3042, "step": 42430 }, { "epoch": 11.51071331705994, "grad_norm": 17.65265653789052, "learning_rate": 1.5599940119560936e-06, "loss": 0.2773, "step": 42440 }, { "epoch": 11.513425549227014, "grad_norm": 14.403801628984821, "learning_rate": 1.557704461659863e-06, "loss": 0.2904, "step": 42450 }, { "epoch": 11.516137781394088, "grad_norm": 22.07931512680624, "learning_rate": 1.5554162826962244e-06, "loss": 0.311, "step": 42460 }, { "epoch": 11.518850013561162, "grad_norm": 15.556659796211113, "learning_rate": 1.5531294759767369e-06, "loss": 0.2838, "step": 42470 }, { "epoch": 11.521562245728234, "grad_norm": 13.899181588273255, "learning_rate": 1.5508440424124126e-06, "loss": 0.279, "step": 42480 }, { "epoch": 11.524274477895307, "grad_norm": 14.593607745655513, "learning_rate": 1.5485599829137181e-06, "loss": 0.3288, "step": 42490 }, { "epoch": 11.526986710062381, "grad_norm": 15.581735746149901, "learning_rate": 1.546277298390566e-06, "loss": 0.2985, "step": 42500 }, { "epoch": 11.529698942229455, "grad_norm": 12.6371592049382, "learning_rate": 1.5439959897523333e-06, "loss": 0.2816, "step": 42510 }, { "epoch": 11.532411174396529, "grad_norm": 14.541523718135862, "learning_rate": 1.5417160579078354e-06, "loss": 0.2729, "step": 42520 }, { "epoch": 11.535123406563601, "grad_norm": 11.549007607156762, "learning_rate": 1.539437503765348e-06, "loss": 0.3011, "step": 42530 }, { "epoch": 11.537835638730675, "grad_norm": 15.20204400533371, "learning_rate": 1.5371603282325958e-06, "loss": 0.3014, "step": 42540 }, { "epoch": 11.540547870897749, "grad_norm": 14.155834041432076, "learning_rate": 1.5348845322167533e-06, "loss": 0.2679, "step": 42550 }, { "epoch": 11.543260103064823, "grad_norm": 17.06143894434815, "learning_rate": 1.5326101166244484e-06, "loss": 0.2729, "step": 42560 }, { "epoch": 11.545972335231896, "grad_norm": 14.958188517406969, "learning_rate": 1.530337082361753e-06, "loss": 0.2545, "step": 42570 }, { "epoch": 11.548684567398968, "grad_norm": 17.13881072850571, "learning_rate": 1.528065430334194e-06, "loss": 0.2725, "step": 42580 }, { "epoch": 11.551396799566042, "grad_norm": 15.100313537344762, "learning_rate": 1.5257951614467463e-06, "loss": 0.244, "step": 42590 }, { "epoch": 11.554109031733116, "grad_norm": 13.058445123384383, "learning_rate": 1.5235262766038338e-06, "loss": 0.2799, "step": 42600 }, { "epoch": 11.55682126390019, "grad_norm": 16.553102650268663, "learning_rate": 1.5212587767093307e-06, "loss": 0.2957, "step": 42610 }, { "epoch": 11.559533496067264, "grad_norm": 14.440500107382032, "learning_rate": 1.5189926626665519e-06, "loss": 0.2651, "step": 42620 }, { "epoch": 11.562245728234338, "grad_norm": 15.641334681466112, "learning_rate": 1.5167279353782733e-06, "loss": 0.278, "step": 42630 }, { "epoch": 11.56495796040141, "grad_norm": 14.30767382545366, "learning_rate": 1.5144645957467051e-06, "loss": 0.2947, "step": 42640 }, { "epoch": 11.567670192568484, "grad_norm": 11.929595198434505, "learning_rate": 1.5122026446735133e-06, "loss": 0.2948, "step": 42650 }, { "epoch": 11.570382424735557, "grad_norm": 9.033390885967757, "learning_rate": 1.5099420830598071e-06, "loss": 0.2588, "step": 42660 }, { "epoch": 11.573094656902631, "grad_norm": 15.734125942637158, "learning_rate": 1.5076829118061436e-06, "loss": 0.2864, "step": 42670 }, { "epoch": 11.575806889069705, "grad_norm": 13.40514827570904, "learning_rate": 1.5054251318125274e-06, "loss": 0.283, "step": 42680 }, { "epoch": 11.578519121236777, "grad_norm": 11.25212889964365, "learning_rate": 1.5031687439784025e-06, "loss": 0.2732, "step": 42690 }, { "epoch": 11.581231353403851, "grad_norm": 10.734317404412188, "learning_rate": 1.5009137492026654e-06, "loss": 0.2552, "step": 42700 }, { "epoch": 11.583943585570925, "grad_norm": 15.663355499685894, "learning_rate": 1.4986601483836543e-06, "loss": 0.3036, "step": 42710 }, { "epoch": 11.586655817737999, "grad_norm": 16.034502210964256, "learning_rate": 1.496407942419153e-06, "loss": 0.3029, "step": 42720 }, { "epoch": 11.589368049905072, "grad_norm": 20.079049612136444, "learning_rate": 1.4941571322063914e-06, "loss": 0.2876, "step": 42730 }, { "epoch": 11.592080282072146, "grad_norm": 9.212472789076207, "learning_rate": 1.4919077186420356e-06, "loss": 0.2969, "step": 42740 }, { "epoch": 11.594792514239218, "grad_norm": 13.64802610191178, "learning_rate": 1.4896597026222082e-06, "loss": 0.2846, "step": 42750 }, { "epoch": 11.597504746406292, "grad_norm": 13.667138383620062, "learning_rate": 1.4874130850424623e-06, "loss": 0.2807, "step": 42760 }, { "epoch": 11.600216978573366, "grad_norm": 15.335138144666768, "learning_rate": 1.4851678667978037e-06, "loss": 0.3149, "step": 42770 }, { "epoch": 11.60292921074044, "grad_norm": 12.872481381913127, "learning_rate": 1.4829240487826712e-06, "loss": 0.2859, "step": 42780 }, { "epoch": 11.605641442907514, "grad_norm": 14.28431202125344, "learning_rate": 1.4806816318909557e-06, "loss": 0.2881, "step": 42790 }, { "epoch": 11.608353675074586, "grad_norm": 20.25790602913667, "learning_rate": 1.4784406170159854e-06, "loss": 0.28, "step": 42800 }, { "epoch": 11.61106590724166, "grad_norm": 14.59415772750334, "learning_rate": 1.4762010050505265e-06, "loss": 0.2996, "step": 42810 }, { "epoch": 11.613778139408733, "grad_norm": 15.766782473849076, "learning_rate": 1.473962796886792e-06, "loss": 0.309, "step": 42820 }, { "epoch": 11.616490371575807, "grad_norm": 14.464494665796225, "learning_rate": 1.4717259934164325e-06, "loss": 0.282, "step": 42830 }, { "epoch": 11.619202603742881, "grad_norm": 13.486889829144078, "learning_rate": 1.4694905955305422e-06, "loss": 0.2943, "step": 42840 }, { "epoch": 11.621914835909953, "grad_norm": 12.644090106690587, "learning_rate": 1.467256604119649e-06, "loss": 0.3072, "step": 42850 }, { "epoch": 11.624627068077027, "grad_norm": 14.713537992452025, "learning_rate": 1.4650240200737254e-06, "loss": 0.286, "step": 42860 }, { "epoch": 11.6273393002441, "grad_norm": 16.716656595939746, "learning_rate": 1.4627928442821866e-06, "loss": 0.3066, "step": 42870 }, { "epoch": 11.630051532411175, "grad_norm": 14.614057820061147, "learning_rate": 1.4605630776338776e-06, "loss": 0.306, "step": 42880 }, { "epoch": 11.632763764578248, "grad_norm": 15.695340216087935, "learning_rate": 1.45833472101709e-06, "loss": 0.275, "step": 42890 }, { "epoch": 11.63547599674532, "grad_norm": 15.128016766619476, "learning_rate": 1.4561077753195462e-06, "loss": 0.2745, "step": 42900 }, { "epoch": 11.638188228912394, "grad_norm": 16.487771490803418, "learning_rate": 1.4538822414284176e-06, "loss": 0.2937, "step": 42910 }, { "epoch": 11.640900461079468, "grad_norm": 22.289051655982423, "learning_rate": 1.4516581202303014e-06, "loss": 0.3057, "step": 42920 }, { "epoch": 11.643612693246542, "grad_norm": 14.497586904306047, "learning_rate": 1.4494354126112386e-06, "loss": 0.2703, "step": 42930 }, { "epoch": 11.646324925413616, "grad_norm": 17.82327258394486, "learning_rate": 1.4472141194567053e-06, "loss": 0.316, "step": 42940 }, { "epoch": 11.64903715758069, "grad_norm": 12.90104684774581, "learning_rate": 1.444994241651615e-06, "loss": 0.281, "step": 42950 }, { "epoch": 11.651749389747762, "grad_norm": 13.469943441831735, "learning_rate": 1.4427757800803182e-06, "loss": 0.3141, "step": 42960 }, { "epoch": 11.654461621914836, "grad_norm": 16.250666152141186, "learning_rate": 1.4405587356265966e-06, "loss": 0.2721, "step": 42970 }, { "epoch": 11.65717385408191, "grad_norm": 14.789063485459323, "learning_rate": 1.4383431091736717e-06, "loss": 0.2661, "step": 42980 }, { "epoch": 11.659886086248983, "grad_norm": 18.556635235463425, "learning_rate": 1.436128901604199e-06, "loss": 0.3167, "step": 42990 }, { "epoch": 11.662598318416057, "grad_norm": 16.569086137565318, "learning_rate": 1.433916113800269e-06, "loss": 0.3233, "step": 43000 }, { "epoch": 11.66531055058313, "grad_norm": 12.988325389890576, "learning_rate": 1.4317047466434075e-06, "loss": 0.2949, "step": 43010 }, { "epoch": 11.668022782750203, "grad_norm": 12.865731292923902, "learning_rate": 1.4294948010145687e-06, "loss": 0.2875, "step": 43020 }, { "epoch": 11.670735014917277, "grad_norm": 13.10589105088249, "learning_rate": 1.427286277794151e-06, "loss": 0.2757, "step": 43030 }, { "epoch": 11.67344724708435, "grad_norm": 13.638078649662242, "learning_rate": 1.4250791778619755e-06, "loss": 0.2533, "step": 43040 }, { "epoch": 11.676159479251425, "grad_norm": 13.505218471665746, "learning_rate": 1.4228735020973023e-06, "loss": 0.2965, "step": 43050 }, { "epoch": 11.678871711418498, "grad_norm": 13.053897590990715, "learning_rate": 1.4206692513788223e-06, "loss": 0.2627, "step": 43060 }, { "epoch": 11.68158394358557, "grad_norm": 13.425846339525481, "learning_rate": 1.4184664265846592e-06, "loss": 0.2787, "step": 43070 }, { "epoch": 11.684296175752644, "grad_norm": 11.757533725473266, "learning_rate": 1.4162650285923707e-06, "loss": 0.291, "step": 43080 }, { "epoch": 11.687008407919718, "grad_norm": 13.965649402434822, "learning_rate": 1.4140650582789407e-06, "loss": 0.2503, "step": 43090 }, { "epoch": 11.689720640086792, "grad_norm": 16.57642503976932, "learning_rate": 1.411866516520788e-06, "loss": 0.2976, "step": 43100 }, { "epoch": 11.692432872253866, "grad_norm": 14.451818649426318, "learning_rate": 1.409669404193763e-06, "loss": 0.2499, "step": 43110 }, { "epoch": 11.695145104420938, "grad_norm": 15.267865806060309, "learning_rate": 1.407473722173145e-06, "loss": 0.2531, "step": 43120 }, { "epoch": 11.697857336588012, "grad_norm": 14.997885276130562, "learning_rate": 1.405279471333646e-06, "loss": 0.288, "step": 43130 }, { "epoch": 11.700569568755085, "grad_norm": 14.006385332169497, "learning_rate": 1.4030866525494002e-06, "loss": 0.2857, "step": 43140 }, { "epoch": 11.70328180092216, "grad_norm": 12.626497453967492, "learning_rate": 1.4008952666939847e-06, "loss": 0.2969, "step": 43150 }, { "epoch": 11.705994033089233, "grad_norm": 12.364393216798886, "learning_rate": 1.3987053146403917e-06, "loss": 0.287, "step": 43160 }, { "epoch": 11.708706265256305, "grad_norm": 16.018772579046747, "learning_rate": 1.396516797261051e-06, "loss": 0.2956, "step": 43170 }, { "epoch": 11.711418497423379, "grad_norm": 17.352669197213107, "learning_rate": 1.3943297154278184e-06, "loss": 0.2908, "step": 43180 }, { "epoch": 11.714130729590453, "grad_norm": 19.264272912290345, "learning_rate": 1.3921440700119776e-06, "loss": 0.303, "step": 43190 }, { "epoch": 11.716842961757527, "grad_norm": 16.41964560588941, "learning_rate": 1.3899598618842414e-06, "loss": 0.2946, "step": 43200 }, { "epoch": 11.7195551939246, "grad_norm": 15.698815117332812, "learning_rate": 1.3877770919147465e-06, "loss": 0.2946, "step": 43210 }, { "epoch": 11.722267426091673, "grad_norm": 13.611058615843618, "learning_rate": 1.3855957609730598e-06, "loss": 0.2755, "step": 43220 }, { "epoch": 11.724979658258746, "grad_norm": 15.05928042331593, "learning_rate": 1.3834158699281747e-06, "loss": 0.2756, "step": 43230 }, { "epoch": 11.72769189042582, "grad_norm": 16.547285293162716, "learning_rate": 1.3812374196485106e-06, "loss": 0.288, "step": 43240 }, { "epoch": 11.730404122592894, "grad_norm": 13.369943260991313, "learning_rate": 1.3790604110019134e-06, "loss": 0.2959, "step": 43250 }, { "epoch": 11.733116354759968, "grad_norm": 16.964842164597332, "learning_rate": 1.3768848448556504e-06, "loss": 0.3173, "step": 43260 }, { "epoch": 11.735828586927042, "grad_norm": 13.507797148550797, "learning_rate": 1.3747107220764232e-06, "loss": 0.2805, "step": 43270 }, { "epoch": 11.738540819094114, "grad_norm": 15.083547028437154, "learning_rate": 1.3725380435303498e-06, "loss": 0.3148, "step": 43280 }, { "epoch": 11.741253051261188, "grad_norm": 9.900001457053584, "learning_rate": 1.370366810082976e-06, "loss": 0.2672, "step": 43290 }, { "epoch": 11.743965283428262, "grad_norm": 12.754573218934825, "learning_rate": 1.368197022599273e-06, "loss": 0.2696, "step": 43300 }, { "epoch": 11.746677515595335, "grad_norm": 15.277789646245655, "learning_rate": 1.3660286819436346e-06, "loss": 0.3069, "step": 43310 }, { "epoch": 11.74938974776241, "grad_norm": 12.875918374284195, "learning_rate": 1.363861788979881e-06, "loss": 0.3033, "step": 43320 }, { "epoch": 11.752101979929481, "grad_norm": 13.5186573079034, "learning_rate": 1.3616963445712489e-06, "loss": 0.2973, "step": 43330 }, { "epoch": 11.754814212096555, "grad_norm": 10.985195151420779, "learning_rate": 1.3595323495804052e-06, "loss": 0.2808, "step": 43340 }, { "epoch": 11.757526444263629, "grad_norm": 12.4113488347914, "learning_rate": 1.3573698048694351e-06, "loss": 0.3284, "step": 43350 }, { "epoch": 11.760238676430703, "grad_norm": 19.517454675392738, "learning_rate": 1.3552087112998485e-06, "loss": 0.3048, "step": 43360 }, { "epoch": 11.762950908597777, "grad_norm": 11.960447122052777, "learning_rate": 1.3530490697325776e-06, "loss": 0.2663, "step": 43370 }, { "epoch": 11.765663140764849, "grad_norm": 13.040531585028617, "learning_rate": 1.3508908810279698e-06, "loss": 0.2944, "step": 43380 }, { "epoch": 11.768375372931922, "grad_norm": 11.531932569652922, "learning_rate": 1.3487341460458047e-06, "loss": 0.2731, "step": 43390 }, { "epoch": 11.771087605098996, "grad_norm": 14.209476528202567, "learning_rate": 1.3465788656452723e-06, "loss": 0.2751, "step": 43400 }, { "epoch": 11.77379983726607, "grad_norm": 17.36197168339422, "learning_rate": 1.3444250406849885e-06, "loss": 0.2958, "step": 43410 }, { "epoch": 11.776512069433144, "grad_norm": 15.010292238347802, "learning_rate": 1.342272672022989e-06, "loss": 0.3091, "step": 43420 }, { "epoch": 11.779224301600216, "grad_norm": 12.718022339348027, "learning_rate": 1.3401217605167283e-06, "loss": 0.2737, "step": 43430 }, { "epoch": 11.78193653376729, "grad_norm": 12.698280203121143, "learning_rate": 1.3379723070230833e-06, "loss": 0.265, "step": 43440 }, { "epoch": 11.784648765934364, "grad_norm": 18.277461899800922, "learning_rate": 1.3358243123983434e-06, "loss": 0.2968, "step": 43450 }, { "epoch": 11.787360998101438, "grad_norm": 16.39886469136035, "learning_rate": 1.3336777774982229e-06, "loss": 0.3035, "step": 43460 }, { "epoch": 11.790073230268511, "grad_norm": 13.044814312403314, "learning_rate": 1.331532703177853e-06, "loss": 0.2936, "step": 43470 }, { "epoch": 11.792785462435585, "grad_norm": 14.92645241472099, "learning_rate": 1.3293890902917823e-06, "loss": 0.2876, "step": 43480 }, { "epoch": 11.795497694602657, "grad_norm": 18.471748971083834, "learning_rate": 1.327246939693979e-06, "loss": 0.2973, "step": 43490 }, { "epoch": 11.798209926769731, "grad_norm": 12.882122644048444, "learning_rate": 1.3251062522378238e-06, "loss": 0.2876, "step": 43500 }, { "epoch": 11.800922158936805, "grad_norm": 15.409171981355346, "learning_rate": 1.3229670287761237e-06, "loss": 0.2838, "step": 43510 }, { "epoch": 11.803634391103879, "grad_norm": 16.327918708994073, "learning_rate": 1.320829270161092e-06, "loss": 0.287, "step": 43520 }, { "epoch": 11.806346623270953, "grad_norm": 12.040860841395206, "learning_rate": 1.3186929772443652e-06, "loss": 0.2969, "step": 43530 }, { "epoch": 11.809058855438025, "grad_norm": 13.161098110238465, "learning_rate": 1.3165581508769936e-06, "loss": 0.29, "step": 43540 }, { "epoch": 11.811771087605099, "grad_norm": 15.931851804942339, "learning_rate": 1.314424791909444e-06, "loss": 0.3128, "step": 43550 }, { "epoch": 11.814483319772172, "grad_norm": 12.786596005067294, "learning_rate": 1.3122929011915996e-06, "loss": 0.2782, "step": 43560 }, { "epoch": 11.817195551939246, "grad_norm": 11.654484483071027, "learning_rate": 1.3101624795727547e-06, "loss": 0.283, "step": 43570 }, { "epoch": 11.81990778410632, "grad_norm": 16.045377654270244, "learning_rate": 1.3080335279016227e-06, "loss": 0.3135, "step": 43580 }, { "epoch": 11.822620016273394, "grad_norm": 14.736367074125559, "learning_rate": 1.305906047026329e-06, "loss": 0.2626, "step": 43590 }, { "epoch": 11.825332248440466, "grad_norm": 12.12480152785402, "learning_rate": 1.3037800377944142e-06, "loss": 0.307, "step": 43600 }, { "epoch": 11.82804448060754, "grad_norm": 14.726607807185552, "learning_rate": 1.3016555010528343e-06, "loss": 0.2786, "step": 43610 }, { "epoch": 11.830756712774614, "grad_norm": 15.9711847342541, "learning_rate": 1.2995324376479512e-06, "loss": 0.3169, "step": 43620 }, { "epoch": 11.833468944941687, "grad_norm": 18.998082162388684, "learning_rate": 1.2974108484255515e-06, "loss": 0.2883, "step": 43630 }, { "epoch": 11.836181177108761, "grad_norm": 20.938875151472256, "learning_rate": 1.2952907342308247e-06, "loss": 0.3213, "step": 43640 }, { "epoch": 11.838893409275833, "grad_norm": 13.33988396233098, "learning_rate": 1.293172095908376e-06, "loss": 0.2675, "step": 43650 }, { "epoch": 11.841605641442907, "grad_norm": 17.63169426900461, "learning_rate": 1.291054934302225e-06, "loss": 0.2912, "step": 43660 }, { "epoch": 11.844317873609981, "grad_norm": 14.517862069156159, "learning_rate": 1.288939250255799e-06, "loss": 0.2847, "step": 43670 }, { "epoch": 11.847030105777055, "grad_norm": 15.517734024860259, "learning_rate": 1.2868250446119418e-06, "loss": 0.3197, "step": 43680 }, { "epoch": 11.849742337944129, "grad_norm": 12.482193265016964, "learning_rate": 1.2847123182129006e-06, "loss": 0.2837, "step": 43690 }, { "epoch": 11.8524545701112, "grad_norm": 14.646886337897918, "learning_rate": 1.2826010719003407e-06, "loss": 0.3083, "step": 43700 }, { "epoch": 11.855166802278275, "grad_norm": 19.0473760674483, "learning_rate": 1.2804913065153334e-06, "loss": 0.2777, "step": 43710 }, { "epoch": 11.857879034445348, "grad_norm": 14.587264788966284, "learning_rate": 1.2783830228983629e-06, "loss": 0.313, "step": 43720 }, { "epoch": 11.860591266612422, "grad_norm": 12.306414448375637, "learning_rate": 1.2762762218893215e-06, "loss": 0.2547, "step": 43730 }, { "epoch": 11.863303498779496, "grad_norm": 14.072318475608872, "learning_rate": 1.274170904327508e-06, "loss": 0.2924, "step": 43740 }, { "epoch": 11.866015730946568, "grad_norm": 13.90382981995175, "learning_rate": 1.2720670710516387e-06, "loss": 0.2648, "step": 43750 }, { "epoch": 11.868727963113642, "grad_norm": 17.317882668227153, "learning_rate": 1.269964722899828e-06, "loss": 0.2712, "step": 43760 }, { "epoch": 11.871440195280716, "grad_norm": 19.332356605974066, "learning_rate": 1.2678638607096077e-06, "loss": 0.2873, "step": 43770 }, { "epoch": 11.87415242744779, "grad_norm": 11.467594125600913, "learning_rate": 1.265764485317909e-06, "loss": 0.2719, "step": 43780 }, { "epoch": 11.876864659614863, "grad_norm": 14.401616746831653, "learning_rate": 1.2636665975610796e-06, "loss": 0.2892, "step": 43790 }, { "epoch": 11.879576891781937, "grad_norm": 14.225532092535131, "learning_rate": 1.2615701982748713e-06, "loss": 0.275, "step": 43800 }, { "epoch": 11.88228912394901, "grad_norm": 17.92934371068315, "learning_rate": 1.2594752882944379e-06, "loss": 0.2979, "step": 43810 }, { "epoch": 11.885001356116083, "grad_norm": 12.763454174924473, "learning_rate": 1.2573818684543465e-06, "loss": 0.3147, "step": 43820 }, { "epoch": 11.887713588283157, "grad_norm": 10.94260519146805, "learning_rate": 1.2552899395885682e-06, "loss": 0.2443, "step": 43830 }, { "epoch": 11.89042582045023, "grad_norm": 14.798418353829769, "learning_rate": 1.253199502530481e-06, "loss": 0.301, "step": 43840 }, { "epoch": 11.893138052617305, "grad_norm": 14.0723372143287, "learning_rate": 1.2511105581128658e-06, "loss": 0.2968, "step": 43850 }, { "epoch": 11.895850284784377, "grad_norm": 16.559899627553133, "learning_rate": 1.2490231071679093e-06, "loss": 0.3393, "step": 43860 }, { "epoch": 11.89856251695145, "grad_norm": 14.13726243944508, "learning_rate": 1.2469371505272098e-06, "loss": 0.2606, "step": 43870 }, { "epoch": 11.901274749118524, "grad_norm": 16.502759733582394, "learning_rate": 1.244852689021761e-06, "loss": 0.2822, "step": 43880 }, { "epoch": 11.903986981285598, "grad_norm": 15.15229649343249, "learning_rate": 1.242769723481968e-06, "loss": 0.2724, "step": 43890 }, { "epoch": 11.906699213452672, "grad_norm": 14.23963477130373, "learning_rate": 1.240688254737632e-06, "loss": 0.305, "step": 43900 }, { "epoch": 11.909411445619746, "grad_norm": 14.920842984412033, "learning_rate": 1.2386082836179698e-06, "loss": 0.2804, "step": 43910 }, { "epoch": 11.912123677786818, "grad_norm": 14.401421577253988, "learning_rate": 1.23652981095159e-06, "loss": 0.2481, "step": 43920 }, { "epoch": 11.914835909953892, "grad_norm": 13.672977997272097, "learning_rate": 1.234452837566511e-06, "loss": 0.3039, "step": 43930 }, { "epoch": 11.917548142120966, "grad_norm": 16.680799320959327, "learning_rate": 1.232377364290151e-06, "loss": 0.2999, "step": 43940 }, { "epoch": 11.92026037428804, "grad_norm": 10.324592901573203, "learning_rate": 1.2303033919493323e-06, "loss": 0.2801, "step": 43950 }, { "epoch": 11.922972606455113, "grad_norm": 14.30265554214487, "learning_rate": 1.2282309213702798e-06, "loss": 0.3049, "step": 43960 }, { "epoch": 11.925684838622185, "grad_norm": 14.522330752343889, "learning_rate": 1.2261599533786156e-06, "loss": 0.2901, "step": 43970 }, { "epoch": 11.92839707078926, "grad_norm": 16.135586000956476, "learning_rate": 1.224090488799367e-06, "loss": 0.2889, "step": 43980 }, { "epoch": 11.931109302956333, "grad_norm": 11.839557589427226, "learning_rate": 1.222022528456966e-06, "loss": 0.2492, "step": 43990 }, { "epoch": 11.933821535123407, "grad_norm": 12.088569919639962, "learning_rate": 1.2199560731752363e-06, "loss": 0.2814, "step": 44000 }, { "epoch": 11.93653376729048, "grad_norm": 13.118787980375432, "learning_rate": 1.2178911237774106e-06, "loss": 0.2949, "step": 44010 }, { "epoch": 11.939245999457553, "grad_norm": 12.93953335320753, "learning_rate": 1.2158276810861136e-06, "loss": 0.2828, "step": 44020 }, { "epoch": 11.941958231624627, "grad_norm": 16.25031325267574, "learning_rate": 1.2137657459233786e-06, "loss": 0.3173, "step": 44030 }, { "epoch": 11.9446704637917, "grad_norm": 17.688426866309676, "learning_rate": 1.2117053191106315e-06, "loss": 0.303, "step": 44040 }, { "epoch": 11.947382695958774, "grad_norm": 11.757429852899802, "learning_rate": 1.2096464014686999e-06, "loss": 0.3154, "step": 44050 }, { "epoch": 11.950094928125848, "grad_norm": 15.304356483926561, "learning_rate": 1.2075889938178098e-06, "loss": 0.3027, "step": 44060 }, { "epoch": 11.95280716029292, "grad_norm": 11.827018981760743, "learning_rate": 1.205533096977587e-06, "loss": 0.267, "step": 44070 }, { "epoch": 11.955519392459994, "grad_norm": 14.445906537455967, "learning_rate": 1.2034787117670544e-06, "loss": 0.285, "step": 44080 }, { "epoch": 11.958231624627068, "grad_norm": 17.58300845587234, "learning_rate": 1.2014258390046302e-06, "loss": 0.2723, "step": 44090 }, { "epoch": 11.960943856794142, "grad_norm": 13.18253746963128, "learning_rate": 1.1993744795081342e-06, "loss": 0.3151, "step": 44100 }, { "epoch": 11.963656088961216, "grad_norm": 14.09472376970088, "learning_rate": 1.1973246340947808e-06, "loss": 0.2982, "step": 44110 }, { "epoch": 11.96636832112829, "grad_norm": 13.686241846262027, "learning_rate": 1.1952763035811827e-06, "loss": 0.3046, "step": 44120 }, { "epoch": 11.969080553295361, "grad_norm": 17.913863387441285, "learning_rate": 1.1932294887833495e-06, "loss": 0.3409, "step": 44130 }, { "epoch": 11.971792785462435, "grad_norm": 12.281308462300231, "learning_rate": 1.1911841905166816e-06, "loss": 0.2426, "step": 44140 }, { "epoch": 11.97450501762951, "grad_norm": 14.94724259003633, "learning_rate": 1.1891404095959852e-06, "loss": 0.2985, "step": 44150 }, { "epoch": 11.977217249796583, "grad_norm": 13.492730778146218, "learning_rate": 1.1870981468354526e-06, "loss": 0.2999, "step": 44160 }, { "epoch": 11.979929481963657, "grad_norm": 14.280607502288442, "learning_rate": 1.1850574030486756e-06, "loss": 0.281, "step": 44170 }, { "epoch": 11.982641714130729, "grad_norm": 18.727386060859423, "learning_rate": 1.1830181790486406e-06, "loss": 0.2802, "step": 44180 }, { "epoch": 11.985353946297803, "grad_norm": 13.234556484182876, "learning_rate": 1.1809804756477284e-06, "loss": 0.2636, "step": 44190 }, { "epoch": 11.988066178464877, "grad_norm": 16.835276985750394, "learning_rate": 1.178944293657715e-06, "loss": 0.2956, "step": 44200 }, { "epoch": 11.99077841063195, "grad_norm": 15.176280618476445, "learning_rate": 1.1769096338897673e-06, "loss": 0.2827, "step": 44210 }, { "epoch": 11.993490642799024, "grad_norm": 15.6469775447597, "learning_rate": 1.174876497154448e-06, "loss": 0.3096, "step": 44220 }, { "epoch": 11.996202874966096, "grad_norm": 9.744843772144089, "learning_rate": 1.1728448842617128e-06, "loss": 0.3033, "step": 44230 }, { "epoch": 11.99891510713317, "grad_norm": 17.433289938735385, "learning_rate": 1.1708147960209109e-06, "loss": 0.3238, "step": 44240 }, { "epoch": 12.001627339300244, "grad_norm": 12.230925080957855, "learning_rate": 1.1687862332407834e-06, "loss": 0.2471, "step": 44250 }, { "epoch": 12.004339571467318, "grad_norm": 8.735358596432167, "learning_rate": 1.166759196729464e-06, "loss": 0.1672, "step": 44260 }, { "epoch": 12.007051803634392, "grad_norm": 12.13336504329554, "learning_rate": 1.1647336872944786e-06, "loss": 0.1924, "step": 44270 }, { "epoch": 12.009764035801465, "grad_norm": 10.323039173687079, "learning_rate": 1.1627097057427416e-06, "loss": 0.1979, "step": 44280 }, { "epoch": 12.012476267968538, "grad_norm": 9.691020166426688, "learning_rate": 1.1606872528805641e-06, "loss": 0.1987, "step": 44290 }, { "epoch": 12.015188500135611, "grad_norm": 9.711683049324028, "learning_rate": 1.1586663295136441e-06, "loss": 0.191, "step": 44300 }, { "epoch": 12.017900732302685, "grad_norm": 12.361374161794073, "learning_rate": 1.1566469364470723e-06, "loss": 0.2001, "step": 44310 }, { "epoch": 12.020612964469759, "grad_norm": 12.086319615712593, "learning_rate": 1.1546290744853305e-06, "loss": 0.1971, "step": 44320 }, { "epoch": 12.023325196636833, "grad_norm": 16.72494511327802, "learning_rate": 1.1526127444322854e-06, "loss": 0.1861, "step": 44330 }, { "epoch": 12.026037428803905, "grad_norm": 10.227465151837968, "learning_rate": 1.150597947091202e-06, "loss": 0.1934, "step": 44340 }, { "epoch": 12.028749660970979, "grad_norm": 8.912552272062435, "learning_rate": 1.1485846832647267e-06, "loss": 0.198, "step": 44350 }, { "epoch": 12.031461893138053, "grad_norm": 9.962192464678667, "learning_rate": 1.1465729537548993e-06, "loss": 0.1829, "step": 44360 }, { "epoch": 12.034174125305126, "grad_norm": 12.352363460064954, "learning_rate": 1.1445627593631465e-06, "loss": 0.1856, "step": 44370 }, { "epoch": 12.0368863574722, "grad_norm": 9.787891711985283, "learning_rate": 1.1425541008902852e-06, "loss": 0.211, "step": 44380 }, { "epoch": 12.039598589639272, "grad_norm": 10.08451617438391, "learning_rate": 1.1405469791365203e-06, "loss": 0.1857, "step": 44390 }, { "epoch": 12.042310821806346, "grad_norm": 13.813951954096392, "learning_rate": 1.1385413949014413e-06, "loss": 0.2177, "step": 44400 }, { "epoch": 12.04502305397342, "grad_norm": 10.75612257742302, "learning_rate": 1.136537348984028e-06, "loss": 0.1958, "step": 44410 }, { "epoch": 12.047735286140494, "grad_norm": 12.686829330355168, "learning_rate": 1.1345348421826485e-06, "loss": 0.1846, "step": 44420 }, { "epoch": 12.050447518307568, "grad_norm": 7.528043302398663, "learning_rate": 1.132533875295055e-06, "loss": 0.1875, "step": 44430 }, { "epoch": 12.053159750474641, "grad_norm": 9.354527419235582, "learning_rate": 1.1305344491183889e-06, "loss": 0.1883, "step": 44440 }, { "epoch": 12.055871982641714, "grad_norm": 10.95698565979496, "learning_rate": 1.128536564449172e-06, "loss": 0.166, "step": 44450 }, { "epoch": 12.058584214808787, "grad_norm": 12.535628745023516, "learning_rate": 1.1265402220833228e-06, "loss": 0.1952, "step": 44460 }, { "epoch": 12.061296446975861, "grad_norm": 10.08268415559441, "learning_rate": 1.1245454228161334e-06, "loss": 0.2135, "step": 44470 }, { "epoch": 12.064008679142935, "grad_norm": 15.374554698726584, "learning_rate": 1.1225521674422896e-06, "loss": 0.1885, "step": 44480 }, { "epoch": 12.066720911310009, "grad_norm": 7.226872814116802, "learning_rate": 1.1205604567558575e-06, "loss": 0.1875, "step": 44490 }, { "epoch": 12.069433143477081, "grad_norm": 13.798685489108918, "learning_rate": 1.1185702915502911e-06, "loss": 0.1954, "step": 44500 }, { "epoch": 12.072145375644155, "grad_norm": 18.886158480241292, "learning_rate": 1.116581672618427e-06, "loss": 0.2055, "step": 44510 }, { "epoch": 12.074857607811229, "grad_norm": 8.664272592983725, "learning_rate": 1.114594600752485e-06, "loss": 0.1712, "step": 44520 }, { "epoch": 12.077569839978302, "grad_norm": 10.423002134615626, "learning_rate": 1.112609076744069e-06, "loss": 0.1814, "step": 44530 }, { "epoch": 12.080282072145376, "grad_norm": 8.751433615954653, "learning_rate": 1.1106251013841679e-06, "loss": 0.2074, "step": 44540 }, { "epoch": 12.082994304312448, "grad_norm": 8.122756669265419, "learning_rate": 1.108642675463153e-06, "loss": 0.1769, "step": 44550 }, { "epoch": 12.085706536479522, "grad_norm": 9.576209410518057, "learning_rate": 1.106661799770778e-06, "loss": 0.1576, "step": 44560 }, { "epoch": 12.088418768646596, "grad_norm": 16.00752105089599, "learning_rate": 1.1046824750961753e-06, "loss": 0.1881, "step": 44570 }, { "epoch": 12.09113100081367, "grad_norm": 10.046922767768748, "learning_rate": 1.1027047022278692e-06, "loss": 0.199, "step": 44580 }, { "epoch": 12.093843232980744, "grad_norm": 12.528715350063855, "learning_rate": 1.1007284819537544e-06, "loss": 0.1822, "step": 44590 }, { "epoch": 12.096555465147816, "grad_norm": 10.213707159024722, "learning_rate": 1.098753815061115e-06, "loss": 0.1685, "step": 44600 }, { "epoch": 12.09926769731489, "grad_norm": 11.152277314723824, "learning_rate": 1.0967807023366123e-06, "loss": 0.1962, "step": 44610 }, { "epoch": 12.101979929481963, "grad_norm": 11.849815324387055, "learning_rate": 1.09480914456629e-06, "loss": 0.2071, "step": 44620 }, { "epoch": 12.104692161649037, "grad_norm": 8.66991007407195, "learning_rate": 1.0928391425355745e-06, "loss": 0.1865, "step": 44630 }, { "epoch": 12.107404393816111, "grad_norm": 15.486296768428922, "learning_rate": 1.090870697029266e-06, "loss": 0.2016, "step": 44640 }, { "epoch": 12.110116625983185, "grad_norm": 7.465746176526196, "learning_rate": 1.0889038088315502e-06, "loss": 0.1886, "step": 44650 }, { "epoch": 12.112828858150257, "grad_norm": 10.413626680897796, "learning_rate": 1.0869384787259913e-06, "loss": 0.1643, "step": 44660 }, { "epoch": 12.11554109031733, "grad_norm": 11.193640381746452, "learning_rate": 1.0849747074955319e-06, "loss": 0.1572, "step": 44670 }, { "epoch": 12.118253322484405, "grad_norm": 9.710001550898395, "learning_rate": 1.0830124959224957e-06, "loss": 0.2174, "step": 44680 }, { "epoch": 12.120965554651479, "grad_norm": 8.370112283481978, "learning_rate": 1.0810518447885787e-06, "loss": 0.1785, "step": 44690 }, { "epoch": 12.123677786818552, "grad_norm": 15.861048852133312, "learning_rate": 1.0790927548748664e-06, "loss": 0.1925, "step": 44700 }, { "epoch": 12.126390018985624, "grad_norm": 10.032912761451733, "learning_rate": 1.0771352269618107e-06, "loss": 0.1832, "step": 44710 }, { "epoch": 12.129102251152698, "grad_norm": 14.408699736250894, "learning_rate": 1.0751792618292484e-06, "loss": 0.1945, "step": 44720 }, { "epoch": 12.131814483319772, "grad_norm": 11.316887631907735, "learning_rate": 1.073224860256391e-06, "loss": 0.1834, "step": 44730 }, { "epoch": 12.134526715486846, "grad_norm": 10.794754050945263, "learning_rate": 1.0712720230218287e-06, "loss": 0.1865, "step": 44740 }, { "epoch": 12.13723894765392, "grad_norm": 8.732366641589918, "learning_rate": 1.0693207509035285e-06, "loss": 0.1813, "step": 44750 }, { "epoch": 12.139951179820992, "grad_norm": 9.486399825188059, "learning_rate": 1.0673710446788306e-06, "loss": 0.215, "step": 44760 }, { "epoch": 12.142663411988066, "grad_norm": 8.77558560512729, "learning_rate": 1.0654229051244547e-06, "loss": 0.1725, "step": 44770 }, { "epoch": 12.14537564415514, "grad_norm": 10.637999357386809, "learning_rate": 1.063476333016496e-06, "loss": 0.1904, "step": 44780 }, { "epoch": 12.148087876322213, "grad_norm": 11.464205251788963, "learning_rate": 1.0615313291304242e-06, "loss": 0.2032, "step": 44790 }, { "epoch": 12.150800108489287, "grad_norm": 8.595071711424088, "learning_rate": 1.0595878942410864e-06, "loss": 0.1957, "step": 44800 }, { "epoch": 12.153512340656361, "grad_norm": 10.20484057176617, "learning_rate": 1.0576460291226998e-06, "loss": 0.1942, "step": 44810 }, { "epoch": 12.156224572823433, "grad_norm": 16.00655348085281, "learning_rate": 1.0557057345488635e-06, "loss": 0.2068, "step": 44820 }, { "epoch": 12.158936804990507, "grad_norm": 12.476476759208497, "learning_rate": 1.0537670112925436e-06, "loss": 0.2301, "step": 44830 }, { "epoch": 12.16164903715758, "grad_norm": 10.17606142633162, "learning_rate": 1.0518298601260863e-06, "loss": 0.1867, "step": 44840 }, { "epoch": 12.164361269324655, "grad_norm": 15.366487906878824, "learning_rate": 1.0498942818212048e-06, "loss": 0.2022, "step": 44850 }, { "epoch": 12.167073501491728, "grad_norm": 11.30860662603357, "learning_rate": 1.0479602771489933e-06, "loss": 0.2046, "step": 44860 }, { "epoch": 12.1697857336588, "grad_norm": 6.271933874242235, "learning_rate": 1.0460278468799156e-06, "loss": 0.2155, "step": 44870 }, { "epoch": 12.172497965825874, "grad_norm": 10.806864954222663, "learning_rate": 1.0440969917838056e-06, "loss": 0.1801, "step": 44880 }, { "epoch": 12.175210197992948, "grad_norm": 12.529274504721473, "learning_rate": 1.0421677126298735e-06, "loss": 0.1972, "step": 44890 }, { "epoch": 12.177922430160022, "grad_norm": 9.26597860560092, "learning_rate": 1.0402400101867006e-06, "loss": 0.1825, "step": 44900 }, { "epoch": 12.180634662327096, "grad_norm": 9.450641300974645, "learning_rate": 1.0383138852222408e-06, "loss": 0.1939, "step": 44910 }, { "epoch": 12.183346894494168, "grad_norm": 10.892593082409071, "learning_rate": 1.036389338503816e-06, "loss": 0.1887, "step": 44920 }, { "epoch": 12.186059126661242, "grad_norm": 7.657663718276867, "learning_rate": 1.034466370798123e-06, "loss": 0.186, "step": 44930 }, { "epoch": 12.188771358828316, "grad_norm": 12.745289860222302, "learning_rate": 1.0325449828712314e-06, "loss": 0.2075, "step": 44940 }, { "epoch": 12.19148359099539, "grad_norm": 15.000574249927803, "learning_rate": 1.0306251754885755e-06, "loss": 0.2025, "step": 44950 }, { "epoch": 12.194195823162463, "grad_norm": 11.952535340910645, "learning_rate": 1.0287069494149654e-06, "loss": 0.2255, "step": 44960 }, { "epoch": 12.196908055329537, "grad_norm": 11.240683666469778, "learning_rate": 1.0267903054145745e-06, "loss": 0.2069, "step": 44970 }, { "epoch": 12.199620287496609, "grad_norm": 10.6996632398934, "learning_rate": 1.0248752442509552e-06, "loss": 0.1973, "step": 44980 }, { "epoch": 12.202332519663683, "grad_norm": 9.863606000175238, "learning_rate": 1.022961766687024e-06, "loss": 0.1787, "step": 44990 }, { "epoch": 12.205044751830757, "grad_norm": 9.958994776345776, "learning_rate": 1.0210498734850644e-06, "loss": 0.1814, "step": 45000 }, { "epoch": 12.20775698399783, "grad_norm": 11.303060452235561, "learning_rate": 1.0191395654067327e-06, "loss": 0.1879, "step": 45010 }, { "epoch": 12.210469216164904, "grad_norm": 13.499411134128534, "learning_rate": 1.0172308432130529e-06, "loss": 0.1839, "step": 45020 }, { "epoch": 12.213181448331976, "grad_norm": 9.20882139644251, "learning_rate": 1.0153237076644168e-06, "loss": 0.1828, "step": 45030 }, { "epoch": 12.21589368049905, "grad_norm": 9.373555549862234, "learning_rate": 1.0134181595205827e-06, "loss": 0.1852, "step": 45040 }, { "epoch": 12.218605912666124, "grad_norm": 12.016131155436005, "learning_rate": 1.011514199540677e-06, "loss": 0.1974, "step": 45050 }, { "epoch": 12.221318144833198, "grad_norm": 10.49906502486142, "learning_rate": 1.0096118284831986e-06, "loss": 0.2052, "step": 45060 }, { "epoch": 12.224030377000272, "grad_norm": 8.544604447253763, "learning_rate": 1.0077110471060047e-06, "loss": 0.1675, "step": 45070 }, { "epoch": 12.226742609167344, "grad_norm": 12.483870321581888, "learning_rate": 1.0058118561663266e-06, "loss": 0.2004, "step": 45080 }, { "epoch": 12.229454841334418, "grad_norm": 12.069788211687698, "learning_rate": 1.0039142564207548e-06, "loss": 0.2256, "step": 45090 }, { "epoch": 12.232167073501492, "grad_norm": 10.887398811620573, "learning_rate": 1.002018248625256e-06, "loss": 0.2043, "step": 45100 }, { "epoch": 12.234879305668565, "grad_norm": 13.467725271470991, "learning_rate": 1.0001238335351515e-06, "loss": 0.1857, "step": 45110 }, { "epoch": 12.23759153783564, "grad_norm": 16.670136057181928, "learning_rate": 9.982310119051358e-07, "loss": 0.2276, "step": 45120 }, { "epoch": 12.240303770002713, "grad_norm": 10.130543353055275, "learning_rate": 9.963397844892653e-07, "loss": 0.1874, "step": 45130 }, { "epoch": 12.243016002169785, "grad_norm": 11.327832356018618, "learning_rate": 9.944501520409627e-07, "loss": 0.1849, "step": 45140 }, { "epoch": 12.245728234336859, "grad_norm": 7.030696728282471, "learning_rate": 9.925621153130165e-07, "loss": 0.1828, "step": 45150 }, { "epoch": 12.248440466503933, "grad_norm": 12.419005551456785, "learning_rate": 9.906756750575746e-07, "loss": 0.1987, "step": 45160 }, { "epoch": 12.251152698671007, "grad_norm": 13.848308895268506, "learning_rate": 9.887908320261536e-07, "loss": 0.2223, "step": 45170 }, { "epoch": 12.25386493083808, "grad_norm": 13.207070990229813, "learning_rate": 9.869075869696315e-07, "loss": 0.2014, "step": 45180 }, { "epoch": 12.256577163005153, "grad_norm": 15.909463262788265, "learning_rate": 9.850259406382518e-07, "loss": 0.2159, "step": 45190 }, { "epoch": 12.259289395172226, "grad_norm": 11.364005113337885, "learning_rate": 9.8314589378162e-07, "loss": 0.1903, "step": 45200 }, { "epoch": 12.2620016273393, "grad_norm": 11.188654151398406, "learning_rate": 9.81267447148701e-07, "loss": 0.1712, "step": 45210 }, { "epoch": 12.264713859506374, "grad_norm": 11.166708939838875, "learning_rate": 9.793906014878306e-07, "loss": 0.2047, "step": 45220 }, { "epoch": 12.267426091673448, "grad_norm": 10.104583324738703, "learning_rate": 9.775153575466972e-07, "loss": 0.1872, "step": 45230 }, { "epoch": 12.27013832384052, "grad_norm": 10.313502166381133, "learning_rate": 9.756417160723569e-07, "loss": 0.1975, "step": 45240 }, { "epoch": 12.272850556007594, "grad_norm": 6.567422736324697, "learning_rate": 9.737696778112267e-07, "loss": 0.1745, "step": 45250 }, { "epoch": 12.275562788174668, "grad_norm": 11.760725864088737, "learning_rate": 9.71899243509083e-07, "loss": 0.219, "step": 45260 }, { "epoch": 12.278275020341741, "grad_norm": 9.174459303709586, "learning_rate": 9.700304139110673e-07, "loss": 0.1869, "step": 45270 }, { "epoch": 12.280987252508815, "grad_norm": 10.348900646006454, "learning_rate": 9.68163189761675e-07, "loss": 0.1881, "step": 45280 }, { "epoch": 12.28369948467589, "grad_norm": 10.742963624970388, "learning_rate": 9.662975718047678e-07, "loss": 0.1834, "step": 45290 }, { "epoch": 12.286411716842961, "grad_norm": 11.555563861893617, "learning_rate": 9.644335607835665e-07, "loss": 0.1896, "step": 45300 }, { "epoch": 12.289123949010035, "grad_norm": 7.427857692536217, "learning_rate": 9.625711574406494e-07, "loss": 0.1943, "step": 45310 }, { "epoch": 12.291836181177109, "grad_norm": 9.374633666720069, "learning_rate": 9.607103625179582e-07, "loss": 0.1981, "step": 45320 }, { "epoch": 12.294548413344183, "grad_norm": 9.525370590303877, "learning_rate": 9.588511767567864e-07, "loss": 0.1751, "step": 45330 }, { "epoch": 12.297260645511257, "grad_norm": 10.60250251961374, "learning_rate": 9.569936008977982e-07, "loss": 0.2368, "step": 45340 }, { "epoch": 12.299972877678329, "grad_norm": 9.43184212213012, "learning_rate": 9.55137635681005e-07, "loss": 0.2136, "step": 45350 }, { "epoch": 12.302685109845402, "grad_norm": 8.15582669940344, "learning_rate": 9.532832818457822e-07, "loss": 0.185, "step": 45360 }, { "epoch": 12.305397342012476, "grad_norm": 10.781673104484602, "learning_rate": 9.514305401308632e-07, "loss": 0.1976, "step": 45370 }, { "epoch": 12.30810957417955, "grad_norm": 10.584665226490593, "learning_rate": 9.495794112743383e-07, "loss": 0.1972, "step": 45380 }, { "epoch": 12.310821806346624, "grad_norm": 12.096895523682912, "learning_rate": 9.47729896013656e-07, "loss": 0.1993, "step": 45390 }, { "epoch": 12.313534038513696, "grad_norm": 13.193559025537201, "learning_rate": 9.458819950856196e-07, "loss": 0.203, "step": 45400 }, { "epoch": 12.31624627068077, "grad_norm": 10.798619142044167, "learning_rate": 9.440357092263913e-07, "loss": 0.2002, "step": 45410 }, { "epoch": 12.318958502847844, "grad_norm": 9.079496185396929, "learning_rate": 9.4219103917149e-07, "loss": 0.1822, "step": 45420 }, { "epoch": 12.321670735014917, "grad_norm": 9.353376670906597, "learning_rate": 9.403479856557906e-07, "loss": 0.1928, "step": 45430 }, { "epoch": 12.324382967181991, "grad_norm": 10.549317594940558, "learning_rate": 9.385065494135253e-07, "loss": 0.1985, "step": 45440 }, { "epoch": 12.327095199349063, "grad_norm": 12.906570312810054, "learning_rate": 9.366667311782762e-07, "loss": 0.1871, "step": 45450 }, { "epoch": 12.329807431516137, "grad_norm": 10.196718935886484, "learning_rate": 9.348285316829908e-07, "loss": 0.1994, "step": 45460 }, { "epoch": 12.332519663683211, "grad_norm": 10.434254837510123, "learning_rate": 9.329919516599623e-07, "loss": 0.2, "step": 45470 }, { "epoch": 12.335231895850285, "grad_norm": 11.488725859575881, "learning_rate": 9.311569918408436e-07, "loss": 0.2016, "step": 45480 }, { "epoch": 12.337944128017359, "grad_norm": 8.718872791922374, "learning_rate": 9.293236529566413e-07, "loss": 0.1929, "step": 45490 }, { "epoch": 12.340656360184433, "grad_norm": 9.591913283172342, "learning_rate": 9.274919357377165e-07, "loss": 0.1997, "step": 45500 }, { "epoch": 12.343368592351505, "grad_norm": 12.434248947620844, "learning_rate": 9.256618409137841e-07, "loss": 0.2016, "step": 45510 }, { "epoch": 12.346080824518578, "grad_norm": 12.851119130064525, "learning_rate": 9.238333692139112e-07, "loss": 0.1907, "step": 45520 }, { "epoch": 12.348793056685652, "grad_norm": 11.206129049020422, "learning_rate": 9.220065213665197e-07, "loss": 0.1981, "step": 45530 }, { "epoch": 12.351505288852726, "grad_norm": 11.699759706330017, "learning_rate": 9.201812980993846e-07, "loss": 0.2066, "step": 45540 }, { "epoch": 12.3542175210198, "grad_norm": 11.505242237091798, "learning_rate": 9.183577001396337e-07, "loss": 0.1825, "step": 45550 }, { "epoch": 12.356929753186872, "grad_norm": 9.683271610163803, "learning_rate": 9.165357282137488e-07, "loss": 0.2004, "step": 45560 }, { "epoch": 12.359641985353946, "grad_norm": 13.766217622723417, "learning_rate": 9.14715383047558e-07, "loss": 0.1821, "step": 45570 }, { "epoch": 12.36235421752102, "grad_norm": 13.356775156757507, "learning_rate": 9.128966653662507e-07, "loss": 0.1864, "step": 45580 }, { "epoch": 12.365066449688094, "grad_norm": 7.800901892636366, "learning_rate": 9.110795758943592e-07, "loss": 0.1847, "step": 45590 }, { "epoch": 12.367778681855167, "grad_norm": 9.689247659016582, "learning_rate": 9.092641153557718e-07, "loss": 0.203, "step": 45600 }, { "epoch": 12.370490914022241, "grad_norm": 8.964153154187514, "learning_rate": 9.074502844737271e-07, "loss": 0.177, "step": 45610 }, { "epoch": 12.373203146189313, "grad_norm": 10.339982035692534, "learning_rate": 9.056380839708146e-07, "loss": 0.1862, "step": 45620 }, { "epoch": 12.375915378356387, "grad_norm": 10.114993152786122, "learning_rate": 9.038275145689739e-07, "loss": 0.1926, "step": 45630 }, { "epoch": 12.378627610523461, "grad_norm": 9.97298690462432, "learning_rate": 9.02018576989494e-07, "loss": 0.2205, "step": 45640 }, { "epoch": 12.381339842690535, "grad_norm": 8.90898116364994, "learning_rate": 9.002112719530149e-07, "loss": 0.1964, "step": 45650 }, { "epoch": 12.384052074857609, "grad_norm": 10.421569694988637, "learning_rate": 8.98405600179526e-07, "loss": 0.1964, "step": 45660 }, { "epoch": 12.38676430702468, "grad_norm": 10.21132787290626, "learning_rate": 8.966015623883667e-07, "loss": 0.1791, "step": 45670 }, { "epoch": 12.389476539191755, "grad_norm": 7.443654447073346, "learning_rate": 8.947991592982253e-07, "loss": 0.1807, "step": 45680 }, { "epoch": 12.392188771358828, "grad_norm": 11.270491813251107, "learning_rate": 8.929983916271351e-07, "loss": 0.1882, "step": 45690 }, { "epoch": 12.394901003525902, "grad_norm": 8.785808439003466, "learning_rate": 8.911992600924857e-07, "loss": 0.2113, "step": 45700 }, { "epoch": 12.397613235692976, "grad_norm": 10.691568002069685, "learning_rate": 8.894017654110077e-07, "loss": 0.1995, "step": 45710 }, { "epoch": 12.400325467860048, "grad_norm": 13.622298517749522, "learning_rate": 8.87605908298782e-07, "loss": 0.1793, "step": 45720 }, { "epoch": 12.403037700027122, "grad_norm": 13.184522579667036, "learning_rate": 8.858116894712387e-07, "loss": 0.243, "step": 45730 }, { "epoch": 12.405749932194196, "grad_norm": 9.361787851967001, "learning_rate": 8.840191096431533e-07, "loss": 0.1958, "step": 45740 }, { "epoch": 12.40846216436127, "grad_norm": 9.287738595952174, "learning_rate": 8.8222816952865e-07, "loss": 0.1803, "step": 45750 }, { "epoch": 12.411174396528343, "grad_norm": 9.114346950686047, "learning_rate": 8.804388698411969e-07, "loss": 0.1939, "step": 45760 }, { "epoch": 12.413886628695415, "grad_norm": 7.68327980323163, "learning_rate": 8.786512112936119e-07, "loss": 0.1641, "step": 45770 }, { "epoch": 12.41659886086249, "grad_norm": 11.121375905351, "learning_rate": 8.768651945980566e-07, "loss": 0.1874, "step": 45780 }, { "epoch": 12.419311093029563, "grad_norm": 12.019130844900427, "learning_rate": 8.750808204660405e-07, "loss": 0.2018, "step": 45790 }, { "epoch": 12.422023325196637, "grad_norm": 11.591432613761123, "learning_rate": 8.732980896084192e-07, "loss": 0.1862, "step": 45800 }, { "epoch": 12.42473555736371, "grad_norm": 11.805847623845933, "learning_rate": 8.715170027353881e-07, "loss": 0.2111, "step": 45810 }, { "epoch": 12.427447789530785, "grad_norm": 9.995626405291889, "learning_rate": 8.697375605564973e-07, "loss": 0.195, "step": 45820 }, { "epoch": 12.430160021697857, "grad_norm": 12.48539594590092, "learning_rate": 8.679597637806314e-07, "loss": 0.2144, "step": 45830 }, { "epoch": 12.43287225386493, "grad_norm": 7.441781290501989, "learning_rate": 8.661836131160284e-07, "loss": 0.1845, "step": 45840 }, { "epoch": 12.435584486032004, "grad_norm": 10.650786315068366, "learning_rate": 8.644091092702622e-07, "loss": 0.1822, "step": 45850 }, { "epoch": 12.438296718199078, "grad_norm": 12.804870291856544, "learning_rate": 8.62636252950258e-07, "loss": 0.1975, "step": 45860 }, { "epoch": 12.441008950366152, "grad_norm": 9.673885384668104, "learning_rate": 8.608650448622824e-07, "loss": 0.1727, "step": 45870 }, { "epoch": 12.443721182533224, "grad_norm": 12.049858127754478, "learning_rate": 8.590954857119421e-07, "loss": 0.1981, "step": 45880 }, { "epoch": 12.446433414700298, "grad_norm": 11.540674296537777, "learning_rate": 8.573275762041905e-07, "loss": 0.1785, "step": 45890 }, { "epoch": 12.449145646867372, "grad_norm": 11.765595003732368, "learning_rate": 8.555613170433224e-07, "loss": 0.1924, "step": 45900 }, { "epoch": 12.451857879034446, "grad_norm": 12.785561430687107, "learning_rate": 8.537967089329758e-07, "loss": 0.2011, "step": 45910 }, { "epoch": 12.45457011120152, "grad_norm": 7.969814702623539, "learning_rate": 8.520337525761313e-07, "loss": 0.191, "step": 45920 }, { "epoch": 12.457282343368592, "grad_norm": 10.758580559731405, "learning_rate": 8.50272448675108e-07, "loss": 0.1974, "step": 45930 }, { "epoch": 12.459994575535665, "grad_norm": 10.321151846922541, "learning_rate": 8.485127979315738e-07, "loss": 0.1809, "step": 45940 }, { "epoch": 12.46270680770274, "grad_norm": 7.1392823116134485, "learning_rate": 8.467548010465299e-07, "loss": 0.2015, "step": 45950 }, { "epoch": 12.465419039869813, "grad_norm": 12.816343353269076, "learning_rate": 8.449984587203258e-07, "loss": 0.2064, "step": 45960 }, { "epoch": 12.468131272036887, "grad_norm": 9.356180738695803, "learning_rate": 8.432437716526437e-07, "loss": 0.1819, "step": 45970 }, { "epoch": 12.47084350420396, "grad_norm": 8.071444156759588, "learning_rate": 8.414907405425155e-07, "loss": 0.2086, "step": 45980 }, { "epoch": 12.473555736371033, "grad_norm": 9.367328734362976, "learning_rate": 8.397393660883091e-07, "loss": 0.1875, "step": 45990 }, { "epoch": 12.476267968538107, "grad_norm": 14.030954798091201, "learning_rate": 8.379896489877292e-07, "loss": 0.2067, "step": 46000 }, { "epoch": 12.47898020070518, "grad_norm": 9.441140712492022, "learning_rate": 8.362415899378262e-07, "loss": 0.188, "step": 46010 }, { "epoch": 12.481692432872254, "grad_norm": 11.462398235483581, "learning_rate": 8.344951896349857e-07, "loss": 0.2158, "step": 46020 }, { "epoch": 12.484404665039328, "grad_norm": 7.234423563937429, "learning_rate": 8.327504487749366e-07, "loss": 0.2086, "step": 46030 }, { "epoch": 12.4871168972064, "grad_norm": 9.980672589490778, "learning_rate": 8.310073680527403e-07, "loss": 0.1712, "step": 46040 }, { "epoch": 12.489829129373474, "grad_norm": 9.949104645512293, "learning_rate": 8.292659481628018e-07, "loss": 0.2092, "step": 46050 }, { "epoch": 12.492541361540548, "grad_norm": 10.80292516195307, "learning_rate": 8.275261897988668e-07, "loss": 0.2033, "step": 46060 }, { "epoch": 12.495253593707622, "grad_norm": 8.830357646793823, "learning_rate": 8.257880936540114e-07, "loss": 0.2004, "step": 46070 }, { "epoch": 12.497965825874696, "grad_norm": 8.467768634076462, "learning_rate": 8.240516604206572e-07, "loss": 0.1891, "step": 46080 }, { "epoch": 12.500678058041768, "grad_norm": 12.66328038046013, "learning_rate": 8.223168907905549e-07, "loss": 0.2053, "step": 46090 }, { "epoch": 12.503390290208841, "grad_norm": 13.108762759158498, "learning_rate": 8.205837854548037e-07, "loss": 0.2009, "step": 46100 }, { "epoch": 12.506102522375915, "grad_norm": 12.726225307594667, "learning_rate": 8.188523451038288e-07, "loss": 0.229, "step": 46110 }, { "epoch": 12.508814754542989, "grad_norm": 11.15740310664281, "learning_rate": 8.171225704273983e-07, "loss": 0.215, "step": 46120 }, { "epoch": 12.511526986710063, "grad_norm": 10.896267125078948, "learning_rate": 8.153944621146159e-07, "loss": 0.2042, "step": 46130 }, { "epoch": 12.514239218877137, "grad_norm": 16.346764134596896, "learning_rate": 8.136680208539199e-07, "loss": 0.2358, "step": 46140 }, { "epoch": 12.516951451044209, "grad_norm": 10.905395003045362, "learning_rate": 8.119432473330874e-07, "loss": 0.1906, "step": 46150 }, { "epoch": 12.519663683211283, "grad_norm": 9.98342564105916, "learning_rate": 8.102201422392258e-07, "loss": 0.1894, "step": 46160 }, { "epoch": 12.522375915378356, "grad_norm": 8.031780006849505, "learning_rate": 8.084987062587834e-07, "loss": 0.1783, "step": 46170 }, { "epoch": 12.52508814754543, "grad_norm": 12.883457248890068, "learning_rate": 8.067789400775406e-07, "loss": 0.2004, "step": 46180 }, { "epoch": 12.527800379712504, "grad_norm": 11.277709390792426, "learning_rate": 8.050608443806135e-07, "loss": 0.1786, "step": 46190 }, { "epoch": 12.530512611879576, "grad_norm": 10.824853141986063, "learning_rate": 8.033444198524537e-07, "loss": 0.1986, "step": 46200 }, { "epoch": 12.53322484404665, "grad_norm": 10.632731583041673, "learning_rate": 8.016296671768425e-07, "loss": 0.1843, "step": 46210 }, { "epoch": 12.535937076213724, "grad_norm": 9.854707097772469, "learning_rate": 7.99916587036903e-07, "loss": 0.1911, "step": 46220 }, { "epoch": 12.538649308380798, "grad_norm": 11.18069921702747, "learning_rate": 7.982051801150836e-07, "loss": 0.1886, "step": 46230 }, { "epoch": 12.541361540547872, "grad_norm": 9.203014563015566, "learning_rate": 7.964954470931724e-07, "loss": 0.1943, "step": 46240 }, { "epoch": 12.544073772714944, "grad_norm": 7.902384028601576, "learning_rate": 7.947873886522872e-07, "loss": 0.1965, "step": 46250 }, { "epoch": 12.546786004882017, "grad_norm": 9.814423161911883, "learning_rate": 7.930810054728805e-07, "loss": 0.1846, "step": 46260 }, { "epoch": 12.549498237049091, "grad_norm": 15.043189804650872, "learning_rate": 7.913762982347379e-07, "loss": 0.2042, "step": 46270 }, { "epoch": 12.552210469216165, "grad_norm": 11.51824285109618, "learning_rate": 7.896732676169727e-07, "loss": 0.1939, "step": 46280 }, { "epoch": 12.554922701383239, "grad_norm": 7.996047233751053, "learning_rate": 7.879719142980364e-07, "loss": 0.1546, "step": 46290 }, { "epoch": 12.557634933550311, "grad_norm": 9.291917010176862, "learning_rate": 7.862722389557093e-07, "loss": 0.2139, "step": 46300 }, { "epoch": 12.560347165717385, "grad_norm": 9.822135707410377, "learning_rate": 7.845742422671027e-07, "loss": 0.1914, "step": 46310 }, { "epoch": 12.563059397884459, "grad_norm": 9.677659392760717, "learning_rate": 7.828779249086615e-07, "loss": 0.215, "step": 46320 }, { "epoch": 12.565771630051533, "grad_norm": 9.591673146336593, "learning_rate": 7.811832875561576e-07, "loss": 0.1932, "step": 46330 }, { "epoch": 12.568483862218606, "grad_norm": 9.702520456772119, "learning_rate": 7.794903308846996e-07, "loss": 0.1966, "step": 46340 }, { "epoch": 12.57119609438568, "grad_norm": 14.595691200867273, "learning_rate": 7.777990555687204e-07, "loss": 0.1922, "step": 46350 }, { "epoch": 12.573908326552752, "grad_norm": 11.339874766207624, "learning_rate": 7.761094622819865e-07, "loss": 0.1846, "step": 46360 }, { "epoch": 12.576620558719826, "grad_norm": 9.970639246238493, "learning_rate": 7.744215516975939e-07, "loss": 0.2052, "step": 46370 }, { "epoch": 12.5793327908869, "grad_norm": 8.278283558954584, "learning_rate": 7.727353244879676e-07, "loss": 0.1821, "step": 46380 }, { "epoch": 12.582045023053974, "grad_norm": 13.029612582006639, "learning_rate": 7.710507813248636e-07, "loss": 0.1908, "step": 46390 }, { "epoch": 12.584757255221048, "grad_norm": 9.20855117941277, "learning_rate": 7.693679228793638e-07, "loss": 0.1865, "step": 46400 }, { "epoch": 12.58746948738812, "grad_norm": 11.646661878622107, "learning_rate": 7.676867498218815e-07, "loss": 0.1815, "step": 46410 }, { "epoch": 12.590181719555193, "grad_norm": 11.371874962172152, "learning_rate": 7.660072628221588e-07, "loss": 0.1915, "step": 46420 }, { "epoch": 12.592893951722267, "grad_norm": 10.361211342517002, "learning_rate": 7.643294625492642e-07, "loss": 0.1854, "step": 46430 }, { "epoch": 12.595606183889341, "grad_norm": 11.877866209618409, "learning_rate": 7.626533496715977e-07, "loss": 0.2162, "step": 46440 }, { "epoch": 12.598318416056415, "grad_norm": 12.261538956257263, "learning_rate": 7.6097892485688e-07, "loss": 0.1934, "step": 46450 }, { "epoch": 12.601030648223489, "grad_norm": 18.618423004960636, "learning_rate": 7.593061887721692e-07, "loss": 0.208, "step": 46460 }, { "epoch": 12.603742880390561, "grad_norm": 13.104683483139896, "learning_rate": 7.576351420838423e-07, "loss": 0.1814, "step": 46470 }, { "epoch": 12.606455112557635, "grad_norm": 10.455447072565818, "learning_rate": 7.559657854576075e-07, "loss": 0.179, "step": 46480 }, { "epoch": 12.609167344724709, "grad_norm": 11.172916396089828, "learning_rate": 7.54298119558498e-07, "loss": 0.2109, "step": 46490 }, { "epoch": 12.611879576891782, "grad_norm": 11.591211510873148, "learning_rate": 7.526321450508756e-07, "loss": 0.2059, "step": 46500 }, { "epoch": 12.614591809058856, "grad_norm": 12.029131457950514, "learning_rate": 7.509678625984262e-07, "loss": 0.1827, "step": 46510 }, { "epoch": 12.617304041225928, "grad_norm": 7.475678215012219, "learning_rate": 7.493052728641615e-07, "loss": 0.202, "step": 46520 }, { "epoch": 12.620016273393002, "grad_norm": 9.345619435726904, "learning_rate": 7.476443765104197e-07, "loss": 0.2101, "step": 46530 }, { "epoch": 12.622728505560076, "grad_norm": 10.144099775777, "learning_rate": 7.459851741988644e-07, "loss": 0.18, "step": 46540 }, { "epoch": 12.62544073772715, "grad_norm": 10.821514237669437, "learning_rate": 7.443276665904848e-07, "loss": 0.1995, "step": 46550 }, { "epoch": 12.628152969894224, "grad_norm": 9.02752474526774, "learning_rate": 7.426718543455952e-07, "loss": 0.2072, "step": 46560 }, { "epoch": 12.630865202061296, "grad_norm": 12.829125093129832, "learning_rate": 7.410177381238298e-07, "loss": 0.1836, "step": 46570 }, { "epoch": 12.63357743422837, "grad_norm": 11.984877092095223, "learning_rate": 7.393653185841565e-07, "loss": 0.1739, "step": 46580 }, { "epoch": 12.636289666395443, "grad_norm": 14.1149176380384, "learning_rate": 7.37714596384858e-07, "loss": 0.204, "step": 46590 }, { "epoch": 12.639001898562517, "grad_norm": 12.916447777661546, "learning_rate": 7.360655721835458e-07, "loss": 0.2235, "step": 46600 }, { "epoch": 12.641714130729591, "grad_norm": 13.289178482346502, "learning_rate": 7.344182466371541e-07, "loss": 0.2031, "step": 46610 }, { "epoch": 12.644426362896663, "grad_norm": 11.091953602714362, "learning_rate": 7.327726204019391e-07, "loss": 0.1692, "step": 46620 }, { "epoch": 12.647138595063737, "grad_norm": 7.291829297985623, "learning_rate": 7.311286941334833e-07, "loss": 0.2013, "step": 46630 }, { "epoch": 12.64985082723081, "grad_norm": 8.93525927916297, "learning_rate": 7.29486468486687e-07, "loss": 0.1874, "step": 46640 }, { "epoch": 12.652563059397885, "grad_norm": 11.281587683453044, "learning_rate": 7.278459441157764e-07, "loss": 0.2044, "step": 46650 }, { "epoch": 12.655275291564958, "grad_norm": 13.176376376222965, "learning_rate": 7.262071216743e-07, "loss": 0.2212, "step": 46660 }, { "epoch": 12.657987523732032, "grad_norm": 11.253166705114058, "learning_rate": 7.245700018151275e-07, "loss": 0.1881, "step": 46670 }, { "epoch": 12.660699755899104, "grad_norm": 5.951070924867108, "learning_rate": 7.229345851904507e-07, "loss": 0.1792, "step": 46680 }, { "epoch": 12.663411988066178, "grad_norm": 13.00896960247789, "learning_rate": 7.2130087245178e-07, "loss": 0.219, "step": 46690 }, { "epoch": 12.666124220233252, "grad_norm": 12.087956703248013, "learning_rate": 7.196688642499533e-07, "loss": 0.1922, "step": 46700 }, { "epoch": 12.668836452400326, "grad_norm": 12.968313407269797, "learning_rate": 7.180385612351232e-07, "loss": 0.2091, "step": 46710 }, { "epoch": 12.6715486845674, "grad_norm": 10.066602156269349, "learning_rate": 7.164099640567662e-07, "loss": 0.1941, "step": 46720 }, { "epoch": 12.674260916734472, "grad_norm": 10.167911044992863, "learning_rate": 7.14783073363679e-07, "loss": 0.1911, "step": 46730 }, { "epoch": 12.676973148901546, "grad_norm": 13.906707040917452, "learning_rate": 7.13157889803977e-07, "loss": 0.206, "step": 46740 }, { "epoch": 12.67968538106862, "grad_norm": 7.954648737223228, "learning_rate": 7.115344140250985e-07, "loss": 0.1828, "step": 46750 }, { "epoch": 12.682397613235693, "grad_norm": 16.56551895881404, "learning_rate": 7.099126466737966e-07, "loss": 0.2132, "step": 46760 }, { "epoch": 12.685109845402767, "grad_norm": 7.3136949058079725, "learning_rate": 7.082925883961484e-07, "loss": 0.1653, "step": 46770 }, { "epoch": 12.68782207756984, "grad_norm": 10.792297277534248, "learning_rate": 7.066742398375482e-07, "loss": 0.1734, "step": 46780 }, { "epoch": 12.690534309736913, "grad_norm": 6.763810929702317, "learning_rate": 7.050576016427096e-07, "loss": 0.1852, "step": 46790 }, { "epoch": 12.693246541903987, "grad_norm": 13.930029994646997, "learning_rate": 7.034426744556655e-07, "loss": 0.1769, "step": 46800 }, { "epoch": 12.69595877407106, "grad_norm": 10.884021254598833, "learning_rate": 7.018294589197627e-07, "loss": 0.1939, "step": 46810 }, { "epoch": 12.698671006238134, "grad_norm": 10.334071375290648, "learning_rate": 7.002179556776751e-07, "loss": 0.1729, "step": 46820 }, { "epoch": 12.701383238405207, "grad_norm": 7.708969542601456, "learning_rate": 6.98608165371385e-07, "loss": 0.1751, "step": 46830 }, { "epoch": 12.70409547057228, "grad_norm": 9.218747752355663, "learning_rate": 6.970000886421979e-07, "loss": 0.1863, "step": 46840 }, { "epoch": 12.706807702739354, "grad_norm": 12.782146153805543, "learning_rate": 6.953937261307347e-07, "loss": 0.1794, "step": 46850 }, { "epoch": 12.709519934906428, "grad_norm": 10.032500054501309, "learning_rate": 6.937890784769341e-07, "loss": 0.1707, "step": 46860 }, { "epoch": 12.712232167073502, "grad_norm": 10.296872833048795, "learning_rate": 6.921861463200529e-07, "loss": 0.1817, "step": 46870 }, { "epoch": 12.714944399240576, "grad_norm": 12.405439014910272, "learning_rate": 6.905849302986606e-07, "loss": 0.1845, "step": 46880 }, { "epoch": 12.717656631407648, "grad_norm": 8.142731692265071, "learning_rate": 6.889854310506456e-07, "loss": 0.1668, "step": 46890 }, { "epoch": 12.720368863574722, "grad_norm": 9.490867949957963, "learning_rate": 6.873876492132136e-07, "loss": 0.2122, "step": 46900 }, { "epoch": 12.723081095741795, "grad_norm": 21.13702540224404, "learning_rate": 6.857915854228836e-07, "loss": 0.2303, "step": 46910 }, { "epoch": 12.72579332790887, "grad_norm": 12.396286061948034, "learning_rate": 6.841972403154929e-07, "loss": 0.1877, "step": 46920 }, { "epoch": 12.728505560075943, "grad_norm": 13.133447263398851, "learning_rate": 6.826046145261889e-07, "loss": 0.1967, "step": 46930 }, { "epoch": 12.731217792243015, "grad_norm": 12.237368867557016, "learning_rate": 6.810137086894419e-07, "loss": 0.1881, "step": 46940 }, { "epoch": 12.733930024410089, "grad_norm": 7.435333077859408, "learning_rate": 6.794245234390307e-07, "loss": 0.1952, "step": 46950 }, { "epoch": 12.736642256577163, "grad_norm": 8.015098664559567, "learning_rate": 6.778370594080513e-07, "loss": 0.2079, "step": 46960 }, { "epoch": 12.739354488744237, "grad_norm": 18.009856390445034, "learning_rate": 6.762513172289109e-07, "loss": 0.2177, "step": 46970 }, { "epoch": 12.74206672091131, "grad_norm": 10.289733826991164, "learning_rate": 6.746672975333362e-07, "loss": 0.1906, "step": 46980 }, { "epoch": 12.744778953078384, "grad_norm": 10.54259340141562, "learning_rate": 6.730850009523654e-07, "loss": 0.2116, "step": 46990 }, { "epoch": 12.747491185245456, "grad_norm": 12.099386207943077, "learning_rate": 6.715044281163474e-07, "loss": 0.2023, "step": 47000 }, { "epoch": 12.75020341741253, "grad_norm": 11.816468268968588, "learning_rate": 6.69925579654947e-07, "loss": 0.2263, "step": 47010 }, { "epoch": 12.752915649579604, "grad_norm": 9.406329731097822, "learning_rate": 6.683484561971421e-07, "loss": 0.1888, "step": 47020 }, { "epoch": 12.755627881746678, "grad_norm": 11.725139422629058, "learning_rate": 6.66773058371224e-07, "loss": 0.2159, "step": 47030 }, { "epoch": 12.758340113913752, "grad_norm": 12.47938504898295, "learning_rate": 6.651993868047934e-07, "loss": 0.1769, "step": 47040 }, { "epoch": 12.761052346080824, "grad_norm": 8.562881797110261, "learning_rate": 6.636274421247657e-07, "loss": 0.1945, "step": 47050 }, { "epoch": 12.763764578247898, "grad_norm": 10.004395351940232, "learning_rate": 6.620572249573709e-07, "loss": 0.2013, "step": 47060 }, { "epoch": 12.766476810414972, "grad_norm": 6.5064769278782215, "learning_rate": 6.604887359281443e-07, "loss": 0.1935, "step": 47070 }, { "epoch": 12.769189042582045, "grad_norm": 9.837742897311955, "learning_rate": 6.589219756619397e-07, "loss": 0.1931, "step": 47080 }, { "epoch": 12.77190127474912, "grad_norm": 10.119827775466689, "learning_rate": 6.573569447829148e-07, "loss": 0.2019, "step": 47090 }, { "epoch": 12.774613506916191, "grad_norm": 16.107238580362274, "learning_rate": 6.557936439145474e-07, "loss": 0.2048, "step": 47100 }, { "epoch": 12.777325739083265, "grad_norm": 14.215993283556307, "learning_rate": 6.542320736796171e-07, "loss": 0.2005, "step": 47110 }, { "epoch": 12.780037971250339, "grad_norm": 11.977546309426092, "learning_rate": 6.526722347002201e-07, "loss": 0.1978, "step": 47120 }, { "epoch": 12.782750203417413, "grad_norm": 10.700252374017323, "learning_rate": 6.511141275977606e-07, "loss": 0.1949, "step": 47130 }, { "epoch": 12.785462435584487, "grad_norm": 17.952319356040995, "learning_rate": 6.495577529929525e-07, "loss": 0.1925, "step": 47140 }, { "epoch": 12.788174667751559, "grad_norm": 12.037607285847118, "learning_rate": 6.480031115058222e-07, "loss": 0.2025, "step": 47150 }, { "epoch": 12.790886899918632, "grad_norm": 15.779464761096902, "learning_rate": 6.464502037557014e-07, "loss": 0.1952, "step": 47160 }, { "epoch": 12.793599132085706, "grad_norm": 10.74649033307021, "learning_rate": 6.448990303612341e-07, "loss": 0.1835, "step": 47170 }, { "epoch": 12.79631136425278, "grad_norm": 10.22125385328076, "learning_rate": 6.433495919403726e-07, "loss": 0.1981, "step": 47180 }, { "epoch": 12.799023596419854, "grad_norm": 11.47562004393764, "learning_rate": 6.418018891103789e-07, "loss": 0.2097, "step": 47190 }, { "epoch": 12.801735828586928, "grad_norm": 7.10205467752413, "learning_rate": 6.402559224878224e-07, "loss": 0.1942, "step": 47200 }, { "epoch": 12.804448060754, "grad_norm": 12.234753900800643, "learning_rate": 6.387116926885794e-07, "loss": 0.1704, "step": 47210 }, { "epoch": 12.807160292921074, "grad_norm": 12.296327737610245, "learning_rate": 6.371692003278395e-07, "loss": 0.1939, "step": 47220 }, { "epoch": 12.809872525088148, "grad_norm": 10.369606764743045, "learning_rate": 6.35628446020094e-07, "loss": 0.1944, "step": 47230 }, { "epoch": 12.812584757255221, "grad_norm": 11.61616057551094, "learning_rate": 6.340894303791451e-07, "loss": 0.1788, "step": 47240 }, { "epoch": 12.815296989422295, "grad_norm": 7.676984359185376, "learning_rate": 6.325521540181018e-07, "loss": 0.175, "step": 47250 }, { "epoch": 12.818009221589367, "grad_norm": 11.660446816311314, "learning_rate": 6.310166175493809e-07, "loss": 0.1973, "step": 47260 }, { "epoch": 12.820721453756441, "grad_norm": 9.541664157818657, "learning_rate": 6.29482821584706e-07, "loss": 0.2164, "step": 47270 }, { "epoch": 12.823433685923515, "grad_norm": 7.830677207750361, "learning_rate": 6.279507667351042e-07, "loss": 0.1977, "step": 47280 }, { "epoch": 12.826145918090589, "grad_norm": 11.824805012369788, "learning_rate": 6.264204536109131e-07, "loss": 0.2039, "step": 47290 }, { "epoch": 12.828858150257663, "grad_norm": 8.866560636803865, "learning_rate": 6.248918828217742e-07, "loss": 0.1915, "step": 47300 }, { "epoch": 12.831570382424736, "grad_norm": 7.652658385074741, "learning_rate": 6.233650549766362e-07, "loss": 0.1847, "step": 47310 }, { "epoch": 12.834282614591809, "grad_norm": 8.810673955644402, "learning_rate": 6.218399706837535e-07, "loss": 0.1869, "step": 47320 }, { "epoch": 12.836994846758882, "grad_norm": 6.775553377060615, "learning_rate": 6.203166305506814e-07, "loss": 0.177, "step": 47330 }, { "epoch": 12.839707078925956, "grad_norm": 12.906507337670154, "learning_rate": 6.187950351842892e-07, "loss": 0.2195, "step": 47340 }, { "epoch": 12.84241931109303, "grad_norm": 10.261473741782703, "learning_rate": 6.172751851907422e-07, "loss": 0.1912, "step": 47350 }, { "epoch": 12.845131543260104, "grad_norm": 10.017828021374612, "learning_rate": 6.157570811755159e-07, "loss": 0.1955, "step": 47360 }, { "epoch": 12.847843775427176, "grad_norm": 10.040929609519425, "learning_rate": 6.142407237433884e-07, "loss": 0.1739, "step": 47370 }, { "epoch": 12.85055600759425, "grad_norm": 7.320749514152863, "learning_rate": 6.127261134984413e-07, "loss": 0.2184, "step": 47380 }, { "epoch": 12.853268239761324, "grad_norm": 13.1863928306637, "learning_rate": 6.112132510440633e-07, "loss": 0.1877, "step": 47390 }, { "epoch": 12.855980471928397, "grad_norm": 9.514540142979547, "learning_rate": 6.09702136982942e-07, "loss": 0.1882, "step": 47400 }, { "epoch": 12.858692704095471, "grad_norm": 6.697247213419113, "learning_rate": 6.081927719170715e-07, "loss": 0.1852, "step": 47410 }, { "epoch": 12.861404936262543, "grad_norm": 13.219369807196674, "learning_rate": 6.0668515644775e-07, "loss": 0.1929, "step": 47420 }, { "epoch": 12.864117168429617, "grad_norm": 7.944098354964244, "learning_rate": 6.05179291175576e-07, "loss": 0.1966, "step": 47430 }, { "epoch": 12.866829400596691, "grad_norm": 19.090134109038893, "learning_rate": 6.036751767004534e-07, "loss": 0.2308, "step": 47440 }, { "epoch": 12.869541632763765, "grad_norm": 9.675412640347593, "learning_rate": 6.021728136215848e-07, "loss": 0.2054, "step": 47450 }, { "epoch": 12.872253864930839, "grad_norm": 12.426472001228326, "learning_rate": 6.006722025374817e-07, "loss": 0.1957, "step": 47460 }, { "epoch": 12.87496609709791, "grad_norm": 12.84980643051701, "learning_rate": 5.991733440459502e-07, "loss": 0.2118, "step": 47470 }, { "epoch": 12.877678329264985, "grad_norm": 14.996915419787312, "learning_rate": 5.976762387441027e-07, "loss": 0.1682, "step": 47480 }, { "epoch": 12.880390561432058, "grad_norm": 10.862356788191976, "learning_rate": 5.961808872283514e-07, "loss": 0.2045, "step": 47490 }, { "epoch": 12.883102793599132, "grad_norm": 8.70993834891894, "learning_rate": 5.946872900944117e-07, "loss": 0.1782, "step": 47500 }, { "epoch": 12.885815025766206, "grad_norm": 15.70855591004715, "learning_rate": 5.931954479372992e-07, "loss": 0.2256, "step": 47510 }, { "epoch": 12.88852725793328, "grad_norm": 10.471703278981655, "learning_rate": 5.917053613513285e-07, "loss": 0.1799, "step": 47520 }, { "epoch": 12.891239490100352, "grad_norm": 9.408202685512615, "learning_rate": 5.902170309301164e-07, "loss": 0.1783, "step": 47530 }, { "epoch": 12.893951722267426, "grad_norm": 11.605573912435087, "learning_rate": 5.887304572665809e-07, "loss": 0.181, "step": 47540 }, { "epoch": 12.8966639544345, "grad_norm": 10.16149150463249, "learning_rate": 5.872456409529387e-07, "loss": 0.1777, "step": 47550 }, { "epoch": 12.899376186601573, "grad_norm": 13.98938509110446, "learning_rate": 5.857625825807084e-07, "loss": 0.1892, "step": 47560 }, { "epoch": 12.902088418768647, "grad_norm": 9.146085563386352, "learning_rate": 5.842812827407041e-07, "loss": 0.1918, "step": 47570 }, { "epoch": 12.90480065093572, "grad_norm": 8.70219286423646, "learning_rate": 5.828017420230452e-07, "loss": 0.1913, "step": 47580 }, { "epoch": 12.907512883102793, "grad_norm": 12.452186558499983, "learning_rate": 5.813239610171456e-07, "loss": 0.1932, "step": 47590 }, { "epoch": 12.910225115269867, "grad_norm": 14.411865152484019, "learning_rate": 5.798479403117191e-07, "loss": 0.2206, "step": 47600 }, { "epoch": 12.91293734743694, "grad_norm": 14.295538830168592, "learning_rate": 5.783736804947804e-07, "loss": 0.1883, "step": 47610 }, { "epoch": 12.915649579604015, "grad_norm": 13.855563931467335, "learning_rate": 5.769011821536407e-07, "loss": 0.2041, "step": 47620 }, { "epoch": 12.918361811771087, "grad_norm": 11.405279626077775, "learning_rate": 5.754304458749105e-07, "loss": 0.212, "step": 47630 }, { "epoch": 12.92107404393816, "grad_norm": 9.980674202709404, "learning_rate": 5.739614722444964e-07, "loss": 0.1811, "step": 47640 }, { "epoch": 12.923786276105234, "grad_norm": 10.820006711897882, "learning_rate": 5.724942618476054e-07, "loss": 0.2078, "step": 47650 }, { "epoch": 12.926498508272308, "grad_norm": 10.667896514383411, "learning_rate": 5.710288152687404e-07, "loss": 0.1787, "step": 47660 }, { "epoch": 12.929210740439382, "grad_norm": 9.123705819578253, "learning_rate": 5.695651330917018e-07, "loss": 0.1753, "step": 47670 }, { "epoch": 12.931922972606454, "grad_norm": 14.375756087534219, "learning_rate": 5.681032158995897e-07, "loss": 0.1947, "step": 47680 }, { "epoch": 12.934635204773528, "grad_norm": 11.79355438337705, "learning_rate": 5.666430642747945e-07, "loss": 0.19, "step": 47690 }, { "epoch": 12.937347436940602, "grad_norm": 15.917616369099877, "learning_rate": 5.651846787990123e-07, "loss": 0.2026, "step": 47700 }, { "epoch": 12.940059669107676, "grad_norm": 10.40144974054267, "learning_rate": 5.63728060053228e-07, "loss": 0.186, "step": 47710 }, { "epoch": 12.94277190127475, "grad_norm": 10.503025854375291, "learning_rate": 5.622732086177257e-07, "loss": 0.1807, "step": 47720 }, { "epoch": 12.945484133441823, "grad_norm": 10.191566458235334, "learning_rate": 5.608201250720863e-07, "loss": 0.2092, "step": 47730 }, { "epoch": 12.948196365608895, "grad_norm": 8.193725681772309, "learning_rate": 5.593688099951844e-07, "loss": 0.2018, "step": 47740 }, { "epoch": 12.95090859777597, "grad_norm": 14.122960966606787, "learning_rate": 5.579192639651925e-07, "loss": 0.2142, "step": 47750 }, { "epoch": 12.953620829943043, "grad_norm": 11.066542042672612, "learning_rate": 5.564714875595751e-07, "loss": 0.1976, "step": 47760 }, { "epoch": 12.956333062110117, "grad_norm": 11.875422241215956, "learning_rate": 5.550254813550937e-07, "loss": 0.1894, "step": 47770 }, { "epoch": 12.95904529427719, "grad_norm": 10.73854973414532, "learning_rate": 5.535812459278056e-07, "loss": 0.185, "step": 47780 }, { "epoch": 12.961757526444263, "grad_norm": 7.511178644482383, "learning_rate": 5.52138781853061e-07, "loss": 0.1898, "step": 47790 }, { "epoch": 12.964469758611337, "grad_norm": 15.029666328472523, "learning_rate": 5.506980897055053e-07, "loss": 0.1926, "step": 47800 }, { "epoch": 12.96718199077841, "grad_norm": 11.802458248447195, "learning_rate": 5.49259170059076e-07, "loss": 0.2192, "step": 47810 }, { "epoch": 12.969894222945484, "grad_norm": 11.337029012292302, "learning_rate": 5.47822023487009e-07, "loss": 0.1786, "step": 47820 }, { "epoch": 12.972606455112558, "grad_norm": 10.783795220827997, "learning_rate": 5.463866505618282e-07, "loss": 0.2142, "step": 47830 }, { "epoch": 12.975318687279632, "grad_norm": 10.303462258157948, "learning_rate": 5.449530518553548e-07, "loss": 0.1783, "step": 47840 }, { "epoch": 12.978030919446704, "grad_norm": 7.720443918612764, "learning_rate": 5.435212279387014e-07, "loss": 0.1913, "step": 47850 }, { "epoch": 12.980743151613778, "grad_norm": 7.636207726460097, "learning_rate": 5.420911793822753e-07, "loss": 0.1705, "step": 47860 }, { "epoch": 12.983455383780852, "grad_norm": 13.974479777120443, "learning_rate": 5.406629067557761e-07, "loss": 0.2016, "step": 47870 }, { "epoch": 12.986167615947926, "grad_norm": 13.584487109936969, "learning_rate": 5.392364106281928e-07, "loss": 0.1942, "step": 47880 }, { "epoch": 12.988879848115, "grad_norm": 13.662028211937013, "learning_rate": 5.378116915678111e-07, "loss": 0.2041, "step": 47890 }, { "epoch": 12.991592080282071, "grad_norm": 8.69750027946428, "learning_rate": 5.363887501422055e-07, "loss": 0.1898, "step": 47900 }, { "epoch": 12.994304312449145, "grad_norm": 8.437095287133735, "learning_rate": 5.349675869182447e-07, "loss": 0.2013, "step": 47910 }, { "epoch": 12.99701654461622, "grad_norm": 8.960351253021399, "learning_rate": 5.335482024620886e-07, "loss": 0.178, "step": 47920 }, { "epoch": 12.999728776783293, "grad_norm": 11.911007436908255, "learning_rate": 5.321305973391849e-07, "loss": 0.1763, "step": 47930 }, { "epoch": 13.002441008950367, "grad_norm": 7.74730124352275, "learning_rate": 5.307147721142791e-07, "loss": 0.1537, "step": 47940 }, { "epoch": 13.005153241117439, "grad_norm": 7.031548153647507, "learning_rate": 5.293007273514011e-07, "loss": 0.1385, "step": 47950 }, { "epoch": 13.007865473284513, "grad_norm": 7.195616410689402, "learning_rate": 5.278884636138765e-07, "loss": 0.1474, "step": 47960 }, { "epoch": 13.010577705451587, "grad_norm": 8.330666536155812, "learning_rate": 5.264779814643156e-07, "loss": 0.1499, "step": 47970 }, { "epoch": 13.01328993761866, "grad_norm": 9.41198820245778, "learning_rate": 5.250692814646263e-07, "loss": 0.141, "step": 47980 }, { "epoch": 13.016002169785734, "grad_norm": 9.180212677034529, "learning_rate": 5.236623641760025e-07, "loss": 0.1342, "step": 47990 }, { "epoch": 13.018714401952808, "grad_norm": 10.00757266655243, "learning_rate": 5.222572301589263e-07, "loss": 0.1381, "step": 48000 }, { "epoch": 13.02142663411988, "grad_norm": 9.043970770937523, "learning_rate": 5.208538799731716e-07, "loss": 0.1441, "step": 48010 }, { "epoch": 13.024138866286954, "grad_norm": 7.732945157162469, "learning_rate": 5.194523141778024e-07, "loss": 0.1422, "step": 48020 }, { "epoch": 13.026851098454028, "grad_norm": 5.936701084341815, "learning_rate": 5.180525333311709e-07, "loss": 0.1413, "step": 48030 }, { "epoch": 13.029563330621102, "grad_norm": 8.100822087252148, "learning_rate": 5.16654537990916e-07, "loss": 0.1533, "step": 48040 }, { "epoch": 13.032275562788175, "grad_norm": 6.756048920600022, "learning_rate": 5.152583287139684e-07, "loss": 0.1294, "step": 48050 }, { "epoch": 13.034987794955248, "grad_norm": 8.896720847973276, "learning_rate": 5.138639060565487e-07, "loss": 0.1458, "step": 48060 }, { "epoch": 13.037700027122321, "grad_norm": 6.328688534598329, "learning_rate": 5.124712705741597e-07, "loss": 0.1424, "step": 48070 }, { "epoch": 13.040412259289395, "grad_norm": 6.4638591582720055, "learning_rate": 5.110804228215988e-07, "loss": 0.1436, "step": 48080 }, { "epoch": 13.043124491456469, "grad_norm": 5.846127529107079, "learning_rate": 5.096913633529448e-07, "loss": 0.1404, "step": 48090 }, { "epoch": 13.045836723623543, "grad_norm": 6.229890038098446, "learning_rate": 5.083040927215716e-07, "loss": 0.1368, "step": 48100 }, { "epoch": 13.048548955790615, "grad_norm": 7.785984971072877, "learning_rate": 5.069186114801334e-07, "loss": 0.1373, "step": 48110 }, { "epoch": 13.051261187957689, "grad_norm": 8.177999230852569, "learning_rate": 5.055349201805754e-07, "loss": 0.1461, "step": 48120 }, { "epoch": 13.053973420124763, "grad_norm": 8.377423043014646, "learning_rate": 5.04153019374129e-07, "loss": 0.167, "step": 48130 }, { "epoch": 13.056685652291836, "grad_norm": 9.231966424690253, "learning_rate": 5.027729096113126e-07, "loss": 0.1631, "step": 48140 }, { "epoch": 13.05939788445891, "grad_norm": 8.577503800727069, "learning_rate": 5.01394591441931e-07, "loss": 0.1375, "step": 48150 }, { "epoch": 13.062110116625982, "grad_norm": 16.092789603309324, "learning_rate": 5.000180654150732e-07, "loss": 0.1368, "step": 48160 }, { "epoch": 13.064822348793056, "grad_norm": 5.89271515185072, "learning_rate": 4.986433320791173e-07, "loss": 0.1384, "step": 48170 }, { "epoch": 13.06753458096013, "grad_norm": 8.792725716437982, "learning_rate": 4.972703919817251e-07, "loss": 0.1565, "step": 48180 }, { "epoch": 13.070246813127204, "grad_norm": 7.265293813546686, "learning_rate": 4.958992456698458e-07, "loss": 0.157, "step": 48190 }, { "epoch": 13.072959045294278, "grad_norm": 6.610943126961778, "learning_rate": 4.945298936897136e-07, "loss": 0.1557, "step": 48200 }, { "epoch": 13.075671277461351, "grad_norm": 8.605817806433855, "learning_rate": 4.931623365868449e-07, "loss": 0.1544, "step": 48210 }, { "epoch": 13.078383509628424, "grad_norm": 9.864176188716053, "learning_rate": 4.91796574906046e-07, "loss": 0.1546, "step": 48220 }, { "epoch": 13.081095741795497, "grad_norm": 6.209052848183088, "learning_rate": 4.904326091914042e-07, "loss": 0.1464, "step": 48230 }, { "epoch": 13.083807973962571, "grad_norm": 5.95855962773441, "learning_rate": 4.890704399862922e-07, "loss": 0.1522, "step": 48240 }, { "epoch": 13.086520206129645, "grad_norm": 8.035561723083594, "learning_rate": 4.87710067833368e-07, "loss": 0.1416, "step": 48250 }, { "epoch": 13.089232438296719, "grad_norm": 6.4179142645529055, "learning_rate": 4.863514932745733e-07, "loss": 0.1531, "step": 48260 }, { "epoch": 13.091944670463791, "grad_norm": 5.626638513866864, "learning_rate": 4.849947168511338e-07, "loss": 0.1226, "step": 48270 }, { "epoch": 13.094656902630865, "grad_norm": 8.212210897626838, "learning_rate": 4.836397391035568e-07, "loss": 0.1475, "step": 48280 }, { "epoch": 13.097369134797939, "grad_norm": 8.664707465799237, "learning_rate": 4.822865605716359e-07, "loss": 0.1429, "step": 48290 }, { "epoch": 13.100081366965012, "grad_norm": 10.339205208307186, "learning_rate": 4.809351817944463e-07, "loss": 0.1476, "step": 48300 }, { "epoch": 13.102793599132086, "grad_norm": 7.362207901914995, "learning_rate": 4.795856033103474e-07, "loss": 0.1543, "step": 48310 }, { "epoch": 13.105505831299158, "grad_norm": 10.812094590389203, "learning_rate": 4.782378256569814e-07, "loss": 0.1409, "step": 48320 }, { "epoch": 13.108218063466232, "grad_norm": 8.46618926446702, "learning_rate": 4.768918493712693e-07, "loss": 0.1546, "step": 48330 }, { "epoch": 13.110930295633306, "grad_norm": 9.827198515053505, "learning_rate": 4.755476749894211e-07, "loss": 0.1627, "step": 48340 }, { "epoch": 13.11364252780038, "grad_norm": 9.270451965804233, "learning_rate": 4.7420530304692347e-07, "loss": 0.1587, "step": 48350 }, { "epoch": 13.116354759967454, "grad_norm": 15.684724575027216, "learning_rate": 4.728647340785475e-07, "loss": 0.1406, "step": 48360 }, { "epoch": 13.119066992134528, "grad_norm": 7.753138228288748, "learning_rate": 4.715259686183454e-07, "loss": 0.157, "step": 48370 }, { "epoch": 13.1217792243016, "grad_norm": 6.226993963783908, "learning_rate": 4.701890071996512e-07, "loss": 0.1419, "step": 48380 }, { "epoch": 13.124491456468673, "grad_norm": 6.46653156217802, "learning_rate": 4.6885385035508104e-07, "loss": 0.1292, "step": 48390 }, { "epoch": 13.127203688635747, "grad_norm": 8.169529742470365, "learning_rate": 4.675204986165288e-07, "loss": 0.1392, "step": 48400 }, { "epoch": 13.129915920802821, "grad_norm": 8.348442487224986, "learning_rate": 4.6618895251517304e-07, "loss": 0.1431, "step": 48410 }, { "epoch": 13.132628152969895, "grad_norm": 5.997320120659237, "learning_rate": 4.6485921258147107e-07, "loss": 0.154, "step": 48420 }, { "epoch": 13.135340385136967, "grad_norm": 7.492776972751454, "learning_rate": 4.6353127934516185e-07, "loss": 0.1588, "step": 48430 }, { "epoch": 13.13805261730404, "grad_norm": 8.021840620354034, "learning_rate": 4.6220515333526506e-07, "loss": 0.1483, "step": 48440 }, { "epoch": 13.140764849471115, "grad_norm": 4.9646237609519615, "learning_rate": 4.608808350800753e-07, "loss": 0.1426, "step": 48450 }, { "epoch": 13.143477081638189, "grad_norm": 7.187301255775015, "learning_rate": 4.595583251071756e-07, "loss": 0.1495, "step": 48460 }, { "epoch": 13.146189313805262, "grad_norm": 6.174852969839417, "learning_rate": 4.5823762394342054e-07, "loss": 0.1526, "step": 48470 }, { "epoch": 13.148901545972334, "grad_norm": 8.50020737193217, "learning_rate": 4.569187321149493e-07, "loss": 0.154, "step": 48480 }, { "epoch": 13.151613778139408, "grad_norm": 6.127822427495343, "learning_rate": 4.5560165014717825e-07, "loss": 0.143, "step": 48490 }, { "epoch": 13.154326010306482, "grad_norm": 7.922489978294722, "learning_rate": 4.5428637856480275e-07, "loss": 0.151, "step": 48500 }, { "epoch": 13.157038242473556, "grad_norm": 8.528804370581982, "learning_rate": 4.5297291789179796e-07, "loss": 0.144, "step": 48510 }, { "epoch": 13.15975047464063, "grad_norm": 7.545614400128218, "learning_rate": 4.5166126865141546e-07, "loss": 0.1372, "step": 48520 }, { "epoch": 13.162462706807704, "grad_norm": 9.450350920592951, "learning_rate": 4.5035143136618776e-07, "loss": 0.1542, "step": 48530 }, { "epoch": 13.165174938974776, "grad_norm": 11.502056512780216, "learning_rate": 4.490434065579236e-07, "loss": 0.1656, "step": 48540 }, { "epoch": 13.16788717114185, "grad_norm": 6.5793350378903295, "learning_rate": 4.47737194747711e-07, "loss": 0.1381, "step": 48550 }, { "epoch": 13.170599403308923, "grad_norm": 6.61148114634675, "learning_rate": 4.4643279645591563e-07, "loss": 0.1338, "step": 48560 }, { "epoch": 13.173311635475997, "grad_norm": 6.114753540793644, "learning_rate": 4.4513021220217834e-07, "loss": 0.1482, "step": 48570 }, { "epoch": 13.176023867643071, "grad_norm": 10.390379035862951, "learning_rate": 4.4382944250542235e-07, "loss": 0.1507, "step": 48580 }, { "epoch": 13.178736099810143, "grad_norm": 10.885032244268038, "learning_rate": 4.4253048788384156e-07, "loss": 0.1409, "step": 48590 }, { "epoch": 13.181448331977217, "grad_norm": 9.691896095433737, "learning_rate": 4.41233348854912e-07, "loss": 0.1426, "step": 48600 }, { "epoch": 13.18416056414429, "grad_norm": 9.91102090633215, "learning_rate": 4.399380259353836e-07, "loss": 0.1532, "step": 48610 }, { "epoch": 13.186872796311365, "grad_norm": 11.287564994507433, "learning_rate": 4.386445196412842e-07, "loss": 0.1453, "step": 48620 }, { "epoch": 13.189585028478438, "grad_norm": 7.54837602209012, "learning_rate": 4.373528304879182e-07, "loss": 0.14, "step": 48630 }, { "epoch": 13.19229726064551, "grad_norm": 8.576381400436933, "learning_rate": 4.360629589898635e-07, "loss": 0.1501, "step": 48640 }, { "epoch": 13.195009492812584, "grad_norm": 10.18239313573493, "learning_rate": 4.3477490566097626e-07, "loss": 0.1587, "step": 48650 }, { "epoch": 13.197721724979658, "grad_norm": 9.38637896328262, "learning_rate": 4.3348867101438876e-07, "loss": 0.1507, "step": 48660 }, { "epoch": 13.200433957146732, "grad_norm": 8.12190460979681, "learning_rate": 4.322042555625067e-07, "loss": 0.1433, "step": 48670 }, { "epoch": 13.203146189313806, "grad_norm": 9.061462223972756, "learning_rate": 4.3092165981701297e-07, "loss": 0.1412, "step": 48680 }, { "epoch": 13.20585842148088, "grad_norm": 7.764513200345701, "learning_rate": 4.2964088428886275e-07, "loss": 0.1354, "step": 48690 }, { "epoch": 13.208570653647952, "grad_norm": 11.215913849060783, "learning_rate": 4.283619294882918e-07, "loss": 0.1585, "step": 48700 }, { "epoch": 13.211282885815026, "grad_norm": 6.521584195854624, "learning_rate": 4.2708479592480255e-07, "loss": 0.1467, "step": 48710 }, { "epoch": 13.2139951179821, "grad_norm": 8.341016501700603, "learning_rate": 4.258094841071786e-07, "loss": 0.1309, "step": 48720 }, { "epoch": 13.216707350149173, "grad_norm": 7.660498600075948, "learning_rate": 4.245359945434746e-07, "loss": 0.1299, "step": 48730 }, { "epoch": 13.219419582316247, "grad_norm": 8.305401241932833, "learning_rate": 4.2326432774101975e-07, "loss": 0.1374, "step": 48740 }, { "epoch": 13.222131814483319, "grad_norm": 6.751512556034302, "learning_rate": 4.2199448420641885e-07, "loss": 0.1346, "step": 48750 }, { "epoch": 13.224844046650393, "grad_norm": 10.825998456573712, "learning_rate": 4.2072646444554613e-07, "loss": 0.1499, "step": 48760 }, { "epoch": 13.227556278817467, "grad_norm": 9.884454055416871, "learning_rate": 4.1946026896355354e-07, "loss": 0.1411, "step": 48770 }, { "epoch": 13.23026851098454, "grad_norm": 8.46798880428063, "learning_rate": 4.1819589826486426e-07, "loss": 0.154, "step": 48780 }, { "epoch": 13.232980743151614, "grad_norm": 7.186139591383252, "learning_rate": 4.1693335285317547e-07, "loss": 0.1452, "step": 48790 }, { "epoch": 13.235692975318686, "grad_norm": 8.00303862326183, "learning_rate": 4.1567263323145756e-07, "loss": 0.1368, "step": 48800 }, { "epoch": 13.23840520748576, "grad_norm": 6.09655226932417, "learning_rate": 4.1441373990194935e-07, "loss": 0.1477, "step": 48810 }, { "epoch": 13.241117439652834, "grad_norm": 10.37110867204025, "learning_rate": 4.1315667336616904e-07, "loss": 0.1539, "step": 48820 }, { "epoch": 13.243829671819908, "grad_norm": 5.754145075135332, "learning_rate": 4.1190143412490167e-07, "loss": 0.1511, "step": 48830 }, { "epoch": 13.246541903986982, "grad_norm": 7.456861467194758, "learning_rate": 4.106480226782061e-07, "loss": 0.1506, "step": 48840 }, { "epoch": 13.249254136154054, "grad_norm": 10.525551388477554, "learning_rate": 4.0939643952541353e-07, "loss": 0.1603, "step": 48850 }, { "epoch": 13.251966368321128, "grad_norm": 8.650755614334834, "learning_rate": 4.081466851651261e-07, "loss": 0.1411, "step": 48860 }, { "epoch": 13.254678600488202, "grad_norm": 10.583502217372082, "learning_rate": 4.06898760095219e-07, "loss": 0.1583, "step": 48870 }, { "epoch": 13.257390832655275, "grad_norm": 7.392721958422807, "learning_rate": 4.0565266481283506e-07, "loss": 0.1319, "step": 48880 }, { "epoch": 13.26010306482235, "grad_norm": 8.29604770881447, "learning_rate": 4.0440839981439107e-07, "loss": 0.1385, "step": 48890 }, { "epoch": 13.262815296989423, "grad_norm": 8.423673514810906, "learning_rate": 4.0316596559557497e-07, "loss": 0.1363, "step": 48900 }, { "epoch": 13.265527529156495, "grad_norm": 14.618501326756945, "learning_rate": 4.019253626513436e-07, "loss": 0.1412, "step": 48910 }, { "epoch": 13.268239761323569, "grad_norm": 12.13290112711075, "learning_rate": 4.006865914759267e-07, "loss": 0.1549, "step": 48920 }, { "epoch": 13.270951993490643, "grad_norm": 7.494474221152689, "learning_rate": 3.994496525628189e-07, "loss": 0.1515, "step": 48930 }, { "epoch": 13.273664225657717, "grad_norm": 17.053792185001203, "learning_rate": 3.9821454640479317e-07, "loss": 0.1422, "step": 48940 }, { "epoch": 13.27637645782479, "grad_norm": 7.119172219952544, "learning_rate": 3.969812734938849e-07, "loss": 0.1388, "step": 48950 }, { "epoch": 13.279088689991863, "grad_norm": 7.370971019897568, "learning_rate": 3.957498343214039e-07, "loss": 0.1396, "step": 48960 }, { "epoch": 13.281800922158936, "grad_norm": 9.287167912057958, "learning_rate": 3.9452022937792434e-07, "loss": 0.148, "step": 48970 }, { "epoch": 13.28451315432601, "grad_norm": 8.878672730878565, "learning_rate": 3.9329245915329663e-07, "loss": 0.1449, "step": 48980 }, { "epoch": 13.287225386493084, "grad_norm": 6.971901953578355, "learning_rate": 3.9206652413663614e-07, "loss": 0.1386, "step": 48990 }, { "epoch": 13.289937618660158, "grad_norm": 8.78689832431388, "learning_rate": 3.908424248163262e-07, "loss": 0.1448, "step": 49000 }, { "epoch": 13.292649850827232, "grad_norm": 10.74924432490106, "learning_rate": 3.8962016168002046e-07, "loss": 0.1395, "step": 49010 }, { "epoch": 13.295362082994304, "grad_norm": 7.490500069236302, "learning_rate": 3.883997352146413e-07, "loss": 0.1358, "step": 49020 }, { "epoch": 13.298074315161378, "grad_norm": 8.626593102827151, "learning_rate": 3.8718114590638013e-07, "loss": 0.1508, "step": 49030 }, { "epoch": 13.300786547328451, "grad_norm": 7.231336422591378, "learning_rate": 3.859643942406932e-07, "loss": 0.1411, "step": 49040 }, { "epoch": 13.303498779495525, "grad_norm": 7.208498498599044, "learning_rate": 3.8474948070230664e-07, "loss": 0.1395, "step": 49050 }, { "epoch": 13.3062110116626, "grad_norm": 6.857046797007351, "learning_rate": 3.835364057752178e-07, "loss": 0.1574, "step": 49060 }, { "epoch": 13.308923243829671, "grad_norm": 7.037024668617298, "learning_rate": 3.82325169942685e-07, "loss": 0.1503, "step": 49070 }, { "epoch": 13.311635475996745, "grad_norm": 12.281411257487045, "learning_rate": 3.8111577368723894e-07, "loss": 0.1526, "step": 49080 }, { "epoch": 13.314347708163819, "grad_norm": 6.918305355781998, "learning_rate": 3.7990821749067317e-07, "loss": 0.1544, "step": 49090 }, { "epoch": 13.317059940330893, "grad_norm": 6.705856727510585, "learning_rate": 3.787025018340551e-07, "loss": 0.1426, "step": 49100 }, { "epoch": 13.319772172497967, "grad_norm": 5.501160589902557, "learning_rate": 3.7749862719771044e-07, "loss": 0.1362, "step": 49110 }, { "epoch": 13.322484404665039, "grad_norm": 6.038104141157242, "learning_rate": 3.762965940612373e-07, "loss": 0.1392, "step": 49120 }, { "epoch": 13.325196636832112, "grad_norm": 8.135394894685934, "learning_rate": 3.750964029034987e-07, "loss": 0.1373, "step": 49130 }, { "epoch": 13.327908868999186, "grad_norm": 10.653162287642987, "learning_rate": 3.738980542026227e-07, "loss": 0.1391, "step": 49140 }, { "epoch": 13.33062110116626, "grad_norm": 6.04836526405824, "learning_rate": 3.7270154843600634e-07, "loss": 0.1405, "step": 49150 }, { "epoch": 13.333333333333334, "grad_norm": 7.929368363587046, "learning_rate": 3.7150688608030783e-07, "loss": 0.1454, "step": 49160 }, { "epoch": 13.336045565500406, "grad_norm": 11.255240041041795, "learning_rate": 3.7031406761145415e-07, "loss": 0.1647, "step": 49170 }, { "epoch": 13.33875779766748, "grad_norm": 9.852461921098305, "learning_rate": 3.691230935046386e-07, "loss": 0.1447, "step": 49180 }, { "epoch": 13.341470029834554, "grad_norm": 11.166103909866466, "learning_rate": 3.679339642343166e-07, "loss": 0.1393, "step": 49190 }, { "epoch": 13.344182262001627, "grad_norm": 9.339785994163032, "learning_rate": 3.6674668027421257e-07, "loss": 0.1447, "step": 49200 }, { "epoch": 13.346894494168701, "grad_norm": 8.134577985474015, "learning_rate": 3.655612420973104e-07, "loss": 0.1438, "step": 49210 }, { "epoch": 13.349606726335775, "grad_norm": 9.307443757108635, "learning_rate": 3.6437765017586504e-07, "loss": 0.1666, "step": 49220 }, { "epoch": 13.352318958502847, "grad_norm": 8.131166121662778, "learning_rate": 3.6319590498139047e-07, "loss": 0.138, "step": 49230 }, { "epoch": 13.355031190669921, "grad_norm": 7.54877426374268, "learning_rate": 3.620160069846673e-07, "loss": 0.1664, "step": 49240 }, { "epoch": 13.357743422836995, "grad_norm": 8.369277679153246, "learning_rate": 3.608379566557418e-07, "loss": 0.1678, "step": 49250 }, { "epoch": 13.360455655004069, "grad_norm": 8.211951021586188, "learning_rate": 3.5966175446392083e-07, "loss": 0.1758, "step": 49260 }, { "epoch": 13.363167887171143, "grad_norm": 6.385388580780119, "learning_rate": 3.584874008777789e-07, "loss": 0.1615, "step": 49270 }, { "epoch": 13.365880119338215, "grad_norm": 5.937695184142141, "learning_rate": 3.5731489636514906e-07, "loss": 0.1545, "step": 49280 }, { "epoch": 13.368592351505288, "grad_norm": 6.0077665844582135, "learning_rate": 3.5614424139313207e-07, "loss": 0.1486, "step": 49290 }, { "epoch": 13.371304583672362, "grad_norm": 8.314777168379404, "learning_rate": 3.5497543642809054e-07, "loss": 0.1365, "step": 49300 }, { "epoch": 13.374016815839436, "grad_norm": 6.208606873319824, "learning_rate": 3.5380848193564907e-07, "loss": 0.1455, "step": 49310 }, { "epoch": 13.37672904800651, "grad_norm": 10.39662536519351, "learning_rate": 3.526433783806976e-07, "loss": 0.1444, "step": 49320 }, { "epoch": 13.379441280173582, "grad_norm": 23.324396077055006, "learning_rate": 3.5148012622738425e-07, "loss": 0.131, "step": 49330 }, { "epoch": 13.382153512340656, "grad_norm": 8.80619923044451, "learning_rate": 3.5031872593912497e-07, "loss": 0.1455, "step": 49340 }, { "epoch": 13.38486574450773, "grad_norm": 9.902839133517306, "learning_rate": 3.491591779785941e-07, "loss": 0.1464, "step": 49350 }, { "epoch": 13.387577976674804, "grad_norm": 8.262136913981793, "learning_rate": 3.4800148280772884e-07, "loss": 0.1523, "step": 49360 }, { "epoch": 13.390290208841877, "grad_norm": 7.205585185393193, "learning_rate": 3.468456408877297e-07, "loss": 0.1376, "step": 49370 }, { "epoch": 13.393002441008951, "grad_norm": 6.99925912025387, "learning_rate": 3.4569165267905766e-07, "loss": 0.1417, "step": 49380 }, { "epoch": 13.395714673176023, "grad_norm": 8.01162684258721, "learning_rate": 3.4453951864143633e-07, "loss": 0.15, "step": 49390 }, { "epoch": 13.398426905343097, "grad_norm": 11.89041999961258, "learning_rate": 3.433892392338478e-07, "loss": 0.1495, "step": 49400 }, { "epoch": 13.401139137510171, "grad_norm": 7.159780040763281, "learning_rate": 3.422408149145384e-07, "loss": 0.1542, "step": 49410 }, { "epoch": 13.403851369677245, "grad_norm": 8.388700482191714, "learning_rate": 3.410942461410144e-07, "loss": 0.1423, "step": 49420 }, { "epoch": 13.406563601844319, "grad_norm": 5.7112800404296475, "learning_rate": 3.399495333700431e-07, "loss": 0.1481, "step": 49430 }, { "epoch": 13.40927583401139, "grad_norm": 7.286128964005208, "learning_rate": 3.388066770576526e-07, "loss": 0.1394, "step": 49440 }, { "epoch": 13.411988066178465, "grad_norm": 9.46372961625417, "learning_rate": 3.376656776591286e-07, "loss": 0.15, "step": 49450 }, { "epoch": 13.414700298345538, "grad_norm": 8.218809266810114, "learning_rate": 3.365265356290226e-07, "loss": 0.1402, "step": 49460 }, { "epoch": 13.417412530512612, "grad_norm": 8.617860728067896, "learning_rate": 3.35389251421141e-07, "loss": 0.1484, "step": 49470 }, { "epoch": 13.420124762679686, "grad_norm": 12.22725835361358, "learning_rate": 3.342538254885519e-07, "loss": 0.1406, "step": 49480 }, { "epoch": 13.422836994846758, "grad_norm": 9.476439828586987, "learning_rate": 3.331202582835846e-07, "loss": 0.1472, "step": 49490 }, { "epoch": 13.425549227013832, "grad_norm": 7.8154234222638586, "learning_rate": 3.3198855025782574e-07, "loss": 0.1601, "step": 49500 }, { "epoch": 13.428261459180906, "grad_norm": 7.9412698647707805, "learning_rate": 3.308587018621229e-07, "loss": 0.1437, "step": 49510 }, { "epoch": 13.43097369134798, "grad_norm": 9.406295617568981, "learning_rate": 3.2973071354658106e-07, "loss": 0.1464, "step": 49520 }, { "epoch": 13.433685923515053, "grad_norm": 7.196706994276284, "learning_rate": 3.2860458576056573e-07, "loss": 0.1411, "step": 49530 }, { "epoch": 13.436398155682127, "grad_norm": 6.157568141744944, "learning_rate": 3.2748031895270037e-07, "loss": 0.1289, "step": 49540 }, { "epoch": 13.4391103878492, "grad_norm": 7.195404649309619, "learning_rate": 3.2635791357086777e-07, "loss": 0.1452, "step": 49550 }, { "epoch": 13.441822620016273, "grad_norm": 7.15410309739374, "learning_rate": 3.2523737006221024e-07, "loss": 0.1616, "step": 49560 }, { "epoch": 13.444534852183347, "grad_norm": 9.855115636849881, "learning_rate": 3.241186888731235e-07, "loss": 0.1483, "step": 49570 }, { "epoch": 13.44724708435042, "grad_norm": 8.780705276887225, "learning_rate": 3.230018704492682e-07, "loss": 0.1468, "step": 49580 }, { "epoch": 13.449959316517495, "grad_norm": 9.686521072262982, "learning_rate": 3.218869152355575e-07, "loss": 0.1479, "step": 49590 }, { "epoch": 13.452671548684567, "grad_norm": 6.238593257279134, "learning_rate": 3.2077382367616483e-07, "loss": 0.1523, "step": 49600 }, { "epoch": 13.45538378085164, "grad_norm": 8.974440294220685, "learning_rate": 3.1966259621452114e-07, "loss": 0.1592, "step": 49610 }, { "epoch": 13.458096013018714, "grad_norm": 8.123150525473251, "learning_rate": 3.1855323329331335e-07, "loss": 0.1319, "step": 49620 }, { "epoch": 13.460808245185788, "grad_norm": 6.8677000257375544, "learning_rate": 3.1744573535448854e-07, "loss": 0.138, "step": 49630 }, { "epoch": 13.463520477352862, "grad_norm": 6.659369041150408, "learning_rate": 3.16340102839246e-07, "loss": 0.1486, "step": 49640 }, { "epoch": 13.466232709519934, "grad_norm": 11.991110589445086, "learning_rate": 3.152363361880462e-07, "loss": 0.1563, "step": 49650 }, { "epoch": 13.468944941687008, "grad_norm": 10.112414316386324, "learning_rate": 3.1413443584060456e-07, "loss": 0.1595, "step": 49660 }, { "epoch": 13.471657173854082, "grad_norm": 9.314494986768663, "learning_rate": 3.130344022358933e-07, "loss": 0.1535, "step": 49670 }, { "epoch": 13.474369406021156, "grad_norm": 5.34199489262054, "learning_rate": 3.1193623581214194e-07, "loss": 0.1345, "step": 49680 }, { "epoch": 13.47708163818823, "grad_norm": 6.288285853820844, "learning_rate": 3.1083993700683215e-07, "loss": 0.1453, "step": 49690 }, { "epoch": 13.479793870355302, "grad_norm": 8.168525100214342, "learning_rate": 3.097455062567084e-07, "loss": 0.1385, "step": 49700 }, { "epoch": 13.482506102522375, "grad_norm": 7.026535352648565, "learning_rate": 3.0865294399776424e-07, "loss": 0.1373, "step": 49710 }, { "epoch": 13.48521833468945, "grad_norm": 7.5195652127725365, "learning_rate": 3.075622506652526e-07, "loss": 0.1475, "step": 49720 }, { "epoch": 13.487930566856523, "grad_norm": 7.575525992378784, "learning_rate": 3.0647342669368087e-07, "loss": 0.1412, "step": 49730 }, { "epoch": 13.490642799023597, "grad_norm": 11.459047855894978, "learning_rate": 3.053864725168121e-07, "loss": 0.137, "step": 49740 }, { "epoch": 13.49335503119067, "grad_norm": 11.721173926876121, "learning_rate": 3.0430138856766544e-07, "loss": 0.1448, "step": 49750 }, { "epoch": 13.496067263357743, "grad_norm": 7.951395911011389, "learning_rate": 3.032181752785113e-07, "loss": 0.1588, "step": 49760 }, { "epoch": 13.498779495524817, "grad_norm": 9.934569745948217, "learning_rate": 3.021368330808788e-07, "loss": 0.1408, "step": 49770 }, { "epoch": 13.50149172769189, "grad_norm": 8.37739475045549, "learning_rate": 3.010573624055496e-07, "loss": 0.133, "step": 49780 }, { "epoch": 13.504203959858964, "grad_norm": 10.926073125883448, "learning_rate": 2.9997976368256075e-07, "loss": 0.1538, "step": 49790 }, { "epoch": 13.506916192026038, "grad_norm": 6.8406202151186895, "learning_rate": 2.9890403734120334e-07, "loss": 0.1389, "step": 49800 }, { "epoch": 13.50962842419311, "grad_norm": 6.951648280523901, "learning_rate": 2.9783018381002226e-07, "loss": 0.1418, "step": 49810 }, { "epoch": 13.512340656360184, "grad_norm": 8.730529468705482, "learning_rate": 2.9675820351681706e-07, "loss": 0.1723, "step": 49820 }, { "epoch": 13.515052888527258, "grad_norm": 7.301539330626089, "learning_rate": 2.956880968886389e-07, "loss": 0.1601, "step": 49830 }, { "epoch": 13.517765120694332, "grad_norm": 8.384912083202751, "learning_rate": 2.9461986435179445e-07, "loss": 0.1462, "step": 49840 }, { "epoch": 13.520477352861406, "grad_norm": 8.499620240212595, "learning_rate": 2.93553506331844e-07, "loss": 0.1414, "step": 49850 }, { "epoch": 13.52318958502848, "grad_norm": 5.326684243067182, "learning_rate": 2.924890232536004e-07, "loss": 0.1426, "step": 49860 }, { "epoch": 13.525901817195551, "grad_norm": 6.544507153976101, "learning_rate": 2.9142641554113004e-07, "loss": 0.1406, "step": 49870 }, { "epoch": 13.528614049362625, "grad_norm": 9.79748776452419, "learning_rate": 2.903656836177493e-07, "loss": 0.1454, "step": 49880 }, { "epoch": 13.531326281529699, "grad_norm": 10.300760159144, "learning_rate": 2.8930682790603347e-07, "loss": 0.1459, "step": 49890 }, { "epoch": 13.534038513696773, "grad_norm": 10.153905990243917, "learning_rate": 2.8824984882780406e-07, "loss": 0.1452, "step": 49900 }, { "epoch": 13.536750745863847, "grad_norm": 5.75424037483558, "learning_rate": 2.8719474680413806e-07, "loss": 0.1357, "step": 49910 }, { "epoch": 13.539462978030919, "grad_norm": 9.549451954995414, "learning_rate": 2.861415222553648e-07, "loss": 0.1574, "step": 49920 }, { "epoch": 13.542175210197993, "grad_norm": 6.2929742959650925, "learning_rate": 2.8509017560106434e-07, "loss": 0.1196, "step": 49930 }, { "epoch": 13.544887442365066, "grad_norm": 6.367612782153681, "learning_rate": 2.8404070726007095e-07, "loss": 0.1307, "step": 49940 }, { "epoch": 13.54759967453214, "grad_norm": 9.535333307987187, "learning_rate": 2.829931176504669e-07, "loss": 0.1506, "step": 49950 }, { "epoch": 13.550311906699214, "grad_norm": 7.9672769581853276, "learning_rate": 2.8194740718958945e-07, "loss": 0.1403, "step": 49960 }, { "epoch": 13.553024138866286, "grad_norm": 10.603801199221248, "learning_rate": 2.8090357629402545e-07, "loss": 0.1462, "step": 49970 }, { "epoch": 13.55573637103336, "grad_norm": 8.913999201803351, "learning_rate": 2.798616253796138e-07, "loss": 0.1391, "step": 49980 }, { "epoch": 13.558448603200434, "grad_norm": 7.65634740588336, "learning_rate": 2.7882155486144535e-07, "loss": 0.1421, "step": 49990 }, { "epoch": 13.561160835367508, "grad_norm": 7.135422797621485, "learning_rate": 2.777833651538575e-07, "loss": 0.1356, "step": 50000 }, { "epoch": 13.563873067534582, "grad_norm": 7.8002193703840135, "learning_rate": 2.76747056670445e-07, "loss": 0.1468, "step": 50010 }, { "epoch": 13.566585299701654, "grad_norm": 8.837594731318273, "learning_rate": 2.75712629824047e-07, "loss": 0.1442, "step": 50020 }, { "epoch": 13.569297531868727, "grad_norm": 7.6550503871083935, "learning_rate": 2.746800850267578e-07, "loss": 0.1313, "step": 50030 }, { "epoch": 13.572009764035801, "grad_norm": 7.743551884945781, "learning_rate": 2.736494226899178e-07, "loss": 0.1381, "step": 50040 }, { "epoch": 13.574721996202875, "grad_norm": 8.735723756981068, "learning_rate": 2.7262064322412075e-07, "loss": 0.1321, "step": 50050 }, { "epoch": 13.577434228369949, "grad_norm": 6.034652223196258, "learning_rate": 2.7159374703920995e-07, "loss": 0.1488, "step": 50060 }, { "epoch": 13.580146460537023, "grad_norm": 7.892258504995161, "learning_rate": 2.705687345442759e-07, "loss": 0.1268, "step": 50070 }, { "epoch": 13.582858692704095, "grad_norm": 10.171599484921249, "learning_rate": 2.695456061476609e-07, "loss": 0.1543, "step": 50080 }, { "epoch": 13.585570924871169, "grad_norm": 9.608618697336567, "learning_rate": 2.685243622569567e-07, "loss": 0.1416, "step": 50090 }, { "epoch": 13.588283157038243, "grad_norm": 9.313314362805274, "learning_rate": 2.6750500327900383e-07, "loss": 0.1466, "step": 50100 }, { "epoch": 13.590995389205316, "grad_norm": 6.587930241584889, "learning_rate": 2.66487529619891e-07, "loss": 0.1432, "step": 50110 }, { "epoch": 13.59370762137239, "grad_norm": 7.3227297952346095, "learning_rate": 2.654719416849555e-07, "loss": 0.1424, "step": 50120 }, { "epoch": 13.596419853539462, "grad_norm": 8.003266597413802, "learning_rate": 2.6445823987878815e-07, "loss": 0.1656, "step": 50130 }, { "epoch": 13.599132085706536, "grad_norm": 7.337083955888947, "learning_rate": 2.634464246052221e-07, "loss": 0.1327, "step": 50140 }, { "epoch": 13.60184431787361, "grad_norm": 6.355880426798901, "learning_rate": 2.6243649626734256e-07, "loss": 0.141, "step": 50150 }, { "epoch": 13.604556550040684, "grad_norm": 8.08947952297106, "learning_rate": 2.614284552674806e-07, "loss": 0.1539, "step": 50160 }, { "epoch": 13.607268782207758, "grad_norm": 8.97847444226663, "learning_rate": 2.604223020072194e-07, "loss": 0.1455, "step": 50170 }, { "epoch": 13.60998101437483, "grad_norm": 8.076961002892782, "learning_rate": 2.5941803688738607e-07, "loss": 0.1419, "step": 50180 }, { "epoch": 13.612693246541903, "grad_norm": 5.638500347268924, "learning_rate": 2.584156603080573e-07, "loss": 0.1403, "step": 50190 }, { "epoch": 13.615405478708977, "grad_norm": 7.403866941347626, "learning_rate": 2.5741517266855696e-07, "loss": 0.1392, "step": 50200 }, { "epoch": 13.618117710876051, "grad_norm": 7.260391263351391, "learning_rate": 2.564165743674574e-07, "loss": 0.1631, "step": 50210 }, { "epoch": 13.620829943043125, "grad_norm": 8.715538403895433, "learning_rate": 2.554198658025775e-07, "loss": 0.1261, "step": 50220 }, { "epoch": 13.623542175210197, "grad_norm": 9.656893747469292, "learning_rate": 2.544250473709831e-07, "loss": 0.1533, "step": 50230 }, { "epoch": 13.626254407377271, "grad_norm": 6.922996186830688, "learning_rate": 2.534321194689865e-07, "loss": 0.1362, "step": 50240 }, { "epoch": 13.628966639544345, "grad_norm": 9.679792406667884, "learning_rate": 2.524410824921492e-07, "loss": 0.1594, "step": 50250 }, { "epoch": 13.631678871711419, "grad_norm": 12.57097111674985, "learning_rate": 2.514519368352769e-07, "loss": 0.1576, "step": 50260 }, { "epoch": 13.634391103878492, "grad_norm": 12.16951506543268, "learning_rate": 2.50464682892424e-07, "loss": 0.1494, "step": 50270 }, { "epoch": 13.637103336045566, "grad_norm": 7.6901173881361675, "learning_rate": 2.4947932105688743e-07, "loss": 0.1462, "step": 50280 }, { "epoch": 13.639815568212638, "grad_norm": 9.425205514232697, "learning_rate": 2.484958517212166e-07, "loss": 0.144, "step": 50290 }, { "epoch": 13.642527800379712, "grad_norm": 8.181274472575929, "learning_rate": 2.4751427527720087e-07, "loss": 0.1589, "step": 50300 }, { "epoch": 13.645240032546786, "grad_norm": 6.969120077036741, "learning_rate": 2.4653459211587904e-07, "loss": 0.148, "step": 50310 }, { "epoch": 13.64795226471386, "grad_norm": 7.3007258440525185, "learning_rate": 2.4555680262753455e-07, "loss": 0.1248, "step": 50320 }, { "epoch": 13.650664496880934, "grad_norm": 9.413377368828002, "learning_rate": 2.445809072016964e-07, "loss": 0.137, "step": 50330 }, { "epoch": 13.653376729048006, "grad_norm": 9.607474513570665, "learning_rate": 2.4360690622714014e-07, "loss": 0.1515, "step": 50340 }, { "epoch": 13.65608896121508, "grad_norm": 6.024960034601547, "learning_rate": 2.426348000918843e-07, "loss": 0.1829, "step": 50350 }, { "epoch": 13.658801193382153, "grad_norm": 8.96211605832958, "learning_rate": 2.4166458918319425e-07, "loss": 0.1426, "step": 50360 }, { "epoch": 13.661513425549227, "grad_norm": 6.004393069313836, "learning_rate": 2.406962738875812e-07, "loss": 0.1333, "step": 50370 }, { "epoch": 13.664225657716301, "grad_norm": 7.775376010341581, "learning_rate": 2.397298545907983e-07, "loss": 0.1614, "step": 50380 }, { "epoch": 13.666937889883375, "grad_norm": 9.473590491979353, "learning_rate": 2.3876533167784766e-07, "loss": 0.1524, "step": 50390 }, { "epoch": 13.669650122050447, "grad_norm": 8.27864976374795, "learning_rate": 2.3780270553296959e-07, "loss": 0.1455, "step": 50400 }, { "epoch": 13.67236235421752, "grad_norm": 5.30202746950802, "learning_rate": 2.368419765396568e-07, "loss": 0.1446, "step": 50410 }, { "epoch": 13.675074586384595, "grad_norm": 8.18862669376248, "learning_rate": 2.358831450806387e-07, "loss": 0.1458, "step": 50420 }, { "epoch": 13.677786818551668, "grad_norm": 8.100517862767365, "learning_rate": 2.3492621153789363e-07, "loss": 0.1561, "step": 50430 }, { "epoch": 13.680499050718742, "grad_norm": 8.076054413487244, "learning_rate": 2.3397117629264165e-07, "loss": 0.1561, "step": 50440 }, { "epoch": 13.683211282885814, "grad_norm": 8.981319203257685, "learning_rate": 2.330180397253473e-07, "loss": 0.1547, "step": 50450 }, { "epoch": 13.685923515052888, "grad_norm": 7.629896701917481, "learning_rate": 2.320668022157191e-07, "loss": 0.1322, "step": 50460 }, { "epoch": 13.688635747219962, "grad_norm": 8.496903014884476, "learning_rate": 2.3111746414270775e-07, "loss": 0.1338, "step": 50470 }, { "epoch": 13.691347979387036, "grad_norm": 8.142087210401447, "learning_rate": 2.3017002588450853e-07, "loss": 0.1493, "step": 50480 }, { "epoch": 13.69406021155411, "grad_norm": 9.018688910079284, "learning_rate": 2.29224487818559e-07, "loss": 0.1485, "step": 50490 }, { "epoch": 13.696772443721182, "grad_norm": 4.5655954058649835, "learning_rate": 2.2828085032154058e-07, "loss": 0.1509, "step": 50500 }, { "epoch": 13.699484675888256, "grad_norm": 7.2738991388047705, "learning_rate": 2.27339113769377e-07, "loss": 0.1402, "step": 50510 }, { "epoch": 13.70219690805533, "grad_norm": 8.635843748206355, "learning_rate": 2.2639927853723376e-07, "loss": 0.1742, "step": 50520 }, { "epoch": 13.704909140222403, "grad_norm": 6.175224439594989, "learning_rate": 2.254613449995219e-07, "loss": 0.1373, "step": 50530 }, { "epoch": 13.707621372389477, "grad_norm": 10.657145268736752, "learning_rate": 2.2452531352989138e-07, "loss": 0.1422, "step": 50540 }, { "epoch": 13.71033360455655, "grad_norm": 9.119034439656366, "learning_rate": 2.2359118450123618e-07, "loss": 0.1608, "step": 50550 }, { "epoch": 13.713045836723623, "grad_norm": 6.864396484023607, "learning_rate": 2.226589582856925e-07, "loss": 0.1366, "step": 50560 }, { "epoch": 13.715758068890697, "grad_norm": 8.333895659192148, "learning_rate": 2.2172863525463829e-07, "loss": 0.1544, "step": 50570 }, { "epoch": 13.71847030105777, "grad_norm": 5.979378435878456, "learning_rate": 2.2080021577869314e-07, "loss": 0.1521, "step": 50580 }, { "epoch": 13.721182533224844, "grad_norm": 16.860601711732947, "learning_rate": 2.1987370022771737e-07, "loss": 0.1495, "step": 50590 }, { "epoch": 13.723894765391918, "grad_norm": 7.843357850343881, "learning_rate": 2.1894908897081457e-07, "loss": 0.1435, "step": 50600 }, { "epoch": 13.72660699755899, "grad_norm": 8.254740369322903, "learning_rate": 2.1802638237632901e-07, "loss": 0.1452, "step": 50610 }, { "epoch": 13.729319229726064, "grad_norm": 10.508237116298293, "learning_rate": 2.1710558081184608e-07, "loss": 0.1381, "step": 50620 }, { "epoch": 13.732031461893138, "grad_norm": 7.959745172604789, "learning_rate": 2.161866846441929e-07, "loss": 0.1458, "step": 50630 }, { "epoch": 13.734743694060212, "grad_norm": 9.073340948095073, "learning_rate": 2.1526969423943445e-07, "loss": 0.1711, "step": 50640 }, { "epoch": 13.737455926227286, "grad_norm": 9.72203872839206, "learning_rate": 2.1435460996288239e-07, "loss": 0.1416, "step": 50650 }, { "epoch": 13.740168158394358, "grad_norm": 8.013163582847167, "learning_rate": 2.1344143217908287e-07, "loss": 0.1434, "step": 50660 }, { "epoch": 13.742880390561432, "grad_norm": 8.846187074503646, "learning_rate": 2.12530161251826e-07, "loss": 0.1621, "step": 50670 }, { "epoch": 13.745592622728505, "grad_norm": 9.35456923042917, "learning_rate": 2.11620797544142e-07, "loss": 0.1418, "step": 50680 }, { "epoch": 13.74830485489558, "grad_norm": 6.748454934460878, "learning_rate": 2.1071334141830045e-07, "loss": 0.1695, "step": 50690 }, { "epoch": 13.751017087062653, "grad_norm": 7.866854044957723, "learning_rate": 2.0980779323581168e-07, "loss": 0.1387, "step": 50700 }, { "epoch": 13.753729319229727, "grad_norm": 9.256141505018714, "learning_rate": 2.0890415335742488e-07, "loss": 0.1532, "step": 50710 }, { "epoch": 13.756441551396799, "grad_norm": 9.22913062748607, "learning_rate": 2.080024221431304e-07, "loss": 0.1398, "step": 50720 }, { "epoch": 13.759153783563873, "grad_norm": 12.572524321969047, "learning_rate": 2.0710259995215643e-07, "loss": 0.1741, "step": 50730 }, { "epoch": 13.761866015730947, "grad_norm": 12.070169965631203, "learning_rate": 2.0620468714297349e-07, "loss": 0.128, "step": 50740 }, { "epoch": 13.76457824789802, "grad_norm": 10.040525808713717, "learning_rate": 2.0530868407328875e-07, "loss": 0.1404, "step": 50750 }, { "epoch": 13.767290480065094, "grad_norm": 6.396743933993828, "learning_rate": 2.0441459110004835e-07, "loss": 0.1465, "step": 50760 }, { "epoch": 13.770002712232166, "grad_norm": 9.936651577257571, "learning_rate": 2.0352240857944128e-07, "loss": 0.1482, "step": 50770 }, { "epoch": 13.77271494439924, "grad_norm": 7.0805268642610955, "learning_rate": 2.0263213686689042e-07, "loss": 0.1295, "step": 50780 }, { "epoch": 13.775427176566314, "grad_norm": 8.543308973805011, "learning_rate": 2.01743776317061e-07, "loss": 0.1405, "step": 50790 }, { "epoch": 13.778139408733388, "grad_norm": 11.114348584481156, "learning_rate": 2.00857327283856e-07, "loss": 0.145, "step": 50800 }, { "epoch": 13.780851640900462, "grad_norm": 14.858435826675583, "learning_rate": 1.9997279012041626e-07, "loss": 0.1618, "step": 50810 }, { "epoch": 13.783563873067534, "grad_norm": 10.341284523644733, "learning_rate": 1.9909016517912161e-07, "loss": 0.1533, "step": 50820 }, { "epoch": 13.786276105234608, "grad_norm": 12.572218171589121, "learning_rate": 1.982094528115891e-07, "loss": 0.1547, "step": 50830 }, { "epoch": 13.788988337401682, "grad_norm": 10.439685476402234, "learning_rate": 1.973306533686753e-07, "loss": 0.1606, "step": 50840 }, { "epoch": 13.791700569568755, "grad_norm": 9.53039992009724, "learning_rate": 1.9645376720047403e-07, "loss": 0.1557, "step": 50850 }, { "epoch": 13.79441280173583, "grad_norm": 10.347988369186167, "learning_rate": 1.9557879465631647e-07, "loss": 0.1459, "step": 50860 }, { "epoch": 13.797125033902901, "grad_norm": 9.63677495537171, "learning_rate": 1.9470573608477206e-07, "loss": 0.1423, "step": 50870 }, { "epoch": 13.799837266069975, "grad_norm": 9.633068737134819, "learning_rate": 1.9383459183364706e-07, "loss": 0.1449, "step": 50880 }, { "epoch": 13.802549498237049, "grad_norm": 9.733037051062098, "learning_rate": 1.929653622499872e-07, "loss": 0.1683, "step": 50890 }, { "epoch": 13.805261730404123, "grad_norm": 8.742647460531206, "learning_rate": 1.9209804768007213e-07, "loss": 0.1514, "step": 50900 }, { "epoch": 13.807973962571197, "grad_norm": 8.06270200140869, "learning_rate": 1.9123264846942102e-07, "loss": 0.1382, "step": 50910 }, { "epoch": 13.81068619473827, "grad_norm": 7.989305713452646, "learning_rate": 1.9036916496278867e-07, "loss": 0.1509, "step": 50920 }, { "epoch": 13.813398426905342, "grad_norm": 7.792813300904974, "learning_rate": 1.895075975041688e-07, "loss": 0.1415, "step": 50930 }, { "epoch": 13.816110659072416, "grad_norm": 10.228041888301375, "learning_rate": 1.8864794643678964e-07, "loss": 0.1439, "step": 50940 }, { "epoch": 13.81882289123949, "grad_norm": 9.26321200248154, "learning_rate": 1.877902121031161e-07, "loss": 0.1509, "step": 50950 }, { "epoch": 13.821535123406564, "grad_norm": 7.993841097491354, "learning_rate": 1.86934394844851e-07, "loss": 0.174, "step": 50960 }, { "epoch": 13.824247355573638, "grad_norm": 6.608207970152219, "learning_rate": 1.8608049500293212e-07, "loss": 0.1537, "step": 50970 }, { "epoch": 13.82695958774071, "grad_norm": 10.269599358600262, "learning_rate": 1.852285129175335e-07, "loss": 0.1315, "step": 50980 }, { "epoch": 13.829671819907784, "grad_norm": 6.631533911152521, "learning_rate": 1.8437844892806644e-07, "loss": 0.1309, "step": 50990 }, { "epoch": 13.832384052074858, "grad_norm": 9.41564590402637, "learning_rate": 1.8353030337317558e-07, "loss": 0.1527, "step": 51000 }, { "epoch": 13.835096284241931, "grad_norm": 11.23612876057676, "learning_rate": 1.8268407659074561e-07, "loss": 0.1501, "step": 51010 }, { "epoch": 13.837808516409005, "grad_norm": 10.666224548899004, "learning_rate": 1.8183976891789135e-07, "loss": 0.1512, "step": 51020 }, { "epoch": 13.840520748576077, "grad_norm": 6.276372747003758, "learning_rate": 1.8099738069096762e-07, "loss": 0.1481, "step": 51030 }, { "epoch": 13.843232980743151, "grad_norm": 7.1187933559969565, "learning_rate": 1.8015691224556098e-07, "loss": 0.149, "step": 51040 }, { "epoch": 13.845945212910225, "grad_norm": 9.702545104524829, "learning_rate": 1.7931836391649694e-07, "loss": 0.1423, "step": 51050 }, { "epoch": 13.848657445077299, "grad_norm": 6.509423373852284, "learning_rate": 1.7848173603783436e-07, "loss": 0.1387, "step": 51060 }, { "epoch": 13.851369677244373, "grad_norm": 10.226179328519663, "learning_rate": 1.77647028942865e-07, "loss": 0.1359, "step": 51070 }, { "epoch": 13.854081909411445, "grad_norm": 5.929846535904961, "learning_rate": 1.7681424296411788e-07, "loss": 0.1544, "step": 51080 }, { "epoch": 13.856794141578519, "grad_norm": 8.166002295460615, "learning_rate": 1.7598337843335645e-07, "loss": 0.1474, "step": 51090 }, { "epoch": 13.859506373745592, "grad_norm": 8.02293615615719, "learning_rate": 1.751544356815793e-07, "loss": 0.1482, "step": 51100 }, { "epoch": 13.862218605912666, "grad_norm": 9.011155588272983, "learning_rate": 1.7432741503901674e-07, "loss": 0.1614, "step": 51110 }, { "epoch": 13.86493083807974, "grad_norm": 6.129258211461797, "learning_rate": 1.7350231683513464e-07, "loss": 0.1403, "step": 51120 }, { "epoch": 13.867643070246814, "grad_norm": 6.042801519333062, "learning_rate": 1.7267914139863617e-07, "loss": 0.1399, "step": 51130 }, { "epoch": 13.870355302413886, "grad_norm": 8.492060334255354, "learning_rate": 1.7185788905745347e-07, "loss": 0.1476, "step": 51140 }, { "epoch": 13.87306753458096, "grad_norm": 10.360932692688454, "learning_rate": 1.7103856013875598e-07, "loss": 0.1564, "step": 51150 }, { "epoch": 13.875779766748034, "grad_norm": 10.505247431007184, "learning_rate": 1.702211549689442e-07, "loss": 0.1694, "step": 51160 }, { "epoch": 13.878491998915107, "grad_norm": 9.056295384965178, "learning_rate": 1.6940567387365658e-07, "loss": 0.1516, "step": 51170 }, { "epoch": 13.881204231082181, "grad_norm": 8.79389265806395, "learning_rate": 1.6859211717775993e-07, "loss": 0.1451, "step": 51180 }, { "epoch": 13.883916463249253, "grad_norm": 6.958897109102988, "learning_rate": 1.6778048520535773e-07, "loss": 0.1514, "step": 51190 }, { "epoch": 13.886628695416327, "grad_norm": 6.938108514513996, "learning_rate": 1.6697077827978526e-07, "loss": 0.1181, "step": 51200 }, { "epoch": 13.889340927583401, "grad_norm": 9.274960526273226, "learning_rate": 1.6616299672361224e-07, "loss": 0.1536, "step": 51210 }, { "epoch": 13.892053159750475, "grad_norm": 6.193877822158684, "learning_rate": 1.6535714085864018e-07, "loss": 0.1519, "step": 51220 }, { "epoch": 13.894765391917549, "grad_norm": 10.03920427314818, "learning_rate": 1.6455321100590393e-07, "loss": 0.145, "step": 51230 }, { "epoch": 13.897477624084623, "grad_norm": 9.946031996284766, "learning_rate": 1.6375120748566953e-07, "loss": 0.1324, "step": 51240 }, { "epoch": 13.900189856251695, "grad_norm": 11.08023920989169, "learning_rate": 1.6295113061744028e-07, "loss": 0.1455, "step": 51250 }, { "epoch": 13.902902088418768, "grad_norm": 6.464633036309455, "learning_rate": 1.6215298071994512e-07, "loss": 0.1403, "step": 51260 }, { "epoch": 13.905614320585842, "grad_norm": 5.977296345397785, "learning_rate": 1.6135675811115193e-07, "loss": 0.1443, "step": 51270 }, { "epoch": 13.908326552752916, "grad_norm": 9.58974460716802, "learning_rate": 1.6056246310825475e-07, "loss": 0.1527, "step": 51280 }, { "epoch": 13.91103878491999, "grad_norm": 12.03803940368802, "learning_rate": 1.5977009602768544e-07, "loss": 0.1507, "step": 51290 }, { "epoch": 13.913751017087062, "grad_norm": 9.987891813009226, "learning_rate": 1.5897965718510323e-07, "loss": 0.1444, "step": 51300 }, { "epoch": 13.916463249254136, "grad_norm": 7.047759535620499, "learning_rate": 1.5819114689540172e-07, "loss": 0.1581, "step": 51310 }, { "epoch": 13.91917548142121, "grad_norm": 5.834443931948584, "learning_rate": 1.5740456547270578e-07, "loss": 0.1518, "step": 51320 }, { "epoch": 13.921887713588283, "grad_norm": 4.582952018357969, "learning_rate": 1.566199132303714e-07, "loss": 0.1387, "step": 51330 }, { "epoch": 13.924599945755357, "grad_norm": 5.874701343589035, "learning_rate": 1.558371904809869e-07, "loss": 0.1381, "step": 51340 }, { "epoch": 13.92731217792243, "grad_norm": 7.313041839927052, "learning_rate": 1.5505639753637003e-07, "loss": 0.1486, "step": 51350 }, { "epoch": 13.930024410089503, "grad_norm": 8.228933040829132, "learning_rate": 1.5427753470757146e-07, "loss": 0.1329, "step": 51360 }, { "epoch": 13.932736642256577, "grad_norm": 8.393032711938291, "learning_rate": 1.5350060230487353e-07, "loss": 0.1368, "step": 51370 }, { "epoch": 13.93544887442365, "grad_norm": 7.454151297282343, "learning_rate": 1.5272560063778698e-07, "loss": 0.1505, "step": 51380 }, { "epoch": 13.938161106590725, "grad_norm": 8.547618536990774, "learning_rate": 1.519525300150565e-07, "loss": 0.1483, "step": 51390 }, { "epoch": 13.940873338757797, "grad_norm": 7.081378720686046, "learning_rate": 1.511813907446541e-07, "loss": 0.1494, "step": 51400 }, { "epoch": 13.94358557092487, "grad_norm": 6.709465861884068, "learning_rate": 1.5041218313378625e-07, "loss": 0.1382, "step": 51410 }, { "epoch": 13.946297803091944, "grad_norm": 9.558127130463388, "learning_rate": 1.4964490748888616e-07, "loss": 0.1378, "step": 51420 }, { "epoch": 13.949010035259018, "grad_norm": 8.865239325778527, "learning_rate": 1.488795641156199e-07, "loss": 0.1653, "step": 51430 }, { "epoch": 13.951722267426092, "grad_norm": 6.28399244729545, "learning_rate": 1.4811615331888241e-07, "loss": 0.1415, "step": 51440 }, { "epoch": 13.954434499593166, "grad_norm": 6.92492990308179, "learning_rate": 1.473546754027999e-07, "loss": 0.1424, "step": 51450 }, { "epoch": 13.957146731760238, "grad_norm": 6.023647796728843, "learning_rate": 1.4659513067072806e-07, "loss": 0.1387, "step": 51460 }, { "epoch": 13.959858963927312, "grad_norm": 6.023598682908134, "learning_rate": 1.4583751942525092e-07, "loss": 0.1347, "step": 51470 }, { "epoch": 13.962571196094386, "grad_norm": 8.754668355377877, "learning_rate": 1.4508184196818487e-07, "loss": 0.164, "step": 51480 }, { "epoch": 13.96528342826146, "grad_norm": 6.29873847824635, "learning_rate": 1.443280986005746e-07, "loss": 0.1501, "step": 51490 }, { "epoch": 13.967995660428533, "grad_norm": 6.479138822338982, "learning_rate": 1.4357628962269442e-07, "loss": 0.1486, "step": 51500 }, { "epoch": 13.970707892595605, "grad_norm": 10.059448152800071, "learning_rate": 1.4282641533404807e-07, "loss": 0.1684, "step": 51510 }, { "epoch": 13.97342012476268, "grad_norm": 7.752569949594296, "learning_rate": 1.420784760333671e-07, "loss": 0.1563, "step": 51520 }, { "epoch": 13.976132356929753, "grad_norm": 10.28799635243489, "learning_rate": 1.41332472018616e-07, "loss": 0.1401, "step": 51530 }, { "epoch": 13.978844589096827, "grad_norm": 9.095321004266452, "learning_rate": 1.4058840358698366e-07, "loss": 0.1609, "step": 51540 }, { "epoch": 13.9815568212639, "grad_norm": 6.629683523686832, "learning_rate": 1.3984627103489135e-07, "loss": 0.1306, "step": 51550 }, { "epoch": 13.984269053430975, "grad_norm": 9.66652983792498, "learning_rate": 1.3910607465798753e-07, "loss": 0.1472, "step": 51560 }, { "epoch": 13.986981285598047, "grad_norm": 8.477395181631088, "learning_rate": 1.383678147511497e-07, "loss": 0.1427, "step": 51570 }, { "epoch": 13.98969351776512, "grad_norm": 8.158376131916361, "learning_rate": 1.376314916084842e-07, "loss": 0.1421, "step": 51580 }, { "epoch": 13.992405749932194, "grad_norm": 8.615167737253199, "learning_rate": 1.368971055233248e-07, "loss": 0.1467, "step": 51590 }, { "epoch": 13.995117982099268, "grad_norm": 9.124379091141885, "learning_rate": 1.3616465678823464e-07, "loss": 0.1608, "step": 51600 }, { "epoch": 13.997830214266342, "grad_norm": 7.84245959699827, "learning_rate": 1.354341456950048e-07, "loss": 0.1673, "step": 51610 }, { "epoch": 14.000542446433414, "grad_norm": 8.092821006105662, "learning_rate": 1.347055725346541e-07, "loss": 0.1438, "step": 51620 }, { "epoch": 14.003254678600488, "grad_norm": 8.16158776157729, "learning_rate": 1.339789375974304e-07, "loss": 0.1234, "step": 51630 }, { "epoch": 14.005966910767562, "grad_norm": 5.6958874436439, "learning_rate": 1.3325424117280717e-07, "loss": 0.1182, "step": 51640 }, { "epoch": 14.008679142934636, "grad_norm": 6.864422075648273, "learning_rate": 1.3253148354948898e-07, "loss": 0.1249, "step": 51650 }, { "epoch": 14.01139137510171, "grad_norm": 6.018120120259627, "learning_rate": 1.3181066501540384e-07, "loss": 0.1435, "step": 51660 }, { "epoch": 14.014103607268781, "grad_norm": 5.438163980979239, "learning_rate": 1.31091785857711e-07, "loss": 0.1263, "step": 51670 }, { "epoch": 14.016815839435855, "grad_norm": 5.874481801442702, "learning_rate": 1.3037484636279575e-07, "loss": 0.1341, "step": 51680 }, { "epoch": 14.01952807160293, "grad_norm": 7.752912167038096, "learning_rate": 1.2965984681626965e-07, "loss": 0.1352, "step": 51690 }, { "epoch": 14.022240303770003, "grad_norm": 5.815083901155653, "learning_rate": 1.2894678750297318e-07, "loss": 0.1224, "step": 51700 }, { "epoch": 14.024952535937077, "grad_norm": 7.397514575140568, "learning_rate": 1.2823566870697245e-07, "loss": 0.1283, "step": 51710 }, { "epoch": 14.027664768104149, "grad_norm": 6.199071522652719, "learning_rate": 1.275264907115614e-07, "loss": 0.1206, "step": 51720 }, { "epoch": 14.030377000271223, "grad_norm": 6.427113596577363, "learning_rate": 1.2681925379925964e-07, "loss": 0.1219, "step": 51730 }, { "epoch": 14.033089232438297, "grad_norm": 8.290436965540996, "learning_rate": 1.261139582518156e-07, "loss": 0.1428, "step": 51740 }, { "epoch": 14.03580146460537, "grad_norm": 5.658520591088887, "learning_rate": 1.2541060435020292e-07, "loss": 0.1383, "step": 51750 }, { "epoch": 14.038513696772444, "grad_norm": 6.941584302434458, "learning_rate": 1.2470919237462075e-07, "loss": 0.1406, "step": 51760 }, { "epoch": 14.041225928939518, "grad_norm": 6.261985763440922, "learning_rate": 1.2400972260449727e-07, "loss": 0.1311, "step": 51770 }, { "epoch": 14.04393816110659, "grad_norm": 9.508209331559343, "learning_rate": 1.2331219531848405e-07, "loss": 0.1207, "step": 51780 }, { "epoch": 14.046650393273664, "grad_norm": 4.546615632293128, "learning_rate": 1.226166107944604e-07, "loss": 0.1218, "step": 51790 }, { "epoch": 14.049362625440738, "grad_norm": 5.6709284545411345, "learning_rate": 1.2192296930953196e-07, "loss": 0.1199, "step": 51800 }, { "epoch": 14.052074857607812, "grad_norm": 7.385695304203951, "learning_rate": 1.2123127114002986e-07, "loss": 0.1227, "step": 51810 }, { "epoch": 14.054787089774885, "grad_norm": 9.002591015956254, "learning_rate": 1.205415165615109e-07, "loss": 0.1255, "step": 51820 }, { "epoch": 14.057499321941958, "grad_norm": 4.8076157745797525, "learning_rate": 1.19853705848757e-07, "loss": 0.1302, "step": 51830 }, { "epoch": 14.060211554109031, "grad_norm": 8.391951335205556, "learning_rate": 1.1916783927577669e-07, "loss": 0.1276, "step": 51840 }, { "epoch": 14.062923786276105, "grad_norm": 6.782087265022005, "learning_rate": 1.1848391711580421e-07, "loss": 0.1205, "step": 51850 }, { "epoch": 14.065636018443179, "grad_norm": 5.903924145338906, "learning_rate": 1.1780193964129827e-07, "loss": 0.113, "step": 51860 }, { "epoch": 14.068348250610253, "grad_norm": 5.376631064644241, "learning_rate": 1.1712190712394322e-07, "loss": 0.1168, "step": 51870 }, { "epoch": 14.071060482777325, "grad_norm": 6.386620194011005, "learning_rate": 1.1644381983464848e-07, "loss": 0.1288, "step": 51880 }, { "epoch": 14.073772714944399, "grad_norm": 6.90731234806098, "learning_rate": 1.1576767804354905e-07, "loss": 0.1241, "step": 51890 }, { "epoch": 14.076484947111473, "grad_norm": 6.748306338004663, "learning_rate": 1.1509348202000448e-07, "loss": 0.1207, "step": 51900 }, { "epoch": 14.079197179278546, "grad_norm": 5.512785300351578, "learning_rate": 1.144212320325988e-07, "loss": 0.1154, "step": 51910 }, { "epoch": 14.08190941144562, "grad_norm": 9.18645370311189, "learning_rate": 1.1375092834914115e-07, "loss": 0.138, "step": 51920 }, { "epoch": 14.084621643612694, "grad_norm": 5.185594428333865, "learning_rate": 1.1308257123666622e-07, "loss": 0.1323, "step": 51930 }, { "epoch": 14.087333875779766, "grad_norm": 7.531253009317517, "learning_rate": 1.1241616096143159e-07, "loss": 0.1464, "step": 51940 }, { "epoch": 14.09004610794684, "grad_norm": 4.77770011675362, "learning_rate": 1.1175169778891992e-07, "loss": 0.1306, "step": 51950 }, { "epoch": 14.092758340113914, "grad_norm": 6.7880615368078345, "learning_rate": 1.1108918198383888e-07, "loss": 0.1194, "step": 51960 }, { "epoch": 14.095470572280988, "grad_norm": 5.92868692840815, "learning_rate": 1.1042861381011904e-07, "loss": 0.1265, "step": 51970 }, { "epoch": 14.098182804448061, "grad_norm": 6.615085933174338, "learning_rate": 1.09769993530916e-07, "loss": 0.1331, "step": 51980 }, { "epoch": 14.100895036615134, "grad_norm": 4.965640624038981, "learning_rate": 1.0911332140861042e-07, "loss": 0.127, "step": 51990 }, { "epoch": 14.103607268782207, "grad_norm": 5.587841927376556, "learning_rate": 1.0845859770480305e-07, "loss": 0.13, "step": 52000 }, { "epoch": 14.106319500949281, "grad_norm": 5.73129230980022, "learning_rate": 1.0780582268032358e-07, "loss": 0.1192, "step": 52010 }, { "epoch": 14.109031733116355, "grad_norm": 5.622672213778571, "learning_rate": 1.0715499659522122e-07, "loss": 0.1097, "step": 52020 }, { "epoch": 14.111743965283429, "grad_norm": 8.341114522919852, "learning_rate": 1.0650611970877134e-07, "loss": 0.1317, "step": 52030 }, { "epoch": 14.114456197450501, "grad_norm": 5.814519703680938, "learning_rate": 1.058591922794705e-07, "loss": 0.1191, "step": 52040 }, { "epoch": 14.117168429617575, "grad_norm": 6.078996155967206, "learning_rate": 1.0521421456504144e-07, "loss": 0.1165, "step": 52050 }, { "epoch": 14.119880661784649, "grad_norm": 7.012200042381726, "learning_rate": 1.045711868224275e-07, "loss": 0.1166, "step": 52060 }, { "epoch": 14.122592893951722, "grad_norm": 5.593033892648063, "learning_rate": 1.0393010930779713e-07, "loss": 0.1203, "step": 52070 }, { "epoch": 14.125305126118796, "grad_norm": 6.129573095331448, "learning_rate": 1.0329098227654044e-07, "loss": 0.1198, "step": 52080 }, { "epoch": 14.12801735828587, "grad_norm": 5.774055589888153, "learning_rate": 1.0265380598327157e-07, "loss": 0.1143, "step": 52090 }, { "epoch": 14.130729590452942, "grad_norm": 6.0474372461363375, "learning_rate": 1.0201858068182746e-07, "loss": 0.1195, "step": 52100 }, { "epoch": 14.133441822620016, "grad_norm": 6.759050704682038, "learning_rate": 1.013853066252668e-07, "loss": 0.1426, "step": 52110 }, { "epoch": 14.13615405478709, "grad_norm": 6.10684666304033, "learning_rate": 1.0075398406587167e-07, "loss": 0.1271, "step": 52120 }, { "epoch": 14.138866286954164, "grad_norm": 7.035984162543115, "learning_rate": 1.0012461325514756e-07, "loss": 0.1199, "step": 52130 }, { "epoch": 14.141578519121238, "grad_norm": 6.790043544970545, "learning_rate": 9.949719444382055e-08, "loss": 0.1314, "step": 52140 }, { "epoch": 14.14429075128831, "grad_norm": 8.32781954722367, "learning_rate": 9.887172788184074e-08, "loss": 0.1328, "step": 52150 }, { "epoch": 14.147002983455383, "grad_norm": 5.483800777750003, "learning_rate": 9.824821381837879e-08, "loss": 0.1339, "step": 52160 }, { "epoch": 14.149715215622457, "grad_norm": 5.510243188211718, "learning_rate": 9.762665250183045e-08, "loss": 0.117, "step": 52170 }, { "epoch": 14.152427447789531, "grad_norm": 5.614607903377166, "learning_rate": 9.700704417980988e-08, "loss": 0.136, "step": 52180 }, { "epoch": 14.155139679956605, "grad_norm": 6.167669772649386, "learning_rate": 9.638938909915574e-08, "loss": 0.1381, "step": 52190 }, { "epoch": 14.157851912123677, "grad_norm": 6.789573987941597, "learning_rate": 9.577368750592785e-08, "loss": 0.1183, "step": 52200 }, { "epoch": 14.16056414429075, "grad_norm": 7.237046789362337, "learning_rate": 9.515993964540782e-08, "loss": 0.1184, "step": 52210 }, { "epoch": 14.163276376457825, "grad_norm": 7.208026346615744, "learning_rate": 9.454814576209892e-08, "loss": 0.1376, "step": 52220 }, { "epoch": 14.165988608624899, "grad_norm": 6.943565681241819, "learning_rate": 9.393830609972565e-08, "loss": 0.1236, "step": 52230 }, { "epoch": 14.168700840791972, "grad_norm": 7.394697243368423, "learning_rate": 9.333042090123423e-08, "loss": 0.1263, "step": 52240 }, { "epoch": 14.171413072959044, "grad_norm": 6.523462851565584, "learning_rate": 9.272449040879317e-08, "loss": 0.135, "step": 52250 }, { "epoch": 14.174125305126118, "grad_norm": 9.110411690550709, "learning_rate": 9.212051486379104e-08, "loss": 0.1295, "step": 52260 }, { "epoch": 14.176837537293192, "grad_norm": 5.432016482944168, "learning_rate": 9.151849450683814e-08, "loss": 0.133, "step": 52270 }, { "epoch": 14.179549769460266, "grad_norm": 7.404683497267898, "learning_rate": 9.091842957776487e-08, "loss": 0.1334, "step": 52280 }, { "epoch": 14.18226200162734, "grad_norm": 7.89337928964462, "learning_rate": 9.032032031562443e-08, "loss": 0.1222, "step": 52290 }, { "epoch": 14.184974233794414, "grad_norm": 9.192335086974412, "learning_rate": 8.972416695869012e-08, "loss": 0.135, "step": 52300 }, { "epoch": 14.187686465961486, "grad_norm": 5.732470813054365, "learning_rate": 8.912996974445475e-08, "loss": 0.117, "step": 52310 }, { "epoch": 14.19039869812856, "grad_norm": 5.577202538417793, "learning_rate": 8.853772890963452e-08, "loss": 0.1178, "step": 52320 }, { "epoch": 14.193110930295633, "grad_norm": 5.730114135908758, "learning_rate": 8.794744469016403e-08, "loss": 0.1222, "step": 52330 }, { "epoch": 14.195823162462707, "grad_norm": 6.105789304877115, "learning_rate": 8.735911732119961e-08, "loss": 0.1256, "step": 52340 }, { "epoch": 14.198535394629781, "grad_norm": 4.412862068919845, "learning_rate": 8.677274703711657e-08, "loss": 0.1142, "step": 52350 }, { "epoch": 14.201247626796853, "grad_norm": 6.487192092324537, "learning_rate": 8.618833407151195e-08, "loss": 0.119, "step": 52360 }, { "epoch": 14.203959858963927, "grad_norm": 7.056795820887286, "learning_rate": 8.560587865720337e-08, "loss": 0.1196, "step": 52370 }, { "epoch": 14.206672091131, "grad_norm": 7.80068764735899, "learning_rate": 8.502538102622693e-08, "loss": 0.1304, "step": 52380 }, { "epoch": 14.209384323298075, "grad_norm": 6.053398734652391, "learning_rate": 8.444684140984094e-08, "loss": 0.1397, "step": 52390 }, { "epoch": 14.212096555465148, "grad_norm": 6.325136931571901, "learning_rate": 8.38702600385205e-08, "loss": 0.1253, "step": 52400 }, { "epoch": 14.214808787632222, "grad_norm": 7.975034120581596, "learning_rate": 8.32956371419652e-08, "loss": 0.1161, "step": 52410 }, { "epoch": 14.217521019799294, "grad_norm": 7.224191397393836, "learning_rate": 8.272297294908971e-08, "loss": 0.1217, "step": 52420 }, { "epoch": 14.220233251966368, "grad_norm": 5.221071640170199, "learning_rate": 8.215226768803098e-08, "loss": 0.1239, "step": 52430 }, { "epoch": 14.222945484133442, "grad_norm": 8.01112893031923, "learning_rate": 8.158352158614491e-08, "loss": 0.1307, "step": 52440 }, { "epoch": 14.225657716300516, "grad_norm": 6.522700690356823, "learning_rate": 8.101673487000805e-08, "loss": 0.1292, "step": 52450 }, { "epoch": 14.22836994846759, "grad_norm": 5.8446531239440835, "learning_rate": 8.045190776541478e-08, "loss": 0.1088, "step": 52460 }, { "epoch": 14.231082180634662, "grad_norm": 6.399217069939955, "learning_rate": 7.9889040497379e-08, "loss": 0.1349, "step": 52470 }, { "epoch": 14.233794412801736, "grad_norm": 5.997522174990348, "learning_rate": 7.932813329013467e-08, "loss": 0.1266, "step": 52480 }, { "epoch": 14.23650664496881, "grad_norm": 7.314072361384652, "learning_rate": 7.876918636713471e-08, "loss": 0.1448, "step": 52490 }, { "epoch": 14.239218877135883, "grad_norm": 7.848057918414941, "learning_rate": 7.821219995105045e-08, "loss": 0.1473, "step": 52500 }, { "epoch": 14.241931109302957, "grad_norm": 6.811895676389257, "learning_rate": 7.765717426377329e-08, "loss": 0.1218, "step": 52510 }, { "epoch": 14.244643341470029, "grad_norm": 9.1937891418902, "learning_rate": 7.710410952641189e-08, "loss": 0.1401, "step": 52520 }, { "epoch": 14.247355573637103, "grad_norm": 11.053982350150932, "learning_rate": 7.655300595929616e-08, "loss": 0.1502, "step": 52530 }, { "epoch": 14.250067805804177, "grad_norm": 7.123579659698091, "learning_rate": 7.600386378197211e-08, "loss": 0.1311, "step": 52540 }, { "epoch": 14.25278003797125, "grad_norm": 6.5097636200320865, "learning_rate": 7.545668321320587e-08, "loss": 0.1229, "step": 52550 }, { "epoch": 14.255492270138324, "grad_norm": 5.467427828884543, "learning_rate": 7.491146447098141e-08, "loss": 0.1205, "step": 52560 }, { "epoch": 14.258204502305396, "grad_norm": 6.626902174793014, "learning_rate": 7.436820777250275e-08, "loss": 0.1317, "step": 52570 }, { "epoch": 14.26091673447247, "grad_norm": 6.477314053711326, "learning_rate": 7.382691333419012e-08, "loss": 0.1392, "step": 52580 }, { "epoch": 14.263628966639544, "grad_norm": 4.532334471530733, "learning_rate": 7.328758137168268e-08, "loss": 0.1188, "step": 52590 }, { "epoch": 14.266341198806618, "grad_norm": 5.5246102278549705, "learning_rate": 7.275021209983857e-08, "loss": 0.131, "step": 52600 }, { "epoch": 14.269053430973692, "grad_norm": 6.604936796272448, "learning_rate": 7.221480573273376e-08, "loss": 0.1363, "step": 52610 }, { "epoch": 14.271765663140766, "grad_norm": 5.862734503521474, "learning_rate": 7.168136248366154e-08, "loss": 0.1265, "step": 52620 }, { "epoch": 14.274477895307838, "grad_norm": 5.681980537443351, "learning_rate": 7.11498825651341e-08, "loss": 0.1243, "step": 52630 }, { "epoch": 14.277190127474912, "grad_norm": 5.806930975493854, "learning_rate": 7.062036618888046e-08, "loss": 0.1213, "step": 52640 }, { "epoch": 14.279902359641985, "grad_norm": 6.177321084013237, "learning_rate": 7.009281356584852e-08, "loss": 0.1156, "step": 52650 }, { "epoch": 14.28261459180906, "grad_norm": 4.908356826712993, "learning_rate": 6.956722490620238e-08, "loss": 0.1304, "step": 52660 }, { "epoch": 14.285326823976133, "grad_norm": 6.967429167994589, "learning_rate": 6.904360041932567e-08, "loss": 0.1194, "step": 52670 }, { "epoch": 14.288039056143205, "grad_norm": 7.308934891271957, "learning_rate": 6.85219403138182e-08, "loss": 0.1388, "step": 52680 }, { "epoch": 14.290751288310279, "grad_norm": 6.095124691968673, "learning_rate": 6.800224479749706e-08, "loss": 0.1224, "step": 52690 }, { "epoch": 14.293463520477353, "grad_norm": 5.218883487526973, "learning_rate": 6.74845140773983e-08, "loss": 0.1307, "step": 52700 }, { "epoch": 14.296175752644427, "grad_norm": 6.920712744930396, "learning_rate": 6.696874835977307e-08, "loss": 0.1322, "step": 52710 }, { "epoch": 14.2988879848115, "grad_norm": 6.389592585590594, "learning_rate": 6.645494785009088e-08, "loss": 0.1204, "step": 52720 }, { "epoch": 14.301600216978573, "grad_norm": 6.978472594930368, "learning_rate": 6.594311275303855e-08, "loss": 0.1288, "step": 52730 }, { "epoch": 14.304312449145646, "grad_norm": 7.5713923841321025, "learning_rate": 6.54332432725191e-08, "loss": 0.1278, "step": 52740 }, { "epoch": 14.30702468131272, "grad_norm": 8.661260975611345, "learning_rate": 6.492533961165447e-08, "loss": 0.1272, "step": 52750 }, { "epoch": 14.309736913479794, "grad_norm": 6.0433091331198465, "learning_rate": 6.441940197277952e-08, "loss": 0.1214, "step": 52760 }, { "epoch": 14.312449145646868, "grad_norm": 6.559536909484446, "learning_rate": 6.391543055745075e-08, "loss": 0.1222, "step": 52770 }, { "epoch": 14.315161377813942, "grad_norm": 6.454205444489659, "learning_rate": 6.341342556643704e-08, "loss": 0.124, "step": 52780 }, { "epoch": 14.317873609981014, "grad_norm": 6.0372061095355996, "learning_rate": 6.291338719972729e-08, "loss": 0.114, "step": 52790 }, { "epoch": 14.320585842148088, "grad_norm": 6.821418897463696, "learning_rate": 6.241531565652437e-08, "loss": 0.1297, "step": 52800 }, { "epoch": 14.323298074315161, "grad_norm": 6.429926375300048, "learning_rate": 6.191921113524957e-08, "loss": 0.1438, "step": 52810 }, { "epoch": 14.326010306482235, "grad_norm": 5.572923703614413, "learning_rate": 6.142507383353925e-08, "loss": 0.1242, "step": 52820 }, { "epoch": 14.32872253864931, "grad_norm": 5.979122436561509, "learning_rate": 6.093290394824648e-08, "loss": 0.1204, "step": 52830 }, { "epoch": 14.331434770816381, "grad_norm": 8.397179129566778, "learning_rate": 6.044270167544053e-08, "loss": 0.1389, "step": 52840 }, { "epoch": 14.334147002983455, "grad_norm": 6.106289056089121, "learning_rate": 5.995446721040687e-08, "loss": 0.1195, "step": 52850 }, { "epoch": 14.336859235150529, "grad_norm": 7.690995335704652, "learning_rate": 5.94682007476477e-08, "loss": 0.1145, "step": 52860 }, { "epoch": 14.339571467317603, "grad_norm": 7.824397953657134, "learning_rate": 5.898390248088026e-08, "loss": 0.1367, "step": 52870 }, { "epoch": 14.342283699484677, "grad_norm": 6.942290819412055, "learning_rate": 5.850157260303691e-08, "loss": 0.1226, "step": 52880 }, { "epoch": 14.344995931651749, "grad_norm": 5.913822562815095, "learning_rate": 5.802121130626892e-08, "loss": 0.1127, "step": 52890 }, { "epoch": 14.347708163818822, "grad_norm": 7.10618897886289, "learning_rate": 5.7542818781939904e-08, "loss": 0.1265, "step": 52900 }, { "epoch": 14.350420395985896, "grad_norm": 7.831533697638606, "learning_rate": 5.706639522063129e-08, "loss": 0.1288, "step": 52910 }, { "epoch": 14.35313262815297, "grad_norm": 7.186983870507468, "learning_rate": 5.659194081213848e-08, "loss": 0.1281, "step": 52920 }, { "epoch": 14.355844860320044, "grad_norm": 6.426264984459145, "learning_rate": 5.6119455745474726e-08, "loss": 0.1326, "step": 52930 }, { "epoch": 14.358557092487118, "grad_norm": 6.62325539823425, "learning_rate": 5.564894020886669e-08, "loss": 0.1077, "step": 52940 }, { "epoch": 14.36126932465419, "grad_norm": 5.689438080504189, "learning_rate": 5.518039438975664e-08, "loss": 0.1268, "step": 52950 }, { "epoch": 14.363981556821264, "grad_norm": 5.61002914498547, "learning_rate": 5.4713818474803615e-08, "loss": 0.1125, "step": 52960 }, { "epoch": 14.366693788988337, "grad_norm": 6.544916024762375, "learning_rate": 5.424921264988003e-08, "loss": 0.1292, "step": 52970 }, { "epoch": 14.369406021155411, "grad_norm": 11.214686683649989, "learning_rate": 5.378657710007507e-08, "loss": 0.133, "step": 52980 }, { "epoch": 14.372118253322485, "grad_norm": 5.870397168833424, "learning_rate": 5.332591200969184e-08, "loss": 0.1352, "step": 52990 }, { "epoch": 14.374830485489557, "grad_norm": 7.262378626047544, "learning_rate": 5.2867217562247996e-08, "loss": 0.1332, "step": 53000 }, { "epoch": 14.377542717656631, "grad_norm": 7.514540008672622, "learning_rate": 5.241049394047904e-08, "loss": 0.1223, "step": 53010 }, { "epoch": 14.380254949823705, "grad_norm": 7.823247264377376, "learning_rate": 5.195574132633163e-08, "loss": 0.1251, "step": 53020 }, { "epoch": 14.382967181990779, "grad_norm": 6.687226123265081, "learning_rate": 5.1502959900970316e-08, "loss": 0.1278, "step": 53030 }, { "epoch": 14.385679414157853, "grad_norm": 5.500401636260465, "learning_rate": 5.105214984477136e-08, "loss": 0.1258, "step": 53040 }, { "epoch": 14.388391646324925, "grad_norm": 5.586669878686618, "learning_rate": 5.0603311337327764e-08, "loss": 0.1406, "step": 53050 }, { "epoch": 14.391103878491998, "grad_norm": 5.2793056245114895, "learning_rate": 5.015644455744817e-08, "loss": 0.1201, "step": 53060 }, { "epoch": 14.393816110659072, "grad_norm": 6.380619516943116, "learning_rate": 4.971154968315184e-08, "loss": 0.1295, "step": 53070 }, { "epoch": 14.396528342826146, "grad_norm": 6.212683875804313, "learning_rate": 4.9268626891676443e-08, "loss": 0.1295, "step": 53080 }, { "epoch": 14.39924057499322, "grad_norm": 6.499659229797424, "learning_rate": 4.882767635947139e-08, "loss": 0.1291, "step": 53090 }, { "epoch": 14.401952807160292, "grad_norm": 9.358752169092348, "learning_rate": 4.838869826220283e-08, "loss": 0.1341, "step": 53100 }, { "epoch": 14.404665039327366, "grad_norm": 7.465070849055684, "learning_rate": 4.7951692774748094e-08, "loss": 0.1352, "step": 53110 }, { "epoch": 14.40737727149444, "grad_norm": 6.314104031804899, "learning_rate": 4.751666007120015e-08, "loss": 0.1229, "step": 53120 }, { "epoch": 14.410089503661514, "grad_norm": 5.692628223540839, "learning_rate": 4.7083600324868136e-08, "loss": 0.1188, "step": 53130 }, { "epoch": 14.412801735828587, "grad_norm": 6.166593300214851, "learning_rate": 4.665251370827184e-08, "loss": 0.1301, "step": 53140 }, { "epoch": 14.415513967995661, "grad_norm": 5.992166852386888, "learning_rate": 4.6223400393146655e-08, "loss": 0.1293, "step": 53150 }, { "epoch": 14.418226200162733, "grad_norm": 5.552860078462938, "learning_rate": 4.579626055044084e-08, "loss": 0.1298, "step": 53160 }, { "epoch": 14.420938432329807, "grad_norm": 4.463790012479482, "learning_rate": 4.537109435031828e-08, "loss": 0.1326, "step": 53170 }, { "epoch": 14.423650664496881, "grad_norm": 7.4448753878783025, "learning_rate": 4.49479019621557e-08, "loss": 0.1303, "step": 53180 }, { "epoch": 14.426362896663955, "grad_norm": 6.369093078542063, "learning_rate": 4.4526683554542704e-08, "loss": 0.1353, "step": 53190 }, { "epoch": 14.429075128831029, "grad_norm": 5.66759359745127, "learning_rate": 4.4107439295283385e-08, "loss": 0.1175, "step": 53200 }, { "epoch": 14.4317873609981, "grad_norm": 6.078309950231089, "learning_rate": 4.3690169351394696e-08, "loss": 0.1257, "step": 53210 }, { "epoch": 14.434499593165175, "grad_norm": 7.033047594624968, "learning_rate": 4.327487388910867e-08, "loss": 0.1227, "step": 53220 }, { "epoch": 14.437211825332248, "grad_norm": 9.47650673011404, "learning_rate": 4.286155307386852e-08, "loss": 0.1258, "step": 53230 }, { "epoch": 14.439924057499322, "grad_norm": 7.373164066549338, "learning_rate": 4.245020707033198e-08, "loss": 0.1189, "step": 53240 }, { "epoch": 14.442636289666396, "grad_norm": 6.971184404112092, "learning_rate": 4.20408360423713e-08, "loss": 0.1176, "step": 53250 }, { "epoch": 14.44534852183347, "grad_norm": 6.293941689980358, "learning_rate": 4.163344015306881e-08, "loss": 0.1337, "step": 53260 }, { "epoch": 14.448060754000542, "grad_norm": 6.966482594671781, "learning_rate": 4.1228019564723556e-08, "loss": 0.1386, "step": 53270 }, { "epoch": 14.450772986167616, "grad_norm": 5.982216244698459, "learning_rate": 4.082457443884358e-08, "loss": 0.1175, "step": 53280 }, { "epoch": 14.45348521833469, "grad_norm": 7.791458936819274, "learning_rate": 4.042310493615531e-08, "loss": 0.1352, "step": 53290 }, { "epoch": 14.456197450501763, "grad_norm": 5.5881980053532025, "learning_rate": 4.0023611216592485e-08, "loss": 0.1207, "step": 53300 }, { "epoch": 14.458909682668837, "grad_norm": 6.792143364254526, "learning_rate": 3.962609343930557e-08, "loss": 0.1414, "step": 53310 }, { "epoch": 14.46162191483591, "grad_norm": 5.315191899122436, "learning_rate": 3.923055176265622e-08, "loss": 0.1246, "step": 53320 }, { "epoch": 14.464334147002983, "grad_norm": 6.778416449390237, "learning_rate": 3.883698634422006e-08, "loss": 0.1264, "step": 53330 }, { "epoch": 14.467046379170057, "grad_norm": 6.263758115694987, "learning_rate": 3.8445397340783896e-08, "loss": 0.1207, "step": 53340 }, { "epoch": 14.46975861133713, "grad_norm": 4.1570878170376435, "learning_rate": 3.805578490834794e-08, "loss": 0.1213, "step": 53350 }, { "epoch": 14.472470843504205, "grad_norm": 10.069326449273055, "learning_rate": 3.766814920212525e-08, "loss": 0.1226, "step": 53360 }, { "epoch": 14.475183075671277, "grad_norm": 7.950616440014156, "learning_rate": 3.728249037654119e-08, "loss": 0.1352, "step": 53370 }, { "epoch": 14.47789530783835, "grad_norm": 6.967750961754619, "learning_rate": 3.689880858523287e-08, "loss": 0.1265, "step": 53380 }, { "epoch": 14.480607540005424, "grad_norm": 7.365096873946075, "learning_rate": 3.651710398105135e-08, "loss": 0.1267, "step": 53390 }, { "epoch": 14.483319772172498, "grad_norm": 6.575399339563677, "learning_rate": 3.613737671605832e-08, "loss": 0.1193, "step": 53400 }, { "epoch": 14.486032004339572, "grad_norm": 5.9426958268796515, "learning_rate": 3.57596269415289e-08, "loss": 0.1236, "step": 53410 }, { "epoch": 14.488744236506644, "grad_norm": 5.047405206772884, "learning_rate": 3.538385480794992e-08, "loss": 0.1377, "step": 53420 }, { "epoch": 14.491456468673718, "grad_norm": 6.114548664219787, "learning_rate": 3.501006046502053e-08, "loss": 0.1202, "step": 53430 }, { "epoch": 14.494168700840792, "grad_norm": 8.233522567242924, "learning_rate": 3.4638244061652174e-08, "loss": 0.1159, "step": 53440 }, { "epoch": 14.496880933007866, "grad_norm": 7.238389547308745, "learning_rate": 3.426840574596746e-08, "loss": 0.1235, "step": 53450 }, { "epoch": 14.49959316517494, "grad_norm": 6.35084986884011, "learning_rate": 3.390054566530243e-08, "loss": 0.13, "step": 53460 }, { "epoch": 14.502305397342013, "grad_norm": 4.325135535823192, "learning_rate": 3.353466396620375e-08, "loss": 0.1107, "step": 53470 }, { "epoch": 14.505017629509085, "grad_norm": 7.1380561844212345, "learning_rate": 3.317076079443038e-08, "loss": 0.1187, "step": 53480 }, { "epoch": 14.50772986167616, "grad_norm": 8.937207488176515, "learning_rate": 3.280883629495302e-08, "loss": 0.1261, "step": 53490 }, { "epoch": 14.510442093843233, "grad_norm": 5.31100809894644, "learning_rate": 3.2448890611954106e-08, "loss": 0.1171, "step": 53500 }, { "epoch": 14.513154326010307, "grad_norm": 5.349731407857855, "learning_rate": 3.209092388882895e-08, "loss": 0.1204, "step": 53510 }, { "epoch": 14.51586655817738, "grad_norm": 6.513873721978427, "learning_rate": 3.173493626818125e-08, "loss": 0.1285, "step": 53520 }, { "epoch": 14.518578790344453, "grad_norm": 8.370365720243845, "learning_rate": 3.1380927891830893e-08, "loss": 0.1308, "step": 53530 }, { "epoch": 14.521291022511527, "grad_norm": 8.570849926820662, "learning_rate": 3.102889890080452e-08, "loss": 0.1237, "step": 53540 }, { "epoch": 14.5240032546786, "grad_norm": 6.710859212597673, "learning_rate": 3.067884943534438e-08, "loss": 0.1379, "step": 53550 }, { "epoch": 14.526715486845674, "grad_norm": 8.452612942750285, "learning_rate": 3.0330779634901163e-08, "loss": 0.1354, "step": 53560 }, { "epoch": 14.529427719012748, "grad_norm": 7.135734684289642, "learning_rate": 2.998468963813783e-08, "loss": 0.1286, "step": 53570 }, { "epoch": 14.532139951179822, "grad_norm": 5.737474921708063, "learning_rate": 2.964057958292965e-08, "loss": 0.1254, "step": 53580 }, { "epoch": 14.534852183346894, "grad_norm": 6.384284967780397, "learning_rate": 2.9298449606361968e-08, "loss": 0.126, "step": 53590 }, { "epoch": 14.537564415513968, "grad_norm": 5.465619387173103, "learning_rate": 2.8958299844731307e-08, "loss": 0.1293, "step": 53600 }, { "epoch": 14.540276647681042, "grad_norm": 4.996920761484813, "learning_rate": 2.8620130433545945e-08, "loss": 0.1261, "step": 53610 }, { "epoch": 14.542988879848116, "grad_norm": 4.911406275793165, "learning_rate": 2.8283941507524783e-08, "loss": 0.1147, "step": 53620 }, { "epoch": 14.545701112015188, "grad_norm": 6.629065255169855, "learning_rate": 2.7949733200598462e-08, "loss": 0.1375, "step": 53630 }, { "epoch": 14.548413344182261, "grad_norm": 9.716299812652345, "learning_rate": 2.7617505645906596e-08, "loss": 0.1396, "step": 53640 }, { "epoch": 14.551125576349335, "grad_norm": 5.663956077010168, "learning_rate": 2.72872589758022e-08, "loss": 0.1392, "step": 53650 }, { "epoch": 14.553837808516409, "grad_norm": 8.407647552971207, "learning_rate": 2.6958993321847815e-08, "loss": 0.1413, "step": 53660 }, { "epoch": 14.556550040683483, "grad_norm": 10.09004847914302, "learning_rate": 2.6632708814817165e-08, "loss": 0.1285, "step": 53670 }, { "epoch": 14.559262272850557, "grad_norm": 5.315202714737616, "learning_rate": 2.6308405584694608e-08, "loss": 0.1188, "step": 53680 }, { "epoch": 14.561974505017629, "grad_norm": 8.004372103376653, "learning_rate": 2.5986083760675128e-08, "loss": 0.1368, "step": 53690 }, { "epoch": 14.564686737184703, "grad_norm": 6.143418232113829, "learning_rate": 2.566574347116435e-08, "loss": 0.1119, "step": 53700 }, { "epoch": 14.567398969351776, "grad_norm": 6.668416742462567, "learning_rate": 2.5347384843778523e-08, "loss": 0.1138, "step": 53710 }, { "epoch": 14.57011120151885, "grad_norm": 6.826725009076117, "learning_rate": 2.5031008005345082e-08, "loss": 0.1229, "step": 53720 }, { "epoch": 14.572823433685924, "grad_norm": 5.548034820424933, "learning_rate": 2.471661308190043e-08, "loss": 0.126, "step": 53730 }, { "epoch": 14.575535665852996, "grad_norm": 7.627843405924242, "learning_rate": 2.4404200198693272e-08, "loss": 0.1259, "step": 53740 }, { "epoch": 14.57824789802007, "grad_norm": 6.631485733027213, "learning_rate": 2.409376948018183e-08, "loss": 0.1392, "step": 53750 }, { "epoch": 14.580960130187144, "grad_norm": 5.826545068914147, "learning_rate": 2.3785321050033284e-08, "loss": 0.1342, "step": 53760 }, { "epoch": 14.583672362354218, "grad_norm": 6.901334137178568, "learning_rate": 2.3478855031128234e-08, "loss": 0.1275, "step": 53770 }, { "epoch": 14.586384594521292, "grad_norm": 7.6136637476651945, "learning_rate": 2.317437154555513e-08, "loss": 0.1252, "step": 53780 }, { "epoch": 14.589096826688365, "grad_norm": 6.276491062219857, "learning_rate": 2.2871870714613054e-08, "loss": 0.1217, "step": 53790 }, { "epoch": 14.591809058855437, "grad_norm": 7.144509291402416, "learning_rate": 2.2571352658812274e-08, "loss": 0.1249, "step": 53800 }, { "epoch": 14.594521291022511, "grad_norm": 6.962527095838553, "learning_rate": 2.2272817497870913e-08, "loss": 0.1365, "step": 53810 }, { "epoch": 14.597233523189585, "grad_norm": 9.157326586981299, "learning_rate": 2.1976265350720506e-08, "loss": 0.1328, "step": 53820 }, { "epoch": 14.599945755356659, "grad_norm": 5.300608977870804, "learning_rate": 2.168169633549877e-08, "loss": 0.1332, "step": 53830 }, { "epoch": 14.602657987523733, "grad_norm": 7.127818428437295, "learning_rate": 2.1389110569556838e-08, "loss": 0.1317, "step": 53840 }, { "epoch": 14.605370219690805, "grad_norm": 6.373106448015858, "learning_rate": 2.109850816945369e-08, "loss": 0.123, "step": 53850 }, { "epoch": 14.608082451857879, "grad_norm": 5.593806723095649, "learning_rate": 2.0809889250958948e-08, "loss": 0.1252, "step": 53860 }, { "epoch": 14.610794684024953, "grad_norm": 6.425313368312958, "learning_rate": 2.052325392905119e-08, "loss": 0.1395, "step": 53870 }, { "epoch": 14.613506916192026, "grad_norm": 9.042098250126264, "learning_rate": 2.0238602317920187e-08, "loss": 0.1457, "step": 53880 }, { "epoch": 14.6162191483591, "grad_norm": 8.056665068792089, "learning_rate": 1.9955934530964117e-08, "loss": 0.1342, "step": 53890 }, { "epoch": 14.618931380526172, "grad_norm": 5.077006637352937, "learning_rate": 1.9675250680792345e-08, "loss": 0.1253, "step": 53900 }, { "epoch": 14.621643612693246, "grad_norm": 6.592412378674997, "learning_rate": 1.9396550879221543e-08, "loss": 0.1216, "step": 53910 }, { "epoch": 14.62435584486032, "grad_norm": 7.659490928455684, "learning_rate": 1.9119835237280672e-08, "loss": 0.1214, "step": 53920 }, { "epoch": 14.627068077027394, "grad_norm": 7.154746347982565, "learning_rate": 1.884510386520655e-08, "loss": 0.1416, "step": 53930 }, { "epoch": 14.629780309194468, "grad_norm": 6.763933327183984, "learning_rate": 1.8572356872446073e-08, "loss": 0.1258, "step": 53940 }, { "epoch": 14.63249254136154, "grad_norm": 7.73102172255719, "learning_rate": 1.8301594367654552e-08, "loss": 0.1234, "step": 53950 }, { "epoch": 14.635204773528613, "grad_norm": 7.68026487354313, "learning_rate": 1.803281645869903e-08, "loss": 0.1177, "step": 53960 }, { "epoch": 14.637917005695687, "grad_norm": 6.473116396027083, "learning_rate": 1.7766023252653863e-08, "loss": 0.1263, "step": 53970 }, { "epoch": 14.640629237862761, "grad_norm": 7.877070200026878, "learning_rate": 1.7501214855802916e-08, "loss": 0.1265, "step": 53980 }, { "epoch": 14.643341470029835, "grad_norm": 6.349132575137432, "learning_rate": 1.723839137364125e-08, "loss": 0.1241, "step": 53990 }, { "epoch": 14.646053702196909, "grad_norm": 6.075173661750412, "learning_rate": 1.697755291087011e-08, "loss": 0.1283, "step": 54000 }, { "epoch": 14.648765934363981, "grad_norm": 7.224273503747136, "learning_rate": 1.6718699571402486e-08, "loss": 0.1394, "step": 54010 }, { "epoch": 14.651478166531055, "grad_norm": 6.834586933562682, "learning_rate": 1.6461831458359777e-08, "loss": 0.1278, "step": 54020 }, { "epoch": 14.654190398698129, "grad_norm": 6.690388676375711, "learning_rate": 1.620694867407291e-08, "loss": 0.1204, "step": 54030 }, { "epoch": 14.656902630865202, "grad_norm": 5.997810998412364, "learning_rate": 1.59540513200801e-08, "loss": 0.125, "step": 54040 }, { "epoch": 14.659614863032276, "grad_norm": 6.40011032730201, "learning_rate": 1.570313949713076e-08, "loss": 0.1197, "step": 54050 }, { "epoch": 14.662327095199348, "grad_norm": 8.407910896619448, "learning_rate": 1.545421330518215e-08, "loss": 0.1247, "step": 54060 }, { "epoch": 14.665039327366422, "grad_norm": 6.842568067896191, "learning_rate": 1.520727284340162e-08, "loss": 0.1341, "step": 54070 }, { "epoch": 14.667751559533496, "grad_norm": 5.329381920436189, "learning_rate": 1.496231821016325e-08, "loss": 0.1243, "step": 54080 }, { "epoch": 14.67046379170057, "grad_norm": 8.349707853477888, "learning_rate": 1.471934950305287e-08, "loss": 0.1278, "step": 54090 }, { "epoch": 14.673176023867644, "grad_norm": 6.234877580053887, "learning_rate": 1.4478366818862499e-08, "loss": 0.1184, "step": 54100 }, { "epoch": 14.675888256034717, "grad_norm": 6.438131077071799, "learning_rate": 1.4239370253595341e-08, "loss": 0.1294, "step": 54110 }, { "epoch": 14.67860048820179, "grad_norm": 7.4159247161479565, "learning_rate": 1.400235990246135e-08, "loss": 0.1095, "step": 54120 }, { "epoch": 14.681312720368863, "grad_norm": 8.56125419853443, "learning_rate": 1.3767335859880549e-08, "loss": 0.127, "step": 54130 }, { "epoch": 14.684024952535937, "grad_norm": 8.652143334510415, "learning_rate": 1.3534298219480824e-08, "loss": 0.1521, "step": 54140 }, { "epoch": 14.686737184703011, "grad_norm": 5.486254570872626, "learning_rate": 1.3303247074099024e-08, "loss": 0.12, "step": 54150 }, { "epoch": 14.689449416870085, "grad_norm": 6.069804260676115, "learning_rate": 1.307418251578152e-08, "loss": 0.1397, "step": 54160 }, { "epoch": 14.692161649037157, "grad_norm": 5.120807077658321, "learning_rate": 1.2847104635781427e-08, "loss": 0.1297, "step": 54170 }, { "epoch": 14.69487388120423, "grad_norm": 6.967952075507266, "learning_rate": 1.2622013524561939e-08, "loss": 0.1327, "step": 54180 }, { "epoch": 14.697586113371305, "grad_norm": 7.849309215168197, "learning_rate": 1.2398909271794102e-08, "loss": 0.1404, "step": 54190 }, { "epoch": 14.700298345538378, "grad_norm": 7.436197811181908, "learning_rate": 1.2177791966357932e-08, "loss": 0.1341, "step": 54200 }, { "epoch": 14.703010577705452, "grad_norm": 23.63294624673642, "learning_rate": 1.19586616963413e-08, "loss": 0.1314, "step": 54210 }, { "epoch": 14.705722809872524, "grad_norm": 7.031988945755344, "learning_rate": 1.1741518549040487e-08, "loss": 0.1188, "step": 54220 }, { "epoch": 14.708435042039598, "grad_norm": 6.6149353639163575, "learning_rate": 1.152636261096074e-08, "loss": 0.1288, "step": 54230 }, { "epoch": 14.711147274206672, "grad_norm": 7.127214680125232, "learning_rate": 1.131319396781516e-08, "loss": 0.1243, "step": 54240 }, { "epoch": 14.713859506373746, "grad_norm": 6.952321367560483, "learning_rate": 1.1102012704525823e-08, "loss": 0.1313, "step": 54250 }, { "epoch": 14.71657173854082, "grad_norm": 9.249869311820921, "learning_rate": 1.0892818905221537e-08, "loss": 0.1314, "step": 54260 }, { "epoch": 14.719283970707892, "grad_norm": 5.066257376684322, "learning_rate": 1.0685612653241195e-08, "loss": 0.1138, "step": 54270 }, { "epoch": 14.721996202874966, "grad_norm": 7.177427495311676, "learning_rate": 1.0480394031130992e-08, "loss": 0.1245, "step": 54280 }, { "epoch": 14.72470843504204, "grad_norm": 5.963008194128325, "learning_rate": 1.0277163120644973e-08, "loss": 0.1205, "step": 54290 }, { "epoch": 14.727420667209113, "grad_norm": 9.771794458575522, "learning_rate": 1.0075920002746153e-08, "loss": 0.1216, "step": 54300 }, { "epoch": 14.730132899376187, "grad_norm": 9.937511143554008, "learning_rate": 9.876664757605402e-09, "loss": 0.1327, "step": 54310 }, { "epoch": 14.732845131543261, "grad_norm": 5.0751583036100305, "learning_rate": 9.679397464600893e-09, "loss": 0.1199, "step": 54320 }, { "epoch": 14.735557363710333, "grad_norm": 5.809182438983171, "learning_rate": 9.484118202319759e-09, "loss": 0.1309, "step": 54330 }, { "epoch": 14.738269595877407, "grad_norm": 7.753022245157958, "learning_rate": 9.29082704855644e-09, "loss": 0.142, "step": 54340 }, { "epoch": 14.74098182804448, "grad_norm": 6.309895117711023, "learning_rate": 9.099524080314336e-09, "loss": 0.1174, "step": 54350 }, { "epoch": 14.743694060211554, "grad_norm": 7.346420433955735, "learning_rate": 8.910209373804157e-09, "loss": 0.1261, "step": 54360 }, { "epoch": 14.746406292378628, "grad_norm": 6.552060815396897, "learning_rate": 8.72288300444446e-09, "loss": 0.1251, "step": 54370 }, { "epoch": 14.7491185245457, "grad_norm": 6.22879719875449, "learning_rate": 8.537545046861662e-09, "loss": 0.1346, "step": 54380 }, { "epoch": 14.751830756712774, "grad_norm": 5.818767876343092, "learning_rate": 8.354195574890034e-09, "loss": 0.1228, "step": 54390 }, { "epoch": 14.754542988879848, "grad_norm": 6.626319281404391, "learning_rate": 8.172834661572815e-09, "loss": 0.1255, "step": 54400 }, { "epoch": 14.757255221046922, "grad_norm": 5.710776233774212, "learning_rate": 7.993462379158878e-09, "loss": 0.1211, "step": 54410 }, { "epoch": 14.759967453213996, "grad_norm": 9.827221356473887, "learning_rate": 7.816078799107173e-09, "loss": 0.1288, "step": 54420 }, { "epoch": 14.76267968538107, "grad_norm": 7.247825287991807, "learning_rate": 7.640683992082287e-09, "loss": 0.1298, "step": 54430 }, { "epoch": 14.765391917548142, "grad_norm": 7.511924308032306, "learning_rate": 7.467278027958325e-09, "loss": 0.1178, "step": 54440 }, { "epoch": 14.768104149715215, "grad_norm": 6.19565883465969, "learning_rate": 7.295860975815583e-09, "loss": 0.1325, "step": 54450 }, { "epoch": 14.77081638188229, "grad_norm": 6.172511576052896, "learning_rate": 7.126432903943326e-09, "loss": 0.1328, "step": 54460 }, { "epoch": 14.773528614049363, "grad_norm": 7.952815967408122, "learning_rate": 6.958993879838116e-09, "loss": 0.1246, "step": 54470 }, { "epoch": 14.776240846216435, "grad_norm": 5.733836514275817, "learning_rate": 6.79354397020271e-09, "loss": 0.1221, "step": 54480 }, { "epoch": 14.778953078383509, "grad_norm": 7.091334633620665, "learning_rate": 6.630083240949936e-09, "loss": 0.1084, "step": 54490 }, { "epoch": 14.781665310550583, "grad_norm": 6.462600293993146, "learning_rate": 6.468611757197707e-09, "loss": 0.1244, "step": 54500 }, { "epoch": 14.784377542717657, "grad_norm": 7.067034753028, "learning_rate": 6.30912958327401e-09, "loss": 0.137, "step": 54510 }, { "epoch": 14.78708977488473, "grad_norm": 6.2837009681014075, "learning_rate": 6.151636782710801e-09, "loss": 0.1358, "step": 54520 }, { "epoch": 14.789802007051804, "grad_norm": 5.193981003898874, "learning_rate": 5.9961334182517815e-09, "loss": 0.1221, "step": 54530 }, { "epoch": 14.792514239218876, "grad_norm": 5.677654077095513, "learning_rate": 5.8426195518451745e-09, "loss": 0.119, "step": 54540 }, { "epoch": 14.79522647138595, "grad_norm": 5.737667902522157, "learning_rate": 5.69109524464706e-09, "loss": 0.1227, "step": 54550 }, { "epoch": 14.797938703553024, "grad_norm": 5.142977187512028, "learning_rate": 5.541560557021375e-09, "loss": 0.1139, "step": 54560 }, { "epoch": 14.800650935720098, "grad_norm": 8.154163113795638, "learning_rate": 5.394015548539911e-09, "loss": 0.1372, "step": 54570 }, { "epoch": 14.803363167887172, "grad_norm": 7.487467610196422, "learning_rate": 5.248460277980649e-09, "loss": 0.1462, "step": 54580 }, { "epoch": 14.806075400054244, "grad_norm": 10.56272006187231, "learning_rate": 5.104894803329985e-09, "loss": 0.1317, "step": 54590 }, { "epoch": 14.808787632221318, "grad_norm": 7.215763968143249, "learning_rate": 4.963319181780501e-09, "loss": 0.1408, "step": 54600 }, { "epoch": 14.811499864388392, "grad_norm": 5.610649103447081, "learning_rate": 4.8237334697331935e-09, "loss": 0.1168, "step": 54610 }, { "epoch": 14.814212096555465, "grad_norm": 6.13622028792654, "learning_rate": 4.686137722796358e-09, "loss": 0.126, "step": 54620 }, { "epoch": 14.81692432872254, "grad_norm": 6.276730144937998, "learning_rate": 4.550531995783924e-09, "loss": 0.1294, "step": 54630 }, { "epoch": 14.819636560889613, "grad_norm": 5.6685523648032845, "learning_rate": 4.416916342718791e-09, "loss": 0.1351, "step": 54640 }, { "epoch": 14.822348793056685, "grad_norm": 6.382981498537089, "learning_rate": 4.285290816830601e-09, "loss": 0.1187, "step": 54650 }, { "epoch": 14.825061025223759, "grad_norm": 6.301430305834408, "learning_rate": 4.155655470555742e-09, "loss": 0.1193, "step": 54660 }, { "epoch": 14.827773257390833, "grad_norm": 6.301392636712552, "learning_rate": 4.0280103555373486e-09, "loss": 0.1291, "step": 54670 }, { "epoch": 14.830485489557907, "grad_norm": 7.508767417518294, "learning_rate": 3.9023555226275215e-09, "loss": 0.136, "step": 54680 }, { "epoch": 14.83319772172498, "grad_norm": 7.150891785140059, "learning_rate": 3.7786910218834405e-09, "loss": 0.1302, "step": 54690 }, { "epoch": 14.835909953892052, "grad_norm": 6.298722537260108, "learning_rate": 3.657016902570698e-09, "loss": 0.1274, "step": 54700 }, { "epoch": 14.838622186059126, "grad_norm": 6.897672064304108, "learning_rate": 3.537333213161076e-09, "loss": 0.1199, "step": 54710 }, { "epoch": 14.8413344182262, "grad_norm": 7.788402342214462, "learning_rate": 3.4196400013342125e-09, "loss": 0.1268, "step": 54720 }, { "epoch": 14.844046650393274, "grad_norm": 7.908707303423157, "learning_rate": 3.3039373139764908e-09, "loss": 0.1294, "step": 54730 }, { "epoch": 14.846758882560348, "grad_norm": 5.156033326194416, "learning_rate": 3.1902251971810406e-09, "loss": 0.1368, "step": 54740 }, { "epoch": 14.84947111472742, "grad_norm": 9.919228716542431, "learning_rate": 3.0785036962482917e-09, "loss": 0.1226, "step": 54750 }, { "epoch": 14.852183346894494, "grad_norm": 5.631202965037279, "learning_rate": 2.968772855684865e-09, "loss": 0.1325, "step": 54760 }, { "epoch": 14.854895579061568, "grad_norm": 6.892572426848919, "learning_rate": 2.8610327192063468e-09, "loss": 0.1247, "step": 54770 }, { "epoch": 14.857607811228641, "grad_norm": 6.375327794905223, "learning_rate": 2.7552833297328498e-09, "loss": 0.1235, "step": 54780 }, { "epoch": 14.860320043395715, "grad_norm": 6.374645089813492, "learning_rate": 2.651524729393451e-09, "loss": 0.119, "step": 54790 }, { "epoch": 14.863032275562787, "grad_norm": 6.137543135916607, "learning_rate": 2.549756959522309e-09, "loss": 0.1237, "step": 54800 }, { "epoch": 14.865744507729861, "grad_norm": 11.050462628597131, "learning_rate": 2.449980060661994e-09, "loss": 0.1285, "step": 54810 }, { "epoch": 14.868456739896935, "grad_norm": 8.012278141258863, "learning_rate": 2.3521940725612647e-09, "loss": 0.141, "step": 54820 }, { "epoch": 14.871168972064009, "grad_norm": 6.572423256274998, "learning_rate": 2.256399034175627e-09, "loss": 0.1314, "step": 54830 }, { "epoch": 14.873881204231083, "grad_norm": 4.99910963828626, "learning_rate": 2.162594983667887e-09, "loss": 0.1148, "step": 54840 }, { "epoch": 14.876593436398156, "grad_norm": 7.082192689977743, "learning_rate": 2.0707819584070423e-09, "loss": 0.1086, "step": 54850 }, { "epoch": 14.879305668565229, "grad_norm": 8.021574321346149, "learning_rate": 1.9809599949699442e-09, "loss": 0.1412, "step": 54860 }, { "epoch": 14.882017900732302, "grad_norm": 5.724290075643509, "learning_rate": 1.893129129139637e-09, "loss": 0.1204, "step": 54870 }, { "epoch": 14.884730132899376, "grad_norm": 5.085554988264016, "learning_rate": 1.807289395905354e-09, "loss": 0.1186, "step": 54880 }, { "epoch": 14.88744236506645, "grad_norm": 6.571476801855149, "learning_rate": 1.7234408294641846e-09, "loss": 0.116, "step": 54890 }, { "epoch": 14.890154597233524, "grad_norm": 7.012795598328354, "learning_rate": 1.6415834632188544e-09, "loss": 0.1382, "step": 54900 }, { "epoch": 14.892866829400596, "grad_norm": 5.105479629351602, "learning_rate": 1.5617173297804989e-09, "loss": 0.1237, "step": 54910 }, { "epoch": 14.89557906156767, "grad_norm": 5.189546459395873, "learning_rate": 1.4838424609647795e-09, "loss": 0.1285, "step": 54920 }, { "epoch": 14.898291293734744, "grad_norm": 5.955384613208829, "learning_rate": 1.4079588877963234e-09, "loss": 0.1227, "step": 54930 }, { "epoch": 14.901003525901817, "grad_norm": 7.587512099572979, "learning_rate": 1.3340666405053938e-09, "loss": 0.1284, "step": 54940 }, { "epoch": 14.903715758068891, "grad_norm": 8.15199084831458, "learning_rate": 1.2621657485278883e-09, "loss": 0.1191, "step": 54950 }, { "epoch": 14.906427990235965, "grad_norm": 6.962887433224575, "learning_rate": 1.1922562405086712e-09, "loss": 0.1189, "step": 54960 }, { "epoch": 14.909140222403037, "grad_norm": 7.0737451178973405, "learning_rate": 1.1243381442976874e-09, "loss": 0.1331, "step": 54970 }, { "epoch": 14.911852454570111, "grad_norm": 6.879596940258811, "learning_rate": 1.0584114869516272e-09, "loss": 0.1272, "step": 54980 }, { "epoch": 14.914564686737185, "grad_norm": 5.203490142490021, "learning_rate": 9.94476294735036e-10, "loss": 0.1202, "step": 54990 }, { "epoch": 14.917276918904259, "grad_norm": 3.8014912305101185, "learning_rate": 9.32532593117541e-10, "loss": 0.124, "step": 55000 }, { "epoch": 14.919989151071333, "grad_norm": 8.896331180536148, "learning_rate": 8.72580406776069e-10, "loss": 0.1272, "step": 55010 }, { "epoch": 14.922701383238405, "grad_norm": 6.956440870519984, "learning_rate": 8.146197595942928e-10, "loss": 0.1516, "step": 55020 }, { "epoch": 14.925413615405478, "grad_norm": 16.24356343475164, "learning_rate": 7.586506746626309e-10, "loss": 0.136, "step": 55030 }, { "epoch": 14.928125847572552, "grad_norm": 6.968466005735997, "learning_rate": 7.046731742782476e-10, "loss": 0.1392, "step": 55040 }, { "epoch": 14.930838079739626, "grad_norm": 7.847781057682088, "learning_rate": 6.52687279943387e-10, "loss": 0.1339, "step": 55050 }, { "epoch": 14.9335503119067, "grad_norm": 7.722718791147108, "learning_rate": 6.026930123692598e-10, "loss": 0.1311, "step": 55060 }, { "epoch": 14.936262544073772, "grad_norm": 7.458890471878592, "learning_rate": 5.546903914721569e-10, "loss": 0.1275, "step": 55070 }, { "epoch": 14.938974776240846, "grad_norm": 9.083644116085104, "learning_rate": 5.086794363751147e-10, "loss": 0.1379, "step": 55080 }, { "epoch": 14.94168700840792, "grad_norm": 5.375271485593029, "learning_rate": 4.646601654073601e-10, "loss": 0.1199, "step": 55090 }, { "epoch": 14.944399240574993, "grad_norm": 6.322407826829851, "learning_rate": 4.226325961059763e-10, "loss": 0.1241, "step": 55100 }, { "epoch": 14.947111472742067, "grad_norm": 4.872340710829426, "learning_rate": 3.8259674521312627e-10, "loss": 0.1214, "step": 55110 }, { "epoch": 14.94982370490914, "grad_norm": 7.119602845160124, "learning_rate": 3.445526286782741e-10, "loss": 0.1316, "step": 55120 }, { "epoch": 14.952535937076213, "grad_norm": 6.538630492186604, "learning_rate": 3.085002616581845e-10, "loss": 0.1255, "step": 55130 }, { "epoch": 14.955248169243287, "grad_norm": 7.40773448713109, "learning_rate": 2.7443965851414733e-10, "loss": 0.1272, "step": 55140 }, { "epoch": 14.95796040141036, "grad_norm": 6.352806017654528, "learning_rate": 2.423708328164187e-10, "loss": 0.122, "step": 55150 }, { "epoch": 14.960672633577435, "grad_norm": 5.655601228826103, "learning_rate": 2.1229379733922474e-10, "loss": 0.1071, "step": 55160 }, { "epoch": 14.963384865744509, "grad_norm": 5.848250147477777, "learning_rate": 1.8420856406464737e-10, "loss": 0.1378, "step": 55170 }, { "epoch": 14.96609709791158, "grad_norm": 5.28875597606381, "learning_rate": 1.5811514418206941e-10, "loss": 0.1251, "step": 55180 }, { "epoch": 14.968809330078654, "grad_norm": 7.373661793861163, "learning_rate": 1.3401354808650902e-10, "loss": 0.1284, "step": 55190 }, { "epoch": 14.971521562245728, "grad_norm": 4.9336260901549505, "learning_rate": 1.1190378537861979e-10, "loss": 0.1304, "step": 55200 }, { "epoch": 14.974233794412802, "grad_norm": 6.759954868086495, "learning_rate": 9.178586486691121e-11, "loss": 0.1175, "step": 55210 }, { "epoch": 14.976946026579876, "grad_norm": 6.268045486785555, "learning_rate": 7.365979456608329e-11, "loss": 0.1362, "step": 55220 }, { "epoch": 14.979658258746948, "grad_norm": 6.352464051743145, "learning_rate": 5.75255816970266e-11, "loss": 0.1317, "step": 55230 }, { "epoch": 14.982370490914022, "grad_norm": 5.486977738909114, "learning_rate": 4.338323268682221e-11, "loss": 0.114, "step": 55240 }, { "epoch": 14.985082723081096, "grad_norm": 7.293743246136093, "learning_rate": 3.123275316985197e-11, "loss": 0.1297, "step": 55250 }, { "epoch": 14.98779495524817, "grad_norm": 8.119090374026696, "learning_rate": 2.1074147986688274e-11, "loss": 0.1227, "step": 55260 }, { "epoch": 14.990507187415243, "grad_norm": 6.8906984769328155, "learning_rate": 1.2907421184094049e-11, "loss": 0.1425, "step": 55270 }, { "epoch": 14.993219419582317, "grad_norm": 5.60855775450967, "learning_rate": 6.732576015577863e-12, "loss": 0.1338, "step": 55280 }, { "epoch": 14.99593165174939, "grad_norm": 6.552248911957402, "learning_rate": 2.5496149413939407e-12, "loss": 0.1145, "step": 55290 }, { "epoch": 14.998643883916463, "grad_norm": 8.6867033605831, "learning_rate": 3.5853962743193083e-13, "loss": 0.1207, "step": 55300 } ], "logging_steps": 10, "max_steps": 55305, "num_input_tokens_seen": 0, "num_train_epochs": 15, "save_steps": 10000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 55788809336832.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }