{
  "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": 1.713125243161643,
      "learning_rate": 9.99998046979289e-06,
      "loss": 0.1091,
      "step": 1
    },
    {
      "epoch": 0.0035587188612099642,
      "grad_norm": 1.6302122706369235,
      "learning_rate": 9.999921879324127e-06,
      "loss": 0.105,
      "step": 2
    },
    {
      "epoch": 0.005338078291814947,
      "grad_norm": 1.809710022557897,
      "learning_rate": 9.999824229051425e-06,
      "loss": 0.1137,
      "step": 3
    },
    {
      "epoch": 0.0071174377224199285,
      "grad_norm": 2.202959405561451,
      "learning_rate": 9.999687519737639e-06,
      "loss": 0.1366,
      "step": 4
    },
    {
      "epoch": 0.008896797153024912,
      "grad_norm": 2.56865774769625,
      "learning_rate": 9.99951175245075e-06,
      "loss": 0.0998,
      "step": 5
    },
    {
      "epoch": 0.010676156583629894,
      "grad_norm": 2.1461290592083517,
      "learning_rate": 9.999296928563868e-06,
      "loss": 0.1289,
      "step": 6
    },
    {
      "epoch": 0.012455516014234875,
      "grad_norm": 2.5711621281004087,
      "learning_rate": 9.999043049755216e-06,
      "loss": 0.1266,
      "step": 7
    },
    {
      "epoch": 0.014234875444839857,
      "grad_norm": 1.7981967241630354,
      "learning_rate": 9.998750118008117e-06,
      "loss": 0.0809,
      "step": 8
    },
    {
      "epoch": 0.01601423487544484,
      "grad_norm": 1.973917848730113,
      "learning_rate": 9.998418135610974e-06,
      "loss": 0.0821,
      "step": 9
    },
    {
      "epoch": 0.017793594306049824,
      "grad_norm": 2.9625604235753817,
      "learning_rate": 9.998047105157265e-06,
      "loss": 0.1591,
      "step": 10
    },
    {
      "epoch": 0.019572953736654804,
      "grad_norm": 1.9602621665143425,
      "learning_rate": 9.997637029545509e-06,
      "loss": 0.0911,
      "step": 11
    },
    {
      "epoch": 0.021352313167259787,
      "grad_norm": 2.455283390741342,
      "learning_rate": 9.997187911979252e-06,
      "loss": 0.1191,
      "step": 12
    },
    {
      "epoch": 0.023131672597864767,
      "grad_norm": 2.9291596500061425,
      "learning_rate": 9.996699755967035e-06,
      "loss": 0.1527,
      "step": 13
    },
    {
      "epoch": 0.02491103202846975,
      "grad_norm": 2.3788876217664625,
      "learning_rate": 9.996172565322375e-06,
      "loss": 0.1147,
      "step": 14
    },
    {
      "epoch": 0.026690391459074734,
      "grad_norm": 1.9511478291744484,
      "learning_rate": 9.995606344163728e-06,
      "loss": 0.1019,
      "step": 15
    },
    {
      "epoch": 0.028469750889679714,
      "grad_norm": 1.8633939175740353,
      "learning_rate": 9.995001096914462e-06,
      "loss": 0.101,
      "step": 16
    },
    {
      "epoch": 0.030249110320284697,
      "grad_norm": 1.9196819989568994,
      "learning_rate": 9.994356828302818e-06,
      "loss": 0.121,
      "step": 17
    },
    {
      "epoch": 0.03202846975088968,
      "grad_norm": 1.5823196958090517,
      "learning_rate": 9.993673543361874e-06,
      "loss": 0.0909,
      "step": 18
    },
    {
      "epoch": 0.033807829181494664,
      "grad_norm": 2.327516890254515,
      "learning_rate": 9.992951247429512e-06,
      "loss": 0.149,
      "step": 19
    },
    {
      "epoch": 0.03558718861209965,
      "grad_norm": 1.8162688672610439,
      "learning_rate": 9.992189946148366e-06,
      "loss": 0.0911,
      "step": 20
    },
    {
      "epoch": 0.037366548042704624,
      "grad_norm": 1.4041247319897927,
      "learning_rate": 9.991389645465786e-06,
      "loss": 0.0739,
      "step": 21
    },
    {
      "epoch": 0.03914590747330961,
      "grad_norm": 1.8851290416161448,
      "learning_rate": 9.990550351633784e-06,
      "loss": 0.1262,
      "step": 22
    },
    {
      "epoch": 0.04092526690391459,
      "grad_norm": 1.63210744195948,
      "learning_rate": 9.989672071208993e-06,
      "loss": 0.1225,
      "step": 23
    },
    {
      "epoch": 0.042704626334519574,
      "grad_norm": 2.0050146620293154,
      "learning_rate": 9.988754811052616e-06,
      "loss": 0.1629,
      "step": 24
    },
    {
      "epoch": 0.04448398576512456,
      "grad_norm": 2.2192029017519266,
      "learning_rate": 9.987798578330365e-06,
      "loss": 0.1528,
      "step": 25
    },
    {
      "epoch": 0.046263345195729534,
      "grad_norm": 1.8738385921945637,
      "learning_rate": 9.986803380512406e-06,
      "loss": 0.1094,
      "step": 26
    },
    {
      "epoch": 0.04804270462633452,
      "grad_norm": 1.6732757112809902,
      "learning_rate": 9.98576922537331e-06,
      "loss": 0.1277,
      "step": 27
    },
    {
      "epoch": 0.0498220640569395,
      "grad_norm": 1.8916312858303854,
      "learning_rate": 9.984696120991979e-06,
      "loss": 0.1285,
      "step": 28
    },
    {
      "epoch": 0.051601423487544484,
      "grad_norm": 1.7052411499616327,
      "learning_rate": 9.983584075751598e-06,
      "loss": 0.1025,
      "step": 29
    },
    {
      "epoch": 0.05338078291814947,
      "grad_norm": 1.5828027832105331,
      "learning_rate": 9.982433098339553e-06,
      "loss": 0.1087,
      "step": 30
    },
    {
      "epoch": 0.05516014234875445,
      "grad_norm": 2.1560097406788143,
      "learning_rate": 9.981243197747375e-06,
      "loss": 0.1455,
      "step": 31
    },
    {
      "epoch": 0.05693950177935943,
      "grad_norm": 2.607390787096197,
      "learning_rate": 9.980014383270668e-06,
      "loss": 0.1533,
      "step": 32
    },
    {
      "epoch": 0.05871886120996441,
      "grad_norm": 1.9030143959070778,
      "learning_rate": 9.978746664509032e-06,
      "loss": 0.1253,
      "step": 33
    },
    {
      "epoch": 0.060498220640569395,
      "grad_norm": 1.7640275048348333,
      "learning_rate": 9.97744005136599e-06,
      "loss": 0.1047,
      "step": 34
    },
    {
      "epoch": 0.06227758007117438,
      "grad_norm": 1.781622258616041,
      "learning_rate": 9.976094554048912e-06,
      "loss": 0.1192,
      "step": 35
    },
    {
      "epoch": 0.06405693950177936,
      "grad_norm": 2.0819045291855685,
      "learning_rate": 9.974710183068935e-06,
      "loss": 0.1138,
      "step": 36
    },
    {
      "epoch": 0.06583629893238434,
      "grad_norm": 1.7462875295258162,
      "learning_rate": 9.97328694924088e-06,
      "loss": 0.1096,
      "step": 37
    },
    {
      "epoch": 0.06761565836298933,
      "grad_norm": 2.006516306117893,
      "learning_rate": 9.971824863683168e-06,
      "loss": 0.1356,
      "step": 38
    },
    {
      "epoch": 0.0693950177935943,
      "grad_norm": 2.0810563155345636,
      "learning_rate": 9.970323937817732e-06,
      "loss": 0.1126,
      "step": 39
    },
    {
      "epoch": 0.0711743772241993,
      "grad_norm": 1.8092099241779336,
      "learning_rate": 9.968784183369929e-06,
      "loss": 0.0951,
      "step": 40
    },
    {
      "epoch": 0.07295373665480427,
      "grad_norm": 1.9570918946441245,
      "learning_rate": 9.96720561236845e-06,
      "loss": 0.1246,
      "step": 41
    },
    {
      "epoch": 0.07473309608540925,
      "grad_norm": 2.4556981444616035,
      "learning_rate": 9.965588237145219e-06,
      "loss": 0.1515,
      "step": 42
    },
    {
      "epoch": 0.07651245551601424,
      "grad_norm": 1.8564819639482109,
      "learning_rate": 9.963932070335307e-06,
      "loss": 0.1102,
      "step": 43
    },
    {
      "epoch": 0.07829181494661921,
      "grad_norm": 2.1107286592911834,
      "learning_rate": 9.962237124876828e-06,
      "loss": 0.1218,
      "step": 44
    },
    {
      "epoch": 0.0800711743772242,
      "grad_norm": 1.850645723956103,
      "learning_rate": 9.960503414010833e-06,
      "loss": 0.1234,
      "step": 45
    },
    {
      "epoch": 0.08185053380782918,
      "grad_norm": 2.016274659018856,
      "learning_rate": 9.958730951281218e-06,
      "loss": 0.1202,
      "step": 46
    },
    {
      "epoch": 0.08362989323843416,
      "grad_norm": 2.008275852911467,
      "learning_rate": 9.956919750534607e-06,
      "loss": 0.1291,
      "step": 47
    },
    {
      "epoch": 0.08540925266903915,
      "grad_norm": 2.007279817801552,
      "learning_rate": 9.955069825920249e-06,
      "loss": 0.1349,
      "step": 48
    },
    {
      "epoch": 0.08718861209964412,
      "grad_norm": 1.7487451811256158,
      "learning_rate": 9.953181191889913e-06,
      "loss": 0.1361,
      "step": 49
    },
    {
      "epoch": 0.08896797153024912,
      "grad_norm": 1.517484712784115,
      "learning_rate": 9.95125386319776e-06,
      "loss": 0.0999,
      "step": 50
    },
    {
      "epoch": 0.09074733096085409,
      "grad_norm": 2.2499753814457515,
      "learning_rate": 9.949287854900243e-06,
      "loss": 0.1518,
      "step": 51
    },
    {
      "epoch": 0.09252669039145907,
      "grad_norm": 1.941191469639605,
      "learning_rate": 9.947283182355982e-06,
      "loss": 0.1227,
      "step": 52
    },
    {
      "epoch": 0.09430604982206406,
      "grad_norm": 1.8110278448455164,
      "learning_rate": 9.945239861225644e-06,
      "loss": 0.1217,
      "step": 53
    },
    {
      "epoch": 0.09608540925266904,
      "grad_norm": 1.6861188866251153,
      "learning_rate": 9.943157907471825e-06,
      "loss": 0.1088,
      "step": 54
    },
    {
      "epoch": 0.09786476868327403,
      "grad_norm": 1.5380016763892475,
      "learning_rate": 9.941037337358918e-06,
      "loss": 0.0897,
      "step": 55
    },
    {
      "epoch": 0.099644128113879,
      "grad_norm": 2.0206087949865372,
      "learning_rate": 9.938878167452991e-06,
      "loss": 0.1297,
      "step": 56
    },
    {
      "epoch": 0.10142348754448399,
      "grad_norm": 2.2039660026655317,
      "learning_rate": 9.936680414621663e-06,
      "loss": 0.1313,
      "step": 57
    },
    {
      "epoch": 0.10320284697508897,
      "grad_norm": 1.6159185919011219,
      "learning_rate": 9.934444096033958e-06,
      "loss": 0.1039,
      "step": 58
    },
    {
      "epoch": 0.10498220640569395,
      "grad_norm": 1.8014661668111245,
      "learning_rate": 9.932169229160183e-06,
      "loss": 0.1277,
      "step": 59
    },
    {
      "epoch": 0.10676156583629894,
      "grad_norm": 1.8882786008462558,
      "learning_rate": 9.929855831771787e-06,
      "loss": 0.1103,
      "step": 60
    },
    {
      "epoch": 0.10854092526690391,
      "grad_norm": 2.0518090104822924,
      "learning_rate": 9.927503921941218e-06,
      "loss": 0.1201,
      "step": 61
    },
    {
      "epoch": 0.1103202846975089,
      "grad_norm": 2.077298073157288,
      "learning_rate": 9.925113518041796e-06,
      "loss": 0.1434,
      "step": 62
    },
    {
      "epoch": 0.11209964412811388,
      "grad_norm": 2.232364693475648,
      "learning_rate": 9.922684638747551e-06,
      "loss": 0.1643,
      "step": 63
    },
    {
      "epoch": 0.11387900355871886,
      "grad_norm": 1.7031115509859254,
      "learning_rate": 9.920217303033091e-06,
      "loss": 0.11,
      "step": 64
    },
    {
      "epoch": 0.11565836298932385,
      "grad_norm": 1.986748690687823,
      "learning_rate": 9.917711530173444e-06,
      "loss": 0.118,
      "step": 65
    },
    {
      "epoch": 0.11743772241992882,
      "grad_norm": 2.149403485535034,
      "learning_rate": 9.91516733974392e-06,
      "loss": 0.1354,
      "step": 66
    },
    {
      "epoch": 0.11921708185053381,
      "grad_norm": 1.739273816537123,
      "learning_rate": 9.912584751619943e-06,
      "loss": 0.1299,
      "step": 67
    },
    {
      "epoch": 0.12099644128113879,
      "grad_norm": 1.5525247406222797,
      "learning_rate": 9.909963785976902e-06,
      "loss": 0.1059,
      "step": 68
    },
    {
      "epoch": 0.12277580071174377,
      "grad_norm": 1.5786375625670799,
      "learning_rate": 9.907304463290004e-06,
      "loss": 0.1115,
      "step": 69
    },
    {
      "epoch": 0.12455516014234876,
      "grad_norm": 1.5125376174848446,
      "learning_rate": 9.904606804334094e-06,
      "loss": 0.0935,
      "step": 70
    },
    {
      "epoch": 0.12633451957295375,
      "grad_norm": 1.6487502063928734,
      "learning_rate": 9.901870830183506e-06,
      "loss": 0.1154,
      "step": 71
    },
    {
      "epoch": 0.12811387900355872,
      "grad_norm": 1.6202582447353944,
      "learning_rate": 9.899096562211902e-06,
      "loss": 0.1133,
      "step": 72
    },
    {
      "epoch": 0.1298932384341637,
      "grad_norm": 2.0925196722374335,
      "learning_rate": 9.896284022092088e-06,
      "loss": 0.1596,
      "step": 73
    },
    {
      "epoch": 0.13167259786476868,
      "grad_norm": 2.1751876400604253,
      "learning_rate": 9.893433231795864e-06,
      "loss": 0.1687,
      "step": 74
    },
    {
      "epoch": 0.13345195729537365,
      "grad_norm": 2.254229114257366,
      "learning_rate": 9.890544213593838e-06,
      "loss": 0.1478,
      "step": 75
    },
    {
      "epoch": 0.13523131672597866,
      "grad_norm": 1.7064942356188921,
      "learning_rate": 9.887616990055262e-06,
      "loss": 0.1268,
      "step": 76
    },
    {
      "epoch": 0.13701067615658363,
      "grad_norm": 2.251141278398778,
      "learning_rate": 9.884651584047845e-06,
      "loss": 0.1368,
      "step": 77
    },
    {
      "epoch": 0.1387900355871886,
      "grad_norm": 1.9266767368483468,
      "learning_rate": 9.881648018737587e-06,
      "loss": 0.1437,
      "step": 78
    },
    {
      "epoch": 0.14056939501779359,
      "grad_norm": 1.7848749816394225,
      "learning_rate": 9.878606317588588e-06,
      "loss": 0.1048,
      "step": 79
    },
    {
      "epoch": 0.1423487544483986,
      "grad_norm": 2.2510801493075365,
      "learning_rate": 9.875526504362868e-06,
      "loss": 0.1569,
      "step": 80
    },
    {
      "epoch": 0.14412811387900357,
      "grad_norm": 2.2791687325251475,
      "learning_rate": 9.872408603120187e-06,
      "loss": 0.164,
      "step": 81
    },
    {
      "epoch": 0.14590747330960854,
      "grad_norm": 1.8163005919527195,
      "learning_rate": 9.869252638217846e-06,
      "loss": 0.1163,
      "step": 82
    },
    {
      "epoch": 0.14768683274021352,
      "grad_norm": 1.8087620419227883,
      "learning_rate": 9.866058634310503e-06,
      "loss": 0.1373,
      "step": 83
    },
    {
      "epoch": 0.1494661921708185,
      "grad_norm": 1.7281912802627122,
      "learning_rate": 9.862826616349981e-06,
      "loss": 0.1277,
      "step": 84
    },
    {
      "epoch": 0.1512455516014235,
      "grad_norm": 1.923607729784607,
      "learning_rate": 9.859556609585075e-06,
      "loss": 0.1269,
      "step": 85
    },
    {
      "epoch": 0.15302491103202848,
      "grad_norm": 1.831229049893517,
      "learning_rate": 9.856248639561346e-06,
      "loss": 0.1154,
      "step": 86
    },
    {
      "epoch": 0.15480427046263345,
      "grad_norm": 1.9074686739935474,
      "learning_rate": 9.85290273212093e-06,
      "loss": 0.11,
      "step": 87
    },
    {
      "epoch": 0.15658362989323843,
      "grad_norm": 1.8045565678093964,
      "learning_rate": 9.849518913402334e-06,
      "loss": 0.1214,
      "step": 88
    },
    {
      "epoch": 0.1583629893238434,
      "grad_norm": 1.8032943168404858,
      "learning_rate": 9.84609720984023e-06,
      "loss": 0.1226,
      "step": 89
    },
    {
      "epoch": 0.1601423487544484,
      "grad_norm": 1.8975578748071806,
      "learning_rate": 9.84263764816525e-06,
      "loss": 0.1166,
      "step": 90
    },
    {
      "epoch": 0.1619217081850534,
      "grad_norm": 1.8048566163814908,
      "learning_rate": 9.839140255403776e-06,
      "loss": 0.1155,
      "step": 91
    },
    {
      "epoch": 0.16370106761565836,
      "grad_norm": 1.893223930862753,
      "learning_rate": 9.83560505887773e-06,
      "loss": 0.1217,
      "step": 92
    },
    {
      "epoch": 0.16548042704626334,
      "grad_norm": 1.6588280399929534,
      "learning_rate": 9.83203208620436e-06,
      "loss": 0.1203,
      "step": 93
    },
    {
      "epoch": 0.16725978647686832,
      "grad_norm": 2.14320081947194,
      "learning_rate": 9.828421365296023e-06,
      "loss": 0.1286,
      "step": 94
    },
    {
      "epoch": 0.16903914590747332,
      "grad_norm": 2.025436310275238,
      "learning_rate": 9.824772924359974e-06,
      "loss": 0.1271,
      "step": 95
    },
    {
      "epoch": 0.1708185053380783,
      "grad_norm": 1.7723627067257361,
      "learning_rate": 9.821086791898133e-06,
      "loss": 0.1109,
      "step": 96
    },
    {
      "epoch": 0.17259786476868327,
      "grad_norm": 1.4989446449146937,
      "learning_rate": 9.817362996706872e-06,
      "loss": 0.1137,
      "step": 97
    },
    {
      "epoch": 0.17437722419928825,
      "grad_norm": 1.5044874236782473,
      "learning_rate": 9.81360156787679e-06,
      "loss": 0.0965,
      "step": 98
    },
    {
      "epoch": 0.17615658362989323,
      "grad_norm": 1.5268666027615458,
      "learning_rate": 9.809802534792477e-06,
      "loss": 0.1034,
      "step": 99
    },
    {
      "epoch": 0.17793594306049823,
      "grad_norm": 1.8910934973911067,
      "learning_rate": 9.805965927132294e-06,
      "loss": 0.1244,
      "step": 100
    },
    {
      "epoch": 0.1797153024911032,
      "grad_norm": 1.7285625650593492,
      "learning_rate": 9.802091774868143e-06,
      "loss": 0.1126,
      "step": 101
    },
    {
      "epoch": 0.18149466192170818,
      "grad_norm": 1.7589212356828077,
      "learning_rate": 9.798180108265218e-06,
      "loss": 0.1282,
      "step": 102
    },
    {
      "epoch": 0.18327402135231316,
      "grad_norm": 1.5602263250455246,
      "learning_rate": 9.794230957881785e-06,
      "loss": 0.0926,
      "step": 103
    },
    {
      "epoch": 0.18505338078291814,
      "grad_norm": 2.0016073161665893,
      "learning_rate": 9.79024435456893e-06,
      "loss": 0.122,
      "step": 104
    },
    {
      "epoch": 0.18683274021352314,
      "grad_norm": 1.7692510987473,
      "learning_rate": 9.786220329470334e-06,
      "loss": 0.1116,
      "step": 105
    },
    {
      "epoch": 0.18861209964412812,
      "grad_norm": 1.7462432785380835,
      "learning_rate": 9.782158914022011e-06,
      "loss": 0.1179,
      "step": 106
    },
    {
      "epoch": 0.1903914590747331,
      "grad_norm": 1.9729204164994196,
      "learning_rate": 9.778060139952075e-06,
      "loss": 0.1473,
      "step": 107
    },
    {
      "epoch": 0.19217081850533807,
      "grad_norm": 1.7216004683071324,
      "learning_rate": 9.773924039280488e-06,
      "loss": 0.112,
      "step": 108
    },
    {
      "epoch": 0.19395017793594305,
      "grad_norm": 1.2936195343213597,
      "learning_rate": 9.769750644318814e-06,
      "loss": 0.0826,
      "step": 109
    },
    {
      "epoch": 0.19572953736654805,
      "grad_norm": 2.091647632796318,
      "learning_rate": 9.765539987669956e-06,
      "loss": 0.1218,
      "step": 110
    },
    {
      "epoch": 0.19750889679715303,
      "grad_norm": 2.1594912942713913,
      "learning_rate": 9.761292102227917e-06,
      "loss": 0.1584,
      "step": 111
    },
    {
      "epoch": 0.199288256227758,
      "grad_norm": 2.1954918321202035,
      "learning_rate": 9.757007021177529e-06,
      "loss": 0.1585,
      "step": 112
    },
    {
      "epoch": 0.20106761565836298,
      "grad_norm": 1.423979053916681,
      "learning_rate": 9.752684777994197e-06,
      "loss": 0.1024,
      "step": 113
    },
    {
      "epoch": 0.20284697508896798,
      "grad_norm": 2.0610055996363172,
      "learning_rate": 9.748325406443647e-06,
      "loss": 0.1308,
      "step": 114
    },
    {
      "epoch": 0.20462633451957296,
      "grad_norm": 1.9436245682010498,
      "learning_rate": 9.743928940581646e-06,
      "loss": 0.1462,
      "step": 115
    },
    {
      "epoch": 0.20640569395017794,
      "grad_norm": 2.1693767257709387,
      "learning_rate": 9.739495414753754e-06,
      "loss": 0.148,
      "step": 116
    },
    {
      "epoch": 0.20818505338078291,
      "grad_norm": 1.6213065354324263,
      "learning_rate": 9.73502486359504e-06,
      "loss": 0.1148,
      "step": 117
    },
    {
      "epoch": 0.2099644128113879,
      "grad_norm": 2.3237060074988456,
      "learning_rate": 9.73051732202982e-06,
      "loss": 0.1366,
      "step": 118
    },
    {
      "epoch": 0.2117437722419929,
      "grad_norm": 1.7197723438397614,
      "learning_rate": 9.725972825271381e-06,
      "loss": 0.1187,
      "step": 119
    },
    {
      "epoch": 0.21352313167259787,
      "grad_norm": 1.74187600691338,
      "learning_rate": 9.721391408821713e-06,
      "loss": 0.1161,
      "step": 120
    },
    {
      "epoch": 0.21530249110320285,
      "grad_norm": 1.6654992677313791,
      "learning_rate": 9.716773108471213e-06,
      "loss": 0.1265,
      "step": 121
    },
    {
      "epoch": 0.21708185053380782,
      "grad_norm": 1.659156824791211,
      "learning_rate": 9.712117960298433e-06,
      "loss": 0.124,
      "step": 122
    },
    {
      "epoch": 0.2188612099644128,
      "grad_norm": 1.7685127112396284,
      "learning_rate": 9.707426000669773e-06,
      "loss": 0.115,
      "step": 123
    },
    {
      "epoch": 0.2206405693950178,
      "grad_norm": 1.8320851831926341,
      "learning_rate": 9.702697266239211e-06,
      "loss": 0.1225,
      "step": 124
    },
    {
      "epoch": 0.22241992882562278,
      "grad_norm": 2.0034924904385965,
      "learning_rate": 9.697931793948012e-06,
      "loss": 0.1419,
      "step": 125
    },
    {
      "epoch": 0.22419928825622776,
      "grad_norm": 1.3778277109253918,
      "learning_rate": 9.693129621024441e-06,
      "loss": 0.1123,
      "step": 126
    },
    {
      "epoch": 0.22597864768683273,
      "grad_norm": 1.572867202970663,
      "learning_rate": 9.68829078498347e-06,
      "loss": 0.1101,
      "step": 127
    },
    {
      "epoch": 0.2277580071174377,
      "grad_norm": 1.9911478485545366,
      "learning_rate": 9.683415323626487e-06,
      "loss": 0.1333,
      "step": 128
    },
    {
      "epoch": 0.22953736654804271,
      "grad_norm": 2.1526420394512735,
      "learning_rate": 9.678503275040997e-06,
      "loss": 0.1406,
      "step": 129
    },
    {
      "epoch": 0.2313167259786477,
      "grad_norm": 1.7509157608496215,
      "learning_rate": 9.673554677600336e-06,
      "loss": 0.1177,
      "step": 130
    },
    {
      "epoch": 0.23309608540925267,
      "grad_norm": 1.637913603488545,
      "learning_rate": 9.668569569963355e-06,
      "loss": 0.127,
      "step": 131
    },
    {
      "epoch": 0.23487544483985764,
      "grad_norm": 1.3008401042020348,
      "learning_rate": 9.663547991074129e-06,
      "loss": 0.0858,
      "step": 132
    },
    {
      "epoch": 0.23665480427046262,
      "grad_norm": 1.9909063025228493,
      "learning_rate": 9.658489980161643e-06,
      "loss": 0.1322,
      "step": 133
    },
    {
      "epoch": 0.23843416370106763,
      "grad_norm": 1.7908617586454427,
      "learning_rate": 9.653395576739504e-06,
      "loss": 0.1279,
      "step": 134
    },
    {
      "epoch": 0.2402135231316726,
      "grad_norm": 2.1475772083420788,
      "learning_rate": 9.648264820605611e-06,
      "loss": 0.1292,
      "step": 135
    },
    {
      "epoch": 0.24199288256227758,
      "grad_norm": 1.8450806226526224,
      "learning_rate": 9.643097751841854e-06,
      "loss": 0.1728,
      "step": 136
    },
    {
      "epoch": 0.24377224199288255,
      "grad_norm": 1.810413803319839,
      "learning_rate": 9.637894410813803e-06,
      "loss": 0.1364,
      "step": 137
    },
    {
      "epoch": 0.24555160142348753,
      "grad_norm": 2.196555501000083,
      "learning_rate": 9.632654838170393e-06,
      "loss": 0.1372,
      "step": 138
    },
    {
      "epoch": 0.24733096085409254,
      "grad_norm": 1.9545632962479074,
      "learning_rate": 9.627379074843595e-06,
      "loss": 0.1344,
      "step": 139
    },
    {
      "epoch": 0.2491103202846975,
      "grad_norm": 1.7909181406263304,
      "learning_rate": 9.622067162048111e-06,
      "loss": 0.1324,
      "step": 140
    },
    {
      "epoch": 0.2508896797153025,
      "grad_norm": 1.4775863824204931,
      "learning_rate": 9.616719141281044e-06,
      "loss": 0.1226,
      "step": 141
    },
    {
      "epoch": 0.2526690391459075,
      "grad_norm": 2.1594279710190247,
      "learning_rate": 9.611335054321576e-06,
      "loss": 0.1624,
      "step": 142
    },
    {
      "epoch": 0.25444839857651247,
      "grad_norm": 2.0008277111563317,
      "learning_rate": 9.605914943230637e-06,
      "loss": 0.1368,
      "step": 143
    },
    {
      "epoch": 0.25622775800711745,
      "grad_norm": 1.5584971412440352,
      "learning_rate": 9.600458850350588e-06,
      "loss": 0.1116,
      "step": 144
    },
    {
      "epoch": 0.2580071174377224,
      "grad_norm": 1.8962770343238244,
      "learning_rate": 9.594966818304875e-06,
      "loss": 0.1274,
      "step": 145
    },
    {
      "epoch": 0.2597864768683274,
      "grad_norm": 1.66993816626153,
      "learning_rate": 9.589438889997712e-06,
      "loss": 0.0981,
      "step": 146
    },
    {
      "epoch": 0.2615658362989324,
      "grad_norm": 1.5888037677062683,
      "learning_rate": 9.583875108613727e-06,
      "loss": 0.1001,
      "step": 147
    },
    {
      "epoch": 0.26334519572953735,
      "grad_norm": 1.7876632713109166,
      "learning_rate": 9.578275517617646e-06,
      "loss": 0.1265,
      "step": 148
    },
    {
      "epoch": 0.26512455516014233,
      "grad_norm": 1.8250752120968232,
      "learning_rate": 9.572640160753936e-06,
      "loss": 0.1286,
      "step": 149
    },
    {
      "epoch": 0.2669039145907473,
      "grad_norm": 1.8585340687105398,
      "learning_rate": 9.566969082046471e-06,
      "loss": 0.1291,
      "step": 150
    },
    {
      "epoch": 0.26868327402135234,
      "grad_norm": 1.392071488839272,
      "learning_rate": 9.561262325798188e-06,
      "loss": 0.0974,
      "step": 151
    },
    {
      "epoch": 0.2704626334519573,
      "grad_norm": 1.6352044589439714,
      "learning_rate": 9.555519936590739e-06,
      "loss": 0.1034,
      "step": 152
    },
    {
      "epoch": 0.2722419928825623,
      "grad_norm": 1.569559059368265,
      "learning_rate": 9.549741959284147e-06,
      "loss": 0.1023,
      "step": 153
    },
    {
      "epoch": 0.27402135231316727,
      "grad_norm": 1.7620804533134524,
      "learning_rate": 9.543928439016445e-06,
      "loss": 0.1206,
      "step": 154
    },
    {
      "epoch": 0.27580071174377224,
      "grad_norm": 1.653313741667142,
      "learning_rate": 9.538079421203339e-06,
      "loss": 0.1128,
      "step": 155
    },
    {
      "epoch": 0.2775800711743772,
      "grad_norm": 1.6113086994597163,
      "learning_rate": 9.532194951537838e-06,
      "loss": 0.1077,
      "step": 156
    },
    {
      "epoch": 0.2793594306049822,
      "grad_norm": 1.8876239574264841,
      "learning_rate": 9.52627507598991e-06,
      "loss": 0.1261,
      "step": 157
    },
    {
      "epoch": 0.28113879003558717,
      "grad_norm": 1.4699207754893826,
      "learning_rate": 9.52031984080611e-06,
      "loss": 0.0981,
      "step": 158
    },
    {
      "epoch": 0.28291814946619215,
      "grad_norm": 1.671507339053813,
      "learning_rate": 9.514329292509227e-06,
      "loss": 0.0999,
      "step": 159
    },
    {
      "epoch": 0.2846975088967972,
      "grad_norm": 1.6963881236457152,
      "learning_rate": 9.508303477897925e-06,
      "loss": 0.1122,
      "step": 160
    },
    {
      "epoch": 0.28647686832740216,
      "grad_norm": 1.9001224033992847,
      "learning_rate": 9.502242444046365e-06,
      "loss": 0.1194,
      "step": 161
    },
    {
      "epoch": 0.28825622775800713,
      "grad_norm": 1.8042859957541233,
      "learning_rate": 9.496146238303846e-06,
      "loss": 0.1338,
      "step": 162
    },
    {
      "epoch": 0.2900355871886121,
      "grad_norm": 1.8537826618863582,
      "learning_rate": 9.49001490829443e-06,
      "loss": 0.1257,
      "step": 163
    },
    {
      "epoch": 0.2918149466192171,
      "grad_norm": 2.2437080318564484,
      "learning_rate": 9.483848501916578e-06,
      "loss": 0.1646,
      "step": 164
    },
    {
      "epoch": 0.29359430604982206,
      "grad_norm": 1.8856490764519114,
      "learning_rate": 9.477647067342766e-06,
      "loss": 0.1305,
      "step": 165
    },
    {
      "epoch": 0.29537366548042704,
      "grad_norm": 1.871953057517294,
      "learning_rate": 9.471410653019115e-06,
      "loss": 0.1296,
      "step": 166
    },
    {
      "epoch": 0.297153024911032,
      "grad_norm": 1.8890262553041286,
      "learning_rate": 9.46513930766501e-06,
      "loss": 0.1461,
      "step": 167
    },
    {
      "epoch": 0.298932384341637,
      "grad_norm": 1.702604333086193,
      "learning_rate": 9.458833080272723e-06,
      "loss": 0.1193,
      "step": 168
    },
    {
      "epoch": 0.30071174377224197,
      "grad_norm": 2.1060131708011713,
      "learning_rate": 9.45249202010702e-06,
      "loss": 0.1247,
      "step": 169
    },
    {
      "epoch": 0.302491103202847,
      "grad_norm": 1.7844858825551164,
      "learning_rate": 9.446116176704791e-06,
      "loss": 0.1214,
      "step": 170
    },
    {
      "epoch": 0.304270462633452,
      "grad_norm": 1.809965711742156,
      "learning_rate": 9.439705599874653e-06,
      "loss": 0.1267,
      "step": 171
    },
    {
      "epoch": 0.30604982206405695,
      "grad_norm": 1.8383203200257763,
      "learning_rate": 9.433260339696564e-06,
      "loss": 0.1569,
      "step": 172
    },
    {
      "epoch": 0.30782918149466193,
      "grad_norm": 1.851579423022255,
      "learning_rate": 9.426780446521429e-06,
      "loss": 0.1427,
      "step": 173
    },
    {
      "epoch": 0.3096085409252669,
      "grad_norm": 1.7486749889832602,
      "learning_rate": 9.42026597097071e-06,
      "loss": 0.1375,
      "step": 174
    },
    {
      "epoch": 0.3113879003558719,
      "grad_norm": 1.5426125916987024,
      "learning_rate": 9.413716963936033e-06,
      "loss": 0.1067,
      "step": 175
    },
    {
      "epoch": 0.31316725978647686,
      "grad_norm": 1.7767166152296758,
      "learning_rate": 9.407133476578778e-06,
      "loss": 0.1304,
      "step": 176
    },
    {
      "epoch": 0.31494661921708184,
      "grad_norm": 2.0825001597829824,
      "learning_rate": 9.400515560329698e-06,
      "loss": 0.1614,
      "step": 177
    },
    {
      "epoch": 0.3167259786476868,
      "grad_norm": 1.936853644459691,
      "learning_rate": 9.393863266888501e-06,
      "loss": 0.1285,
      "step": 178
    },
    {
      "epoch": 0.3185053380782918,
      "grad_norm": 1.8691308371098896,
      "learning_rate": 9.387176648223457e-06,
      "loss": 0.1293,
      "step": 179
    },
    {
      "epoch": 0.3202846975088968,
      "grad_norm": 2.0845945821205594,
      "learning_rate": 9.38045575657098e-06,
      "loss": 0.1385,
      "step": 180
    },
    {
      "epoch": 0.3220640569395018,
      "grad_norm": 1.8922278351620145,
      "learning_rate": 9.37370064443524e-06,
      "loss": 0.1326,
      "step": 181
    },
    {
      "epoch": 0.3238434163701068,
      "grad_norm": 1.4484109298708199,
      "learning_rate": 9.366911364587726e-06,
      "loss": 0.1088,
      "step": 182
    },
    {
      "epoch": 0.32562277580071175,
      "grad_norm": 1.6068722883178463,
      "learning_rate": 9.360087970066854e-06,
      "loss": 0.1158,
      "step": 183
    },
    {
      "epoch": 0.3274021352313167,
      "grad_norm": 1.6483460514222545,
      "learning_rate": 9.353230514177553e-06,
      "loss": 0.1084,
      "step": 184
    },
    {
      "epoch": 0.3291814946619217,
      "grad_norm": 1.5900280334030124,
      "learning_rate": 9.346339050490832e-06,
      "loss": 0.1084,
      "step": 185
    },
    {
      "epoch": 0.3309608540925267,
      "grad_norm": 1.9398841007100078,
      "learning_rate": 9.33941363284338e-06,
      "loss": 0.1243,
      "step": 186
    },
    {
      "epoch": 0.33274021352313166,
      "grad_norm": 1.768852204916369,
      "learning_rate": 9.332454315337129e-06,
      "loss": 0.1211,
      "step": 187
    },
    {
      "epoch": 0.33451957295373663,
      "grad_norm": 1.8111870820232214,
      "learning_rate": 9.325461152338846e-06,
      "loss": 0.1186,
      "step": 188
    },
    {
      "epoch": 0.33629893238434166,
      "grad_norm": 1.8399631044999671,
      "learning_rate": 9.3184341984797e-06,
      "loss": 0.1254,
      "step": 189
    },
    {
      "epoch": 0.33807829181494664,
      "grad_norm": 1.8969509563880365,
      "learning_rate": 9.311373508654838e-06,
      "loss": 0.1282,
      "step": 190
    },
    {
      "epoch": 0.3398576512455516,
      "grad_norm": 2.063335135899934,
      "learning_rate": 9.30427913802295e-06,
      "loss": 0.1707,
      "step": 191
    },
    {
      "epoch": 0.3416370106761566,
      "grad_norm": 2.1666854721298567,
      "learning_rate": 9.297151142005852e-06,
      "loss": 0.148,
      "step": 192
    },
    {
      "epoch": 0.34341637010676157,
      "grad_norm": 1.650369001188152,
      "learning_rate": 9.289989576288035e-06,
      "loss": 0.1191,
      "step": 193
    },
    {
      "epoch": 0.34519572953736655,
      "grad_norm": 1.8993536752951423,
      "learning_rate": 9.282794496816244e-06,
      "loss": 0.1292,
      "step": 194
    },
    {
      "epoch": 0.3469750889679715,
      "grad_norm": 2.2609471080912584,
      "learning_rate": 9.27556595979904e-06,
      "loss": 0.1272,
      "step": 195
    },
    {
      "epoch": 0.3487544483985765,
      "grad_norm": 1.7117829576423016,
      "learning_rate": 9.26830402170635e-06,
      "loss": 0.1287,
      "step": 196
    },
    {
      "epoch": 0.3505338078291815,
      "grad_norm": 1.6290750835204475,
      "learning_rate": 9.261008739269035e-06,
      "loss": 0.1064,
      "step": 197
    },
    {
      "epoch": 0.35231316725978645,
      "grad_norm": 1.7563570331974911,
      "learning_rate": 9.253680169478448e-06,
      "loss": 0.1256,
      "step": 198
    },
    {
      "epoch": 0.3540925266903915,
      "grad_norm": 1.7616635408950143,
      "learning_rate": 9.246318369585983e-06,
      "loss": 0.1236,
      "step": 199
    },
    {
      "epoch": 0.35587188612099646,
      "grad_norm": 2.239621202888575,
      "learning_rate": 9.238923397102629e-06,
      "loss": 0.1506,
      "step": 200
    },
    {
      "epoch": 0.35587188612099646,
      "eval_loss": 0.13475362956523895,
      "eval_runtime": 1.9305,
      "eval_samples_per_second": 23.828,
      "eval_steps_per_second": 6.216,
      "step": 200
    },
    {
      "epoch": 0.35765124555160144,
      "grad_norm": 1.722022941441938,
      "learning_rate": 9.231495309798525e-06,
      "loss": 0.1094,
      "step": 201
    },
    {
      "epoch": 0.3594306049822064,
      "grad_norm": 1.8851030530255979,
      "learning_rate": 9.224034165702506e-06,
      "loss": 0.1297,
      "step": 202
    },
    {
      "epoch": 0.3612099644128114,
      "grad_norm": 1.312223289817984,
      "learning_rate": 9.216540023101646e-06,
      "loss": 0.0999,
      "step": 203
    },
    {
      "epoch": 0.36298932384341637,
      "grad_norm": 1.4473604929562314,
      "learning_rate": 9.209012940540806e-06,
      "loss": 0.1123,
      "step": 204
    },
    {
      "epoch": 0.36476868327402134,
      "grad_norm": 1.4571406246983327,
      "learning_rate": 9.20145297682218e-06,
      "loss": 0.1176,
      "step": 205
    },
    {
      "epoch": 0.3665480427046263,
      "grad_norm": 2.3012201580614335,
      "learning_rate": 9.193860191004833e-06,
      "loss": 0.1607,
      "step": 206
    },
    {
      "epoch": 0.3683274021352313,
      "grad_norm": 1.7914329813601564,
      "learning_rate": 9.186234642404234e-06,
      "loss": 0.1425,
      "step": 207
    },
    {
      "epoch": 0.3701067615658363,
      "grad_norm": 2.1066136590560833,
      "learning_rate": 9.178576390591803e-06,
      "loss": 0.1376,
      "step": 208
    },
    {
      "epoch": 0.3718861209964413,
      "grad_norm": 1.4796833660326827,
      "learning_rate": 9.170885495394435e-06,
      "loss": 0.1075,
      "step": 209
    },
    {
      "epoch": 0.3736654804270463,
      "grad_norm": 2.1451912781621703,
      "learning_rate": 9.16316201689404e-06,
      "loss": 0.1352,
      "step": 210
    },
    {
      "epoch": 0.37544483985765126,
      "grad_norm": 1.9950963378711903,
      "learning_rate": 9.155406015427076e-06,
      "loss": 0.136,
      "step": 211
    },
    {
      "epoch": 0.37722419928825623,
      "grad_norm": 2.014416096651257,
      "learning_rate": 9.147617551584066e-06,
      "loss": 0.125,
      "step": 212
    },
    {
      "epoch": 0.3790035587188612,
      "grad_norm": 1.6357866840912252,
      "learning_rate": 9.139796686209135e-06,
      "loss": 0.1385,
      "step": 213
    },
    {
      "epoch": 0.3807829181494662,
      "grad_norm": 1.5071862190810121,
      "learning_rate": 9.131943480399531e-06,
      "loss": 0.1035,
      "step": 214
    },
    {
      "epoch": 0.38256227758007116,
      "grad_norm": 1.5458548378488257,
      "learning_rate": 9.124057995505148e-06,
      "loss": 0.1179,
      "step": 215
    },
    {
      "epoch": 0.38434163701067614,
      "grad_norm": 2.437466435728138,
      "learning_rate": 9.11614029312805e-06,
      "loss": 0.179,
      "step": 216
    },
    {
      "epoch": 0.3861209964412811,
      "grad_norm": 2.4650250923800145,
      "learning_rate": 9.108190435121982e-06,
      "loss": 0.145,
      "step": 217
    },
    {
      "epoch": 0.3879003558718861,
      "grad_norm": 1.9026122449272178,
      "learning_rate": 9.100208483591892e-06,
      "loss": 0.1169,
      "step": 218
    },
    {
      "epoch": 0.3896797153024911,
      "grad_norm": 2.538495019627512,
      "learning_rate": 9.092194500893448e-06,
      "loss": 0.1915,
      "step": 219
    },
    {
      "epoch": 0.3914590747330961,
      "grad_norm": 1.8352767388204005,
      "learning_rate": 9.084148549632547e-06,
      "loss": 0.1093,
      "step": 220
    },
    {
      "epoch": 0.3932384341637011,
      "grad_norm": 1.7395462886077808,
      "learning_rate": 9.076070692664827e-06,
      "loss": 0.1323,
      "step": 221
    },
    {
      "epoch": 0.39501779359430605,
      "grad_norm": 1.9492465490294997,
      "learning_rate": 9.067960993095176e-06,
      "loss": 0.1371,
      "step": 222
    },
    {
      "epoch": 0.39679715302491103,
      "grad_norm": 1.9129506900488067,
      "learning_rate": 9.059819514277238e-06,
      "loss": 0.129,
      "step": 223
    },
    {
      "epoch": 0.398576512455516,
      "grad_norm": 1.5683853142980035,
      "learning_rate": 9.05164631981292e-06,
      "loss": 0.1138,
      "step": 224
    },
    {
      "epoch": 0.400355871886121,
      "grad_norm": 1.8327399507256446,
      "learning_rate": 9.043441473551893e-06,
      "loss": 0.1354,
      "step": 225
    },
    {
      "epoch": 0.40213523131672596,
      "grad_norm": 1.8097839744674817,
      "learning_rate": 9.035205039591099e-06,
      "loss": 0.1185,
      "step": 226
    },
    {
      "epoch": 0.40391459074733094,
      "grad_norm": 1.8950697374562147,
      "learning_rate": 9.02693708227424e-06,
      "loss": 0.1308,
      "step": 227
    },
    {
      "epoch": 0.40569395017793597,
      "grad_norm": 1.5329358084281786,
      "learning_rate": 9.018637666191284e-06,
      "loss": 0.1414,
      "step": 228
    },
    {
      "epoch": 0.40747330960854095,
      "grad_norm": 1.8451527794408986,
      "learning_rate": 9.010306856177958e-06,
      "loss": 0.1526,
      "step": 229
    },
    {
      "epoch": 0.4092526690391459,
      "grad_norm": 1.8746813968819265,
      "learning_rate": 9.001944717315236e-06,
      "loss": 0.1521,
      "step": 230
    },
    {
      "epoch": 0.4110320284697509,
      "grad_norm": 1.8654814285834336,
      "learning_rate": 8.993551314928846e-06,
      "loss": 0.1353,
      "step": 231
    },
    {
      "epoch": 0.4128113879003559,
      "grad_norm": 1.588468323963603,
      "learning_rate": 8.985126714588739e-06,
      "loss": 0.101,
      "step": 232
    },
    {
      "epoch": 0.41459074733096085,
      "grad_norm": 1.6435632650135055,
      "learning_rate": 8.976670982108591e-06,
      "loss": 0.1296,
      "step": 233
    },
    {
      "epoch": 0.41637010676156583,
      "grad_norm": 2.074509978038458,
      "learning_rate": 8.968184183545285e-06,
      "loss": 0.161,
      "step": 234
    },
    {
      "epoch": 0.4181494661921708,
      "grad_norm": 1.800921231336751,
      "learning_rate": 8.959666385198396e-06,
      "loss": 0.127,
      "step": 235
    },
    {
      "epoch": 0.4199288256227758,
      "grad_norm": 1.5481816813962979,
      "learning_rate": 8.951117653609666e-06,
      "loss": 0.1066,
      "step": 236
    },
    {
      "epoch": 0.42170818505338076,
      "grad_norm": 1.8174378992361993,
      "learning_rate": 8.9425380555625e-06,
      "loss": 0.1265,
      "step": 237
    },
    {
      "epoch": 0.4234875444839858,
      "grad_norm": 1.213418675768833,
      "learning_rate": 8.933927658081423e-06,
      "loss": 0.0879,
      "step": 238
    },
    {
      "epoch": 0.42526690391459077,
      "grad_norm": 1.7336578526752655,
      "learning_rate": 8.925286528431578e-06,
      "loss": 0.1282,
      "step": 239
    },
    {
      "epoch": 0.42704626334519574,
      "grad_norm": 1.7581123861292776,
      "learning_rate": 8.916614734118184e-06,
      "loss": 0.1074,
      "step": 240
    },
    {
      "epoch": 0.4288256227758007,
      "grad_norm": 1.4649653987511146,
      "learning_rate": 8.907912342886016e-06,
      "loss": 0.1049,
      "step": 241
    },
    {
      "epoch": 0.4306049822064057,
      "grad_norm": 1.8770703024666036,
      "learning_rate": 8.899179422718877e-06,
      "loss": 0.1299,
      "step": 242
    },
    {
      "epoch": 0.43238434163701067,
      "grad_norm": 1.7042374945879506,
      "learning_rate": 8.890416041839061e-06,
      "loss": 0.1205,
      "step": 243
    },
    {
      "epoch": 0.43416370106761565,
      "grad_norm": 1.541039271792107,
      "learning_rate": 8.881622268706825e-06,
      "loss": 0.1048,
      "step": 244
    },
    {
      "epoch": 0.4359430604982206,
      "grad_norm": 1.623341393251503,
      "learning_rate": 8.872798172019856e-06,
      "loss": 0.1039,
      "step": 245
    },
    {
      "epoch": 0.4377224199288256,
      "grad_norm": 1.8832552058089562,
      "learning_rate": 8.863943820712726e-06,
      "loss": 0.1323,
      "step": 246
    },
    {
      "epoch": 0.4395017793594306,
      "grad_norm": 1.8056962073705773,
      "learning_rate": 8.855059283956363e-06,
      "loss": 0.1565,
      "step": 247
    },
    {
      "epoch": 0.4412811387900356,
      "grad_norm": 2.064823148363225,
      "learning_rate": 8.8461446311575e-06,
      "loss": 0.1172,
      "step": 248
    },
    {
      "epoch": 0.4430604982206406,
      "grad_norm": 1.5565441163419895,
      "learning_rate": 8.837199931958147e-06,
      "loss": 0.0999,
      "step": 249
    },
    {
      "epoch": 0.44483985765124556,
      "grad_norm": 1.8960372885941401,
      "learning_rate": 8.828225256235035e-06,
      "loss": 0.1519,
      "step": 250
    },
    {
      "epoch": 0.44661921708185054,
      "grad_norm": 1.969709272424476,
      "learning_rate": 8.819220674099074e-06,
      "loss": 0.1401,
      "step": 251
    },
    {
      "epoch": 0.4483985765124555,
      "grad_norm": 1.6753926221638498,
      "learning_rate": 8.810186255894804e-06,
      "loss": 0.1131,
      "step": 252
    },
    {
      "epoch": 0.4501779359430605,
      "grad_norm": 1.790313522333547,
      "learning_rate": 8.801122072199848e-06,
      "loss": 0.1128,
      "step": 253
    },
    {
      "epoch": 0.45195729537366547,
      "grad_norm": 1.5256328049106158,
      "learning_rate": 8.792028193824364e-06,
      "loss": 0.0996,
      "step": 254
    },
    {
      "epoch": 0.45373665480427045,
      "grad_norm": 1.5352367306579815,
      "learning_rate": 8.782904691810478e-06,
      "loss": 0.0973,
      "step": 255
    },
    {
      "epoch": 0.4555160142348754,
      "grad_norm": 1.5787792820803526,
      "learning_rate": 8.77375163743175e-06,
      "loss": 0.107,
      "step": 256
    },
    {
      "epoch": 0.45729537366548045,
      "grad_norm": 1.6006892743569767,
      "learning_rate": 8.764569102192593e-06,
      "loss": 0.1243,
      "step": 257
    },
    {
      "epoch": 0.45907473309608543,
      "grad_norm": 1.486918683494662,
      "learning_rate": 8.755357157827735e-06,
      "loss": 0.0855,
      "step": 258
    },
    {
      "epoch": 0.4608540925266904,
      "grad_norm": 1.9745134700331965,
      "learning_rate": 8.746115876301651e-06,
      "loss": 0.1381,
      "step": 259
    },
    {
      "epoch": 0.4626334519572954,
      "grad_norm": 1.7190356287429387,
      "learning_rate": 8.736845329807994e-06,
      "loss": 0.1159,
      "step": 260
    },
    {
      "epoch": 0.46441281138790036,
      "grad_norm": 1.821394554170655,
      "learning_rate": 8.727545590769044e-06,
      "loss": 0.1225,
      "step": 261
    },
    {
      "epoch": 0.46619217081850534,
      "grad_norm": 1.563725148737111,
      "learning_rate": 8.718216731835131e-06,
      "loss": 0.1181,
      "step": 262
    },
    {
      "epoch": 0.4679715302491103,
      "grad_norm": 1.7469646688639315,
      "learning_rate": 8.708858825884075e-06,
      "loss": 0.1463,
      "step": 263
    },
    {
      "epoch": 0.4697508896797153,
      "grad_norm": 1.2802876928340927,
      "learning_rate": 8.699471946020612e-06,
      "loss": 0.1046,
      "step": 264
    },
    {
      "epoch": 0.47153024911032027,
      "grad_norm": 1.726895684120543,
      "learning_rate": 8.690056165575825e-06,
      "loss": 0.1203,
      "step": 265
    },
    {
      "epoch": 0.47330960854092524,
      "grad_norm": 1.6877644560065082,
      "learning_rate": 8.680611558106571e-06,
      "loss": 0.1204,
      "step": 266
    },
    {
      "epoch": 0.4750889679715303,
      "grad_norm": 1.6351820175443095,
      "learning_rate": 8.671138197394907e-06,
      "loss": 0.1112,
      "step": 267
    },
    {
      "epoch": 0.47686832740213525,
      "grad_norm": 1.391943383044968,
      "learning_rate": 8.661636157447511e-06,
      "loss": 0.11,
      "step": 268
    },
    {
      "epoch": 0.4786476868327402,
      "grad_norm": 2.241489032548888,
      "learning_rate": 8.652105512495106e-06,
      "loss": 0.1403,
      "step": 269
    },
    {
      "epoch": 0.4804270462633452,
      "grad_norm": 1.6491124607865364,
      "learning_rate": 8.64254633699188e-06,
      "loss": 0.1063,
      "step": 270
    },
    {
      "epoch": 0.4822064056939502,
      "grad_norm": 2.040840488179136,
      "learning_rate": 8.632958705614905e-06,
      "loss": 0.1401,
      "step": 271
    },
    {
      "epoch": 0.48398576512455516,
      "grad_norm": 1.598519190543473,
      "learning_rate": 8.623342693263549e-06,
      "loss": 0.1076,
      "step": 272
    },
    {
      "epoch": 0.48576512455516013,
      "grad_norm": 1.885951067799148,
      "learning_rate": 8.6136983750589e-06,
      "loss": 0.1299,
      "step": 273
    },
    {
      "epoch": 0.4875444839857651,
      "grad_norm": 1.9080426833799893,
      "learning_rate": 8.604025826343167e-06,
      "loss": 0.1321,
      "step": 274
    },
    {
      "epoch": 0.4893238434163701,
      "grad_norm": 2.167665140654122,
      "learning_rate": 8.594325122679107e-06,
      "loss": 0.1496,
      "step": 275
    },
    {
      "epoch": 0.49110320284697506,
      "grad_norm": 1.6546589653225403,
      "learning_rate": 8.584596339849419e-06,
      "loss": 0.1268,
      "step": 276
    },
    {
      "epoch": 0.4928825622775801,
      "grad_norm": 1.3032213408818085,
      "learning_rate": 8.574839553856157e-06,
      "loss": 0.0915,
      "step": 277
    },
    {
      "epoch": 0.49466192170818507,
      "grad_norm": 2.006909044611022,
      "learning_rate": 8.565054840920145e-06,
      "loss": 0.1563,
      "step": 278
    },
    {
      "epoch": 0.49644128113879005,
      "grad_norm": 1.4707698991652156,
      "learning_rate": 8.55524227748037e-06,
      "loss": 0.1003,
      "step": 279
    },
    {
      "epoch": 0.498220640569395,
      "grad_norm": 1.6284224803578924,
      "learning_rate": 8.545401940193392e-06,
      "loss": 0.1041,
      "step": 280
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7660225764099606,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.1181,
      "step": 281
    },
    {
      "epoch": 0.501779359430605,
      "grad_norm": 2.3295025011304404,
      "learning_rate": 8.525638251788312e-06,
      "loss": 0.1707,
      "step": 282
    },
    {
      "epoch": 0.50355871886121,
      "grad_norm": 1.8460447919155434,
      "learning_rate": 8.515715055065783e-06,
      "loss": 0.1392,
      "step": 283
    },
    {
      "epoch": 0.505338078291815,
      "grad_norm": 1.7877505252596202,
      "learning_rate": 8.505764393285985e-06,
      "loss": 0.1502,
      "step": 284
    },
    {
      "epoch": 0.5071174377224199,
      "grad_norm": 1.8027468359852232,
      "learning_rate": 8.495786344184314e-06,
      "loss": 0.1155,
      "step": 285
    },
    {
      "epoch": 0.5088967971530249,
      "grad_norm": 1.8283957307006067,
      "learning_rate": 8.485780985710113e-06,
      "loss": 0.1325,
      "step": 286
    },
    {
      "epoch": 0.5106761565836299,
      "grad_norm": 1.617620230902481,
      "learning_rate": 8.475748396026074e-06,
      "loss": 0.1143,
      "step": 287
    },
    {
      "epoch": 0.5124555160142349,
      "grad_norm": 1.7737423091093987,
      "learning_rate": 8.46568865350762e-06,
      "loss": 0.1288,
      "step": 288
    },
    {
      "epoch": 0.5142348754448398,
      "grad_norm": 2.16238464487307,
      "learning_rate": 8.45560183674229e-06,
      "loss": 0.1449,
      "step": 289
    },
    {
      "epoch": 0.5160142348754448,
      "grad_norm": 1.9859827985295293,
      "learning_rate": 8.445488024529133e-06,
      "loss": 0.1336,
      "step": 290
    },
    {
      "epoch": 0.5177935943060499,
      "grad_norm": 1.4852303020995183,
      "learning_rate": 8.435347295878087e-06,
      "loss": 0.0909,
      "step": 291
    },
    {
      "epoch": 0.5195729537366548,
      "grad_norm": 1.4904683254821536,
      "learning_rate": 8.425179730009368e-06,
      "loss": 0.0935,
      "step": 292
    },
    {
      "epoch": 0.5213523131672598,
      "grad_norm": 1.8017255674921966,
      "learning_rate": 8.41498540635284e-06,
      "loss": 0.1289,
      "step": 293
    },
    {
      "epoch": 0.5231316725978647,
      "grad_norm": 1.868536635319127,
      "learning_rate": 8.404764404547404e-06,
      "loss": 0.1225,
      "step": 294
    },
    {
      "epoch": 0.5249110320284698,
      "grad_norm": 1.925292982995918,
      "learning_rate": 8.394516804440374e-06,
      "loss": 0.1255,
      "step": 295
    },
    {
      "epoch": 0.5266903914590747,
      "grad_norm": 2.0014473918154194,
      "learning_rate": 8.384242686086848e-06,
      "loss": 0.1284,
      "step": 296
    },
    {
      "epoch": 0.5284697508896797,
      "grad_norm": 1.9207087771935707,
      "learning_rate": 8.373942129749094e-06,
      "loss": 0.1563,
      "step": 297
    },
    {
      "epoch": 0.5302491103202847,
      "grad_norm": 1.4839026560301476,
      "learning_rate": 8.363615215895908e-06,
      "loss": 0.113,
      "step": 298
    },
    {
      "epoch": 0.5320284697508897,
      "grad_norm": 2.01462825124946,
      "learning_rate": 8.353262025202e-06,
      "loss": 0.1341,
      "step": 299
    },
    {
      "epoch": 0.5338078291814946,
      "grad_norm": 1.6455438560729794,
      "learning_rate": 8.342882638547351e-06,
      "loss": 0.1044,
      "step": 300
    },
    {
      "epoch": 0.5355871886120996,
      "grad_norm": 1.7305094065473785,
      "learning_rate": 8.332477137016587e-06,
      "loss": 0.1188,
      "step": 301
    },
    {
      "epoch": 0.5373665480427047,
      "grad_norm": 1.542702868065484,
      "learning_rate": 8.322045601898354e-06,
      "loss": 0.1015,
      "step": 302
    },
    {
      "epoch": 0.5391459074733096,
      "grad_norm": 1.5192625113596616,
      "learning_rate": 8.311588114684665e-06,
      "loss": 0.1072,
      "step": 303
    },
    {
      "epoch": 0.5409252669039146,
      "grad_norm": 1.9929013945889749,
      "learning_rate": 8.301104757070276e-06,
      "loss": 0.1891,
      "step": 304
    },
    {
      "epoch": 0.5427046263345195,
      "grad_norm": 1.917073137249212,
      "learning_rate": 8.290595610952045e-06,
      "loss": 0.1301,
      "step": 305
    },
    {
      "epoch": 0.5444839857651246,
      "grad_norm": 2.105582590022374,
      "learning_rate": 8.280060758428294e-06,
      "loss": 0.151,
      "step": 306
    },
    {
      "epoch": 0.5462633451957295,
      "grad_norm": 1.4383491073063088,
      "learning_rate": 8.269500281798164e-06,
      "loss": 0.0958,
      "step": 307
    },
    {
      "epoch": 0.5480427046263345,
      "grad_norm": 1.71817539102519,
      "learning_rate": 8.258914263560971e-06,
      "loss": 0.1178,
      "step": 308
    },
    {
      "epoch": 0.5498220640569395,
      "grad_norm": 1.6918612604010685,
      "learning_rate": 8.248302786415567e-06,
      "loss": 0.125,
      "step": 309
    },
    {
      "epoch": 0.5516014234875445,
      "grad_norm": 1.974800223194389,
      "learning_rate": 8.237665933259693e-06,
      "loss": 0.1265,
      "step": 310
    },
    {
      "epoch": 0.5533807829181495,
      "grad_norm": 2.015176371319247,
      "learning_rate": 8.227003787189323e-06,
      "loss": 0.1352,
      "step": 311
    },
    {
      "epoch": 0.5551601423487544,
      "grad_norm": 1.5231634085500674,
      "learning_rate": 8.216316431498028e-06,
      "loss": 0.1125,
      "step": 312
    },
    {
      "epoch": 0.5569395017793595,
      "grad_norm": 1.686994913410353,
      "learning_rate": 8.205603949676317e-06,
      "loss": 0.1421,
      "step": 313
    },
    {
      "epoch": 0.5587188612099644,
      "grad_norm": 1.6346089206372874,
      "learning_rate": 8.194866425410984e-06,
      "loss": 0.1245,
      "step": 314
    },
    {
      "epoch": 0.5604982206405694,
      "grad_norm": 1.818534252631312,
      "learning_rate": 8.184103942584456e-06,
      "loss": 0.1229,
      "step": 315
    },
    {
      "epoch": 0.5622775800711743,
      "grad_norm": 1.5560887578457265,
      "learning_rate": 8.173316585274144e-06,
      "loss": 0.1099,
      "step": 316
    },
    {
      "epoch": 0.5640569395017794,
      "grad_norm": 1.3145157648633286,
      "learning_rate": 8.162504437751775e-06,
      "loss": 0.1061,
      "step": 317
    },
    {
      "epoch": 0.5658362989323843,
      "grad_norm": 1.5183108662488272,
      "learning_rate": 8.151667584482742e-06,
      "loss": 0.1129,
      "step": 318
    },
    {
      "epoch": 0.5676156583629893,
      "grad_norm": 1.932761087650258,
      "learning_rate": 8.140806110125442e-06,
      "loss": 0.1329,
      "step": 319
    },
    {
      "epoch": 0.5693950177935944,
      "grad_norm": 1.5180293915846248,
      "learning_rate": 8.129920099530608e-06,
      "loss": 0.1273,
      "step": 320
    },
    {
      "epoch": 0.5711743772241993,
      "grad_norm": 1.2786794962721737,
      "learning_rate": 8.119009637740663e-06,
      "loss": 0.0786,
      "step": 321
    },
    {
      "epoch": 0.5729537366548043,
      "grad_norm": 1.5302896723393862,
      "learning_rate": 8.108074809989032e-06,
      "loss": 0.1186,
      "step": 322
    },
    {
      "epoch": 0.5747330960854092,
      "grad_norm": 1.4684058764049013,
      "learning_rate": 8.097115701699498e-06,
      "loss": 0.0832,
      "step": 323
    },
    {
      "epoch": 0.5765124555160143,
      "grad_norm": 1.8338256122637822,
      "learning_rate": 8.086132398485525e-06,
      "loss": 0.1372,
      "step": 324
    },
    {
      "epoch": 0.5782918149466192,
      "grad_norm": 1.2459604312864998,
      "learning_rate": 8.075124986149583e-06,
      "loss": 0.0926,
      "step": 325
    },
    {
      "epoch": 0.5800711743772242,
      "grad_norm": 1.5228442272884104,
      "learning_rate": 8.064093550682494e-06,
      "loss": 0.1074,
      "step": 326
    },
    {
      "epoch": 0.5818505338078291,
      "grad_norm": 1.4784309568139622,
      "learning_rate": 8.053038178262742e-06,
      "loss": 0.088,
      "step": 327
    },
    {
      "epoch": 0.5836298932384342,
      "grad_norm": 1.9491523552105292,
      "learning_rate": 8.041958955255815e-06,
      "loss": 0.1309,
      "step": 328
    },
    {
      "epoch": 0.5854092526690391,
      "grad_norm": 1.7028424152601356,
      "learning_rate": 8.030855968213518e-06,
      "loss": 0.116,
      "step": 329
    },
    {
      "epoch": 0.5871886120996441,
      "grad_norm": 1.832716229289757,
      "learning_rate": 8.019729303873307e-06,
      "loss": 0.1222,
      "step": 330
    },
    {
      "epoch": 0.5889679715302492,
      "grad_norm": 1.4787109520975619,
      "learning_rate": 8.008579049157607e-06,
      "loss": 0.0942,
      "step": 331
    },
    {
      "epoch": 0.5907473309608541,
      "grad_norm": 1.3870666218321572,
      "learning_rate": 7.99740529117313e-06,
      "loss": 0.1034,
      "step": 332
    },
    {
      "epoch": 0.5925266903914591,
      "grad_norm": 1.8729180024029073,
      "learning_rate": 7.986208117210198e-06,
      "loss": 0.1215,
      "step": 333
    },
    {
      "epoch": 0.594306049822064,
      "grad_norm": 1.9494037432879063,
      "learning_rate": 7.974987614742066e-06,
      "loss": 0.1291,
      "step": 334
    },
    {
      "epoch": 0.5960854092526691,
      "grad_norm": 2.074788964054358,
      "learning_rate": 7.963743871424224e-06,
      "loss": 0.1354,
      "step": 335
    },
    {
      "epoch": 0.597864768683274,
      "grad_norm": 1.7358856083117655,
      "learning_rate": 7.952476975093729e-06,
      "loss": 0.1096,
      "step": 336
    },
    {
      "epoch": 0.599644128113879,
      "grad_norm": 2.18549630435684,
      "learning_rate": 7.941187013768508e-06,
      "loss": 0.1467,
      "step": 337
    },
    {
      "epoch": 0.6014234875444839,
      "grad_norm": 1.8549801001780495,
      "learning_rate": 7.929874075646673e-06,
      "loss": 0.1158,
      "step": 338
    },
    {
      "epoch": 0.603202846975089,
      "grad_norm": 2.281363378969548,
      "learning_rate": 7.918538249105835e-06,
      "loss": 0.1327,
      "step": 339
    },
    {
      "epoch": 0.604982206405694,
      "grad_norm": 1.9737692145790846,
      "learning_rate": 7.907179622702409e-06,
      "loss": 0.1371,
      "step": 340
    },
    {
      "epoch": 0.6067615658362989,
      "grad_norm": 1.541775896502389,
      "learning_rate": 7.895798285170927e-06,
      "loss": 0.1055,
      "step": 341
    },
    {
      "epoch": 0.608540925266904,
      "grad_norm": 1.5424414645659343,
      "learning_rate": 7.88439432542334e-06,
      "loss": 0.092,
      "step": 342
    },
    {
      "epoch": 0.6103202846975089,
      "grad_norm": 1.7641424692067884,
      "learning_rate": 7.872967832548327e-06,
      "loss": 0.1322,
      "step": 343
    },
    {
      "epoch": 0.6120996441281139,
      "grad_norm": 1.7068160648312833,
      "learning_rate": 7.861518895810597e-06,
      "loss": 0.1342,
      "step": 344
    },
    {
      "epoch": 0.6138790035587188,
      "grad_norm": 1.6198357548760545,
      "learning_rate": 7.850047604650188e-06,
      "loss": 0.1291,
      "step": 345
    },
    {
      "epoch": 0.6156583629893239,
      "grad_norm": 1.9424546799664637,
      "learning_rate": 7.838554048681783e-06,
      "loss": 0.1416,
      "step": 346
    },
    {
      "epoch": 0.6174377224199288,
      "grad_norm": 1.8134114400606622,
      "learning_rate": 7.827038317693988e-06,
      "loss": 0.1739,
      "step": 347
    },
    {
      "epoch": 0.6192170818505338,
      "grad_norm": 1.6179820531330302,
      "learning_rate": 7.815500501648654e-06,
      "loss": 0.1026,
      "step": 348
    },
    {
      "epoch": 0.6209964412811388,
      "grad_norm": 1.7337222934014782,
      "learning_rate": 7.80394069068015e-06,
      "loss": 0.125,
      "step": 349
    },
    {
      "epoch": 0.6227758007117438,
      "grad_norm": 1.667295539855105,
      "learning_rate": 7.79235897509468e-06,
      "loss": 0.1177,
      "step": 350
    },
    {
      "epoch": 0.6245551601423488,
      "grad_norm": 1.955546157211665,
      "learning_rate": 7.780755445369563e-06,
      "loss": 0.1282,
      "step": 351
    },
    {
      "epoch": 0.6263345195729537,
      "grad_norm": 1.9462021045750202,
      "learning_rate": 7.769130192152538e-06,
      "loss": 0.1451,
      "step": 352
    },
    {
      "epoch": 0.6281138790035588,
      "grad_norm": 1.800382141356206,
      "learning_rate": 7.757483306261042e-06,
      "loss": 0.1343,
      "step": 353
    },
    {
      "epoch": 0.6298932384341637,
      "grad_norm": 2.26985913926227,
      "learning_rate": 7.745814878681516e-06,
      "loss": 0.1489,
      "step": 354
    },
    {
      "epoch": 0.6316725978647687,
      "grad_norm": 1.799130259991671,
      "learning_rate": 7.734125000568684e-06,
      "loss": 0.124,
      "step": 355
    },
    {
      "epoch": 0.6334519572953736,
      "grad_norm": 1.7795729814205064,
      "learning_rate": 7.722413763244837e-06,
      "loss": 0.1153,
      "step": 356
    },
    {
      "epoch": 0.6352313167259787,
      "grad_norm": 1.3982291886331777,
      "learning_rate": 7.710681258199136e-06,
      "loss": 0.0891,
      "step": 357
    },
    {
      "epoch": 0.6370106761565836,
      "grad_norm": 1.8767939872776536,
      "learning_rate": 7.69892757708688e-06,
      "loss": 0.1137,
      "step": 358
    },
    {
      "epoch": 0.6387900355871886,
      "grad_norm": 1.7380204088883748,
      "learning_rate": 7.687152811728799e-06,
      "loss": 0.1225,
      "step": 359
    },
    {
      "epoch": 0.6405693950177936,
      "grad_norm": 1.8623478448351884,
      "learning_rate": 7.675357054110337e-06,
      "loss": 0.1296,
      "step": 360
    },
    {
      "epoch": 0.6423487544483986,
      "grad_norm": 1.6357070391019277,
      "learning_rate": 7.663540396380931e-06,
      "loss": 0.1163,
      "step": 361
    },
    {
      "epoch": 0.6441281138790036,
      "grad_norm": 1.613704121919402,
      "learning_rate": 7.651702930853287e-06,
      "loss": 0.1066,
      "step": 362
    },
    {
      "epoch": 0.6459074733096085,
      "grad_norm": 1.804431133157121,
      "learning_rate": 7.639844750002668e-06,
      "loss": 0.1176,
      "step": 363
    },
    {
      "epoch": 0.6476868327402135,
      "grad_norm": 1.575922045491728,
      "learning_rate": 7.627965946466167e-06,
      "loss": 0.1354,
      "step": 364
    },
    {
      "epoch": 0.6494661921708185,
      "grad_norm": 1.7194160828765315,
      "learning_rate": 7.616066613041977e-06,
      "loss": 0.1213,
      "step": 365
    },
    {
      "epoch": 0.6512455516014235,
      "grad_norm": 1.7672637643995626,
      "learning_rate": 7.6041468426886785e-06,
      "loss": 0.1265,
      "step": 366
    },
    {
      "epoch": 0.6530249110320284,
      "grad_norm": 1.2188856249208775,
      "learning_rate": 7.592206728524507e-06,
      "loss": 0.0782,
      "step": 367
    },
    {
      "epoch": 0.6548042704626335,
      "grad_norm": 1.6574272174437377,
      "learning_rate": 7.580246363826621e-06,
      "loss": 0.119,
      "step": 368
    },
    {
      "epoch": 0.6565836298932385,
      "grad_norm": 1.8707252165943433,
      "learning_rate": 7.568265842030381e-06,
      "loss": 0.1299,
      "step": 369
    },
    {
      "epoch": 0.6583629893238434,
      "grad_norm": 1.8017615193515297,
      "learning_rate": 7.556265256728618e-06,
      "loss": 0.133,
      "step": 370
    },
    {
      "epoch": 0.6601423487544484,
      "grad_norm": 1.5559132921238767,
      "learning_rate": 7.544244701670894e-06,
      "loss": 0.1121,
      "step": 371
    },
    {
      "epoch": 0.6619217081850534,
      "grad_norm": 1.6450753225373973,
      "learning_rate": 7.532204270762786e-06,
      "loss": 0.1179,
      "step": 372
    },
    {
      "epoch": 0.6637010676156584,
      "grad_norm": 1.4763378163352459,
      "learning_rate": 7.520144058065133e-06,
      "loss": 0.0935,
      "step": 373
    },
    {
      "epoch": 0.6654804270462633,
      "grad_norm": 1.5678797250102572,
      "learning_rate": 7.50806415779332e-06,
      "loss": 0.0986,
      "step": 374
    },
    {
      "epoch": 0.6672597864768683,
      "grad_norm": 1.8100227241796532,
      "learning_rate": 7.495964664316525e-06,
      "loss": 0.1228,
      "step": 375
    },
    {
      "epoch": 0.6690391459074733,
      "grad_norm": 1.803251843053676,
      "learning_rate": 7.4838456721569975e-06,
      "loss": 0.126,
      "step": 376
    },
    {
      "epoch": 0.6708185053380783,
      "grad_norm": 1.5683511146446225,
      "learning_rate": 7.471707275989304e-06,
      "loss": 0.1094,
      "step": 377
    },
    {
      "epoch": 0.6725978647686833,
      "grad_norm": 1.5339313950162716,
      "learning_rate": 7.459549570639602e-06,
      "loss": 0.1076,
      "step": 378
    },
    {
      "epoch": 0.6743772241992882,
      "grad_norm": 1.7413818738133184,
      "learning_rate": 7.447372651084896e-06,
      "loss": 0.1125,
      "step": 379
    },
    {
      "epoch": 0.6761565836298933,
      "grad_norm": 1.658372874930133,
      "learning_rate": 7.435176612452286e-06,
      "loss": 0.1035,
      "step": 380
    },
    {
      "epoch": 0.6779359430604982,
      "grad_norm": 1.6938524137294504,
      "learning_rate": 7.4229615500182396e-06,
      "loss": 0.1205,
      "step": 381
    },
    {
      "epoch": 0.6797153024911032,
      "grad_norm": 1.3779729634980662,
      "learning_rate": 7.4107275592078345e-06,
      "loss": 0.0919,
      "step": 382
    },
    {
      "epoch": 0.6814946619217082,
      "grad_norm": 1.5361671194524793,
      "learning_rate": 7.398474735594022e-06,
      "loss": 0.0935,
      "step": 383
    },
    {
      "epoch": 0.6832740213523132,
      "grad_norm": 1.7830401687214017,
      "learning_rate": 7.386203174896872e-06,
      "loss": 0.119,
      "step": 384
    },
    {
      "epoch": 0.6850533807829181,
      "grad_norm": 1.643113393999474,
      "learning_rate": 7.373912972982838e-06,
      "loss": 0.1127,
      "step": 385
    },
    {
      "epoch": 0.6868327402135231,
      "grad_norm": 1.6146921438882946,
      "learning_rate": 7.361604225863992e-06,
      "loss": 0.111,
      "step": 386
    },
    {
      "epoch": 0.6886120996441281,
      "grad_norm": 2.038729594815312,
      "learning_rate": 7.349277029697287e-06,
      "loss": 0.1229,
      "step": 387
    },
    {
      "epoch": 0.6903914590747331,
      "grad_norm": 1.8394395989354648,
      "learning_rate": 7.336931480783801e-06,
      "loss": 0.1135,
      "step": 388
    },
    {
      "epoch": 0.6921708185053381,
      "grad_norm": 2.0657047204677257,
      "learning_rate": 7.3245676755679854e-06,
      "loss": 0.1371,
      "step": 389
    },
    {
      "epoch": 0.693950177935943,
      "grad_norm": 1.8046034252430407,
      "learning_rate": 7.312185710636911e-06,
      "loss": 0.1303,
      "step": 390
    },
    {
      "epoch": 0.6957295373665481,
      "grad_norm": 1.605467186751234,
      "learning_rate": 7.299785682719512e-06,
      "loss": 0.0894,
      "step": 391
    },
    {
      "epoch": 0.697508896797153,
      "grad_norm": 1.7075429581097383,
      "learning_rate": 7.287367688685835e-06,
      "loss": 0.1116,
      "step": 392
    },
    {
      "epoch": 0.699288256227758,
      "grad_norm": 2.32565431520281,
      "learning_rate": 7.274931825546279e-06,
      "loss": 0.1418,
      "step": 393
    },
    {
      "epoch": 0.701067615658363,
      "grad_norm": 1.5879770982216497,
      "learning_rate": 7.262478190450834e-06,
      "loss": 0.1082,
      "step": 394
    },
    {
      "epoch": 0.702846975088968,
      "grad_norm": 1.9778471493382452,
      "learning_rate": 7.250006880688332e-06,
      "loss": 0.1339,
      "step": 395
    },
    {
      "epoch": 0.7046263345195729,
      "grad_norm": 1.8536563807341508,
      "learning_rate": 7.2375179936856775e-06,
      "loss": 0.1191,
      "step": 396
    },
    {
      "epoch": 0.7064056939501779,
      "grad_norm": 1.5238892089449574,
      "learning_rate": 7.22501162700709e-06,
      "loss": 0.1168,
      "step": 397
    },
    {
      "epoch": 0.708185053380783,
      "grad_norm": 1.5633979883321871,
      "learning_rate": 7.21248787835334e-06,
      "loss": 0.1034,
      "step": 398
    },
    {
      "epoch": 0.7099644128113879,
      "grad_norm": 1.4512875858323726,
      "learning_rate": 7.199946845560994e-06,
      "loss": 0.0983,
      "step": 399
    },
    {
      "epoch": 0.7117437722419929,
      "grad_norm": 1.9721738682199543,
      "learning_rate": 7.1873886266016365e-06,
      "loss": 0.1648,
      "step": 400
    },
    {
      "epoch": 0.7117437722419929,
      "eval_loss": 0.12959618866443634,
      "eval_runtime": 1.8938,
      "eval_samples_per_second": 24.29,
      "eval_steps_per_second": 6.337,
      "step": 400
    },
    {
      "epoch": 0.7135231316725978,
      "grad_norm": 1.650786272208085,
      "learning_rate": 7.174813319581115e-06,
      "loss": 0.1312,
      "step": 401
    },
    {
      "epoch": 0.7153024911032029,
      "grad_norm": 2.2854099811118678,
      "learning_rate": 7.162221022738768e-06,
      "loss": 0.1264,
      "step": 402
    },
    {
      "epoch": 0.7170818505338078,
      "grad_norm": 1.8413188198287198,
      "learning_rate": 7.149611834446664e-06,
      "loss": 0.1443,
      "step": 403
    },
    {
      "epoch": 0.7188612099644128,
      "grad_norm": 1.8316665572391215,
      "learning_rate": 7.136985853208824e-06,
      "loss": 0.1269,
      "step": 404
    },
    {
      "epoch": 0.7206405693950177,
      "grad_norm": 1.7696567334329205,
      "learning_rate": 7.124343177660462e-06,
      "loss": 0.1233,
      "step": 405
    },
    {
      "epoch": 0.7224199288256228,
      "grad_norm": 2.0410226214676617,
      "learning_rate": 7.111683906567206e-06,
      "loss": 0.1421,
      "step": 406
    },
    {
      "epoch": 0.7241992882562278,
      "grad_norm": 1.998853048286752,
      "learning_rate": 7.099008138824329e-06,
      "loss": 0.1165,
      "step": 407
    },
    {
      "epoch": 0.7259786476868327,
      "grad_norm": 1.4819468358435748,
      "learning_rate": 7.086315973455982e-06,
      "loss": 0.111,
      "step": 408
    },
    {
      "epoch": 0.7277580071174378,
      "grad_norm": 1.8351818346938322,
      "learning_rate": 7.0736075096144084e-06,
      "loss": 0.1334,
      "step": 409
    },
    {
      "epoch": 0.7295373665480427,
      "grad_norm": 1.4198255763046972,
      "learning_rate": 7.060882846579182e-06,
      "loss": 0.0995,
      "step": 410
    },
    {
      "epoch": 0.7313167259786477,
      "grad_norm": 1.600917979892676,
      "learning_rate": 7.048142083756427e-06,
      "loss": 0.1228,
      "step": 411
    },
    {
      "epoch": 0.7330960854092526,
      "grad_norm": 1.90864411785868,
      "learning_rate": 7.035385320678035e-06,
      "loss": 0.1269,
      "step": 412
    },
    {
      "epoch": 0.7348754448398577,
      "grad_norm": 1.6237496523624368,
      "learning_rate": 7.022612657000898e-06,
      "loss": 0.1178,
      "step": 413
    },
    {
      "epoch": 0.7366548042704626,
      "grad_norm": 1.6251548811313257,
      "learning_rate": 7.0098241925061215e-06,
      "loss": 0.1239,
      "step": 414
    },
    {
      "epoch": 0.7384341637010676,
      "grad_norm": 1.822242923118626,
      "learning_rate": 6.997020027098249e-06,
      "loss": 0.1188,
      "step": 415
    },
    {
      "epoch": 0.7402135231316725,
      "grad_norm": 1.7162826090987764,
      "learning_rate": 6.9842002608044844e-06,
      "loss": 0.1277,
      "step": 416
    },
    {
      "epoch": 0.7419928825622776,
      "grad_norm": 1.610396787883105,
      "learning_rate": 6.971364993773901e-06,
      "loss": 0.1116,
      "step": 417
    },
    {
      "epoch": 0.7437722419928826,
      "grad_norm": 1.7175436316671284,
      "learning_rate": 6.958514326276669e-06,
      "loss": 0.115,
      "step": 418
    },
    {
      "epoch": 0.7455516014234875,
      "grad_norm": 1.7443156287546546,
      "learning_rate": 6.945648358703269e-06,
      "loss": 0.1147,
      "step": 419
    },
    {
      "epoch": 0.7473309608540926,
      "grad_norm": 1.5876555242050159,
      "learning_rate": 6.932767191563703e-06,
      "loss": 0.1323,
      "step": 420
    },
    {
      "epoch": 0.7491103202846975,
      "grad_norm": 1.5695174495576405,
      "learning_rate": 6.919870925486718e-06,
      "loss": 0.0915,
      "step": 421
    },
    {
      "epoch": 0.7508896797153025,
      "grad_norm": 1.3651172743631523,
      "learning_rate": 6.906959661219011e-06,
      "loss": 0.0967,
      "step": 422
    },
    {
      "epoch": 0.7526690391459074,
      "grad_norm": 1.6444213992770682,
      "learning_rate": 6.8940334996244505e-06,
      "loss": 0.118,
      "step": 423
    },
    {
      "epoch": 0.7544483985765125,
      "grad_norm": 1.5930505187965764,
      "learning_rate": 6.881092541683279e-06,
      "loss": 0.1092,
      "step": 424
    },
    {
      "epoch": 0.7562277580071174,
      "grad_norm": 1.5479644418825749,
      "learning_rate": 6.8681368884913345e-06,
      "loss": 0.0997,
      "step": 425
    },
    {
      "epoch": 0.7580071174377224,
      "grad_norm": 1.3562907458800957,
      "learning_rate": 6.855166641259252e-06,
      "loss": 0.0983,
      "step": 426
    },
    {
      "epoch": 0.7597864768683275,
      "grad_norm": 1.4937038917842138,
      "learning_rate": 6.8421819013116766e-06,
      "loss": 0.1151,
      "step": 427
    },
    {
      "epoch": 0.7615658362989324,
      "grad_norm": 1.7104421483054004,
      "learning_rate": 6.829182770086474e-06,
      "loss": 0.1279,
      "step": 428
    },
    {
      "epoch": 0.7633451957295374,
      "grad_norm": 1.7244890467183474,
      "learning_rate": 6.816169349133934e-06,
      "loss": 0.1175,
      "step": 429
    },
    {
      "epoch": 0.7651245551601423,
      "grad_norm": 1.503448338372118,
      "learning_rate": 6.803141740115979e-06,
      "loss": 0.0904,
      "step": 430
    },
    {
      "epoch": 0.7669039145907474,
      "grad_norm": 2.3929780319512255,
      "learning_rate": 6.7901000448053676e-06,
      "loss": 0.1545,
      "step": 431
    },
    {
      "epoch": 0.7686832740213523,
      "grad_norm": 1.43897184818364,
      "learning_rate": 6.777044365084907e-06,
      "loss": 0.0883,
      "step": 432
    },
    {
      "epoch": 0.7704626334519573,
      "grad_norm": 1.4530826753752788,
      "learning_rate": 6.763974802946649e-06,
      "loss": 0.1057,
      "step": 433
    },
    {
      "epoch": 0.7722419928825622,
      "grad_norm": 1.7940307230352617,
      "learning_rate": 6.750891460491093e-06,
      "loss": 0.1408,
      "step": 434
    },
    {
      "epoch": 0.7740213523131673,
      "grad_norm": 1.9240804867547232,
      "learning_rate": 6.737794439926395e-06,
      "loss": 0.1542,
      "step": 435
    },
    {
      "epoch": 0.7758007117437722,
      "grad_norm": 1.6937236315495823,
      "learning_rate": 6.724683843567567e-06,
      "loss": 0.1263,
      "step": 436
    },
    {
      "epoch": 0.7775800711743772,
      "grad_norm": 1.7995741664274278,
      "learning_rate": 6.711559773835672e-06,
      "loss": 0.1212,
      "step": 437
    },
    {
      "epoch": 0.7793594306049823,
      "grad_norm": 1.984915526739746,
      "learning_rate": 6.69842233325703e-06,
      "loss": 0.1291,
      "step": 438
    },
    {
      "epoch": 0.7811387900355872,
      "grad_norm": 2.1011550852549568,
      "learning_rate": 6.685271624462416e-06,
      "loss": 0.1444,
      "step": 439
    },
    {
      "epoch": 0.7829181494661922,
      "grad_norm": 1.49523653959074,
      "learning_rate": 6.672107750186255e-06,
      "loss": 0.1088,
      "step": 440
    },
    {
      "epoch": 0.7846975088967971,
      "grad_norm": 1.1980117468822036,
      "learning_rate": 6.658930813265825e-06,
      "loss": 0.0905,
      "step": 441
    },
    {
      "epoch": 0.7864768683274022,
      "grad_norm": 1.5409102171679387,
      "learning_rate": 6.645740916640449e-06,
      "loss": 0.1086,
      "step": 442
    },
    {
      "epoch": 0.7882562277580071,
      "grad_norm": 1.336133139289894,
      "learning_rate": 6.63253816335069e-06,
      "loss": 0.091,
      "step": 443
    },
    {
      "epoch": 0.7900355871886121,
      "grad_norm": 1.5049065226881317,
      "learning_rate": 6.619322656537552e-06,
      "loss": 0.1115,
      "step": 444
    },
    {
      "epoch": 0.791814946619217,
      "grad_norm": 1.6763895935577287,
      "learning_rate": 6.606094499441671e-06,
      "loss": 0.1126,
      "step": 445
    },
    {
      "epoch": 0.7935943060498221,
      "grad_norm": 1.701011457351025,
      "learning_rate": 6.592853795402502e-06,
      "loss": 0.1165,
      "step": 446
    },
    {
      "epoch": 0.7953736654804271,
      "grad_norm": 1.7173326356205285,
      "learning_rate": 6.579600647857525e-06,
      "loss": 0.1154,
      "step": 447
    },
    {
      "epoch": 0.797153024911032,
      "grad_norm": 2.0689784607077684,
      "learning_rate": 6.566335160341425e-06,
      "loss": 0.1571,
      "step": 448
    },
    {
      "epoch": 0.798932384341637,
      "grad_norm": 1.4893826484487036,
      "learning_rate": 6.553057436485289e-06,
      "loss": 0.1119,
      "step": 449
    },
    {
      "epoch": 0.800711743772242,
      "grad_norm": 1.4973328833983912,
      "learning_rate": 6.539767580015799e-06,
      "loss": 0.1303,
      "step": 450
    },
    {
      "epoch": 0.802491103202847,
      "grad_norm": 2.121646356038464,
      "learning_rate": 6.52646569475441e-06,
      "loss": 0.1245,
      "step": 451
    },
    {
      "epoch": 0.8042704626334519,
      "grad_norm": 1.7539319055288605,
      "learning_rate": 6.513151884616556e-06,
      "loss": 0.1421,
      "step": 452
    },
    {
      "epoch": 0.806049822064057,
      "grad_norm": 1.8502044144055405,
      "learning_rate": 6.499826253610823e-06,
      "loss": 0.1223,
      "step": 453
    },
    {
      "epoch": 0.8078291814946619,
      "grad_norm": 1.5319280024206472,
      "learning_rate": 6.486488905838143e-06,
      "loss": 0.104,
      "step": 454
    },
    {
      "epoch": 0.8096085409252669,
      "grad_norm": 1.4797150090237958,
      "learning_rate": 6.473139945490984e-06,
      "loss": 0.0916,
      "step": 455
    },
    {
      "epoch": 0.8113879003558719,
      "grad_norm": 1.662988573862244,
      "learning_rate": 6.459779476852528e-06,
      "loss": 0.1137,
      "step": 456
    },
    {
      "epoch": 0.8131672597864769,
      "grad_norm": 1.639421479562668,
      "learning_rate": 6.446407604295863e-06,
      "loss": 0.1235,
      "step": 457
    },
    {
      "epoch": 0.8149466192170819,
      "grad_norm": 1.6822467563938035,
      "learning_rate": 6.433024432283169e-06,
      "loss": 0.0969,
      "step": 458
    },
    {
      "epoch": 0.8167259786476868,
      "grad_norm": 1.8318802686757292,
      "learning_rate": 6.41963006536489e-06,
      "loss": 0.1142,
      "step": 459
    },
    {
      "epoch": 0.8185053380782918,
      "grad_norm": 1.5892619636637637,
      "learning_rate": 6.4062246081789316e-06,
      "loss": 0.116,
      "step": 460
    },
    {
      "epoch": 0.8202846975088968,
      "grad_norm": 1.4199562438109783,
      "learning_rate": 6.392808165449836e-06,
      "loss": 0.0763,
      "step": 461
    },
    {
      "epoch": 0.8220640569395018,
      "grad_norm": 1.5896441433111637,
      "learning_rate": 6.379380841987965e-06,
      "loss": 0.1022,
      "step": 462
    },
    {
      "epoch": 0.8238434163701067,
      "grad_norm": 1.4918014202567753,
      "learning_rate": 6.365942742688684e-06,
      "loss": 0.0879,
      "step": 463
    },
    {
      "epoch": 0.8256227758007118,
      "grad_norm": 2.1081867590383774,
      "learning_rate": 6.352493972531535e-06,
      "loss": 0.1398,
      "step": 464
    },
    {
      "epoch": 0.8274021352313167,
      "grad_norm": 2.236482421245564,
      "learning_rate": 6.339034636579425e-06,
      "loss": 0.1402,
      "step": 465
    },
    {
      "epoch": 0.8291814946619217,
      "grad_norm": 1.7557563854463891,
      "learning_rate": 6.325564839977802e-06,
      "loss": 0.0986,
      "step": 466
    },
    {
      "epoch": 0.8309608540925267,
      "grad_norm": 1.5462814737486743,
      "learning_rate": 6.312084687953835e-06,
      "loss": 0.0995,
      "step": 467
    },
    {
      "epoch": 0.8327402135231317,
      "grad_norm": 2.0033068933866676,
      "learning_rate": 6.298594285815585e-06,
      "loss": 0.1415,
      "step": 468
    },
    {
      "epoch": 0.8345195729537367,
      "grad_norm": 1.9131492658503364,
      "learning_rate": 6.2850937389511936e-06,
      "loss": 0.1388,
      "step": 469
    },
    {
      "epoch": 0.8362989323843416,
      "grad_norm": 1.4654295413644622,
      "learning_rate": 6.271583152828049e-06,
      "loss": 0.087,
      "step": 470
    },
    {
      "epoch": 0.8380782918149466,
      "grad_norm": 1.5753128564647159,
      "learning_rate": 6.258062632991972e-06,
      "loss": 0.0877,
      "step": 471
    },
    {
      "epoch": 0.8398576512455516,
      "grad_norm": 1.6548802244524703,
      "learning_rate": 6.244532285066382e-06,
      "loss": 0.1163,
      "step": 472
    },
    {
      "epoch": 0.8416370106761566,
      "grad_norm": 1.530428330922915,
      "learning_rate": 6.2309922147514775e-06,
      "loss": 0.1005,
      "step": 473
    },
    {
      "epoch": 0.8434163701067615,
      "grad_norm": 1.490709317766687,
      "learning_rate": 6.2174425278234115e-06,
      "loss": 0.1281,
      "step": 474
    },
    {
      "epoch": 0.8451957295373665,
      "grad_norm": 1.7010977681402695,
      "learning_rate": 6.20388333013346e-06,
      "loss": 0.0892,
      "step": 475
    },
    {
      "epoch": 0.8469750889679716,
      "grad_norm": 1.6598493946668496,
      "learning_rate": 6.190314727607196e-06,
      "loss": 0.1296,
      "step": 476
    },
    {
      "epoch": 0.8487544483985765,
      "grad_norm": 1.6584750831655197,
      "learning_rate": 6.176736826243671e-06,
      "loss": 0.1138,
      "step": 477
    },
    {
      "epoch": 0.8505338078291815,
      "grad_norm": 1.6311767798481245,
      "learning_rate": 6.163149732114571e-06,
      "loss": 0.115,
      "step": 478
    },
    {
      "epoch": 0.8523131672597865,
      "grad_norm": 1.664554587358972,
      "learning_rate": 6.149553551363404e-06,
      "loss": 0.1021,
      "step": 479
    },
    {
      "epoch": 0.8540925266903915,
      "grad_norm": 1.789706761510048,
      "learning_rate": 6.1359483902046605e-06,
      "loss": 0.1334,
      "step": 480
    },
    {
      "epoch": 0.8558718861209964,
      "grad_norm": 1.6481955793107783,
      "learning_rate": 6.122334354922984e-06,
      "loss": 0.1095,
      "step": 481
    },
    {
      "epoch": 0.8576512455516014,
      "grad_norm": 1.6933883078833007,
      "learning_rate": 6.108711551872347e-06,
      "loss": 0.1197,
      "step": 482
    },
    {
      "epoch": 0.8594306049822064,
      "grad_norm": 1.9982362806330403,
      "learning_rate": 6.095080087475218e-06,
      "loss": 0.1191,
      "step": 483
    },
    {
      "epoch": 0.8612099644128114,
      "grad_norm": 1.6279749091088431,
      "learning_rate": 6.0814400682217236e-06,
      "loss": 0.1195,
      "step": 484
    },
    {
      "epoch": 0.8629893238434164,
      "grad_norm": 2.0702920307947217,
      "learning_rate": 6.067791600668823e-06,
      "loss": 0.0989,
      "step": 485
    },
    {
      "epoch": 0.8647686832740213,
      "grad_norm": 1.366280914619679,
      "learning_rate": 6.054134791439479e-06,
      "loss": 0.0811,
      "step": 486
    },
    {
      "epoch": 0.8665480427046264,
      "grad_norm": 1.620040269330004,
      "learning_rate": 6.040469747221815e-06,
      "loss": 0.0888,
      "step": 487
    },
    {
      "epoch": 0.8683274021352313,
      "grad_norm": 1.6550005458585226,
      "learning_rate": 6.026796574768288e-06,
      "loss": 0.0949,
      "step": 488
    },
    {
      "epoch": 0.8701067615658363,
      "grad_norm": 1.8368090577979796,
      "learning_rate": 6.013115380894854e-06,
      "loss": 0.1096,
      "step": 489
    },
    {
      "epoch": 0.8718861209964412,
      "grad_norm": 1.4081492155967914,
      "learning_rate": 5.999426272480133e-06,
      "loss": 0.0864,
      "step": 490
    },
    {
      "epoch": 0.8736654804270463,
      "grad_norm": 1.5439311233007154,
      "learning_rate": 5.985729356464575e-06,
      "loss": 0.1144,
      "step": 491
    },
    {
      "epoch": 0.8754448398576512,
      "grad_norm": 1.6390509124621109,
      "learning_rate": 5.972024739849622e-06,
      "loss": 0.1086,
      "step": 492
    },
    {
      "epoch": 0.8772241992882562,
      "grad_norm": 1.1804966375226666,
      "learning_rate": 5.958312529696874e-06,
      "loss": 0.0789,
      "step": 493
    },
    {
      "epoch": 0.8790035587188612,
      "grad_norm": 1.5359055122663885,
      "learning_rate": 5.944592833127253e-06,
      "loss": 0.1306,
      "step": 494
    },
    {
      "epoch": 0.8807829181494662,
      "grad_norm": 1.701310220891771,
      "learning_rate": 5.9308657573201645e-06,
      "loss": 0.1385,
      "step": 495
    },
    {
      "epoch": 0.8825622775800712,
      "grad_norm": 1.7232981407665273,
      "learning_rate": 5.917131409512663e-06,
      "loss": 0.1038,
      "step": 496
    },
    {
      "epoch": 0.8843416370106761,
      "grad_norm": 1.5134955064921805,
      "learning_rate": 5.903389896998611e-06,
      "loss": 0.112,
      "step": 497
    },
    {
      "epoch": 0.8861209964412812,
      "grad_norm": 1.7101215835872345,
      "learning_rate": 5.889641327127843e-06,
      "loss": 0.1076,
      "step": 498
    },
    {
      "epoch": 0.8879003558718861,
      "grad_norm": 1.5969144507758808,
      "learning_rate": 5.875885807305326e-06,
      "loss": 0.1127,
      "step": 499
    },
    {
      "epoch": 0.8896797153024911,
      "grad_norm": 1.5293635939910035,
      "learning_rate": 5.862123444990319e-06,
      "loss": 0.102,
      "step": 500
    },
    {
      "epoch": 0.891459074733096,
      "grad_norm": 1.9129248797631837,
      "learning_rate": 5.848354347695537e-06,
      "loss": 0.1657,
      "step": 501
    },
    {
      "epoch": 0.8932384341637011,
      "grad_norm": 1.779995432700007,
      "learning_rate": 5.83457862298631e-06,
      "loss": 0.1301,
      "step": 502
    },
    {
      "epoch": 0.895017793594306,
      "grad_norm": 1.9085016153186696,
      "learning_rate": 5.8207963784797396e-06,
      "loss": 0.1237,
      "step": 503
    },
    {
      "epoch": 0.896797153024911,
      "grad_norm": 1.6577275898251052,
      "learning_rate": 5.807007721843862e-06,
      "loss": 0.1224,
      "step": 504
    },
    {
      "epoch": 0.8985765124555161,
      "grad_norm": 1.9893076829870855,
      "learning_rate": 5.793212760796804e-06,
      "loss": 0.1478,
      "step": 505
    },
    {
      "epoch": 0.900355871886121,
      "grad_norm": 1.8021286071340172,
      "learning_rate": 5.779411603105947e-06,
      "loss": 0.1319,
      "step": 506
    },
    {
      "epoch": 0.902135231316726,
      "grad_norm": 1.6187095597458239,
      "learning_rate": 5.765604356587076e-06,
      "loss": 0.1136,
      "step": 507
    },
    {
      "epoch": 0.9039145907473309,
      "grad_norm": 1.9304740865856787,
      "learning_rate": 5.751791129103545e-06,
      "loss": 0.1244,
      "step": 508
    },
    {
      "epoch": 0.905693950177936,
      "grad_norm": 1.9072697594835601,
      "learning_rate": 5.737972028565431e-06,
      "loss": 0.1423,
      "step": 509
    },
    {
      "epoch": 0.9074733096085409,
      "grad_norm": 1.7973013201222665,
      "learning_rate": 5.7241471629286934e-06,
      "loss": 0.1263,
      "step": 510
    },
    {
      "epoch": 0.9092526690391459,
      "grad_norm": 2.0208042621710183,
      "learning_rate": 5.7103166401943276e-06,
      "loss": 0.1315,
      "step": 511
    },
    {
      "epoch": 0.9110320284697508,
      "grad_norm": 1.7789311400391496,
      "learning_rate": 5.696480568407523e-06,
      "loss": 0.112,
      "step": 512
    },
    {
      "epoch": 0.9128113879003559,
      "grad_norm": 1.9380071988357526,
      "learning_rate": 5.682639055656817e-06,
      "loss": 0.1407,
      "step": 513
    },
    {
      "epoch": 0.9145907473309609,
      "grad_norm": 1.7941071122408592,
      "learning_rate": 5.668792210073255e-06,
      "loss": 0.1419,
      "step": 514
    },
    {
      "epoch": 0.9163701067615658,
      "grad_norm": 1.782245032916222,
      "learning_rate": 5.654940139829544e-06,
      "loss": 0.1222,
      "step": 515
    },
    {
      "epoch": 0.9181494661921709,
      "grad_norm": 1.555276532221455,
      "learning_rate": 5.641082953139201e-06,
      "loss": 0.1048,
      "step": 516
    },
    {
      "epoch": 0.9199288256227758,
      "grad_norm": 1.186415095228362,
      "learning_rate": 5.6272207582557195e-06,
      "loss": 0.0795,
      "step": 517
    },
    {
      "epoch": 0.9217081850533808,
      "grad_norm": 1.5599270090425488,
      "learning_rate": 5.61335366347171e-06,
      "loss": 0.0968,
      "step": 518
    },
    {
      "epoch": 0.9234875444839857,
      "grad_norm": 1.734183135234115,
      "learning_rate": 5.599481777118071e-06,
      "loss": 0.125,
      "step": 519
    },
    {
      "epoch": 0.9252669039145908,
      "grad_norm": 1.826092599683182,
      "learning_rate": 5.585605207563124e-06,
      "loss": 0.0988,
      "step": 520
    },
    {
      "epoch": 0.9270462633451957,
      "grad_norm": 1.5856036507290956,
      "learning_rate": 5.571724063211782e-06,
      "loss": 0.095,
      "step": 521
    },
    {
      "epoch": 0.9288256227758007,
      "grad_norm": 1.5692105352719896,
      "learning_rate": 5.557838452504692e-06,
      "loss": 0.0892,
      "step": 522
    },
    {
      "epoch": 0.9306049822064056,
      "grad_norm": 1.9410504415194334,
      "learning_rate": 5.5439484839173996e-06,
      "loss": 0.1264,
      "step": 523
    },
    {
      "epoch": 0.9323843416370107,
      "grad_norm": 1.573827854269837,
      "learning_rate": 5.530054265959486e-06,
      "loss": 0.1106,
      "step": 524
    },
    {
      "epoch": 0.9341637010676157,
      "grad_norm": 1.7669043670382616,
      "learning_rate": 5.516155907173735e-06,
      "loss": 0.1306,
      "step": 525
    },
    {
      "epoch": 0.9359430604982206,
      "grad_norm": 2.0723297874693345,
      "learning_rate": 5.5022535161352764e-06,
      "loss": 0.1024,
      "step": 526
    },
    {
      "epoch": 0.9377224199288257,
      "grad_norm": 1.8477927933322909,
      "learning_rate": 5.488347201450741e-06,
      "loss": 0.1155,
      "step": 527
    },
    {
      "epoch": 0.9395017793594306,
      "grad_norm": 1.4335122848552868,
      "learning_rate": 5.47443707175741e-06,
      "loss": 0.1034,
      "step": 528
    },
    {
      "epoch": 0.9412811387900356,
      "grad_norm": 1.6848783449366886,
      "learning_rate": 5.46052323572237e-06,
      "loss": 0.1035,
      "step": 529
    },
    {
      "epoch": 0.9430604982206405,
      "grad_norm": 1.5026758325433949,
      "learning_rate": 5.446605802041662e-06,
      "loss": 0.0904,
      "step": 530
    },
    {
      "epoch": 0.9448398576512456,
      "grad_norm": 1.3646921844987656,
      "learning_rate": 5.432684879439428e-06,
      "loss": 0.093,
      "step": 531
    },
    {
      "epoch": 0.9466192170818505,
      "grad_norm": 1.3970690383199675,
      "learning_rate": 5.418760576667071e-06,
      "loss": 0.0787,
      "step": 532
    },
    {
      "epoch": 0.9483985765124555,
      "grad_norm": 1.7828424411384753,
      "learning_rate": 5.404833002502398e-06,
      "loss": 0.135,
      "step": 533
    },
    {
      "epoch": 0.9501779359430605,
      "grad_norm": 1.846733318748555,
      "learning_rate": 5.39090226574877e-06,
      "loss": 0.1151,
      "step": 534
    },
    {
      "epoch": 0.9519572953736655,
      "grad_norm": 2.081735210052978,
      "learning_rate": 5.376968475234258e-06,
      "loss": 0.1481,
      "step": 535
    },
    {
      "epoch": 0.9537366548042705,
      "grad_norm": 1.2654872607931265,
      "learning_rate": 5.363031739810787e-06,
      "loss": 0.0769,
      "step": 536
    },
    {
      "epoch": 0.9555160142348754,
      "grad_norm": 1.7040244513445832,
      "learning_rate": 5.349092168353291e-06,
      "loss": 0.125,
      "step": 537
    },
    {
      "epoch": 0.9572953736654805,
      "grad_norm": 2.2736701514547386,
      "learning_rate": 5.335149869758855e-06,
      "loss": 0.1011,
      "step": 538
    },
    {
      "epoch": 0.9590747330960854,
      "grad_norm": 1.8704870222474606,
      "learning_rate": 5.32120495294587e-06,
      "loss": 0.1243,
      "step": 539
    },
    {
      "epoch": 0.9608540925266904,
      "grad_norm": 1.5757509714900944,
      "learning_rate": 5.3072575268531835e-06,
      "loss": 0.1296,
      "step": 540
    },
    {
      "epoch": 0.9626334519572953,
      "grad_norm": 1.9597413650086275,
      "learning_rate": 5.293307700439242e-06,
      "loss": 0.1315,
      "step": 541
    },
    {
      "epoch": 0.9644128113879004,
      "grad_norm": 1.9584643962285975,
      "learning_rate": 5.2793555826812456e-06,
      "loss": 0.1242,
      "step": 542
    },
    {
      "epoch": 0.9661921708185054,
      "grad_norm": 1.6703671916927374,
      "learning_rate": 5.265401282574294e-06,
      "loss": 0.0931,
      "step": 543
    },
    {
      "epoch": 0.9679715302491103,
      "grad_norm": 1.7775044285702417,
      "learning_rate": 5.2514449091305375e-06,
      "loss": 0.1127,
      "step": 544
    },
    {
      "epoch": 0.9697508896797153,
      "grad_norm": 1.900198783977832,
      "learning_rate": 5.237486571378317e-06,
      "loss": 0.122,
      "step": 545
    },
    {
      "epoch": 0.9715302491103203,
      "grad_norm": 1.6252752512728061,
      "learning_rate": 5.22352637836133e-06,
      "loss": 0.1395,
      "step": 546
    },
    {
      "epoch": 0.9733096085409253,
      "grad_norm": 1.699565228191691,
      "learning_rate": 5.209564439137755e-06,
      "loss": 0.1106,
      "step": 547
    },
    {
      "epoch": 0.9750889679715302,
      "grad_norm": 1.7727298452584392,
      "learning_rate": 5.195600862779421e-06,
      "loss": 0.1455,
      "step": 548
    },
    {
      "epoch": 0.9768683274021353,
      "grad_norm": 1.1591710836347968,
      "learning_rate": 5.181635758370942e-06,
      "loss": 0.0672,
      "step": 549
    },
    {
      "epoch": 0.9786476868327402,
      "grad_norm": 1.8120422070928126,
      "learning_rate": 5.167669235008871e-06,
      "loss": 0.1278,
      "step": 550
    },
    {
      "epoch": 0.9804270462633452,
      "grad_norm": 2.091801828124558,
      "learning_rate": 5.153701401800845e-06,
      "loss": 0.1506,
      "step": 551
    },
    {
      "epoch": 0.9822064056939501,
      "grad_norm": 1.6050350521689105,
      "learning_rate": 5.139732367864736e-06,
      "loss": 0.1084,
      "step": 552
    },
    {
      "epoch": 0.9839857651245552,
      "grad_norm": 1.4294153326893941,
      "learning_rate": 5.1257622423277934e-06,
      "loss": 0.0976,
      "step": 553
    },
    {
      "epoch": 0.9857651245551602,
      "grad_norm": 1.587042660396233,
      "learning_rate": 5.111791134325793e-06,
      "loss": 0.1118,
      "step": 554
    },
    {
      "epoch": 0.9875444839857651,
      "grad_norm": 1.35922627926222,
      "learning_rate": 5.097819153002192e-06,
      "loss": 0.0976,
      "step": 555
    },
    {
      "epoch": 0.9893238434163701,
      "grad_norm": 1.3856363204998594,
      "learning_rate": 5.083846407507263e-06,
      "loss": 0.0921,
      "step": 556
    },
    {
      "epoch": 0.9911032028469751,
      "grad_norm": 1.9578614122627025,
      "learning_rate": 5.0698730069972535e-06,
      "loss": 0.138,
      "step": 557
    },
    {
      "epoch": 0.9928825622775801,
      "grad_norm": 1.4729540412888569,
      "learning_rate": 5.055899060633524e-06,
      "loss": 0.0993,
      "step": 558
    },
    {
      "epoch": 0.994661921708185,
      "grad_norm": 1.4805535422577267,
      "learning_rate": 5.041924677581702e-06,
      "loss": 0.1191,
      "step": 559
    },
    {
      "epoch": 0.99644128113879,
      "grad_norm": 1.5224552225206658,
      "learning_rate": 5.0279499670108245e-06,
      "loss": 0.1125,
      "step": 560
    },
    {
      "epoch": 0.998220640569395,
      "grad_norm": 1.5788176034634458,
      "learning_rate": 5.013975038092491e-06,
      "loss": 0.1103,
      "step": 561
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.289171803628575,
      "learning_rate": 5e-06,
      "loss": 0.0774,
      "step": 562
    },
    {
      "epoch": 1.001779359430605,
      "grad_norm": 1.321547667551879,
      "learning_rate": 4.98602496190751e-06,
      "loss": 0.0648,
      "step": 563
    },
    {
      "epoch": 1.00355871886121,
      "grad_norm": 1.0341965320750677,
      "learning_rate": 4.9720500329891755e-06,
      "loss": 0.0473,
      "step": 564
    },
    {
      "epoch": 1.0053380782918149,
      "grad_norm": 1.3887310631222791,
      "learning_rate": 4.9580753224183005e-06,
      "loss": 0.066,
      "step": 565
    },
    {
      "epoch": 1.00711743772242,
      "grad_norm": 1.0190005857961302,
      "learning_rate": 4.944100939366478e-06,
      "loss": 0.0458,
      "step": 566
    },
    {
      "epoch": 1.008896797153025,
      "grad_norm": 0.8996172927014492,
      "learning_rate": 4.930126993002748e-06,
      "loss": 0.0442,
      "step": 567
    },
    {
      "epoch": 1.01067615658363,
      "grad_norm": 1.097170668718373,
      "learning_rate": 4.9161535924927375e-06,
      "loss": 0.0503,
      "step": 568
    },
    {
      "epoch": 1.0124555160142348,
      "grad_norm": 0.9670152388080503,
      "learning_rate": 4.90218084699781e-06,
      "loss": 0.0371,
      "step": 569
    },
    {
      "epoch": 1.0142348754448398,
      "grad_norm": 1.40278326941056,
      "learning_rate": 4.888208865674208e-06,
      "loss": 0.0642,
      "step": 570
    },
    {
      "epoch": 1.0160142348754448,
      "grad_norm": 1.1623749611062457,
      "learning_rate": 4.874237757672209e-06,
      "loss": 0.0518,
      "step": 571
    },
    {
      "epoch": 1.0177935943060499,
      "grad_norm": 1.6036458955615824,
      "learning_rate": 4.8602676321352646e-06,
      "loss": 0.0773,
      "step": 572
    },
    {
      "epoch": 1.019572953736655,
      "grad_norm": 1.365741105836663,
      "learning_rate": 4.846298598199155e-06,
      "loss": 0.0503,
      "step": 573
    },
    {
      "epoch": 1.0213523131672597,
      "grad_norm": 1.3357812339388766,
      "learning_rate": 4.832330764991131e-06,
      "loss": 0.0446,
      "step": 574
    },
    {
      "epoch": 1.0231316725978647,
      "grad_norm": 1.4334544851379145,
      "learning_rate": 4.81836424162906e-06,
      "loss": 0.0576,
      "step": 575
    },
    {
      "epoch": 1.0249110320284698,
      "grad_norm": 1.3829268803494221,
      "learning_rate": 4.80439913722058e-06,
      "loss": 0.0418,
      "step": 576
    },
    {
      "epoch": 1.0266903914590748,
      "grad_norm": 1.2483367813256785,
      "learning_rate": 4.790435560862247e-06,
      "loss": 0.0449,
      "step": 577
    },
    {
      "epoch": 1.0284697508896796,
      "grad_norm": 1.695500284079523,
      "learning_rate": 4.776473621638673e-06,
      "loss": 0.0435,
      "step": 578
    },
    {
      "epoch": 1.0302491103202847,
      "grad_norm": 1.8991173686037819,
      "learning_rate": 4.762513428621684e-06,
      "loss": 0.058,
      "step": 579
    },
    {
      "epoch": 1.0320284697508897,
      "grad_norm": 1.5521630263117991,
      "learning_rate": 4.748555090869464e-06,
      "loss": 0.0438,
      "step": 580
    },
    {
      "epoch": 1.0338078291814947,
      "grad_norm": 1.3477409399835327,
      "learning_rate": 4.734598717425706e-06,
      "loss": 0.0448,
      "step": 581
    },
    {
      "epoch": 1.0355871886120998,
      "grad_norm": 1.7106433757366346,
      "learning_rate": 4.720644417318755e-06,
      "loss": 0.0635,
      "step": 582
    },
    {
      "epoch": 1.0373665480427046,
      "grad_norm": 2.851090566805398,
      "learning_rate": 4.70669229956076e-06,
      "loss": 0.0709,
      "step": 583
    },
    {
      "epoch": 1.0391459074733096,
      "grad_norm": 1.5953113364569276,
      "learning_rate": 4.692742473146818e-06,
      "loss": 0.0477,
      "step": 584
    },
    {
      "epoch": 1.0409252669039146,
      "grad_norm": 1.4761057849508938,
      "learning_rate": 4.678795047054131e-06,
      "loss": 0.0522,
      "step": 585
    },
    {
      "epoch": 1.0427046263345197,
      "grad_norm": 1.6559294062055907,
      "learning_rate": 4.664850130241146e-06,
      "loss": 0.0413,
      "step": 586
    },
    {
      "epoch": 1.0444839857651245,
      "grad_norm": 1.4698096409217896,
      "learning_rate": 4.650907831646711e-06,
      "loss": 0.0437,
      "step": 587
    },
    {
      "epoch": 1.0462633451957295,
      "grad_norm": 1.6443599758137117,
      "learning_rate": 4.636968260189214e-06,
      "loss": 0.0701,
      "step": 588
    },
    {
      "epoch": 1.0480427046263345,
      "grad_norm": 1.5694553791769974,
      "learning_rate": 4.623031524765744e-06,
      "loss": 0.0429,
      "step": 589
    },
    {
      "epoch": 1.0498220640569396,
      "grad_norm": 1.4177615456583517,
      "learning_rate": 4.609097734251231e-06,
      "loss": 0.0431,
      "step": 590
    },
    {
      "epoch": 1.0516014234875444,
      "grad_norm": 1.8222158396214463,
      "learning_rate": 4.595166997497605e-06,
      "loss": 0.0469,
      "step": 591
    },
    {
      "epoch": 1.0533807829181494,
      "grad_norm": 1.595194163281614,
      "learning_rate": 4.58123942333293e-06,
      "loss": 0.0531,
      "step": 592
    },
    {
      "epoch": 1.0551601423487544,
      "grad_norm": 1.2829348617654672,
      "learning_rate": 4.567315120560573e-06,
      "loss": 0.0398,
      "step": 593
    },
    {
      "epoch": 1.0569395017793595,
      "grad_norm": 2.5773510148852994,
      "learning_rate": 4.553394197958339e-06,
      "loss": 0.054,
      "step": 594
    },
    {
      "epoch": 1.0587188612099645,
      "grad_norm": 1.3295098546028743,
      "learning_rate": 4.539476764277631e-06,
      "loss": 0.0375,
      "step": 595
    },
    {
      "epoch": 1.0604982206405693,
      "grad_norm": 1.1136857786265548,
      "learning_rate": 4.525562928242592e-06,
      "loss": 0.0314,
      "step": 596
    },
    {
      "epoch": 1.0622775800711743,
      "grad_norm": 1.8457059181272801,
      "learning_rate": 4.511652798549261e-06,
      "loss": 0.0571,
      "step": 597
    },
    {
      "epoch": 1.0640569395017794,
      "grad_norm": 1.754973123612851,
      "learning_rate": 4.497746483864725e-06,
      "loss": 0.0527,
      "step": 598
    },
    {
      "epoch": 1.0658362989323844,
      "grad_norm": 1.4884897747797308,
      "learning_rate": 4.483844092826267e-06,
      "loss": 0.034,
      "step": 599
    },
    {
      "epoch": 1.0676156583629894,
      "grad_norm": 1.1921017691419724,
      "learning_rate": 4.469945734040516e-06,
      "loss": 0.0422,
      "step": 600
    },
    {
      "epoch": 1.0676156583629894,
      "eval_loss": 0.14088106155395508,
      "eval_runtime": 1.9014,
      "eval_samples_per_second": 24.193,
      "eval_steps_per_second": 6.311,
      "step": 600
    },
    {
      "epoch": 1.0693950177935942,
      "grad_norm": 1.4520627009161615,
      "learning_rate": 4.456051516082603e-06,
      "loss": 0.0446,
      "step": 601
    },
    {
      "epoch": 1.0711743772241993,
      "grad_norm": 2.065566049470662,
      "learning_rate": 4.442161547495309e-06,
      "loss": 0.0636,
      "step": 602
    },
    {
      "epoch": 1.0729537366548043,
      "grad_norm": 1.8731193389363843,
      "learning_rate": 4.42827593678822e-06,
      "loss": 0.0541,
      "step": 603
    },
    {
      "epoch": 1.0747330960854093,
      "grad_norm": 1.5421526602172757,
      "learning_rate": 4.414394792436877e-06,
      "loss": 0.0469,
      "step": 604
    },
    {
      "epoch": 1.0765124555160142,
      "grad_norm": 1.5147767598172146,
      "learning_rate": 4.400518222881931e-06,
      "loss": 0.0584,
      "step": 605
    },
    {
      "epoch": 1.0782918149466192,
      "grad_norm": 1.4569848874928137,
      "learning_rate": 4.386646336528291e-06,
      "loss": 0.0471,
      "step": 606
    },
    {
      "epoch": 1.0800711743772242,
      "grad_norm": 1.2472822600183657,
      "learning_rate": 4.372779241744282e-06,
      "loss": 0.0369,
      "step": 607
    },
    {
      "epoch": 1.0818505338078293,
      "grad_norm": 1.4365896007681718,
      "learning_rate": 4.358917046860799e-06,
      "loss": 0.0467,
      "step": 608
    },
    {
      "epoch": 1.083629893238434,
      "grad_norm": 1.401699727521943,
      "learning_rate": 4.345059860170458e-06,
      "loss": 0.0454,
      "step": 609
    },
    {
      "epoch": 1.085409252669039,
      "grad_norm": 1.5809770179560945,
      "learning_rate": 4.331207789926746e-06,
      "loss": 0.0463,
      "step": 610
    },
    {
      "epoch": 1.0871886120996441,
      "grad_norm": 1.3652320905173538,
      "learning_rate": 4.317360944343184e-06,
      "loss": 0.0394,
      "step": 611
    },
    {
      "epoch": 1.0889679715302492,
      "grad_norm": 1.0954675718060185,
      "learning_rate": 4.303519431592479e-06,
      "loss": 0.0319,
      "step": 612
    },
    {
      "epoch": 1.0907473309608542,
      "grad_norm": 1.8123638898111263,
      "learning_rate": 4.289683359805673e-06,
      "loss": 0.0676,
      "step": 613
    },
    {
      "epoch": 1.092526690391459,
      "grad_norm": 1.5165089008513655,
      "learning_rate": 4.275852837071309e-06,
      "loss": 0.036,
      "step": 614
    },
    {
      "epoch": 1.094306049822064,
      "grad_norm": 1.349530319256878,
      "learning_rate": 4.26202797143457e-06,
      "loss": 0.0418,
      "step": 615
    },
    {
      "epoch": 1.096085409252669,
      "grad_norm": 1.3926375697865283,
      "learning_rate": 4.248208870896456e-06,
      "loss": 0.0398,
      "step": 616
    },
    {
      "epoch": 1.097864768683274,
      "grad_norm": 1.6140068888073313,
      "learning_rate": 4.234395643412925e-06,
      "loss": 0.0528,
      "step": 617
    },
    {
      "epoch": 1.099644128113879,
      "grad_norm": 1.4936700700182268,
      "learning_rate": 4.220588396894055e-06,
      "loss": 0.0449,
      "step": 618
    },
    {
      "epoch": 1.101423487544484,
      "grad_norm": 1.605327651042368,
      "learning_rate": 4.2067872392031965e-06,
      "loss": 0.046,
      "step": 619
    },
    {
      "epoch": 1.103202846975089,
      "grad_norm": 1.6306908072739312,
      "learning_rate": 4.192992278156141e-06,
      "loss": 0.0494,
      "step": 620
    },
    {
      "epoch": 1.104982206405694,
      "grad_norm": 1.5710135845017419,
      "learning_rate": 4.179203621520262e-06,
      "loss": 0.0519,
      "step": 621
    },
    {
      "epoch": 1.106761565836299,
      "grad_norm": 1.5232535291378393,
      "learning_rate": 4.165421377013691e-06,
      "loss": 0.0374,
      "step": 622
    },
    {
      "epoch": 1.1085409252669038,
      "grad_norm": 1.3194104666712834,
      "learning_rate": 4.151645652304465e-06,
      "loss": 0.0403,
      "step": 623
    },
    {
      "epoch": 1.1103202846975089,
      "grad_norm": 1.7240847723214994,
      "learning_rate": 4.137876555009684e-06,
      "loss": 0.0615,
      "step": 624
    },
    {
      "epoch": 1.112099644128114,
      "grad_norm": 1.7096274273606844,
      "learning_rate": 4.124114192694676e-06,
      "loss": 0.0474,
      "step": 625
    },
    {
      "epoch": 1.113879003558719,
      "grad_norm": 1.5625773874625453,
      "learning_rate": 4.110358672872158e-06,
      "loss": 0.0504,
      "step": 626
    },
    {
      "epoch": 1.1156583629893237,
      "grad_norm": 1.396076023321496,
      "learning_rate": 4.0966101030013915e-06,
      "loss": 0.0479,
      "step": 627
    },
    {
      "epoch": 1.1174377224199288,
      "grad_norm": 1.280866591548747,
      "learning_rate": 4.082868590487339e-06,
      "loss": 0.0433,
      "step": 628
    },
    {
      "epoch": 1.1192170818505338,
      "grad_norm": 0.9376285586247859,
      "learning_rate": 4.069134242679837e-06,
      "loss": 0.0285,
      "step": 629
    },
    {
      "epoch": 1.1209964412811388,
      "grad_norm": 1.9861569745921686,
      "learning_rate": 4.055407166872748e-06,
      "loss": 0.0504,
      "step": 630
    },
    {
      "epoch": 1.1227758007117439,
      "grad_norm": 1.2950339400408617,
      "learning_rate": 4.041687470303127e-06,
      "loss": 0.0498,
      "step": 631
    },
    {
      "epoch": 1.1245551601423487,
      "grad_norm": 1.2224045265981505,
      "learning_rate": 4.02797526015038e-06,
      "loss": 0.0488,
      "step": 632
    },
    {
      "epoch": 1.1263345195729537,
      "grad_norm": 1.415136249691723,
      "learning_rate": 4.014270643535427e-06,
      "loss": 0.0389,
      "step": 633
    },
    {
      "epoch": 1.1281138790035588,
      "grad_norm": 1.7473902552323264,
      "learning_rate": 4.000573727519868e-06,
      "loss": 0.0603,
      "step": 634
    },
    {
      "epoch": 1.1298932384341638,
      "grad_norm": 1.4250341245930906,
      "learning_rate": 3.9868846191051465e-06,
      "loss": 0.0458,
      "step": 635
    },
    {
      "epoch": 1.1316725978647686,
      "grad_norm": 1.5697987257601447,
      "learning_rate": 3.973203425231715e-06,
      "loss": 0.0537,
      "step": 636
    },
    {
      "epoch": 1.1334519572953736,
      "grad_norm": 1.3820206518714562,
      "learning_rate": 3.959530252778187e-06,
      "loss": 0.0559,
      "step": 637
    },
    {
      "epoch": 1.1352313167259787,
      "grad_norm": 1.890111172118723,
      "learning_rate": 3.945865208560522e-06,
      "loss": 0.0689,
      "step": 638
    },
    {
      "epoch": 1.1370106761565837,
      "grad_norm": 1.6690475978184052,
      "learning_rate": 3.932208399331177e-06,
      "loss": 0.0588,
      "step": 639
    },
    {
      "epoch": 1.1387900355871885,
      "grad_norm": 1.537753983006339,
      "learning_rate": 3.918559931778277e-06,
      "loss": 0.0543,
      "step": 640
    },
    {
      "epoch": 1.1405693950177935,
      "grad_norm": 1.2252836821683382,
      "learning_rate": 3.904919912524784e-06,
      "loss": 0.0334,
      "step": 641
    },
    {
      "epoch": 1.1423487544483986,
      "grad_norm": 1.3024566427865079,
      "learning_rate": 3.891288448127654e-06,
      "loss": 0.0477,
      "step": 642
    },
    {
      "epoch": 1.1441281138790036,
      "grad_norm": 1.6214233572235495,
      "learning_rate": 3.877665645077017e-06,
      "loss": 0.0481,
      "step": 643
    },
    {
      "epoch": 1.1459074733096086,
      "grad_norm": 1.5322825992787603,
      "learning_rate": 3.86405160979534e-06,
      "loss": 0.0531,
      "step": 644
    },
    {
      "epoch": 1.1476868327402134,
      "grad_norm": 1.4602298716456994,
      "learning_rate": 3.850446448636597e-06,
      "loss": 0.0475,
      "step": 645
    },
    {
      "epoch": 1.1494661921708185,
      "grad_norm": 0.9381534926052785,
      "learning_rate": 3.8368502678854296e-06,
      "loss": 0.0279,
      "step": 646
    },
    {
      "epoch": 1.1512455516014235,
      "grad_norm": 1.32324909822255,
      "learning_rate": 3.8232631737563306e-06,
      "loss": 0.0437,
      "step": 647
    },
    {
      "epoch": 1.1530249110320285,
      "grad_norm": 1.6095475209235117,
      "learning_rate": 3.809685272392804e-06,
      "loss": 0.0439,
      "step": 648
    },
    {
      "epoch": 1.1548042704626336,
      "grad_norm": 1.388044010745066,
      "learning_rate": 3.796116669866543e-06,
      "loss": 0.0463,
      "step": 649
    },
    {
      "epoch": 1.1565836298932384,
      "grad_norm": 1.4902338129431416,
      "learning_rate": 3.78255747217659e-06,
      "loss": 0.0469,
      "step": 650
    },
    {
      "epoch": 1.1583629893238434,
      "grad_norm": 1.2950571221168115,
      "learning_rate": 3.769007785248523e-06,
      "loss": 0.0398,
      "step": 651
    },
    {
      "epoch": 1.1601423487544484,
      "grad_norm": 1.404075114771851,
      "learning_rate": 3.7554677149336186e-06,
      "loss": 0.0456,
      "step": 652
    },
    {
      "epoch": 1.1619217081850535,
      "grad_norm": 1.4870402316668874,
      "learning_rate": 3.7419373670080284e-06,
      "loss": 0.046,
      "step": 653
    },
    {
      "epoch": 1.1637010676156583,
      "grad_norm": 1.8098481304060874,
      "learning_rate": 3.7284168471719527e-06,
      "loss": 0.0577,
      "step": 654
    },
    {
      "epoch": 1.1654804270462633,
      "grad_norm": 1.427817357295298,
      "learning_rate": 3.7149062610488085e-06,
      "loss": 0.0434,
      "step": 655
    },
    {
      "epoch": 1.1672597864768683,
      "grad_norm": 1.3560241301938276,
      "learning_rate": 3.701405714184416e-06,
      "loss": 0.0408,
      "step": 656
    },
    {
      "epoch": 1.1690391459074734,
      "grad_norm": 1.1882146638495006,
      "learning_rate": 3.687915312046166e-06,
      "loss": 0.0428,
      "step": 657
    },
    {
      "epoch": 1.1708185053380782,
      "grad_norm": 1.549165743586227,
      "learning_rate": 3.6744351600221994e-06,
      "loss": 0.0361,
      "step": 658
    },
    {
      "epoch": 1.1725978647686832,
      "grad_norm": 1.711958118064137,
      "learning_rate": 3.6609653634205773e-06,
      "loss": 0.0664,
      "step": 659
    },
    {
      "epoch": 1.1743772241992882,
      "grad_norm": 2.379429750352705,
      "learning_rate": 3.647506027468467e-06,
      "loss": 0.0529,
      "step": 660
    },
    {
      "epoch": 1.1761565836298933,
      "grad_norm": 1.3263718740247306,
      "learning_rate": 3.6340572573113176e-06,
      "loss": 0.0439,
      "step": 661
    },
    {
      "epoch": 1.1779359430604983,
      "grad_norm": 1.0138072445617423,
      "learning_rate": 3.6206191580120346e-06,
      "loss": 0.0346,
      "step": 662
    },
    {
      "epoch": 1.1797153024911031,
      "grad_norm": 1.187321952035311,
      "learning_rate": 3.6071918345501655e-06,
      "loss": 0.0343,
      "step": 663
    },
    {
      "epoch": 1.1814946619217082,
      "grad_norm": 1.5660777663999803,
      "learning_rate": 3.5937753918210705e-06,
      "loss": 0.0422,
      "step": 664
    },
    {
      "epoch": 1.1832740213523132,
      "grad_norm": 1.665330796338174,
      "learning_rate": 3.5803699346351117e-06,
      "loss": 0.042,
      "step": 665
    },
    {
      "epoch": 1.1850533807829182,
      "grad_norm": 1.280733162874042,
      "learning_rate": 3.566975567716833e-06,
      "loss": 0.0378,
      "step": 666
    },
    {
      "epoch": 1.1868327402135233,
      "grad_norm": 1.568241267913538,
      "learning_rate": 3.5535923957041374e-06,
      "loss": 0.0509,
      "step": 667
    },
    {
      "epoch": 1.188612099644128,
      "grad_norm": 1.2754536567847237,
      "learning_rate": 3.540220523147474e-06,
      "loss": 0.0416,
      "step": 668
    },
    {
      "epoch": 1.190391459074733,
      "grad_norm": 1.4117015620711075,
      "learning_rate": 3.5268600545090183e-06,
      "loss": 0.0413,
      "step": 669
    },
    {
      "epoch": 1.1921708185053381,
      "grad_norm": 1.4766697587328628,
      "learning_rate": 3.513511094161858e-06,
      "loss": 0.0431,
      "step": 670
    },
    {
      "epoch": 1.193950177935943,
      "grad_norm": 1.4252410807316729,
      "learning_rate": 3.5001737463891793e-06,
      "loss": 0.0356,
      "step": 671
    },
    {
      "epoch": 1.195729537366548,
      "grad_norm": 1.4613171450739462,
      "learning_rate": 3.4868481153834454e-06,
      "loss": 0.052,
      "step": 672
    },
    {
      "epoch": 1.197508896797153,
      "grad_norm": 1.3907823994123085,
      "learning_rate": 3.4735343052455905e-06,
      "loss": 0.036,
      "step": 673
    },
    {
      "epoch": 1.199288256227758,
      "grad_norm": 1.4184575748732666,
      "learning_rate": 3.4602324199842026e-06,
      "loss": 0.0369,
      "step": 674
    },
    {
      "epoch": 1.201067615658363,
      "grad_norm": 1.36001916499385,
      "learning_rate": 3.446942563514711e-06,
      "loss": 0.0447,
      "step": 675
    },
    {
      "epoch": 1.2028469750889679,
      "grad_norm": 1.5849240899708212,
      "learning_rate": 3.4336648396585777e-06,
      "loss": 0.04,
      "step": 676
    },
    {
      "epoch": 1.204626334519573,
      "grad_norm": 1.8093553492261332,
      "learning_rate": 3.4203993521424774e-06,
      "loss": 0.0494,
      "step": 677
    },
    {
      "epoch": 1.206405693950178,
      "grad_norm": 1.6175915096827707,
      "learning_rate": 3.407146204597499e-06,
      "loss": 0.0449,
      "step": 678
    },
    {
      "epoch": 1.208185053380783,
      "grad_norm": 1.4313997499435398,
      "learning_rate": 3.3939055005583305e-06,
      "loss": 0.0407,
      "step": 679
    },
    {
      "epoch": 1.209964412811388,
      "grad_norm": 1.8171800594362804,
      "learning_rate": 3.3806773434624475e-06,
      "loss": 0.0544,
      "step": 680
    },
    {
      "epoch": 1.2117437722419928,
      "grad_norm": 1.88415600137481,
      "learning_rate": 3.3674618366493117e-06,
      "loss": 0.0664,
      "step": 681
    },
    {
      "epoch": 1.2135231316725978,
      "grad_norm": 1.509965419009144,
      "learning_rate": 3.3542590833595533e-06,
      "loss": 0.0441,
      "step": 682
    },
    {
      "epoch": 1.2153024911032029,
      "grad_norm": 1.3740017262996316,
      "learning_rate": 3.341069186734176e-06,
      "loss": 0.0423,
      "step": 683
    },
    {
      "epoch": 1.217081850533808,
      "grad_norm": 1.1351552579207862,
      "learning_rate": 3.3278922498137455e-06,
      "loss": 0.0284,
      "step": 684
    },
    {
      "epoch": 1.2188612099644127,
      "grad_norm": 1.500731680019629,
      "learning_rate": 3.314728375537587e-06,
      "loss": 0.0412,
      "step": 685
    },
    {
      "epoch": 1.2206405693950177,
      "grad_norm": 1.3947257433457336,
      "learning_rate": 3.3015776667429724e-06,
      "loss": 0.0514,
      "step": 686
    },
    {
      "epoch": 1.2224199288256228,
      "grad_norm": 1.5673957109623669,
      "learning_rate": 3.2884402261643296e-06,
      "loss": 0.0435,
      "step": 687
    },
    {
      "epoch": 1.2241992882562278,
      "grad_norm": 1.646194599971042,
      "learning_rate": 3.2753161564324344e-06,
      "loss": 0.0498,
      "step": 688
    },
    {
      "epoch": 1.2259786476868326,
      "grad_norm": 1.7331102573507795,
      "learning_rate": 3.262205560073605e-06,
      "loss": 0.0406,
      "step": 689
    },
    {
      "epoch": 1.2277580071174377,
      "grad_norm": 1.3195416424228423,
      "learning_rate": 3.249108539508909e-06,
      "loss": 0.0233,
      "step": 690
    },
    {
      "epoch": 1.2295373665480427,
      "grad_norm": 1.645362672136927,
      "learning_rate": 3.2360251970533527e-06,
      "loss": 0.059,
      "step": 691
    },
    {
      "epoch": 1.2313167259786477,
      "grad_norm": 1.4355758651820663,
      "learning_rate": 3.2229556349150947e-06,
      "loss": 0.0469,
      "step": 692
    },
    {
      "epoch": 1.2330960854092528,
      "grad_norm": 1.2130450326062125,
      "learning_rate": 3.2098999551946337e-06,
      "loss": 0.0334,
      "step": 693
    },
    {
      "epoch": 1.2348754448398576,
      "grad_norm": 1.553424758251328,
      "learning_rate": 3.1968582598840234e-06,
      "loss": 0.0482,
      "step": 694
    },
    {
      "epoch": 1.2366548042704626,
      "grad_norm": 1.8096017032291503,
      "learning_rate": 3.183830650866068e-06,
      "loss": 0.0502,
      "step": 695
    },
    {
      "epoch": 1.2384341637010676,
      "grad_norm": 1.3611557146325757,
      "learning_rate": 3.1708172299135266e-06,
      "loss": 0.0431,
      "step": 696
    },
    {
      "epoch": 1.2402135231316727,
      "grad_norm": 2.1315854303653445,
      "learning_rate": 3.1578180986883234e-06,
      "loss": 0.0718,
      "step": 697
    },
    {
      "epoch": 1.2419928825622777,
      "grad_norm": 1.5995732681828845,
      "learning_rate": 3.1448333587407486e-06,
      "loss": 0.0397,
      "step": 698
    },
    {
      "epoch": 1.2437722419928825,
      "grad_norm": 1.9099934919827357,
      "learning_rate": 3.131863111508667e-06,
      "loss": 0.0481,
      "step": 699
    },
    {
      "epoch": 1.2455516014234875,
      "grad_norm": 1.7796727726121742,
      "learning_rate": 3.118907458316722e-06,
      "loss": 0.0458,
      "step": 700
    },
    {
      "epoch": 1.2473309608540926,
      "grad_norm": 1.5413449983837406,
      "learning_rate": 3.105966500375551e-06,
      "loss": 0.0304,
      "step": 701
    },
    {
      "epoch": 1.2491103202846976,
      "grad_norm": 1.5340534676383317,
      "learning_rate": 3.0930403387809892e-06,
      "loss": 0.0563,
      "step": 702
    },
    {
      "epoch": 1.2508896797153026,
      "grad_norm": 1.7857283319423334,
      "learning_rate": 3.080129074513285e-06,
      "loss": 0.0604,
      "step": 703
    },
    {
      "epoch": 1.2526690391459074,
      "grad_norm": 1.4470212376702312,
      "learning_rate": 3.067232808436299e-06,
      "loss": 0.0416,
      "step": 704
    },
    {
      "epoch": 1.2544483985765125,
      "grad_norm": 1.581230153366629,
      "learning_rate": 3.0543516412967327e-06,
      "loss": 0.0412,
      "step": 705
    },
    {
      "epoch": 1.2562277580071175,
      "grad_norm": 1.2528928269572963,
      "learning_rate": 3.041485673723331e-06,
      "loss": 0.0359,
      "step": 706
    },
    {
      "epoch": 1.2580071174377223,
      "grad_norm": 1.414387324032504,
      "learning_rate": 3.0286350062261017e-06,
      "loss": 0.0431,
      "step": 707
    },
    {
      "epoch": 1.2597864768683273,
      "grad_norm": 1.3825588629970618,
      "learning_rate": 3.0157997391955172e-06,
      "loss": 0.0427,
      "step": 708
    },
    {
      "epoch": 1.2615658362989324,
      "grad_norm": 1.5765392660994293,
      "learning_rate": 3.0029799729017518e-06,
      "loss": 0.0504,
      "step": 709
    },
    {
      "epoch": 1.2633451957295374,
      "grad_norm": 1.2090760592199181,
      "learning_rate": 2.9901758074938797e-06,
      "loss": 0.0355,
      "step": 710
    },
    {
      "epoch": 1.2651245551601424,
      "grad_norm": 1.437616375079449,
      "learning_rate": 2.977387342999103e-06,
      "loss": 0.0392,
      "step": 711
    },
    {
      "epoch": 1.2669039145907472,
      "grad_norm": 1.4974122591348888,
      "learning_rate": 2.964614679321966e-06,
      "loss": 0.0427,
      "step": 712
    },
    {
      "epoch": 1.2686832740213523,
      "grad_norm": 1.5168013370760074,
      "learning_rate": 2.951857916243574e-06,
      "loss": 0.0516,
      "step": 713
    },
    {
      "epoch": 1.2704626334519573,
      "grad_norm": 2.003028011281055,
      "learning_rate": 2.9391171534208185e-06,
      "loss": 0.0693,
      "step": 714
    },
    {
      "epoch": 1.2722419928825623,
      "grad_norm": 1.7991683612987106,
      "learning_rate": 2.9263924903855932e-06,
      "loss": 0.0406,
      "step": 715
    },
    {
      "epoch": 1.2740213523131674,
      "grad_norm": 1.3303723493564288,
      "learning_rate": 2.9136840265440213e-06,
      "loss": 0.0413,
      "step": 716
    },
    {
      "epoch": 1.2758007117437722,
      "grad_norm": 1.526024916039291,
      "learning_rate": 2.9009918611756732e-06,
      "loss": 0.0467,
      "step": 717
    },
    {
      "epoch": 1.2775800711743772,
      "grad_norm": 1.5867906173093136,
      "learning_rate": 2.8883160934327968e-06,
      "loss": 0.0416,
      "step": 718
    },
    {
      "epoch": 1.2793594306049823,
      "grad_norm": 1.9563296293164805,
      "learning_rate": 2.8756568223395396e-06,
      "loss": 0.0512,
      "step": 719
    },
    {
      "epoch": 1.281138790035587,
      "grad_norm": 1.8578308384644335,
      "learning_rate": 2.8630141467911777e-06,
      "loss": 0.051,
      "step": 720
    },
    {
      "epoch": 1.282918149466192,
      "grad_norm": 1.3561542441838423,
      "learning_rate": 2.8503881655533395e-06,
      "loss": 0.0354,
      "step": 721
    },
    {
      "epoch": 1.2846975088967971,
      "grad_norm": 1.315049816981894,
      "learning_rate": 2.837778977261235e-06,
      "loss": 0.0423,
      "step": 722
    },
    {
      "epoch": 1.2864768683274022,
      "grad_norm": 1.2376947248706438,
      "learning_rate": 2.8251866804188875e-06,
      "loss": 0.036,
      "step": 723
    },
    {
      "epoch": 1.2882562277580072,
      "grad_norm": 1.2549310049286888,
      "learning_rate": 2.812611373398365e-06,
      "loss": 0.0433,
      "step": 724
    },
    {
      "epoch": 1.290035587188612,
      "grad_norm": 1.451092755538499,
      "learning_rate": 2.8000531544390064e-06,
      "loss": 0.0373,
      "step": 725
    },
    {
      "epoch": 1.291814946619217,
      "grad_norm": 1.5962826162874273,
      "learning_rate": 2.7875121216466595e-06,
      "loss": 0.0501,
      "step": 726
    },
    {
      "epoch": 1.293594306049822,
      "grad_norm": 1.548895078757718,
      "learning_rate": 2.7749883729929105e-06,
      "loss": 0.0493,
      "step": 727
    },
    {
      "epoch": 1.295373665480427,
      "grad_norm": 1.17990257829568,
      "learning_rate": 2.762482006314324e-06,
      "loss": 0.0331,
      "step": 728
    },
    {
      "epoch": 1.2971530249110321,
      "grad_norm": 1.418124895350595,
      "learning_rate": 2.7499931193116692e-06,
      "loss": 0.0333,
      "step": 729
    },
    {
      "epoch": 1.298932384341637,
      "grad_norm": 1.5240534414358369,
      "learning_rate": 2.737521809549167e-06,
      "loss": 0.0483,
      "step": 730
    },
    {
      "epoch": 1.300711743772242,
      "grad_norm": 1.7782199464705766,
      "learning_rate": 2.725068174453722e-06,
      "loss": 0.0555,
      "step": 731
    },
    {
      "epoch": 1.302491103202847,
      "grad_norm": 1.536452691578479,
      "learning_rate": 2.712632311314165e-06,
      "loss": 0.0443,
      "step": 732
    },
    {
      "epoch": 1.304270462633452,
      "grad_norm": 1.7260553220223356,
      "learning_rate": 2.7002143172804875e-06,
      "loss": 0.0399,
      "step": 733
    },
    {
      "epoch": 1.306049822064057,
      "grad_norm": 1.6688231505615356,
      "learning_rate": 2.6878142893630904e-06,
      "loss": 0.0484,
      "step": 734
    },
    {
      "epoch": 1.3078291814946619,
      "grad_norm": 1.3634331026927937,
      "learning_rate": 2.6754323244320154e-06,
      "loss": 0.0506,
      "step": 735
    },
    {
      "epoch": 1.309608540925267,
      "grad_norm": 1.2266569925734165,
      "learning_rate": 2.6630685192161995e-06,
      "loss": 0.0386,
      "step": 736
    },
    {
      "epoch": 1.311387900355872,
      "grad_norm": 1.547088872548096,
      "learning_rate": 2.650722970302714e-06,
      "loss": 0.0373,
      "step": 737
    },
    {
      "epoch": 1.3131672597864767,
      "grad_norm": 1.7992473615890379,
      "learning_rate": 2.638395774136009e-06,
      "loss": 0.0536,
      "step": 738
    },
    {
      "epoch": 1.3149466192170818,
      "grad_norm": 1.5739426478457876,
      "learning_rate": 2.6260870270171645e-06,
      "loss": 0.0437,
      "step": 739
    },
    {
      "epoch": 1.3167259786476868,
      "grad_norm": 1.5956055301638785,
      "learning_rate": 2.613796825103129e-06,
      "loss": 0.0519,
      "step": 740
    },
    {
      "epoch": 1.3185053380782918,
      "grad_norm": 1.41793678602245,
      "learning_rate": 2.60152526440598e-06,
      "loss": 0.0391,
      "step": 741
    },
    {
      "epoch": 1.3202846975088969,
      "grad_norm": 1.220700573283321,
      "learning_rate": 2.5892724407921667e-06,
      "loss": 0.03,
      "step": 742
    },
    {
      "epoch": 1.3220640569395017,
      "grad_norm": 1.8057026056641117,
      "learning_rate": 2.577038449981763e-06,
      "loss": 0.0545,
      "step": 743
    },
    {
      "epoch": 1.3238434163701067,
      "grad_norm": 1.5437830629091855,
      "learning_rate": 2.564823387547716e-06,
      "loss": 0.0401,
      "step": 744
    },
    {
      "epoch": 1.3256227758007118,
      "grad_norm": 1.7115235307639178,
      "learning_rate": 2.552627348915106e-06,
      "loss": 0.0503,
      "step": 745
    },
    {
      "epoch": 1.3274021352313168,
      "grad_norm": 1.7122787035013294,
      "learning_rate": 2.5404504293603983e-06,
      "loss": 0.0586,
      "step": 746
    },
    {
      "epoch": 1.3291814946619218,
      "grad_norm": 1.874841166084849,
      "learning_rate": 2.528292724010697e-06,
      "loss": 0.0617,
      "step": 747
    },
    {
      "epoch": 1.3309608540925266,
      "grad_norm": 1.395531322757944,
      "learning_rate": 2.5161543278430055e-06,
      "loss": 0.0431,
      "step": 748
    },
    {
      "epoch": 1.3327402135231317,
      "grad_norm": 1.5827454142328405,
      "learning_rate": 2.5040353356834756e-06,
      "loss": 0.0452,
      "step": 749
    },
    {
      "epoch": 1.3345195729537367,
      "grad_norm": 1.6750835011708034,
      "learning_rate": 2.4919358422066816e-06,
      "loss": 0.0389,
      "step": 750
    },
    {
      "epoch": 1.3362989323843417,
      "grad_norm": 1.7128735446274321,
      "learning_rate": 2.4798559419348672e-06,
      "loss": 0.0421,
      "step": 751
    },
    {
      "epoch": 1.3380782918149468,
      "grad_norm": 1.2099184361355606,
      "learning_rate": 2.4677957292372166e-06,
      "loss": 0.0357,
      "step": 752
    },
    {
      "epoch": 1.3398576512455516,
      "grad_norm": 1.1929806757750958,
      "learning_rate": 2.455755298329107e-06,
      "loss": 0.0381,
      "step": 753
    },
    {
      "epoch": 1.3416370106761566,
      "grad_norm": 1.2743550325713942,
      "learning_rate": 2.4437347432713838e-06,
      "loss": 0.0365,
      "step": 754
    },
    {
      "epoch": 1.3434163701067616,
      "grad_norm": 1.6149355720674208,
      "learning_rate": 2.431734157969619e-06,
      "loss": 0.0438,
      "step": 755
    },
    {
      "epoch": 1.3451957295373664,
      "grad_norm": 1.7415000691153033,
      "learning_rate": 2.4197536361733792e-06,
      "loss": 0.0599,
      "step": 756
    },
    {
      "epoch": 1.3469750889679715,
      "grad_norm": 1.1977328149862356,
      "learning_rate": 2.407793271475495e-06,
      "loss": 0.0334,
      "step": 757
    },
    {
      "epoch": 1.3487544483985765,
      "grad_norm": 1.3909866005431668,
      "learning_rate": 2.3958531573113223e-06,
      "loss": 0.0528,
      "step": 758
    },
    {
      "epoch": 1.3505338078291815,
      "grad_norm": 1.4244544483594586,
      "learning_rate": 2.3839333869580243e-06,
      "loss": 0.0425,
      "step": 759
    },
    {
      "epoch": 1.3523131672597866,
      "grad_norm": 1.6973049135550342,
      "learning_rate": 2.372034053533835e-06,
      "loss": 0.0421,
      "step": 760
    },
    {
      "epoch": 1.3540925266903914,
      "grad_norm": 1.2874358350573702,
      "learning_rate": 2.360155249997334e-06,
      "loss": 0.0442,
      "step": 761
    },
    {
      "epoch": 1.3558718861209964,
      "grad_norm": 1.2434951030469052,
      "learning_rate": 2.348297069146715e-06,
      "loss": 0.0319,
      "step": 762
    },
    {
      "epoch": 1.3576512455516014,
      "grad_norm": 1.1665491752588282,
      "learning_rate": 2.3364596036190706e-06,
      "loss": 0.0359,
      "step": 763
    },
    {
      "epoch": 1.3594306049822065,
      "grad_norm": 1.461756045564582,
      "learning_rate": 2.3246429458896637e-06,
      "loss": 0.0459,
      "step": 764
    },
    {
      "epoch": 1.3612099644128115,
      "grad_norm": 1.6343361792210032,
      "learning_rate": 2.312847188271203e-06,
      "loss": 0.0524,
      "step": 765
    },
    {
      "epoch": 1.3629893238434163,
      "grad_norm": 1.3548274991759919,
      "learning_rate": 2.301072422913123e-06,
      "loss": 0.0321,
      "step": 766
    },
    {
      "epoch": 1.3647686832740213,
      "grad_norm": 1.4953383228875219,
      "learning_rate": 2.2893187418008666e-06,
      "loss": 0.0389,
      "step": 767
    },
    {
      "epoch": 1.3665480427046264,
      "grad_norm": 1.5567290941348062,
      "learning_rate": 2.2775862367551642e-06,
      "loss": 0.0447,
      "step": 768
    },
    {
      "epoch": 1.3683274021352312,
      "grad_norm": 1.5235519266079038,
      "learning_rate": 2.265874999431318e-06,
      "loss": 0.0428,
      "step": 769
    },
    {
      "epoch": 1.3701067615658362,
      "grad_norm": 1.4150161766049616,
      "learning_rate": 2.254185121318484e-06,
      "loss": 0.0329,
      "step": 770
    },
    {
      "epoch": 1.3718861209964412,
      "grad_norm": 1.4845114426548112,
      "learning_rate": 2.2425166937389596e-06,
      "loss": 0.0421,
      "step": 771
    },
    {
      "epoch": 1.3736654804270463,
      "grad_norm": 1.2681985823538768,
      "learning_rate": 2.2308698078474645e-06,
      "loss": 0.0418,
      "step": 772
    },
    {
      "epoch": 1.3754448398576513,
      "grad_norm": 1.3516636841672425,
      "learning_rate": 2.219244554630438e-06,
      "loss": 0.0455,
      "step": 773
    },
    {
      "epoch": 1.3772241992882561,
      "grad_norm": 1.3300362415228972,
      "learning_rate": 2.207641024905322e-06,
      "loss": 0.0386,
      "step": 774
    },
    {
      "epoch": 1.3790035587188612,
      "grad_norm": 1.6374240147172454,
      "learning_rate": 2.1960593093198508e-06,
      "loss": 0.0413,
      "step": 775
    },
    {
      "epoch": 1.3807829181494662,
      "grad_norm": 1.5008763316056877,
      "learning_rate": 2.184499498351347e-06,
      "loss": 0.042,
      "step": 776
    },
    {
      "epoch": 1.3825622775800712,
      "grad_norm": 1.3756810833501245,
      "learning_rate": 2.172961682306011e-06,
      "loss": 0.0454,
      "step": 777
    },
    {
      "epoch": 1.3843416370106763,
      "grad_norm": 1.7291217702441377,
      "learning_rate": 2.1614459513182173e-06,
      "loss": 0.0467,
      "step": 778
    },
    {
      "epoch": 1.386120996441281,
      "grad_norm": 1.4328660208411663,
      "learning_rate": 2.149952395349813e-06,
      "loss": 0.0353,
      "step": 779
    },
    {
      "epoch": 1.387900355871886,
      "grad_norm": 1.7034968289059418,
      "learning_rate": 2.1384811041894055e-06,
      "loss": 0.0402,
      "step": 780
    },
    {
      "epoch": 1.3896797153024911,
      "grad_norm": 1.5883094217308575,
      "learning_rate": 2.1270321674516736e-06,
      "loss": 0.0434,
      "step": 781
    },
    {
      "epoch": 1.3914590747330962,
      "grad_norm": 1.5960922211126674,
      "learning_rate": 2.1156056745766593e-06,
      "loss": 0.0443,
      "step": 782
    },
    {
      "epoch": 1.3932384341637012,
      "grad_norm": 1.492197946369974,
      "learning_rate": 2.104201714829074e-06,
      "loss": 0.045,
      "step": 783
    },
    {
      "epoch": 1.395017793594306,
      "grad_norm": 1.5295778778391644,
      "learning_rate": 2.0928203772975917e-06,
      "loss": 0.0459,
      "step": 784
    },
    {
      "epoch": 1.396797153024911,
      "grad_norm": 1.3529469158844694,
      "learning_rate": 2.081461750894166e-06,
      "loss": 0.0382,
      "step": 785
    },
    {
      "epoch": 1.398576512455516,
      "grad_norm": 1.6374373924539984,
      "learning_rate": 2.070125924353328e-06,
      "loss": 0.0413,
      "step": 786
    },
    {
      "epoch": 1.4003558718861209,
      "grad_norm": 1.2104674520430558,
      "learning_rate": 2.058812986231493e-06,
      "loss": 0.0402,
      "step": 787
    },
    {
      "epoch": 1.402135231316726,
      "grad_norm": 1.6742851795060303,
      "learning_rate": 2.0475230249062727e-06,
      "loss": 0.0696,
      "step": 788
    },
    {
      "epoch": 1.403914590747331,
      "grad_norm": 1.6002299302582397,
      "learning_rate": 2.0362561285757766e-06,
      "loss": 0.0407,
      "step": 789
    },
    {
      "epoch": 1.405693950177936,
      "grad_norm": 1.5373713699855578,
      "learning_rate": 2.0250123852579347e-06,
      "loss": 0.0432,
      "step": 790
    },
    {
      "epoch": 1.407473309608541,
      "grad_norm": 1.032138062275992,
      "learning_rate": 2.013791882789801e-06,
      "loss": 0.0243,
      "step": 791
    },
    {
      "epoch": 1.4092526690391458,
      "grad_norm": 1.274112710997587,
      "learning_rate": 2.0025947088268714e-06,
      "loss": 0.029,
      "step": 792
    },
    {
      "epoch": 1.4110320284697508,
      "grad_norm": 1.324044563801138,
      "learning_rate": 1.9914209508423943e-06,
      "loss": 0.039,
      "step": 793
    },
    {
      "epoch": 1.4128113879003559,
      "grad_norm": 1.2108667213796944,
      "learning_rate": 1.9802706961266936e-06,
      "loss": 0.0345,
      "step": 794
    },
    {
      "epoch": 1.414590747330961,
      "grad_norm": 1.801072446974817,
      "learning_rate": 1.969144031786483e-06,
      "loss": 0.0562,
      "step": 795
    },
    {
      "epoch": 1.416370106761566,
      "grad_norm": 1.477022032529501,
      "learning_rate": 1.958041044744186e-06,
      "loss": 0.0542,
      "step": 796
    },
    {
      "epoch": 1.4181494661921707,
      "grad_norm": 1.1339480073256372,
      "learning_rate": 1.94696182173726e-06,
      "loss": 0.0287,
      "step": 797
    },
    {
      "epoch": 1.4199288256227758,
      "grad_norm": 1.4188145835215062,
      "learning_rate": 1.9359064493175077e-06,
      "loss": 0.045,
      "step": 798
    },
    {
      "epoch": 1.4217081850533808,
      "grad_norm": 1.5010449192896431,
      "learning_rate": 1.9248750138504176e-06,
      "loss": 0.0438,
      "step": 799
    },
    {
      "epoch": 1.4234875444839858,
      "grad_norm": 1.6148218770935596,
      "learning_rate": 1.9138676015144765e-06,
      "loss": 0.0392,
      "step": 800
    },
    {
      "epoch": 1.4234875444839858,
      "eval_loss": 0.13553155958652496,
      "eval_runtime": 1.9004,
      "eval_samples_per_second": 24.206,
      "eval_steps_per_second": 6.315,
      "step": 800
    },
    {
      "epoch": 1.4252669039145909,
      "grad_norm": 1.6045264551089624,
      "learning_rate": 1.9028842983005036e-06,
      "loss": 0.046,
      "step": 801
    },
    {
      "epoch": 1.4270462633451957,
      "grad_norm": 1.5663550316366912,
      "learning_rate": 1.8919251900109697e-06,
      "loss": 0.05,
      "step": 802
    },
    {
      "epoch": 1.4288256227758007,
      "grad_norm": 1.3392919230009432,
      "learning_rate": 1.8809903622593395e-06,
      "loss": 0.0349,
      "step": 803
    },
    {
      "epoch": 1.4306049822064058,
      "grad_norm": 1.0197400883600565,
      "learning_rate": 1.870079900469392e-06,
      "loss": 0.0253,
      "step": 804
    },
    {
      "epoch": 1.4323843416370106,
      "grad_norm": 1.3083944846556463,
      "learning_rate": 1.8591938898745593e-06,
      "loss": 0.0359,
      "step": 805
    },
    {
      "epoch": 1.4341637010676156,
      "grad_norm": 1.6177889716123028,
      "learning_rate": 1.8483324155172594e-06,
      "loss": 0.0431,
      "step": 806
    },
    {
      "epoch": 1.4359430604982206,
      "grad_norm": 1.2746615320017745,
      "learning_rate": 1.837495562248226e-06,
      "loss": 0.0391,
      "step": 807
    },
    {
      "epoch": 1.4377224199288257,
      "grad_norm": 1.4729114621843828,
      "learning_rate": 1.8266834147258577e-06,
      "loss": 0.04,
      "step": 808
    },
    {
      "epoch": 1.4395017793594307,
      "grad_norm": 1.4598157425135223,
      "learning_rate": 1.8158960574155455e-06,
      "loss": 0.0376,
      "step": 809
    },
    {
      "epoch": 1.4412811387900355,
      "grad_norm": 1.720682633776246,
      "learning_rate": 1.8051335745890196e-06,
      "loss": 0.0441,
      "step": 810
    },
    {
      "epoch": 1.4430604982206405,
      "grad_norm": 1.4838556201963446,
      "learning_rate": 1.7943960503236856e-06,
      "loss": 0.0547,
      "step": 811
    },
    {
      "epoch": 1.4448398576512456,
      "grad_norm": 1.3314051570537497,
      "learning_rate": 1.7836835685019732e-06,
      "loss": 0.0395,
      "step": 812
    },
    {
      "epoch": 1.4466192170818506,
      "grad_norm": 2.6112852624779865,
      "learning_rate": 1.7729962128106787e-06,
      "loss": 0.0474,
      "step": 813
    },
    {
      "epoch": 1.4483985765124556,
      "grad_norm": 1.2722427860827188,
      "learning_rate": 1.7623340667403089e-06,
      "loss": 0.0349,
      "step": 814
    },
    {
      "epoch": 1.4501779359430604,
      "grad_norm": 1.9655154853037475,
      "learning_rate": 1.7516972135844352e-06,
      "loss": 0.066,
      "step": 815
    },
    {
      "epoch": 1.4519572953736655,
      "grad_norm": 1.6141416011186334,
      "learning_rate": 1.741085736439031e-06,
      "loss": 0.0404,
      "step": 816
    },
    {
      "epoch": 1.4537366548042705,
      "grad_norm": 1.5342062022939358,
      "learning_rate": 1.730499718201838e-06,
      "loss": 0.0467,
      "step": 817
    },
    {
      "epoch": 1.4555160142348753,
      "grad_norm": 1.4924465392368111,
      "learning_rate": 1.7199392415717064e-06,
      "loss": 0.0401,
      "step": 818
    },
    {
      "epoch": 1.4572953736654806,
      "grad_norm": 1.6401729800161722,
      "learning_rate": 1.7094043890479557e-06,
      "loss": 0.0474,
      "step": 819
    },
    {
      "epoch": 1.4590747330960854,
      "grad_norm": 1.285145366368902,
      "learning_rate": 1.698895242929725e-06,
      "loss": 0.0409,
      "step": 820
    },
    {
      "epoch": 1.4608540925266904,
      "grad_norm": 1.4781898012303594,
      "learning_rate": 1.6884118853153358e-06,
      "loss": 0.0327,
      "step": 821
    },
    {
      "epoch": 1.4626334519572954,
      "grad_norm": 1.5716182633825058,
      "learning_rate": 1.6779543981016478e-06,
      "loss": 0.0427,
      "step": 822
    },
    {
      "epoch": 1.4644128113879002,
      "grad_norm": 1.4583203708664216,
      "learning_rate": 1.6675228629834133e-06,
      "loss": 0.0395,
      "step": 823
    },
    {
      "epoch": 1.4661921708185053,
      "grad_norm": 1.3613801155561407,
      "learning_rate": 1.657117361452651e-06,
      "loss": 0.0412,
      "step": 824
    },
    {
      "epoch": 1.4679715302491103,
      "grad_norm": 1.3332552202413799,
      "learning_rate": 1.6467379747980011e-06,
      "loss": 0.0399,
      "step": 825
    },
    {
      "epoch": 1.4697508896797153,
      "grad_norm": 1.2983964530965444,
      "learning_rate": 1.6363847841040914e-06,
      "loss": 0.0345,
      "step": 826
    },
    {
      "epoch": 1.4715302491103204,
      "grad_norm": 1.4073663064364275,
      "learning_rate": 1.626057870250906e-06,
      "loss": 0.031,
      "step": 827
    },
    {
      "epoch": 1.4733096085409252,
      "grad_norm": 1.2993154536578022,
      "learning_rate": 1.6157573139131527e-06,
      "loss": 0.0295,
      "step": 828
    },
    {
      "epoch": 1.4750889679715302,
      "grad_norm": 1.2768133277723346,
      "learning_rate": 1.605483195559628e-06,
      "loss": 0.0441,
      "step": 829
    },
    {
      "epoch": 1.4768683274021353,
      "grad_norm": 1.5200745193794887,
      "learning_rate": 1.5952355954525966e-06,
      "loss": 0.0412,
      "step": 830
    },
    {
      "epoch": 1.4786476868327403,
      "grad_norm": 1.5704689191771652,
      "learning_rate": 1.5850145936471607e-06,
      "loss": 0.0382,
      "step": 831
    },
    {
      "epoch": 1.4804270462633453,
      "grad_norm": 1.628460117627646,
      "learning_rate": 1.5748202699906335e-06,
      "loss": 0.0475,
      "step": 832
    },
    {
      "epoch": 1.4822064056939501,
      "grad_norm": 0.9182008751415289,
      "learning_rate": 1.5646527041219128e-06,
      "loss": 0.0256,
      "step": 833
    },
    {
      "epoch": 1.4839857651245552,
      "grad_norm": 1.439259606190967,
      "learning_rate": 1.5545119754708682e-06,
      "loss": 0.0388,
      "step": 834
    },
    {
      "epoch": 1.4857651245551602,
      "grad_norm": 1.571297799533974,
      "learning_rate": 1.544398163257711e-06,
      "loss": 0.0423,
      "step": 835
    },
    {
      "epoch": 1.487544483985765,
      "grad_norm": 1.3212867468780074,
      "learning_rate": 1.5343113464923808e-06,
      "loss": 0.0397,
      "step": 836
    },
    {
      "epoch": 1.48932384341637,
      "grad_norm": 1.0278976202774528,
      "learning_rate": 1.524251603973927e-06,
      "loss": 0.0278,
      "step": 837
    },
    {
      "epoch": 1.491103202846975,
      "grad_norm": 1.2427517728362785,
      "learning_rate": 1.5142190142898883e-06,
      "loss": 0.0396,
      "step": 838
    },
    {
      "epoch": 1.49288256227758,
      "grad_norm": 1.4576960725750074,
      "learning_rate": 1.5042136558156883e-06,
      "loss": 0.0453,
      "step": 839
    },
    {
      "epoch": 1.4946619217081851,
      "grad_norm": 1.5951738011737673,
      "learning_rate": 1.4942356067140162e-06,
      "loss": 0.0475,
      "step": 840
    },
    {
      "epoch": 1.49644128113879,
      "grad_norm": 1.0994539356718618,
      "learning_rate": 1.4842849449342195e-06,
      "loss": 0.0354,
      "step": 841
    },
    {
      "epoch": 1.498220640569395,
      "grad_norm": 2.1389617046432385,
      "learning_rate": 1.4743617482116896e-06,
      "loss": 0.0436,
      "step": 842
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7674007226468915,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0347,
      "step": 843
    },
    {
      "epoch": 1.501779359430605,
      "grad_norm": 1.3372047597755599,
      "learning_rate": 1.454598059806609e-06,
      "loss": 0.0438,
      "step": 844
    },
    {
      "epoch": 1.50355871886121,
      "grad_norm": 1.8690478799924626,
      "learning_rate": 1.4447577225196296e-06,
      "loss": 0.048,
      "step": 845
    },
    {
      "epoch": 1.5053380782918149,
      "grad_norm": 1.5607126430748184,
      "learning_rate": 1.4349451590798564e-06,
      "loss": 0.0462,
      "step": 846
    },
    {
      "epoch": 1.50711743772242,
      "grad_norm": 1.3630895620251622,
      "learning_rate": 1.4251604461438444e-06,
      "loss": 0.0403,
      "step": 847
    },
    {
      "epoch": 1.508896797153025,
      "grad_norm": 1.1778397710106534,
      "learning_rate": 1.4154036601505834e-06,
      "loss": 0.0295,
      "step": 848
    },
    {
      "epoch": 1.5106761565836297,
      "grad_norm": 1.54919239226997,
      "learning_rate": 1.4056748773208933e-06,
      "loss": 0.0418,
      "step": 849
    },
    {
      "epoch": 1.512455516014235,
      "grad_norm": 1.4460624576031404,
      "learning_rate": 1.3959741736568339e-06,
      "loss": 0.0413,
      "step": 850
    },
    {
      "epoch": 1.5142348754448398,
      "grad_norm": 1.2016844961666362,
      "learning_rate": 1.3863016249411027e-06,
      "loss": 0.0368,
      "step": 851
    },
    {
      "epoch": 1.5160142348754448,
      "grad_norm": 1.3648649143091593,
      "learning_rate": 1.376657306736453e-06,
      "loss": 0.0406,
      "step": 852
    },
    {
      "epoch": 1.5177935943060499,
      "grad_norm": 1.569101439175862,
      "learning_rate": 1.3670412943850975e-06,
      "loss": 0.0508,
      "step": 853
    },
    {
      "epoch": 1.5195729537366547,
      "grad_norm": 1.0369296172387779,
      "learning_rate": 1.3574536630081208e-06,
      "loss": 0.0236,
      "step": 854
    },
    {
      "epoch": 1.52135231316726,
      "grad_norm": 1.3753090570781858,
      "learning_rate": 1.347894487504896e-06,
      "loss": 0.0393,
      "step": 855
    },
    {
      "epoch": 1.5231316725978647,
      "grad_norm": 1.4364217227404474,
      "learning_rate": 1.3383638425524909e-06,
      "loss": 0.0372,
      "step": 856
    },
    {
      "epoch": 1.5249110320284698,
      "grad_norm": 1.2082339988983573,
      "learning_rate": 1.3288618026050943e-06,
      "loss": 0.0359,
      "step": 857
    },
    {
      "epoch": 1.5266903914590748,
      "grad_norm": 1.7697427845149132,
      "learning_rate": 1.31938844189343e-06,
      "loss": 0.0411,
      "step": 858
    },
    {
      "epoch": 1.5284697508896796,
      "grad_norm": 1.5496943208028782,
      "learning_rate": 1.3099438344241777e-06,
      "loss": 0.0431,
      "step": 859
    },
    {
      "epoch": 1.5302491103202847,
      "grad_norm": 1.3876393495649573,
      "learning_rate": 1.3005280539793908e-06,
      "loss": 0.038,
      "step": 860
    },
    {
      "epoch": 1.5320284697508897,
      "grad_norm": 1.5283606760759147,
      "learning_rate": 1.2911411741159273e-06,
      "loss": 0.0341,
      "step": 861
    },
    {
      "epoch": 1.5338078291814945,
      "grad_norm": 1.6602618583439275,
      "learning_rate": 1.2817832681648712e-06,
      "loss": 0.0623,
      "step": 862
    },
    {
      "epoch": 1.5355871886120998,
      "grad_norm": 1.5411359409223566,
      "learning_rate": 1.2724544092309581e-06,
      "loss": 0.042,
      "step": 863
    },
    {
      "epoch": 1.5373665480427046,
      "grad_norm": 1.655744290136215,
      "learning_rate": 1.2631546701920073e-06,
      "loss": 0.0532,
      "step": 864
    },
    {
      "epoch": 1.5391459074733096,
      "grad_norm": 1.3563301578406313,
      "learning_rate": 1.2538841236983519e-06,
      "loss": 0.0399,
      "step": 865
    },
    {
      "epoch": 1.5409252669039146,
      "grad_norm": 1.39309908268714,
      "learning_rate": 1.244642842172266e-06,
      "loss": 0.0328,
      "step": 866
    },
    {
      "epoch": 1.5427046263345194,
      "grad_norm": 1.4010073149803954,
      "learning_rate": 1.2354308978074088e-06,
      "loss": 0.0407,
      "step": 867
    },
    {
      "epoch": 1.5444839857651247,
      "grad_norm": 1.349733650119393,
      "learning_rate": 1.2262483625682514e-06,
      "loss": 0.0426,
      "step": 868
    },
    {
      "epoch": 1.5462633451957295,
      "grad_norm": 1.5329016214906686,
      "learning_rate": 1.2170953081895214e-06,
      "loss": 0.0418,
      "step": 869
    },
    {
      "epoch": 1.5480427046263345,
      "grad_norm": 1.6663962662447735,
      "learning_rate": 1.2079718061756369e-06,
      "loss": 0.0359,
      "step": 870
    },
    {
      "epoch": 1.5498220640569396,
      "grad_norm": 1.3192206265069957,
      "learning_rate": 1.1988779278001517e-06,
      "loss": 0.0375,
      "step": 871
    },
    {
      "epoch": 1.5516014234875444,
      "grad_norm": 1.3072180362983923,
      "learning_rate": 1.1898137441051982e-06,
      "loss": 0.04,
      "step": 872
    },
    {
      "epoch": 1.5533807829181496,
      "grad_norm": 1.4269989776940648,
      "learning_rate": 1.1807793259009282e-06,
      "loss": 0.0453,
      "step": 873
    },
    {
      "epoch": 1.5551601423487544,
      "grad_norm": 1.2378646218489764,
      "learning_rate": 1.1717747437649657e-06,
      "loss": 0.0341,
      "step": 874
    },
    {
      "epoch": 1.5569395017793595,
      "grad_norm": 1.7443064937890536,
      "learning_rate": 1.1628000680418533e-06,
      "loss": 0.0459,
      "step": 875
    },
    {
      "epoch": 1.5587188612099645,
      "grad_norm": 1.6102197030681638,
      "learning_rate": 1.1538553688425002e-06,
      "loss": 0.0466,
      "step": 876
    },
    {
      "epoch": 1.5604982206405693,
      "grad_norm": 1.397834534714242,
      "learning_rate": 1.14494071604364e-06,
      "loss": 0.0381,
      "step": 877
    },
    {
      "epoch": 1.5622775800711743,
      "grad_norm": 1.4037906893710472,
      "learning_rate": 1.1360561792872754e-06,
      "loss": 0.0408,
      "step": 878
    },
    {
      "epoch": 1.5640569395017794,
      "grad_norm": 1.7743651590701754,
      "learning_rate": 1.127201827980145e-06,
      "loss": 0.0624,
      "step": 879
    },
    {
      "epoch": 1.5658362989323842,
      "grad_norm": 1.164655492700238,
      "learning_rate": 1.1183777312931748e-06,
      "loss": 0.0257,
      "step": 880
    },
    {
      "epoch": 1.5676156583629894,
      "grad_norm": 1.3555269499971025,
      "learning_rate": 1.1095839581609407e-06,
      "loss": 0.0432,
      "step": 881
    },
    {
      "epoch": 1.5693950177935942,
      "grad_norm": 1.7212104120336336,
      "learning_rate": 1.1008205772811248e-06,
      "loss": 0.0554,
      "step": 882
    },
    {
      "epoch": 1.5711743772241993,
      "grad_norm": 1.6447106126467486,
      "learning_rate": 1.0920876571139843e-06,
      "loss": 0.0473,
      "step": 883
    },
    {
      "epoch": 1.5729537366548043,
      "grad_norm": 1.3228366873077255,
      "learning_rate": 1.0833852658818167e-06,
      "loss": 0.0423,
      "step": 884
    },
    {
      "epoch": 1.5747330960854091,
      "grad_norm": 1.440481606522488,
      "learning_rate": 1.0747134715684221e-06,
      "loss": 0.0306,
      "step": 885
    },
    {
      "epoch": 1.5765124555160144,
      "grad_norm": 1.2889642040008586,
      "learning_rate": 1.0660723419185776e-06,
      "loss": 0.0354,
      "step": 886
    },
    {
      "epoch": 1.5782918149466192,
      "grad_norm": 1.5231752699982202,
      "learning_rate": 1.0574619444375017e-06,
      "loss": 0.0405,
      "step": 887
    },
    {
      "epoch": 1.5800711743772242,
      "grad_norm": 1.3381006034070078,
      "learning_rate": 1.0488823463903341e-06,
      "loss": 0.0355,
      "step": 888
    },
    {
      "epoch": 1.5818505338078293,
      "grad_norm": 1.470805921997831,
      "learning_rate": 1.0403336148016053e-06,
      "loss": 0.0484,
      "step": 889
    },
    {
      "epoch": 1.583629893238434,
      "grad_norm": 1.3067202932834707,
      "learning_rate": 1.0318158164547159e-06,
      "loss": 0.0389,
      "step": 890
    },
    {
      "epoch": 1.585409252669039,
      "grad_norm": 1.3902165526739316,
      "learning_rate": 1.0233290178914096e-06,
      "loss": 0.0343,
      "step": 891
    },
    {
      "epoch": 1.5871886120996441,
      "grad_norm": 1.0650989482283322,
      "learning_rate": 1.014873285411262e-06,
      "loss": 0.0305,
      "step": 892
    },
    {
      "epoch": 1.5889679715302492,
      "grad_norm": 1.5776228251020936,
      "learning_rate": 1.006448685071154e-06,
      "loss": 0.049,
      "step": 893
    },
    {
      "epoch": 1.5907473309608542,
      "grad_norm": 1.6847511370413242,
      "learning_rate": 9.980552826847635e-07,
      "loss": 0.0436,
      "step": 894
    },
    {
      "epoch": 1.592526690391459,
      "grad_norm": 1.5267170572384654,
      "learning_rate": 9.896931438220453e-07,
      "loss": 0.0453,
      "step": 895
    },
    {
      "epoch": 1.594306049822064,
      "grad_norm": 1.6853099289986873,
      "learning_rate": 9.813623338087181e-07,
      "loss": 0.0451,
      "step": 896
    },
    {
      "epoch": 1.596085409252669,
      "grad_norm": 1.8621945073557655,
      "learning_rate": 9.730629177257623e-07,
      "loss": 0.0532,
      "step": 897
    },
    {
      "epoch": 1.5978647686832739,
      "grad_norm": 1.1570214892546553,
      "learning_rate": 9.64794960408903e-07,
      "loss": 0.029,
      "step": 898
    },
    {
      "epoch": 1.5996441281138791,
      "grad_norm": 1.570142854946665,
      "learning_rate": 9.565585264481092e-07,
      "loss": 0.0467,
      "step": 899
    },
    {
      "epoch": 1.601423487544484,
      "grad_norm": 1.2747258900459593,
      "learning_rate": 9.483536801870835e-07,
      "loss": 0.0357,
      "step": 900
    },
    {
      "epoch": 1.603202846975089,
      "grad_norm": 1.768207167854998,
      "learning_rate": 9.401804857227648e-07,
      "loss": 0.0376,
      "step": 901
    },
    {
      "epoch": 1.604982206405694,
      "grad_norm": 1.3816828381796138,
      "learning_rate": 9.320390069048258e-07,
      "loss": 0.0419,
      "step": 902
    },
    {
      "epoch": 1.6067615658362988,
      "grad_norm": 1.3282762248575404,
      "learning_rate": 9.239293073351735e-07,
      "loss": 0.0368,
      "step": 903
    },
    {
      "epoch": 1.608540925266904,
      "grad_norm": 1.5077127417736853,
      "learning_rate": 9.158514503674543e-07,
      "loss": 0.0366,
      "step": 904
    },
    {
      "epoch": 1.6103202846975089,
      "grad_norm": 1.1844435993103961,
      "learning_rate": 9.078054991065532e-07,
      "loss": 0.0302,
      "step": 905
    },
    {
      "epoch": 1.612099644128114,
      "grad_norm": 1.5361179282683537,
      "learning_rate": 8.997915164081095e-07,
      "loss": 0.0516,
      "step": 906
    },
    {
      "epoch": 1.613879003558719,
      "grad_norm": 1.5967997549831903,
      "learning_rate": 8.918095648780195e-07,
      "loss": 0.0512,
      "step": 907
    },
    {
      "epoch": 1.6156583629893237,
      "grad_norm": 1.488194638702583,
      "learning_rate": 8.838597068719518e-07,
      "loss": 0.0357,
      "step": 908
    },
    {
      "epoch": 1.6174377224199288,
      "grad_norm": 1.751026076647405,
      "learning_rate": 8.75942004494853e-07,
      "loss": 0.0446,
      "step": 909
    },
    {
      "epoch": 1.6192170818505338,
      "grad_norm": 1.8228721992338053,
      "learning_rate": 8.680565196004704e-07,
      "loss": 0.0567,
      "step": 910
    },
    {
      "epoch": 1.6209964412811388,
      "grad_norm": 1.328560267532672,
      "learning_rate": 8.602033137908666e-07,
      "loss": 0.0361,
      "step": 911
    },
    {
      "epoch": 1.6227758007117439,
      "grad_norm": 1.3252873462126447,
      "learning_rate": 8.523824484159348e-07,
      "loss": 0.0302,
      "step": 912
    },
    {
      "epoch": 1.6245551601423487,
      "grad_norm": 1.3494694773324976,
      "learning_rate": 8.445939845729245e-07,
      "loss": 0.0342,
      "step": 913
    },
    {
      "epoch": 1.6263345195729537,
      "grad_norm": 1.1524568157816502,
      "learning_rate": 8.368379831059592e-07,
      "loss": 0.0378,
      "step": 914
    },
    {
      "epoch": 1.6281138790035588,
      "grad_norm": 1.2377124974985947,
      "learning_rate": 8.29114504605566e-07,
      "loss": 0.0381,
      "step": 915
    },
    {
      "epoch": 1.6298932384341636,
      "grad_norm": 1.2509629067651231,
      "learning_rate": 8.21423609408199e-07,
      "loss": 0.042,
      "step": 916
    },
    {
      "epoch": 1.6316725978647688,
      "grad_norm": 1.347754497325353,
      "learning_rate": 8.137653575957666e-07,
      "loss": 0.0325,
      "step": 917
    },
    {
      "epoch": 1.6334519572953736,
      "grad_norm": 1.3991680648985156,
      "learning_rate": 8.061398089951678e-07,
      "loss": 0.0474,
      "step": 918
    },
    {
      "epoch": 1.6352313167259787,
      "grad_norm": 1.3702972693459372,
      "learning_rate": 7.985470231778203e-07,
      "loss": 0.0375,
      "step": 919
    },
    {
      "epoch": 1.6370106761565837,
      "grad_norm": 1.4493468169237673,
      "learning_rate": 7.909870594591951e-07,
      "loss": 0.0364,
      "step": 920
    },
    {
      "epoch": 1.6387900355871885,
      "grad_norm": 1.6728531935037791,
      "learning_rate": 7.834599768983553e-07,
      "loss": 0.0459,
      "step": 921
    },
    {
      "epoch": 1.6405693950177938,
      "grad_norm": 1.2858460794736846,
      "learning_rate": 7.759658342974951e-07,
      "loss": 0.0301,
      "step": 922
    },
    {
      "epoch": 1.6423487544483986,
      "grad_norm": 1.43482025502732,
      "learning_rate": 7.685046902014747e-07,
      "loss": 0.0448,
      "step": 923
    },
    {
      "epoch": 1.6441281138790036,
      "grad_norm": 1.5661414126628719,
      "learning_rate": 7.61076602897371e-07,
      "loss": 0.0405,
      "step": 924
    },
    {
      "epoch": 1.6459074733096086,
      "grad_norm": 0.9422703119014241,
      "learning_rate": 7.536816304140177e-07,
      "loss": 0.0235,
      "step": 925
    },
    {
      "epoch": 1.6476868327402134,
      "grad_norm": 1.2618684227905703,
      "learning_rate": 7.46319830521553e-07,
      "loss": 0.0272,
      "step": 926
    },
    {
      "epoch": 1.6494661921708185,
      "grad_norm": 1.2743567720453082,
      "learning_rate": 7.389912607309662e-07,
      "loss": 0.0382,
      "step": 927
    },
    {
      "epoch": 1.6512455516014235,
      "grad_norm": 1.5516278790516005,
      "learning_rate": 7.316959782936516e-07,
      "loss": 0.0439,
      "step": 928
    },
    {
      "epoch": 1.6530249110320283,
      "grad_norm": 1.047201487462478,
      "learning_rate": 7.244340402009608e-07,
      "loss": 0.0279,
      "step": 929
    },
    {
      "epoch": 1.6548042704626336,
      "grad_norm": 1.9607395531967122,
      "learning_rate": 7.172055031837572e-07,
      "loss": 0.0558,
      "step": 930
    },
    {
      "epoch": 1.6565836298932384,
      "grad_norm": 1.2367918261541342,
      "learning_rate": 7.100104237119676e-07,
      "loss": 0.0358,
      "step": 931
    },
    {
      "epoch": 1.6583629893238434,
      "grad_norm": 1.2604371813311994,
      "learning_rate": 7.028488579941506e-07,
      "loss": 0.0435,
      "step": 932
    },
    {
      "epoch": 1.6601423487544484,
      "grad_norm": 1.1764718538407726,
      "learning_rate": 6.957208619770505e-07,
      "loss": 0.0314,
      "step": 933
    },
    {
      "epoch": 1.6619217081850532,
      "grad_norm": 1.917288342025507,
      "learning_rate": 6.886264913451635e-07,
      "loss": 0.0522,
      "step": 934
    },
    {
      "epoch": 1.6637010676156585,
      "grad_norm": 1.4495493967970827,
      "learning_rate": 6.815658015203014e-07,
      "loss": 0.0404,
      "step": 935
    },
    {
      "epoch": 1.6654804270462633,
      "grad_norm": 1.567577585358252,
      "learning_rate": 6.745388476611553e-07,
      "loss": 0.0494,
      "step": 936
    },
    {
      "epoch": 1.6672597864768683,
      "grad_norm": 1.3496135255298065,
      "learning_rate": 6.67545684662873e-07,
      "loss": 0.0392,
      "step": 937
    },
    {
      "epoch": 1.6690391459074734,
      "grad_norm": 1.2624866082491015,
      "learning_rate": 6.605863671566221e-07,
      "loss": 0.0397,
      "step": 938
    },
    {
      "epoch": 1.6708185053380782,
      "grad_norm": 1.4932679808167775,
      "learning_rate": 6.536609495091695e-07,
      "loss": 0.0357,
      "step": 939
    },
    {
      "epoch": 1.6725978647686834,
      "grad_norm": 2.0994499945272773,
      "learning_rate": 6.467694858224488e-07,
      "loss": 0.0605,
      "step": 940
    },
    {
      "epoch": 1.6743772241992882,
      "grad_norm": 1.2202425474352014,
      "learning_rate": 6.399120299331468e-07,
      "loss": 0.028,
      "step": 941
    },
    {
      "epoch": 1.6761565836298933,
      "grad_norm": 1.2547013509656346,
      "learning_rate": 6.330886354122768e-07,
      "loss": 0.0351,
      "step": 942
    },
    {
      "epoch": 1.6779359430604983,
      "grad_norm": 1.2243512640917917,
      "learning_rate": 6.262993555647617e-07,
      "loss": 0.0386,
      "step": 943
    },
    {
      "epoch": 1.6797153024911031,
      "grad_norm": 1.0270397840602432,
      "learning_rate": 6.1954424342902e-07,
      "loss": 0.0304,
      "step": 944
    },
    {
      "epoch": 1.6814946619217082,
      "grad_norm": 2.0627083576047265,
      "learning_rate": 6.128233517765448e-07,
      "loss": 0.0532,
      "step": 945
    },
    {
      "epoch": 1.6832740213523132,
      "grad_norm": 1.7232388309851174,
      "learning_rate": 6.061367331114992e-07,
      "loss": 0.041,
      "step": 946
    },
    {
      "epoch": 1.685053380782918,
      "grad_norm": 1.73200461230147,
      "learning_rate": 5.994844396703025e-07,
      "loss": 0.0543,
      "step": 947
    },
    {
      "epoch": 1.6868327402135233,
      "grad_norm": 1.4191648146634566,
      "learning_rate": 5.928665234212233e-07,
      "loss": 0.033,
      "step": 948
    },
    {
      "epoch": 1.688612099644128,
      "grad_norm": 1.769010986702388,
      "learning_rate": 5.862830360639698e-07,
      "loss": 0.0416,
      "step": 949
    },
    {
      "epoch": 1.690391459074733,
      "grad_norm": 1.2365859730297384,
      "learning_rate": 5.797340290292907e-07,
      "loss": 0.0327,
      "step": 950
    },
    {
      "epoch": 1.6921708185053381,
      "grad_norm": 1.5944536984646245,
      "learning_rate": 5.732195534785723e-07,
      "loss": 0.0425,
      "step": 951
    },
    {
      "epoch": 1.693950177935943,
      "grad_norm": 1.0978948243138542,
      "learning_rate": 5.667396603034369e-07,
      "loss": 0.0281,
      "step": 952
    },
    {
      "epoch": 1.6957295373665482,
      "grad_norm": 1.6489451622459412,
      "learning_rate": 5.602944001253486e-07,
      "loss": 0.0471,
      "step": 953
    },
    {
      "epoch": 1.697508896797153,
      "grad_norm": 1.5034551647306265,
      "learning_rate": 5.538838232952104e-07,
      "loss": 0.0443,
      "step": 954
    },
    {
      "epoch": 1.699288256227758,
      "grad_norm": 1.7753753659867093,
      "learning_rate": 5.475079798929816e-07,
      "loss": 0.0561,
      "step": 955
    },
    {
      "epoch": 1.701067615658363,
      "grad_norm": 1.4311690621571131,
      "learning_rate": 5.411669197272795e-07,
      "loss": 0.0356,
      "step": 956
    },
    {
      "epoch": 1.7028469750889679,
      "grad_norm": 1.2084248674878706,
      "learning_rate": 5.348606923349903e-07,
      "loss": 0.0314,
      "step": 957
    },
    {
      "epoch": 1.704626334519573,
      "grad_norm": 1.4653829807321628,
      "learning_rate": 5.285893469808855e-07,
      "loss": 0.0342,
      "step": 958
    },
    {
      "epoch": 1.706405693950178,
      "grad_norm": 1.4861717769330527,
      "learning_rate": 5.223529326572352e-07,
      "loss": 0.0396,
      "step": 959
    },
    {
      "epoch": 1.708185053380783,
      "grad_norm": 1.6382009093594376,
      "learning_rate": 5.161514980834232e-07,
      "loss": 0.053,
      "step": 960
    },
    {
      "epoch": 1.709964412811388,
      "grad_norm": 1.3917549305779011,
      "learning_rate": 5.099850917055709e-07,
      "loss": 0.0416,
      "step": 961
    },
    {
      "epoch": 1.7117437722419928,
      "grad_norm": 1.5941820214039433,
      "learning_rate": 5.038537616961559e-07,
      "loss": 0.0329,
      "step": 962
    },
    {
      "epoch": 1.7135231316725978,
      "grad_norm": 1.3505572045437793,
      "learning_rate": 4.977575559536358e-07,
      "loss": 0.0393,
      "step": 963
    },
    {
      "epoch": 1.7153024911032029,
      "grad_norm": 1.4906267614713944,
      "learning_rate": 4.916965221020753e-07,
      "loss": 0.0353,
      "step": 964
    },
    {
      "epoch": 1.7170818505338077,
      "grad_norm": 1.1664280571560683,
      "learning_rate": 4.856707074907729e-07,
      "loss": 0.0297,
      "step": 965
    },
    {
      "epoch": 1.718861209964413,
      "grad_norm": 1.5848671896442588,
      "learning_rate": 4.796801591938922e-07,
      "loss": 0.0416,
      "step": 966
    },
    {
      "epoch": 1.7206405693950177,
      "grad_norm": 1.6248722133837754,
      "learning_rate": 4.737249240100911e-07,
      "loss": 0.0424,
      "step": 967
    },
    {
      "epoch": 1.7224199288256228,
      "grad_norm": 1.5499587224919618,
      "learning_rate": 4.6780504846216155e-07,
      "loss": 0.0331,
      "step": 968
    },
    {
      "epoch": 1.7241992882562278,
      "grad_norm": 1.3626836640644866,
      "learning_rate": 4.619205787966613e-07,
      "loss": 0.0373,
      "step": 969
    },
    {
      "epoch": 1.7259786476868326,
      "grad_norm": 1.054552356505648,
      "learning_rate": 4.560715609835548e-07,
      "loss": 0.0313,
      "step": 970
    },
    {
      "epoch": 1.7277580071174379,
      "grad_norm": 1.32690819345873,
      "learning_rate": 4.5025804071585464e-07,
      "loss": 0.0363,
      "step": 971
    },
    {
      "epoch": 1.7295373665480427,
      "grad_norm": 1.8155269019307265,
      "learning_rate": 4.4448006340926163e-07,
      "loss": 0.0564,
      "step": 972
    },
    {
      "epoch": 1.7313167259786477,
      "grad_norm": 1.5408465686558235,
      "learning_rate": 4.3873767420181344e-07,
      "loss": 0.0362,
      "step": 973
    },
    {
      "epoch": 1.7330960854092528,
      "grad_norm": 1.6773239432537133,
      "learning_rate": 4.3303091795353024e-07,
      "loss": 0.0455,
      "step": 974
    },
    {
      "epoch": 1.7348754448398576,
      "grad_norm": 1.7322358515920886,
      "learning_rate": 4.2735983924606596e-07,
      "loss": 0.0408,
      "step": 975
    },
    {
      "epoch": 1.7366548042704626,
      "grad_norm": 1.4106575440753382,
      "learning_rate": 4.2172448238235464e-07,
      "loss": 0.0305,
      "step": 976
    },
    {
      "epoch": 1.7384341637010676,
      "grad_norm": 1.3733644626019057,
      "learning_rate": 4.161248913862731e-07,
      "loss": 0.0301,
      "step": 977
    },
    {
      "epoch": 1.7402135231316724,
      "grad_norm": 1.3994999537006558,
      "learning_rate": 4.1056111000228937e-07,
      "loss": 0.0361,
      "step": 978
    },
    {
      "epoch": 1.7419928825622777,
      "grad_norm": 1.8229650102075292,
      "learning_rate": 4.0503318169512417e-07,
      "loss": 0.041,
      "step": 979
    },
    {
      "epoch": 1.7437722419928825,
      "grad_norm": 1.1810685268091181,
      "learning_rate": 3.9954114964941336e-07,
      "loss": 0.0304,
      "step": 980
    },
    {
      "epoch": 1.7455516014234875,
      "grad_norm": 1.3189393973510206,
      "learning_rate": 3.9408505676936327e-07,
      "loss": 0.0338,
      "step": 981
    },
    {
      "epoch": 1.7473309608540926,
      "grad_norm": 1.6684414853575773,
      "learning_rate": 3.886649456784253e-07,
      "loss": 0.047,
      "step": 982
    },
    {
      "epoch": 1.7491103202846974,
      "grad_norm": 1.6906975429443252,
      "learning_rate": 3.8328085871895624e-07,
      "loss": 0.0558,
      "step": 983
    },
    {
      "epoch": 1.7508896797153026,
      "grad_norm": 1.3184724212662704,
      "learning_rate": 3.779328379518898e-07,
      "loss": 0.0345,
      "step": 984
    },
    {
      "epoch": 1.7526690391459074,
      "grad_norm": 1.4655669673179201,
      "learning_rate": 3.7262092515640556e-07,
      "loss": 0.0391,
      "step": 985
    },
    {
      "epoch": 1.7544483985765125,
      "grad_norm": 1.1582462516823915,
      "learning_rate": 3.673451618296081e-07,
      "loss": 0.0284,
      "step": 986
    },
    {
      "epoch": 1.7562277580071175,
      "grad_norm": 1.2884297196498398,
      "learning_rate": 3.621055891861963e-07,
      "loss": 0.0389,
      "step": 987
    },
    {
      "epoch": 1.7580071174377223,
      "grad_norm": 1.5602060082103053,
      "learning_rate": 3.56902248158148e-07,
      "loss": 0.0373,
      "step": 988
    },
    {
      "epoch": 1.7597864768683276,
      "grad_norm": 1.3896690008295718,
      "learning_rate": 3.517351793943913e-07,
      "loss": 0.0339,
      "step": 989
    },
    {
      "epoch": 1.7615658362989324,
      "grad_norm": 1.3806573572348462,
      "learning_rate": 3.4660442326049704e-07,
      "loss": 0.0296,
      "step": 990
    },
    {
      "epoch": 1.7633451957295374,
      "grad_norm": 1.2620526912546866,
      "learning_rate": 3.4151001983835696e-07,
      "loss": 0.0363,
      "step": 991
    },
    {
      "epoch": 1.7651245551601424,
      "grad_norm": 1.532310867131612,
      "learning_rate": 3.364520089258727e-07,
      "loss": 0.036,
      "step": 992
    },
    {
      "epoch": 1.7669039145907472,
      "grad_norm": 1.3687563826799665,
      "learning_rate": 3.314304300366461e-07,
      "loss": 0.0365,
      "step": 993
    },
    {
      "epoch": 1.7686832740213523,
      "grad_norm": 1.411437401308879,
      "learning_rate": 3.2644532239966444e-07,
      "loss": 0.0421,
      "step": 994
    },
    {
      "epoch": 1.7704626334519573,
      "grad_norm": 1.1120626082359164,
      "learning_rate": 3.2149672495900286e-07,
      "loss": 0.0305,
      "step": 995
    },
    {
      "epoch": 1.7722419928825621,
      "grad_norm": 1.3261442032732178,
      "learning_rate": 3.165846763735153e-07,
      "loss": 0.0404,
      "step": 996
    },
    {
      "epoch": 1.7740213523131674,
      "grad_norm": 1.3314927487277297,
      "learning_rate": 3.117092150165324e-07,
      "loss": 0.0395,
      "step": 997
    },
    {
      "epoch": 1.7758007117437722,
      "grad_norm": 1.354501013095216,
      "learning_rate": 3.068703789755606e-07,
      "loss": 0.0392,
      "step": 998
    },
    {
      "epoch": 1.7775800711743772,
      "grad_norm": 1.494535900064754,
      "learning_rate": 3.020682060519886e-07,
      "loss": 0.0388,
      "step": 999
    },
    {
      "epoch": 1.7793594306049823,
      "grad_norm": 1.5874497599489694,
      "learning_rate": 2.9730273376078923e-07,
      "loss": 0.0414,
      "step": 1000
    },
    {
      "epoch": 1.7793594306049823,
      "eval_loss": 0.13214601576328278,
      "eval_runtime": 1.901,
      "eval_samples_per_second": 24.198,
      "eval_steps_per_second": 6.312,
      "step": 1000
    },
    {
      "epoch": 1.781138790035587,
      "grad_norm": 2.3994380111667395,
      "learning_rate": 2.9257399933022737e-07,
      "loss": 0.0501,
      "step": 1001
    },
    {
      "epoch": 1.7829181494661923,
      "grad_norm": 1.2655566792616009,
      "learning_rate": 2.8788203970156805e-07,
      "loss": 0.0312,
      "step": 1002
    },
    {
      "epoch": 1.7846975088967971,
      "grad_norm": 1.3444020777696293,
      "learning_rate": 2.832268915287878e-07,
      "loss": 0.0437,
      "step": 1003
    },
    {
      "epoch": 1.7864768683274022,
      "grad_norm": 1.5542906543407722,
      "learning_rate": 2.7860859117828985e-07,
      "loss": 0.0428,
      "step": 1004
    },
    {
      "epoch": 1.7882562277580072,
      "grad_norm": 2.321897861204451,
      "learning_rate": 2.740271747286194e-07,
      "loss": 0.0747,
      "step": 1005
    },
    {
      "epoch": 1.790035587188612,
      "grad_norm": 1.330710360419337,
      "learning_rate": 2.6948267797018145e-07,
      "loss": 0.0356,
      "step": 1006
    },
    {
      "epoch": 1.791814946619217,
      "grad_norm": 1.3689667536089076,
      "learning_rate": 2.649751364049613e-07,
      "loss": 0.0329,
      "step": 1007
    },
    {
      "epoch": 1.793594306049822,
      "grad_norm": 1.382354181385714,
      "learning_rate": 2.6050458524624735e-07,
      "loss": 0.0349,
      "step": 1008
    },
    {
      "epoch": 1.795373665480427,
      "grad_norm": 1.2218244187209417,
      "learning_rate": 2.560710594183552e-07,
      "loss": 0.0298,
      "step": 1009
    },
    {
      "epoch": 1.7971530249110321,
      "grad_norm": 1.2892594694947963,
      "learning_rate": 2.5167459355635524e-07,
      "loss": 0.0407,
      "step": 1010
    },
    {
      "epoch": 1.798932384341637,
      "grad_norm": 1.6023679642626898,
      "learning_rate": 2.473152220058039e-07,
      "loss": 0.0387,
      "step": 1011
    },
    {
      "epoch": 1.800711743772242,
      "grad_norm": 1.665045377876774,
      "learning_rate": 2.429929788224722e-07,
      "loss": 0.0433,
      "step": 1012
    },
    {
      "epoch": 1.802491103202847,
      "grad_norm": 1.3909161579328713,
      "learning_rate": 2.38707897772083e-07,
      "loss": 0.0449,
      "step": 1013
    },
    {
      "epoch": 1.8042704626334518,
      "grad_norm": 1.5830892876212914,
      "learning_rate": 2.3446001233004333e-07,
      "loss": 0.0525,
      "step": 1014
    },
    {
      "epoch": 1.806049822064057,
      "grad_norm": 1.8525572439643299,
      "learning_rate": 2.3024935568118745e-07,
      "loss": 0.0409,
      "step": 1015
    },
    {
      "epoch": 1.8078291814946619,
      "grad_norm": 1.2135030517745455,
      "learning_rate": 2.2607596071951288e-07,
      "loss": 0.0292,
      "step": 1016
    },
    {
      "epoch": 1.809608540925267,
      "grad_norm": 1.1214741935267347,
      "learning_rate": 2.2193986004792667e-07,
      "loss": 0.032,
      "step": 1017
    },
    {
      "epoch": 1.811387900355872,
      "grad_norm": 1.068768510245524,
      "learning_rate": 2.1784108597799058e-07,
      "loss": 0.0209,
      "step": 1018
    },
    {
      "epoch": 1.8131672597864767,
      "grad_norm": 1.384011261046797,
      "learning_rate": 2.1377967052966685e-07,
      "loss": 0.0358,
      "step": 1019
    },
    {
      "epoch": 1.814946619217082,
      "grad_norm": 1.2099586837907217,
      "learning_rate": 2.0975564543107007e-07,
      "loss": 0.0331,
      "step": 1020
    },
    {
      "epoch": 1.8167259786476868,
      "grad_norm": 1.2940151547770433,
      "learning_rate": 2.057690421182168e-07,
      "loss": 0.0334,
      "step": 1021
    },
    {
      "epoch": 1.8185053380782918,
      "grad_norm": 1.3827063274317817,
      "learning_rate": 2.01819891734783e-07,
      "loss": 0.0397,
      "step": 1022
    },
    {
      "epoch": 1.8202846975088969,
      "grad_norm": 1.3215914412490635,
      "learning_rate": 1.979082251318576e-07,
      "loss": 0.0343,
      "step": 1023
    },
    {
      "epoch": 1.8220640569395017,
      "grad_norm": 1.2162108890219305,
      "learning_rate": 1.9403407286770592e-07,
      "loss": 0.0293,
      "step": 1024
    },
    {
      "epoch": 1.8238434163701067,
      "grad_norm": 1.193119010928115,
      "learning_rate": 1.9019746520752502e-07,
      "loss": 0.0382,
      "step": 1025
    },
    {
      "epoch": 1.8256227758007118,
      "grad_norm": 1.2590633332237346,
      "learning_rate": 1.8639843212321206e-07,
      "loss": 0.0379,
      "step": 1026
    },
    {
      "epoch": 1.8274021352313166,
      "grad_norm": 1.4827732789423547,
      "learning_rate": 1.826370032931285e-07,
      "loss": 0.0386,
      "step": 1027
    },
    {
      "epoch": 1.8291814946619218,
      "grad_norm": 1.4019662312978882,
      "learning_rate": 1.789132081018674e-07,
      "loss": 0.0344,
      "step": 1028
    },
    {
      "epoch": 1.8309608540925266,
      "grad_norm": 1.277130084295682,
      "learning_rate": 1.7522707564002706e-07,
      "loss": 0.0311,
      "step": 1029
    },
    {
      "epoch": 1.8327402135231317,
      "grad_norm": 1.1965932241938335,
      "learning_rate": 1.7157863470397718e-07,
      "loss": 0.0262,
      "step": 1030
    },
    {
      "epoch": 1.8345195729537367,
      "grad_norm": 1.5200736373961061,
      "learning_rate": 1.6796791379564138e-07,
      "loss": 0.0416,
      "step": 1031
    },
    {
      "epoch": 1.8362989323843415,
      "grad_norm": 1.2800099516950374,
      "learning_rate": 1.6439494112227173e-07,
      "loss": 0.0278,
      "step": 1032
    },
    {
      "epoch": 1.8380782918149468,
      "grad_norm": 1.5304018078565587,
      "learning_rate": 1.6085974459622567e-07,
      "loss": 0.0485,
      "step": 1033
    },
    {
      "epoch": 1.8398576512455516,
      "grad_norm": 1.071069909759288,
      "learning_rate": 1.573623518347517e-07,
      "loss": 0.0266,
      "step": 1034
    },
    {
      "epoch": 1.8416370106761566,
      "grad_norm": 1.493163724386636,
      "learning_rate": 1.5390279015977117e-07,
      "loss": 0.0453,
      "step": 1035
    },
    {
      "epoch": 1.8434163701067616,
      "grad_norm": 0.9977415418263439,
      "learning_rate": 1.5048108659766693e-07,
      "loss": 0.0246,
      "step": 1036
    },
    {
      "epoch": 1.8451957295373664,
      "grad_norm": 1.2758345593726452,
      "learning_rate": 1.470972678790711e-07,
      "loss": 0.0402,
      "step": 1037
    },
    {
      "epoch": 1.8469750889679717,
      "grad_norm": 1.321818491410688,
      "learning_rate": 1.437513604386559e-07,
      "loss": 0.0488,
      "step": 1038
    },
    {
      "epoch": 1.8487544483985765,
      "grad_norm": 1.3564010771277493,
      "learning_rate": 1.404433904149266e-07,
      "loss": 0.0319,
      "step": 1039
    },
    {
      "epoch": 1.8505338078291815,
      "grad_norm": 2.1447861358818656,
      "learning_rate": 1.3717338365001943e-07,
      "loss": 0.0353,
      "step": 1040
    },
    {
      "epoch": 1.8523131672597866,
      "grad_norm": 1.621539034569892,
      "learning_rate": 1.3394136568949834e-07,
      "loss": 0.0486,
      "step": 1041
    },
    {
      "epoch": 1.8540925266903914,
      "grad_norm": 1.7080237630248185,
      "learning_rate": 1.307473617821553e-07,
      "loss": 0.0437,
      "step": 1042
    },
    {
      "epoch": 1.8558718861209964,
      "grad_norm": 1.6764300595040629,
      "learning_rate": 1.275913968798137e-07,
      "loss": 0.0406,
      "step": 1043
    },
    {
      "epoch": 1.8576512455516014,
      "grad_norm": 1.5347915498756959,
      "learning_rate": 1.2447349563713186e-07,
      "loss": 0.0478,
      "step": 1044
    },
    {
      "epoch": 1.8594306049822062,
      "grad_norm": 1.0943355790240306,
      "learning_rate": 1.213936824114137e-07,
      "loss": 0.0298,
      "step": 1045
    },
    {
      "epoch": 1.8612099644128115,
      "grad_norm": 1.613150778817604,
      "learning_rate": 1.1835198126241509e-07,
      "loss": 0.0405,
      "step": 1046
    },
    {
      "epoch": 1.8629893238434163,
      "grad_norm": 1.5848325689123757,
      "learning_rate": 1.1534841595215617e-07,
      "loss": 0.0401,
      "step": 1047
    },
    {
      "epoch": 1.8647686832740213,
      "grad_norm": 1.0431301216835898,
      "learning_rate": 1.1238300994473983e-07,
      "loss": 0.023,
      "step": 1048
    },
    {
      "epoch": 1.8665480427046264,
      "grad_norm": 1.3741342743919496,
      "learning_rate": 1.0945578640616183e-07,
      "loss": 0.0453,
      "step": 1049
    },
    {
      "epoch": 1.8683274021352312,
      "grad_norm": 1.2455640295474606,
      "learning_rate": 1.0656676820413603e-07,
      "loss": 0.0241,
      "step": 1050
    },
    {
      "epoch": 1.8701067615658364,
      "grad_norm": 1.1284197355327303,
      "learning_rate": 1.0371597790791166e-07,
      "loss": 0.0297,
      "step": 1051
    },
    {
      "epoch": 1.8718861209964412,
      "grad_norm": 1.5640890914227485,
      "learning_rate": 1.0090343778809908e-07,
      "loss": 0.0455,
      "step": 1052
    },
    {
      "epoch": 1.8736654804270463,
      "grad_norm": 1.853926089016026,
      "learning_rate": 9.812916981649433e-08,
      "loss": 0.0522,
      "step": 1053
    },
    {
      "epoch": 1.8754448398576513,
      "grad_norm": 1.2087726984754221,
      "learning_rate": 9.539319566590766e-08,
      "loss": 0.0386,
      "step": 1054
    },
    {
      "epoch": 1.8772241992882561,
      "grad_norm": 1.3864583312632892,
      "learning_rate": 9.269553670999743e-08,
      "loss": 0.0414,
      "step": 1055
    },
    {
      "epoch": 1.8790035587188612,
      "grad_norm": 1.7413407084585752,
      "learning_rate": 9.003621402309815e-08,
      "loss": 0.0432,
      "step": 1056
    },
    {
      "epoch": 1.8807829181494662,
      "grad_norm": 1.7263371896066506,
      "learning_rate": 8.741524838005888e-08,
      "loss": 0.0462,
      "step": 1057
    },
    {
      "epoch": 1.8825622775800712,
      "grad_norm": 1.3344573548942262,
      "learning_rate": 8.483266025608061e-08,
      "loss": 0.0281,
      "step": 1058
    },
    {
      "epoch": 1.8843416370106763,
      "grad_norm": 1.2682211863586654,
      "learning_rate": 8.228846982655525e-08,
      "loss": 0.0336,
      "step": 1059
    },
    {
      "epoch": 1.886120996441281,
      "grad_norm": 1.3925211114661775,
      "learning_rate": 7.978269696691021e-08,
      "loss": 0.0443,
      "step": 1060
    },
    {
      "epoch": 1.887900355871886,
      "grad_norm": 1.4338224449866763,
      "learning_rate": 7.731536125244965e-08,
      "loss": 0.0389,
      "step": 1061
    },
    {
      "epoch": 1.8896797153024911,
      "grad_norm": 1.5427573154466507,
      "learning_rate": 7.488648195820513e-08,
      "loss": 0.0544,
      "step": 1062
    },
    {
      "epoch": 1.891459074733096,
      "grad_norm": 1.9957246447587085,
      "learning_rate": 7.249607805878245e-08,
      "loss": 0.0492,
      "step": 1063
    },
    {
      "epoch": 1.8932384341637012,
      "grad_norm": 1.5083653751763368,
      "learning_rate": 7.014416822821557e-08,
      "loss": 0.037,
      "step": 1064
    },
    {
      "epoch": 1.895017793594306,
      "grad_norm": 1.095822277468113,
      "learning_rate": 6.783077083981793e-08,
      "loss": 0.0235,
      "step": 1065
    },
    {
      "epoch": 1.896797153024911,
      "grad_norm": 1.4578622440819116,
      "learning_rate": 6.55559039660425e-08,
      "loss": 0.0352,
      "step": 1066
    },
    {
      "epoch": 1.898576512455516,
      "grad_norm": 1.6579309111167273,
      "learning_rate": 6.331958537833693e-08,
      "loss": 0.0428,
      "step": 1067
    },
    {
      "epoch": 1.9003558718861209,
      "grad_norm": 1.5358610493871263,
      "learning_rate": 6.112183254700866e-08,
      "loss": 0.0405,
      "step": 1068
    },
    {
      "epoch": 1.9021352313167261,
      "grad_norm": 1.260740212875412,
      "learning_rate": 5.8962662641083856e-08,
      "loss": 0.0292,
      "step": 1069
    },
    {
      "epoch": 1.903914590747331,
      "grad_norm": 1.4404732905947126,
      "learning_rate": 5.6842092528176516e-08,
      "loss": 0.039,
      "step": 1070
    },
    {
      "epoch": 1.905693950177936,
      "grad_norm": 1.3400934906594497,
      "learning_rate": 5.476013877435626e-08,
      "loss": 0.0332,
      "step": 1071
    },
    {
      "epoch": 1.907473309608541,
      "grad_norm": 1.558440412579011,
      "learning_rate": 5.271681764401848e-08,
      "loss": 0.0396,
      "step": 1072
    },
    {
      "epoch": 1.9092526690391458,
      "grad_norm": 1.0564459406252982,
      "learning_rate": 5.071214509975775e-08,
      "loss": 0.0298,
      "step": 1073
    },
    {
      "epoch": 1.9110320284697508,
      "grad_norm": 1.5703824080677984,
      "learning_rate": 4.8746136802240716e-08,
      "loss": 0.0388,
      "step": 1074
    },
    {
      "epoch": 1.9128113879003559,
      "grad_norm": 1.5587546168148023,
      "learning_rate": 4.6818808110087875e-08,
      "loss": 0.0445,
      "step": 1075
    },
    {
      "epoch": 1.914590747330961,
      "grad_norm": 1.7881392555936386,
      "learning_rate": 4.493017407975087e-08,
      "loss": 0.0562,
      "step": 1076
    },
    {
      "epoch": 1.916370106761566,
      "grad_norm": 1.2093607732055727,
      "learning_rate": 4.308024946539424e-08,
      "loss": 0.0259,
      "step": 1077
    },
    {
      "epoch": 1.9181494661921707,
      "grad_norm": 1.4963960688310853,
      "learning_rate": 4.1269048718783344e-08,
      "loss": 0.0353,
      "step": 1078
    },
    {
      "epoch": 1.9199288256227758,
      "grad_norm": 1.2644520299129427,
      "learning_rate": 3.9496585989167726e-08,
      "loss": 0.0297,
      "step": 1079
    },
    {
      "epoch": 1.9217081850533808,
      "grad_norm": 1.6218651221441036,
      "learning_rate": 3.776287512317345e-08,
      "loss": 0.0399,
      "step": 1080
    },
    {
      "epoch": 1.9234875444839856,
      "grad_norm": 1.3533436539813684,
      "learning_rate": 3.606792966469375e-08,
      "loss": 0.0333,
      "step": 1081
    },
    {
      "epoch": 1.9252669039145909,
      "grad_norm": 1.3436293271314022,
      "learning_rate": 3.4411762854782426e-08,
      "loss": 0.0339,
      "step": 1082
    },
    {
      "epoch": 1.9270462633451957,
      "grad_norm": 1.2424682477808022,
      "learning_rate": 3.279438763155174e-08,
      "loss": 0.0222,
      "step": 1083
    },
    {
      "epoch": 1.9288256227758007,
      "grad_norm": 1.4625796682975236,
      "learning_rate": 3.121581663007134e-08,
      "loss": 0.0441,
      "step": 1084
    },
    {
      "epoch": 1.9306049822064058,
      "grad_norm": 1.6645394110511582,
      "learning_rate": 2.967606218226837e-08,
      "loss": 0.0464,
      "step": 1085
    },
    {
      "epoch": 1.9323843416370106,
      "grad_norm": 1.106478723782067,
      "learning_rate": 2.8175136316832e-08,
      "loss": 0.027,
      "step": 1086
    },
    {
      "epoch": 1.9341637010676158,
      "grad_norm": 1.8103239757671263,
      "learning_rate": 2.6713050759120117e-08,
      "loss": 0.0592,
      "step": 1087
    },
    {
      "epoch": 1.9359430604982206,
      "grad_norm": 1.2901120841696607,
      "learning_rate": 2.528981693106558e-08,
      "loss": 0.03,
      "step": 1088
    },
    {
      "epoch": 1.9377224199288257,
      "grad_norm": 1.422552779123322,
      "learning_rate": 2.3905445951089013e-08,
      "loss": 0.0384,
      "step": 1089
    },
    {
      "epoch": 1.9395017793594307,
      "grad_norm": 1.8539314540512652,
      "learning_rate": 2.2559948634011673e-08,
      "loss": 0.0365,
      "step": 1090
    },
    {
      "epoch": 1.9412811387900355,
      "grad_norm": 1.5661200081043165,
      "learning_rate": 2.125333549096942e-08,
      "loss": 0.0381,
      "step": 1091
    },
    {
      "epoch": 1.9430604982206405,
      "grad_norm": 1.1789336601585891,
      "learning_rate": 1.9985616729332747e-08,
      "loss": 0.0343,
      "step": 1092
    },
    {
      "epoch": 1.9448398576512456,
      "grad_norm": 1.4173486867137657,
      "learning_rate": 1.8756802252625773e-08,
      "loss": 0.04,
      "step": 1093
    },
    {
      "epoch": 1.9466192170818504,
      "grad_norm": 1.2219710901824574,
      "learning_rate": 1.75669016604485e-08,
      "loss": 0.0337,
      "step": 1094
    },
    {
      "epoch": 1.9483985765124556,
      "grad_norm": 1.5078105550310874,
      "learning_rate": 1.6415924248403547e-08,
      "loss": 0.0292,
      "step": 1095
    },
    {
      "epoch": 1.9501779359430604,
      "grad_norm": 1.5769468954516352,
      "learning_rate": 1.5303879008021773e-08,
      "loss": 0.046,
      "step": 1096
    },
    {
      "epoch": 1.9519572953736655,
      "grad_norm": 1.5417189576459365,
      "learning_rate": 1.4230774626691756e-08,
      "loss": 0.0338,
      "step": 1097
    },
    {
      "epoch": 1.9537366548042705,
      "grad_norm": 1.4625492420378,
      "learning_rate": 1.3196619487594875e-08,
      "loss": 0.0419,
      "step": 1098
    },
    {
      "epoch": 1.9555160142348753,
      "grad_norm": 1.6253293099375465,
      "learning_rate": 1.2201421669636448e-08,
      "loss": 0.0449,
      "step": 1099
    },
    {
      "epoch": 1.9572953736654806,
      "grad_norm": 1.2845684493583356,
      "learning_rate": 1.1245188947384133e-08,
      "loss": 0.0321,
      "step": 1100
    },
    {
      "epoch": 1.9590747330960854,
      "grad_norm": 1.5211741158195164,
      "learning_rate": 1.0327928791006858e-08,
      "loss": 0.0394,
      "step": 1101
    },
    {
      "epoch": 1.9608540925266904,
      "grad_norm": 1.32502918966484,
      "learning_rate": 9.449648366217645e-09,
      "loss": 0.0388,
      "step": 1102
    },
    {
      "epoch": 1.9626334519572954,
      "grad_norm": 1.2306329274971302,
      "learning_rate": 8.61035453421588e-09,
      "loss": 0.0326,
      "step": 1103
    },
    {
      "epoch": 1.9644128113879002,
      "grad_norm": 1.259497000375209,
      "learning_rate": 7.81005385163458e-09,
      "loss": 0.0337,
      "step": 1104
    },
    {
      "epoch": 1.9661921708185055,
      "grad_norm": 1.0711614937711706,
      "learning_rate": 7.048752570488205e-09,
      "loss": 0.0349,
      "step": 1105
    },
    {
      "epoch": 1.9679715302491103,
      "grad_norm": 1.671578527538187,
      "learning_rate": 6.326456638125478e-09,
      "loss": 0.0491,
      "step": 1106
    },
    {
      "epoch": 1.9697508896797153,
      "grad_norm": 1.2763717849779994,
      "learning_rate": 5.643171697183314e-09,
      "loss": 0.0346,
      "step": 1107
    },
    {
      "epoch": 1.9715302491103204,
      "grad_norm": 1.3179124608704447,
      "learning_rate": 4.998903085539075e-09,
      "loss": 0.039,
      "step": 1108
    },
    {
      "epoch": 1.9733096085409252,
      "grad_norm": 1.3806194230293778,
      "learning_rate": 4.393655836272825e-09,
      "loss": 0.0332,
      "step": 1109
    },
    {
      "epoch": 1.9750889679715302,
      "grad_norm": 1.2918039555657261,
      "learning_rate": 3.8274346776262514e-09,
      "loss": 0.0331,
      "step": 1110
    },
    {
      "epoch": 1.9768683274021353,
      "grad_norm": 1.4506145536156116,
      "learning_rate": 3.300244032966582e-09,
      "loss": 0.0428,
      "step": 1111
    },
    {
      "epoch": 1.97864768683274,
      "grad_norm": 1.4951087403195147,
      "learning_rate": 2.8120880207493928e-09,
      "loss": 0.0332,
      "step": 1112
    },
    {
      "epoch": 1.9804270462633453,
      "grad_norm": 1.1081908867353776,
      "learning_rate": 2.362970454491409e-09,
      "loss": 0.0236,
      "step": 1113
    },
    {
      "epoch": 1.9822064056939501,
      "grad_norm": 1.5718787811958137,
      "learning_rate": 1.952894842735531e-09,
      "loss": 0.046,
      "step": 1114
    },
    {
      "epoch": 1.9839857651245552,
      "grad_norm": 1.048518531118998,
      "learning_rate": 1.5818643890258555e-09,
      "loss": 0.0284,
      "step": 1115
    },
    {
      "epoch": 1.9857651245551602,
      "grad_norm": 1.8576913911909543,
      "learning_rate": 1.2498819918843609e-09,
      "loss": 0.0395,
      "step": 1116
    },
    {
      "epoch": 1.987544483985765,
      "grad_norm": 1.7624658512036215,
      "learning_rate": 9.569502447837053e-10,
      "loss": 0.041,
      "step": 1117
    },
    {
      "epoch": 1.9893238434163703,
      "grad_norm": 1.4732334931781161,
      "learning_rate": 7.03071436131686e-10,
      "loss": 0.0331,
      "step": 1118
    },
    {
      "epoch": 1.991103202846975,
      "grad_norm": 1.3765988284402233,
      "learning_rate": 4.882475492506977e-10,
      "loss": 0.0368,
      "step": 1119
    },
    {
      "epoch": 1.99288256227758,
      "grad_norm": 1.4137741396402363,
      "learning_rate": 3.124802623627465e-10,
      "loss": 0.0376,
      "step": 1120
    },
    {
      "epoch": 1.9946619217081851,
      "grad_norm": 1.447737573132864,
      "learning_rate": 1.7577094857557097e-10,
      "loss": 0.0476,
      "step": 1121
    },
    {
      "epoch": 1.99644128113879,
      "grad_norm": 1.1791771499083685,
      "learning_rate": 7.812067587487093e-11,
      "loss": 0.0344,
      "step": 1122
    },
    {
      "epoch": 1.998220640569395,
      "grad_norm": 1.5023715918920821,
      "learning_rate": 1.9530207111539967e-11,
      "loss": 0.0332,
      "step": 1123
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1718081619906127,
      "learning_rate": 0.0,
      "loss": 0.0332,
      "step": 1124
    },
    {
      "epoch": 2.0,
      "step": 1124,
      "total_flos": 4998105219072.0,
      "train_loss": 0.08123247806327517,
      "train_runtime": 863.7561,
      "train_samples_per_second": 10.403,
      "train_steps_per_second": 1.301
    }
  ],
  "logging_steps": 1,
  "max_steps": 1124,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "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
}