diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7950 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 1124,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017793594306049821,
+      "grad_norm": 1.850117504015236,
+      "learning_rate": 9.99998046979289e-06,
+      "loss": 0.1899,
+      "step": 1
+    },
+    {
+      "epoch": 0.0035587188612099642,
+      "grad_norm": 1.3242181565053093,
+      "learning_rate": 9.999921879324127e-06,
+      "loss": 0.1604,
+      "step": 2
+    },
+    {
+      "epoch": 0.005338078291814947,
+      "grad_norm": 1.6844951018295002,
+      "learning_rate": 9.999824229051425e-06,
+      "loss": 0.1869,
+      "step": 3
+    },
+    {
+      "epoch": 0.0071174377224199285,
+      "grad_norm": 1.482072176428284,
+      "learning_rate": 9.999687519737639e-06,
+      "loss": 0.1971,
+      "step": 4
+    },
+    {
+      "epoch": 0.008896797153024912,
+      "grad_norm": 1.8068000550688117,
+      "learning_rate": 9.99951175245075e-06,
+      "loss": 0.1691,
+      "step": 5
+    },
+    {
+      "epoch": 0.010676156583629894,
+      "grad_norm": 1.076831085259329,
+      "learning_rate": 9.999296928563868e-06,
+      "loss": 0.1617,
+      "step": 6
+    },
+    {
+      "epoch": 0.012455516014234875,
+      "grad_norm": 1.6543768823577851,
+      "learning_rate": 9.999043049755216e-06,
+      "loss": 0.2022,
+      "step": 7
+    },
+    {
+      "epoch": 0.014234875444839857,
+      "grad_norm": 0.7753371450542343,
+      "learning_rate": 9.998750118008117e-06,
+      "loss": 0.134,
+      "step": 8
+    },
+    {
+      "epoch": 0.01601423487544484,
+      "grad_norm": 0.8255895406318958,
+      "learning_rate": 9.998418135610974e-06,
+      "loss": 0.1212,
+      "step": 9
+    },
+    {
+      "epoch": 0.017793594306049824,
+      "grad_norm": 1.007873843010065,
+      "learning_rate": 9.998047105157265e-06,
+      "loss": 0.1624,
+      "step": 10
+    },
+    {
+      "epoch": 0.019572953736654804,
+      "grad_norm": 0.6748774695574393,
+      "learning_rate": 9.997637029545509e-06,
+      "loss": 0.0976,
+      "step": 11
+    },
+    {
+      "epoch": 0.021352313167259787,
+      "grad_norm": 0.9985515270653275,
+      "learning_rate": 9.997187911979252e-06,
+      "loss": 0.1624,
+      "step": 12
+    },
+    {
+      "epoch": 0.023131672597864767,
+      "grad_norm": 0.8751432464926944,
+      "learning_rate": 9.996699755967035e-06,
+      "loss": 0.1576,
+      "step": 13
+    },
+    {
+      "epoch": 0.02491103202846975,
+      "grad_norm": 0.7590943868234059,
+      "learning_rate": 9.996172565322375e-06,
+      "loss": 0.1164,
+      "step": 14
+    },
+    {
+      "epoch": 0.026690391459074734,
+      "grad_norm": 0.7008756422019338,
+      "learning_rate": 9.995606344163728e-06,
+      "loss": 0.11,
+      "step": 15
+    },
+    {
+      "epoch": 0.028469750889679714,
+      "grad_norm": 0.621434356000605,
+      "learning_rate": 9.995001096914462e-06,
+      "loss": 0.1008,
+      "step": 16
+    },
+    {
+      "epoch": 0.030249110320284697,
+      "grad_norm": 0.6506134281620294,
+      "learning_rate": 9.994356828302818e-06,
+      "loss": 0.1112,
+      "step": 17
+    },
+    {
+      "epoch": 0.03202846975088968,
+      "grad_norm": 0.5549824133725161,
+      "learning_rate": 9.993673543361874e-06,
+      "loss": 0.0881,
+      "step": 18
+    },
+    {
+      "epoch": 0.033807829181494664,
+      "grad_norm": 0.8199102769003839,
+      "learning_rate": 9.992951247429512e-06,
+      "loss": 0.1429,
+      "step": 19
+    },
+    {
+      "epoch": 0.03558718861209965,
+      "grad_norm": 0.6230438351979487,
+      "learning_rate": 9.992189946148366e-06,
+      "loss": 0.0937,
+      "step": 20
+    },
+    {
+      "epoch": 0.037366548042704624,
+      "grad_norm": 0.49532494731557897,
+      "learning_rate": 9.991389645465786e-06,
+      "loss": 0.0747,
+      "step": 21
+    },
+    {
+      "epoch": 0.03914590747330961,
+      "grad_norm": 0.715858463925942,
+      "learning_rate": 9.990550351633784e-06,
+      "loss": 0.1221,
+      "step": 22
+    },
+    {
+      "epoch": 0.04092526690391459,
+      "grad_norm": 0.5927031804034358,
+      "learning_rate": 9.989672071208993e-06,
+      "loss": 0.1115,
+      "step": 23
+    },
+    {
+      "epoch": 0.042704626334519574,
+      "grad_norm": 0.7016033126105914,
+      "learning_rate": 9.988754811052616e-06,
+      "loss": 0.1388,
+      "step": 24
+    },
+    {
+      "epoch": 0.04448398576512456,
+      "grad_norm": 0.8389829180519494,
+      "learning_rate": 9.987798578330365e-06,
+      "loss": 0.1465,
+      "step": 25
+    },
+    {
+      "epoch": 0.046263345195729534,
+      "grad_norm": 0.7593714046854071,
+      "learning_rate": 9.986803380512406e-06,
+      "loss": 0.1151,
+      "step": 26
+    },
+    {
+      "epoch": 0.04804270462633452,
+      "grad_norm": 0.5960366050603942,
+      "learning_rate": 9.98576922537331e-06,
+      "loss": 0.1092,
+      "step": 27
+    },
+    {
+      "epoch": 0.0498220640569395,
+      "grad_norm": 0.7129023729746609,
+      "learning_rate": 9.984696120991979e-06,
+      "loss": 0.1154,
+      "step": 28
+    },
+    {
+      "epoch": 0.051601423487544484,
+      "grad_norm": 0.7124388681355925,
+      "learning_rate": 9.983584075751598e-06,
+      "loss": 0.1141,
+      "step": 29
+    },
+    {
+      "epoch": 0.05338078291814947,
+      "grad_norm": 0.6082846294284628,
+      "learning_rate": 9.982433098339553e-06,
+      "loss": 0.1061,
+      "step": 30
+    },
+    {
+      "epoch": 0.05516014234875445,
+      "grad_norm": 0.8575006526888903,
+      "learning_rate": 9.981243197747375e-06,
+      "loss": 0.1417,
+      "step": 31
+    },
+    {
+      "epoch": 0.05693950177935943,
+      "grad_norm": 1.0680093705119493,
+      "learning_rate": 9.980014383270668e-06,
+      "loss": 0.1755,
+      "step": 32
+    },
+    {
+      "epoch": 0.05871886120996441,
+      "grad_norm": 0.6694632665569926,
+      "learning_rate": 9.978746664509032e-06,
+      "loss": 0.1111,
+      "step": 33
+    },
+    {
+      "epoch": 0.060498220640569395,
+      "grad_norm": 0.6026340169014638,
+      "learning_rate": 9.97744005136599e-06,
+      "loss": 0.0956,
+      "step": 34
+    },
+    {
+      "epoch": 0.06227758007117438,
+      "grad_norm": 0.6183244394801065,
+      "learning_rate": 9.976094554048912e-06,
+      "loss": 0.1095,
+      "step": 35
+    },
+    {
+      "epoch": 0.06405693950177936,
+      "grad_norm": 0.8047165339343109,
+      "learning_rate": 9.974710183068935e-06,
+      "loss": 0.1189,
+      "step": 36
+    },
+    {
+      "epoch": 0.06583629893238434,
+      "grad_norm": 0.549475889211002,
+      "learning_rate": 9.97328694924088e-06,
+      "loss": 0.0825,
+      "step": 37
+    },
+    {
+      "epoch": 0.06761565836298933,
+      "grad_norm": 0.6270364534262443,
+      "learning_rate": 9.971824863683168e-06,
+      "loss": 0.1131,
+      "step": 38
+    },
+    {
+      "epoch": 0.0693950177935943,
+      "grad_norm": 0.6771958186856858,
+      "learning_rate": 9.970323937817732e-06,
+      "loss": 0.1077,
+      "step": 39
+    },
+    {
+      "epoch": 0.0711743772241993,
+      "grad_norm": 0.6107729197632482,
+      "learning_rate": 9.968784183369929e-06,
+      "loss": 0.0879,
+      "step": 40
+    },
+    {
+      "epoch": 0.07295373665480427,
+      "grad_norm": 0.7427018257468275,
+      "learning_rate": 9.96720561236845e-06,
+      "loss": 0.1221,
+      "step": 41
+    },
+    {
+      "epoch": 0.07473309608540925,
+      "grad_norm": 0.7733632040387297,
+      "learning_rate": 9.965588237145219e-06,
+      "loss": 0.1183,
+      "step": 42
+    },
+    {
+      "epoch": 0.07651245551601424,
+      "grad_norm": 0.642583102094258,
+      "learning_rate": 9.963932070335307e-06,
+      "loss": 0.0985,
+      "step": 43
+    },
+    {
+      "epoch": 0.07829181494661921,
+      "grad_norm": 0.7694154983355995,
+      "learning_rate": 9.962237124876828e-06,
+      "loss": 0.1221,
+      "step": 44
+    },
+    {
+      "epoch": 0.0800711743772242,
+      "grad_norm": 0.6214418816806236,
+      "learning_rate": 9.960503414010833e-06,
+      "loss": 0.101,
+      "step": 45
+    },
+    {
+      "epoch": 0.08185053380782918,
+      "grad_norm": 0.694919763716097,
+      "learning_rate": 9.958730951281218e-06,
+      "loss": 0.1032,
+      "step": 46
+    },
+    {
+      "epoch": 0.08362989323843416,
+      "grad_norm": 0.7267926921693514,
+      "learning_rate": 9.956919750534607e-06,
+      "loss": 0.1041,
+      "step": 47
+    },
+    {
+      "epoch": 0.08540925266903915,
+      "grad_norm": 0.7600537591386981,
+      "learning_rate": 9.955069825920249e-06,
+      "loss": 0.1209,
+      "step": 48
+    },
+    {
+      "epoch": 0.08718861209964412,
+      "grad_norm": 0.5864985154716256,
+      "learning_rate": 9.953181191889913e-06,
+      "loss": 0.1075,
+      "step": 49
+    },
+    {
+      "epoch": 0.08896797153024912,
+      "grad_norm": 0.5258150803037892,
+      "learning_rate": 9.95125386319776e-06,
+      "loss": 0.0788,
+      "step": 50
+    },
+    {
+      "epoch": 0.09074733096085409,
+      "grad_norm": 0.8564907816232372,
+      "learning_rate": 9.949287854900243e-06,
+      "loss": 0.1336,
+      "step": 51
+    },
+    {
+      "epoch": 0.09252669039145907,
+      "grad_norm": 0.8243172885579434,
+      "learning_rate": 9.947283182355982e-06,
+      "loss": 0.123,
+      "step": 52
+    },
+    {
+      "epoch": 0.09430604982206406,
+      "grad_norm": 0.687563037070661,
+      "learning_rate": 9.945239861225644e-06,
+      "loss": 0.1054,
+      "step": 53
+    },
+    {
+      "epoch": 0.09608540925266904,
+      "grad_norm": 0.640308385467676,
+      "learning_rate": 9.943157907471825e-06,
+      "loss": 0.0973,
+      "step": 54
+    },
+    {
+      "epoch": 0.09786476868327403,
+      "grad_norm": 0.5939368232103092,
+      "learning_rate": 9.941037337358918e-06,
+      "loss": 0.0836,
+      "step": 55
+    },
+    {
+      "epoch": 0.099644128113879,
+      "grad_norm": 0.8233833437191558,
+      "learning_rate": 9.938878167452991e-06,
+      "loss": 0.1225,
+      "step": 56
+    },
+    {
+      "epoch": 0.10142348754448399,
+      "grad_norm": 0.9849389149391184,
+      "learning_rate": 9.936680414621663e-06,
+      "loss": 0.1325,
+      "step": 57
+    },
+    {
+      "epoch": 0.10320284697508897,
+      "grad_norm": 0.6389684265861103,
+      "learning_rate": 9.934444096033958e-06,
+      "loss": 0.0879,
+      "step": 58
+    },
+    {
+      "epoch": 0.10498220640569395,
+      "grad_norm": 0.632783982922516,
+      "learning_rate": 9.932169229160183e-06,
+      "loss": 0.1006,
+      "step": 59
+    },
+    {
+      "epoch": 0.10676156583629894,
+      "grad_norm": 0.6538263180129327,
+      "learning_rate": 9.929855831771787e-06,
+      "loss": 0.0899,
+      "step": 60
+    },
+    {
+      "epoch": 0.10854092526690391,
+      "grad_norm": 0.7488778758989755,
+      "learning_rate": 9.927503921941218e-06,
+      "loss": 0.1065,
+      "step": 61
+    },
+    {
+      "epoch": 0.1103202846975089,
+      "grad_norm": 0.6974578016718342,
+      "learning_rate": 9.925113518041796e-06,
+      "loss": 0.1114,
+      "step": 62
+    },
+    {
+      "epoch": 0.11209964412811388,
+      "grad_norm": 0.8259150828853346,
+      "learning_rate": 9.922684638747551e-06,
+      "loss": 0.1286,
+      "step": 63
+    },
+    {
+      "epoch": 0.11387900355871886,
+      "grad_norm": 0.6085539351439517,
+      "learning_rate": 9.920217303033091e-06,
+      "loss": 0.0877,
+      "step": 64
+    },
+    {
+      "epoch": 0.11565836298932385,
+      "grad_norm": 0.7075111007307412,
+      "learning_rate": 9.917711530173444e-06,
+      "loss": 0.1014,
+      "step": 65
+    },
+    {
+      "epoch": 0.11743772241992882,
+      "grad_norm": 0.7819497785303041,
+      "learning_rate": 9.91516733974392e-06,
+      "loss": 0.1065,
+      "step": 66
+    },
+    {
+      "epoch": 0.11921708185053381,
+      "grad_norm": 0.7402592942502634,
+      "learning_rate": 9.912584751619943e-06,
+      "loss": 0.1178,
+      "step": 67
+    },
+    {
+      "epoch": 0.12099644128113879,
+      "grad_norm": 0.5752840801262433,
+      "learning_rate": 9.909963785976902e-06,
+      "loss": 0.0847,
+      "step": 68
+    },
+    {
+      "epoch": 0.12277580071174377,
+      "grad_norm": 0.6564295906063625,
+      "learning_rate": 9.907304463290004e-06,
+      "loss": 0.0944,
+      "step": 69
+    },
+    {
+      "epoch": 0.12455516014234876,
+      "grad_norm": 0.6446921176074449,
+      "learning_rate": 9.904606804334094e-06,
+      "loss": 0.0861,
+      "step": 70
+    },
+    {
+      "epoch": 0.12633451957295375,
+      "grad_norm": 0.628843903964402,
+      "learning_rate": 9.901870830183506e-06,
+      "loss": 0.0901,
+      "step": 71
+    },
+    {
+      "epoch": 0.12811387900355872,
+      "grad_norm": 0.5987948243429673,
+      "learning_rate": 9.899096562211902e-06,
+      "loss": 0.0898,
+      "step": 72
+    },
+    {
+      "epoch": 0.1298932384341637,
+      "grad_norm": 0.7260953310256757,
+      "learning_rate": 9.896284022092088e-06,
+      "loss": 0.1213,
+      "step": 73
+    },
+    {
+      "epoch": 0.13167259786476868,
+      "grad_norm": 0.8917438817816882,
+      "learning_rate": 9.893433231795864e-06,
+      "loss": 0.1484,
+      "step": 74
+    },
+    {
+      "epoch": 0.13345195729537365,
+      "grad_norm": 0.9195784724738812,
+      "learning_rate": 9.890544213593838e-06,
+      "loss": 0.1393,
+      "step": 75
+    },
+    {
+      "epoch": 0.13523131672597866,
+      "grad_norm": 0.6944466877154212,
+      "learning_rate": 9.887616990055262e-06,
+      "loss": 0.1055,
+      "step": 76
+    },
+    {
+      "epoch": 0.13701067615658363,
+      "grad_norm": 0.8545513247764966,
+      "learning_rate": 9.884651584047845e-06,
+      "loss": 0.1199,
+      "step": 77
+    },
+    {
+      "epoch": 0.1387900355871886,
+      "grad_norm": 0.6343869807913708,
+      "learning_rate": 9.881648018737587e-06,
+      "loss": 0.1055,
+      "step": 78
+    },
+    {
+      "epoch": 0.14056939501779359,
+      "grad_norm": 0.6346625568791502,
+      "learning_rate": 9.878606317588588e-06,
+      "loss": 0.091,
+      "step": 79
+    },
+    {
+      "epoch": 0.1423487544483986,
+      "grad_norm": 0.8913412271772487,
+      "learning_rate": 9.875526504362868e-06,
+      "loss": 0.1387,
+      "step": 80
+    },
+    {
+      "epoch": 0.14412811387900357,
+      "grad_norm": 0.8134513840397917,
+      "learning_rate": 9.872408603120187e-06,
+      "loss": 0.1281,
+      "step": 81
+    },
+    {
+      "epoch": 0.14590747330960854,
+      "grad_norm": 0.6541117756472733,
+      "learning_rate": 9.869252638217846e-06,
+      "loss": 0.0847,
+      "step": 82
+    },
+    {
+      "epoch": 0.14768683274021352,
+      "grad_norm": 0.6516258750206984,
+      "learning_rate": 9.866058634310503e-06,
+      "loss": 0.1056,
+      "step": 83
+    },
+    {
+      "epoch": 0.1494661921708185,
+      "grad_norm": 0.6226798145474893,
+      "learning_rate": 9.862826616349981e-06,
+      "loss": 0.0966,
+      "step": 84
+    },
+    {
+      "epoch": 0.1512455516014235,
+      "grad_norm": 0.8349836367712182,
+      "learning_rate": 9.859556609585075e-06,
+      "loss": 0.1057,
+      "step": 85
+    },
+    {
+      "epoch": 0.15302491103202848,
+      "grad_norm": 0.6445224952984742,
+      "learning_rate": 9.856248639561346e-06,
+      "loss": 0.0852,
+      "step": 86
+    },
+    {
+      "epoch": 0.15480427046263345,
+      "grad_norm": 0.7818149975223446,
+      "learning_rate": 9.85290273212093e-06,
+      "loss": 0.0989,
+      "step": 87
+    },
+    {
+      "epoch": 0.15658362989323843,
+      "grad_norm": 0.6703861318699004,
+      "learning_rate": 9.849518913402334e-06,
+      "loss": 0.0904,
+      "step": 88
+    },
+    {
+      "epoch": 0.1583629893238434,
+      "grad_norm": 0.6757370171889933,
+      "learning_rate": 9.84609720984023e-06,
+      "loss": 0.0948,
+      "step": 89
+    },
+    {
+      "epoch": 0.1601423487544484,
+      "grad_norm": 0.7853354536103218,
+      "learning_rate": 9.84263764816525e-06,
+      "loss": 0.1045,
+      "step": 90
+    },
+    {
+      "epoch": 0.1619217081850534,
+      "grad_norm": 0.6618679395032799,
+      "learning_rate": 9.839140255403776e-06,
+      "loss": 0.085,
+      "step": 91
+    },
+    {
+      "epoch": 0.16370106761565836,
+      "grad_norm": 0.728046316153083,
+      "learning_rate": 9.83560505887773e-06,
+      "loss": 0.1038,
+      "step": 92
+    },
+    {
+      "epoch": 0.16548042704626334,
+      "grad_norm": 0.6607749578368273,
+      "learning_rate": 9.83203208620436e-06,
+      "loss": 0.0951,
+      "step": 93
+    },
+    {
+      "epoch": 0.16725978647686832,
+      "grad_norm": 0.7511016803015494,
+      "learning_rate": 9.828421365296023e-06,
+      "loss": 0.089,
+      "step": 94
+    },
+    {
+      "epoch": 0.16903914590747332,
+      "grad_norm": 0.8097609006354567,
+      "learning_rate": 9.824772924359974e-06,
+      "loss": 0.1068,
+      "step": 95
+    },
+    {
+      "epoch": 0.1708185053380783,
+      "grad_norm": 0.744233392622452,
+      "learning_rate": 9.821086791898133e-06,
+      "loss": 0.0974,
+      "step": 96
+    },
+    {
+      "epoch": 0.17259786476868327,
+      "grad_norm": 0.5615410319115791,
+      "learning_rate": 9.817362996706872e-06,
+      "loss": 0.0868,
+      "step": 97
+    },
+    {
+      "epoch": 0.17437722419928825,
+      "grad_norm": 0.5688174355438915,
+      "learning_rate": 9.81360156787679e-06,
+      "loss": 0.0711,
+      "step": 98
+    },
+    {
+      "epoch": 0.17615658362989323,
+      "grad_norm": 0.6394335890962268,
+      "learning_rate": 9.809802534792477e-06,
+      "loss": 0.0854,
+      "step": 99
+    },
+    {
+      "epoch": 0.17793594306049823,
+      "grad_norm": 0.7199182483223772,
+      "learning_rate": 9.805965927132294e-06,
+      "loss": 0.0981,
+      "step": 100
+    },
+    {
+      "epoch": 0.1797153024911032,
+      "grad_norm": 0.6253502238572606,
+      "learning_rate": 9.802091774868143e-06,
+      "loss": 0.0868,
+      "step": 101
+    },
+    {
+      "epoch": 0.18149466192170818,
+      "grad_norm": 0.6011550945586328,
+      "learning_rate": 9.798180108265218e-06,
+      "loss": 0.0925,
+      "step": 102
+    },
+    {
+      "epoch": 0.18327402135231316,
+      "grad_norm": 0.6109657743148987,
+      "learning_rate": 9.794230957881785e-06,
+      "loss": 0.0749,
+      "step": 103
+    },
+    {
+      "epoch": 0.18505338078291814,
+      "grad_norm": 0.8011987162358997,
+      "learning_rate": 9.79024435456893e-06,
+      "loss": 0.0912,
+      "step": 104
+    },
+    {
+      "epoch": 0.18683274021352314,
+      "grad_norm": 0.6792391775776118,
+      "learning_rate": 9.786220329470334e-06,
+      "loss": 0.0782,
+      "step": 105
+    },
+    {
+      "epoch": 0.18861209964412812,
+      "grad_norm": 0.6974473015111701,
+      "learning_rate": 9.782158914022011e-06,
+      "loss": 0.0933,
+      "step": 106
+    },
+    {
+      "epoch": 0.1903914590747331,
+      "grad_norm": 0.6999037301345081,
+      "learning_rate": 9.778060139952075e-06,
+      "loss": 0.1095,
+      "step": 107
+    },
+    {
+      "epoch": 0.19217081850533807,
+      "grad_norm": 0.6852089985154038,
+      "learning_rate": 9.773924039280488e-06,
+      "loss": 0.0852,
+      "step": 108
+    },
+    {
+      "epoch": 0.19395017793594305,
+      "grad_norm": 0.47126123895871164,
+      "learning_rate": 9.769750644318814e-06,
+      "loss": 0.0557,
+      "step": 109
+    },
+    {
+      "epoch": 0.19572953736654805,
+      "grad_norm": 0.7544192059276259,
+      "learning_rate": 9.765539987669956e-06,
+      "loss": 0.0942,
+      "step": 110
+    },
+    {
+      "epoch": 0.19750889679715303,
+      "grad_norm": 0.8279699211381575,
+      "learning_rate": 9.761292102227917e-06,
+      "loss": 0.1231,
+      "step": 111
+    },
+    {
+      "epoch": 0.199288256227758,
+      "grad_norm": 0.8842382500513801,
+      "learning_rate": 9.757007021177529e-06,
+      "loss": 0.1239,
+      "step": 112
+    },
+    {
+      "epoch": 0.20106761565836298,
+      "grad_norm": 0.5347581512933909,
+      "learning_rate": 9.752684777994197e-06,
+      "loss": 0.0766,
+      "step": 113
+    },
+    {
+      "epoch": 0.20284697508896798,
+      "grad_norm": 0.8038794814335862,
+      "learning_rate": 9.748325406443647e-06,
+      "loss": 0.1026,
+      "step": 114
+    },
+    {
+      "epoch": 0.20462633451957296,
+      "grad_norm": 0.7564353543142572,
+      "learning_rate": 9.743928940581646e-06,
+      "loss": 0.1114,
+      "step": 115
+    },
+    {
+      "epoch": 0.20640569395017794,
+      "grad_norm": 0.776560373803886,
+      "learning_rate": 9.739495414753754e-06,
+      "loss": 0.1109,
+      "step": 116
+    },
+    {
+      "epoch": 0.20818505338078291,
+      "grad_norm": 0.576126774936746,
+      "learning_rate": 9.73502486359504e-06,
+      "loss": 0.0788,
+      "step": 117
+    },
+    {
+      "epoch": 0.2099644128113879,
+      "grad_norm": 0.7866276137958961,
+      "learning_rate": 9.73051732202982e-06,
+      "loss": 0.0929,
+      "step": 118
+    },
+    {
+      "epoch": 0.2117437722419929,
+      "grad_norm": 0.6516198953302067,
+      "learning_rate": 9.725972825271381e-06,
+      "loss": 0.094,
+      "step": 119
+    },
+    {
+      "epoch": 0.21352313167259787,
+      "grad_norm": 0.6528611110632259,
+      "learning_rate": 9.721391408821713e-06,
+      "loss": 0.0852,
+      "step": 120
+    },
+    {
+      "epoch": 0.21530249110320285,
+      "grad_norm": 0.6297819585212988,
+      "learning_rate": 9.716773108471213e-06,
+      "loss": 0.0975,
+      "step": 121
+    },
+    {
+      "epoch": 0.21708185053380782,
+      "grad_norm": 0.5830482215666739,
+      "learning_rate": 9.712117960298433e-06,
+      "loss": 0.0889,
+      "step": 122
+    },
+    {
+      "epoch": 0.2188612099644128,
+      "grad_norm": 0.7588511383189046,
+      "learning_rate": 9.707426000669773e-06,
+      "loss": 0.0982,
+      "step": 123
+    },
+    {
+      "epoch": 0.2206405693950178,
+      "grad_norm": 0.7068978079636339,
+      "learning_rate": 9.702697266239211e-06,
+      "loss": 0.0928,
+      "step": 124
+    },
+    {
+      "epoch": 0.22241992882562278,
+      "grad_norm": 0.7839249497631552,
+      "learning_rate": 9.697931793948012e-06,
+      "loss": 0.1159,
+      "step": 125
+    },
+    {
+      "epoch": 0.22419928825622776,
+      "grad_norm": 0.5259986353587656,
+      "learning_rate": 9.693129621024441e-06,
+      "loss": 0.0801,
+      "step": 126
+    },
+    {
+      "epoch": 0.22597864768683273,
+      "grad_norm": 0.6479411960069293,
+      "learning_rate": 9.68829078498347e-06,
+      "loss": 0.0898,
+      "step": 127
+    },
+    {
+      "epoch": 0.2277580071174377,
+      "grad_norm": 0.7659163732430799,
+      "learning_rate": 9.683415323626487e-06,
+      "loss": 0.0993,
+      "step": 128
+    },
+    {
+      "epoch": 0.22953736654804271,
+      "grad_norm": 0.7444181593280138,
+      "learning_rate": 9.678503275040997e-06,
+      "loss": 0.0917,
+      "step": 129
+    },
+    {
+      "epoch": 0.2313167259786477,
+      "grad_norm": 0.6720754654245071,
+      "learning_rate": 9.673554677600336e-06,
+      "loss": 0.0923,
+      "step": 130
+    },
+    {
+      "epoch": 0.23309608540925267,
+      "grad_norm": 0.6150542932919436,
+      "learning_rate": 9.668569569963355e-06,
+      "loss": 0.0912,
+      "step": 131
+    },
+    {
+      "epoch": 0.23487544483985764,
+      "grad_norm": 0.4195368111494265,
+      "learning_rate": 9.663547991074129e-06,
+      "loss": 0.0589,
+      "step": 132
+    },
+    {
+      "epoch": 0.23665480427046262,
+      "grad_norm": 0.8051306209414907,
+      "learning_rate": 9.658489980161643e-06,
+      "loss": 0.1027,
+      "step": 133
+    },
+    {
+      "epoch": 0.23843416370106763,
+      "grad_norm": 0.6894834376260431,
+      "learning_rate": 9.653395576739504e-06,
+      "loss": 0.0976,
+      "step": 134
+    },
+    {
+      "epoch": 0.2402135231316726,
+      "grad_norm": 0.801288298645482,
+      "learning_rate": 9.648264820605611e-06,
+      "loss": 0.1044,
+      "step": 135
+    },
+    {
+      "epoch": 0.24199288256227758,
+      "grad_norm": 0.5871780489625489,
+      "learning_rate": 9.643097751841854e-06,
+      "loss": 0.1158,
+      "step": 136
+    },
+    {
+      "epoch": 0.24377224199288255,
+      "grad_norm": 0.6959804215583536,
+      "learning_rate": 9.637894410813803e-06,
+      "loss": 0.0979,
+      "step": 137
+    },
+    {
+      "epoch": 0.24555160142348753,
+      "grad_norm": 0.7577161295911691,
+      "learning_rate": 9.632654838170393e-06,
+      "loss": 0.0978,
+      "step": 138
+    },
+    {
+      "epoch": 0.24733096085409254,
+      "grad_norm": 0.792649676823859,
+      "learning_rate": 9.627379074843595e-06,
+      "loss": 0.1053,
+      "step": 139
+    },
+    {
+      "epoch": 0.2491103202846975,
+      "grad_norm": 0.6832972710414728,
+      "learning_rate": 9.622067162048111e-06,
+      "loss": 0.0967,
+      "step": 140
+    },
+    {
+      "epoch": 0.2508896797153025,
+      "grad_norm": 0.5314901541816573,
+      "learning_rate": 9.616719141281044e-06,
+      "loss": 0.085,
+      "step": 141
+    },
+    {
+      "epoch": 0.2526690391459075,
+      "grad_norm": 0.7844050126301578,
+      "learning_rate": 9.611335054321576e-06,
+      "loss": 0.1206,
+      "step": 142
+    },
+    {
+      "epoch": 0.25444839857651247,
+      "grad_norm": 0.7459927569956005,
+      "learning_rate": 9.605914943230637e-06,
+      "loss": 0.1011,
+      "step": 143
+    },
+    {
+      "epoch": 0.25622775800711745,
+      "grad_norm": 0.548617640096374,
+      "learning_rate": 9.600458850350588e-06,
+      "loss": 0.0749,
+      "step": 144
+    },
+    {
+      "epoch": 0.2580071174377224,
+      "grad_norm": 0.7073189058202478,
+      "learning_rate": 9.594966818304875e-06,
+      "loss": 0.0937,
+      "step": 145
+    },
+    {
+      "epoch": 0.2597864768683274,
+      "grad_norm": 0.6637121557718688,
+      "learning_rate": 9.589438889997712e-06,
+      "loss": 0.0645,
+      "step": 146
+    },
+    {
+      "epoch": 0.2615658362989324,
+      "grad_norm": 0.6075819112473875,
+      "learning_rate": 9.583875108613727e-06,
+      "loss": 0.0788,
+      "step": 147
+    },
+    {
+      "epoch": 0.26334519572953735,
+      "grad_norm": 0.5788525993664415,
+      "learning_rate": 9.578275517617646e-06,
+      "loss": 0.0803,
+      "step": 148
+    },
+    {
+      "epoch": 0.26512455516014233,
+      "grad_norm": 0.678075585849508,
+      "learning_rate": 9.572640160753936e-06,
+      "loss": 0.0935,
+      "step": 149
+    },
+    {
+      "epoch": 0.2669039145907473,
+      "grad_norm": 0.5970275103319939,
+      "learning_rate": 9.566969082046471e-06,
+      "loss": 0.0862,
+      "step": 150
+    },
+    {
+      "epoch": 0.26868327402135234,
+      "grad_norm": 0.4583173640758323,
+      "learning_rate": 9.561262325798188e-06,
+      "loss": 0.0685,
+      "step": 151
+    },
+    {
+      "epoch": 0.2704626334519573,
+      "grad_norm": 0.5684691359262456,
+      "learning_rate": 9.555519936590739e-06,
+      "loss": 0.0658,
+      "step": 152
+    },
+    {
+      "epoch": 0.2722419928825623,
+      "grad_norm": 0.5693116087660894,
+      "learning_rate": 9.549741959284147e-06,
+      "loss": 0.0722,
+      "step": 153
+    },
+    {
+      "epoch": 0.27402135231316727,
+      "grad_norm": 0.5979140383676496,
+      "learning_rate": 9.543928439016445e-06,
+      "loss": 0.0784,
+      "step": 154
+    },
+    {
+      "epoch": 0.27580071174377224,
+      "grad_norm": 0.6152117674361749,
+      "learning_rate": 9.538079421203339e-06,
+      "loss": 0.0808,
+      "step": 155
+    },
+    {
+      "epoch": 0.2775800711743772,
+      "grad_norm": 0.6244747595287627,
+      "learning_rate": 9.532194951537838e-06,
+      "loss": 0.0808,
+      "step": 156
+    },
+    {
+      "epoch": 0.2793594306049822,
+      "grad_norm": 0.6623086375720926,
+      "learning_rate": 9.52627507598991e-06,
+      "loss": 0.0863,
+      "step": 157
+    },
+    {
+      "epoch": 0.28113879003558717,
+      "grad_norm": 0.5323261989884099,
+      "learning_rate": 9.52031984080611e-06,
+      "loss": 0.0725,
+      "step": 158
+    },
+    {
+      "epoch": 0.28291814946619215,
+      "grad_norm": 0.6040059887255236,
+      "learning_rate": 9.514329292509227e-06,
+      "loss": 0.0712,
+      "step": 159
+    },
+    {
+      "epoch": 0.2846975088967972,
+      "grad_norm": 0.6486204936676071,
+      "learning_rate": 9.508303477897925e-06,
+      "loss": 0.0852,
+      "step": 160
+    },
+    {
+      "epoch": 0.28647686832740216,
+      "grad_norm": 0.7542464545844152,
+      "learning_rate": 9.502242444046365e-06,
+      "loss": 0.0884,
+      "step": 161
+    },
+    {
+      "epoch": 0.28825622775800713,
+      "grad_norm": 0.7210698561490735,
+      "learning_rate": 9.496146238303846e-06,
+      "loss": 0.1053,
+      "step": 162
+    },
+    {
+      "epoch": 0.2900355871886121,
+      "grad_norm": 0.6919197427285332,
+      "learning_rate": 9.49001490829443e-06,
+      "loss": 0.0852,
+      "step": 163
+    },
+    {
+      "epoch": 0.2918149466192171,
+      "grad_norm": 0.7307213394932671,
+      "learning_rate": 9.483848501916578e-06,
+      "loss": 0.1123,
+      "step": 164
+    },
+    {
+      "epoch": 0.29359430604982206,
+      "grad_norm": 0.6737178049112171,
+      "learning_rate": 9.477647067342766e-06,
+      "loss": 0.0916,
+      "step": 165
+    },
+    {
+      "epoch": 0.29537366548042704,
+      "grad_norm": 0.6556180849779367,
+      "learning_rate": 9.471410653019115e-06,
+      "loss": 0.0909,
+      "step": 166
+    },
+    {
+      "epoch": 0.297153024911032,
+      "grad_norm": 0.674264549881035,
+      "learning_rate": 9.46513930766501e-06,
+      "loss": 0.0994,
+      "step": 167
+    },
+    {
+      "epoch": 0.298932384341637,
+      "grad_norm": 0.6074772289852925,
+      "learning_rate": 9.458833080272723e-06,
+      "loss": 0.0833,
+      "step": 168
+    },
+    {
+      "epoch": 0.30071174377224197,
+      "grad_norm": 0.6734630459304292,
+      "learning_rate": 9.45249202010702e-06,
+      "loss": 0.0816,
+      "step": 169
+    },
+    {
+      "epoch": 0.302491103202847,
+      "grad_norm": 0.6719592318255239,
+      "learning_rate": 9.446116176704791e-06,
+      "loss": 0.0781,
+      "step": 170
+    },
+    {
+      "epoch": 0.304270462633452,
+      "grad_norm": 0.6349233926358508,
+      "learning_rate": 9.439705599874653e-06,
+      "loss": 0.0821,
+      "step": 171
+    },
+    {
+      "epoch": 0.30604982206405695,
+      "grad_norm": 0.6629073766546002,
+      "learning_rate": 9.433260339696564e-06,
+      "loss": 0.1016,
+      "step": 172
+    },
+    {
+      "epoch": 0.30782918149466193,
+      "grad_norm": 0.7329910922558953,
+      "learning_rate": 9.426780446521429e-06,
+      "loss": 0.0975,
+      "step": 173
+    },
+    {
+      "epoch": 0.3096085409252669,
+      "grad_norm": 0.6999034501567389,
+      "learning_rate": 9.42026597097071e-06,
+      "loss": 0.0951,
+      "step": 174
+    },
+    {
+      "epoch": 0.3113879003558719,
+      "grad_norm": 0.5661201948845255,
+      "learning_rate": 9.413716963936033e-06,
+      "loss": 0.0736,
+      "step": 175
+    },
+    {
+      "epoch": 0.31316725978647686,
+      "grad_norm": 0.559435121368133,
+      "learning_rate": 9.407133476578778e-06,
+      "loss": 0.0777,
+      "step": 176
+    },
+    {
+      "epoch": 0.31494661921708184,
+      "grad_norm": 0.9040593482233773,
+      "learning_rate": 9.400515560329698e-06,
+      "loss": 0.1196,
+      "step": 177
+    },
+    {
+      "epoch": 0.3167259786476868,
+      "grad_norm": 0.7782287907193186,
+      "learning_rate": 9.393863266888501e-06,
+      "loss": 0.0942,
+      "step": 178
+    },
+    {
+      "epoch": 0.3185053380782918,
+      "grad_norm": 0.723879662922279,
+      "learning_rate": 9.387176648223457e-06,
+      "loss": 0.0885,
+      "step": 179
+    },
+    {
+      "epoch": 0.3202846975088968,
+      "grad_norm": 0.7401619082235917,
+      "learning_rate": 9.38045575657098e-06,
+      "loss": 0.0932,
+      "step": 180
+    },
+    {
+      "epoch": 0.3220640569395018,
+      "grad_norm": 0.7240797342344476,
+      "learning_rate": 9.37370064443524e-06,
+      "loss": 0.0948,
+      "step": 181
+    },
+    {
+      "epoch": 0.3238434163701068,
+      "grad_norm": 0.4908908319509219,
+      "learning_rate": 9.366911364587726e-06,
+      "loss": 0.065,
+      "step": 182
+    },
+    {
+      "epoch": 0.32562277580071175,
+      "grad_norm": 0.5970056883250735,
+      "learning_rate": 9.360087970066854e-06,
+      "loss": 0.0796,
+      "step": 183
+    },
+    {
+      "epoch": 0.3274021352313167,
+      "grad_norm": 0.5727871481805177,
+      "learning_rate": 9.353230514177553e-06,
+      "loss": 0.0723,
+      "step": 184
+    },
+    {
+      "epoch": 0.3291814946619217,
+      "grad_norm": 0.5661263769545877,
+      "learning_rate": 9.346339050490832e-06,
+      "loss": 0.0749,
+      "step": 185
+    },
+    {
+      "epoch": 0.3309608540925267,
+      "grad_norm": 0.7343812706985552,
+      "learning_rate": 9.33941363284338e-06,
+      "loss": 0.0911,
+      "step": 186
+    },
+    {
+      "epoch": 0.33274021352313166,
+      "grad_norm": 0.6255882493467256,
+      "learning_rate": 9.332454315337129e-06,
+      "loss": 0.0774,
+      "step": 187
+    },
+    {
+      "epoch": 0.33451957295373663,
+      "grad_norm": 0.659020405837201,
+      "learning_rate": 9.325461152338846e-06,
+      "loss": 0.0862,
+      "step": 188
+    },
+    {
+      "epoch": 0.33629893238434166,
+      "grad_norm": 0.6273549342248239,
+      "learning_rate": 9.3184341984797e-06,
+      "loss": 0.0842,
+      "step": 189
+    },
+    {
+      "epoch": 0.33807829181494664,
+      "grad_norm": 0.6253907881479073,
+      "learning_rate": 9.311373508654838e-06,
+      "loss": 0.0822,
+      "step": 190
+    },
+    {
+      "epoch": 0.3398576512455516,
+      "grad_norm": 0.7138766085704715,
+      "learning_rate": 9.30427913802295e-06,
+      "loss": 0.1111,
+      "step": 191
+    },
+    {
+      "epoch": 0.3416370106761566,
+      "grad_norm": 0.6881652870610153,
+      "learning_rate": 9.297151142005852e-06,
+      "loss": 0.085,
+      "step": 192
+    },
+    {
+      "epoch": 0.34341637010676157,
+      "grad_norm": 0.5897145531140318,
+      "learning_rate": 9.289989576288035e-06,
+      "loss": 0.0828,
+      "step": 193
+    },
+    {
+      "epoch": 0.34519572953736655,
+      "grad_norm": 0.6476608159716881,
+      "learning_rate": 9.282794496816244e-06,
+      "loss": 0.092,
+      "step": 194
+    },
+    {
+      "epoch": 0.3469750889679715,
+      "grad_norm": 0.6724767144045319,
+      "learning_rate": 9.27556595979904e-06,
+      "loss": 0.079,
+      "step": 195
+    },
+    {
+      "epoch": 0.3487544483985765,
+      "grad_norm": 0.6427256525852164,
+      "learning_rate": 9.26830402170635e-06,
+      "loss": 0.0914,
+      "step": 196
+    },
+    {
+      "epoch": 0.3505338078291815,
+      "grad_norm": 0.6085957633145633,
+      "learning_rate": 9.261008739269035e-06,
+      "loss": 0.0762,
+      "step": 197
+    },
+    {
+      "epoch": 0.35231316725978645,
+      "grad_norm": 0.6500484784291933,
+      "learning_rate": 9.253680169478448e-06,
+      "loss": 0.0874,
+      "step": 198
+    },
+    {
+      "epoch": 0.3540925266903915,
+      "grad_norm": 0.5855077318889239,
+      "learning_rate": 9.246318369585983e-06,
+      "loss": 0.0731,
+      "step": 199
+    },
+    {
+      "epoch": 0.35587188612099646,
+      "grad_norm": 0.8550757328242855,
+      "learning_rate": 9.238923397102629e-06,
+      "loss": 0.1163,
+      "step": 200
+    },
+    {
+      "epoch": 0.35587188612099646,
+      "eval_loss": 0.09775189310312271,
+      "eval_runtime": 1.9319,
+      "eval_samples_per_second": 23.811,
+      "eval_steps_per_second": 6.212,
+      "step": 200
+    },
+    {
+      "epoch": 0.35765124555160144,
+      "grad_norm": 0.5827757104748117,
+      "learning_rate": 9.231495309798525e-06,
+      "loss": 0.076,
+      "step": 201
+    },
+    {
+      "epoch": 0.3594306049822064,
+      "grad_norm": 0.7910040544379672,
+      "learning_rate": 9.224034165702506e-06,
+      "loss": 0.1056,
+      "step": 202
+    },
+    {
+      "epoch": 0.3612099644128114,
+      "grad_norm": 0.47020652771103016,
+      "learning_rate": 9.216540023101646e-06,
+      "loss": 0.0663,
+      "step": 203
+    },
+    {
+      "epoch": 0.36298932384341637,
+      "grad_norm": 0.5457605328526042,
+      "learning_rate": 9.209012940540806e-06,
+      "loss": 0.0767,
+      "step": 204
+    },
+    {
+      "epoch": 0.36476868327402134,
+      "grad_norm": 0.5945070019844252,
+      "learning_rate": 9.20145297682218e-06,
+      "loss": 0.0895,
+      "step": 205
+    },
+    {
+      "epoch": 0.3665480427046263,
+      "grad_norm": 0.831097298745982,
+      "learning_rate": 9.193860191004833e-06,
+      "loss": 0.1163,
+      "step": 206
+    },
+    {
+      "epoch": 0.3683274021352313,
+      "grad_norm": 0.5939208524858341,
+      "learning_rate": 9.186234642404234e-06,
+      "loss": 0.0866,
+      "step": 207
+    },
+    {
+      "epoch": 0.3701067615658363,
+      "grad_norm": 0.7303461400925312,
+      "learning_rate": 9.178576390591803e-06,
+      "loss": 0.0952,
+      "step": 208
+    },
+    {
+      "epoch": 0.3718861209964413,
+      "grad_norm": 0.5613025944912394,
+      "learning_rate": 9.170885495394435e-06,
+      "loss": 0.0677,
+      "step": 209
+    },
+    {
+      "epoch": 0.3736654804270463,
+      "grad_norm": 0.6768219719543376,
+      "learning_rate": 9.16316201689404e-06,
+      "loss": 0.0883,
+      "step": 210
+    },
+    {
+      "epoch": 0.37544483985765126,
+      "grad_norm": 0.6259175546472245,
+      "learning_rate": 9.155406015427076e-06,
+      "loss": 0.0792,
+      "step": 211
+    },
+    {
+      "epoch": 0.37722419928825623,
+      "grad_norm": 0.734745718156219,
+      "learning_rate": 9.147617551584066e-06,
+      "loss": 0.0891,
+      "step": 212
+    },
+    {
+      "epoch": 0.3790035587188612,
+      "grad_norm": 0.5739749459529792,
+      "learning_rate": 9.139796686209135e-06,
+      "loss": 0.0994,
+      "step": 213
+    },
+    {
+      "epoch": 0.3807829181494662,
+      "grad_norm": 0.5326279587618431,
+      "learning_rate": 9.131943480399531e-06,
+      "loss": 0.0713,
+      "step": 214
+    },
+    {
+      "epoch": 0.38256227758007116,
+      "grad_norm": 0.5758309530074663,
+      "learning_rate": 9.124057995505148e-06,
+      "loss": 0.0748,
+      "step": 215
+    },
+    {
+      "epoch": 0.38434163701067614,
+      "grad_norm": 0.8393287295992246,
+      "learning_rate": 9.11614029312805e-06,
+      "loss": 0.1209,
+      "step": 216
+    },
+    {
+      "epoch": 0.3861209964412811,
+      "grad_norm": 0.7311480098939347,
+      "learning_rate": 9.108190435121982e-06,
+      "loss": 0.0876,
+      "step": 217
+    },
+    {
+      "epoch": 0.3879003558718861,
+      "grad_norm": 0.7099465287200297,
+      "learning_rate": 9.100208483591892e-06,
+      "loss": 0.0821,
+      "step": 218
+    },
+    {
+      "epoch": 0.3896797153024911,
+      "grad_norm": 0.8700041252199241,
+      "learning_rate": 9.092194500893448e-06,
+      "loss": 0.1371,
+      "step": 219
+    },
+    {
+      "epoch": 0.3914590747330961,
+      "grad_norm": 0.48633235996160695,
+      "learning_rate": 9.084148549632547e-06,
+      "loss": 0.0625,
+      "step": 220
+    },
+    {
+      "epoch": 0.3932384341637011,
+      "grad_norm": 0.6366463887419176,
+      "learning_rate": 9.076070692664827e-06,
+      "loss": 0.0891,
+      "step": 221
+    },
+    {
+      "epoch": 0.39501779359430605,
+      "grad_norm": 0.6191802975890209,
+      "learning_rate": 9.067960993095176e-06,
+      "loss": 0.0811,
+      "step": 222
+    },
+    {
+      "epoch": 0.39679715302491103,
+      "grad_norm": 0.7142357966133903,
+      "learning_rate": 9.059819514277238e-06,
+      "loss": 0.0908,
+      "step": 223
+    },
+    {
+      "epoch": 0.398576512455516,
+      "grad_norm": 0.5485185495335182,
+      "learning_rate": 9.05164631981292e-06,
+      "loss": 0.0809,
+      "step": 224
+    },
+    {
+      "epoch": 0.400355871886121,
+      "grad_norm": 0.6342671033966139,
+      "learning_rate": 9.043441473551893e-06,
+      "loss": 0.095,
+      "step": 225
+    },
+    {
+      "epoch": 0.40213523131672596,
+      "grad_norm": 0.6666697229095734,
+      "learning_rate": 9.035205039591099e-06,
+      "loss": 0.087,
+      "step": 226
+    },
+    {
+      "epoch": 0.40391459074733094,
+      "grad_norm": 0.7882851422684577,
+      "learning_rate": 9.02693708227424e-06,
+      "loss": 0.0985,
+      "step": 227
+    },
+    {
+      "epoch": 0.40569395017793597,
+      "grad_norm": 0.5817505478200452,
+      "learning_rate": 9.018637666191284e-06,
+      "loss": 0.0995,
+      "step": 228
+    },
+    {
+      "epoch": 0.40747330960854095,
+      "grad_norm": 0.6732707820733036,
+      "learning_rate": 9.010306856177958e-06,
+      "loss": 0.1051,
+      "step": 229
+    },
+    {
+      "epoch": 0.4092526690391459,
+      "grad_norm": 0.7213337413590702,
+      "learning_rate": 9.001944717315236e-06,
+      "loss": 0.1105,
+      "step": 230
+    },
+    {
+      "epoch": 0.4110320284697509,
+      "grad_norm": 0.705425724855184,
+      "learning_rate": 8.993551314928846e-06,
+      "loss": 0.0971,
+      "step": 231
+    },
+    {
+      "epoch": 0.4128113879003559,
+      "grad_norm": 0.5644042838059969,
+      "learning_rate": 8.985126714588739e-06,
+      "loss": 0.069,
+      "step": 232
+    },
+    {
+      "epoch": 0.41459074733096085,
+      "grad_norm": 0.6040660433867808,
+      "learning_rate": 8.976670982108591e-06,
+      "loss": 0.0883,
+      "step": 233
+    },
+    {
+      "epoch": 0.41637010676156583,
+      "grad_norm": 0.7152774714049163,
+      "learning_rate": 8.968184183545285e-06,
+      "loss": 0.1106,
+      "step": 234
+    },
+    {
+      "epoch": 0.4181494661921708,
+      "grad_norm": 0.6794860700282848,
+      "learning_rate": 8.959666385198396e-06,
+      "loss": 0.094,
+      "step": 235
+    },
+    {
+      "epoch": 0.4199288256227758,
+      "grad_norm": 0.5168332741493902,
+      "learning_rate": 8.951117653609666e-06,
+      "loss": 0.0663,
+      "step": 236
+    },
+    {
+      "epoch": 0.42170818505338076,
+      "grad_norm": 0.7237957776846629,
+      "learning_rate": 8.9425380555625e-06,
+      "loss": 0.0912,
+      "step": 237
+    },
+    {
+      "epoch": 0.4234875444839858,
+      "grad_norm": 0.4249680455752117,
+      "learning_rate": 8.933927658081423e-06,
+      "loss": 0.06,
+      "step": 238
+    },
+    {
+      "epoch": 0.42526690391459077,
+      "grad_norm": 0.5877668209736876,
+      "learning_rate": 8.925286528431578e-06,
+      "loss": 0.0862,
+      "step": 239
+    },
+    {
+      "epoch": 0.42704626334519574,
+      "grad_norm": 0.6182690992242768,
+      "learning_rate": 8.916614734118184e-06,
+      "loss": 0.0714,
+      "step": 240
+    },
+    {
+      "epoch": 0.4288256227758007,
+      "grad_norm": 0.5629458245161718,
+      "learning_rate": 8.907912342886016e-06,
+      "loss": 0.0761,
+      "step": 241
+    },
+    {
+      "epoch": 0.4306049822064057,
+      "grad_norm": 0.5500939739622357,
+      "learning_rate": 8.899179422718877e-06,
+      "loss": 0.0743,
+      "step": 242
+    },
+    {
+      "epoch": 0.43238434163701067,
+      "grad_norm": 0.6502559218257742,
+      "learning_rate": 8.890416041839061e-06,
+      "loss": 0.0821,
+      "step": 243
+    },
+    {
+      "epoch": 0.43416370106761565,
+      "grad_norm": 0.5775018575963887,
+      "learning_rate": 8.881622268706825e-06,
+      "loss": 0.0817,
+      "step": 244
+    },
+    {
+      "epoch": 0.4359430604982206,
+      "grad_norm": 0.5329054174202055,
+      "learning_rate": 8.872798172019856e-06,
+      "loss": 0.0682,
+      "step": 245
+    },
+    {
+      "epoch": 0.4377224199288256,
+      "grad_norm": 0.6295860570816871,
+      "learning_rate": 8.863943820712726e-06,
+      "loss": 0.0809,
+      "step": 246
+    },
+    {
+      "epoch": 0.4395017793594306,
+      "grad_norm": 0.6450487888416644,
+      "learning_rate": 8.855059283956363e-06,
+      "loss": 0.1075,
+      "step": 247
+    },
+    {
+      "epoch": 0.4412811387900356,
+      "grad_norm": 0.6247274267077918,
+      "learning_rate": 8.8461446311575e-06,
+      "loss": 0.079,
+      "step": 248
+    },
+    {
+      "epoch": 0.4430604982206406,
+      "grad_norm": 0.5583190678626496,
+      "learning_rate": 8.837199931958147e-06,
+      "loss": 0.0682,
+      "step": 249
+    },
+    {
+      "epoch": 0.44483985765124556,
+      "grad_norm": 0.6635980939676087,
+      "learning_rate": 8.828225256235035e-06,
+      "loss": 0.1014,
+      "step": 250
+    },
+    {
+      "epoch": 0.44661921708185054,
+      "grad_norm": 0.714701104726756,
+      "learning_rate": 8.819220674099074e-06,
+      "loss": 0.0994,
+      "step": 251
+    },
+    {
+      "epoch": 0.4483985765124555,
+      "grad_norm": 0.5670674841344918,
+      "learning_rate": 8.810186255894804e-06,
+      "loss": 0.0717,
+      "step": 252
+    },
+    {
+      "epoch": 0.4501779359430605,
+      "grad_norm": 0.6367810407716749,
+      "learning_rate": 8.801122072199848e-06,
+      "loss": 0.0793,
+      "step": 253
+    },
+    {
+      "epoch": 0.45195729537366547,
+      "grad_norm": 0.5633294435643387,
+      "learning_rate": 8.792028193824364e-06,
+      "loss": 0.0656,
+      "step": 254
+    },
+    {
+      "epoch": 0.45373665480427045,
+      "grad_norm": 0.5482574235979996,
+      "learning_rate": 8.782904691810478e-06,
+      "loss": 0.064,
+      "step": 255
+    },
+    {
+      "epoch": 0.4555160142348754,
+      "grad_norm": 0.5453362862480027,
+      "learning_rate": 8.77375163743175e-06,
+      "loss": 0.0742,
+      "step": 256
+    },
+    {
+      "epoch": 0.45729537366548045,
+      "grad_norm": 0.5470461633803484,
+      "learning_rate": 8.764569102192593e-06,
+      "loss": 0.0821,
+      "step": 257
+    },
+    {
+      "epoch": 0.45907473309608543,
+      "grad_norm": 0.5673337606010138,
+      "learning_rate": 8.755357157827735e-06,
+      "loss": 0.0595,
+      "step": 258
+    },
+    {
+      "epoch": 0.4608540925266904,
+      "grad_norm": 0.6707023423065681,
+      "learning_rate": 8.746115876301651e-06,
+      "loss": 0.0994,
+      "step": 259
+    },
+    {
+      "epoch": 0.4626334519572954,
+      "grad_norm": 0.5530120276506165,
+      "learning_rate": 8.736845329807994e-06,
+      "loss": 0.0737,
+      "step": 260
+    },
+    {
+      "epoch": 0.46441281138790036,
+      "grad_norm": 0.6630059306733089,
+      "learning_rate": 8.727545590769044e-06,
+      "loss": 0.0841,
+      "step": 261
+    },
+    {
+      "epoch": 0.46619217081850534,
+      "grad_norm": 0.5396028366537927,
+      "learning_rate": 8.718216731835131e-06,
+      "loss": 0.072,
+      "step": 262
+    },
+    {
+      "epoch": 0.4679715302491103,
+      "grad_norm": 0.6308080025504466,
+      "learning_rate": 8.708858825884075e-06,
+      "loss": 0.1018,
+      "step": 263
+    },
+    {
+      "epoch": 0.4697508896797153,
+      "grad_norm": 0.4856629897550936,
+      "learning_rate": 8.699471946020612e-06,
+      "loss": 0.0711,
+      "step": 264
+    },
+    {
+      "epoch": 0.47153024911032027,
+      "grad_norm": 0.6188788665108327,
+      "learning_rate": 8.690056165575825e-06,
+      "loss": 0.0795,
+      "step": 265
+    },
+    {
+      "epoch": 0.47330960854092524,
+      "grad_norm": 0.5817833173352943,
+      "learning_rate": 8.680611558106571e-06,
+      "loss": 0.078,
+      "step": 266
+    },
+    {
+      "epoch": 0.4750889679715303,
+      "grad_norm": 0.5516357749066444,
+      "learning_rate": 8.671138197394907e-06,
+      "loss": 0.0701,
+      "step": 267
+    },
+    {
+      "epoch": 0.47686832740213525,
+      "grad_norm": 0.478730052454012,
+      "learning_rate": 8.661636157447511e-06,
+      "loss": 0.075,
+      "step": 268
+    },
+    {
+      "epoch": 0.4786476868327402,
+      "grad_norm": 0.8801401368067325,
+      "learning_rate": 8.652105512495106e-06,
+      "loss": 0.0981,
+      "step": 269
+    },
+    {
+      "epoch": 0.4804270462633452,
+      "grad_norm": 0.613381294333874,
+      "learning_rate": 8.64254633699188e-06,
+      "loss": 0.0746,
+      "step": 270
+    },
+    {
+      "epoch": 0.4822064056939502,
+      "grad_norm": 0.6633199768679631,
+      "learning_rate": 8.632958705614905e-06,
+      "loss": 0.0887,
+      "step": 271
+    },
+    {
+      "epoch": 0.48398576512455516,
+      "grad_norm": 0.6066399926602792,
+      "learning_rate": 8.623342693263549e-06,
+      "loss": 0.0761,
+      "step": 272
+    },
+    {
+      "epoch": 0.48576512455516013,
+      "grad_norm": 0.615589252036537,
+      "learning_rate": 8.6136983750589e-06,
+      "loss": 0.0849,
+      "step": 273
+    },
+    {
+      "epoch": 0.4875444839857651,
+      "grad_norm": 0.6148054492313724,
+      "learning_rate": 8.604025826343167e-06,
+      "loss": 0.0799,
+      "step": 274
+    },
+    {
+      "epoch": 0.4893238434163701,
+      "grad_norm": 0.7707762697948145,
+      "learning_rate": 8.594325122679107e-06,
+      "loss": 0.1019,
+      "step": 275
+    },
+    {
+      "epoch": 0.49110320284697506,
+      "grad_norm": 0.5826593809661291,
+      "learning_rate": 8.584596339849419e-06,
+      "loss": 0.0851,
+      "step": 276
+    },
+    {
+      "epoch": 0.4928825622775801,
+      "grad_norm": 0.4450172230041677,
+      "learning_rate": 8.574839553856157e-06,
+      "loss": 0.0598,
+      "step": 277
+    },
+    {
+      "epoch": 0.49466192170818507,
+      "grad_norm": 0.6846744777066885,
+      "learning_rate": 8.565054840920145e-06,
+      "loss": 0.1028,
+      "step": 278
+    },
+    {
+      "epoch": 0.49644128113879005,
+      "grad_norm": 0.48587612893981863,
+      "learning_rate": 8.55524227748037e-06,
+      "loss": 0.0578,
+      "step": 279
+    },
+    {
+      "epoch": 0.498220640569395,
+      "grad_norm": 0.648463167587547,
+      "learning_rate": 8.545401940193392e-06,
+      "loss": 0.0782,
+      "step": 280
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6336176861253136,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.0824,
+      "step": 281
+    },
+    {
+      "epoch": 0.501779359430605,
+      "grad_norm": 0.8221327250150554,
+      "learning_rate": 8.525638251788312e-06,
+      "loss": 0.1201,
+      "step": 282
+    },
+    {
+      "epoch": 0.50355871886121,
+      "grad_norm": 0.6402201040615425,
+      "learning_rate": 8.515715055065783e-06,
+      "loss": 0.0896,
+      "step": 283
+    },
+    {
+      "epoch": 0.505338078291815,
+      "grad_norm": 0.6134541678932172,
+      "learning_rate": 8.505764393285985e-06,
+      "loss": 0.0987,
+      "step": 284
+    },
+    {
+      "epoch": 0.5071174377224199,
+      "grad_norm": 0.6527781314220821,
+      "learning_rate": 8.495786344184314e-06,
+      "loss": 0.0798,
+      "step": 285
+    },
+    {
+      "epoch": 0.5088967971530249,
+      "grad_norm": 0.6669453789195827,
+      "learning_rate": 8.485780985710113e-06,
+      "loss": 0.0983,
+      "step": 286
+    },
+    {
+      "epoch": 0.5106761565836299,
+      "grad_norm": 0.5620429533070573,
+      "learning_rate": 8.475748396026074e-06,
+      "loss": 0.0796,
+      "step": 287
+    },
+    {
+      "epoch": 0.5124555160142349,
+      "grad_norm": 0.670053036143723,
+      "learning_rate": 8.46568865350762e-06,
+      "loss": 0.0844,
+      "step": 288
+    },
+    {
+      "epoch": 0.5142348754448398,
+      "grad_norm": 0.8169563577982049,
+      "learning_rate": 8.45560183674229e-06,
+      "loss": 0.1069,
+      "step": 289
+    },
+    {
+      "epoch": 0.5160142348754448,
+      "grad_norm": 0.6921516382175138,
+      "learning_rate": 8.445488024529133e-06,
+      "loss": 0.0946,
+      "step": 290
+    },
+    {
+      "epoch": 0.5177935943060499,
+      "grad_norm": 0.526282680545529,
+      "learning_rate": 8.435347295878087e-06,
+      "loss": 0.0624,
+      "step": 291
+    },
+    {
+      "epoch": 0.5195729537366548,
+      "grad_norm": 0.48187582064852913,
+      "learning_rate": 8.425179730009368e-06,
+      "loss": 0.0584,
+      "step": 292
+    },
+    {
+      "epoch": 0.5213523131672598,
+      "grad_norm": 0.6181180575142583,
+      "learning_rate": 8.41498540635284e-06,
+      "loss": 0.0833,
+      "step": 293
+    },
+    {
+      "epoch": 0.5231316725978647,
+      "grad_norm": 0.7263869102966996,
+      "learning_rate": 8.404764404547404e-06,
+      "loss": 0.09,
+      "step": 294
+    },
+    {
+      "epoch": 0.5249110320284698,
+      "grad_norm": 0.6818957144999949,
+      "learning_rate": 8.394516804440374e-06,
+      "loss": 0.0863,
+      "step": 295
+    },
+    {
+      "epoch": 0.5266903914590747,
+      "grad_norm": 0.6766803291187631,
+      "learning_rate": 8.384242686086848e-06,
+      "loss": 0.0815,
+      "step": 296
+    },
+    {
+      "epoch": 0.5284697508896797,
+      "grad_norm": 0.6874252990146803,
+      "learning_rate": 8.373942129749094e-06,
+      "loss": 0.1071,
+      "step": 297
+    },
+    {
+      "epoch": 0.5302491103202847,
+      "grad_norm": 0.5065780358128282,
+      "learning_rate": 8.363615215895908e-06,
+      "loss": 0.0739,
+      "step": 298
+    },
+    {
+      "epoch": 0.5320284697508897,
+      "grad_norm": 0.7169575374595328,
+      "learning_rate": 8.353262025202e-06,
+      "loss": 0.0925,
+      "step": 299
+    },
+    {
+      "epoch": 0.5338078291814946,
+      "grad_norm": 0.5798506663394333,
+      "learning_rate": 8.342882638547351e-06,
+      "loss": 0.0703,
+      "step": 300
+    },
+    {
+      "epoch": 0.5355871886120996,
+      "grad_norm": 0.559469231036395,
+      "learning_rate": 8.332477137016587e-06,
+      "loss": 0.0719,
+      "step": 301
+    },
+    {
+      "epoch": 0.5373665480427047,
+      "grad_norm": 0.5294399405365486,
+      "learning_rate": 8.322045601898354e-06,
+      "loss": 0.0716,
+      "step": 302
+    },
+    {
+      "epoch": 0.5391459074733096,
+      "grad_norm": 0.47526356096146516,
+      "learning_rate": 8.311588114684665e-06,
+      "loss": 0.0627,
+      "step": 303
+    },
+    {
+      "epoch": 0.5409252669039146,
+      "grad_norm": 0.7007198956042635,
+      "learning_rate": 8.301104757070276e-06,
+      "loss": 0.1302,
+      "step": 304
+    },
+    {
+      "epoch": 0.5427046263345195,
+      "grad_norm": 0.6435137578911594,
+      "learning_rate": 8.290595610952045e-06,
+      "loss": 0.08,
+      "step": 305
+    },
+    {
+      "epoch": 0.5444839857651246,
+      "grad_norm": 0.730931680146527,
+      "learning_rate": 8.280060758428294e-06,
+      "loss": 0.1076,
+      "step": 306
+    },
+    {
+      "epoch": 0.5462633451957295,
+      "grad_norm": 0.49081853948760507,
+      "learning_rate": 8.269500281798164e-06,
+      "loss": 0.0588,
+      "step": 307
+    },
+    {
+      "epoch": 0.5480427046263345,
+      "grad_norm": 0.6770080271014127,
+      "learning_rate": 8.258914263560971e-06,
+      "loss": 0.0843,
+      "step": 308
+    },
+    {
+      "epoch": 0.5498220640569395,
+      "grad_norm": 0.5754753326301075,
+      "learning_rate": 8.248302786415567e-06,
+      "loss": 0.0756,
+      "step": 309
+    },
+    {
+      "epoch": 0.5516014234875445,
+      "grad_norm": 0.6768621505158106,
+      "learning_rate": 8.237665933259693e-06,
+      "loss": 0.0819,
+      "step": 310
+    },
+    {
+      "epoch": 0.5533807829181495,
+      "grad_norm": 0.6588591864484897,
+      "learning_rate": 8.227003787189323e-06,
+      "loss": 0.0774,
+      "step": 311
+    },
+    {
+      "epoch": 0.5551601423487544,
+      "grad_norm": 0.5417797803615313,
+      "learning_rate": 8.216316431498028e-06,
+      "loss": 0.0794,
+      "step": 312
+    },
+    {
+      "epoch": 0.5569395017793595,
+      "grad_norm": 0.6076925656033024,
+      "learning_rate": 8.205603949676317e-06,
+      "loss": 0.0932,
+      "step": 313
+    },
+    {
+      "epoch": 0.5587188612099644,
+      "grad_norm": 0.5768262019613668,
+      "learning_rate": 8.194866425410984e-06,
+      "loss": 0.0812,
+      "step": 314
+    },
+    {
+      "epoch": 0.5604982206405694,
+      "grad_norm": 0.6988406288623876,
+      "learning_rate": 8.184103942584456e-06,
+      "loss": 0.0842,
+      "step": 315
+    },
+    {
+      "epoch": 0.5622775800711743,
+      "grad_norm": 0.6359858483020949,
+      "learning_rate": 8.173316585274144e-06,
+      "loss": 0.0779,
+      "step": 316
+    },
+    {
+      "epoch": 0.5640569395017794,
+      "grad_norm": 0.5527909874623004,
+      "learning_rate": 8.162504437751775e-06,
+      "loss": 0.0739,
+      "step": 317
+    },
+    {
+      "epoch": 0.5658362989323843,
+      "grad_norm": 0.5537037611104253,
+      "learning_rate": 8.151667584482742e-06,
+      "loss": 0.0725,
+      "step": 318
+    },
+    {
+      "epoch": 0.5676156583629893,
+      "grad_norm": 0.7068561963318866,
+      "learning_rate": 8.140806110125442e-06,
+      "loss": 0.0969,
+      "step": 319
+    },
+    {
+      "epoch": 0.5693950177935944,
+      "grad_norm": 0.5609851270749815,
+      "learning_rate": 8.129920099530608e-06,
+      "loss": 0.0877,
+      "step": 320
+    },
+    {
+      "epoch": 0.5711743772241993,
+      "grad_norm": 0.4655168011801215,
+      "learning_rate": 8.119009637740663e-06,
+      "loss": 0.0519,
+      "step": 321
+    },
+    {
+      "epoch": 0.5729537366548043,
+      "grad_norm": 0.5355322988860772,
+      "learning_rate": 8.108074809989032e-06,
+      "loss": 0.0738,
+      "step": 322
+    },
+    {
+      "epoch": 0.5747330960854092,
+      "grad_norm": 0.5947891132033333,
+      "learning_rate": 8.097115701699498e-06,
+      "loss": 0.0655,
+      "step": 323
+    },
+    {
+      "epoch": 0.5765124555160143,
+      "grad_norm": 0.6328637555120441,
+      "learning_rate": 8.086132398485525e-06,
+      "loss": 0.0892,
+      "step": 324
+    },
+    {
+      "epoch": 0.5782918149466192,
+      "grad_norm": 0.4183862512736257,
+      "learning_rate": 8.075124986149583e-06,
+      "loss": 0.0581,
+      "step": 325
+    },
+    {
+      "epoch": 0.5800711743772242,
+      "grad_norm": 0.5332379610141228,
+      "learning_rate": 8.064093550682494e-06,
+      "loss": 0.0696,
+      "step": 326
+    },
+    {
+      "epoch": 0.5818505338078291,
+      "grad_norm": 0.6764235311806219,
+      "learning_rate": 8.053038178262742e-06,
+      "loss": 0.0718,
+      "step": 327
+    },
+    {
+      "epoch": 0.5836298932384342,
+      "grad_norm": 0.6283247020704513,
+      "learning_rate": 8.041958955255815e-06,
+      "loss": 0.0891,
+      "step": 328
+    },
+    {
+      "epoch": 0.5854092526690391,
+      "grad_norm": 0.615283153887637,
+      "learning_rate": 8.030855968213518e-06,
+      "loss": 0.0774,
+      "step": 329
+    },
+    {
+      "epoch": 0.5871886120996441,
+      "grad_norm": 0.6674004981859706,
+      "learning_rate": 8.019729303873307e-06,
+      "loss": 0.0922,
+      "step": 330
+    },
+    {
+      "epoch": 0.5889679715302492,
+      "grad_norm": 0.5338918570899855,
+      "learning_rate": 8.008579049157607e-06,
+      "loss": 0.0643,
+      "step": 331
+    },
+    {
+      "epoch": 0.5907473309608541,
+      "grad_norm": 0.46951495630918333,
+      "learning_rate": 7.99740529117313e-06,
+      "loss": 0.0687,
+      "step": 332
+    },
+    {
+      "epoch": 0.5925266903914591,
+      "grad_norm": 0.6450024637575998,
+      "learning_rate": 7.986208117210198e-06,
+      "loss": 0.0881,
+      "step": 333
+    },
+    {
+      "epoch": 0.594306049822064,
+      "grad_norm": 0.6141195944510839,
+      "learning_rate": 7.974987614742066e-06,
+      "loss": 0.0796,
+      "step": 334
+    },
+    {
+      "epoch": 0.5960854092526691,
+      "grad_norm": 0.8023385376103279,
+      "learning_rate": 7.963743871424224e-06,
+      "loss": 0.1086,
+      "step": 335
+    },
+    {
+      "epoch": 0.597864768683274,
+      "grad_norm": 0.5724038241885593,
+      "learning_rate": 7.952476975093729e-06,
+      "loss": 0.069,
+      "step": 336
+    },
+    {
+      "epoch": 0.599644128113879,
+      "grad_norm": 0.7438664476942817,
+      "learning_rate": 7.941187013768508e-06,
+      "loss": 0.1026,
+      "step": 337
+    },
+    {
+      "epoch": 0.6014234875444839,
+      "grad_norm": 0.6488736528342951,
+      "learning_rate": 7.929874075646673e-06,
+      "loss": 0.0785,
+      "step": 338
+    },
+    {
+      "epoch": 0.603202846975089,
+      "grad_norm": 0.6409090448347048,
+      "learning_rate": 7.918538249105835e-06,
+      "loss": 0.0722,
+      "step": 339
+    },
+    {
+      "epoch": 0.604982206405694,
+      "grad_norm": 0.6899222158655678,
+      "learning_rate": 7.907179622702409e-06,
+      "loss": 0.0935,
+      "step": 340
+    },
+    {
+      "epoch": 0.6067615658362989,
+      "grad_norm": 0.5113893894283309,
+      "learning_rate": 7.895798285170927e-06,
+      "loss": 0.0725,
+      "step": 341
+    },
+    {
+      "epoch": 0.608540925266904,
+      "grad_norm": 0.5412264458405445,
+      "learning_rate": 7.88439432542334e-06,
+      "loss": 0.0609,
+      "step": 342
+    },
+    {
+      "epoch": 0.6103202846975089,
+      "grad_norm": 0.6123342276863017,
+      "learning_rate": 7.872967832548327e-06,
+      "loss": 0.0892,
+      "step": 343
+    },
+    {
+      "epoch": 0.6120996441281139,
+      "grad_norm": 0.7052381103397575,
+      "learning_rate": 7.861518895810597e-06,
+      "loss": 0.1108,
+      "step": 344
+    },
+    {
+      "epoch": 0.6138790035587188,
+      "grad_norm": 0.5868954658849361,
+      "learning_rate": 7.850047604650188e-06,
+      "loss": 0.0908,
+      "step": 345
+    },
+    {
+      "epoch": 0.6156583629893239,
+      "grad_norm": 0.7183887693689837,
+      "learning_rate": 7.838554048681783e-06,
+      "loss": 0.1014,
+      "step": 346
+    },
+    {
+      "epoch": 0.6174377224199288,
+      "grad_norm": 0.6926236981647969,
+      "learning_rate": 7.827038317693988e-06,
+      "loss": 0.1197,
+      "step": 347
+    },
+    {
+      "epoch": 0.6192170818505338,
+      "grad_norm": 0.5583766040997787,
+      "learning_rate": 7.815500501648654e-06,
+      "loss": 0.0667,
+      "step": 348
+    },
+    {
+      "epoch": 0.6209964412811388,
+      "grad_norm": 0.6448890818580357,
+      "learning_rate": 7.80394069068015e-06,
+      "loss": 0.0901,
+      "step": 349
+    },
+    {
+      "epoch": 0.6227758007117438,
+      "grad_norm": 0.6644576773681511,
+      "learning_rate": 7.79235897509468e-06,
+      "loss": 0.0841,
+      "step": 350
+    },
+    {
+      "epoch": 0.6245551601423488,
+      "grad_norm": 0.6554253477295041,
+      "learning_rate": 7.780755445369563e-06,
+      "loss": 0.0868,
+      "step": 351
+    },
+    {
+      "epoch": 0.6263345195729537,
+      "grad_norm": 0.7440996867202055,
+      "learning_rate": 7.769130192152538e-06,
+      "loss": 0.1066,
+      "step": 352
+    },
+    {
+      "epoch": 0.6281138790035588,
+      "grad_norm": 0.6677510957052545,
+      "learning_rate": 7.757483306261042e-06,
+      "loss": 0.0982,
+      "step": 353
+    },
+    {
+      "epoch": 0.6298932384341637,
+      "grad_norm": 0.8220111924883771,
+      "learning_rate": 7.745814878681516e-06,
+      "loss": 0.1102,
+      "step": 354
+    },
+    {
+      "epoch": 0.6316725978647687,
+      "grad_norm": 0.5992600939349928,
+      "learning_rate": 7.734125000568684e-06,
+      "loss": 0.0832,
+      "step": 355
+    },
+    {
+      "epoch": 0.6334519572953736,
+      "grad_norm": 0.5913819704745602,
+      "learning_rate": 7.722413763244837e-06,
+      "loss": 0.0766,
+      "step": 356
+    },
+    {
+      "epoch": 0.6352313167259787,
+      "grad_norm": 0.4575752837241832,
+      "learning_rate": 7.710681258199136e-06,
+      "loss": 0.0675,
+      "step": 357
+    },
+    {
+      "epoch": 0.6370106761565836,
+      "grad_norm": 0.6193667069029222,
+      "learning_rate": 7.69892757708688e-06,
+      "loss": 0.067,
+      "step": 358
+    },
+    {
+      "epoch": 0.6387900355871886,
+      "grad_norm": 0.58104391221812,
+      "learning_rate": 7.687152811728799e-06,
+      "loss": 0.0775,
+      "step": 359
+    },
+    {
+      "epoch": 0.6405693950177936,
+      "grad_norm": 0.6256158730816243,
+      "learning_rate": 7.675357054110337e-06,
+      "loss": 0.0867,
+      "step": 360
+    },
+    {
+      "epoch": 0.6423487544483986,
+      "grad_norm": 0.5481591302548908,
+      "learning_rate": 7.663540396380931e-06,
+      "loss": 0.0727,
+      "step": 361
+    },
+    {
+      "epoch": 0.6441281138790036,
+      "grad_norm": 0.535754939441803,
+      "learning_rate": 7.651702930853287e-06,
+      "loss": 0.0642,
+      "step": 362
+    },
+    {
+      "epoch": 0.6459074733096085,
+      "grad_norm": 0.6133191042647672,
+      "learning_rate": 7.639844750002668e-06,
+      "loss": 0.0771,
+      "step": 363
+    },
+    {
+      "epoch": 0.6476868327402135,
+      "grad_norm": 0.594623046355906,
+      "learning_rate": 7.627965946466167e-06,
+      "loss": 0.0908,
+      "step": 364
+    },
+    {
+      "epoch": 0.6494661921708185,
+      "grad_norm": 0.6130316019108981,
+      "learning_rate": 7.616066613041977e-06,
+      "loss": 0.083,
+      "step": 365
+    },
+    {
+      "epoch": 0.6512455516014235,
+      "grad_norm": 0.6052953871726992,
+      "learning_rate": 7.6041468426886785e-06,
+      "loss": 0.0756,
+      "step": 366
+    },
+    {
+      "epoch": 0.6530249110320284,
+      "grad_norm": 0.47099821392322444,
+      "learning_rate": 7.592206728524507e-06,
+      "loss": 0.0537,
+      "step": 367
+    },
+    {
+      "epoch": 0.6548042704626335,
+      "grad_norm": 0.6695703323107953,
+      "learning_rate": 7.580246363826621e-06,
+      "loss": 0.0837,
+      "step": 368
+    },
+    {
+      "epoch": 0.6565836298932385,
+      "grad_norm": 0.7525684301360943,
+      "learning_rate": 7.568265842030381e-06,
+      "loss": 0.094,
+      "step": 369
+    },
+    {
+      "epoch": 0.6583629893238434,
+      "grad_norm": 0.5699097841883088,
+      "learning_rate": 7.556265256728618e-06,
+      "loss": 0.0844,
+      "step": 370
+    },
+    {
+      "epoch": 0.6601423487544484,
+      "grad_norm": 0.5363897803060782,
+      "learning_rate": 7.544244701670894e-06,
+      "loss": 0.0774,
+      "step": 371
+    },
+    {
+      "epoch": 0.6619217081850534,
+      "grad_norm": 0.6555054569157386,
+      "learning_rate": 7.532204270762786e-06,
+      "loss": 0.086,
+      "step": 372
+    },
+    {
+      "epoch": 0.6637010676156584,
+      "grad_norm": 0.5240413201814093,
+      "learning_rate": 7.520144058065133e-06,
+      "loss": 0.0625,
+      "step": 373
+    },
+    {
+      "epoch": 0.6654804270462633,
+      "grad_norm": 0.5768911338079254,
+      "learning_rate": 7.50806415779332e-06,
+      "loss": 0.0677,
+      "step": 374
+    },
+    {
+      "epoch": 0.6672597864768683,
+      "grad_norm": 0.6579420235799222,
+      "learning_rate": 7.495964664316525e-06,
+      "loss": 0.0875,
+      "step": 375
+    },
+    {
+      "epoch": 0.6690391459074733,
+      "grad_norm": 0.6742118122452695,
+      "learning_rate": 7.4838456721569975e-06,
+      "loss": 0.0867,
+      "step": 376
+    },
+    {
+      "epoch": 0.6708185053380783,
+      "grad_norm": 0.4999372791533043,
+      "learning_rate": 7.471707275989304e-06,
+      "loss": 0.0674,
+      "step": 377
+    },
+    {
+      "epoch": 0.6725978647686833,
+      "grad_norm": 0.5598430424260302,
+      "learning_rate": 7.459549570639602e-06,
+      "loss": 0.0766,
+      "step": 378
+    },
+    {
+      "epoch": 0.6743772241992882,
+      "grad_norm": 0.6422859491137244,
+      "learning_rate": 7.447372651084896e-06,
+      "loss": 0.0846,
+      "step": 379
+    },
+    {
+      "epoch": 0.6761565836298933,
+      "grad_norm": 0.5900710727978363,
+      "learning_rate": 7.435176612452286e-06,
+      "loss": 0.0704,
+      "step": 380
+    },
+    {
+      "epoch": 0.6779359430604982,
+      "grad_norm": 0.5500655493435561,
+      "learning_rate": 7.4229615500182396e-06,
+      "loss": 0.0773,
+      "step": 381
+    },
+    {
+      "epoch": 0.6797153024911032,
+      "grad_norm": 0.49174002715768744,
+      "learning_rate": 7.4107275592078345e-06,
+      "loss": 0.0622,
+      "step": 382
+    },
+    {
+      "epoch": 0.6814946619217082,
+      "grad_norm": 0.46349679059114945,
+      "learning_rate": 7.398474735594022e-06,
+      "loss": 0.0528,
+      "step": 383
+    },
+    {
+      "epoch": 0.6832740213523132,
+      "grad_norm": 0.623276564216402,
+      "learning_rate": 7.386203174896872e-06,
+      "loss": 0.0774,
+      "step": 384
+    },
+    {
+      "epoch": 0.6850533807829181,
+      "grad_norm": 0.6228253714651386,
+      "learning_rate": 7.373912972982838e-06,
+      "loss": 0.0759,
+      "step": 385
+    },
+    {
+      "epoch": 0.6868327402135231,
+      "grad_norm": 0.5512703147099375,
+      "learning_rate": 7.361604225863992e-06,
+      "loss": 0.0637,
+      "step": 386
+    },
+    {
+      "epoch": 0.6886120996441281,
+      "grad_norm": 0.6808629833111032,
+      "learning_rate": 7.349277029697287e-06,
+      "loss": 0.0809,
+      "step": 387
+    },
+    {
+      "epoch": 0.6903914590747331,
+      "grad_norm": 0.6430323304489203,
+      "learning_rate": 7.336931480783801e-06,
+      "loss": 0.075,
+      "step": 388
+    },
+    {
+      "epoch": 0.6921708185053381,
+      "grad_norm": 0.755589634737508,
+      "learning_rate": 7.3245676755679854e-06,
+      "loss": 0.0883,
+      "step": 389
+    },
+    {
+      "epoch": 0.693950177935943,
+      "grad_norm": 0.5996697136060419,
+      "learning_rate": 7.312185710636911e-06,
+      "loss": 0.0816,
+      "step": 390
+    },
+    {
+      "epoch": 0.6957295373665481,
+      "grad_norm": 0.568082212674168,
+      "learning_rate": 7.299785682719512e-06,
+      "loss": 0.0587,
+      "step": 391
+    },
+    {
+      "epoch": 0.697508896797153,
+      "grad_norm": 0.5573848563576075,
+      "learning_rate": 7.287367688685835e-06,
+      "loss": 0.0705,
+      "step": 392
+    },
+    {
+      "epoch": 0.699288256227758,
+      "grad_norm": 0.8119069939159559,
+      "learning_rate": 7.274931825546279e-06,
+      "loss": 0.0914,
+      "step": 393
+    },
+    {
+      "epoch": 0.701067615658363,
+      "grad_norm": 0.5970507724112131,
+      "learning_rate": 7.262478190450834e-06,
+      "loss": 0.0722,
+      "step": 394
+    },
+    {
+      "epoch": 0.702846975088968,
+      "grad_norm": 0.6689271324290902,
+      "learning_rate": 7.250006880688332e-06,
+      "loss": 0.0939,
+      "step": 395
+    },
+    {
+      "epoch": 0.7046263345195729,
+      "grad_norm": 0.6435531744179841,
+      "learning_rate": 7.2375179936856775e-06,
+      "loss": 0.0775,
+      "step": 396
+    },
+    {
+      "epoch": 0.7064056939501779,
+      "grad_norm": 0.5799223482447977,
+      "learning_rate": 7.22501162700709e-06,
+      "loss": 0.0815,
+      "step": 397
+    },
+    {
+      "epoch": 0.708185053380783,
+      "grad_norm": 0.5816851433415442,
+      "learning_rate": 7.21248787835334e-06,
+      "loss": 0.0661,
+      "step": 398
+    },
+    {
+      "epoch": 0.7099644128113879,
+      "grad_norm": 0.5686561981171016,
+      "learning_rate": 7.199946845560994e-06,
+      "loss": 0.0661,
+      "step": 399
+    },
+    {
+      "epoch": 0.7117437722419929,
+      "grad_norm": 0.8096878162451965,
+      "learning_rate": 7.1873886266016365e-06,
+      "loss": 0.1274,
+      "step": 400
+    },
+    {
+      "epoch": 0.7117437722419929,
+      "eval_loss": 0.09619853645563126,
+      "eval_runtime": 1.902,
+      "eval_samples_per_second": 24.185,
+      "eval_steps_per_second": 6.309,
+      "step": 400
+    },
+    {
+      "epoch": 0.7135231316725978,
+      "grad_norm": 0.6308288006427037,
+      "learning_rate": 7.174813319581115e-06,
+      "loss": 0.0876,
+      "step": 401
+    },
+    {
+      "epoch": 0.7153024911032029,
+      "grad_norm": 1.0609043381975727,
+      "learning_rate": 7.162221022738768e-06,
+      "loss": 0.1054,
+      "step": 402
+    },
+    {
+      "epoch": 0.7170818505338078,
+      "grad_norm": 0.6540996209556451,
+      "learning_rate": 7.149611834446664e-06,
+      "loss": 0.0997,
+      "step": 403
+    },
+    {
+      "epoch": 0.7188612099644128,
+      "grad_norm": 0.6663683512803799,
+      "learning_rate": 7.136985853208824e-06,
+      "loss": 0.0863,
+      "step": 404
+    },
+    {
+      "epoch": 0.7206405693950177,
+      "grad_norm": 0.6207898889399628,
+      "learning_rate": 7.124343177660462e-06,
+      "loss": 0.0845,
+      "step": 405
+    },
+    {
+      "epoch": 0.7224199288256228,
+      "grad_norm": 0.8761810121212872,
+      "learning_rate": 7.111683906567206e-06,
+      "loss": 0.1094,
+      "step": 406
+    },
+    {
+      "epoch": 0.7241992882562278,
+      "grad_norm": 0.6899582396747612,
+      "learning_rate": 7.099008138824329e-06,
+      "loss": 0.0837,
+      "step": 407
+    },
+    {
+      "epoch": 0.7259786476868327,
+      "grad_norm": 0.5015205633733023,
+      "learning_rate": 7.086315973455982e-06,
+      "loss": 0.0728,
+      "step": 408
+    },
+    {
+      "epoch": 0.7277580071174378,
+      "grad_norm": 0.6873783556101611,
+      "learning_rate": 7.0736075096144084e-06,
+      "loss": 0.0956,
+      "step": 409
+    },
+    {
+      "epoch": 0.7295373665480427,
+      "grad_norm": 0.5598391468094712,
+      "learning_rate": 7.060882846579182e-06,
+      "loss": 0.0703,
+      "step": 410
+    },
+    {
+      "epoch": 0.7313167259786477,
+      "grad_norm": 0.596357202385122,
+      "learning_rate": 7.048142083756427e-06,
+      "loss": 0.0849,
+      "step": 411
+    },
+    {
+      "epoch": 0.7330960854092526,
+      "grad_norm": 0.6276227649322608,
+      "learning_rate": 7.035385320678035e-06,
+      "loss": 0.086,
+      "step": 412
+    },
+    {
+      "epoch": 0.7348754448398577,
+      "grad_norm": 0.591332078902182,
+      "learning_rate": 7.022612657000898e-06,
+      "loss": 0.0812,
+      "step": 413
+    },
+    {
+      "epoch": 0.7366548042704626,
+      "grad_norm": 0.582266003308539,
+      "learning_rate": 7.0098241925061215e-06,
+      "loss": 0.0803,
+      "step": 414
+    },
+    {
+      "epoch": 0.7384341637010676,
+      "grad_norm": 0.7813817236673202,
+      "learning_rate": 6.997020027098249e-06,
+      "loss": 0.0918,
+      "step": 415
+    },
+    {
+      "epoch": 0.7402135231316725,
+      "grad_norm": 0.5692349269379092,
+      "learning_rate": 6.9842002608044844e-06,
+      "loss": 0.0842,
+      "step": 416
+    },
+    {
+      "epoch": 0.7419928825622776,
+      "grad_norm": 0.5596027770547295,
+      "learning_rate": 6.971364993773901e-06,
+      "loss": 0.0767,
+      "step": 417
+    },
+    {
+      "epoch": 0.7437722419928826,
+      "grad_norm": 0.6106199025093823,
+      "learning_rate": 6.958514326276669e-06,
+      "loss": 0.0723,
+      "step": 418
+    },
+    {
+      "epoch": 0.7455516014234875,
+      "grad_norm": 0.5907693066110888,
+      "learning_rate": 6.945648358703269e-06,
+      "loss": 0.0683,
+      "step": 419
+    },
+    {
+      "epoch": 0.7473309608540926,
+      "grad_norm": 0.6040945105664361,
+      "learning_rate": 6.932767191563703e-06,
+      "loss": 0.1002,
+      "step": 420
+    },
+    {
+      "epoch": 0.7491103202846975,
+      "grad_norm": 0.5885777668275425,
+      "learning_rate": 6.919870925486718e-06,
+      "loss": 0.063,
+      "step": 421
+    },
+    {
+      "epoch": 0.7508896797153025,
+      "grad_norm": 0.4882635440212406,
+      "learning_rate": 6.906959661219011e-06,
+      "loss": 0.0702,
+      "step": 422
+    },
+    {
+      "epoch": 0.7526690391459074,
+      "grad_norm": 0.5559806306949177,
+      "learning_rate": 6.8940334996244505e-06,
+      "loss": 0.0758,
+      "step": 423
+    },
+    {
+      "epoch": 0.7544483985765125,
+      "grad_norm": 0.5634294498054987,
+      "learning_rate": 6.881092541683279e-06,
+      "loss": 0.0691,
+      "step": 424
+    },
+    {
+      "epoch": 0.7562277580071174,
+      "grad_norm": 0.6225537930949692,
+      "learning_rate": 6.8681368884913345e-06,
+      "loss": 0.0678,
+      "step": 425
+    },
+    {
+      "epoch": 0.7580071174377224,
+      "grad_norm": 0.5075541365177242,
+      "learning_rate": 6.855166641259252e-06,
+      "loss": 0.0719,
+      "step": 426
+    },
+    {
+      "epoch": 0.7597864768683275,
+      "grad_norm": 0.5393158595361945,
+      "learning_rate": 6.8421819013116766e-06,
+      "loss": 0.0783,
+      "step": 427
+    },
+    {
+      "epoch": 0.7615658362989324,
+      "grad_norm": 0.5613388688697452,
+      "learning_rate": 6.829182770086474e-06,
+      "loss": 0.0837,
+      "step": 428
+    },
+    {
+      "epoch": 0.7633451957295374,
+      "grad_norm": 0.5892299718956402,
+      "learning_rate": 6.816169349133934e-06,
+      "loss": 0.075,
+      "step": 429
+    },
+    {
+      "epoch": 0.7651245551601423,
+      "grad_norm": 0.48296551597226545,
+      "learning_rate": 6.803141740115979e-06,
+      "loss": 0.0537,
+      "step": 430
+    },
+    {
+      "epoch": 0.7669039145907474,
+      "grad_norm": 0.8689291679790655,
+      "learning_rate": 6.7901000448053676e-06,
+      "loss": 0.1132,
+      "step": 431
+    },
+    {
+      "epoch": 0.7686832740213523,
+      "grad_norm": 0.4699852303251046,
+      "learning_rate": 6.777044365084907e-06,
+      "loss": 0.0547,
+      "step": 432
+    },
+    {
+      "epoch": 0.7704626334519573,
+      "grad_norm": 0.5415176369615972,
+      "learning_rate": 6.763974802946649e-06,
+      "loss": 0.0708,
+      "step": 433
+    },
+    {
+      "epoch": 0.7722419928825622,
+      "grad_norm": 0.6246160739181148,
+      "learning_rate": 6.750891460491093e-06,
+      "loss": 0.0975,
+      "step": 434
+    },
+    {
+      "epoch": 0.7740213523131673,
+      "grad_norm": 0.653075432757108,
+      "learning_rate": 6.737794439926395e-06,
+      "loss": 0.0989,
+      "step": 435
+    },
+    {
+      "epoch": 0.7758007117437722,
+      "grad_norm": 0.5778833564472982,
+      "learning_rate": 6.724683843567567e-06,
+      "loss": 0.0854,
+      "step": 436
+    },
+    {
+      "epoch": 0.7775800711743772,
+      "grad_norm": 0.6528192216481189,
+      "learning_rate": 6.711559773835672e-06,
+      "loss": 0.082,
+      "step": 437
+    },
+    {
+      "epoch": 0.7793594306049823,
+      "grad_norm": 0.69609065721773,
+      "learning_rate": 6.69842233325703e-06,
+      "loss": 0.0893,
+      "step": 438
+    },
+    {
+      "epoch": 0.7811387900355872,
+      "grad_norm": 0.733027771758474,
+      "learning_rate": 6.685271624462416e-06,
+      "loss": 0.0965,
+      "step": 439
+    },
+    {
+      "epoch": 0.7829181494661922,
+      "grad_norm": 0.5814951186497005,
+      "learning_rate": 6.672107750186255e-06,
+      "loss": 0.0762,
+      "step": 440
+    },
+    {
+      "epoch": 0.7846975088967971,
+      "grad_norm": 0.43083535009005236,
+      "learning_rate": 6.658930813265825e-06,
+      "loss": 0.0605,
+      "step": 441
+    },
+    {
+      "epoch": 0.7864768683274022,
+      "grad_norm": 0.5828700920748653,
+      "learning_rate": 6.645740916640449e-06,
+      "loss": 0.0716,
+      "step": 442
+    },
+    {
+      "epoch": 0.7882562277580071,
+      "grad_norm": 0.47346142035086036,
+      "learning_rate": 6.63253816335069e-06,
+      "loss": 0.0582,
+      "step": 443
+    },
+    {
+      "epoch": 0.7900355871886121,
+      "grad_norm": 0.5541391972925614,
+      "learning_rate": 6.619322656537552e-06,
+      "loss": 0.074,
+      "step": 444
+    },
+    {
+      "epoch": 0.791814946619217,
+      "grad_norm": 0.6017300902633588,
+      "learning_rate": 6.606094499441671e-06,
+      "loss": 0.0775,
+      "step": 445
+    },
+    {
+      "epoch": 0.7935943060498221,
+      "grad_norm": 0.5820072495888647,
+      "learning_rate": 6.592853795402502e-06,
+      "loss": 0.0709,
+      "step": 446
+    },
+    {
+      "epoch": 0.7953736654804271,
+      "grad_norm": 0.6003721329145758,
+      "learning_rate": 6.579600647857525e-06,
+      "loss": 0.0775,
+      "step": 447
+    },
+    {
+      "epoch": 0.797153024911032,
+      "grad_norm": 0.8004333977530397,
+      "learning_rate": 6.566335160341425e-06,
+      "loss": 0.1097,
+      "step": 448
+    },
+    {
+      "epoch": 0.798932384341637,
+      "grad_norm": 0.49847691592505233,
+      "learning_rate": 6.553057436485289e-06,
+      "loss": 0.0684,
+      "step": 449
+    },
+    {
+      "epoch": 0.800711743772242,
+      "grad_norm": 0.551872115687283,
+      "learning_rate": 6.539767580015799e-06,
+      "loss": 0.0912,
+      "step": 450
+    },
+    {
+      "epoch": 0.802491103202847,
+      "grad_norm": 0.7739372493913477,
+      "learning_rate": 6.52646569475441e-06,
+      "loss": 0.0889,
+      "step": 451
+    },
+    {
+      "epoch": 0.8042704626334519,
+      "grad_norm": 0.6449890312051941,
+      "learning_rate": 6.513151884616556e-06,
+      "loss": 0.0946,
+      "step": 452
+    },
+    {
+      "epoch": 0.806049822064057,
+      "grad_norm": 0.7179500571965539,
+      "learning_rate": 6.499826253610823e-06,
+      "loss": 0.0836,
+      "step": 453
+    },
+    {
+      "epoch": 0.8078291814946619,
+      "grad_norm": 0.56933951755536,
+      "learning_rate": 6.486488905838143e-06,
+      "loss": 0.0648,
+      "step": 454
+    },
+    {
+      "epoch": 0.8096085409252669,
+      "grad_norm": 0.5488266152215692,
+      "learning_rate": 6.473139945490984e-06,
+      "loss": 0.0645,
+      "step": 455
+    },
+    {
+      "epoch": 0.8113879003558719,
+      "grad_norm": 0.6694489074626839,
+      "learning_rate": 6.459779476852528e-06,
+      "loss": 0.0902,
+      "step": 456
+    },
+    {
+      "epoch": 0.8131672597864769,
+      "grad_norm": 0.6092097866858663,
+      "learning_rate": 6.446407604295863e-06,
+      "loss": 0.0915,
+      "step": 457
+    },
+    {
+      "epoch": 0.8149466192170819,
+      "grad_norm": 0.5913140106453859,
+      "learning_rate": 6.433024432283169e-06,
+      "loss": 0.0673,
+      "step": 458
+    },
+    {
+      "epoch": 0.8167259786476868,
+      "grad_norm": 0.6893172625244806,
+      "learning_rate": 6.41963006536489e-06,
+      "loss": 0.083,
+      "step": 459
+    },
+    {
+      "epoch": 0.8185053380782918,
+      "grad_norm": 0.5598448746539126,
+      "learning_rate": 6.4062246081789316e-06,
+      "loss": 0.0786,
+      "step": 460
+    },
+    {
+      "epoch": 0.8202846975088968,
+      "grad_norm": 0.5177335294634069,
+      "learning_rate": 6.392808165449836e-06,
+      "loss": 0.0563,
+      "step": 461
+    },
+    {
+      "epoch": 0.8220640569395018,
+      "grad_norm": 0.5842558783867281,
+      "learning_rate": 6.379380841987965e-06,
+      "loss": 0.0771,
+      "step": 462
+    },
+    {
+      "epoch": 0.8238434163701067,
+      "grad_norm": 0.5688880899797458,
+      "learning_rate": 6.365942742688684e-06,
+      "loss": 0.0643,
+      "step": 463
+    },
+    {
+      "epoch": 0.8256227758007118,
+      "grad_norm": 0.7234733036501532,
+      "learning_rate": 6.352493972531535e-06,
+      "loss": 0.0955,
+      "step": 464
+    },
+    {
+      "epoch": 0.8274021352313167,
+      "grad_norm": 0.7462013341152287,
+      "learning_rate": 6.339034636579425e-06,
+      "loss": 0.0927,
+      "step": 465
+    },
+    {
+      "epoch": 0.8291814946619217,
+      "grad_norm": 0.5405760975761205,
+      "learning_rate": 6.325564839977802e-06,
+      "loss": 0.0638,
+      "step": 466
+    },
+    {
+      "epoch": 0.8309608540925267,
+      "grad_norm": 0.5128923165474947,
+      "learning_rate": 6.312084687953835e-06,
+      "loss": 0.059,
+      "step": 467
+    },
+    {
+      "epoch": 0.8327402135231317,
+      "grad_norm": 0.6793493365157661,
+      "learning_rate": 6.298594285815585e-06,
+      "loss": 0.0868,
+      "step": 468
+    },
+    {
+      "epoch": 0.8345195729537367,
+      "grad_norm": 0.6922815755694199,
+      "learning_rate": 6.2850937389511936e-06,
+      "loss": 0.0989,
+      "step": 469
+    },
+    {
+      "epoch": 0.8362989323843416,
+      "grad_norm": 0.5426327245961869,
+      "learning_rate": 6.271583152828049e-06,
+      "loss": 0.0592,
+      "step": 470
+    },
+    {
+      "epoch": 0.8380782918149466,
+      "grad_norm": 0.5437526609838257,
+      "learning_rate": 6.258062632991972e-06,
+      "loss": 0.0587,
+      "step": 471
+    },
+    {
+      "epoch": 0.8398576512455516,
+      "grad_norm": 0.6266086864737455,
+      "learning_rate": 6.244532285066382e-06,
+      "loss": 0.0798,
+      "step": 472
+    },
+    {
+      "epoch": 0.8416370106761566,
+      "grad_norm": 0.5360768999254263,
+      "learning_rate": 6.2309922147514775e-06,
+      "loss": 0.0717,
+      "step": 473
+    },
+    {
+      "epoch": 0.8434163701067615,
+      "grad_norm": 0.5266686440757316,
+      "learning_rate": 6.2174425278234115e-06,
+      "loss": 0.0898,
+      "step": 474
+    },
+    {
+      "epoch": 0.8451957295373665,
+      "grad_norm": 0.5994448330310465,
+      "learning_rate": 6.20388333013346e-06,
+      "loss": 0.0609,
+      "step": 475
+    },
+    {
+      "epoch": 0.8469750889679716,
+      "grad_norm": 0.6229884218165104,
+      "learning_rate": 6.190314727607196e-06,
+      "loss": 0.091,
+      "step": 476
+    },
+    {
+      "epoch": 0.8487544483985765,
+      "grad_norm": 0.614298623698568,
+      "learning_rate": 6.176736826243671e-06,
+      "loss": 0.0822,
+      "step": 477
+    },
+    {
+      "epoch": 0.8505338078291815,
+      "grad_norm": 0.6243400643280597,
+      "learning_rate": 6.163149732114571e-06,
+      "loss": 0.0847,
+      "step": 478
+    },
+    {
+      "epoch": 0.8523131672597865,
+      "grad_norm": 0.5356652919838362,
+      "learning_rate": 6.149553551363404e-06,
+      "loss": 0.0609,
+      "step": 479
+    },
+    {
+      "epoch": 0.8540925266903915,
+      "grad_norm": 0.6643524587935227,
+      "learning_rate": 6.1359483902046605e-06,
+      "loss": 0.0907,
+      "step": 480
+    },
+    {
+      "epoch": 0.8558718861209964,
+      "grad_norm": 0.5974234300279159,
+      "learning_rate": 6.122334354922984e-06,
+      "loss": 0.0732,
+      "step": 481
+    },
+    {
+      "epoch": 0.8576512455516014,
+      "grad_norm": 0.5426476099437977,
+      "learning_rate": 6.108711551872347e-06,
+      "loss": 0.0753,
+      "step": 482
+    },
+    {
+      "epoch": 0.8594306049822064,
+      "grad_norm": 0.6560339520192112,
+      "learning_rate": 6.095080087475218e-06,
+      "loss": 0.075,
+      "step": 483
+    },
+    {
+      "epoch": 0.8612099644128114,
+      "grad_norm": 0.6236602827051071,
+      "learning_rate": 6.0814400682217236e-06,
+      "loss": 0.092,
+      "step": 484
+    },
+    {
+      "epoch": 0.8629893238434164,
+      "grad_norm": 0.6404470960030519,
+      "learning_rate": 6.067791600668823e-06,
+      "loss": 0.0644,
+      "step": 485
+    },
+    {
+      "epoch": 0.8647686832740213,
+      "grad_norm": 0.4935887263225689,
+      "learning_rate": 6.054134791439479e-06,
+      "loss": 0.0523,
+      "step": 486
+    },
+    {
+      "epoch": 0.8665480427046264,
+      "grad_norm": 0.5860720171524344,
+      "learning_rate": 6.040469747221815e-06,
+      "loss": 0.0621,
+      "step": 487
+    },
+    {
+      "epoch": 0.8683274021352313,
+      "grad_norm": 0.6193003216324234,
+      "learning_rate": 6.026796574768288e-06,
+      "loss": 0.0681,
+      "step": 488
+    },
+    {
+      "epoch": 0.8701067615658363,
+      "grad_norm": 0.6539081548840181,
+      "learning_rate": 6.013115380894854e-06,
+      "loss": 0.0772,
+      "step": 489
+    },
+    {
+      "epoch": 0.8718861209964412,
+      "grad_norm": 0.5386596984492199,
+      "learning_rate": 5.999426272480133e-06,
+      "loss": 0.0591,
+      "step": 490
+    },
+    {
+      "epoch": 0.8736654804270463,
+      "grad_norm": 0.5886914048623416,
+      "learning_rate": 5.985729356464575e-06,
+      "loss": 0.0766,
+      "step": 491
+    },
+    {
+      "epoch": 0.8754448398576512,
+      "grad_norm": 0.5219309291432573,
+      "learning_rate": 5.972024739849622e-06,
+      "loss": 0.0668,
+      "step": 492
+    },
+    {
+      "epoch": 0.8772241992882562,
+      "grad_norm": 0.40521287017057617,
+      "learning_rate": 5.958312529696874e-06,
+      "loss": 0.0495,
+      "step": 493
+    },
+    {
+      "epoch": 0.8790035587188612,
+      "grad_norm": 0.6323331029978013,
+      "learning_rate": 5.944592833127253e-06,
+      "loss": 0.1008,
+      "step": 494
+    },
+    {
+      "epoch": 0.8807829181494662,
+      "grad_norm": 0.6769001268578118,
+      "learning_rate": 5.9308657573201645e-06,
+      "loss": 0.0962,
+      "step": 495
+    },
+    {
+      "epoch": 0.8825622775800712,
+      "grad_norm": 0.535870726814628,
+      "learning_rate": 5.917131409512663e-06,
+      "loss": 0.0664,
+      "step": 496
+    },
+    {
+      "epoch": 0.8843416370106761,
+      "grad_norm": 0.5317076659036012,
+      "learning_rate": 5.903389896998611e-06,
+      "loss": 0.072,
+      "step": 497
+    },
+    {
+      "epoch": 0.8861209964412812,
+      "grad_norm": 0.7071330357247503,
+      "learning_rate": 5.889641327127843e-06,
+      "loss": 0.087,
+      "step": 498
+    },
+    {
+      "epoch": 0.8879003558718861,
+      "grad_norm": 0.5154601670345001,
+      "learning_rate": 5.875885807305326e-06,
+      "loss": 0.0686,
+      "step": 499
+    },
+    {
+      "epoch": 0.8896797153024911,
+      "grad_norm": 0.5866084497446581,
+      "learning_rate": 5.862123444990319e-06,
+      "loss": 0.0779,
+      "step": 500
+    },
+    {
+      "epoch": 0.891459074733096,
+      "grad_norm": 0.6726656671012712,
+      "learning_rate": 5.848354347695537e-06,
+      "loss": 0.1204,
+      "step": 501
+    },
+    {
+      "epoch": 0.8932384341637011,
+      "grad_norm": 0.6720362671054517,
+      "learning_rate": 5.83457862298631e-06,
+      "loss": 0.0968,
+      "step": 502
+    },
+    {
+      "epoch": 0.895017793594306,
+      "grad_norm": 0.7652311603687231,
+      "learning_rate": 5.8207963784797396e-06,
+      "loss": 0.0881,
+      "step": 503
+    },
+    {
+      "epoch": 0.896797153024911,
+      "grad_norm": 0.5772072959034297,
+      "learning_rate": 5.807007721843862e-06,
+      "loss": 0.08,
+      "step": 504
+    },
+    {
+      "epoch": 0.8985765124555161,
+      "grad_norm": 0.7371408703339905,
+      "learning_rate": 5.793212760796804e-06,
+      "loss": 0.1153,
+      "step": 505
+    },
+    {
+      "epoch": 0.900355871886121,
+      "grad_norm": 0.6199296471856783,
+      "learning_rate": 5.779411603105947e-06,
+      "loss": 0.0897,
+      "step": 506
+    },
+    {
+      "epoch": 0.902135231316726,
+      "grad_norm": 0.6478732074999847,
+      "learning_rate": 5.765604356587076e-06,
+      "loss": 0.0819,
+      "step": 507
+    },
+    {
+      "epoch": 0.9039145907473309,
+      "grad_norm": 0.6138084668644911,
+      "learning_rate": 5.751791129103545e-06,
+      "loss": 0.0821,
+      "step": 508
+    },
+    {
+      "epoch": 0.905693950177936,
+      "grad_norm": 0.695629504201085,
+      "learning_rate": 5.737972028565431e-06,
+      "loss": 0.099,
+      "step": 509
+    },
+    {
+      "epoch": 0.9074733096085409,
+      "grad_norm": 0.6675649550597265,
+      "learning_rate": 5.7241471629286934e-06,
+      "loss": 0.093,
+      "step": 510
+    },
+    {
+      "epoch": 0.9092526690391459,
+      "grad_norm": 0.7118507219874773,
+      "learning_rate": 5.7103166401943276e-06,
+      "loss": 0.0852,
+      "step": 511
+    },
+    {
+      "epoch": 0.9110320284697508,
+      "grad_norm": 0.6861425623474775,
+      "learning_rate": 5.696480568407523e-06,
+      "loss": 0.0805,
+      "step": 512
+    },
+    {
+      "epoch": 0.9128113879003559,
+      "grad_norm": 0.6929027641132477,
+      "learning_rate": 5.682639055656817e-06,
+      "loss": 0.0967,
+      "step": 513
+    },
+    {
+      "epoch": 0.9145907473309609,
+      "grad_norm": 0.6867679728608744,
+      "learning_rate": 5.668792210073255e-06,
+      "loss": 0.105,
+      "step": 514
+    },
+    {
+      "epoch": 0.9163701067615658,
+      "grad_norm": 0.6776218280370291,
+      "learning_rate": 5.654940139829544e-06,
+      "loss": 0.0891,
+      "step": 515
+    },
+    {
+      "epoch": 0.9181494661921709,
+      "grad_norm": 0.5207733951637723,
+      "learning_rate": 5.641082953139201e-06,
+      "loss": 0.0648,
+      "step": 516
+    },
+    {
+      "epoch": 0.9199288256227758,
+      "grad_norm": 0.4290319117100716,
+      "learning_rate": 5.6272207582557195e-06,
+      "loss": 0.0512,
+      "step": 517
+    },
+    {
+      "epoch": 0.9217081850533808,
+      "grad_norm": 0.5451688233257064,
+      "learning_rate": 5.61335366347171e-06,
+      "loss": 0.0581,
+      "step": 518
+    },
+    {
+      "epoch": 0.9234875444839857,
+      "grad_norm": 0.6027630127454737,
+      "learning_rate": 5.599481777118071e-06,
+      "loss": 0.0866,
+      "step": 519
+    },
+    {
+      "epoch": 0.9252669039145908,
+      "grad_norm": 0.6904048077912736,
+      "learning_rate": 5.585605207563124e-06,
+      "loss": 0.0744,
+      "step": 520
+    },
+    {
+      "epoch": 0.9270462633451957,
+      "grad_norm": 0.5283052650955237,
+      "learning_rate": 5.571724063211782e-06,
+      "loss": 0.0654,
+      "step": 521
+    },
+    {
+      "epoch": 0.9288256227758007,
+      "grad_norm": 0.5731146064244245,
+      "learning_rate": 5.557838452504692e-06,
+      "loss": 0.0571,
+      "step": 522
+    },
+    {
+      "epoch": 0.9306049822064056,
+      "grad_norm": 0.698094542185011,
+      "learning_rate": 5.5439484839173996e-06,
+      "loss": 0.0854,
+      "step": 523
+    },
+    {
+      "epoch": 0.9323843416370107,
+      "grad_norm": 0.5728243170232641,
+      "learning_rate": 5.530054265959486e-06,
+      "loss": 0.0756,
+      "step": 524
+    },
+    {
+      "epoch": 0.9341637010676157,
+      "grad_norm": 0.593737755566197,
+      "learning_rate": 5.516155907173735e-06,
+      "loss": 0.0842,
+      "step": 525
+    },
+    {
+      "epoch": 0.9359430604982206,
+      "grad_norm": 0.4695045031586955,
+      "learning_rate": 5.5022535161352764e-06,
+      "loss": 0.061,
+      "step": 526
+    },
+    {
+      "epoch": 0.9377224199288257,
+      "grad_norm": 0.7433234884667963,
+      "learning_rate": 5.488347201450741e-06,
+      "loss": 0.0867,
+      "step": 527
+    },
+    {
+      "epoch": 0.9395017793594306,
+      "grad_norm": 0.5182533411782908,
+      "learning_rate": 5.47443707175741e-06,
+      "loss": 0.0764,
+      "step": 528
+    },
+    {
+      "epoch": 0.9412811387900356,
+      "grad_norm": 0.6094435429205892,
+      "learning_rate": 5.46052323572237e-06,
+      "loss": 0.0769,
+      "step": 529
+    },
+    {
+      "epoch": 0.9430604982206405,
+      "grad_norm": 0.49378910463153747,
+      "learning_rate": 5.446605802041662e-06,
+      "loss": 0.0588,
+      "step": 530
+    },
+    {
+      "epoch": 0.9448398576512456,
+      "grad_norm": 0.4454207888718773,
+      "learning_rate": 5.432684879439428e-06,
+      "loss": 0.0553,
+      "step": 531
+    },
+    {
+      "epoch": 0.9466192170818505,
+      "grad_norm": 0.5422828036366653,
+      "learning_rate": 5.418760576667071e-06,
+      "loss": 0.0548,
+      "step": 532
+    },
+    {
+      "epoch": 0.9483985765124555,
+      "grad_norm": 0.6250838944912477,
+      "learning_rate": 5.404833002502398e-06,
+      "loss": 0.0944,
+      "step": 533
+    },
+    {
+      "epoch": 0.9501779359430605,
+      "grad_norm": 0.6911488471271517,
+      "learning_rate": 5.39090226574877e-06,
+      "loss": 0.081,
+      "step": 534
+    },
+    {
+      "epoch": 0.9519572953736655,
+      "grad_norm": 0.8757181768393908,
+      "learning_rate": 5.376968475234258e-06,
+      "loss": 0.1138,
+      "step": 535
+    },
+    {
+      "epoch": 0.9537366548042705,
+      "grad_norm": 0.42064942629606794,
+      "learning_rate": 5.363031739810787e-06,
+      "loss": 0.0519,
+      "step": 536
+    },
+    {
+      "epoch": 0.9555160142348754,
+      "grad_norm": 0.5499240365100218,
+      "learning_rate": 5.349092168353291e-06,
+      "loss": 0.0765,
+      "step": 537
+    },
+    {
+      "epoch": 0.9572953736654805,
+      "grad_norm": 0.5599222870068542,
+      "learning_rate": 5.335149869758855e-06,
+      "loss": 0.0656,
+      "step": 538
+    },
+    {
+      "epoch": 0.9590747330960854,
+      "grad_norm": 0.6808637143886486,
+      "learning_rate": 5.32120495294587e-06,
+      "loss": 0.0902,
+      "step": 539
+    },
+    {
+      "epoch": 0.9608540925266904,
+      "grad_norm": 0.5687310432860996,
+      "learning_rate": 5.3072575268531835e-06,
+      "loss": 0.0914,
+      "step": 540
+    },
+    {
+      "epoch": 0.9626334519572953,
+      "grad_norm": 0.6180476062009892,
+      "learning_rate": 5.293307700439242e-06,
+      "loss": 0.0845,
+      "step": 541
+    },
+    {
+      "epoch": 0.9644128113879004,
+      "grad_norm": 0.7811554479377757,
+      "learning_rate": 5.2793555826812456e-06,
+      "loss": 0.1047,
+      "step": 542
+    },
+    {
+      "epoch": 0.9661921708185054,
+      "grad_norm": 0.5611606077633421,
+      "learning_rate": 5.265401282574294e-06,
+      "loss": 0.0631,
+      "step": 543
+    },
+    {
+      "epoch": 0.9679715302491103,
+      "grad_norm": 0.6694019888974924,
+      "learning_rate": 5.2514449091305375e-06,
+      "loss": 0.0796,
+      "step": 544
+    },
+    {
+      "epoch": 0.9697508896797153,
+      "grad_norm": 0.6882733604233107,
+      "learning_rate": 5.237486571378317e-06,
+      "loss": 0.0867,
+      "step": 545
+    },
+    {
+      "epoch": 0.9715302491103203,
+      "grad_norm": 0.64843841264274,
+      "learning_rate": 5.22352637836133e-06,
+      "loss": 0.1048,
+      "step": 546
+    },
+    {
+      "epoch": 0.9733096085409253,
+      "grad_norm": 0.5649177716893614,
+      "learning_rate": 5.209564439137755e-06,
+      "loss": 0.0724,
+      "step": 547
+    },
+    {
+      "epoch": 0.9750889679715302,
+      "grad_norm": 0.6517410439575472,
+      "learning_rate": 5.195600862779421e-06,
+      "loss": 0.106,
+      "step": 548
+    },
+    {
+      "epoch": 0.9768683274021353,
+      "grad_norm": 0.4361812381162346,
+      "learning_rate": 5.181635758370942e-06,
+      "loss": 0.051,
+      "step": 549
+    },
+    {
+      "epoch": 0.9786476868327402,
+      "grad_norm": 0.6658029563798565,
+      "learning_rate": 5.167669235008871e-06,
+      "loss": 0.0877,
+      "step": 550
+    },
+    {
+      "epoch": 0.9804270462633452,
+      "grad_norm": 0.8284049743532196,
+      "learning_rate": 5.153701401800845e-06,
+      "loss": 0.114,
+      "step": 551
+    },
+    {
+      "epoch": 0.9822064056939501,
+      "grad_norm": 0.5330839242256337,
+      "learning_rate": 5.139732367864736e-06,
+      "loss": 0.0647,
+      "step": 552
+    },
+    {
+      "epoch": 0.9839857651245552,
+      "grad_norm": 0.5892479677125644,
+      "learning_rate": 5.1257622423277934e-06,
+      "loss": 0.0792,
+      "step": 553
+    },
+    {
+      "epoch": 0.9857651245551602,
+      "grad_norm": 0.5756038799049453,
+      "learning_rate": 5.111791134325793e-06,
+      "loss": 0.0732,
+      "step": 554
+    },
+    {
+      "epoch": 0.9875444839857651,
+      "grad_norm": 0.4980512963062255,
+      "learning_rate": 5.097819153002192e-06,
+      "loss": 0.0645,
+      "step": 555
+    },
+    {
+      "epoch": 0.9893238434163701,
+      "grad_norm": 0.4629502136532782,
+      "learning_rate": 5.083846407507263e-06,
+      "loss": 0.0559,
+      "step": 556
+    },
+    {
+      "epoch": 0.9911032028469751,
+      "grad_norm": 0.7669925781901444,
+      "learning_rate": 5.0698730069972535e-06,
+      "loss": 0.095,
+      "step": 557
+    },
+    {
+      "epoch": 0.9928825622775801,
+      "grad_norm": 0.4914404899817638,
+      "learning_rate": 5.055899060633524e-06,
+      "loss": 0.0585,
+      "step": 558
+    },
+    {
+      "epoch": 0.994661921708185,
+      "grad_norm": 0.578987389662384,
+      "learning_rate": 5.041924677581702e-06,
+      "loss": 0.0844,
+      "step": 559
+    },
+    {
+      "epoch": 0.99644128113879,
+      "grad_norm": 0.5921493312483826,
+      "learning_rate": 5.0279499670108245e-06,
+      "loss": 0.0843,
+      "step": 560
+    },
+    {
+      "epoch": 0.998220640569395,
+      "grad_norm": 0.5405541186328452,
+      "learning_rate": 5.013975038092491e-06,
+      "loss": 0.0728,
+      "step": 561
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5297177692326421,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "step": 562
+    },
+    {
+      "epoch": 1.001779359430605,
+      "grad_norm": 0.37231571331449936,
+      "learning_rate": 4.98602496190751e-06,
+      "loss": 0.0385,
+      "step": 563
+    },
+    {
+      "epoch": 1.00355871886121,
+      "grad_norm": 0.33290219285664524,
+      "learning_rate": 4.9720500329891755e-06,
+      "loss": 0.0313,
+      "step": 564
+    },
+    {
+      "epoch": 1.0053380782918149,
+      "grad_norm": 0.40470037025187544,
+      "learning_rate": 4.9580753224183005e-06,
+      "loss": 0.0401,
+      "step": 565
+    },
+    {
+      "epoch": 1.00711743772242,
+      "grad_norm": 0.34435050798627803,
+      "learning_rate": 4.944100939366478e-06,
+      "loss": 0.0323,
+      "step": 566
+    },
+    {
+      "epoch": 1.008896797153025,
+      "grad_norm": 0.3136698775037879,
+      "learning_rate": 4.930126993002748e-06,
+      "loss": 0.0353,
+      "step": 567
+    },
+    {
+      "epoch": 1.01067615658363,
+      "grad_norm": 0.32901559193646895,
+      "learning_rate": 4.9161535924927375e-06,
+      "loss": 0.0339,
+      "step": 568
+    },
+    {
+      "epoch": 1.0124555160142348,
+      "grad_norm": 0.3284888422968096,
+      "learning_rate": 4.90218084699781e-06,
+      "loss": 0.03,
+      "step": 569
+    },
+    {
+      "epoch": 1.0142348754448398,
+      "grad_norm": 0.4654635550262594,
+      "learning_rate": 4.888208865674208e-06,
+      "loss": 0.0537,
+      "step": 570
+    },
+    {
+      "epoch": 1.0160142348754448,
+      "grad_norm": 0.350820536491868,
+      "learning_rate": 4.874237757672209e-06,
+      "loss": 0.0338,
+      "step": 571
+    },
+    {
+      "epoch": 1.0177935943060499,
+      "grad_norm": 0.45907868868096996,
+      "learning_rate": 4.8602676321352646e-06,
+      "loss": 0.0517,
+      "step": 572
+    },
+    {
+      "epoch": 1.019572953736655,
+      "grad_norm": 0.3767960041118899,
+      "learning_rate": 4.846298598199155e-06,
+      "loss": 0.0353,
+      "step": 573
+    },
+    {
+      "epoch": 1.0213523131672597,
+      "grad_norm": 0.3799565544749443,
+      "learning_rate": 4.832330764991131e-06,
+      "loss": 0.0319,
+      "step": 574
+    },
+    {
+      "epoch": 1.0231316725978647,
+      "grad_norm": 0.41274624207299293,
+      "learning_rate": 4.81836424162906e-06,
+      "loss": 0.0435,
+      "step": 575
+    },
+    {
+      "epoch": 1.0249110320284698,
+      "grad_norm": 0.3684728252454572,
+      "learning_rate": 4.80439913722058e-06,
+      "loss": 0.0327,
+      "step": 576
+    },
+    {
+      "epoch": 1.0266903914590748,
+      "grad_norm": 0.37389221894566355,
+      "learning_rate": 4.790435560862247e-06,
+      "loss": 0.0418,
+      "step": 577
+    },
+    {
+      "epoch": 1.0284697508896796,
+      "grad_norm": 0.4387833156061662,
+      "learning_rate": 4.776473621638673e-06,
+      "loss": 0.0331,
+      "step": 578
+    },
+    {
+      "epoch": 1.0302491103202847,
+      "grad_norm": 0.37722817837486,
+      "learning_rate": 4.762513428621684e-06,
+      "loss": 0.0328,
+      "step": 579
+    },
+    {
+      "epoch": 1.0320284697508897,
+      "grad_norm": 0.4105154919583334,
+      "learning_rate": 4.748555090869464e-06,
+      "loss": 0.029,
+      "step": 580
+    },
+    {
+      "epoch": 1.0338078291814947,
+      "grad_norm": 0.4004609842551577,
+      "learning_rate": 4.734598717425706e-06,
+      "loss": 0.038,
+      "step": 581
+    },
+    {
+      "epoch": 1.0355871886120998,
+      "grad_norm": 0.5052396548372445,
+      "learning_rate": 4.720644417318755e-06,
+      "loss": 0.0487,
+      "step": 582
+    },
+    {
+      "epoch": 1.0373665480427046,
+      "grad_norm": 0.5525044303059882,
+      "learning_rate": 4.70669229956076e-06,
+      "loss": 0.038,
+      "step": 583
+    },
+    {
+      "epoch": 1.0391459074733096,
+      "grad_norm": 0.39308195605537394,
+      "learning_rate": 4.692742473146818e-06,
+      "loss": 0.0302,
+      "step": 584
+    },
+    {
+      "epoch": 1.0409252669039146,
+      "grad_norm": 0.42384562293392075,
+      "learning_rate": 4.678795047054131e-06,
+      "loss": 0.0429,
+      "step": 585
+    },
+    {
+      "epoch": 1.0427046263345197,
+      "grad_norm": 0.421125689110126,
+      "learning_rate": 4.664850130241146e-06,
+      "loss": 0.0267,
+      "step": 586
+    },
+    {
+      "epoch": 1.0444839857651245,
+      "grad_norm": 0.42633334791790684,
+      "learning_rate": 4.650907831646711e-06,
+      "loss": 0.033,
+      "step": 587
+    },
+    {
+      "epoch": 1.0462633451957295,
+      "grad_norm": 0.49276679304538995,
+      "learning_rate": 4.636968260189214e-06,
+      "loss": 0.0573,
+      "step": 588
+    },
+    {
+      "epoch": 1.0480427046263345,
+      "grad_norm": 0.4164340915488122,
+      "learning_rate": 4.623031524765744e-06,
+      "loss": 0.0325,
+      "step": 589
+    },
+    {
+      "epoch": 1.0498220640569396,
+      "grad_norm": 0.46840095331341436,
+      "learning_rate": 4.609097734251231e-06,
+      "loss": 0.038,
+      "step": 590
+    },
+    {
+      "epoch": 1.0516014234875444,
+      "grad_norm": 0.5682818829025402,
+      "learning_rate": 4.595166997497605e-06,
+      "loss": 0.038,
+      "step": 591
+    },
+    {
+      "epoch": 1.0533807829181494,
+      "grad_norm": 0.494897369043239,
+      "learning_rate": 4.58123942333293e-06,
+      "loss": 0.0375,
+      "step": 592
+    },
+    {
+      "epoch": 1.0551601423487544,
+      "grad_norm": 0.39327813276424867,
+      "learning_rate": 4.567315120560573e-06,
+      "loss": 0.0292,
+      "step": 593
+    },
+    {
+      "epoch": 1.0569395017793595,
+      "grad_norm": 0.46150445116801964,
+      "learning_rate": 4.553394197958339e-06,
+      "loss": 0.0295,
+      "step": 594
+    },
+    {
+      "epoch": 1.0587188612099645,
+      "grad_norm": 0.37987895338294614,
+      "learning_rate": 4.539476764277631e-06,
+      "loss": 0.0214,
+      "step": 595
+    },
+    {
+      "epoch": 1.0604982206405693,
+      "grad_norm": 0.36773400599364514,
+      "learning_rate": 4.525562928242592e-06,
+      "loss": 0.0274,
+      "step": 596
+    },
+    {
+      "epoch": 1.0622775800711743,
+      "grad_norm": 0.5445032635689646,
+      "learning_rate": 4.511652798549261e-06,
+      "loss": 0.0385,
+      "step": 597
+    },
+    {
+      "epoch": 1.0640569395017794,
+      "grad_norm": 0.48046751777982516,
+      "learning_rate": 4.497746483864725e-06,
+      "loss": 0.0295,
+      "step": 598
+    },
+    {
+      "epoch": 1.0658362989323844,
+      "grad_norm": 0.42859559023482485,
+      "learning_rate": 4.483844092826267e-06,
+      "loss": 0.0253,
+      "step": 599
+    },
+    {
+      "epoch": 1.0676156583629894,
+      "grad_norm": 0.39255377119763224,
+      "learning_rate": 4.469945734040516e-06,
+      "loss": 0.0292,
+      "step": 600
+    },
+    {
+      "epoch": 1.0676156583629894,
+      "eval_loss": 0.10132589191198349,
+      "eval_runtime": 1.9009,
+      "eval_samples_per_second": 24.2,
+      "eval_steps_per_second": 6.313,
+      "step": 600
+    },
+    {
+      "epoch": 1.0693950177935942,
+      "grad_norm": 0.41982302603821053,
+      "learning_rate": 4.456051516082603e-06,
+      "loss": 0.0268,
+      "step": 601
+    },
+    {
+      "epoch": 1.0711743772241993,
+      "grad_norm": 0.5411098262992029,
+      "learning_rate": 4.442161547495309e-06,
+      "loss": 0.0316,
+      "step": 602
+    },
+    {
+      "epoch": 1.0729537366548043,
+      "grad_norm": 0.5722972912255525,
+      "learning_rate": 4.42827593678822e-06,
+      "loss": 0.0341,
+      "step": 603
+    },
+    {
+      "epoch": 1.0747330960854093,
+      "grad_norm": 0.44271601257289406,
+      "learning_rate": 4.414394792436877e-06,
+      "loss": 0.0244,
+      "step": 604
+    },
+    {
+      "epoch": 1.0765124555160142,
+      "grad_norm": 0.5242190644338142,
+      "learning_rate": 4.400518222881931e-06,
+      "loss": 0.0398,
+      "step": 605
+    },
+    {
+      "epoch": 1.0782918149466192,
+      "grad_norm": 0.4698262063294386,
+      "learning_rate": 4.386646336528291e-06,
+      "loss": 0.0292,
+      "step": 606
+    },
+    {
+      "epoch": 1.0800711743772242,
+      "grad_norm": 0.44443234939542753,
+      "learning_rate": 4.372779241744282e-06,
+      "loss": 0.0258,
+      "step": 607
+    },
+    {
+      "epoch": 1.0818505338078293,
+      "grad_norm": 0.397682890295574,
+      "learning_rate": 4.358917046860799e-06,
+      "loss": 0.025,
+      "step": 608
+    },
+    {
+      "epoch": 1.083629893238434,
+      "grad_norm": 0.5454455369317831,
+      "learning_rate": 4.345059860170458e-06,
+      "loss": 0.0397,
+      "step": 609
+    },
+    {
+      "epoch": 1.085409252669039,
+      "grad_norm": 0.4393105152340805,
+      "learning_rate": 4.331207789926746e-06,
+      "loss": 0.0248,
+      "step": 610
+    },
+    {
+      "epoch": 1.0871886120996441,
+      "grad_norm": 0.5229522694135293,
+      "learning_rate": 4.317360944343184e-06,
+      "loss": 0.0352,
+      "step": 611
+    },
+    {
+      "epoch": 1.0889679715302492,
+      "grad_norm": 0.40631810770413984,
+      "learning_rate": 4.303519431592479e-06,
+      "loss": 0.0207,
+      "step": 612
+    },
+    {
+      "epoch": 1.0907473309608542,
+      "grad_norm": 0.7279279703395911,
+      "learning_rate": 4.289683359805673e-06,
+      "loss": 0.0547,
+      "step": 613
+    },
+    {
+      "epoch": 1.092526690391459,
+      "grad_norm": 0.4882159079703789,
+      "learning_rate": 4.275852837071309e-06,
+      "loss": 0.0262,
+      "step": 614
+    },
+    {
+      "epoch": 1.094306049822064,
+      "grad_norm": 0.4403471854241194,
+      "learning_rate": 4.26202797143457e-06,
+      "loss": 0.0314,
+      "step": 615
+    },
+    {
+      "epoch": 1.096085409252669,
+      "grad_norm": 0.6219621205175584,
+      "learning_rate": 4.248208870896456e-06,
+      "loss": 0.0359,
+      "step": 616
+    },
+    {
+      "epoch": 1.097864768683274,
+      "grad_norm": 0.5230243934718025,
+      "learning_rate": 4.234395643412925e-06,
+      "loss": 0.0292,
+      "step": 617
+    },
+    {
+      "epoch": 1.099644128113879,
+      "grad_norm": 0.5761941524891295,
+      "learning_rate": 4.220588396894055e-06,
+      "loss": 0.0274,
+      "step": 618
+    },
+    {
+      "epoch": 1.101423487544484,
+      "grad_norm": 0.5489102528661693,
+      "learning_rate": 4.2067872392031965e-06,
+      "loss": 0.0329,
+      "step": 619
+    },
+    {
+      "epoch": 1.103202846975089,
+      "grad_norm": 0.6669442727099592,
+      "learning_rate": 4.192992278156141e-06,
+      "loss": 0.0347,
+      "step": 620
+    },
+    {
+      "epoch": 1.104982206405694,
+      "grad_norm": 0.5668858098491039,
+      "learning_rate": 4.179203621520262e-06,
+      "loss": 0.0318,
+      "step": 621
+    },
+    {
+      "epoch": 1.106761565836299,
+      "grad_norm": 0.4328830713959674,
+      "learning_rate": 4.165421377013691e-06,
+      "loss": 0.0174,
+      "step": 622
+    },
+    {
+      "epoch": 1.1085409252669038,
+      "grad_norm": 0.5028507905361573,
+      "learning_rate": 4.151645652304465e-06,
+      "loss": 0.0279,
+      "step": 623
+    },
+    {
+      "epoch": 1.1103202846975089,
+      "grad_norm": 0.5511850799525075,
+      "learning_rate": 4.137876555009684e-06,
+      "loss": 0.0364,
+      "step": 624
+    },
+    {
+      "epoch": 1.112099644128114,
+      "grad_norm": 0.5614986797133694,
+      "learning_rate": 4.124114192694676e-06,
+      "loss": 0.0323,
+      "step": 625
+    },
+    {
+      "epoch": 1.113879003558719,
+      "grad_norm": 0.5066348541953499,
+      "learning_rate": 4.110358672872158e-06,
+      "loss": 0.0274,
+      "step": 626
+    },
+    {
+      "epoch": 1.1156583629893237,
+      "grad_norm": 0.5613464361555829,
+      "learning_rate": 4.0966101030013915e-06,
+      "loss": 0.0403,
+      "step": 627
+    },
+    {
+      "epoch": 1.1174377224199288,
+      "grad_norm": 0.4747076239155826,
+      "learning_rate": 4.082868590487339e-06,
+      "loss": 0.0293,
+      "step": 628
+    },
+    {
+      "epoch": 1.1192170818505338,
+      "grad_norm": 0.34591339797950243,
+      "learning_rate": 4.069134242679837e-06,
+      "loss": 0.0238,
+      "step": 629
+    },
+    {
+      "epoch": 1.1209964412811388,
+      "grad_norm": 0.5482550232191747,
+      "learning_rate": 4.055407166872748e-06,
+      "loss": 0.0297,
+      "step": 630
+    },
+    {
+      "epoch": 1.1227758007117439,
+      "grad_norm": 0.5384365324308098,
+      "learning_rate": 4.041687470303127e-06,
+      "loss": 0.0393,
+      "step": 631
+    },
+    {
+      "epoch": 1.1245551601423487,
+      "grad_norm": 0.51319728545892,
+      "learning_rate": 4.02797526015038e-06,
+      "loss": 0.0455,
+      "step": 632
+    },
+    {
+      "epoch": 1.1263345195729537,
+      "grad_norm": 0.46208051847600173,
+      "learning_rate": 4.014270643535427e-06,
+      "loss": 0.0258,
+      "step": 633
+    },
+    {
+      "epoch": 1.1281138790035588,
+      "grad_norm": 0.6632496900416235,
+      "learning_rate": 4.000573727519868e-06,
+      "loss": 0.0371,
+      "step": 634
+    },
+    {
+      "epoch": 1.1298932384341638,
+      "grad_norm": 0.4654635253396662,
+      "learning_rate": 3.9868846191051465e-06,
+      "loss": 0.0259,
+      "step": 635
+    },
+    {
+      "epoch": 1.1316725978647686,
+      "grad_norm": 0.6166580209214856,
+      "learning_rate": 3.973203425231715e-06,
+      "loss": 0.0384,
+      "step": 636
+    },
+    {
+      "epoch": 1.1334519572953736,
+      "grad_norm": 0.5030023800898313,
+      "learning_rate": 3.959530252778187e-06,
+      "loss": 0.0432,
+      "step": 637
+    },
+    {
+      "epoch": 1.1352313167259787,
+      "grad_norm": 0.5517001917546517,
+      "learning_rate": 3.945865208560522e-06,
+      "loss": 0.0336,
+      "step": 638
+    },
+    {
+      "epoch": 1.1370106761565837,
+      "grad_norm": 0.6078287080558146,
+      "learning_rate": 3.932208399331177e-06,
+      "loss": 0.0403,
+      "step": 639
+    },
+    {
+      "epoch": 1.1387900355871885,
+      "grad_norm": 0.5896009844084492,
+      "learning_rate": 3.918559931778277e-06,
+      "loss": 0.0348,
+      "step": 640
+    },
+    {
+      "epoch": 1.1405693950177935,
+      "grad_norm": 0.4756242119363783,
+      "learning_rate": 3.904919912524784e-06,
+      "loss": 0.0231,
+      "step": 641
+    },
+    {
+      "epoch": 1.1423487544483986,
+      "grad_norm": 0.4680870550754033,
+      "learning_rate": 3.891288448127654e-06,
+      "loss": 0.0287,
+      "step": 642
+    },
+    {
+      "epoch": 1.1441281138790036,
+      "grad_norm": 0.6392194479536875,
+      "learning_rate": 3.877665645077017e-06,
+      "loss": 0.03,
+      "step": 643
+    },
+    {
+      "epoch": 1.1459074733096086,
+      "grad_norm": 0.622293208056315,
+      "learning_rate": 3.86405160979534e-06,
+      "loss": 0.0407,
+      "step": 644
+    },
+    {
+      "epoch": 1.1476868327402134,
+      "grad_norm": 0.5562876642196498,
+      "learning_rate": 3.850446448636597e-06,
+      "loss": 0.0254,
+      "step": 645
+    },
+    {
+      "epoch": 1.1494661921708185,
+      "grad_norm": 0.37076465741572157,
+      "learning_rate": 3.8368502678854296e-06,
+      "loss": 0.0165,
+      "step": 646
+    },
+    {
+      "epoch": 1.1512455516014235,
+      "grad_norm": 0.5277168187317152,
+      "learning_rate": 3.8232631737563306e-06,
+      "loss": 0.0322,
+      "step": 647
+    },
+    {
+      "epoch": 1.1530249110320285,
+      "grad_norm": 0.44117008880884667,
+      "learning_rate": 3.809685272392804e-06,
+      "loss": 0.024,
+      "step": 648
+    },
+    {
+      "epoch": 1.1548042704626336,
+      "grad_norm": 0.48659107555209974,
+      "learning_rate": 3.796116669866543e-06,
+      "loss": 0.0325,
+      "step": 649
+    },
+    {
+      "epoch": 1.1565836298932384,
+      "grad_norm": 0.6014690700387714,
+      "learning_rate": 3.78255747217659e-06,
+      "loss": 0.0306,
+      "step": 650
+    },
+    {
+      "epoch": 1.1583629893238434,
+      "grad_norm": 0.5689988351140361,
+      "learning_rate": 3.769007785248523e-06,
+      "loss": 0.0248,
+      "step": 651
+    },
+    {
+      "epoch": 1.1601423487544484,
+      "grad_norm": 0.4775010506299335,
+      "learning_rate": 3.7554677149336186e-06,
+      "loss": 0.0263,
+      "step": 652
+    },
+    {
+      "epoch": 1.1619217081850535,
+      "grad_norm": 0.5193772851085521,
+      "learning_rate": 3.7419373670080284e-06,
+      "loss": 0.0263,
+      "step": 653
+    },
+    {
+      "epoch": 1.1637010676156583,
+      "grad_norm": 0.5794416324922071,
+      "learning_rate": 3.7284168471719527e-06,
+      "loss": 0.0329,
+      "step": 654
+    },
+    {
+      "epoch": 1.1654804270462633,
+      "grad_norm": 0.5308889524911553,
+      "learning_rate": 3.7149062610488085e-06,
+      "loss": 0.0306,
+      "step": 655
+    },
+    {
+      "epoch": 1.1672597864768683,
+      "grad_norm": 0.5288293137641096,
+      "learning_rate": 3.701405714184416e-06,
+      "loss": 0.0236,
+      "step": 656
+    },
+    {
+      "epoch": 1.1690391459074734,
+      "grad_norm": 0.4969193422165757,
+      "learning_rate": 3.687915312046166e-06,
+      "loss": 0.0397,
+      "step": 657
+    },
+    {
+      "epoch": 1.1708185053380782,
+      "grad_norm": 0.5596094436258661,
+      "learning_rate": 3.6744351600221994e-06,
+      "loss": 0.021,
+      "step": 658
+    },
+    {
+      "epoch": 1.1725978647686832,
+      "grad_norm": 0.5392735445614594,
+      "learning_rate": 3.6609653634205773e-06,
+      "loss": 0.0441,
+      "step": 659
+    },
+    {
+      "epoch": 1.1743772241992882,
+      "grad_norm": 0.7729441851865585,
+      "learning_rate": 3.647506027468467e-06,
+      "loss": 0.0359,
+      "step": 660
+    },
+    {
+      "epoch": 1.1761565836298933,
+      "grad_norm": 0.5933554565530665,
+      "learning_rate": 3.6340572573113176e-06,
+      "loss": 0.0319,
+      "step": 661
+    },
+    {
+      "epoch": 1.1779359430604983,
+      "grad_norm": 0.5587745596854016,
+      "learning_rate": 3.6206191580120346e-06,
+      "loss": 0.0379,
+      "step": 662
+    },
+    {
+      "epoch": 1.1797153024911031,
+      "grad_norm": 0.5159699232898955,
+      "learning_rate": 3.6071918345501655e-06,
+      "loss": 0.0275,
+      "step": 663
+    },
+    {
+      "epoch": 1.1814946619217082,
+      "grad_norm": 0.5209866174849147,
+      "learning_rate": 3.5937753918210705e-06,
+      "loss": 0.0313,
+      "step": 664
+    },
+    {
+      "epoch": 1.1832740213523132,
+      "grad_norm": 0.5496657964594364,
+      "learning_rate": 3.5803699346351117e-06,
+      "loss": 0.0259,
+      "step": 665
+    },
+    {
+      "epoch": 1.1850533807829182,
+      "grad_norm": 0.43481536669564436,
+      "learning_rate": 3.566975567716833e-06,
+      "loss": 0.0267,
+      "step": 666
+    },
+    {
+      "epoch": 1.1868327402135233,
+      "grad_norm": 0.6106568189994596,
+      "learning_rate": 3.5535923957041374e-06,
+      "loss": 0.0316,
+      "step": 667
+    },
+    {
+      "epoch": 1.188612099644128,
+      "grad_norm": 0.6200920676704657,
+      "learning_rate": 3.540220523147474e-06,
+      "loss": 0.0319,
+      "step": 668
+    },
+    {
+      "epoch": 1.190391459074733,
+      "grad_norm": 0.6096281395569259,
+      "learning_rate": 3.5268600545090183e-06,
+      "loss": 0.0291,
+      "step": 669
+    },
+    {
+      "epoch": 1.1921708185053381,
+      "grad_norm": 0.5933955090787452,
+      "learning_rate": 3.513511094161858e-06,
+      "loss": 0.0331,
+      "step": 670
+    },
+    {
+      "epoch": 1.193950177935943,
+      "grad_norm": 0.5394904060031113,
+      "learning_rate": 3.5001737463891793e-06,
+      "loss": 0.0264,
+      "step": 671
+    },
+    {
+      "epoch": 1.195729537366548,
+      "grad_norm": 0.5109184799871422,
+      "learning_rate": 3.4868481153834454e-06,
+      "loss": 0.0392,
+      "step": 672
+    },
+    {
+      "epoch": 1.197508896797153,
+      "grad_norm": 0.3890777327820562,
+      "learning_rate": 3.4735343052455905e-06,
+      "loss": 0.018,
+      "step": 673
+    },
+    {
+      "epoch": 1.199288256227758,
+      "grad_norm": 0.5773846616051793,
+      "learning_rate": 3.4602324199842026e-06,
+      "loss": 0.0267,
+      "step": 674
+    },
+    {
+      "epoch": 1.201067615658363,
+      "grad_norm": 0.530347741913844,
+      "learning_rate": 3.446942563514711e-06,
+      "loss": 0.033,
+      "step": 675
+    },
+    {
+      "epoch": 1.2028469750889679,
+      "grad_norm": 0.5888170108390139,
+      "learning_rate": 3.4336648396585777e-06,
+      "loss": 0.0257,
+      "step": 676
+    },
+    {
+      "epoch": 1.204626334519573,
+      "grad_norm": 0.4468523820386189,
+      "learning_rate": 3.4203993521424774e-06,
+      "loss": 0.0299,
+      "step": 677
+    },
+    {
+      "epoch": 1.206405693950178,
+      "grad_norm": 0.4726029343329746,
+      "learning_rate": 3.407146204597499e-06,
+      "loss": 0.0252,
+      "step": 678
+    },
+    {
+      "epoch": 1.208185053380783,
+      "grad_norm": 0.6786115232129615,
+      "learning_rate": 3.3939055005583305e-06,
+      "loss": 0.038,
+      "step": 679
+    },
+    {
+      "epoch": 1.209964412811388,
+      "grad_norm": 0.6271110017244763,
+      "learning_rate": 3.3806773434624475e-06,
+      "loss": 0.0373,
+      "step": 680
+    },
+    {
+      "epoch": 1.2117437722419928,
+      "grad_norm": 0.6091761721264172,
+      "learning_rate": 3.3674618366493117e-06,
+      "loss": 0.0454,
+      "step": 681
+    },
+    {
+      "epoch": 1.2135231316725978,
+      "grad_norm": 0.4804538021951033,
+      "learning_rate": 3.3542590833595533e-06,
+      "loss": 0.0259,
+      "step": 682
+    },
+    {
+      "epoch": 1.2153024911032029,
+      "grad_norm": 0.5216126739964431,
+      "learning_rate": 3.341069186734176e-06,
+      "loss": 0.0294,
+      "step": 683
+    },
+    {
+      "epoch": 1.217081850533808,
+      "grad_norm": 0.48564372243177023,
+      "learning_rate": 3.3278922498137455e-06,
+      "loss": 0.0215,
+      "step": 684
+    },
+    {
+      "epoch": 1.2188612099644127,
+      "grad_norm": 0.5829324347251167,
+      "learning_rate": 3.314728375537587e-06,
+      "loss": 0.0259,
+      "step": 685
+    },
+    {
+      "epoch": 1.2206405693950177,
+      "grad_norm": 0.4844189706447932,
+      "learning_rate": 3.3015776667429724e-06,
+      "loss": 0.0387,
+      "step": 686
+    },
+    {
+      "epoch": 1.2224199288256228,
+      "grad_norm": 0.5133354574125267,
+      "learning_rate": 3.2884402261643296e-06,
+      "loss": 0.0277,
+      "step": 687
+    },
+    {
+      "epoch": 1.2241992882562278,
+      "grad_norm": 0.6650537146764004,
+      "learning_rate": 3.2753161564324344e-06,
+      "loss": 0.0358,
+      "step": 688
+    },
+    {
+      "epoch": 1.2259786476868326,
+      "grad_norm": 0.7656333551742341,
+      "learning_rate": 3.262205560073605e-06,
+      "loss": 0.0343,
+      "step": 689
+    },
+    {
+      "epoch": 1.2277580071174377,
+      "grad_norm": 0.4300670962342256,
+      "learning_rate": 3.249108539508909e-06,
+      "loss": 0.0182,
+      "step": 690
+    },
+    {
+      "epoch": 1.2295373665480427,
+      "grad_norm": 0.6278622969257219,
+      "learning_rate": 3.2360251970533527e-06,
+      "loss": 0.044,
+      "step": 691
+    },
+    {
+      "epoch": 1.2313167259786477,
+      "grad_norm": 0.4906449642956863,
+      "learning_rate": 3.2229556349150947e-06,
+      "loss": 0.0312,
+      "step": 692
+    },
+    {
+      "epoch": 1.2330960854092528,
+      "grad_norm": 0.48032904484620403,
+      "learning_rate": 3.2098999551946337e-06,
+      "loss": 0.0223,
+      "step": 693
+    },
+    {
+      "epoch": 1.2348754448398576,
+      "grad_norm": 0.5717655298093528,
+      "learning_rate": 3.1968582598840234e-06,
+      "loss": 0.0296,
+      "step": 694
+    },
+    {
+      "epoch": 1.2366548042704626,
+      "grad_norm": 0.46403373773959444,
+      "learning_rate": 3.183830650866068e-06,
+      "loss": 0.0212,
+      "step": 695
+    },
+    {
+      "epoch": 1.2384341637010676,
+      "grad_norm": 0.5158708941291148,
+      "learning_rate": 3.1708172299135266e-06,
+      "loss": 0.0301,
+      "step": 696
+    },
+    {
+      "epoch": 1.2402135231316727,
+      "grad_norm": 0.713835522055893,
+      "learning_rate": 3.1578180986883234e-06,
+      "loss": 0.0455,
+      "step": 697
+    },
+    {
+      "epoch": 1.2419928825622777,
+      "grad_norm": 0.584812610035929,
+      "learning_rate": 3.1448333587407486e-06,
+      "loss": 0.0292,
+      "step": 698
+    },
+    {
+      "epoch": 1.2437722419928825,
+      "grad_norm": 0.47034924316665355,
+      "learning_rate": 3.131863111508667e-06,
+      "loss": 0.0247,
+      "step": 699
+    },
+    {
+      "epoch": 1.2455516014234875,
+      "grad_norm": 0.5674732033417192,
+      "learning_rate": 3.118907458316722e-06,
+      "loss": 0.0267,
+      "step": 700
+    },
+    {
+      "epoch": 1.2473309608540926,
+      "grad_norm": 0.5121130860950636,
+      "learning_rate": 3.105966500375551e-06,
+      "loss": 0.025,
+      "step": 701
+    },
+    {
+      "epoch": 1.2491103202846976,
+      "grad_norm": 0.669884869505064,
+      "learning_rate": 3.0930403387809892e-06,
+      "loss": 0.0446,
+      "step": 702
+    },
+    {
+      "epoch": 1.2508896797153026,
+      "grad_norm": 0.6070756572461927,
+      "learning_rate": 3.080129074513285e-06,
+      "loss": 0.0344,
+      "step": 703
+    },
+    {
+      "epoch": 1.2526690391459074,
+      "grad_norm": 0.5092051656790545,
+      "learning_rate": 3.067232808436299e-06,
+      "loss": 0.026,
+      "step": 704
+    },
+    {
+      "epoch": 1.2544483985765125,
+      "grad_norm": 0.5214283435400325,
+      "learning_rate": 3.0543516412967327e-06,
+      "loss": 0.0258,
+      "step": 705
+    },
+    {
+      "epoch": 1.2562277580071175,
+      "grad_norm": 0.5061791936684357,
+      "learning_rate": 3.041485673723331e-06,
+      "loss": 0.0268,
+      "step": 706
+    },
+    {
+      "epoch": 1.2580071174377223,
+      "grad_norm": 0.5017957108850589,
+      "learning_rate": 3.0286350062261017e-06,
+      "loss": 0.0231,
+      "step": 707
+    },
+    {
+      "epoch": 1.2597864768683273,
+      "grad_norm": 0.4929547543051328,
+      "learning_rate": 3.0157997391955172e-06,
+      "loss": 0.0277,
+      "step": 708
+    },
+    {
+      "epoch": 1.2615658362989324,
+      "grad_norm": 0.6091803586415299,
+      "learning_rate": 3.0029799729017518e-06,
+      "loss": 0.0338,
+      "step": 709
+    },
+    {
+      "epoch": 1.2633451957295374,
+      "grad_norm": 0.5008439695979372,
+      "learning_rate": 2.9901758074938797e-06,
+      "loss": 0.0301,
+      "step": 710
+    },
+    {
+      "epoch": 1.2651245551601424,
+      "grad_norm": 0.5777913879936855,
+      "learning_rate": 2.977387342999103e-06,
+      "loss": 0.0266,
+      "step": 711
+    },
+    {
+      "epoch": 1.2669039145907472,
+      "grad_norm": 0.7046890763652409,
+      "learning_rate": 2.964614679321966e-06,
+      "loss": 0.0468,
+      "step": 712
+    },
+    {
+      "epoch": 1.2686832740213523,
+      "grad_norm": 0.5273334941063192,
+      "learning_rate": 2.951857916243574e-06,
+      "loss": 0.0329,
+      "step": 713
+    },
+    {
+      "epoch": 1.2704626334519573,
+      "grad_norm": 0.8341457043185303,
+      "learning_rate": 2.9391171534208185e-06,
+      "loss": 0.0499,
+      "step": 714
+    },
+    {
+      "epoch": 1.2722419928825623,
+      "grad_norm": 0.6780952467306876,
+      "learning_rate": 2.9263924903855932e-06,
+      "loss": 0.0299,
+      "step": 715
+    },
+    {
+      "epoch": 1.2740213523131674,
+      "grad_norm": 0.47572326746507165,
+      "learning_rate": 2.9136840265440213e-06,
+      "loss": 0.0217,
+      "step": 716
+    },
+    {
+      "epoch": 1.2758007117437722,
+      "grad_norm": 0.5645590556805177,
+      "learning_rate": 2.9009918611756732e-06,
+      "loss": 0.03,
+      "step": 717
+    },
+    {
+      "epoch": 1.2775800711743772,
+      "grad_norm": 0.5607967438422882,
+      "learning_rate": 2.8883160934327968e-06,
+      "loss": 0.0241,
+      "step": 718
+    },
+    {
+      "epoch": 1.2793594306049823,
+      "grad_norm": 0.7680013137247107,
+      "learning_rate": 2.8756568223395396e-06,
+      "loss": 0.0399,
+      "step": 719
+    },
+    {
+      "epoch": 1.281138790035587,
+      "grad_norm": 0.5290992655416129,
+      "learning_rate": 2.8630141467911777e-06,
+      "loss": 0.0247,
+      "step": 720
+    },
+    {
+      "epoch": 1.282918149466192,
+      "grad_norm": 0.5201662095762271,
+      "learning_rate": 2.8503881655533395e-06,
+      "loss": 0.0242,
+      "step": 721
+    },
+    {
+      "epoch": 1.2846975088967971,
+      "grad_norm": 0.5464649329916923,
+      "learning_rate": 2.837778977261235e-06,
+      "loss": 0.0361,
+      "step": 722
+    },
+    {
+      "epoch": 1.2864768683274022,
+      "grad_norm": 0.5343221036154335,
+      "learning_rate": 2.8251866804188875e-06,
+      "loss": 0.0309,
+      "step": 723
+    },
+    {
+      "epoch": 1.2882562277580072,
+      "grad_norm": 0.5454331280658008,
+      "learning_rate": 2.812611373398365e-06,
+      "loss": 0.0347,
+      "step": 724
+    },
+    {
+      "epoch": 1.290035587188612,
+      "grad_norm": 0.608672589185765,
+      "learning_rate": 2.8000531544390064e-06,
+      "loss": 0.0289,
+      "step": 725
+    },
+    {
+      "epoch": 1.291814946619217,
+      "grad_norm": 0.7258360439096404,
+      "learning_rate": 2.7875121216466595e-06,
+      "loss": 0.0428,
+      "step": 726
+    },
+    {
+      "epoch": 1.293594306049822,
+      "grad_norm": 0.6630072927506494,
+      "learning_rate": 2.7749883729929105e-06,
+      "loss": 0.0375,
+      "step": 727
+    },
+    {
+      "epoch": 1.295373665480427,
+      "grad_norm": 0.6047532282495882,
+      "learning_rate": 2.762482006314324e-06,
+      "loss": 0.0296,
+      "step": 728
+    },
+    {
+      "epoch": 1.2971530249110321,
+      "grad_norm": 0.592232045467234,
+      "learning_rate": 2.7499931193116692e-06,
+      "loss": 0.0275,
+      "step": 729
+    },
+    {
+      "epoch": 1.298932384341637,
+      "grad_norm": 0.5699824966826516,
+      "learning_rate": 2.737521809549167e-06,
+      "loss": 0.0326,
+      "step": 730
+    },
+    {
+      "epoch": 1.300711743772242,
+      "grad_norm": 0.5091173852911484,
+      "learning_rate": 2.725068174453722e-06,
+      "loss": 0.0298,
+      "step": 731
+    },
+    {
+      "epoch": 1.302491103202847,
+      "grad_norm": 0.6043390966207082,
+      "learning_rate": 2.712632311314165e-06,
+      "loss": 0.0275,
+      "step": 732
+    },
+    {
+      "epoch": 1.304270462633452,
+      "grad_norm": 0.8275102741011439,
+      "learning_rate": 2.7002143172804875e-06,
+      "loss": 0.0381,
+      "step": 733
+    },
+    {
+      "epoch": 1.306049822064057,
+      "grad_norm": 0.5893884216460472,
+      "learning_rate": 2.6878142893630904e-06,
+      "loss": 0.0345,
+      "step": 734
+    },
+    {
+      "epoch": 1.3078291814946619,
+      "grad_norm": 0.5321531772824266,
+      "learning_rate": 2.6754323244320154e-06,
+      "loss": 0.0355,
+      "step": 735
+    },
+    {
+      "epoch": 1.309608540925267,
+      "grad_norm": 0.6588685079069572,
+      "learning_rate": 2.6630685192161995e-06,
+      "loss": 0.0383,
+      "step": 736
+    },
+    {
+      "epoch": 1.311387900355872,
+      "grad_norm": 0.5911013661546711,
+      "learning_rate": 2.650722970302714e-06,
+      "loss": 0.0284,
+      "step": 737
+    },
+    {
+      "epoch": 1.3131672597864767,
+      "grad_norm": 0.6677144198716436,
+      "learning_rate": 2.638395774136009e-06,
+      "loss": 0.037,
+      "step": 738
+    },
+    {
+      "epoch": 1.3149466192170818,
+      "grad_norm": 0.6111546023515535,
+      "learning_rate": 2.6260870270171645e-06,
+      "loss": 0.0324,
+      "step": 739
+    },
+    {
+      "epoch": 1.3167259786476868,
+      "grad_norm": 0.5224366905001679,
+      "learning_rate": 2.613796825103129e-06,
+      "loss": 0.0357,
+      "step": 740
+    },
+    {
+      "epoch": 1.3185053380782918,
+      "grad_norm": 0.6137618372785942,
+      "learning_rate": 2.60152526440598e-06,
+      "loss": 0.0302,
+      "step": 741
+    },
+    {
+      "epoch": 1.3202846975088969,
+      "grad_norm": 0.44710040077805935,
+      "learning_rate": 2.5892724407921667e-06,
+      "loss": 0.019,
+      "step": 742
+    },
+    {
+      "epoch": 1.3220640569395017,
+      "grad_norm": 0.7750042599220767,
+      "learning_rate": 2.577038449981763e-06,
+      "loss": 0.0398,
+      "step": 743
+    },
+    {
+      "epoch": 1.3238434163701067,
+      "grad_norm": 0.6740594243534924,
+      "learning_rate": 2.564823387547716e-06,
+      "loss": 0.0376,
+      "step": 744
+    },
+    {
+      "epoch": 1.3256227758007118,
+      "grad_norm": 0.5686128243233131,
+      "learning_rate": 2.552627348915106e-06,
+      "loss": 0.0366,
+      "step": 745
+    },
+    {
+      "epoch": 1.3274021352313168,
+      "grad_norm": 0.7647805522047961,
+      "learning_rate": 2.5404504293603983e-06,
+      "loss": 0.0476,
+      "step": 746
+    },
+    {
+      "epoch": 1.3291814946619218,
+      "grad_norm": 0.7225875925284596,
+      "learning_rate": 2.528292724010697e-06,
+      "loss": 0.0499,
+      "step": 747
+    },
+    {
+      "epoch": 1.3309608540925266,
+      "grad_norm": 0.5385502947698766,
+      "learning_rate": 2.5161543278430055e-06,
+      "loss": 0.033,
+      "step": 748
+    },
+    {
+      "epoch": 1.3327402135231317,
+      "grad_norm": 0.7176207299997726,
+      "learning_rate": 2.5040353356834756e-06,
+      "loss": 0.0404,
+      "step": 749
+    },
+    {
+      "epoch": 1.3345195729537367,
+      "grad_norm": 0.5633544534189135,
+      "learning_rate": 2.4919358422066816e-06,
+      "loss": 0.0277,
+      "step": 750
+    },
+    {
+      "epoch": 1.3362989323843417,
+      "grad_norm": 0.47376558087374704,
+      "learning_rate": 2.4798559419348672e-06,
+      "loss": 0.0219,
+      "step": 751
+    },
+    {
+      "epoch": 1.3380782918149468,
+      "grad_norm": 0.4431153157508507,
+      "learning_rate": 2.4677957292372166e-06,
+      "loss": 0.0233,
+      "step": 752
+    },
+    {
+      "epoch": 1.3398576512455516,
+      "grad_norm": 0.48747659617260786,
+      "learning_rate": 2.455755298329107e-06,
+      "loss": 0.0257,
+      "step": 753
+    },
+    {
+      "epoch": 1.3416370106761566,
+      "grad_norm": 0.5533858346909668,
+      "learning_rate": 2.4437347432713838e-06,
+      "loss": 0.0303,
+      "step": 754
+    },
+    {
+      "epoch": 1.3434163701067616,
+      "grad_norm": 0.5721842794425006,
+      "learning_rate": 2.431734157969619e-06,
+      "loss": 0.0279,
+      "step": 755
+    },
+    {
+      "epoch": 1.3451957295373664,
+      "grad_norm": 0.6553462119225486,
+      "learning_rate": 2.4197536361733792e-06,
+      "loss": 0.0411,
+      "step": 756
+    },
+    {
+      "epoch": 1.3469750889679715,
+      "grad_norm": 0.46611418401937593,
+      "learning_rate": 2.407793271475495e-06,
+      "loss": 0.0283,
+      "step": 757
+    },
+    {
+      "epoch": 1.3487544483985765,
+      "grad_norm": 0.5393248523056013,
+      "learning_rate": 2.3958531573113223e-06,
+      "loss": 0.0429,
+      "step": 758
+    },
+    {
+      "epoch": 1.3505338078291815,
+      "grad_norm": 0.5513670033925337,
+      "learning_rate": 2.3839333869580243e-06,
+      "loss": 0.0363,
+      "step": 759
+    },
+    {
+      "epoch": 1.3523131672597866,
+      "grad_norm": 0.5004589620656431,
+      "learning_rate": 2.372034053533835e-06,
+      "loss": 0.0274,
+      "step": 760
+    },
+    {
+      "epoch": 1.3540925266903914,
+      "grad_norm": 0.3990498324321676,
+      "learning_rate": 2.360155249997334e-06,
+      "loss": 0.0303,
+      "step": 761
+    },
+    {
+      "epoch": 1.3558718861209964,
+      "grad_norm": 0.4563630642805391,
+      "learning_rate": 2.348297069146715e-06,
+      "loss": 0.0235,
+      "step": 762
+    },
+    {
+      "epoch": 1.3576512455516014,
+      "grad_norm": 0.4957616146268126,
+      "learning_rate": 2.3364596036190706e-06,
+      "loss": 0.0266,
+      "step": 763
+    },
+    {
+      "epoch": 1.3594306049822065,
+      "grad_norm": 0.5476841760593381,
+      "learning_rate": 2.3246429458896637e-06,
+      "loss": 0.037,
+      "step": 764
+    },
+    {
+      "epoch": 1.3612099644128115,
+      "grad_norm": 0.6575783862688775,
+      "learning_rate": 2.312847188271203e-06,
+      "loss": 0.0425,
+      "step": 765
+    },
+    {
+      "epoch": 1.3629893238434163,
+      "grad_norm": 0.5771476743858435,
+      "learning_rate": 2.301072422913123e-06,
+      "loss": 0.0272,
+      "step": 766
+    },
+    {
+      "epoch": 1.3647686832740213,
+      "grad_norm": 0.5795858816918574,
+      "learning_rate": 2.2893187418008666e-06,
+      "loss": 0.0279,
+      "step": 767
+    },
+    {
+      "epoch": 1.3665480427046264,
+      "grad_norm": 0.631047476081519,
+      "learning_rate": 2.2775862367551642e-06,
+      "loss": 0.0342,
+      "step": 768
+    },
+    {
+      "epoch": 1.3683274021352312,
+      "grad_norm": 0.6505230123904633,
+      "learning_rate": 2.265874999431318e-06,
+      "loss": 0.0347,
+      "step": 769
+    },
+    {
+      "epoch": 1.3701067615658362,
+      "grad_norm": 0.5575570299354918,
+      "learning_rate": 2.254185121318484e-06,
+      "loss": 0.0255,
+      "step": 770
+    },
+    {
+      "epoch": 1.3718861209964412,
+      "grad_norm": 0.6228271772420306,
+      "learning_rate": 2.2425166937389596e-06,
+      "loss": 0.0346,
+      "step": 771
+    },
+    {
+      "epoch": 1.3736654804270463,
+      "grad_norm": 0.5181242061773714,
+      "learning_rate": 2.2308698078474645e-06,
+      "loss": 0.0319,
+      "step": 772
+    },
+    {
+      "epoch": 1.3754448398576513,
+      "grad_norm": 0.5798039287937309,
+      "learning_rate": 2.219244554630438e-06,
+      "loss": 0.0351,
+      "step": 773
+    },
+    {
+      "epoch": 1.3772241992882561,
+      "grad_norm": 0.548955425936005,
+      "learning_rate": 2.207641024905322e-06,
+      "loss": 0.0309,
+      "step": 774
+    },
+    {
+      "epoch": 1.3790035587188612,
+      "grad_norm": 0.5991790962444122,
+      "learning_rate": 2.1960593093198508e-06,
+      "loss": 0.0308,
+      "step": 775
+    },
+    {
+      "epoch": 1.3807829181494662,
+      "grad_norm": 0.5475701382598392,
+      "learning_rate": 2.184499498351347e-06,
+      "loss": 0.0273,
+      "step": 776
+    },
+    {
+      "epoch": 1.3825622775800712,
+      "grad_norm": 0.47004287037346615,
+      "learning_rate": 2.172961682306011e-06,
+      "loss": 0.0281,
+      "step": 777
+    },
+    {
+      "epoch": 1.3843416370106763,
+      "grad_norm": 0.47426615282421464,
+      "learning_rate": 2.1614459513182173e-06,
+      "loss": 0.0272,
+      "step": 778
+    },
+    {
+      "epoch": 1.386120996441281,
+      "grad_norm": 0.5211254930227172,
+      "learning_rate": 2.149952395349813e-06,
+      "loss": 0.0243,
+      "step": 779
+    },
+    {
+      "epoch": 1.387900355871886,
+      "grad_norm": 0.5991031776407816,
+      "learning_rate": 2.1384811041894055e-06,
+      "loss": 0.0278,
+      "step": 780
+    },
+    {
+      "epoch": 1.3896797153024911,
+      "grad_norm": 0.6181129581484839,
+      "learning_rate": 2.1270321674516736e-06,
+      "loss": 0.034,
+      "step": 781
+    },
+    {
+      "epoch": 1.3914590747330962,
+      "grad_norm": 0.5193874899718711,
+      "learning_rate": 2.1156056745766593e-06,
+      "loss": 0.0252,
+      "step": 782
+    },
+    {
+      "epoch": 1.3932384341637012,
+      "grad_norm": 0.5220850523384505,
+      "learning_rate": 2.104201714829074e-06,
+      "loss": 0.0289,
+      "step": 783
+    },
+    {
+      "epoch": 1.395017793594306,
+      "grad_norm": 0.504280384412821,
+      "learning_rate": 2.0928203772975917e-06,
+      "loss": 0.0303,
+      "step": 784
+    },
+    {
+      "epoch": 1.396797153024911,
+      "grad_norm": 0.521563094542065,
+      "learning_rate": 2.081461750894166e-06,
+      "loss": 0.0294,
+      "step": 785
+    },
+    {
+      "epoch": 1.398576512455516,
+      "grad_norm": 0.5827930463180666,
+      "learning_rate": 2.070125924353328e-06,
+      "loss": 0.0236,
+      "step": 786
+    },
+    {
+      "epoch": 1.4003558718861209,
+      "grad_norm": 0.4714411350415353,
+      "learning_rate": 2.058812986231493e-06,
+      "loss": 0.0321,
+      "step": 787
+    },
+    {
+      "epoch": 1.402135231316726,
+      "grad_norm": 0.6566988221180272,
+      "learning_rate": 2.0475230249062727e-06,
+      "loss": 0.0546,
+      "step": 788
+    },
+    {
+      "epoch": 1.403914590747331,
+      "grad_norm": 0.555464519745758,
+      "learning_rate": 2.0362561285757766e-06,
+      "loss": 0.0283,
+      "step": 789
+    },
+    {
+      "epoch": 1.405693950177936,
+      "grad_norm": 0.6375046837873795,
+      "learning_rate": 2.0250123852579347e-06,
+      "loss": 0.028,
+      "step": 790
+    },
+    {
+      "epoch": 1.407473309608541,
+      "grad_norm": 0.45069498425647225,
+      "learning_rate": 2.013791882789801e-06,
+      "loss": 0.0199,
+      "step": 791
+    },
+    {
+      "epoch": 1.4092526690391458,
+      "grad_norm": 0.5452636683420757,
+      "learning_rate": 2.0025947088268714e-06,
+      "loss": 0.0222,
+      "step": 792
+    },
+    {
+      "epoch": 1.4110320284697508,
+      "grad_norm": 0.5149639562315755,
+      "learning_rate": 1.9914209508423943e-06,
+      "loss": 0.0292,
+      "step": 793
+    },
+    {
+      "epoch": 1.4128113879003559,
+      "grad_norm": 0.532869743711829,
+      "learning_rate": 1.9802706961266936e-06,
+      "loss": 0.0323,
+      "step": 794
+    },
+    {
+      "epoch": 1.414590747330961,
+      "grad_norm": 0.7152251845775712,
+      "learning_rate": 1.969144031786483e-06,
+      "loss": 0.0413,
+      "step": 795
+    },
+    {
+      "epoch": 1.416370106761566,
+      "grad_norm": 0.5509582517202217,
+      "learning_rate": 1.958041044744186e-06,
+      "loss": 0.0352,
+      "step": 796
+    },
+    {
+      "epoch": 1.4181494661921707,
+      "grad_norm": 0.5160483721103603,
+      "learning_rate": 1.94696182173726e-06,
+      "loss": 0.0322,
+      "step": 797
+    },
+    {
+      "epoch": 1.4199288256227758,
+      "grad_norm": 0.5346773436249085,
+      "learning_rate": 1.9359064493175077e-06,
+      "loss": 0.0314,
+      "step": 798
+    },
+    {
+      "epoch": 1.4217081850533808,
+      "grad_norm": 0.5821721590900316,
+      "learning_rate": 1.9248750138504176e-06,
+      "loss": 0.0313,
+      "step": 799
+    },
+    {
+      "epoch": 1.4234875444839858,
+      "grad_norm": 0.6268323389881963,
+      "learning_rate": 1.9138676015144765e-06,
+      "loss": 0.0337,
+      "step": 800
+    },
+    {
+      "epoch": 1.4234875444839858,
+      "eval_loss": 0.10909521579742432,
+      "eval_runtime": 1.903,
+      "eval_samples_per_second": 24.173,
+      "eval_steps_per_second": 6.306,
+      "step": 800
+    },
+    {
+      "epoch": 1.4252669039145909,
+      "grad_norm": 0.6540528546925491,
+      "learning_rate": 1.9028842983005036e-06,
+      "loss": 0.0379,
+      "step": 801
+    },
+    {
+      "epoch": 1.4270462633451957,
+      "grad_norm": 0.6363023110009841,
+      "learning_rate": 1.8919251900109697e-06,
+      "loss": 0.0376,
+      "step": 802
+    },
+    {
+      "epoch": 1.4288256227758007,
+      "grad_norm": 0.4658581295153636,
+      "learning_rate": 1.8809903622593395e-06,
+      "loss": 0.0252,
+      "step": 803
+    },
+    {
+      "epoch": 1.4306049822064058,
+      "grad_norm": 0.46196243197885767,
+      "learning_rate": 1.870079900469392e-06,
+      "loss": 0.0207,
+      "step": 804
+    },
+    {
+      "epoch": 1.4323843416370106,
+      "grad_norm": 0.4988549367276664,
+      "learning_rate": 1.8591938898745593e-06,
+      "loss": 0.0273,
+      "step": 805
+    },
+    {
+      "epoch": 1.4341637010676156,
+      "grad_norm": 0.4521542573559007,
+      "learning_rate": 1.8483324155172594e-06,
+      "loss": 0.0247,
+      "step": 806
+    },
+    {
+      "epoch": 1.4359430604982206,
+      "grad_norm": 0.47412790030717633,
+      "learning_rate": 1.837495562248226e-06,
+      "loss": 0.0281,
+      "step": 807
+    },
+    {
+      "epoch": 1.4377224199288257,
+      "grad_norm": 0.5266325201621073,
+      "learning_rate": 1.8266834147258577e-06,
+      "loss": 0.03,
+      "step": 808
+    },
+    {
+      "epoch": 1.4395017793594307,
+      "grad_norm": 0.5878059625456945,
+      "learning_rate": 1.8158960574155455e-06,
+      "loss": 0.034,
+      "step": 809
+    },
+    {
+      "epoch": 1.4412811387900355,
+      "grad_norm": 0.5933800554756186,
+      "learning_rate": 1.8051335745890196e-06,
+      "loss": 0.0319,
+      "step": 810
+    },
+    {
+      "epoch": 1.4430604982206405,
+      "grad_norm": 0.562837427887017,
+      "learning_rate": 1.7943960503236856e-06,
+      "loss": 0.0399,
+      "step": 811
+    },
+    {
+      "epoch": 1.4448398576512456,
+      "grad_norm": 0.4758570720457722,
+      "learning_rate": 1.7836835685019732e-06,
+      "loss": 0.0256,
+      "step": 812
+    },
+    {
+      "epoch": 1.4466192170818506,
+      "grad_norm": 0.6328667032294728,
+      "learning_rate": 1.7729962128106787e-06,
+      "loss": 0.031,
+      "step": 813
+    },
+    {
+      "epoch": 1.4483985765124556,
+      "grad_norm": 0.5104103333255997,
+      "learning_rate": 1.7623340667403089e-06,
+      "loss": 0.0246,
+      "step": 814
+    },
+    {
+      "epoch": 1.4501779359430604,
+      "grad_norm": 0.7828964860965184,
+      "learning_rate": 1.7516972135844352e-06,
+      "loss": 0.0483,
+      "step": 815
+    },
+    {
+      "epoch": 1.4519572953736655,
+      "grad_norm": 0.6394325697318681,
+      "learning_rate": 1.741085736439031e-06,
+      "loss": 0.0311,
+      "step": 816
+    },
+    {
+      "epoch": 1.4537366548042705,
+      "grad_norm": 0.6006271736363957,
+      "learning_rate": 1.730499718201838e-06,
+      "loss": 0.0362,
+      "step": 817
+    },
+    {
+      "epoch": 1.4555160142348753,
+      "grad_norm": 0.5706527226628239,
+      "learning_rate": 1.7199392415717064e-06,
+      "loss": 0.031,
+      "step": 818
+    },
+    {
+      "epoch": 1.4572953736654806,
+      "grad_norm": 0.5533539085681772,
+      "learning_rate": 1.7094043890479557e-06,
+      "loss": 0.0378,
+      "step": 819
+    },
+    {
+      "epoch": 1.4590747330960854,
+      "grad_norm": 0.4531577011649327,
+      "learning_rate": 1.698895242929725e-06,
+      "loss": 0.0256,
+      "step": 820
+    },
+    {
+      "epoch": 1.4608540925266904,
+      "grad_norm": 0.5973047534221524,
+      "learning_rate": 1.6884118853153358e-06,
+      "loss": 0.0256,
+      "step": 821
+    },
+    {
+      "epoch": 1.4626334519572954,
+      "grad_norm": 0.5265224300995814,
+      "learning_rate": 1.6779543981016478e-06,
+      "loss": 0.0272,
+      "step": 822
+    },
+    {
+      "epoch": 1.4644128113879002,
+      "grad_norm": 0.5390737293098566,
+      "learning_rate": 1.6675228629834133e-06,
+      "loss": 0.0337,
+      "step": 823
+    },
+    {
+      "epoch": 1.4661921708185053,
+      "grad_norm": 0.45937605422817024,
+      "learning_rate": 1.657117361452651e-06,
+      "loss": 0.0224,
+      "step": 824
+    },
+    {
+      "epoch": 1.4679715302491103,
+      "grad_norm": 0.4464493512984618,
+      "learning_rate": 1.6467379747980011e-06,
+      "loss": 0.0274,
+      "step": 825
+    },
+    {
+      "epoch": 1.4697508896797153,
+      "grad_norm": 0.46139514829440215,
+      "learning_rate": 1.6363847841040914e-06,
+      "loss": 0.0255,
+      "step": 826
+    },
+    {
+      "epoch": 1.4715302491103204,
+      "grad_norm": 0.4236571274212341,
+      "learning_rate": 1.626057870250906e-06,
+      "loss": 0.0228,
+      "step": 827
+    },
+    {
+      "epoch": 1.4733096085409252,
+      "grad_norm": 0.47879598721645045,
+      "learning_rate": 1.6157573139131527e-06,
+      "loss": 0.0216,
+      "step": 828
+    },
+    {
+      "epoch": 1.4750889679715302,
+      "grad_norm": 0.5866221554250787,
+      "learning_rate": 1.605483195559628e-06,
+      "loss": 0.0365,
+      "step": 829
+    },
+    {
+      "epoch": 1.4768683274021353,
+      "grad_norm": 0.4965045925956703,
+      "learning_rate": 1.5952355954525966e-06,
+      "loss": 0.0243,
+      "step": 830
+    },
+    {
+      "epoch": 1.4786476868327403,
+      "grad_norm": 0.566095657670681,
+      "learning_rate": 1.5850145936471607e-06,
+      "loss": 0.0285,
+      "step": 831
+    },
+    {
+      "epoch": 1.4804270462633453,
+      "grad_norm": 0.5947924806696321,
+      "learning_rate": 1.5748202699906335e-06,
+      "loss": 0.0281,
+      "step": 832
+    },
+    {
+      "epoch": 1.4822064056939501,
+      "grad_norm": 0.4880024684583829,
+      "learning_rate": 1.5646527041219128e-06,
+      "loss": 0.0229,
+      "step": 833
+    },
+    {
+      "epoch": 1.4839857651245552,
+      "grad_norm": 0.42166752613543057,
+      "learning_rate": 1.5545119754708682e-06,
+      "loss": 0.0198,
+      "step": 834
+    },
+    {
+      "epoch": 1.4857651245551602,
+      "grad_norm": 0.5533395505543854,
+      "learning_rate": 1.544398163257711e-06,
+      "loss": 0.0345,
+      "step": 835
+    },
+    {
+      "epoch": 1.487544483985765,
+      "grad_norm": 0.5564318355103515,
+      "learning_rate": 1.5343113464923808e-06,
+      "loss": 0.0289,
+      "step": 836
+    },
+    {
+      "epoch": 1.48932384341637,
+      "grad_norm": 0.42156174636794386,
+      "learning_rate": 1.524251603973927e-06,
+      "loss": 0.0174,
+      "step": 837
+    },
+    {
+      "epoch": 1.491103202846975,
+      "grad_norm": 0.5435587659292476,
+      "learning_rate": 1.5142190142898883e-06,
+      "loss": 0.0322,
+      "step": 838
+    },
+    {
+      "epoch": 1.49288256227758,
+      "grad_norm": 0.6349234666147967,
+      "learning_rate": 1.5042136558156883e-06,
+      "loss": 0.0375,
+      "step": 839
+    },
+    {
+      "epoch": 1.4946619217081851,
+      "grad_norm": 0.6520204010327719,
+      "learning_rate": 1.4942356067140162e-06,
+      "loss": 0.0355,
+      "step": 840
+    },
+    {
+      "epoch": 1.49644128113879,
+      "grad_norm": 0.4904477135197103,
+      "learning_rate": 1.4842849449342195e-06,
+      "loss": 0.0345,
+      "step": 841
+    },
+    {
+      "epoch": 1.498220640569395,
+      "grad_norm": 0.6026605866286852,
+      "learning_rate": 1.4743617482116896e-06,
+      "loss": 0.0408,
+      "step": 842
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6427782982675385,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.0283,
+      "step": 843
+    },
+    {
+      "epoch": 1.501779359430605,
+      "grad_norm": 0.5264112453735525,
+      "learning_rate": 1.454598059806609e-06,
+      "loss": 0.0292,
+      "step": 844
+    },
+    {
+      "epoch": 1.50355871886121,
+      "grad_norm": 0.9685150894825538,
+      "learning_rate": 1.4447577225196296e-06,
+      "loss": 0.0448,
+      "step": 845
+    },
+    {
+      "epoch": 1.5053380782918149,
+      "grad_norm": 0.6340825081407843,
+      "learning_rate": 1.4349451590798564e-06,
+      "loss": 0.0321,
+      "step": 846
+    },
+    {
+      "epoch": 1.50711743772242,
+      "grad_norm": 0.46941783770658396,
+      "learning_rate": 1.4251604461438444e-06,
+      "loss": 0.0256,
+      "step": 847
+    },
+    {
+      "epoch": 1.508896797153025,
+      "grad_norm": 0.6257226618519582,
+      "learning_rate": 1.4154036601505834e-06,
+      "loss": 0.0357,
+      "step": 848
+    },
+    {
+      "epoch": 1.5106761565836297,
+      "grad_norm": 0.5184599761838135,
+      "learning_rate": 1.4056748773208933e-06,
+      "loss": 0.0253,
+      "step": 849
+    },
+    {
+      "epoch": 1.512455516014235,
+      "grad_norm": 0.6036303241474837,
+      "learning_rate": 1.3959741736568339e-06,
+      "loss": 0.0305,
+      "step": 850
+    },
+    {
+      "epoch": 1.5142348754448398,
+      "grad_norm": 0.43430192725541933,
+      "learning_rate": 1.3863016249411027e-06,
+      "loss": 0.0204,
+      "step": 851
+    },
+    {
+      "epoch": 1.5160142348754448,
+      "grad_norm": 0.4661616992468015,
+      "learning_rate": 1.376657306736453e-06,
+      "loss": 0.0267,
+      "step": 852
+    },
+    {
+      "epoch": 1.5177935943060499,
+      "grad_norm": 0.5703917529364232,
+      "learning_rate": 1.3670412943850975e-06,
+      "loss": 0.0356,
+      "step": 853
+    },
+    {
+      "epoch": 1.5195729537366547,
+      "grad_norm": 0.5382879217607484,
+      "learning_rate": 1.3574536630081208e-06,
+      "loss": 0.0205,
+      "step": 854
+    },
+    {
+      "epoch": 1.52135231316726,
+      "grad_norm": 0.6724666870092444,
+      "learning_rate": 1.347894487504896e-06,
+      "loss": 0.0321,
+      "step": 855
+    },
+    {
+      "epoch": 1.5231316725978647,
+      "grad_norm": 0.47783003074778535,
+      "learning_rate": 1.3383638425524909e-06,
+      "loss": 0.0255,
+      "step": 856
+    },
+    {
+      "epoch": 1.5249110320284698,
+      "grad_norm": 0.5174114634621159,
+      "learning_rate": 1.3288618026050943e-06,
+      "loss": 0.0258,
+      "step": 857
+    },
+    {
+      "epoch": 1.5266903914590748,
+      "grad_norm": 0.6878772716666101,
+      "learning_rate": 1.31938844189343e-06,
+      "loss": 0.0366,
+      "step": 858
+    },
+    {
+      "epoch": 1.5284697508896796,
+      "grad_norm": 0.5826104090247197,
+      "learning_rate": 1.3099438344241777e-06,
+      "loss": 0.0332,
+      "step": 859
+    },
+    {
+      "epoch": 1.5302491103202847,
+      "grad_norm": 0.4638446299445823,
+      "learning_rate": 1.3005280539793908e-06,
+      "loss": 0.0241,
+      "step": 860
+    },
+    {
+      "epoch": 1.5320284697508897,
+      "grad_norm": 0.5619534686110462,
+      "learning_rate": 1.2911411741159273e-06,
+      "loss": 0.0196,
+      "step": 861
+    },
+    {
+      "epoch": 1.5338078291814945,
+      "grad_norm": 0.6336238375125608,
+      "learning_rate": 1.2817832681648712e-06,
+      "loss": 0.0428,
+      "step": 862
+    },
+    {
+      "epoch": 1.5355871886120998,
+      "grad_norm": 0.6325112253247881,
+      "learning_rate": 1.2724544092309581e-06,
+      "loss": 0.0335,
+      "step": 863
+    },
+    {
+      "epoch": 1.5373665480427046,
+      "grad_norm": 0.5614379138249233,
+      "learning_rate": 1.2631546701920073e-06,
+      "loss": 0.0345,
+      "step": 864
+    },
+    {
+      "epoch": 1.5391459074733096,
+      "grad_norm": 0.43298690490211045,
+      "learning_rate": 1.2538841236983519e-06,
+      "loss": 0.0225,
+      "step": 865
+    },
+    {
+      "epoch": 1.5409252669039146,
+      "grad_norm": 0.6571668286720693,
+      "learning_rate": 1.244642842172266e-06,
+      "loss": 0.0267,
+      "step": 866
+    },
+    {
+      "epoch": 1.5427046263345194,
+      "grad_norm": 0.5615346865470033,
+      "learning_rate": 1.2354308978074088e-06,
+      "loss": 0.0253,
+      "step": 867
+    },
+    {
+      "epoch": 1.5444839857651247,
+      "grad_norm": 0.6007861480383755,
+      "learning_rate": 1.2262483625682514e-06,
+      "loss": 0.0376,
+      "step": 868
+    },
+    {
+      "epoch": 1.5462633451957295,
+      "grad_norm": 0.49354539806250364,
+      "learning_rate": 1.2170953081895214e-06,
+      "loss": 0.0273,
+      "step": 869
+    },
+    {
+      "epoch": 1.5480427046263345,
+      "grad_norm": 0.4931708418219992,
+      "learning_rate": 1.2079718061756369e-06,
+      "loss": 0.0223,
+      "step": 870
+    },
+    {
+      "epoch": 1.5498220640569396,
+      "grad_norm": 0.5023781916796228,
+      "learning_rate": 1.1988779278001517e-06,
+      "loss": 0.0291,
+      "step": 871
+    },
+    {
+      "epoch": 1.5516014234875444,
+      "grad_norm": 0.487836444715539,
+      "learning_rate": 1.1898137441051982e-06,
+      "loss": 0.0295,
+      "step": 872
+    },
+    {
+      "epoch": 1.5533807829181496,
+      "grad_norm": 0.579657188679387,
+      "learning_rate": 1.1807793259009282e-06,
+      "loss": 0.0387,
+      "step": 873
+    },
+    {
+      "epoch": 1.5551601423487544,
+      "grad_norm": 0.591518676066802,
+      "learning_rate": 1.1717747437649657e-06,
+      "loss": 0.0372,
+      "step": 874
+    },
+    {
+      "epoch": 1.5569395017793595,
+      "grad_norm": 0.5957302265496146,
+      "learning_rate": 1.1628000680418533e-06,
+      "loss": 0.0351,
+      "step": 875
+    },
+    {
+      "epoch": 1.5587188612099645,
+      "grad_norm": 0.6143280015707291,
+      "learning_rate": 1.1538553688425002e-06,
+      "loss": 0.0287,
+      "step": 876
+    },
+    {
+      "epoch": 1.5604982206405693,
+      "grad_norm": 0.5216049919179628,
+      "learning_rate": 1.14494071604364e-06,
+      "loss": 0.0268,
+      "step": 877
+    },
+    {
+      "epoch": 1.5622775800711743,
+      "grad_norm": 0.46167224333357065,
+      "learning_rate": 1.1360561792872754e-06,
+      "loss": 0.0197,
+      "step": 878
+    },
+    {
+      "epoch": 1.5640569395017794,
+      "grad_norm": 0.7102351969249161,
+      "learning_rate": 1.127201827980145e-06,
+      "loss": 0.0443,
+      "step": 879
+    },
+    {
+      "epoch": 1.5658362989323842,
+      "grad_norm": 0.44287453780740843,
+      "learning_rate": 1.1183777312931748e-06,
+      "loss": 0.0193,
+      "step": 880
+    },
+    {
+      "epoch": 1.5676156583629894,
+      "grad_norm": 0.5079174251675422,
+      "learning_rate": 1.1095839581609407e-06,
+      "loss": 0.0368,
+      "step": 881
+    },
+    {
+      "epoch": 1.5693950177935942,
+      "grad_norm": 0.6701106920259725,
+      "learning_rate": 1.1008205772811248e-06,
+      "loss": 0.0417,
+      "step": 882
+    },
+    {
+      "epoch": 1.5711743772241993,
+      "grad_norm": 0.5666474471923277,
+      "learning_rate": 1.0920876571139843e-06,
+      "loss": 0.0353,
+      "step": 883
+    },
+    {
+      "epoch": 1.5729537366548043,
+      "grad_norm": 0.4772762793956462,
+      "learning_rate": 1.0833852658818167e-06,
+      "loss": 0.0278,
+      "step": 884
+    },
+    {
+      "epoch": 1.5747330960854091,
+      "grad_norm": 0.5442519089451494,
+      "learning_rate": 1.0747134715684221e-06,
+      "loss": 0.024,
+      "step": 885
+    },
+    {
+      "epoch": 1.5765124555160144,
+      "grad_norm": 0.5565328990261434,
+      "learning_rate": 1.0660723419185776e-06,
+      "loss": 0.031,
+      "step": 886
+    },
+    {
+      "epoch": 1.5782918149466192,
+      "grad_norm": 0.4724702129945966,
+      "learning_rate": 1.0574619444375017e-06,
+      "loss": 0.0244,
+      "step": 887
+    },
+    {
+      "epoch": 1.5800711743772242,
+      "grad_norm": 0.4678087634041709,
+      "learning_rate": 1.0488823463903341e-06,
+      "loss": 0.0247,
+      "step": 888
+    },
+    {
+      "epoch": 1.5818505338078293,
+      "grad_norm": 0.5269451450623034,
+      "learning_rate": 1.0403336148016053e-06,
+      "loss": 0.0363,
+      "step": 889
+    },
+    {
+      "epoch": 1.583629893238434,
+      "grad_norm": 0.4532049288036021,
+      "learning_rate": 1.0318158164547159e-06,
+      "loss": 0.0288,
+      "step": 890
+    },
+    {
+      "epoch": 1.585409252669039,
+      "grad_norm": 0.564319606141916,
+      "learning_rate": 1.0233290178914096e-06,
+      "loss": 0.0281,
+      "step": 891
+    },
+    {
+      "epoch": 1.5871886120996441,
+      "grad_norm": 0.4566043691096667,
+      "learning_rate": 1.014873285411262e-06,
+      "loss": 0.0261,
+      "step": 892
+    },
+    {
+      "epoch": 1.5889679715302492,
+      "grad_norm": 0.5673992363431041,
+      "learning_rate": 1.006448685071154e-06,
+      "loss": 0.0311,
+      "step": 893
+    },
+    {
+      "epoch": 1.5907473309608542,
+      "grad_norm": 0.6093723235170314,
+      "learning_rate": 9.980552826847635e-07,
+      "loss": 0.0333,
+      "step": 894
+    },
+    {
+      "epoch": 1.592526690391459,
+      "grad_norm": 0.657601058626444,
+      "learning_rate": 9.896931438220453e-07,
+      "loss": 0.0445,
+      "step": 895
+    },
+    {
+      "epoch": 1.594306049822064,
+      "grad_norm": 0.5646130664730646,
+      "learning_rate": 9.813623338087181e-07,
+      "loss": 0.0258,
+      "step": 896
+    },
+    {
+      "epoch": 1.596085409252669,
+      "grad_norm": 0.6127166797557376,
+      "learning_rate": 9.730629177257623e-07,
+      "loss": 0.0375,
+      "step": 897
+    },
+    {
+      "epoch": 1.5978647686832739,
+      "grad_norm": 0.4375042072095259,
+      "learning_rate": 9.64794960408903e-07,
+      "loss": 0.0206,
+      "step": 898
+    },
+    {
+      "epoch": 1.5996441281138791,
+      "grad_norm": 0.5486855351634862,
+      "learning_rate": 9.565585264481092e-07,
+      "loss": 0.0286,
+      "step": 899
+    },
+    {
+      "epoch": 1.601423487544484,
+      "grad_norm": 0.618260314386264,
+      "learning_rate": 9.483536801870835e-07,
+      "loss": 0.0352,
+      "step": 900
+    },
+    {
+      "epoch": 1.603202846975089,
+      "grad_norm": 0.5315533899280454,
+      "learning_rate": 9.401804857227648e-07,
+      "loss": 0.0232,
+      "step": 901
+    },
+    {
+      "epoch": 1.604982206405694,
+      "grad_norm": 0.6563575648306289,
+      "learning_rate": 9.320390069048258e-07,
+      "loss": 0.0369,
+      "step": 902
+    },
+    {
+      "epoch": 1.6067615658362988,
+      "grad_norm": 0.5970829961178313,
+      "learning_rate": 9.239293073351735e-07,
+      "loss": 0.0286,
+      "step": 903
+    },
+    {
+      "epoch": 1.608540925266904,
+      "grad_norm": 0.5268364867084087,
+      "learning_rate": 9.158514503674543e-07,
+      "loss": 0.025,
+      "step": 904
+    },
+    {
+      "epoch": 1.6103202846975089,
+      "grad_norm": 0.477387269954479,
+      "learning_rate": 9.078054991065532e-07,
+      "loss": 0.026,
+      "step": 905
+    },
+    {
+      "epoch": 1.612099644128114,
+      "grad_norm": 0.5647010800996194,
+      "learning_rate": 8.997915164081095e-07,
+      "loss": 0.0328,
+      "step": 906
+    },
+    {
+      "epoch": 1.613879003558719,
+      "grad_norm": 0.6271409318160498,
+      "learning_rate": 8.918095648780195e-07,
+      "loss": 0.0313,
+      "step": 907
+    },
+    {
+      "epoch": 1.6156583629893237,
+      "grad_norm": 0.5861591552337049,
+      "learning_rate": 8.838597068719518e-07,
+      "loss": 0.0321,
+      "step": 908
+    },
+    {
+      "epoch": 1.6174377224199288,
+      "grad_norm": 0.5348555151079447,
+      "learning_rate": 8.75942004494853e-07,
+      "loss": 0.0234,
+      "step": 909
+    },
+    {
+      "epoch": 1.6192170818505338,
+      "grad_norm": 0.7719770171120217,
+      "learning_rate": 8.680565196004704e-07,
+      "loss": 0.0472,
+      "step": 910
+    },
+    {
+      "epoch": 1.6209964412811388,
+      "grad_norm": 0.535882757195425,
+      "learning_rate": 8.602033137908666e-07,
+      "loss": 0.0294,
+      "step": 911
+    },
+    {
+      "epoch": 1.6227758007117439,
+      "grad_norm": 0.5451422102090946,
+      "learning_rate": 8.523824484159348e-07,
+      "loss": 0.0238,
+      "step": 912
+    },
+    {
+      "epoch": 1.6245551601423487,
+      "grad_norm": 0.46006495968318883,
+      "learning_rate": 8.445939845729245e-07,
+      "loss": 0.021,
+      "step": 913
+    },
+    {
+      "epoch": 1.6263345195729537,
+      "grad_norm": 0.504186947188953,
+      "learning_rate": 8.368379831059592e-07,
+      "loss": 0.03,
+      "step": 914
+    },
+    {
+      "epoch": 1.6281138790035588,
+      "grad_norm": 0.4482605871625423,
+      "learning_rate": 8.29114504605566e-07,
+      "loss": 0.0247,
+      "step": 915
+    },
+    {
+      "epoch": 1.6298932384341636,
+      "grad_norm": 0.5013499179171815,
+      "learning_rate": 8.21423609408199e-07,
+      "loss": 0.0306,
+      "step": 916
+    },
+    {
+      "epoch": 1.6316725978647688,
+      "grad_norm": 0.5656027821511815,
+      "learning_rate": 8.137653575957666e-07,
+      "loss": 0.0295,
+      "step": 917
+    },
+    {
+      "epoch": 1.6334519572953736,
+      "grad_norm": 0.5820898086053026,
+      "learning_rate": 8.061398089951678e-07,
+      "loss": 0.0357,
+      "step": 918
+    },
+    {
+      "epoch": 1.6352313167259787,
+      "grad_norm": 0.5305931322393914,
+      "learning_rate": 7.985470231778203e-07,
+      "loss": 0.0269,
+      "step": 919
+    },
+    {
+      "epoch": 1.6370106761565837,
+      "grad_norm": 0.5380435311245324,
+      "learning_rate": 7.909870594591951e-07,
+      "loss": 0.0237,
+      "step": 920
+    },
+    {
+      "epoch": 1.6387900355871885,
+      "grad_norm": 0.7144326342673704,
+      "learning_rate": 7.834599768983553e-07,
+      "loss": 0.0378,
+      "step": 921
+    },
+    {
+      "epoch": 1.6405693950177938,
+      "grad_norm": 0.42524466098017927,
+      "learning_rate": 7.759658342974951e-07,
+      "loss": 0.0187,
+      "step": 922
+    },
+    {
+      "epoch": 1.6423487544483986,
+      "grad_norm": 0.5731172641352889,
+      "learning_rate": 7.685046902014747e-07,
+      "loss": 0.035,
+      "step": 923
+    },
+    {
+      "epoch": 1.6441281138790036,
+      "grad_norm": 0.5901192946651156,
+      "learning_rate": 7.61076602897371e-07,
+      "loss": 0.0317,
+      "step": 924
+    },
+    {
+      "epoch": 1.6459074733096086,
+      "grad_norm": 0.40041792419688366,
+      "learning_rate": 7.536816304140177e-07,
+      "loss": 0.0196,
+      "step": 925
+    },
+    {
+      "epoch": 1.6476868327402134,
+      "grad_norm": 0.49310606410969055,
+      "learning_rate": 7.46319830521553e-07,
+      "loss": 0.0226,
+      "step": 926
+    },
+    {
+      "epoch": 1.6494661921708185,
+      "grad_norm": 0.5474552892778177,
+      "learning_rate": 7.389912607309662e-07,
+      "loss": 0.0345,
+      "step": 927
+    },
+    {
+      "epoch": 1.6512455516014235,
+      "grad_norm": 0.5756288894474917,
+      "learning_rate": 7.316959782936516e-07,
+      "loss": 0.0338,
+      "step": 928
+    },
+    {
+      "epoch": 1.6530249110320283,
+      "grad_norm": 0.417011229644052,
+      "learning_rate": 7.244340402009608e-07,
+      "loss": 0.0228,
+      "step": 929
+    },
+    {
+      "epoch": 1.6548042704626336,
+      "grad_norm": 0.7068228875808372,
+      "learning_rate": 7.172055031837572e-07,
+      "loss": 0.0443,
+      "step": 930
+    },
+    {
+      "epoch": 1.6565836298932384,
+      "grad_norm": 0.4802528787546097,
+      "learning_rate": 7.100104237119676e-07,
+      "loss": 0.0253,
+      "step": 931
+    },
+    {
+      "epoch": 1.6583629893238434,
+      "grad_norm": 0.488730442256627,
+      "learning_rate": 7.028488579941506e-07,
+      "loss": 0.035,
+      "step": 932
+    },
+    {
+      "epoch": 1.6601423487544484,
+      "grad_norm": 0.5308489642378459,
+      "learning_rate": 6.957208619770505e-07,
+      "loss": 0.0309,
+      "step": 933
+    },
+    {
+      "epoch": 1.6619217081850532,
+      "grad_norm": 0.5859144193986185,
+      "learning_rate": 6.886264913451635e-07,
+      "loss": 0.0295,
+      "step": 934
+    },
+    {
+      "epoch": 1.6637010676156585,
+      "grad_norm": 0.4889835968066027,
+      "learning_rate": 6.815658015203014e-07,
+      "loss": 0.0244,
+      "step": 935
+    },
+    {
+      "epoch": 1.6654804270462633,
+      "grad_norm": 0.6603384037819555,
+      "learning_rate": 6.745388476611553e-07,
+      "loss": 0.0373,
+      "step": 936
+    },
+    {
+      "epoch": 1.6672597864768683,
+      "grad_norm": 0.5780593035406629,
+      "learning_rate": 6.67545684662873e-07,
+      "loss": 0.0296,
+      "step": 937
+    },
+    {
+      "epoch": 1.6690391459074734,
+      "grad_norm": 0.5099798539099286,
+      "learning_rate": 6.605863671566221e-07,
+      "loss": 0.0261,
+      "step": 938
+    },
+    {
+      "epoch": 1.6708185053380782,
+      "grad_norm": 0.5186562986690287,
+      "learning_rate": 6.536609495091695e-07,
+      "loss": 0.0229,
+      "step": 939
+    },
+    {
+      "epoch": 1.6725978647686834,
+      "grad_norm": 0.6843756380540322,
+      "learning_rate": 6.467694858224488e-07,
+      "loss": 0.0337,
+      "step": 940
+    },
+    {
+      "epoch": 1.6743772241992882,
+      "grad_norm": 0.661303230946439,
+      "learning_rate": 6.399120299331468e-07,
+      "loss": 0.0265,
+      "step": 941
+    },
+    {
+      "epoch": 1.6761565836298933,
+      "grad_norm": 0.5726885563698805,
+      "learning_rate": 6.330886354122768e-07,
+      "loss": 0.0284,
+      "step": 942
+    },
+    {
+      "epoch": 1.6779359430604983,
+      "grad_norm": 0.5243124931556882,
+      "learning_rate": 6.262993555647617e-07,
+      "loss": 0.0349,
+      "step": 943
+    },
+    {
+      "epoch": 1.6797153024911031,
+      "grad_norm": 0.5084663565385805,
+      "learning_rate": 6.1954424342902e-07,
+      "loss": 0.029,
+      "step": 944
+    },
+    {
+      "epoch": 1.6814946619217082,
+      "grad_norm": 0.7657956595822966,
+      "learning_rate": 6.128233517765448e-07,
+      "loss": 0.0422,
+      "step": 945
+    },
+    {
+      "epoch": 1.6832740213523132,
+      "grad_norm": 0.5089808182200171,
+      "learning_rate": 6.061367331114992e-07,
+      "loss": 0.0257,
+      "step": 946
+    },
+    {
+      "epoch": 1.685053380782918,
+      "grad_norm": 0.5674246032773146,
+      "learning_rate": 5.994844396703025e-07,
+      "loss": 0.0381,
+      "step": 947
+    },
+    {
+      "epoch": 1.6868327402135233,
+      "grad_norm": 0.5456484302274213,
+      "learning_rate": 5.928665234212233e-07,
+      "loss": 0.0256,
+      "step": 948
+    },
+    {
+      "epoch": 1.688612099644128,
+      "grad_norm": 0.5409947844064108,
+      "learning_rate": 5.862830360639698e-07,
+      "loss": 0.0291,
+      "step": 949
+    },
+    {
+      "epoch": 1.690391459074733,
+      "grad_norm": 0.6581305811920027,
+      "learning_rate": 5.797340290292907e-07,
+      "loss": 0.0294,
+      "step": 950
+    },
+    {
+      "epoch": 1.6921708185053381,
+      "grad_norm": 0.7748156119582554,
+      "learning_rate": 5.732195534785723e-07,
+      "loss": 0.0407,
+      "step": 951
+    },
+    {
+      "epoch": 1.693950177935943,
+      "grad_norm": 0.4569319304190757,
+      "learning_rate": 5.667396603034369e-07,
+      "loss": 0.0241,
+      "step": 952
+    },
+    {
+      "epoch": 1.6957295373665482,
+      "grad_norm": 0.5986243248537888,
+      "learning_rate": 5.602944001253486e-07,
+      "loss": 0.0321,
+      "step": 953
+    },
+    {
+      "epoch": 1.697508896797153,
+      "grad_norm": 0.5959495489710426,
+      "learning_rate": 5.538838232952104e-07,
+      "loss": 0.037,
+      "step": 954
+    },
+    {
+      "epoch": 1.699288256227758,
+      "grad_norm": 0.7297810878073716,
+      "learning_rate": 5.475079798929816e-07,
+      "loss": 0.0499,
+      "step": 955
+    },
+    {
+      "epoch": 1.701067615658363,
+      "grad_norm": 0.4361515543722615,
+      "learning_rate": 5.411669197272795e-07,
+      "loss": 0.0193,
+      "step": 956
+    },
+    {
+      "epoch": 1.7028469750889679,
+      "grad_norm": 0.45807297555729887,
+      "learning_rate": 5.348606923349903e-07,
+      "loss": 0.024,
+      "step": 957
+    },
+    {
+      "epoch": 1.704626334519573,
+      "grad_norm": 0.6621201864718516,
+      "learning_rate": 5.285893469808855e-07,
+      "loss": 0.031,
+      "step": 958
+    },
+    {
+      "epoch": 1.706405693950178,
+      "grad_norm": 0.531021723938786,
+      "learning_rate": 5.223529326572352e-07,
+      "loss": 0.025,
+      "step": 959
+    },
+    {
+      "epoch": 1.708185053380783,
+      "grad_norm": 0.7569972873953904,
+      "learning_rate": 5.161514980834232e-07,
+      "loss": 0.048,
+      "step": 960
+    },
+    {
+      "epoch": 1.709964412811388,
+      "grad_norm": 0.6779061071175689,
+      "learning_rate": 5.099850917055709e-07,
+      "loss": 0.0348,
+      "step": 961
+    },
+    {
+      "epoch": 1.7117437722419928,
+      "grad_norm": 0.5302304346672033,
+      "learning_rate": 5.038537616961559e-07,
+      "loss": 0.0216,
+      "step": 962
+    },
+    {
+      "epoch": 1.7135231316725978,
+      "grad_norm": 0.546054448969863,
+      "learning_rate": 4.977575559536358e-07,
+      "loss": 0.0246,
+      "step": 963
+    },
+    {
+      "epoch": 1.7153024911032029,
+      "grad_norm": 0.5978800476335613,
+      "learning_rate": 4.916965221020753e-07,
+      "loss": 0.0267,
+      "step": 964
+    },
+    {
+      "epoch": 1.7170818505338077,
+      "grad_norm": 0.5052882213023883,
+      "learning_rate": 4.856707074907729e-07,
+      "loss": 0.0258,
+      "step": 965
+    },
+    {
+      "epoch": 1.718861209964413,
+      "grad_norm": 0.7863194108649,
+      "learning_rate": 4.796801591938922e-07,
+      "loss": 0.0401,
+      "step": 966
+    },
+    {
+      "epoch": 1.7206405693950177,
+      "grad_norm": 0.5475310043033773,
+      "learning_rate": 4.737249240100911e-07,
+      "loss": 0.0312,
+      "step": 967
+    },
+    {
+      "epoch": 1.7224199288256228,
+      "grad_norm": 0.4917581029969849,
+      "learning_rate": 4.6780504846216155e-07,
+      "loss": 0.0219,
+      "step": 968
+    },
+    {
+      "epoch": 1.7241992882562278,
+      "grad_norm": 0.5819690046514756,
+      "learning_rate": 4.619205787966613e-07,
+      "loss": 0.0303,
+      "step": 969
+    },
+    {
+      "epoch": 1.7259786476868326,
+      "grad_norm": 0.5065038876137672,
+      "learning_rate": 4.560715609835548e-07,
+      "loss": 0.0265,
+      "step": 970
+    },
+    {
+      "epoch": 1.7277580071174379,
+      "grad_norm": 0.55910683680331,
+      "learning_rate": 4.5025804071585464e-07,
+      "loss": 0.027,
+      "step": 971
+    },
+    {
+      "epoch": 1.7295373665480427,
+      "grad_norm": 0.5702523556265029,
+      "learning_rate": 4.4448006340926163e-07,
+      "loss": 0.0321,
+      "step": 972
+    },
+    {
+      "epoch": 1.7313167259786477,
+      "grad_norm": 0.5803756285463108,
+      "learning_rate": 4.3873767420181344e-07,
+      "loss": 0.0302,
+      "step": 973
+    },
+    {
+      "epoch": 1.7330960854092528,
+      "grad_norm": 0.6676158022227648,
+      "learning_rate": 4.3303091795353024e-07,
+      "loss": 0.0406,
+      "step": 974
+    },
+    {
+      "epoch": 1.7348754448398576,
+      "grad_norm": 0.5781104433601609,
+      "learning_rate": 4.2735983924606596e-07,
+      "loss": 0.0271,
+      "step": 975
+    },
+    {
+      "epoch": 1.7366548042704626,
+      "grad_norm": 0.5676289003437387,
+      "learning_rate": 4.2172448238235464e-07,
+      "loss": 0.0243,
+      "step": 976
+    },
+    {
+      "epoch": 1.7384341637010676,
+      "grad_norm": 0.5421582987099706,
+      "learning_rate": 4.161248913862731e-07,
+      "loss": 0.0222,
+      "step": 977
+    },
+    {
+      "epoch": 1.7402135231316724,
+      "grad_norm": 0.5274379744897273,
+      "learning_rate": 4.1056111000228937e-07,
+      "loss": 0.028,
+      "step": 978
+    },
+    {
+      "epoch": 1.7419928825622777,
+      "grad_norm": 0.5383743255720211,
+      "learning_rate": 4.0503318169512417e-07,
+      "loss": 0.0238,
+      "step": 979
+    },
+    {
+      "epoch": 1.7437722419928825,
+      "grad_norm": 0.4568917183113237,
+      "learning_rate": 3.9954114964941336e-07,
+      "loss": 0.0238,
+      "step": 980
+    },
+    {
+      "epoch": 1.7455516014234875,
+      "grad_norm": 0.5983213611490273,
+      "learning_rate": 3.9408505676936327e-07,
+      "loss": 0.0309,
+      "step": 981
+    },
+    {
+      "epoch": 1.7473309608540926,
+      "grad_norm": 0.7062525101639744,
+      "learning_rate": 3.886649456784253e-07,
+      "loss": 0.0395,
+      "step": 982
+    },
+    {
+      "epoch": 1.7491103202846974,
+      "grad_norm": 0.5210853129335417,
+      "learning_rate": 3.8328085871895624e-07,
+      "loss": 0.0328,
+      "step": 983
+    },
+    {
+      "epoch": 1.7508896797153026,
+      "grad_norm": 0.5419591978052065,
+      "learning_rate": 3.779328379518898e-07,
+      "loss": 0.0273,
+      "step": 984
+    },
+    {
+      "epoch": 1.7526690391459074,
+      "grad_norm": 0.5723908258672803,
+      "learning_rate": 3.7262092515640556e-07,
+      "loss": 0.0279,
+      "step": 985
+    },
+    {
+      "epoch": 1.7544483985765125,
+      "grad_norm": 0.4485465693900342,
+      "learning_rate": 3.673451618296081e-07,
+      "loss": 0.0212,
+      "step": 986
+    },
+    {
+      "epoch": 1.7562277580071175,
+      "grad_norm": 0.6408565267179657,
+      "learning_rate": 3.621055891861963e-07,
+      "loss": 0.0352,
+      "step": 987
+    },
+    {
+      "epoch": 1.7580071174377223,
+      "grad_norm": 0.532814477134847,
+      "learning_rate": 3.56902248158148e-07,
+      "loss": 0.0228,
+      "step": 988
+    },
+    {
+      "epoch": 1.7597864768683276,
+      "grad_norm": 0.535832717235761,
+      "learning_rate": 3.517351793943913e-07,
+      "loss": 0.0232,
+      "step": 989
+    },
+    {
+      "epoch": 1.7615658362989324,
+      "grad_norm": 0.666958902596261,
+      "learning_rate": 3.4660442326049704e-07,
+      "loss": 0.027,
+      "step": 990
+    },
+    {
+      "epoch": 1.7633451957295374,
+      "grad_norm": 0.5287144789540672,
+      "learning_rate": 3.4151001983835696e-07,
+      "loss": 0.0342,
+      "step": 991
+    },
+    {
+      "epoch": 1.7651245551601424,
+      "grad_norm": 0.5031541876244054,
+      "learning_rate": 3.364520089258727e-07,
+      "loss": 0.023,
+      "step": 992
+    },
+    {
+      "epoch": 1.7669039145907472,
+      "grad_norm": 0.6407730451235067,
+      "learning_rate": 3.314304300366461e-07,
+      "loss": 0.0318,
+      "step": 993
+    },
+    {
+      "epoch": 1.7686832740213523,
+      "grad_norm": 0.5391324129593671,
+      "learning_rate": 3.2644532239966444e-07,
+      "loss": 0.031,
+      "step": 994
+    },
+    {
+      "epoch": 1.7704626334519573,
+      "grad_norm": 0.43139291301576776,
+      "learning_rate": 3.2149672495900286e-07,
+      "loss": 0.0254,
+      "step": 995
+    },
+    {
+      "epoch": 1.7722419928825621,
+      "grad_norm": 0.6076928449490513,
+      "learning_rate": 3.165846763735153e-07,
+      "loss": 0.0345,
+      "step": 996
+    },
+    {
+      "epoch": 1.7740213523131674,
+      "grad_norm": 0.516757188718613,
+      "learning_rate": 3.117092150165324e-07,
+      "loss": 0.0282,
+      "step": 997
+    },
+    {
+      "epoch": 1.7758007117437722,
+      "grad_norm": 0.5441207000245784,
+      "learning_rate": 3.068703789755606e-07,
+      "loss": 0.0266,
+      "step": 998
+    },
+    {
+      "epoch": 1.7775800711743772,
+      "grad_norm": 0.6430652925224465,
+      "learning_rate": 3.020682060519886e-07,
+      "loss": 0.0343,
+      "step": 999
+    },
+    {
+      "epoch": 1.7793594306049823,
+      "grad_norm": 0.7274642284317804,
+      "learning_rate": 2.9730273376078923e-07,
+      "loss": 0.0303,
+      "step": 1000
+    },
+    {
+      "epoch": 1.7793594306049823,
+      "eval_loss": 0.10914742201566696,
+      "eval_runtime": 1.9077,
+      "eval_samples_per_second": 24.113,
+      "eval_steps_per_second": 6.29,
+      "step": 1000
+    },
+    {
+      "epoch": 1.781138790035587,
+      "grad_norm": 0.6267070777299288,
+      "learning_rate": 2.9257399933022737e-07,
+      "loss": 0.029,
+      "step": 1001
+    },
+    {
+      "epoch": 1.7829181494661923,
+      "grad_norm": 0.5375800784905966,
+      "learning_rate": 2.8788203970156805e-07,
+      "loss": 0.0286,
+      "step": 1002
+    },
+    {
+      "epoch": 1.7846975088967971,
+      "grad_norm": 0.5597691647106762,
+      "learning_rate": 2.832268915287878e-07,
+      "loss": 0.0358,
+      "step": 1003
+    },
+    {
+      "epoch": 1.7864768683274022,
+      "grad_norm": 0.6702740841316541,
+      "learning_rate": 2.7860859117828985e-07,
+      "loss": 0.0383,
+      "step": 1004
+    },
+    {
+      "epoch": 1.7882562277580072,
+      "grad_norm": 0.8530198517538259,
+      "learning_rate": 2.740271747286194e-07,
+      "loss": 0.059,
+      "step": 1005
+    },
+    {
+      "epoch": 1.790035587188612,
+      "grad_norm": 0.6544945027088211,
+      "learning_rate": 2.6948267797018145e-07,
+      "loss": 0.0323,
+      "step": 1006
+    },
+    {
+      "epoch": 1.791814946619217,
+      "grad_norm": 0.40496543442602423,
+      "learning_rate": 2.649751364049613e-07,
+      "loss": 0.014,
+      "step": 1007
+    },
+    {
+      "epoch": 1.793594306049822,
+      "grad_norm": 0.5125664512440179,
+      "learning_rate": 2.6050458524624735e-07,
+      "loss": 0.0197,
+      "step": 1008
+    },
+    {
+      "epoch": 1.795373665480427,
+      "grad_norm": 0.6033856074152585,
+      "learning_rate": 2.560710594183552e-07,
+      "loss": 0.0293,
+      "step": 1009
+    },
+    {
+      "epoch": 1.7971530249110321,
+      "grad_norm": 0.5423512214621586,
+      "learning_rate": 2.5167459355635524e-07,
+      "loss": 0.0304,
+      "step": 1010
+    },
+    {
+      "epoch": 1.798932384341637,
+      "grad_norm": 0.5884913724630706,
+      "learning_rate": 2.473152220058039e-07,
+      "loss": 0.0285,
+      "step": 1011
+    },
+    {
+      "epoch": 1.800711743772242,
+      "grad_norm": 0.6409067029649802,
+      "learning_rate": 2.429929788224722e-07,
+      "loss": 0.0337,
+      "step": 1012
+    },
+    {
+      "epoch": 1.802491103202847,
+      "grad_norm": 0.5442336729867471,
+      "learning_rate": 2.38707897772083e-07,
+      "loss": 0.0328,
+      "step": 1013
+    },
+    {
+      "epoch": 1.8042704626334518,
+      "grad_norm": 0.6423724382221596,
+      "learning_rate": 2.3446001233004333e-07,
+      "loss": 0.0494,
+      "step": 1014
+    },
+    {
+      "epoch": 1.806049822064057,
+      "grad_norm": 0.6283064159506893,
+      "learning_rate": 2.3024935568118745e-07,
+      "loss": 0.0306,
+      "step": 1015
+    },
+    {
+      "epoch": 1.8078291814946619,
+      "grad_norm": 0.49475791702995275,
+      "learning_rate": 2.2607596071951288e-07,
+      "loss": 0.0236,
+      "step": 1016
+    },
+    {
+      "epoch": 1.809608540925267,
+      "grad_norm": 0.5275808845123406,
+      "learning_rate": 2.2193986004792667e-07,
+      "loss": 0.0336,
+      "step": 1017
+    },
+    {
+      "epoch": 1.811387900355872,
+      "grad_norm": 0.4908832663374769,
+      "learning_rate": 2.1784108597799058e-07,
+      "loss": 0.0197,
+      "step": 1018
+    },
+    {
+      "epoch": 1.8131672597864767,
+      "grad_norm": 0.5227427635196124,
+      "learning_rate": 2.1377967052966685e-07,
+      "loss": 0.0308,
+      "step": 1019
+    },
+    {
+      "epoch": 1.814946619217082,
+      "grad_norm": 0.5400601247782824,
+      "learning_rate": 2.0975564543107007e-07,
+      "loss": 0.0303,
+      "step": 1020
+    },
+    {
+      "epoch": 1.8167259786476868,
+      "grad_norm": 0.481658814241904,
+      "learning_rate": 2.057690421182168e-07,
+      "loss": 0.0226,
+      "step": 1021
+    },
+    {
+      "epoch": 1.8185053380782918,
+      "grad_norm": 0.5128321430461603,
+      "learning_rate": 2.01819891734783e-07,
+      "loss": 0.0277,
+      "step": 1022
+    },
+    {
+      "epoch": 1.8202846975088969,
+      "grad_norm": 0.49221138183757146,
+      "learning_rate": 1.979082251318576e-07,
+      "loss": 0.0229,
+      "step": 1023
+    },
+    {
+      "epoch": 1.8220640569395017,
+      "grad_norm": 0.46524746215858104,
+      "learning_rate": 1.9403407286770592e-07,
+      "loss": 0.0253,
+      "step": 1024
+    },
+    {
+      "epoch": 1.8238434163701067,
+      "grad_norm": 0.4743767165702309,
+      "learning_rate": 1.9019746520752502e-07,
+      "loss": 0.0258,
+      "step": 1025
+    },
+    {
+      "epoch": 1.8256227758007118,
+      "grad_norm": 0.4232238911293423,
+      "learning_rate": 1.8639843212321206e-07,
+      "loss": 0.0208,
+      "step": 1026
+    },
+    {
+      "epoch": 1.8274021352313166,
+      "grad_norm": 0.6570919069692968,
+      "learning_rate": 1.826370032931285e-07,
+      "loss": 0.0329,
+      "step": 1027
+    },
+    {
+      "epoch": 1.8291814946619218,
+      "grad_norm": 0.4582249757293188,
+      "learning_rate": 1.789132081018674e-07,
+      "loss": 0.0199,
+      "step": 1028
+    },
+    {
+      "epoch": 1.8309608540925266,
+      "grad_norm": 0.5558054222081665,
+      "learning_rate": 1.7522707564002706e-07,
+      "loss": 0.0266,
+      "step": 1029
+    },
+    {
+      "epoch": 1.8327402135231317,
+      "grad_norm": 0.5366499403380625,
+      "learning_rate": 1.7157863470397718e-07,
+      "loss": 0.0248,
+      "step": 1030
+    },
+    {
+      "epoch": 1.8345195729537367,
+      "grad_norm": 0.5779785294808715,
+      "learning_rate": 1.6796791379564138e-07,
+      "loss": 0.0365,
+      "step": 1031
+    },
+    {
+      "epoch": 1.8362989323843415,
+      "grad_norm": 0.5361471348639685,
+      "learning_rate": 1.6439494112227173e-07,
+      "loss": 0.0187,
+      "step": 1032
+    },
+    {
+      "epoch": 1.8380782918149468,
+      "grad_norm": 0.5498921017791115,
+      "learning_rate": 1.6085974459622567e-07,
+      "loss": 0.0264,
+      "step": 1033
+    },
+    {
+      "epoch": 1.8398576512455516,
+      "grad_norm": 0.5478384989237817,
+      "learning_rate": 1.573623518347517e-07,
+      "loss": 0.0255,
+      "step": 1034
+    },
+    {
+      "epoch": 1.8416370106761566,
+      "grad_norm": 0.6256643434892188,
+      "learning_rate": 1.5390279015977117e-07,
+      "loss": 0.0281,
+      "step": 1035
+    },
+    {
+      "epoch": 1.8434163701067616,
+      "grad_norm": 0.4106996962512219,
+      "learning_rate": 1.5048108659766693e-07,
+      "loss": 0.0204,
+      "step": 1036
+    },
+    {
+      "epoch": 1.8451957295373664,
+      "grad_norm": 0.4660681817315904,
+      "learning_rate": 1.470972678790711e-07,
+      "loss": 0.0296,
+      "step": 1037
+    },
+    {
+      "epoch": 1.8469750889679717,
+      "grad_norm": 0.6241004270673808,
+      "learning_rate": 1.437513604386559e-07,
+      "loss": 0.0359,
+      "step": 1038
+    },
+    {
+      "epoch": 1.8487544483985765,
+      "grad_norm": 0.6265079011562581,
+      "learning_rate": 1.404433904149266e-07,
+      "loss": 0.032,
+      "step": 1039
+    },
+    {
+      "epoch": 1.8505338078291815,
+      "grad_norm": 0.605027449857589,
+      "learning_rate": 1.3717338365001943e-07,
+      "loss": 0.0243,
+      "step": 1040
+    },
+    {
+      "epoch": 1.8523131672597866,
+      "grad_norm": 0.5608926286421247,
+      "learning_rate": 1.3394136568949834e-07,
+      "loss": 0.0379,
+      "step": 1041
+    },
+    {
+      "epoch": 1.8540925266903914,
+      "grad_norm": 0.6590296201749801,
+      "learning_rate": 1.307473617821553e-07,
+      "loss": 0.0331,
+      "step": 1042
+    },
+    {
+      "epoch": 1.8558718861209964,
+      "grad_norm": 0.6953750895461714,
+      "learning_rate": 1.275913968798137e-07,
+      "loss": 0.028,
+      "step": 1043
+    },
+    {
+      "epoch": 1.8576512455516014,
+      "grad_norm": 0.6120255716451495,
+      "learning_rate": 1.2447349563713186e-07,
+      "loss": 0.0335,
+      "step": 1044
+    },
+    {
+      "epoch": 1.8594306049822062,
+      "grad_norm": 0.5337868131654809,
+      "learning_rate": 1.213936824114137e-07,
+      "loss": 0.0294,
+      "step": 1045
+    },
+    {
+      "epoch": 1.8612099644128115,
+      "grad_norm": 0.4986554527093418,
+      "learning_rate": 1.1835198126241509e-07,
+      "loss": 0.024,
+      "step": 1046
+    },
+    {
+      "epoch": 1.8629893238434163,
+      "grad_norm": 0.5932396782159736,
+      "learning_rate": 1.1534841595215617e-07,
+      "loss": 0.0338,
+      "step": 1047
+    },
+    {
+      "epoch": 1.8647686832740213,
+      "grad_norm": 0.47192930725594395,
+      "learning_rate": 1.1238300994473983e-07,
+      "loss": 0.0181,
+      "step": 1048
+    },
+    {
+      "epoch": 1.8665480427046264,
+      "grad_norm": 0.527213594505284,
+      "learning_rate": 1.0945578640616183e-07,
+      "loss": 0.0374,
+      "step": 1049
+    },
+    {
+      "epoch": 1.8683274021352312,
+      "grad_norm": 0.6230131238045177,
+      "learning_rate": 1.0656676820413603e-07,
+      "loss": 0.0203,
+      "step": 1050
+    },
+    {
+      "epoch": 1.8701067615658364,
+      "grad_norm": 0.5079900354844558,
+      "learning_rate": 1.0371597790791166e-07,
+      "loss": 0.0288,
+      "step": 1051
+    },
+    {
+      "epoch": 1.8718861209964412,
+      "grad_norm": 0.5648855496273179,
+      "learning_rate": 1.0090343778809908e-07,
+      "loss": 0.0245,
+      "step": 1052
+    },
+    {
+      "epoch": 1.8736654804270463,
+      "grad_norm": 0.8154245401960067,
+      "learning_rate": 9.812916981649433e-08,
+      "loss": 0.0442,
+      "step": 1053
+    },
+    {
+      "epoch": 1.8754448398576513,
+      "grad_norm": 0.5209021174339272,
+      "learning_rate": 9.539319566590766e-08,
+      "loss": 0.0333,
+      "step": 1054
+    },
+    {
+      "epoch": 1.8772241992882561,
+      "grad_norm": 0.5541761197201868,
+      "learning_rate": 9.269553670999743e-08,
+      "loss": 0.0327,
+      "step": 1055
+    },
+    {
+      "epoch": 1.8790035587188612,
+      "grad_norm": 0.870759391971751,
+      "learning_rate": 9.003621402309815e-08,
+      "loss": 0.0355,
+      "step": 1056
+    },
+    {
+      "epoch": 1.8807829181494662,
+      "grad_norm": 0.562100318264173,
+      "learning_rate": 8.741524838005888e-08,
+      "loss": 0.0258,
+      "step": 1057
+    },
+    {
+      "epoch": 1.8825622775800712,
+      "grad_norm": 0.6081273178657819,
+      "learning_rate": 8.483266025608061e-08,
+      "loss": 0.0257,
+      "step": 1058
+    },
+    {
+      "epoch": 1.8843416370106763,
+      "grad_norm": 0.49278719954195993,
+      "learning_rate": 8.228846982655525e-08,
+      "loss": 0.0256,
+      "step": 1059
+    },
+    {
+      "epoch": 1.886120996441281,
+      "grad_norm": 0.5481935103430626,
+      "learning_rate": 7.978269696691021e-08,
+      "loss": 0.0363,
+      "step": 1060
+    },
+    {
+      "epoch": 1.887900355871886,
+      "grad_norm": 0.5651959530930333,
+      "learning_rate": 7.731536125244965e-08,
+      "loss": 0.0297,
+      "step": 1061
+    },
+    {
+      "epoch": 1.8896797153024911,
+      "grad_norm": 0.6087128339997628,
+      "learning_rate": 7.488648195820513e-08,
+      "loss": 0.0394,
+      "step": 1062
+    },
+    {
+      "epoch": 1.891459074733096,
+      "grad_norm": 0.6200395551927195,
+      "learning_rate": 7.249607805878245e-08,
+      "loss": 0.0291,
+      "step": 1063
+    },
+    {
+      "epoch": 1.8932384341637012,
+      "grad_norm": 0.5568364964365429,
+      "learning_rate": 7.014416822821557e-08,
+      "loss": 0.0318,
+      "step": 1064
+    },
+    {
+      "epoch": 1.895017793594306,
+      "grad_norm": 0.47850363736380264,
+      "learning_rate": 6.783077083981793e-08,
+      "loss": 0.0192,
+      "step": 1065
+    },
+    {
+      "epoch": 1.896797153024911,
+      "grad_norm": 0.5568868037104718,
+      "learning_rate": 6.55559039660425e-08,
+      "loss": 0.031,
+      "step": 1066
+    },
+    {
+      "epoch": 1.898576512455516,
+      "grad_norm": 0.5856082372692212,
+      "learning_rate": 6.331958537833693e-08,
+      "loss": 0.0267,
+      "step": 1067
+    },
+    {
+      "epoch": 1.9003558718861209,
+      "grad_norm": 0.5763438169679147,
+      "learning_rate": 6.112183254700866e-08,
+      "loss": 0.0306,
+      "step": 1068
+    },
+    {
+      "epoch": 1.9021352313167261,
+      "grad_norm": 0.4385864617055261,
+      "learning_rate": 5.8962662641083856e-08,
+      "loss": 0.0232,
+      "step": 1069
+    },
+    {
+      "epoch": 1.903914590747331,
+      "grad_norm": 0.5918581330216593,
+      "learning_rate": 5.6842092528176516e-08,
+      "loss": 0.0286,
+      "step": 1070
+    },
+    {
+      "epoch": 1.905693950177936,
+      "grad_norm": 0.5097272028457035,
+      "learning_rate": 5.476013877435626e-08,
+      "loss": 0.0282,
+      "step": 1071
+    },
+    {
+      "epoch": 1.907473309608541,
+      "grad_norm": 0.5361282213378592,
+      "learning_rate": 5.271681764401848e-08,
+      "loss": 0.0306,
+      "step": 1072
+    },
+    {
+      "epoch": 1.9092526690391458,
+      "grad_norm": 0.4563645831543143,
+      "learning_rate": 5.071214509975775e-08,
+      "loss": 0.0237,
+      "step": 1073
+    },
+    {
+      "epoch": 1.9110320284697508,
+      "grad_norm": 0.6400703462597346,
+      "learning_rate": 4.8746136802240716e-08,
+      "loss": 0.0338,
+      "step": 1074
+    },
+    {
+      "epoch": 1.9128113879003559,
+      "grad_norm": 0.5837169104700998,
+      "learning_rate": 4.6818808110087875e-08,
+      "loss": 0.0341,
+      "step": 1075
+    },
+    {
+      "epoch": 1.914590747330961,
+      "grad_norm": 0.6383099107452931,
+      "learning_rate": 4.493017407975087e-08,
+      "loss": 0.0379,
+      "step": 1076
+    },
+    {
+      "epoch": 1.916370106761566,
+      "grad_norm": 0.6540327209501722,
+      "learning_rate": 4.308024946539424e-08,
+      "loss": 0.0288,
+      "step": 1077
+    },
+    {
+      "epoch": 1.9181494661921707,
+      "grad_norm": 0.543093782634597,
+      "learning_rate": 4.1269048718783344e-08,
+      "loss": 0.0251,
+      "step": 1078
+    },
+    {
+      "epoch": 1.9199288256227758,
+      "grad_norm": 0.5525641221326526,
+      "learning_rate": 3.9496585989167726e-08,
+      "loss": 0.0272,
+      "step": 1079
+    },
+    {
+      "epoch": 1.9217081850533808,
+      "grad_norm": 0.5694370077498606,
+      "learning_rate": 3.776287512317345e-08,
+      "loss": 0.0313,
+      "step": 1080
+    },
+    {
+      "epoch": 1.9234875444839856,
+      "grad_norm": 0.5584974726124532,
+      "learning_rate": 3.606792966469375e-08,
+      "loss": 0.0293,
+      "step": 1081
+    },
+    {
+      "epoch": 1.9252669039145909,
+      "grad_norm": 0.5371647833967312,
+      "learning_rate": 3.4411762854782426e-08,
+      "loss": 0.0261,
+      "step": 1082
+    },
+    {
+      "epoch": 1.9270462633451957,
+      "grad_norm": 0.5114294226885525,
+      "learning_rate": 3.279438763155174e-08,
+      "loss": 0.0182,
+      "step": 1083
+    },
+    {
+      "epoch": 1.9288256227758007,
+      "grad_norm": 0.6128606060124004,
+      "learning_rate": 3.121581663007134e-08,
+      "loss": 0.0338,
+      "step": 1084
+    },
+    {
+      "epoch": 1.9306049822064058,
+      "grad_norm": 0.4791490079343616,
+      "learning_rate": 2.967606218226837e-08,
+      "loss": 0.0314,
+      "step": 1085
+    },
+    {
+      "epoch": 1.9323843416370106,
+      "grad_norm": 0.44347976185089155,
+      "learning_rate": 2.8175136316832e-08,
+      "loss": 0.0198,
+      "step": 1086
+    },
+    {
+      "epoch": 1.9341637010676158,
+      "grad_norm": 0.7278884007141428,
+      "learning_rate": 2.6713050759120117e-08,
+      "loss": 0.0552,
+      "step": 1087
+    },
+    {
+      "epoch": 1.9359430604982206,
+      "grad_norm": 0.437966917276108,
+      "learning_rate": 2.528981693106558e-08,
+      "loss": 0.0231,
+      "step": 1088
+    },
+    {
+      "epoch": 1.9377224199288257,
+      "grad_norm": 0.5402238331284621,
+      "learning_rate": 2.3905445951089013e-08,
+      "loss": 0.03,
+      "step": 1089
+    },
+    {
+      "epoch": 1.9395017793594307,
+      "grad_norm": 0.5679480641764137,
+      "learning_rate": 2.2559948634011673e-08,
+      "loss": 0.0261,
+      "step": 1090
+    },
+    {
+      "epoch": 1.9412811387900355,
+      "grad_norm": 0.5610437276588442,
+      "learning_rate": 2.125333549096942e-08,
+      "loss": 0.0287,
+      "step": 1091
+    },
+    {
+      "epoch": 1.9430604982206405,
+      "grad_norm": 0.48838186084587387,
+      "learning_rate": 1.9985616729332747e-08,
+      "loss": 0.0269,
+      "step": 1092
+    },
+    {
+      "epoch": 1.9448398576512456,
+      "grad_norm": 0.5359273947813544,
+      "learning_rate": 1.8756802252625773e-08,
+      "loss": 0.0314,
+      "step": 1093
+    },
+    {
+      "epoch": 1.9466192170818504,
+      "grad_norm": 0.4836531070849993,
+      "learning_rate": 1.75669016604485e-08,
+      "loss": 0.0236,
+      "step": 1094
+    },
+    {
+      "epoch": 1.9483985765124556,
+      "grad_norm": 0.7679305729893142,
+      "learning_rate": 1.6415924248403547e-08,
+      "loss": 0.0236,
+      "step": 1095
+    },
+    {
+      "epoch": 1.9501779359430604,
+      "grad_norm": 0.5262346670378306,
+      "learning_rate": 1.5303879008021773e-08,
+      "loss": 0.029,
+      "step": 1096
+    },
+    {
+      "epoch": 1.9519572953736655,
+      "grad_norm": 0.5331124425550651,
+      "learning_rate": 1.4230774626691756e-08,
+      "loss": 0.0241,
+      "step": 1097
+    },
+    {
+      "epoch": 1.9537366548042705,
+      "grad_norm": 0.5360098553396363,
+      "learning_rate": 1.3196619487594875e-08,
+      "loss": 0.0302,
+      "step": 1098
+    },
+    {
+      "epoch": 1.9555160142348753,
+      "grad_norm": 0.7136334745765697,
+      "learning_rate": 1.2201421669636448e-08,
+      "loss": 0.0414,
+      "step": 1099
+    },
+    {
+      "epoch": 1.9572953736654806,
+      "grad_norm": 0.5473115383759471,
+      "learning_rate": 1.1245188947384133e-08,
+      "loss": 0.029,
+      "step": 1100
+    },
+    {
+      "epoch": 1.9590747330960854,
+      "grad_norm": 0.5662935971783153,
+      "learning_rate": 1.0327928791006858e-08,
+      "loss": 0.0297,
+      "step": 1101
+    },
+    {
+      "epoch": 1.9608540925266904,
+      "grad_norm": 0.4929988429991171,
+      "learning_rate": 9.449648366217645e-09,
+      "loss": 0.0304,
+      "step": 1102
+    },
+    {
+      "epoch": 1.9626334519572954,
+      "grad_norm": 0.5261686689921419,
+      "learning_rate": 8.61035453421588e-09,
+      "loss": 0.0269,
+      "step": 1103
+    },
+    {
+      "epoch": 1.9644128113879002,
+      "grad_norm": 0.5064395250046319,
+      "learning_rate": 7.81005385163458e-09,
+      "loss": 0.0309,
+      "step": 1104
+    },
+    {
+      "epoch": 1.9661921708185055,
+      "grad_norm": 0.4401106683095485,
+      "learning_rate": 7.048752570488205e-09,
+      "loss": 0.0308,
+      "step": 1105
+    },
+    {
+      "epoch": 1.9679715302491103,
+      "grad_norm": 0.6830942479701549,
+      "learning_rate": 6.326456638125478e-09,
+      "loss": 0.0395,
+      "step": 1106
+    },
+    {
+      "epoch": 1.9697508896797153,
+      "grad_norm": 0.614250780649773,
+      "learning_rate": 5.643171697183314e-09,
+      "loss": 0.0282,
+      "step": 1107
+    },
+    {
+      "epoch": 1.9715302491103204,
+      "grad_norm": 0.6627976370297659,
+      "learning_rate": 4.998903085539075e-09,
+      "loss": 0.038,
+      "step": 1108
+    },
+    {
+      "epoch": 1.9733096085409252,
+      "grad_norm": 0.44609610959807516,
+      "learning_rate": 4.393655836272825e-09,
+      "loss": 0.0185,
+      "step": 1109
+    },
+    {
+      "epoch": 1.9750889679715302,
+      "grad_norm": 0.6989476336260286,
+      "learning_rate": 3.8274346776262514e-09,
+      "loss": 0.0368,
+      "step": 1110
+    },
+    {
+      "epoch": 1.9768683274021353,
+      "grad_norm": 0.6019554822530829,
+      "learning_rate": 3.300244032966582e-09,
+      "loss": 0.0326,
+      "step": 1111
+    },
+    {
+      "epoch": 1.97864768683274,
+      "grad_norm": 0.5659339819867732,
+      "learning_rate": 2.8120880207493928e-09,
+      "loss": 0.0236,
+      "step": 1112
+    },
+    {
+      "epoch": 1.9804270462633453,
+      "grad_norm": 0.5138329513422308,
+      "learning_rate": 2.362970454491409e-09,
+      "loss": 0.0233,
+      "step": 1113
+    },
+    {
+      "epoch": 1.9822064056939501,
+      "grad_norm": 0.540048076423124,
+      "learning_rate": 1.952894842735531e-09,
+      "loss": 0.0329,
+      "step": 1114
+    },
+    {
+      "epoch": 1.9839857651245552,
+      "grad_norm": 0.39364613991722036,
+      "learning_rate": 1.5818643890258555e-09,
+      "loss": 0.0231,
+      "step": 1115
+    },
+    {
+      "epoch": 1.9857651245551602,
+      "grad_norm": 0.7366129499575088,
+      "learning_rate": 1.2498819918843609e-09,
+      "loss": 0.0311,
+      "step": 1116
+    },
+    {
+      "epoch": 1.987544483985765,
+      "grad_norm": 0.7062862262912368,
+      "learning_rate": 9.569502447837053e-10,
+      "loss": 0.0301,
+      "step": 1117
+    },
+    {
+      "epoch": 1.9893238434163703,
+      "grad_norm": 0.5928477386244514,
+      "learning_rate": 7.03071436131686e-10,
+      "loss": 0.0294,
+      "step": 1118
+    },
+    {
+      "epoch": 1.991103202846975,
+      "grad_norm": 0.5457497613600347,
+      "learning_rate": 4.882475492506977e-10,
+      "loss": 0.03,
+      "step": 1119
+    },
+    {
+      "epoch": 1.99288256227758,
+      "grad_norm": 0.5115553471236458,
+      "learning_rate": 3.124802623627465e-10,
+      "loss": 0.0244,
+      "step": 1120
+    },
+    {
+      "epoch": 1.9946619217081851,
+      "grad_norm": 0.5894693561388302,
+      "learning_rate": 1.7577094857557097e-10,
+      "loss": 0.042,
+      "step": 1121
+    },
+    {
+      "epoch": 1.99644128113879,
+      "grad_norm": 0.6432251727424743,
+      "learning_rate": 7.812067587487093e-11,
+      "loss": 0.0361,
+      "step": 1122
+    },
+    {
+      "epoch": 1.998220640569395,
+      "grad_norm": 0.48465548838581035,
+      "learning_rate": 1.9530207111539967e-11,
+      "loss": 0.0236,
+      "step": 1123
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.43288703756228897,
+      "learning_rate": 0.0,
+      "loss": 0.0255,
+      "step": 1124
+    },
+    {
+      "epoch": 2.0,
+      "step": 1124,
+      "total_flos": 4998105219072.0,
+      "train_loss": 0.05959913034151139,
+      "train_runtime": 860.8955,
+      "train_samples_per_second": 10.438,
+      "train_steps_per_second": 1.306
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1124,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4998105219072.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}