{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999549082382648,
  "eval_steps": 500,
  "global_step": 11088,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.018352347026198e-05,
      "grad_norm": 66.02296371930943,
      "learning_rate": 0.0,
      "loss": 2.1194,
      "step": 1
    },
    {
      "epoch": 0.00018036704694052397,
      "grad_norm": 60.49348856036594,
      "learning_rate": 4.773623799730706e-07,
      "loss": 2.0397,
      "step": 2
    },
    {
      "epoch": 0.0002705505704107859,
      "grad_norm": 36.67498721881168,
      "learning_rate": 7.566014715123208e-07,
      "loss": 1.7394,
      "step": 3
    },
    {
      "epoch": 0.00036073409388104793,
      "grad_norm": 52.48251662012074,
      "learning_rate": 9.547247599461412e-07,
      "loss": 2.1386,
      "step": 4
    },
    {
      "epoch": 0.0004509176173513099,
      "grad_norm": 45.11166700533244,
      "learning_rate": 1.108401121501769e-06,
      "loss": 1.6264,
      "step": 5
    },
    {
      "epoch": 0.0005411011408215718,
      "grad_norm": 48.47355764760438,
      "learning_rate": 1.2339638514853914e-06,
      "loss": 1.8425,
      "step": 6
    },
    {
      "epoch": 0.0006312846642918339,
      "grad_norm": 38.37405046412793,
      "learning_rate": 1.3401256270225321e-06,
      "loss": 1.5285,
      "step": 7
    },
    {
      "epoch": 0.0007214681877620959,
      "grad_norm": 24.79345710337142,
      "learning_rate": 1.4320871399192119e-06,
      "loss": 1.4702,
      "step": 8
    },
    {
      "epoch": 0.0008116517112323579,
      "grad_norm": 14.442534654657283,
      "learning_rate": 1.5132029430246416e-06,
      "loss": 1.3272,
      "step": 9
    },
    {
      "epoch": 0.0009018352347026198,
      "grad_norm": 16.00263764574746,
      "learning_rate": 1.5857635014748399e-06,
      "loss": 1.4375,
      "step": 10
    },
    {
      "epoch": 0.0009920187581728818,
      "grad_norm": 16.700138406693952,
      "learning_rate": 1.6514025108267924e-06,
      "loss": 1.5854,
      "step": 11
    },
    {
      "epoch": 0.0010822022816431437,
      "grad_norm": 11.664910175159141,
      "learning_rate": 1.711326231458462e-06,
      "loss": 1.4252,
      "step": 12
    },
    {
      "epoch": 0.0011723858051134058,
      "grad_norm": 5.90237509556948,
      "learning_rate": 1.7664507107987104e-06,
      "loss": 1.3125,
      "step": 13
    },
    {
      "epoch": 0.0012625693285836677,
      "grad_norm": 4.731143465442639,
      "learning_rate": 1.8174880069956024e-06,
      "loss": 1.263,
      "step": 14
    },
    {
      "epoch": 0.0013527528520539298,
      "grad_norm": 4.038155016602303,
      "learning_rate": 1.8650025930140899e-06,
      "loss": 1.2224,
      "step": 15
    },
    {
      "epoch": 0.0014429363755241917,
      "grad_norm": 4.534637447783109,
      "learning_rate": 1.9094495198922823e-06,
      "loss": 1.3112,
      "step": 16
    },
    {
      "epoch": 0.0015331198989944536,
      "grad_norm": 3.7201595680305015,
      "learning_rate": 1.9512009899507514e-06,
      "loss": 1.1645,
      "step": 17
    },
    {
      "epoch": 0.0016233034224647158,
      "grad_norm": 3.2848393744415185,
      "learning_rate": 1.990565322997712e-06,
      "loss": 1.2201,
      "step": 18
    },
    {
      "epoch": 0.0017134869459349777,
      "grad_norm": 2.991184953109061,
      "learning_rate": 2.027800787770518e-06,
      "loss": 1.1098,
      "step": 19
    },
    {
      "epoch": 0.0018036704694052396,
      "grad_norm": 3.4548499860643003,
      "learning_rate": 2.06312588144791e-06,
      "loss": 1.2607,
      "step": 20
    },
    {
      "epoch": 0.0018938539928755017,
      "grad_norm": 2.9046790686654798,
      "learning_rate": 2.0967270985348526e-06,
      "loss": 1.1201,
      "step": 21
    },
    {
      "epoch": 0.0019840375163457636,
      "grad_norm": 4.576912468748122,
      "learning_rate": 2.128764890799863e-06,
      "loss": 1.0585,
      "step": 22
    },
    {
      "epoch": 0.0020742210398160257,
      "grad_norm": 2.256751738578352,
      "learning_rate": 2.1593783012990145e-06,
      "loss": 1.0679,
      "step": 23
    },
    {
      "epoch": 0.0021644045632862874,
      "grad_norm": 2.18881414993715,
      "learning_rate": 2.188688611431533e-06,
      "loss": 1.189,
      "step": 24
    },
    {
      "epoch": 0.0022545880867565495,
      "grad_norm": 2.2330777330411493,
      "learning_rate": 2.216802243003538e-06,
      "loss": 1.2215,
      "step": 25
    },
    {
      "epoch": 0.0023447716102268116,
      "grad_norm": 2.0820329676866156,
      "learning_rate": 2.243813090771781e-06,
      "loss": 1.0678,
      "step": 26
    },
    {
      "epoch": 0.0024349551336970737,
      "grad_norm": 2.2198059363214564,
      "learning_rate": 2.269804414536962e-06,
      "loss": 1.1955,
      "step": 27
    },
    {
      "epoch": 0.0025251386571673354,
      "grad_norm": 1.8676160376473958,
      "learning_rate": 2.2948503869686733e-06,
      "loss": 1.1228,
      "step": 28
    },
    {
      "epoch": 0.0026153221806375975,
      "grad_norm": 2.4714909077885547,
      "learning_rate": 2.3190173696980436e-06,
      "loss": 0.698,
      "step": 29
    },
    {
      "epoch": 0.0027055057041078597,
      "grad_norm": 2.140920688227387,
      "learning_rate": 2.3423649729871604e-06,
      "loss": 1.1151,
      "step": 30
    },
    {
      "epoch": 0.0027956892275781214,
      "grad_norm": 1.7424667203070374,
      "learning_rate": 2.364946941580084e-06,
      "loss": 1.1525,
      "step": 31
    },
    {
      "epoch": 0.0028858727510483835,
      "grad_norm": 1.9299507116042631,
      "learning_rate": 2.3868118998653532e-06,
      "loss": 1.2207,
      "step": 32
    },
    {
      "epoch": 0.0029760562745186456,
      "grad_norm": 2.2759694685203846,
      "learning_rate": 2.408003982339113e-06,
      "loss": 0.6451,
      "step": 33
    },
    {
      "epoch": 0.0030662397979889073,
      "grad_norm": 1.509968298648748,
      "learning_rate": 2.4285633699238223e-06,
      "loss": 1.0106,
      "step": 34
    },
    {
      "epoch": 0.0031564233214591694,
      "grad_norm": 2.0406752014995275,
      "learning_rate": 2.4485267485243007e-06,
      "loss": 1.1093,
      "step": 35
    },
    {
      "epoch": 0.0032466068449294315,
      "grad_norm": 1.532303428302214,
      "learning_rate": 2.467927702970783e-06,
      "loss": 1.0668,
      "step": 36
    },
    {
      "epoch": 0.003336790368399693,
      "grad_norm": 1.4794440031860727,
      "learning_rate": 2.4867970569753584e-06,
      "loss": 1.0816,
      "step": 37
    },
    {
      "epoch": 0.0034269738918699553,
      "grad_norm": 1.887697333321474,
      "learning_rate": 2.5051631677435883e-06,
      "loss": 0.9768,
      "step": 38
    },
    {
      "epoch": 0.0035171574153402174,
      "grad_norm": 1.6430300568276772,
      "learning_rate": 2.523052182311031e-06,
      "loss": 0.9466,
      "step": 39
    },
    {
      "epoch": 0.003607340938810479,
      "grad_norm": 1.7602849907319529,
      "learning_rate": 2.540488261420981e-06,
      "loss": 1.1248,
      "step": 40
    },
    {
      "epoch": 0.0036975244622807412,
      "grad_norm": 1.7329230201912291,
      "learning_rate": 2.557493775753984e-06,
      "loss": 0.9319,
      "step": 41
    },
    {
      "epoch": 0.0037877079857510034,
      "grad_norm": 1.429045319294377,
      "learning_rate": 2.5740894785079235e-06,
      "loss": 1.0066,
      "step": 42
    },
    {
      "epoch": 0.0038778915092212655,
      "grad_norm": 1.3916859811976787,
      "learning_rate": 2.5902946576685834e-06,
      "loss": 1.0534,
      "step": 43
    },
    {
      "epoch": 0.003968075032691527,
      "grad_norm": 1.7406517067851646,
      "learning_rate": 2.606127270772933e-06,
      "loss": 0.9967,
      "step": 44
    },
    {
      "epoch": 0.004058258556161789,
      "grad_norm": 1.7641069756333576,
      "learning_rate": 2.62160406452641e-06,
      "loss": 0.9453,
      "step": 45
    },
    {
      "epoch": 0.004148442079632051,
      "grad_norm": 1.5659129443324444,
      "learning_rate": 2.636740681272085e-06,
      "loss": 1.0826,
      "step": 46
    },
    {
      "epoch": 0.004238625603102313,
      "grad_norm": 1.6587246814856667,
      "learning_rate": 2.651551754008722e-06,
      "loss": 1.1542,
      "step": 47
    },
    {
      "epoch": 0.004328809126572575,
      "grad_norm": 1.5336112675805589,
      "learning_rate": 2.6660509914046035e-06,
      "loss": 1.0475,
      "step": 48
    },
    {
      "epoch": 0.004418992650042837,
      "grad_norm": 1.6370489474781165,
      "learning_rate": 2.6802512540450642e-06,
      "loss": 1.0133,
      "step": 49
    },
    {
      "epoch": 0.004509176173513099,
      "grad_norm": 1.733145613164302,
      "learning_rate": 2.694164622976609e-06,
      "loss": 1.0843,
      "step": 50
    },
    {
      "epoch": 0.0045993596969833616,
      "grad_norm": 1.4998985892029721,
      "learning_rate": 2.707802461463072e-06,
      "loss": 0.9705,
      "step": 51
    },
    {
      "epoch": 0.004689543220453623,
      "grad_norm": 1.50826964094594,
      "learning_rate": 2.7211754707448516e-06,
      "loss": 0.9997,
      "step": 52
    },
    {
      "epoch": 0.004779726743923885,
      "grad_norm": 1.372029085999296,
      "learning_rate": 2.734293740486721e-06,
      "loss": 0.9664,
      "step": 53
    },
    {
      "epoch": 0.0048699102673941475,
      "grad_norm": 1.4251831472389183,
      "learning_rate": 2.747166794510033e-06,
      "loss": 1.0324,
      "step": 54
    },
    {
      "epoch": 0.004960093790864409,
      "grad_norm": 1.5558223703682974,
      "learning_rate": 2.759803632328562e-06,
      "loss": 0.9857,
      "step": 55
    },
    {
      "epoch": 0.005050277314334671,
      "grad_norm": 1.5265646115525635,
      "learning_rate": 2.772212766941744e-06,
      "loss": 1.1301,
      "step": 56
    },
    {
      "epoch": 0.005140460837804933,
      "grad_norm": 1.529281596647021,
      "learning_rate": 2.7844022592828385e-06,
      "loss": 1.0702,
      "step": 57
    },
    {
      "epoch": 0.005230644361275195,
      "grad_norm": 1.5177975057859154,
      "learning_rate": 2.7963797496711145e-06,
      "loss": 0.9553,
      "step": 58
    },
    {
      "epoch": 0.005320827884745457,
      "grad_norm": 1.5686136802017456,
      "learning_rate": 2.80815248657541e-06,
      "loss": 1.055,
      "step": 59
    },
    {
      "epoch": 0.005411011408215719,
      "grad_norm": 1.416505252767802,
      "learning_rate": 2.819727352960231e-06,
      "loss": 0.6207,
      "step": 60
    },
    {
      "epoch": 0.005501194931685981,
      "grad_norm": 1.7102699545209337,
      "learning_rate": 2.8311108904541717e-06,
      "loss": 1.0041,
      "step": 61
    },
    {
      "epoch": 0.005591378455156243,
      "grad_norm": 1.5901352523762278,
      "learning_rate": 2.842309321553155e-06,
      "loss": 0.9974,
      "step": 62
    },
    {
      "epoch": 0.005681561978626505,
      "grad_norm": 1.3391872833048108,
      "learning_rate": 2.8533285700471737e-06,
      "loss": 0.9267,
      "step": 63
    },
    {
      "epoch": 0.005771745502096767,
      "grad_norm": 1.4323162654729502,
      "learning_rate": 2.8641742798384237e-06,
      "loss": 1.0717,
      "step": 64
    },
    {
      "epoch": 0.005861929025567029,
      "grad_norm": 1.37883509811669,
      "learning_rate": 2.874851832300479e-06,
      "loss": 1.0083,
      "step": 65
    },
    {
      "epoch": 0.005952112549037291,
      "grad_norm": 1.609741909830849,
      "learning_rate": 2.8853663623121834e-06,
      "loss": 1.0416,
      "step": 66
    },
    {
      "epoch": 0.006042296072507553,
      "grad_norm": 1.479463663864893,
      "learning_rate": 2.895722773085839e-06,
      "loss": 1.0427,
      "step": 67
    },
    {
      "epoch": 0.0061324795959778146,
      "grad_norm": 1.3251863618462696,
      "learning_rate": 2.905925749896893e-06,
      "loss": 1.0159,
      "step": 68
    },
    {
      "epoch": 0.006222663119448077,
      "grad_norm": 1.4638206062260204,
      "learning_rate": 2.915979772811335e-06,
      "loss": 1.0086,
      "step": 69
    },
    {
      "epoch": 0.006312846642918339,
      "grad_norm": 1.4380954670101382,
      "learning_rate": 2.925889128497372e-06,
      "loss": 0.9551,
      "step": 70
    },
    {
      "epoch": 0.0064030301663886005,
      "grad_norm": 1.6057626868193275,
      "learning_rate": 2.9356579211992906e-06,
      "loss": 1.0809,
      "step": 71
    },
    {
      "epoch": 0.006493213689858863,
      "grad_norm": 1.3981864320221378,
      "learning_rate": 2.9452900829438533e-06,
      "loss": 0.9879,
      "step": 72
    },
    {
      "epoch": 0.006583397213329125,
      "grad_norm": 1.6462499920056457,
      "learning_rate": 2.954789383042727e-06,
      "loss": 1.0289,
      "step": 73
    },
    {
      "epoch": 0.006673580736799386,
      "grad_norm": 1.298145177736712,
      "learning_rate": 2.9641594369484293e-06,
      "loss": 0.6913,
      "step": 74
    },
    {
      "epoch": 0.006763764260269649,
      "grad_norm": 1.5941823485115454,
      "learning_rate": 2.9734037145158586e-06,
      "loss": 0.9994,
      "step": 75
    },
    {
      "epoch": 0.006853947783739911,
      "grad_norm": 1.585123388740365,
      "learning_rate": 2.982525547716659e-06,
      "loss": 0.9789,
      "step": 76
    },
    {
      "epoch": 0.006944131307210172,
      "grad_norm": 2.3170044819636617,
      "learning_rate": 2.9915281378493246e-06,
      "loss": 1.0538,
      "step": 77
    },
    {
      "epoch": 0.007034314830680435,
      "grad_norm": 1.6136549546835155,
      "learning_rate": 3.000414562284102e-06,
      "loss": 1.033,
      "step": 78
    },
    {
      "epoch": 0.0071244983541506966,
      "grad_norm": 1.3822294396099242,
      "learning_rate": 3.009187780778246e-06,
      "loss": 0.9971,
      "step": 79
    },
    {
      "epoch": 0.007214681877620958,
      "grad_norm": 1.8241264761024305,
      "learning_rate": 3.017850641394051e-06,
      "loss": 1.0066,
      "step": 80
    },
    {
      "epoch": 0.007304865401091221,
      "grad_norm": 1.6106731031322925,
      "learning_rate": 3.0264058860492832e-06,
      "loss": 0.9055,
      "step": 81
    },
    {
      "epoch": 0.0073950489245614825,
      "grad_norm": 1.4644992270632053,
      "learning_rate": 3.0348561557270548e-06,
      "loss": 1.0557,
      "step": 82
    },
    {
      "epoch": 0.007485232448031745,
      "grad_norm": 0.9382787331222585,
      "learning_rate": 3.043203995369939e-06,
      "loss": 0.6496,
      "step": 83
    },
    {
      "epoch": 0.007575415971502007,
      "grad_norm": 1.5454446363896404,
      "learning_rate": 3.051451858480994e-06,
      "loss": 1.0919,
      "step": 84
    },
    {
      "epoch": 0.007665599494972268,
      "grad_norm": 1.7436981494263468,
      "learning_rate": 3.05960211145252e-06,
      "loss": 0.9249,
      "step": 85
    },
    {
      "epoch": 0.007755783018442531,
      "grad_norm": 1.3440523029322848,
      "learning_rate": 3.0676570376416543e-06,
      "loss": 0.8969,
      "step": 86
    },
    {
      "epoch": 0.007845966541912792,
      "grad_norm": 1.6008230595786073,
      "learning_rate": 3.0756188412103647e-06,
      "loss": 1.0483,
      "step": 87
    },
    {
      "epoch": 0.007936150065383054,
      "grad_norm": 1.383891702855259,
      "learning_rate": 3.083489650746004e-06,
      "loss": 0.978,
      "step": 88
    },
    {
      "epoch": 0.008026333588853317,
      "grad_norm": 1.4356016810328944,
      "learning_rate": 3.0912715226772975e-06,
      "loss": 1.033,
      "step": 89
    },
    {
      "epoch": 0.008116517112323578,
      "grad_norm": 1.253676063586227,
      "learning_rate": 3.098966444499481e-06,
      "loss": 0.9033,
      "step": 90
    },
    {
      "epoch": 0.00820670063579384,
      "grad_norm": 1.419167256440096,
      "learning_rate": 3.1065763378212426e-06,
      "loss": 1.0298,
      "step": 91
    },
    {
      "epoch": 0.008296884159264103,
      "grad_norm": 1.309482921440906,
      "learning_rate": 3.1141030612451554e-06,
      "loss": 0.9717,
      "step": 92
    },
    {
      "epoch": 0.008387067682734364,
      "grad_norm": 0.9250591375159557,
      "learning_rate": 3.1215484130924052e-06,
      "loss": 0.6803,
      "step": 93
    },
    {
      "epoch": 0.008477251206204626,
      "grad_norm": 1.5707042452170916,
      "learning_rate": 3.128914133981793e-06,
      "loss": 1.0117,
      "step": 94
    },
    {
      "epoch": 0.008567434729674889,
      "grad_norm": 1.538473878407507,
      "learning_rate": 3.136201909272287e-06,
      "loss": 1.0416,
      "step": 95
    },
    {
      "epoch": 0.00865761825314515,
      "grad_norm": 1.6884272112013075,
      "learning_rate": 3.1434133713776735e-06,
      "loss": 0.9467,
      "step": 96
    },
    {
      "epoch": 0.008747801776615412,
      "grad_norm": 1.4869600774694374,
      "learning_rate": 3.15055010196128e-06,
      "loss": 1.0108,
      "step": 97
    },
    {
      "epoch": 0.008837985300085675,
      "grad_norm": 1.626966973253826,
      "learning_rate": 3.157613634018135e-06,
      "loss": 0.9812,
      "step": 98
    },
    {
      "epoch": 0.008928168823555935,
      "grad_norm": 1.2750057089434699,
      "learning_rate": 3.1646054538514336e-06,
      "loss": 0.978,
      "step": 99
    },
    {
      "epoch": 0.009018352347026198,
      "grad_norm": 1.5881112521378564,
      "learning_rate": 3.1715270029496797e-06,
      "loss": 1.027,
      "step": 100
    },
    {
      "epoch": 0.00910853587049646,
      "grad_norm": 1.3067705805405634,
      "learning_rate": 3.1783796797704243e-06,
      "loss": 1.0168,
      "step": 101
    },
    {
      "epoch": 0.009198719393966723,
      "grad_norm": 1.6077350297508795,
      "learning_rate": 3.185164841436142e-06,
      "loss": 0.9562,
      "step": 102
    },
    {
      "epoch": 0.009288902917436984,
      "grad_norm": 2.130492461142175,
      "learning_rate": 3.1918838053473723e-06,
      "loss": 0.9869,
      "step": 103
    },
    {
      "epoch": 0.009379086440907246,
      "grad_norm": 1.3983632499546934,
      "learning_rate": 3.198537850717922e-06,
      "loss": 0.8763,
      "step": 104
    },
    {
      "epoch": 0.009469269964377509,
      "grad_norm": 1.4128274740544848,
      "learning_rate": 3.205128220036622e-06,
      "loss": 0.9601,
      "step": 105
    },
    {
      "epoch": 0.00955945348784777,
      "grad_norm": 1.2900550488991016,
      "learning_rate": 3.2116561204597917e-06,
      "loss": 0.9292,
      "step": 106
    },
    {
      "epoch": 0.009649637011318032,
      "grad_norm": 0.9087963832787456,
      "learning_rate": 3.218122725138335e-06,
      "loss": 0.672,
      "step": 107
    },
    {
      "epoch": 0.009739820534788295,
      "grad_norm": 1.6345539206604685,
      "learning_rate": 3.224529174483104e-06,
      "loss": 0.974,
      "step": 108
    },
    {
      "epoch": 0.009830004058258556,
      "grad_norm": 1.713343283195153,
      "learning_rate": 3.2308765773719435e-06,
      "loss": 0.9356,
      "step": 109
    },
    {
      "epoch": 0.009920187581728818,
      "grad_norm": 1.3499036984180564,
      "learning_rate": 3.2371660123016323e-06,
      "loss": 0.9814,
      "step": 110
    },
    {
      "epoch": 0.010010371105199081,
      "grad_norm": 1.4695111189684726,
      "learning_rate": 3.2433985284876787e-06,
      "loss": 0.9445,
      "step": 111
    },
    {
      "epoch": 0.010100554628669342,
      "grad_norm": 1.3054492576874321,
      "learning_rate": 3.2495751469148143e-06,
      "loss": 0.9712,
      "step": 112
    },
    {
      "epoch": 0.010190738152139604,
      "grad_norm": 2.122659965570853,
      "learning_rate": 3.2556968613407816e-06,
      "loss": 0.9999,
      "step": 113
    },
    {
      "epoch": 0.010280921675609867,
      "grad_norm": 1.5224278160735303,
      "learning_rate": 3.2617646392559094e-06,
      "loss": 1.0274,
      "step": 114
    },
    {
      "epoch": 0.010371105199080128,
      "grad_norm": 1.5665664400682378,
      "learning_rate": 3.2677794228007836e-06,
      "loss": 0.9938,
      "step": 115
    },
    {
      "epoch": 0.01046128872255039,
      "grad_norm": 1.5009374867839762,
      "learning_rate": 3.273742129644185e-06,
      "loss": 1.0027,
      "step": 116
    },
    {
      "epoch": 0.010551472246020653,
      "grad_norm": 1.483008950729265,
      "learning_rate": 3.279653653823352e-06,
      "loss": 0.9702,
      "step": 117
    },
    {
      "epoch": 0.010641655769490914,
      "grad_norm": 1.7124109064023514,
      "learning_rate": 3.285514866548481e-06,
      "loss": 0.9553,
      "step": 118
    },
    {
      "epoch": 0.010731839292961176,
      "grad_norm": 1.6421754323005993,
      "learning_rate": 3.2913266169732838e-06,
      "loss": 1.0205,
      "step": 119
    },
    {
      "epoch": 0.010822022816431439,
      "grad_norm": 1.6933978823251077,
      "learning_rate": 3.2970897329333017e-06,
      "loss": 0.96,
      "step": 120
    },
    {
      "epoch": 0.0109122063399017,
      "grad_norm": 1.5044091114698361,
      "learning_rate": 3.302805021653585e-06,
      "loss": 0.977,
      "step": 121
    },
    {
      "epoch": 0.011002389863371962,
      "grad_norm": 1.4929744382996442,
      "learning_rate": 3.3084732704272426e-06,
      "loss": 0.9945,
      "step": 122
    },
    {
      "epoch": 0.011092573386842225,
      "grad_norm": 1.5214451208180368,
      "learning_rate": 3.314095247266304e-06,
      "loss": 0.9832,
      "step": 123
    },
    {
      "epoch": 0.011182756910312485,
      "grad_norm": 4.41161241979121,
      "learning_rate": 3.3196717015262255e-06,
      "loss": 0.9796,
      "step": 124
    },
    {
      "epoch": 0.011272940433782748,
      "grad_norm": 1.3804973467642803,
      "learning_rate": 3.325203364505307e-06,
      "loss": 0.9445,
      "step": 125
    },
    {
      "epoch": 0.01136312395725301,
      "grad_norm": 1.6193718647309976,
      "learning_rate": 3.3306909500202442e-06,
      "loss": 1.0167,
      "step": 126
    },
    {
      "epoch": 0.011453307480723271,
      "grad_norm": 1.840943220573555,
      "learning_rate": 3.3361351549589145e-06,
      "loss": 1.0127,
      "step": 127
    },
    {
      "epoch": 0.011543491004193534,
      "grad_norm": 1.7126562812943686,
      "learning_rate": 3.341536659811494e-06,
      "loss": 0.942,
      "step": 128
    },
    {
      "epoch": 0.011633674527663796,
      "grad_norm": 1.5179987144163214,
      "learning_rate": 3.346896129180904e-06,
      "loss": 1.0228,
      "step": 129
    },
    {
      "epoch": 0.011723858051134057,
      "grad_norm": 1.1848101518554086,
      "learning_rate": 3.35221421227355e-06,
      "loss": 0.9521,
      "step": 130
    },
    {
      "epoch": 0.01181404157460432,
      "grad_norm": 1.700822095682767,
      "learning_rate": 3.357491543371255e-06,
      "loss": 1.0363,
      "step": 131
    },
    {
      "epoch": 0.011904225098074582,
      "grad_norm": 1.3564199139034572,
      "learning_rate": 3.3627287422852543e-06,
      "loss": 0.923,
      "step": 132
    },
    {
      "epoch": 0.011994408621544843,
      "grad_norm": 1.2296430936015161,
      "learning_rate": 3.3679264147930497e-06,
      "loss": 0.9523,
      "step": 133
    },
    {
      "epoch": 0.012084592145015106,
      "grad_norm": 1.6067241577425437,
      "learning_rate": 3.37308515305891e-06,
      "loss": 0.9083,
      "step": 134
    },
    {
      "epoch": 0.012174775668485368,
      "grad_norm": 1.3291846592003231,
      "learning_rate": 3.3782055360387313e-06,
      "loss": 1.0255,
      "step": 135
    },
    {
      "epoch": 0.012264959191955629,
      "grad_norm": 1.4234698228065954,
      "learning_rate": 3.3832881298699633e-06,
      "loss": 0.9396,
      "step": 136
    },
    {
      "epoch": 0.012355142715425892,
      "grad_norm": 1.3603456035628638,
      "learning_rate": 3.388333488247249e-06,
      "loss": 0.9408,
      "step": 137
    },
    {
      "epoch": 0.012445326238896154,
      "grad_norm": 1.3832218835738874,
      "learning_rate": 3.393342152784406e-06,
      "loss": 0.9931,
      "step": 138
    },
    {
      "epoch": 0.012535509762366415,
      "grad_norm": 1.5337012775520038,
      "learning_rate": 3.3983146533633376e-06,
      "loss": 0.9874,
      "step": 139
    },
    {
      "epoch": 0.012625693285836678,
      "grad_norm": 1.8065828221327407,
      "learning_rate": 3.403251508470442e-06,
      "loss": 1.0248,
      "step": 140
    },
    {
      "epoch": 0.01271587680930694,
      "grad_norm": 1.4716125351457232,
      "learning_rate": 3.408153225521043e-06,
      "loss": 0.9081,
      "step": 141
    },
    {
      "epoch": 0.012806060332777201,
      "grad_norm": 1.5394326166792025,
      "learning_rate": 3.413020301172361e-06,
      "loss": 0.8845,
      "step": 142
    },
    {
      "epoch": 0.012896243856247463,
      "grad_norm": 1.3258926929218728,
      "learning_rate": 3.4178532216255024e-06,
      "loss": 0.9521,
      "step": 143
    },
    {
      "epoch": 0.012986427379717726,
      "grad_norm": 1.4256441233287624,
      "learning_rate": 3.422652462916924e-06,
      "loss": 1.0253,
      "step": 144
    },
    {
      "epoch": 0.013076610903187987,
      "grad_norm": 1.3345485851295062,
      "learning_rate": 3.4274184911998124e-06,
      "loss": 0.8527,
      "step": 145
    },
    {
      "epoch": 0.01316679442665825,
      "grad_norm": 1.5979471478130869,
      "learning_rate": 3.4321517630157976e-06,
      "loss": 1.0822,
      "step": 146
    },
    {
      "epoch": 0.013256977950128512,
      "grad_norm": 1.5910008224446817,
      "learning_rate": 3.4368527255573845e-06,
      "loss": 1.0337,
      "step": 147
    },
    {
      "epoch": 0.013347161473598773,
      "grad_norm": 1.3902575832572395,
      "learning_rate": 3.4415218169214994e-06,
      "loss": 0.9371,
      "step": 148
    },
    {
      "epoch": 0.013437344997069035,
      "grad_norm": 1.2876660730602671,
      "learning_rate": 3.4461594663544882e-06,
      "loss": 0.9571,
      "step": 149
    },
    {
      "epoch": 0.013527528520539298,
      "grad_norm": 1.5190416346476383,
      "learning_rate": 3.450766094488929e-06,
      "loss": 0.934,
      "step": 150
    },
    {
      "epoch": 0.013617712044009559,
      "grad_norm": 1.5130216912569732,
      "learning_rate": 3.4553421135725735e-06,
      "loss": 1.0081,
      "step": 151
    },
    {
      "epoch": 0.013707895567479821,
      "grad_norm": 1.3633965386116058,
      "learning_rate": 3.45988792768973e-06,
      "loss": 0.968,
      "step": 152
    },
    {
      "epoch": 0.013798079090950084,
      "grad_norm": 1.350288275563416,
      "learning_rate": 3.464403932975393e-06,
      "loss": 1.0017,
      "step": 153
    },
    {
      "epoch": 0.013888262614420345,
      "grad_norm": 1.5173813386679915,
      "learning_rate": 3.468890517822395e-06,
      "loss": 1.0006,
      "step": 154
    },
    {
      "epoch": 0.013978446137890607,
      "grad_norm": 1.7178274279551795,
      "learning_rate": 3.473348063081853e-06,
      "loss": 1.0086,
      "step": 155
    },
    {
      "epoch": 0.01406862966136087,
      "grad_norm": 1.2847037402819743,
      "learning_rate": 3.4777769422571727e-06,
      "loss": 0.8852,
      "step": 156
    },
    {
      "epoch": 0.01415881318483113,
      "grad_norm": 1.3896423468385941,
      "learning_rate": 3.4821775216918497e-06,
      "loss": 0.9487,
      "step": 157
    },
    {
      "epoch": 0.014248996708301393,
      "grad_norm": 1.3453327438833975,
      "learning_rate": 3.4865501607513164e-06,
      "loss": 0.8987,
      "step": 158
    },
    {
      "epoch": 0.014339180231771656,
      "grad_norm": 1.5046818306282326,
      "learning_rate": 3.4908952119990423e-06,
      "loss": 1.0259,
      "step": 159
    },
    {
      "epoch": 0.014429363755241916,
      "grad_norm": 0.9475109608501866,
      "learning_rate": 3.495213021367122e-06,
      "loss": 0.6297,
      "step": 160
    },
    {
      "epoch": 0.014519547278712179,
      "grad_norm": 1.6736653804114394,
      "learning_rate": 3.4995039283215464e-06,
      "loss": 1.0623,
      "step": 161
    },
    {
      "epoch": 0.014609730802182442,
      "grad_norm": 1.494153869171285,
      "learning_rate": 3.5037682660223533e-06,
      "loss": 0.9835,
      "step": 162
    },
    {
      "epoch": 0.014699914325652702,
      "grad_norm": 1.5899601924159634,
      "learning_rate": 3.508006361478857e-06,
      "loss": 1.0224,
      "step": 163
    },
    {
      "epoch": 0.014790097849122965,
      "grad_norm": 1.4361330249982842,
      "learning_rate": 3.5122185357001253e-06,
      "loss": 0.9507,
      "step": 164
    },
    {
      "epoch": 0.014880281372593228,
      "grad_norm": 1.9340598421237254,
      "learning_rate": 3.5164051038408817e-06,
      "loss": 1.0753,
      "step": 165
    },
    {
      "epoch": 0.01497046489606349,
      "grad_norm": 1.4550099606517142,
      "learning_rate": 3.5205663753430093e-06,
      "loss": 0.9908,
      "step": 166
    },
    {
      "epoch": 0.015060648419533751,
      "grad_norm": 1.3785686565952964,
      "learning_rate": 3.5247026540727915e-06,
      "loss": 1.055,
      "step": 167
    },
    {
      "epoch": 0.015150831943004013,
      "grad_norm": 1.446334477548058,
      "learning_rate": 3.5288142384540645e-06,
      "loss": 0.877,
      "step": 168
    },
    {
      "epoch": 0.015241015466474276,
      "grad_norm": 1.3613614881414315,
      "learning_rate": 3.532901421597421e-06,
      "loss": 0.9799,
      "step": 169
    },
    {
      "epoch": 0.015331198989944537,
      "grad_norm": 2.0360050591291663,
      "learning_rate": 3.5369644914255915e-06,
      "loss": 0.9765,
      "step": 170
    },
    {
      "epoch": 0.0154213825134148,
      "grad_norm": 1.4611267130509888,
      "learning_rate": 3.5410037307951596e-06,
      "loss": 0.9807,
      "step": 171
    },
    {
      "epoch": 0.015511566036885062,
      "grad_norm": 1.240548017597279,
      "learning_rate": 3.545019417614725e-06,
      "loss": 0.9877,
      "step": 172
    },
    {
      "epoch": 0.015601749560355323,
      "grad_norm": 1.4009077535207348,
      "learning_rate": 3.5490118249596387e-06,
      "loss": 1.021,
      "step": 173
    },
    {
      "epoch": 0.015691933083825584,
      "grad_norm": 1.388231510780462,
      "learning_rate": 3.5529812211834352e-06,
      "loss": 1.0069,
      "step": 174
    },
    {
      "epoch": 0.015782116607295848,
      "grad_norm": 1.356025339671894,
      "learning_rate": 3.5569278700260707e-06,
      "loss": 0.971,
      "step": 175
    },
    {
      "epoch": 0.01587230013076611,
      "grad_norm": 1.3586156192925642,
      "learning_rate": 3.5608520307190746e-06,
      "loss": 0.9631,
      "step": 176
    },
    {
      "epoch": 0.01596248365423637,
      "grad_norm": 2.988045716820703,
      "learning_rate": 3.564753958087731e-06,
      "loss": 0.9307,
      "step": 177
    },
    {
      "epoch": 0.016052667177706634,
      "grad_norm": 1.3237516766646504,
      "learning_rate": 3.5686339026503684e-06,
      "loss": 0.9394,
      "step": 178
    },
    {
      "epoch": 0.016142850701176895,
      "grad_norm": 1.3934036525902387,
      "learning_rate": 3.5724921107148806e-06,
      "loss": 1.0542,
      "step": 179
    },
    {
      "epoch": 0.016233034224647155,
      "grad_norm": 1.4700284223338957,
      "learning_rate": 3.576328824472552e-06,
      "loss": 1.0231,
      "step": 180
    },
    {
      "epoch": 0.01632321774811742,
      "grad_norm": 1.527209732146749,
      "learning_rate": 3.5801442820892838e-06,
      "loss": 1.0392,
      "step": 181
    },
    {
      "epoch": 0.01641340127158768,
      "grad_norm": 1.2098239824250996,
      "learning_rate": 3.583938717794313e-06,
      "loss": 0.9668,
      "step": 182
    },
    {
      "epoch": 0.01650358479505794,
      "grad_norm": 1.3215529804874606,
      "learning_rate": 3.5877123619664928e-06,
      "loss": 0.9875,
      "step": 183
    },
    {
      "epoch": 0.016593768318528206,
      "grad_norm": 1.4172414380358462,
      "learning_rate": 3.5914654412182268e-06,
      "loss": 0.9483,
      "step": 184
    },
    {
      "epoch": 0.016683951841998466,
      "grad_norm": 1.346879021800115,
      "learning_rate": 3.595198178477127e-06,
      "loss": 1.0047,
      "step": 185
    },
    {
      "epoch": 0.016774135365468727,
      "grad_norm": 1.3312247958953913,
      "learning_rate": 3.5989107930654757e-06,
      "loss": 0.9068,
      "step": 186
    },
    {
      "epoch": 0.01686431888893899,
      "grad_norm": 1.3847372592877778,
      "learning_rate": 3.6026035007775437e-06,
      "loss": 1.0045,
      "step": 187
    },
    {
      "epoch": 0.016954502412409252,
      "grad_norm": 1.594654948302073,
      "learning_rate": 3.6062765139548636e-06,
      "loss": 0.9863,
      "step": 188
    },
    {
      "epoch": 0.017044685935879513,
      "grad_norm": 1.4829616845416485,
      "learning_rate": 3.6099300415594945e-06,
      "loss": 0.8718,
      "step": 189
    },
    {
      "epoch": 0.017134869459349777,
      "grad_norm": 1.2584916171043654,
      "learning_rate": 3.6135642892453575e-06,
      "loss": 0.8647,
      "step": 190
    },
    {
      "epoch": 0.01722505298282004,
      "grad_norm": 1.258116833375595,
      "learning_rate": 3.6171794594277004e-06,
      "loss": 0.9538,
      "step": 191
    },
    {
      "epoch": 0.0173152365062903,
      "grad_norm": 1.395854119482943,
      "learning_rate": 3.620775751350745e-06,
      "loss": 0.9283,
      "step": 192
    },
    {
      "epoch": 0.017405420029760563,
      "grad_norm": 1.7938841766256675,
      "learning_rate": 3.6243533611535794e-06,
      "loss": 1.0456,
      "step": 193
    },
    {
      "epoch": 0.017495603553230824,
      "grad_norm": 1.4689516781259577,
      "learning_rate": 3.627912481934351e-06,
      "loss": 0.9406,
      "step": 194
    },
    {
      "epoch": 0.017585787076701085,
      "grad_norm": 1.475713495867052,
      "learning_rate": 3.6314533038128e-06,
      "loss": 0.9445,
      "step": 195
    },
    {
      "epoch": 0.01767597060017135,
      "grad_norm": 0.833829521436084,
      "learning_rate": 3.6349760139912048e-06,
      "loss": 0.6147,
      "step": 196
    },
    {
      "epoch": 0.01776615412364161,
      "grad_norm": 1.6006458171189377,
      "learning_rate": 3.638480796813769e-06,
      "loss": 1.0549,
      "step": 197
    },
    {
      "epoch": 0.01785633764711187,
      "grad_norm": 1.1963856187117423,
      "learning_rate": 3.641967833824504e-06,
      "loss": 0.9579,
      "step": 198
    },
    {
      "epoch": 0.017946521170582135,
      "grad_norm": 1.216434136905871,
      "learning_rate": 3.645437303823663e-06,
      "loss": 0.9203,
      "step": 199
    },
    {
      "epoch": 0.018036704694052396,
      "grad_norm": 1.5177188842378768,
      "learning_rate": 3.64888938292275e-06,
      "loss": 0.9405,
      "step": 200
    },
    {
      "epoch": 0.01812688821752266,
      "grad_norm": 1.391308905630317,
      "learning_rate": 3.6523242445981603e-06,
      "loss": 0.9531,
      "step": 201
    },
    {
      "epoch": 0.01821707174099292,
      "grad_norm": 1.1862753526331964,
      "learning_rate": 3.655742059743495e-06,
      "loss": 0.9241,
      "step": 202
    },
    {
      "epoch": 0.018307255264463182,
      "grad_norm": 1.4581264439921908,
      "learning_rate": 3.659142996720576e-06,
      "loss": 0.9758,
      "step": 203
    },
    {
      "epoch": 0.018397438787933446,
      "grad_norm": 1.5206787954050505,
      "learning_rate": 3.6625272214092135e-06,
      "loss": 0.9354,
      "step": 204
    },
    {
      "epoch": 0.018487622311403707,
      "grad_norm": 0.7063738680553531,
      "learning_rate": 3.6658948972557535e-06,
      "loss": 0.5904,
      "step": 205
    },
    {
      "epoch": 0.018577805834873968,
      "grad_norm": 1.4875942456818547,
      "learning_rate": 3.6692461853204432e-06,
      "loss": 0.8882,
      "step": 206
    },
    {
      "epoch": 0.018667989358344232,
      "grad_norm": 1.4506692953056708,
      "learning_rate": 3.672581244323656e-06,
      "loss": 0.8877,
      "step": 207
    },
    {
      "epoch": 0.018758172881814493,
      "grad_norm": 1.4888580085784715,
      "learning_rate": 3.6759002306909926e-06,
      "loss": 0.8934,
      "step": 208
    },
    {
      "epoch": 0.018848356405284754,
      "grad_norm": 1.331568349464172,
      "learning_rate": 3.67920329859731e-06,
      "loss": 0.9301,
      "step": 209
    },
    {
      "epoch": 0.018938539928755018,
      "grad_norm": 1.4571598599614628,
      "learning_rate": 3.6824906000096923e-06,
      "loss": 0.9465,
      "step": 210
    },
    {
      "epoch": 0.01902872345222528,
      "grad_norm": 1.61711238373347,
      "learning_rate": 3.6857622847294067e-06,
      "loss": 0.9711,
      "step": 211
    },
    {
      "epoch": 0.01911890697569554,
      "grad_norm": 1.6931927344882085,
      "learning_rate": 3.6890185004328626e-06,
      "loss": 1.0121,
      "step": 212
    },
    {
      "epoch": 0.019209090499165804,
      "grad_norm": 1.4884032352146828,
      "learning_rate": 3.6922593927116113e-06,
      "loss": 0.8794,
      "step": 213
    },
    {
      "epoch": 0.019299274022636065,
      "grad_norm": 2.07804801626234,
      "learning_rate": 3.695485105111406e-06,
      "loss": 0.9607,
      "step": 214
    },
    {
      "epoch": 0.019389457546106326,
      "grad_norm": 1.2850059259033122,
      "learning_rate": 3.698695779170352e-06,
      "loss": 0.8821,
      "step": 215
    },
    {
      "epoch": 0.01947964106957659,
      "grad_norm": 1.6144344466278264,
      "learning_rate": 3.7018915544561744e-06,
      "loss": 0.9855,
      "step": 216
    },
    {
      "epoch": 0.01956982459304685,
      "grad_norm": 1.5677244960758614,
      "learning_rate": 3.7050725686026164e-06,
      "loss": 0.9486,
      "step": 217
    },
    {
      "epoch": 0.01966000811651711,
      "grad_norm": 1.6149337470339022,
      "learning_rate": 3.708238957345014e-06,
      "loss": 1.0304,
      "step": 218
    },
    {
      "epoch": 0.019750191639987376,
      "grad_norm": 1.360131853247465,
      "learning_rate": 3.7113908545550482e-06,
      "loss": 0.9645,
      "step": 219
    },
    {
      "epoch": 0.019840375163457637,
      "grad_norm": 1.4955724699469295,
      "learning_rate": 3.7145283922747028e-06,
      "loss": 0.9663,
      "step": 220
    },
    {
      "epoch": 0.019930558686927898,
      "grad_norm": 1.3760414081181473,
      "learning_rate": 3.7176517007494612e-06,
      "loss": 1.023,
      "step": 221
    },
    {
      "epoch": 0.020020742210398162,
      "grad_norm": 1.4613194093951618,
      "learning_rate": 3.7207609084607496e-06,
      "loss": 0.8795,
      "step": 222
    },
    {
      "epoch": 0.020110925733868423,
      "grad_norm": 1.5767625271620258,
      "learning_rate": 3.723856142157645e-06,
      "loss": 0.9929,
      "step": 223
    },
    {
      "epoch": 0.020201109257338683,
      "grad_norm": 1.3462833211276368,
      "learning_rate": 3.726937526887885e-06,
      "loss": 0.926,
      "step": 224
    },
    {
      "epoch": 0.020291292780808948,
      "grad_norm": 1.4897786614813215,
      "learning_rate": 3.7300051860281798e-06,
      "loss": 0.9234,
      "step": 225
    },
    {
      "epoch": 0.02038147630427921,
      "grad_norm": 1.7381438125969448,
      "learning_rate": 3.733059241313852e-06,
      "loss": 1.056,
      "step": 226
    },
    {
      "epoch": 0.02047165982774947,
      "grad_norm": 1.4383429667893206,
      "learning_rate": 3.736099812867827e-06,
      "loss": 0.9423,
      "step": 227
    },
    {
      "epoch": 0.020561843351219734,
      "grad_norm": 1.6350804690002878,
      "learning_rate": 3.73912701922898e-06,
      "loss": 1.0147,
      "step": 228
    },
    {
      "epoch": 0.020652026874689994,
      "grad_norm": 0.9426568616159745,
      "learning_rate": 3.742140977379868e-06,
      "loss": 0.6216,
      "step": 229
    },
    {
      "epoch": 0.020742210398160255,
      "grad_norm": 1.3622617653268492,
      "learning_rate": 3.745141802773854e-06,
      "loss": 1.0029,
      "step": 230
    },
    {
      "epoch": 0.02083239392163052,
      "grad_norm": 1.5892746115184941,
      "learning_rate": 3.748129609361645e-06,
      "loss": 0.9385,
      "step": 231
    },
    {
      "epoch": 0.02092257744510078,
      "grad_norm": 1.2419449191480154,
      "learning_rate": 3.7511045096172555e-06,
      "loss": 0.9252,
      "step": 232
    },
    {
      "epoch": 0.02101276096857104,
      "grad_norm": 1.6828892200659438,
      "learning_rate": 3.7540666145634137e-06,
      "loss": 0.9576,
      "step": 233
    },
    {
      "epoch": 0.021102944492041305,
      "grad_norm": 1.4056937495101143,
      "learning_rate": 3.7570160337964225e-06,
      "loss": 0.9191,
      "step": 234
    },
    {
      "epoch": 0.021193128015511566,
      "grad_norm": 2.4936149120789346,
      "learning_rate": 3.7599528755104913e-06,
      "loss": 1.0097,
      "step": 235
    },
    {
      "epoch": 0.021283311538981827,
      "grad_norm": 1.478197111342628,
      "learning_rate": 3.7628772465215515e-06,
      "loss": 0.9327,
      "step": 236
    },
    {
      "epoch": 0.02137349506245209,
      "grad_norm": 1.4359883362713461,
      "learning_rate": 3.7657892522905666e-06,
      "loss": 0.9353,
      "step": 237
    },
    {
      "epoch": 0.021463678585922352,
      "grad_norm": 1.5381320522600845,
      "learning_rate": 3.7686889969463542e-06,
      "loss": 0.8809,
      "step": 238
    },
    {
      "epoch": 0.021553862109392613,
      "grad_norm": 1.5795800328134266,
      "learning_rate": 3.771576583307928e-06,
      "loss": 0.9183,
      "step": 239
    },
    {
      "epoch": 0.021644045632862877,
      "grad_norm": 1.8773776555740083,
      "learning_rate": 3.7744521129063722e-06,
      "loss": 0.9739,
      "step": 240
    },
    {
      "epoch": 0.021734229156333138,
      "grad_norm": 1.4258280707209616,
      "learning_rate": 3.7773156860062653e-06,
      "loss": 0.9626,
      "step": 241
    },
    {
      "epoch": 0.0218244126798034,
      "grad_norm": 1.639646038260417,
      "learning_rate": 3.7801674016266554e-06,
      "loss": 1.0741,
      "step": 242
    },
    {
      "epoch": 0.021914596203273663,
      "grad_norm": 0.6332671098426979,
      "learning_rate": 3.7830073575616035e-06,
      "loss": 0.5657,
      "step": 243
    },
    {
      "epoch": 0.022004779726743924,
      "grad_norm": 1.5322241506563634,
      "learning_rate": 3.785835650400313e-06,
      "loss": 0.9382,
      "step": 244
    },
    {
      "epoch": 0.022094963250214185,
      "grad_norm": 1.800688635705629,
      "learning_rate": 3.7886523755468334e-06,
      "loss": 0.8538,
      "step": 245
    },
    {
      "epoch": 0.02218514677368445,
      "grad_norm": 1.2779937863325717,
      "learning_rate": 3.7914576272393746e-06,
      "loss": 0.8944,
      "step": 246
    },
    {
      "epoch": 0.02227533029715471,
      "grad_norm": 1.8514721683405804,
      "learning_rate": 3.7942514985692284e-06,
      "loss": 1.0779,
      "step": 247
    },
    {
      "epoch": 0.02236551382062497,
      "grad_norm": 1.4833327800592542,
      "learning_rate": 3.797034081499296e-06,
      "loss": 0.967,
      "step": 248
    },
    {
      "epoch": 0.022455697344095235,
      "grad_norm": 1.407537930182467,
      "learning_rate": 3.7998054668822595e-06,
      "loss": 0.9291,
      "step": 249
    },
    {
      "epoch": 0.022545880867565496,
      "grad_norm": 2.862891197959248,
      "learning_rate": 3.8025657444783776e-06,
      "loss": 0.9063,
      "step": 250
    },
    {
      "epoch": 0.022636064391035757,
      "grad_norm": 1.6068861150783726,
      "learning_rate": 3.80531500297293e-06,
      "loss": 0.8569,
      "step": 251
    },
    {
      "epoch": 0.02272624791450602,
      "grad_norm": 1.971047166295798,
      "learning_rate": 3.8080533299933147e-06,
      "loss": 1.0156,
      "step": 252
    },
    {
      "epoch": 0.022816431437976282,
      "grad_norm": 1.4523154844393016,
      "learning_rate": 3.8107808121258067e-06,
      "loss": 0.8925,
      "step": 253
    },
    {
      "epoch": 0.022906614961446543,
      "grad_norm": 1.5321602743189668,
      "learning_rate": 3.813497534931985e-06,
      "loss": 1.0204,
      "step": 254
    },
    {
      "epoch": 0.022996798484916807,
      "grad_norm": 1.7138495202526003,
      "learning_rate": 3.816203582964841e-06,
      "loss": 0.9673,
      "step": 255
    },
    {
      "epoch": 0.023086982008387068,
      "grad_norm": 1.7209257746044708,
      "learning_rate": 3.818899039784565e-06,
      "loss": 1.0049,
      "step": 256
    },
    {
      "epoch": 0.02317716553185733,
      "grad_norm": 1.5308093001227439,
      "learning_rate": 3.821583987974031e-06,
      "loss": 0.9565,
      "step": 257
    },
    {
      "epoch": 0.023267349055327593,
      "grad_norm": 1.668556485046183,
      "learning_rate": 3.8242585091539755e-06,
      "loss": 0.8794,
      "step": 258
    },
    {
      "epoch": 0.023357532578797854,
      "grad_norm": 1.4973716756005326,
      "learning_rate": 3.8269226839978895e-06,
      "loss": 1.0111,
      "step": 259
    },
    {
      "epoch": 0.023447716102268115,
      "grad_norm": 1.8197710271522671,
      "learning_rate": 3.82957659224662e-06,
      "loss": 0.9,
      "step": 260
    },
    {
      "epoch": 0.02353789962573838,
      "grad_norm": 1.8785648154681587,
      "learning_rate": 3.8322203127226855e-06,
      "loss": 0.876,
      "step": 261
    },
    {
      "epoch": 0.02362808314920864,
      "grad_norm": 1.658874160270832,
      "learning_rate": 3.834853923344326e-06,
      "loss": 1.0282,
      "step": 262
    },
    {
      "epoch": 0.0237182666726789,
      "grad_norm": 1.4107223600187546,
      "learning_rate": 3.837477501139285e-06,
      "loss": 0.9277,
      "step": 263
    },
    {
      "epoch": 0.023808450196149165,
      "grad_norm": 1.8620267919749685,
      "learning_rate": 3.840091122258324e-06,
      "loss": 0.9439,
      "step": 264
    },
    {
      "epoch": 0.023898633719619426,
      "grad_norm": 1.5838467200657278,
      "learning_rate": 3.84269486198849e-06,
      "loss": 0.8345,
      "step": 265
    },
    {
      "epoch": 0.023988817243089686,
      "grad_norm": 1.618260861288885,
      "learning_rate": 3.845288794766121e-06,
      "loss": 0.9555,
      "step": 266
    },
    {
      "epoch": 0.02407900076655995,
      "grad_norm": 1.9133911200955995,
      "learning_rate": 3.847872994189619e-06,
      "loss": 0.9385,
      "step": 267
    },
    {
      "epoch": 0.02416918429003021,
      "grad_norm": 1.8240604699551435,
      "learning_rate": 3.8504475330319805e-06,
      "loss": 0.9761,
      "step": 268
    },
    {
      "epoch": 0.024259367813500472,
      "grad_norm": 2.285918488491891,
      "learning_rate": 3.853012483253093e-06,
      "loss": 0.8726,
      "step": 269
    },
    {
      "epoch": 0.024349551336970737,
      "grad_norm": 2.0221708007651187,
      "learning_rate": 3.855567916011802e-06,
      "loss": 0.928,
      "step": 270
    },
    {
      "epoch": 0.024439734860440997,
      "grad_norm": 1.934305202722428,
      "learning_rate": 3.858113901677755e-06,
      "loss": 0.9461,
      "step": 271
    },
    {
      "epoch": 0.024529918383911258,
      "grad_norm": 1.7048422179391345,
      "learning_rate": 3.860650509843034e-06,
      "loss": 0.8741,
      "step": 272
    },
    {
      "epoch": 0.024620101907381522,
      "grad_norm": 2.3919794571482402,
      "learning_rate": 3.863177809333563e-06,
      "loss": 0.9472,
      "step": 273
    },
    {
      "epoch": 0.024710285430851783,
      "grad_norm": 1.5169453326479605,
      "learning_rate": 3.86569586822032e-06,
      "loss": 0.8834,
      "step": 274
    },
    {
      "epoch": 0.024800468954322044,
      "grad_norm": 2.0658375871600936,
      "learning_rate": 3.868204753830331e-06,
      "loss": 0.9185,
      "step": 275
    },
    {
      "epoch": 0.02489065247779231,
      "grad_norm": 1.799615710350628,
      "learning_rate": 3.870704532757476e-06,
      "loss": 0.842,
      "step": 276
    },
    {
      "epoch": 0.02498083600126257,
      "grad_norm": 1.7194584860131636,
      "learning_rate": 3.8731952708730974e-06,
      "loss": 0.9383,
      "step": 277
    },
    {
      "epoch": 0.02507101952473283,
      "grad_norm": 1.698936561742791,
      "learning_rate": 3.8756770333364085e-06,
      "loss": 0.9234,
      "step": 278
    },
    {
      "epoch": 0.025161203048203094,
      "grad_norm": 2.1086614461507143,
      "learning_rate": 3.878149884604725e-06,
      "loss": 0.8173,
      "step": 279
    },
    {
      "epoch": 0.025251386571673355,
      "grad_norm": 2.221787214087483,
      "learning_rate": 3.8806138884435125e-06,
      "loss": 0.9067,
      "step": 280
    },
    {
      "epoch": 0.025341570095143616,
      "grad_norm": 1.6782569648669752,
      "learning_rate": 3.883069107936248e-06,
      "loss": 0.8248,
      "step": 281
    },
    {
      "epoch": 0.02543175361861388,
      "grad_norm": 2.419364661842831,
      "learning_rate": 3.885515605494114e-06,
      "loss": 0.9568,
      "step": 282
    },
    {
      "epoch": 0.02552193714208414,
      "grad_norm": 2.386031381175261,
      "learning_rate": 3.8879534428655145e-06,
      "loss": 0.9673,
      "step": 283
    },
    {
      "epoch": 0.025612120665554402,
      "grad_norm": 1.8500671142570069,
      "learning_rate": 3.890382681145432e-06,
      "loss": 0.9302,
      "step": 284
    },
    {
      "epoch": 0.025702304189024666,
      "grad_norm": 1.7906887209540838,
      "learning_rate": 3.892803380784608e-06,
      "loss": 0.9322,
      "step": 285
    },
    {
      "epoch": 0.025792487712494927,
      "grad_norm": 0.9745743922986421,
      "learning_rate": 3.8952156015985725e-06,
      "loss": 0.6442,
      "step": 286
    },
    {
      "epoch": 0.025882671235965188,
      "grad_norm": 2.032525351766248,
      "learning_rate": 3.897619402776516e-06,
      "loss": 0.9015,
      "step": 287
    },
    {
      "epoch": 0.025972854759435452,
      "grad_norm": 1.82678904176837,
      "learning_rate": 3.900014842889995e-06,
      "loss": 0.9667,
      "step": 288
    },
    {
      "epoch": 0.026063038282905713,
      "grad_norm": 1.525166239823479,
      "learning_rate": 3.902401979901503e-06,
      "loss": 0.7947,
      "step": 289
    },
    {
      "epoch": 0.026153221806375974,
      "grad_norm": 2.0276112756637983,
      "learning_rate": 3.904780871172884e-06,
      "loss": 0.9362,
      "step": 290
    },
    {
      "epoch": 0.026243405329846238,
      "grad_norm": 2.895952748276716,
      "learning_rate": 3.907151573473601e-06,
      "loss": 0.7918,
      "step": 291
    },
    {
      "epoch": 0.0263335888533165,
      "grad_norm": 1.3264601201785882,
      "learning_rate": 3.909514142988868e-06,
      "loss": 0.8897,
      "step": 292
    },
    {
      "epoch": 0.02642377237678676,
      "grad_norm": 2.09489250326114,
      "learning_rate": 3.911868635327639e-06,
      "loss": 0.8606,
      "step": 293
    },
    {
      "epoch": 0.026513955900257024,
      "grad_norm": 1.754605907740964,
      "learning_rate": 3.914215105530455e-06,
      "loss": 0.8629,
      "step": 294
    },
    {
      "epoch": 0.026604139423727285,
      "grad_norm": 1.7411451428916294,
      "learning_rate": 3.916553608077179e-06,
      "loss": 0.8703,
      "step": 295
    },
    {
      "epoch": 0.026694322947197546,
      "grad_norm": 1.7601091767174113,
      "learning_rate": 3.91888419689457e-06,
      "loss": 0.868,
      "step": 296
    },
    {
      "epoch": 0.02678450647066781,
      "grad_norm": 2.179170171128246,
      "learning_rate": 3.921206925363754e-06,
      "loss": 0.819,
      "step": 297
    },
    {
      "epoch": 0.02687468999413807,
      "grad_norm": 3.107441977022896,
      "learning_rate": 3.923521846327559e-06,
      "loss": 0.8202,
      "step": 298
    },
    {
      "epoch": 0.02696487351760833,
      "grad_norm": 2.0139712861257757,
      "learning_rate": 3.925829012097725e-06,
      "loss": 0.8682,
      "step": 299
    },
    {
      "epoch": 0.027055057041078596,
      "grad_norm": 2.5510507064499253,
      "learning_rate": 3.928128474462e-06,
      "loss": 0.8231,
      "step": 300
    },
    {
      "epoch": 0.027145240564548857,
      "grad_norm": 4.1324686581622565,
      "learning_rate": 3.930420284691115e-06,
      "loss": 0.9135,
      "step": 301
    },
    {
      "epoch": 0.027235424088019117,
      "grad_norm": 2.4759197183144055,
      "learning_rate": 3.932704493545644e-06,
      "loss": 0.9113,
      "step": 302
    },
    {
      "epoch": 0.02732560761148938,
      "grad_norm": 2.0995137605409218,
      "learning_rate": 3.934981151282745e-06,
      "loss": 0.9583,
      "step": 303
    },
    {
      "epoch": 0.027415791134959643,
      "grad_norm": 2.1715264349410313,
      "learning_rate": 3.9372503076628006e-06,
      "loss": 0.8935,
      "step": 304
    },
    {
      "epoch": 0.027505974658429903,
      "grad_norm": 2.070101514897862,
      "learning_rate": 3.939512011955941e-06,
      "loss": 0.8924,
      "step": 305
    },
    {
      "epoch": 0.027596158181900168,
      "grad_norm": 1.9765793268143375,
      "learning_rate": 3.941766312948463e-06,
      "loss": 0.8891,
      "step": 306
    },
    {
      "epoch": 0.02768634170537043,
      "grad_norm": 1.8934481011896798,
      "learning_rate": 3.944013258949147e-06,
      "loss": 0.8377,
      "step": 307
    },
    {
      "epoch": 0.02777652522884069,
      "grad_norm": 2.150158667808711,
      "learning_rate": 3.946252897795465e-06,
      "loss": 0.6945,
      "step": 308
    },
    {
      "epoch": 0.027866708752310954,
      "grad_norm": 2.913384161737814,
      "learning_rate": 3.9484852768596935e-06,
      "loss": 0.8945,
      "step": 309
    },
    {
      "epoch": 0.027956892275781214,
      "grad_norm": 2.8575747299768635,
      "learning_rate": 3.950710443054923e-06,
      "loss": 0.7795,
      "step": 310
    },
    {
      "epoch": 0.028047075799251475,
      "grad_norm": 2.6780516855894465,
      "learning_rate": 3.952928442840981e-06,
      "loss": 0.7842,
      "step": 311
    },
    {
      "epoch": 0.02813725932272174,
      "grad_norm": 2.047786713438822,
      "learning_rate": 3.955139322230243e-06,
      "loss": 0.8938,
      "step": 312
    },
    {
      "epoch": 0.028227442846192,
      "grad_norm": 2.288866262519546,
      "learning_rate": 3.957343126793365e-06,
      "loss": 0.8049,
      "step": 313
    },
    {
      "epoch": 0.02831762636966226,
      "grad_norm": 1.8699047794680197,
      "learning_rate": 3.959539901664921e-06,
      "loss": 0.8445,
      "step": 314
    },
    {
      "epoch": 0.028407809893132525,
      "grad_norm": 2.378029848217825,
      "learning_rate": 3.9617296915489425e-06,
      "loss": 0.9702,
      "step": 315
    },
    {
      "epoch": 0.028497993416602786,
      "grad_norm": 2.575708632505666,
      "learning_rate": 3.963912540724387e-06,
      "loss": 0.8352,
      "step": 316
    },
    {
      "epoch": 0.028588176940073047,
      "grad_norm": 2.2626042716093298,
      "learning_rate": 3.966088493050501e-06,
      "loss": 0.8872,
      "step": 317
    },
    {
      "epoch": 0.02867836046354331,
      "grad_norm": 1.9481359620163745,
      "learning_rate": 3.968257591972113e-06,
      "loss": 0.8551,
      "step": 318
    },
    {
      "epoch": 0.028768543987013572,
      "grad_norm": 1.6713824393295347,
      "learning_rate": 3.970419880524835e-06,
      "loss": 0.9345,
      "step": 319
    },
    {
      "epoch": 0.028858727510483833,
      "grad_norm": 2.24591488333318,
      "learning_rate": 3.972575401340192e-06,
      "loss": 0.837,
      "step": 320
    },
    {
      "epoch": 0.028948911033954097,
      "grad_norm": 1.660227910907417,
      "learning_rate": 3.974724196650656e-06,
      "loss": 0.8896,
      "step": 321
    },
    {
      "epoch": 0.029039094557424358,
      "grad_norm": 2.186956828664131,
      "learning_rate": 3.976866308294617e-06,
      "loss": 0.856,
      "step": 322
    },
    {
      "epoch": 0.02912927808089462,
      "grad_norm": 2.1929385561982775,
      "learning_rate": 3.979001777721269e-06,
      "loss": 0.8125,
      "step": 323
    },
    {
      "epoch": 0.029219461604364883,
      "grad_norm": 2.0359351485290764,
      "learning_rate": 3.981130645995424e-06,
      "loss": 0.8263,
      "step": 324
    },
    {
      "epoch": 0.029309645127835144,
      "grad_norm": 1.8751139924048161,
      "learning_rate": 3.983252953802248e-06,
      "loss": 0.9346,
      "step": 325
    },
    {
      "epoch": 0.029399828651305405,
      "grad_norm": 0.7312465341599974,
      "learning_rate": 3.9853687414519285e-06,
      "loss": 0.6369,
      "step": 326
    },
    {
      "epoch": 0.02949001217477567,
      "grad_norm": 2.0357174383344594,
      "learning_rate": 3.987478048884265e-06,
      "loss": 0.9061,
      "step": 327
    },
    {
      "epoch": 0.02958019569824593,
      "grad_norm": 2.3297199699856215,
      "learning_rate": 3.989580915673196e-06,
      "loss": 0.8356,
      "step": 328
    },
    {
      "epoch": 0.02967037922171619,
      "grad_norm": 1.795595659015598,
      "learning_rate": 3.991677381031255e-06,
      "loss": 0.8337,
      "step": 329
    },
    {
      "epoch": 0.029760562745186455,
      "grad_norm": 1.7753236126396639,
      "learning_rate": 3.993767483813953e-06,
      "loss": 0.9294,
      "step": 330
    },
    {
      "epoch": 0.029850746268656716,
      "grad_norm": 1.9416936588730564,
      "learning_rate": 3.995851262524104e-06,
      "loss": 0.7664,
      "step": 331
    },
    {
      "epoch": 0.02994092979212698,
      "grad_norm": 2.0092006293260702,
      "learning_rate": 3.997928755316079e-06,
      "loss": 0.8621,
      "step": 332
    },
    {
      "epoch": 0.03003111331559724,
      "grad_norm": 0.8739310615355309,
      "learning_rate": 4e-06,
      "loss": 0.6243,
      "step": 333
    },
    {
      "epoch": 0.030121296839067502,
      "grad_norm": 1.6216363573263988,
      "learning_rate": 3.999999914674486e-06,
      "loss": 0.8219,
      "step": 334
    },
    {
      "epoch": 0.030211480362537766,
      "grad_norm": 1.6674952672522323,
      "learning_rate": 3.999999658697952e-06,
      "loss": 0.8427,
      "step": 335
    },
    {
      "epoch": 0.030301663886008027,
      "grad_norm": 0.6927725659373812,
      "learning_rate": 3.9999992320704185e-06,
      "loss": 0.5606,
      "step": 336
    },
    {
      "epoch": 0.030391847409478288,
      "grad_norm": 2.0799820712123798,
      "learning_rate": 3.999998634791922e-06,
      "loss": 0.8753,
      "step": 337
    },
    {
      "epoch": 0.030482030932948552,
      "grad_norm": 1.6800526066899595,
      "learning_rate": 3.999997866862515e-06,
      "loss": 0.8357,
      "step": 338
    },
    {
      "epoch": 0.030572214456418813,
      "grad_norm": 1.6903582491178364,
      "learning_rate": 3.999996928282262e-06,
      "loss": 0.939,
      "step": 339
    },
    {
      "epoch": 0.030662397979889074,
      "grad_norm": 2.52736544710306,
      "learning_rate": 3.999995819051244e-06,
      "loss": 0.8753,
      "step": 340
    },
    {
      "epoch": 0.030752581503359338,
      "grad_norm": 1.612868064450966,
      "learning_rate": 3.9999945391695536e-06,
      "loss": 0.8099,
      "step": 341
    },
    {
      "epoch": 0.0308427650268296,
      "grad_norm": 3.3506347510379704,
      "learning_rate": 3.999993088637302e-06,
      "loss": 0.9329,
      "step": 342
    },
    {
      "epoch": 0.03093294855029986,
      "grad_norm": 1.9688104287213493,
      "learning_rate": 3.999991467454612e-06,
      "loss": 0.9132,
      "step": 343
    },
    {
      "epoch": 0.031023132073770124,
      "grad_norm": 1.5433872862435192,
      "learning_rate": 3.999989675621622e-06,
      "loss": 0.7652,
      "step": 344
    },
    {
      "epoch": 0.031113315597240385,
      "grad_norm": 1.8884769563973371,
      "learning_rate": 3.999987713138485e-06,
      "loss": 0.8526,
      "step": 345
    },
    {
      "epoch": 0.031203499120710645,
      "grad_norm": 0.8007130663616862,
      "learning_rate": 3.999985580005369e-06,
      "loss": 0.6682,
      "step": 346
    },
    {
      "epoch": 0.031293682644180906,
      "grad_norm": 2.036671372869687,
      "learning_rate": 3.999983276222455e-06,
      "loss": 0.9187,
      "step": 347
    },
    {
      "epoch": 0.03138386616765117,
      "grad_norm": 1.675634062022911,
      "learning_rate": 3.999980801789941e-06,
      "loss": 0.8477,
      "step": 348
    },
    {
      "epoch": 0.031474049691121435,
      "grad_norm": 1.8567067082171984,
      "learning_rate": 3.999978156708036e-06,
      "loss": 0.8443,
      "step": 349
    },
    {
      "epoch": 0.031564233214591696,
      "grad_norm": 4.197359008602281,
      "learning_rate": 3.9999753409769675e-06,
      "loss": 0.9703,
      "step": 350
    },
    {
      "epoch": 0.031654416738061956,
      "grad_norm": 0.7125797795921733,
      "learning_rate": 3.999972354596975e-06,
      "loss": 0.6108,
      "step": 351
    },
    {
      "epoch": 0.03174460026153222,
      "grad_norm": 2.162600026546031,
      "learning_rate": 3.999969197568314e-06,
      "loss": 0.8002,
      "step": 352
    },
    {
      "epoch": 0.03183478378500248,
      "grad_norm": 1.9631677839929806,
      "learning_rate": 3.999965869891253e-06,
      "loss": 0.9036,
      "step": 353
    },
    {
      "epoch": 0.03192496730847274,
      "grad_norm": 1.5692896002318222,
      "learning_rate": 3.999962371566075e-06,
      "loss": 0.872,
      "step": 354
    },
    {
      "epoch": 0.03201515083194301,
      "grad_norm": 1.67128907391244,
      "learning_rate": 3.999958702593082e-06,
      "loss": 0.9183,
      "step": 355
    },
    {
      "epoch": 0.03210533435541327,
      "grad_norm": 1.5645375508781854,
      "learning_rate": 3.999954862972583e-06,
      "loss": 0.8732,
      "step": 356
    },
    {
      "epoch": 0.03219551787888353,
      "grad_norm": 1.7661385253587063,
      "learning_rate": 3.999950852704908e-06,
      "loss": 0.7944,
      "step": 357
    },
    {
      "epoch": 0.03228570140235379,
      "grad_norm": 2.2771460838315525,
      "learning_rate": 3.9999466717903995e-06,
      "loss": 0.8986,
      "step": 358
    },
    {
      "epoch": 0.03237588492582405,
      "grad_norm": 0.7222988997820593,
      "learning_rate": 3.999942320229413e-06,
      "loss": 0.5969,
      "step": 359
    },
    {
      "epoch": 0.03246606844929431,
      "grad_norm": 2.0724289925884767,
      "learning_rate": 3.99993779802232e-06,
      "loss": 0.9823,
      "step": 360
    },
    {
      "epoch": 0.03255625197276458,
      "grad_norm": 1.8721182612575284,
      "learning_rate": 3.999933105169506e-06,
      "loss": 0.8329,
      "step": 361
    },
    {
      "epoch": 0.03264643549623484,
      "grad_norm": 1.6248669936693554,
      "learning_rate": 3.999928241671373e-06,
      "loss": 0.8416,
      "step": 362
    },
    {
      "epoch": 0.0327366190197051,
      "grad_norm": 1.8539875744112113,
      "learning_rate": 3.999923207528334e-06,
      "loss": 0.7488,
      "step": 363
    },
    {
      "epoch": 0.03282680254317536,
      "grad_norm": 1.458617545997769,
      "learning_rate": 3.9999180027408196e-06,
      "loss": 0.8356,
      "step": 364
    },
    {
      "epoch": 0.03291698606664562,
      "grad_norm": 1.862287660309876,
      "learning_rate": 3.9999126273092735e-06,
      "loss": 0.8277,
      "step": 365
    },
    {
      "epoch": 0.03300716959011588,
      "grad_norm": 2.553540453499073,
      "learning_rate": 3.999907081234156e-06,
      "loss": 0.9366,
      "step": 366
    },
    {
      "epoch": 0.03309735311358615,
      "grad_norm": 2.214333342554147,
      "learning_rate": 3.999901364515938e-06,
      "loss": 0.8254,
      "step": 367
    },
    {
      "epoch": 0.03318753663705641,
      "grad_norm": 1.8901283936057294,
      "learning_rate": 3.999895477155108e-06,
      "loss": 0.8944,
      "step": 368
    },
    {
      "epoch": 0.03327772016052667,
      "grad_norm": 0.7537759065367093,
      "learning_rate": 3.999889419152169e-06,
      "loss": 0.5777,
      "step": 369
    },
    {
      "epoch": 0.03336790368399693,
      "grad_norm": 2.3043396768120736,
      "learning_rate": 3.999883190507638e-06,
      "loss": 0.8581,
      "step": 370
    },
    {
      "epoch": 0.033458087207467194,
      "grad_norm": 1.742729684258471,
      "learning_rate": 3.999876791222044e-06,
      "loss": 0.8391,
      "step": 371
    },
    {
      "epoch": 0.033548270730937454,
      "grad_norm": 1.8944406389413886,
      "learning_rate": 3.999870221295936e-06,
      "loss": 0.8026,
      "step": 372
    },
    {
      "epoch": 0.03363845425440772,
      "grad_norm": 3.422360877810274,
      "learning_rate": 3.999863480729875e-06,
      "loss": 0.8227,
      "step": 373
    },
    {
      "epoch": 0.03372863777787798,
      "grad_norm": 2.3489861553656217,
      "learning_rate": 3.999856569524433e-06,
      "loss": 0.7878,
      "step": 374
    },
    {
      "epoch": 0.033818821301348244,
      "grad_norm": 1.5372453990500783,
      "learning_rate": 3.999849487680202e-06,
      "loss": 0.789,
      "step": 375
    },
    {
      "epoch": 0.033909004824818505,
      "grad_norm": 1.8360390493025918,
      "learning_rate": 3.999842235197786e-06,
      "loss": 0.7783,
      "step": 376
    },
    {
      "epoch": 0.033999188348288766,
      "grad_norm": 2.3475313327961085,
      "learning_rate": 3.999834812077803e-06,
      "loss": 0.7807,
      "step": 377
    },
    {
      "epoch": 0.034089371871759026,
      "grad_norm": 2.279556481908581,
      "learning_rate": 3.999827218320886e-06,
      "loss": 0.9011,
      "step": 378
    },
    {
      "epoch": 0.034179555395229294,
      "grad_norm": 2.0361931149268813,
      "learning_rate": 3.999819453927685e-06,
      "loss": 0.7596,
      "step": 379
    },
    {
      "epoch": 0.034269738918699555,
      "grad_norm": 2.151370955440981,
      "learning_rate": 3.999811518898861e-06,
      "loss": 0.8475,
      "step": 380
    },
    {
      "epoch": 0.034359922442169816,
      "grad_norm": 2.329265136093088,
      "learning_rate": 3.999803413235092e-06,
      "loss": 0.8758,
      "step": 381
    },
    {
      "epoch": 0.03445010596564008,
      "grad_norm": 1.7340002256435085,
      "learning_rate": 3.999795136937068e-06,
      "loss": 0.954,
      "step": 382
    },
    {
      "epoch": 0.03454028948911034,
      "grad_norm": 1.8729045603031658,
      "learning_rate": 3.999786690005496e-06,
      "loss": 0.8385,
      "step": 383
    },
    {
      "epoch": 0.0346304730125806,
      "grad_norm": 1.757243288916031,
      "learning_rate": 3.999778072441098e-06,
      "loss": 0.8339,
      "step": 384
    },
    {
      "epoch": 0.034720656536050866,
      "grad_norm": 2.7958331165417865,
      "learning_rate": 3.999769284244608e-06,
      "loss": 0.9044,
      "step": 385
    },
    {
      "epoch": 0.03481084005952113,
      "grad_norm": 1.944800129405323,
      "learning_rate": 3.999760325416775e-06,
      "loss": 0.8599,
      "step": 386
    },
    {
      "epoch": 0.03490102358299139,
      "grad_norm": 1.7009794386978352,
      "learning_rate": 3.999751195958366e-06,
      "loss": 0.9315,
      "step": 387
    },
    {
      "epoch": 0.03499120710646165,
      "grad_norm": 0.7875190974750641,
      "learning_rate": 3.999741895870157e-06,
      "loss": 0.6016,
      "step": 388
    },
    {
      "epoch": 0.03508139062993191,
      "grad_norm": 2.1661190906025576,
      "learning_rate": 3.999732425152944e-06,
      "loss": 0.8912,
      "step": 389
    },
    {
      "epoch": 0.03517157415340217,
      "grad_norm": 1.9187623536150245,
      "learning_rate": 3.999722783807533e-06,
      "loss": 0.8998,
      "step": 390
    },
    {
      "epoch": 0.03526175767687244,
      "grad_norm": 2.4497649878066667,
      "learning_rate": 3.999712971834748e-06,
      "loss": 0.7829,
      "step": 391
    },
    {
      "epoch": 0.0353519412003427,
      "grad_norm": 1.7239733999771387,
      "learning_rate": 3.999702989235427e-06,
      "loss": 0.8606,
      "step": 392
    },
    {
      "epoch": 0.03544212472381296,
      "grad_norm": 1.7810330175116218,
      "learning_rate": 3.999692836010419e-06,
      "loss": 0.9324,
      "step": 393
    },
    {
      "epoch": 0.03553230824728322,
      "grad_norm": 2.2095055138268163,
      "learning_rate": 3.999682512160593e-06,
      "loss": 0.9836,
      "step": 394
    },
    {
      "epoch": 0.03562249177075348,
      "grad_norm": 0.7859839636110727,
      "learning_rate": 3.99967201768683e-06,
      "loss": 0.5794,
      "step": 395
    },
    {
      "epoch": 0.03571267529422374,
      "grad_norm": 4.14063444316435,
      "learning_rate": 3.999661352590023e-06,
      "loss": 0.8388,
      "step": 396
    },
    {
      "epoch": 0.03580285881769401,
      "grad_norm": 2.39954114739633,
      "learning_rate": 3.999650516871083e-06,
      "loss": 0.8498,
      "step": 397
    },
    {
      "epoch": 0.03589304234116427,
      "grad_norm": 1.7793098725598955,
      "learning_rate": 3.9996395105309365e-06,
      "loss": 0.8731,
      "step": 398
    },
    {
      "epoch": 0.03598322586463453,
      "grad_norm": 1.6904958340701537,
      "learning_rate": 3.99962833357052e-06,
      "loss": 0.8707,
      "step": 399
    },
    {
      "epoch": 0.03607340938810479,
      "grad_norm": 1.728918203251367,
      "learning_rate": 3.999616985990789e-06,
      "loss": 0.8263,
      "step": 400
    },
    {
      "epoch": 0.03616359291157505,
      "grad_norm": 2.0619292191901133,
      "learning_rate": 3.9996054677927104e-06,
      "loss": 0.7821,
      "step": 401
    },
    {
      "epoch": 0.03625377643504532,
      "grad_norm": 2.3056830890521605,
      "learning_rate": 3.9995937789772675e-06,
      "loss": 0.79,
      "step": 402
    },
    {
      "epoch": 0.03634395995851558,
      "grad_norm": 0.7259559150163184,
      "learning_rate": 3.999581919545458e-06,
      "loss": 0.6044,
      "step": 403
    },
    {
      "epoch": 0.03643414348198584,
      "grad_norm": 2.496224413393452,
      "learning_rate": 3.9995698894982935e-06,
      "loss": 0.8853,
      "step": 404
    },
    {
      "epoch": 0.0365243270054561,
      "grad_norm": 1.8960485301305527,
      "learning_rate": 3.9995576888368e-06,
      "loss": 0.8916,
      "step": 405
    },
    {
      "epoch": 0.036614510528926364,
      "grad_norm": 1.9223219189058856,
      "learning_rate": 3.9995453175620194e-06,
      "loss": 0.8384,
      "step": 406
    },
    {
      "epoch": 0.036704694052396625,
      "grad_norm": 2.290525001747497,
      "learning_rate": 3.999532775675007e-06,
      "loss": 0.7109,
      "step": 407
    },
    {
      "epoch": 0.03679487757586689,
      "grad_norm": 2.0704884580296468,
      "learning_rate": 3.9995200631768326e-06,
      "loss": 0.8258,
      "step": 408
    },
    {
      "epoch": 0.03688506109933715,
      "grad_norm": 3.0568133125640258,
      "learning_rate": 3.9995071800685815e-06,
      "loss": 0.9871,
      "step": 409
    },
    {
      "epoch": 0.036975244622807414,
      "grad_norm": 2.429635786383193,
      "learning_rate": 3.999494126351352e-06,
      "loss": 0.8227,
      "step": 410
    },
    {
      "epoch": 0.037065428146277675,
      "grad_norm": 2.3336107792981657,
      "learning_rate": 3.99948090202626e-06,
      "loss": 0.8568,
      "step": 411
    },
    {
      "epoch": 0.037155611669747936,
      "grad_norm": 2.107864177910211,
      "learning_rate": 3.999467507094431e-06,
      "loss": 0.8458,
      "step": 412
    },
    {
      "epoch": 0.0372457951932182,
      "grad_norm": 1.4791098928525483,
      "learning_rate": 3.999453941557011e-06,
      "loss": 0.7535,
      "step": 413
    },
    {
      "epoch": 0.037335978716688464,
      "grad_norm": 0.7561830128408047,
      "learning_rate": 3.999440205415154e-06,
      "loss": 0.5571,
      "step": 414
    },
    {
      "epoch": 0.037426162240158725,
      "grad_norm": 3.031022840030591,
      "learning_rate": 3.999426298670035e-06,
      "loss": 0.8963,
      "step": 415
    },
    {
      "epoch": 0.037516345763628986,
      "grad_norm": 1.542620501270663,
      "learning_rate": 3.9994122213228385e-06,
      "loss": 0.8878,
      "step": 416
    },
    {
      "epoch": 0.03760652928709925,
      "grad_norm": 2.1571180282650766,
      "learning_rate": 3.9993979733747675e-06,
      "loss": 0.8744,
      "step": 417
    },
    {
      "epoch": 0.03769671281056951,
      "grad_norm": 1.7283635476664687,
      "learning_rate": 3.999383554827037e-06,
      "loss": 0.8438,
      "step": 418
    },
    {
      "epoch": 0.03778689633403977,
      "grad_norm": 1.7775892151066777,
      "learning_rate": 3.999368965680876e-06,
      "loss": 0.8146,
      "step": 419
    },
    {
      "epoch": 0.037877079857510036,
      "grad_norm": 1.836823334955398,
      "learning_rate": 3.999354205937531e-06,
      "loss": 0.8963,
      "step": 420
    },
    {
      "epoch": 0.0379672633809803,
      "grad_norm": 1.7478124024397796,
      "learning_rate": 3.999339275598261e-06,
      "loss": 0.8958,
      "step": 421
    },
    {
      "epoch": 0.03805744690445056,
      "grad_norm": 1.5404200883819958,
      "learning_rate": 3.99932417466434e-06,
      "loss": 0.8204,
      "step": 422
    },
    {
      "epoch": 0.03814763042792082,
      "grad_norm": 2.4257078297879127,
      "learning_rate": 3.999308903137056e-06,
      "loss": 0.7853,
      "step": 423
    },
    {
      "epoch": 0.03823781395139108,
      "grad_norm": 1.8520390022158713,
      "learning_rate": 3.999293461017711e-06,
      "loss": 0.8946,
      "step": 424
    },
    {
      "epoch": 0.03832799747486134,
      "grad_norm": 2.3063609109483045,
      "learning_rate": 3.9992778483076255e-06,
      "loss": 0.8438,
      "step": 425
    },
    {
      "epoch": 0.03841818099833161,
      "grad_norm": 1.5706215929084737,
      "learning_rate": 3.99926206500813e-06,
      "loss": 0.9004,
      "step": 426
    },
    {
      "epoch": 0.03850836452180187,
      "grad_norm": 1.773656755917356,
      "learning_rate": 3.999246111120571e-06,
      "loss": 0.778,
      "step": 427
    },
    {
      "epoch": 0.03859854804527213,
      "grad_norm": 1.7163962459836972,
      "learning_rate": 3.999229986646311e-06,
      "loss": 0.9115,
      "step": 428
    },
    {
      "epoch": 0.03868873156874239,
      "grad_norm": 1.944612737214957,
      "learning_rate": 3.999213691586723e-06,
      "loss": 0.8086,
      "step": 429
    },
    {
      "epoch": 0.03877891509221265,
      "grad_norm": 1.617926461013881,
      "learning_rate": 3.9991972259432e-06,
      "loss": 0.8782,
      "step": 430
    },
    {
      "epoch": 0.03886909861568291,
      "grad_norm": 1.6581188670252378,
      "learning_rate": 3.999180589717147e-06,
      "loss": 0.9223,
      "step": 431
    },
    {
      "epoch": 0.03895928213915318,
      "grad_norm": 1.8315448562665226,
      "learning_rate": 3.999163782909983e-06,
      "loss": 0.7662,
      "step": 432
    },
    {
      "epoch": 0.03904946566262344,
      "grad_norm": 1.7965416640544671,
      "learning_rate": 3.99914680552314e-06,
      "loss": 0.8788,
      "step": 433
    },
    {
      "epoch": 0.0391396491860937,
      "grad_norm": 2.0808215321715555,
      "learning_rate": 3.999129657558069e-06,
      "loss": 0.7789,
      "step": 434
    },
    {
      "epoch": 0.03922983270956396,
      "grad_norm": 0.7411148350752504,
      "learning_rate": 3.999112339016234e-06,
      "loss": 0.5712,
      "step": 435
    },
    {
      "epoch": 0.03932001623303422,
      "grad_norm": 1.7480224607670083,
      "learning_rate": 3.999094849899109e-06,
      "loss": 0.869,
      "step": 436
    },
    {
      "epoch": 0.039410199756504484,
      "grad_norm": 3.3918561677847503,
      "learning_rate": 3.99907719020819e-06,
      "loss": 0.7643,
      "step": 437
    },
    {
      "epoch": 0.03950038327997475,
      "grad_norm": 1.6112468167222498,
      "learning_rate": 3.999059359944982e-06,
      "loss": 0.867,
      "step": 438
    },
    {
      "epoch": 0.03959056680344501,
      "grad_norm": 1.6020761550061249,
      "learning_rate": 3.999041359111007e-06,
      "loss": 0.8966,
      "step": 439
    },
    {
      "epoch": 0.03968075032691527,
      "grad_norm": 1.8507177945068165,
      "learning_rate": 3.999023187707801e-06,
      "loss": 0.9186,
      "step": 440
    },
    {
      "epoch": 0.039770933850385534,
      "grad_norm": 1.6887668516728656,
      "learning_rate": 3.999004845736913e-06,
      "loss": 0.8941,
      "step": 441
    },
    {
      "epoch": 0.039861117373855795,
      "grad_norm": 1.6852581532757447,
      "learning_rate": 3.9989863331999096e-06,
      "loss": 0.9161,
      "step": 442
    },
    {
      "epoch": 0.039951300897326056,
      "grad_norm": 2.2104286713653187,
      "learning_rate": 3.99896765009837e-06,
      "loss": 0.7807,
      "step": 443
    },
    {
      "epoch": 0.040041484420796324,
      "grad_norm": 1.6734589533494786,
      "learning_rate": 3.998948796433888e-06,
      "loss": 0.8156,
      "step": 444
    },
    {
      "epoch": 0.040131667944266584,
      "grad_norm": 1.4613982101475411,
      "learning_rate": 3.998929772208073e-06,
      "loss": 0.8857,
      "step": 445
    },
    {
      "epoch": 0.040221851467736845,
      "grad_norm": 0.6900600755149858,
      "learning_rate": 3.998910577422547e-06,
      "loss": 0.5982,
      "step": 446
    },
    {
      "epoch": 0.040312034991207106,
      "grad_norm": 1.5001723667292606,
      "learning_rate": 3.99889121207895e-06,
      "loss": 0.8889,
      "step": 447
    },
    {
      "epoch": 0.04040221851467737,
      "grad_norm": 1.6929605931823575,
      "learning_rate": 3.9988716761789324e-06,
      "loss": 0.8672,
      "step": 448
    },
    {
      "epoch": 0.04049240203814763,
      "grad_norm": 2.310451940938551,
      "learning_rate": 3.998851969724161e-06,
      "loss": 0.7806,
      "step": 449
    },
    {
      "epoch": 0.040582585561617895,
      "grad_norm": 1.7410390173162418,
      "learning_rate": 3.998832092716319e-06,
      "loss": 0.8917,
      "step": 450
    },
    {
      "epoch": 0.040672769085088156,
      "grad_norm": 1.436971650260061,
      "learning_rate": 3.998812045157102e-06,
      "loss": 0.8233,
      "step": 451
    },
    {
      "epoch": 0.04076295260855842,
      "grad_norm": 2.251734700656339,
      "learning_rate": 3.998791827048219e-06,
      "loss": 0.8982,
      "step": 452
    },
    {
      "epoch": 0.04085313613202868,
      "grad_norm": 1.6461360105820793,
      "learning_rate": 3.998771438391396e-06,
      "loss": 0.7886,
      "step": 453
    },
    {
      "epoch": 0.04094331965549894,
      "grad_norm": 0.6401343443802504,
      "learning_rate": 3.9987508791883725e-06,
      "loss": 0.5753,
      "step": 454
    },
    {
      "epoch": 0.0410335031789692,
      "grad_norm": 1.7171653378195875,
      "learning_rate": 3.998730149440904e-06,
      "loss": 0.9693,
      "step": 455
    },
    {
      "epoch": 0.04112368670243947,
      "grad_norm": 1.7657855475697222,
      "learning_rate": 3.998709249150758e-06,
      "loss": 0.8915,
      "step": 456
    },
    {
      "epoch": 0.04121387022590973,
      "grad_norm": 1.676987505084299,
      "learning_rate": 3.998688178319717e-06,
      "loss": 0.8593,
      "step": 457
    },
    {
      "epoch": 0.04130405374937999,
      "grad_norm": 1.9063794451679184,
      "learning_rate": 3.9986669369495805e-06,
      "loss": 0.8345,
      "step": 458
    },
    {
      "epoch": 0.04139423727285025,
      "grad_norm": 1.9296495688722037,
      "learning_rate": 3.998645525042161e-06,
      "loss": 0.898,
      "step": 459
    },
    {
      "epoch": 0.04148442079632051,
      "grad_norm": 1.8814874949043496,
      "learning_rate": 3.998623942599284e-06,
      "loss": 0.9153,
      "step": 460
    },
    {
      "epoch": 0.04157460431979077,
      "grad_norm": 2.6838369023361057,
      "learning_rate": 3.998602189622793e-06,
      "loss": 0.841,
      "step": 461
    },
    {
      "epoch": 0.04166478784326104,
      "grad_norm": 2.098349799328171,
      "learning_rate": 3.998580266114542e-06,
      "loss": 0.8703,
      "step": 462
    },
    {
      "epoch": 0.0417549713667313,
      "grad_norm": 1.5703505326405247,
      "learning_rate": 3.998558172076404e-06,
      "loss": 0.7465,
      "step": 463
    },
    {
      "epoch": 0.04184515489020156,
      "grad_norm": 1.777146347318016,
      "learning_rate": 3.998535907510262e-06,
      "loss": 0.7778,
      "step": 464
    },
    {
      "epoch": 0.04193533841367182,
      "grad_norm": 1.5818557453528546,
      "learning_rate": 3.998513472418016e-06,
      "loss": 0.8545,
      "step": 465
    },
    {
      "epoch": 0.04202552193714208,
      "grad_norm": 1.662703124522791,
      "learning_rate": 3.998490866801582e-06,
      "loss": 0.8742,
      "step": 466
    },
    {
      "epoch": 0.04211570546061234,
      "grad_norm": 1.5932659087153762,
      "learning_rate": 3.998468090662886e-06,
      "loss": 0.8347,
      "step": 467
    },
    {
      "epoch": 0.04220588898408261,
      "grad_norm": 1.5787189234727803,
      "learning_rate": 3.998445144003874e-06,
      "loss": 0.7357,
      "step": 468
    },
    {
      "epoch": 0.04229607250755287,
      "grad_norm": 4.023486564855566,
      "learning_rate": 3.998422026826504e-06,
      "loss": 0.8682,
      "step": 469
    },
    {
      "epoch": 0.04238625603102313,
      "grad_norm": 2.3886356428990143,
      "learning_rate": 3.998398739132746e-06,
      "loss": 0.8855,
      "step": 470
    },
    {
      "epoch": 0.04247643955449339,
      "grad_norm": 1.807790161627835,
      "learning_rate": 3.99837528092459e-06,
      "loss": 0.796,
      "step": 471
    },
    {
      "epoch": 0.042566623077963654,
      "grad_norm": 2.1765222024028357,
      "learning_rate": 3.998351652204034e-06,
      "loss": 0.8858,
      "step": 472
    },
    {
      "epoch": 0.042656806601433915,
      "grad_norm": 1.55019184894134,
      "learning_rate": 3.998327852973098e-06,
      "loss": 0.8812,
      "step": 473
    },
    {
      "epoch": 0.04274699012490418,
      "grad_norm": 2.0415001107093693,
      "learning_rate": 3.99830388323381e-06,
      "loss": 0.7994,
      "step": 474
    },
    {
      "epoch": 0.042837173648374444,
      "grad_norm": 1.7293730080292882,
      "learning_rate": 3.998279742988216e-06,
      "loss": 0.8719,
      "step": 475
    },
    {
      "epoch": 0.042927357171844704,
      "grad_norm": 1.7427501368672393,
      "learning_rate": 3.998255432238377e-06,
      "loss": 0.8141,
      "step": 476
    },
    {
      "epoch": 0.043017540695314965,
      "grad_norm": 1.7233512324222064,
      "learning_rate": 3.9982309509863656e-06,
      "loss": 0.9178,
      "step": 477
    },
    {
      "epoch": 0.043107724218785226,
      "grad_norm": 2.3935942317788594,
      "learning_rate": 3.998206299234272e-06,
      "loss": 0.8017,
      "step": 478
    },
    {
      "epoch": 0.04319790774225549,
      "grad_norm": 1.4936757280727115,
      "learning_rate": 3.998181476984198e-06,
      "loss": 0.8552,
      "step": 479
    },
    {
      "epoch": 0.043288091265725755,
      "grad_norm": 1.6533230524102371,
      "learning_rate": 3.998156484238263e-06,
      "loss": 0.7704,
      "step": 480
    },
    {
      "epoch": 0.043378274789196015,
      "grad_norm": 1.7555088121828069,
      "learning_rate": 3.998131320998599e-06,
      "loss": 0.7896,
      "step": 481
    },
    {
      "epoch": 0.043468458312666276,
      "grad_norm": 2.268720839972459,
      "learning_rate": 3.998105987267353e-06,
      "loss": 0.8985,
      "step": 482
    },
    {
      "epoch": 0.04355864183613654,
      "grad_norm": 1.5681973916437704,
      "learning_rate": 3.998080483046687e-06,
      "loss": 0.7666,
      "step": 483
    },
    {
      "epoch": 0.0436488253596068,
      "grad_norm": 1.3614820177177467,
      "learning_rate": 3.998054808338776e-06,
      "loss": 0.7859,
      "step": 484
    },
    {
      "epoch": 0.04373900888307706,
      "grad_norm": 1.621588721027491,
      "learning_rate": 3.998028963145812e-06,
      "loss": 0.7972,
      "step": 485
    },
    {
      "epoch": 0.043829192406547327,
      "grad_norm": 1.72909022549108,
      "learning_rate": 3.99800294747e-06,
      "loss": 0.8671,
      "step": 486
    },
    {
      "epoch": 0.04391937593001759,
      "grad_norm": 1.5929100216623746,
      "learning_rate": 3.99797676131356e-06,
      "loss": 0.8685,
      "step": 487
    },
    {
      "epoch": 0.04400955945348785,
      "grad_norm": 1.906919455621144,
      "learning_rate": 3.997950404678726e-06,
      "loss": 0.7045,
      "step": 488
    },
    {
      "epoch": 0.04409974297695811,
      "grad_norm": 1.8366586121222719,
      "learning_rate": 3.997923877567746e-06,
      "loss": 0.9575,
      "step": 489
    },
    {
      "epoch": 0.04418992650042837,
      "grad_norm": 1.7430952640922779,
      "learning_rate": 3.9978971799828855e-06,
      "loss": 0.6756,
      "step": 490
    },
    {
      "epoch": 0.04428011002389863,
      "grad_norm": 3.5080685936236136,
      "learning_rate": 3.997870311926421e-06,
      "loss": 0.8365,
      "step": 491
    },
    {
      "epoch": 0.0443702935473689,
      "grad_norm": 1.6890376820536286,
      "learning_rate": 3.997843273400645e-06,
      "loss": 0.8014,
      "step": 492
    },
    {
      "epoch": 0.04446047707083916,
      "grad_norm": 1.5266252862050302,
      "learning_rate": 3.997816064407865e-06,
      "loss": 0.8249,
      "step": 493
    },
    {
      "epoch": 0.04455066059430942,
      "grad_norm": 1.5019932060697585,
      "learning_rate": 3.997788684950402e-06,
      "loss": 0.864,
      "step": 494
    },
    {
      "epoch": 0.04464084411777968,
      "grad_norm": 1.9667687513505654,
      "learning_rate": 3.997761135030593e-06,
      "loss": 0.8893,
      "step": 495
    },
    {
      "epoch": 0.04473102764124994,
      "grad_norm": 1.6816552988934041,
      "learning_rate": 3.997733414650789e-06,
      "loss": 0.7412,
      "step": 496
    },
    {
      "epoch": 0.0448212111647202,
      "grad_norm": 1.989355550246059,
      "learning_rate": 3.9977055238133554e-06,
      "loss": 0.7959,
      "step": 497
    },
    {
      "epoch": 0.04491139468819047,
      "grad_norm": 1.4009238362544774,
      "learning_rate": 3.99767746252067e-06,
      "loss": 0.7154,
      "step": 498
    },
    {
      "epoch": 0.04500157821166073,
      "grad_norm": 1.6227864080542982,
      "learning_rate": 3.997649230775129e-06,
      "loss": 0.8468,
      "step": 499
    },
    {
      "epoch": 0.04509176173513099,
      "grad_norm": 1.6453127782229109,
      "learning_rate": 3.9976208285791395e-06,
      "loss": 0.7835,
      "step": 500
    },
    {
      "epoch": 0.04518194525860125,
      "grad_norm": 0.7900732007095296,
      "learning_rate": 3.997592255935127e-06,
      "loss": 0.6725,
      "step": 501
    },
    {
      "epoch": 0.045272128782071513,
      "grad_norm": 1.6945748548230712,
      "learning_rate": 3.997563512845529e-06,
      "loss": 0.8687,
      "step": 502
    },
    {
      "epoch": 0.045362312305541774,
      "grad_norm": 2.4048426091645823,
      "learning_rate": 3.9975345993127975e-06,
      "loss": 0.8145,
      "step": 503
    },
    {
      "epoch": 0.04545249582901204,
      "grad_norm": 1.8881849195713925,
      "learning_rate": 3.9975055153393985e-06,
      "loss": 0.8555,
      "step": 504
    },
    {
      "epoch": 0.0455426793524823,
      "grad_norm": 1.689357159485432,
      "learning_rate": 3.997476260927816e-06,
      "loss": 0.8566,
      "step": 505
    },
    {
      "epoch": 0.045632862875952564,
      "grad_norm": 2.0262849427963174,
      "learning_rate": 3.997446836080545e-06,
      "loss": 0.816,
      "step": 506
    },
    {
      "epoch": 0.045723046399422825,
      "grad_norm": 2.246194800716018,
      "learning_rate": 3.997417240800095e-06,
      "loss": 0.8958,
      "step": 507
    },
    {
      "epoch": 0.045813229922893085,
      "grad_norm": 1.586056709507168,
      "learning_rate": 3.997387475088994e-06,
      "loss": 0.9375,
      "step": 508
    },
    {
      "epoch": 0.045903413446363346,
      "grad_norm": 2.001964319709554,
      "learning_rate": 3.99735753894978e-06,
      "loss": 0.9435,
      "step": 509
    },
    {
      "epoch": 0.045993596969833614,
      "grad_norm": 1.742041611657312,
      "learning_rate": 3.997327432385006e-06,
      "loss": 0.8439,
      "step": 510
    },
    {
      "epoch": 0.046083780493303875,
      "grad_norm": 1.89039717633589,
      "learning_rate": 3.997297155397244e-06,
      "loss": 0.774,
      "step": 511
    },
    {
      "epoch": 0.046173964016774136,
      "grad_norm": 2.129421234838152,
      "learning_rate": 3.997266707989074e-06,
      "loss": 0.7554,
      "step": 512
    },
    {
      "epoch": 0.046264147540244396,
      "grad_norm": 2.213583505272492,
      "learning_rate": 3.997236090163097e-06,
      "loss": 0.7697,
      "step": 513
    },
    {
      "epoch": 0.04635433106371466,
      "grad_norm": 1.6968998663048032,
      "learning_rate": 3.9972053019219235e-06,
      "loss": 0.9742,
      "step": 514
    },
    {
      "epoch": 0.04644451458718492,
      "grad_norm": 1.90599927660437,
      "learning_rate": 3.997174343268181e-06,
      "loss": 0.8593,
      "step": 515
    },
    {
      "epoch": 0.046534698110655186,
      "grad_norm": 1.5922765465450686,
      "learning_rate": 3.9971432142045115e-06,
      "loss": 0.8442,
      "step": 516
    },
    {
      "epoch": 0.04662488163412545,
      "grad_norm": 0.8090849043542454,
      "learning_rate": 3.99711191473357e-06,
      "loss": 0.664,
      "step": 517
    },
    {
      "epoch": 0.04671506515759571,
      "grad_norm": 2.2575871444222577,
      "learning_rate": 3.99708044485803e-06,
      "loss": 0.8565,
      "step": 518
    },
    {
      "epoch": 0.04680524868106597,
      "grad_norm": 2.288845533692851,
      "learning_rate": 3.997048804580574e-06,
      "loss": 0.8466,
      "step": 519
    },
    {
      "epoch": 0.04689543220453623,
      "grad_norm": 1.7631798075717449,
      "learning_rate": 3.997016993903901e-06,
      "loss": 0.7902,
      "step": 520
    },
    {
      "epoch": 0.04698561572800649,
      "grad_norm": 2.714043149291669,
      "learning_rate": 3.996985012830728e-06,
      "loss": 0.9104,
      "step": 521
    },
    {
      "epoch": 0.04707579925147676,
      "grad_norm": 1.5366921543288699,
      "learning_rate": 3.996952861363782e-06,
      "loss": 0.7825,
      "step": 522
    },
    {
      "epoch": 0.04716598277494702,
      "grad_norm": 1.533259416263613,
      "learning_rate": 3.9969205395058064e-06,
      "loss": 0.8476,
      "step": 523
    },
    {
      "epoch": 0.04725616629841728,
      "grad_norm": 1.8802267339621457,
      "learning_rate": 3.99688804725956e-06,
      "loss": 0.841,
      "step": 524
    },
    {
      "epoch": 0.04734634982188754,
      "grad_norm": 1.2209738469333915,
      "learning_rate": 3.996855384627815e-06,
      "loss": 0.7902,
      "step": 525
    },
    {
      "epoch": 0.0474365333453578,
      "grad_norm": 1.719372723216328,
      "learning_rate": 3.996822551613357e-06,
      "loss": 0.8712,
      "step": 526
    },
    {
      "epoch": 0.04752671686882806,
      "grad_norm": 1.6092417393297989,
      "learning_rate": 3.996789548218989e-06,
      "loss": 0.767,
      "step": 527
    },
    {
      "epoch": 0.04761690039229833,
      "grad_norm": 1.521565700801069,
      "learning_rate": 3.996756374447526e-06,
      "loss": 0.7276,
      "step": 528
    },
    {
      "epoch": 0.04770708391576859,
      "grad_norm": 2.1085290413390485,
      "learning_rate": 3.9967230303018005e-06,
      "loss": 0.8555,
      "step": 529
    },
    {
      "epoch": 0.04779726743923885,
      "grad_norm": 1.8467497910659154,
      "learning_rate": 3.996689515784655e-06,
      "loss": 0.8268,
      "step": 530
    },
    {
      "epoch": 0.04788745096270911,
      "grad_norm": 2.0968909545837233,
      "learning_rate": 3.996655830898951e-06,
      "loss": 0.8188,
      "step": 531
    },
    {
      "epoch": 0.04797763448617937,
      "grad_norm": 2.2989452141020297,
      "learning_rate": 3.996621975647562e-06,
      "loss": 0.84,
      "step": 532
    },
    {
      "epoch": 0.04806781800964964,
      "grad_norm": 1.3745344847831549,
      "learning_rate": 3.996587950033377e-06,
      "loss": 0.7687,
      "step": 533
    },
    {
      "epoch": 0.0481580015331199,
      "grad_norm": 2.231731852157988,
      "learning_rate": 3.996553754059299e-06,
      "loss": 0.7754,
      "step": 534
    },
    {
      "epoch": 0.04824818505659016,
      "grad_norm": 1.840579717885437,
      "learning_rate": 3.996519387728245e-06,
      "loss": 0.8495,
      "step": 535
    },
    {
      "epoch": 0.04833836858006042,
      "grad_norm": 1.4833632384010988,
      "learning_rate": 3.9964848510431495e-06,
      "loss": 0.8132,
      "step": 536
    },
    {
      "epoch": 0.048428552103530684,
      "grad_norm": 1.7273216132268565,
      "learning_rate": 3.996450144006957e-06,
      "loss": 0.899,
      "step": 537
    },
    {
      "epoch": 0.048518735627000945,
      "grad_norm": 1.987908348983456,
      "learning_rate": 3.99641526662263e-06,
      "loss": 0.8282,
      "step": 538
    },
    {
      "epoch": 0.04860891915047121,
      "grad_norm": 1.8802827800971542,
      "learning_rate": 3.996380218893145e-06,
      "loss": 0.9246,
      "step": 539
    },
    {
      "epoch": 0.04869910267394147,
      "grad_norm": 1.782510746285466,
      "learning_rate": 3.996345000821491e-06,
      "loss": 0.8636,
      "step": 540
    },
    {
      "epoch": 0.048789286197411734,
      "grad_norm": 2.2180641081259007,
      "learning_rate": 3.996309612410674e-06,
      "loss": 0.9471,
      "step": 541
    },
    {
      "epoch": 0.048879469720881995,
      "grad_norm": 2.1336527416408746,
      "learning_rate": 3.996274053663713e-06,
      "loss": 0.7724,
      "step": 542
    },
    {
      "epoch": 0.048969653244352256,
      "grad_norm": 0.8012232607279285,
      "learning_rate": 3.996238324583643e-06,
      "loss": 0.6195,
      "step": 543
    },
    {
      "epoch": 0.049059836767822516,
      "grad_norm": 1.980479704997746,
      "learning_rate": 3.996202425173512e-06,
      "loss": 0.9155,
      "step": 544
    },
    {
      "epoch": 0.049150020291292784,
      "grad_norm": 1.7999723193901285,
      "learning_rate": 3.996166355436383e-06,
      "loss": 0.8045,
      "step": 545
    },
    {
      "epoch": 0.049240203814763045,
      "grad_norm": 2.4904765890170286,
      "learning_rate": 3.996130115375333e-06,
      "loss": 0.8485,
      "step": 546
    },
    {
      "epoch": 0.049330387338233306,
      "grad_norm": 0.8651560108528268,
      "learning_rate": 3.996093704993456e-06,
      "loss": 0.6346,
      "step": 547
    },
    {
      "epoch": 0.04942057086170357,
      "grad_norm": 1.727971878304182,
      "learning_rate": 3.996057124293857e-06,
      "loss": 0.9237,
      "step": 548
    },
    {
      "epoch": 0.04951075438517383,
      "grad_norm": 1.4708433451685503,
      "learning_rate": 3.996020373279659e-06,
      "loss": 0.8834,
      "step": 549
    },
    {
      "epoch": 0.04960093790864409,
      "grad_norm": 1.5166024271317158,
      "learning_rate": 3.995983451953996e-06,
      "loss": 0.8517,
      "step": 550
    },
    {
      "epoch": 0.049691121432114356,
      "grad_norm": 1.7908374360778496,
      "learning_rate": 3.99594636032002e-06,
      "loss": 0.8503,
      "step": 551
    },
    {
      "epoch": 0.04978130495558462,
      "grad_norm": 2.0957994821088595,
      "learning_rate": 3.995909098380894e-06,
      "loss": 0.7208,
      "step": 552
    },
    {
      "epoch": 0.04987148847905488,
      "grad_norm": 1.4340279984287998,
      "learning_rate": 3.995871666139799e-06,
      "loss": 0.7155,
      "step": 553
    },
    {
      "epoch": 0.04996167200252514,
      "grad_norm": 1.5635742309486447,
      "learning_rate": 3.995834063599928e-06,
      "loss": 0.8114,
      "step": 554
    },
    {
      "epoch": 0.0500518555259954,
      "grad_norm": 1.491609950914281,
      "learning_rate": 3.99579629076449e-06,
      "loss": 0.8562,
      "step": 555
    },
    {
      "epoch": 0.05014203904946566,
      "grad_norm": 1.832556220281204,
      "learning_rate": 3.9957583476367084e-06,
      "loss": 0.8072,
      "step": 556
    },
    {
      "epoch": 0.05023222257293593,
      "grad_norm": 1.8742227850835185,
      "learning_rate": 3.995720234219819e-06,
      "loss": 0.8367,
      "step": 557
    },
    {
      "epoch": 0.05032240609640619,
      "grad_norm": 1.728277950864836,
      "learning_rate": 3.995681950517075e-06,
      "loss": 0.752,
      "step": 558
    },
    {
      "epoch": 0.05041258961987645,
      "grad_norm": 1.7823634097718142,
      "learning_rate": 3.995643496531743e-06,
      "loss": 0.8604,
      "step": 559
    },
    {
      "epoch": 0.05050277314334671,
      "grad_norm": 2.792938445667108,
      "learning_rate": 3.9956048722671044e-06,
      "loss": 0.8475,
      "step": 560
    },
    {
      "epoch": 0.05059295666681697,
      "grad_norm": 1.6479876297606844,
      "learning_rate": 3.995566077726454e-06,
      "loss": 0.8138,
      "step": 561
    },
    {
      "epoch": 0.05068314019028723,
      "grad_norm": 2.184221262835891,
      "learning_rate": 3.995527112913103e-06,
      "loss": 0.8045,
      "step": 562
    },
    {
      "epoch": 0.0507733237137575,
      "grad_norm": 1.791495322380805,
      "learning_rate": 3.995487977830375e-06,
      "loss": 0.7857,
      "step": 563
    },
    {
      "epoch": 0.05086350723722776,
      "grad_norm": 2.084293131671774,
      "learning_rate": 3.9954486724816105e-06,
      "loss": 0.9276,
      "step": 564
    },
    {
      "epoch": 0.05095369076069802,
      "grad_norm": 2.8962231014683595,
      "learning_rate": 3.995409196870161e-06,
      "loss": 0.8838,
      "step": 565
    },
    {
      "epoch": 0.05104387428416828,
      "grad_norm": 0.7994102092084325,
      "learning_rate": 3.995369550999398e-06,
      "loss": 0.5797,
      "step": 566
    },
    {
      "epoch": 0.05113405780763854,
      "grad_norm": 1.9728904056379009,
      "learning_rate": 3.995329734872702e-06,
      "loss": 0.868,
      "step": 567
    },
    {
      "epoch": 0.051224241331108804,
      "grad_norm": 1.4884323083505504,
      "learning_rate": 3.9952897484934706e-06,
      "loss": 0.9494,
      "step": 568
    },
    {
      "epoch": 0.05131442485457907,
      "grad_norm": 2.080915714029,
      "learning_rate": 3.995249591865115e-06,
      "loss": 0.8501,
      "step": 569
    },
    {
      "epoch": 0.05140460837804933,
      "grad_norm": 1.9138205239066803,
      "learning_rate": 3.995209264991063e-06,
      "loss": 0.7977,
      "step": 570
    },
    {
      "epoch": 0.05149479190151959,
      "grad_norm": 1.7931643443745693,
      "learning_rate": 3.995168767874756e-06,
      "loss": 0.8893,
      "step": 571
    },
    {
      "epoch": 0.051584975424989854,
      "grad_norm": 1.567336341430447,
      "learning_rate": 3.995128100519648e-06,
      "loss": 0.828,
      "step": 572
    },
    {
      "epoch": 0.051675158948460115,
      "grad_norm": 1.9177060073542818,
      "learning_rate": 3.995087262929209e-06,
      "loss": 0.8471,
      "step": 573
    },
    {
      "epoch": 0.051765342471930376,
      "grad_norm": 1.9009914095432163,
      "learning_rate": 3.995046255106925e-06,
      "loss": 0.848,
      "step": 574
    },
    {
      "epoch": 0.05185552599540064,
      "grad_norm": 1.9603700462686873,
      "learning_rate": 3.995005077056293e-06,
      "loss": 0.9041,
      "step": 575
    },
    {
      "epoch": 0.051945709518870904,
      "grad_norm": 3.763564563086439,
      "learning_rate": 3.9949637287808284e-06,
      "loss": 0.7331,
      "step": 576
    },
    {
      "epoch": 0.052035893042341165,
      "grad_norm": 2.1628181929031802,
      "learning_rate": 3.994922210284057e-06,
      "loss": 0.7827,
      "step": 577
    },
    {
      "epoch": 0.052126076565811426,
      "grad_norm": 2.1884743836446092,
      "learning_rate": 3.994880521569524e-06,
      "loss": 0.8757,
      "step": 578
    },
    {
      "epoch": 0.05221626008928169,
      "grad_norm": 1.626217825920974,
      "learning_rate": 3.994838662640785e-06,
      "loss": 0.845,
      "step": 579
    },
    {
      "epoch": 0.05230644361275195,
      "grad_norm": 2.1452130899644866,
      "learning_rate": 3.9947966335014116e-06,
      "loss": 0.8805,
      "step": 580
    },
    {
      "epoch": 0.052396627136222215,
      "grad_norm": 2.186385824652905,
      "learning_rate": 3.99475443415499e-06,
      "loss": 0.824,
      "step": 581
    },
    {
      "epoch": 0.052486810659692476,
      "grad_norm": 2.0761129634677533,
      "learning_rate": 3.994712064605121e-06,
      "loss": 0.8325,
      "step": 582
    },
    {
      "epoch": 0.05257699418316274,
      "grad_norm": 1.9113799312079638,
      "learning_rate": 3.99466952485542e-06,
      "loss": 0.9855,
      "step": 583
    },
    {
      "epoch": 0.052667177706633,
      "grad_norm": 2.505178428889106,
      "learning_rate": 3.994626814909518e-06,
      "loss": 0.8521,
      "step": 584
    },
    {
      "epoch": 0.05275736123010326,
      "grad_norm": 1.9172445821930515,
      "learning_rate": 3.994583934771056e-06,
      "loss": 0.824,
      "step": 585
    },
    {
      "epoch": 0.05284754475357352,
      "grad_norm": 1.7567572923310508,
      "learning_rate": 3.9945408844436955e-06,
      "loss": 0.8944,
      "step": 586
    },
    {
      "epoch": 0.05293772827704379,
      "grad_norm": 2.1834533318698646,
      "learning_rate": 3.994497663931109e-06,
      "loss": 0.8709,
      "step": 587
    },
    {
      "epoch": 0.05302791180051405,
      "grad_norm": 1.823421982255428,
      "learning_rate": 3.994454273236984e-06,
      "loss": 0.856,
      "step": 588
    },
    {
      "epoch": 0.05311809532398431,
      "grad_norm": 1.635426377512616,
      "learning_rate": 3.994410712365023e-06,
      "loss": 0.8508,
      "step": 589
    },
    {
      "epoch": 0.05320827884745457,
      "grad_norm": 2.128904346150238,
      "learning_rate": 3.994366981318943e-06,
      "loss": 0.9349,
      "step": 590
    },
    {
      "epoch": 0.05329846237092483,
      "grad_norm": 1.764295837269606,
      "learning_rate": 3.9943230801024765e-06,
      "loss": 0.8419,
      "step": 591
    },
    {
      "epoch": 0.05338864589439509,
      "grad_norm": 1.7542788103490554,
      "learning_rate": 3.9942790087193666e-06,
      "loss": 0.8746,
      "step": 592
    },
    {
      "epoch": 0.05347882941786536,
      "grad_norm": 1.790959447813763,
      "learning_rate": 3.994234767173376e-06,
      "loss": 0.8313,
      "step": 593
    },
    {
      "epoch": 0.05356901294133562,
      "grad_norm": 3.972048732700636,
      "learning_rate": 3.994190355468279e-06,
      "loss": 0.7353,
      "step": 594
    },
    {
      "epoch": 0.05365919646480588,
      "grad_norm": 1.7800880791806661,
      "learning_rate": 3.994145773607865e-06,
      "loss": 0.8415,
      "step": 595
    },
    {
      "epoch": 0.05374937998827614,
      "grad_norm": 1.7544651967486198,
      "learning_rate": 3.994101021595938e-06,
      "loss": 0.8093,
      "step": 596
    },
    {
      "epoch": 0.0538395635117464,
      "grad_norm": 1.7213610150253411,
      "learning_rate": 3.9940560994363165e-06,
      "loss": 0.8512,
      "step": 597
    },
    {
      "epoch": 0.05392974703521666,
      "grad_norm": 1.9336557089386568,
      "learning_rate": 3.994011007132833e-06,
      "loss": 0.7232,
      "step": 598
    },
    {
      "epoch": 0.05401993055868693,
      "grad_norm": 0.7428637585830059,
      "learning_rate": 3.993965744689337e-06,
      "loss": 0.6501,
      "step": 599
    },
    {
      "epoch": 0.05411011408215719,
      "grad_norm": 1.7786994877584226,
      "learning_rate": 3.993920312109687e-06,
      "loss": 0.8293,
      "step": 600
    },
    {
      "epoch": 0.05420029760562745,
      "grad_norm": 1.7108092238184023,
      "learning_rate": 3.993874709397764e-06,
      "loss": 0.8267,
      "step": 601
    },
    {
      "epoch": 0.05429048112909771,
      "grad_norm": 2.2318663485788526,
      "learning_rate": 3.993828936557454e-06,
      "loss": 0.8254,
      "step": 602
    },
    {
      "epoch": 0.054380664652567974,
      "grad_norm": 1.672554163376511,
      "learning_rate": 3.993782993592667e-06,
      "loss": 0.8008,
      "step": 603
    },
    {
      "epoch": 0.054470848176038235,
      "grad_norm": 1.7808783294010428,
      "learning_rate": 3.993736880507321e-06,
      "loss": 0.8848,
      "step": 604
    },
    {
      "epoch": 0.0545610316995085,
      "grad_norm": 1.9859915929310963,
      "learning_rate": 3.99369059730535e-06,
      "loss": 0.7803,
      "step": 605
    },
    {
      "epoch": 0.05465121522297876,
      "grad_norm": 1.3370359043416136,
      "learning_rate": 3.993644143990706e-06,
      "loss": 0.8339,
      "step": 606
    },
    {
      "epoch": 0.054741398746449024,
      "grad_norm": 1.7798471782746081,
      "learning_rate": 3.99359752056735e-06,
      "loss": 0.7608,
      "step": 607
    },
    {
      "epoch": 0.054831582269919285,
      "grad_norm": 2.1472597535365194,
      "learning_rate": 3.993550727039261e-06,
      "loss": 0.6934,
      "step": 608
    },
    {
      "epoch": 0.054921765793389546,
      "grad_norm": 2.0592832044133718,
      "learning_rate": 3.993503763410431e-06,
      "loss": 0.8592,
      "step": 609
    },
    {
      "epoch": 0.05501194931685981,
      "grad_norm": 1.702607732492725,
      "learning_rate": 3.9934566296848686e-06,
      "loss": 0.7506,
      "step": 610
    },
    {
      "epoch": 0.055102132840330074,
      "grad_norm": 1.861834078028839,
      "learning_rate": 3.993409325866595e-06,
      "loss": 0.9209,
      "step": 611
    },
    {
      "epoch": 0.055192316363800335,
      "grad_norm": 0.6357401014255689,
      "learning_rate": 3.993361851959645e-06,
      "loss": 0.5977,
      "step": 612
    },
    {
      "epoch": 0.055282499887270596,
      "grad_norm": 1.5874277263619687,
      "learning_rate": 3.993314207968071e-06,
      "loss": 0.8773,
      "step": 613
    },
    {
      "epoch": 0.05537268341074086,
      "grad_norm": 2.01224156965399,
      "learning_rate": 3.993266393895938e-06,
      "loss": 0.8194,
      "step": 614
    },
    {
      "epoch": 0.05546286693421112,
      "grad_norm": 3.1137719431762063,
      "learning_rate": 3.993218409747326e-06,
      "loss": 0.8463,
      "step": 615
    },
    {
      "epoch": 0.05555305045768138,
      "grad_norm": 1.8291744252288726,
      "learning_rate": 3.993170255526328e-06,
      "loss": 0.8318,
      "step": 616
    },
    {
      "epoch": 0.055643233981151646,
      "grad_norm": 1.9759351013545496,
      "learning_rate": 3.993121931237054e-06,
      "loss": 0.8668,
      "step": 617
    },
    {
      "epoch": 0.05573341750462191,
      "grad_norm": 1.8419701989214157,
      "learning_rate": 3.993073436883627e-06,
      "loss": 0.857,
      "step": 618
    },
    {
      "epoch": 0.05582360102809217,
      "grad_norm": 1.5280376916450762,
      "learning_rate": 3.993024772470184e-06,
      "loss": 0.8192,
      "step": 619
    },
    {
      "epoch": 0.05591378455156243,
      "grad_norm": 1.832770485164712,
      "learning_rate": 3.992975938000878e-06,
      "loss": 0.8205,
      "step": 620
    },
    {
      "epoch": 0.05600396807503269,
      "grad_norm": 0.6685435850005554,
      "learning_rate": 3.992926933479876e-06,
      "loss": 0.5881,
      "step": 621
    },
    {
      "epoch": 0.05609415159850295,
      "grad_norm": 1.7332149440731228,
      "learning_rate": 3.9928777589113595e-06,
      "loss": 0.8577,
      "step": 622
    },
    {
      "epoch": 0.05618433512197322,
      "grad_norm": 1.7779604334510597,
      "learning_rate": 3.992828414299524e-06,
      "loss": 0.7927,
      "step": 623
    },
    {
      "epoch": 0.05627451864544348,
      "grad_norm": 1.705718419475526,
      "learning_rate": 3.992778899648579e-06,
      "loss": 0.8315,
      "step": 624
    },
    {
      "epoch": 0.05636470216891374,
      "grad_norm": 1.8068514974062475,
      "learning_rate": 3.992729214962751e-06,
      "loss": 0.8605,
      "step": 625
    },
    {
      "epoch": 0.056454885692384,
      "grad_norm": 2.3477876597867144,
      "learning_rate": 3.992679360246279e-06,
      "loss": 0.8783,
      "step": 626
    },
    {
      "epoch": 0.05654506921585426,
      "grad_norm": 1.8680049112569872,
      "learning_rate": 3.992629335503416e-06,
      "loss": 0.8685,
      "step": 627
    },
    {
      "epoch": 0.05663525273932452,
      "grad_norm": 1.4769895228774297,
      "learning_rate": 3.9925791407384304e-06,
      "loss": 0.817,
      "step": 628
    },
    {
      "epoch": 0.05672543626279479,
      "grad_norm": 1.9601477138170735,
      "learning_rate": 3.992528775955606e-06,
      "loss": 0.8807,
      "step": 629
    },
    {
      "epoch": 0.05681561978626505,
      "grad_norm": 1.834799555918954,
      "learning_rate": 3.992478241159239e-06,
      "loss": 0.8097,
      "step": 630
    },
    {
      "epoch": 0.05690580330973531,
      "grad_norm": 1.9274769836318202,
      "learning_rate": 3.992427536353643e-06,
      "loss": 0.8545,
      "step": 631
    },
    {
      "epoch": 0.05699598683320557,
      "grad_norm": 2.037084328779993,
      "learning_rate": 3.992376661543143e-06,
      "loss": 0.8321,
      "step": 632
    },
    {
      "epoch": 0.05708617035667583,
      "grad_norm": 2.374928925103232,
      "learning_rate": 3.992325616732081e-06,
      "loss": 0.8283,
      "step": 633
    },
    {
      "epoch": 0.057176353880146094,
      "grad_norm": 1.8797772900578635,
      "learning_rate": 3.992274401924811e-06,
      "loss": 0.9358,
      "step": 634
    },
    {
      "epoch": 0.05726653740361636,
      "grad_norm": 2.652929317235903,
      "learning_rate": 3.992223017125704e-06,
      "loss": 0.7571,
      "step": 635
    },
    {
      "epoch": 0.05735672092708662,
      "grad_norm": 1.6428134672507837,
      "learning_rate": 3.992171462339145e-06,
      "loss": 0.8609,
      "step": 636
    },
    {
      "epoch": 0.057446904450556883,
      "grad_norm": 1.7740674349494636,
      "learning_rate": 3.992119737569532e-06,
      "loss": 0.9037,
      "step": 637
    },
    {
      "epoch": 0.057537087974027144,
      "grad_norm": 2.10979677857831,
      "learning_rate": 3.992067842821277e-06,
      "loss": 0.8318,
      "step": 638
    },
    {
      "epoch": 0.057627271497497405,
      "grad_norm": 2.580640825337182,
      "learning_rate": 3.99201577809881e-06,
      "loss": 0.9665,
      "step": 639
    },
    {
      "epoch": 0.057717455020967666,
      "grad_norm": 2.230763003419031,
      "learning_rate": 3.991963543406574e-06,
      "loss": 0.8587,
      "step": 640
    },
    {
      "epoch": 0.057807638544437934,
      "grad_norm": 1.95546832082746,
      "learning_rate": 3.991911138749024e-06,
      "loss": 0.8328,
      "step": 641
    },
    {
      "epoch": 0.057897822067908195,
      "grad_norm": 2.333976509368284,
      "learning_rate": 3.991858564130633e-06,
      "loss": 0.8571,
      "step": 642
    },
    {
      "epoch": 0.057988005591378455,
      "grad_norm": 2.7628781745939697,
      "learning_rate": 3.991805819555885e-06,
      "loss": 0.8195,
      "step": 643
    },
    {
      "epoch": 0.058078189114848716,
      "grad_norm": 1.7445112392099567,
      "learning_rate": 3.991752905029283e-06,
      "loss": 0.8321,
      "step": 644
    },
    {
      "epoch": 0.05816837263831898,
      "grad_norm": 1.9291595478077563,
      "learning_rate": 3.991699820555341e-06,
      "loss": 0.7067,
      "step": 645
    },
    {
      "epoch": 0.05825855616178924,
      "grad_norm": 3.1286110423218454,
      "learning_rate": 3.991646566138588e-06,
      "loss": 0.8364,
      "step": 646
    },
    {
      "epoch": 0.058348739685259506,
      "grad_norm": 1.9094123874131825,
      "learning_rate": 3.991593141783567e-06,
      "loss": 0.7599,
      "step": 647
    },
    {
      "epoch": 0.058438923208729766,
      "grad_norm": 1.7525230339811226,
      "learning_rate": 3.991539547494839e-06,
      "loss": 0.8499,
      "step": 648
    },
    {
      "epoch": 0.05852910673220003,
      "grad_norm": 2.9434714799072443,
      "learning_rate": 3.991485783276974e-06,
      "loss": 0.8328,
      "step": 649
    },
    {
      "epoch": 0.05861929025567029,
      "grad_norm": 1.5456762581920414,
      "learning_rate": 3.991431849134563e-06,
      "loss": 0.8852,
      "step": 650
    },
    {
      "epoch": 0.05870947377914055,
      "grad_norm": 1.9824992648648876,
      "learning_rate": 3.991377745072205e-06,
      "loss": 0.7286,
      "step": 651
    },
    {
      "epoch": 0.05879965730261081,
      "grad_norm": 2.116281463335079,
      "learning_rate": 3.991323471094517e-06,
      "loss": 0.8662,
      "step": 652
    },
    {
      "epoch": 0.05888984082608108,
      "grad_norm": 1.8225462473586196,
      "learning_rate": 3.991269027206131e-06,
      "loss": 0.8026,
      "step": 653
    },
    {
      "epoch": 0.05898002434955134,
      "grad_norm": 0.8019105716941335,
      "learning_rate": 3.9912144134116916e-06,
      "loss": 0.6398,
      "step": 654
    },
    {
      "epoch": 0.0590702078730216,
      "grad_norm": 1.7605578875657484,
      "learning_rate": 3.99115962971586e-06,
      "loss": 0.7611,
      "step": 655
    },
    {
      "epoch": 0.05916039139649186,
      "grad_norm": 2.7374882876350295,
      "learning_rate": 3.991104676123308e-06,
      "loss": 0.8625,
      "step": 656
    },
    {
      "epoch": 0.05925057491996212,
      "grad_norm": 0.6548607972424014,
      "learning_rate": 3.991049552638727e-06,
      "loss": 0.5742,
      "step": 657
    },
    {
      "epoch": 0.05934075844343238,
      "grad_norm": 2.291141022017147,
      "learning_rate": 3.99099425926682e-06,
      "loss": 0.7994,
      "step": 658
    },
    {
      "epoch": 0.05943094196690265,
      "grad_norm": 1.9313723972879013,
      "learning_rate": 3.990938796012304e-06,
      "loss": 0.809,
      "step": 659
    },
    {
      "epoch": 0.05952112549037291,
      "grad_norm": 1.9397629474060785,
      "learning_rate": 3.990883162879912e-06,
      "loss": 0.9845,
      "step": 660
    },
    {
      "epoch": 0.05961130901384317,
      "grad_norm": 2.6320230902264803,
      "learning_rate": 3.990827359874391e-06,
      "loss": 0.9007,
      "step": 661
    },
    {
      "epoch": 0.05970149253731343,
      "grad_norm": 1.9510592102576365,
      "learning_rate": 3.990771387000503e-06,
      "loss": 0.8328,
      "step": 662
    },
    {
      "epoch": 0.05979167606078369,
      "grad_norm": 1.9596075890340592,
      "learning_rate": 3.990715244263023e-06,
      "loss": 0.8602,
      "step": 663
    },
    {
      "epoch": 0.05988185958425396,
      "grad_norm": 0.7431389181567802,
      "learning_rate": 3.990658931666741e-06,
      "loss": 0.6072,
      "step": 664
    },
    {
      "epoch": 0.05997204310772422,
      "grad_norm": 1.722046828653354,
      "learning_rate": 3.990602449216463e-06,
      "loss": 0.7643,
      "step": 665
    },
    {
      "epoch": 0.06006222663119448,
      "grad_norm": 1.9546608951261348,
      "learning_rate": 3.990545796917008e-06,
      "loss": 0.7372,
      "step": 666
    },
    {
      "epoch": 0.06015241015466474,
      "grad_norm": 3.591405618006593,
      "learning_rate": 3.99048897477321e-06,
      "loss": 0.8827,
      "step": 667
    },
    {
      "epoch": 0.060242593678135004,
      "grad_norm": 2.5639228592157437,
      "learning_rate": 3.990431982789917e-06,
      "loss": 0.8519,
      "step": 668
    },
    {
      "epoch": 0.060332777201605264,
      "grad_norm": 2.0153998907196904,
      "learning_rate": 3.990374820971992e-06,
      "loss": 0.7686,
      "step": 669
    },
    {
      "epoch": 0.06042296072507553,
      "grad_norm": 1.9782652625906805,
      "learning_rate": 3.990317489324312e-06,
      "loss": 0.7244,
      "step": 670
    },
    {
      "epoch": 0.06051314424854579,
      "grad_norm": 2.2595426800608034,
      "learning_rate": 3.99025998785177e-06,
      "loss": 0.8255,
      "step": 671
    },
    {
      "epoch": 0.060603327772016054,
      "grad_norm": 1.656942834433219,
      "learning_rate": 3.990202316559271e-06,
      "loss": 0.7771,
      "step": 672
    },
    {
      "epoch": 0.060693511295486315,
      "grad_norm": 2.163553226190896,
      "learning_rate": 3.990144475451738e-06,
      "loss": 0.8777,
      "step": 673
    },
    {
      "epoch": 0.060783694818956575,
      "grad_norm": 1.638942618845282,
      "learning_rate": 3.9900864645341036e-06,
      "loss": 0.8296,
      "step": 674
    },
    {
      "epoch": 0.060873878342426836,
      "grad_norm": 1.8491157428914857,
      "learning_rate": 3.990028283811319e-06,
      "loss": 0.8429,
      "step": 675
    },
    {
      "epoch": 0.060964061865897104,
      "grad_norm": 1.684283405141982,
      "learning_rate": 3.989969933288348e-06,
      "loss": 0.8231,
      "step": 676
    },
    {
      "epoch": 0.061054245389367365,
      "grad_norm": 0.6732877471886547,
      "learning_rate": 3.98991141297017e-06,
      "loss": 0.5512,
      "step": 677
    },
    {
      "epoch": 0.061144428912837626,
      "grad_norm": 1.8272788543072225,
      "learning_rate": 3.989852722861778e-06,
      "loss": 0.7819,
      "step": 678
    },
    {
      "epoch": 0.061234612436307886,
      "grad_norm": 0.7261433776878155,
      "learning_rate": 3.98979386296818e-06,
      "loss": 0.5798,
      "step": 679
    },
    {
      "epoch": 0.06132479595977815,
      "grad_norm": 2.136228069195081,
      "learning_rate": 3.989734833294398e-06,
      "loss": 0.8103,
      "step": 680
    },
    {
      "epoch": 0.06141497948324841,
      "grad_norm": 1.6748698853016701,
      "learning_rate": 3.989675633845469e-06,
      "loss": 0.8959,
      "step": 681
    },
    {
      "epoch": 0.061505163006718676,
      "grad_norm": 2.017504503323092,
      "learning_rate": 3.989616264626443e-06,
      "loss": 0.8122,
      "step": 682
    },
    {
      "epoch": 0.06159534653018894,
      "grad_norm": 1.5708216510746416,
      "learning_rate": 3.989556725642388e-06,
      "loss": 0.8646,
      "step": 683
    },
    {
      "epoch": 0.0616855300536592,
      "grad_norm": 1.7333669396345983,
      "learning_rate": 3.989497016898382e-06,
      "loss": 0.886,
      "step": 684
    },
    {
      "epoch": 0.06177571357712946,
      "grad_norm": 2.127131458999647,
      "learning_rate": 3.98943713839952e-06,
      "loss": 0.8247,
      "step": 685
    },
    {
      "epoch": 0.06186589710059972,
      "grad_norm": 0.6880080773217647,
      "learning_rate": 3.9893770901509125e-06,
      "loss": 0.6085,
      "step": 686
    },
    {
      "epoch": 0.06195608062406998,
      "grad_norm": 1.5143573611653471,
      "learning_rate": 3.989316872157682e-06,
      "loss": 0.8172,
      "step": 687
    },
    {
      "epoch": 0.06204626414754025,
      "grad_norm": 5.221632475503089,
      "learning_rate": 3.989256484424968e-06,
      "loss": 0.9045,
      "step": 688
    },
    {
      "epoch": 0.06213644767101051,
      "grad_norm": 1.995454211717879,
      "learning_rate": 3.98919592695792e-06,
      "loss": 0.8056,
      "step": 689
    },
    {
      "epoch": 0.06222663119448077,
      "grad_norm": 2.6436476703436353,
      "learning_rate": 3.9891351997617096e-06,
      "loss": 0.8572,
      "step": 690
    },
    {
      "epoch": 0.06231681471795103,
      "grad_norm": 2.012062532253287,
      "learning_rate": 3.989074302841514e-06,
      "loss": 0.7802,
      "step": 691
    },
    {
      "epoch": 0.06240699824142129,
      "grad_norm": 2.411883608488095,
      "learning_rate": 3.989013236202533e-06,
      "loss": 0.9462,
      "step": 692
    },
    {
      "epoch": 0.06249718176489155,
      "grad_norm": 1.6076191332678664,
      "learning_rate": 3.988951999849974e-06,
      "loss": 0.8696,
      "step": 693
    },
    {
      "epoch": 0.06258736528836181,
      "grad_norm": 2.061658861129569,
      "learning_rate": 3.988890593789064e-06,
      "loss": 0.7526,
      "step": 694
    },
    {
      "epoch": 0.06267754881183207,
      "grad_norm": 1.5125538493256976,
      "learning_rate": 3.9888290180250415e-06,
      "loss": 0.8176,
      "step": 695
    },
    {
      "epoch": 0.06276773233530233,
      "grad_norm": 2.509113295270546,
      "learning_rate": 3.988767272563161e-06,
      "loss": 0.7952,
      "step": 696
    },
    {
      "epoch": 0.06285791585877261,
      "grad_norm": 1.8831659515516348,
      "learning_rate": 3.988705357408691e-06,
      "loss": 0.7735,
      "step": 697
    },
    {
      "epoch": 0.06294809938224287,
      "grad_norm": 1.787059679017872,
      "learning_rate": 3.9886432725669146e-06,
      "loss": 0.79,
      "step": 698
    },
    {
      "epoch": 0.06303828290571313,
      "grad_norm": 1.9917711009759282,
      "learning_rate": 3.988581018043128e-06,
      "loss": 0.7898,
      "step": 699
    },
    {
      "epoch": 0.06312846642918339,
      "grad_norm": 2.734750950717665,
      "learning_rate": 3.988518593842645e-06,
      "loss": 0.8509,
      "step": 700
    },
    {
      "epoch": 0.06321864995265365,
      "grad_norm": 2.0274848668372116,
      "learning_rate": 3.9884559999707906e-06,
      "loss": 0.8273,
      "step": 701
    },
    {
      "epoch": 0.06330883347612391,
      "grad_norm": 2.586999680804044,
      "learning_rate": 3.988393236432906e-06,
      "loss": 0.8412,
      "step": 702
    },
    {
      "epoch": 0.06339901699959417,
      "grad_norm": 1.9405658215331518,
      "learning_rate": 3.988330303234347e-06,
      "loss": 0.9076,
      "step": 703
    },
    {
      "epoch": 0.06348920052306443,
      "grad_norm": 2.4087705415523883,
      "learning_rate": 3.988267200380483e-06,
      "loss": 0.778,
      "step": 704
    },
    {
      "epoch": 0.0635793840465347,
      "grad_norm": 1.5623168838011565,
      "learning_rate": 3.988203927876698e-06,
      "loss": 0.8803,
      "step": 705
    },
    {
      "epoch": 0.06366956757000496,
      "grad_norm": 1.7236324666917147,
      "learning_rate": 3.988140485728391e-06,
      "loss": 0.7203,
      "step": 706
    },
    {
      "epoch": 0.06375975109347522,
      "grad_norm": 1.8810013413452362,
      "learning_rate": 3.988076873940975e-06,
      "loss": 0.7784,
      "step": 707
    },
    {
      "epoch": 0.06384993461694548,
      "grad_norm": 2.095210121378825,
      "learning_rate": 3.9880130925198786e-06,
      "loss": 0.8233,
      "step": 708
    },
    {
      "epoch": 0.06394011814041575,
      "grad_norm": 2.8906517233773177,
      "learning_rate": 3.987949141470543e-06,
      "loss": 0.8418,
      "step": 709
    },
    {
      "epoch": 0.06403030166388601,
      "grad_norm": 1.6901167608026002,
      "learning_rate": 3.987885020798425e-06,
      "loss": 0.834,
      "step": 710
    },
    {
      "epoch": 0.06412048518735627,
      "grad_norm": 3.3026835540885937,
      "learning_rate": 3.987820730508996e-06,
      "loss": 0.8001,
      "step": 711
    },
    {
      "epoch": 0.06421066871082654,
      "grad_norm": 1.916183479881443,
      "learning_rate": 3.987756270607742e-06,
      "loss": 0.9289,
      "step": 712
    },
    {
      "epoch": 0.0643008522342968,
      "grad_norm": 0.6637457597097813,
      "learning_rate": 3.987691641100162e-06,
      "loss": 0.566,
      "step": 713
    },
    {
      "epoch": 0.06439103575776706,
      "grad_norm": 2.247994164612008,
      "learning_rate": 3.987626841991771e-06,
      "loss": 0.9219,
      "step": 714
    },
    {
      "epoch": 0.06448121928123732,
      "grad_norm": 2.519627106491717,
      "learning_rate": 3.987561873288099e-06,
      "loss": 0.8665,
      "step": 715
    },
    {
      "epoch": 0.06457140280470758,
      "grad_norm": 1.8942974772208299,
      "learning_rate": 3.987496734994688e-06,
      "loss": 0.7806,
      "step": 716
    },
    {
      "epoch": 0.06466158632817784,
      "grad_norm": 1.993756203955681,
      "learning_rate": 3.987431427117097e-06,
      "loss": 0.8642,
      "step": 717
    },
    {
      "epoch": 0.0647517698516481,
      "grad_norm": 2.3122811084972206,
      "learning_rate": 3.9873659496608985e-06,
      "loss": 0.8462,
      "step": 718
    },
    {
      "epoch": 0.06484195337511836,
      "grad_norm": 1.9117016615764462,
      "learning_rate": 3.987300302631678e-06,
      "loss": 0.7817,
      "step": 719
    },
    {
      "epoch": 0.06493213689858862,
      "grad_norm": 2.609139437521719,
      "learning_rate": 3.987234486035039e-06,
      "loss": 0.87,
      "step": 720
    },
    {
      "epoch": 0.0650223204220589,
      "grad_norm": 2.6841518926662946,
      "learning_rate": 3.987168499876595e-06,
      "loss": 0.8891,
      "step": 721
    },
    {
      "epoch": 0.06511250394552916,
      "grad_norm": 2.1852982475943006,
      "learning_rate": 3.987102344161978e-06,
      "loss": 0.8429,
      "step": 722
    },
    {
      "epoch": 0.06520268746899942,
      "grad_norm": 1.8537108757638805,
      "learning_rate": 3.987036018896832e-06,
      "loss": 0.8651,
      "step": 723
    },
    {
      "epoch": 0.06529287099246968,
      "grad_norm": 1.8283000112375727,
      "learning_rate": 3.986969524086817e-06,
      "loss": 0.8949,
      "step": 724
    },
    {
      "epoch": 0.06538305451593994,
      "grad_norm": 2.385191683313922,
      "learning_rate": 3.986902859737605e-06,
      "loss": 0.8122,
      "step": 725
    },
    {
      "epoch": 0.0654732380394102,
      "grad_norm": 1.7210860589436077,
      "learning_rate": 3.986836025854886e-06,
      "loss": 0.7639,
      "step": 726
    },
    {
      "epoch": 0.06556342156288046,
      "grad_norm": 1.888697500755251,
      "learning_rate": 3.986769022444362e-06,
      "loss": 0.7908,
      "step": 727
    },
    {
      "epoch": 0.06565360508635072,
      "grad_norm": 6.748365416233447,
      "learning_rate": 3.986701849511751e-06,
      "loss": 0.8231,
      "step": 728
    },
    {
      "epoch": 0.06574378860982098,
      "grad_norm": 2.310280327404583,
      "learning_rate": 3.986634507062782e-06,
      "loss": 0.8071,
      "step": 729
    },
    {
      "epoch": 0.06583397213329124,
      "grad_norm": 1.853660457246493,
      "learning_rate": 3.986566995103204e-06,
      "loss": 0.8512,
      "step": 730
    },
    {
      "epoch": 0.0659241556567615,
      "grad_norm": 2.236333348963604,
      "learning_rate": 3.986499313638776e-06,
      "loss": 0.7657,
      "step": 731
    },
    {
      "epoch": 0.06601433918023177,
      "grad_norm": 1.7119414985055477,
      "learning_rate": 3.986431462675272e-06,
      "loss": 0.7605,
      "step": 732
    },
    {
      "epoch": 0.06610452270370204,
      "grad_norm": 1.4489237441380083,
      "learning_rate": 3.9863634422184835e-06,
      "loss": 0.8427,
      "step": 733
    },
    {
      "epoch": 0.0661947062271723,
      "grad_norm": 1.8408795658359824,
      "learning_rate": 3.986295252274213e-06,
      "loss": 0.862,
      "step": 734
    },
    {
      "epoch": 0.06628488975064256,
      "grad_norm": 1.809876944537132,
      "learning_rate": 3.9862268928482796e-06,
      "loss": 0.8716,
      "step": 735
    },
    {
      "epoch": 0.06637507327411282,
      "grad_norm": 1.6573051114818735,
      "learning_rate": 3.986158363946515e-06,
      "loss": 0.7371,
      "step": 736
    },
    {
      "epoch": 0.06646525679758308,
      "grad_norm": 1.9504754367192105,
      "learning_rate": 3.9860896655747685e-06,
      "loss": 0.8358,
      "step": 737
    },
    {
      "epoch": 0.06655544032105334,
      "grad_norm": 2.1319035979371153,
      "learning_rate": 3.9860207977388994e-06,
      "loss": 0.9007,
      "step": 738
    },
    {
      "epoch": 0.0666456238445236,
      "grad_norm": 1.7414419862786406,
      "learning_rate": 3.9859517604447854e-06,
      "loss": 0.7613,
      "step": 739
    },
    {
      "epoch": 0.06673580736799387,
      "grad_norm": 1.6207494898719814,
      "learning_rate": 3.985882553698317e-06,
      "loss": 0.7956,
      "step": 740
    },
    {
      "epoch": 0.06682599089146413,
      "grad_norm": 2.0499625179306706,
      "learning_rate": 3.985813177505399e-06,
      "loss": 0.8946,
      "step": 741
    },
    {
      "epoch": 0.06691617441493439,
      "grad_norm": 1.774769891679272,
      "learning_rate": 3.985743631871951e-06,
      "loss": 0.7754,
      "step": 742
    },
    {
      "epoch": 0.06700635793840465,
      "grad_norm": 1.9450910204695762,
      "learning_rate": 3.985673916803907e-06,
      "loss": 0.8234,
      "step": 743
    },
    {
      "epoch": 0.06709654146187491,
      "grad_norm": 1.9624831958221394,
      "learning_rate": 3.985604032307215e-06,
      "loss": 0.8521,
      "step": 744
    },
    {
      "epoch": 0.06718672498534518,
      "grad_norm": 1.9689685155309056,
      "learning_rate": 3.985533978387839e-06,
      "loss": 0.8347,
      "step": 745
    },
    {
      "epoch": 0.06727690850881544,
      "grad_norm": 3.3627922243106663,
      "learning_rate": 3.985463755051756e-06,
      "loss": 0.871,
      "step": 746
    },
    {
      "epoch": 0.0673670920322857,
      "grad_norm": 8.13859195912588,
      "learning_rate": 3.9853933623049575e-06,
      "loss": 0.8249,
      "step": 747
    },
    {
      "epoch": 0.06745727555575597,
      "grad_norm": 1.6698542709497457,
      "learning_rate": 3.98532280015345e-06,
      "loss": 0.8263,
      "step": 748
    },
    {
      "epoch": 0.06754745907922623,
      "grad_norm": 1.722761293698909,
      "learning_rate": 3.985252068603254e-06,
      "loss": 0.7534,
      "step": 749
    },
    {
      "epoch": 0.06763764260269649,
      "grad_norm": 1.6884906121767624,
      "learning_rate": 3.985181167660406e-06,
      "loss": 0.8602,
      "step": 750
    },
    {
      "epoch": 0.06772782612616675,
      "grad_norm": 1.8883263983042213,
      "learning_rate": 3.985110097330953e-06,
      "loss": 0.8485,
      "step": 751
    },
    {
      "epoch": 0.06781800964963701,
      "grad_norm": 1.62872254870359,
      "learning_rate": 3.985038857620962e-06,
      "loss": 0.8179,
      "step": 752
    },
    {
      "epoch": 0.06790819317310727,
      "grad_norm": 0.6909351360223669,
      "learning_rate": 3.9849674485365094e-06,
      "loss": 0.5906,
      "step": 753
    },
    {
      "epoch": 0.06799837669657753,
      "grad_norm": 2.3906944115229596,
      "learning_rate": 3.98489587008369e-06,
      "loss": 0.7594,
      "step": 754
    },
    {
      "epoch": 0.06808856022004779,
      "grad_norm": 2.506198637095939,
      "learning_rate": 3.98482412226861e-06,
      "loss": 0.6741,
      "step": 755
    },
    {
      "epoch": 0.06817874374351805,
      "grad_norm": 1.756351457156368,
      "learning_rate": 3.984752205097391e-06,
      "loss": 0.8582,
      "step": 756
    },
    {
      "epoch": 0.06826892726698833,
      "grad_norm": 1.6852087077260136,
      "learning_rate": 3.984680118576171e-06,
      "loss": 0.8348,
      "step": 757
    },
    {
      "epoch": 0.06835911079045859,
      "grad_norm": 1.729882272231625,
      "learning_rate": 3.984607862711099e-06,
      "loss": 0.7861,
      "step": 758
    },
    {
      "epoch": 0.06844929431392885,
      "grad_norm": 4.822446378624342,
      "learning_rate": 3.984535437508341e-06,
      "loss": 0.8734,
      "step": 759
    },
    {
      "epoch": 0.06853947783739911,
      "grad_norm": 0.6694648228991199,
      "learning_rate": 3.984462842974078e-06,
      "loss": 0.5951,
      "step": 760
    },
    {
      "epoch": 0.06862966136086937,
      "grad_norm": 2.1871514178876748,
      "learning_rate": 3.984390079114502e-06,
      "loss": 0.838,
      "step": 761
    },
    {
      "epoch": 0.06871984488433963,
      "grad_norm": 1.5138826088186927,
      "learning_rate": 3.984317145935824e-06,
      "loss": 0.9445,
      "step": 762
    },
    {
      "epoch": 0.06881002840780989,
      "grad_norm": 1.4845295875764049,
      "learning_rate": 3.984244043444264e-06,
      "loss": 0.7998,
      "step": 763
    },
    {
      "epoch": 0.06890021193128015,
      "grad_norm": 1.5461117952054164,
      "learning_rate": 3.984170771646062e-06,
      "loss": 0.7782,
      "step": 764
    },
    {
      "epoch": 0.06899039545475041,
      "grad_norm": 2.2933022875495674,
      "learning_rate": 3.9840973305474695e-06,
      "loss": 0.8698,
      "step": 765
    },
    {
      "epoch": 0.06908057897822067,
      "grad_norm": 4.040288447667622,
      "learning_rate": 3.984023720154752e-06,
      "loss": 0.8798,
      "step": 766
    },
    {
      "epoch": 0.06917076250169094,
      "grad_norm": 1.3777421570902317,
      "learning_rate": 3.9839499404741915e-06,
      "loss": 0.839,
      "step": 767
    },
    {
      "epoch": 0.0692609460251612,
      "grad_norm": 0.7470464166546997,
      "learning_rate": 3.983875991512082e-06,
      "loss": 0.6111,
      "step": 768
    },
    {
      "epoch": 0.06935112954863147,
      "grad_norm": 2.653211314096975,
      "learning_rate": 3.9838018732747345e-06,
      "loss": 0.841,
      "step": 769
    },
    {
      "epoch": 0.06944131307210173,
      "grad_norm": 1.736215731541454,
      "learning_rate": 3.9837275857684716e-06,
      "loss": 0.8112,
      "step": 770
    },
    {
      "epoch": 0.06953149659557199,
      "grad_norm": 2.4760489427724153,
      "learning_rate": 3.983653128999634e-06,
      "loss": 0.959,
      "step": 771
    },
    {
      "epoch": 0.06962168011904225,
      "grad_norm": 1.8628927341608421,
      "learning_rate": 3.983578502974574e-06,
      "loss": 0.8535,
      "step": 772
    },
    {
      "epoch": 0.06971186364251251,
      "grad_norm": 2.1889798199806947,
      "learning_rate": 3.983503707699658e-06,
      "loss": 0.8206,
      "step": 773
    },
    {
      "epoch": 0.06980204716598278,
      "grad_norm": 1.7089228504550382,
      "learning_rate": 3.983428743181268e-06,
      "loss": 0.8576,
      "step": 774
    },
    {
      "epoch": 0.06989223068945304,
      "grad_norm": 1.8530087516565648,
      "learning_rate": 3.983353609425802e-06,
      "loss": 0.8971,
      "step": 775
    },
    {
      "epoch": 0.0699824142129233,
      "grad_norm": 1.7180115240374447,
      "learning_rate": 3.983278306439671e-06,
      "loss": 0.8244,
      "step": 776
    },
    {
      "epoch": 0.07007259773639356,
      "grad_norm": 2.1275032109775363,
      "learning_rate": 3.983202834229297e-06,
      "loss": 0.863,
      "step": 777
    },
    {
      "epoch": 0.07016278125986382,
      "grad_norm": 2.180627326297148,
      "learning_rate": 3.983127192801123e-06,
      "loss": 0.911,
      "step": 778
    },
    {
      "epoch": 0.07025296478333408,
      "grad_norm": 1.9171060459333933,
      "learning_rate": 3.983051382161602e-06,
      "loss": 0.8642,
      "step": 779
    },
    {
      "epoch": 0.07034314830680434,
      "grad_norm": 3.8043325521032183,
      "learning_rate": 3.982975402317203e-06,
      "loss": 0.8492,
      "step": 780
    },
    {
      "epoch": 0.07043333183027461,
      "grad_norm": 3.213420121642022,
      "learning_rate": 3.982899253274409e-06,
      "loss": 0.8474,
      "step": 781
    },
    {
      "epoch": 0.07052351535374488,
      "grad_norm": 3.8049411571629665,
      "learning_rate": 3.982822935039717e-06,
      "loss": 0.8157,
      "step": 782
    },
    {
      "epoch": 0.07061369887721514,
      "grad_norm": 1.9981635722427702,
      "learning_rate": 3.982746447619638e-06,
      "loss": 0.7602,
      "step": 783
    },
    {
      "epoch": 0.0707038824006854,
      "grad_norm": 0.71591560593375,
      "learning_rate": 3.9826697910207e-06,
      "loss": 0.5852,
      "step": 784
    },
    {
      "epoch": 0.07079406592415566,
      "grad_norm": 2.101797324749399,
      "learning_rate": 3.982592965249442e-06,
      "loss": 0.8138,
      "step": 785
    },
    {
      "epoch": 0.07088424944762592,
      "grad_norm": 3.11479933041832,
      "learning_rate": 3.982515970312422e-06,
      "loss": 0.8692,
      "step": 786
    },
    {
      "epoch": 0.07097443297109618,
      "grad_norm": 3.4446339879814847,
      "learning_rate": 3.982438806216207e-06,
      "loss": 0.8044,
      "step": 787
    },
    {
      "epoch": 0.07106461649456644,
      "grad_norm": 2.103873304809591,
      "learning_rate": 3.982361472967382e-06,
      "loss": 0.821,
      "step": 788
    },
    {
      "epoch": 0.0711548000180367,
      "grad_norm": 5.855052442047048,
      "learning_rate": 3.982283970572546e-06,
      "loss": 0.8975,
      "step": 789
    },
    {
      "epoch": 0.07124498354150696,
      "grad_norm": 1.9264049332560298,
      "learning_rate": 3.982206299038311e-06,
      "loss": 0.8003,
      "step": 790
    },
    {
      "epoch": 0.07133516706497722,
      "grad_norm": 1.6867793628426164,
      "learning_rate": 3.9821284583713054e-06,
      "loss": 0.9019,
      "step": 791
    },
    {
      "epoch": 0.07142535058844748,
      "grad_norm": 2.0316311992066485,
      "learning_rate": 3.98205044857817e-06,
      "loss": 0.819,
      "step": 792
    },
    {
      "epoch": 0.07151553411191776,
      "grad_norm": 0.7577438323381275,
      "learning_rate": 3.981972269665561e-06,
      "loss": 0.6138,
      "step": 793
    },
    {
      "epoch": 0.07160571763538802,
      "grad_norm": 1.7396372788644103,
      "learning_rate": 3.98189392164015e-06,
      "loss": 0.8169,
      "step": 794
    },
    {
      "epoch": 0.07169590115885828,
      "grad_norm": 1.8142342656160142,
      "learning_rate": 3.981815404508621e-06,
      "loss": 0.8679,
      "step": 795
    },
    {
      "epoch": 0.07178608468232854,
      "grad_norm": 1.8516448441234408,
      "learning_rate": 3.981736718277674e-06,
      "loss": 0.8345,
      "step": 796
    },
    {
      "epoch": 0.0718762682057988,
      "grad_norm": 1.8412112197883854,
      "learning_rate": 3.9816578629540235e-06,
      "loss": 0.8644,
      "step": 797
    },
    {
      "epoch": 0.07196645172926906,
      "grad_norm": 1.8587403055642422,
      "learning_rate": 3.981578838544398e-06,
      "loss": 0.838,
      "step": 798
    },
    {
      "epoch": 0.07205663525273932,
      "grad_norm": 0.7021785936525011,
      "learning_rate": 3.981499645055539e-06,
      "loss": 0.5673,
      "step": 799
    },
    {
      "epoch": 0.07214681877620958,
      "grad_norm": 0.6721793306103447,
      "learning_rate": 3.981420282494204e-06,
      "loss": 0.6018,
      "step": 800
    },
    {
      "epoch": 0.07223700229967984,
      "grad_norm": 0.744937900156803,
      "learning_rate": 3.981340750867166e-06,
      "loss": 0.6008,
      "step": 801
    },
    {
      "epoch": 0.0723271858231501,
      "grad_norm": 1.7952699873682416,
      "learning_rate": 3.981261050181209e-06,
      "loss": 0.8136,
      "step": 802
    },
    {
      "epoch": 0.07241736934662037,
      "grad_norm": 2.396917103543075,
      "learning_rate": 3.9811811804431355e-06,
      "loss": 0.7879,
      "step": 803
    },
    {
      "epoch": 0.07250755287009064,
      "grad_norm": 1.6803996860671846,
      "learning_rate": 3.981101141659759e-06,
      "loss": 0.8725,
      "step": 804
    },
    {
      "epoch": 0.0725977363935609,
      "grad_norm": 1.5823516897938592,
      "learning_rate": 3.98102093383791e-06,
      "loss": 0.8571,
      "step": 805
    },
    {
      "epoch": 0.07268791991703116,
      "grad_norm": 2.257830662637803,
      "learning_rate": 3.9809405569844315e-06,
      "loss": 0.7472,
      "step": 806
    },
    {
      "epoch": 0.07277810344050142,
      "grad_norm": 1.9989751335685104,
      "learning_rate": 3.980860011106182e-06,
      "loss": 0.7863,
      "step": 807
    },
    {
      "epoch": 0.07286828696397168,
      "grad_norm": 1.3728768689760145,
      "learning_rate": 3.980779296210033e-06,
      "loss": 0.76,
      "step": 808
    },
    {
      "epoch": 0.07295847048744195,
      "grad_norm": 3.498334624672031,
      "learning_rate": 3.980698412302874e-06,
      "loss": 0.9013,
      "step": 809
    },
    {
      "epoch": 0.0730486540109122,
      "grad_norm": 2.5473547173608924,
      "learning_rate": 3.980617359391604e-06,
      "loss": 0.8351,
      "step": 810
    },
    {
      "epoch": 0.07313883753438247,
      "grad_norm": 1.7025935892605257,
      "learning_rate": 3.98053613748314e-06,
      "loss": 0.7823,
      "step": 811
    },
    {
      "epoch": 0.07322902105785273,
      "grad_norm": 1.7576583455626273,
      "learning_rate": 3.980454746584413e-06,
      "loss": 0.6868,
      "step": 812
    },
    {
      "epoch": 0.07331920458132299,
      "grad_norm": 1.977148881260446,
      "learning_rate": 3.9803731867023665e-06,
      "loss": 0.9047,
      "step": 813
    },
    {
      "epoch": 0.07340938810479325,
      "grad_norm": 1.8815122681358933,
      "learning_rate": 3.9802914578439596e-06,
      "loss": 0.806,
      "step": 814
    },
    {
      "epoch": 0.07349957162826351,
      "grad_norm": 2.0497633960303348,
      "learning_rate": 3.980209560016167e-06,
      "loss": 0.9214,
      "step": 815
    },
    {
      "epoch": 0.07358975515173379,
      "grad_norm": 1.8551412514223498,
      "learning_rate": 3.980127493225975e-06,
      "loss": 0.9165,
      "step": 816
    },
    {
      "epoch": 0.07367993867520405,
      "grad_norm": 2.0042133058869718,
      "learning_rate": 3.980045257480387e-06,
      "loss": 0.8553,
      "step": 817
    },
    {
      "epoch": 0.0737701221986743,
      "grad_norm": 2.1178214049718482,
      "learning_rate": 3.9799628527864205e-06,
      "loss": 0.7782,
      "step": 818
    },
    {
      "epoch": 0.07386030572214457,
      "grad_norm": 2.26259394498069,
      "learning_rate": 3.979880279151106e-06,
      "loss": 0.8096,
      "step": 819
    },
    {
      "epoch": 0.07395048924561483,
      "grad_norm": 2.422919952399097,
      "learning_rate": 3.979797536581489e-06,
      "loss": 0.7694,
      "step": 820
    },
    {
      "epoch": 0.07404067276908509,
      "grad_norm": 2.2416129090435035,
      "learning_rate": 3.97971462508463e-06,
      "loss": 0.6755,
      "step": 821
    },
    {
      "epoch": 0.07413085629255535,
      "grad_norm": 1.51701220708775,
      "learning_rate": 3.979631544667603e-06,
      "loss": 0.8068,
      "step": 822
    },
    {
      "epoch": 0.07422103981602561,
      "grad_norm": 1.6789894250260677,
      "learning_rate": 3.979548295337496e-06,
      "loss": 0.8295,
      "step": 823
    },
    {
      "epoch": 0.07431122333949587,
      "grad_norm": 1.54470194759477,
      "learning_rate": 3.9794648771014146e-06,
      "loss": 0.7913,
      "step": 824
    },
    {
      "epoch": 0.07440140686296613,
      "grad_norm": 1.6467602047370218,
      "learning_rate": 3.9793812899664745e-06,
      "loss": 0.8276,
      "step": 825
    },
    {
      "epoch": 0.0744915903864364,
      "grad_norm": 1.885778283554552,
      "learning_rate": 3.979297533939809e-06,
      "loss": 0.7822,
      "step": 826
    },
    {
      "epoch": 0.07458177390990665,
      "grad_norm": 1.9379799156171797,
      "learning_rate": 3.979213609028564e-06,
      "loss": 0.7966,
      "step": 827
    },
    {
      "epoch": 0.07467195743337693,
      "grad_norm": 1.8125564961014176,
      "learning_rate": 3.979129515239901e-06,
      "loss": 0.8344,
      "step": 828
    },
    {
      "epoch": 0.07476214095684719,
      "grad_norm": 2.019210347222539,
      "learning_rate": 3.979045252580994e-06,
      "loss": 0.8315,
      "step": 829
    },
    {
      "epoch": 0.07485232448031745,
      "grad_norm": 2.1998873248290147,
      "learning_rate": 3.978960821059034e-06,
      "loss": 0.768,
      "step": 830
    },
    {
      "epoch": 0.07494250800378771,
      "grad_norm": 2.0945534374480492,
      "learning_rate": 3.978876220681225e-06,
      "loss": 0.8599,
      "step": 831
    },
    {
      "epoch": 0.07503269152725797,
      "grad_norm": 2.5737953590371974,
      "learning_rate": 3.978791451454786e-06,
      "loss": 0.7508,
      "step": 832
    },
    {
      "epoch": 0.07512287505072823,
      "grad_norm": 2.968340393720566,
      "learning_rate": 3.978706513386949e-06,
      "loss": 0.7816,
      "step": 833
    },
    {
      "epoch": 0.0752130585741985,
      "grad_norm": 2.401555820671726,
      "learning_rate": 3.978621406484962e-06,
      "loss": 0.8331,
      "step": 834
    },
    {
      "epoch": 0.07530324209766875,
      "grad_norm": 2.0651302327983987,
      "learning_rate": 3.978536130756086e-06,
      "loss": 0.8089,
      "step": 835
    },
    {
      "epoch": 0.07539342562113902,
      "grad_norm": 1.801824834140493,
      "learning_rate": 3.978450686207599e-06,
      "loss": 0.8847,
      "step": 836
    },
    {
      "epoch": 0.07548360914460928,
      "grad_norm": 2.208460306320008,
      "learning_rate": 3.978365072846789e-06,
      "loss": 0.8069,
      "step": 837
    },
    {
      "epoch": 0.07557379266807954,
      "grad_norm": 2.883696795213754,
      "learning_rate": 3.9782792906809625e-06,
      "loss": 0.7708,
      "step": 838
    },
    {
      "epoch": 0.0756639761915498,
      "grad_norm": 2.356446020740414,
      "learning_rate": 3.97819333971744e-06,
      "loss": 0.8073,
      "step": 839
    },
    {
      "epoch": 0.07575415971502007,
      "grad_norm": 2.9350639143943,
      "learning_rate": 3.978107219963553e-06,
      "loss": 0.8434,
      "step": 840
    },
    {
      "epoch": 0.07584434323849033,
      "grad_norm": 1.9073250037598866,
      "learning_rate": 3.978020931426651e-06,
      "loss": 0.7307,
      "step": 841
    },
    {
      "epoch": 0.0759345267619606,
      "grad_norm": 1.6191135307594342,
      "learning_rate": 3.977934474114096e-06,
      "loss": 0.762,
      "step": 842
    },
    {
      "epoch": 0.07602471028543085,
      "grad_norm": 0.6477291650531511,
      "learning_rate": 3.977847848033267e-06,
      "loss": 0.5796,
      "step": 843
    },
    {
      "epoch": 0.07611489380890112,
      "grad_norm": 1.7352505784928909,
      "learning_rate": 3.977761053191553e-06,
      "loss": 0.8634,
      "step": 844
    },
    {
      "epoch": 0.07620507733237138,
      "grad_norm": 1.7619670684456237,
      "learning_rate": 3.977674089596361e-06,
      "loss": 0.7996,
      "step": 845
    },
    {
      "epoch": 0.07629526085584164,
      "grad_norm": 0.6967902179001819,
      "learning_rate": 3.97758695725511e-06,
      "loss": 0.5765,
      "step": 846
    },
    {
      "epoch": 0.0763854443793119,
      "grad_norm": 1.391986009464441,
      "learning_rate": 3.977499656175236e-06,
      "loss": 0.8222,
      "step": 847
    },
    {
      "epoch": 0.07647562790278216,
      "grad_norm": 2.1411211315955003,
      "learning_rate": 3.977412186364187e-06,
      "loss": 0.7333,
      "step": 848
    },
    {
      "epoch": 0.07656581142625242,
      "grad_norm": 2.0877199022884696,
      "learning_rate": 3.977324547829428e-06,
      "loss": 0.8476,
      "step": 849
    },
    {
      "epoch": 0.07665599494972268,
      "grad_norm": 2.1087064460638865,
      "learning_rate": 3.977236740578435e-06,
      "loss": 0.8494,
      "step": 850
    },
    {
      "epoch": 0.07674617847319294,
      "grad_norm": 1.9763583956192183,
      "learning_rate": 3.9771487646187015e-06,
      "loss": 0.8113,
      "step": 851
    },
    {
      "epoch": 0.07683636199666322,
      "grad_norm": 2.057342090874008,
      "learning_rate": 3.9770606199577325e-06,
      "loss": 0.8544,
      "step": 852
    },
    {
      "epoch": 0.07692654552013348,
      "grad_norm": 2.1305871329798935,
      "learning_rate": 3.9769723066030505e-06,
      "loss": 0.7592,
      "step": 853
    },
    {
      "epoch": 0.07701672904360374,
      "grad_norm": 3.245814783111046,
      "learning_rate": 3.976883824562191e-06,
      "loss": 0.8809,
      "step": 854
    },
    {
      "epoch": 0.077106912567074,
      "grad_norm": 2.019683655055366,
      "learning_rate": 3.976795173842703e-06,
      "loss": 0.8218,
      "step": 855
    },
    {
      "epoch": 0.07719709609054426,
      "grad_norm": 3.9276258395776162,
      "learning_rate": 3.97670635445215e-06,
      "loss": 0.7965,
      "step": 856
    },
    {
      "epoch": 0.07728727961401452,
      "grad_norm": 2.1331035708835637,
      "learning_rate": 3.976617366398112e-06,
      "loss": 0.7981,
      "step": 857
    },
    {
      "epoch": 0.07737746313748478,
      "grad_norm": 1.4170855669502946,
      "learning_rate": 3.976528209688181e-06,
      "loss": 0.8508,
      "step": 858
    },
    {
      "epoch": 0.07746764666095504,
      "grad_norm": 1.5412347635349832,
      "learning_rate": 3.976438884329965e-06,
      "loss": 0.8237,
      "step": 859
    },
    {
      "epoch": 0.0775578301844253,
      "grad_norm": 1.6380246581339695,
      "learning_rate": 3.976349390331085e-06,
      "loss": 0.9441,
      "step": 860
    },
    {
      "epoch": 0.07764801370789556,
      "grad_norm": 1.8544534254316922,
      "learning_rate": 3.976259727699178e-06,
      "loss": 0.789,
      "step": 861
    },
    {
      "epoch": 0.07773819723136582,
      "grad_norm": 2.1033622664543765,
      "learning_rate": 3.976169896441895e-06,
      "loss": 0.7899,
      "step": 862
    },
    {
      "epoch": 0.07782838075483609,
      "grad_norm": 1.895933532861596,
      "learning_rate": 3.976079896566898e-06,
      "loss": 0.852,
      "step": 863
    },
    {
      "epoch": 0.07791856427830636,
      "grad_norm": 1.8346526499566729,
      "learning_rate": 3.97598972808187e-06,
      "loss": 0.885,
      "step": 864
    },
    {
      "epoch": 0.07800874780177662,
      "grad_norm": 3.259857926110476,
      "learning_rate": 3.975899390994501e-06,
      "loss": 0.8176,
      "step": 865
    },
    {
      "epoch": 0.07809893132524688,
      "grad_norm": 1.5285564788576238,
      "learning_rate": 3.975808885312502e-06,
      "loss": 0.8837,
      "step": 866
    },
    {
      "epoch": 0.07818911484871714,
      "grad_norm": 1.954711330887561,
      "learning_rate": 3.975718211043594e-06,
      "loss": 0.872,
      "step": 867
    },
    {
      "epoch": 0.0782792983721874,
      "grad_norm": 1.9025072971359447,
      "learning_rate": 3.975627368195515e-06,
      "loss": 0.7366,
      "step": 868
    },
    {
      "epoch": 0.07836948189565766,
      "grad_norm": 1.370249866198147,
      "learning_rate": 3.975536356776015e-06,
      "loss": 0.7837,
      "step": 869
    },
    {
      "epoch": 0.07845966541912792,
      "grad_norm": 0.7403482233834734,
      "learning_rate": 3.975445176792861e-06,
      "loss": 0.6535,
      "step": 870
    },
    {
      "epoch": 0.07854984894259819,
      "grad_norm": 1.714803569209996,
      "learning_rate": 3.975353828253831e-06,
      "loss": 0.8266,
      "step": 871
    },
    {
      "epoch": 0.07864003246606845,
      "grad_norm": 1.7583749507058777,
      "learning_rate": 3.97526231116672e-06,
      "loss": 0.8258,
      "step": 872
    },
    {
      "epoch": 0.07873021598953871,
      "grad_norm": 1.7364422274710098,
      "learning_rate": 3.975170625539338e-06,
      "loss": 0.7644,
      "step": 873
    },
    {
      "epoch": 0.07882039951300897,
      "grad_norm": 2.3603687340992057,
      "learning_rate": 3.975078771379507e-06,
      "loss": 0.8161,
      "step": 874
    },
    {
      "epoch": 0.07891058303647923,
      "grad_norm": 2.0281879526956965,
      "learning_rate": 3.974986748695064e-06,
      "loss": 0.7388,
      "step": 875
    },
    {
      "epoch": 0.0790007665599495,
      "grad_norm": 0.72295765133054,
      "learning_rate": 3.974894557493862e-06,
      "loss": 0.5909,
      "step": 876
    },
    {
      "epoch": 0.07909095008341976,
      "grad_norm": 1.5883735776261294,
      "learning_rate": 3.974802197783768e-06,
      "loss": 0.8654,
      "step": 877
    },
    {
      "epoch": 0.07918113360689003,
      "grad_norm": 1.9303558057213517,
      "learning_rate": 3.974709669572661e-06,
      "loss": 0.883,
      "step": 878
    },
    {
      "epoch": 0.07927131713036029,
      "grad_norm": 2.966432650762324,
      "learning_rate": 3.974616972868436e-06,
      "loss": 0.812,
      "step": 879
    },
    {
      "epoch": 0.07936150065383055,
      "grad_norm": 1.9056762707432502,
      "learning_rate": 3.974524107679003e-06,
      "loss": 0.8242,
      "step": 880
    },
    {
      "epoch": 0.07945168417730081,
      "grad_norm": 2.5288975463000902,
      "learning_rate": 3.974431074012286e-06,
      "loss": 0.8132,
      "step": 881
    },
    {
      "epoch": 0.07954186770077107,
      "grad_norm": 2.774534970058829,
      "learning_rate": 3.974337871876223e-06,
      "loss": 0.8528,
      "step": 882
    },
    {
      "epoch": 0.07963205122424133,
      "grad_norm": 1.5582141844270192,
      "learning_rate": 3.974244501278766e-06,
      "loss": 0.7875,
      "step": 883
    },
    {
      "epoch": 0.07972223474771159,
      "grad_norm": 1.683864916849633,
      "learning_rate": 3.974150962227883e-06,
      "loss": 0.7675,
      "step": 884
    },
    {
      "epoch": 0.07981241827118185,
      "grad_norm": 1.7924750818199153,
      "learning_rate": 3.9740572547315535e-06,
      "loss": 0.7584,
      "step": 885
    },
    {
      "epoch": 0.07990260179465211,
      "grad_norm": 2.401918649666445,
      "learning_rate": 3.973963378797775e-06,
      "loss": 0.8517,
      "step": 886
    },
    {
      "epoch": 0.07999278531812237,
      "grad_norm": 1.58465313699093,
      "learning_rate": 3.973869334434556e-06,
      "loss": 0.8334,
      "step": 887
    },
    {
      "epoch": 0.08008296884159265,
      "grad_norm": 1.7707109371596157,
      "learning_rate": 3.973775121649922e-06,
      "loss": 0.8426,
      "step": 888
    },
    {
      "epoch": 0.08017315236506291,
      "grad_norm": 1.5776199439103533,
      "learning_rate": 3.973680740451911e-06,
      "loss": 0.7329,
      "step": 889
    },
    {
      "epoch": 0.08026333588853317,
      "grad_norm": 3.8966953459490035,
      "learning_rate": 3.9735861908485776e-06,
      "loss": 0.7842,
      "step": 890
    },
    {
      "epoch": 0.08035351941200343,
      "grad_norm": 1.8461004243359564,
      "learning_rate": 3.973491472847987e-06,
      "loss": 0.8348,
      "step": 891
    },
    {
      "epoch": 0.08044370293547369,
      "grad_norm": 1.7464748709088793,
      "learning_rate": 3.973396586458222e-06,
      "loss": 0.7519,
      "step": 892
    },
    {
      "epoch": 0.08053388645894395,
      "grad_norm": 1.6353020921489307,
      "learning_rate": 3.97330153168738e-06,
      "loss": 0.8012,
      "step": 893
    },
    {
      "epoch": 0.08062406998241421,
      "grad_norm": 1.9121353724332324,
      "learning_rate": 3.973206308543571e-06,
      "loss": 0.764,
      "step": 894
    },
    {
      "epoch": 0.08071425350588447,
      "grad_norm": 2.319075848903772,
      "learning_rate": 3.973110917034918e-06,
      "loss": 0.7951,
      "step": 895
    },
    {
      "epoch": 0.08080443702935473,
      "grad_norm": 1.566290948671965,
      "learning_rate": 3.973015357169563e-06,
      "loss": 0.7267,
      "step": 896
    },
    {
      "epoch": 0.080894620552825,
      "grad_norm": 1.7048194225814182,
      "learning_rate": 3.972919628955659e-06,
      "loss": 0.7108,
      "step": 897
    },
    {
      "epoch": 0.08098480407629526,
      "grad_norm": 2.2099821904703316,
      "learning_rate": 3.972823732401373e-06,
      "loss": 0.8078,
      "step": 898
    },
    {
      "epoch": 0.08107498759976552,
      "grad_norm": 1.778825548810047,
      "learning_rate": 3.972727667514888e-06,
      "loss": 0.8133,
      "step": 899
    },
    {
      "epoch": 0.08116517112323579,
      "grad_norm": 1.3935047293077363,
      "learning_rate": 3.972631434304402e-06,
      "loss": 0.7773,
      "step": 900
    },
    {
      "epoch": 0.08125535464670605,
      "grad_norm": 1.9839990214606917,
      "learning_rate": 3.972535032778124e-06,
      "loss": 0.9289,
      "step": 901
    },
    {
      "epoch": 0.08134553817017631,
      "grad_norm": 1.824508807302446,
      "learning_rate": 3.97243846294428e-06,
      "loss": 0.8985,
      "step": 902
    },
    {
      "epoch": 0.08143572169364657,
      "grad_norm": 1.555325861086489,
      "learning_rate": 3.972341724811111e-06,
      "loss": 0.9542,
      "step": 903
    },
    {
      "epoch": 0.08152590521711683,
      "grad_norm": 2.0905787806398313,
      "learning_rate": 3.972244818386872e-06,
      "loss": 0.7609,
      "step": 904
    },
    {
      "epoch": 0.0816160887405871,
      "grad_norm": 2.183660025094129,
      "learning_rate": 3.972147743679828e-06,
      "loss": 0.7601,
      "step": 905
    },
    {
      "epoch": 0.08170627226405736,
      "grad_norm": 1.38323985247369,
      "learning_rate": 3.972050500698265e-06,
      "loss": 0.782,
      "step": 906
    },
    {
      "epoch": 0.08179645578752762,
      "grad_norm": 1.9373714650495544,
      "learning_rate": 3.971953089450481e-06,
      "loss": 0.8147,
      "step": 907
    },
    {
      "epoch": 0.08188663931099788,
      "grad_norm": 1.9303865595093956,
      "learning_rate": 3.971855509944784e-06,
      "loss": 0.8912,
      "step": 908
    },
    {
      "epoch": 0.08197682283446814,
      "grad_norm": 1.599561827981158,
      "learning_rate": 3.971757762189504e-06,
      "loss": 0.8,
      "step": 909
    },
    {
      "epoch": 0.0820670063579384,
      "grad_norm": 1.9341871795925216,
      "learning_rate": 3.9716598461929785e-06,
      "loss": 0.8136,
      "step": 910
    },
    {
      "epoch": 0.08215718988140866,
      "grad_norm": 1.6892713504089567,
      "learning_rate": 3.971561761963563e-06,
      "loss": 0.7879,
      "step": 911
    },
    {
      "epoch": 0.08224737340487893,
      "grad_norm": 2.103183050615655,
      "learning_rate": 3.971463509509628e-06,
      "loss": 0.8931,
      "step": 912
    },
    {
      "epoch": 0.0823375569283492,
      "grad_norm": 2.2388269847623765,
      "learning_rate": 3.9713650888395555e-06,
      "loss": 0.8284,
      "step": 913
    },
    {
      "epoch": 0.08242774045181946,
      "grad_norm": 2.0124476029108607,
      "learning_rate": 3.9712664999617425e-06,
      "loss": 0.8067,
      "step": 914
    },
    {
      "epoch": 0.08251792397528972,
      "grad_norm": 1.5462583603629076,
      "learning_rate": 3.971167742884603e-06,
      "loss": 0.7875,
      "step": 915
    },
    {
      "epoch": 0.08260810749875998,
      "grad_norm": 1.8160474197187457,
      "learning_rate": 3.971068817616564e-06,
      "loss": 0.8667,
      "step": 916
    },
    {
      "epoch": 0.08269829102223024,
      "grad_norm": 1.5752108266294669,
      "learning_rate": 3.970969724166064e-06,
      "loss": 0.8395,
      "step": 917
    },
    {
      "epoch": 0.0827884745457005,
      "grad_norm": 0.8056944685313467,
      "learning_rate": 3.970870462541559e-06,
      "loss": 0.5794,
      "step": 918
    },
    {
      "epoch": 0.08287865806917076,
      "grad_norm": 1.5550671597908436,
      "learning_rate": 3.97077103275152e-06,
      "loss": 0.8304,
      "step": 919
    },
    {
      "epoch": 0.08296884159264102,
      "grad_norm": 1.434756605128791,
      "learning_rate": 3.970671434804428e-06,
      "loss": 0.8001,
      "step": 920
    },
    {
      "epoch": 0.08305902511611128,
      "grad_norm": 2.509719646344402,
      "learning_rate": 3.970571668708784e-06,
      "loss": 0.8108,
      "step": 921
    },
    {
      "epoch": 0.08314920863958154,
      "grad_norm": 0.6531359037374347,
      "learning_rate": 3.9704717344731e-06,
      "loss": 0.5887,
      "step": 922
    },
    {
      "epoch": 0.0832393921630518,
      "grad_norm": 0.7141530603457684,
      "learning_rate": 3.9703716321059026e-06,
      "loss": 0.6049,
      "step": 923
    },
    {
      "epoch": 0.08332957568652208,
      "grad_norm": 1.755845796492269,
      "learning_rate": 3.9702713616157325e-06,
      "loss": 0.905,
      "step": 924
    },
    {
      "epoch": 0.08341975920999234,
      "grad_norm": 2.1310124329195763,
      "learning_rate": 3.9701709230111455e-06,
      "loss": 0.8307,
      "step": 925
    },
    {
      "epoch": 0.0835099427334626,
      "grad_norm": 1.6059931805799599,
      "learning_rate": 3.970070316300713e-06,
      "loss": 0.7793,
      "step": 926
    },
    {
      "epoch": 0.08360012625693286,
      "grad_norm": 2.067670640834097,
      "learning_rate": 3.969969541493017e-06,
      "loss": 0.844,
      "step": 927
    },
    {
      "epoch": 0.08369030978040312,
      "grad_norm": 1.5710326104619108,
      "learning_rate": 3.969868598596658e-06,
      "loss": 0.8675,
      "step": 928
    },
    {
      "epoch": 0.08378049330387338,
      "grad_norm": 1.279687393803295,
      "learning_rate": 3.969767487620249e-06,
      "loss": 0.7566,
      "step": 929
    },
    {
      "epoch": 0.08387067682734364,
      "grad_norm": 1.3457535289964038,
      "learning_rate": 3.969666208572416e-06,
      "loss": 0.8635,
      "step": 930
    },
    {
      "epoch": 0.0839608603508139,
      "grad_norm": 2.03249379737724,
      "learning_rate": 3.969564761461802e-06,
      "loss": 0.881,
      "step": 931
    },
    {
      "epoch": 0.08405104387428416,
      "grad_norm": 1.7483140453563215,
      "learning_rate": 3.969463146297062e-06,
      "loss": 0.8,
      "step": 932
    },
    {
      "epoch": 0.08414122739775443,
      "grad_norm": 5.076331564316786,
      "learning_rate": 3.969361363086867e-06,
      "loss": 0.8416,
      "step": 933
    },
    {
      "epoch": 0.08423141092122469,
      "grad_norm": 1.8081767382110816,
      "learning_rate": 3.9692594118399014e-06,
      "loss": 0.7838,
      "step": 934
    },
    {
      "epoch": 0.08432159444469496,
      "grad_norm": 1.8701071796199251,
      "learning_rate": 3.969157292564865e-06,
      "loss": 0.847,
      "step": 935
    },
    {
      "epoch": 0.08441177796816522,
      "grad_norm": 2.2500196032199957,
      "learning_rate": 3.96905500527047e-06,
      "loss": 0.8016,
      "step": 936
    },
    {
      "epoch": 0.08450196149163548,
      "grad_norm": 1.6774828259289634,
      "learning_rate": 3.968952549965445e-06,
      "loss": 0.8082,
      "step": 937
    },
    {
      "epoch": 0.08459214501510574,
      "grad_norm": 1.7048285127912788,
      "learning_rate": 3.968849926658532e-06,
      "loss": 0.8483,
      "step": 938
    },
    {
      "epoch": 0.084682328538576,
      "grad_norm": 2.1083784221708015,
      "learning_rate": 3.9687471353584866e-06,
      "loss": 0.7277,
      "step": 939
    },
    {
      "epoch": 0.08477251206204627,
      "grad_norm": 1.4900794992424062,
      "learning_rate": 3.9686441760740795e-06,
      "loss": 0.8292,
      "step": 940
    },
    {
      "epoch": 0.08486269558551653,
      "grad_norm": 2.1965503393040926,
      "learning_rate": 3.968541048814098e-06,
      "loss": 0.7567,
      "step": 941
    },
    {
      "epoch": 0.08495287910898679,
      "grad_norm": 2.422102492938234,
      "learning_rate": 3.968437753587339e-06,
      "loss": 0.7506,
      "step": 942
    },
    {
      "epoch": 0.08504306263245705,
      "grad_norm": 2.0104000769475885,
      "learning_rate": 3.968334290402616e-06,
      "loss": 0.8616,
      "step": 943
    },
    {
      "epoch": 0.08513324615592731,
      "grad_norm": 1.504460775024406,
      "learning_rate": 3.968230659268759e-06,
      "loss": 0.7099,
      "step": 944
    },
    {
      "epoch": 0.08522342967939757,
      "grad_norm": 1.9576374264899923,
      "learning_rate": 3.968126860194609e-06,
      "loss": 0.893,
      "step": 945
    },
    {
      "epoch": 0.08531361320286783,
      "grad_norm": 1.7234514611129192,
      "learning_rate": 3.968022893189025e-06,
      "loss": 0.8655,
      "step": 946
    },
    {
      "epoch": 0.0854037967263381,
      "grad_norm": 1.8422047880362928,
      "learning_rate": 3.967918758260874e-06,
      "loss": 0.8519,
      "step": 947
    },
    {
      "epoch": 0.08549398024980837,
      "grad_norm": 1.6092468506985353,
      "learning_rate": 3.967814455419044e-06,
      "loss": 0.8659,
      "step": 948
    },
    {
      "epoch": 0.08558416377327863,
      "grad_norm": 2.640105518052992,
      "learning_rate": 3.967709984672434e-06,
      "loss": 0.8644,
      "step": 949
    },
    {
      "epoch": 0.08567434729674889,
      "grad_norm": 2.191656142621706,
      "learning_rate": 3.967605346029959e-06,
      "loss": 0.8483,
      "step": 950
    },
    {
      "epoch": 0.08576453082021915,
      "grad_norm": 1.5369393667895621,
      "learning_rate": 3.9675005395005466e-06,
      "loss": 0.8938,
      "step": 951
    },
    {
      "epoch": 0.08585471434368941,
      "grad_norm": 1.6515449053820512,
      "learning_rate": 3.967395565093139e-06,
      "loss": 0.8484,
      "step": 952
    },
    {
      "epoch": 0.08594489786715967,
      "grad_norm": 1.5186886841755123,
      "learning_rate": 3.967290422816693e-06,
      "loss": 0.8294,
      "step": 953
    },
    {
      "epoch": 0.08603508139062993,
      "grad_norm": 2.2190208806984737,
      "learning_rate": 3.967185112680183e-06,
      "loss": 0.7858,
      "step": 954
    },
    {
      "epoch": 0.08612526491410019,
      "grad_norm": 2.025139994550519,
      "learning_rate": 3.96707963469259e-06,
      "loss": 0.8995,
      "step": 955
    },
    {
      "epoch": 0.08621544843757045,
      "grad_norm": 1.9918148515020628,
      "learning_rate": 3.966973988862917e-06,
      "loss": 0.8133,
      "step": 956
    },
    {
      "epoch": 0.08630563196104071,
      "grad_norm": 3.007841511375916,
      "learning_rate": 3.966868175200178e-06,
      "loss": 0.8713,
      "step": 957
    },
    {
      "epoch": 0.08639581548451097,
      "grad_norm": 1.8554084205858252,
      "learning_rate": 3.9667621937134e-06,
      "loss": 0.8763,
      "step": 958
    },
    {
      "epoch": 0.08648599900798125,
      "grad_norm": 1.8346667498008207,
      "learning_rate": 3.966656044411627e-06,
      "loss": 0.7429,
      "step": 959
    },
    {
      "epoch": 0.08657618253145151,
      "grad_norm": 1.485226597692598,
      "learning_rate": 3.966549727303918e-06,
      "loss": 0.791,
      "step": 960
    },
    {
      "epoch": 0.08666636605492177,
      "grad_norm": 3.810293465873927,
      "learning_rate": 3.966443242399341e-06,
      "loss": 0.8646,
      "step": 961
    },
    {
      "epoch": 0.08675654957839203,
      "grad_norm": 0.7027180235626528,
      "learning_rate": 3.966336589706985e-06,
      "loss": 0.5998,
      "step": 962
    },
    {
      "epoch": 0.08684673310186229,
      "grad_norm": 2.078315439318011,
      "learning_rate": 3.966229769235948e-06,
      "loss": 0.8384,
      "step": 963
    },
    {
      "epoch": 0.08693691662533255,
      "grad_norm": 1.7713836338456517,
      "learning_rate": 3.966122780995345e-06,
      "loss": 0.9013,
      "step": 964
    },
    {
      "epoch": 0.08702710014880281,
      "grad_norm": 2.8089218266366043,
      "learning_rate": 3.966015624994306e-06,
      "loss": 0.8668,
      "step": 965
    },
    {
      "epoch": 0.08711728367227307,
      "grad_norm": 1.8935073445280688,
      "learning_rate": 3.9659083012419735e-06,
      "loss": 0.7949,
      "step": 966
    },
    {
      "epoch": 0.08720746719574334,
      "grad_norm": 1.6956979339808387,
      "learning_rate": 3.965800809747505e-06,
      "loss": 0.7556,
      "step": 967
    },
    {
      "epoch": 0.0872976507192136,
      "grad_norm": 1.8832960338893692,
      "learning_rate": 3.965693150520071e-06,
      "loss": 0.7873,
      "step": 968
    },
    {
      "epoch": 0.08738783424268386,
      "grad_norm": 2.402091259597269,
      "learning_rate": 3.96558532356886e-06,
      "loss": 0.7616,
      "step": 969
    },
    {
      "epoch": 0.08747801776615412,
      "grad_norm": 2.003310682049156,
      "learning_rate": 3.9654773289030704e-06,
      "loss": 0.8687,
      "step": 970
    },
    {
      "epoch": 0.08756820128962439,
      "grad_norm": 1.9267385088363977,
      "learning_rate": 3.9653691665319176e-06,
      "loss": 0.8416,
      "step": 971
    },
    {
      "epoch": 0.08765838481309465,
      "grad_norm": 1.8747969835363318,
      "learning_rate": 3.96526083646463e-06,
      "loss": 0.7502,
      "step": 972
    },
    {
      "epoch": 0.08774856833656491,
      "grad_norm": 1.6112577665746999,
      "learning_rate": 3.9651523387104526e-06,
      "loss": 0.7722,
      "step": 973
    },
    {
      "epoch": 0.08783875186003517,
      "grad_norm": 2.1219039128486825,
      "learning_rate": 3.965043673278641e-06,
      "loss": 0.8405,
      "step": 974
    },
    {
      "epoch": 0.08792893538350544,
      "grad_norm": 1.7110878516896644,
      "learning_rate": 3.964934840178469e-06,
      "loss": 0.8277,
      "step": 975
    },
    {
      "epoch": 0.0880191189069757,
      "grad_norm": 1.8843566134490493,
      "learning_rate": 3.964825839419221e-06,
      "loss": 0.8691,
      "step": 976
    },
    {
      "epoch": 0.08810930243044596,
      "grad_norm": 7.047617938681065,
      "learning_rate": 3.964716671010199e-06,
      "loss": 0.8121,
      "step": 977
    },
    {
      "epoch": 0.08819948595391622,
      "grad_norm": 1.489416455433491,
      "learning_rate": 3.9646073349607165e-06,
      "loss": 0.7854,
      "step": 978
    },
    {
      "epoch": 0.08828966947738648,
      "grad_norm": 2.144005771871171,
      "learning_rate": 3.964497831280105e-06,
      "loss": 0.9522,
      "step": 979
    },
    {
      "epoch": 0.08837985300085674,
      "grad_norm": 1.8322053025454488,
      "learning_rate": 3.964388159977705e-06,
      "loss": 0.8261,
      "step": 980
    },
    {
      "epoch": 0.088470036524327,
      "grad_norm": 1.6218157160219653,
      "learning_rate": 3.964278321062876e-06,
      "loss": 0.7302,
      "step": 981
    },
    {
      "epoch": 0.08856022004779726,
      "grad_norm": 1.8770977365475352,
      "learning_rate": 3.96416831454499e-06,
      "loss": 0.8458,
      "step": 982
    },
    {
      "epoch": 0.08865040357126754,
      "grad_norm": 0.7100886632737706,
      "learning_rate": 3.964058140433434e-06,
      "loss": 0.6312,
      "step": 983
    },
    {
      "epoch": 0.0887405870947378,
      "grad_norm": 1.9453305163181156,
      "learning_rate": 3.963947798737606e-06,
      "loss": 0.8846,
      "step": 984
    },
    {
      "epoch": 0.08883077061820806,
      "grad_norm": 1.9006130484318433,
      "learning_rate": 3.963837289466923e-06,
      "loss": 0.8502,
      "step": 985
    },
    {
      "epoch": 0.08892095414167832,
      "grad_norm": 2.1783041443896414,
      "learning_rate": 3.9637266126308145e-06,
      "loss": 0.7875,
      "step": 986
    },
    {
      "epoch": 0.08901113766514858,
      "grad_norm": 0.6949610732861418,
      "learning_rate": 3.963615768238724e-06,
      "loss": 0.6201,
      "step": 987
    },
    {
      "epoch": 0.08910132118861884,
      "grad_norm": 1.7052715668074365,
      "learning_rate": 3.963504756300107e-06,
      "loss": 0.7173,
      "step": 988
    },
    {
      "epoch": 0.0891915047120891,
      "grad_norm": 1.7143454555478723,
      "learning_rate": 3.96339357682444e-06,
      "loss": 0.8582,
      "step": 989
    },
    {
      "epoch": 0.08928168823555936,
      "grad_norm": 1.5773394300714356,
      "learning_rate": 3.963282229821206e-06,
      "loss": 0.8389,
      "step": 990
    },
    {
      "epoch": 0.08937187175902962,
      "grad_norm": 1.5388623412436382,
      "learning_rate": 3.963170715299906e-06,
      "loss": 0.8362,
      "step": 991
    },
    {
      "epoch": 0.08946205528249988,
      "grad_norm": 1.6571563543982093,
      "learning_rate": 3.963059033270056e-06,
      "loss": 0.7492,
      "step": 992
    },
    {
      "epoch": 0.08955223880597014,
      "grad_norm": 1.9620636521726307,
      "learning_rate": 3.9629471837411855e-06,
      "loss": 0.7682,
      "step": 993
    },
    {
      "epoch": 0.0896424223294404,
      "grad_norm": 2.3781736150877957,
      "learning_rate": 3.962835166722838e-06,
      "loss": 0.7686,
      "step": 994
    },
    {
      "epoch": 0.08973260585291068,
      "grad_norm": 1.6461010324890075,
      "learning_rate": 3.96272298222457e-06,
      "loss": 0.8295,
      "step": 995
    },
    {
      "epoch": 0.08982278937638094,
      "grad_norm": 4.095798591370701,
      "learning_rate": 3.962610630255956e-06,
      "loss": 0.9481,
      "step": 996
    },
    {
      "epoch": 0.0899129728998512,
      "grad_norm": 1.7907907725454517,
      "learning_rate": 3.96249811082658e-06,
      "loss": 0.7971,
      "step": 997
    },
    {
      "epoch": 0.09000315642332146,
      "grad_norm": 0.8050398934590617,
      "learning_rate": 3.962385423946046e-06,
      "loss": 0.6741,
      "step": 998
    },
    {
      "epoch": 0.09009333994679172,
      "grad_norm": 1.4901138196906083,
      "learning_rate": 3.962272569623966e-06,
      "loss": 0.8011,
      "step": 999
    },
    {
      "epoch": 0.09018352347026198,
      "grad_norm": 3.5058359809456245,
      "learning_rate": 3.9621595478699704e-06,
      "loss": 0.7143,
      "step": 1000
    },
    {
      "epoch": 0.09027370699373224,
      "grad_norm": 1.8166663855222898,
      "learning_rate": 3.962046358693703e-06,
      "loss": 0.8135,
      "step": 1001
    },
    {
      "epoch": 0.0903638905172025,
      "grad_norm": 6.6414989659266155,
      "learning_rate": 3.961933002104822e-06,
      "loss": 0.8946,
      "step": 1002
    },
    {
      "epoch": 0.09045407404067277,
      "grad_norm": 1.6985256842467165,
      "learning_rate": 3.961819478112999e-06,
      "loss": 0.7484,
      "step": 1003
    },
    {
      "epoch": 0.09054425756414303,
      "grad_norm": 2.266388652618795,
      "learning_rate": 3.961705786727921e-06,
      "loss": 0.8431,
      "step": 1004
    },
    {
      "epoch": 0.09063444108761329,
      "grad_norm": 1.6941088785403502,
      "learning_rate": 3.961591927959288e-06,
      "loss": 0.7503,
      "step": 1005
    },
    {
      "epoch": 0.09072462461108355,
      "grad_norm": 0.7210607700927277,
      "learning_rate": 3.961477901816816e-06,
      "loss": 0.5955,
      "step": 1006
    },
    {
      "epoch": 0.09081480813455382,
      "grad_norm": 1.7746512003157078,
      "learning_rate": 3.961363708310233e-06,
      "loss": 0.8871,
      "step": 1007
    },
    {
      "epoch": 0.09090499165802408,
      "grad_norm": 1.7748558657514995,
      "learning_rate": 3.961249347449286e-06,
      "loss": 0.8506,
      "step": 1008
    },
    {
      "epoch": 0.09099517518149434,
      "grad_norm": 1.6559431403575697,
      "learning_rate": 3.961134819243728e-06,
      "loss": 0.8092,
      "step": 1009
    },
    {
      "epoch": 0.0910853587049646,
      "grad_norm": 2.4866101269093814,
      "learning_rate": 3.961020123703335e-06,
      "loss": 0.8276,
      "step": 1010
    },
    {
      "epoch": 0.09117554222843487,
      "grad_norm": 1.6411891149077622,
      "learning_rate": 3.960905260837892e-06,
      "loss": 0.8323,
      "step": 1011
    },
    {
      "epoch": 0.09126572575190513,
      "grad_norm": 2.0490212872182743,
      "learning_rate": 3.960790230657199e-06,
      "loss": 0.7947,
      "step": 1012
    },
    {
      "epoch": 0.09135590927537539,
      "grad_norm": 2.2221009645228103,
      "learning_rate": 3.960675033171072e-06,
      "loss": 0.7707,
      "step": 1013
    },
    {
      "epoch": 0.09144609279884565,
      "grad_norm": 1.8291298476607043,
      "learning_rate": 3.960559668389341e-06,
      "loss": 0.7627,
      "step": 1014
    },
    {
      "epoch": 0.09153627632231591,
      "grad_norm": 1.7750736704837158,
      "learning_rate": 3.960444136321847e-06,
      "loss": 0.8238,
      "step": 1015
    },
    {
      "epoch": 0.09162645984578617,
      "grad_norm": 1.8568537311497877,
      "learning_rate": 3.960328436978451e-06,
      "loss": 0.9081,
      "step": 1016
    },
    {
      "epoch": 0.09171664336925643,
      "grad_norm": 0.6657227350396939,
      "learning_rate": 3.960212570369024e-06,
      "loss": 0.5959,
      "step": 1017
    },
    {
      "epoch": 0.09180682689272669,
      "grad_norm": 2.0548830120636605,
      "learning_rate": 3.9600965365034515e-06,
      "loss": 0.7541,
      "step": 1018
    },
    {
      "epoch": 0.09189701041619697,
      "grad_norm": 1.6437626943351908,
      "learning_rate": 3.959980335391634e-06,
      "loss": 0.8746,
      "step": 1019
    },
    {
      "epoch": 0.09198719393966723,
      "grad_norm": 1.7112494754720793,
      "learning_rate": 3.959863967043487e-06,
      "loss": 0.7586,
      "step": 1020
    },
    {
      "epoch": 0.09207737746313749,
      "grad_norm": 1.5437112687537997,
      "learning_rate": 3.9597474314689405e-06,
      "loss": 0.8462,
      "step": 1021
    },
    {
      "epoch": 0.09216756098660775,
      "grad_norm": 1.728413965716415,
      "learning_rate": 3.959630728677937e-06,
      "loss": 0.8378,
      "step": 1022
    },
    {
      "epoch": 0.09225774451007801,
      "grad_norm": 2.141603900370646,
      "learning_rate": 3.959513858680434e-06,
      "loss": 0.8002,
      "step": 1023
    },
    {
      "epoch": 0.09234792803354827,
      "grad_norm": 2.696242815979931,
      "learning_rate": 3.959396821486405e-06,
      "loss": 0.9106,
      "step": 1024
    },
    {
      "epoch": 0.09243811155701853,
      "grad_norm": 0.8238641780743866,
      "learning_rate": 3.959279617105835e-06,
      "loss": 0.6251,
      "step": 1025
    },
    {
      "epoch": 0.09252829508048879,
      "grad_norm": 2.0712681130334953,
      "learning_rate": 3.9591622455487235e-06,
      "loss": 0.8318,
      "step": 1026
    },
    {
      "epoch": 0.09261847860395905,
      "grad_norm": 1.888360993007942,
      "learning_rate": 3.959044706825087e-06,
      "loss": 0.8947,
      "step": 1027
    },
    {
      "epoch": 0.09270866212742931,
      "grad_norm": 1.8605473132536423,
      "learning_rate": 3.958927000944954e-06,
      "loss": 0.7246,
      "step": 1028
    },
    {
      "epoch": 0.09279884565089958,
      "grad_norm": 1.6673667232205904,
      "learning_rate": 3.958809127918368e-06,
      "loss": 0.906,
      "step": 1029
    },
    {
      "epoch": 0.09288902917436984,
      "grad_norm": 1.4799127264323546,
      "learning_rate": 3.958691087755387e-06,
      "loss": 0.7568,
      "step": 1030
    },
    {
      "epoch": 0.09297921269784011,
      "grad_norm": 2.1575393139206716,
      "learning_rate": 3.958572880466081e-06,
      "loss": 0.7012,
      "step": 1031
    },
    {
      "epoch": 0.09306939622131037,
      "grad_norm": 1.3884302457542237,
      "learning_rate": 3.9584545060605385e-06,
      "loss": 0.8055,
      "step": 1032
    },
    {
      "epoch": 0.09315957974478063,
      "grad_norm": 1.608154426599025,
      "learning_rate": 3.958335964548859e-06,
      "loss": 0.7819,
      "step": 1033
    },
    {
      "epoch": 0.0932497632682509,
      "grad_norm": 1.800693600852969,
      "learning_rate": 3.958217255941156e-06,
      "loss": 0.8743,
      "step": 1034
    },
    {
      "epoch": 0.09333994679172115,
      "grad_norm": 1.4190343630650988,
      "learning_rate": 3.95809838024756e-06,
      "loss": 0.7677,
      "step": 1035
    },
    {
      "epoch": 0.09343013031519141,
      "grad_norm": 1.6040110409373287,
      "learning_rate": 3.957979337478212e-06,
      "loss": 0.7809,
      "step": 1036
    },
    {
      "epoch": 0.09352031383866168,
      "grad_norm": 1.9285438020937555,
      "learning_rate": 3.957860127643272e-06,
      "loss": 0.843,
      "step": 1037
    },
    {
      "epoch": 0.09361049736213194,
      "grad_norm": 1.7251395321127445,
      "learning_rate": 3.95774075075291e-06,
      "loss": 0.8253,
      "step": 1038
    },
    {
      "epoch": 0.0937006808856022,
      "grad_norm": 1.5539715522558117,
      "learning_rate": 3.957621206817312e-06,
      "loss": 0.8764,
      "step": 1039
    },
    {
      "epoch": 0.09379086440907246,
      "grad_norm": 2.0952904571352304,
      "learning_rate": 3.957501495846679e-06,
      "loss": 0.8005,
      "step": 1040
    },
    {
      "epoch": 0.09388104793254272,
      "grad_norm": 0.6860211462579138,
      "learning_rate": 3.957381617851225e-06,
      "loss": 0.601,
      "step": 1041
    },
    {
      "epoch": 0.09397123145601298,
      "grad_norm": 2.8859090740660656,
      "learning_rate": 3.9572615728411776e-06,
      "loss": 0.7512,
      "step": 1042
    },
    {
      "epoch": 0.09406141497948325,
      "grad_norm": 2.5442555068200945,
      "learning_rate": 3.957141360826781e-06,
      "loss": 0.7081,
      "step": 1043
    },
    {
      "epoch": 0.09415159850295352,
      "grad_norm": 2.503601531770344,
      "learning_rate": 3.957020981818292e-06,
      "loss": 0.7997,
      "step": 1044
    },
    {
      "epoch": 0.09424178202642378,
      "grad_norm": 1.7698597718850047,
      "learning_rate": 3.956900435825982e-06,
      "loss": 0.8107,
      "step": 1045
    },
    {
      "epoch": 0.09433196554989404,
      "grad_norm": 1.4896631108330687,
      "learning_rate": 3.9567797228601364e-06,
      "loss": 0.7644,
      "step": 1046
    },
    {
      "epoch": 0.0944221490733643,
      "grad_norm": 1.8105102667044684,
      "learning_rate": 3.956658842931055e-06,
      "loss": 0.8905,
      "step": 1047
    },
    {
      "epoch": 0.09451233259683456,
      "grad_norm": 1.5413528672886763,
      "learning_rate": 3.956537796049052e-06,
      "loss": 0.8297,
      "step": 1048
    },
    {
      "epoch": 0.09460251612030482,
      "grad_norm": 1.488758321077424,
      "learning_rate": 3.956416582224457e-06,
      "loss": 0.6197,
      "step": 1049
    },
    {
      "epoch": 0.09469269964377508,
      "grad_norm": 1.890092128344963,
      "learning_rate": 3.956295201467611e-06,
      "loss": 0.8575,
      "step": 1050
    },
    {
      "epoch": 0.09478288316724534,
      "grad_norm": 1.4525055744372763,
      "learning_rate": 3.956173653788872e-06,
      "loss": 0.8479,
      "step": 1051
    },
    {
      "epoch": 0.0948730666907156,
      "grad_norm": 1.962940667117086,
      "learning_rate": 3.95605193919861e-06,
      "loss": 0.8613,
      "step": 1052
    },
    {
      "epoch": 0.09496325021418586,
      "grad_norm": 2.1854992026958353,
      "learning_rate": 3.955930057707211e-06,
      "loss": 0.8123,
      "step": 1053
    },
    {
      "epoch": 0.09505343373765612,
      "grad_norm": 2.2333528541546874,
      "learning_rate": 3.955808009325075e-06,
      "loss": 0.8312,
      "step": 1054
    },
    {
      "epoch": 0.0951436172611264,
      "grad_norm": 1.945446392798215,
      "learning_rate": 3.955685794062615e-06,
      "loss": 0.7978,
      "step": 1055
    },
    {
      "epoch": 0.09523380078459666,
      "grad_norm": 1.758914449523048,
      "learning_rate": 3.95556341193026e-06,
      "loss": 0.9204,
      "step": 1056
    },
    {
      "epoch": 0.09532398430806692,
      "grad_norm": 1.523092299256297,
      "learning_rate": 3.955440862938452e-06,
      "loss": 0.9152,
      "step": 1057
    },
    {
      "epoch": 0.09541416783153718,
      "grad_norm": 2.5158207029430155,
      "learning_rate": 3.955318147097647e-06,
      "loss": 0.8207,
      "step": 1058
    },
    {
      "epoch": 0.09550435135500744,
      "grad_norm": 2.051462173042515,
      "learning_rate": 3.955195264418316e-06,
      "loss": 0.7886,
      "step": 1059
    },
    {
      "epoch": 0.0955945348784777,
      "grad_norm": 3.0663454474963103,
      "learning_rate": 3.955072214910944e-06,
      "loss": 0.6775,
      "step": 1060
    },
    {
      "epoch": 0.09568471840194796,
      "grad_norm": 2.4565634956935316,
      "learning_rate": 3.954948998586032e-06,
      "loss": 0.7507,
      "step": 1061
    },
    {
      "epoch": 0.09577490192541822,
      "grad_norm": 1.5270841333960519,
      "learning_rate": 3.954825615454089e-06,
      "loss": 0.8284,
      "step": 1062
    },
    {
      "epoch": 0.09586508544888848,
      "grad_norm": 1.6330091002164595,
      "learning_rate": 3.954702065525649e-06,
      "loss": 0.8077,
      "step": 1063
    },
    {
      "epoch": 0.09595526897235875,
      "grad_norm": 2.2629520828696337,
      "learning_rate": 3.954578348811248e-06,
      "loss": 0.8486,
      "step": 1064
    },
    {
      "epoch": 0.096045452495829,
      "grad_norm": 2.7815044158126994,
      "learning_rate": 3.954454465321447e-06,
      "loss": 0.7969,
      "step": 1065
    },
    {
      "epoch": 0.09613563601929928,
      "grad_norm": 2.379053070598853,
      "learning_rate": 3.954330415066813e-06,
      "loss": 0.9008,
      "step": 1066
    },
    {
      "epoch": 0.09622581954276954,
      "grad_norm": 2.056779152170802,
      "learning_rate": 3.954206198057932e-06,
      "loss": 0.8091,
      "step": 1067
    },
    {
      "epoch": 0.0963160030662398,
      "grad_norm": 1.722872835068407,
      "learning_rate": 3.954081814305403e-06,
      "loss": 0.8795,
      "step": 1068
    },
    {
      "epoch": 0.09640618658971006,
      "grad_norm": 2.1000056402948113,
      "learning_rate": 3.953957263819839e-06,
      "loss": 0.7537,
      "step": 1069
    },
    {
      "epoch": 0.09649637011318032,
      "grad_norm": 1.7042844133185966,
      "learning_rate": 3.953832546611867e-06,
      "loss": 0.7558,
      "step": 1070
    },
    {
      "epoch": 0.09658655363665059,
      "grad_norm": 2.0424581148482392,
      "learning_rate": 3.953707662692129e-06,
      "loss": 0.8489,
      "step": 1071
    },
    {
      "epoch": 0.09667673716012085,
      "grad_norm": 1.746217386191025,
      "learning_rate": 3.95358261207128e-06,
      "loss": 0.753,
      "step": 1072
    },
    {
      "epoch": 0.0967669206835911,
      "grad_norm": 4.09176168130748,
      "learning_rate": 3.953457394759992e-06,
      "loss": 0.8086,
      "step": 1073
    },
    {
      "epoch": 0.09685710420706137,
      "grad_norm": 2.043417186644846,
      "learning_rate": 3.953332010768947e-06,
      "loss": 0.8328,
      "step": 1074
    },
    {
      "epoch": 0.09694728773053163,
      "grad_norm": 2.6416750293041757,
      "learning_rate": 3.9532064601088436e-06,
      "loss": 0.7869,
      "step": 1075
    },
    {
      "epoch": 0.09703747125400189,
      "grad_norm": 2.181637565043736,
      "learning_rate": 3.953080742790396e-06,
      "loss": 0.9729,
      "step": 1076
    },
    {
      "epoch": 0.09712765477747215,
      "grad_norm": 0.7671044904936222,
      "learning_rate": 3.95295485882433e-06,
      "loss": 0.6616,
      "step": 1077
    },
    {
      "epoch": 0.09721783830094242,
      "grad_norm": 0.6803872682683745,
      "learning_rate": 3.952828808221387e-06,
      "loss": 0.6291,
      "step": 1078
    },
    {
      "epoch": 0.09730802182441269,
      "grad_norm": 2.199422877647784,
      "learning_rate": 3.9527025909923225e-06,
      "loss": 0.801,
      "step": 1079
    },
    {
      "epoch": 0.09739820534788295,
      "grad_norm": 1.5430424539853158,
      "learning_rate": 3.952576207147906e-06,
      "loss": 0.905,
      "step": 1080
    },
    {
      "epoch": 0.09748838887135321,
      "grad_norm": 1.656465912186938,
      "learning_rate": 3.95244965669892e-06,
      "loss": 0.8561,
      "step": 1081
    },
    {
      "epoch": 0.09757857239482347,
      "grad_norm": 1.709533951811856,
      "learning_rate": 3.952322939656165e-06,
      "loss": 0.7756,
      "step": 1082
    },
    {
      "epoch": 0.09766875591829373,
      "grad_norm": 1.7201797780629824,
      "learning_rate": 3.952196056030451e-06,
      "loss": 0.812,
      "step": 1083
    },
    {
      "epoch": 0.09775893944176399,
      "grad_norm": 1.6818667446970121,
      "learning_rate": 3.952069005832605e-06,
      "loss": 0.7854,
      "step": 1084
    },
    {
      "epoch": 0.09784912296523425,
      "grad_norm": 1.6040855816581894,
      "learning_rate": 3.951941789073468e-06,
      "loss": 0.8463,
      "step": 1085
    },
    {
      "epoch": 0.09793930648870451,
      "grad_norm": 1.7380529360901908,
      "learning_rate": 3.9518144057638955e-06,
      "loss": 0.8463,
      "step": 1086
    },
    {
      "epoch": 0.09802949001217477,
      "grad_norm": 2.786375316678553,
      "learning_rate": 3.951686855914755e-06,
      "loss": 0.8152,
      "step": 1087
    },
    {
      "epoch": 0.09811967353564503,
      "grad_norm": 2.029356205709007,
      "learning_rate": 3.9515591395369305e-06,
      "loss": 0.8448,
      "step": 1088
    },
    {
      "epoch": 0.0982098570591153,
      "grad_norm": 2.2007918753348847,
      "learning_rate": 3.95143125664132e-06,
      "loss": 0.8041,
      "step": 1089
    },
    {
      "epoch": 0.09830004058258557,
      "grad_norm": 3.5742638423555846,
      "learning_rate": 3.951303207238833e-06,
      "loss": 0.8439,
      "step": 1090
    },
    {
      "epoch": 0.09839022410605583,
      "grad_norm": 1.5812799624291956,
      "learning_rate": 3.951174991340399e-06,
      "loss": 0.7854,
      "step": 1091
    },
    {
      "epoch": 0.09848040762952609,
      "grad_norm": 1.8181655731342419,
      "learning_rate": 3.9510466089569546e-06,
      "loss": 0.7776,
      "step": 1092
    },
    {
      "epoch": 0.09857059115299635,
      "grad_norm": 1.8788076521605768,
      "learning_rate": 3.950918060099456e-06,
      "loss": 0.7869,
      "step": 1093
    },
    {
      "epoch": 0.09866077467646661,
      "grad_norm": 1.734260211617017,
      "learning_rate": 3.950789344778871e-06,
      "loss": 0.8684,
      "step": 1094
    },
    {
      "epoch": 0.09875095819993687,
      "grad_norm": 2.1757433355510534,
      "learning_rate": 3.950660463006184e-06,
      "loss": 0.7605,
      "step": 1095
    },
    {
      "epoch": 0.09884114172340713,
      "grad_norm": 1.816295398898252,
      "learning_rate": 3.950531414792389e-06,
      "loss": 0.8884,
      "step": 1096
    },
    {
      "epoch": 0.0989313252468774,
      "grad_norm": 1.6626659855406547,
      "learning_rate": 3.950402200148498e-06,
      "loss": 0.8231,
      "step": 1097
    },
    {
      "epoch": 0.09902150877034765,
      "grad_norm": 1.6863552025058524,
      "learning_rate": 3.950272819085538e-06,
      "loss": 0.7957,
      "step": 1098
    },
    {
      "epoch": 0.09911169229381792,
      "grad_norm": 2.173140781313469,
      "learning_rate": 3.9501432716145474e-06,
      "loss": 0.9055,
      "step": 1099
    },
    {
      "epoch": 0.09920187581728818,
      "grad_norm": 1.8671021382120452,
      "learning_rate": 3.950013557746579e-06,
      "loss": 0.7146,
      "step": 1100
    },
    {
      "epoch": 0.09929205934075844,
      "grad_norm": 1.5111837697247534,
      "learning_rate": 3.949883677492703e-06,
      "loss": 0.8536,
      "step": 1101
    },
    {
      "epoch": 0.09938224286422871,
      "grad_norm": 1.8361021576761059,
      "learning_rate": 3.9497536308639994e-06,
      "loss": 0.8329,
      "step": 1102
    },
    {
      "epoch": 0.09947242638769897,
      "grad_norm": 1.644450206073485,
      "learning_rate": 3.949623417871565e-06,
      "loss": 0.8655,
      "step": 1103
    },
    {
      "epoch": 0.09956260991116923,
      "grad_norm": 2.0590197938446564,
      "learning_rate": 3.949493038526511e-06,
      "loss": 0.7993,
      "step": 1104
    },
    {
      "epoch": 0.0996527934346395,
      "grad_norm": 1.7464619020139425,
      "learning_rate": 3.949362492839961e-06,
      "loss": 0.7689,
      "step": 1105
    },
    {
      "epoch": 0.09974297695810976,
      "grad_norm": 1.7133434919261472,
      "learning_rate": 3.949231780823054e-06,
      "loss": 0.7812,
      "step": 1106
    },
    {
      "epoch": 0.09983316048158002,
      "grad_norm": 1.7538896657588703,
      "learning_rate": 3.949100902486945e-06,
      "loss": 0.758,
      "step": 1107
    },
    {
      "epoch": 0.09992334400505028,
      "grad_norm": 1.6564459774743314,
      "learning_rate": 3.948969857842799e-06,
      "loss": 0.7866,
      "step": 1108
    },
    {
      "epoch": 0.10001352752852054,
      "grad_norm": 1.61804559765518,
      "learning_rate": 3.948838646901798e-06,
      "loss": 0.8092,
      "step": 1109
    },
    {
      "epoch": 0.1001037110519908,
      "grad_norm": 1.9200325174359612,
      "learning_rate": 3.948707269675138e-06,
      "loss": 0.8117,
      "step": 1110
    },
    {
      "epoch": 0.10019389457546106,
      "grad_norm": 1.7070970326194337,
      "learning_rate": 3.948575726174028e-06,
      "loss": 0.8076,
      "step": 1111
    },
    {
      "epoch": 0.10028407809893132,
      "grad_norm": 1.8340772578857405,
      "learning_rate": 3.9484440164096935e-06,
      "loss": 0.8951,
      "step": 1112
    },
    {
      "epoch": 0.10037426162240158,
      "grad_norm": 1.6881336152903064,
      "learning_rate": 3.948312140393372e-06,
      "loss": 0.8386,
      "step": 1113
    },
    {
      "epoch": 0.10046444514587186,
      "grad_norm": 2.1471540466820507,
      "learning_rate": 3.948180098136316e-06,
      "loss": 0.786,
      "step": 1114
    },
    {
      "epoch": 0.10055462866934212,
      "grad_norm": 12.60181496279188,
      "learning_rate": 3.948047889649791e-06,
      "loss": 0.7869,
      "step": 1115
    },
    {
      "epoch": 0.10064481219281238,
      "grad_norm": 1.605243305053913,
      "learning_rate": 3.947915514945079e-06,
      "loss": 0.8014,
      "step": 1116
    },
    {
      "epoch": 0.10073499571628264,
      "grad_norm": 1.8496622885654919,
      "learning_rate": 3.947782974033474e-06,
      "loss": 0.8788,
      "step": 1117
    },
    {
      "epoch": 0.1008251792397529,
      "grad_norm": 2.2260072249974767,
      "learning_rate": 3.9476502669262866e-06,
      "loss": 0.7776,
      "step": 1118
    },
    {
      "epoch": 0.10091536276322316,
      "grad_norm": 1.9566792002032256,
      "learning_rate": 3.947517393634839e-06,
      "loss": 0.7926,
      "step": 1119
    },
    {
      "epoch": 0.10100554628669342,
      "grad_norm": 1.7301476311949133,
      "learning_rate": 3.947384354170469e-06,
      "loss": 0.8547,
      "step": 1120
    },
    {
      "epoch": 0.10109572981016368,
      "grad_norm": 1.58137306367766,
      "learning_rate": 3.947251148544528e-06,
      "loss": 0.8463,
      "step": 1121
    },
    {
      "epoch": 0.10118591333363394,
      "grad_norm": 1.8588116217059258,
      "learning_rate": 3.947117776768382e-06,
      "loss": 0.8473,
      "step": 1122
    },
    {
      "epoch": 0.1012760968571042,
      "grad_norm": 1.860930105045721,
      "learning_rate": 3.9469842388534105e-06,
      "loss": 0.7923,
      "step": 1123
    },
    {
      "epoch": 0.10136628038057446,
      "grad_norm": 1.8248671496387254,
      "learning_rate": 3.946850534811009e-06,
      "loss": 0.8241,
      "step": 1124
    },
    {
      "epoch": 0.10145646390404472,
      "grad_norm": 1.8215613276722777,
      "learning_rate": 3.946716664652585e-06,
      "loss": 0.8041,
      "step": 1125
    },
    {
      "epoch": 0.101546647427515,
      "grad_norm": 4.287347836477379,
      "learning_rate": 3.94658262838956e-06,
      "loss": 0.7416,
      "step": 1126
    },
    {
      "epoch": 0.10163683095098526,
      "grad_norm": 1.9838256676321824,
      "learning_rate": 3.946448426033373e-06,
      "loss": 0.737,
      "step": 1127
    },
    {
      "epoch": 0.10172701447445552,
      "grad_norm": 1.6521097653340218,
      "learning_rate": 3.946314057595473e-06,
      "loss": 0.8495,
      "step": 1128
    },
    {
      "epoch": 0.10181719799792578,
      "grad_norm": 1.9118424600436503,
      "learning_rate": 3.946179523087326e-06,
      "loss": 0.8766,
      "step": 1129
    },
    {
      "epoch": 0.10190738152139604,
      "grad_norm": 1.5299766072498264,
      "learning_rate": 3.9460448225204104e-06,
      "loss": 0.7288,
      "step": 1130
    },
    {
      "epoch": 0.1019975650448663,
      "grad_norm": 2.5206823277498533,
      "learning_rate": 3.945909955906221e-06,
      "loss": 0.8129,
      "step": 1131
    },
    {
      "epoch": 0.10208774856833656,
      "grad_norm": 1.618320676215216,
      "learning_rate": 3.945774923256264e-06,
      "loss": 0.8146,
      "step": 1132
    },
    {
      "epoch": 0.10217793209180683,
      "grad_norm": 2.869357667611798,
      "learning_rate": 3.945639724582062e-06,
      "loss": 0.8898,
      "step": 1133
    },
    {
      "epoch": 0.10226811561527709,
      "grad_norm": 2.2792692599548214,
      "learning_rate": 3.94550435989515e-06,
      "loss": 0.7877,
      "step": 1134
    },
    {
      "epoch": 0.10235829913874735,
      "grad_norm": 1.4660411772520026,
      "learning_rate": 3.945368829207079e-06,
      "loss": 0.839,
      "step": 1135
    },
    {
      "epoch": 0.10244848266221761,
      "grad_norm": 2.0488686207568483,
      "learning_rate": 3.945233132529414e-06,
      "loss": 0.7715,
      "step": 1136
    },
    {
      "epoch": 0.10253866618568787,
      "grad_norm": 2.082492111758969,
      "learning_rate": 3.9450972698737304e-06,
      "loss": 0.8159,
      "step": 1137
    },
    {
      "epoch": 0.10262884970915814,
      "grad_norm": 1.8789302324759063,
      "learning_rate": 3.944961241251623e-06,
      "loss": 0.793,
      "step": 1138
    },
    {
      "epoch": 0.1027190332326284,
      "grad_norm": 1.4568337314280622,
      "learning_rate": 3.9448250466746985e-06,
      "loss": 0.7828,
      "step": 1139
    },
    {
      "epoch": 0.10280921675609866,
      "grad_norm": 1.7334879763406008,
      "learning_rate": 3.944688686154578e-06,
      "loss": 0.8474,
      "step": 1140
    },
    {
      "epoch": 0.10289940027956893,
      "grad_norm": 2.047357749795026,
      "learning_rate": 3.944552159702894e-06,
      "loss": 0.7728,
      "step": 1141
    },
    {
      "epoch": 0.10298958380303919,
      "grad_norm": 1.7035522624945838,
      "learning_rate": 3.944415467331299e-06,
      "loss": 0.8026,
      "step": 1142
    },
    {
      "epoch": 0.10307976732650945,
      "grad_norm": 3.2848444551428035,
      "learning_rate": 3.944278609051455e-06,
      "loss": 0.7682,
      "step": 1143
    },
    {
      "epoch": 0.10316995084997971,
      "grad_norm": 2.449180395869449,
      "learning_rate": 3.944141584875039e-06,
      "loss": 0.7521,
      "step": 1144
    },
    {
      "epoch": 0.10326013437344997,
      "grad_norm": 1.7137031672316063,
      "learning_rate": 3.944004394813743e-06,
      "loss": 0.8241,
      "step": 1145
    },
    {
      "epoch": 0.10335031789692023,
      "grad_norm": 1.6202849427163304,
      "learning_rate": 3.943867038879273e-06,
      "loss": 0.8362,
      "step": 1146
    },
    {
      "epoch": 0.10344050142039049,
      "grad_norm": 3.044026926590757,
      "learning_rate": 3.943729517083349e-06,
      "loss": 0.8559,
      "step": 1147
    },
    {
      "epoch": 0.10353068494386075,
      "grad_norm": 1.5858930008381467,
      "learning_rate": 3.943591829437705e-06,
      "loss": 0.8502,
      "step": 1148
    },
    {
      "epoch": 0.10362086846733101,
      "grad_norm": 1.93039952782737,
      "learning_rate": 3.9434539759540895e-06,
      "loss": 0.7556,
      "step": 1149
    },
    {
      "epoch": 0.10371105199080129,
      "grad_norm": 1.6952073728250752,
      "learning_rate": 3.943315956644264e-06,
      "loss": 0.7497,
      "step": 1150
    },
    {
      "epoch": 0.10380123551427155,
      "grad_norm": 1.7546521748004247,
      "learning_rate": 3.943177771520006e-06,
      "loss": 0.883,
      "step": 1151
    },
    {
      "epoch": 0.10389141903774181,
      "grad_norm": 1.9236393095199769,
      "learning_rate": 3.9430394205931065e-06,
      "loss": 0.8602,
      "step": 1152
    },
    {
      "epoch": 0.10398160256121207,
      "grad_norm": 1.8011524537759607,
      "learning_rate": 3.942900903875369e-06,
      "loss": 0.6915,
      "step": 1153
    },
    {
      "epoch": 0.10407178608468233,
      "grad_norm": 2.1788063592674045,
      "learning_rate": 3.942762221378614e-06,
      "loss": 0.8178,
      "step": 1154
    },
    {
      "epoch": 0.10416196960815259,
      "grad_norm": 1.4476711513289668,
      "learning_rate": 3.942623373114673e-06,
      "loss": 0.8444,
      "step": 1155
    },
    {
      "epoch": 0.10425215313162285,
      "grad_norm": 2.0124311352433106,
      "learning_rate": 3.942484359095396e-06,
      "loss": 0.7998,
      "step": 1156
    },
    {
      "epoch": 0.10434233665509311,
      "grad_norm": 1.9529594046011036,
      "learning_rate": 3.942345179332642e-06,
      "loss": 0.7458,
      "step": 1157
    },
    {
      "epoch": 0.10443252017856337,
      "grad_norm": 1.705888169038851,
      "learning_rate": 3.942205833838287e-06,
      "loss": 0.7694,
      "step": 1158
    },
    {
      "epoch": 0.10452270370203363,
      "grad_norm": 1.8531559393167119,
      "learning_rate": 3.9420663226242204e-06,
      "loss": 0.9101,
      "step": 1159
    },
    {
      "epoch": 0.1046128872255039,
      "grad_norm": 2.396901586352816,
      "learning_rate": 3.941926645702348e-06,
      "loss": 0.8629,
      "step": 1160
    },
    {
      "epoch": 0.10470307074897416,
      "grad_norm": 3.0148428263271128,
      "learning_rate": 3.941786803084586e-06,
      "loss": 0.7974,
      "step": 1161
    },
    {
      "epoch": 0.10479325427244443,
      "grad_norm": 2.0678580075404995,
      "learning_rate": 3.941646794782867e-06,
      "loss": 0.7633,
      "step": 1162
    },
    {
      "epoch": 0.10488343779591469,
      "grad_norm": 2.0963818534460184,
      "learning_rate": 3.941506620809137e-06,
      "loss": 0.8259,
      "step": 1163
    },
    {
      "epoch": 0.10497362131938495,
      "grad_norm": 2.137991182878261,
      "learning_rate": 3.941366281175357e-06,
      "loss": 0.7961,
      "step": 1164
    },
    {
      "epoch": 0.10506380484285521,
      "grad_norm": 1.5030015478373646,
      "learning_rate": 3.941225775893502e-06,
      "loss": 0.7527,
      "step": 1165
    },
    {
      "epoch": 0.10515398836632547,
      "grad_norm": 1.4540044932761678,
      "learning_rate": 3.941085104975559e-06,
      "loss": 0.7874,
      "step": 1166
    },
    {
      "epoch": 0.10524417188979573,
      "grad_norm": 0.7624734514336108,
      "learning_rate": 3.9409442684335325e-06,
      "loss": 0.6244,
      "step": 1167
    },
    {
      "epoch": 0.105334355413266,
      "grad_norm": 1.944216745153747,
      "learning_rate": 3.940803266279438e-06,
      "loss": 0.8076,
      "step": 1168
    },
    {
      "epoch": 0.10542453893673626,
      "grad_norm": 3.6583046641924515,
      "learning_rate": 3.9406620985253076e-06,
      "loss": 0.7563,
      "step": 1169
    },
    {
      "epoch": 0.10551472246020652,
      "grad_norm": 2.6424689541728,
      "learning_rate": 3.940520765183187e-06,
      "loss": 0.784,
      "step": 1170
    },
    {
      "epoch": 0.10560490598367678,
      "grad_norm": 8.162750343117356,
      "learning_rate": 3.940379266265134e-06,
      "loss": 0.8576,
      "step": 1171
    },
    {
      "epoch": 0.10569508950714704,
      "grad_norm": 2.199526289310811,
      "learning_rate": 3.940237601783223e-06,
      "loss": 0.7797,
      "step": 1172
    },
    {
      "epoch": 0.1057852730306173,
      "grad_norm": 1.4932407037639448,
      "learning_rate": 3.940095771749542e-06,
      "loss": 0.8108,
      "step": 1173
    },
    {
      "epoch": 0.10587545655408757,
      "grad_norm": 1.934748264468221,
      "learning_rate": 3.939953776176192e-06,
      "loss": 0.8734,
      "step": 1174
    },
    {
      "epoch": 0.10596564007755783,
      "grad_norm": 1.2721147747050532,
      "learning_rate": 3.939811615075288e-06,
      "loss": 0.8376,
      "step": 1175
    },
    {
      "epoch": 0.1060558236010281,
      "grad_norm": 5.164858388815096,
      "learning_rate": 3.9396692884589616e-06,
      "loss": 0.8763,
      "step": 1176
    },
    {
      "epoch": 0.10614600712449836,
      "grad_norm": 2.1357235447645992,
      "learning_rate": 3.9395267963393565e-06,
      "loss": 0.7957,
      "step": 1177
    },
    {
      "epoch": 0.10623619064796862,
      "grad_norm": 2.88082511234922,
      "learning_rate": 3.939384138728631e-06,
      "loss": 0.8521,
      "step": 1178
    },
    {
      "epoch": 0.10632637417143888,
      "grad_norm": 1.5832118774142014,
      "learning_rate": 3.939241315638956e-06,
      "loss": 0.7988,
      "step": 1179
    },
    {
      "epoch": 0.10641655769490914,
      "grad_norm": 2.2402816275258526,
      "learning_rate": 3.93909832708252e-06,
      "loss": 0.7998,
      "step": 1180
    },
    {
      "epoch": 0.1065067412183794,
      "grad_norm": 1.4901295796315894,
      "learning_rate": 3.938955173071523e-06,
      "loss": 0.8188,
      "step": 1181
    },
    {
      "epoch": 0.10659692474184966,
      "grad_norm": 0.6150090607309072,
      "learning_rate": 3.938811853618179e-06,
      "loss": 0.6097,
      "step": 1182
    },
    {
      "epoch": 0.10668710826531992,
      "grad_norm": 1.922722668296897,
      "learning_rate": 3.938668368734717e-06,
      "loss": 0.852,
      "step": 1183
    },
    {
      "epoch": 0.10677729178879018,
      "grad_norm": 2.066298945332435,
      "learning_rate": 3.93852471843338e-06,
      "loss": 0.6919,
      "step": 1184
    },
    {
      "epoch": 0.10686747531226044,
      "grad_norm": 1.755149621481491,
      "learning_rate": 3.9383809027264254e-06,
      "loss": 0.7926,
      "step": 1185
    },
    {
      "epoch": 0.10695765883573072,
      "grad_norm": 2.2853918231273354,
      "learning_rate": 3.938236921626124e-06,
      "loss": 0.852,
      "step": 1186
    },
    {
      "epoch": 0.10704784235920098,
      "grad_norm": 2.660521270758688,
      "learning_rate": 3.938092775144761e-06,
      "loss": 0.8246,
      "step": 1187
    },
    {
      "epoch": 0.10713802588267124,
      "grad_norm": 1.5604094443601852,
      "learning_rate": 3.9379484632946355e-06,
      "loss": 0.734,
      "step": 1188
    },
    {
      "epoch": 0.1072282094061415,
      "grad_norm": 1.6495672756697282,
      "learning_rate": 3.937803986088062e-06,
      "loss": 0.8566,
      "step": 1189
    },
    {
      "epoch": 0.10731839292961176,
      "grad_norm": 2.080058868748896,
      "learning_rate": 3.937659343537367e-06,
      "loss": 0.7744,
      "step": 1190
    },
    {
      "epoch": 0.10740857645308202,
      "grad_norm": 1.757781236158608,
      "learning_rate": 3.937514535654893e-06,
      "loss": 0.8851,
      "step": 1191
    },
    {
      "epoch": 0.10749875997655228,
      "grad_norm": 3.2778136651899623,
      "learning_rate": 3.937369562452996e-06,
      "loss": 0.8223,
      "step": 1192
    },
    {
      "epoch": 0.10758894350002254,
      "grad_norm": 2.545549668045743,
      "learning_rate": 3.937224423944044e-06,
      "loss": 0.8715,
      "step": 1193
    },
    {
      "epoch": 0.1076791270234928,
      "grad_norm": 1.7342002711904192,
      "learning_rate": 3.937079120140423e-06,
      "loss": 0.748,
      "step": 1194
    },
    {
      "epoch": 0.10776931054696307,
      "grad_norm": 1.931732082998214,
      "learning_rate": 3.936933651054531e-06,
      "loss": 0.8483,
      "step": 1195
    },
    {
      "epoch": 0.10785949407043333,
      "grad_norm": 1.6202422697909664,
      "learning_rate": 3.936788016698779e-06,
      "loss": 0.9269,
      "step": 1196
    },
    {
      "epoch": 0.1079496775939036,
      "grad_norm": 2.557364737922809,
      "learning_rate": 3.936642217085594e-06,
      "loss": 0.846,
      "step": 1197
    },
    {
      "epoch": 0.10803986111737386,
      "grad_norm": 2.011073807229307,
      "learning_rate": 3.936496252227417e-06,
      "loss": 0.7261,
      "step": 1198
    },
    {
      "epoch": 0.10813004464084412,
      "grad_norm": 5.085533583111134,
      "learning_rate": 3.936350122136703e-06,
      "loss": 0.798,
      "step": 1199
    },
    {
      "epoch": 0.10822022816431438,
      "grad_norm": 0.6954516796587367,
      "learning_rate": 3.936203826825919e-06,
      "loss": 0.6312,
      "step": 1200
    },
    {
      "epoch": 0.10831041168778464,
      "grad_norm": 1.9861171013920338,
      "learning_rate": 3.9360573663075475e-06,
      "loss": 0.7719,
      "step": 1201
    },
    {
      "epoch": 0.1084005952112549,
      "grad_norm": 2.413190171735779,
      "learning_rate": 3.935910740594087e-06,
      "loss": 0.8443,
      "step": 1202
    },
    {
      "epoch": 0.10849077873472517,
      "grad_norm": 2.167095386442514,
      "learning_rate": 3.935763949698047e-06,
      "loss": 0.8214,
      "step": 1203
    },
    {
      "epoch": 0.10858096225819543,
      "grad_norm": 1.716766461308037,
      "learning_rate": 3.935616993631954e-06,
      "loss": 0.8814,
      "step": 1204
    },
    {
      "epoch": 0.10867114578166569,
      "grad_norm": 4.52158793584155,
      "learning_rate": 3.935469872408345e-06,
      "loss": 0.8057,
      "step": 1205
    },
    {
      "epoch": 0.10876132930513595,
      "grad_norm": 2.1389339478055303,
      "learning_rate": 3.935322586039776e-06,
      "loss": 0.832,
      "step": 1206
    },
    {
      "epoch": 0.10885151282860621,
      "grad_norm": 1.5377067364898578,
      "learning_rate": 3.935175134538811e-06,
      "loss": 0.7436,
      "step": 1207
    },
    {
      "epoch": 0.10894169635207647,
      "grad_norm": 2.2107961225295845,
      "learning_rate": 3.935027517918034e-06,
      "loss": 0.7566,
      "step": 1208
    },
    {
      "epoch": 0.10903187987554674,
      "grad_norm": 3.0338072067795796,
      "learning_rate": 3.93487973619004e-06,
      "loss": 0.8454,
      "step": 1209
    },
    {
      "epoch": 0.109122063399017,
      "grad_norm": 4.9065938452486,
      "learning_rate": 3.934731789367438e-06,
      "loss": 0.8277,
      "step": 1210
    },
    {
      "epoch": 0.10921224692248727,
      "grad_norm": 1.0232223401053406,
      "learning_rate": 3.9345836774628505e-06,
      "loss": 0.6088,
      "step": 1211
    },
    {
      "epoch": 0.10930243044595753,
      "grad_norm": 1.5553231018330502,
      "learning_rate": 3.934435400488917e-06,
      "loss": 0.8014,
      "step": 1212
    },
    {
      "epoch": 0.10939261396942779,
      "grad_norm": 1.622464402547424,
      "learning_rate": 3.934286958458289e-06,
      "loss": 0.8433,
      "step": 1213
    },
    {
      "epoch": 0.10948279749289805,
      "grad_norm": 1.6065308707506765,
      "learning_rate": 3.934138351383632e-06,
      "loss": 0.8724,
      "step": 1214
    },
    {
      "epoch": 0.10957298101636831,
      "grad_norm": 3.5324139322539394,
      "learning_rate": 3.933989579277626e-06,
      "loss": 0.7473,
      "step": 1215
    },
    {
      "epoch": 0.10966316453983857,
      "grad_norm": 1.8002604561030466,
      "learning_rate": 3.933840642152966e-06,
      "loss": 0.8071,
      "step": 1216
    },
    {
      "epoch": 0.10975334806330883,
      "grad_norm": 2.2195103376800485,
      "learning_rate": 3.933691540022359e-06,
      "loss": 0.8399,
      "step": 1217
    },
    {
      "epoch": 0.10984353158677909,
      "grad_norm": 1.7733536082855927,
      "learning_rate": 3.933542272898527e-06,
      "loss": 0.7824,
      "step": 1218
    },
    {
      "epoch": 0.10993371511024935,
      "grad_norm": 0.7526594535469616,
      "learning_rate": 3.933392840794207e-06,
      "loss": 0.567,
      "step": 1219
    },
    {
      "epoch": 0.11002389863371961,
      "grad_norm": 1.7884463119489713,
      "learning_rate": 3.93324324372215e-06,
      "loss": 0.9197,
      "step": 1220
    },
    {
      "epoch": 0.11011408215718989,
      "grad_norm": 1.699932781461319,
      "learning_rate": 3.9330934816951185e-06,
      "loss": 0.8382,
      "step": 1221
    },
    {
      "epoch": 0.11020426568066015,
      "grad_norm": 1.7797666697856787,
      "learning_rate": 3.932943554725893e-06,
      "loss": 0.8445,
      "step": 1222
    },
    {
      "epoch": 0.11029444920413041,
      "grad_norm": 0.7070543696380228,
      "learning_rate": 3.932793462827265e-06,
      "loss": 0.641,
      "step": 1223
    },
    {
      "epoch": 0.11038463272760067,
      "grad_norm": 1.8073221766583536,
      "learning_rate": 3.932643206012041e-06,
      "loss": 0.8711,
      "step": 1224
    },
    {
      "epoch": 0.11047481625107093,
      "grad_norm": 2.115730037363798,
      "learning_rate": 3.932492784293043e-06,
      "loss": 0.7381,
      "step": 1225
    },
    {
      "epoch": 0.11056499977454119,
      "grad_norm": 1.5144156123945,
      "learning_rate": 3.932342197683104e-06,
      "loss": 0.8032,
      "step": 1226
    },
    {
      "epoch": 0.11065518329801145,
      "grad_norm": 2.056875709943901,
      "learning_rate": 3.932191446195075e-06,
      "loss": 0.6854,
      "step": 1227
    },
    {
      "epoch": 0.11074536682148171,
      "grad_norm": 2.428398845454799,
      "learning_rate": 3.9320405298418175e-06,
      "loss": 0.8129,
      "step": 1228
    },
    {
      "epoch": 0.11083555034495197,
      "grad_norm": 1.6947060480356637,
      "learning_rate": 3.9318894486362076e-06,
      "loss": 0.7381,
      "step": 1229
    },
    {
      "epoch": 0.11092573386842224,
      "grad_norm": 1.9142695723648127,
      "learning_rate": 3.9317382025911395e-06,
      "loss": 0.8857,
      "step": 1230
    },
    {
      "epoch": 0.1110159173918925,
      "grad_norm": 1.8092078861418082,
      "learning_rate": 3.9315867917195145e-06,
      "loss": 0.9117,
      "step": 1231
    },
    {
      "epoch": 0.11110610091536276,
      "grad_norm": 1.7184998763758348,
      "learning_rate": 3.931435216034256e-06,
      "loss": 0.8121,
      "step": 1232
    },
    {
      "epoch": 0.11119628443883303,
      "grad_norm": 3.2721052349301103,
      "learning_rate": 3.931283475548293e-06,
      "loss": 0.8253,
      "step": 1233
    },
    {
      "epoch": 0.11128646796230329,
      "grad_norm": 1.352898615340822,
      "learning_rate": 3.931131570274576e-06,
      "loss": 0.8169,
      "step": 1234
    },
    {
      "epoch": 0.11137665148577355,
      "grad_norm": 1.5574705431557452,
      "learning_rate": 3.930979500226065e-06,
      "loss": 0.847,
      "step": 1235
    },
    {
      "epoch": 0.11146683500924381,
      "grad_norm": 2.8065173627084223,
      "learning_rate": 3.930827265415736e-06,
      "loss": 0.65,
      "step": 1236
    },
    {
      "epoch": 0.11155701853271408,
      "grad_norm": 1.7343415265676416,
      "learning_rate": 3.930674865856578e-06,
      "loss": 0.8236,
      "step": 1237
    },
    {
      "epoch": 0.11164720205618434,
      "grad_norm": 1.9646313288931916,
      "learning_rate": 3.930522301561595e-06,
      "loss": 0.8486,
      "step": 1238
    },
    {
      "epoch": 0.1117373855796546,
      "grad_norm": 2.4088857507783,
      "learning_rate": 3.930369572543804e-06,
      "loss": 0.8575,
      "step": 1239
    },
    {
      "epoch": 0.11182756910312486,
      "grad_norm": 1.635268194501253,
      "learning_rate": 3.930216678816237e-06,
      "loss": 0.7594,
      "step": 1240
    },
    {
      "epoch": 0.11191775262659512,
      "grad_norm": 0.6841361591454287,
      "learning_rate": 3.930063620391941e-06,
      "loss": 0.6152,
      "step": 1241
    },
    {
      "epoch": 0.11200793615006538,
      "grad_norm": 1.8249518740853583,
      "learning_rate": 3.9299103972839735e-06,
      "loss": 0.8247,
      "step": 1242
    },
    {
      "epoch": 0.11209811967353564,
      "grad_norm": 0.654133448901395,
      "learning_rate": 3.92975700950541e-06,
      "loss": 0.5802,
      "step": 1243
    },
    {
      "epoch": 0.1121883031970059,
      "grad_norm": 1.8952893185750657,
      "learning_rate": 3.929603457069338e-06,
      "loss": 0.8419,
      "step": 1244
    },
    {
      "epoch": 0.11227848672047618,
      "grad_norm": 1.6489425536328652,
      "learning_rate": 3.929449739988859e-06,
      "loss": 0.7771,
      "step": 1245
    },
    {
      "epoch": 0.11236867024394644,
      "grad_norm": 1.327081068067948,
      "learning_rate": 3.929295858277089e-06,
      "loss": 0.81,
      "step": 1246
    },
    {
      "epoch": 0.1124588537674167,
      "grad_norm": 1.7679343186083951,
      "learning_rate": 3.9291418119471585e-06,
      "loss": 0.8103,
      "step": 1247
    },
    {
      "epoch": 0.11254903729088696,
      "grad_norm": 2.304006805760875,
      "learning_rate": 3.928987601012212e-06,
      "loss": 0.8425,
      "step": 1248
    },
    {
      "epoch": 0.11263922081435722,
      "grad_norm": 1.8496378621502787,
      "learning_rate": 3.928833225485407e-06,
      "loss": 0.8017,
      "step": 1249
    },
    {
      "epoch": 0.11272940433782748,
      "grad_norm": 1.7314402176232853,
      "learning_rate": 3.928678685379915e-06,
      "loss": 0.8035,
      "step": 1250
    },
    {
      "epoch": 0.11281958786129774,
      "grad_norm": 1.6686715782683559,
      "learning_rate": 3.928523980708924e-06,
      "loss": 0.8562,
      "step": 1251
    },
    {
      "epoch": 0.112909771384768,
      "grad_norm": 2.170087998535248,
      "learning_rate": 3.928369111485632e-06,
      "loss": 0.7648,
      "step": 1252
    },
    {
      "epoch": 0.11299995490823826,
      "grad_norm": 1.9537821770359522,
      "learning_rate": 3.928214077723255e-06,
      "loss": 0.7995,
      "step": 1253
    },
    {
      "epoch": 0.11309013843170852,
      "grad_norm": 1.8641644509918638,
      "learning_rate": 3.928058879435021e-06,
      "loss": 0.7971,
      "step": 1254
    },
    {
      "epoch": 0.11318032195517878,
      "grad_norm": 1.7154414844856516,
      "learning_rate": 3.9279035166341725e-06,
      "loss": 0.795,
      "step": 1255
    },
    {
      "epoch": 0.11327050547864904,
      "grad_norm": 12.151428364752473,
      "learning_rate": 3.927747989333965e-06,
      "loss": 0.794,
      "step": 1256
    },
    {
      "epoch": 0.11336068900211932,
      "grad_norm": 2.311780663035396,
      "learning_rate": 3.927592297547669e-06,
      "loss": 0.7842,
      "step": 1257
    },
    {
      "epoch": 0.11345087252558958,
      "grad_norm": 1.878153057174443,
      "learning_rate": 3.927436441288571e-06,
      "loss": 0.8376,
      "step": 1258
    },
    {
      "epoch": 0.11354105604905984,
      "grad_norm": 2.220719121843494,
      "learning_rate": 3.927280420569968e-06,
      "loss": 0.826,
      "step": 1259
    },
    {
      "epoch": 0.1136312395725301,
      "grad_norm": 1.8236060083451373,
      "learning_rate": 3.927124235405171e-06,
      "loss": 0.8384,
      "step": 1260
    },
    {
      "epoch": 0.11372142309600036,
      "grad_norm": 1.9168336698779853,
      "learning_rate": 3.92696788580751e-06,
      "loss": 0.6738,
      "step": 1261
    },
    {
      "epoch": 0.11381160661947062,
      "grad_norm": 1.9619456583443862,
      "learning_rate": 3.9268113717903225e-06,
      "loss": 0.651,
      "step": 1262
    },
    {
      "epoch": 0.11390179014294088,
      "grad_norm": 1.9431846041661947,
      "learning_rate": 3.926654693366965e-06,
      "loss": 0.8827,
      "step": 1263
    },
    {
      "epoch": 0.11399197366641114,
      "grad_norm": 2.346361358552524,
      "learning_rate": 3.926497850550805e-06,
      "loss": 0.8225,
      "step": 1264
    },
    {
      "epoch": 0.1140821571898814,
      "grad_norm": 1.9270844605588624,
      "learning_rate": 3.926340843355226e-06,
      "loss": 0.8116,
      "step": 1265
    },
    {
      "epoch": 0.11417234071335167,
      "grad_norm": 1.905730379918635,
      "learning_rate": 3.926183671793625e-06,
      "loss": 0.8235,
      "step": 1266
    },
    {
      "epoch": 0.11426252423682193,
      "grad_norm": 1.5562212102593995,
      "learning_rate": 3.926026335879412e-06,
      "loss": 0.7794,
      "step": 1267
    },
    {
      "epoch": 0.11435270776029219,
      "grad_norm": 0.6332168464656507,
      "learning_rate": 3.925868835626012e-06,
      "loss": 0.5517,
      "step": 1268
    },
    {
      "epoch": 0.11444289128376246,
      "grad_norm": 2.008914748855751,
      "learning_rate": 3.925711171046864e-06,
      "loss": 0.821,
      "step": 1269
    },
    {
      "epoch": 0.11453307480723272,
      "grad_norm": 1.4501718024714692,
      "learning_rate": 3.925553342155421e-06,
      "loss": 0.7659,
      "step": 1270
    },
    {
      "epoch": 0.11462325833070298,
      "grad_norm": 0.6937499115058912,
      "learning_rate": 3.9253953489651485e-06,
      "loss": 0.6257,
      "step": 1271
    },
    {
      "epoch": 0.11471344185417325,
      "grad_norm": 0.7380979199331464,
      "learning_rate": 3.925237191489529e-06,
      "loss": 0.601,
      "step": 1272
    },
    {
      "epoch": 0.1148036253776435,
      "grad_norm": 1.5977257151530126,
      "learning_rate": 3.925078869742056e-06,
      "loss": 0.7789,
      "step": 1273
    },
    {
      "epoch": 0.11489380890111377,
      "grad_norm": 1.6639194337478556,
      "learning_rate": 3.92492038373624e-06,
      "loss": 0.7461,
      "step": 1274
    },
    {
      "epoch": 0.11498399242458403,
      "grad_norm": 2.182736141368373,
      "learning_rate": 3.924761733485602e-06,
      "loss": 0.8671,
      "step": 1275
    },
    {
      "epoch": 0.11507417594805429,
      "grad_norm": 1.4887661682070903,
      "learning_rate": 3.92460291900368e-06,
      "loss": 0.7994,
      "step": 1276
    },
    {
      "epoch": 0.11516435947152455,
      "grad_norm": 2.0714132585570244,
      "learning_rate": 3.924443940304025e-06,
      "loss": 0.8303,
      "step": 1277
    },
    {
      "epoch": 0.11525454299499481,
      "grad_norm": 1.9127354595373054,
      "learning_rate": 3.924284797400202e-06,
      "loss": 0.8649,
      "step": 1278
    },
    {
      "epoch": 0.11534472651846507,
      "grad_norm": 2.3049707271016286,
      "learning_rate": 3.924125490305789e-06,
      "loss": 0.8362,
      "step": 1279
    },
    {
      "epoch": 0.11543491004193533,
      "grad_norm": 2.0652291707853028,
      "learning_rate": 3.923966019034381e-06,
      "loss": 0.7904,
      "step": 1280
    },
    {
      "epoch": 0.1155250935654056,
      "grad_norm": 1.8572237408494512,
      "learning_rate": 3.923806383599583e-06,
      "loss": 0.8221,
      "step": 1281
    },
    {
      "epoch": 0.11561527708887587,
      "grad_norm": 1.65427597735288,
      "learning_rate": 3.923646584015017e-06,
      "loss": 0.778,
      "step": 1282
    },
    {
      "epoch": 0.11570546061234613,
      "grad_norm": 2.104845057242062,
      "learning_rate": 3.923486620294316e-06,
      "loss": 0.8136,
      "step": 1283
    },
    {
      "epoch": 0.11579564413581639,
      "grad_norm": 1.8671336146284465,
      "learning_rate": 3.923326492451132e-06,
      "loss": 0.8257,
      "step": 1284
    },
    {
      "epoch": 0.11588582765928665,
      "grad_norm": 2.1911890802875247,
      "learning_rate": 3.923166200499125e-06,
      "loss": 0.7877,
      "step": 1285
    },
    {
      "epoch": 0.11597601118275691,
      "grad_norm": 1.6052405573411679,
      "learning_rate": 3.923005744451975e-06,
      "loss": 0.7763,
      "step": 1286
    },
    {
      "epoch": 0.11606619470622717,
      "grad_norm": 2.097298191987902,
      "learning_rate": 3.9228451243233715e-06,
      "loss": 0.7585,
      "step": 1287
    },
    {
      "epoch": 0.11615637822969743,
      "grad_norm": 2.2373497640796782,
      "learning_rate": 3.9226843401270195e-06,
      "loss": 0.8383,
      "step": 1288
    },
    {
      "epoch": 0.1162465617531677,
      "grad_norm": 1.7945916969040066,
      "learning_rate": 3.9225233918766376e-06,
      "loss": 0.8136,
      "step": 1289
    },
    {
      "epoch": 0.11633674527663795,
      "grad_norm": 1.5547368794176046,
      "learning_rate": 3.92236227958596e-06,
      "loss": 0.8225,
      "step": 1290
    },
    {
      "epoch": 0.11642692880010821,
      "grad_norm": 1.9531301269463959,
      "learning_rate": 3.922201003268731e-06,
      "loss": 0.8307,
      "step": 1291
    },
    {
      "epoch": 0.11651711232357848,
      "grad_norm": 0.6919338131712399,
      "learning_rate": 3.922039562938715e-06,
      "loss": 0.6274,
      "step": 1292
    },
    {
      "epoch": 0.11660729584704875,
      "grad_norm": 1.6609191577720377,
      "learning_rate": 3.921877958609685e-06,
      "loss": 0.8572,
      "step": 1293
    },
    {
      "epoch": 0.11669747937051901,
      "grad_norm": 0.7834649540184359,
      "learning_rate": 3.921716190295431e-06,
      "loss": 0.6919,
      "step": 1294
    },
    {
      "epoch": 0.11678766289398927,
      "grad_norm": 1.820819767683927,
      "learning_rate": 3.921554258009755e-06,
      "loss": 0.7677,
      "step": 1295
    },
    {
      "epoch": 0.11687784641745953,
      "grad_norm": 2.2510190351538024,
      "learning_rate": 3.921392161766474e-06,
      "loss": 0.7955,
      "step": 1296
    },
    {
      "epoch": 0.1169680299409298,
      "grad_norm": 2.151171467138376,
      "learning_rate": 3.92122990157942e-06,
      "loss": 0.792,
      "step": 1297
    },
    {
      "epoch": 0.11705821346440005,
      "grad_norm": 1.8695852932609054,
      "learning_rate": 3.921067477462437e-06,
      "loss": 0.8542,
      "step": 1298
    },
    {
      "epoch": 0.11714839698787032,
      "grad_norm": 2.2264716949685424,
      "learning_rate": 3.920904889429385e-06,
      "loss": 0.866,
      "step": 1299
    },
    {
      "epoch": 0.11723858051134058,
      "grad_norm": 2.3372244973435192,
      "learning_rate": 3.920742137494135e-06,
      "loss": 0.8518,
      "step": 1300
    },
    {
      "epoch": 0.11732876403481084,
      "grad_norm": 1.738068093935826,
      "learning_rate": 3.920579221670575e-06,
      "loss": 0.8681,
      "step": 1301
    },
    {
      "epoch": 0.1174189475582811,
      "grad_norm": 1.762171042294946,
      "learning_rate": 3.920416141972606e-06,
      "loss": 0.8142,
      "step": 1302
    },
    {
      "epoch": 0.11750913108175136,
      "grad_norm": 1.506508614273334,
      "learning_rate": 3.920252898414143e-06,
      "loss": 0.8041,
      "step": 1303
    },
    {
      "epoch": 0.11759931460522162,
      "grad_norm": 2.351370566003107,
      "learning_rate": 3.920089491009114e-06,
      "loss": 0.8233,
      "step": 1304
    },
    {
      "epoch": 0.1176894981286919,
      "grad_norm": 1.8303031009577984,
      "learning_rate": 3.919925919771463e-06,
      "loss": 0.8904,
      "step": 1305
    },
    {
      "epoch": 0.11777968165216215,
      "grad_norm": 1.8846780232839733,
      "learning_rate": 3.919762184715146e-06,
      "loss": 0.7248,
      "step": 1306
    },
    {
      "epoch": 0.11786986517563242,
      "grad_norm": 2.1567456325399674,
      "learning_rate": 3.919598285854134e-06,
      "loss": 0.8067,
      "step": 1307
    },
    {
      "epoch": 0.11796004869910268,
      "grad_norm": 1.9170608391617006,
      "learning_rate": 3.919434223202411e-06,
      "loss": 0.8636,
      "step": 1308
    },
    {
      "epoch": 0.11805023222257294,
      "grad_norm": 1.8961647777983295,
      "learning_rate": 3.919269996773977e-06,
      "loss": 0.8614,
      "step": 1309
    },
    {
      "epoch": 0.1181404157460432,
      "grad_norm": 1.4804040094477469,
      "learning_rate": 3.919105606582844e-06,
      "loss": 0.8242,
      "step": 1310
    },
    {
      "epoch": 0.11823059926951346,
      "grad_norm": 1.806022645573882,
      "learning_rate": 3.918941052643039e-06,
      "loss": 0.7847,
      "step": 1311
    },
    {
      "epoch": 0.11832078279298372,
      "grad_norm": 1.6891635889226988,
      "learning_rate": 3.918776334968602e-06,
      "loss": 0.821,
      "step": 1312
    },
    {
      "epoch": 0.11841096631645398,
      "grad_norm": 1.9454683295039539,
      "learning_rate": 3.918611453573589e-06,
      "loss": 0.9191,
      "step": 1313
    },
    {
      "epoch": 0.11850114983992424,
      "grad_norm": 2.009582333228031,
      "learning_rate": 3.918446408472066e-06,
      "loss": 0.7448,
      "step": 1314
    },
    {
      "epoch": 0.1185913333633945,
      "grad_norm": 1.5979166354992136,
      "learning_rate": 3.918281199678119e-06,
      "loss": 0.8487,
      "step": 1315
    },
    {
      "epoch": 0.11868151688686476,
      "grad_norm": 0.7330946210467959,
      "learning_rate": 3.9181158272058414e-06,
      "loss": 0.6094,
      "step": 1316
    },
    {
      "epoch": 0.11877170041033504,
      "grad_norm": 1.5890086837252797,
      "learning_rate": 3.9179502910693455e-06,
      "loss": 0.8375,
      "step": 1317
    },
    {
      "epoch": 0.1188618839338053,
      "grad_norm": 1.752158264602047,
      "learning_rate": 3.917784591282756e-06,
      "loss": 0.7672,
      "step": 1318
    },
    {
      "epoch": 0.11895206745727556,
      "grad_norm": 1.7641386003864012,
      "learning_rate": 3.9176187278602105e-06,
      "loss": 0.8323,
      "step": 1319
    },
    {
      "epoch": 0.11904225098074582,
      "grad_norm": 1.51256385857574,
      "learning_rate": 3.9174527008158606e-06,
      "loss": 0.7514,
      "step": 1320
    },
    {
      "epoch": 0.11913243450421608,
      "grad_norm": 1.5129963998140312,
      "learning_rate": 3.917286510163874e-06,
      "loss": 0.733,
      "step": 1321
    },
    {
      "epoch": 0.11922261802768634,
      "grad_norm": 1.316759826084641,
      "learning_rate": 3.917120155918431e-06,
      "loss": 0.7797,
      "step": 1322
    },
    {
      "epoch": 0.1193128015511566,
      "grad_norm": 0.7407789818729613,
      "learning_rate": 3.916953638093725e-06,
      "loss": 0.6411,
      "step": 1323
    },
    {
      "epoch": 0.11940298507462686,
      "grad_norm": 1.9296382635221312,
      "learning_rate": 3.916786956703964e-06,
      "loss": 0.8027,
      "step": 1324
    },
    {
      "epoch": 0.11949316859809712,
      "grad_norm": 3.8251576659176307,
      "learning_rate": 3.916620111763372e-06,
      "loss": 0.8077,
      "step": 1325
    },
    {
      "epoch": 0.11958335212156739,
      "grad_norm": 1.6842952957121144,
      "learning_rate": 3.916453103286183e-06,
      "loss": 0.8699,
      "step": 1326
    },
    {
      "epoch": 0.11967353564503765,
      "grad_norm": 2.3803307789054013,
      "learning_rate": 3.916285931286648e-06,
      "loss": 0.8544,
      "step": 1327
    },
    {
      "epoch": 0.11976371916850792,
      "grad_norm": 1.724605918872174,
      "learning_rate": 3.916118595779031e-06,
      "loss": 0.7558,
      "step": 1328
    },
    {
      "epoch": 0.11985390269197818,
      "grad_norm": 1.7433076003359356,
      "learning_rate": 3.915951096777611e-06,
      "loss": 0.7867,
      "step": 1329
    },
    {
      "epoch": 0.11994408621544844,
      "grad_norm": 1.7584441003968767,
      "learning_rate": 3.915783434296678e-06,
      "loss": 0.7634,
      "step": 1330
    },
    {
      "epoch": 0.1200342697389187,
      "grad_norm": 2.0520821069497512,
      "learning_rate": 3.91561560835054e-06,
      "loss": 0.7908,
      "step": 1331
    },
    {
      "epoch": 0.12012445326238896,
      "grad_norm": 1.599295001874651,
      "learning_rate": 3.915447618953515e-06,
      "loss": 0.8261,
      "step": 1332
    },
    {
      "epoch": 0.12021463678585922,
      "grad_norm": 2.231341884227758,
      "learning_rate": 3.915279466119937e-06,
      "loss": 0.7653,
      "step": 1333
    },
    {
      "epoch": 0.12030482030932949,
      "grad_norm": 0.725687751455233,
      "learning_rate": 3.9151111498641546e-06,
      "loss": 0.6257,
      "step": 1334
    },
    {
      "epoch": 0.12039500383279975,
      "grad_norm": 2.095293529404707,
      "learning_rate": 3.914942670200529e-06,
      "loss": 0.6913,
      "step": 1335
    },
    {
      "epoch": 0.12048518735627001,
      "grad_norm": 2.097787863582317,
      "learning_rate": 3.914774027143436e-06,
      "loss": 0.8662,
      "step": 1336
    },
    {
      "epoch": 0.12057537087974027,
      "grad_norm": 3.175436191182632,
      "learning_rate": 3.914605220707265e-06,
      "loss": 0.7978,
      "step": 1337
    },
    {
      "epoch": 0.12066555440321053,
      "grad_norm": 1.808361986112001,
      "learning_rate": 3.9144362509064194e-06,
      "loss": 0.9561,
      "step": 1338
    },
    {
      "epoch": 0.12075573792668079,
      "grad_norm": 3.4757545475310274,
      "learning_rate": 3.914267117755317e-06,
      "loss": 0.8783,
      "step": 1339
    },
    {
      "epoch": 0.12084592145015106,
      "grad_norm": 1.859310052843358,
      "learning_rate": 3.914097821268389e-06,
      "loss": 0.7738,
      "step": 1340
    },
    {
      "epoch": 0.12093610497362133,
      "grad_norm": 1.7276414632772523,
      "learning_rate": 3.913928361460081e-06,
      "loss": 0.8231,
      "step": 1341
    },
    {
      "epoch": 0.12102628849709159,
      "grad_norm": 1.9711116371144526,
      "learning_rate": 3.913758738344851e-06,
      "loss": 0.7919,
      "step": 1342
    },
    {
      "epoch": 0.12111647202056185,
      "grad_norm": 0.7460283180824441,
      "learning_rate": 3.913588951937174e-06,
      "loss": 0.6028,
      "step": 1343
    },
    {
      "epoch": 0.12120665554403211,
      "grad_norm": 0.7113735831753825,
      "learning_rate": 3.9134190022515355e-06,
      "loss": 0.6021,
      "step": 1344
    },
    {
      "epoch": 0.12129683906750237,
      "grad_norm": 1.7684453533380442,
      "learning_rate": 3.913248889302438e-06,
      "loss": 0.77,
      "step": 1345
    },
    {
      "epoch": 0.12138702259097263,
      "grad_norm": 1.7139208141316353,
      "learning_rate": 3.913078613104395e-06,
      "loss": 0.7441,
      "step": 1346
    },
    {
      "epoch": 0.12147720611444289,
      "grad_norm": 1.6727519364762689,
      "learning_rate": 3.912908173671936e-06,
      "loss": 0.6937,
      "step": 1347
    },
    {
      "epoch": 0.12156738963791315,
      "grad_norm": 3.0352395143220177,
      "learning_rate": 3.9127375710196044e-06,
      "loss": 0.7329,
      "step": 1348
    },
    {
      "epoch": 0.12165757316138341,
      "grad_norm": 1.9852167581931974,
      "learning_rate": 3.912566805161957e-06,
      "loss": 0.751,
      "step": 1349
    },
    {
      "epoch": 0.12174775668485367,
      "grad_norm": 3.2113516510302493,
      "learning_rate": 3.912395876113564e-06,
      "loss": 0.8278,
      "step": 1350
    },
    {
      "epoch": 0.12183794020832393,
      "grad_norm": 1.4400123534732323,
      "learning_rate": 3.912224783889009e-06,
      "loss": 0.8101,
      "step": 1351
    },
    {
      "epoch": 0.12192812373179421,
      "grad_norm": 1.6968490739109852,
      "learning_rate": 3.912053528502892e-06,
      "loss": 0.818,
      "step": 1352
    },
    {
      "epoch": 0.12201830725526447,
      "grad_norm": 1.600464085984857,
      "learning_rate": 3.911882109969825e-06,
      "loss": 0.7317,
      "step": 1353
    },
    {
      "epoch": 0.12210849077873473,
      "grad_norm": 1.8084068119338443,
      "learning_rate": 3.911710528304435e-06,
      "loss": 0.8199,
      "step": 1354
    },
    {
      "epoch": 0.12219867430220499,
      "grad_norm": 8.581959685494743,
      "learning_rate": 3.911538783521361e-06,
      "loss": 0.7607,
      "step": 1355
    },
    {
      "epoch": 0.12228885782567525,
      "grad_norm": 2.2019727141987766,
      "learning_rate": 3.9113668756352575e-06,
      "loss": 0.8246,
      "step": 1356
    },
    {
      "epoch": 0.12237904134914551,
      "grad_norm": 1.403871304645486,
      "learning_rate": 3.911194804660793e-06,
      "loss": 0.7349,
      "step": 1357
    },
    {
      "epoch": 0.12246922487261577,
      "grad_norm": 1.7097462726880925,
      "learning_rate": 3.91102257061265e-06,
      "loss": 0.8896,
      "step": 1358
    },
    {
      "epoch": 0.12255940839608603,
      "grad_norm": 1.4897646582442932,
      "learning_rate": 3.910850173505524e-06,
      "loss": 0.8482,
      "step": 1359
    },
    {
      "epoch": 0.1226495919195563,
      "grad_norm": 2.5134248767779104,
      "learning_rate": 3.9106776133541255e-06,
      "loss": 0.8838,
      "step": 1360
    },
    {
      "epoch": 0.12273977544302656,
      "grad_norm": 1.8355981391393557,
      "learning_rate": 3.9105048901731766e-06,
      "loss": 0.7633,
      "step": 1361
    },
    {
      "epoch": 0.12282995896649682,
      "grad_norm": 2.2390043944587874,
      "learning_rate": 3.9103320039774165e-06,
      "loss": 0.7952,
      "step": 1362
    },
    {
      "epoch": 0.12292014248996708,
      "grad_norm": 1.6737409499979963,
      "learning_rate": 3.9101589547815965e-06,
      "loss": 0.8637,
      "step": 1363
    },
    {
      "epoch": 0.12301032601343735,
      "grad_norm": 1.426190377676727,
      "learning_rate": 3.909985742600482e-06,
      "loss": 0.7953,
      "step": 1364
    },
    {
      "epoch": 0.12310050953690761,
      "grad_norm": 1.6482431740141392,
      "learning_rate": 3.909812367448852e-06,
      "loss": 0.7713,
      "step": 1365
    },
    {
      "epoch": 0.12319069306037787,
      "grad_norm": 2.5192084527049228,
      "learning_rate": 3.909638829341501e-06,
      "loss": 0.785,
      "step": 1366
    },
    {
      "epoch": 0.12328087658384813,
      "grad_norm": 1.5866199396903693,
      "learning_rate": 3.909465128293234e-06,
      "loss": 0.8632,
      "step": 1367
    },
    {
      "epoch": 0.1233710601073184,
      "grad_norm": 1.9970400483839184,
      "learning_rate": 3.9092912643188745e-06,
      "loss": 0.8648,
      "step": 1368
    },
    {
      "epoch": 0.12346124363078866,
      "grad_norm": 5.553943181770318,
      "learning_rate": 3.909117237433256e-06,
      "loss": 0.8044,
      "step": 1369
    },
    {
      "epoch": 0.12355142715425892,
      "grad_norm": 2.9651314617173576,
      "learning_rate": 3.908943047651229e-06,
      "loss": 0.852,
      "step": 1370
    },
    {
      "epoch": 0.12364161067772918,
      "grad_norm": 1.6507701550249414,
      "learning_rate": 3.908768694987655e-06,
      "loss": 0.793,
      "step": 1371
    },
    {
      "epoch": 0.12373179420119944,
      "grad_norm": 2.500468496293895,
      "learning_rate": 3.908594179457411e-06,
      "loss": 0.7288,
      "step": 1372
    },
    {
      "epoch": 0.1238219777246697,
      "grad_norm": 1.760711923551785,
      "learning_rate": 3.908419501075388e-06,
      "loss": 0.809,
      "step": 1373
    },
    {
      "epoch": 0.12391216124813996,
      "grad_norm": 2.229173915768064,
      "learning_rate": 3.90824465985649e-06,
      "loss": 0.8815,
      "step": 1374
    },
    {
      "epoch": 0.12400234477161022,
      "grad_norm": 1.764840618534358,
      "learning_rate": 3.908069655815636e-06,
      "loss": 0.8491,
      "step": 1375
    },
    {
      "epoch": 0.1240925282950805,
      "grad_norm": 2.094486050258579,
      "learning_rate": 3.907894488967758e-06,
      "loss": 0.7469,
      "step": 1376
    },
    {
      "epoch": 0.12418271181855076,
      "grad_norm": 2.5062833027517746,
      "learning_rate": 3.9077191593278005e-06,
      "loss": 0.8672,
      "step": 1377
    },
    {
      "epoch": 0.12427289534202102,
      "grad_norm": 0.7155432275420278,
      "learning_rate": 3.9075436669107265e-06,
      "loss": 0.5779,
      "step": 1378
    },
    {
      "epoch": 0.12436307886549128,
      "grad_norm": 1.8234228975288267,
      "learning_rate": 3.90736801173151e-06,
      "loss": 0.8242,
      "step": 1379
    },
    {
      "epoch": 0.12445326238896154,
      "grad_norm": 1.4001963120152443,
      "learning_rate": 3.907192193805136e-06,
      "loss": 0.8192,
      "step": 1380
    },
    {
      "epoch": 0.1245434459124318,
      "grad_norm": 1.9588583452145847,
      "learning_rate": 3.907016213146608e-06,
      "loss": 0.8651,
      "step": 1381
    },
    {
      "epoch": 0.12463362943590206,
      "grad_norm": 1.7540457145449142,
      "learning_rate": 3.906840069770942e-06,
      "loss": 0.9213,
      "step": 1382
    },
    {
      "epoch": 0.12472381295937232,
      "grad_norm": 5.416763754732614,
      "learning_rate": 3.906663763693167e-06,
      "loss": 0.8357,
      "step": 1383
    },
    {
      "epoch": 0.12481399648284258,
      "grad_norm": 2.537911867437698,
      "learning_rate": 3.906487294928327e-06,
      "loss": 0.791,
      "step": 1384
    },
    {
      "epoch": 0.12490418000631284,
      "grad_norm": 1.9679482053477104,
      "learning_rate": 3.906310663491478e-06,
      "loss": 0.8095,
      "step": 1385
    },
    {
      "epoch": 0.1249943635297831,
      "grad_norm": 1.7318644180442628,
      "learning_rate": 3.906133869397692e-06,
      "loss": 0.8587,
      "step": 1386
    },
    {
      "epoch": 0.12508454705325336,
      "grad_norm": 1.784626821729833,
      "learning_rate": 3.905956912662054e-06,
      "loss": 0.7709,
      "step": 1387
    },
    {
      "epoch": 0.12517473057672363,
      "grad_norm": 3.2159390814216366,
      "learning_rate": 3.905779793299662e-06,
      "loss": 0.8758,
      "step": 1388
    },
    {
      "epoch": 0.12526491410019389,
      "grad_norm": 1.781204089192122,
      "learning_rate": 3.905602511325631e-06,
      "loss": 0.838,
      "step": 1389
    },
    {
      "epoch": 0.12535509762366415,
      "grad_norm": 1.8463949854490127,
      "learning_rate": 3.905425066755086e-06,
      "loss": 0.7733,
      "step": 1390
    },
    {
      "epoch": 0.1254452811471344,
      "grad_norm": 1.5501171006299115,
      "learning_rate": 3.905247459603168e-06,
      "loss": 0.7633,
      "step": 1391
    },
    {
      "epoch": 0.12553546467060467,
      "grad_norm": 2.176340575724349,
      "learning_rate": 3.905069689885031e-06,
      "loss": 0.6872,
      "step": 1392
    },
    {
      "epoch": 0.12562564819407493,
      "grad_norm": 1.462061322273834,
      "learning_rate": 3.904891757615843e-06,
      "loss": 0.6918,
      "step": 1393
    },
    {
      "epoch": 0.12571583171754522,
      "grad_norm": 0.6161450140703891,
      "learning_rate": 3.9047136628107874e-06,
      "loss": 0.5642,
      "step": 1394
    },
    {
      "epoch": 0.12580601524101548,
      "grad_norm": 1.72451163303326,
      "learning_rate": 3.904535405485059e-06,
      "loss": 0.9325,
      "step": 1395
    },
    {
      "epoch": 0.12589619876448574,
      "grad_norm": 4.325220999831997,
      "learning_rate": 3.90435698565387e-06,
      "loss": 0.754,
      "step": 1396
    },
    {
      "epoch": 0.125986382287956,
      "grad_norm": 4.054917757115152,
      "learning_rate": 3.904178403332441e-06,
      "loss": 0.7692,
      "step": 1397
    },
    {
      "epoch": 0.12607656581142626,
      "grad_norm": 2.2332298705584783,
      "learning_rate": 3.903999658536012e-06,
      "loss": 0.8942,
      "step": 1398
    },
    {
      "epoch": 0.12616674933489652,
      "grad_norm": 2.1198789936295506,
      "learning_rate": 3.903820751279833e-06,
      "loss": 0.807,
      "step": 1399
    },
    {
      "epoch": 0.12625693285836678,
      "grad_norm": 2.801191100455003,
      "learning_rate": 3.90364168157917e-06,
      "loss": 0.8444,
      "step": 1400
    },
    {
      "epoch": 0.12634711638183704,
      "grad_norm": 1.7414901090259471,
      "learning_rate": 3.903462449449302e-06,
      "loss": 0.841,
      "step": 1401
    },
    {
      "epoch": 0.1264372999053073,
      "grad_norm": 3.012458047535132,
      "learning_rate": 3.903283054905522e-06,
      "loss": 0.7786,
      "step": 1402
    },
    {
      "epoch": 0.12652748342877757,
      "grad_norm": 2.4167082772016,
      "learning_rate": 3.9031034979631385e-06,
      "loss": 0.7236,
      "step": 1403
    },
    {
      "epoch": 0.12661766695224783,
      "grad_norm": 1.6986210615951916,
      "learning_rate": 3.902923778637469e-06,
      "loss": 0.8402,
      "step": 1404
    },
    {
      "epoch": 0.1267078504757181,
      "grad_norm": 1.6287014693014605,
      "learning_rate": 3.902743896943852e-06,
      "loss": 0.7405,
      "step": 1405
    },
    {
      "epoch": 0.12679803399918835,
      "grad_norm": 0.7239077891171432,
      "learning_rate": 3.902563852897633e-06,
      "loss": 0.6326,
      "step": 1406
    },
    {
      "epoch": 0.1268882175226586,
      "grad_norm": 0.702148734611268,
      "learning_rate": 3.9023836465141755e-06,
      "loss": 0.6071,
      "step": 1407
    },
    {
      "epoch": 0.12697840104612887,
      "grad_norm": 1.8613506688036672,
      "learning_rate": 3.902203277808856e-06,
      "loss": 0.8548,
      "step": 1408
    },
    {
      "epoch": 0.12706858456959913,
      "grad_norm": 1.6721361615850094,
      "learning_rate": 3.902022746797064e-06,
      "loss": 0.8643,
      "step": 1409
    },
    {
      "epoch": 0.1271587680930694,
      "grad_norm": 1.9560498119946728,
      "learning_rate": 3.9018420534942035e-06,
      "loss": 0.9179,
      "step": 1410
    },
    {
      "epoch": 0.12724895161653965,
      "grad_norm": 2.168404310307514,
      "learning_rate": 3.9016611979156935e-06,
      "loss": 0.804,
      "step": 1411
    },
    {
      "epoch": 0.1273391351400099,
      "grad_norm": 2.7122938297516166,
      "learning_rate": 3.9014801800769635e-06,
      "loss": 0.7234,
      "step": 1412
    },
    {
      "epoch": 0.12742931866348017,
      "grad_norm": 1.7816844878749174,
      "learning_rate": 3.901298999993459e-06,
      "loss": 0.8373,
      "step": 1413
    },
    {
      "epoch": 0.12751950218695043,
      "grad_norm": 2.781361352641635,
      "learning_rate": 3.901117657680642e-06,
      "loss": 0.7422,
      "step": 1414
    },
    {
      "epoch": 0.1276096857104207,
      "grad_norm": 1.6435813788310976,
      "learning_rate": 3.900936153153982e-06,
      "loss": 0.7398,
      "step": 1415
    },
    {
      "epoch": 0.12769986923389096,
      "grad_norm": 0.7122390068279678,
      "learning_rate": 3.900754486428968e-06,
      "loss": 0.5602,
      "step": 1416
    },
    {
      "epoch": 0.12779005275736122,
      "grad_norm": 3.1963035333750485,
      "learning_rate": 3.900572657521102e-06,
      "loss": 0.8955,
      "step": 1417
    },
    {
      "epoch": 0.1278802362808315,
      "grad_norm": 2.1097871730913917,
      "learning_rate": 3.900390666445896e-06,
      "loss": 0.7853,
      "step": 1418
    },
    {
      "epoch": 0.12797041980430177,
      "grad_norm": 7.843081016432398,
      "learning_rate": 3.9002085132188795e-06,
      "loss": 0.7763,
      "step": 1419
    },
    {
      "epoch": 0.12806060332777203,
      "grad_norm": 1.7983403077695912,
      "learning_rate": 3.9000261978555964e-06,
      "loss": 0.8471,
      "step": 1420
    },
    {
      "epoch": 0.1281507868512423,
      "grad_norm": 1.7464432675910382,
      "learning_rate": 3.8998437203716e-06,
      "loss": 0.8671,
      "step": 1421
    },
    {
      "epoch": 0.12824097037471255,
      "grad_norm": 4.971004910104614,
      "learning_rate": 3.899661080782462e-06,
      "loss": 0.8326,
      "step": 1422
    },
    {
      "epoch": 0.1283311538981828,
      "grad_norm": 1.5861949993485203,
      "learning_rate": 3.899478279103767e-06,
      "loss": 0.8421,
      "step": 1423
    },
    {
      "epoch": 0.12842133742165307,
      "grad_norm": 2.0428848395246892,
      "learning_rate": 3.8992953153511105e-06,
      "loss": 0.8385,
      "step": 1424
    },
    {
      "epoch": 0.12851152094512333,
      "grad_norm": 1.6801911774350897,
      "learning_rate": 3.899112189540106e-06,
      "loss": 0.8693,
      "step": 1425
    },
    {
      "epoch": 0.1286017044685936,
      "grad_norm": 2.172704414785762,
      "learning_rate": 3.898928901686377e-06,
      "loss": 0.846,
      "step": 1426
    },
    {
      "epoch": 0.12869188799206385,
      "grad_norm": 0.7092917960689188,
      "learning_rate": 3.898745451805564e-06,
      "loss": 0.5647,
      "step": 1427
    },
    {
      "epoch": 0.1287820715155341,
      "grad_norm": 1.704563128639838,
      "learning_rate": 3.898561839913319e-06,
      "loss": 0.7402,
      "step": 1428
    },
    {
      "epoch": 0.12887225503900437,
      "grad_norm": 1.9718856666143443,
      "learning_rate": 3.89837806602531e-06,
      "loss": 0.6582,
      "step": 1429
    },
    {
      "epoch": 0.12896243856247463,
      "grad_norm": 1.8493374359761015,
      "learning_rate": 3.898194130157217e-06,
      "loss": 0.8168,
      "step": 1430
    },
    {
      "epoch": 0.1290526220859449,
      "grad_norm": 1.8485330513436296,
      "learning_rate": 3.8980100323247335e-06,
      "loss": 0.7478,
      "step": 1431
    },
    {
      "epoch": 0.12914280560941516,
      "grad_norm": 1.728960882910426,
      "learning_rate": 3.897825772543568e-06,
      "loss": 0.7431,
      "step": 1432
    },
    {
      "epoch": 0.12923298913288542,
      "grad_norm": 1.972809557103585,
      "learning_rate": 3.897641350829444e-06,
      "loss": 0.8666,
      "step": 1433
    },
    {
      "epoch": 0.12932317265635568,
      "grad_norm": 1.9126427192044215,
      "learning_rate": 3.897456767198096e-06,
      "loss": 0.8906,
      "step": 1434
    },
    {
      "epoch": 0.12941335617982594,
      "grad_norm": 2.7613772946015307,
      "learning_rate": 3.897272021665275e-06,
      "loss": 0.746,
      "step": 1435
    },
    {
      "epoch": 0.1295035397032962,
      "grad_norm": 1.6608629584967984,
      "learning_rate": 3.897087114246743e-06,
      "loss": 0.914,
      "step": 1436
    },
    {
      "epoch": 0.12959372322676646,
      "grad_norm": 2.01532938364096,
      "learning_rate": 3.896902044958279e-06,
      "loss": 0.9237,
      "step": 1437
    },
    {
      "epoch": 0.12968390675023672,
      "grad_norm": 3.088018946722935,
      "learning_rate": 3.896716813815672e-06,
      "loss": 0.8692,
      "step": 1438
    },
    {
      "epoch": 0.12977409027370698,
      "grad_norm": 1.6037413871456558,
      "learning_rate": 3.896531420834728e-06,
      "loss": 0.7547,
      "step": 1439
    },
    {
      "epoch": 0.12986427379717724,
      "grad_norm": 1.590274911662281,
      "learning_rate": 3.896345866031266e-06,
      "loss": 0.7145,
      "step": 1440
    },
    {
      "epoch": 0.1299544573206475,
      "grad_norm": 1.7132185966700795,
      "learning_rate": 3.896160149421119e-06,
      "loss": 0.8685,
      "step": 1441
    },
    {
      "epoch": 0.1300446408441178,
      "grad_norm": 1.9930540107724017,
      "learning_rate": 3.8959742710201314e-06,
      "loss": 0.8769,
      "step": 1442
    },
    {
      "epoch": 0.13013482436758805,
      "grad_norm": 2.060219515428322,
      "learning_rate": 3.895788230844166e-06,
      "loss": 0.7112,
      "step": 1443
    },
    {
      "epoch": 0.13022500789105831,
      "grad_norm": 1.799069664857879,
      "learning_rate": 3.895602028909095e-06,
      "loss": 0.9684,
      "step": 1444
    },
    {
      "epoch": 0.13031519141452858,
      "grad_norm": 0.6429899339674072,
      "learning_rate": 3.895415665230807e-06,
      "loss": 0.5987,
      "step": 1445
    },
    {
      "epoch": 0.13040537493799884,
      "grad_norm": 1.2620819797216851,
      "learning_rate": 3.895229139825203e-06,
      "loss": 0.8323,
      "step": 1446
    },
    {
      "epoch": 0.1304955584614691,
      "grad_norm": 1.7352784012038092,
      "learning_rate": 3.895042452708198e-06,
      "loss": 0.726,
      "step": 1447
    },
    {
      "epoch": 0.13058574198493936,
      "grad_norm": 1.7749841474443058,
      "learning_rate": 3.894855603895723e-06,
      "loss": 0.718,
      "step": 1448
    },
    {
      "epoch": 0.13067592550840962,
      "grad_norm": 1.7871624047588055,
      "learning_rate": 3.894668593403718e-06,
      "loss": 0.8871,
      "step": 1449
    },
    {
      "epoch": 0.13076610903187988,
      "grad_norm": 1.7714885471887887,
      "learning_rate": 3.8944814212481425e-06,
      "loss": 0.8123,
      "step": 1450
    },
    {
      "epoch": 0.13085629255535014,
      "grad_norm": 1.9636451903017984,
      "learning_rate": 3.894294087444966e-06,
      "loss": 0.7904,
      "step": 1451
    },
    {
      "epoch": 0.1309464760788204,
      "grad_norm": 1.964874146000754,
      "learning_rate": 3.894106592010173e-06,
      "loss": 0.7062,
      "step": 1452
    },
    {
      "epoch": 0.13103665960229066,
      "grad_norm": 1.5972570093024696,
      "learning_rate": 3.893918934959762e-06,
      "loss": 0.6865,
      "step": 1453
    },
    {
      "epoch": 0.13112684312576092,
      "grad_norm": 3.6593563870967314,
      "learning_rate": 3.893731116309743e-06,
      "loss": 0.7462,
      "step": 1454
    },
    {
      "epoch": 0.13121702664923118,
      "grad_norm": 1.965725649635681,
      "learning_rate": 3.893543136076145e-06,
      "loss": 0.8158,
      "step": 1455
    },
    {
      "epoch": 0.13130721017270144,
      "grad_norm": 2.289553124249041,
      "learning_rate": 3.893354994275006e-06,
      "loss": 0.8361,
      "step": 1456
    },
    {
      "epoch": 0.1313973936961717,
      "grad_norm": 13.046424720897557,
      "learning_rate": 3.893166690922378e-06,
      "loss": 0.7763,
      "step": 1457
    },
    {
      "epoch": 0.13148757721964197,
      "grad_norm": 2.3249650029655604,
      "learning_rate": 3.892978226034329e-06,
      "loss": 0.8439,
      "step": 1458
    },
    {
      "epoch": 0.13157776074311223,
      "grad_norm": 2.0545878222985148,
      "learning_rate": 3.89278959962694e-06,
      "loss": 0.852,
      "step": 1459
    },
    {
      "epoch": 0.1316679442665825,
      "grad_norm": 1.7388090211132754,
      "learning_rate": 3.8926008117163056e-06,
      "loss": 0.7839,
      "step": 1460
    },
    {
      "epoch": 0.13175812779005275,
      "grad_norm": 1.8281655592372374,
      "learning_rate": 3.892411862318535e-06,
      "loss": 0.7922,
      "step": 1461
    },
    {
      "epoch": 0.131848311313523,
      "grad_norm": 1.8632765795891086,
      "learning_rate": 3.892222751449749e-06,
      "loss": 0.8275,
      "step": 1462
    },
    {
      "epoch": 0.13193849483699327,
      "grad_norm": 2.0037198045041524,
      "learning_rate": 3.892033479126084e-06,
      "loss": 0.76,
      "step": 1463
    },
    {
      "epoch": 0.13202867836046353,
      "grad_norm": 2.805715048592591,
      "learning_rate": 3.891844045363691e-06,
      "loss": 0.7912,
      "step": 1464
    },
    {
      "epoch": 0.13211886188393382,
      "grad_norm": 1.861595174679734,
      "learning_rate": 3.891654450178732e-06,
      "loss": 0.733,
      "step": 1465
    },
    {
      "epoch": 0.13220904540740408,
      "grad_norm": 1.5001048210394616,
      "learning_rate": 3.891464693587385e-06,
      "loss": 0.8529,
      "step": 1466
    },
    {
      "epoch": 0.13229922893087434,
      "grad_norm": 2.5280823844269693,
      "learning_rate": 3.89127477560584e-06,
      "loss": 0.8481,
      "step": 1467
    },
    {
      "epoch": 0.1323894124543446,
      "grad_norm": 1.6533911160913426,
      "learning_rate": 3.891084696250304e-06,
      "loss": 0.8432,
      "step": 1468
    },
    {
      "epoch": 0.13247959597781486,
      "grad_norm": 2.3374354430071067,
      "learning_rate": 3.890894455536993e-06,
      "loss": 0.849,
      "step": 1469
    },
    {
      "epoch": 0.13256977950128512,
      "grad_norm": 2.9386857257208816,
      "learning_rate": 3.890704053482142e-06,
      "loss": 0.8649,
      "step": 1470
    },
    {
      "epoch": 0.13265996302475538,
      "grad_norm": 1.7490773493531913,
      "learning_rate": 3.890513490101995e-06,
      "loss": 0.7571,
      "step": 1471
    },
    {
      "epoch": 0.13275014654822564,
      "grad_norm": 0.6184931356763047,
      "learning_rate": 3.890322765412814e-06,
      "loss": 0.6064,
      "step": 1472
    },
    {
      "epoch": 0.1328403300716959,
      "grad_norm": 1.6345547228988324,
      "learning_rate": 3.890131879430871e-06,
      "loss": 0.833,
      "step": 1473
    },
    {
      "epoch": 0.13293051359516617,
      "grad_norm": 1.64694920468591,
      "learning_rate": 3.889940832172454e-06,
      "loss": 0.8339,
      "step": 1474
    },
    {
      "epoch": 0.13302069711863643,
      "grad_norm": 6.649351171520216,
      "learning_rate": 3.889749623653864e-06,
      "loss": 0.8515,
      "step": 1475
    },
    {
      "epoch": 0.1331108806421067,
      "grad_norm": 1.437467906427788,
      "learning_rate": 3.889558253891416e-06,
      "loss": 0.7483,
      "step": 1476
    },
    {
      "epoch": 0.13320106416557695,
      "grad_norm": 0.6294075289050791,
      "learning_rate": 3.8893667229014385e-06,
      "loss": 0.5826,
      "step": 1477
    },
    {
      "epoch": 0.1332912476890472,
      "grad_norm": 1.9506132530926958,
      "learning_rate": 3.8891750307002746e-06,
      "loss": 0.7159,
      "step": 1478
    },
    {
      "epoch": 0.13338143121251747,
      "grad_norm": 1.3745487946681432,
      "learning_rate": 3.888983177304281e-06,
      "loss": 0.7409,
      "step": 1479
    },
    {
      "epoch": 0.13347161473598773,
      "grad_norm": 0.7477515212757446,
      "learning_rate": 3.888791162729826e-06,
      "loss": 0.5809,
      "step": 1480
    },
    {
      "epoch": 0.133561798259458,
      "grad_norm": 3.48335476979743,
      "learning_rate": 3.888598986993295e-06,
      "loss": 0.8329,
      "step": 1481
    },
    {
      "epoch": 0.13365198178292825,
      "grad_norm": 1.7194074587044403,
      "learning_rate": 3.888406650111085e-06,
      "loss": 0.8381,
      "step": 1482
    },
    {
      "epoch": 0.1337421653063985,
      "grad_norm": 2.0364792636533973,
      "learning_rate": 3.888214152099607e-06,
      "loss": 0.8091,
      "step": 1483
    },
    {
      "epoch": 0.13383234882986877,
      "grad_norm": 5.0280036641352615,
      "learning_rate": 3.888021492975285e-06,
      "loss": 0.8815,
      "step": 1484
    },
    {
      "epoch": 0.13392253235333904,
      "grad_norm": 2.2608921470917736,
      "learning_rate": 3.88782867275456e-06,
      "loss": 0.8611,
      "step": 1485
    },
    {
      "epoch": 0.1340127158768093,
      "grad_norm": 1.464475132656513,
      "learning_rate": 3.8876356914538824e-06,
      "loss": 0.7706,
      "step": 1486
    },
    {
      "epoch": 0.13410289940027956,
      "grad_norm": 2.0513775639733276,
      "learning_rate": 3.88744254908972e-06,
      "loss": 0.7649,
      "step": 1487
    },
    {
      "epoch": 0.13419308292374982,
      "grad_norm": 1.9408420533502608,
      "learning_rate": 3.887249245678552e-06,
      "loss": 0.7837,
      "step": 1488
    },
    {
      "epoch": 0.1342832664472201,
      "grad_norm": 1.5861049620061367,
      "learning_rate": 3.887055781236872e-06,
      "loss": 0.8815,
      "step": 1489
    },
    {
      "epoch": 0.13437344997069037,
      "grad_norm": 1.8831548735651116,
      "learning_rate": 3.886862155781186e-06,
      "loss": 0.6807,
      "step": 1490
    },
    {
      "epoch": 0.13446363349416063,
      "grad_norm": 0.6801621325479685,
      "learning_rate": 3.886668369328019e-06,
      "loss": 0.605,
      "step": 1491
    },
    {
      "epoch": 0.1345538170176309,
      "grad_norm": 2.3152470124454747,
      "learning_rate": 3.886474421893904e-06,
      "loss": 0.8379,
      "step": 1492
    },
    {
      "epoch": 0.13464400054110115,
      "grad_norm": 1.598036070353316,
      "learning_rate": 3.886280313495388e-06,
      "loss": 0.7282,
      "step": 1493
    },
    {
      "epoch": 0.1347341840645714,
      "grad_norm": 2.718472959938582,
      "learning_rate": 3.886086044149035e-06,
      "loss": 0.7774,
      "step": 1494
    },
    {
      "epoch": 0.13482436758804167,
      "grad_norm": 1.507166037061642,
      "learning_rate": 3.885891613871421e-06,
      "loss": 0.8152,
      "step": 1495
    },
    {
      "epoch": 0.13491455111151193,
      "grad_norm": 2.277341132710492,
      "learning_rate": 3.885697022679136e-06,
      "loss": 0.8575,
      "step": 1496
    },
    {
      "epoch": 0.1350047346349822,
      "grad_norm": 1.7333109572995908,
      "learning_rate": 3.885502270588784e-06,
      "loss": 0.8349,
      "step": 1497
    },
    {
      "epoch": 0.13509491815845245,
      "grad_norm": 1.9317125204841024,
      "learning_rate": 3.885307357616981e-06,
      "loss": 0.8036,
      "step": 1498
    },
    {
      "epoch": 0.13518510168192271,
      "grad_norm": 1.7489518023655428,
      "learning_rate": 3.885112283780359e-06,
      "loss": 0.8034,
      "step": 1499
    },
    {
      "epoch": 0.13527528520539298,
      "grad_norm": 2.9502150311737574,
      "learning_rate": 3.8849170490955624e-06,
      "loss": 0.6941,
      "step": 1500
    },
    {
      "epoch": 0.13536546872886324,
      "grad_norm": 2.0999016420854346,
      "learning_rate": 3.88472165357925e-06,
      "loss": 0.7993,
      "step": 1501
    },
    {
      "epoch": 0.1354556522523335,
      "grad_norm": 1.9164894754631858,
      "learning_rate": 3.884526097248093e-06,
      "loss": 0.761,
      "step": 1502
    },
    {
      "epoch": 0.13554583577580376,
      "grad_norm": 2.034158004227629,
      "learning_rate": 3.884330380118779e-06,
      "loss": 0.8198,
      "step": 1503
    },
    {
      "epoch": 0.13563601929927402,
      "grad_norm": 3.7673933380358045,
      "learning_rate": 3.884134502208007e-06,
      "loss": 0.7574,
      "step": 1504
    },
    {
      "epoch": 0.13572620282274428,
      "grad_norm": 3.2656596058741316,
      "learning_rate": 3.88393846353249e-06,
      "loss": 0.9053,
      "step": 1505
    },
    {
      "epoch": 0.13581638634621454,
      "grad_norm": 1.5555355500266943,
      "learning_rate": 3.883742264108955e-06,
      "loss": 0.8571,
      "step": 1506
    },
    {
      "epoch": 0.1359065698696848,
      "grad_norm": 1.7951256238921551,
      "learning_rate": 3.883545903954145e-06,
      "loss": 0.8104,
      "step": 1507
    },
    {
      "epoch": 0.13599675339315506,
      "grad_norm": 1.9134640750787095,
      "learning_rate": 3.883349383084811e-06,
      "loss": 0.8008,
      "step": 1508
    },
    {
      "epoch": 0.13608693691662532,
      "grad_norm": 2.0050804460706417,
      "learning_rate": 3.883152701517723e-06,
      "loss": 0.7288,
      "step": 1509
    },
    {
      "epoch": 0.13617712044009558,
      "grad_norm": 1.849651718863408,
      "learning_rate": 3.882955859269664e-06,
      "loss": 0.7923,
      "step": 1510
    },
    {
      "epoch": 0.13626730396356584,
      "grad_norm": 1.6707992658051387,
      "learning_rate": 3.882758856357428e-06,
      "loss": 0.8097,
      "step": 1511
    },
    {
      "epoch": 0.1363574874870361,
      "grad_norm": 2.23645498923727,
      "learning_rate": 3.882561692797824e-06,
      "loss": 0.8007,
      "step": 1512
    },
    {
      "epoch": 0.1364476710105064,
      "grad_norm": 2.194916930484175,
      "learning_rate": 3.882364368607677e-06,
      "loss": 0.8066,
      "step": 1513
    },
    {
      "epoch": 0.13653785453397665,
      "grad_norm": 4.870045222516969,
      "learning_rate": 3.8821668838038225e-06,
      "loss": 0.8177,
      "step": 1514
    },
    {
      "epoch": 0.13662803805744692,
      "grad_norm": 3.3474850338180366,
      "learning_rate": 3.881969238403111e-06,
      "loss": 0.7179,
      "step": 1515
    },
    {
      "epoch": 0.13671822158091718,
      "grad_norm": 3.2990424818880775,
      "learning_rate": 3.881771432422408e-06,
      "loss": 0.9335,
      "step": 1516
    },
    {
      "epoch": 0.13680840510438744,
      "grad_norm": 1.7514118221649333,
      "learning_rate": 3.88157346587859e-06,
      "loss": 0.8645,
      "step": 1517
    },
    {
      "epoch": 0.1368985886278577,
      "grad_norm": 1.540601628805243,
      "learning_rate": 3.881375338788549e-06,
      "loss": 0.7798,
      "step": 1518
    },
    {
      "epoch": 0.13698877215132796,
      "grad_norm": 3.1506402046307422,
      "learning_rate": 3.88117705116919e-06,
      "loss": 0.8263,
      "step": 1519
    },
    {
      "epoch": 0.13707895567479822,
      "grad_norm": 0.8156739011991138,
      "learning_rate": 3.880978603037432e-06,
      "loss": 0.6563,
      "step": 1520
    },
    {
      "epoch": 0.13716913919826848,
      "grad_norm": 1.7247230086254237,
      "learning_rate": 3.880779994410209e-06,
      "loss": 0.8075,
      "step": 1521
    },
    {
      "epoch": 0.13725932272173874,
      "grad_norm": 2.300659694601744,
      "learning_rate": 3.880581225304466e-06,
      "loss": 0.6833,
      "step": 1522
    },
    {
      "epoch": 0.137349506245209,
      "grad_norm": 1.92741896997078,
      "learning_rate": 3.880382295737163e-06,
      "loss": 0.8317,
      "step": 1523
    },
    {
      "epoch": 0.13743968976867926,
      "grad_norm": 2.7332764080756844,
      "learning_rate": 3.880183205725274e-06,
      "loss": 0.7861,
      "step": 1524
    },
    {
      "epoch": 0.13752987329214952,
      "grad_norm": 2.172953824840771,
      "learning_rate": 3.879983955285788e-06,
      "loss": 0.7419,
      "step": 1525
    },
    {
      "epoch": 0.13762005681561978,
      "grad_norm": 4.982931567851168,
      "learning_rate": 3.879784544435703e-06,
      "loss": 0.7812,
      "step": 1526
    },
    {
      "epoch": 0.13771024033909005,
      "grad_norm": 2.457854746984734,
      "learning_rate": 3.879584973192037e-06,
      "loss": 0.7509,
      "step": 1527
    },
    {
      "epoch": 0.1378004238625603,
      "grad_norm": 1.9022515057204252,
      "learning_rate": 3.8793852415718165e-06,
      "loss": 0.7402,
      "step": 1528
    },
    {
      "epoch": 0.13789060738603057,
      "grad_norm": 2.108731998350938,
      "learning_rate": 3.879185349592085e-06,
      "loss": 0.8135,
      "step": 1529
    },
    {
      "epoch": 0.13798079090950083,
      "grad_norm": 4.261639084670156,
      "learning_rate": 3.878985297269897e-06,
      "loss": 0.8272,
      "step": 1530
    },
    {
      "epoch": 0.1380709744329711,
      "grad_norm": 2.849131428390104,
      "learning_rate": 3.878785084622323e-06,
      "loss": 0.8472,
      "step": 1531
    },
    {
      "epoch": 0.13816115795644135,
      "grad_norm": 2.466141686978156,
      "learning_rate": 3.878584711666447e-06,
      "loss": 0.9883,
      "step": 1532
    },
    {
      "epoch": 0.1382513414799116,
      "grad_norm": 1.8980262216660007,
      "learning_rate": 3.8783841784193635e-06,
      "loss": 0.8243,
      "step": 1533
    },
    {
      "epoch": 0.13834152500338187,
      "grad_norm": 1.808298239242894,
      "learning_rate": 3.8781834848981855e-06,
      "loss": 0.7829,
      "step": 1534
    },
    {
      "epoch": 0.13843170852685213,
      "grad_norm": 7.7301825363893215,
      "learning_rate": 3.877982631120037e-06,
      "loss": 0.7434,
      "step": 1535
    },
    {
      "epoch": 0.1385218920503224,
      "grad_norm": 2.414530983314309,
      "learning_rate": 3.877781617102053e-06,
      "loss": 0.8698,
      "step": 1536
    },
    {
      "epoch": 0.13861207557379268,
      "grad_norm": 1.8357123054209572,
      "learning_rate": 3.877580442861389e-06,
      "loss": 0.8399,
      "step": 1537
    },
    {
      "epoch": 0.13870225909726294,
      "grad_norm": 3.262791915052601,
      "learning_rate": 3.877379108415209e-06,
      "loss": 0.8351,
      "step": 1538
    },
    {
      "epoch": 0.1387924426207332,
      "grad_norm": 1.526318999034394,
      "learning_rate": 3.8771776137806915e-06,
      "loss": 0.8331,
      "step": 1539
    },
    {
      "epoch": 0.13888262614420346,
      "grad_norm": 2.123604821355983,
      "learning_rate": 3.8769759589750295e-06,
      "loss": 0.7554,
      "step": 1540
    },
    {
      "epoch": 0.13897280966767372,
      "grad_norm": 1.8494345110013461,
      "learning_rate": 3.876774144015429e-06,
      "loss": 0.807,
      "step": 1541
    },
    {
      "epoch": 0.13906299319114399,
      "grad_norm": 4.053700938955689,
      "learning_rate": 3.87657216891911e-06,
      "loss": 0.8464,
      "step": 1542
    },
    {
      "epoch": 0.13915317671461425,
      "grad_norm": 2.539195083677824,
      "learning_rate": 3.876370033703307e-06,
      "loss": 0.8029,
      "step": 1543
    },
    {
      "epoch": 0.1392433602380845,
      "grad_norm": 1.774498710229856,
      "learning_rate": 3.876167738385265e-06,
      "loss": 0.708,
      "step": 1544
    },
    {
      "epoch": 0.13933354376155477,
      "grad_norm": 1.8071990928429789,
      "learning_rate": 3.875965282982247e-06,
      "loss": 0.8532,
      "step": 1545
    },
    {
      "epoch": 0.13942372728502503,
      "grad_norm": 3.062037530351363,
      "learning_rate": 3.875762667511528e-06,
      "loss": 0.8241,
      "step": 1546
    },
    {
      "epoch": 0.1395139108084953,
      "grad_norm": 1.4819758250520885,
      "learning_rate": 3.875559891990394e-06,
      "loss": 0.8142,
      "step": 1547
    },
    {
      "epoch": 0.13960409433196555,
      "grad_norm": 1.7356522127407665,
      "learning_rate": 3.875356956436149e-06,
      "loss": 0.7723,
      "step": 1548
    },
    {
      "epoch": 0.1396942778554358,
      "grad_norm": 2.1743378431495746,
      "learning_rate": 3.875153860866108e-06,
      "loss": 0.7908,
      "step": 1549
    },
    {
      "epoch": 0.13978446137890607,
      "grad_norm": 2.797595991408738,
      "learning_rate": 3.8749506052976e-06,
      "loss": 0.8504,
      "step": 1550
    },
    {
      "epoch": 0.13987464490237633,
      "grad_norm": 1.8893835231981209,
      "learning_rate": 3.874747189747968e-06,
      "loss": 0.8344,
      "step": 1551
    },
    {
      "epoch": 0.1399648284258466,
      "grad_norm": 2.004724762972378,
      "learning_rate": 3.874543614234568e-06,
      "loss": 0.8177,
      "step": 1552
    },
    {
      "epoch": 0.14005501194931685,
      "grad_norm": 1.6241022711342015,
      "learning_rate": 3.874339878774771e-06,
      "loss": 0.7893,
      "step": 1553
    },
    {
      "epoch": 0.14014519547278712,
      "grad_norm": 1.7801095088747843,
      "learning_rate": 3.874135983385961e-06,
      "loss": 0.8135,
      "step": 1554
    },
    {
      "epoch": 0.14023537899625738,
      "grad_norm": 1.8205767918551263,
      "learning_rate": 3.873931928085535e-06,
      "loss": 0.8075,
      "step": 1555
    },
    {
      "epoch": 0.14032556251972764,
      "grad_norm": 2.299004741615692,
      "learning_rate": 3.873727712890904e-06,
      "loss": 0.8955,
      "step": 1556
    },
    {
      "epoch": 0.1404157460431979,
      "grad_norm": 2.0604355623814286,
      "learning_rate": 3.873523337819493e-06,
      "loss": 0.781,
      "step": 1557
    },
    {
      "epoch": 0.14050592956666816,
      "grad_norm": 1.6466302592130482,
      "learning_rate": 3.873318802888739e-06,
      "loss": 0.7371,
      "step": 1558
    },
    {
      "epoch": 0.14059611309013842,
      "grad_norm": 0.5891346995985618,
      "learning_rate": 3.873114108116097e-06,
      "loss": 0.5318,
      "step": 1559
    },
    {
      "epoch": 0.14068629661360868,
      "grad_norm": 2.0709798636938013,
      "learning_rate": 3.872909253519031e-06,
      "loss": 0.9014,
      "step": 1560
    },
    {
      "epoch": 0.14077648013707897,
      "grad_norm": 2.5259561635651333,
      "learning_rate": 3.8727042391150195e-06,
      "loss": 0.8835,
      "step": 1561
    },
    {
      "epoch": 0.14086666366054923,
      "grad_norm": 1.7806748247924944,
      "learning_rate": 3.872499064921556e-06,
      "loss": 0.7503,
      "step": 1562
    },
    {
      "epoch": 0.1409568471840195,
      "grad_norm": 1.9098411877501118,
      "learning_rate": 3.872293730956149e-06,
      "loss": 0.8248,
      "step": 1563
    },
    {
      "epoch": 0.14104703070748975,
      "grad_norm": 1.923005926582301,
      "learning_rate": 3.872088237236316e-06,
      "loss": 0.8066,
      "step": 1564
    },
    {
      "epoch": 0.14113721423096,
      "grad_norm": 1.9004497773706053,
      "learning_rate": 3.871882583779592e-06,
      "loss": 0.7169,
      "step": 1565
    },
    {
      "epoch": 0.14122739775443027,
      "grad_norm": 1.8803514406756652,
      "learning_rate": 3.871676770603525e-06,
      "loss": 0.8085,
      "step": 1566
    },
    {
      "epoch": 0.14131758127790053,
      "grad_norm": 6.051775383350521,
      "learning_rate": 3.871470797725676e-06,
      "loss": 0.8069,
      "step": 1567
    },
    {
      "epoch": 0.1414077648013708,
      "grad_norm": 1.7528946641913798,
      "learning_rate": 3.8712646651636185e-06,
      "loss": 0.881,
      "step": 1568
    },
    {
      "epoch": 0.14149794832484106,
      "grad_norm": 1.5106450489183125,
      "learning_rate": 3.871058372934942e-06,
      "loss": 0.8049,
      "step": 1569
    },
    {
      "epoch": 0.14158813184831132,
      "grad_norm": 2.3099148709022654,
      "learning_rate": 3.8708519210572485e-06,
      "loss": 0.6448,
      "step": 1570
    },
    {
      "epoch": 0.14167831537178158,
      "grad_norm": 1.8631509859306925,
      "learning_rate": 3.870645309548153e-06,
      "loss": 0.7547,
      "step": 1571
    },
    {
      "epoch": 0.14176849889525184,
      "grad_norm": 0.6076472315139877,
      "learning_rate": 3.870438538425284e-06,
      "loss": 0.5448,
      "step": 1572
    },
    {
      "epoch": 0.1418586824187221,
      "grad_norm": 2.2737179088372605,
      "learning_rate": 3.870231607706287e-06,
      "loss": 0.7616,
      "step": 1573
    },
    {
      "epoch": 0.14194886594219236,
      "grad_norm": 1.8930802599360896,
      "learning_rate": 3.870024517408817e-06,
      "loss": 0.8156,
      "step": 1574
    },
    {
      "epoch": 0.14203904946566262,
      "grad_norm": 3.083165722664464,
      "learning_rate": 3.8698172675505425e-06,
      "loss": 0.7368,
      "step": 1575
    },
    {
      "epoch": 0.14212923298913288,
      "grad_norm": 0.7479887459183716,
      "learning_rate": 3.86960985814915e-06,
      "loss": 0.5592,
      "step": 1576
    },
    {
      "epoch": 0.14221941651260314,
      "grad_norm": 1.7298153575861623,
      "learning_rate": 3.869402289222335e-06,
      "loss": 0.7395,
      "step": 1577
    },
    {
      "epoch": 0.1423096000360734,
      "grad_norm": 2.7093978159294694,
      "learning_rate": 3.869194560787808e-06,
      "loss": 0.7885,
      "step": 1578
    },
    {
      "epoch": 0.14239978355954366,
      "grad_norm": 1.7633901985796845,
      "learning_rate": 3.868986672863296e-06,
      "loss": 0.7369,
      "step": 1579
    },
    {
      "epoch": 0.14248996708301392,
      "grad_norm": 1.582426045453959,
      "learning_rate": 3.868778625466535e-06,
      "loss": 0.9078,
      "step": 1580
    },
    {
      "epoch": 0.14258015060648419,
      "grad_norm": 1.7607998703633754,
      "learning_rate": 3.868570418615278e-06,
      "loss": 0.7346,
      "step": 1581
    },
    {
      "epoch": 0.14267033412995445,
      "grad_norm": 0.7530433974634446,
      "learning_rate": 3.8683620523272885e-06,
      "loss": 0.5675,
      "step": 1582
    },
    {
      "epoch": 0.1427605176534247,
      "grad_norm": 2.0086319138214868,
      "learning_rate": 3.8681535266203464e-06,
      "loss": 0.9171,
      "step": 1583
    },
    {
      "epoch": 0.14285070117689497,
      "grad_norm": 2.0247773554346193,
      "learning_rate": 3.867944841512246e-06,
      "loss": 0.8331,
      "step": 1584
    },
    {
      "epoch": 0.14294088470036526,
      "grad_norm": 2.045024821869947,
      "learning_rate": 3.867735997020791e-06,
      "loss": 0.8107,
      "step": 1585
    },
    {
      "epoch": 0.14303106822383552,
      "grad_norm": 2.0254624297933534,
      "learning_rate": 3.867526993163802e-06,
      "loss": 0.7124,
      "step": 1586
    },
    {
      "epoch": 0.14312125174730578,
      "grad_norm": 1.6264720264964485,
      "learning_rate": 3.867317829959113e-06,
      "loss": 0.6844,
      "step": 1587
    },
    {
      "epoch": 0.14321143527077604,
      "grad_norm": 2.2828297827847766,
      "learning_rate": 3.8671085074245704e-06,
      "loss": 0.6068,
      "step": 1588
    },
    {
      "epoch": 0.1433016187942463,
      "grad_norm": 4.326427360895644,
      "learning_rate": 3.866899025578035e-06,
      "loss": 0.876,
      "step": 1589
    },
    {
      "epoch": 0.14339180231771656,
      "grad_norm": 1.8592424105004215,
      "learning_rate": 3.86668938443738e-06,
      "loss": 0.7739,
      "step": 1590
    },
    {
      "epoch": 0.14348198584118682,
      "grad_norm": 2.9173951102183566,
      "learning_rate": 3.866479584020495e-06,
      "loss": 0.8209,
      "step": 1591
    },
    {
      "epoch": 0.14357216936465708,
      "grad_norm": 1.8240502747333032,
      "learning_rate": 3.866269624345279e-06,
      "loss": 0.8407,
      "step": 1592
    },
    {
      "epoch": 0.14366235288812734,
      "grad_norm": 1.6823981126427323,
      "learning_rate": 3.866059505429649e-06,
      "loss": 0.8871,
      "step": 1593
    },
    {
      "epoch": 0.1437525364115976,
      "grad_norm": 0.6557160657725851,
      "learning_rate": 3.865849227291532e-06,
      "loss": 0.5951,
      "step": 1594
    },
    {
      "epoch": 0.14384271993506786,
      "grad_norm": 1.9429428804437021,
      "learning_rate": 3.865638789948872e-06,
      "loss": 0.7962,
      "step": 1595
    },
    {
      "epoch": 0.14393290345853813,
      "grad_norm": 2.027884997257801,
      "learning_rate": 3.865428193419622e-06,
      "loss": 0.762,
      "step": 1596
    },
    {
      "epoch": 0.14402308698200839,
      "grad_norm": 1.866326552582051,
      "learning_rate": 3.865217437721753e-06,
      "loss": 0.7969,
      "step": 1597
    },
    {
      "epoch": 0.14411327050547865,
      "grad_norm": 2.3466095836516856,
      "learning_rate": 3.865006522873249e-06,
      "loss": 0.7361,
      "step": 1598
    },
    {
      "epoch": 0.1442034540289489,
      "grad_norm": 1.572969532598193,
      "learning_rate": 3.864795448892103e-06,
      "loss": 0.8013,
      "step": 1599
    },
    {
      "epoch": 0.14429363755241917,
      "grad_norm": 2.109818475517216,
      "learning_rate": 3.864584215796327e-06,
      "loss": 0.8538,
      "step": 1600
    },
    {
      "epoch": 0.14438382107588943,
      "grad_norm": 1.7942813896972287,
      "learning_rate": 3.8643728236039455e-06,
      "loss": 0.8266,
      "step": 1601
    },
    {
      "epoch": 0.1444740045993597,
      "grad_norm": 2.3728027218150336,
      "learning_rate": 3.864161272332994e-06,
      "loss": 0.7867,
      "step": 1602
    },
    {
      "epoch": 0.14456418812282995,
      "grad_norm": 1.8474805558850056,
      "learning_rate": 3.863949562001524e-06,
      "loss": 0.8459,
      "step": 1603
    },
    {
      "epoch": 0.1446543716463002,
      "grad_norm": 1.7638473338975127,
      "learning_rate": 3.8637376926276005e-06,
      "loss": 0.7765,
      "step": 1604
    },
    {
      "epoch": 0.14474455516977047,
      "grad_norm": 7.806194232530734,
      "learning_rate": 3.8635256642293e-06,
      "loss": 0.8227,
      "step": 1605
    },
    {
      "epoch": 0.14483473869324073,
      "grad_norm": 1.9575592364316319,
      "learning_rate": 3.863313476824714e-06,
      "loss": 0.7833,
      "step": 1606
    },
    {
      "epoch": 0.144924922216711,
      "grad_norm": 1.9656699777378717,
      "learning_rate": 3.863101130431948e-06,
      "loss": 0.7886,
      "step": 1607
    },
    {
      "epoch": 0.14501510574018128,
      "grad_norm": 1.5966366823336908,
      "learning_rate": 3.862888625069121e-06,
      "loss": 0.8469,
      "step": 1608
    },
    {
      "epoch": 0.14510528926365154,
      "grad_norm": 1.6295535770444536,
      "learning_rate": 3.8626759607543645e-06,
      "loss": 0.7644,
      "step": 1609
    },
    {
      "epoch": 0.1451954727871218,
      "grad_norm": 1.8534868758678282,
      "learning_rate": 3.862463137505825e-06,
      "loss": 0.7677,
      "step": 1610
    },
    {
      "epoch": 0.14528565631059207,
      "grad_norm": 2.868099389858626,
      "learning_rate": 3.862250155341659e-06,
      "loss": 0.7707,
      "step": 1611
    },
    {
      "epoch": 0.14537583983406233,
      "grad_norm": 1.7445816900577948,
      "learning_rate": 3.862037014280043e-06,
      "loss": 0.9309,
      "step": 1612
    },
    {
      "epoch": 0.1454660233575326,
      "grad_norm": 1.7356376519461514,
      "learning_rate": 3.861823714339162e-06,
      "loss": 0.7999,
      "step": 1613
    },
    {
      "epoch": 0.14555620688100285,
      "grad_norm": 1.596755764983237,
      "learning_rate": 3.861610255537215e-06,
      "loss": 0.8152,
      "step": 1614
    },
    {
      "epoch": 0.1456463904044731,
      "grad_norm": 1.7439182733538785,
      "learning_rate": 3.8613966378924165e-06,
      "loss": 0.8393,
      "step": 1615
    },
    {
      "epoch": 0.14573657392794337,
      "grad_norm": 1.978753906843771,
      "learning_rate": 3.861182861422993e-06,
      "loss": 0.7515,
      "step": 1616
    },
    {
      "epoch": 0.14582675745141363,
      "grad_norm": 1.667485473089123,
      "learning_rate": 3.860968926147185e-06,
      "loss": 0.8772,
      "step": 1617
    },
    {
      "epoch": 0.1459169409748839,
      "grad_norm": 1.4660127984824454,
      "learning_rate": 3.860754832083247e-06,
      "loss": 0.6391,
      "step": 1618
    },
    {
      "epoch": 0.14600712449835415,
      "grad_norm": 2.2755596310960677,
      "learning_rate": 3.8605405792494475e-06,
      "loss": 0.8741,
      "step": 1619
    },
    {
      "epoch": 0.1460973080218244,
      "grad_norm": 1.891321038363625,
      "learning_rate": 3.860326167664066e-06,
      "loss": 0.8204,
      "step": 1620
    },
    {
      "epoch": 0.14618749154529467,
      "grad_norm": 1.7403773325075875,
      "learning_rate": 3.860111597345399e-06,
      "loss": 0.7913,
      "step": 1621
    },
    {
      "epoch": 0.14627767506876493,
      "grad_norm": 2.0464002917819677,
      "learning_rate": 3.859896868311753e-06,
      "loss": 0.6642,
      "step": 1622
    },
    {
      "epoch": 0.1463678585922352,
      "grad_norm": 1.5958787501606233,
      "learning_rate": 3.859681980581452e-06,
      "loss": 0.8953,
      "step": 1623
    },
    {
      "epoch": 0.14645804211570546,
      "grad_norm": 1.7413679170903285,
      "learning_rate": 3.859466934172829e-06,
      "loss": 0.777,
      "step": 1624
    },
    {
      "epoch": 0.14654822563917572,
      "grad_norm": 1.8630739492583952,
      "learning_rate": 3.859251729104235e-06,
      "loss": 0.7335,
      "step": 1625
    },
    {
      "epoch": 0.14663840916264598,
      "grad_norm": 1.5994150791311719,
      "learning_rate": 3.859036365394031e-06,
      "loss": 0.7554,
      "step": 1626
    },
    {
      "epoch": 0.14672859268611624,
      "grad_norm": 1.9192458366377705,
      "learning_rate": 3.858820843060594e-06,
      "loss": 0.7868,
      "step": 1627
    },
    {
      "epoch": 0.1468187762095865,
      "grad_norm": 1.6663639826897922,
      "learning_rate": 3.858605162122314e-06,
      "loss": 0.7769,
      "step": 1628
    },
    {
      "epoch": 0.14690895973305676,
      "grad_norm": 1.7586827137431638,
      "learning_rate": 3.858389322597592e-06,
      "loss": 0.76,
      "step": 1629
    },
    {
      "epoch": 0.14699914325652702,
      "grad_norm": 2.1273422795969026,
      "learning_rate": 3.858173324504847e-06,
      "loss": 0.819,
      "step": 1630
    },
    {
      "epoch": 0.14708932677999728,
      "grad_norm": 3.5869919633182437,
      "learning_rate": 3.857957167862508e-06,
      "loss": 0.7645,
      "step": 1631
    },
    {
      "epoch": 0.14717951030346757,
      "grad_norm": 0.6742616889514658,
      "learning_rate": 3.857740852689018e-06,
      "loss": 0.58,
      "step": 1632
    },
    {
      "epoch": 0.14726969382693783,
      "grad_norm": 3.4033692103284245,
      "learning_rate": 3.857524379002835e-06,
      "loss": 0.8055,
      "step": 1633
    },
    {
      "epoch": 0.1473598773504081,
      "grad_norm": 2.652870901136526,
      "learning_rate": 3.85730774682243e-06,
      "loss": 0.8199,
      "step": 1634
    },
    {
      "epoch": 0.14745006087387835,
      "grad_norm": 1.4536590415202357,
      "learning_rate": 3.8570909561662875e-06,
      "loss": 0.86,
      "step": 1635
    },
    {
      "epoch": 0.1475402443973486,
      "grad_norm": 1.5380859375,
      "learning_rate": 3.8568740070529045e-06,
      "loss": 0.862,
      "step": 1636
    },
    {
      "epoch": 0.14763042792081887,
      "grad_norm": 1.7113963169494706,
      "learning_rate": 3.856656899500792e-06,
      "loss": 0.7676,
      "step": 1637
    },
    {
      "epoch": 0.14772061144428913,
      "grad_norm": 1.73117485020997,
      "learning_rate": 3.856439633528476e-06,
      "loss": 0.7313,
      "step": 1638
    },
    {
      "epoch": 0.1478107949677594,
      "grad_norm": 2.904330696570889,
      "learning_rate": 3.856222209154494e-06,
      "loss": 0.8551,
      "step": 1639
    },
    {
      "epoch": 0.14790097849122966,
      "grad_norm": 1.7271673601583426,
      "learning_rate": 3.856004626397397e-06,
      "loss": 0.8248,
      "step": 1640
    },
    {
      "epoch": 0.14799116201469992,
      "grad_norm": 1.787106306563281,
      "learning_rate": 3.855786885275753e-06,
      "loss": 0.7765,
      "step": 1641
    },
    {
      "epoch": 0.14808134553817018,
      "grad_norm": 3.389729324382424,
      "learning_rate": 3.855568985808138e-06,
      "loss": 0.7184,
      "step": 1642
    },
    {
      "epoch": 0.14817152906164044,
      "grad_norm": 3.1304959605638283,
      "learning_rate": 3.855350928013145e-06,
      "loss": 0.7724,
      "step": 1643
    },
    {
      "epoch": 0.1482617125851107,
      "grad_norm": 2.1135707240945383,
      "learning_rate": 3.8551327119093825e-06,
      "loss": 0.8019,
      "step": 1644
    },
    {
      "epoch": 0.14835189610858096,
      "grad_norm": 1.9394568129213425,
      "learning_rate": 3.854914337515467e-06,
      "loss": 0.7313,
      "step": 1645
    },
    {
      "epoch": 0.14844207963205122,
      "grad_norm": 1.7108423912055069,
      "learning_rate": 3.8546958048500324e-06,
      "loss": 0.6957,
      "step": 1646
    },
    {
      "epoch": 0.14853226315552148,
      "grad_norm": 0.6857827805010739,
      "learning_rate": 3.854477113931725e-06,
      "loss": 0.6321,
      "step": 1647
    },
    {
      "epoch": 0.14862244667899174,
      "grad_norm": 2.0107200143340944,
      "learning_rate": 3.854258264779205e-06,
      "loss": 0.821,
      "step": 1648
    },
    {
      "epoch": 0.148712630202462,
      "grad_norm": 1.7287125814695323,
      "learning_rate": 3.854039257411145e-06,
      "loss": 0.7774,
      "step": 1649
    },
    {
      "epoch": 0.14880281372593226,
      "grad_norm": 2.2272274697470507,
      "learning_rate": 3.853820091846232e-06,
      "loss": 0.7503,
      "step": 1650
    },
    {
      "epoch": 0.14889299724940253,
      "grad_norm": 1.9317987915330697,
      "learning_rate": 3.853600768103169e-06,
      "loss": 0.8006,
      "step": 1651
    },
    {
      "epoch": 0.1489831807728728,
      "grad_norm": 2.215092182523155,
      "learning_rate": 3.853381286200667e-06,
      "loss": 0.784,
      "step": 1652
    },
    {
      "epoch": 0.14907336429634305,
      "grad_norm": 1.854954248671297,
      "learning_rate": 3.853161646157453e-06,
      "loss": 0.8907,
      "step": 1653
    },
    {
      "epoch": 0.1491635478198133,
      "grad_norm": 1.691011096351712,
      "learning_rate": 3.852941847992269e-06,
      "loss": 0.768,
      "step": 1654
    },
    {
      "epoch": 0.14925373134328357,
      "grad_norm": 3.6071302691594775,
      "learning_rate": 3.852721891723871e-06,
      "loss": 0.8145,
      "step": 1655
    },
    {
      "epoch": 0.14934391486675386,
      "grad_norm": 1.8408771698364474,
      "learning_rate": 3.852501777371025e-06,
      "loss": 0.7836,
      "step": 1656
    },
    {
      "epoch": 0.14943409839022412,
      "grad_norm": 1.8847071188903521,
      "learning_rate": 3.8522815049525125e-06,
      "loss": 0.8369,
      "step": 1657
    },
    {
      "epoch": 0.14952428191369438,
      "grad_norm": 2.2319135646217636,
      "learning_rate": 3.852061074487129e-06,
      "loss": 0.8503,
      "step": 1658
    },
    {
      "epoch": 0.14961446543716464,
      "grad_norm": 2.089475110776845,
      "learning_rate": 3.851840485993682e-06,
      "loss": 0.7068,
      "step": 1659
    },
    {
      "epoch": 0.1497046489606349,
      "grad_norm": 2.0466673068605763,
      "learning_rate": 3.851619739490994e-06,
      "loss": 0.8054,
      "step": 1660
    },
    {
      "epoch": 0.14979483248410516,
      "grad_norm": 4.575070348313411,
      "learning_rate": 3.8513988349978996e-06,
      "loss": 0.8249,
      "step": 1661
    },
    {
      "epoch": 0.14988501600757542,
      "grad_norm": 3.0423464255022163,
      "learning_rate": 3.851177772533249e-06,
      "loss": 0.8226,
      "step": 1662
    },
    {
      "epoch": 0.14997519953104568,
      "grad_norm": 2.285696216920552,
      "learning_rate": 3.850956552115903e-06,
      "loss": 0.9791,
      "step": 1663
    },
    {
      "epoch": 0.15006538305451594,
      "grad_norm": 0.7082651236360519,
      "learning_rate": 3.850735173764738e-06,
      "loss": 0.5829,
      "step": 1664
    },
    {
      "epoch": 0.1501555665779862,
      "grad_norm": 2.6729860728393744,
      "learning_rate": 3.850513637498642e-06,
      "loss": 0.8201,
      "step": 1665
    },
    {
      "epoch": 0.15024575010145647,
      "grad_norm": 1.812281628312052,
      "learning_rate": 3.850291943336521e-06,
      "loss": 0.8225,
      "step": 1666
    },
    {
      "epoch": 0.15033593362492673,
      "grad_norm": 1.9694417918799716,
      "learning_rate": 3.850070091297287e-06,
      "loss": 0.8547,
      "step": 1667
    },
    {
      "epoch": 0.150426117148397,
      "grad_norm": 0.7381297768511749,
      "learning_rate": 3.8498480813998735e-06,
      "loss": 0.6744,
      "step": 1668
    },
    {
      "epoch": 0.15051630067186725,
      "grad_norm": 1.8278278207096077,
      "learning_rate": 3.84962591366322e-06,
      "loss": 0.7956,
      "step": 1669
    },
    {
      "epoch": 0.1506064841953375,
      "grad_norm": 2.1738362410703163,
      "learning_rate": 3.8494035881062855e-06,
      "loss": 0.8518,
      "step": 1670
    },
    {
      "epoch": 0.15069666771880777,
      "grad_norm": 6.2888001570865235,
      "learning_rate": 3.84918110474804e-06,
      "loss": 0.7786,
      "step": 1671
    },
    {
      "epoch": 0.15078685124227803,
      "grad_norm": 1.6916943452321491,
      "learning_rate": 3.8489584636074655e-06,
      "loss": 0.7871,
      "step": 1672
    },
    {
      "epoch": 0.1508770347657483,
      "grad_norm": 0.7054585833149604,
      "learning_rate": 3.848735664703561e-06,
      "loss": 0.5664,
      "step": 1673
    },
    {
      "epoch": 0.15096721828921855,
      "grad_norm": 1.8779648545930432,
      "learning_rate": 3.8485127080553346e-06,
      "loss": 0.7956,
      "step": 1674
    },
    {
      "epoch": 0.1510574018126888,
      "grad_norm": 2.1470687667714805,
      "learning_rate": 3.8482895936818115e-06,
      "loss": 0.7987,
      "step": 1675
    },
    {
      "epoch": 0.15114758533615907,
      "grad_norm": 2.0342541121091675,
      "learning_rate": 3.848066321602029e-06,
      "loss": 0.8141,
      "step": 1676
    },
    {
      "epoch": 0.15123776885962933,
      "grad_norm": 1.9203233735635687,
      "learning_rate": 3.847842891835038e-06,
      "loss": 0.7649,
      "step": 1677
    },
    {
      "epoch": 0.1513279523830996,
      "grad_norm": 0.6318773023221929,
      "learning_rate": 3.847619304399902e-06,
      "loss": 0.5813,
      "step": 1678
    },
    {
      "epoch": 0.15141813590656986,
      "grad_norm": 2.167835018397892,
      "learning_rate": 3.8473955593157e-06,
      "loss": 0.8252,
      "step": 1679
    },
    {
      "epoch": 0.15150831943004014,
      "grad_norm": 2.16077236336184,
      "learning_rate": 3.847171656601522e-06,
      "loss": 0.763,
      "step": 1680
    },
    {
      "epoch": 0.1515985029535104,
      "grad_norm": 1.78990067083861,
      "learning_rate": 3.846947596276473e-06,
      "loss": 0.8284,
      "step": 1681
    },
    {
      "epoch": 0.15168868647698067,
      "grad_norm": 4.164387244753732,
      "learning_rate": 3.846723378359672e-06,
      "loss": 0.8448,
      "step": 1682
    },
    {
      "epoch": 0.15177887000045093,
      "grad_norm": 2.08790227246225,
      "learning_rate": 3.846499002870249e-06,
      "loss": 0.8072,
      "step": 1683
    },
    {
      "epoch": 0.1518690535239212,
      "grad_norm": 2.3634450808399343,
      "learning_rate": 3.846274469827349e-06,
      "loss": 0.8174,
      "step": 1684
    },
    {
      "epoch": 0.15195923704739145,
      "grad_norm": 2.0331346425186343,
      "learning_rate": 3.846049779250132e-06,
      "loss": 0.8089,
      "step": 1685
    },
    {
      "epoch": 0.1520494205708617,
      "grad_norm": 0.7878108152971485,
      "learning_rate": 3.845824931157769e-06,
      "loss": 0.6239,
      "step": 1686
    },
    {
      "epoch": 0.15213960409433197,
      "grad_norm": 1.7414825107858318,
      "learning_rate": 3.845599925569444e-06,
      "loss": 0.7906,
      "step": 1687
    },
    {
      "epoch": 0.15222978761780223,
      "grad_norm": 2.2498895829976084,
      "learning_rate": 3.845374762504357e-06,
      "loss": 0.7636,
      "step": 1688
    },
    {
      "epoch": 0.1523199711412725,
      "grad_norm": 2.23294950242436,
      "learning_rate": 3.8451494419817204e-06,
      "loss": 0.8573,
      "step": 1689
    },
    {
      "epoch": 0.15241015466474275,
      "grad_norm": 2.42841671402768,
      "learning_rate": 3.8449239640207594e-06,
      "loss": 0.7376,
      "step": 1690
    },
    {
      "epoch": 0.152500338188213,
      "grad_norm": 1.9063072196002087,
      "learning_rate": 3.844698328640713e-06,
      "loss": 0.8433,
      "step": 1691
    },
    {
      "epoch": 0.15259052171168327,
      "grad_norm": 1.5615535921640862,
      "learning_rate": 3.844472535860833e-06,
      "loss": 0.8164,
      "step": 1692
    },
    {
      "epoch": 0.15268070523515354,
      "grad_norm": 2.1528141456184624,
      "learning_rate": 3.8442465857003864e-06,
      "loss": 0.8478,
      "step": 1693
    },
    {
      "epoch": 0.1527708887586238,
      "grad_norm": 1.6014354376282933,
      "learning_rate": 3.844020478178653e-06,
      "loss": 0.7427,
      "step": 1694
    },
    {
      "epoch": 0.15286107228209406,
      "grad_norm": 2.031276761392221,
      "learning_rate": 3.843794213314923e-06,
      "loss": 0.8314,
      "step": 1695
    },
    {
      "epoch": 0.15295125580556432,
      "grad_norm": 2.03025359143589,
      "learning_rate": 3.843567791128505e-06,
      "loss": 0.8286,
      "step": 1696
    },
    {
      "epoch": 0.15304143932903458,
      "grad_norm": 2.365930809099206,
      "learning_rate": 3.843341211638717e-06,
      "loss": 0.7857,
      "step": 1697
    },
    {
      "epoch": 0.15313162285250484,
      "grad_norm": 1.7559715294582843,
      "learning_rate": 3.843114474864894e-06,
      "loss": 0.7606,
      "step": 1698
    },
    {
      "epoch": 0.1532218063759751,
      "grad_norm": 2.0138008083798407,
      "learning_rate": 3.84288758082638e-06,
      "loss": 0.7965,
      "step": 1699
    },
    {
      "epoch": 0.15331198989944536,
      "grad_norm": 2.189397370498847,
      "learning_rate": 3.842660529542536e-06,
      "loss": 0.8866,
      "step": 1700
    },
    {
      "epoch": 0.15340217342291562,
      "grad_norm": 1.6378396504624104,
      "learning_rate": 3.842433321032736e-06,
      "loss": 0.8143,
      "step": 1701
    },
    {
      "epoch": 0.15349235694638588,
      "grad_norm": 1.7441063820957885,
      "learning_rate": 3.842205955316365e-06,
      "loss": 0.7216,
      "step": 1702
    },
    {
      "epoch": 0.15358254046985614,
      "grad_norm": 1.7867959670114189,
      "learning_rate": 3.8419784324128256e-06,
      "loss": 0.704,
      "step": 1703
    },
    {
      "epoch": 0.15367272399332643,
      "grad_norm": 1.6848333977083454,
      "learning_rate": 3.841750752341529e-06,
      "loss": 0.7608,
      "step": 1704
    },
    {
      "epoch": 0.1537629075167967,
      "grad_norm": 1.8034938811771333,
      "learning_rate": 3.841522915121902e-06,
      "loss": 0.8093,
      "step": 1705
    },
    {
      "epoch": 0.15385309104026695,
      "grad_norm": 2.2249951651874262,
      "learning_rate": 3.841294920773387e-06,
      "loss": 0.733,
      "step": 1706
    },
    {
      "epoch": 0.15394327456373721,
      "grad_norm": 2.2382438616600813,
      "learning_rate": 3.841066769315436e-06,
      "loss": 0.8155,
      "step": 1707
    },
    {
      "epoch": 0.15403345808720748,
      "grad_norm": 3.4338871783839418,
      "learning_rate": 3.840838460767518e-06,
      "loss": 0.8107,
      "step": 1708
    },
    {
      "epoch": 0.15412364161067774,
      "grad_norm": 1.7243742042641983,
      "learning_rate": 3.840609995149111e-06,
      "loss": 0.7748,
      "step": 1709
    },
    {
      "epoch": 0.154213825134148,
      "grad_norm": 1.8872140427788573,
      "learning_rate": 3.84038137247971e-06,
      "loss": 0.8085,
      "step": 1710
    },
    {
      "epoch": 0.15430400865761826,
      "grad_norm": 2.067059995794655,
      "learning_rate": 3.840152592778823e-06,
      "loss": 0.7862,
      "step": 1711
    },
    {
      "epoch": 0.15439419218108852,
      "grad_norm": 1.5999776302204387,
      "learning_rate": 3.83992365606597e-06,
      "loss": 0.8427,
      "step": 1712
    },
    {
      "epoch": 0.15448437570455878,
      "grad_norm": 1.7925447116313813,
      "learning_rate": 3.8396945623606855e-06,
      "loss": 0.7536,
      "step": 1713
    },
    {
      "epoch": 0.15457455922802904,
      "grad_norm": 1.7770677960138828,
      "learning_rate": 3.8394653116825174e-06,
      "loss": 0.7611,
      "step": 1714
    },
    {
      "epoch": 0.1546647427514993,
      "grad_norm": 1.6047805833423483,
      "learning_rate": 3.839235904051026e-06,
      "loss": 0.789,
      "step": 1715
    },
    {
      "epoch": 0.15475492627496956,
      "grad_norm": 1.943017241298888,
      "learning_rate": 3.8390063394857855e-06,
      "loss": 0.7796,
      "step": 1716
    },
    {
      "epoch": 0.15484510979843982,
      "grad_norm": 1.704465985613107,
      "learning_rate": 3.838776618006385e-06,
      "loss": 0.884,
      "step": 1717
    },
    {
      "epoch": 0.15493529332191008,
      "grad_norm": 1.8003475966208435,
      "learning_rate": 3.838546739632423e-06,
      "loss": 0.8102,
      "step": 1718
    },
    {
      "epoch": 0.15502547684538034,
      "grad_norm": 2.4193470220542066,
      "learning_rate": 3.838316704383517e-06,
      "loss": 0.8436,
      "step": 1719
    },
    {
      "epoch": 0.1551156603688506,
      "grad_norm": 1.732424971482048,
      "learning_rate": 3.838086512279292e-06,
      "loss": 0.8028,
      "step": 1720
    },
    {
      "epoch": 0.15520584389232087,
      "grad_norm": 2.1112498675584046,
      "learning_rate": 3.837856163339391e-06,
      "loss": 0.7995,
      "step": 1721
    },
    {
      "epoch": 0.15529602741579113,
      "grad_norm": 1.490724095663406,
      "learning_rate": 3.837625657583469e-06,
      "loss": 0.7283,
      "step": 1722
    },
    {
      "epoch": 0.1553862109392614,
      "grad_norm": 1.6237927867723083,
      "learning_rate": 3.837394995031193e-06,
      "loss": 0.7624,
      "step": 1723
    },
    {
      "epoch": 0.15547639446273165,
      "grad_norm": 1.8141619856281486,
      "learning_rate": 3.837164175702245e-06,
      "loss": 0.7998,
      "step": 1724
    },
    {
      "epoch": 0.1555665779862019,
      "grad_norm": 1.9543681346551922,
      "learning_rate": 3.836933199616319e-06,
      "loss": 0.7141,
      "step": 1725
    },
    {
      "epoch": 0.15565676150967217,
      "grad_norm": 2.6985180355174014,
      "learning_rate": 3.836702066793124e-06,
      "loss": 0.8678,
      "step": 1726
    },
    {
      "epoch": 0.15574694503314246,
      "grad_norm": 1.996724008693109,
      "learning_rate": 3.836470777252381e-06,
      "loss": 0.8546,
      "step": 1727
    },
    {
      "epoch": 0.15583712855661272,
      "grad_norm": 2.747712397605032,
      "learning_rate": 3.836239331013825e-06,
      "loss": 0.8153,
      "step": 1728
    },
    {
      "epoch": 0.15592731208008298,
      "grad_norm": 2.3557016437842013,
      "learning_rate": 3.836007728097205e-06,
      "loss": 0.769,
      "step": 1729
    },
    {
      "epoch": 0.15601749560355324,
      "grad_norm": 1.7190517507444827,
      "learning_rate": 3.835775968522282e-06,
      "loss": 0.7499,
      "step": 1730
    },
    {
      "epoch": 0.1561076791270235,
      "grad_norm": 3.352444619354999,
      "learning_rate": 3.83554405230883e-06,
      "loss": 0.8227,
      "step": 1731
    },
    {
      "epoch": 0.15619786265049376,
      "grad_norm": 2.3304357134388916,
      "learning_rate": 3.835311979476639e-06,
      "loss": 0.7795,
      "step": 1732
    },
    {
      "epoch": 0.15628804617396402,
      "grad_norm": 4.098462602754447,
      "learning_rate": 3.83507975004551e-06,
      "loss": 0.6895,
      "step": 1733
    },
    {
      "epoch": 0.15637822969743428,
      "grad_norm": 1.6242123675714095,
      "learning_rate": 3.834847364035258e-06,
      "loss": 0.773,
      "step": 1734
    },
    {
      "epoch": 0.15646841322090455,
      "grad_norm": 6.976503720094625,
      "learning_rate": 3.834614821465712e-06,
      "loss": 0.7433,
      "step": 1735
    },
    {
      "epoch": 0.1565585967443748,
      "grad_norm": 2.0652132394343004,
      "learning_rate": 3.834382122356713e-06,
      "loss": 0.8335,
      "step": 1736
    },
    {
      "epoch": 0.15664878026784507,
      "grad_norm": 1.4886943414054372,
      "learning_rate": 3.834149266728117e-06,
      "loss": 0.7595,
      "step": 1737
    },
    {
      "epoch": 0.15673896379131533,
      "grad_norm": 0.6909105711631911,
      "learning_rate": 3.833916254599792e-06,
      "loss": 0.6201,
      "step": 1738
    },
    {
      "epoch": 0.1568291473147856,
      "grad_norm": 0.6437938749305808,
      "learning_rate": 3.83368308599162e-06,
      "loss": 0.5663,
      "step": 1739
    },
    {
      "epoch": 0.15691933083825585,
      "grad_norm": 0.5789914211368956,
      "learning_rate": 3.833449760923498e-06,
      "loss": 0.5382,
      "step": 1740
    },
    {
      "epoch": 0.1570095143617261,
      "grad_norm": 1.6597318937196306,
      "learning_rate": 3.83321627941533e-06,
      "loss": 0.8359,
      "step": 1741
    },
    {
      "epoch": 0.15709969788519637,
      "grad_norm": 2.6669633124180647,
      "learning_rate": 3.832982641487043e-06,
      "loss": 0.851,
      "step": 1742
    },
    {
      "epoch": 0.15718988140866663,
      "grad_norm": 1.7966746342941766,
      "learning_rate": 3.832748847158568e-06,
      "loss": 0.8869,
      "step": 1743
    },
    {
      "epoch": 0.1572800649321369,
      "grad_norm": 1.589728573837862,
      "learning_rate": 3.832514896449858e-06,
      "loss": 0.8169,
      "step": 1744
    },
    {
      "epoch": 0.15737024845560715,
      "grad_norm": 1.7838221934625111,
      "learning_rate": 3.832280789380871e-06,
      "loss": 0.7319,
      "step": 1745
    },
    {
      "epoch": 0.15746043197907741,
      "grad_norm": 1.8994087278001883,
      "learning_rate": 3.832046525971584e-06,
      "loss": 0.7017,
      "step": 1746
    },
    {
      "epoch": 0.15755061550254768,
      "grad_norm": 2.0248799135369993,
      "learning_rate": 3.831812106241987e-06,
      "loss": 0.8551,
      "step": 1747
    },
    {
      "epoch": 0.15764079902601794,
      "grad_norm": 1.6246801941962905,
      "learning_rate": 3.8315775302120796e-06,
      "loss": 0.6985,
      "step": 1748
    },
    {
      "epoch": 0.1577309825494882,
      "grad_norm": 1.5818803126540966,
      "learning_rate": 3.831342797901878e-06,
      "loss": 0.8087,
      "step": 1749
    },
    {
      "epoch": 0.15782116607295846,
      "grad_norm": 0.6316044897950266,
      "learning_rate": 3.831107909331411e-06,
      "loss": 0.5787,
      "step": 1750
    },
    {
      "epoch": 0.15791134959642875,
      "grad_norm": 3.8834743741554005,
      "learning_rate": 3.830872864520721e-06,
      "loss": 0.8388,
      "step": 1751
    },
    {
      "epoch": 0.158001533119899,
      "grad_norm": 0.7127983974214,
      "learning_rate": 3.830637663489862e-06,
      "loss": 0.6589,
      "step": 1752
    },
    {
      "epoch": 0.15809171664336927,
      "grad_norm": 0.7746234625123497,
      "learning_rate": 3.830402306258904e-06,
      "loss": 0.6892,
      "step": 1753
    },
    {
      "epoch": 0.15818190016683953,
      "grad_norm": 3.5548141289258623,
      "learning_rate": 3.830166792847929e-06,
      "loss": 0.7796,
      "step": 1754
    },
    {
      "epoch": 0.1582720836903098,
      "grad_norm": 1.8337207153727921,
      "learning_rate": 3.829931123277031e-06,
      "loss": 0.733,
      "step": 1755
    },
    {
      "epoch": 0.15836226721378005,
      "grad_norm": 1.7608866619093029,
      "learning_rate": 3.8296952975663204e-06,
      "loss": 0.7865,
      "step": 1756
    },
    {
      "epoch": 0.1584524507372503,
      "grad_norm": 1.6341369222099429,
      "learning_rate": 3.829459315735918e-06,
      "loss": 0.8594,
      "step": 1757
    },
    {
      "epoch": 0.15854263426072057,
      "grad_norm": 1.7734146452788067,
      "learning_rate": 3.829223177805959e-06,
      "loss": 0.7182,
      "step": 1758
    },
    {
      "epoch": 0.15863281778419083,
      "grad_norm": 1.738532161819992,
      "learning_rate": 3.828986883796591e-06,
      "loss": 0.8322,
      "step": 1759
    },
    {
      "epoch": 0.1587230013076611,
      "grad_norm": 2.255975735128056,
      "learning_rate": 3.828750433727979e-06,
      "loss": 0.7874,
      "step": 1760
    },
    {
      "epoch": 0.15881318483113135,
      "grad_norm": 1.7718762486905872,
      "learning_rate": 3.828513827620296e-06,
      "loss": 0.8437,
      "step": 1761
    },
    {
      "epoch": 0.15890336835460162,
      "grad_norm": 2.001303248176575,
      "learning_rate": 3.82827706549373e-06,
      "loss": 0.8505,
      "step": 1762
    },
    {
      "epoch": 0.15899355187807188,
      "grad_norm": 2.4985961787331465,
      "learning_rate": 3.828040147368484e-06,
      "loss": 0.7882,
      "step": 1763
    },
    {
      "epoch": 0.15908373540154214,
      "grad_norm": 4.060957395674377,
      "learning_rate": 3.827803073264774e-06,
      "loss": 0.7918,
      "step": 1764
    },
    {
      "epoch": 0.1591739189250124,
      "grad_norm": 1.8157766912390134,
      "learning_rate": 3.827565843202826e-06,
      "loss": 0.7927,
      "step": 1765
    },
    {
      "epoch": 0.15926410244848266,
      "grad_norm": 2.070283939956308,
      "learning_rate": 3.827328457202884e-06,
      "loss": 0.6664,
      "step": 1766
    },
    {
      "epoch": 0.15935428597195292,
      "grad_norm": 1.4993601069917235,
      "learning_rate": 3.8270909152852014e-06,
      "loss": 0.7257,
      "step": 1767
    },
    {
      "epoch": 0.15944446949542318,
      "grad_norm": 2.024705879609624,
      "learning_rate": 3.826853217470048e-06,
      "loss": 0.7897,
      "step": 1768
    },
    {
      "epoch": 0.15953465301889344,
      "grad_norm": 1.7037628799141948,
      "learning_rate": 3.826615363777705e-06,
      "loss": 0.8759,
      "step": 1769
    },
    {
      "epoch": 0.1596248365423637,
      "grad_norm": 1.8805649508127773,
      "learning_rate": 3.826377354228468e-06,
      "loss": 0.7732,
      "step": 1770
    },
    {
      "epoch": 0.15971502006583396,
      "grad_norm": 0.6481596799674829,
      "learning_rate": 3.826139188842643e-06,
      "loss": 0.5921,
      "step": 1771
    },
    {
      "epoch": 0.15980520358930422,
      "grad_norm": 1.474433292208436,
      "learning_rate": 3.825900867640554e-06,
      "loss": 0.8473,
      "step": 1772
    },
    {
      "epoch": 0.15989538711277448,
      "grad_norm": 1.4320446656842694,
      "learning_rate": 3.825662390642535e-06,
      "loss": 0.8312,
      "step": 1773
    },
    {
      "epoch": 0.15998557063624474,
      "grad_norm": 1.6201145385941274,
      "learning_rate": 3.825423757868934e-06,
      "loss": 0.8974,
      "step": 1774
    },
    {
      "epoch": 0.16007575415971503,
      "grad_norm": 2.1085834289501713,
      "learning_rate": 3.825184969340114e-06,
      "loss": 0.8238,
      "step": 1775
    },
    {
      "epoch": 0.1601659376831853,
      "grad_norm": 2.0868410589491866,
      "learning_rate": 3.824946025076447e-06,
      "loss": 0.8184,
      "step": 1776
    },
    {
      "epoch": 0.16025612120665556,
      "grad_norm": 1.8955091331984386,
      "learning_rate": 3.824706925098323e-06,
      "loss": 0.7438,
      "step": 1777
    },
    {
      "epoch": 0.16034630473012582,
      "grad_norm": 2.3513414652919655,
      "learning_rate": 3.824467669426143e-06,
      "loss": 0.8646,
      "step": 1778
    },
    {
      "epoch": 0.16043648825359608,
      "grad_norm": 1.5818040472702588,
      "learning_rate": 3.824228258080321e-06,
      "loss": 0.7678,
      "step": 1779
    },
    {
      "epoch": 0.16052667177706634,
      "grad_norm": 1.7089169908590436,
      "learning_rate": 3.823988691081285e-06,
      "loss": 0.711,
      "step": 1780
    },
    {
      "epoch": 0.1606168553005366,
      "grad_norm": 2.4618473369267337,
      "learning_rate": 3.823748968449478e-06,
      "loss": 0.8252,
      "step": 1781
    },
    {
      "epoch": 0.16070703882400686,
      "grad_norm": 2.4408416339302903,
      "learning_rate": 3.823509090205352e-06,
      "loss": 0.6814,
      "step": 1782
    },
    {
      "epoch": 0.16079722234747712,
      "grad_norm": 2.51752253456978,
      "learning_rate": 3.823269056369376e-06,
      "loss": 0.7506,
      "step": 1783
    },
    {
      "epoch": 0.16088740587094738,
      "grad_norm": 1.8287774250427713,
      "learning_rate": 3.8230288669620295e-06,
      "loss": 0.8252,
      "step": 1784
    },
    {
      "epoch": 0.16097758939441764,
      "grad_norm": 1.6823726748713945,
      "learning_rate": 3.822788522003809e-06,
      "loss": 0.8021,
      "step": 1785
    },
    {
      "epoch": 0.1610677729178879,
      "grad_norm": 2.290850152871767,
      "learning_rate": 3.822548021515221e-06,
      "loss": 0.8725,
      "step": 1786
    },
    {
      "epoch": 0.16115795644135816,
      "grad_norm": 1.705370411503829,
      "learning_rate": 3.822307365516787e-06,
      "loss": 0.7691,
      "step": 1787
    },
    {
      "epoch": 0.16124813996482842,
      "grad_norm": 1.5950354086075689,
      "learning_rate": 3.8220665540290395e-06,
      "loss": 0.8146,
      "step": 1788
    },
    {
      "epoch": 0.16133832348829868,
      "grad_norm": 1.5362416863585104,
      "learning_rate": 3.8218255870725265e-06,
      "loss": 0.8567,
      "step": 1789
    },
    {
      "epoch": 0.16142850701176895,
      "grad_norm": 2.6467964241768116,
      "learning_rate": 3.82158446466781e-06,
      "loss": 0.7372,
      "step": 1790
    },
    {
      "epoch": 0.1615186905352392,
      "grad_norm": 1.5630187127279989,
      "learning_rate": 3.821343186835462e-06,
      "loss": 0.8074,
      "step": 1791
    },
    {
      "epoch": 0.16160887405870947,
      "grad_norm": 1.847700961259527,
      "learning_rate": 3.821101753596072e-06,
      "loss": 0.8624,
      "step": 1792
    },
    {
      "epoch": 0.16169905758217973,
      "grad_norm": 3.4751655429406934,
      "learning_rate": 3.820860164970237e-06,
      "loss": 0.7871,
      "step": 1793
    },
    {
      "epoch": 0.16178924110565,
      "grad_norm": 1.4451377144366324,
      "learning_rate": 3.820618420978574e-06,
      "loss": 0.8579,
      "step": 1794
    },
    {
      "epoch": 0.16187942462912025,
      "grad_norm": 2.1057870373041694,
      "learning_rate": 3.820376521641708e-06,
      "loss": 0.7932,
      "step": 1795
    },
    {
      "epoch": 0.1619696081525905,
      "grad_norm": 1.6130002864735697,
      "learning_rate": 3.82013446698028e-06,
      "loss": 0.8062,
      "step": 1796
    },
    {
      "epoch": 0.16205979167606077,
      "grad_norm": 1.5543421836244926,
      "learning_rate": 3.819892257014943e-06,
      "loss": 0.7549,
      "step": 1797
    },
    {
      "epoch": 0.16214997519953103,
      "grad_norm": 2.454058912812097,
      "learning_rate": 3.819649891766364e-06,
      "loss": 0.8242,
      "step": 1798
    },
    {
      "epoch": 0.16224015872300132,
      "grad_norm": 1.5098856376136158,
      "learning_rate": 3.819407371255222e-06,
      "loss": 0.7866,
      "step": 1799
    },
    {
      "epoch": 0.16233034224647158,
      "grad_norm": 5.144092630907935,
      "learning_rate": 3.819164695502212e-06,
      "loss": 0.8027,
      "step": 1800
    },
    {
      "epoch": 0.16242052576994184,
      "grad_norm": 1.689651389423138,
      "learning_rate": 3.818921864528039e-06,
      "loss": 0.7352,
      "step": 1801
    },
    {
      "epoch": 0.1625107092934121,
      "grad_norm": 4.6409515368728576,
      "learning_rate": 3.818678878353423e-06,
      "loss": 0.8477,
      "step": 1802
    },
    {
      "epoch": 0.16260089281688236,
      "grad_norm": 3.4430378089178304,
      "learning_rate": 3.818435736999097e-06,
      "loss": 0.7401,
      "step": 1803
    },
    {
      "epoch": 0.16269107634035262,
      "grad_norm": 3.2111413895595873,
      "learning_rate": 3.818192440485807e-06,
      "loss": 0.7838,
      "step": 1804
    },
    {
      "epoch": 0.16278125986382289,
      "grad_norm": 1.9663994092977644,
      "learning_rate": 3.817948988834314e-06,
      "loss": 0.7313,
      "step": 1805
    },
    {
      "epoch": 0.16287144338729315,
      "grad_norm": 1.9924532245616307,
      "learning_rate": 3.817705382065388e-06,
      "loss": 0.7928,
      "step": 1806
    },
    {
      "epoch": 0.1629616269107634,
      "grad_norm": 1.3979761652026588,
      "learning_rate": 3.8174616201998155e-06,
      "loss": 0.7606,
      "step": 1807
    },
    {
      "epoch": 0.16305181043423367,
      "grad_norm": 13.917321347734044,
      "learning_rate": 3.817217703258397e-06,
      "loss": 0.8457,
      "step": 1808
    },
    {
      "epoch": 0.16314199395770393,
      "grad_norm": 1.665098915254889,
      "learning_rate": 3.816973631261943e-06,
      "loss": 0.7824,
      "step": 1809
    },
    {
      "epoch": 0.1632321774811742,
      "grad_norm": 2.02553870488394,
      "learning_rate": 3.816729404231281e-06,
      "loss": 0.8277,
      "step": 1810
    },
    {
      "epoch": 0.16332236100464445,
      "grad_norm": 2.191673548098979,
      "learning_rate": 3.816485022187249e-06,
      "loss": 0.8576,
      "step": 1811
    },
    {
      "epoch": 0.1634125445281147,
      "grad_norm": 2.1390451880591113,
      "learning_rate": 3.816240485150698e-06,
      "loss": 0.7492,
      "step": 1812
    },
    {
      "epoch": 0.16350272805158497,
      "grad_norm": 1.4383111406708666,
      "learning_rate": 3.815995793142495e-06,
      "loss": 0.7951,
      "step": 1813
    },
    {
      "epoch": 0.16359291157505523,
      "grad_norm": 1.65932186925281,
      "learning_rate": 3.815750946183518e-06,
      "loss": 0.7337,
      "step": 1814
    },
    {
      "epoch": 0.1636830950985255,
      "grad_norm": 2.2508280607875215,
      "learning_rate": 3.815505944294658e-06,
      "loss": 0.8463,
      "step": 1815
    },
    {
      "epoch": 0.16377327862199575,
      "grad_norm": 1.9494085123961358,
      "learning_rate": 3.81526078749682e-06,
      "loss": 0.7522,
      "step": 1816
    },
    {
      "epoch": 0.16386346214546602,
      "grad_norm": 3.434822565685054,
      "learning_rate": 3.8150154758109225e-06,
      "loss": 0.7326,
      "step": 1817
    },
    {
      "epoch": 0.16395364566893628,
      "grad_norm": 1.5300836597583072,
      "learning_rate": 3.814770009257896e-06,
      "loss": 0.7634,
      "step": 1818
    },
    {
      "epoch": 0.16404382919240654,
      "grad_norm": 1.9010055190181332,
      "learning_rate": 3.814524387858687e-06,
      "loss": 0.7997,
      "step": 1819
    },
    {
      "epoch": 0.1641340127158768,
      "grad_norm": 1.719593257383982,
      "learning_rate": 3.814278611634251e-06,
      "loss": 0.8151,
      "step": 1820
    },
    {
      "epoch": 0.16422419623934706,
      "grad_norm": 1.795281674059836,
      "learning_rate": 3.8140326806055606e-06,
      "loss": 0.7704,
      "step": 1821
    },
    {
      "epoch": 0.16431437976281732,
      "grad_norm": 1.7340502735668664,
      "learning_rate": 3.8137865947935992e-06,
      "loss": 0.854,
      "step": 1822
    },
    {
      "epoch": 0.1644045632862876,
      "grad_norm": 1.7547315257418301,
      "learning_rate": 3.8135403542193646e-06,
      "loss": 0.8157,
      "step": 1823
    },
    {
      "epoch": 0.16449474680975787,
      "grad_norm": 2.2036201515826037,
      "learning_rate": 3.813293958903867e-06,
      "loss": 0.733,
      "step": 1824
    },
    {
      "epoch": 0.16458493033322813,
      "grad_norm": 1.7717383222662058,
      "learning_rate": 3.8130474088681306e-06,
      "loss": 0.7645,
      "step": 1825
    },
    {
      "epoch": 0.1646751138566984,
      "grad_norm": 2.0426866617435935,
      "learning_rate": 3.8128007041331927e-06,
      "loss": 0.7171,
      "step": 1826
    },
    {
      "epoch": 0.16476529738016865,
      "grad_norm": 1.6387542231059418,
      "learning_rate": 3.812553844720102e-06,
      "loss": 0.7072,
      "step": 1827
    },
    {
      "epoch": 0.1648554809036389,
      "grad_norm": 1.6798439130480516,
      "learning_rate": 3.8123068306499236e-06,
      "loss": 0.8082,
      "step": 1828
    },
    {
      "epoch": 0.16494566442710917,
      "grad_norm": 1.9582989331193281,
      "learning_rate": 3.812059661943733e-06,
      "loss": 0.7892,
      "step": 1829
    },
    {
      "epoch": 0.16503584795057943,
      "grad_norm": 5.641021799095109,
      "learning_rate": 3.811812338622621e-06,
      "loss": 0.7622,
      "step": 1830
    },
    {
      "epoch": 0.1651260314740497,
      "grad_norm": 1.4913536258323594,
      "learning_rate": 3.81156486070769e-06,
      "loss": 0.7814,
      "step": 1831
    },
    {
      "epoch": 0.16521621499751996,
      "grad_norm": 1.6580678212889532,
      "learning_rate": 3.811317228220056e-06,
      "loss": 0.8402,
      "step": 1832
    },
    {
      "epoch": 0.16530639852099022,
      "grad_norm": 3.885574801229505,
      "learning_rate": 3.811069441180849e-06,
      "loss": 0.8393,
      "step": 1833
    },
    {
      "epoch": 0.16539658204446048,
      "grad_norm": 1.8278389079233655,
      "learning_rate": 3.8108214996112107e-06,
      "loss": 0.7966,
      "step": 1834
    },
    {
      "epoch": 0.16548676556793074,
      "grad_norm": 2.060255968431239,
      "learning_rate": 3.810573403532297e-06,
      "loss": 0.8277,
      "step": 1835
    },
    {
      "epoch": 0.165576949091401,
      "grad_norm": 5.442969366746771,
      "learning_rate": 3.8103251529652774e-06,
      "loss": 0.7687,
      "step": 1836
    },
    {
      "epoch": 0.16566713261487126,
      "grad_norm": 1.7881571382089814,
      "learning_rate": 3.810076747931334e-06,
      "loss": 0.7704,
      "step": 1837
    },
    {
      "epoch": 0.16575731613834152,
      "grad_norm": 4.394679033973415,
      "learning_rate": 3.809828188451662e-06,
      "loss": 0.7538,
      "step": 1838
    },
    {
      "epoch": 0.16584749966181178,
      "grad_norm": 1.9392636793410727,
      "learning_rate": 3.809579474547469e-06,
      "loss": 0.7884,
      "step": 1839
    },
    {
      "epoch": 0.16593768318528204,
      "grad_norm": 1.646525281802965,
      "learning_rate": 3.809330606239977e-06,
      "loss": 0.8382,
      "step": 1840
    },
    {
      "epoch": 0.1660278667087523,
      "grad_norm": 1.8707307531133177,
      "learning_rate": 3.809081583550422e-06,
      "loss": 0.8574,
      "step": 1841
    },
    {
      "epoch": 0.16611805023222256,
      "grad_norm": 2.182486101348773,
      "learning_rate": 3.808832406500051e-06,
      "loss": 0.8763,
      "step": 1842
    },
    {
      "epoch": 0.16620823375569282,
      "grad_norm": 2.172912788814353,
      "learning_rate": 3.8085830751101253e-06,
      "loss": 0.8671,
      "step": 1843
    },
    {
      "epoch": 0.16629841727916309,
      "grad_norm": 1.6572521704525618,
      "learning_rate": 3.808333589401919e-06,
      "loss": 0.8267,
      "step": 1844
    },
    {
      "epoch": 0.16638860080263335,
      "grad_norm": 1.7135508188237099,
      "learning_rate": 3.8080839493967194e-06,
      "loss": 0.7253,
      "step": 1845
    },
    {
      "epoch": 0.1664787843261036,
      "grad_norm": 0.6466345252532504,
      "learning_rate": 3.807834155115828e-06,
      "loss": 0.5461,
      "step": 1846
    },
    {
      "epoch": 0.1665689678495739,
      "grad_norm": 0.6520370173433466,
      "learning_rate": 3.8075842065805584e-06,
      "loss": 0.6501,
      "step": 1847
    },
    {
      "epoch": 0.16665915137304416,
      "grad_norm": 3.033315367872533,
      "learning_rate": 3.8073341038122374e-06,
      "loss": 0.7821,
      "step": 1848
    },
    {
      "epoch": 0.16674933489651442,
      "grad_norm": 1.9841197893526934,
      "learning_rate": 3.8070838468322048e-06,
      "loss": 0.8607,
      "step": 1849
    },
    {
      "epoch": 0.16683951841998468,
      "grad_norm": 1.9374302882297154,
      "learning_rate": 3.8068334356618143e-06,
      "loss": 0.7296,
      "step": 1850
    },
    {
      "epoch": 0.16692970194345494,
      "grad_norm": 3.374178468534884,
      "learning_rate": 3.8065828703224324e-06,
      "loss": 0.835,
      "step": 1851
    },
    {
      "epoch": 0.1670198854669252,
      "grad_norm": 2.346306284177047,
      "learning_rate": 3.8063321508354386e-06,
      "loss": 0.7929,
      "step": 1852
    },
    {
      "epoch": 0.16711006899039546,
      "grad_norm": 1.5509918607969226,
      "learning_rate": 3.8060812772222255e-06,
      "loss": 0.7773,
      "step": 1853
    },
    {
      "epoch": 0.16720025251386572,
      "grad_norm": 1.4254160792262118,
      "learning_rate": 3.8058302495041993e-06,
      "loss": 0.8586,
      "step": 1854
    },
    {
      "epoch": 0.16729043603733598,
      "grad_norm": 1.92684708942859,
      "learning_rate": 3.805579067702779e-06,
      "loss": 0.8079,
      "step": 1855
    },
    {
      "epoch": 0.16738061956080624,
      "grad_norm": 1.7634965338417385,
      "learning_rate": 3.8053277318393967e-06,
      "loss": 0.7956,
      "step": 1856
    },
    {
      "epoch": 0.1674708030842765,
      "grad_norm": 1.5521921038650548,
      "learning_rate": 3.805076241935498e-06,
      "loss": 0.8101,
      "step": 1857
    },
    {
      "epoch": 0.16756098660774676,
      "grad_norm": 1.8914261058598676,
      "learning_rate": 3.804824598012541e-06,
      "loss": 0.6681,
      "step": 1858
    },
    {
      "epoch": 0.16765117013121703,
      "grad_norm": 1.5163820086924042,
      "learning_rate": 3.8045728000919975e-06,
      "loss": 0.8338,
      "step": 1859
    },
    {
      "epoch": 0.1677413536546873,
      "grad_norm": 1.3923050236787848,
      "learning_rate": 3.8043208481953524e-06,
      "loss": 0.826,
      "step": 1860
    },
    {
      "epoch": 0.16783153717815755,
      "grad_norm": 15.23022729995841,
      "learning_rate": 3.804068742344104e-06,
      "loss": 0.7204,
      "step": 1861
    },
    {
      "epoch": 0.1679217207016278,
      "grad_norm": 1.7907154825993357,
      "learning_rate": 3.8038164825597628e-06,
      "loss": 0.7613,
      "step": 1862
    },
    {
      "epoch": 0.16801190422509807,
      "grad_norm": 2.6009888345844656,
      "learning_rate": 3.8035640688638537e-06,
      "loss": 0.8784,
      "step": 1863
    },
    {
      "epoch": 0.16810208774856833,
      "grad_norm": 1.6776590563528673,
      "learning_rate": 3.8033115012779125e-06,
      "loss": 0.8677,
      "step": 1864
    },
    {
      "epoch": 0.1681922712720386,
      "grad_norm": 1.4876050639514324,
      "learning_rate": 3.8030587798234915e-06,
      "loss": 0.8347,
      "step": 1865
    },
    {
      "epoch": 0.16828245479550885,
      "grad_norm": 2.3833620532068265,
      "learning_rate": 3.802805904522153e-06,
      "loss": 0.7334,
      "step": 1866
    },
    {
      "epoch": 0.1683726383189791,
      "grad_norm": 1.4407125318170577,
      "learning_rate": 3.8025528753954742e-06,
      "loss": 0.783,
      "step": 1867
    },
    {
      "epoch": 0.16846282184244937,
      "grad_norm": 2.2002985795020504,
      "learning_rate": 3.802299692465045e-06,
      "loss": 0.7475,
      "step": 1868
    },
    {
      "epoch": 0.16855300536591963,
      "grad_norm": 1.9018606085456864,
      "learning_rate": 3.802046355752468e-06,
      "loss": 0.7833,
      "step": 1869
    },
    {
      "epoch": 0.16864318888938992,
      "grad_norm": 2.3211874805221284,
      "learning_rate": 3.80179286527936e-06,
      "loss": 0.8529,
      "step": 1870
    },
    {
      "epoch": 0.16873337241286018,
      "grad_norm": 1.7204016638991881,
      "learning_rate": 3.801539221067349e-06,
      "loss": 0.7073,
      "step": 1871
    },
    {
      "epoch": 0.16882355593633044,
      "grad_norm": 1.5100603494521094,
      "learning_rate": 3.801285423138079e-06,
      "loss": 0.746,
      "step": 1872
    },
    {
      "epoch": 0.1689137394598007,
      "grad_norm": 3.4470851083667386,
      "learning_rate": 3.8010314715132037e-06,
      "loss": 0.7487,
      "step": 1873
    },
    {
      "epoch": 0.16900392298327097,
      "grad_norm": 2.0085346986272357,
      "learning_rate": 3.800777366214393e-06,
      "loss": 0.7503,
      "step": 1874
    },
    {
      "epoch": 0.16909410650674123,
      "grad_norm": 2.2079217515224463,
      "learning_rate": 3.800523107263328e-06,
      "loss": 0.7662,
      "step": 1875
    },
    {
      "epoch": 0.1691842900302115,
      "grad_norm": 1.5849629682432693,
      "learning_rate": 3.800268694681703e-06,
      "loss": 0.8456,
      "step": 1876
    },
    {
      "epoch": 0.16927447355368175,
      "grad_norm": 0.6560381592840218,
      "learning_rate": 3.800014128491227e-06,
      "loss": 0.5989,
      "step": 1877
    },
    {
      "epoch": 0.169364657077152,
      "grad_norm": 1.5503503311238072,
      "learning_rate": 3.79975940871362e-06,
      "loss": 0.8468,
      "step": 1878
    },
    {
      "epoch": 0.16945484060062227,
      "grad_norm": 3.051637653884463,
      "learning_rate": 3.799504535370617e-06,
      "loss": 0.8015,
      "step": 1879
    },
    {
      "epoch": 0.16954502412409253,
      "grad_norm": 2.332875774570441,
      "learning_rate": 3.799249508483964e-06,
      "loss": 0.7304,
      "step": 1880
    },
    {
      "epoch": 0.1696352076475628,
      "grad_norm": 0.7260923197374896,
      "learning_rate": 3.798994328075422e-06,
      "loss": 0.6595,
      "step": 1881
    },
    {
      "epoch": 0.16972539117103305,
      "grad_norm": 1.7239796232846996,
      "learning_rate": 3.798738994166765e-06,
      "loss": 0.7213,
      "step": 1882
    },
    {
      "epoch": 0.1698155746945033,
      "grad_norm": 1.728450795482702,
      "learning_rate": 3.7984835067797788e-06,
      "loss": 0.792,
      "step": 1883
    },
    {
      "epoch": 0.16990575821797357,
      "grad_norm": 2.346599829906827,
      "learning_rate": 3.798227865936263e-06,
      "loss": 0.8836,
      "step": 1884
    },
    {
      "epoch": 0.16999594174144383,
      "grad_norm": 2.0789369990136595,
      "learning_rate": 3.7979720716580297e-06,
      "loss": 0.6567,
      "step": 1885
    },
    {
      "epoch": 0.1700861252649141,
      "grad_norm": 1.821842647415728,
      "learning_rate": 3.7977161239669057e-06,
      "loss": 0.7827,
      "step": 1886
    },
    {
      "epoch": 0.17017630878838436,
      "grad_norm": 2.0390007623556,
      "learning_rate": 3.7974600228847294e-06,
      "loss": 0.7753,
      "step": 1887
    },
    {
      "epoch": 0.17026649231185462,
      "grad_norm": 2.386409758250087,
      "learning_rate": 3.7972037684333534e-06,
      "loss": 0.7334,
      "step": 1888
    },
    {
      "epoch": 0.17035667583532488,
      "grad_norm": 1.4859247249210552,
      "learning_rate": 3.796947360634642e-06,
      "loss": 0.719,
      "step": 1889
    },
    {
      "epoch": 0.17044685935879514,
      "grad_norm": 2.1454915882551107,
      "learning_rate": 3.796690799510473e-06,
      "loss": 0.8479,
      "step": 1890
    },
    {
      "epoch": 0.1705370428822654,
      "grad_norm": 1.5517491315787313,
      "learning_rate": 3.7964340850827387e-06,
      "loss": 0.7546,
      "step": 1891
    },
    {
      "epoch": 0.17062722640573566,
      "grad_norm": 0.6797795945838627,
      "learning_rate": 3.7961772173733425e-06,
      "loss": 0.5677,
      "step": 1892
    },
    {
      "epoch": 0.17071740992920592,
      "grad_norm": 1.4742709346201504,
      "learning_rate": 3.7959201964042024e-06,
      "loss": 0.7759,
      "step": 1893
    },
    {
      "epoch": 0.1708075934526762,
      "grad_norm": 1.5886931133515512,
      "learning_rate": 3.795663022197248e-06,
      "loss": 0.8216,
      "step": 1894
    },
    {
      "epoch": 0.17089777697614647,
      "grad_norm": 2.139682088750095,
      "learning_rate": 3.7954056947744242e-06,
      "loss": 0.7964,
      "step": 1895
    },
    {
      "epoch": 0.17098796049961673,
      "grad_norm": 2.8399636349566126,
      "learning_rate": 3.7951482141576863e-06,
      "loss": 0.7645,
      "step": 1896
    },
    {
      "epoch": 0.171078144023087,
      "grad_norm": 1.85486151144878,
      "learning_rate": 3.794890580369004e-06,
      "loss": 0.8434,
      "step": 1897
    },
    {
      "epoch": 0.17116832754655725,
      "grad_norm": 1.6899856345564965,
      "learning_rate": 3.7946327934303612e-06,
      "loss": 0.8205,
      "step": 1898
    },
    {
      "epoch": 0.1712585110700275,
      "grad_norm": 1.6609363832227768,
      "learning_rate": 3.794374853363752e-06,
      "loss": 0.8846,
      "step": 1899
    },
    {
      "epoch": 0.17134869459349777,
      "grad_norm": 1.6858388000389588,
      "learning_rate": 3.794116760191187e-06,
      "loss": 0.812,
      "step": 1900
    },
    {
      "epoch": 0.17143887811696804,
      "grad_norm": 1.7534208241935814,
      "learning_rate": 3.7938585139346877e-06,
      "loss": 0.8898,
      "step": 1901
    },
    {
      "epoch": 0.1715290616404383,
      "grad_norm": 2.039127626511732,
      "learning_rate": 3.793600114616288e-06,
      "loss": 0.774,
      "step": 1902
    },
    {
      "epoch": 0.17161924516390856,
      "grad_norm": 0.7403012046741297,
      "learning_rate": 3.793341562258037e-06,
      "loss": 0.5576,
      "step": 1903
    },
    {
      "epoch": 0.17170942868737882,
      "grad_norm": 1.6213752961694101,
      "learning_rate": 3.7930828568819953e-06,
      "loss": 0.806,
      "step": 1904
    },
    {
      "epoch": 0.17179961221084908,
      "grad_norm": 2.1084074839141267,
      "learning_rate": 3.7928239985102378e-06,
      "loss": 0.7867,
      "step": 1905
    },
    {
      "epoch": 0.17188979573431934,
      "grad_norm": 3.241010410855906,
      "learning_rate": 3.7925649871648505e-06,
      "loss": 0.7919,
      "step": 1906
    },
    {
      "epoch": 0.1719799792577896,
      "grad_norm": 1.628442858458586,
      "learning_rate": 3.792305822867935e-06,
      "loss": 0.7125,
      "step": 1907
    },
    {
      "epoch": 0.17207016278125986,
      "grad_norm": 2.2633741049635994,
      "learning_rate": 3.792046505641604e-06,
      "loss": 0.7565,
      "step": 1908
    },
    {
      "epoch": 0.17216034630473012,
      "grad_norm": 1.7019214402165639,
      "learning_rate": 3.791787035507984e-06,
      "loss": 0.748,
      "step": 1909
    },
    {
      "epoch": 0.17225052982820038,
      "grad_norm": 3.175022611762801,
      "learning_rate": 3.7915274124892136e-06,
      "loss": 0.7498,
      "step": 1910
    },
    {
      "epoch": 0.17234071335167064,
      "grad_norm": 4.334072612398862,
      "learning_rate": 3.7912676366074466e-06,
      "loss": 0.7723,
      "step": 1911
    },
    {
      "epoch": 0.1724308968751409,
      "grad_norm": 2.48167253254884,
      "learning_rate": 3.7910077078848478e-06,
      "loss": 0.755,
      "step": 1912
    },
    {
      "epoch": 0.17252108039861117,
      "grad_norm": 2.1293614452077314,
      "learning_rate": 3.7907476263435957e-06,
      "loss": 0.7541,
      "step": 1913
    },
    {
      "epoch": 0.17261126392208143,
      "grad_norm": 3.5252356159654474,
      "learning_rate": 3.7904873920058826e-06,
      "loss": 0.759,
      "step": 1914
    },
    {
      "epoch": 0.1727014474455517,
      "grad_norm": 1.994172188474495,
      "learning_rate": 3.7902270048939114e-06,
      "loss": 0.6678,
      "step": 1915
    },
    {
      "epoch": 0.17279163096902195,
      "grad_norm": 1.9749616088033075,
      "learning_rate": 3.7899664650299023e-06,
      "loss": 0.7525,
      "step": 1916
    },
    {
      "epoch": 0.1728818144924922,
      "grad_norm": 1.785737658756425,
      "learning_rate": 3.7897057724360836e-06,
      "loss": 0.8003,
      "step": 1917
    },
    {
      "epoch": 0.1729719980159625,
      "grad_norm": 2.852790296461683,
      "learning_rate": 3.7894449271347004e-06,
      "loss": 0.8113,
      "step": 1918
    },
    {
      "epoch": 0.17306218153943276,
      "grad_norm": 2.1416871714321633,
      "learning_rate": 3.789183929148009e-06,
      "loss": 0.849,
      "step": 1919
    },
    {
      "epoch": 0.17315236506290302,
      "grad_norm": 4.396810606787935,
      "learning_rate": 3.7889227784982795e-06,
      "loss": 0.8173,
      "step": 1920
    },
    {
      "epoch": 0.17324254858637328,
      "grad_norm": 1.6428598350597619,
      "learning_rate": 3.7886614752077945e-06,
      "loss": 0.7837,
      "step": 1921
    },
    {
      "epoch": 0.17333273210984354,
      "grad_norm": 2.0110841213028716,
      "learning_rate": 3.7884000192988495e-06,
      "loss": 0.7929,
      "step": 1922
    },
    {
      "epoch": 0.1734229156333138,
      "grad_norm": 2.405335388748717,
      "learning_rate": 3.7881384107937546e-06,
      "loss": 0.8448,
      "step": 1923
    },
    {
      "epoch": 0.17351309915678406,
      "grad_norm": 1.7997824987822761,
      "learning_rate": 3.78787664971483e-06,
      "loss": 0.7821,
      "step": 1924
    },
    {
      "epoch": 0.17360328268025432,
      "grad_norm": 1.499352553828891,
      "learning_rate": 3.7876147360844115e-06,
      "loss": 0.7567,
      "step": 1925
    },
    {
      "epoch": 0.17369346620372458,
      "grad_norm": 2.228698241552273,
      "learning_rate": 3.7873526699248474e-06,
      "loss": 0.8581,
      "step": 1926
    },
    {
      "epoch": 0.17378364972719484,
      "grad_norm": 1.4406388884100403,
      "learning_rate": 3.7870904512584974e-06,
      "loss": 0.7947,
      "step": 1927
    },
    {
      "epoch": 0.1738738332506651,
      "grad_norm": 1.8298498640332064,
      "learning_rate": 3.7868280801077368e-06,
      "loss": 0.79,
      "step": 1928
    },
    {
      "epoch": 0.17396401677413537,
      "grad_norm": 1.495615511719464,
      "learning_rate": 3.7865655564949517e-06,
      "loss": 0.7795,
      "step": 1929
    },
    {
      "epoch": 0.17405420029760563,
      "grad_norm": 1.4500885278708744,
      "learning_rate": 3.786302880442542e-06,
      "loss": 0.7592,
      "step": 1930
    },
    {
      "epoch": 0.1741443838210759,
      "grad_norm": 1.8617578832921229,
      "learning_rate": 3.7860400519729215e-06,
      "loss": 0.7858,
      "step": 1931
    },
    {
      "epoch": 0.17423456734454615,
      "grad_norm": 1.9105235489704528,
      "learning_rate": 3.7857770711085157e-06,
      "loss": 0.8215,
      "step": 1932
    },
    {
      "epoch": 0.1743247508680164,
      "grad_norm": 2.8314201590160235,
      "learning_rate": 3.785513937871763e-06,
      "loss": 0.8352,
      "step": 1933
    },
    {
      "epoch": 0.17441493439148667,
      "grad_norm": 1.7787904992388774,
      "learning_rate": 3.785250652285116e-06,
      "loss": 0.7613,
      "step": 1934
    },
    {
      "epoch": 0.17450511791495693,
      "grad_norm": 1.6435171158157278,
      "learning_rate": 3.78498721437104e-06,
      "loss": 0.863,
      "step": 1935
    },
    {
      "epoch": 0.1745953014384272,
      "grad_norm": 2.123320308367885,
      "learning_rate": 3.784723624152012e-06,
      "loss": 0.7947,
      "step": 1936
    },
    {
      "epoch": 0.17468548496189745,
      "grad_norm": 3.7299726385999925,
      "learning_rate": 3.784459881650524e-06,
      "loss": 0.7404,
      "step": 1937
    },
    {
      "epoch": 0.1747756684853677,
      "grad_norm": 1.6603442276390008,
      "learning_rate": 3.784195986889079e-06,
      "loss": 0.7604,
      "step": 1938
    },
    {
      "epoch": 0.17486585200883797,
      "grad_norm": 1.7346545157426034,
      "learning_rate": 3.7839319398901946e-06,
      "loss": 0.7157,
      "step": 1939
    },
    {
      "epoch": 0.17495603553230824,
      "grad_norm": 5.843414460203121,
      "learning_rate": 3.7836677406764013e-06,
      "loss": 0.8578,
      "step": 1940
    },
    {
      "epoch": 0.1750462190557785,
      "grad_norm": 1.4921570620003821,
      "learning_rate": 3.7834033892702407e-06,
      "loss": 0.8362,
      "step": 1941
    },
    {
      "epoch": 0.17513640257924878,
      "grad_norm": 1.697029334368308,
      "learning_rate": 3.783138885694269e-06,
      "loss": 0.7574,
      "step": 1942
    },
    {
      "epoch": 0.17522658610271905,
      "grad_norm": 1.8337959948429117,
      "learning_rate": 3.7828742299710558e-06,
      "loss": 0.8701,
      "step": 1943
    },
    {
      "epoch": 0.1753167696261893,
      "grad_norm": 1.6958554009478732,
      "learning_rate": 3.782609422123183e-06,
      "loss": 0.7732,
      "step": 1944
    },
    {
      "epoch": 0.17540695314965957,
      "grad_norm": 1.8959093253307497,
      "learning_rate": 3.7823444621732444e-06,
      "loss": 0.7953,
      "step": 1945
    },
    {
      "epoch": 0.17549713667312983,
      "grad_norm": 2.2108540333875055,
      "learning_rate": 3.782079350143849e-06,
      "loss": 0.8065,
      "step": 1946
    },
    {
      "epoch": 0.1755873201966001,
      "grad_norm": 2.5404455555587204,
      "learning_rate": 3.781814086057617e-06,
      "loss": 0.889,
      "step": 1947
    },
    {
      "epoch": 0.17567750372007035,
      "grad_norm": 3.926536798601781,
      "learning_rate": 3.7815486699371826e-06,
      "loss": 0.7767,
      "step": 1948
    },
    {
      "epoch": 0.1757676872435406,
      "grad_norm": 1.5548665312619125,
      "learning_rate": 3.7812831018051918e-06,
      "loss": 0.7635,
      "step": 1949
    },
    {
      "epoch": 0.17585787076701087,
      "grad_norm": 1.5018497823375263,
      "learning_rate": 3.7810173816843058e-06,
      "loss": 0.7987,
      "step": 1950
    },
    {
      "epoch": 0.17594805429048113,
      "grad_norm": 1.5707540318365838,
      "learning_rate": 3.7807515095971955e-06,
      "loss": 0.8627,
      "step": 1951
    },
    {
      "epoch": 0.1760382378139514,
      "grad_norm": 1.5084198830292799,
      "learning_rate": 3.7804854855665475e-06,
      "loss": 0.783,
      "step": 1952
    },
    {
      "epoch": 0.17612842133742165,
      "grad_norm": 3.1161587578383636,
      "learning_rate": 3.7802193096150606e-06,
      "loss": 0.7988,
      "step": 1953
    },
    {
      "epoch": 0.17621860486089191,
      "grad_norm": 1.933345216133607,
      "learning_rate": 3.779952981765446e-06,
      "loss": 0.7475,
      "step": 1954
    },
    {
      "epoch": 0.17630878838436218,
      "grad_norm": 1.4438968633291516,
      "learning_rate": 3.779686502040429e-06,
      "loss": 0.806,
      "step": 1955
    },
    {
      "epoch": 0.17639897190783244,
      "grad_norm": 0.7003870532598444,
      "learning_rate": 3.779419870462746e-06,
      "loss": 0.5762,
      "step": 1956
    },
    {
      "epoch": 0.1764891554313027,
      "grad_norm": 2.0088694124745223,
      "learning_rate": 3.779153087055148e-06,
      "loss": 0.7702,
      "step": 1957
    },
    {
      "epoch": 0.17657933895477296,
      "grad_norm": 2.062091324516388,
      "learning_rate": 3.7788861518403988e-06,
      "loss": 0.7999,
      "step": 1958
    },
    {
      "epoch": 0.17666952247824322,
      "grad_norm": 1.6624198291923196,
      "learning_rate": 3.7786190648412742e-06,
      "loss": 0.6825,
      "step": 1959
    },
    {
      "epoch": 0.17675970600171348,
      "grad_norm": 0.6475048021558217,
      "learning_rate": 3.778351826080564e-06,
      "loss": 0.5871,
      "step": 1960
    },
    {
      "epoch": 0.17684988952518374,
      "grad_norm": 1.5335005745005146,
      "learning_rate": 3.7780844355810704e-06,
      "loss": 0.8926,
      "step": 1961
    },
    {
      "epoch": 0.176940073048654,
      "grad_norm": 1.5129242262690366,
      "learning_rate": 3.777816893365608e-06,
      "loss": 0.8012,
      "step": 1962
    },
    {
      "epoch": 0.17703025657212426,
      "grad_norm": 1.5400705666908816,
      "learning_rate": 3.7775491994570057e-06,
      "loss": 0.8644,
      "step": 1963
    },
    {
      "epoch": 0.17712044009559452,
      "grad_norm": 1.6617502974442064,
      "learning_rate": 3.777281353878105e-06,
      "loss": 0.7624,
      "step": 1964
    },
    {
      "epoch": 0.17721062361906478,
      "grad_norm": 0.6201857401284572,
      "learning_rate": 3.777013356651758e-06,
      "loss": 0.594,
      "step": 1965
    },
    {
      "epoch": 0.17730080714253507,
      "grad_norm": 1.9058160757004317,
      "learning_rate": 3.776745207800834e-06,
      "loss": 0.8053,
      "step": 1966
    },
    {
      "epoch": 0.17739099066600533,
      "grad_norm": 2.0542461661017173,
      "learning_rate": 3.7764769073482122e-06,
      "loss": 0.8181,
      "step": 1967
    },
    {
      "epoch": 0.1774811741894756,
      "grad_norm": 1.5307643665248234,
      "learning_rate": 3.7762084553167846e-06,
      "loss": 0.8571,
      "step": 1968
    },
    {
      "epoch": 0.17757135771294585,
      "grad_norm": 2.0830254390809535,
      "learning_rate": 3.775939851729458e-06,
      "loss": 0.8799,
      "step": 1969
    },
    {
      "epoch": 0.17766154123641612,
      "grad_norm": 4.031717199413893,
      "learning_rate": 3.775671096609151e-06,
      "loss": 0.7484,
      "step": 1970
    },
    {
      "epoch": 0.17775172475988638,
      "grad_norm": 1.716536744148651,
      "learning_rate": 3.775402189978795e-06,
      "loss": 0.7741,
      "step": 1971
    },
    {
      "epoch": 0.17784190828335664,
      "grad_norm": 1.3417337061242907,
      "learning_rate": 3.7751331318613343e-06,
      "loss": 0.7717,
      "step": 1972
    },
    {
      "epoch": 0.1779320918068269,
      "grad_norm": 2.230659971040541,
      "learning_rate": 3.774863922279727e-06,
      "loss": 0.8114,
      "step": 1973
    },
    {
      "epoch": 0.17802227533029716,
      "grad_norm": 1.4662095683345395,
      "learning_rate": 3.7745945612569435e-06,
      "loss": 0.8474,
      "step": 1974
    },
    {
      "epoch": 0.17811245885376742,
      "grad_norm": 1.4991641894933745,
      "learning_rate": 3.7743250488159674e-06,
      "loss": 0.6698,
      "step": 1975
    },
    {
      "epoch": 0.17820264237723768,
      "grad_norm": 1.7583309511045013,
      "learning_rate": 3.774055384979794e-06,
      "loss": 0.6951,
      "step": 1976
    },
    {
      "epoch": 0.17829282590070794,
      "grad_norm": 1.6091744010867166,
      "learning_rate": 3.773785569771433e-06,
      "loss": 0.7727,
      "step": 1977
    },
    {
      "epoch": 0.1783830094241782,
      "grad_norm": 1.670161215601429,
      "learning_rate": 3.7735156032139066e-06,
      "loss": 0.7808,
      "step": 1978
    },
    {
      "epoch": 0.17847319294764846,
      "grad_norm": 1.6323006297605684,
      "learning_rate": 3.773245485330251e-06,
      "loss": 0.8937,
      "step": 1979
    },
    {
      "epoch": 0.17856337647111872,
      "grad_norm": 2.144810170307968,
      "learning_rate": 3.7729752161435115e-06,
      "loss": 0.7595,
      "step": 1980
    },
    {
      "epoch": 0.17865355999458898,
      "grad_norm": 1.739805431997465,
      "learning_rate": 3.7727047956767514e-06,
      "loss": 0.8573,
      "step": 1981
    },
    {
      "epoch": 0.17874374351805924,
      "grad_norm": 1.4871701876664083,
      "learning_rate": 3.7724342239530436e-06,
      "loss": 0.9164,
      "step": 1982
    },
    {
      "epoch": 0.1788339270415295,
      "grad_norm": 2.1016986334858037,
      "learning_rate": 3.772163500995474e-06,
      "loss": 0.7408,
      "step": 1983
    },
    {
      "epoch": 0.17892411056499977,
      "grad_norm": 2.0319578844647053,
      "learning_rate": 3.7718926268271437e-06,
      "loss": 0.8883,
      "step": 1984
    },
    {
      "epoch": 0.17901429408847003,
      "grad_norm": 1.8494292255069675,
      "learning_rate": 3.771621601471164e-06,
      "loss": 0.8156,
      "step": 1985
    },
    {
      "epoch": 0.1791044776119403,
      "grad_norm": 2.9965469833196656,
      "learning_rate": 3.771350424950661e-06,
      "loss": 0.8376,
      "step": 1986
    },
    {
      "epoch": 0.17919466113541055,
      "grad_norm": 1.713701219483555,
      "learning_rate": 3.771079097288772e-06,
      "loss": 0.7394,
      "step": 1987
    },
    {
      "epoch": 0.1792848446588808,
      "grad_norm": 1.751238452837911,
      "learning_rate": 3.770807618508649e-06,
      "loss": 0.8065,
      "step": 1988
    },
    {
      "epoch": 0.17937502818235107,
      "grad_norm": 1.703230268392377,
      "learning_rate": 3.7705359886334555e-06,
      "loss": 0.7802,
      "step": 1989
    },
    {
      "epoch": 0.17946521170582136,
      "grad_norm": 1.8486484178306601,
      "learning_rate": 3.7702642076863694e-06,
      "loss": 0.772,
      "step": 1990
    },
    {
      "epoch": 0.17955539522929162,
      "grad_norm": 1.5442447842421414,
      "learning_rate": 3.7699922756905795e-06,
      "loss": 0.8482,
      "step": 1991
    },
    {
      "epoch": 0.17964557875276188,
      "grad_norm": 2.028270707972517,
      "learning_rate": 3.7697201926692895e-06,
      "loss": 0.7474,
      "step": 1992
    },
    {
      "epoch": 0.17973576227623214,
      "grad_norm": 1.8279310596329261,
      "learning_rate": 3.7694479586457144e-06,
      "loss": 0.8032,
      "step": 1993
    },
    {
      "epoch": 0.1798259457997024,
      "grad_norm": 2.9665585108840404,
      "learning_rate": 3.7691755736430827e-06,
      "loss": 0.8058,
      "step": 1994
    },
    {
      "epoch": 0.17991612932317266,
      "grad_norm": 1.5894371468787856,
      "learning_rate": 3.768903037684636e-06,
      "loss": 0.8322,
      "step": 1995
    },
    {
      "epoch": 0.18000631284664292,
      "grad_norm": 1.6456826861362084,
      "learning_rate": 3.7686303507936284e-06,
      "loss": 0.8363,
      "step": 1996
    },
    {
      "epoch": 0.18009649637011318,
      "grad_norm": 2.6556930855749328,
      "learning_rate": 3.7683575129933272e-06,
      "loss": 0.8344,
      "step": 1997
    },
    {
      "epoch": 0.18018667989358345,
      "grad_norm": 1.5508840992835025,
      "learning_rate": 3.7680845243070128e-06,
      "loss": 0.7987,
      "step": 1998
    },
    {
      "epoch": 0.1802768634170537,
      "grad_norm": 0.7069361290778869,
      "learning_rate": 3.7678113847579767e-06,
      "loss": 0.5541,
      "step": 1999
    },
    {
      "epoch": 0.18036704694052397,
      "grad_norm": 2.001604985926004,
      "learning_rate": 3.7675380943695264e-06,
      "loss": 0.8623,
      "step": 2000
    },
    {
      "epoch": 0.18045723046399423,
      "grad_norm": 1.655653216250084,
      "learning_rate": 3.7672646531649795e-06,
      "loss": 0.7569,
      "step": 2001
    },
    {
      "epoch": 0.1805474139874645,
      "grad_norm": 0.5906039511749632,
      "learning_rate": 3.7669910611676682e-06,
      "loss": 0.5771,
      "step": 2002
    },
    {
      "epoch": 0.18063759751093475,
      "grad_norm": 2.123099093948709,
      "learning_rate": 3.7667173184009356e-06,
      "loss": 0.78,
      "step": 2003
    },
    {
      "epoch": 0.180727781034405,
      "grad_norm": 3.6720757612171027,
      "learning_rate": 3.7664434248881403e-06,
      "loss": 0.853,
      "step": 2004
    },
    {
      "epoch": 0.18081796455787527,
      "grad_norm": 2.28540705421209,
      "learning_rate": 3.766169380652652e-06,
      "loss": 0.8053,
      "step": 2005
    },
    {
      "epoch": 0.18090814808134553,
      "grad_norm": 1.8132679726354093,
      "learning_rate": 3.7658951857178537e-06,
      "loss": 0.8355,
      "step": 2006
    },
    {
      "epoch": 0.1809983316048158,
      "grad_norm": 1.6232716832735965,
      "learning_rate": 3.7656208401071414e-06,
      "loss": 0.7999,
      "step": 2007
    },
    {
      "epoch": 0.18108851512828605,
      "grad_norm": 2.4131552957580897,
      "learning_rate": 3.7653463438439225e-06,
      "loss": 0.7355,
      "step": 2008
    },
    {
      "epoch": 0.18117869865175631,
      "grad_norm": 1.8206674778150895,
      "learning_rate": 3.7650716969516203e-06,
      "loss": 0.8091,
      "step": 2009
    },
    {
      "epoch": 0.18126888217522658,
      "grad_norm": 1.8482250281759527,
      "learning_rate": 3.764796899453668e-06,
      "loss": 0.697,
      "step": 2010
    },
    {
      "epoch": 0.18135906569869684,
      "grad_norm": 1.6638508693460432,
      "learning_rate": 3.7645219513735134e-06,
      "loss": 0.8321,
      "step": 2011
    },
    {
      "epoch": 0.1814492492221671,
      "grad_norm": 1.6518410353068063,
      "learning_rate": 3.764246852734617e-06,
      "loss": 0.8372,
      "step": 2012
    },
    {
      "epoch": 0.18153943274563739,
      "grad_norm": 1.8496208472535582,
      "learning_rate": 3.7639716035604502e-06,
      "loss": 0.7665,
      "step": 2013
    },
    {
      "epoch": 0.18162961626910765,
      "grad_norm": 0.6421790695565218,
      "learning_rate": 3.7636962038745e-06,
      "loss": 0.5487,
      "step": 2014
    },
    {
      "epoch": 0.1817197997925779,
      "grad_norm": 1.893253296528244,
      "learning_rate": 3.763420653700265e-06,
      "loss": 0.7499,
      "step": 2015
    },
    {
      "epoch": 0.18180998331604817,
      "grad_norm": 1.5455904935262086,
      "learning_rate": 3.7631449530612565e-06,
      "loss": 0.7524,
      "step": 2016
    },
    {
      "epoch": 0.18190016683951843,
      "grad_norm": 1.526149897811352,
      "learning_rate": 3.762869101980999e-06,
      "loss": 0.8812,
      "step": 2017
    },
    {
      "epoch": 0.1819903503629887,
      "grad_norm": 1.899810294669234,
      "learning_rate": 3.7625931004830287e-06,
      "loss": 0.7986,
      "step": 2018
    },
    {
      "epoch": 0.18208053388645895,
      "grad_norm": 1.6338272227079274,
      "learning_rate": 3.7623169485908966e-06,
      "loss": 0.7498,
      "step": 2019
    },
    {
      "epoch": 0.1821707174099292,
      "grad_norm": 0.7217517037393025,
      "learning_rate": 3.7620406463281647e-06,
      "loss": 0.5952,
      "step": 2020
    },
    {
      "epoch": 0.18226090093339947,
      "grad_norm": 4.851952087190936,
      "learning_rate": 3.7617641937184095e-06,
      "loss": 0.8129,
      "step": 2021
    },
    {
      "epoch": 0.18235108445686973,
      "grad_norm": 1.96781311752409,
      "learning_rate": 3.761487590785219e-06,
      "loss": 0.826,
      "step": 2022
    },
    {
      "epoch": 0.18244126798034,
      "grad_norm": 1.6193108371403786,
      "learning_rate": 3.7612108375521942e-06,
      "loss": 0.7695,
      "step": 2023
    },
    {
      "epoch": 0.18253145150381025,
      "grad_norm": 1.717492493192434,
      "learning_rate": 3.76093393404295e-06,
      "loss": 0.842,
      "step": 2024
    },
    {
      "epoch": 0.18262163502728052,
      "grad_norm": 1.855026867098717,
      "learning_rate": 3.7606568802811126e-06,
      "loss": 0.7866,
      "step": 2025
    },
    {
      "epoch": 0.18271181855075078,
      "grad_norm": 2.062212028051169,
      "learning_rate": 3.760379676290322e-06,
      "loss": 0.804,
      "step": 2026
    },
    {
      "epoch": 0.18280200207422104,
      "grad_norm": 1.6250115907695852,
      "learning_rate": 3.760102322094231e-06,
      "loss": 0.7296,
      "step": 2027
    },
    {
      "epoch": 0.1828921855976913,
      "grad_norm": 2.1908800441927423,
      "learning_rate": 3.759824817716504e-06,
      "loss": 0.7588,
      "step": 2028
    },
    {
      "epoch": 0.18298236912116156,
      "grad_norm": 1.3245003099912782,
      "learning_rate": 3.759547163180821e-06,
      "loss": 0.8428,
      "step": 2029
    },
    {
      "epoch": 0.18307255264463182,
      "grad_norm": 1.8776375339486646,
      "learning_rate": 3.759269358510871e-06,
      "loss": 0.5932,
      "step": 2030
    },
    {
      "epoch": 0.18316273616810208,
      "grad_norm": 1.8907915152370467,
      "learning_rate": 3.75899140373036e-06,
      "loss": 0.6979,
      "step": 2031
    },
    {
      "epoch": 0.18325291969157234,
      "grad_norm": 2.3967983347979924,
      "learning_rate": 3.7587132988630028e-06,
      "loss": 0.8574,
      "step": 2032
    },
    {
      "epoch": 0.1833431032150426,
      "grad_norm": 2.1773492994398578,
      "learning_rate": 3.7584350439325295e-06,
      "loss": 0.8284,
      "step": 2033
    },
    {
      "epoch": 0.18343328673851286,
      "grad_norm": 1.8884922324806879,
      "learning_rate": 3.758156638962682e-06,
      "loss": 0.7832,
      "step": 2034
    },
    {
      "epoch": 0.18352347026198312,
      "grad_norm": 1.8230079700674497,
      "learning_rate": 3.757878083977216e-06,
      "loss": 0.7531,
      "step": 2035
    },
    {
      "epoch": 0.18361365378545338,
      "grad_norm": 2.14173726609925,
      "learning_rate": 3.7575993789999e-06,
      "loss": 0.7799,
      "step": 2036
    },
    {
      "epoch": 0.18370383730892367,
      "grad_norm": 4.8285178607990735,
      "learning_rate": 3.757320524054512e-06,
      "loss": 0.8271,
      "step": 2037
    },
    {
      "epoch": 0.18379402083239393,
      "grad_norm": 1.6169875016681938,
      "learning_rate": 3.757041519164848e-06,
      "loss": 0.727,
      "step": 2038
    },
    {
      "epoch": 0.1838842043558642,
      "grad_norm": 1.8443753749804372,
      "learning_rate": 3.7567623643547133e-06,
      "loss": 0.7975,
      "step": 2039
    },
    {
      "epoch": 0.18397438787933446,
      "grad_norm": 1.7807078289290688,
      "learning_rate": 3.756483059647927e-06,
      "loss": 0.7953,
      "step": 2040
    },
    {
      "epoch": 0.18406457140280472,
      "grad_norm": 1.455873740380981,
      "learning_rate": 3.756203605068321e-06,
      "loss": 0.7614,
      "step": 2041
    },
    {
      "epoch": 0.18415475492627498,
      "grad_norm": 1.8253851027010277,
      "learning_rate": 3.7559240006397396e-06,
      "loss": 0.7845,
      "step": 2042
    },
    {
      "epoch": 0.18424493844974524,
      "grad_norm": 1.6830735254756348,
      "learning_rate": 3.7556442463860406e-06,
      "loss": 0.9088,
      "step": 2043
    },
    {
      "epoch": 0.1843351219732155,
      "grad_norm": 2.030948675253733,
      "learning_rate": 3.7553643423310934e-06,
      "loss": 0.8991,
      "step": 2044
    },
    {
      "epoch": 0.18442530549668576,
      "grad_norm": 3.6282041135273113,
      "learning_rate": 3.755084288498782e-06,
      "loss": 0.6492,
      "step": 2045
    },
    {
      "epoch": 0.18451548902015602,
      "grad_norm": 1.8068485284724727,
      "learning_rate": 3.754804084913002e-06,
      "loss": 0.8011,
      "step": 2046
    },
    {
      "epoch": 0.18460567254362628,
      "grad_norm": 1.5689528163187683,
      "learning_rate": 3.754523731597661e-06,
      "loss": 0.8575,
      "step": 2047
    },
    {
      "epoch": 0.18469585606709654,
      "grad_norm": 1.855705744013546,
      "learning_rate": 3.754243228576681e-06,
      "loss": 0.6928,
      "step": 2048
    },
    {
      "epoch": 0.1847860395905668,
      "grad_norm": 2.1075990335493757,
      "learning_rate": 3.753962575873996e-06,
      "loss": 0.851,
      "step": 2049
    },
    {
      "epoch": 0.18487622311403706,
      "grad_norm": 1.9587231782343917,
      "learning_rate": 3.7536817735135527e-06,
      "loss": 0.7504,
      "step": 2050
    },
    {
      "epoch": 0.18496640663750732,
      "grad_norm": 1.4657281672559463,
      "learning_rate": 3.753400821519311e-06,
      "loss": 0.8352,
      "step": 2051
    },
    {
      "epoch": 0.18505659016097759,
      "grad_norm": 1.5950724780796477,
      "learning_rate": 3.7531197199152426e-06,
      "loss": 0.7266,
      "step": 2052
    },
    {
      "epoch": 0.18514677368444785,
      "grad_norm": 0.7116055546175485,
      "learning_rate": 3.7528384687253335e-06,
      "loss": 0.6184,
      "step": 2053
    },
    {
      "epoch": 0.1852369572079181,
      "grad_norm": 1.720258709502268,
      "learning_rate": 3.7525570679735815e-06,
      "loss": 0.9038,
      "step": 2054
    },
    {
      "epoch": 0.18532714073138837,
      "grad_norm": 1.6879521929673862,
      "learning_rate": 3.7522755176839965e-06,
      "loss": 0.8073,
      "step": 2055
    },
    {
      "epoch": 0.18541732425485863,
      "grad_norm": 2.643609521619146,
      "learning_rate": 3.7519938178806027e-06,
      "loss": 0.8039,
      "step": 2056
    },
    {
      "epoch": 0.1855075077783289,
      "grad_norm": 2.5528757746439914,
      "learning_rate": 3.7517119685874358e-06,
      "loss": 0.7511,
      "step": 2057
    },
    {
      "epoch": 0.18559769130179915,
      "grad_norm": 1.5083368210023351,
      "learning_rate": 3.7514299698285447e-06,
      "loss": 0.7826,
      "step": 2058
    },
    {
      "epoch": 0.1856878748252694,
      "grad_norm": 1.5911344120928914,
      "learning_rate": 3.751147821627991e-06,
      "loss": 0.7719,
      "step": 2059
    },
    {
      "epoch": 0.18577805834873967,
      "grad_norm": 2.0948629125932308,
      "learning_rate": 3.75086552400985e-06,
      "loss": 0.8102,
      "step": 2060
    },
    {
      "epoch": 0.18586824187220996,
      "grad_norm": 1.9203938304790102,
      "learning_rate": 3.750583076998208e-06,
      "loss": 0.7689,
      "step": 2061
    },
    {
      "epoch": 0.18595842539568022,
      "grad_norm": 1.6122433923984298,
      "learning_rate": 3.7503004806171655e-06,
      "loss": 0.7686,
      "step": 2062
    },
    {
      "epoch": 0.18604860891915048,
      "grad_norm": 2.3232638408971473,
      "learning_rate": 3.7500177348908354e-06,
      "loss": 0.7251,
      "step": 2063
    },
    {
      "epoch": 0.18613879244262074,
      "grad_norm": 1.4404532737740547,
      "learning_rate": 3.749734839843342e-06,
      "loss": 0.8153,
      "step": 2064
    },
    {
      "epoch": 0.186228975966091,
      "grad_norm": 2.53325846544698,
      "learning_rate": 3.7494517954988245e-06,
      "loss": 0.7944,
      "step": 2065
    },
    {
      "epoch": 0.18631915948956126,
      "grad_norm": 1.7823643461294651,
      "learning_rate": 3.749168601881433e-06,
      "loss": 0.7894,
      "step": 2066
    },
    {
      "epoch": 0.18640934301303153,
      "grad_norm": 2.2748315476313574,
      "learning_rate": 3.7488852590153315e-06,
      "loss": 0.7762,
      "step": 2067
    },
    {
      "epoch": 0.1864995265365018,
      "grad_norm": 1.5369882305640254,
      "learning_rate": 3.748601766924697e-06,
      "loss": 0.7844,
      "step": 2068
    },
    {
      "epoch": 0.18658971005997205,
      "grad_norm": 1.8036269998150034,
      "learning_rate": 3.7483181256337176e-06,
      "loss": 0.757,
      "step": 2069
    },
    {
      "epoch": 0.1866798935834423,
      "grad_norm": 1.6834301791643567,
      "learning_rate": 3.7480343351665962e-06,
      "loss": 0.7937,
      "step": 2070
    },
    {
      "epoch": 0.18677007710691257,
      "grad_norm": 1.6744898375017407,
      "learning_rate": 3.747750395547546e-06,
      "loss": 0.7781,
      "step": 2071
    },
    {
      "epoch": 0.18686026063038283,
      "grad_norm": 1.3069130195914294,
      "learning_rate": 3.7474663068007956e-06,
      "loss": 0.7753,
      "step": 2072
    },
    {
      "epoch": 0.1869504441538531,
      "grad_norm": 1.2183199881950386,
      "learning_rate": 3.747182068950584e-06,
      "loss": 0.7707,
      "step": 2073
    },
    {
      "epoch": 0.18704062767732335,
      "grad_norm": 3.0129230629574297,
      "learning_rate": 3.7468976820211643e-06,
      "loss": 0.7916,
      "step": 2074
    },
    {
      "epoch": 0.1871308112007936,
      "grad_norm": 2.448171583389344,
      "learning_rate": 3.746613146036803e-06,
      "loss": 0.7914,
      "step": 2075
    },
    {
      "epoch": 0.18722099472426387,
      "grad_norm": 1.6152299468084705,
      "learning_rate": 3.7463284610217766e-06,
      "loss": 0.8353,
      "step": 2076
    },
    {
      "epoch": 0.18731117824773413,
      "grad_norm": 1.8276677008002138,
      "learning_rate": 3.746043627000377e-06,
      "loss": 0.729,
      "step": 2077
    },
    {
      "epoch": 0.1874013617712044,
      "grad_norm": 1.6450071393817547,
      "learning_rate": 3.7457586439969076e-06,
      "loss": 0.7394,
      "step": 2078
    },
    {
      "epoch": 0.18749154529467466,
      "grad_norm": 1.3342192011697402,
      "learning_rate": 3.7454735120356842e-06,
      "loss": 0.7809,
      "step": 2079
    },
    {
      "epoch": 0.18758172881814492,
      "grad_norm": 1.7373919666103241,
      "learning_rate": 3.7451882311410373e-06,
      "loss": 0.8605,
      "step": 2080
    },
    {
      "epoch": 0.18767191234161518,
      "grad_norm": 1.678889382172456,
      "learning_rate": 3.7449028013373074e-06,
      "loss": 0.8269,
      "step": 2081
    },
    {
      "epoch": 0.18776209586508544,
      "grad_norm": 2.036206229004654,
      "learning_rate": 3.7446172226488485e-06,
      "loss": 0.7872,
      "step": 2082
    },
    {
      "epoch": 0.1878522793885557,
      "grad_norm": 2.7436625542750255,
      "learning_rate": 3.7443314951000285e-06,
      "loss": 0.8535,
      "step": 2083
    },
    {
      "epoch": 0.18794246291202596,
      "grad_norm": 1.7788481999819719,
      "learning_rate": 3.7440456187152276e-06,
      "loss": 0.782,
      "step": 2084
    },
    {
      "epoch": 0.18803264643549625,
      "grad_norm": 1.9668843951635766,
      "learning_rate": 3.7437595935188377e-06,
      "loss": 0.8767,
      "step": 2085
    },
    {
      "epoch": 0.1881228299589665,
      "grad_norm": 1.990119189522082,
      "learning_rate": 3.7434734195352647e-06,
      "loss": 0.7906,
      "step": 2086
    },
    {
      "epoch": 0.18821301348243677,
      "grad_norm": 0.6639479370029406,
      "learning_rate": 3.743187096788926e-06,
      "loss": 0.5809,
      "step": 2087
    },
    {
      "epoch": 0.18830319700590703,
      "grad_norm": 1.606389730587654,
      "learning_rate": 3.7429006253042524e-06,
      "loss": 0.7869,
      "step": 2088
    },
    {
      "epoch": 0.1883933805293773,
      "grad_norm": 1.6492781823100804,
      "learning_rate": 3.7426140051056867e-06,
      "loss": 0.733,
      "step": 2089
    },
    {
      "epoch": 0.18848356405284755,
      "grad_norm": 1.8221293538485055,
      "learning_rate": 3.7423272362176856e-06,
      "loss": 0.7329,
      "step": 2090
    },
    {
      "epoch": 0.1885737475763178,
      "grad_norm": 1.5470755090711708,
      "learning_rate": 3.742040318664718e-06,
      "loss": 0.833,
      "step": 2091
    },
    {
      "epoch": 0.18866393109978807,
      "grad_norm": 2.2456939606837056,
      "learning_rate": 3.7417532524712643e-06,
      "loss": 0.8473,
      "step": 2092
    },
    {
      "epoch": 0.18875411462325833,
      "grad_norm": 1.4831819257007355,
      "learning_rate": 3.7414660376618195e-06,
      "loss": 0.8221,
      "step": 2093
    },
    {
      "epoch": 0.1888442981467286,
      "grad_norm": 0.6474685093133445,
      "learning_rate": 3.74117867426089e-06,
      "loss": 0.5904,
      "step": 2094
    },
    {
      "epoch": 0.18893448167019886,
      "grad_norm": 2.3269231369445365,
      "learning_rate": 3.7408911622929954e-06,
      "loss": 0.7689,
      "step": 2095
    },
    {
      "epoch": 0.18902466519366912,
      "grad_norm": 1.4911049316140499,
      "learning_rate": 3.740603501782668e-06,
      "loss": 0.8283,
      "step": 2096
    },
    {
      "epoch": 0.18911484871713938,
      "grad_norm": 1.7186676005638866,
      "learning_rate": 3.7403156927544516e-06,
      "loss": 0.7354,
      "step": 2097
    },
    {
      "epoch": 0.18920503224060964,
      "grad_norm": 2.109076245662036,
      "learning_rate": 3.740027735232904e-06,
      "loss": 0.7216,
      "step": 2098
    },
    {
      "epoch": 0.1892952157640799,
      "grad_norm": 1.7644599514116204,
      "learning_rate": 3.7397396292425966e-06,
      "loss": 0.7757,
      "step": 2099
    },
    {
      "epoch": 0.18938539928755016,
      "grad_norm": 1.835966199285328,
      "learning_rate": 3.7394513748081105e-06,
      "loss": 0.7685,
      "step": 2100
    },
    {
      "epoch": 0.18947558281102042,
      "grad_norm": 1.6892364186566204,
      "learning_rate": 3.7391629719540418e-06,
      "loss": 0.8186,
      "step": 2101
    },
    {
      "epoch": 0.18956576633449068,
      "grad_norm": 1.4591275187058623,
      "learning_rate": 3.7388744207049998e-06,
      "loss": 0.7788,
      "step": 2102
    },
    {
      "epoch": 0.18965594985796094,
      "grad_norm": 2.1642435046355524,
      "learning_rate": 3.7385857210856027e-06,
      "loss": 0.7663,
      "step": 2103
    },
    {
      "epoch": 0.1897461333814312,
      "grad_norm": 3.240717469591017,
      "learning_rate": 3.738296873120486e-06,
      "loss": 0.7707,
      "step": 2104
    },
    {
      "epoch": 0.18983631690490146,
      "grad_norm": 1.713027583250358,
      "learning_rate": 3.7380078768342955e-06,
      "loss": 0.7603,
      "step": 2105
    },
    {
      "epoch": 0.18992650042837173,
      "grad_norm": 2.067038311408102,
      "learning_rate": 3.7377187322516895e-06,
      "loss": 0.8061,
      "step": 2106
    },
    {
      "epoch": 0.19001668395184199,
      "grad_norm": 1.5931452183359,
      "learning_rate": 3.7374294393973395e-06,
      "loss": 0.7461,
      "step": 2107
    },
    {
      "epoch": 0.19010686747531225,
      "grad_norm": 1.5223356517164066,
      "learning_rate": 3.7371399982959294e-06,
      "loss": 0.8369,
      "step": 2108
    },
    {
      "epoch": 0.19019705099878254,
      "grad_norm": 2.233518636427191,
      "learning_rate": 3.7368504089721565e-06,
      "loss": 0.7066,
      "step": 2109
    },
    {
      "epoch": 0.1902872345222528,
      "grad_norm": 1.5502567509302876,
      "learning_rate": 3.73656067145073e-06,
      "loss": 0.7343,
      "step": 2110
    },
    {
      "epoch": 0.19037741804572306,
      "grad_norm": 1.6593853046794125,
      "learning_rate": 3.736270785756371e-06,
      "loss": 0.8398,
      "step": 2111
    },
    {
      "epoch": 0.19046760156919332,
      "grad_norm": 2.0361482688262273,
      "learning_rate": 3.7359807519138156e-06,
      "loss": 0.7994,
      "step": 2112
    },
    {
      "epoch": 0.19055778509266358,
      "grad_norm": 1.4423269417285172,
      "learning_rate": 3.73569056994781e-06,
      "loss": 0.7818,
      "step": 2113
    },
    {
      "epoch": 0.19064796861613384,
      "grad_norm": 2.107695073051502,
      "learning_rate": 3.7354002398831144e-06,
      "loss": 0.8287,
      "step": 2114
    },
    {
      "epoch": 0.1907381521396041,
      "grad_norm": 1.9259340794335003,
      "learning_rate": 3.7351097617445015e-06,
      "loss": 0.7919,
      "step": 2115
    },
    {
      "epoch": 0.19082833566307436,
      "grad_norm": 0.6950818868807335,
      "learning_rate": 3.7348191355567567e-06,
      "loss": 0.6118,
      "step": 2116
    },
    {
      "epoch": 0.19091851918654462,
      "grad_norm": 1.6233173608614233,
      "learning_rate": 3.734528361344677e-06,
      "loss": 0.8154,
      "step": 2117
    },
    {
      "epoch": 0.19100870271001488,
      "grad_norm": 1.7106957112346917,
      "learning_rate": 3.734237439133074e-06,
      "loss": 0.7843,
      "step": 2118
    },
    {
      "epoch": 0.19109888623348514,
      "grad_norm": 1.733332277566637,
      "learning_rate": 3.7339463689467702e-06,
      "loss": 0.8372,
      "step": 2119
    },
    {
      "epoch": 0.1911890697569554,
      "grad_norm": 1.7954280168704697,
      "learning_rate": 3.733655150810601e-06,
      "loss": 0.7575,
      "step": 2120
    },
    {
      "epoch": 0.19127925328042567,
      "grad_norm": 2.0711228422224703,
      "learning_rate": 3.7333637847494154e-06,
      "loss": 0.9205,
      "step": 2121
    },
    {
      "epoch": 0.19136943680389593,
      "grad_norm": 1.7393777539268787,
      "learning_rate": 3.7330722707880734e-06,
      "loss": 0.9419,
      "step": 2122
    },
    {
      "epoch": 0.1914596203273662,
      "grad_norm": 1.39998485693235,
      "learning_rate": 3.7327806089514497e-06,
      "loss": 0.6945,
      "step": 2123
    },
    {
      "epoch": 0.19154980385083645,
      "grad_norm": 1.5006513770731278,
      "learning_rate": 3.7324887992644297e-06,
      "loss": 0.884,
      "step": 2124
    },
    {
      "epoch": 0.1916399873743067,
      "grad_norm": 0.7271809560680491,
      "learning_rate": 3.7321968417519123e-06,
      "loss": 0.5864,
      "step": 2125
    },
    {
      "epoch": 0.19173017089777697,
      "grad_norm": 1.9991969045896711,
      "learning_rate": 3.7319047364388097e-06,
      "loss": 0.7749,
      "step": 2126
    },
    {
      "epoch": 0.19182035442124723,
      "grad_norm": 0.6351733732085492,
      "learning_rate": 3.7316124833500453e-06,
      "loss": 0.5723,
      "step": 2127
    },
    {
      "epoch": 0.1919105379447175,
      "grad_norm": 2.2876420023883344,
      "learning_rate": 3.731320082510556e-06,
      "loss": 0.7734,
      "step": 2128
    },
    {
      "epoch": 0.19200072146818775,
      "grad_norm": 2.165769599171518,
      "learning_rate": 3.7310275339452906e-06,
      "loss": 0.8022,
      "step": 2129
    },
    {
      "epoch": 0.192090904991658,
      "grad_norm": 6.0468459806312715,
      "learning_rate": 3.7307348376792113e-06,
      "loss": 0.7778,
      "step": 2130
    },
    {
      "epoch": 0.19218108851512827,
      "grad_norm": 1.532207267644606,
      "learning_rate": 3.730441993737292e-06,
      "loss": 0.7823,
      "step": 2131
    },
    {
      "epoch": 0.19227127203859856,
      "grad_norm": 2.038036804630576,
      "learning_rate": 3.7301490021445205e-06,
      "loss": 0.7735,
      "step": 2132
    },
    {
      "epoch": 0.19236145556206882,
      "grad_norm": 2.1095765194176,
      "learning_rate": 3.7298558629258966e-06,
      "loss": 0.7326,
      "step": 2133
    },
    {
      "epoch": 0.19245163908553908,
      "grad_norm": 1.5598184081293283,
      "learning_rate": 3.7295625761064314e-06,
      "loss": 0.825,
      "step": 2134
    },
    {
      "epoch": 0.19254182260900934,
      "grad_norm": 3.4688580470397885,
      "learning_rate": 3.7292691417111504e-06,
      "loss": 0.755,
      "step": 2135
    },
    {
      "epoch": 0.1926320061324796,
      "grad_norm": 1.6013258596373463,
      "learning_rate": 3.728975559765092e-06,
      "loss": 0.7468,
      "step": 2136
    },
    {
      "epoch": 0.19272218965594987,
      "grad_norm": 1.9910765897694958,
      "learning_rate": 3.728681830293305e-06,
      "loss": 0.8163,
      "step": 2137
    },
    {
      "epoch": 0.19281237317942013,
      "grad_norm": 1.6241678161121187,
      "learning_rate": 3.7283879533208523e-06,
      "loss": 0.7474,
      "step": 2138
    },
    {
      "epoch": 0.1929025567028904,
      "grad_norm": 1.3743417204487536,
      "learning_rate": 3.7280939288728094e-06,
      "loss": 0.7722,
      "step": 2139
    },
    {
      "epoch": 0.19299274022636065,
      "grad_norm": 1.860878600910839,
      "learning_rate": 3.7277997569742637e-06,
      "loss": 0.7513,
      "step": 2140
    },
    {
      "epoch": 0.1930829237498309,
      "grad_norm": 4.647079961960151,
      "learning_rate": 3.7275054376503155e-06,
      "loss": 0.7565,
      "step": 2141
    },
    {
      "epoch": 0.19317310727330117,
      "grad_norm": 1.9867619253635653,
      "learning_rate": 3.7272109709260783e-06,
      "loss": 0.7306,
      "step": 2142
    },
    {
      "epoch": 0.19326329079677143,
      "grad_norm": 0.7681947517229645,
      "learning_rate": 3.7269163568266774e-06,
      "loss": 0.5883,
      "step": 2143
    },
    {
      "epoch": 0.1933534743202417,
      "grad_norm": 1.7172389237094834,
      "learning_rate": 3.7266215953772512e-06,
      "loss": 0.774,
      "step": 2144
    },
    {
      "epoch": 0.19344365784371195,
      "grad_norm": 2.9015492806161745,
      "learning_rate": 3.7263266866029492e-06,
      "loss": 0.8206,
      "step": 2145
    },
    {
      "epoch": 0.1935338413671822,
      "grad_norm": 1.4411250307902217,
      "learning_rate": 3.726031630528936e-06,
      "loss": 0.7586,
      "step": 2146
    },
    {
      "epoch": 0.19362402489065247,
      "grad_norm": 2.6835791797048056,
      "learning_rate": 3.7257364271803865e-06,
      "loss": 0.6715,
      "step": 2147
    },
    {
      "epoch": 0.19371420841412273,
      "grad_norm": 2.077801672083547,
      "learning_rate": 3.7254410765824896e-06,
      "loss": 0.7126,
      "step": 2148
    },
    {
      "epoch": 0.193804391937593,
      "grad_norm": 1.8966545869323037,
      "learning_rate": 3.725145578760446e-06,
      "loss": 0.7029,
      "step": 2149
    },
    {
      "epoch": 0.19389457546106326,
      "grad_norm": 2.4336604664874266,
      "learning_rate": 3.7248499337394696e-06,
      "loss": 0.7034,
      "step": 2150
    },
    {
      "epoch": 0.19398475898453352,
      "grad_norm": 1.9306239894322164,
      "learning_rate": 3.7245541415447848e-06,
      "loss": 0.8737,
      "step": 2151
    },
    {
      "epoch": 0.19407494250800378,
      "grad_norm": 1.740058521843184,
      "learning_rate": 3.724258202201633e-06,
      "loss": 0.8033,
      "step": 2152
    },
    {
      "epoch": 0.19416512603147404,
      "grad_norm": 1.8002761496536015,
      "learning_rate": 3.7239621157352633e-06,
      "loss": 0.7983,
      "step": 2153
    },
    {
      "epoch": 0.1942553095549443,
      "grad_norm": 1.9122735207290993,
      "learning_rate": 3.7236658821709403e-06,
      "loss": 0.7718,
      "step": 2154
    },
    {
      "epoch": 0.19434549307841456,
      "grad_norm": 1.7341207369444906,
      "learning_rate": 3.7233695015339404e-06,
      "loss": 0.8259,
      "step": 2155
    },
    {
      "epoch": 0.19443567660188485,
      "grad_norm": 1.6506077369674086,
      "learning_rate": 3.7230729738495513e-06,
      "loss": 0.7686,
      "step": 2156
    },
    {
      "epoch": 0.1945258601253551,
      "grad_norm": 1.8122480809320896,
      "learning_rate": 3.722776299143075e-06,
      "loss": 0.7791,
      "step": 2157
    },
    {
      "epoch": 0.19461604364882537,
      "grad_norm": 0.6666803929783444,
      "learning_rate": 3.722479477439826e-06,
      "loss": 0.5869,
      "step": 2158
    },
    {
      "epoch": 0.19470622717229563,
      "grad_norm": 1.3170184967115066,
      "learning_rate": 3.7221825087651306e-06,
      "loss": 0.7624,
      "step": 2159
    },
    {
      "epoch": 0.1947964106957659,
      "grad_norm": 2.1866747798226425,
      "learning_rate": 3.7218853931443274e-06,
      "loss": 0.7215,
      "step": 2160
    },
    {
      "epoch": 0.19488659421923615,
      "grad_norm": 1.3110941669389347,
      "learning_rate": 3.721588130602768e-06,
      "loss": 0.7245,
      "step": 2161
    },
    {
      "epoch": 0.19497677774270641,
      "grad_norm": 2.02281597710565,
      "learning_rate": 3.7212907211658164e-06,
      "loss": 0.7617,
      "step": 2162
    },
    {
      "epoch": 0.19506696126617667,
      "grad_norm": 1.5197291698857516,
      "learning_rate": 3.72099316485885e-06,
      "loss": 0.752,
      "step": 2163
    },
    {
      "epoch": 0.19515714478964694,
      "grad_norm": 1.5027953168997599,
      "learning_rate": 3.720695461707256e-06,
      "loss": 0.8165,
      "step": 2164
    },
    {
      "epoch": 0.1952473283131172,
      "grad_norm": 3.545116644098424,
      "learning_rate": 3.7203976117364383e-06,
      "loss": 0.8052,
      "step": 2165
    },
    {
      "epoch": 0.19533751183658746,
      "grad_norm": 1.9952013742065897,
      "learning_rate": 3.7200996149718105e-06,
      "loss": 0.7683,
      "step": 2166
    },
    {
      "epoch": 0.19542769536005772,
      "grad_norm": 1.3723233787139133,
      "learning_rate": 3.7198014714387985e-06,
      "loss": 0.7692,
      "step": 2167
    },
    {
      "epoch": 0.19551787888352798,
      "grad_norm": 2.1864247540468074,
      "learning_rate": 3.7195031811628422e-06,
      "loss": 0.7129,
      "step": 2168
    },
    {
      "epoch": 0.19560806240699824,
      "grad_norm": 2.6111054307681942,
      "learning_rate": 3.719204744169393e-06,
      "loss": 0.813,
      "step": 2169
    },
    {
      "epoch": 0.1956982459304685,
      "grad_norm": 1.9171408674791393,
      "learning_rate": 3.718906160483916e-06,
      "loss": 0.7577,
      "step": 2170
    },
    {
      "epoch": 0.19578842945393876,
      "grad_norm": 1.5633104130027902,
      "learning_rate": 3.7186074301318868e-06,
      "loss": 0.7541,
      "step": 2171
    },
    {
      "epoch": 0.19587861297740902,
      "grad_norm": 1.55497302016621,
      "learning_rate": 3.7183085531387957e-06,
      "loss": 0.8552,
      "step": 2172
    },
    {
      "epoch": 0.19596879650087928,
      "grad_norm": 1.6657539332087337,
      "learning_rate": 3.7180095295301443e-06,
      "loss": 0.7562,
      "step": 2173
    },
    {
      "epoch": 0.19605898002434954,
      "grad_norm": 4.0363212447318695,
      "learning_rate": 3.7177103593314465e-06,
      "loss": 0.8446,
      "step": 2174
    },
    {
      "epoch": 0.1961491635478198,
      "grad_norm": 1.7299861880947303,
      "learning_rate": 3.7174110425682297e-06,
      "loss": 0.8157,
      "step": 2175
    },
    {
      "epoch": 0.19623934707129007,
      "grad_norm": 1.9121002103529414,
      "learning_rate": 3.7171115792660333e-06,
      "loss": 0.9035,
      "step": 2176
    },
    {
      "epoch": 0.19632953059476033,
      "grad_norm": 1.632596234322789,
      "learning_rate": 3.7168119694504083e-06,
      "loss": 0.8487,
      "step": 2177
    },
    {
      "epoch": 0.1964197141182306,
      "grad_norm": 1.8932866048867003,
      "learning_rate": 3.71651221314692e-06,
      "loss": 0.7792,
      "step": 2178
    },
    {
      "epoch": 0.19650989764170085,
      "grad_norm": 1.6391589926362418,
      "learning_rate": 3.716212310381145e-06,
      "loss": 0.7801,
      "step": 2179
    },
    {
      "epoch": 0.19660008116517114,
      "grad_norm": 1.5998946274391608,
      "learning_rate": 3.7159122611786725e-06,
      "loss": 0.8204,
      "step": 2180
    },
    {
      "epoch": 0.1966902646886414,
      "grad_norm": 1.6361541866870388,
      "learning_rate": 3.7156120655651045e-06,
      "loss": 0.7675,
      "step": 2181
    },
    {
      "epoch": 0.19678044821211166,
      "grad_norm": 1.3583878188331822,
      "learning_rate": 3.7153117235660553e-06,
      "loss": 0.7615,
      "step": 2182
    },
    {
      "epoch": 0.19687063173558192,
      "grad_norm": 0.7091888889397645,
      "learning_rate": 3.7150112352071514e-06,
      "loss": 0.5862,
      "step": 2183
    },
    {
      "epoch": 0.19696081525905218,
      "grad_norm": 1.9025739651912832,
      "learning_rate": 3.7147106005140326e-06,
      "loss": 0.747,
      "step": 2184
    },
    {
      "epoch": 0.19705099878252244,
      "grad_norm": 0.6773159483381553,
      "learning_rate": 3.714409819512351e-06,
      "loss": 0.5947,
      "step": 2185
    },
    {
      "epoch": 0.1971411823059927,
      "grad_norm": 2.472638604702295,
      "learning_rate": 3.7141088922277695e-06,
      "loss": 0.8075,
      "step": 2186
    },
    {
      "epoch": 0.19723136582946296,
      "grad_norm": 1.764095082417144,
      "learning_rate": 3.7138078186859664e-06,
      "loss": 0.8289,
      "step": 2187
    },
    {
      "epoch": 0.19732154935293322,
      "grad_norm": 1.5935946370113092,
      "learning_rate": 3.7135065989126303e-06,
      "loss": 0.7943,
      "step": 2188
    },
    {
      "epoch": 0.19741173287640348,
      "grad_norm": 0.7530651242842608,
      "learning_rate": 3.713205232933463e-06,
      "loss": 0.6498,
      "step": 2189
    },
    {
      "epoch": 0.19750191639987374,
      "grad_norm": 1.5537810630485969,
      "learning_rate": 3.7129037207741792e-06,
      "loss": 0.7884,
      "step": 2190
    },
    {
      "epoch": 0.197592099923344,
      "grad_norm": 1.8396713921224226,
      "learning_rate": 3.7126020624605046e-06,
      "loss": 0.7949,
      "step": 2191
    },
    {
      "epoch": 0.19768228344681427,
      "grad_norm": 1.8594205193597715,
      "learning_rate": 3.7123002580181785e-06,
      "loss": 0.7393,
      "step": 2192
    },
    {
      "epoch": 0.19777246697028453,
      "grad_norm": 1.3597794731243722,
      "learning_rate": 3.7119983074729532e-06,
      "loss": 0.8537,
      "step": 2193
    },
    {
      "epoch": 0.1978626504937548,
      "grad_norm": 1.6885199819776975,
      "learning_rate": 3.7116962108505926e-06,
      "loss": 0.8368,
      "step": 2194
    },
    {
      "epoch": 0.19795283401722505,
      "grad_norm": 1.3979867816212819,
      "learning_rate": 3.711393968176873e-06,
      "loss": 0.8786,
      "step": 2195
    },
    {
      "epoch": 0.1980430175406953,
      "grad_norm": 1.7572194094934719,
      "learning_rate": 3.711091579477584e-06,
      "loss": 0.788,
      "step": 2196
    },
    {
      "epoch": 0.19813320106416557,
      "grad_norm": 2.0546398084784174,
      "learning_rate": 3.7107890447785255e-06,
      "loss": 0.8178,
      "step": 2197
    },
    {
      "epoch": 0.19822338458763583,
      "grad_norm": 1.8683554854881896,
      "learning_rate": 3.710486364105513e-06,
      "loss": 0.835,
      "step": 2198
    },
    {
      "epoch": 0.1983135681111061,
      "grad_norm": 2.100319420272853,
      "learning_rate": 3.7101835374843728e-06,
      "loss": 0.7808,
      "step": 2199
    },
    {
      "epoch": 0.19840375163457635,
      "grad_norm": 1.5033615909523015,
      "learning_rate": 3.7098805649409427e-06,
      "loss": 0.807,
      "step": 2200
    },
    {
      "epoch": 0.1984939351580466,
      "grad_norm": 1.7666153957316677,
      "learning_rate": 3.7095774465010748e-06,
      "loss": 0.8499,
      "step": 2201
    },
    {
      "epoch": 0.19858411868151687,
      "grad_norm": 1.656575440876357,
      "learning_rate": 3.7092741821906328e-06,
      "loss": 0.7855,
      "step": 2202
    },
    {
      "epoch": 0.19867430220498714,
      "grad_norm": 1.8529411366228514,
      "learning_rate": 3.708970772035493e-06,
      "loss": 0.7643,
      "step": 2203
    },
    {
      "epoch": 0.19876448572845742,
      "grad_norm": 1.3986446717314431,
      "learning_rate": 3.7086672160615427e-06,
      "loss": 0.7528,
      "step": 2204
    },
    {
      "epoch": 0.19885466925192768,
      "grad_norm": 2.8371522261797306,
      "learning_rate": 3.7083635142946852e-06,
      "loss": 0.7667,
      "step": 2205
    },
    {
      "epoch": 0.19894485277539795,
      "grad_norm": 1.6787402653642696,
      "learning_rate": 3.7080596667608327e-06,
      "loss": 0.82,
      "step": 2206
    },
    {
      "epoch": 0.1990350362988682,
      "grad_norm": 1.5636135710825299,
      "learning_rate": 3.707755673485911e-06,
      "loss": 0.829,
      "step": 2207
    },
    {
      "epoch": 0.19912521982233847,
      "grad_norm": 1.6605409423320128,
      "learning_rate": 3.7074515344958584e-06,
      "loss": 0.6966,
      "step": 2208
    },
    {
      "epoch": 0.19921540334580873,
      "grad_norm": 0.6597692269316712,
      "learning_rate": 3.707147249816627e-06,
      "loss": 0.614,
      "step": 2209
    },
    {
      "epoch": 0.199305586869279,
      "grad_norm": 1.5551845196967997,
      "learning_rate": 3.706842819474178e-06,
      "loss": 0.8062,
      "step": 2210
    },
    {
      "epoch": 0.19939577039274925,
      "grad_norm": 1.7533089825108055,
      "learning_rate": 3.706538243494489e-06,
      "loss": 0.7293,
      "step": 2211
    },
    {
      "epoch": 0.1994859539162195,
      "grad_norm": 1.7237739656482245,
      "learning_rate": 3.706233521903547e-06,
      "loss": 0.7613,
      "step": 2212
    },
    {
      "epoch": 0.19957613743968977,
      "grad_norm": 2.209471954851408,
      "learning_rate": 3.705928654727353e-06,
      "loss": 0.8654,
      "step": 2213
    },
    {
      "epoch": 0.19966632096316003,
      "grad_norm": 6.2177101684438085,
      "learning_rate": 3.7056236419919195e-06,
      "loss": 0.7678,
      "step": 2214
    },
    {
      "epoch": 0.1997565044866303,
      "grad_norm": 1.8847180612521748,
      "learning_rate": 3.705318483723273e-06,
      "loss": 0.8341,
      "step": 2215
    },
    {
      "epoch": 0.19984668801010055,
      "grad_norm": 1.6320317057914335,
      "learning_rate": 3.7050131799474493e-06,
      "loss": 0.7619,
      "step": 2216
    },
    {
      "epoch": 0.19993687153357081,
      "grad_norm": 1.4271222222962636,
      "learning_rate": 3.7047077306905e-06,
      "loss": 0.786,
      "step": 2217
    },
    {
      "epoch": 0.20002705505704108,
      "grad_norm": 1.5177843893871212,
      "learning_rate": 3.704402135978488e-06,
      "loss": 0.8185,
      "step": 2218
    },
    {
      "epoch": 0.20011723858051134,
      "grad_norm": 2.3937459283301084,
      "learning_rate": 3.7040963958374877e-06,
      "loss": 0.8096,
      "step": 2219
    },
    {
      "epoch": 0.2002074221039816,
      "grad_norm": 1.4204588324742216,
      "learning_rate": 3.7037905102935864e-06,
      "loss": 0.8178,
      "step": 2220
    },
    {
      "epoch": 0.20029760562745186,
      "grad_norm": 1.5988069974210275,
      "learning_rate": 3.7034844793728837e-06,
      "loss": 0.7964,
      "step": 2221
    },
    {
      "epoch": 0.20038778915092212,
      "grad_norm": 1.540051447516782,
      "learning_rate": 3.7031783031014933e-06,
      "loss": 0.6715,
      "step": 2222
    },
    {
      "epoch": 0.20047797267439238,
      "grad_norm": 2.218759026307564,
      "learning_rate": 3.702871981505538e-06,
      "loss": 0.7625,
      "step": 2223
    },
    {
      "epoch": 0.20056815619786264,
      "grad_norm": 2.1063854366417685,
      "learning_rate": 3.7025655146111563e-06,
      "loss": 0.7296,
      "step": 2224
    },
    {
      "epoch": 0.2006583397213329,
      "grad_norm": 0.6834960649721717,
      "learning_rate": 3.702258902444497e-06,
      "loss": 0.6651,
      "step": 2225
    },
    {
      "epoch": 0.20074852324480316,
      "grad_norm": 1.4329365151418962,
      "learning_rate": 3.701952145031722e-06,
      "loss": 0.842,
      "step": 2226
    },
    {
      "epoch": 0.20083870676827342,
      "grad_norm": 1.6758620591675861,
      "learning_rate": 3.701645242399005e-06,
      "loss": 0.8646,
      "step": 2227
    },
    {
      "epoch": 0.2009288902917437,
      "grad_norm": 2.028931571088565,
      "learning_rate": 3.701338194572533e-06,
      "loss": 0.8527,
      "step": 2228
    },
    {
      "epoch": 0.20101907381521397,
      "grad_norm": 1.4824292731822968,
      "learning_rate": 3.7010310015785056e-06,
      "loss": 0.7674,
      "step": 2229
    },
    {
      "epoch": 0.20110925733868423,
      "grad_norm": 1.708317252602915,
      "learning_rate": 3.700723663443134e-06,
      "loss": 0.731,
      "step": 2230
    },
    {
      "epoch": 0.2011994408621545,
      "grad_norm": 0.6965001282592568,
      "learning_rate": 3.7004161801926416e-06,
      "loss": 0.5919,
      "step": 2231
    },
    {
      "epoch": 0.20128962438562475,
      "grad_norm": 1.7730960601284391,
      "learning_rate": 3.7001085518532643e-06,
      "loss": 0.8091,
      "step": 2232
    },
    {
      "epoch": 0.20137980790909502,
      "grad_norm": 1.671137210232659,
      "learning_rate": 3.6998007784512515e-06,
      "loss": 0.7809,
      "step": 2233
    },
    {
      "epoch": 0.20146999143256528,
      "grad_norm": 1.7368107780341757,
      "learning_rate": 3.6994928600128637e-06,
      "loss": 0.6587,
      "step": 2234
    },
    {
      "epoch": 0.20156017495603554,
      "grad_norm": 1.6089868910867755,
      "learning_rate": 3.6991847965643742e-06,
      "loss": 0.7535,
      "step": 2235
    },
    {
      "epoch": 0.2016503584795058,
      "grad_norm": 1.6735232063236773,
      "learning_rate": 3.698876588132068e-06,
      "loss": 0.7434,
      "step": 2236
    },
    {
      "epoch": 0.20174054200297606,
      "grad_norm": 1.5857752825077547,
      "learning_rate": 3.6985682347422446e-06,
      "loss": 0.7447,
      "step": 2237
    },
    {
      "epoch": 0.20183072552644632,
      "grad_norm": 2.3198566438516113,
      "learning_rate": 3.698259736421213e-06,
      "loss": 0.7943,
      "step": 2238
    },
    {
      "epoch": 0.20192090904991658,
      "grad_norm": 1.74427198575349,
      "learning_rate": 3.697951093195297e-06,
      "loss": 0.7671,
      "step": 2239
    },
    {
      "epoch": 0.20201109257338684,
      "grad_norm": 2.240714090829265,
      "learning_rate": 3.6976423050908307e-06,
      "loss": 0.7648,
      "step": 2240
    },
    {
      "epoch": 0.2021012760968571,
      "grad_norm": 1.4638191403576777,
      "learning_rate": 3.697333372134163e-06,
      "loss": 0.7728,
      "step": 2241
    },
    {
      "epoch": 0.20219145962032736,
      "grad_norm": 3.557783786287568,
      "learning_rate": 3.697024294351653e-06,
      "loss": 0.8828,
      "step": 2242
    },
    {
      "epoch": 0.20228164314379762,
      "grad_norm": 1.6941679857545675,
      "learning_rate": 3.696715071769672e-06,
      "loss": 0.7446,
      "step": 2243
    },
    {
      "epoch": 0.20237182666726788,
      "grad_norm": 1.735685729577368,
      "learning_rate": 3.696405704414606e-06,
      "loss": 0.7849,
      "step": 2244
    },
    {
      "epoch": 0.20246201019073815,
      "grad_norm": 1.452268696816149,
      "learning_rate": 3.6960961923128514e-06,
      "loss": 0.8869,
      "step": 2245
    },
    {
      "epoch": 0.2025521937142084,
      "grad_norm": 1.985127584167109,
      "learning_rate": 3.6957865354908177e-06,
      "loss": 0.7268,
      "step": 2246
    },
    {
      "epoch": 0.20264237723767867,
      "grad_norm": 1.6664811587408825,
      "learning_rate": 3.6954767339749262e-06,
      "loss": 0.8022,
      "step": 2247
    },
    {
      "epoch": 0.20273256076114893,
      "grad_norm": 2.16537479865409,
      "learning_rate": 3.6951667877916113e-06,
      "loss": 0.7686,
      "step": 2248
    },
    {
      "epoch": 0.2028227442846192,
      "grad_norm": 1.6058044082710325,
      "learning_rate": 3.694856696967319e-06,
      "loss": 0.7912,
      "step": 2249
    },
    {
      "epoch": 0.20291292780808945,
      "grad_norm": 2.286173589148037,
      "learning_rate": 3.6945464615285077e-06,
      "loss": 0.7686,
      "step": 2250
    },
    {
      "epoch": 0.2030031113315597,
      "grad_norm": 0.704109075455919,
      "learning_rate": 3.694236081501648e-06,
      "loss": 0.693,
      "step": 2251
    },
    {
      "epoch": 0.20309329485503,
      "grad_norm": 4.398930090090089,
      "learning_rate": 3.6939255569132246e-06,
      "loss": 0.744,
      "step": 2252
    },
    {
      "epoch": 0.20318347837850026,
      "grad_norm": 0.8445169354124615,
      "learning_rate": 3.693614887789733e-06,
      "loss": 0.6198,
      "step": 2253
    },
    {
      "epoch": 0.20327366190197052,
      "grad_norm": 0.8806294820006804,
      "learning_rate": 3.69330407415768e-06,
      "loss": 0.6208,
      "step": 2254
    },
    {
      "epoch": 0.20336384542544078,
      "grad_norm": 1.778030028262343,
      "learning_rate": 3.6929931160435867e-06,
      "loss": 0.7865,
      "step": 2255
    },
    {
      "epoch": 0.20345402894891104,
      "grad_norm": 1.9419300746566435,
      "learning_rate": 3.6926820134739858e-06,
      "loss": 0.84,
      "step": 2256
    },
    {
      "epoch": 0.2035442124723813,
      "grad_norm": 1.692178034733923,
      "learning_rate": 3.692370766475422e-06,
      "loss": 0.7179,
      "step": 2257
    },
    {
      "epoch": 0.20363439599585156,
      "grad_norm": 1.7415209123624673,
      "learning_rate": 3.692059375074453e-06,
      "loss": 0.8708,
      "step": 2258
    },
    {
      "epoch": 0.20372457951932182,
      "grad_norm": 1.9662114079805362,
      "learning_rate": 3.6917478392976475e-06,
      "loss": 0.8493,
      "step": 2259
    },
    {
      "epoch": 0.20381476304279209,
      "grad_norm": 1.4879611414480798,
      "learning_rate": 3.691436159171589e-06,
      "loss": 0.8453,
      "step": 2260
    },
    {
      "epoch": 0.20390494656626235,
      "grad_norm": 1.6030983608478997,
      "learning_rate": 3.6911243347228703e-06,
      "loss": 0.7664,
      "step": 2261
    },
    {
      "epoch": 0.2039951300897326,
      "grad_norm": 0.6408816498035114,
      "learning_rate": 3.690812365978099e-06,
      "loss": 0.5335,
      "step": 2262
    },
    {
      "epoch": 0.20408531361320287,
      "grad_norm": 1.5551620602563865,
      "learning_rate": 3.690500252963893e-06,
      "loss": 0.7397,
      "step": 2263
    },
    {
      "epoch": 0.20417549713667313,
      "grad_norm": 1.710105310398514,
      "learning_rate": 3.6901879957068846e-06,
      "loss": 0.7721,
      "step": 2264
    },
    {
      "epoch": 0.2042656806601434,
      "grad_norm": 1.419747909950079,
      "learning_rate": 3.689875594233717e-06,
      "loss": 0.872,
      "step": 2265
    },
    {
      "epoch": 0.20435586418361365,
      "grad_norm": 1.70115211378986,
      "learning_rate": 3.689563048571046e-06,
      "loss": 0.7695,
      "step": 2266
    },
    {
      "epoch": 0.2044460477070839,
      "grad_norm": 1.394668102227804,
      "learning_rate": 3.6892503587455395e-06,
      "loss": 0.7776,
      "step": 2267
    },
    {
      "epoch": 0.20453623123055417,
      "grad_norm": 1.4653296720084723,
      "learning_rate": 3.6889375247838766e-06,
      "loss": 0.7917,
      "step": 2268
    },
    {
      "epoch": 0.20462641475402443,
      "grad_norm": 1.5724481133177457,
      "learning_rate": 3.688624546712753e-06,
      "loss": 0.7856,
      "step": 2269
    },
    {
      "epoch": 0.2047165982774947,
      "grad_norm": 1.5615049627092685,
      "learning_rate": 3.688311424558871e-06,
      "loss": 0.746,
      "step": 2270
    },
    {
      "epoch": 0.20480678180096495,
      "grad_norm": 3.326020484296946,
      "learning_rate": 3.6879981583489496e-06,
      "loss": 0.6943,
      "step": 2271
    },
    {
      "epoch": 0.20489696532443522,
      "grad_norm": 1.8694964700947625,
      "learning_rate": 3.687684748109718e-06,
      "loss": 0.8043,
      "step": 2272
    },
    {
      "epoch": 0.20498714884790548,
      "grad_norm": 2.01496841027272,
      "learning_rate": 3.6873711938679174e-06,
      "loss": 0.8457,
      "step": 2273
    },
    {
      "epoch": 0.20507733237137574,
      "grad_norm": 1.935340970478414,
      "learning_rate": 3.6870574956503027e-06,
      "loss": 0.7769,
      "step": 2274
    },
    {
      "epoch": 0.20516751589484603,
      "grad_norm": 1.7752659316385528,
      "learning_rate": 3.68674365348364e-06,
      "loss": 0.7775,
      "step": 2275
    },
    {
      "epoch": 0.2052576994183163,
      "grad_norm": 1.643548014599496,
      "learning_rate": 3.6864296673947086e-06,
      "loss": 0.7904,
      "step": 2276
    },
    {
      "epoch": 0.20534788294178655,
      "grad_norm": 1.6124538119302891,
      "learning_rate": 3.686115537410298e-06,
      "loss": 0.7671,
      "step": 2277
    },
    {
      "epoch": 0.2054380664652568,
      "grad_norm": 1.7093059128323984,
      "learning_rate": 3.685801263557214e-06,
      "loss": 0.733,
      "step": 2278
    },
    {
      "epoch": 0.20552824998872707,
      "grad_norm": 1.846294974242261,
      "learning_rate": 3.68548684586227e-06,
      "loss": 0.8767,
      "step": 2279
    },
    {
      "epoch": 0.20561843351219733,
      "grad_norm": 1.76573909964742,
      "learning_rate": 3.685172284352295e-06,
      "loss": 0.7881,
      "step": 2280
    },
    {
      "epoch": 0.2057086170356676,
      "grad_norm": 1.2878269169386227,
      "learning_rate": 3.684857579054128e-06,
      "loss": 0.7157,
      "step": 2281
    },
    {
      "epoch": 0.20579880055913785,
      "grad_norm": 1.7709167984384746,
      "learning_rate": 3.6845427299946233e-06,
      "loss": 0.8491,
      "step": 2282
    },
    {
      "epoch": 0.2058889840826081,
      "grad_norm": 1.65182999363762,
      "learning_rate": 3.6842277372006434e-06,
      "loss": 0.8785,
      "step": 2283
    },
    {
      "epoch": 0.20597916760607837,
      "grad_norm": 1.4769792725458781,
      "learning_rate": 3.6839126006990664e-06,
      "loss": 0.7546,
      "step": 2284
    },
    {
      "epoch": 0.20606935112954863,
      "grad_norm": 1.6944590587020483,
      "learning_rate": 3.6835973205167818e-06,
      "loss": 0.8363,
      "step": 2285
    },
    {
      "epoch": 0.2061595346530189,
      "grad_norm": 1.9572419987728278,
      "learning_rate": 3.6832818966806904e-06,
      "loss": 0.8238,
      "step": 2286
    },
    {
      "epoch": 0.20624971817648916,
      "grad_norm": 1.589332592121267,
      "learning_rate": 3.682966329217706e-06,
      "loss": 0.8631,
      "step": 2287
    },
    {
      "epoch": 0.20633990169995942,
      "grad_norm": 2.091581559697809,
      "learning_rate": 3.6826506181547543e-06,
      "loss": 0.8192,
      "step": 2288
    },
    {
      "epoch": 0.20643008522342968,
      "grad_norm": 1.6855049349096722,
      "learning_rate": 3.682334763518774e-06,
      "loss": 0.8077,
      "step": 2289
    },
    {
      "epoch": 0.20652026874689994,
      "grad_norm": 1.7291777124013812,
      "learning_rate": 3.6820187653367158e-06,
      "loss": 0.7487,
      "step": 2290
    },
    {
      "epoch": 0.2066104522703702,
      "grad_norm": 2.3412633292841667,
      "learning_rate": 3.6817026236355412e-06,
      "loss": 0.672,
      "step": 2291
    },
    {
      "epoch": 0.20670063579384046,
      "grad_norm": 1.9822525929301604,
      "learning_rate": 3.681386338442227e-06,
      "loss": 0.7681,
      "step": 2292
    },
    {
      "epoch": 0.20679081931731072,
      "grad_norm": 1.3939836212192083,
      "learning_rate": 3.681069909783758e-06,
      "loss": 0.7441,
      "step": 2293
    },
    {
      "epoch": 0.20688100284078098,
      "grad_norm": 1.6451956885914105,
      "learning_rate": 3.680753337687136e-06,
      "loss": 0.7935,
      "step": 2294
    },
    {
      "epoch": 0.20697118636425124,
      "grad_norm": 1.77329875386495,
      "learning_rate": 3.680436622179371e-06,
      "loss": 0.7131,
      "step": 2295
    },
    {
      "epoch": 0.2070613698877215,
      "grad_norm": 2.4647081340836814,
      "learning_rate": 3.680119763287488e-06,
      "loss": 0.7712,
      "step": 2296
    },
    {
      "epoch": 0.20715155341119176,
      "grad_norm": 1.891362322329112,
      "learning_rate": 3.6798027610385227e-06,
      "loss": 0.7244,
      "step": 2297
    },
    {
      "epoch": 0.20724173693466202,
      "grad_norm": 2.015029464392669,
      "learning_rate": 3.6794856154595235e-06,
      "loss": 0.8305,
      "step": 2298
    },
    {
      "epoch": 0.2073319204581323,
      "grad_norm": 2.007458013568158,
      "learning_rate": 3.6791683265775506e-06,
      "loss": 0.7945,
      "step": 2299
    },
    {
      "epoch": 0.20742210398160257,
      "grad_norm": 1.950909510052261,
      "learning_rate": 3.6788508944196773e-06,
      "loss": 0.7411,
      "step": 2300
    },
    {
      "epoch": 0.20751228750507283,
      "grad_norm": 1.6084268980860823,
      "learning_rate": 3.678533319012989e-06,
      "loss": 0.8556,
      "step": 2301
    },
    {
      "epoch": 0.2076024710285431,
      "grad_norm": 3.6161096668408206,
      "learning_rate": 3.6782156003845826e-06,
      "loss": 0.7305,
      "step": 2302
    },
    {
      "epoch": 0.20769265455201336,
      "grad_norm": 0.6465087609431344,
      "learning_rate": 3.6778977385615676e-06,
      "loss": 0.5951,
      "step": 2303
    },
    {
      "epoch": 0.20778283807548362,
      "grad_norm": 1.7158018316360304,
      "learning_rate": 3.6775797335710656e-06,
      "loss": 0.8317,
      "step": 2304
    },
    {
      "epoch": 0.20787302159895388,
      "grad_norm": 1.8550492304312596,
      "learning_rate": 3.6772615854402105e-06,
      "loss": 0.7365,
      "step": 2305
    },
    {
      "epoch": 0.20796320512242414,
      "grad_norm": 1.6716963057522654,
      "learning_rate": 3.6769432941961487e-06,
      "loss": 0.665,
      "step": 2306
    },
    {
      "epoch": 0.2080533886458944,
      "grad_norm": 1.7827526246296146,
      "learning_rate": 3.676624859866038e-06,
      "loss": 0.7832,
      "step": 2307
    },
    {
      "epoch": 0.20814357216936466,
      "grad_norm": 2.2639341161327864,
      "learning_rate": 3.67630628247705e-06,
      "loss": 0.7074,
      "step": 2308
    },
    {
      "epoch": 0.20823375569283492,
      "grad_norm": 1.778045783934901,
      "learning_rate": 3.675987562056367e-06,
      "loss": 0.8457,
      "step": 2309
    },
    {
      "epoch": 0.20832393921630518,
      "grad_norm": 2.142654681859249,
      "learning_rate": 3.675668698631184e-06,
      "loss": 0.8196,
      "step": 2310
    },
    {
      "epoch": 0.20841412273977544,
      "grad_norm": 1.4909470280295054,
      "learning_rate": 3.675349692228708e-06,
      "loss": 0.8478,
      "step": 2311
    },
    {
      "epoch": 0.2085043062632457,
      "grad_norm": 2.452478894717429,
      "learning_rate": 3.6750305428761578e-06,
      "loss": 0.7601,
      "step": 2312
    },
    {
      "epoch": 0.20859448978671596,
      "grad_norm": 2.1482911216114497,
      "learning_rate": 3.674711250600766e-06,
      "loss": 0.7915,
      "step": 2313
    },
    {
      "epoch": 0.20868467331018623,
      "grad_norm": 0.8581935043386061,
      "learning_rate": 3.6743918154297765e-06,
      "loss": 0.5764,
      "step": 2314
    },
    {
      "epoch": 0.20877485683365649,
      "grad_norm": 1.6940737650640456,
      "learning_rate": 3.6740722373904446e-06,
      "loss": 0.7783,
      "step": 2315
    },
    {
      "epoch": 0.20886504035712675,
      "grad_norm": 1.4529557488431344,
      "learning_rate": 3.6737525165100383e-06,
      "loss": 0.7616,
      "step": 2316
    },
    {
      "epoch": 0.208955223880597,
      "grad_norm": 2.133577873201638,
      "learning_rate": 3.6734326528158385e-06,
      "loss": 0.8062,
      "step": 2317
    },
    {
      "epoch": 0.20904540740406727,
      "grad_norm": 1.3664011846658946,
      "learning_rate": 3.673112646335138e-06,
      "loss": 0.7674,
      "step": 2318
    },
    {
      "epoch": 0.20913559092753753,
      "grad_norm": 1.574284933133039,
      "learning_rate": 3.672792497095241e-06,
      "loss": 0.7371,
      "step": 2319
    },
    {
      "epoch": 0.2092257744510078,
      "grad_norm": 2.0969946473234784,
      "learning_rate": 3.672472205123464e-06,
      "loss": 0.7906,
      "step": 2320
    },
    {
      "epoch": 0.20931595797447805,
      "grad_norm": 1.9369552215575903,
      "learning_rate": 3.6721517704471363e-06,
      "loss": 0.7103,
      "step": 2321
    },
    {
      "epoch": 0.2094061414979483,
      "grad_norm": 1.6002658712502602,
      "learning_rate": 3.6718311930936e-06,
      "loss": 0.818,
      "step": 2322
    },
    {
      "epoch": 0.2094963250214186,
      "grad_norm": 1.7610776288686802,
      "learning_rate": 3.6715104730902074e-06,
      "loss": 0.7318,
      "step": 2323
    },
    {
      "epoch": 0.20958650854488886,
      "grad_norm": 1.7038031811436065,
      "learning_rate": 3.671189610464325e-06,
      "loss": 0.8276,
      "step": 2324
    },
    {
      "epoch": 0.20967669206835912,
      "grad_norm": 1.8799146456314497,
      "learning_rate": 3.6708686052433303e-06,
      "loss": 0.7299,
      "step": 2325
    },
    {
      "epoch": 0.20976687559182938,
      "grad_norm": 1.5859966548918074,
      "learning_rate": 3.6705474574546127e-06,
      "loss": 0.7984,
      "step": 2326
    },
    {
      "epoch": 0.20985705911529964,
      "grad_norm": 3.10641138897109,
      "learning_rate": 3.670226167125575e-06,
      "loss": 0.8031,
      "step": 2327
    },
    {
      "epoch": 0.2099472426387699,
      "grad_norm": 1.8873648790916433,
      "learning_rate": 3.6699047342836313e-06,
      "loss": 0.7862,
      "step": 2328
    },
    {
      "epoch": 0.21003742616224017,
      "grad_norm": 1.552430628211037,
      "learning_rate": 3.669583158956208e-06,
      "loss": 0.8174,
      "step": 2329
    },
    {
      "epoch": 0.21012760968571043,
      "grad_norm": 0.6728826108725644,
      "learning_rate": 3.669261441170743e-06,
      "loss": 0.6077,
      "step": 2330
    },
    {
      "epoch": 0.2102177932091807,
      "grad_norm": 2.047513905758616,
      "learning_rate": 3.668939580954688e-06,
      "loss": 0.7934,
      "step": 2331
    },
    {
      "epoch": 0.21030797673265095,
      "grad_norm": 1.4802552726044802,
      "learning_rate": 3.668617578335506e-06,
      "loss": 0.7586,
      "step": 2332
    },
    {
      "epoch": 0.2103981602561212,
      "grad_norm": 2.764401181239551,
      "learning_rate": 3.6682954333406707e-06,
      "loss": 0.8126,
      "step": 2333
    },
    {
      "epoch": 0.21048834377959147,
      "grad_norm": 1.9461707860644923,
      "learning_rate": 3.6679731459976707e-06,
      "loss": 0.9244,
      "step": 2334
    },
    {
      "epoch": 0.21057852730306173,
      "grad_norm": 1.5279439932060703,
      "learning_rate": 3.6676507163340046e-06,
      "loss": 0.7794,
      "step": 2335
    },
    {
      "epoch": 0.210668710826532,
      "grad_norm": 4.044549104246979,
      "learning_rate": 3.6673281443771842e-06,
      "loss": 0.8182,
      "step": 2336
    },
    {
      "epoch": 0.21075889435000225,
      "grad_norm": 1.979999554373951,
      "learning_rate": 3.667005430154733e-06,
      "loss": 0.674,
      "step": 2337
    },
    {
      "epoch": 0.2108490778734725,
      "grad_norm": 0.6644859310351425,
      "learning_rate": 3.666682573694186e-06,
      "loss": 0.6814,
      "step": 2338
    },
    {
      "epoch": 0.21093926139694277,
      "grad_norm": 0.6642342289296792,
      "learning_rate": 3.6663595750230924e-06,
      "loss": 0.5508,
      "step": 2339
    },
    {
      "epoch": 0.21102944492041303,
      "grad_norm": 1.9157292796590846,
      "learning_rate": 3.666036434169012e-06,
      "loss": 0.8388,
      "step": 2340
    },
    {
      "epoch": 0.2111196284438833,
      "grad_norm": 1.4162494755716186,
      "learning_rate": 3.665713151159516e-06,
      "loss": 0.764,
      "step": 2341
    },
    {
      "epoch": 0.21120981196735356,
      "grad_norm": 2.057613132003894,
      "learning_rate": 3.665389726022189e-06,
      "loss": 0.8824,
      "step": 2342
    },
    {
      "epoch": 0.21129999549082382,
      "grad_norm": 1.6405319914112464,
      "learning_rate": 3.6650661587846283e-06,
      "loss": 0.7535,
      "step": 2343
    },
    {
      "epoch": 0.21139017901429408,
      "grad_norm": 1.7891216018442997,
      "learning_rate": 3.6647424494744418e-06,
      "loss": 0.7457,
      "step": 2344
    },
    {
      "epoch": 0.21148036253776434,
      "grad_norm": 1.6700489375088134,
      "learning_rate": 3.6644185981192503e-06,
      "loss": 0.8358,
      "step": 2345
    },
    {
      "epoch": 0.2115705460612346,
      "grad_norm": 1.9558236361332144,
      "learning_rate": 3.6640946047466868e-06,
      "loss": 0.7199,
      "step": 2346
    },
    {
      "epoch": 0.2116607295847049,
      "grad_norm": 1.6374092193366334,
      "learning_rate": 3.6637704693843953e-06,
      "loss": 0.749,
      "step": 2347
    },
    {
      "epoch": 0.21175091310817515,
      "grad_norm": 1.4805893660154583,
      "learning_rate": 3.6634461920600337e-06,
      "loss": 0.7173,
      "step": 2348
    },
    {
      "epoch": 0.2118410966316454,
      "grad_norm": 3.9328197941430174,
      "learning_rate": 3.66312177280127e-06,
      "loss": 0.8079,
      "step": 2349
    },
    {
      "epoch": 0.21193128015511567,
      "grad_norm": 1.4198208737558609,
      "learning_rate": 3.6627972116357872e-06,
      "loss": 0.8204,
      "step": 2350
    },
    {
      "epoch": 0.21202146367858593,
      "grad_norm": 2.1043863716871893,
      "learning_rate": 3.662472508591278e-06,
      "loss": 0.8468,
      "step": 2351
    },
    {
      "epoch": 0.2121116472020562,
      "grad_norm": 1.9323147329374362,
      "learning_rate": 3.662147663695447e-06,
      "loss": 0.7722,
      "step": 2352
    },
    {
      "epoch": 0.21220183072552645,
      "grad_norm": 3.7293148473073012,
      "learning_rate": 3.6618226769760127e-06,
      "loss": 0.7729,
      "step": 2353
    },
    {
      "epoch": 0.2122920142489967,
      "grad_norm": 2.405040982188678,
      "learning_rate": 3.661497548460704e-06,
      "loss": 0.6711,
      "step": 2354
    },
    {
      "epoch": 0.21238219777246697,
      "grad_norm": 3.259521767812909,
      "learning_rate": 3.6611722781772635e-06,
      "loss": 0.8028,
      "step": 2355
    },
    {
      "epoch": 0.21247238129593723,
      "grad_norm": 0.8301129502115647,
      "learning_rate": 3.6608468661534444e-06,
      "loss": 0.6451,
      "step": 2356
    },
    {
      "epoch": 0.2125625648194075,
      "grad_norm": 2.015603826840404,
      "learning_rate": 3.660521312417013e-06,
      "loss": 0.7132,
      "step": 2357
    },
    {
      "epoch": 0.21265274834287776,
      "grad_norm": 2.01257661472227,
      "learning_rate": 3.660195616995747e-06,
      "loss": 0.7502,
      "step": 2358
    },
    {
      "epoch": 0.21274293186634802,
      "grad_norm": 2.4261031129250625,
      "learning_rate": 3.6598697799174367e-06,
      "loss": 0.813,
      "step": 2359
    },
    {
      "epoch": 0.21283311538981828,
      "grad_norm": 2.1669502684022905,
      "learning_rate": 3.6595438012098844e-06,
      "loss": 0.8494,
      "step": 2360
    },
    {
      "epoch": 0.21292329891328854,
      "grad_norm": 2.0455040453691558,
      "learning_rate": 3.6592176809009045e-06,
      "loss": 0.7584,
      "step": 2361
    },
    {
      "epoch": 0.2130134824367588,
      "grad_norm": 1.7477158216982265,
      "learning_rate": 3.6588914190183227e-06,
      "loss": 0.8175,
      "step": 2362
    },
    {
      "epoch": 0.21310366596022906,
      "grad_norm": 1.948639135626706,
      "learning_rate": 3.658565015589978e-06,
      "loss": 0.656,
      "step": 2363
    },
    {
      "epoch": 0.21319384948369932,
      "grad_norm": 2.7029197482644323,
      "learning_rate": 3.6582384706437217e-06,
      "loss": 0.7875,
      "step": 2364
    },
    {
      "epoch": 0.21328403300716958,
      "grad_norm": 1.9873278417339717,
      "learning_rate": 3.6579117842074156e-06,
      "loss": 0.8023,
      "step": 2365
    },
    {
      "epoch": 0.21337421653063984,
      "grad_norm": 1.7097802276371217,
      "learning_rate": 3.657584956308934e-06,
      "loss": 0.7225,
      "step": 2366
    },
    {
      "epoch": 0.2134644000541101,
      "grad_norm": 3.45914193930197,
      "learning_rate": 3.6572579869761648e-06,
      "loss": 0.8465,
      "step": 2367
    },
    {
      "epoch": 0.21355458357758036,
      "grad_norm": 2.645270330247302,
      "learning_rate": 3.6569308762370056e-06,
      "loss": 0.834,
      "step": 2368
    },
    {
      "epoch": 0.21364476710105063,
      "grad_norm": 2.0693325962617553,
      "learning_rate": 3.6566036241193676e-06,
      "loss": 0.8368,
      "step": 2369
    },
    {
      "epoch": 0.2137349506245209,
      "grad_norm": 0.6199434292272539,
      "learning_rate": 3.656276230651174e-06,
      "loss": 0.6085,
      "step": 2370
    },
    {
      "epoch": 0.21382513414799117,
      "grad_norm": 0.7116851229213816,
      "learning_rate": 3.65594869586036e-06,
      "loss": 0.6429,
      "step": 2371
    },
    {
      "epoch": 0.21391531767146144,
      "grad_norm": 1.760779153496109,
      "learning_rate": 3.6556210197748724e-06,
      "loss": 0.705,
      "step": 2372
    },
    {
      "epoch": 0.2140055011949317,
      "grad_norm": 1.6355824315643326,
      "learning_rate": 3.655293202422671e-06,
      "loss": 0.7527,
      "step": 2373
    },
    {
      "epoch": 0.21409568471840196,
      "grad_norm": 1.3654329649381574,
      "learning_rate": 3.654965243831725e-06,
      "loss": 0.7965,
      "step": 2374
    },
    {
      "epoch": 0.21418586824187222,
      "grad_norm": 1.4859167023335473,
      "learning_rate": 3.65463714403002e-06,
      "loss": 0.8259,
      "step": 2375
    },
    {
      "epoch": 0.21427605176534248,
      "grad_norm": 1.4677444424687245,
      "learning_rate": 3.65430890304555e-06,
      "loss": 0.7378,
      "step": 2376
    },
    {
      "epoch": 0.21436623528881274,
      "grad_norm": 1.504087125131577,
      "learning_rate": 3.653980520906323e-06,
      "loss": 0.8112,
      "step": 2377
    },
    {
      "epoch": 0.214456418812283,
      "grad_norm": 2.178271527608976,
      "learning_rate": 3.653651997640358e-06,
      "loss": 0.763,
      "step": 2378
    },
    {
      "epoch": 0.21454660233575326,
      "grad_norm": 1.6934052739418184,
      "learning_rate": 3.653323333275686e-06,
      "loss": 0.7858,
      "step": 2379
    },
    {
      "epoch": 0.21463678585922352,
      "grad_norm": 1.6440029428321135,
      "learning_rate": 3.652994527840351e-06,
      "loss": 0.762,
      "step": 2380
    },
    {
      "epoch": 0.21472696938269378,
      "grad_norm": 1.8219803794365508,
      "learning_rate": 3.6526655813624087e-06,
      "loss": 0.7467,
      "step": 2381
    },
    {
      "epoch": 0.21481715290616404,
      "grad_norm": 0.5774253529446612,
      "learning_rate": 3.652336493869925e-06,
      "loss": 0.6027,
      "step": 2382
    },
    {
      "epoch": 0.2149073364296343,
      "grad_norm": 1.6477865949851986,
      "learning_rate": 3.6520072653909823e-06,
      "loss": 0.7865,
      "step": 2383
    },
    {
      "epoch": 0.21499751995310457,
      "grad_norm": 1.809351916651823,
      "learning_rate": 3.6516778959536702e-06,
      "loss": 0.7797,
      "step": 2384
    },
    {
      "epoch": 0.21508770347657483,
      "grad_norm": 1.7038871389766386,
      "learning_rate": 3.6513483855860923e-06,
      "loss": 0.8307,
      "step": 2385
    },
    {
      "epoch": 0.2151778870000451,
      "grad_norm": 1.6042332408351243,
      "learning_rate": 3.6510187343163654e-06,
      "loss": 0.7284,
      "step": 2386
    },
    {
      "epoch": 0.21526807052351535,
      "grad_norm": 1.9424650501853067,
      "learning_rate": 3.650688942172616e-06,
      "loss": 0.818,
      "step": 2387
    },
    {
      "epoch": 0.2153582540469856,
      "grad_norm": 1.9280893603822575,
      "learning_rate": 3.650359009182984e-06,
      "loss": 0.7218,
      "step": 2388
    },
    {
      "epoch": 0.21544843757045587,
      "grad_norm": 2.6967744704657197,
      "learning_rate": 3.650028935375622e-06,
      "loss": 0.866,
      "step": 2389
    },
    {
      "epoch": 0.21553862109392613,
      "grad_norm": 1.540421481356716,
      "learning_rate": 3.6496987207786926e-06,
      "loss": 0.745,
      "step": 2390
    },
    {
      "epoch": 0.2156288046173964,
      "grad_norm": 1.8270129872796408,
      "learning_rate": 3.6493683654203724e-06,
      "loss": 0.7333,
      "step": 2391
    },
    {
      "epoch": 0.21571898814086665,
      "grad_norm": 2.014740269204931,
      "learning_rate": 3.6490378693288484e-06,
      "loss": 0.767,
      "step": 2392
    },
    {
      "epoch": 0.2158091716643369,
      "grad_norm": 1.824716896150916,
      "learning_rate": 3.648707232532321e-06,
      "loss": 0.8309,
      "step": 2393
    },
    {
      "epoch": 0.2158993551878072,
      "grad_norm": 5.127591664105451,
      "learning_rate": 3.6483764550590017e-06,
      "loss": 0.7921,
      "step": 2394
    },
    {
      "epoch": 0.21598953871127746,
      "grad_norm": 1.699419031726695,
      "learning_rate": 3.6480455369371133e-06,
      "loss": 0.7984,
      "step": 2395
    },
    {
      "epoch": 0.21607972223474772,
      "grad_norm": 5.608646951251193,
      "learning_rate": 3.647714478194893e-06,
      "loss": 0.7857,
      "step": 2396
    },
    {
      "epoch": 0.21616990575821798,
      "grad_norm": 2.0918793933595468,
      "learning_rate": 3.647383278860588e-06,
      "loss": 0.7277,
      "step": 2397
    },
    {
      "epoch": 0.21626008928168824,
      "grad_norm": 1.8359676277448231,
      "learning_rate": 3.6470519389624587e-06,
      "loss": 0.769,
      "step": 2398
    },
    {
      "epoch": 0.2163502728051585,
      "grad_norm": 1.5194869243364255,
      "learning_rate": 3.646720458528776e-06,
      "loss": 0.8399,
      "step": 2399
    },
    {
      "epoch": 0.21644045632862877,
      "grad_norm": 2.452738057130886,
      "learning_rate": 3.6463888375878235e-06,
      "loss": 0.7523,
      "step": 2400
    },
    {
      "epoch": 0.21653063985209903,
      "grad_norm": 2.1472754092563493,
      "learning_rate": 3.646057076167897e-06,
      "loss": 0.7969,
      "step": 2401
    },
    {
      "epoch": 0.2166208233755693,
      "grad_norm": 1.6554313651846264,
      "learning_rate": 3.645725174297305e-06,
      "loss": 0.7877,
      "step": 2402
    },
    {
      "epoch": 0.21671100689903955,
      "grad_norm": 1.600260954676497,
      "learning_rate": 3.645393132004367e-06,
      "loss": 0.7647,
      "step": 2403
    },
    {
      "epoch": 0.2168011904225098,
      "grad_norm": 1.6541645668202227,
      "learning_rate": 3.6450609493174135e-06,
      "loss": 0.7442,
      "step": 2404
    },
    {
      "epoch": 0.21689137394598007,
      "grad_norm": 1.748039032476297,
      "learning_rate": 3.6447286262647896e-06,
      "loss": 0.7435,
      "step": 2405
    },
    {
      "epoch": 0.21698155746945033,
      "grad_norm": 1.6858178691120174,
      "learning_rate": 3.64439616287485e-06,
      "loss": 0.7477,
      "step": 2406
    },
    {
      "epoch": 0.2170717409929206,
      "grad_norm": 1.9473012540230332,
      "learning_rate": 3.644063559175963e-06,
      "loss": 0.7449,
      "step": 2407
    },
    {
      "epoch": 0.21716192451639085,
      "grad_norm": 1.996438491225366,
      "learning_rate": 3.6437308151965074e-06,
      "loss": 0.762,
      "step": 2408
    },
    {
      "epoch": 0.2172521080398611,
      "grad_norm": 1.3527239621472424,
      "learning_rate": 3.643397930964876e-06,
      "loss": 0.7941,
      "step": 2409
    },
    {
      "epoch": 0.21734229156333137,
      "grad_norm": 2.0887296469956387,
      "learning_rate": 3.6430649065094707e-06,
      "loss": 0.8476,
      "step": 2410
    },
    {
      "epoch": 0.21743247508680164,
      "grad_norm": 1.9967884862841405,
      "learning_rate": 3.6427317418587086e-06,
      "loss": 0.8529,
      "step": 2411
    },
    {
      "epoch": 0.2175226586102719,
      "grad_norm": 5.010896634636142,
      "learning_rate": 3.6423984370410157e-06,
      "loss": 0.7202,
      "step": 2412
    },
    {
      "epoch": 0.21761284213374216,
      "grad_norm": 1.7147628313233256,
      "learning_rate": 3.6420649920848324e-06,
      "loss": 0.7501,
      "step": 2413
    },
    {
      "epoch": 0.21770302565721242,
      "grad_norm": 1.677064986355652,
      "learning_rate": 3.6417314070186096e-06,
      "loss": 0.7912,
      "step": 2414
    },
    {
      "epoch": 0.21779320918068268,
      "grad_norm": 1.4553510781110797,
      "learning_rate": 3.641397681870811e-06,
      "loss": 0.7886,
      "step": 2415
    },
    {
      "epoch": 0.21788339270415294,
      "grad_norm": 1.5337247507866387,
      "learning_rate": 3.641063816669911e-06,
      "loss": 0.7988,
      "step": 2416
    },
    {
      "epoch": 0.2179735762276232,
      "grad_norm": 1.6445827158941335,
      "learning_rate": 3.640729811444398e-06,
      "loss": 0.8408,
      "step": 2417
    },
    {
      "epoch": 0.2180637597510935,
      "grad_norm": 1.7815572908596664,
      "learning_rate": 3.6403956662227706e-06,
      "loss": 0.7842,
      "step": 2418
    },
    {
      "epoch": 0.21815394327456375,
      "grad_norm": 2.213614740100529,
      "learning_rate": 3.6400613810335396e-06,
      "loss": 0.8373,
      "step": 2419
    },
    {
      "epoch": 0.218244126798034,
      "grad_norm": 7.699322797051902,
      "learning_rate": 3.639726955905228e-06,
      "loss": 0.7006,
      "step": 2420
    },
    {
      "epoch": 0.21833431032150427,
      "grad_norm": 1.8886230209959294,
      "learning_rate": 3.639392390866372e-06,
      "loss": 0.8107,
      "step": 2421
    },
    {
      "epoch": 0.21842449384497453,
      "grad_norm": 1.6113384084662385,
      "learning_rate": 3.639057685945517e-06,
      "loss": 0.755,
      "step": 2422
    },
    {
      "epoch": 0.2185146773684448,
      "grad_norm": 1.79679273334097,
      "learning_rate": 3.638722841171223e-06,
      "loss": 0.6791,
      "step": 2423
    },
    {
      "epoch": 0.21860486089191505,
      "grad_norm": 1.6875043091895574,
      "learning_rate": 3.638387856572061e-06,
      "loss": 0.8192,
      "step": 2424
    },
    {
      "epoch": 0.21869504441538531,
      "grad_norm": 0.6549012083432998,
      "learning_rate": 3.638052732176612e-06,
      "loss": 0.6084,
      "step": 2425
    },
    {
      "epoch": 0.21878522793885558,
      "grad_norm": 1.7668056760236037,
      "learning_rate": 3.637717468013472e-06,
      "loss": 0.7964,
      "step": 2426
    },
    {
      "epoch": 0.21887541146232584,
      "grad_norm": 0.6019398632729984,
      "learning_rate": 3.6373820641112475e-06,
      "loss": 0.5599,
      "step": 2427
    },
    {
      "epoch": 0.2189655949857961,
      "grad_norm": 1.7719359237719323,
      "learning_rate": 3.6370465204985567e-06,
      "loss": 0.7939,
      "step": 2428
    },
    {
      "epoch": 0.21905577850926636,
      "grad_norm": 0.6394753957329539,
      "learning_rate": 3.6367108372040304e-06,
      "loss": 0.6211,
      "step": 2429
    },
    {
      "epoch": 0.21914596203273662,
      "grad_norm": 1.8940070931443804,
      "learning_rate": 3.6363750142563107e-06,
      "loss": 0.786,
      "step": 2430
    },
    {
      "epoch": 0.21923614555620688,
      "grad_norm": 0.7317433380355434,
      "learning_rate": 3.636039051684052e-06,
      "loss": 0.6569,
      "step": 2431
    },
    {
      "epoch": 0.21932632907967714,
      "grad_norm": 2.2076937871011415,
      "learning_rate": 3.6357029495159203e-06,
      "loss": 0.7536,
      "step": 2432
    },
    {
      "epoch": 0.2194165126031474,
      "grad_norm": 1.5747072583575417,
      "learning_rate": 3.6353667077805934e-06,
      "loss": 0.7596,
      "step": 2433
    },
    {
      "epoch": 0.21950669612661766,
      "grad_norm": 2.212993191852948,
      "learning_rate": 3.6350303265067625e-06,
      "loss": 0.8265,
      "step": 2434
    },
    {
      "epoch": 0.21959687965008792,
      "grad_norm": 1.9978432947819935,
      "learning_rate": 3.6346938057231285e-06,
      "loss": 0.8143,
      "step": 2435
    },
    {
      "epoch": 0.21968706317355818,
      "grad_norm": 1.561798472275575,
      "learning_rate": 3.6343571454584047e-06,
      "loss": 0.7288,
      "step": 2436
    },
    {
      "epoch": 0.21977724669702844,
      "grad_norm": 1.7624163885101778,
      "learning_rate": 3.6340203457413176e-06,
      "loss": 0.8224,
      "step": 2437
    },
    {
      "epoch": 0.2198674302204987,
      "grad_norm": 1.5723465987495,
      "learning_rate": 3.633683406600605e-06,
      "loss": 0.803,
      "step": 2438
    },
    {
      "epoch": 0.21995761374396897,
      "grad_norm": 2.0028314812817705,
      "learning_rate": 3.6333463280650165e-06,
      "loss": 0.8435,
      "step": 2439
    },
    {
      "epoch": 0.22004779726743923,
      "grad_norm": 0.7043119161293506,
      "learning_rate": 3.6330091101633126e-06,
      "loss": 0.5694,
      "step": 2440
    },
    {
      "epoch": 0.2201379807909095,
      "grad_norm": 1.6906125362592572,
      "learning_rate": 3.632671752924267e-06,
      "loss": 0.8172,
      "step": 2441
    },
    {
      "epoch": 0.22022816431437978,
      "grad_norm": 2.437365992848276,
      "learning_rate": 3.632334256376665e-06,
      "loss": 0.7907,
      "step": 2442
    },
    {
      "epoch": 0.22031834783785004,
      "grad_norm": 1.4456815506181606,
      "learning_rate": 3.6319966205493044e-06,
      "loss": 0.8409,
      "step": 2443
    },
    {
      "epoch": 0.2204085313613203,
      "grad_norm": 0.7022628797026995,
      "learning_rate": 3.6316588454709922e-06,
      "loss": 0.6534,
      "step": 2444
    },
    {
      "epoch": 0.22049871488479056,
      "grad_norm": 1.5439864653206594,
      "learning_rate": 3.6313209311705514e-06,
      "loss": 0.7364,
      "step": 2445
    },
    {
      "epoch": 0.22058889840826082,
      "grad_norm": 1.7198446775877336,
      "learning_rate": 3.6309828776768133e-06,
      "loss": 0.8259,
      "step": 2446
    },
    {
      "epoch": 0.22067908193173108,
      "grad_norm": 1.6184719715914602,
      "learning_rate": 3.630644685018623e-06,
      "loss": 0.7487,
      "step": 2447
    },
    {
      "epoch": 0.22076926545520134,
      "grad_norm": 2.243767689924716,
      "learning_rate": 3.6303063532248367e-06,
      "loss": 0.864,
      "step": 2448
    },
    {
      "epoch": 0.2208594489786716,
      "grad_norm": 1.5392351392555548,
      "learning_rate": 3.6299678823243236e-06,
      "loss": 0.8607,
      "step": 2449
    },
    {
      "epoch": 0.22094963250214186,
      "grad_norm": 1.733113629220941,
      "learning_rate": 3.629629272345963e-06,
      "loss": 0.6916,
      "step": 2450
    },
    {
      "epoch": 0.22103981602561212,
      "grad_norm": 1.660484227712928,
      "learning_rate": 3.6292905233186468e-06,
      "loss": 0.9046,
      "step": 2451
    },
    {
      "epoch": 0.22112999954908238,
      "grad_norm": 1.627854261384407,
      "learning_rate": 3.6289516352712796e-06,
      "loss": 0.7423,
      "step": 2452
    },
    {
      "epoch": 0.22122018307255265,
      "grad_norm": 0.7101881668422059,
      "learning_rate": 3.6286126082327764e-06,
      "loss": 0.6045,
      "step": 2453
    },
    {
      "epoch": 0.2213103665960229,
      "grad_norm": 1.559599129082749,
      "learning_rate": 3.628273442232066e-06,
      "loss": 0.8148,
      "step": 2454
    },
    {
      "epoch": 0.22140055011949317,
      "grad_norm": 1.3947025482371078,
      "learning_rate": 3.627934137298087e-06,
      "loss": 0.7793,
      "step": 2455
    },
    {
      "epoch": 0.22149073364296343,
      "grad_norm": 1.809847831546924,
      "learning_rate": 3.627594693459792e-06,
      "loss": 0.7591,
      "step": 2456
    },
    {
      "epoch": 0.2215809171664337,
      "grad_norm": 1.571724176470495,
      "learning_rate": 3.6272551107461424e-06,
      "loss": 0.7443,
      "step": 2457
    },
    {
      "epoch": 0.22167110068990395,
      "grad_norm": 2.160704172464278,
      "learning_rate": 3.6269153891861137e-06,
      "loss": 0.7764,
      "step": 2458
    },
    {
      "epoch": 0.2217612842133742,
      "grad_norm": 1.9917734950104204,
      "learning_rate": 3.6265755288086944e-06,
      "loss": 0.7511,
      "step": 2459
    },
    {
      "epoch": 0.22185146773684447,
      "grad_norm": 2.222356524647664,
      "learning_rate": 3.626235529642881e-06,
      "loss": 0.7674,
      "step": 2460
    },
    {
      "epoch": 0.22194165126031473,
      "grad_norm": 9.723965830596557,
      "learning_rate": 3.625895391717686e-06,
      "loss": 0.7156,
      "step": 2461
    },
    {
      "epoch": 0.222031834783785,
      "grad_norm": 1.456223169169655,
      "learning_rate": 3.625555115062131e-06,
      "loss": 0.7557,
      "step": 2462
    },
    {
      "epoch": 0.22212201830725525,
      "grad_norm": 1.8038971726828104,
      "learning_rate": 3.6252146997052507e-06,
      "loss": 0.8559,
      "step": 2463
    },
    {
      "epoch": 0.22221220183072551,
      "grad_norm": 0.6586762717101058,
      "learning_rate": 3.6248741456760898e-06,
      "loss": 0.5603,
      "step": 2464
    },
    {
      "epoch": 0.22230238535419578,
      "grad_norm": 2.4366293110138835,
      "learning_rate": 3.624533453003708e-06,
      "loss": 0.7773,
      "step": 2465
    },
    {
      "epoch": 0.22239256887766606,
      "grad_norm": 1.8830968238502754,
      "learning_rate": 3.6241926217171745e-06,
      "loss": 0.7928,
      "step": 2466
    },
    {
      "epoch": 0.22248275240113632,
      "grad_norm": 2.1050448583684407,
      "learning_rate": 3.6238516518455703e-06,
      "loss": 0.6987,
      "step": 2467
    },
    {
      "epoch": 0.22257293592460659,
      "grad_norm": 3.38193153380624,
      "learning_rate": 3.62351054341799e-06,
      "loss": 0.9089,
      "step": 2468
    },
    {
      "epoch": 0.22266311944807685,
      "grad_norm": 1.776175649591533,
      "learning_rate": 3.623169296463538e-06,
      "loss": 0.8111,
      "step": 2469
    },
    {
      "epoch": 0.2227533029715471,
      "grad_norm": 1.5697559490160997,
      "learning_rate": 3.6228279110113316e-06,
      "loss": 0.7858,
      "step": 2470
    },
    {
      "epoch": 0.22284348649501737,
      "grad_norm": 1.8604695920973544,
      "learning_rate": 3.6224863870904994e-06,
      "loss": 0.7895,
      "step": 2471
    },
    {
      "epoch": 0.22293367001848763,
      "grad_norm": 1.8299140978803725,
      "learning_rate": 3.6221447247301827e-06,
      "loss": 0.7946,
      "step": 2472
    },
    {
      "epoch": 0.2230238535419579,
      "grad_norm": 1.554926714872134,
      "learning_rate": 3.6218029239595332e-06,
      "loss": 0.7343,
      "step": 2473
    },
    {
      "epoch": 0.22311403706542815,
      "grad_norm": 1.592889291559842,
      "learning_rate": 3.621460984807716e-06,
      "loss": 0.8084,
      "step": 2474
    },
    {
      "epoch": 0.2232042205888984,
      "grad_norm": 2.1881117918895625,
      "learning_rate": 3.621118907303907e-06,
      "loss": 0.7592,
      "step": 2475
    },
    {
      "epoch": 0.22329440411236867,
      "grad_norm": 2.0679859836553462,
      "learning_rate": 3.620776691477294e-06,
      "loss": 0.7847,
      "step": 2476
    },
    {
      "epoch": 0.22338458763583893,
      "grad_norm": 0.6747266666339614,
      "learning_rate": 3.6204343373570765e-06,
      "loss": 0.551,
      "step": 2477
    },
    {
      "epoch": 0.2234747711593092,
      "grad_norm": 1.5076500997893296,
      "learning_rate": 3.620091844972467e-06,
      "loss": 0.8075,
      "step": 2478
    },
    {
      "epoch": 0.22356495468277945,
      "grad_norm": 1.5988209403422053,
      "learning_rate": 3.619749214352688e-06,
      "loss": 0.8108,
      "step": 2479
    },
    {
      "epoch": 0.22365513820624972,
      "grad_norm": 1.9511202479836232,
      "learning_rate": 3.6194064455269744e-06,
      "loss": 0.8899,
      "step": 2480
    },
    {
      "epoch": 0.22374532172971998,
      "grad_norm": 2.1045260610310432,
      "learning_rate": 3.6190635385245737e-06,
      "loss": 0.7069,
      "step": 2481
    },
    {
      "epoch": 0.22383550525319024,
      "grad_norm": 1.822931656548767,
      "learning_rate": 3.618720493374745e-06,
      "loss": 0.7193,
      "step": 2482
    },
    {
      "epoch": 0.2239256887766605,
      "grad_norm": 2.337113100559301,
      "learning_rate": 3.6183773101067575e-06,
      "loss": 0.6424,
      "step": 2483
    },
    {
      "epoch": 0.22401587230013076,
      "grad_norm": 1.504187461025957,
      "learning_rate": 3.6180339887498948e-06,
      "loss": 0.7075,
      "step": 2484
    },
    {
      "epoch": 0.22410605582360102,
      "grad_norm": 1.810678619233421,
      "learning_rate": 3.61769052933345e-06,
      "loss": 0.786,
      "step": 2485
    },
    {
      "epoch": 0.22419623934707128,
      "grad_norm": 8.78483188110047,
      "learning_rate": 3.6173469318867297e-06,
      "loss": 0.9383,
      "step": 2486
    },
    {
      "epoch": 0.22428642287054154,
      "grad_norm": 1.6840125180456493,
      "learning_rate": 3.617003196439051e-06,
      "loss": 0.8094,
      "step": 2487
    },
    {
      "epoch": 0.2243766063940118,
      "grad_norm": 1.6011339312019974,
      "learning_rate": 3.616659323019744e-06,
      "loss": 0.7923,
      "step": 2488
    },
    {
      "epoch": 0.22446678991748206,
      "grad_norm": 1.8847352020496388,
      "learning_rate": 3.616315311658149e-06,
      "loss": 0.8135,
      "step": 2489
    },
    {
      "epoch": 0.22455697344095235,
      "grad_norm": 1.885967033806039,
      "learning_rate": 3.6159711623836195e-06,
      "loss": 0.844,
      "step": 2490
    },
    {
      "epoch": 0.2246471569644226,
      "grad_norm": 0.8258303394000318,
      "learning_rate": 3.6156268752255203e-06,
      "loss": 0.5651,
      "step": 2491
    },
    {
      "epoch": 0.22473734048789287,
      "grad_norm": 1.4774082728537345,
      "learning_rate": 3.615282450213227e-06,
      "loss": 0.7479,
      "step": 2492
    },
    {
      "epoch": 0.22482752401136313,
      "grad_norm": 1.9174424204828966,
      "learning_rate": 3.614937887376128e-06,
      "loss": 0.7978,
      "step": 2493
    },
    {
      "epoch": 0.2249177075348334,
      "grad_norm": 2.1594688604909154,
      "learning_rate": 3.614593186743625e-06,
      "loss": 0.7997,
      "step": 2494
    },
    {
      "epoch": 0.22500789105830366,
      "grad_norm": 2.045983273151595,
      "learning_rate": 3.614248348345128e-06,
      "loss": 0.7269,
      "step": 2495
    },
    {
      "epoch": 0.22509807458177392,
      "grad_norm": 1.7422910924656103,
      "learning_rate": 3.6139033722100614e-06,
      "loss": 0.8364,
      "step": 2496
    },
    {
      "epoch": 0.22518825810524418,
      "grad_norm": 2.165307523457714,
      "learning_rate": 3.6135582583678596e-06,
      "loss": 0.8608,
      "step": 2497
    },
    {
      "epoch": 0.22527844162871444,
      "grad_norm": 1.8198457700754678,
      "learning_rate": 3.61321300684797e-06,
      "loss": 0.8048,
      "step": 2498
    },
    {
      "epoch": 0.2253686251521847,
      "grad_norm": 1.2230972128602977,
      "learning_rate": 3.6128676176798527e-06,
      "loss": 0.8079,
      "step": 2499
    },
    {
      "epoch": 0.22545880867565496,
      "grad_norm": 1.8355395595991701,
      "learning_rate": 3.612522090892976e-06,
      "loss": 0.8058,
      "step": 2500
    },
    {
      "epoch": 0.22554899219912522,
      "grad_norm": 1.3623958652843513,
      "learning_rate": 3.6121764265168232e-06,
      "loss": 0.7468,
      "step": 2501
    },
    {
      "epoch": 0.22563917572259548,
      "grad_norm": 1.4569949229930217,
      "learning_rate": 3.611830624580888e-06,
      "loss": 0.8121,
      "step": 2502
    },
    {
      "epoch": 0.22572935924606574,
      "grad_norm": 1.7520841041348547,
      "learning_rate": 3.6114846851146767e-06,
      "loss": 0.795,
      "step": 2503
    },
    {
      "epoch": 0.225819542769536,
      "grad_norm": 1.647333001108315,
      "learning_rate": 3.6111386081477068e-06,
      "loss": 0.6835,
      "step": 2504
    },
    {
      "epoch": 0.22590972629300626,
      "grad_norm": 2.442032536857725,
      "learning_rate": 3.6107923937095066e-06,
      "loss": 0.8046,
      "step": 2505
    },
    {
      "epoch": 0.22599990981647652,
      "grad_norm": 3.910553171779405,
      "learning_rate": 3.6104460418296173e-06,
      "loss": 0.7162,
      "step": 2506
    },
    {
      "epoch": 0.22609009333994678,
      "grad_norm": 1.8922867841054567,
      "learning_rate": 3.6100995525375924e-06,
      "loss": 0.8363,
      "step": 2507
    },
    {
      "epoch": 0.22618027686341705,
      "grad_norm": 2.0563052230886614,
      "learning_rate": 3.6097529258629952e-06,
      "loss": 0.7299,
      "step": 2508
    },
    {
      "epoch": 0.2262704603868873,
      "grad_norm": 1.70081018047408,
      "learning_rate": 3.6094061618354027e-06,
      "loss": 0.816,
      "step": 2509
    },
    {
      "epoch": 0.22636064391035757,
      "grad_norm": 1.9378741733657776,
      "learning_rate": 3.609059260484402e-06,
      "loss": 0.7825,
      "step": 2510
    },
    {
      "epoch": 0.22645082743382783,
      "grad_norm": 2.8892730025520894,
      "learning_rate": 3.6087122218395935e-06,
      "loss": 0.7307,
      "step": 2511
    },
    {
      "epoch": 0.2265410109572981,
      "grad_norm": 1.857194086300967,
      "learning_rate": 3.608365045930587e-06,
      "loss": 0.7537,
      "step": 2512
    },
    {
      "epoch": 0.22663119448076835,
      "grad_norm": 1.8718700351553719,
      "learning_rate": 3.608017732787007e-06,
      "loss": 0.7756,
      "step": 2513
    },
    {
      "epoch": 0.22672137800423864,
      "grad_norm": 1.589232231085208,
      "learning_rate": 3.6076702824384875e-06,
      "loss": 0.7156,
      "step": 2514
    },
    {
      "epoch": 0.2268115615277089,
      "grad_norm": 1.6209641738029517,
      "learning_rate": 3.607322694914675e-06,
      "loss": 0.8008,
      "step": 2515
    },
    {
      "epoch": 0.22690174505117916,
      "grad_norm": 2.0001356555709005,
      "learning_rate": 3.606974970245227e-06,
      "loss": 0.7579,
      "step": 2516
    },
    {
      "epoch": 0.22699192857464942,
      "grad_norm": 1.9578139515937605,
      "learning_rate": 3.606627108459814e-06,
      "loss": 0.7359,
      "step": 2517
    },
    {
      "epoch": 0.22708211209811968,
      "grad_norm": 1.3586419858699854,
      "learning_rate": 3.6062791095881174e-06,
      "loss": 0.7127,
      "step": 2518
    },
    {
      "epoch": 0.22717229562158994,
      "grad_norm": 1.869995814436038,
      "learning_rate": 3.6059309736598303e-06,
      "loss": 0.8607,
      "step": 2519
    },
    {
      "epoch": 0.2272624791450602,
      "grad_norm": 1.8695601069339067,
      "learning_rate": 3.605582700704657e-06,
      "loss": 0.7605,
      "step": 2520
    },
    {
      "epoch": 0.22735266266853046,
      "grad_norm": 1.8518383609315636,
      "learning_rate": 3.6052342907523146e-06,
      "loss": 0.7875,
      "step": 2521
    },
    {
      "epoch": 0.22744284619200072,
      "grad_norm": 1.9951739496866072,
      "learning_rate": 3.604885743832532e-06,
      "loss": 0.7186,
      "step": 2522
    },
    {
      "epoch": 0.22753302971547099,
      "grad_norm": 0.6920781077104031,
      "learning_rate": 3.6045370599750482e-06,
      "loss": 0.5812,
      "step": 2523
    },
    {
      "epoch": 0.22762321323894125,
      "grad_norm": 1.6079585684662565,
      "learning_rate": 3.604188239209615e-06,
      "loss": 0.7723,
      "step": 2524
    },
    {
      "epoch": 0.2277133967624115,
      "grad_norm": 2.927347372687577,
      "learning_rate": 3.603839281565996e-06,
      "loss": 0.7095,
      "step": 2525
    },
    {
      "epoch": 0.22780358028588177,
      "grad_norm": 1.7530988776288838,
      "learning_rate": 3.603490187073966e-06,
      "loss": 0.8,
      "step": 2526
    },
    {
      "epoch": 0.22789376380935203,
      "grad_norm": 1.818054567023814,
      "learning_rate": 3.6031409557633117e-06,
      "loss": 0.8507,
      "step": 2527
    },
    {
      "epoch": 0.2279839473328223,
      "grad_norm": 1.8055398752884728,
      "learning_rate": 3.602791587663831e-06,
      "loss": 0.8244,
      "step": 2528
    },
    {
      "epoch": 0.22807413085629255,
      "grad_norm": 2.1704084535813064,
      "learning_rate": 3.6024420828053348e-06,
      "loss": 0.8233,
      "step": 2529
    },
    {
      "epoch": 0.2281643143797628,
      "grad_norm": 1.5203321841809267,
      "learning_rate": 3.6020924412176445e-06,
      "loss": 0.859,
      "step": 2530
    },
    {
      "epoch": 0.22825449790323307,
      "grad_norm": 1.5847324998800736,
      "learning_rate": 3.601742662930593e-06,
      "loss": 0.7588,
      "step": 2531
    },
    {
      "epoch": 0.22834468142670333,
      "grad_norm": 1.8484708188069827,
      "learning_rate": 3.6013927479740248e-06,
      "loss": 0.7491,
      "step": 2532
    },
    {
      "epoch": 0.2284348649501736,
      "grad_norm": 1.8670232213746625,
      "learning_rate": 3.6010426963777985e-06,
      "loss": 0.8596,
      "step": 2533
    },
    {
      "epoch": 0.22852504847364385,
      "grad_norm": 1.6449445264095077,
      "learning_rate": 3.6006925081717804e-06,
      "loss": 0.8331,
      "step": 2534
    },
    {
      "epoch": 0.22861523199711412,
      "grad_norm": 1.469184405428376,
      "learning_rate": 3.600342183385852e-06,
      "loss": 0.7517,
      "step": 2535
    },
    {
      "epoch": 0.22870541552058438,
      "grad_norm": 1.4923573342384926,
      "learning_rate": 3.5999917220499043e-06,
      "loss": 0.7256,
      "step": 2536
    },
    {
      "epoch": 0.22879559904405466,
      "grad_norm": 1.753588539353044,
      "learning_rate": 3.5996411241938404e-06,
      "loss": 0.7905,
      "step": 2537
    },
    {
      "epoch": 0.22888578256752493,
      "grad_norm": 1.6548084967647967,
      "learning_rate": 3.5992903898475752e-06,
      "loss": 0.7457,
      "step": 2538
    },
    {
      "epoch": 0.2289759660909952,
      "grad_norm": 1.4471714745912896,
      "learning_rate": 3.5989395190410365e-06,
      "loss": 0.8856,
      "step": 2539
    },
    {
      "epoch": 0.22906614961446545,
      "grad_norm": 2.079528105464905,
      "learning_rate": 3.598588511804161e-06,
      "loss": 0.7915,
      "step": 2540
    },
    {
      "epoch": 0.2291563331379357,
      "grad_norm": 2.1656387042428045,
      "learning_rate": 3.5982373681668987e-06,
      "loss": 0.7917,
      "step": 2541
    },
    {
      "epoch": 0.22924651666140597,
      "grad_norm": 4.1544155575577575,
      "learning_rate": 3.597886088159212e-06,
      "loss": 0.7889,
      "step": 2542
    },
    {
      "epoch": 0.22933670018487623,
      "grad_norm": 1.5314116100742527,
      "learning_rate": 3.597534671811074e-06,
      "loss": 0.7153,
      "step": 2543
    },
    {
      "epoch": 0.2294268837083465,
      "grad_norm": 1.6401538717378576,
      "learning_rate": 3.5971831191524684e-06,
      "loss": 0.7657,
      "step": 2544
    },
    {
      "epoch": 0.22951706723181675,
      "grad_norm": 1.668762089831959,
      "learning_rate": 3.5968314302133925e-06,
      "loss": 0.8324,
      "step": 2545
    },
    {
      "epoch": 0.229607250755287,
      "grad_norm": 1.8611172119281028,
      "learning_rate": 3.596479605023854e-06,
      "loss": 0.7859,
      "step": 2546
    },
    {
      "epoch": 0.22969743427875727,
      "grad_norm": 5.737417635191417,
      "learning_rate": 3.596127643613873e-06,
      "loss": 0.8002,
      "step": 2547
    },
    {
      "epoch": 0.22978761780222753,
      "grad_norm": 1.5230865856863025,
      "learning_rate": 3.59577554601348e-06,
      "loss": 0.8075,
      "step": 2548
    },
    {
      "epoch": 0.2298778013256978,
      "grad_norm": 1.46292175341339,
      "learning_rate": 3.595423312252719e-06,
      "loss": 0.8328,
      "step": 2549
    },
    {
      "epoch": 0.22996798484916806,
      "grad_norm": 2.525004369901898,
      "learning_rate": 3.5950709423616436e-06,
      "loss": 0.7732,
      "step": 2550
    },
    {
      "epoch": 0.23005816837263832,
      "grad_norm": 2.3400611840098686,
      "learning_rate": 3.5947184363703203e-06,
      "loss": 0.7679,
      "step": 2551
    },
    {
      "epoch": 0.23014835189610858,
      "grad_norm": 1.5607707563661015,
      "learning_rate": 3.5943657943088274e-06,
      "loss": 0.7378,
      "step": 2552
    },
    {
      "epoch": 0.23023853541957884,
      "grad_norm": 1.926741107417763,
      "learning_rate": 3.5940130162072525e-06,
      "loss": 0.7103,
      "step": 2553
    },
    {
      "epoch": 0.2303287189430491,
      "grad_norm": 1.9088030119143926,
      "learning_rate": 3.5936601020956985e-06,
      "loss": 0.7858,
      "step": 2554
    },
    {
      "epoch": 0.23041890246651936,
      "grad_norm": 1.6462442713980328,
      "learning_rate": 3.5933070520042772e-06,
      "loss": 0.8206,
      "step": 2555
    },
    {
      "epoch": 0.23050908598998962,
      "grad_norm": 0.7627721300893873,
      "learning_rate": 3.5929538659631133e-06,
      "loss": 0.6309,
      "step": 2556
    },
    {
      "epoch": 0.23059926951345988,
      "grad_norm": 2.3181569783176994,
      "learning_rate": 3.592600544002341e-06,
      "loss": 0.8156,
      "step": 2557
    },
    {
      "epoch": 0.23068945303693014,
      "grad_norm": 2.5070819681992402,
      "learning_rate": 3.5922470861521098e-06,
      "loss": 0.7985,
      "step": 2558
    },
    {
      "epoch": 0.2307796365604004,
      "grad_norm": 2.082706649620345,
      "learning_rate": 3.591893492442577e-06,
      "loss": 0.7516,
      "step": 2559
    },
    {
      "epoch": 0.23086982008387066,
      "grad_norm": 1.9292660766540726,
      "learning_rate": 3.591539762903914e-06,
      "loss": 0.8484,
      "step": 2560
    },
    {
      "epoch": 0.23096000360734095,
      "grad_norm": 1.6781836586473122,
      "learning_rate": 3.591185897566303e-06,
      "loss": 0.7151,
      "step": 2561
    },
    {
      "epoch": 0.2310501871308112,
      "grad_norm": 2.2821595978288105,
      "learning_rate": 3.590831896459937e-06,
      "loss": 0.7844,
      "step": 2562
    },
    {
      "epoch": 0.23114037065428147,
      "grad_norm": 2.0638656718729647,
      "learning_rate": 3.5904777596150222e-06,
      "loss": 0.7944,
      "step": 2563
    },
    {
      "epoch": 0.23123055417775173,
      "grad_norm": 1.8223873641551482,
      "learning_rate": 3.590123487061775e-06,
      "loss": 0.7806,
      "step": 2564
    },
    {
      "epoch": 0.231320737701222,
      "grad_norm": 2.0745868823139917,
      "learning_rate": 3.589769078830424e-06,
      "loss": 0.8837,
      "step": 2565
    },
    {
      "epoch": 0.23141092122469226,
      "grad_norm": 1.9587714401853689,
      "learning_rate": 3.58941453495121e-06,
      "loss": 0.7655,
      "step": 2566
    },
    {
      "epoch": 0.23150110474816252,
      "grad_norm": 1.6619638453245287,
      "learning_rate": 3.5890598554543834e-06,
      "loss": 0.7981,
      "step": 2567
    },
    {
      "epoch": 0.23159128827163278,
      "grad_norm": 1.7437213936376879,
      "learning_rate": 3.5887050403702073e-06,
      "loss": 0.8373,
      "step": 2568
    },
    {
      "epoch": 0.23168147179510304,
      "grad_norm": 1.845829260012895,
      "learning_rate": 3.588350089728958e-06,
      "loss": 0.7581,
      "step": 2569
    },
    {
      "epoch": 0.2317716553185733,
      "grad_norm": 1.832827787343667,
      "learning_rate": 3.5879950035609204e-06,
      "loss": 0.8108,
      "step": 2570
    },
    {
      "epoch": 0.23186183884204356,
      "grad_norm": 1.8386236141914625,
      "learning_rate": 3.5876397818963933e-06,
      "loss": 0.7821,
      "step": 2571
    },
    {
      "epoch": 0.23195202236551382,
      "grad_norm": 1.6189730457420703,
      "learning_rate": 3.5872844247656858e-06,
      "loss": 0.8369,
      "step": 2572
    },
    {
      "epoch": 0.23204220588898408,
      "grad_norm": 2.3074000026737598,
      "learning_rate": 3.5869289321991195e-06,
      "loss": 0.7893,
      "step": 2573
    },
    {
      "epoch": 0.23213238941245434,
      "grad_norm": 1.5424860331986527,
      "learning_rate": 3.5865733042270263e-06,
      "loss": 0.78,
      "step": 2574
    },
    {
      "epoch": 0.2322225729359246,
      "grad_norm": 1.784737235211049,
      "learning_rate": 3.5862175408797498e-06,
      "loss": 0.8147,
      "step": 2575
    },
    {
      "epoch": 0.23231275645939486,
      "grad_norm": 8.848775884593401,
      "learning_rate": 3.585861642187647e-06,
      "loss": 0.7258,
      "step": 2576
    },
    {
      "epoch": 0.23240293998286513,
      "grad_norm": 1.4157032589196346,
      "learning_rate": 3.5855056081810845e-06,
      "loss": 0.7816,
      "step": 2577
    },
    {
      "epoch": 0.2324931235063354,
      "grad_norm": 1.5888716141370036,
      "learning_rate": 3.5851494388904406e-06,
      "loss": 0.784,
      "step": 2578
    },
    {
      "epoch": 0.23258330702980565,
      "grad_norm": 1.5908693197158998,
      "learning_rate": 3.5847931343461064e-06,
      "loss": 0.7219,
      "step": 2579
    },
    {
      "epoch": 0.2326734905532759,
      "grad_norm": 2.5935564083792015,
      "learning_rate": 3.5844366945784835e-06,
      "loss": 0.7148,
      "step": 2580
    },
    {
      "epoch": 0.23276367407674617,
      "grad_norm": 1.8382200296553586,
      "learning_rate": 3.5840801196179856e-06,
      "loss": 0.846,
      "step": 2581
    },
    {
      "epoch": 0.23285385760021643,
      "grad_norm": 1.6966831977028916,
      "learning_rate": 3.583723409495037e-06,
      "loss": 0.8372,
      "step": 2582
    },
    {
      "epoch": 0.2329440411236867,
      "grad_norm": 4.011748702876253,
      "learning_rate": 3.5833665642400747e-06,
      "loss": 0.8378,
      "step": 2583
    },
    {
      "epoch": 0.23303422464715695,
      "grad_norm": 2.533173571953191,
      "learning_rate": 3.5830095838835472e-06,
      "loss": 0.7112,
      "step": 2584
    },
    {
      "epoch": 0.23312440817062724,
      "grad_norm": 1.8051526679782135,
      "learning_rate": 3.5826524684559125e-06,
      "loss": 0.7552,
      "step": 2585
    },
    {
      "epoch": 0.2332145916940975,
      "grad_norm": 2.296416983059263,
      "learning_rate": 3.5822952179876433e-06,
      "loss": 0.8688,
      "step": 2586
    },
    {
      "epoch": 0.23330477521756776,
      "grad_norm": 1.5288180287306674,
      "learning_rate": 3.5819378325092205e-06,
      "loss": 0.7616,
      "step": 2587
    },
    {
      "epoch": 0.23339495874103802,
      "grad_norm": 1.8666165169155458,
      "learning_rate": 3.581580312051139e-06,
      "loss": 0.8072,
      "step": 2588
    },
    {
      "epoch": 0.23348514226450828,
      "grad_norm": 1.6438148369711414,
      "learning_rate": 3.5812226566439057e-06,
      "loss": 0.7137,
      "step": 2589
    },
    {
      "epoch": 0.23357532578797854,
      "grad_norm": 1.7034695478023425,
      "learning_rate": 3.580864866318036e-06,
      "loss": 0.7618,
      "step": 2590
    },
    {
      "epoch": 0.2336655093114488,
      "grad_norm": 1.9109562795719066,
      "learning_rate": 3.580506941104059e-06,
      "loss": 0.7894,
      "step": 2591
    },
    {
      "epoch": 0.23375569283491907,
      "grad_norm": 1.5285170943352813,
      "learning_rate": 3.580148881032515e-06,
      "loss": 0.7735,
      "step": 2592
    },
    {
      "epoch": 0.23384587635838933,
      "grad_norm": 1.4981279612334089,
      "learning_rate": 3.5797906861339556e-06,
      "loss": 0.7921,
      "step": 2593
    },
    {
      "epoch": 0.2339360598818596,
      "grad_norm": 2.1702415855691934,
      "learning_rate": 3.5794323564389435e-06,
      "loss": 0.8131,
      "step": 2594
    },
    {
      "epoch": 0.23402624340532985,
      "grad_norm": 2.177950587466137,
      "learning_rate": 3.579073891978055e-06,
      "loss": 0.797,
      "step": 2595
    },
    {
      "epoch": 0.2341164269288001,
      "grad_norm": 2.0788693349858667,
      "learning_rate": 3.5787152927818746e-06,
      "loss": 0.7563,
      "step": 2596
    },
    {
      "epoch": 0.23420661045227037,
      "grad_norm": 1.5505138714418027,
      "learning_rate": 3.5783565588810003e-06,
      "loss": 0.8482,
      "step": 2597
    },
    {
      "epoch": 0.23429679397574063,
      "grad_norm": 1.6371933424252516,
      "learning_rate": 3.5779976903060412e-06,
      "loss": 0.7565,
      "step": 2598
    },
    {
      "epoch": 0.2343869774992109,
      "grad_norm": 1.748020142079106,
      "learning_rate": 3.577638687087619e-06,
      "loss": 0.8211,
      "step": 2599
    },
    {
      "epoch": 0.23447716102268115,
      "grad_norm": 1.9470088563706864,
      "learning_rate": 3.577279549256364e-06,
      "loss": 0.7939,
      "step": 2600
    },
    {
      "epoch": 0.2345673445461514,
      "grad_norm": 1.620086283328665,
      "learning_rate": 3.5769202768429213e-06,
      "loss": 0.8304,
      "step": 2601
    },
    {
      "epoch": 0.23465752806962167,
      "grad_norm": 1.7808238406442285,
      "learning_rate": 3.5765608698779454e-06,
      "loss": 0.7637,
      "step": 2602
    },
    {
      "epoch": 0.23474771159309193,
      "grad_norm": 1.808031691020277,
      "learning_rate": 3.5762013283921033e-06,
      "loss": 0.7985,
      "step": 2603
    },
    {
      "epoch": 0.2348378951165622,
      "grad_norm": 1.80721842006332,
      "learning_rate": 3.5758416524160728e-06,
      "loss": 0.7907,
      "step": 2604
    },
    {
      "epoch": 0.23492807864003246,
      "grad_norm": 1.5700050926125901,
      "learning_rate": 3.5754818419805427e-06,
      "loss": 0.7809,
      "step": 2605
    },
    {
      "epoch": 0.23501826216350272,
      "grad_norm": 1.6776073261157411,
      "learning_rate": 3.575121897116216e-06,
      "loss": 0.7632,
      "step": 2606
    },
    {
      "epoch": 0.23510844568697298,
      "grad_norm": 1.473549812057435,
      "learning_rate": 3.574761817853803e-06,
      "loss": 0.8193,
      "step": 2607
    },
    {
      "epoch": 0.23519862921044324,
      "grad_norm": 21.360498695981224,
      "learning_rate": 3.5744016042240287e-06,
      "loss": 0.7232,
      "step": 2608
    },
    {
      "epoch": 0.23528881273391353,
      "grad_norm": 2.1367573568981344,
      "learning_rate": 3.5740412562576286e-06,
      "loss": 0.7207,
      "step": 2609
    },
    {
      "epoch": 0.2353789962573838,
      "grad_norm": 2.1181785825703914,
      "learning_rate": 3.573680773985349e-06,
      "loss": 0.7859,
      "step": 2610
    },
    {
      "epoch": 0.23546917978085405,
      "grad_norm": 1.8825509297243412,
      "learning_rate": 3.5733201574379486e-06,
      "loss": 0.7825,
      "step": 2611
    },
    {
      "epoch": 0.2355593633043243,
      "grad_norm": 1.684272859268469,
      "learning_rate": 3.5729594066461975e-06,
      "loss": 0.7236,
      "step": 2612
    },
    {
      "epoch": 0.23564954682779457,
      "grad_norm": 1.4158712379241911,
      "learning_rate": 3.572598521640876e-06,
      "loss": 0.7994,
      "step": 2613
    },
    {
      "epoch": 0.23573973035126483,
      "grad_norm": 0.7822531934912534,
      "learning_rate": 3.5722375024527782e-06,
      "loss": 0.6365,
      "step": 2614
    },
    {
      "epoch": 0.2358299138747351,
      "grad_norm": 1.3724646602312496,
      "learning_rate": 3.571876349112707e-06,
      "loss": 0.8068,
      "step": 2615
    },
    {
      "epoch": 0.23592009739820535,
      "grad_norm": 1.604724646624007,
      "learning_rate": 3.5715150616514784e-06,
      "loss": 0.7821,
      "step": 2616
    },
    {
      "epoch": 0.2360102809216756,
      "grad_norm": 2.163819557627764,
      "learning_rate": 3.5711536400999196e-06,
      "loss": 0.8388,
      "step": 2617
    },
    {
      "epoch": 0.23610046444514587,
      "grad_norm": 0.7185124128990965,
      "learning_rate": 3.570792084488869e-06,
      "loss": 0.6378,
      "step": 2618
    },
    {
      "epoch": 0.23619064796861614,
      "grad_norm": 0.7176442555547073,
      "learning_rate": 3.5704303948491764e-06,
      "loss": 0.6494,
      "step": 2619
    },
    {
      "epoch": 0.2362808314920864,
      "grad_norm": 2.0073842582733805,
      "learning_rate": 3.5700685712117035e-06,
      "loss": 0.8116,
      "step": 2620
    },
    {
      "epoch": 0.23637101501555666,
      "grad_norm": 2.1175491756057045,
      "learning_rate": 3.5697066136073227e-06,
      "loss": 0.7759,
      "step": 2621
    },
    {
      "epoch": 0.23646119853902692,
      "grad_norm": 1.458772983582437,
      "learning_rate": 3.5693445220669184e-06,
      "loss": 0.7985,
      "step": 2622
    },
    {
      "epoch": 0.23655138206249718,
      "grad_norm": 1.3658541471855523,
      "learning_rate": 3.568982296621386e-06,
      "loss": 0.7815,
      "step": 2623
    },
    {
      "epoch": 0.23664156558596744,
      "grad_norm": 1.7863963554760272,
      "learning_rate": 3.5686199373016325e-06,
      "loss": 0.8598,
      "step": 2624
    },
    {
      "epoch": 0.2367317491094377,
      "grad_norm": 2.1486647607075073,
      "learning_rate": 3.568257444138577e-06,
      "loss": 0.7357,
      "step": 2625
    },
    {
      "epoch": 0.23682193263290796,
      "grad_norm": 2.35166358096837,
      "learning_rate": 3.5678948171631495e-06,
      "loss": 0.7914,
      "step": 2626
    },
    {
      "epoch": 0.23691211615637822,
      "grad_norm": 1.5659417203238457,
      "learning_rate": 3.5675320564062908e-06,
      "loss": 0.795,
      "step": 2627
    },
    {
      "epoch": 0.23700229967984848,
      "grad_norm": 1.819395169795312,
      "learning_rate": 3.5671691618989533e-06,
      "loss": 0.6986,
      "step": 2628
    },
    {
      "epoch": 0.23709248320331874,
      "grad_norm": 1.7357118283051998,
      "learning_rate": 3.5668061336721024e-06,
      "loss": 0.7979,
      "step": 2629
    },
    {
      "epoch": 0.237182666726789,
      "grad_norm": 1.6411173490447206,
      "learning_rate": 3.5664429717567117e-06,
      "loss": 0.7828,
      "step": 2630
    },
    {
      "epoch": 0.23727285025025927,
      "grad_norm": 1.7209313855011492,
      "learning_rate": 3.56607967618377e-06,
      "loss": 0.8169,
      "step": 2631
    },
    {
      "epoch": 0.23736303377372953,
      "grad_norm": 2.132832286030304,
      "learning_rate": 3.5657162469842754e-06,
      "loss": 0.7984,
      "step": 2632
    },
    {
      "epoch": 0.23745321729719981,
      "grad_norm": 1.8044074695864674,
      "learning_rate": 3.5653526841892374e-06,
      "loss": 0.79,
      "step": 2633
    },
    {
      "epoch": 0.23754340082067008,
      "grad_norm": 2.062780418984483,
      "learning_rate": 3.564988987829676e-06,
      "loss": 0.8181,
      "step": 2634
    },
    {
      "epoch": 0.23763358434414034,
      "grad_norm": 1.656981324734654,
      "learning_rate": 3.564625157936626e-06,
      "loss": 0.7333,
      "step": 2635
    },
    {
      "epoch": 0.2377237678676106,
      "grad_norm": 1.7335132141797944,
      "learning_rate": 3.56426119454113e-06,
      "loss": 0.8249,
      "step": 2636
    },
    {
      "epoch": 0.23781395139108086,
      "grad_norm": 1.9311203058143644,
      "learning_rate": 3.5638970976742436e-06,
      "loss": 0.8636,
      "step": 2637
    },
    {
      "epoch": 0.23790413491455112,
      "grad_norm": 2.0009165094392576,
      "learning_rate": 3.5635328673670335e-06,
      "loss": 0.8114,
      "step": 2638
    },
    {
      "epoch": 0.23799431843802138,
      "grad_norm": 1.4780246821550613,
      "learning_rate": 3.5631685036505783e-06,
      "loss": 0.8785,
      "step": 2639
    },
    {
      "epoch": 0.23808450196149164,
      "grad_norm": 1.3668394245110749,
      "learning_rate": 3.562804006555966e-06,
      "loss": 0.8,
      "step": 2640
    },
    {
      "epoch": 0.2381746854849619,
      "grad_norm": 2.8031411331035705,
      "learning_rate": 3.5624393761143e-06,
      "loss": 0.727,
      "step": 2641
    },
    {
      "epoch": 0.23826486900843216,
      "grad_norm": 5.072261300597406,
      "learning_rate": 3.5620746123566906e-06,
      "loss": 0.703,
      "step": 2642
    },
    {
      "epoch": 0.23835505253190242,
      "grad_norm": 1.7398233152987206,
      "learning_rate": 3.5617097153142623e-06,
      "loss": 0.7408,
      "step": 2643
    },
    {
      "epoch": 0.23844523605537268,
      "grad_norm": 1.647791008030521,
      "learning_rate": 3.5613446850181497e-06,
      "loss": 0.8245,
      "step": 2644
    },
    {
      "epoch": 0.23853541957884294,
      "grad_norm": 0.6105868198948975,
      "learning_rate": 3.5609795214994996e-06,
      "loss": 0.5374,
      "step": 2645
    },
    {
      "epoch": 0.2386256031023132,
      "grad_norm": 1.6570398858334092,
      "learning_rate": 3.560614224789469e-06,
      "loss": 0.8436,
      "step": 2646
    },
    {
      "epoch": 0.23871578662578347,
      "grad_norm": 1.8392840463469373,
      "learning_rate": 3.5602487949192285e-06,
      "loss": 0.7329,
      "step": 2647
    },
    {
      "epoch": 0.23880597014925373,
      "grad_norm": 3.377496290677827,
      "learning_rate": 3.559883231919957e-06,
      "loss": 0.7319,
      "step": 2648
    },
    {
      "epoch": 0.238896153672724,
      "grad_norm": 1.7819253410900446,
      "learning_rate": 3.5595175358228473e-06,
      "loss": 0.7608,
      "step": 2649
    },
    {
      "epoch": 0.23898633719619425,
      "grad_norm": 1.7320777181363511,
      "learning_rate": 3.5591517066591027e-06,
      "loss": 0.7852,
      "step": 2650
    },
    {
      "epoch": 0.2390765207196645,
      "grad_norm": 2.238691096039453,
      "learning_rate": 3.5587857444599364e-06,
      "loss": 0.7793,
      "step": 2651
    },
    {
      "epoch": 0.23916670424313477,
      "grad_norm": 1.6042190477150424,
      "learning_rate": 3.5584196492565766e-06,
      "loss": 0.7577,
      "step": 2652
    },
    {
      "epoch": 0.23925688776660503,
      "grad_norm": 3.0803914512477015,
      "learning_rate": 3.5580534210802587e-06,
      "loss": 0.852,
      "step": 2653
    },
    {
      "epoch": 0.2393470712900753,
      "grad_norm": 1.5355986523104868,
      "learning_rate": 3.557687059962232e-06,
      "loss": 0.8312,
      "step": 2654
    },
    {
      "epoch": 0.23943725481354555,
      "grad_norm": 1.5096711083519005,
      "learning_rate": 3.5573205659337558e-06,
      "loss": 0.7808,
      "step": 2655
    },
    {
      "epoch": 0.23952743833701584,
      "grad_norm": 1.5790846522362956,
      "learning_rate": 3.5569539390261025e-06,
      "loss": 0.7745,
      "step": 2656
    },
    {
      "epoch": 0.2396176218604861,
      "grad_norm": 1.2912469099807327,
      "learning_rate": 3.5565871792705543e-06,
      "loss": 0.8001,
      "step": 2657
    },
    {
      "epoch": 0.23970780538395636,
      "grad_norm": 1.4787719181916137,
      "learning_rate": 3.5562202866984045e-06,
      "loss": 0.7775,
      "step": 2658
    },
    {
      "epoch": 0.23979798890742662,
      "grad_norm": 1.8825022968212366,
      "learning_rate": 3.5558532613409594e-06,
      "loss": 0.801,
      "step": 2659
    },
    {
      "epoch": 0.23988817243089688,
      "grad_norm": 1.6007054710789903,
      "learning_rate": 3.555486103229535e-06,
      "loss": 0.727,
      "step": 2660
    },
    {
      "epoch": 0.23997835595436715,
      "grad_norm": 5.936917888816948,
      "learning_rate": 3.5551188123954595e-06,
      "loss": 0.8132,
      "step": 2661
    },
    {
      "epoch": 0.2400685394778374,
      "grad_norm": 1.6058730755599429,
      "learning_rate": 3.5547513888700715e-06,
      "loss": 0.6852,
      "step": 2662
    },
    {
      "epoch": 0.24015872300130767,
      "grad_norm": 1.551000392237404,
      "learning_rate": 3.5543838326847224e-06,
      "loss": 0.8186,
      "step": 2663
    },
    {
      "epoch": 0.24024890652477793,
      "grad_norm": 2.1784164386979374,
      "learning_rate": 3.5540161438707744e-06,
      "loss": 0.7637,
      "step": 2664
    },
    {
      "epoch": 0.2403390900482482,
      "grad_norm": 1.6985574772912777,
      "learning_rate": 3.5536483224596e-06,
      "loss": 0.9424,
      "step": 2665
    },
    {
      "epoch": 0.24042927357171845,
      "grad_norm": 2.0445523661425673,
      "learning_rate": 3.553280368482584e-06,
      "loss": 0.7644,
      "step": 2666
    },
    {
      "epoch": 0.2405194570951887,
      "grad_norm": 1.6836985402785538,
      "learning_rate": 3.5529122819711227e-06,
      "loss": 0.6113,
      "step": 2667
    },
    {
      "epoch": 0.24060964061865897,
      "grad_norm": 1.7098630553616918,
      "learning_rate": 3.5525440629566223e-06,
      "loss": 0.731,
      "step": 2668
    },
    {
      "epoch": 0.24069982414212923,
      "grad_norm": 2.427063926870231,
      "learning_rate": 3.552175711470502e-06,
      "loss": 0.8073,
      "step": 2669
    },
    {
      "epoch": 0.2407900076655995,
      "grad_norm": 3.9951004777669374,
      "learning_rate": 3.5518072275441912e-06,
      "loss": 0.8263,
      "step": 2670
    },
    {
      "epoch": 0.24088019118906975,
      "grad_norm": 1.534058233998456,
      "learning_rate": 3.551438611209131e-06,
      "loss": 0.8139,
      "step": 2671
    },
    {
      "epoch": 0.24097037471254001,
      "grad_norm": 2.3367540852315187,
      "learning_rate": 3.551069862496774e-06,
      "loss": 0.8834,
      "step": 2672
    },
    {
      "epoch": 0.24106055823601027,
      "grad_norm": 0.648316797259696,
      "learning_rate": 3.5507009814385846e-06,
      "loss": 0.599,
      "step": 2673
    },
    {
      "epoch": 0.24115074175948054,
      "grad_norm": 0.6542378640964303,
      "learning_rate": 3.550331968066036e-06,
      "loss": 0.5904,
      "step": 2674
    },
    {
      "epoch": 0.2412409252829508,
      "grad_norm": 1.9801757114164127,
      "learning_rate": 3.549962822410616e-06,
      "loss": 0.8038,
      "step": 2675
    },
    {
      "epoch": 0.24133110880642106,
      "grad_norm": 0.6948041879496969,
      "learning_rate": 3.5495935445038217e-06,
      "loss": 0.5868,
      "step": 2676
    },
    {
      "epoch": 0.24142129232989132,
      "grad_norm": 1.840771937050869,
      "learning_rate": 3.5492241343771612e-06,
      "loss": 0.8268,
      "step": 2677
    },
    {
      "epoch": 0.24151147585336158,
      "grad_norm": 2.7333405199964376,
      "learning_rate": 3.548854592062156e-06,
      "loss": 0.7817,
      "step": 2678
    },
    {
      "epoch": 0.24160165937683184,
      "grad_norm": 3.1324024265007604,
      "learning_rate": 3.548484917590336e-06,
      "loss": 0.7675,
      "step": 2679
    },
    {
      "epoch": 0.24169184290030213,
      "grad_norm": 2.019546479217542,
      "learning_rate": 3.5481151109932447e-06,
      "loss": 0.7032,
      "step": 2680
    },
    {
      "epoch": 0.2417820264237724,
      "grad_norm": 1.7438568191898556,
      "learning_rate": 3.5477451723024364e-06,
      "loss": 0.8131,
      "step": 2681
    },
    {
      "epoch": 0.24187220994724265,
      "grad_norm": 0.5888908470044281,
      "learning_rate": 3.5473751015494757e-06,
      "loss": 0.5711,
      "step": 2682
    },
    {
      "epoch": 0.2419623934707129,
      "grad_norm": 2.0818862721422313,
      "learning_rate": 3.547004898765939e-06,
      "loss": 0.8044,
      "step": 2683
    },
    {
      "epoch": 0.24205257699418317,
      "grad_norm": 1.3047209866732008,
      "learning_rate": 3.546634563983414e-06,
      "loss": 0.8065,
      "step": 2684
    },
    {
      "epoch": 0.24214276051765343,
      "grad_norm": 1.8343827827457477,
      "learning_rate": 3.5462640972335002e-06,
      "loss": 0.7509,
      "step": 2685
    },
    {
      "epoch": 0.2422329440411237,
      "grad_norm": 1.4876597150440327,
      "learning_rate": 3.5458934985478077e-06,
      "loss": 0.8088,
      "step": 2686
    },
    {
      "epoch": 0.24232312756459395,
      "grad_norm": 2.7081644347907456,
      "learning_rate": 3.5455227679579577e-06,
      "loss": 0.7624,
      "step": 2687
    },
    {
      "epoch": 0.24241331108806422,
      "grad_norm": 2.795392032214718,
      "learning_rate": 3.545151905495584e-06,
      "loss": 0.8919,
      "step": 2688
    },
    {
      "epoch": 0.24250349461153448,
      "grad_norm": 2.41389874640548,
      "learning_rate": 3.544780911192329e-06,
      "loss": 0.7737,
      "step": 2689
    },
    {
      "epoch": 0.24259367813500474,
      "grad_norm": 1.7094771889890992,
      "learning_rate": 3.544409785079849e-06,
      "loss": 0.7356,
      "step": 2690
    },
    {
      "epoch": 0.242683861658475,
      "grad_norm": 1.4399617718549638,
      "learning_rate": 3.5440385271898103e-06,
      "loss": 0.728,
      "step": 2691
    },
    {
      "epoch": 0.24277404518194526,
      "grad_norm": 1.6602644582152473,
      "learning_rate": 3.5436671375538903e-06,
      "loss": 0.7122,
      "step": 2692
    },
    {
      "epoch": 0.24286422870541552,
      "grad_norm": 2.352540684022895,
      "learning_rate": 3.543295616203779e-06,
      "loss": 0.7894,
      "step": 2693
    },
    {
      "epoch": 0.24295441222888578,
      "grad_norm": 1.8069568584086406,
      "learning_rate": 3.542923963171176e-06,
      "loss": 0.8822,
      "step": 2694
    },
    {
      "epoch": 0.24304459575235604,
      "grad_norm": 1.9423059109768976,
      "learning_rate": 3.542552178487793e-06,
      "loss": 0.7037,
      "step": 2695
    },
    {
      "epoch": 0.2431347792758263,
      "grad_norm": 2.1842005232955044,
      "learning_rate": 3.5421802621853523e-06,
      "loss": 0.7765,
      "step": 2696
    },
    {
      "epoch": 0.24322496279929656,
      "grad_norm": 1.825827369221717,
      "learning_rate": 3.5418082142955887e-06,
      "loss": 0.783,
      "step": 2697
    },
    {
      "epoch": 0.24331514632276682,
      "grad_norm": 0.6958463740887595,
      "learning_rate": 3.5414360348502463e-06,
      "loss": 0.6159,
      "step": 2698
    },
    {
      "epoch": 0.24340532984623708,
      "grad_norm": 1.5948700054852,
      "learning_rate": 3.5410637238810825e-06,
      "loss": 0.6938,
      "step": 2699
    },
    {
      "epoch": 0.24349551336970734,
      "grad_norm": 1.7417022987383406,
      "learning_rate": 3.5406912814198635e-06,
      "loss": 0.7428,
      "step": 2700
    },
    {
      "epoch": 0.2435856968931776,
      "grad_norm": 1.7537554227890604,
      "learning_rate": 3.54031870749837e-06,
      "loss": 0.75,
      "step": 2701
    },
    {
      "epoch": 0.24367588041664787,
      "grad_norm": 1.4242152144429512,
      "learning_rate": 3.539946002148391e-06,
      "loss": 0.7935,
      "step": 2702
    },
    {
      "epoch": 0.24376606394011813,
      "grad_norm": 2.553666870437187,
      "learning_rate": 3.5395731654017277e-06,
      "loss": 0.792,
      "step": 2703
    },
    {
      "epoch": 0.24385624746358842,
      "grad_norm": 2.1549704251837767,
      "learning_rate": 3.5392001972901923e-06,
      "loss": 0.8261,
      "step": 2704
    },
    {
      "epoch": 0.24394643098705868,
      "grad_norm": 1.6447143452762163,
      "learning_rate": 3.5388270978456098e-06,
      "loss": 0.7468,
      "step": 2705
    },
    {
      "epoch": 0.24403661451052894,
      "grad_norm": 1.8028598795441024,
      "learning_rate": 3.5384538670998137e-06,
      "loss": 0.7591,
      "step": 2706
    },
    {
      "epoch": 0.2441267980339992,
      "grad_norm": 1.8252699639253054,
      "learning_rate": 3.538080505084651e-06,
      "loss": 0.809,
      "step": 2707
    },
    {
      "epoch": 0.24421698155746946,
      "grad_norm": 1.4372291309741305,
      "learning_rate": 3.5377070118319788e-06,
      "loss": 0.7978,
      "step": 2708
    },
    {
      "epoch": 0.24430716508093972,
      "grad_norm": 2.3651280272914903,
      "learning_rate": 3.5373333873736657e-06,
      "loss": 0.7472,
      "step": 2709
    },
    {
      "epoch": 0.24439734860440998,
      "grad_norm": 1.4925414459505804,
      "learning_rate": 3.536959631741591e-06,
      "loss": 0.7928,
      "step": 2710
    },
    {
      "epoch": 0.24448753212788024,
      "grad_norm": 1.7872564536685298,
      "learning_rate": 3.536585744967646e-06,
      "loss": 0.8066,
      "step": 2711
    },
    {
      "epoch": 0.2445777156513505,
      "grad_norm": 1.9617413093002014,
      "learning_rate": 3.5362117270837326e-06,
      "loss": 0.7712,
      "step": 2712
    },
    {
      "epoch": 0.24466789917482076,
      "grad_norm": 1.5424571287605795,
      "learning_rate": 3.5358375781217634e-06,
      "loss": 0.686,
      "step": 2713
    },
    {
      "epoch": 0.24475808269829102,
      "grad_norm": 1.5693658660902403,
      "learning_rate": 3.535463298113664e-06,
      "loss": 0.8224,
      "step": 2714
    },
    {
      "epoch": 0.24484826622176128,
      "grad_norm": 1.6133617168614744,
      "learning_rate": 3.5350888870913697e-06,
      "loss": 0.7458,
      "step": 2715
    },
    {
      "epoch": 0.24493844974523155,
      "grad_norm": 1.668994882739195,
      "learning_rate": 3.5347143450868273e-06,
      "loss": 0.7921,
      "step": 2716
    },
    {
      "epoch": 0.2450286332687018,
      "grad_norm": 1.8712802865655032,
      "learning_rate": 3.534339672131994e-06,
      "loss": 0.8113,
      "step": 2717
    },
    {
      "epoch": 0.24511881679217207,
      "grad_norm": 2.201385243998123,
      "learning_rate": 3.5339648682588397e-06,
      "loss": 0.7203,
      "step": 2718
    },
    {
      "epoch": 0.24520900031564233,
      "grad_norm": 1.7375263074900393,
      "learning_rate": 3.533589933499345e-06,
      "loss": 0.8334,
      "step": 2719
    },
    {
      "epoch": 0.2452991838391126,
      "grad_norm": 2.3582661024919034,
      "learning_rate": 3.533214867885501e-06,
      "loss": 0.7212,
      "step": 2720
    },
    {
      "epoch": 0.24538936736258285,
      "grad_norm": 1.9400433490275306,
      "learning_rate": 3.53283967144931e-06,
      "loss": 0.722,
      "step": 2721
    },
    {
      "epoch": 0.2454795508860531,
      "grad_norm": 2.2621536495597083,
      "learning_rate": 3.532464344222787e-06,
      "loss": 0.7983,
      "step": 2722
    },
    {
      "epoch": 0.24556973440952337,
      "grad_norm": 2.0266027253922716,
      "learning_rate": 3.532088886237956e-06,
      "loss": 0.8294,
      "step": 2723
    },
    {
      "epoch": 0.24565991793299363,
      "grad_norm": 0.6365730727473645,
      "learning_rate": 3.5317132975268535e-06,
      "loss": 0.5272,
      "step": 2724
    },
    {
      "epoch": 0.2457501014564639,
      "grad_norm": 1.5160744384114409,
      "learning_rate": 3.531337578121526e-06,
      "loss": 0.8106,
      "step": 2725
    },
    {
      "epoch": 0.24584028497993415,
      "grad_norm": 2.1297646008288917,
      "learning_rate": 3.530961728054033e-06,
      "loss": 0.8183,
      "step": 2726
    },
    {
      "epoch": 0.24593046850340441,
      "grad_norm": 1.5864163883080271,
      "learning_rate": 3.5305857473564435e-06,
      "loss": 0.7458,
      "step": 2727
    },
    {
      "epoch": 0.2460206520268747,
      "grad_norm": 2.00299765528724,
      "learning_rate": 3.5302096360608385e-06,
      "loss": 0.7226,
      "step": 2728
    },
    {
      "epoch": 0.24611083555034496,
      "grad_norm": 1.6292786389804876,
      "learning_rate": 3.5298333941993105e-06,
      "loss": 0.8365,
      "step": 2729
    },
    {
      "epoch": 0.24620101907381522,
      "grad_norm": 1.804656651881407,
      "learning_rate": 3.529457021803962e-06,
      "loss": 0.7802,
      "step": 2730
    },
    {
      "epoch": 0.24629120259728549,
      "grad_norm": 1.914875963143209,
      "learning_rate": 3.529080518906906e-06,
      "loss": 0.8036,
      "step": 2731
    },
    {
      "epoch": 0.24638138612075575,
      "grad_norm": 1.641896145978136,
      "learning_rate": 3.5287038855402696e-06,
      "loss": 0.7714,
      "step": 2732
    },
    {
      "epoch": 0.246471569644226,
      "grad_norm": 1.884083868662609,
      "learning_rate": 3.528327121736188e-06,
      "loss": 0.8246,
      "step": 2733
    },
    {
      "epoch": 0.24656175316769627,
      "grad_norm": 1.6774892927757599,
      "learning_rate": 3.52795022752681e-06,
      "loss": 0.8665,
      "step": 2734
    },
    {
      "epoch": 0.24665193669116653,
      "grad_norm": 2.1986396528511896,
      "learning_rate": 3.5275732029442925e-06,
      "loss": 0.8593,
      "step": 2735
    },
    {
      "epoch": 0.2467421202146368,
      "grad_norm": 1.992763540723244,
      "learning_rate": 3.5271960480208077e-06,
      "loss": 0.722,
      "step": 2736
    },
    {
      "epoch": 0.24683230373810705,
      "grad_norm": 0.6906701819713651,
      "learning_rate": 3.526818762788534e-06,
      "loss": 0.629,
      "step": 2737
    },
    {
      "epoch": 0.2469224872615773,
      "grad_norm": 1.637462001097568,
      "learning_rate": 3.5264413472796653e-06,
      "loss": 0.8227,
      "step": 2738
    },
    {
      "epoch": 0.24701267078504757,
      "grad_norm": 2.063100091737042,
      "learning_rate": 3.5260638015264037e-06,
      "loss": 0.7687,
      "step": 2739
    },
    {
      "epoch": 0.24710285430851783,
      "grad_norm": 2.510825655427781,
      "learning_rate": 3.5256861255609644e-06,
      "loss": 0.7542,
      "step": 2740
    },
    {
      "epoch": 0.2471930378319881,
      "grad_norm": 1.7071664438026255,
      "learning_rate": 3.5253083194155723e-06,
      "loss": 0.7961,
      "step": 2741
    },
    {
      "epoch": 0.24728322135545835,
      "grad_norm": 1.4709735839515363,
      "learning_rate": 3.5249303831224637e-06,
      "loss": 0.8081,
      "step": 2742
    },
    {
      "epoch": 0.24737340487892862,
      "grad_norm": 1.779334661457304,
      "learning_rate": 3.524552316713887e-06,
      "loss": 0.7747,
      "step": 2743
    },
    {
      "epoch": 0.24746358840239888,
      "grad_norm": 1.857762318567865,
      "learning_rate": 3.5241741202220995e-06,
      "loss": 0.8497,
      "step": 2744
    },
    {
      "epoch": 0.24755377192586914,
      "grad_norm": 1.62017663958126,
      "learning_rate": 3.5237957936793724e-06,
      "loss": 0.7731,
      "step": 2745
    },
    {
      "epoch": 0.2476439554493394,
      "grad_norm": 1.5622466835674136,
      "learning_rate": 3.523417337117986e-06,
      "loss": 0.7467,
      "step": 2746
    },
    {
      "epoch": 0.24773413897280966,
      "grad_norm": 1.6814420540745159,
      "learning_rate": 3.523038750570232e-06,
      "loss": 0.7333,
      "step": 2747
    },
    {
      "epoch": 0.24782432249627992,
      "grad_norm": 1.6434720721818263,
      "learning_rate": 3.522660034068414e-06,
      "loss": 0.8152,
      "step": 2748
    },
    {
      "epoch": 0.24791450601975018,
      "grad_norm": 1.787604658357462,
      "learning_rate": 3.5222811876448464e-06,
      "loss": 0.8019,
      "step": 2749
    },
    {
      "epoch": 0.24800468954322044,
      "grad_norm": 1.3013278397026604,
      "learning_rate": 3.521902211331854e-06,
      "loss": 0.802,
      "step": 2750
    },
    {
      "epoch": 0.2480948730666907,
      "grad_norm": 1.5395527944603076,
      "learning_rate": 3.5215231051617726e-06,
      "loss": 0.7352,
      "step": 2751
    },
    {
      "epoch": 0.248185056590161,
      "grad_norm": 1.7927331042383934,
      "learning_rate": 3.521143869166951e-06,
      "loss": 0.6878,
      "step": 2752
    },
    {
      "epoch": 0.24827524011363125,
      "grad_norm": 2.142602160587608,
      "learning_rate": 3.5207645033797464e-06,
      "loss": 0.8223,
      "step": 2753
    },
    {
      "epoch": 0.2483654236371015,
      "grad_norm": 1.7625397589778051,
      "learning_rate": 3.5203850078325293e-06,
      "loss": 0.7124,
      "step": 2754
    },
    {
      "epoch": 0.24845560716057177,
      "grad_norm": 1.7597845925379019,
      "learning_rate": 3.5200053825576797e-06,
      "loss": 0.7058,
      "step": 2755
    },
    {
      "epoch": 0.24854579068404203,
      "grad_norm": 4.740357298764122,
      "learning_rate": 3.51962562758759e-06,
      "loss": 0.7867,
      "step": 2756
    },
    {
      "epoch": 0.2486359742075123,
      "grad_norm": 1.6582598457475466,
      "learning_rate": 3.5192457429546627e-06,
      "loss": 0.7563,
      "step": 2757
    },
    {
      "epoch": 0.24872615773098256,
      "grad_norm": 1.6656447933659915,
      "learning_rate": 3.5188657286913115e-06,
      "loss": 0.7536,
      "step": 2758
    },
    {
      "epoch": 0.24881634125445282,
      "grad_norm": 2.174595617651999,
      "learning_rate": 3.518485584829961e-06,
      "loss": 0.7484,
      "step": 2759
    },
    {
      "epoch": 0.24890652477792308,
      "grad_norm": 1.6017850395598543,
      "learning_rate": 3.5181053114030485e-06,
      "loss": 0.7952,
      "step": 2760
    },
    {
      "epoch": 0.24899670830139334,
      "grad_norm": 1.5651910115635865,
      "learning_rate": 3.5177249084430198e-06,
      "loss": 0.7808,
      "step": 2761
    },
    {
      "epoch": 0.2490868918248636,
      "grad_norm": 1.8621504129121358,
      "learning_rate": 3.517344375982333e-06,
      "loss": 0.8109,
      "step": 2762
    },
    {
      "epoch": 0.24917707534833386,
      "grad_norm": 1.8246937037494955,
      "learning_rate": 3.5169637140534565e-06,
      "loss": 0.8179,
      "step": 2763
    },
    {
      "epoch": 0.24926725887180412,
      "grad_norm": 1.76930952893572,
      "learning_rate": 3.5165829226888733e-06,
      "loss": 0.7506,
      "step": 2764
    },
    {
      "epoch": 0.24935744239527438,
      "grad_norm": 3.7923724317627148,
      "learning_rate": 3.516202001921072e-06,
      "loss": 0.81,
      "step": 2765
    },
    {
      "epoch": 0.24944762591874464,
      "grad_norm": 1.5957395813231454,
      "learning_rate": 3.515820951782555e-06,
      "loss": 0.8091,
      "step": 2766
    },
    {
      "epoch": 0.2495378094422149,
      "grad_norm": 2.149313675209968,
      "learning_rate": 3.5154397723058366e-06,
      "loss": 0.7185,
      "step": 2767
    },
    {
      "epoch": 0.24962799296568516,
      "grad_norm": 1.6603683515704364,
      "learning_rate": 3.5150584635234416e-06,
      "loss": 0.8432,
      "step": 2768
    },
    {
      "epoch": 0.24971817648915542,
      "grad_norm": 1.3693099196136913,
      "learning_rate": 3.5146770254679035e-06,
      "loss": 0.8239,
      "step": 2769
    },
    {
      "epoch": 0.24980836001262569,
      "grad_norm": 1.6226930381781186,
      "learning_rate": 3.51429545817177e-06,
      "loss": 0.703,
      "step": 2770
    },
    {
      "epoch": 0.24989854353609595,
      "grad_norm": 1.8028198750927444,
      "learning_rate": 3.5139137616675985e-06,
      "loss": 0.7942,
      "step": 2771
    },
    {
      "epoch": 0.2499887270595662,
      "grad_norm": 1.6403728472989052,
      "learning_rate": 3.513531935987957e-06,
      "loss": 0.7259,
      "step": 2772
    },
    {
      "epoch": 0.2500789105830365,
      "grad_norm": 2.188942896579553,
      "learning_rate": 3.5131499811654253e-06,
      "loss": 0.8022,
      "step": 2773
    },
    {
      "epoch": 0.25016909410650673,
      "grad_norm": 1.5184223276822089,
      "learning_rate": 3.512767897232594e-06,
      "loss": 0.7469,
      "step": 2774
    },
    {
      "epoch": 0.250259277629977,
      "grad_norm": 1.8736706789823547,
      "learning_rate": 3.512385684222064e-06,
      "loss": 0.7772,
      "step": 2775
    },
    {
      "epoch": 0.25034946115344725,
      "grad_norm": 1.8017646272525256,
      "learning_rate": 3.512003342166449e-06,
      "loss": 0.8773,
      "step": 2776
    },
    {
      "epoch": 0.25043964467691754,
      "grad_norm": 1.6192155735537241,
      "learning_rate": 3.511620871098371e-06,
      "loss": 0.8183,
      "step": 2777
    },
    {
      "epoch": 0.25052982820038777,
      "grad_norm": 1.6385049837261882,
      "learning_rate": 3.511238271050465e-06,
      "loss": 0.815,
      "step": 2778
    },
    {
      "epoch": 0.25062001172385806,
      "grad_norm": 1.9806468514861815,
      "learning_rate": 3.5108555420553778e-06,
      "loss": 0.6974,
      "step": 2779
    },
    {
      "epoch": 0.2507101952473283,
      "grad_norm": 1.7316736026707007,
      "learning_rate": 3.510472684145764e-06,
      "loss": 0.6023,
      "step": 2780
    },
    {
      "epoch": 0.2508003787707986,
      "grad_norm": 1.6539820131617728,
      "learning_rate": 3.5100896973542926e-06,
      "loss": 0.6746,
      "step": 2781
    },
    {
      "epoch": 0.2508905622942688,
      "grad_norm": 2.0385493659600336,
      "learning_rate": 3.509706581713642e-06,
      "loss": 0.8222,
      "step": 2782
    },
    {
      "epoch": 0.2509807458177391,
      "grad_norm": 1.6010242581170797,
      "learning_rate": 3.509323337256501e-06,
      "loss": 0.808,
      "step": 2783
    },
    {
      "epoch": 0.25107092934120934,
      "grad_norm": 1.736134023727113,
      "learning_rate": 3.5089399640155703e-06,
      "loss": 0.7338,
      "step": 2784
    },
    {
      "epoch": 0.2511611128646796,
      "grad_norm": 1.5255744227457202,
      "learning_rate": 3.508556462023562e-06,
      "loss": 0.7743,
      "step": 2785
    },
    {
      "epoch": 0.25125129638814986,
      "grad_norm": 2.4750867231183613,
      "learning_rate": 3.5081728313131984e-06,
      "loss": 0.7666,
      "step": 2786
    },
    {
      "epoch": 0.25134147991162015,
      "grad_norm": 3.4344563274108726,
      "learning_rate": 3.5077890719172125e-06,
      "loss": 0.7794,
      "step": 2787
    },
    {
      "epoch": 0.25143166343509044,
      "grad_norm": 1.6499747389969348,
      "learning_rate": 3.5074051838683497e-06,
      "loss": 0.7854,
      "step": 2788
    },
    {
      "epoch": 0.25152184695856067,
      "grad_norm": 1.8825314260547201,
      "learning_rate": 3.5070211671993643e-06,
      "loss": 0.7054,
      "step": 2789
    },
    {
      "epoch": 0.25161203048203096,
      "grad_norm": 2.456068471941934,
      "learning_rate": 3.5066370219430238e-06,
      "loss": 0.8035,
      "step": 2790
    },
    {
      "epoch": 0.2517022140055012,
      "grad_norm": 1.5614330463597086,
      "learning_rate": 3.5062527481321044e-06,
      "loss": 0.7811,
      "step": 2791
    },
    {
      "epoch": 0.2517923975289715,
      "grad_norm": 1.98865150346814,
      "learning_rate": 3.5058683457993954e-06,
      "loss": 0.7969,
      "step": 2792
    },
    {
      "epoch": 0.2518825810524417,
      "grad_norm": 0.7782945808168066,
      "learning_rate": 3.5054838149776963e-06,
      "loss": 0.5368,
      "step": 2793
    },
    {
      "epoch": 0.251972764575912,
      "grad_norm": 1.5475752045286442,
      "learning_rate": 3.505099155699816e-06,
      "loss": 0.8192,
      "step": 2794
    },
    {
      "epoch": 0.25206294809938223,
      "grad_norm": 1.6366477350019273,
      "learning_rate": 3.5047143679985775e-06,
      "loss": 0.6609,
      "step": 2795
    },
    {
      "epoch": 0.2521531316228525,
      "grad_norm": 1.5067872826365891,
      "learning_rate": 3.5043294519068126e-06,
      "loss": 0.7966,
      "step": 2796
    },
    {
      "epoch": 0.25224331514632276,
      "grad_norm": 1.9897586152788524,
      "learning_rate": 3.503944407457363e-06,
      "loss": 0.7065,
      "step": 2797
    },
    {
      "epoch": 0.25233349866979304,
      "grad_norm": 1.7535785462360145,
      "learning_rate": 3.5035592346830846e-06,
      "loss": 0.797,
      "step": 2798
    },
    {
      "epoch": 0.2524236821932633,
      "grad_norm": 1.9214194773652777,
      "learning_rate": 3.503173933616841e-06,
      "loss": 0.8516,
      "step": 2799
    },
    {
      "epoch": 0.25251386571673357,
      "grad_norm": 1.6211386800767211,
      "learning_rate": 3.50278850429151e-06,
      "loss": 0.8566,
      "step": 2800
    },
    {
      "epoch": 0.2526040492402038,
      "grad_norm": 1.6607149507957804,
      "learning_rate": 3.502402946739977e-06,
      "loss": 0.8047,
      "step": 2801
    },
    {
      "epoch": 0.2526942327636741,
      "grad_norm": 1.739497070305647,
      "learning_rate": 3.5020172609951405e-06,
      "loss": 0.8015,
      "step": 2802
    },
    {
      "epoch": 0.2527844162871443,
      "grad_norm": 1.4848652080409235,
      "learning_rate": 3.501631447089909e-06,
      "loss": 0.7978,
      "step": 2803
    },
    {
      "epoch": 0.2528745998106146,
      "grad_norm": 2.0515751346568054,
      "learning_rate": 3.501245505057203e-06,
      "loss": 0.7584,
      "step": 2804
    },
    {
      "epoch": 0.25296478333408484,
      "grad_norm": 1.7600820174397056,
      "learning_rate": 3.5008594349299526e-06,
      "loss": 0.8568,
      "step": 2805
    },
    {
      "epoch": 0.25305496685755513,
      "grad_norm": 1.6754899916707116,
      "learning_rate": 3.500473236741099e-06,
      "loss": 0.7672,
      "step": 2806
    },
    {
      "epoch": 0.25314515038102536,
      "grad_norm": 1.495526637068085,
      "learning_rate": 3.500086910523596e-06,
      "loss": 0.7417,
      "step": 2807
    },
    {
      "epoch": 0.25323533390449565,
      "grad_norm": 1.6097855275809134,
      "learning_rate": 3.499700456310406e-06,
      "loss": 0.7302,
      "step": 2808
    },
    {
      "epoch": 0.2533255174279659,
      "grad_norm": 2.2167541768671506,
      "learning_rate": 3.499313874134504e-06,
      "loss": 0.8319,
      "step": 2809
    },
    {
      "epoch": 0.2534157009514362,
      "grad_norm": 2.1633446118962545,
      "learning_rate": 3.498927164028875e-06,
      "loss": 0.6543,
      "step": 2810
    },
    {
      "epoch": 0.25350588447490646,
      "grad_norm": 1.3127741754354811,
      "learning_rate": 3.498540326026515e-06,
      "loss": 0.8171,
      "step": 2811
    },
    {
      "epoch": 0.2535960679983767,
      "grad_norm": 1.5674822148921805,
      "learning_rate": 3.4981533601604323e-06,
      "loss": 0.8525,
      "step": 2812
    },
    {
      "epoch": 0.253686251521847,
      "grad_norm": 1.7593048542464638,
      "learning_rate": 3.4977662664636443e-06,
      "loss": 0.7942,
      "step": 2813
    },
    {
      "epoch": 0.2537764350453172,
      "grad_norm": 1.6888485923917755,
      "learning_rate": 3.497379044969179e-06,
      "loss": 0.7431,
      "step": 2814
    },
    {
      "epoch": 0.2538666185687875,
      "grad_norm": 1.6847399286211668,
      "learning_rate": 3.4969916957100777e-06,
      "loss": 0.776,
      "step": 2815
    },
    {
      "epoch": 0.25395680209225774,
      "grad_norm": 1.4922655494609638,
      "learning_rate": 3.4966042187193905e-06,
      "loss": 0.7748,
      "step": 2816
    },
    {
      "epoch": 0.254046985615728,
      "grad_norm": 1.4409379069009842,
      "learning_rate": 3.496216614030179e-06,
      "loss": 0.7782,
      "step": 2817
    },
    {
      "epoch": 0.25413716913919826,
      "grad_norm": 1.598753461438832,
      "learning_rate": 3.495828881675516e-06,
      "loss": 0.6808,
      "step": 2818
    },
    {
      "epoch": 0.25422735266266855,
      "grad_norm": 6.788710186665443,
      "learning_rate": 3.4954410216884845e-06,
      "loss": 0.8511,
      "step": 2819
    },
    {
      "epoch": 0.2543175361861388,
      "grad_norm": 1.6711133844293076,
      "learning_rate": 3.49505303410218e-06,
      "loss": 0.793,
      "step": 2820
    },
    {
      "epoch": 0.25440771970960907,
      "grad_norm": 3.724896008364841,
      "learning_rate": 3.4946649189497067e-06,
      "loss": 0.8094,
      "step": 2821
    },
    {
      "epoch": 0.2544979032330793,
      "grad_norm": 19.58305551318338,
      "learning_rate": 3.4942766762641805e-06,
      "loss": 0.7743,
      "step": 2822
    },
    {
      "epoch": 0.2545880867565496,
      "grad_norm": 2.025445479383144,
      "learning_rate": 3.49388830607873e-06,
      "loss": 0.8118,
      "step": 2823
    },
    {
      "epoch": 0.2546782702800198,
      "grad_norm": 1.2732861200703645,
      "learning_rate": 3.493499808426491e-06,
      "loss": 0.748,
      "step": 2824
    },
    {
      "epoch": 0.2547684538034901,
      "grad_norm": 1.8112744101070097,
      "learning_rate": 3.493111183340614e-06,
      "loss": 0.8328,
      "step": 2825
    },
    {
      "epoch": 0.25485863732696035,
      "grad_norm": 1.5913551887931254,
      "learning_rate": 3.4927224308542576e-06,
      "loss": 0.6825,
      "step": 2826
    },
    {
      "epoch": 0.25494882085043064,
      "grad_norm": 1.5628758550631057,
      "learning_rate": 3.4923335510005923e-06,
      "loss": 0.8623,
      "step": 2827
    },
    {
      "epoch": 0.25503900437390087,
      "grad_norm": 1.5322513809285014,
      "learning_rate": 3.4919445438128e-06,
      "loss": 0.8026,
      "step": 2828
    },
    {
      "epoch": 0.25512918789737116,
      "grad_norm": 1.4550974594982233,
      "learning_rate": 3.491555409324073e-06,
      "loss": 0.8617,
      "step": 2829
    },
    {
      "epoch": 0.2552193714208414,
      "grad_norm": 0.6677975128377142,
      "learning_rate": 3.4911661475676136e-06,
      "loss": 0.5608,
      "step": 2830
    },
    {
      "epoch": 0.2553095549443117,
      "grad_norm": 1.5265856175931984,
      "learning_rate": 3.490776758576637e-06,
      "loss": 0.7575,
      "step": 2831
    },
    {
      "epoch": 0.2553997384677819,
      "grad_norm": 2.9740689463624217,
      "learning_rate": 3.4903872423843668e-06,
      "loss": 0.8303,
      "step": 2832
    },
    {
      "epoch": 0.2554899219912522,
      "grad_norm": 1.6706439325506457,
      "learning_rate": 3.4899975990240396e-06,
      "loss": 0.7472,
      "step": 2833
    },
    {
      "epoch": 0.25558010551472243,
      "grad_norm": 2.0299992158258737,
      "learning_rate": 3.489607828528901e-06,
      "loss": 0.8057,
      "step": 2834
    },
    {
      "epoch": 0.2556702890381927,
      "grad_norm": 0.6616549437474907,
      "learning_rate": 3.4892179309322093e-06,
      "loss": 0.6698,
      "step": 2835
    },
    {
      "epoch": 0.255760472561663,
      "grad_norm": 1.6484926765934098,
      "learning_rate": 3.488827906267232e-06,
      "loss": 0.6939,
      "step": 2836
    },
    {
      "epoch": 0.25585065608513324,
      "grad_norm": 1.5931096005808547,
      "learning_rate": 3.4884377545672485e-06,
      "loss": 0.8194,
      "step": 2837
    },
    {
      "epoch": 0.25594083960860353,
      "grad_norm": 1.4786276932325093,
      "learning_rate": 3.4880474758655485e-06,
      "loss": 0.7566,
      "step": 2838
    },
    {
      "epoch": 0.25603102313207377,
      "grad_norm": 1.599564585448756,
      "learning_rate": 3.487657070195433e-06,
      "loss": 0.8491,
      "step": 2839
    },
    {
      "epoch": 0.25612120665554405,
      "grad_norm": 1.5947977249890881,
      "learning_rate": 3.487266537590213e-06,
      "loss": 0.857,
      "step": 2840
    },
    {
      "epoch": 0.2562113901790143,
      "grad_norm": 1.495781529768349,
      "learning_rate": 3.4868758780832116e-06,
      "loss": 0.6981,
      "step": 2841
    },
    {
      "epoch": 0.2563015737024846,
      "grad_norm": 1.7733837237166041,
      "learning_rate": 3.486485091707762e-06,
      "loss": 0.793,
      "step": 2842
    },
    {
      "epoch": 0.2563917572259548,
      "grad_norm": 1.812134475682181,
      "learning_rate": 3.4860941784972077e-06,
      "loss": 0.6603,
      "step": 2843
    },
    {
      "epoch": 0.2564819407494251,
      "grad_norm": 2.63202789116201,
      "learning_rate": 3.485703138484904e-06,
      "loss": 0.8779,
      "step": 2844
    },
    {
      "epoch": 0.25657212427289533,
      "grad_norm": 1.7330921686957974,
      "learning_rate": 3.485311971704216e-06,
      "loss": 0.8311,
      "step": 2845
    },
    {
      "epoch": 0.2566623077963656,
      "grad_norm": 1.4717707311544852,
      "learning_rate": 3.484920678188521e-06,
      "loss": 0.8144,
      "step": 2846
    },
    {
      "epoch": 0.25675249131983585,
      "grad_norm": 2.023665485278477,
      "learning_rate": 3.4845292579712063e-06,
      "loss": 0.8056,
      "step": 2847
    },
    {
      "epoch": 0.25684267484330614,
      "grad_norm": 2.3691041948327474,
      "learning_rate": 3.484137711085669e-06,
      "loss": 0.6647,
      "step": 2848
    },
    {
      "epoch": 0.2569328583667764,
      "grad_norm": 2.3731291580063303,
      "learning_rate": 3.4837460375653198e-06,
      "loss": 0.7686,
      "step": 2849
    },
    {
      "epoch": 0.25702304189024666,
      "grad_norm": 3.218370730082226,
      "learning_rate": 3.483354237443576e-06,
      "loss": 0.8099,
      "step": 2850
    },
    {
      "epoch": 0.2571132254137169,
      "grad_norm": 1.465773955952114,
      "learning_rate": 3.48296231075387e-06,
      "loss": 0.7248,
      "step": 2851
    },
    {
      "epoch": 0.2572034089371872,
      "grad_norm": 1.5571086184835852,
      "learning_rate": 3.4825702575296433e-06,
      "loss": 0.7992,
      "step": 2852
    },
    {
      "epoch": 0.2572935924606574,
      "grad_norm": 0.6914814191571244,
      "learning_rate": 3.482178077804347e-06,
      "loss": 0.6136,
      "step": 2853
    },
    {
      "epoch": 0.2573837759841277,
      "grad_norm": 1.8440519829123536,
      "learning_rate": 3.4817857716114443e-06,
      "loss": 0.7482,
      "step": 2854
    },
    {
      "epoch": 0.25747395950759794,
      "grad_norm": 3.0815204927207778,
      "learning_rate": 3.4813933389844094e-06,
      "loss": 0.7376,
      "step": 2855
    },
    {
      "epoch": 0.2575641430310682,
      "grad_norm": 1.739230738862311,
      "learning_rate": 3.4810007799567264e-06,
      "loss": 0.7911,
      "step": 2856
    },
    {
      "epoch": 0.25765432655453846,
      "grad_norm": 2.0218195393191634,
      "learning_rate": 3.480608094561891e-06,
      "loss": 0.7966,
      "step": 2857
    },
    {
      "epoch": 0.25774451007800875,
      "grad_norm": 1.6572287204887983,
      "learning_rate": 3.4802152828334083e-06,
      "loss": 0.7992,
      "step": 2858
    },
    {
      "epoch": 0.25783469360147904,
      "grad_norm": 2.3985099657275932,
      "learning_rate": 3.479822344804796e-06,
      "loss": 0.7252,
      "step": 2859
    },
    {
      "epoch": 0.25792487712494927,
      "grad_norm": 1.7879007890515315,
      "learning_rate": 3.479429280509582e-06,
      "loss": 0.8405,
      "step": 2860
    },
    {
      "epoch": 0.25801506064841956,
      "grad_norm": 1.9432954557248794,
      "learning_rate": 3.4790360899813038e-06,
      "loss": 0.8269,
      "step": 2861
    },
    {
      "epoch": 0.2581052441718898,
      "grad_norm": 2.0077424389679344,
      "learning_rate": 3.4786427732535115e-06,
      "loss": 0.7175,
      "step": 2862
    },
    {
      "epoch": 0.2581954276953601,
      "grad_norm": 2.1862456949059665,
      "learning_rate": 3.478249330359764e-06,
      "loss": 0.8235,
      "step": 2863
    },
    {
      "epoch": 0.2582856112188303,
      "grad_norm": 1.539692860975895,
      "learning_rate": 3.4778557613336333e-06,
      "loss": 0.713,
      "step": 2864
    },
    {
      "epoch": 0.2583757947423006,
      "grad_norm": 1.786889701641349,
      "learning_rate": 3.4774620662087004e-06,
      "loss": 0.7986,
      "step": 2865
    },
    {
      "epoch": 0.25846597826577083,
      "grad_norm": 2.063610124936806,
      "learning_rate": 3.477068245018557e-06,
      "loss": 0.7823,
      "step": 2866
    },
    {
      "epoch": 0.2585561617892411,
      "grad_norm": 1.6836692987825619,
      "learning_rate": 3.476674297796807e-06,
      "loss": 0.7867,
      "step": 2867
    },
    {
      "epoch": 0.25864634531271136,
      "grad_norm": 1.7662695324378614,
      "learning_rate": 3.4762802245770627e-06,
      "loss": 0.8239,
      "step": 2868
    },
    {
      "epoch": 0.25873652883618165,
      "grad_norm": 1.5216237544220195,
      "learning_rate": 3.4758860253929497e-06,
      "loss": 0.7285,
      "step": 2869
    },
    {
      "epoch": 0.2588267123596519,
      "grad_norm": 1.6549904550467283,
      "learning_rate": 3.4754917002781038e-06,
      "loss": 0.804,
      "step": 2870
    },
    {
      "epoch": 0.25891689588312217,
      "grad_norm": 2.956395835585946,
      "learning_rate": 3.475097249266169e-06,
      "loss": 0.8233,
      "step": 2871
    },
    {
      "epoch": 0.2590070794065924,
      "grad_norm": 1.648183721741051,
      "learning_rate": 3.4747026723908044e-06,
      "loss": 0.7681,
      "step": 2872
    },
    {
      "epoch": 0.2590972629300627,
      "grad_norm": 1.6213501509157082,
      "learning_rate": 3.474307969685676e-06,
      "loss": 0.7637,
      "step": 2873
    },
    {
      "epoch": 0.2591874464535329,
      "grad_norm": 1.5399725687382977,
      "learning_rate": 3.473913141184462e-06,
      "loss": 0.7332,
      "step": 2874
    },
    {
      "epoch": 0.2592776299770032,
      "grad_norm": 1.5855253087973828,
      "learning_rate": 3.4735181869208523e-06,
      "loss": 0.6548,
      "step": 2875
    },
    {
      "epoch": 0.25936781350047344,
      "grad_norm": 2.3280623254724473,
      "learning_rate": 3.473123106928546e-06,
      "loss": 0.6582,
      "step": 2876
    },
    {
      "epoch": 0.25945799702394373,
      "grad_norm": 1.86331509413462,
      "learning_rate": 3.4727279012412533e-06,
      "loss": 0.7399,
      "step": 2877
    },
    {
      "epoch": 0.25954818054741396,
      "grad_norm": 0.6337612951276601,
      "learning_rate": 3.4723325698926953e-06,
      "loss": 0.5761,
      "step": 2878
    },
    {
      "epoch": 0.25963836407088425,
      "grad_norm": 1.6060431350910447,
      "learning_rate": 3.4719371129166045e-06,
      "loss": 0.7855,
      "step": 2879
    },
    {
      "epoch": 0.2597285475943545,
      "grad_norm": 1.746100918543905,
      "learning_rate": 3.471541530346723e-06,
      "loss": 0.7058,
      "step": 2880
    },
    {
      "epoch": 0.2598187311178248,
      "grad_norm": 1.8983262896832729,
      "learning_rate": 3.4711458222168037e-06,
      "loss": 0.8489,
      "step": 2881
    },
    {
      "epoch": 0.259908914641295,
      "grad_norm": 1.5521642249957996,
      "learning_rate": 3.4707499885606114e-06,
      "loss": 0.8233,
      "step": 2882
    },
    {
      "epoch": 0.2599990981647653,
      "grad_norm": 1.6090985403286502,
      "learning_rate": 3.4703540294119204e-06,
      "loss": 0.8162,
      "step": 2883
    },
    {
      "epoch": 0.2600892816882356,
      "grad_norm": 1.876680002972571,
      "learning_rate": 3.4699579448045163e-06,
      "loss": 0.8397,
      "step": 2884
    },
    {
      "epoch": 0.2601794652117058,
      "grad_norm": 1.4821256761326567,
      "learning_rate": 3.4695617347721947e-06,
      "loss": 0.6692,
      "step": 2885
    },
    {
      "epoch": 0.2602696487351761,
      "grad_norm": 1.7795241176238883,
      "learning_rate": 3.469165399348763e-06,
      "loss": 0.7524,
      "step": 2886
    },
    {
      "epoch": 0.26035983225864634,
      "grad_norm": 1.946280365047363,
      "learning_rate": 3.4687689385680384e-06,
      "loss": 0.7665,
      "step": 2887
    },
    {
      "epoch": 0.26045001578211663,
      "grad_norm": 1.6235642325873763,
      "learning_rate": 3.4683723524638494e-06,
      "loss": 0.7636,
      "step": 2888
    },
    {
      "epoch": 0.26054019930558686,
      "grad_norm": 1.7024534808958094,
      "learning_rate": 3.4679756410700354e-06,
      "loss": 0.8499,
      "step": 2889
    },
    {
      "epoch": 0.26063038282905715,
      "grad_norm": 2.131799925614869,
      "learning_rate": 3.4675788044204445e-06,
      "loss": 0.7858,
      "step": 2890
    },
    {
      "epoch": 0.2607205663525274,
      "grad_norm": 1.5104880042055941,
      "learning_rate": 3.467181842548938e-06,
      "loss": 0.7128,
      "step": 2891
    },
    {
      "epoch": 0.26081074987599767,
      "grad_norm": 1.6134553311988298,
      "learning_rate": 3.466784755489387e-06,
      "loss": 0.7819,
      "step": 2892
    },
    {
      "epoch": 0.2609009333994679,
      "grad_norm": 1.5790443386450497,
      "learning_rate": 3.4663875432756726e-06,
      "loss": 0.8997,
      "step": 2893
    },
    {
      "epoch": 0.2609911169229382,
      "grad_norm": 3.1452960298640438,
      "learning_rate": 3.465990205941687e-06,
      "loss": 0.9555,
      "step": 2894
    },
    {
      "epoch": 0.2610813004464084,
      "grad_norm": 3.41570586364459,
      "learning_rate": 3.465592743521335e-06,
      "loss": 0.7665,
      "step": 2895
    },
    {
      "epoch": 0.2611714839698787,
      "grad_norm": 1.8121712485142376,
      "learning_rate": 3.465195156048528e-06,
      "loss": 0.8328,
      "step": 2896
    },
    {
      "epoch": 0.26126166749334895,
      "grad_norm": 0.5703837990283788,
      "learning_rate": 3.464797443557191e-06,
      "loss": 0.574,
      "step": 2897
    },
    {
      "epoch": 0.26135185101681924,
      "grad_norm": 2.0802891297393655,
      "learning_rate": 3.46439960608126e-06,
      "loss": 0.7572,
      "step": 2898
    },
    {
      "epoch": 0.26144203454028947,
      "grad_norm": 2.1616058176731796,
      "learning_rate": 3.4640016436546797e-06,
      "loss": 0.7269,
      "step": 2899
    },
    {
      "epoch": 0.26153221806375976,
      "grad_norm": 1.78431758754849,
      "learning_rate": 3.4636035563114065e-06,
      "loss": 0.7615,
      "step": 2900
    },
    {
      "epoch": 0.26162240158723,
      "grad_norm": 1.5764111570035568,
      "learning_rate": 3.4632053440854085e-06,
      "loss": 0.7103,
      "step": 2901
    },
    {
      "epoch": 0.2617125851107003,
      "grad_norm": 1.802607539086621,
      "learning_rate": 3.462807007010662e-06,
      "loss": 0.8377,
      "step": 2902
    },
    {
      "epoch": 0.2618027686341705,
      "grad_norm": 2.299090815259968,
      "learning_rate": 3.462408545121155e-06,
      "loss": 0.8119,
      "step": 2903
    },
    {
      "epoch": 0.2618929521576408,
      "grad_norm": 4.501049978891102,
      "learning_rate": 3.4620099584508883e-06,
      "loss": 0.8422,
      "step": 2904
    },
    {
      "epoch": 0.26198313568111103,
      "grad_norm": 1.7732157967749929,
      "learning_rate": 3.46161124703387e-06,
      "loss": 0.7423,
      "step": 2905
    },
    {
      "epoch": 0.2620733192045813,
      "grad_norm": 2.417628053194902,
      "learning_rate": 3.461212410904122e-06,
      "loss": 0.7847,
      "step": 2906
    },
    {
      "epoch": 0.2621635027280516,
      "grad_norm": 1.6143231664260382,
      "learning_rate": 3.4608134500956726e-06,
      "loss": 0.8422,
      "step": 2907
    },
    {
      "epoch": 0.26225368625152184,
      "grad_norm": 1.5722813954954418,
      "learning_rate": 3.4604143646425655e-06,
      "loss": 0.7698,
      "step": 2908
    },
    {
      "epoch": 0.26234386977499213,
      "grad_norm": 1.5457544603130728,
      "learning_rate": 3.460015154578852e-06,
      "loss": 0.8513,
      "step": 2909
    },
    {
      "epoch": 0.26243405329846237,
      "grad_norm": 1.6627115846382334,
      "learning_rate": 3.459615819938595e-06,
      "loss": 0.7567,
      "step": 2910
    },
    {
      "epoch": 0.26252423682193265,
      "grad_norm": 1.6624916793500295,
      "learning_rate": 3.4592163607558684e-06,
      "loss": 0.7502,
      "step": 2911
    },
    {
      "epoch": 0.2626144203454029,
      "grad_norm": 1.3518075886151746,
      "learning_rate": 3.4588167770647553e-06,
      "loss": 0.7845,
      "step": 2912
    },
    {
      "epoch": 0.2627046038688732,
      "grad_norm": 1.735185657099165,
      "learning_rate": 3.458417068899351e-06,
      "loss": 0.8724,
      "step": 2913
    },
    {
      "epoch": 0.2627947873923434,
      "grad_norm": 1.4337645384792221,
      "learning_rate": 3.4580172362937612e-06,
      "loss": 0.7511,
      "step": 2914
    },
    {
      "epoch": 0.2628849709158137,
      "grad_norm": 1.2916255093241906,
      "learning_rate": 3.457617279282101e-06,
      "loss": 0.746,
      "step": 2915
    },
    {
      "epoch": 0.26297515443928393,
      "grad_norm": 2.3398029916107856,
      "learning_rate": 3.4572171978984975e-06,
      "loss": 0.763,
      "step": 2916
    },
    {
      "epoch": 0.2630653379627542,
      "grad_norm": 1.5497415357870261,
      "learning_rate": 3.456816992177088e-06,
      "loss": 0.7847,
      "step": 2917
    },
    {
      "epoch": 0.26315552148622445,
      "grad_norm": 1.3756818814543543,
      "learning_rate": 3.4564166621520193e-06,
      "loss": 0.7515,
      "step": 2918
    },
    {
      "epoch": 0.26324570500969474,
      "grad_norm": 1.467333171141386,
      "learning_rate": 3.4560162078574507e-06,
      "loss": 0.7765,
      "step": 2919
    },
    {
      "epoch": 0.263335888533165,
      "grad_norm": 1.639530070828045,
      "learning_rate": 3.455615629327551e-06,
      "loss": 0.7849,
      "step": 2920
    },
    {
      "epoch": 0.26342607205663526,
      "grad_norm": 1.8942525442347231,
      "learning_rate": 3.4552149265964994e-06,
      "loss": 0.7635,
      "step": 2921
    },
    {
      "epoch": 0.2635162555801055,
      "grad_norm": 1.6114304388681855,
      "learning_rate": 3.4548140996984866e-06,
      "loss": 0.7299,
      "step": 2922
    },
    {
      "epoch": 0.2636064391035758,
      "grad_norm": 1.5877878453773666,
      "learning_rate": 3.4544131486677124e-06,
      "loss": 0.8069,
      "step": 2923
    },
    {
      "epoch": 0.263696622627046,
      "grad_norm": 1.5165334908560177,
      "learning_rate": 3.454012073538389e-06,
      "loss": 0.7208,
      "step": 2924
    },
    {
      "epoch": 0.2637868061505163,
      "grad_norm": 1.4548197062368344,
      "learning_rate": 3.453610874344738e-06,
      "loss": 0.7921,
      "step": 2925
    },
    {
      "epoch": 0.26387698967398654,
      "grad_norm": 1.6897279372543577,
      "learning_rate": 3.453209551120993e-06,
      "loss": 0.8246,
      "step": 2926
    },
    {
      "epoch": 0.26396717319745683,
      "grad_norm": 2.888907257249912,
      "learning_rate": 3.452808103901395e-06,
      "loss": 0.756,
      "step": 2927
    },
    {
      "epoch": 0.26405735672092706,
      "grad_norm": 1.9411892990378596,
      "learning_rate": 3.4524065327202e-06,
      "loss": 0.8563,
      "step": 2928
    },
    {
      "epoch": 0.26414754024439735,
      "grad_norm": 0.6494796834503622,
      "learning_rate": 3.4520048376116702e-06,
      "loss": 0.5261,
      "step": 2929
    },
    {
      "epoch": 0.26423772376786764,
      "grad_norm": 1.6947696361583084,
      "learning_rate": 3.4516030186100817e-06,
      "loss": 0.8216,
      "step": 2930
    },
    {
      "epoch": 0.26432790729133787,
      "grad_norm": 1.3666286525246598,
      "learning_rate": 3.4512010757497197e-06,
      "loss": 0.696,
      "step": 2931
    },
    {
      "epoch": 0.26441809081480816,
      "grad_norm": 1.7535927541200227,
      "learning_rate": 3.4507990090648804e-06,
      "loss": 0.6991,
      "step": 2932
    },
    {
      "epoch": 0.2645082743382784,
      "grad_norm": 1.7223483478685837,
      "learning_rate": 3.4503968185898696e-06,
      "loss": 0.8053,
      "step": 2933
    },
    {
      "epoch": 0.2645984578617487,
      "grad_norm": 1.5454899916434333,
      "learning_rate": 3.4499945043590047e-06,
      "loss": 0.7875,
      "step": 2934
    },
    {
      "epoch": 0.2646886413852189,
      "grad_norm": 2.0066225555657873,
      "learning_rate": 3.4495920664066137e-06,
      "loss": 0.7973,
      "step": 2935
    },
    {
      "epoch": 0.2647788249086892,
      "grad_norm": 1.9730548720272316,
      "learning_rate": 3.449189504767035e-06,
      "loss": 0.7799,
      "step": 2936
    },
    {
      "epoch": 0.26486900843215944,
      "grad_norm": 3.491886407558037,
      "learning_rate": 3.4487868194746163e-06,
      "loss": 0.8086,
      "step": 2937
    },
    {
      "epoch": 0.2649591919556297,
      "grad_norm": 1.7195893752288727,
      "learning_rate": 3.4483840105637175e-06,
      "loss": 0.823,
      "step": 2938
    },
    {
      "epoch": 0.26504937547909996,
      "grad_norm": 1.8737100614499198,
      "learning_rate": 3.4479810780687097e-06,
      "loss": 0.7686,
      "step": 2939
    },
    {
      "epoch": 0.26513955900257025,
      "grad_norm": 1.4222864247440563,
      "learning_rate": 3.4475780220239714e-06,
      "loss": 0.7745,
      "step": 2940
    },
    {
      "epoch": 0.2652297425260405,
      "grad_norm": 1.4148198412723991,
      "learning_rate": 3.4471748424638948e-06,
      "loss": 0.7718,
      "step": 2941
    },
    {
      "epoch": 0.26531992604951077,
      "grad_norm": 2.1401433472440416,
      "learning_rate": 3.4467715394228803e-06,
      "loss": 0.5925,
      "step": 2942
    },
    {
      "epoch": 0.265410109572981,
      "grad_norm": 1.3860109053894196,
      "learning_rate": 3.4463681129353413e-06,
      "loss": 0.785,
      "step": 2943
    },
    {
      "epoch": 0.2655002930964513,
      "grad_norm": 1.4298987727583004,
      "learning_rate": 3.4459645630357e-06,
      "loss": 0.7724,
      "step": 2944
    },
    {
      "epoch": 0.2655904766199215,
      "grad_norm": 1.6054510779231912,
      "learning_rate": 3.4455608897583884e-06,
      "loss": 0.8185,
      "step": 2945
    },
    {
      "epoch": 0.2656806601433918,
      "grad_norm": 2.329997289107551,
      "learning_rate": 3.4451570931378514e-06,
      "loss": 0.783,
      "step": 2946
    },
    {
      "epoch": 0.26577084366686204,
      "grad_norm": 1.4167985948205633,
      "learning_rate": 3.444753173208543e-06,
      "loss": 0.7888,
      "step": 2947
    },
    {
      "epoch": 0.26586102719033233,
      "grad_norm": 1.9526926401329072,
      "learning_rate": 3.444349130004927e-06,
      "loss": 0.7794,
      "step": 2948
    },
    {
      "epoch": 0.26595121071380257,
      "grad_norm": 1.7125389233920174,
      "learning_rate": 3.4439449635614794e-06,
      "loss": 0.7051,
      "step": 2949
    },
    {
      "epoch": 0.26604139423727285,
      "grad_norm": 1.5888902959098603,
      "learning_rate": 3.4435406739126854e-06,
      "loss": 0.8091,
      "step": 2950
    },
    {
      "epoch": 0.2661315777607431,
      "grad_norm": 2.102452993513839,
      "learning_rate": 3.443136261093042e-06,
      "loss": 0.6004,
      "step": 2951
    },
    {
      "epoch": 0.2662217612842134,
      "grad_norm": 1.5598897111249286,
      "learning_rate": 3.4427317251370553e-06,
      "loss": 0.7261,
      "step": 2952
    },
    {
      "epoch": 0.2663119448076836,
      "grad_norm": 2.7615642153657007,
      "learning_rate": 3.4423270660792422e-06,
      "loss": 0.6973,
      "step": 2953
    },
    {
      "epoch": 0.2664021283311539,
      "grad_norm": 1.6679182201298532,
      "learning_rate": 3.4419222839541314e-06,
      "loss": 0.6672,
      "step": 2954
    },
    {
      "epoch": 0.2664923118546242,
      "grad_norm": 1.419665957637364,
      "learning_rate": 3.4415173787962607e-06,
      "loss": 0.7564,
      "step": 2955
    },
    {
      "epoch": 0.2665824953780944,
      "grad_norm": 2.1394839621859014,
      "learning_rate": 3.4411123506401783e-06,
      "loss": 0.6444,
      "step": 2956
    },
    {
      "epoch": 0.2666726789015647,
      "grad_norm": 1.9098023006719773,
      "learning_rate": 3.440707199520444e-06,
      "loss": 0.6337,
      "step": 2957
    },
    {
      "epoch": 0.26676286242503494,
      "grad_norm": 1.55866778587473,
      "learning_rate": 3.440301925471628e-06,
      "loss": 0.742,
      "step": 2958
    },
    {
      "epoch": 0.26685304594850523,
      "grad_norm": 2.20605879433272,
      "learning_rate": 3.43989652852831e-06,
      "loss": 0.8156,
      "step": 2959
    },
    {
      "epoch": 0.26694322947197546,
      "grad_norm": 1.5616178692766567,
      "learning_rate": 3.4394910087250804e-06,
      "loss": 0.8403,
      "step": 2960
    },
    {
      "epoch": 0.26703341299544575,
      "grad_norm": 1.491301028584865,
      "learning_rate": 3.4390853660965405e-06,
      "loss": 0.7946,
      "step": 2961
    },
    {
      "epoch": 0.267123596518916,
      "grad_norm": 1.8953052945169093,
      "learning_rate": 3.438679600677302e-06,
      "loss": 0.8596,
      "step": 2962
    },
    {
      "epoch": 0.2672137800423863,
      "grad_norm": 2.6789725103274367,
      "learning_rate": 3.4382737125019874e-06,
      "loss": 0.735,
      "step": 2963
    },
    {
      "epoch": 0.2673039635658565,
      "grad_norm": 2.7511823019999766,
      "learning_rate": 3.4378677016052294e-06,
      "loss": 0.7485,
      "step": 2964
    },
    {
      "epoch": 0.2673941470893268,
      "grad_norm": 1.6964419070415218,
      "learning_rate": 3.43746156802167e-06,
      "loss": 0.7954,
      "step": 2965
    },
    {
      "epoch": 0.267484330612797,
      "grad_norm": 1.5268925539398166,
      "learning_rate": 3.4370553117859643e-06,
      "loss": 0.7162,
      "step": 2966
    },
    {
      "epoch": 0.2675745141362673,
      "grad_norm": 1.7465973880850945,
      "learning_rate": 3.4366489329327754e-06,
      "loss": 0.8069,
      "step": 2967
    },
    {
      "epoch": 0.26766469765973755,
      "grad_norm": 2.2111887435234614,
      "learning_rate": 3.4362424314967777e-06,
      "loss": 0.8036,
      "step": 2968
    },
    {
      "epoch": 0.26775488118320784,
      "grad_norm": 2.389017599418646,
      "learning_rate": 3.4358358075126567e-06,
      "loss": 0.8065,
      "step": 2969
    },
    {
      "epoch": 0.26784506470667807,
      "grad_norm": 1.6593583646755825,
      "learning_rate": 3.4354290610151077e-06,
      "loss": 0.8011,
      "step": 2970
    },
    {
      "epoch": 0.26793524823014836,
      "grad_norm": 1.3078643132272378,
      "learning_rate": 3.4350221920388354e-06,
      "loss": 0.7865,
      "step": 2971
    },
    {
      "epoch": 0.2680254317536186,
      "grad_norm": 1.494879087578896,
      "learning_rate": 3.4346152006185574e-06,
      "loss": 0.8779,
      "step": 2972
    },
    {
      "epoch": 0.2681156152770889,
      "grad_norm": 1.7248046349728576,
      "learning_rate": 3.4342080867890006e-06,
      "loss": 0.8772,
      "step": 2973
    },
    {
      "epoch": 0.2682057988005591,
      "grad_norm": 3.8021745784035486,
      "learning_rate": 3.4338008505849016e-06,
      "loss": 0.6987,
      "step": 2974
    },
    {
      "epoch": 0.2682959823240294,
      "grad_norm": 1.6191242065333502,
      "learning_rate": 3.433393492041008e-06,
      "loss": 0.7879,
      "step": 2975
    },
    {
      "epoch": 0.26838616584749964,
      "grad_norm": 1.3956499074104762,
      "learning_rate": 3.432986011192078e-06,
      "loss": 0.8295,
      "step": 2976
    },
    {
      "epoch": 0.2684763493709699,
      "grad_norm": 1.6479983354948953,
      "learning_rate": 3.4325784080728796e-06,
      "loss": 0.8472,
      "step": 2977
    },
    {
      "epoch": 0.2685665328944402,
      "grad_norm": 1.5031663536045792,
      "learning_rate": 3.4321706827181926e-06,
      "loss": 0.8099,
      "step": 2978
    },
    {
      "epoch": 0.26865671641791045,
      "grad_norm": 1.2105842813508372,
      "learning_rate": 3.4317628351628064e-06,
      "loss": 0.7311,
      "step": 2979
    },
    {
      "epoch": 0.26874689994138073,
      "grad_norm": 1.588518644906075,
      "learning_rate": 3.43135486544152e-06,
      "loss": 0.7599,
      "step": 2980
    },
    {
      "epoch": 0.26883708346485097,
      "grad_norm": 1.3954527559091008,
      "learning_rate": 3.4309467735891442e-06,
      "loss": 0.8148,
      "step": 2981
    },
    {
      "epoch": 0.26892726698832126,
      "grad_norm": 1.774430723642197,
      "learning_rate": 3.4305385596405e-06,
      "loss": 0.8648,
      "step": 2982
    },
    {
      "epoch": 0.2690174505117915,
      "grad_norm": 1.954508177218308,
      "learning_rate": 3.4301302236304174e-06,
      "loss": 0.7461,
      "step": 2983
    },
    {
      "epoch": 0.2691076340352618,
      "grad_norm": 0.6631524076559451,
      "learning_rate": 3.429721765593739e-06,
      "loss": 0.5481,
      "step": 2984
    },
    {
      "epoch": 0.269197817558732,
      "grad_norm": 1.7765018690997534,
      "learning_rate": 3.4293131855653155e-06,
      "loss": 0.8051,
      "step": 2985
    },
    {
      "epoch": 0.2692880010822023,
      "grad_norm": 1.6434922367953602,
      "learning_rate": 3.4289044835800102e-06,
      "loss": 0.7175,
      "step": 2986
    },
    {
      "epoch": 0.26937818460567253,
      "grad_norm": 1.7630308555720102,
      "learning_rate": 3.4284956596726953e-06,
      "loss": 0.7843,
      "step": 2987
    },
    {
      "epoch": 0.2694683681291428,
      "grad_norm": 1.4715279625618538,
      "learning_rate": 3.4280867138782544e-06,
      "loss": 0.7726,
      "step": 2988
    },
    {
      "epoch": 0.26955855165261305,
      "grad_norm": 2.282852132820049,
      "learning_rate": 3.4276776462315803e-06,
      "loss": 0.7856,
      "step": 2989
    },
    {
      "epoch": 0.26964873517608334,
      "grad_norm": 1.4308721085132035,
      "learning_rate": 3.427268456767578e-06,
      "loss": 0.7802,
      "step": 2990
    },
    {
      "epoch": 0.2697389186995536,
      "grad_norm": 2.609105627352526,
      "learning_rate": 3.42685914552116e-06,
      "loss": 0.8111,
      "step": 2991
    },
    {
      "epoch": 0.26982910222302386,
      "grad_norm": 1.5599530633565513,
      "learning_rate": 3.426449712527253e-06,
      "loss": 0.7505,
      "step": 2992
    },
    {
      "epoch": 0.2699192857464941,
      "grad_norm": 1.3984789708983767,
      "learning_rate": 3.4260401578207904e-06,
      "loss": 0.6951,
      "step": 2993
    },
    {
      "epoch": 0.2700094692699644,
      "grad_norm": 1.9617696873209425,
      "learning_rate": 3.4256304814367185e-06,
      "loss": 0.7505,
      "step": 2994
    },
    {
      "epoch": 0.2700996527934346,
      "grad_norm": 1.8240580518628107,
      "learning_rate": 3.4252206834099936e-06,
      "loss": 0.7394,
      "step": 2995
    },
    {
      "epoch": 0.2701898363169049,
      "grad_norm": 1.739700046971669,
      "learning_rate": 3.424810763775581e-06,
      "loss": 0.776,
      "step": 2996
    },
    {
      "epoch": 0.27028001984037514,
      "grad_norm": 1.794372540946931,
      "learning_rate": 3.4244007225684587e-06,
      "loss": 0.716,
      "step": 2997
    },
    {
      "epoch": 0.27037020336384543,
      "grad_norm": 1.6738477102622773,
      "learning_rate": 3.4239905598236115e-06,
      "loss": 0.7864,
      "step": 2998
    },
    {
      "epoch": 0.27046038688731566,
      "grad_norm": 1.5183555153497623,
      "learning_rate": 3.4235802755760386e-06,
      "loss": 0.8077,
      "step": 2999
    },
    {
      "epoch": 0.27055057041078595,
      "grad_norm": 0.8496752833967948,
      "learning_rate": 3.4231698698607464e-06,
      "loss": 0.6954,
      "step": 3000
    },
    {
      "epoch": 0.2706407539342562,
      "grad_norm": 1.6677857774348785,
      "learning_rate": 3.4227593427127543e-06,
      "loss": 0.8347,
      "step": 3001
    },
    {
      "epoch": 0.2707309374577265,
      "grad_norm": 1.811444994932368,
      "learning_rate": 3.42234869416709e-06,
      "loss": 0.747,
      "step": 3002
    },
    {
      "epoch": 0.27082112098119676,
      "grad_norm": 1.9316872185158787,
      "learning_rate": 3.421937924258792e-06,
      "loss": 0.7821,
      "step": 3003
    },
    {
      "epoch": 0.270911304504667,
      "grad_norm": 2.055461201206266,
      "learning_rate": 3.4215270330229096e-06,
      "loss": 0.8242,
      "step": 3004
    },
    {
      "epoch": 0.2710014880281373,
      "grad_norm": 2.1133775953517318,
      "learning_rate": 3.421116020494503e-06,
      "loss": 0.7436,
      "step": 3005
    },
    {
      "epoch": 0.2710916715516075,
      "grad_norm": 1.4942517607003443,
      "learning_rate": 3.420704886708642e-06,
      "loss": 0.7256,
      "step": 3006
    },
    {
      "epoch": 0.2711818550750778,
      "grad_norm": 1.142517557739638,
      "learning_rate": 3.4202936317004056e-06,
      "loss": 0.7296,
      "step": 3007
    },
    {
      "epoch": 0.27127203859854804,
      "grad_norm": 1.8882678762584946,
      "learning_rate": 3.4198822555048856e-06,
      "loss": 0.797,
      "step": 3008
    },
    {
      "epoch": 0.2713622221220183,
      "grad_norm": 1.7985721700111246,
      "learning_rate": 3.419470758157182e-06,
      "loss": 0.8652,
      "step": 3009
    },
    {
      "epoch": 0.27145240564548856,
      "grad_norm": 1.6776002912487455,
      "learning_rate": 3.4190591396924068e-06,
      "loss": 0.7802,
      "step": 3010
    },
    {
      "epoch": 0.27154258916895885,
      "grad_norm": 0.6829778021616402,
      "learning_rate": 3.418647400145681e-06,
      "loss": 0.616,
      "step": 3011
    },
    {
      "epoch": 0.2716327726924291,
      "grad_norm": 0.69978516296184,
      "learning_rate": 3.4182355395521367e-06,
      "loss": 0.5765,
      "step": 3012
    },
    {
      "epoch": 0.27172295621589937,
      "grad_norm": 1.6283472506196164,
      "learning_rate": 3.417823557946916e-06,
      "loss": 0.7761,
      "step": 3013
    },
    {
      "epoch": 0.2718131397393696,
      "grad_norm": 1.4518491825299438,
      "learning_rate": 3.417411455365172e-06,
      "loss": 0.6707,
      "step": 3014
    },
    {
      "epoch": 0.2719033232628399,
      "grad_norm": 1.8596022010283102,
      "learning_rate": 3.416999231842066e-06,
      "loss": 0.6895,
      "step": 3015
    },
    {
      "epoch": 0.2719935067863101,
      "grad_norm": 1.8209455314421144,
      "learning_rate": 3.416586887412773e-06,
      "loss": 0.8287,
      "step": 3016
    },
    {
      "epoch": 0.2720836903097804,
      "grad_norm": 2.403247793406342,
      "learning_rate": 3.416174422112476e-06,
      "loss": 0.7439,
      "step": 3017
    },
    {
      "epoch": 0.27217387383325065,
      "grad_norm": 0.6193885908358305,
      "learning_rate": 3.4157618359763687e-06,
      "loss": 0.5407,
      "step": 3018
    },
    {
      "epoch": 0.27226405735672093,
      "grad_norm": 1.3975621472701967,
      "learning_rate": 3.4153491290396542e-06,
      "loss": 0.6995,
      "step": 3019
    },
    {
      "epoch": 0.27235424088019117,
      "grad_norm": 1.705041139950931,
      "learning_rate": 3.4149363013375485e-06,
      "loss": 0.7917,
      "step": 3020
    },
    {
      "epoch": 0.27244442440366146,
      "grad_norm": 0.697458854849512,
      "learning_rate": 3.414523352905276e-06,
      "loss": 0.5558,
      "step": 3021
    },
    {
      "epoch": 0.2725346079271317,
      "grad_norm": 2.4816200768286416,
      "learning_rate": 3.414110283778071e-06,
      "loss": 0.7525,
      "step": 3022
    },
    {
      "epoch": 0.272624791450602,
      "grad_norm": 1.9082387961857479,
      "learning_rate": 3.4136970939911797e-06,
      "loss": 0.7305,
      "step": 3023
    },
    {
      "epoch": 0.2727149749740722,
      "grad_norm": 1.536191014025758,
      "learning_rate": 3.413283783579857e-06,
      "loss": 0.7431,
      "step": 3024
    },
    {
      "epoch": 0.2728051584975425,
      "grad_norm": 3.485462378281751,
      "learning_rate": 3.412870352579369e-06,
      "loss": 0.7214,
      "step": 3025
    },
    {
      "epoch": 0.2728953420210128,
      "grad_norm": 0.6620239607819564,
      "learning_rate": 3.4124568010249915e-06,
      "loss": 0.5969,
      "step": 3026
    },
    {
      "epoch": 0.272985525544483,
      "grad_norm": 1.5730685061835306,
      "learning_rate": 3.4120431289520124e-06,
      "loss": 0.7653,
      "step": 3027
    },
    {
      "epoch": 0.2730757090679533,
      "grad_norm": 1.6859541983214996,
      "learning_rate": 3.4116293363957276e-06,
      "loss": 0.7845,
      "step": 3028
    },
    {
      "epoch": 0.27316589259142354,
      "grad_norm": 1.8203412491652424,
      "learning_rate": 3.4112154233914438e-06,
      "loss": 0.7869,
      "step": 3029
    },
    {
      "epoch": 0.27325607611489383,
      "grad_norm": 1.8507962472226704,
      "learning_rate": 3.410801389974479e-06,
      "loss": 0.6839,
      "step": 3030
    },
    {
      "epoch": 0.27334625963836406,
      "grad_norm": 2.5305178489997746,
      "learning_rate": 3.410387236180161e-06,
      "loss": 0.8116,
      "step": 3031
    },
    {
      "epoch": 0.27343644316183435,
      "grad_norm": 1.760147781453288,
      "learning_rate": 3.409972962043826e-06,
      "loss": 0.6614,
      "step": 3032
    },
    {
      "epoch": 0.2735266266853046,
      "grad_norm": 2.133342746872796,
      "learning_rate": 3.4095585676008234e-06,
      "loss": 0.8343,
      "step": 3033
    },
    {
      "epoch": 0.2736168102087749,
      "grad_norm": 1.493446258563939,
      "learning_rate": 3.4091440528865125e-06,
      "loss": 0.7733,
      "step": 3034
    },
    {
      "epoch": 0.2737069937322451,
      "grad_norm": 3.226661957692685,
      "learning_rate": 3.4087294179362606e-06,
      "loss": 0.7782,
      "step": 3035
    },
    {
      "epoch": 0.2737971772557154,
      "grad_norm": 4.771528146225245,
      "learning_rate": 3.4083146627854474e-06,
      "loss": 0.8239,
      "step": 3036
    },
    {
      "epoch": 0.27388736077918563,
      "grad_norm": 2.0564386716045777,
      "learning_rate": 3.4078997874694614e-06,
      "loss": 0.7097,
      "step": 3037
    },
    {
      "epoch": 0.2739775443026559,
      "grad_norm": 1.6857974329224796,
      "learning_rate": 3.407484792023703e-06,
      "loss": 0.6731,
      "step": 3038
    },
    {
      "epoch": 0.27406772782612615,
      "grad_norm": 1.8316874557694522,
      "learning_rate": 3.407069676483581e-06,
      "loss": 0.7495,
      "step": 3039
    },
    {
      "epoch": 0.27415791134959644,
      "grad_norm": 1.4044570937564522,
      "learning_rate": 3.406654440884516e-06,
      "loss": 0.8156,
      "step": 3040
    },
    {
      "epoch": 0.2742480948730667,
      "grad_norm": 1.712464300187275,
      "learning_rate": 3.4062390852619372e-06,
      "loss": 0.8092,
      "step": 3041
    },
    {
      "epoch": 0.27433827839653696,
      "grad_norm": 1.7688549509352427,
      "learning_rate": 3.4058236096512867e-06,
      "loss": 0.7733,
      "step": 3042
    },
    {
      "epoch": 0.2744284619200072,
      "grad_norm": 1.9821698409100823,
      "learning_rate": 3.405408014088013e-06,
      "loss": 0.6749,
      "step": 3043
    },
    {
      "epoch": 0.2745186454434775,
      "grad_norm": 2.104845963412431,
      "learning_rate": 3.404992298607579e-06,
      "loss": 0.665,
      "step": 3044
    },
    {
      "epoch": 0.2746088289669477,
      "grad_norm": 2.3520193068172532,
      "learning_rate": 3.4045764632454547e-06,
      "loss": 0.7544,
      "step": 3045
    },
    {
      "epoch": 0.274699012490418,
      "grad_norm": 1.781534205735451,
      "learning_rate": 3.4041605080371223e-06,
      "loss": 0.7348,
      "step": 3046
    },
    {
      "epoch": 0.27478919601388824,
      "grad_norm": 3.953647382998306,
      "learning_rate": 3.4037444330180726e-06,
      "loss": 0.7886,
      "step": 3047
    },
    {
      "epoch": 0.2748793795373585,
      "grad_norm": 1.982792621389654,
      "learning_rate": 3.403328238223808e-06,
      "loss": 0.7914,
      "step": 3048
    },
    {
      "epoch": 0.27496956306082876,
      "grad_norm": 4.601680216554567,
      "learning_rate": 3.4029119236898395e-06,
      "loss": 0.8085,
      "step": 3049
    },
    {
      "epoch": 0.27505974658429905,
      "grad_norm": 1.069364420193586,
      "learning_rate": 3.4024954894516906e-06,
      "loss": 0.6095,
      "step": 3050
    },
    {
      "epoch": 0.27514993010776934,
      "grad_norm": 1.9867373844762393,
      "learning_rate": 3.4020789355448933e-06,
      "loss": 0.7798,
      "step": 3051
    },
    {
      "epoch": 0.27524011363123957,
      "grad_norm": 1.593180984944069,
      "learning_rate": 3.40166226200499e-06,
      "loss": 0.8482,
      "step": 3052
    },
    {
      "epoch": 0.27533029715470986,
      "grad_norm": 2.10036584755442,
      "learning_rate": 3.401245468867534e-06,
      "loss": 0.7636,
      "step": 3053
    },
    {
      "epoch": 0.2754204806781801,
      "grad_norm": 0.712993457641119,
      "learning_rate": 3.400828556168088e-06,
      "loss": 0.6333,
      "step": 3054
    },
    {
      "epoch": 0.2755106642016504,
      "grad_norm": 2.023760324302195,
      "learning_rate": 3.4004115239422255e-06,
      "loss": 0.7903,
      "step": 3055
    },
    {
      "epoch": 0.2756008477251206,
      "grad_norm": 1.9008481767680543,
      "learning_rate": 3.3999943722255305e-06,
      "loss": 0.7642,
      "step": 3056
    },
    {
      "epoch": 0.2756910312485909,
      "grad_norm": 1.6153268476259703,
      "learning_rate": 3.3995771010535955e-06,
      "loss": 0.6757,
      "step": 3057
    },
    {
      "epoch": 0.27578121477206113,
      "grad_norm": 0.7013232139542087,
      "learning_rate": 3.3991597104620253e-06,
      "loss": 0.5397,
      "step": 3058
    },
    {
      "epoch": 0.2758713982955314,
      "grad_norm": 1.6670084920665733,
      "learning_rate": 3.398742200486434e-06,
      "loss": 0.7548,
      "step": 3059
    },
    {
      "epoch": 0.27596158181900166,
      "grad_norm": 1.7505892715158478,
      "learning_rate": 3.3983245711624453e-06,
      "loss": 0.7946,
      "step": 3060
    },
    {
      "epoch": 0.27605176534247194,
      "grad_norm": 0.7926906229481142,
      "learning_rate": 3.3979068225256946e-06,
      "loss": 0.6554,
      "step": 3061
    },
    {
      "epoch": 0.2761419488659422,
      "grad_norm": 1.8968703469434602,
      "learning_rate": 3.3974889546118246e-06,
      "loss": 0.7473,
      "step": 3062
    },
    {
      "epoch": 0.27623213238941247,
      "grad_norm": 1.5552659995171727,
      "learning_rate": 3.3970709674564918e-06,
      "loss": 0.7165,
      "step": 3063
    },
    {
      "epoch": 0.2763223159128827,
      "grad_norm": 2.011541325420816,
      "learning_rate": 3.3966528610953607e-06,
      "loss": 0.7697,
      "step": 3064
    },
    {
      "epoch": 0.276412499436353,
      "grad_norm": 1.7132061414349813,
      "learning_rate": 3.3962346355641067e-06,
      "loss": 0.7805,
      "step": 3065
    },
    {
      "epoch": 0.2765026829598232,
      "grad_norm": 1.6399190201188616,
      "learning_rate": 3.3958162908984146e-06,
      "loss": 0.8071,
      "step": 3066
    },
    {
      "epoch": 0.2765928664832935,
      "grad_norm": 1.6024590936638117,
      "learning_rate": 3.39539782713398e-06,
      "loss": 0.7672,
      "step": 3067
    },
    {
      "epoch": 0.27668305000676374,
      "grad_norm": 1.8996743148946054,
      "learning_rate": 3.394979244306509e-06,
      "loss": 0.8019,
      "step": 3068
    },
    {
      "epoch": 0.27677323353023403,
      "grad_norm": 1.5788941727845947,
      "learning_rate": 3.3945605424517166e-06,
      "loss": 0.7839,
      "step": 3069
    },
    {
      "epoch": 0.27686341705370426,
      "grad_norm": 2.173862234226727,
      "learning_rate": 3.3941417216053294e-06,
      "loss": 0.771,
      "step": 3070
    },
    {
      "epoch": 0.27695360057717455,
      "grad_norm": 3.393369205812354,
      "learning_rate": 3.3937227818030835e-06,
      "loss": 0.7402,
      "step": 3071
    },
    {
      "epoch": 0.2770437841006448,
      "grad_norm": 1.7857335866157509,
      "learning_rate": 3.393303723080725e-06,
      "loss": 0.734,
      "step": 3072
    },
    {
      "epoch": 0.2771339676241151,
      "grad_norm": 1.7196973790601253,
      "learning_rate": 3.3928845454740097e-06,
      "loss": 0.7875,
      "step": 3073
    },
    {
      "epoch": 0.27722415114758536,
      "grad_norm": 1.8155019170172637,
      "learning_rate": 3.392465249018705e-06,
      "loss": 0.7612,
      "step": 3074
    },
    {
      "epoch": 0.2773143346710556,
      "grad_norm": 1.6762074459855913,
      "learning_rate": 3.3920458337505872e-06,
      "loss": 0.7721,
      "step": 3075
    },
    {
      "epoch": 0.2774045181945259,
      "grad_norm": 2.231985134546963,
      "learning_rate": 3.391626299705443e-06,
      "loss": 0.6169,
      "step": 3076
    },
    {
      "epoch": 0.2774947017179961,
      "grad_norm": 2.1325691279915975,
      "learning_rate": 3.39120664691907e-06,
      "loss": 0.7831,
      "step": 3077
    },
    {
      "epoch": 0.2775848852414664,
      "grad_norm": 1.6797194366967496,
      "learning_rate": 3.390786875427275e-06,
      "loss": 0.7457,
      "step": 3078
    },
    {
      "epoch": 0.27767506876493664,
      "grad_norm": 2.0868840159643867,
      "learning_rate": 3.390366985265875e-06,
      "loss": 0.7661,
      "step": 3079
    },
    {
      "epoch": 0.2777652522884069,
      "grad_norm": 1.5222639992744158,
      "learning_rate": 3.389946976470697e-06,
      "loss": 0.7285,
      "step": 3080
    },
    {
      "epoch": 0.27785543581187716,
      "grad_norm": 1.5570000553020842,
      "learning_rate": 3.3895268490775787e-06,
      "loss": 0.8654,
      "step": 3081
    },
    {
      "epoch": 0.27794561933534745,
      "grad_norm": 1.7006484057334443,
      "learning_rate": 3.3891066031223685e-06,
      "loss": 0.7291,
      "step": 3082
    },
    {
      "epoch": 0.2780358028588177,
      "grad_norm": 1.6912106583693398,
      "learning_rate": 3.3886862386409237e-06,
      "loss": 0.8053,
      "step": 3083
    },
    {
      "epoch": 0.27812598638228797,
      "grad_norm": 2.4403002377569707,
      "learning_rate": 3.388265755669111e-06,
      "loss": 0.7395,
      "step": 3084
    },
    {
      "epoch": 0.2782161699057582,
      "grad_norm": 2.0149646239097874,
      "learning_rate": 3.3878451542428093e-06,
      "loss": 0.8383,
      "step": 3085
    },
    {
      "epoch": 0.2783063534292285,
      "grad_norm": 1.43948649933219,
      "learning_rate": 3.387424434397907e-06,
      "loss": 0.7556,
      "step": 3086
    },
    {
      "epoch": 0.2783965369526987,
      "grad_norm": 5.21198460068554,
      "learning_rate": 3.3870035961703013e-06,
      "loss": 0.6869,
      "step": 3087
    },
    {
      "epoch": 0.278486720476169,
      "grad_norm": 6.566484386153943,
      "learning_rate": 3.3865826395959014e-06,
      "loss": 0.8976,
      "step": 3088
    },
    {
      "epoch": 0.27857690399963925,
      "grad_norm": 1.724814310998412,
      "learning_rate": 3.3861615647106253e-06,
      "loss": 0.8588,
      "step": 3089
    },
    {
      "epoch": 0.27866708752310954,
      "grad_norm": 2.430067811498569,
      "learning_rate": 3.3857403715504012e-06,
      "loss": 0.7892,
      "step": 3090
    },
    {
      "epoch": 0.27875727104657977,
      "grad_norm": 1.6828762570118412,
      "learning_rate": 3.385319060151167e-06,
      "loss": 0.7381,
      "step": 3091
    },
    {
      "epoch": 0.27884745457005006,
      "grad_norm": 2.1694221092742665,
      "learning_rate": 3.3848976305488728e-06,
      "loss": 0.7696,
      "step": 3092
    },
    {
      "epoch": 0.2789376380935203,
      "grad_norm": 2.3052220645937442,
      "learning_rate": 3.384476082779476e-06,
      "loss": 0.8165,
      "step": 3093
    },
    {
      "epoch": 0.2790278216169906,
      "grad_norm": 2.5416448519244796,
      "learning_rate": 3.3840544168789463e-06,
      "loss": 0.7485,
      "step": 3094
    },
    {
      "epoch": 0.2791180051404608,
      "grad_norm": 1.5991931788380624,
      "learning_rate": 3.3836326328832617e-06,
      "loss": 0.8766,
      "step": 3095
    },
    {
      "epoch": 0.2792081886639311,
      "grad_norm": 2.1788382020694237,
      "learning_rate": 3.383210730828412e-06,
      "loss": 0.728,
      "step": 3096
    },
    {
      "epoch": 0.2792983721874014,
      "grad_norm": 1.5059485106056778,
      "learning_rate": 3.3827887107503953e-06,
      "loss": 0.8409,
      "step": 3097
    },
    {
      "epoch": 0.2793885557108716,
      "grad_norm": 2.8383575286401435,
      "learning_rate": 3.3823665726852216e-06,
      "loss": 0.7191,
      "step": 3098
    },
    {
      "epoch": 0.2794787392343419,
      "grad_norm": 1.501774056701877,
      "learning_rate": 3.3819443166689095e-06,
      "loss": 0.8969,
      "step": 3099
    },
    {
      "epoch": 0.27956892275781214,
      "grad_norm": 1.547713543728277,
      "learning_rate": 3.3815219427374886e-06,
      "loss": 0.7678,
      "step": 3100
    },
    {
      "epoch": 0.27965910628128243,
      "grad_norm": 1.5865743326253443,
      "learning_rate": 3.3810994509269975e-06,
      "loss": 0.7925,
      "step": 3101
    },
    {
      "epoch": 0.27974928980475267,
      "grad_norm": 1.758848300510241,
      "learning_rate": 3.3806768412734864e-06,
      "loss": 0.8032,
      "step": 3102
    },
    {
      "epoch": 0.27983947332822295,
      "grad_norm": 1.5919833395867333,
      "learning_rate": 3.380254113813014e-06,
      "loss": 0.854,
      "step": 3103
    },
    {
      "epoch": 0.2799296568516932,
      "grad_norm": 1.6459451066061017,
      "learning_rate": 3.3798312685816496e-06,
      "loss": 0.804,
      "step": 3104
    },
    {
      "epoch": 0.2800198403751635,
      "grad_norm": 2.577274344007833,
      "learning_rate": 3.3794083056154738e-06,
      "loss": 0.7743,
      "step": 3105
    },
    {
      "epoch": 0.2801100238986337,
      "grad_norm": 1.6180255579680418,
      "learning_rate": 3.3789852249505746e-06,
      "loss": 0.7821,
      "step": 3106
    },
    {
      "epoch": 0.280200207422104,
      "grad_norm": 1.7051366422384522,
      "learning_rate": 3.378562026623053e-06,
      "loss": 0.8299,
      "step": 3107
    },
    {
      "epoch": 0.28029039094557423,
      "grad_norm": 1.6591662518079089,
      "learning_rate": 3.3781387106690175e-06,
      "loss": 0.8473,
      "step": 3108
    },
    {
      "epoch": 0.2803805744690445,
      "grad_norm": 1.4680750493191244,
      "learning_rate": 3.3777152771245885e-06,
      "loss": 0.8011,
      "step": 3109
    },
    {
      "epoch": 0.28047075799251475,
      "grad_norm": 1.7052380114353307,
      "learning_rate": 3.377291726025895e-06,
      "loss": 0.8022,
      "step": 3110
    },
    {
      "epoch": 0.28056094151598504,
      "grad_norm": 1.9522540782837507,
      "learning_rate": 3.3768680574090782e-06,
      "loss": 0.8878,
      "step": 3111
    },
    {
      "epoch": 0.2806511250394553,
      "grad_norm": 2.0426065915331466,
      "learning_rate": 3.3764442713102857e-06,
      "loss": 0.7496,
      "step": 3112
    },
    {
      "epoch": 0.28074130856292556,
      "grad_norm": 1.512972447403512,
      "learning_rate": 3.3760203677656786e-06,
      "loss": 0.7711,
      "step": 3113
    },
    {
      "epoch": 0.2808314920863958,
      "grad_norm": 1.8521933466788627,
      "learning_rate": 3.3755963468114262e-06,
      "loss": 0.8293,
      "step": 3114
    },
    {
      "epoch": 0.2809216756098661,
      "grad_norm": 1.6432319637059156,
      "learning_rate": 3.3751722084837095e-06,
      "loss": 0.6758,
      "step": 3115
    },
    {
      "epoch": 0.2810118591333363,
      "grad_norm": 1.6214190420995456,
      "learning_rate": 3.3747479528187166e-06,
      "loss": 0.7532,
      "step": 3116
    },
    {
      "epoch": 0.2811020426568066,
      "grad_norm": 1.8664001337001472,
      "learning_rate": 3.3743235798526485e-06,
      "loss": 0.8264,
      "step": 3117
    },
    {
      "epoch": 0.28119222618027684,
      "grad_norm": 1.4449662077229286,
      "learning_rate": 3.373899089621714e-06,
      "loss": 0.7053,
      "step": 3118
    },
    {
      "epoch": 0.2812824097037471,
      "grad_norm": 1.7806435606749993,
      "learning_rate": 3.373474482162134e-06,
      "loss": 0.754,
      "step": 3119
    },
    {
      "epoch": 0.28137259322721736,
      "grad_norm": 2.797924079421263,
      "learning_rate": 3.3730497575101376e-06,
      "loss": 0.7791,
      "step": 3120
    },
    {
      "epoch": 0.28146277675068765,
      "grad_norm": 1.615453850481619,
      "learning_rate": 3.3726249157019654e-06,
      "loss": 0.7705,
      "step": 3121
    },
    {
      "epoch": 0.28155296027415794,
      "grad_norm": 2.081430201705143,
      "learning_rate": 3.372199956773866e-06,
      "loss": 0.8607,
      "step": 3122
    },
    {
      "epoch": 0.28164314379762817,
      "grad_norm": 1.7586193352307822,
      "learning_rate": 3.371774880762101e-06,
      "loss": 0.7282,
      "step": 3123
    },
    {
      "epoch": 0.28173332732109846,
      "grad_norm": 1.8150752615759316,
      "learning_rate": 3.3713496877029392e-06,
      "loss": 0.7631,
      "step": 3124
    },
    {
      "epoch": 0.2818235108445687,
      "grad_norm": 1.993117470689669,
      "learning_rate": 3.37092437763266e-06,
      "loss": 0.8383,
      "step": 3125
    },
    {
      "epoch": 0.281913694368039,
      "grad_norm": 1.8872671021527834,
      "learning_rate": 3.3704989505875537e-06,
      "loss": 0.8096,
      "step": 3126
    },
    {
      "epoch": 0.2820038778915092,
      "grad_norm": 1.6175804213705853,
      "learning_rate": 3.3700734066039205e-06,
      "loss": 0.7445,
      "step": 3127
    },
    {
      "epoch": 0.2820940614149795,
      "grad_norm": 2.7586888982093054,
      "learning_rate": 3.36964774571807e-06,
      "loss": 0.8065,
      "step": 3128
    },
    {
      "epoch": 0.28218424493844974,
      "grad_norm": 2.0337240557665788,
      "learning_rate": 3.3692219679663206e-06,
      "loss": 0.774,
      "step": 3129
    },
    {
      "epoch": 0.28227442846192,
      "grad_norm": 2.041017494018283,
      "learning_rate": 3.3687960733850043e-06,
      "loss": 0.7955,
      "step": 3130
    },
    {
      "epoch": 0.28236461198539026,
      "grad_norm": 1.7150734853988014,
      "learning_rate": 3.3683700620104586e-06,
      "loss": 0.7154,
      "step": 3131
    },
    {
      "epoch": 0.28245479550886055,
      "grad_norm": 1.9391137447548161,
      "learning_rate": 3.3679439338790347e-06,
      "loss": 0.79,
      "step": 3132
    },
    {
      "epoch": 0.2825449790323308,
      "grad_norm": 1.5794405599859358,
      "learning_rate": 3.3675176890270916e-06,
      "loss": 0.8988,
      "step": 3133
    },
    {
      "epoch": 0.28263516255580107,
      "grad_norm": 1.5381543728574842,
      "learning_rate": 3.367091327490998e-06,
      "loss": 0.8003,
      "step": 3134
    },
    {
      "epoch": 0.2827253460792713,
      "grad_norm": 1.7242129119649763,
      "learning_rate": 3.3666648493071347e-06,
      "loss": 0.767,
      "step": 3135
    },
    {
      "epoch": 0.2828155296027416,
      "grad_norm": 1.703825220475237,
      "learning_rate": 3.3662382545118914e-06,
      "loss": 0.7546,
      "step": 3136
    },
    {
      "epoch": 0.2829057131262118,
      "grad_norm": 1.7056289604591135,
      "learning_rate": 3.3658115431416663e-06,
      "loss": 0.8574,
      "step": 3137
    },
    {
      "epoch": 0.2829958966496821,
      "grad_norm": 0.6979761643930739,
      "learning_rate": 3.36538471523287e-06,
      "loss": 0.609,
      "step": 3138
    },
    {
      "epoch": 0.28308608017315234,
      "grad_norm": 1.493458630862733,
      "learning_rate": 3.3649577708219204e-06,
      "loss": 0.8322,
      "step": 3139
    },
    {
      "epoch": 0.28317626369662263,
      "grad_norm": 1.678664423734779,
      "learning_rate": 3.3645307099452477e-06,
      "loss": 0.7244,
      "step": 3140
    },
    {
      "epoch": 0.28326644722009287,
      "grad_norm": 1.8340350094306435,
      "learning_rate": 3.3641035326392907e-06,
      "loss": 0.7746,
      "step": 3141
    },
    {
      "epoch": 0.28335663074356315,
      "grad_norm": 2.112186620819049,
      "learning_rate": 3.363676238940499e-06,
      "loss": 0.7902,
      "step": 3142
    },
    {
      "epoch": 0.2834468142670334,
      "grad_norm": 1.808640217853548,
      "learning_rate": 3.363248828885331e-06,
      "loss": 0.8049,
      "step": 3143
    },
    {
      "epoch": 0.2835369977905037,
      "grad_norm": 1.4391575254861617,
      "learning_rate": 3.3628213025102562e-06,
      "loss": 0.802,
      "step": 3144
    },
    {
      "epoch": 0.28362718131397396,
      "grad_norm": 1.7801137947826504,
      "learning_rate": 3.3623936598517536e-06,
      "loss": 0.8164,
      "step": 3145
    },
    {
      "epoch": 0.2837173648374442,
      "grad_norm": 2.0708438983036257,
      "learning_rate": 3.3619659009463117e-06,
      "loss": 0.6257,
      "step": 3146
    },
    {
      "epoch": 0.2838075483609145,
      "grad_norm": 1.5771144993204123,
      "learning_rate": 3.3615380258304287e-06,
      "loss": 0.8261,
      "step": 3147
    },
    {
      "epoch": 0.2838977318843847,
      "grad_norm": 1.8503924829829244,
      "learning_rate": 3.3611100345406146e-06,
      "loss": 0.7647,
      "step": 3148
    },
    {
      "epoch": 0.283987915407855,
      "grad_norm": 1.6396919140211963,
      "learning_rate": 3.3606819271133873e-06,
      "loss": 0.765,
      "step": 3149
    },
    {
      "epoch": 0.28407809893132524,
      "grad_norm": 0.7189409168277952,
      "learning_rate": 3.360253703585275e-06,
      "loss": 0.5934,
      "step": 3150
    },
    {
      "epoch": 0.28416828245479553,
      "grad_norm": 1.907551868078504,
      "learning_rate": 3.3598253639928164e-06,
      "loss": 0.7193,
      "step": 3151
    },
    {
      "epoch": 0.28425846597826576,
      "grad_norm": 1.8914223873138585,
      "learning_rate": 3.3593969083725596e-06,
      "loss": 0.7344,
      "step": 3152
    },
    {
      "epoch": 0.28434864950173605,
      "grad_norm": 3.028544368777242,
      "learning_rate": 3.358968336761063e-06,
      "loss": 0.7351,
      "step": 3153
    },
    {
      "epoch": 0.2844388330252063,
      "grad_norm": 1.6699928313232792,
      "learning_rate": 3.3585396491948945e-06,
      "loss": 0.7399,
      "step": 3154
    },
    {
      "epoch": 0.28452901654867657,
      "grad_norm": 1.7438057538488545,
      "learning_rate": 3.358110845710633e-06,
      "loss": 0.8327,
      "step": 3155
    },
    {
      "epoch": 0.2846192000721468,
      "grad_norm": 1.9111084856094147,
      "learning_rate": 3.357681926344865e-06,
      "loss": 0.7648,
      "step": 3156
    },
    {
      "epoch": 0.2847093835956171,
      "grad_norm": 1.60842600870039,
      "learning_rate": 3.357252891134189e-06,
      "loss": 0.7744,
      "step": 3157
    },
    {
      "epoch": 0.2847995671190873,
      "grad_norm": 1.8382579018941665,
      "learning_rate": 3.356823740115212e-06,
      "loss": 0.8232,
      "step": 3158
    },
    {
      "epoch": 0.2848897506425576,
      "grad_norm": 1.2743467115823703,
      "learning_rate": 3.3563944733245525e-06,
      "loss": 0.7197,
      "step": 3159
    },
    {
      "epoch": 0.28497993416602785,
      "grad_norm": 1.8552697004663516,
      "learning_rate": 3.3559650907988375e-06,
      "loss": 0.767,
      "step": 3160
    },
    {
      "epoch": 0.28507011768949814,
      "grad_norm": 2.804781941388495,
      "learning_rate": 3.3555355925747045e-06,
      "loss": 0.6213,
      "step": 3161
    },
    {
      "epoch": 0.28516030121296837,
      "grad_norm": 1.7517501390254848,
      "learning_rate": 3.3551059786888e-06,
      "loss": 0.7633,
      "step": 3162
    },
    {
      "epoch": 0.28525048473643866,
      "grad_norm": 1.7958554900086652,
      "learning_rate": 3.3546762491777807e-06,
      "loss": 0.8224,
      "step": 3163
    },
    {
      "epoch": 0.2853406682599089,
      "grad_norm": 1.6528378733048528,
      "learning_rate": 3.3542464040783156e-06,
      "loss": 0.7697,
      "step": 3164
    },
    {
      "epoch": 0.2854308517833792,
      "grad_norm": 1.9745488218838885,
      "learning_rate": 3.353816443427079e-06,
      "loss": 0.7923,
      "step": 3165
    },
    {
      "epoch": 0.2855210353068494,
      "grad_norm": 1.893726484700362,
      "learning_rate": 3.3533863672607597e-06,
      "loss": 0.7552,
      "step": 3166
    },
    {
      "epoch": 0.2856112188303197,
      "grad_norm": 1.613282062816646,
      "learning_rate": 3.352956175616052e-06,
      "loss": 0.8033,
      "step": 3167
    },
    {
      "epoch": 0.28570140235378993,
      "grad_norm": 1.5984057321682426,
      "learning_rate": 3.352525868529664e-06,
      "loss": 0.7415,
      "step": 3168
    },
    {
      "epoch": 0.2857915858772602,
      "grad_norm": 1.911132126361176,
      "learning_rate": 3.3520954460383103e-06,
      "loss": 0.7179,
      "step": 3169
    },
    {
      "epoch": 0.2858817694007305,
      "grad_norm": 1.783152785327074,
      "learning_rate": 3.3516649081787182e-06,
      "loss": 0.7114,
      "step": 3170
    },
    {
      "epoch": 0.28597195292420075,
      "grad_norm": 2.2011418717107247,
      "learning_rate": 3.3512342549876236e-06,
      "loss": 0.8743,
      "step": 3171
    },
    {
      "epoch": 0.28606213644767103,
      "grad_norm": 1.8050992421687981,
      "learning_rate": 3.350803486501771e-06,
      "loss": 0.7303,
      "step": 3172
    },
    {
      "epoch": 0.28615231997114127,
      "grad_norm": 1.6610583917182538,
      "learning_rate": 3.3503726027579175e-06,
      "loss": 0.8011,
      "step": 3173
    },
    {
      "epoch": 0.28624250349461156,
      "grad_norm": 1.4256173653672024,
      "learning_rate": 3.349941603792827e-06,
      "loss": 0.8273,
      "step": 3174
    },
    {
      "epoch": 0.2863326870180818,
      "grad_norm": 1.6443668382964793,
      "learning_rate": 3.3495104896432755e-06,
      "loss": 0.7697,
      "step": 3175
    },
    {
      "epoch": 0.2864228705415521,
      "grad_norm": 1.5729734734725476,
      "learning_rate": 3.3490792603460477e-06,
      "loss": 0.7385,
      "step": 3176
    },
    {
      "epoch": 0.2865130540650223,
      "grad_norm": 2.0108165310367028,
      "learning_rate": 3.3486479159379393e-06,
      "loss": 0.7075,
      "step": 3177
    },
    {
      "epoch": 0.2866032375884926,
      "grad_norm": 0.6730158900004742,
      "learning_rate": 3.3482164564557537e-06,
      "loss": 0.5773,
      "step": 3178
    },
    {
      "epoch": 0.28669342111196283,
      "grad_norm": 2.174970438087648,
      "learning_rate": 3.3477848819363065e-06,
      "loss": 0.8309,
      "step": 3179
    },
    {
      "epoch": 0.2867836046354331,
      "grad_norm": 0.6594642985472572,
      "learning_rate": 3.3473531924164213e-06,
      "loss": 0.6053,
      "step": 3180
    },
    {
      "epoch": 0.28687378815890335,
      "grad_norm": 3.6344628307852282,
      "learning_rate": 3.3469213879329325e-06,
      "loss": 0.7278,
      "step": 3181
    },
    {
      "epoch": 0.28696397168237364,
      "grad_norm": 1.8474888796336157,
      "learning_rate": 3.3464894685226837e-06,
      "loss": 0.7346,
      "step": 3182
    },
    {
      "epoch": 0.2870541552058439,
      "grad_norm": 1.6282125409217671,
      "learning_rate": 3.34605743422253e-06,
      "loss": 0.8484,
      "step": 3183
    },
    {
      "epoch": 0.28714433872931416,
      "grad_norm": 1.7997884599571954,
      "learning_rate": 3.345625285069333e-06,
      "loss": 0.7428,
      "step": 3184
    },
    {
      "epoch": 0.2872345222527844,
      "grad_norm": 1.898101086106323,
      "learning_rate": 3.345193021099967e-06,
      "loss": 0.7533,
      "step": 3185
    },
    {
      "epoch": 0.2873247057762547,
      "grad_norm": 1.9773784657011546,
      "learning_rate": 3.3447606423513157e-06,
      "loss": 0.7425,
      "step": 3186
    },
    {
      "epoch": 0.2874148892997249,
      "grad_norm": 3.5478336870334846,
      "learning_rate": 3.344328148860271e-06,
      "loss": 0.7438,
      "step": 3187
    },
    {
      "epoch": 0.2875050728231952,
      "grad_norm": 1.396363025007805,
      "learning_rate": 3.3438955406637365e-06,
      "loss": 0.7841,
      "step": 3188
    },
    {
      "epoch": 0.28759525634666544,
      "grad_norm": 1.5570191960459154,
      "learning_rate": 3.343462817798624e-06,
      "loss": 0.7967,
      "step": 3189
    },
    {
      "epoch": 0.28768543987013573,
      "grad_norm": 1.8848747261371486,
      "learning_rate": 3.343029980301856e-06,
      "loss": 0.739,
      "step": 3190
    },
    {
      "epoch": 0.28777562339360596,
      "grad_norm": 2.0456113918934573,
      "learning_rate": 3.342597028210365e-06,
      "loss": 0.8079,
      "step": 3191
    },
    {
      "epoch": 0.28786580691707625,
      "grad_norm": 2.046581218107899,
      "learning_rate": 3.342163961561092e-06,
      "loss": 0.7187,
      "step": 3192
    },
    {
      "epoch": 0.28795599044054654,
      "grad_norm": 1.5848367563653938,
      "learning_rate": 3.34173078039099e-06,
      "loss": 0.7456,
      "step": 3193
    },
    {
      "epoch": 0.28804617396401677,
      "grad_norm": 1.8504781646347295,
      "learning_rate": 3.3412974847370193e-06,
      "loss": 0.9401,
      "step": 3194
    },
    {
      "epoch": 0.28813635748748706,
      "grad_norm": 1.4152890781545842,
      "learning_rate": 3.3408640746361514e-06,
      "loss": 0.7953,
      "step": 3195
    },
    {
      "epoch": 0.2882265410109573,
      "grad_norm": 0.7557638339594585,
      "learning_rate": 3.3404305501253663e-06,
      "loss": 0.5997,
      "step": 3196
    },
    {
      "epoch": 0.2883167245344276,
      "grad_norm": 1.5214029664968627,
      "learning_rate": 3.3399969112416565e-06,
      "loss": 0.8489,
      "step": 3197
    },
    {
      "epoch": 0.2884069080578978,
      "grad_norm": 1.6806453369090462,
      "learning_rate": 3.3395631580220213e-06,
      "loss": 0.7097,
      "step": 3198
    },
    {
      "epoch": 0.2884970915813681,
      "grad_norm": 1.6529344445283742,
      "learning_rate": 3.3391292905034714e-06,
      "loss": 0.7523,
      "step": 3199
    },
    {
      "epoch": 0.28858727510483834,
      "grad_norm": 1.6184152558973361,
      "learning_rate": 3.338695308723027e-06,
      "loss": 0.8396,
      "step": 3200
    },
    {
      "epoch": 0.2886774586283086,
      "grad_norm": 1.7173220164378287,
      "learning_rate": 3.338261212717716e-06,
      "loss": 0.8288,
      "step": 3201
    },
    {
      "epoch": 0.28876764215177886,
      "grad_norm": 1.815106129656642,
      "learning_rate": 3.33782700252458e-06,
      "loss": 0.8071,
      "step": 3202
    },
    {
      "epoch": 0.28885782567524915,
      "grad_norm": 2.129984563314889,
      "learning_rate": 3.337392678180668e-06,
      "loss": 0.8185,
      "step": 3203
    },
    {
      "epoch": 0.2889480091987194,
      "grad_norm": 1.9262757198490577,
      "learning_rate": 3.3369582397230377e-06,
      "loss": 0.7776,
      "step": 3204
    },
    {
      "epoch": 0.28903819272218967,
      "grad_norm": 1.7504173189940035,
      "learning_rate": 3.336523687188759e-06,
      "loss": 0.8175,
      "step": 3205
    },
    {
      "epoch": 0.2891283762456599,
      "grad_norm": 0.7194434013956463,
      "learning_rate": 3.336089020614909e-06,
      "loss": 0.6138,
      "step": 3206
    },
    {
      "epoch": 0.2892185597691302,
      "grad_norm": 1.6896321990084189,
      "learning_rate": 3.3356542400385774e-06,
      "loss": 0.6548,
      "step": 3207
    },
    {
      "epoch": 0.2893087432926004,
      "grad_norm": 2.0311408380373015,
      "learning_rate": 3.3352193454968607e-06,
      "loss": 0.71,
      "step": 3208
    },
    {
      "epoch": 0.2893989268160707,
      "grad_norm": 2.2526769607924386,
      "learning_rate": 3.3347843370268675e-06,
      "loss": 0.8359,
      "step": 3209
    },
    {
      "epoch": 0.28948911033954094,
      "grad_norm": 1.4205065839078457,
      "learning_rate": 3.334349214665715e-06,
      "loss": 0.7887,
      "step": 3210
    },
    {
      "epoch": 0.28957929386301123,
      "grad_norm": 1.5590267488192546,
      "learning_rate": 3.3339139784505293e-06,
      "loss": 0.8586,
      "step": 3211
    },
    {
      "epoch": 0.28966947738648147,
      "grad_norm": 2.060110846935222,
      "learning_rate": 3.333478628418448e-06,
      "loss": 0.8421,
      "step": 3212
    },
    {
      "epoch": 0.28975966090995176,
      "grad_norm": 2.03658673466207,
      "learning_rate": 3.333043164606618e-06,
      "loss": 0.668,
      "step": 3213
    },
    {
      "epoch": 0.289849844433422,
      "grad_norm": 2.0728616243910953,
      "learning_rate": 3.3326075870521948e-06,
      "loss": 0.8586,
      "step": 3214
    },
    {
      "epoch": 0.2899400279568923,
      "grad_norm": 1.8828300380285685,
      "learning_rate": 3.3321718957923437e-06,
      "loss": 0.7752,
      "step": 3215
    },
    {
      "epoch": 0.29003021148036257,
      "grad_norm": 1.619550222901257,
      "learning_rate": 3.3317360908642413e-06,
      "loss": 0.7718,
      "step": 3216
    },
    {
      "epoch": 0.2901203950038328,
      "grad_norm": 2.326727428444431,
      "learning_rate": 3.331300172305072e-06,
      "loss": 0.7318,
      "step": 3217
    },
    {
      "epoch": 0.2902105785273031,
      "grad_norm": 1.6781546762353599,
      "learning_rate": 3.330864140152032e-06,
      "loss": 0.765,
      "step": 3218
    },
    {
      "epoch": 0.2903007620507733,
      "grad_norm": 1.8240623652144607,
      "learning_rate": 3.330427994442325e-06,
      "loss": 0.7914,
      "step": 3219
    },
    {
      "epoch": 0.2903909455742436,
      "grad_norm": 1.883553260295258,
      "learning_rate": 3.3299917352131657e-06,
      "loss": 0.8136,
      "step": 3220
    },
    {
      "epoch": 0.29048112909771384,
      "grad_norm": 2.5444288617933593,
      "learning_rate": 3.329555362501778e-06,
      "loss": 0.7438,
      "step": 3221
    },
    {
      "epoch": 0.29057131262118413,
      "grad_norm": 1.8809539992950357,
      "learning_rate": 3.3291188763453954e-06,
      "loss": 0.7411,
      "step": 3222
    },
    {
      "epoch": 0.29066149614465436,
      "grad_norm": 2.7382492956019364,
      "learning_rate": 3.3286822767812618e-06,
      "loss": 0.7659,
      "step": 3223
    },
    {
      "epoch": 0.29075167966812465,
      "grad_norm": 4.203395905680617,
      "learning_rate": 3.32824556384663e-06,
      "loss": 0.758,
      "step": 3224
    },
    {
      "epoch": 0.2908418631915949,
      "grad_norm": 2.0974476744316553,
      "learning_rate": 3.3278087375787628e-06,
      "loss": 0.7788,
      "step": 3225
    },
    {
      "epoch": 0.2909320467150652,
      "grad_norm": 2.312078489086663,
      "learning_rate": 3.327371798014933e-06,
      "loss": 0.7804,
      "step": 3226
    },
    {
      "epoch": 0.2910222302385354,
      "grad_norm": 1.7816202548517932,
      "learning_rate": 3.3269347451924218e-06,
      "loss": 0.7757,
      "step": 3227
    },
    {
      "epoch": 0.2911124137620057,
      "grad_norm": 1.6044356463285896,
      "learning_rate": 3.326497579148522e-06,
      "loss": 0.8398,
      "step": 3228
    },
    {
      "epoch": 0.29120259728547593,
      "grad_norm": 1.7788563757779188,
      "learning_rate": 3.3260602999205345e-06,
      "loss": 0.7759,
      "step": 3229
    },
    {
      "epoch": 0.2912927808089462,
      "grad_norm": 1.5548132457391712,
      "learning_rate": 3.32562290754577e-06,
      "loss": 0.787,
      "step": 3230
    },
    {
      "epoch": 0.29138296433241645,
      "grad_norm": 1.7533592271645397,
      "learning_rate": 3.3251854020615494e-06,
      "loss": 0.762,
      "step": 3231
    },
    {
      "epoch": 0.29147314785588674,
      "grad_norm": 1.9085165213513657,
      "learning_rate": 3.324747783505204e-06,
      "loss": 0.7824,
      "step": 3232
    },
    {
      "epoch": 0.29156333137935697,
      "grad_norm": 1.8033385416450431,
      "learning_rate": 3.324310051914073e-06,
      "loss": 0.6942,
      "step": 3233
    },
    {
      "epoch": 0.29165351490282726,
      "grad_norm": 2.1922270789621368,
      "learning_rate": 3.3238722073255056e-06,
      "loss": 0.7466,
      "step": 3234
    },
    {
      "epoch": 0.2917436984262975,
      "grad_norm": 1.9113440691674477,
      "learning_rate": 3.323434249776863e-06,
      "loss": 0.7374,
      "step": 3235
    },
    {
      "epoch": 0.2918338819497678,
      "grad_norm": 0.7221660704262153,
      "learning_rate": 3.3229961793055117e-06,
      "loss": 0.5899,
      "step": 3236
    },
    {
      "epoch": 0.291924065473238,
      "grad_norm": 1.5226100924421089,
      "learning_rate": 3.3225579959488314e-06,
      "loss": 0.7034,
      "step": 3237
    },
    {
      "epoch": 0.2920142489967083,
      "grad_norm": 1.841370485692443,
      "learning_rate": 3.322119699744211e-06,
      "loss": 0.7966,
      "step": 3238
    },
    {
      "epoch": 0.29210443252017854,
      "grad_norm": 1.5137618893346234,
      "learning_rate": 3.3216812907290476e-06,
      "loss": 0.7885,
      "step": 3239
    },
    {
      "epoch": 0.2921946160436488,
      "grad_norm": 2.366459800741477,
      "learning_rate": 3.3212427689407484e-06,
      "loss": 0.7965,
      "step": 3240
    },
    {
      "epoch": 0.2922847995671191,
      "grad_norm": 0.6517131978769983,
      "learning_rate": 3.3208041344167317e-06,
      "loss": 0.5679,
      "step": 3241
    },
    {
      "epoch": 0.29237498309058935,
      "grad_norm": 1.651396856599791,
      "learning_rate": 3.3203653871944224e-06,
      "loss": 0.6845,
      "step": 3242
    },
    {
      "epoch": 0.29246516661405964,
      "grad_norm": 1.5141955841751817,
      "learning_rate": 3.3199265273112587e-06,
      "loss": 0.7336,
      "step": 3243
    },
    {
      "epoch": 0.29255535013752987,
      "grad_norm": 1.8320884451651325,
      "learning_rate": 3.3194875548046852e-06,
      "loss": 0.7149,
      "step": 3244
    },
    {
      "epoch": 0.29264553366100016,
      "grad_norm": 1.7167662529932204,
      "learning_rate": 3.319048469712158e-06,
      "loss": 0.7632,
      "step": 3245
    },
    {
      "epoch": 0.2927357171844704,
      "grad_norm": 2.177421129902762,
      "learning_rate": 3.3186092720711423e-06,
      "loss": 0.8719,
      "step": 3246
    },
    {
      "epoch": 0.2928259007079407,
      "grad_norm": 1.8406799748602918,
      "learning_rate": 3.3181699619191125e-06,
      "loss": 0.7054,
      "step": 3247
    },
    {
      "epoch": 0.2929160842314109,
      "grad_norm": 1.5890257134394654,
      "learning_rate": 3.3177305392935536e-06,
      "loss": 0.8144,
      "step": 3248
    },
    {
      "epoch": 0.2930062677548812,
      "grad_norm": 0.731535148451496,
      "learning_rate": 3.3172910042319595e-06,
      "loss": 0.6558,
      "step": 3249
    },
    {
      "epoch": 0.29309645127835143,
      "grad_norm": 1.4467559205193319,
      "learning_rate": 3.316851356771833e-06,
      "loss": 0.8063,
      "step": 3250
    },
    {
      "epoch": 0.2931866348018217,
      "grad_norm": 1.4215929359436557,
      "learning_rate": 3.3164115969506876e-06,
      "loss": 0.8047,
      "step": 3251
    },
    {
      "epoch": 0.29327681832529195,
      "grad_norm": 2.0342371177532663,
      "learning_rate": 3.315971724806046e-06,
      "loss": 0.863,
      "step": 3252
    },
    {
      "epoch": 0.29336700184876224,
      "grad_norm": 2.0120773440258914,
      "learning_rate": 3.315531740375441e-06,
      "loss": 0.7185,
      "step": 3253
    },
    {
      "epoch": 0.2934571853722325,
      "grad_norm": 1.4610618497588665,
      "learning_rate": 3.315091643696414e-06,
      "loss": 0.7437,
      "step": 3254
    },
    {
      "epoch": 0.29354736889570276,
      "grad_norm": 19.059345121871562,
      "learning_rate": 3.3146514348065164e-06,
      "loss": 0.7039,
      "step": 3255
    },
    {
      "epoch": 0.293637552419173,
      "grad_norm": 2.7811004941382635,
      "learning_rate": 3.31421111374331e-06,
      "loss": 0.7819,
      "step": 3256
    },
    {
      "epoch": 0.2937277359426433,
      "grad_norm": 2.4957292317463766,
      "learning_rate": 3.3137706805443647e-06,
      "loss": 0.789,
      "step": 3257
    },
    {
      "epoch": 0.2938179194661135,
      "grad_norm": 1.764470558511084,
      "learning_rate": 3.313330135247261e-06,
      "loss": 0.8197,
      "step": 3258
    },
    {
      "epoch": 0.2939081029895838,
      "grad_norm": 1.3395857938009224,
      "learning_rate": 3.312889477889588e-06,
      "loss": 0.7471,
      "step": 3259
    },
    {
      "epoch": 0.29399828651305404,
      "grad_norm": 1.7816592633076014,
      "learning_rate": 3.3124487085089464e-06,
      "loss": 0.7885,
      "step": 3260
    },
    {
      "epoch": 0.29408847003652433,
      "grad_norm": 1.9170559888532712,
      "learning_rate": 3.312007827142943e-06,
      "loss": 0.8089,
      "step": 3261
    },
    {
      "epoch": 0.29417865355999456,
      "grad_norm": 1.4374683210779928,
      "learning_rate": 3.3115668338291983e-06,
      "loss": 0.6954,
      "step": 3262
    },
    {
      "epoch": 0.29426883708346485,
      "grad_norm": 1.6523157572798763,
      "learning_rate": 3.3111257286053394e-06,
      "loss": 0.8534,
      "step": 3263
    },
    {
      "epoch": 0.29435902060693514,
      "grad_norm": 1.5756030184745686,
      "learning_rate": 3.3106845115090043e-06,
      "loss": 0.7752,
      "step": 3264
    },
    {
      "epoch": 0.2944492041304054,
      "grad_norm": 1.6141019123948779,
      "learning_rate": 3.310243182577839e-06,
      "loss": 0.7307,
      "step": 3265
    },
    {
      "epoch": 0.29453938765387566,
      "grad_norm": 1.5304954382383849,
      "learning_rate": 3.3098017418495007e-06,
      "loss": 0.667,
      "step": 3266
    },
    {
      "epoch": 0.2946295711773459,
      "grad_norm": 2.0118653475484534,
      "learning_rate": 3.309360189361656e-06,
      "loss": 0.7429,
      "step": 3267
    },
    {
      "epoch": 0.2947197547008162,
      "grad_norm": 1.8414750369159156,
      "learning_rate": 3.3089185251519797e-06,
      "loss": 0.8524,
      "step": 3268
    },
    {
      "epoch": 0.2948099382242864,
      "grad_norm": 2.313959125319348,
      "learning_rate": 3.3084767492581574e-06,
      "loss": 0.7513,
      "step": 3269
    },
    {
      "epoch": 0.2949001217477567,
      "grad_norm": 2.7048001253247054,
      "learning_rate": 3.3080348617178846e-06,
      "loss": 0.7957,
      "step": 3270
    },
    {
      "epoch": 0.29499030527122694,
      "grad_norm": 1.5628271141963168,
      "learning_rate": 3.307592862568865e-06,
      "loss": 0.7477,
      "step": 3271
    },
    {
      "epoch": 0.2950804887946972,
      "grad_norm": 1.570144416581519,
      "learning_rate": 3.307150751848812e-06,
      "loss": 0.8361,
      "step": 3272
    },
    {
      "epoch": 0.29517067231816746,
      "grad_norm": 1.7691835310220982,
      "learning_rate": 3.3067085295954497e-06,
      "loss": 0.6805,
      "step": 3273
    },
    {
      "epoch": 0.29526085584163775,
      "grad_norm": 1.75557467960586,
      "learning_rate": 3.3062661958465098e-06,
      "loss": 0.6322,
      "step": 3274
    },
    {
      "epoch": 0.295351039365108,
      "grad_norm": 1.981217763740791,
      "learning_rate": 3.305823750639736e-06,
      "loss": 0.6642,
      "step": 3275
    },
    {
      "epoch": 0.29544122288857827,
      "grad_norm": 1.8172299722052527,
      "learning_rate": 3.3053811940128795e-06,
      "loss": 0.7653,
      "step": 3276
    },
    {
      "epoch": 0.2955314064120485,
      "grad_norm": 1.623728841830243,
      "learning_rate": 3.3049385260037016e-06,
      "loss": 0.7482,
      "step": 3277
    },
    {
      "epoch": 0.2956215899355188,
      "grad_norm": 2.5394926205697033,
      "learning_rate": 3.3044957466499736e-06,
      "loss": 0.6602,
      "step": 3278
    },
    {
      "epoch": 0.295711773458989,
      "grad_norm": 2.3340287080648348,
      "learning_rate": 3.304052855989475e-06,
      "loss": 0.6896,
      "step": 3279
    },
    {
      "epoch": 0.2958019569824593,
      "grad_norm": 1.9450529607118041,
      "learning_rate": 3.3036098540599966e-06,
      "loss": 0.8309,
      "step": 3280
    },
    {
      "epoch": 0.29589214050592955,
      "grad_norm": 2.735214976312518,
      "learning_rate": 3.3031667408993373e-06,
      "loss": 0.732,
      "step": 3281
    },
    {
      "epoch": 0.29598232402939983,
      "grad_norm": 1.7916739116196936,
      "learning_rate": 3.302723516545306e-06,
      "loss": 0.765,
      "step": 3282
    },
    {
      "epoch": 0.29607250755287007,
      "grad_norm": 2.526569607547159,
      "learning_rate": 3.302280181035722e-06,
      "loss": 0.795,
      "step": 3283
    },
    {
      "epoch": 0.29616269107634036,
      "grad_norm": 0.6073947667392906,
      "learning_rate": 3.3018367344084117e-06,
      "loss": 0.6183,
      "step": 3284
    },
    {
      "epoch": 0.2962528745998106,
      "grad_norm": 1.9939893764186065,
      "learning_rate": 3.3013931767012125e-06,
      "loss": 0.7584,
      "step": 3285
    },
    {
      "epoch": 0.2963430581232809,
      "grad_norm": 1.680459488293864,
      "learning_rate": 3.300949507951972e-06,
      "loss": 0.7605,
      "step": 3286
    },
    {
      "epoch": 0.2964332416467511,
      "grad_norm": 2.3486883398776426,
      "learning_rate": 3.300505728198546e-06,
      "loss": 0.8152,
      "step": 3287
    },
    {
      "epoch": 0.2965234251702214,
      "grad_norm": 2.6220013884679636,
      "learning_rate": 3.3000618374788e-06,
      "loss": 0.7562,
      "step": 3288
    },
    {
      "epoch": 0.2966136086936917,
      "grad_norm": 1.5123897512309106,
      "learning_rate": 3.2996178358306104e-06,
      "loss": 0.806,
      "step": 3289
    },
    {
      "epoch": 0.2967037922171619,
      "grad_norm": 1.496061001013345,
      "learning_rate": 3.2991737232918606e-06,
      "loss": 0.7344,
      "step": 3290
    },
    {
      "epoch": 0.2967939757406322,
      "grad_norm": 2.033624874643262,
      "learning_rate": 3.298729499900445e-06,
      "loss": 0.7235,
      "step": 3291
    },
    {
      "epoch": 0.29688415926410244,
      "grad_norm": 3.9204381005064146,
      "learning_rate": 3.2982851656942677e-06,
      "loss": 0.7193,
      "step": 3292
    },
    {
      "epoch": 0.29697434278757273,
      "grad_norm": 1.3780659392626802,
      "learning_rate": 3.2978407207112416e-06,
      "loss": 0.7926,
      "step": 3293
    },
    {
      "epoch": 0.29706452631104296,
      "grad_norm": 2.1352019287315054,
      "learning_rate": 3.2973961649892888e-06,
      "loss": 0.7734,
      "step": 3294
    },
    {
      "epoch": 0.29715470983451325,
      "grad_norm": 1.658890113461051,
      "learning_rate": 3.296951498566341e-06,
      "loss": 0.8335,
      "step": 3295
    },
    {
      "epoch": 0.2972448933579835,
      "grad_norm": 2.361062059329265,
      "learning_rate": 3.2965067214803404e-06,
      "loss": 0.7668,
      "step": 3296
    },
    {
      "epoch": 0.2973350768814538,
      "grad_norm": 1.2288096065557994,
      "learning_rate": 3.2960618337692372e-06,
      "loss": 0.8012,
      "step": 3297
    },
    {
      "epoch": 0.297425260404924,
      "grad_norm": 1.601337100659549,
      "learning_rate": 3.2956168354709927e-06,
      "loss": 0.7294,
      "step": 3298
    },
    {
      "epoch": 0.2975154439283943,
      "grad_norm": 3.0828280851160628,
      "learning_rate": 3.2951717266235754e-06,
      "loss": 0.8204,
      "step": 3299
    },
    {
      "epoch": 0.29760562745186453,
      "grad_norm": 1.8492036755184718,
      "learning_rate": 3.294726507264964e-06,
      "loss": 0.6441,
      "step": 3300
    },
    {
      "epoch": 0.2976958109753348,
      "grad_norm": 2.947826658207015,
      "learning_rate": 3.2942811774331487e-06,
      "loss": 0.7399,
      "step": 3301
    },
    {
      "epoch": 0.29778599449880505,
      "grad_norm": 2.1953120655860334,
      "learning_rate": 3.293835737166127e-06,
      "loss": 0.7699,
      "step": 3302
    },
    {
      "epoch": 0.29787617802227534,
      "grad_norm": 1.601192450222501,
      "learning_rate": 3.293390186501906e-06,
      "loss": 0.718,
      "step": 3303
    },
    {
      "epoch": 0.2979663615457456,
      "grad_norm": 1.7529825271675727,
      "learning_rate": 3.2929445254785024e-06,
      "loss": 0.7796,
      "step": 3304
    },
    {
      "epoch": 0.29805654506921586,
      "grad_norm": 1.9813982044709706,
      "learning_rate": 3.2924987541339423e-06,
      "loss": 0.7283,
      "step": 3305
    },
    {
      "epoch": 0.2981467285926861,
      "grad_norm": 2.8143239358957963,
      "learning_rate": 3.292052872506262e-06,
      "loss": 0.7888,
      "step": 3306
    },
    {
      "epoch": 0.2982369121161564,
      "grad_norm": 1.6400790078042613,
      "learning_rate": 3.291606880633506e-06,
      "loss": 0.8076,
      "step": 3307
    },
    {
      "epoch": 0.2983270956396266,
      "grad_norm": 2.318480722261255,
      "learning_rate": 3.2911607785537297e-06,
      "loss": 0.8208,
      "step": 3308
    },
    {
      "epoch": 0.2984172791630969,
      "grad_norm": 2.1341857960340302,
      "learning_rate": 3.290714566304997e-06,
      "loss": 0.8565,
      "step": 3309
    },
    {
      "epoch": 0.29850746268656714,
      "grad_norm": 1.9736201886618479,
      "learning_rate": 3.2902682439253794e-06,
      "loss": 0.8317,
      "step": 3310
    },
    {
      "epoch": 0.2985976462100374,
      "grad_norm": 1.985335830798776,
      "learning_rate": 3.289821811452961e-06,
      "loss": 0.7848,
      "step": 3311
    },
    {
      "epoch": 0.2986878297335077,
      "grad_norm": 1.8335833451276937,
      "learning_rate": 3.289375268925834e-06,
      "loss": 0.8244,
      "step": 3312
    },
    {
      "epoch": 0.29877801325697795,
      "grad_norm": 10.4328632220386,
      "learning_rate": 3.288928616382099e-06,
      "loss": 0.7557,
      "step": 3313
    },
    {
      "epoch": 0.29886819678044824,
      "grad_norm": 1.4369942978046983,
      "learning_rate": 3.288481853859868e-06,
      "loss": 0.8035,
      "step": 3314
    },
    {
      "epoch": 0.29895838030391847,
      "grad_norm": 1.4915981227250177,
      "learning_rate": 3.2880349813972604e-06,
      "loss": 0.7426,
      "step": 3315
    },
    {
      "epoch": 0.29904856382738876,
      "grad_norm": 1.6510184092844091,
      "learning_rate": 3.2875879990324052e-06,
      "loss": 0.7679,
      "step": 3316
    },
    {
      "epoch": 0.299138747350859,
      "grad_norm": 1.728227597824665,
      "learning_rate": 3.287140906803443e-06,
      "loss": 0.8216,
      "step": 3317
    },
    {
      "epoch": 0.2992289308743293,
      "grad_norm": 2.0460107740638733,
      "learning_rate": 3.2866937047485216e-06,
      "loss": 0.6901,
      "step": 3318
    },
    {
      "epoch": 0.2993191143977995,
      "grad_norm": 2.330324912931478,
      "learning_rate": 3.2862463929057985e-06,
      "loss": 0.829,
      "step": 3319
    },
    {
      "epoch": 0.2994092979212698,
      "grad_norm": 1.9644410344432752,
      "learning_rate": 3.285798971313441e-06,
      "loss": 0.6189,
      "step": 3320
    },
    {
      "epoch": 0.29949948144474003,
      "grad_norm": 1.4600702659472578,
      "learning_rate": 3.2853514400096248e-06,
      "loss": 0.7906,
      "step": 3321
    },
    {
      "epoch": 0.2995896649682103,
      "grad_norm": 1.6542870747569807,
      "learning_rate": 3.2849037990325367e-06,
      "loss": 0.7975,
      "step": 3322
    },
    {
      "epoch": 0.29967984849168056,
      "grad_norm": 1.6776298517035222,
      "learning_rate": 3.2844560484203717e-06,
      "loss": 0.726,
      "step": 3323
    },
    {
      "epoch": 0.29977003201515084,
      "grad_norm": 1.4508586971747044,
      "learning_rate": 3.2840081882113333e-06,
      "loss": 0.8505,
      "step": 3324
    },
    {
      "epoch": 0.2998602155386211,
      "grad_norm": 1.7173659560800258,
      "learning_rate": 3.283560218443638e-06,
      "loss": 0.7414,
      "step": 3325
    },
    {
      "epoch": 0.29995039906209137,
      "grad_norm": 1.437479599517691,
      "learning_rate": 3.2831121391555064e-06,
      "loss": 0.7876,
      "step": 3326
    },
    {
      "epoch": 0.3000405825855616,
      "grad_norm": 1.6592827866501865,
      "learning_rate": 3.2826639503851724e-06,
      "loss": 0.7203,
      "step": 3327
    },
    {
      "epoch": 0.3001307661090319,
      "grad_norm": 1.6542150844275014,
      "learning_rate": 3.282215652170877e-06,
      "loss": 0.8127,
      "step": 3328
    },
    {
      "epoch": 0.3002209496325021,
      "grad_norm": 6.904609588922422,
      "learning_rate": 3.281767244550873e-06,
      "loss": 0.8063,
      "step": 3329
    },
    {
      "epoch": 0.3003111331559724,
      "grad_norm": 0.6038090014406198,
      "learning_rate": 3.2813187275634193e-06,
      "loss": 0.5532,
      "step": 3330
    },
    {
      "epoch": 0.30040131667944264,
      "grad_norm": 1.5915933865093643,
      "learning_rate": 3.280870101246787e-06,
      "loss": 0.7964,
      "step": 3331
    },
    {
      "epoch": 0.30049150020291293,
      "grad_norm": 2.7973239474428513,
      "learning_rate": 3.280421365639255e-06,
      "loss": 0.6849,
      "step": 3332
    },
    {
      "epoch": 0.30058168372638316,
      "grad_norm": 1.6131451343815464,
      "learning_rate": 3.279972520779112e-06,
      "loss": 0.761,
      "step": 3333
    },
    {
      "epoch": 0.30067186724985345,
      "grad_norm": 1.4476314602256368,
      "learning_rate": 3.279523566704656e-06,
      "loss": 0.8347,
      "step": 3334
    },
    {
      "epoch": 0.30076205077332374,
      "grad_norm": 9.717888797002075,
      "learning_rate": 3.2790745034541935e-06,
      "loss": 0.8248,
      "step": 3335
    },
    {
      "epoch": 0.300852234296794,
      "grad_norm": 2.344785436155655,
      "learning_rate": 3.278625331066042e-06,
      "loss": 0.8175,
      "step": 3336
    },
    {
      "epoch": 0.30094241782026426,
      "grad_norm": 1.8541754461644944,
      "learning_rate": 3.278176049578527e-06,
      "loss": 0.8164,
      "step": 3337
    },
    {
      "epoch": 0.3010326013437345,
      "grad_norm": 2.3016440776551614,
      "learning_rate": 3.2777266590299835e-06,
      "loss": 0.7297,
      "step": 3338
    },
    {
      "epoch": 0.3011227848672048,
      "grad_norm": 2.1096120065419024,
      "learning_rate": 3.2772771594587562e-06,
      "loss": 0.7058,
      "step": 3339
    },
    {
      "epoch": 0.301212968390675,
      "grad_norm": 1.665204081929812,
      "learning_rate": 3.2768275509031988e-06,
      "loss": 0.8737,
      "step": 3340
    },
    {
      "epoch": 0.3013031519141453,
      "grad_norm": 1.6414683036567566,
      "learning_rate": 3.276377833401675e-06,
      "loss": 0.7479,
      "step": 3341
    },
    {
      "epoch": 0.30139333543761554,
      "grad_norm": 1.6515936986598372,
      "learning_rate": 3.2759280069925557e-06,
      "loss": 0.8339,
      "step": 3342
    },
    {
      "epoch": 0.30148351896108583,
      "grad_norm": 1.5998014118215533,
      "learning_rate": 3.2754780717142233e-06,
      "loss": 0.8092,
      "step": 3343
    },
    {
      "epoch": 0.30157370248455606,
      "grad_norm": 2.7255916233201622,
      "learning_rate": 3.27502802760507e-06,
      "loss": 0.7449,
      "step": 3344
    },
    {
      "epoch": 0.30166388600802635,
      "grad_norm": 1.8358666588934094,
      "learning_rate": 3.2745778747034943e-06,
      "loss": 0.6837,
      "step": 3345
    },
    {
      "epoch": 0.3017540695314966,
      "grad_norm": 1.8138009368242929,
      "learning_rate": 3.274127613047906e-06,
      "loss": 0.7534,
      "step": 3346
    },
    {
      "epoch": 0.30184425305496687,
      "grad_norm": 2.411404328259901,
      "learning_rate": 3.273677242676725e-06,
      "loss": 0.7899,
      "step": 3347
    },
    {
      "epoch": 0.3019344365784371,
      "grad_norm": 2.7212653089310854,
      "learning_rate": 3.2732267636283782e-06,
      "loss": 0.7696,
      "step": 3348
    },
    {
      "epoch": 0.3020246201019074,
      "grad_norm": 5.864722819953456,
      "learning_rate": 3.2727761759413034e-06,
      "loss": 0.7351,
      "step": 3349
    },
    {
      "epoch": 0.3021148036253776,
      "grad_norm": 2.543451266356372,
      "learning_rate": 3.2723254796539477e-06,
      "loss": 0.7903,
      "step": 3350
    },
    {
      "epoch": 0.3022049871488479,
      "grad_norm": 1.7923155947523937,
      "learning_rate": 3.271874674804766e-06,
      "loss": 0.7578,
      "step": 3351
    },
    {
      "epoch": 0.30229517067231815,
      "grad_norm": 1.7514915240945985,
      "learning_rate": 3.2714237614322242e-06,
      "loss": 0.7979,
      "step": 3352
    },
    {
      "epoch": 0.30238535419578844,
      "grad_norm": 1.8947909147720006,
      "learning_rate": 3.2709727395747974e-06,
      "loss": 0.6032,
      "step": 3353
    },
    {
      "epoch": 0.30247553771925867,
      "grad_norm": 0.6758856720262612,
      "learning_rate": 3.2705216092709673e-06,
      "loss": 0.5725,
      "step": 3354
    },
    {
      "epoch": 0.30256572124272896,
      "grad_norm": 2.1217171611705496,
      "learning_rate": 3.2700703705592282e-06,
      "loss": 0.8636,
      "step": 3355
    },
    {
      "epoch": 0.3026559047661992,
      "grad_norm": 1.9431082251736647,
      "learning_rate": 3.269619023478082e-06,
      "loss": 0.6782,
      "step": 3356
    },
    {
      "epoch": 0.3027460882896695,
      "grad_norm": 1.6072344572212003,
      "learning_rate": 3.26916756806604e-06,
      "loss": 0.8709,
      "step": 3357
    },
    {
      "epoch": 0.3028362718131397,
      "grad_norm": 1.7918138295733919,
      "learning_rate": 3.268716004361623e-06,
      "loss": 0.7928,
      "step": 3358
    },
    {
      "epoch": 0.30292645533661,
      "grad_norm": 0.6625695219089738,
      "learning_rate": 3.268264332403361e-06,
      "loss": 0.5879,
      "step": 3359
    },
    {
      "epoch": 0.3030166388600803,
      "grad_norm": 1.4455774837589659,
      "learning_rate": 3.2678125522297933e-06,
      "loss": 0.8048,
      "step": 3360
    },
    {
      "epoch": 0.3031068223835505,
      "grad_norm": 2.1321032116725984,
      "learning_rate": 3.267360663879468e-06,
      "loss": 0.7141,
      "step": 3361
    },
    {
      "epoch": 0.3031970059070208,
      "grad_norm": 1.6305924796812632,
      "learning_rate": 3.266908667390942e-06,
      "loss": 0.7653,
      "step": 3362
    },
    {
      "epoch": 0.30328718943049104,
      "grad_norm": 1.9499561696139551,
      "learning_rate": 3.2664565628027833e-06,
      "loss": 0.8258,
      "step": 3363
    },
    {
      "epoch": 0.30337737295396133,
      "grad_norm": 3.1171653192252045,
      "learning_rate": 3.2660043501535675e-06,
      "loss": 0.7692,
      "step": 3364
    },
    {
      "epoch": 0.30346755647743157,
      "grad_norm": 2.0914280100485936,
      "learning_rate": 3.2655520294818797e-06,
      "loss": 0.8454,
      "step": 3365
    },
    {
      "epoch": 0.30355774000090185,
      "grad_norm": 4.785411743816965,
      "learning_rate": 3.2650996008263146e-06,
      "loss": 0.8472,
      "step": 3366
    },
    {
      "epoch": 0.3036479235243721,
      "grad_norm": 1.8390568631846627,
      "learning_rate": 3.2646470642254756e-06,
      "loss": 0.7362,
      "step": 3367
    },
    {
      "epoch": 0.3037381070478424,
      "grad_norm": 1.802601587234352,
      "learning_rate": 3.2641944197179767e-06,
      "loss": 0.797,
      "step": 3368
    },
    {
      "epoch": 0.3038282905713126,
      "grad_norm": 0.6641577484037853,
      "learning_rate": 3.2637416673424383e-06,
      "loss": 0.5734,
      "step": 3369
    },
    {
      "epoch": 0.3039184740947829,
      "grad_norm": 3.2541922893484165,
      "learning_rate": 3.2632888071374937e-06,
      "loss": 0.7844,
      "step": 3370
    },
    {
      "epoch": 0.30400865761825313,
      "grad_norm": 2.2429730571092943,
      "learning_rate": 3.2628358391417815e-06,
      "loss": 0.8071,
      "step": 3371
    },
    {
      "epoch": 0.3040988411417234,
      "grad_norm": 2.1035985903729224,
      "learning_rate": 3.2623827633939526e-06,
      "loss": 0.7646,
      "step": 3372
    },
    {
      "epoch": 0.30418902466519365,
      "grad_norm": 2.4362225241601685,
      "learning_rate": 3.2619295799326657e-06,
      "loss": 0.7675,
      "step": 3373
    },
    {
      "epoch": 0.30427920818866394,
      "grad_norm": 1.8333268454465799,
      "learning_rate": 3.2614762887965883e-06,
      "loss": 0.6712,
      "step": 3374
    },
    {
      "epoch": 0.3043693917121342,
      "grad_norm": 1.672882418283916,
      "learning_rate": 3.2610228900243984e-06,
      "loss": 0.7642,
      "step": 3375
    },
    {
      "epoch": 0.30445957523560446,
      "grad_norm": 0.6146936371520668,
      "learning_rate": 3.260569383654783e-06,
      "loss": 0.6241,
      "step": 3376
    },
    {
      "epoch": 0.3045497587590747,
      "grad_norm": 0.6937626055697748,
      "learning_rate": 3.2601157697264365e-06,
      "loss": 0.5941,
      "step": 3377
    },
    {
      "epoch": 0.304639942282545,
      "grad_norm": 1.9662728241252363,
      "learning_rate": 3.2596620482780647e-06,
      "loss": 0.7553,
      "step": 3378
    },
    {
      "epoch": 0.3047301258060152,
      "grad_norm": 0.5727040821203052,
      "learning_rate": 3.2592082193483803e-06,
      "loss": 0.5936,
      "step": 3379
    },
    {
      "epoch": 0.3048203093294855,
      "grad_norm": 1.5097630983161263,
      "learning_rate": 3.258754282976109e-06,
      "loss": 0.7589,
      "step": 3380
    },
    {
      "epoch": 0.30491049285295574,
      "grad_norm": 1.8603400282614877,
      "learning_rate": 3.25830023919998e-06,
      "loss": 0.847,
      "step": 3381
    },
    {
      "epoch": 0.305000676376426,
      "grad_norm": 2.2985675829415175,
      "learning_rate": 3.2578460880587374e-06,
      "loss": 0.7255,
      "step": 3382
    },
    {
      "epoch": 0.3050908598998963,
      "grad_norm": 1.7473388601946078,
      "learning_rate": 3.2573918295911306e-06,
      "loss": 0.8314,
      "step": 3383
    },
    {
      "epoch": 0.30518104342336655,
      "grad_norm": 1.6039137537688268,
      "learning_rate": 3.2569374638359196e-06,
      "loss": 0.7068,
      "step": 3384
    },
    {
      "epoch": 0.30527122694683684,
      "grad_norm": 1.3784568421695989,
      "learning_rate": 3.2564829908318736e-06,
      "loss": 0.7854,
      "step": 3385
    },
    {
      "epoch": 0.30536141047030707,
      "grad_norm": 1.870316887693429,
      "learning_rate": 3.2560284106177705e-06,
      "loss": 0.832,
      "step": 3386
    },
    {
      "epoch": 0.30545159399377736,
      "grad_norm": 4.058424093392184,
      "learning_rate": 3.2555737232323978e-06,
      "loss": 0.7428,
      "step": 3387
    },
    {
      "epoch": 0.3055417775172476,
      "grad_norm": 1.422488824166911,
      "learning_rate": 3.255118928714552e-06,
      "loss": 0.8063,
      "step": 3388
    },
    {
      "epoch": 0.3056319610407179,
      "grad_norm": 2.144046360361557,
      "learning_rate": 3.2546640271030386e-06,
      "loss": 0.7016,
      "step": 3389
    },
    {
      "epoch": 0.3057221445641881,
      "grad_norm": 0.6365461993295698,
      "learning_rate": 3.2542090184366717e-06,
      "loss": 0.6249,
      "step": 3390
    },
    {
      "epoch": 0.3058123280876584,
      "grad_norm": 2.5064249448529092,
      "learning_rate": 3.253753902754276e-06,
      "loss": 0.8569,
      "step": 3391
    },
    {
      "epoch": 0.30590251161112864,
      "grad_norm": 1.5710715362025514,
      "learning_rate": 3.253298680094685e-06,
      "loss": 0.7056,
      "step": 3392
    },
    {
      "epoch": 0.3059926951345989,
      "grad_norm": 1.8348998038770248,
      "learning_rate": 3.2528433504967394e-06,
      "loss": 0.7191,
      "step": 3393
    },
    {
      "epoch": 0.30608287865806916,
      "grad_norm": 1.6761504080114946,
      "learning_rate": 3.252387913999291e-06,
      "loss": 0.7848,
      "step": 3394
    },
    {
      "epoch": 0.30617306218153945,
      "grad_norm": 1.6291069071834907,
      "learning_rate": 3.2519323706411998e-06,
      "loss": 0.7776,
      "step": 3395
    },
    {
      "epoch": 0.3062632457050097,
      "grad_norm": 1.396699176620882,
      "learning_rate": 3.251476720461336e-06,
      "loss": 0.6632,
      "step": 3396
    },
    {
      "epoch": 0.30635342922847997,
      "grad_norm": 2.3491411384674064,
      "learning_rate": 3.251020963498578e-06,
      "loss": 0.7057,
      "step": 3397
    },
    {
      "epoch": 0.3064436127519502,
      "grad_norm": 1.5015479683545017,
      "learning_rate": 3.250565099791813e-06,
      "loss": 0.8145,
      "step": 3398
    },
    {
      "epoch": 0.3065337962754205,
      "grad_norm": 1.4302568813159102,
      "learning_rate": 3.2501091293799387e-06,
      "loss": 0.8273,
      "step": 3399
    },
    {
      "epoch": 0.3066239797988907,
      "grad_norm": 2.125788934897734,
      "learning_rate": 3.24965305230186e-06,
      "loss": 0.7875,
      "step": 3400
    },
    {
      "epoch": 0.306714163322361,
      "grad_norm": 2.0496924425482246,
      "learning_rate": 3.249196868596492e-06,
      "loss": 0.8422,
      "step": 3401
    },
    {
      "epoch": 0.30680434684583124,
      "grad_norm": 2.136711162435111,
      "learning_rate": 3.24874057830276e-06,
      "loss": 0.7036,
      "step": 3402
    },
    {
      "epoch": 0.30689453036930153,
      "grad_norm": 1.6701083252089155,
      "learning_rate": 3.2482841814595954e-06,
      "loss": 0.6957,
      "step": 3403
    },
    {
      "epoch": 0.30698471389277177,
      "grad_norm": 1.870451305213413,
      "learning_rate": 3.247827678105943e-06,
      "loss": 0.7162,
      "step": 3404
    },
    {
      "epoch": 0.30707489741624205,
      "grad_norm": 1.9534418688273514,
      "learning_rate": 3.247371068280751e-06,
      "loss": 0.769,
      "step": 3405
    },
    {
      "epoch": 0.3071650809397123,
      "grad_norm": 1.3691606506537441,
      "learning_rate": 3.2469143520229823e-06,
      "loss": 0.6349,
      "step": 3406
    },
    {
      "epoch": 0.3072552644631826,
      "grad_norm": 1.4264649306625221,
      "learning_rate": 3.2464575293716054e-06,
      "loss": 0.7229,
      "step": 3407
    },
    {
      "epoch": 0.30734544798665286,
      "grad_norm": 2.289967328800539,
      "learning_rate": 3.2460006003655997e-06,
      "loss": 0.7125,
      "step": 3408
    },
    {
      "epoch": 0.3074356315101231,
      "grad_norm": 2.8594061844589636,
      "learning_rate": 3.245543565043952e-06,
      "loss": 0.8293,
      "step": 3409
    },
    {
      "epoch": 0.3075258150335934,
      "grad_norm": 2.138756709397637,
      "learning_rate": 3.2450864234456592e-06,
      "loss": 0.7688,
      "step": 3410
    },
    {
      "epoch": 0.3076159985570636,
      "grad_norm": 1.930644859639106,
      "learning_rate": 3.244629175609728e-06,
      "loss": 0.7647,
      "step": 3411
    },
    {
      "epoch": 0.3077061820805339,
      "grad_norm": 1.5021917701693828,
      "learning_rate": 3.2441718215751726e-06,
      "loss": 0.7589,
      "step": 3412
    },
    {
      "epoch": 0.30779636560400414,
      "grad_norm": 1.7953493358528285,
      "learning_rate": 3.2437143613810173e-06,
      "loss": 0.736,
      "step": 3413
    },
    {
      "epoch": 0.30788654912747443,
      "grad_norm": 1.6104482386933452,
      "learning_rate": 3.2432567950662947e-06,
      "loss": 0.8156,
      "step": 3414
    },
    {
      "epoch": 0.30797673265094466,
      "grad_norm": 1.5047509929693634,
      "learning_rate": 3.2427991226700468e-06,
      "loss": 0.7568,
      "step": 3415
    },
    {
      "epoch": 0.30806691617441495,
      "grad_norm": 1.7592593193983697,
      "learning_rate": 3.2423413442313246e-06,
      "loss": 0.6956,
      "step": 3416
    },
    {
      "epoch": 0.3081570996978852,
      "grad_norm": 1.8002561519408031,
      "learning_rate": 3.2418834597891904e-06,
      "loss": 0.692,
      "step": 3417
    },
    {
      "epoch": 0.3082472832213555,
      "grad_norm": 2.050456749810397,
      "learning_rate": 3.2414254693827098e-06,
      "loss": 0.8551,
      "step": 3418
    },
    {
      "epoch": 0.3083374667448257,
      "grad_norm": 1.5928422175092425,
      "learning_rate": 3.2409673730509644e-06,
      "loss": 0.7521,
      "step": 3419
    },
    {
      "epoch": 0.308427650268296,
      "grad_norm": 2.3221440648377802,
      "learning_rate": 3.2405091708330393e-06,
      "loss": 0.7414,
      "step": 3420
    },
    {
      "epoch": 0.3085178337917662,
      "grad_norm": 6.670952404833123,
      "learning_rate": 3.2400508627680323e-06,
      "loss": 0.7632,
      "step": 3421
    },
    {
      "epoch": 0.3086080173152365,
      "grad_norm": 10.206820462806277,
      "learning_rate": 3.2395924488950474e-06,
      "loss": 0.8599,
      "step": 3422
    },
    {
      "epoch": 0.30869820083870675,
      "grad_norm": 1.825115563378629,
      "learning_rate": 3.2391339292532004e-06,
      "loss": 0.8963,
      "step": 3423
    },
    {
      "epoch": 0.30878838436217704,
      "grad_norm": 2.7574682263975925,
      "learning_rate": 3.238675303881614e-06,
      "loss": 0.7229,
      "step": 3424
    },
    {
      "epoch": 0.30887856788564727,
      "grad_norm": 1.7773152443412914,
      "learning_rate": 3.2382165728194203e-06,
      "loss": 0.7411,
      "step": 3425
    },
    {
      "epoch": 0.30896875140911756,
      "grad_norm": 1.5432165019399389,
      "learning_rate": 3.237757736105761e-06,
      "loss": 0.752,
      "step": 3426
    },
    {
      "epoch": 0.3090589349325878,
      "grad_norm": 1.6794845813339874,
      "learning_rate": 3.2372987937797867e-06,
      "loss": 0.8488,
      "step": 3427
    },
    {
      "epoch": 0.3091491184560581,
      "grad_norm": 1.3873810811674667,
      "learning_rate": 3.2368397458806573e-06,
      "loss": 0.6719,
      "step": 3428
    },
    {
      "epoch": 0.3092393019795283,
      "grad_norm": 1.8400482906348643,
      "learning_rate": 3.2363805924475412e-06,
      "loss": 0.8023,
      "step": 3429
    },
    {
      "epoch": 0.3093294855029986,
      "grad_norm": 1.8256461789301042,
      "learning_rate": 3.2359213335196153e-06,
      "loss": 0.7132,
      "step": 3430
    },
    {
      "epoch": 0.3094196690264689,
      "grad_norm": 2.679807877130616,
      "learning_rate": 3.2354619691360663e-06,
      "loss": 0.7216,
      "step": 3431
    },
    {
      "epoch": 0.3095098525499391,
      "grad_norm": 1.4013392854365123,
      "learning_rate": 3.2350024993360898e-06,
      "loss": 0.8091,
      "step": 3432
    },
    {
      "epoch": 0.3096000360734094,
      "grad_norm": 1.733058945627931,
      "learning_rate": 3.2345429241588902e-06,
      "loss": 0.765,
      "step": 3433
    },
    {
      "epoch": 0.30969021959687965,
      "grad_norm": 2.4887664178204236,
      "learning_rate": 3.234083243643681e-06,
      "loss": 0.7944,
      "step": 3434
    },
    {
      "epoch": 0.30978040312034993,
      "grad_norm": 1.469140751581813,
      "learning_rate": 3.233623457829686e-06,
      "loss": 0.837,
      "step": 3435
    },
    {
      "epoch": 0.30987058664382017,
      "grad_norm": 2.5556304777837213,
      "learning_rate": 3.2331635667561344e-06,
      "loss": 0.7428,
      "step": 3436
    },
    {
      "epoch": 0.30996077016729046,
      "grad_norm": 1.764563114571362,
      "learning_rate": 3.2327035704622674e-06,
      "loss": 0.7977,
      "step": 3437
    },
    {
      "epoch": 0.3100509536907607,
      "grad_norm": 1.3921331307255191,
      "learning_rate": 3.2322434689873353e-06,
      "loss": 0.6616,
      "step": 3438
    },
    {
      "epoch": 0.310141137214231,
      "grad_norm": 1.7170964696773996,
      "learning_rate": 3.2317832623705957e-06,
      "loss": 0.8482,
      "step": 3439
    },
    {
      "epoch": 0.3102313207377012,
      "grad_norm": 2.0666725261153966,
      "learning_rate": 3.231322950651316e-06,
      "loss": 0.7139,
      "step": 3440
    },
    {
      "epoch": 0.3103215042611715,
      "grad_norm": 1.8971998788061872,
      "learning_rate": 3.2308625338687735e-06,
      "loss": 0.7932,
      "step": 3441
    },
    {
      "epoch": 0.31041168778464173,
      "grad_norm": 2.9217075565114943,
      "learning_rate": 3.230402012062252e-06,
      "loss": 0.8813,
      "step": 3442
    },
    {
      "epoch": 0.310501871308112,
      "grad_norm": 0.6171380276710208,
      "learning_rate": 3.2299413852710466e-06,
      "loss": 0.6138,
      "step": 3443
    },
    {
      "epoch": 0.31059205483158225,
      "grad_norm": 1.691984857347153,
      "learning_rate": 3.2294806535344606e-06,
      "loss": 0.7314,
      "step": 3444
    },
    {
      "epoch": 0.31068223835505254,
      "grad_norm": 2.035432353849164,
      "learning_rate": 3.2290198168918056e-06,
      "loss": 0.8099,
      "step": 3445
    },
    {
      "epoch": 0.3107724218785228,
      "grad_norm": 4.289361106052988,
      "learning_rate": 3.2285588753824035e-06,
      "loss": 0.8062,
      "step": 3446
    },
    {
      "epoch": 0.31086260540199306,
      "grad_norm": 1.7968226549568238,
      "learning_rate": 3.228097829045584e-06,
      "loss": 0.7804,
      "step": 3447
    },
    {
      "epoch": 0.3109527889254633,
      "grad_norm": 1.3481724289574344,
      "learning_rate": 3.227636677920685e-06,
      "loss": 0.8102,
      "step": 3448
    },
    {
      "epoch": 0.3110429724489336,
      "grad_norm": 1.8054339693978938,
      "learning_rate": 3.2271754220470567e-06,
      "loss": 0.8158,
      "step": 3449
    },
    {
      "epoch": 0.3111331559724038,
      "grad_norm": 1.9218911891348442,
      "learning_rate": 3.2267140614640547e-06,
      "loss": 0.8512,
      "step": 3450
    },
    {
      "epoch": 0.3112233394958741,
      "grad_norm": 1.7033555460985121,
      "learning_rate": 3.2262525962110445e-06,
      "loss": 0.8162,
      "step": 3451
    },
    {
      "epoch": 0.31131352301934434,
      "grad_norm": 1.9159423316368265,
      "learning_rate": 3.2257910263274015e-06,
      "loss": 0.6971,
      "step": 3452
    },
    {
      "epoch": 0.31140370654281463,
      "grad_norm": 2.7627415685196235,
      "learning_rate": 3.225329351852509e-06,
      "loss": 0.7166,
      "step": 3453
    },
    {
      "epoch": 0.3114938900662849,
      "grad_norm": 1.9342223099453075,
      "learning_rate": 3.2248675728257596e-06,
      "loss": 0.772,
      "step": 3454
    },
    {
      "epoch": 0.31158407358975515,
      "grad_norm": 1.5870232835009912,
      "learning_rate": 3.2244056892865557e-06,
      "loss": 0.8428,
      "step": 3455
    },
    {
      "epoch": 0.31167425711322544,
      "grad_norm": 1.60222437138148,
      "learning_rate": 3.2239437012743063e-06,
      "loss": 0.7603,
      "step": 3456
    },
    {
      "epoch": 0.31176444063669567,
      "grad_norm": 1.8592816497706388,
      "learning_rate": 3.223481608828432e-06,
      "loss": 0.7845,
      "step": 3457
    },
    {
      "epoch": 0.31185462416016596,
      "grad_norm": 1.978210666723923,
      "learning_rate": 3.223019411988361e-06,
      "loss": 0.7896,
      "step": 3458
    },
    {
      "epoch": 0.3119448076836362,
      "grad_norm": 1.6697271936272422,
      "learning_rate": 3.22255711079353e-06,
      "loss": 0.7687,
      "step": 3459
    },
    {
      "epoch": 0.3120349912071065,
      "grad_norm": 1.5258077327150805,
      "learning_rate": 3.222094705283385e-06,
      "loss": 0.7291,
      "step": 3460
    },
    {
      "epoch": 0.3121251747305767,
      "grad_norm": 2.1644978550129053,
      "learning_rate": 3.2216321954973805e-06,
      "loss": 0.7662,
      "step": 3461
    },
    {
      "epoch": 0.312215358254047,
      "grad_norm": 2.269068028882353,
      "learning_rate": 3.2211695814749816e-06,
      "loss": 0.764,
      "step": 3462
    },
    {
      "epoch": 0.31230554177751724,
      "grad_norm": 2.4907185880339497,
      "learning_rate": 3.220706863255661e-06,
      "loss": 0.88,
      "step": 3463
    },
    {
      "epoch": 0.3123957253009875,
      "grad_norm": 2.453081847376226,
      "learning_rate": 3.2202440408788994e-06,
      "loss": 0.7684,
      "step": 3464
    },
    {
      "epoch": 0.31248590882445776,
      "grad_norm": 1.7838083600105505,
      "learning_rate": 3.2197811143841883e-06,
      "loss": 0.8404,
      "step": 3465
    },
    {
      "epoch": 0.31257609234792805,
      "grad_norm": 1.6986120785266572,
      "learning_rate": 3.2193180838110267e-06,
      "loss": 0.764,
      "step": 3466
    },
    {
      "epoch": 0.3126662758713983,
      "grad_norm": 2.821373483696048,
      "learning_rate": 3.2188549491989225e-06,
      "loss": 0.6647,
      "step": 3467
    },
    {
      "epoch": 0.31275645939486857,
      "grad_norm": 1.7277068061535137,
      "learning_rate": 3.2183917105873934e-06,
      "loss": 0.7933,
      "step": 3468
    },
    {
      "epoch": 0.3128466429183388,
      "grad_norm": 1.7784104049460603,
      "learning_rate": 3.217928368015966e-06,
      "loss": 0.8274,
      "step": 3469
    },
    {
      "epoch": 0.3129368264418091,
      "grad_norm": 1.4614916887598393,
      "learning_rate": 3.217464921524174e-06,
      "loss": 0.7984,
      "step": 3470
    },
    {
      "epoch": 0.3130270099652793,
      "grad_norm": 1.6188385132851693,
      "learning_rate": 3.2170013711515616e-06,
      "loss": 0.7834,
      "step": 3471
    },
    {
      "epoch": 0.3131171934887496,
      "grad_norm": 2.1575894825436803,
      "learning_rate": 3.216537716937682e-06,
      "loss": 0.6943,
      "step": 3472
    },
    {
      "epoch": 0.31320737701221985,
      "grad_norm": 1.799659230506505,
      "learning_rate": 3.2160739589220968e-06,
      "loss": 0.7448,
      "step": 3473
    },
    {
      "epoch": 0.31329756053569013,
      "grad_norm": 1.5548057319502089,
      "learning_rate": 3.215610097144376e-06,
      "loss": 0.825,
      "step": 3474
    },
    {
      "epoch": 0.31338774405916037,
      "grad_norm": 1.9883093691413218,
      "learning_rate": 3.215146131644099e-06,
      "loss": 0.7161,
      "step": 3475
    },
    {
      "epoch": 0.31347792758263066,
      "grad_norm": 1.7192286431807025,
      "learning_rate": 3.214682062460854e-06,
      "loss": 0.8443,
      "step": 3476
    },
    {
      "epoch": 0.3135681111061009,
      "grad_norm": 2.0598873583542536,
      "learning_rate": 3.2142178896342367e-06,
      "loss": 0.689,
      "step": 3477
    },
    {
      "epoch": 0.3136582946295712,
      "grad_norm": 1.6144760916585632,
      "learning_rate": 3.2137536132038552e-06,
      "loss": 0.7567,
      "step": 3478
    },
    {
      "epoch": 0.31374847815304147,
      "grad_norm": 1.8821151518340167,
      "learning_rate": 3.2132892332093226e-06,
      "loss": 0.7725,
      "step": 3479
    },
    {
      "epoch": 0.3138386616765117,
      "grad_norm": 1.7334733285978423,
      "learning_rate": 3.2128247496902623e-06,
      "loss": 0.8848,
      "step": 3480
    },
    {
      "epoch": 0.313928845199982,
      "grad_norm": 1.8704144672333562,
      "learning_rate": 3.2123601626863064e-06,
      "loss": 0.7849,
      "step": 3481
    },
    {
      "epoch": 0.3140190287234522,
      "grad_norm": 0.6416929925951208,
      "learning_rate": 3.2118954722370974e-06,
      "loss": 0.5713,
      "step": 3482
    },
    {
      "epoch": 0.3141092122469225,
      "grad_norm": 1.9384996850006344,
      "learning_rate": 3.2114306783822837e-06,
      "loss": 0.7219,
      "step": 3483
    },
    {
      "epoch": 0.31419939577039274,
      "grad_norm": 1.850467470750509,
      "learning_rate": 3.210965781161525e-06,
      "loss": 0.7444,
      "step": 3484
    },
    {
      "epoch": 0.31428957929386303,
      "grad_norm": 1.9180598033416336,
      "learning_rate": 3.2105007806144892e-06,
      "loss": 0.8831,
      "step": 3485
    },
    {
      "epoch": 0.31437976281733326,
      "grad_norm": 1.855578417297714,
      "learning_rate": 3.2100356767808513e-06,
      "loss": 0.7685,
      "step": 3486
    },
    {
      "epoch": 0.31446994634080355,
      "grad_norm": 1.9395656649081516,
      "learning_rate": 3.2095704697002977e-06,
      "loss": 0.8322,
      "step": 3487
    },
    {
      "epoch": 0.3145601298642738,
      "grad_norm": 1.381545181148605,
      "learning_rate": 3.209105159412522e-06,
      "loss": 0.7697,
      "step": 3488
    },
    {
      "epoch": 0.3146503133877441,
      "grad_norm": 3.0993296852191548,
      "learning_rate": 3.208639745957228e-06,
      "loss": 0.7416,
      "step": 3489
    },
    {
      "epoch": 0.3147404969112143,
      "grad_norm": 1.460608868687373,
      "learning_rate": 3.2081742293741256e-06,
      "loss": 0.7981,
      "step": 3490
    },
    {
      "epoch": 0.3148306804346846,
      "grad_norm": 1.4836180513710817,
      "learning_rate": 3.2077086097029366e-06,
      "loss": 0.833,
      "step": 3491
    },
    {
      "epoch": 0.31492086395815483,
      "grad_norm": 1.955778958583732,
      "learning_rate": 3.2072428869833895e-06,
      "loss": 0.7722,
      "step": 3492
    },
    {
      "epoch": 0.3150110474816251,
      "grad_norm": 4.056885345967661,
      "learning_rate": 3.206777061255223e-06,
      "loss": 0.7869,
      "step": 3493
    },
    {
      "epoch": 0.31510123100509535,
      "grad_norm": 1.865593073841704,
      "learning_rate": 3.206311132558183e-06,
      "loss": 0.8085,
      "step": 3494
    },
    {
      "epoch": 0.31519141452856564,
      "grad_norm": 2.0080554384732343,
      "learning_rate": 3.205845100932026e-06,
      "loss": 0.7858,
      "step": 3495
    },
    {
      "epoch": 0.31528159805203587,
      "grad_norm": 1.6193466147522948,
      "learning_rate": 3.205378966416516e-06,
      "loss": 0.8017,
      "step": 3496
    },
    {
      "epoch": 0.31537178157550616,
      "grad_norm": 3.0896522279355976,
      "learning_rate": 3.204912729051426e-06,
      "loss": 0.7692,
      "step": 3497
    },
    {
      "epoch": 0.3154619650989764,
      "grad_norm": 2.0692670377716835,
      "learning_rate": 3.2044463888765384e-06,
      "loss": 0.7449,
      "step": 3498
    },
    {
      "epoch": 0.3155521486224467,
      "grad_norm": 1.8894874997075009,
      "learning_rate": 3.2039799459316436e-06,
      "loss": 0.6765,
      "step": 3499
    },
    {
      "epoch": 0.3156423321459169,
      "grad_norm": 1.9733411147741737,
      "learning_rate": 3.2035134002565407e-06,
      "loss": 0.8054,
      "step": 3500
    },
    {
      "epoch": 0.3157325156693872,
      "grad_norm": 1.966530108425279,
      "learning_rate": 3.203046751891039e-06,
      "loss": 0.7857,
      "step": 3501
    },
    {
      "epoch": 0.3158226991928575,
      "grad_norm": 1.8010643593759719,
      "learning_rate": 3.2025800008749545e-06,
      "loss": 0.7866,
      "step": 3502
    },
    {
      "epoch": 0.3159128827163277,
      "grad_norm": 2.139044742217938,
      "learning_rate": 3.202113147248114e-06,
      "loss": 0.7512,
      "step": 3503
    },
    {
      "epoch": 0.316003066239798,
      "grad_norm": 2.1241199690708084,
      "learning_rate": 3.20164619105035e-06,
      "loss": 0.745,
      "step": 3504
    },
    {
      "epoch": 0.31609324976326825,
      "grad_norm": 1.5197093241475894,
      "learning_rate": 3.201179132321508e-06,
      "loss": 0.6263,
      "step": 3505
    },
    {
      "epoch": 0.31618343328673854,
      "grad_norm": 1.645266189720445,
      "learning_rate": 3.200711971101439e-06,
      "loss": 0.7157,
      "step": 3506
    },
    {
      "epoch": 0.31627361681020877,
      "grad_norm": 2.490850682007284,
      "learning_rate": 3.2002447074300047e-06,
      "loss": 0.7736,
      "step": 3507
    },
    {
      "epoch": 0.31636380033367906,
      "grad_norm": 1.7756122580866311,
      "learning_rate": 3.1997773413470736e-06,
      "loss": 0.7771,
      "step": 3508
    },
    {
      "epoch": 0.3164539838571493,
      "grad_norm": 5.399448945927514,
      "learning_rate": 3.199309872892524e-06,
      "loss": 0.7663,
      "step": 3509
    },
    {
      "epoch": 0.3165441673806196,
      "grad_norm": 1.9693833800645726,
      "learning_rate": 3.198842302106243e-06,
      "loss": 0.8235,
      "step": 3510
    },
    {
      "epoch": 0.3166343509040898,
      "grad_norm": 2.0920545631207763,
      "learning_rate": 3.1983746290281265e-06,
      "loss": 0.7525,
      "step": 3511
    },
    {
      "epoch": 0.3167245344275601,
      "grad_norm": 1.3469437194428568,
      "learning_rate": 3.197906853698079e-06,
      "loss": 0.782,
      "step": 3512
    },
    {
      "epoch": 0.31681471795103033,
      "grad_norm": 1.5707214733874524,
      "learning_rate": 3.1974389761560137e-06,
      "loss": 0.7234,
      "step": 3513
    },
    {
      "epoch": 0.3169049014745006,
      "grad_norm": 1.9180348184863973,
      "learning_rate": 3.1969709964418525e-06,
      "loss": 0.7253,
      "step": 3514
    },
    {
      "epoch": 0.31699508499797086,
      "grad_norm": 2.7478523538007957,
      "learning_rate": 3.196502914595525e-06,
      "loss": 0.7932,
      "step": 3515
    },
    {
      "epoch": 0.31708526852144114,
      "grad_norm": 1.6617162219316615,
      "learning_rate": 3.1960347306569723e-06,
      "loss": 0.7922,
      "step": 3516
    },
    {
      "epoch": 0.3171754520449114,
      "grad_norm": 2.287759038714517,
      "learning_rate": 3.195566444666141e-06,
      "loss": 0.7232,
      "step": 3517
    },
    {
      "epoch": 0.31726563556838167,
      "grad_norm": 1.708946916442159,
      "learning_rate": 3.1950980566629886e-06,
      "loss": 0.8806,
      "step": 3518
    },
    {
      "epoch": 0.3173558190918519,
      "grad_norm": 1.5621566395192394,
      "learning_rate": 3.1946295666874797e-06,
      "loss": 0.7888,
      "step": 3519
    },
    {
      "epoch": 0.3174460026153222,
      "grad_norm": 1.5557412261887127,
      "learning_rate": 3.19416097477959e-06,
      "loss": 0.7945,
      "step": 3520
    },
    {
      "epoch": 0.3175361861387924,
      "grad_norm": 4.4100027133647846,
      "learning_rate": 3.1936922809793005e-06,
      "loss": 0.8314,
      "step": 3521
    },
    {
      "epoch": 0.3176263696622627,
      "grad_norm": 4.219871993451138,
      "learning_rate": 3.193223485326604e-06,
      "loss": 0.8916,
      "step": 3522
    },
    {
      "epoch": 0.31771655318573294,
      "grad_norm": 2.8310410445622463,
      "learning_rate": 3.1927545878615005e-06,
      "loss": 0.8564,
      "step": 3523
    },
    {
      "epoch": 0.31780673670920323,
      "grad_norm": 2.372980363329738,
      "learning_rate": 3.192285588623999e-06,
      "loss": 0.7298,
      "step": 3524
    },
    {
      "epoch": 0.31789692023267346,
      "grad_norm": 1.3929702185400017,
      "learning_rate": 3.191816487654117e-06,
      "loss": 0.7068,
      "step": 3525
    },
    {
      "epoch": 0.31798710375614375,
      "grad_norm": 3.3623110693284626,
      "learning_rate": 3.19134728499188e-06,
      "loss": 0.745,
      "step": 3526
    },
    {
      "epoch": 0.31807728727961404,
      "grad_norm": 5.616578134052181,
      "learning_rate": 3.1908779806773235e-06,
      "loss": 0.6633,
      "step": 3527
    },
    {
      "epoch": 0.3181674708030843,
      "grad_norm": 1.5724437162578984,
      "learning_rate": 3.190408574750492e-06,
      "loss": 0.8289,
      "step": 3528
    },
    {
      "epoch": 0.31825765432655456,
      "grad_norm": 1.4637017032464046,
      "learning_rate": 3.1899390672514367e-06,
      "loss": 0.8426,
      "step": 3529
    },
    {
      "epoch": 0.3183478378500248,
      "grad_norm": 2.0135965712459383,
      "learning_rate": 3.189469458220219e-06,
      "loss": 0.6951,
      "step": 3530
    },
    {
      "epoch": 0.3184380213734951,
      "grad_norm": 1.5722794241921623,
      "learning_rate": 3.1889997476969086e-06,
      "loss": 0.7721,
      "step": 3531
    },
    {
      "epoch": 0.3185282048969653,
      "grad_norm": 4.2932730819391764,
      "learning_rate": 3.188529935721583e-06,
      "loss": 0.7648,
      "step": 3532
    },
    {
      "epoch": 0.3186183884204356,
      "grad_norm": 1.753137568736453,
      "learning_rate": 3.18806002233433e-06,
      "loss": 0.8344,
      "step": 3533
    },
    {
      "epoch": 0.31870857194390584,
      "grad_norm": 1.6587718981150472,
      "learning_rate": 3.187590007575245e-06,
      "loss": 0.8373,
      "step": 3534
    },
    {
      "epoch": 0.3187987554673761,
      "grad_norm": 1.866666777077172,
      "learning_rate": 3.1871198914844327e-06,
      "loss": 0.7125,
      "step": 3535
    },
    {
      "epoch": 0.31888893899084636,
      "grad_norm": 1.5963193340971453,
      "learning_rate": 3.1866496741020057e-06,
      "loss": 0.7144,
      "step": 3536
    },
    {
      "epoch": 0.31897912251431665,
      "grad_norm": 1.2858128595030425,
      "learning_rate": 3.186179355468085e-06,
      "loss": 0.7387,
      "step": 3537
    },
    {
      "epoch": 0.3190693060377869,
      "grad_norm": 1.8713531632867069,
      "learning_rate": 3.1857089356228015e-06,
      "loss": 0.7684,
      "step": 3538
    },
    {
      "epoch": 0.31915948956125717,
      "grad_norm": 1.662978558033374,
      "learning_rate": 3.1852384146062933e-06,
      "loss": 0.723,
      "step": 3539
    },
    {
      "epoch": 0.3192496730847274,
      "grad_norm": 0.6124292517693577,
      "learning_rate": 3.184767792458708e-06,
      "loss": 0.5574,
      "step": 3540
    },
    {
      "epoch": 0.3193398566081977,
      "grad_norm": 11.843531624182367,
      "learning_rate": 3.1842970692202023e-06,
      "loss": 0.9008,
      "step": 3541
    },
    {
      "epoch": 0.3194300401316679,
      "grad_norm": 2.311765915111379,
      "learning_rate": 3.1838262449309403e-06,
      "loss": 0.7529,
      "step": 3542
    },
    {
      "epoch": 0.3195202236551382,
      "grad_norm": 1.7865386885503645,
      "learning_rate": 3.1833553196310956e-06,
      "loss": 0.6889,
      "step": 3543
    },
    {
      "epoch": 0.31961040717860845,
      "grad_norm": 1.635795606054059,
      "learning_rate": 3.18288429336085e-06,
      "loss": 0.6294,
      "step": 3544
    },
    {
      "epoch": 0.31970059070207874,
      "grad_norm": 1.7949286492918275,
      "learning_rate": 3.182413166160394e-06,
      "loss": 0.8787,
      "step": 3545
    },
    {
      "epoch": 0.31979077422554897,
      "grad_norm": 1.5736743373725646,
      "learning_rate": 3.1819419380699275e-06,
      "loss": 0.8164,
      "step": 3546
    },
    {
      "epoch": 0.31988095774901926,
      "grad_norm": 1.6811424164225481,
      "learning_rate": 3.181470609129658e-06,
      "loss": 0.7782,
      "step": 3547
    },
    {
      "epoch": 0.3199711412724895,
      "grad_norm": 2.561042091789346,
      "learning_rate": 3.1809991793798e-06,
      "loss": 0.8257,
      "step": 3548
    },
    {
      "epoch": 0.3200613247959598,
      "grad_norm": 1.9911097584565616,
      "learning_rate": 3.1805276488605806e-06,
      "loss": 0.7332,
      "step": 3549
    },
    {
      "epoch": 0.32015150831943007,
      "grad_norm": 1.8599705744025716,
      "learning_rate": 3.1800560176122336e-06,
      "loss": 0.7651,
      "step": 3550
    },
    {
      "epoch": 0.3202416918429003,
      "grad_norm": 1.637520677858209,
      "learning_rate": 3.179584285675e-06,
      "loss": 0.7852,
      "step": 3551
    },
    {
      "epoch": 0.3203318753663706,
      "grad_norm": 1.8820574500895615,
      "learning_rate": 3.1791124530891315e-06,
      "loss": 0.692,
      "step": 3552
    },
    {
      "epoch": 0.3204220588898408,
      "grad_norm": 1.9489359985028694,
      "learning_rate": 3.178640519894886e-06,
      "loss": 0.7738,
      "step": 3553
    },
    {
      "epoch": 0.3205122424133111,
      "grad_norm": 1.5504382928933969,
      "learning_rate": 3.1781684861325324e-06,
      "loss": 0.7298,
      "step": 3554
    },
    {
      "epoch": 0.32060242593678134,
      "grad_norm": 1.6984410402599732,
      "learning_rate": 3.177696351842348e-06,
      "loss": 0.7779,
      "step": 3555
    },
    {
      "epoch": 0.32069260946025163,
      "grad_norm": 0.7159855796224921,
      "learning_rate": 3.1772241170646167e-06,
      "loss": 0.5604,
      "step": 3556
    },
    {
      "epoch": 0.32078279298372187,
      "grad_norm": 1.8978614718147186,
      "learning_rate": 3.1767517818396334e-06,
      "loss": 0.8783,
      "step": 3557
    },
    {
      "epoch": 0.32087297650719215,
      "grad_norm": 5.153159585238701,
      "learning_rate": 3.1762793462076986e-06,
      "loss": 0.9283,
      "step": 3558
    },
    {
      "epoch": 0.3209631600306624,
      "grad_norm": 2.11956473470659,
      "learning_rate": 3.1758068102091236e-06,
      "loss": 0.6641,
      "step": 3559
    },
    {
      "epoch": 0.3210533435541327,
      "grad_norm": 3.024324351117406,
      "learning_rate": 3.175334173884229e-06,
      "loss": 0.7414,
      "step": 3560
    },
    {
      "epoch": 0.3211435270776029,
      "grad_norm": 1.4322768147016431,
      "learning_rate": 3.174861437273342e-06,
      "loss": 0.7722,
      "step": 3561
    },
    {
      "epoch": 0.3212337106010732,
      "grad_norm": 1.590076476363195,
      "learning_rate": 3.174388600416799e-06,
      "loss": 0.733,
      "step": 3562
    },
    {
      "epoch": 0.32132389412454343,
      "grad_norm": 0.6654711872458359,
      "learning_rate": 3.1739156633549445e-06,
      "loss": 0.571,
      "step": 3563
    },
    {
      "epoch": 0.3214140776480137,
      "grad_norm": 1.932760348036586,
      "learning_rate": 3.173442626128133e-06,
      "loss": 0.8371,
      "step": 3564
    },
    {
      "epoch": 0.32150426117148395,
      "grad_norm": 2.8718757239206956,
      "learning_rate": 3.1729694887767265e-06,
      "loss": 0.8844,
      "step": 3565
    },
    {
      "epoch": 0.32159444469495424,
      "grad_norm": 2.9487829833795582,
      "learning_rate": 3.172496251341096e-06,
      "loss": 0.7744,
      "step": 3566
    },
    {
      "epoch": 0.3216846282184245,
      "grad_norm": 1.6512901608523436,
      "learning_rate": 3.172022913861619e-06,
      "loss": 0.6271,
      "step": 3567
    },
    {
      "epoch": 0.32177481174189476,
      "grad_norm": 1.9094574631085712,
      "learning_rate": 3.171549476378686e-06,
      "loss": 0.7335,
      "step": 3568
    },
    {
      "epoch": 0.321864995265365,
      "grad_norm": 1.7211745067876039,
      "learning_rate": 3.1710759389326906e-06,
      "loss": 0.7444,
      "step": 3569
    },
    {
      "epoch": 0.3219551787888353,
      "grad_norm": 1.7940679766769552,
      "learning_rate": 3.1706023015640396e-06,
      "loss": 0.7651,
      "step": 3570
    },
    {
      "epoch": 0.3220453623123055,
      "grad_norm": 1.855252737220356,
      "learning_rate": 3.1701285643131453e-06,
      "loss": 0.7908,
      "step": 3571
    },
    {
      "epoch": 0.3221355458357758,
      "grad_norm": 2.0313577476681033,
      "learning_rate": 3.16965472722043e-06,
      "loss": 0.7856,
      "step": 3572
    },
    {
      "epoch": 0.32222572935924604,
      "grad_norm": 2.833825536124172,
      "learning_rate": 3.169180790326324e-06,
      "loss": 0.6989,
      "step": 3573
    },
    {
      "epoch": 0.3223159128827163,
      "grad_norm": 1.5199392207441946,
      "learning_rate": 3.168706753671266e-06,
      "loss": 0.777,
      "step": 3574
    },
    {
      "epoch": 0.3224060964061866,
      "grad_norm": 2.265306016236766,
      "learning_rate": 3.168232617295704e-06,
      "loss": 0.6966,
      "step": 3575
    },
    {
      "epoch": 0.32249627992965685,
      "grad_norm": 1.8879580642964806,
      "learning_rate": 3.167758381240093e-06,
      "loss": 0.7676,
      "step": 3576
    },
    {
      "epoch": 0.32258646345312714,
      "grad_norm": 1.6601515825991375,
      "learning_rate": 3.1672840455448978e-06,
      "loss": 0.8161,
      "step": 3577
    },
    {
      "epoch": 0.32267664697659737,
      "grad_norm": 1.6620439635336817,
      "learning_rate": 3.166809610250592e-06,
      "loss": 0.7158,
      "step": 3578
    },
    {
      "epoch": 0.32276683050006766,
      "grad_norm": 1.5785751456159587,
      "learning_rate": 3.166335075397656e-06,
      "loss": 0.7023,
      "step": 3579
    },
    {
      "epoch": 0.3228570140235379,
      "grad_norm": 1.9013871249195453,
      "learning_rate": 3.1658604410265808e-06,
      "loss": 0.7361,
      "step": 3580
    },
    {
      "epoch": 0.3229471975470082,
      "grad_norm": 0.8849388441410629,
      "learning_rate": 3.1653857071778644e-06,
      "loss": 0.5998,
      "step": 3581
    },
    {
      "epoch": 0.3230373810704784,
      "grad_norm": 1.829491976382334,
      "learning_rate": 3.1649108738920133e-06,
      "loss": 0.7644,
      "step": 3582
    },
    {
      "epoch": 0.3231275645939487,
      "grad_norm": 1.6840137214554676,
      "learning_rate": 3.1644359412095432e-06,
      "loss": 0.7753,
      "step": 3583
    },
    {
      "epoch": 0.32321774811741893,
      "grad_norm": 0.7552423052984548,
      "learning_rate": 3.163960909170978e-06,
      "loss": 0.6094,
      "step": 3584
    },
    {
      "epoch": 0.3233079316408892,
      "grad_norm": 1.6067766862347395,
      "learning_rate": 3.1634857778168496e-06,
      "loss": 0.6702,
      "step": 3585
    },
    {
      "epoch": 0.32339811516435946,
      "grad_norm": 2.6709027222314807,
      "learning_rate": 3.1630105471877002e-06,
      "loss": 0.8073,
      "step": 3586
    },
    {
      "epoch": 0.32348829868782975,
      "grad_norm": 2.1039650949548983,
      "learning_rate": 3.162535217324077e-06,
      "loss": 0.7561,
      "step": 3587
    },
    {
      "epoch": 0.3235784822113,
      "grad_norm": 4.012961844130987,
      "learning_rate": 3.1620597882665393e-06,
      "loss": 0.8015,
      "step": 3588
    },
    {
      "epoch": 0.32366866573477027,
      "grad_norm": 1.8802528552193576,
      "learning_rate": 3.1615842600556535e-06,
      "loss": 0.8307,
      "step": 3589
    },
    {
      "epoch": 0.3237588492582405,
      "grad_norm": 1.6061129797296347,
      "learning_rate": 3.1611086327319932e-06,
      "loss": 0.767,
      "step": 3590
    },
    {
      "epoch": 0.3238490327817108,
      "grad_norm": 1.6758353840369462,
      "learning_rate": 3.160632906336142e-06,
      "loss": 0.8393,
      "step": 3591
    },
    {
      "epoch": 0.323939216305181,
      "grad_norm": 2.275583309806922,
      "learning_rate": 3.160157080908692e-06,
      "loss": 0.7939,
      "step": 3592
    },
    {
      "epoch": 0.3240293998286513,
      "grad_norm": 2.072022160385811,
      "learning_rate": 3.1596811564902426e-06,
      "loss": 0.8296,
      "step": 3593
    },
    {
      "epoch": 0.32411958335212154,
      "grad_norm": 2.3654828370285443,
      "learning_rate": 3.1592051331214023e-06,
      "loss": 0.7166,
      "step": 3594
    },
    {
      "epoch": 0.32420976687559183,
      "grad_norm": 1.6798209203915044,
      "learning_rate": 3.158729010842789e-06,
      "loss": 0.7971,
      "step": 3595
    },
    {
      "epoch": 0.32429995039906206,
      "grad_norm": 1.6915240169695318,
      "learning_rate": 3.1582527896950266e-06,
      "loss": 0.8182,
      "step": 3596
    },
    {
      "epoch": 0.32439013392253235,
      "grad_norm": 3.4347335780963335,
      "learning_rate": 3.157776469718749e-06,
      "loss": 0.7885,
      "step": 3597
    },
    {
      "epoch": 0.32448031744600264,
      "grad_norm": 1.8914107274188021,
      "learning_rate": 3.1573000509546004e-06,
      "loss": 0.7603,
      "step": 3598
    },
    {
      "epoch": 0.3245705009694729,
      "grad_norm": 2.083796742080352,
      "learning_rate": 3.1568235334432296e-06,
      "loss": 0.8241,
      "step": 3599
    },
    {
      "epoch": 0.32466068449294316,
      "grad_norm": 1.8261160612501834,
      "learning_rate": 3.1563469172252964e-06,
      "loss": 0.8444,
      "step": 3600
    },
    {
      "epoch": 0.3247508680164134,
      "grad_norm": 1.9717596754879254,
      "learning_rate": 3.155870202341468e-06,
      "loss": 0.8471,
      "step": 3601
    },
    {
      "epoch": 0.3248410515398837,
      "grad_norm": 2.802430160972599,
      "learning_rate": 3.155393388832421e-06,
      "loss": 0.7641,
      "step": 3602
    },
    {
      "epoch": 0.3249312350633539,
      "grad_norm": 1.8748460706468666,
      "learning_rate": 3.1549164767388386e-06,
      "loss": 0.8023,
      "step": 3603
    },
    {
      "epoch": 0.3250214185868242,
      "grad_norm": 2.0915734664194052,
      "learning_rate": 3.1544394661014145e-06,
      "loss": 0.7741,
      "step": 3604
    },
    {
      "epoch": 0.32511160211029444,
      "grad_norm": 1.4384870871931255,
      "learning_rate": 3.15396235696085e-06,
      "loss": 0.6884,
      "step": 3605
    },
    {
      "epoch": 0.32520178563376473,
      "grad_norm": 1.5572598134060947,
      "learning_rate": 3.153485149357854e-06,
      "loss": 0.7635,
      "step": 3606
    },
    {
      "epoch": 0.32529196915723496,
      "grad_norm": 1.727340729985558,
      "learning_rate": 3.153007843333145e-06,
      "loss": 0.7642,
      "step": 3607
    },
    {
      "epoch": 0.32538215268070525,
      "grad_norm": 1.8106868488149561,
      "learning_rate": 3.152530438927449e-06,
      "loss": 0.8041,
      "step": 3608
    },
    {
      "epoch": 0.3254723362041755,
      "grad_norm": 1.8391882504209431,
      "learning_rate": 3.1520529361815008e-06,
      "loss": 0.8178,
      "step": 3609
    },
    {
      "epoch": 0.32556251972764577,
      "grad_norm": 1.56128821113195,
      "learning_rate": 3.151575335136044e-06,
      "loss": 0.8182,
      "step": 3610
    },
    {
      "epoch": 0.325652703251116,
      "grad_norm": 2.9209456164172902,
      "learning_rate": 3.1510976358318298e-06,
      "loss": 0.7563,
      "step": 3611
    },
    {
      "epoch": 0.3257428867745863,
      "grad_norm": 1.6619098332869415,
      "learning_rate": 3.1506198383096186e-06,
      "loss": 0.7272,
      "step": 3612
    },
    {
      "epoch": 0.3258330702980565,
      "grad_norm": 1.5923777637848489,
      "learning_rate": 3.150141942610178e-06,
      "loss": 0.7913,
      "step": 3613
    },
    {
      "epoch": 0.3259232538215268,
      "grad_norm": 0.6977464530018196,
      "learning_rate": 3.1496639487742853e-06,
      "loss": 0.5873,
      "step": 3614
    },
    {
      "epoch": 0.32601343734499705,
      "grad_norm": 2.242426630221165,
      "learning_rate": 3.1491858568427247e-06,
      "loss": 0.8227,
      "step": 3615
    },
    {
      "epoch": 0.32610362086846734,
      "grad_norm": 2.0437631029181462,
      "learning_rate": 3.1487076668562903e-06,
      "loss": 0.8304,
      "step": 3616
    },
    {
      "epoch": 0.32619380439193757,
      "grad_norm": 3.2022015985122874,
      "learning_rate": 3.1482293788557847e-06,
      "loss": 0.813,
      "step": 3617
    },
    {
      "epoch": 0.32628398791540786,
      "grad_norm": 1.6986253425720754,
      "learning_rate": 3.1477509928820165e-06,
      "loss": 0.7404,
      "step": 3618
    },
    {
      "epoch": 0.3263741714388781,
      "grad_norm": 1.680546172915382,
      "learning_rate": 3.147272508975805e-06,
      "loss": 0.7724,
      "step": 3619
    },
    {
      "epoch": 0.3264643549623484,
      "grad_norm": 0.6838621430078339,
      "learning_rate": 3.1467939271779775e-06,
      "loss": 0.6065,
      "step": 3620
    },
    {
      "epoch": 0.32655453848581867,
      "grad_norm": 1.8206742872644153,
      "learning_rate": 3.146315247529368e-06,
      "loss": 0.735,
      "step": 3621
    },
    {
      "epoch": 0.3266447220092889,
      "grad_norm": 1.8689515786549777,
      "learning_rate": 3.1458364700708212e-06,
      "loss": 0.7355,
      "step": 3622
    },
    {
      "epoch": 0.3267349055327592,
      "grad_norm": 6.424411838788453,
      "learning_rate": 3.1453575948431892e-06,
      "loss": 0.7994,
      "step": 3623
    },
    {
      "epoch": 0.3268250890562294,
      "grad_norm": 2.088230658481007,
      "learning_rate": 3.144878621887331e-06,
      "loss": 0.7794,
      "step": 3624
    },
    {
      "epoch": 0.3269152725796997,
      "grad_norm": 6.850352268489014,
      "learning_rate": 3.1443995512441167e-06,
      "loss": 0.8172,
      "step": 3625
    },
    {
      "epoch": 0.32700545610316994,
      "grad_norm": 1.5791835445149642,
      "learning_rate": 3.1439203829544224e-06,
      "loss": 0.8405,
      "step": 3626
    },
    {
      "epoch": 0.32709563962664023,
      "grad_norm": 1.5156918540446263,
      "learning_rate": 3.143441117059133e-06,
      "loss": 0.8554,
      "step": 3627
    },
    {
      "epoch": 0.32718582315011047,
      "grad_norm": 1.7920841869052728,
      "learning_rate": 3.142961753599143e-06,
      "loss": 0.7274,
      "step": 3628
    },
    {
      "epoch": 0.32727600667358075,
      "grad_norm": 1.5360316918054102,
      "learning_rate": 3.1424822926153543e-06,
      "loss": 0.7425,
      "step": 3629
    },
    {
      "epoch": 0.327366190197051,
      "grad_norm": 1.9764858426996668,
      "learning_rate": 3.142002734148676e-06,
      "loss": 0.8172,
      "step": 3630
    },
    {
      "epoch": 0.3274563737205213,
      "grad_norm": 1.916400897908503,
      "learning_rate": 3.141523078240028e-06,
      "loss": 0.8085,
      "step": 3631
    },
    {
      "epoch": 0.3275465572439915,
      "grad_norm": 2.2641830591257266,
      "learning_rate": 3.1410433249303366e-06,
      "loss": 0.8359,
      "step": 3632
    },
    {
      "epoch": 0.3276367407674618,
      "grad_norm": 2.5296345017779642,
      "learning_rate": 3.1405634742605366e-06,
      "loss": 0.8671,
      "step": 3633
    },
    {
      "epoch": 0.32772692429093203,
      "grad_norm": 1.9794627386461092,
      "learning_rate": 3.1400835262715727e-06,
      "loss": 0.8143,
      "step": 3634
    },
    {
      "epoch": 0.3278171078144023,
      "grad_norm": 2.0938679676130247,
      "learning_rate": 3.139603481004396e-06,
      "loss": 0.7696,
      "step": 3635
    },
    {
      "epoch": 0.32790729133787255,
      "grad_norm": 1.7746335336332622,
      "learning_rate": 3.139123338499966e-06,
      "loss": 0.7887,
      "step": 3636
    },
    {
      "epoch": 0.32799747486134284,
      "grad_norm": 2.555681694299371,
      "learning_rate": 3.1386430987992524e-06,
      "loss": 0.8149,
      "step": 3637
    },
    {
      "epoch": 0.3280876583848131,
      "grad_norm": 1.7406890309471468,
      "learning_rate": 3.1381627619432307e-06,
      "loss": 0.7237,
      "step": 3638
    },
    {
      "epoch": 0.32817784190828336,
      "grad_norm": 2.668652738634723,
      "learning_rate": 3.1376823279728864e-06,
      "loss": 0.7881,
      "step": 3639
    },
    {
      "epoch": 0.3282680254317536,
      "grad_norm": 1.7090792384385571,
      "learning_rate": 3.1372017969292125e-06,
      "loss": 0.8577,
      "step": 3640
    },
    {
      "epoch": 0.3283582089552239,
      "grad_norm": 6.904800470053425,
      "learning_rate": 3.136721168853211e-06,
      "loss": 0.7595,
      "step": 3641
    },
    {
      "epoch": 0.3284483924786941,
      "grad_norm": 2.3196286821828975,
      "learning_rate": 3.1362404437858924e-06,
      "loss": 0.7425,
      "step": 3642
    },
    {
      "epoch": 0.3285385760021644,
      "grad_norm": 1.8444410419177393,
      "learning_rate": 3.135759621768273e-06,
      "loss": 0.8344,
      "step": 3643
    },
    {
      "epoch": 0.32862875952563464,
      "grad_norm": 2.3624915955409165,
      "learning_rate": 3.13527870284138e-06,
      "loss": 0.8274,
      "step": 3644
    },
    {
      "epoch": 0.32871894304910493,
      "grad_norm": 2.60888502808819,
      "learning_rate": 3.134797687046249e-06,
      "loss": 0.8399,
      "step": 3645
    },
    {
      "epoch": 0.3288091265725752,
      "grad_norm": 2.175304860640194,
      "learning_rate": 3.1343165744239218e-06,
      "loss": 0.7184,
      "step": 3646
    },
    {
      "epoch": 0.32889931009604545,
      "grad_norm": 1.869521593514486,
      "learning_rate": 3.13383536501545e-06,
      "loss": 0.7981,
      "step": 3647
    },
    {
      "epoch": 0.32898949361951574,
      "grad_norm": 1.944429863769117,
      "learning_rate": 3.133354058861893e-06,
      "loss": 0.8001,
      "step": 3648
    },
    {
      "epoch": 0.32907967714298597,
      "grad_norm": 1.7350606851655868,
      "learning_rate": 3.132872656004318e-06,
      "loss": 0.746,
      "step": 3649
    },
    {
      "epoch": 0.32916986066645626,
      "grad_norm": 1.6217950613795542,
      "learning_rate": 3.132391156483802e-06,
      "loss": 0.6811,
      "step": 3650
    },
    {
      "epoch": 0.3292600441899265,
      "grad_norm": 5.2550740018474125,
      "learning_rate": 3.131909560341428e-06,
      "loss": 0.7428,
      "step": 3651
    },
    {
      "epoch": 0.3293502277133968,
      "grad_norm": 3.7563655232518176,
      "learning_rate": 3.1314278676182893e-06,
      "loss": 0.7246,
      "step": 3652
    },
    {
      "epoch": 0.329440411236867,
      "grad_norm": 1.690661753256745,
      "learning_rate": 3.130946078355486e-06,
      "loss": 0.8462,
      "step": 3653
    },
    {
      "epoch": 0.3295305947603373,
      "grad_norm": 2.2140557503715153,
      "learning_rate": 3.130464192594128e-06,
      "loss": 0.8134,
      "step": 3654
    },
    {
      "epoch": 0.32962077828380754,
      "grad_norm": 1.5904849397341423,
      "learning_rate": 3.1299822103753315e-06,
      "loss": 0.7055,
      "step": 3655
    },
    {
      "epoch": 0.3297109618072778,
      "grad_norm": 1.8425583948951612,
      "learning_rate": 3.1295001317402217e-06,
      "loss": 0.7798,
      "step": 3656
    },
    {
      "epoch": 0.32980114533074806,
      "grad_norm": 2.5842120819217964,
      "learning_rate": 3.1290179567299335e-06,
      "loss": 0.7215,
      "step": 3657
    },
    {
      "epoch": 0.32989132885421835,
      "grad_norm": 1.6455493231271923,
      "learning_rate": 3.128535685385607e-06,
      "loss": 0.8071,
      "step": 3658
    },
    {
      "epoch": 0.3299815123776886,
      "grad_norm": 1.6562418667575582,
      "learning_rate": 3.1280533177483935e-06,
      "loss": 0.8585,
      "step": 3659
    },
    {
      "epoch": 0.33007169590115887,
      "grad_norm": 1.4064844995587702,
      "learning_rate": 3.127570853859451e-06,
      "loss": 0.7891,
      "step": 3660
    },
    {
      "epoch": 0.3301618794246291,
      "grad_norm": 2.2540773958120948,
      "learning_rate": 3.1270882937599456e-06,
      "loss": 0.6676,
      "step": 3661
    },
    {
      "epoch": 0.3302520629480994,
      "grad_norm": 3.1090064669026787,
      "learning_rate": 3.1266056374910532e-06,
      "loss": 0.8012,
      "step": 3662
    },
    {
      "epoch": 0.3303422464715696,
      "grad_norm": 1.966161388478841,
      "learning_rate": 3.126122885093955e-06,
      "loss": 0.6761,
      "step": 3663
    },
    {
      "epoch": 0.3304324299950399,
      "grad_norm": 1.8480950574413404,
      "learning_rate": 3.1256400366098427e-06,
      "loss": 0.7744,
      "step": 3664
    },
    {
      "epoch": 0.33052261351851014,
      "grad_norm": 1.7543554601610598,
      "learning_rate": 3.125157092079916e-06,
      "loss": 0.7327,
      "step": 3665
    },
    {
      "epoch": 0.33061279704198043,
      "grad_norm": 4.331899161083609,
      "learning_rate": 3.1246740515453824e-06,
      "loss": 0.7511,
      "step": 3666
    },
    {
      "epoch": 0.33070298056545067,
      "grad_norm": 1.5534186624794137,
      "learning_rate": 3.124190915047457e-06,
      "loss": 0.716,
      "step": 3667
    },
    {
      "epoch": 0.33079316408892095,
      "grad_norm": 1.9169673752891503,
      "learning_rate": 3.123707682627364e-06,
      "loss": 0.7461,
      "step": 3668
    },
    {
      "epoch": 0.33088334761239124,
      "grad_norm": 2.0669257334898252,
      "learning_rate": 3.1232243543263356e-06,
      "loss": 0.7584,
      "step": 3669
    },
    {
      "epoch": 0.3309735311358615,
      "grad_norm": 1.7503522790830175,
      "learning_rate": 3.1227409301856122e-06,
      "loss": 0.6829,
      "step": 3670
    },
    {
      "epoch": 0.33106371465933176,
      "grad_norm": 1.5124413786317827,
      "learning_rate": 3.1222574102464413e-06,
      "loss": 0.753,
      "step": 3671
    },
    {
      "epoch": 0.331153898182802,
      "grad_norm": 1.5080676875822219,
      "learning_rate": 3.12177379455008e-06,
      "loss": 0.8046,
      "step": 3672
    },
    {
      "epoch": 0.3312440817062723,
      "grad_norm": 1.9143515855115119,
      "learning_rate": 3.121290083137794e-06,
      "loss": 0.7673,
      "step": 3673
    },
    {
      "epoch": 0.3313342652297425,
      "grad_norm": 2.1411137823367428,
      "learning_rate": 3.1208062760508547e-06,
      "loss": 0.7493,
      "step": 3674
    },
    {
      "epoch": 0.3314244487532128,
      "grad_norm": 1.5096012238170553,
      "learning_rate": 3.1203223733305438e-06,
      "loss": 0.7143,
      "step": 3675
    },
    {
      "epoch": 0.33151463227668304,
      "grad_norm": 2.293757912105042,
      "learning_rate": 3.1198383750181512e-06,
      "loss": 0.5646,
      "step": 3676
    },
    {
      "epoch": 0.33160481580015333,
      "grad_norm": 1.5933149155279835,
      "learning_rate": 3.1193542811549734e-06,
      "loss": 0.6939,
      "step": 3677
    },
    {
      "epoch": 0.33169499932362356,
      "grad_norm": 1.7228301429779271,
      "learning_rate": 3.1188700917823166e-06,
      "loss": 0.6996,
      "step": 3678
    },
    {
      "epoch": 0.33178518284709385,
      "grad_norm": 0.7026733219243546,
      "learning_rate": 3.1183858069414937e-06,
      "loss": 0.6263,
      "step": 3679
    },
    {
      "epoch": 0.3318753663705641,
      "grad_norm": 2.1559210263723356,
      "learning_rate": 3.117901426673827e-06,
      "loss": 0.827,
      "step": 3680
    },
    {
      "epoch": 0.3319655498940344,
      "grad_norm": 2.188283180636365,
      "learning_rate": 3.1174169510206466e-06,
      "loss": 0.7323,
      "step": 3681
    },
    {
      "epoch": 0.3320557334175046,
      "grad_norm": 1.6561780949944598,
      "learning_rate": 3.1169323800232908e-06,
      "loss": 0.7723,
      "step": 3682
    },
    {
      "epoch": 0.3321459169409749,
      "grad_norm": 1.6879806540451778,
      "learning_rate": 3.1164477137231054e-06,
      "loss": 0.8074,
      "step": 3683
    },
    {
      "epoch": 0.3322361004644451,
      "grad_norm": 2.376958340816997,
      "learning_rate": 3.115962952161445e-06,
      "loss": 0.7796,
      "step": 3684
    },
    {
      "epoch": 0.3323262839879154,
      "grad_norm": 2.085904581903182,
      "learning_rate": 3.1154780953796727e-06,
      "loss": 0.8143,
      "step": 3685
    },
    {
      "epoch": 0.33241646751138565,
      "grad_norm": 3.6268344052409605,
      "learning_rate": 3.114993143419158e-06,
      "loss": 0.6803,
      "step": 3686
    },
    {
      "epoch": 0.33250665103485594,
      "grad_norm": 1.3743029908540807,
      "learning_rate": 3.1145080963212806e-06,
      "loss": 0.6814,
      "step": 3687
    },
    {
      "epoch": 0.33259683455832617,
      "grad_norm": 1.5731011676157631,
      "learning_rate": 3.114022954127427e-06,
      "loss": 0.7724,
      "step": 3688
    },
    {
      "epoch": 0.33268701808179646,
      "grad_norm": 1.6332768414258236,
      "learning_rate": 3.1135377168789923e-06,
      "loss": 0.9808,
      "step": 3689
    },
    {
      "epoch": 0.3327772016052667,
      "grad_norm": 1.8876297609301897,
      "learning_rate": 3.1130523846173803e-06,
      "loss": 0.838,
      "step": 3690
    },
    {
      "epoch": 0.332867385128737,
      "grad_norm": 1.8849333535309458,
      "learning_rate": 3.1125669573840006e-06,
      "loss": 0.7163,
      "step": 3691
    },
    {
      "epoch": 0.3329575686522072,
      "grad_norm": 2.850064487731552,
      "learning_rate": 3.112081435220274e-06,
      "loss": 0.7151,
      "step": 3692
    },
    {
      "epoch": 0.3330477521756775,
      "grad_norm": 1.5949058829016622,
      "learning_rate": 3.111595818167627e-06,
      "loss": 0.7618,
      "step": 3693
    },
    {
      "epoch": 0.3331379356991478,
      "grad_norm": 2.1613945886488604,
      "learning_rate": 3.1111101062674953e-06,
      "loss": 0.835,
      "step": 3694
    },
    {
      "epoch": 0.333228119222618,
      "grad_norm": 1.603105573926869,
      "learning_rate": 3.1106242995613233e-06,
      "loss": 0.7592,
      "step": 3695
    },
    {
      "epoch": 0.3333183027460883,
      "grad_norm": 1.6132335886715359,
      "learning_rate": 3.1101383980905616e-06,
      "loss": 0.7725,
      "step": 3696
    },
    {
      "epoch": 0.33340848626955855,
      "grad_norm": 0.647811886593445,
      "learning_rate": 3.109652401896671e-06,
      "loss": 0.5516,
      "step": 3697
    },
    {
      "epoch": 0.33349866979302883,
      "grad_norm": 1.457512988102985,
      "learning_rate": 3.109166311021119e-06,
      "loss": 0.8172,
      "step": 3698
    },
    {
      "epoch": 0.33358885331649907,
      "grad_norm": 2.0367023942917677,
      "learning_rate": 3.1086801255053807e-06,
      "loss": 0.8067,
      "step": 3699
    },
    {
      "epoch": 0.33367903683996936,
      "grad_norm": 1.4265620353840032,
      "learning_rate": 3.108193845390942e-06,
      "loss": 0.8084,
      "step": 3700
    },
    {
      "epoch": 0.3337692203634396,
      "grad_norm": 1.3145982228608208,
      "learning_rate": 3.1077074707192933e-06,
      "loss": 0.7483,
      "step": 3701
    },
    {
      "epoch": 0.3338594038869099,
      "grad_norm": 1.4653680702259733,
      "learning_rate": 3.1072210015319353e-06,
      "loss": 0.792,
      "step": 3702
    },
    {
      "epoch": 0.3339495874103801,
      "grad_norm": 1.8315342470966696,
      "learning_rate": 3.106734437870376e-06,
      "loss": 0.7335,
      "step": 3703
    },
    {
      "epoch": 0.3340397709338504,
      "grad_norm": 1.9125933555805936,
      "learning_rate": 3.1062477797761327e-06,
      "loss": 0.7187,
      "step": 3704
    },
    {
      "epoch": 0.33412995445732063,
      "grad_norm": 2.0576929658714573,
      "learning_rate": 3.105761027290729e-06,
      "loss": 0.8193,
      "step": 3705
    },
    {
      "epoch": 0.3342201379807909,
      "grad_norm": 1.9586010269741971,
      "learning_rate": 3.105274180455697e-06,
      "loss": 0.7414,
      "step": 3706
    },
    {
      "epoch": 0.33431032150426115,
      "grad_norm": 1.6813948359729922,
      "learning_rate": 3.1047872393125775e-06,
      "loss": 0.7964,
      "step": 3707
    },
    {
      "epoch": 0.33440050502773144,
      "grad_norm": 0.6245203084705402,
      "learning_rate": 3.1043002039029186e-06,
      "loss": 0.5792,
      "step": 3708
    },
    {
      "epoch": 0.3344906885512017,
      "grad_norm": 1.4046773806197912,
      "learning_rate": 3.1038130742682782e-06,
      "loss": 0.7617,
      "step": 3709
    },
    {
      "epoch": 0.33458087207467196,
      "grad_norm": 1.417757633350792,
      "learning_rate": 3.103325850450219e-06,
      "loss": 0.7481,
      "step": 3710
    },
    {
      "epoch": 0.3346710555981422,
      "grad_norm": 1.8477414778878174,
      "learning_rate": 3.1028385324903154e-06,
      "loss": 0.8134,
      "step": 3711
    },
    {
      "epoch": 0.3347612391216125,
      "grad_norm": 1.6339207588741589,
      "learning_rate": 3.1023511204301465e-06,
      "loss": 0.7794,
      "step": 3712
    },
    {
      "epoch": 0.3348514226450827,
      "grad_norm": 1.5698846642231437,
      "learning_rate": 3.1018636143113022e-06,
      "loss": 0.7309,
      "step": 3713
    },
    {
      "epoch": 0.334941606168553,
      "grad_norm": 1.7129282756917128,
      "learning_rate": 3.1013760141753787e-06,
      "loss": 0.6811,
      "step": 3714
    },
    {
      "epoch": 0.33503178969202324,
      "grad_norm": 1.574762368819344,
      "learning_rate": 3.100888320063981e-06,
      "loss": 0.7405,
      "step": 3715
    },
    {
      "epoch": 0.33512197321549353,
      "grad_norm": 1.4546015631084208,
      "learning_rate": 3.100400532018721e-06,
      "loss": 0.7773,
      "step": 3716
    },
    {
      "epoch": 0.3352121567389638,
      "grad_norm": 1.98607598639005,
      "learning_rate": 3.0999126500812204e-06,
      "loss": 0.8407,
      "step": 3717
    },
    {
      "epoch": 0.33530234026243405,
      "grad_norm": 0.7213932444503347,
      "learning_rate": 3.0994246742931076e-06,
      "loss": 0.6342,
      "step": 3718
    },
    {
      "epoch": 0.33539252378590434,
      "grad_norm": 1.888553840554485,
      "learning_rate": 3.098936604696019e-06,
      "loss": 0.7123,
      "step": 3719
    },
    {
      "epoch": 0.3354827073093746,
      "grad_norm": 1.8342226501314194,
      "learning_rate": 3.0984484413316e-06,
      "loss": 0.8004,
      "step": 3720
    },
    {
      "epoch": 0.33557289083284486,
      "grad_norm": 2.177963066922882,
      "learning_rate": 3.0979601842415033e-06,
      "loss": 0.7523,
      "step": 3721
    },
    {
      "epoch": 0.3356630743563151,
      "grad_norm": 3.5797074329448986,
      "learning_rate": 3.0974718334673896e-06,
      "loss": 0.6923,
      "step": 3722
    },
    {
      "epoch": 0.3357532578797854,
      "grad_norm": 1.5103933749102108,
      "learning_rate": 3.0969833890509282e-06,
      "loss": 0.7618,
      "step": 3723
    },
    {
      "epoch": 0.3358434414032556,
      "grad_norm": 2.518808849773638,
      "learning_rate": 3.096494851033795e-06,
      "loss": 0.7698,
      "step": 3724
    },
    {
      "epoch": 0.3359336249267259,
      "grad_norm": 1.790930160430209,
      "learning_rate": 3.0960062194576747e-06,
      "loss": 0.7001,
      "step": 3725
    },
    {
      "epoch": 0.33602380845019614,
      "grad_norm": 1.784437439689543,
      "learning_rate": 3.0955174943642606e-06,
      "loss": 0.7231,
      "step": 3726
    },
    {
      "epoch": 0.3361139919736664,
      "grad_norm": 1.9617282443346635,
      "learning_rate": 3.0950286757952534e-06,
      "loss": 0.6928,
      "step": 3727
    },
    {
      "epoch": 0.33620417549713666,
      "grad_norm": 1.8266934389509961,
      "learning_rate": 3.0945397637923617e-06,
      "loss": 0.834,
      "step": 3728
    },
    {
      "epoch": 0.33629435902060695,
      "grad_norm": 2.005714596999458,
      "learning_rate": 3.0940507583973025e-06,
      "loss": 0.7795,
      "step": 3729
    },
    {
      "epoch": 0.3363845425440772,
      "grad_norm": 1.6475714266030068,
      "learning_rate": 3.093561659651799e-06,
      "loss": 0.7415,
      "step": 3730
    },
    {
      "epoch": 0.33647472606754747,
      "grad_norm": 1.8124707975994756,
      "learning_rate": 3.093072467597586e-06,
      "loss": 0.7082,
      "step": 3731
    },
    {
      "epoch": 0.3365649095910177,
      "grad_norm": 0.6420087291697477,
      "learning_rate": 3.092583182276402e-06,
      "loss": 0.5844,
      "step": 3732
    },
    {
      "epoch": 0.336655093114488,
      "grad_norm": 0.6360278681626408,
      "learning_rate": 3.092093803729997e-06,
      "loss": 0.5905,
      "step": 3733
    },
    {
      "epoch": 0.3367452766379582,
      "grad_norm": 1.6311897354394214,
      "learning_rate": 3.0916043320001264e-06,
      "loss": 0.7819,
      "step": 3734
    },
    {
      "epoch": 0.3368354601614285,
      "grad_norm": 1.7275538295634398,
      "learning_rate": 3.0911147671285557e-06,
      "loss": 0.782,
      "step": 3735
    },
    {
      "epoch": 0.33692564368489875,
      "grad_norm": 2.3174779770348946,
      "learning_rate": 3.0906251091570565e-06,
      "loss": 0.799,
      "step": 3736
    },
    {
      "epoch": 0.33701582720836903,
      "grad_norm": 1.7598836268618545,
      "learning_rate": 3.0901353581274094e-06,
      "loss": 0.782,
      "step": 3737
    },
    {
      "epoch": 0.33710601073183927,
      "grad_norm": 1.949922423164538,
      "learning_rate": 3.089645514081402e-06,
      "loss": 0.6499,
      "step": 3738
    },
    {
      "epoch": 0.33719619425530956,
      "grad_norm": 2.72661802568708,
      "learning_rate": 3.0891555770608323e-06,
      "loss": 0.702,
      "step": 3739
    },
    {
      "epoch": 0.33728637777877984,
      "grad_norm": 1.7134897365257122,
      "learning_rate": 3.088665547107503e-06,
      "loss": 0.793,
      "step": 3740
    },
    {
      "epoch": 0.3373765613022501,
      "grad_norm": 1.4262781400854962,
      "learning_rate": 3.0881754242632254e-06,
      "loss": 0.8478,
      "step": 3741
    },
    {
      "epoch": 0.33746674482572037,
      "grad_norm": 1.353262917623807,
      "learning_rate": 3.0876852085698213e-06,
      "loss": 0.8137,
      "step": 3742
    },
    {
      "epoch": 0.3375569283491906,
      "grad_norm": 1.8636973814057416,
      "learning_rate": 3.087194900069117e-06,
      "loss": 0.7978,
      "step": 3743
    },
    {
      "epoch": 0.3376471118726609,
      "grad_norm": 2.3100056985892152,
      "learning_rate": 3.08670449880295e-06,
      "loss": 0.8263,
      "step": 3744
    },
    {
      "epoch": 0.3377372953961311,
      "grad_norm": 1.550241371557704,
      "learning_rate": 3.086214004813163e-06,
      "loss": 0.7912,
      "step": 3745
    },
    {
      "epoch": 0.3378274789196014,
      "grad_norm": 1.6128003602142915,
      "learning_rate": 3.0857234181416074e-06,
      "loss": 0.8347,
      "step": 3746
    },
    {
      "epoch": 0.33791766244307164,
      "grad_norm": 1.6386306266458177,
      "learning_rate": 3.085232738830143e-06,
      "loss": 0.8216,
      "step": 3747
    },
    {
      "epoch": 0.33800784596654193,
      "grad_norm": 1.8652602588395477,
      "learning_rate": 3.084741966920638e-06,
      "loss": 0.8316,
      "step": 3748
    },
    {
      "epoch": 0.33809802949001216,
      "grad_norm": 1.5111621551603398,
      "learning_rate": 3.084251102454966e-06,
      "loss": 0.7294,
      "step": 3749
    },
    {
      "epoch": 0.33818821301348245,
      "grad_norm": 0.7563495475888231,
      "learning_rate": 3.083760145475013e-06,
      "loss": 0.6144,
      "step": 3750
    },
    {
      "epoch": 0.3382783965369527,
      "grad_norm": 1.7947713064439972,
      "learning_rate": 3.0832690960226678e-06,
      "loss": 0.7136,
      "step": 3751
    },
    {
      "epoch": 0.338368580060423,
      "grad_norm": 2.0759053243095713,
      "learning_rate": 3.08277795413983e-06,
      "loss": 0.7444,
      "step": 3752
    },
    {
      "epoch": 0.3384587635838932,
      "grad_norm": 1.5343044723182016,
      "learning_rate": 3.0822867198684073e-06,
      "loss": 0.8241,
      "step": 3753
    },
    {
      "epoch": 0.3385489471073635,
      "grad_norm": 2.2536556428525807,
      "learning_rate": 3.081795393250314e-06,
      "loss": 0.781,
      "step": 3754
    },
    {
      "epoch": 0.33863913063083373,
      "grad_norm": 4.93409147794896,
      "learning_rate": 3.081303974327473e-06,
      "loss": 0.7472,
      "step": 3755
    },
    {
      "epoch": 0.338729314154304,
      "grad_norm": 2.1403441975881288,
      "learning_rate": 3.080812463141814e-06,
      "loss": 0.7704,
      "step": 3756
    },
    {
      "epoch": 0.33881949767777425,
      "grad_norm": 7.484848994972381,
      "learning_rate": 3.080320859735276e-06,
      "loss": 0.6462,
      "step": 3757
    },
    {
      "epoch": 0.33890968120124454,
      "grad_norm": 2.00155019287328,
      "learning_rate": 3.079829164149806e-06,
      "loss": 0.8223,
      "step": 3758
    },
    {
      "epoch": 0.3389998647247148,
      "grad_norm": 1.7602331829570703,
      "learning_rate": 3.0793373764273573e-06,
      "loss": 0.7622,
      "step": 3759
    },
    {
      "epoch": 0.33909004824818506,
      "grad_norm": 5.009126154667501,
      "learning_rate": 3.078845496609892e-06,
      "loss": 0.7974,
      "step": 3760
    },
    {
      "epoch": 0.3391802317716553,
      "grad_norm": 1.58990665845575,
      "learning_rate": 3.078353524739381e-06,
      "loss": 0.8605,
      "step": 3761
    },
    {
      "epoch": 0.3392704152951256,
      "grad_norm": 1.7083289758890297,
      "learning_rate": 3.077861460857801e-06,
      "loss": 0.8614,
      "step": 3762
    },
    {
      "epoch": 0.3393605988185958,
      "grad_norm": 1.694414525037147,
      "learning_rate": 3.077369305007138e-06,
      "loss": 0.7899,
      "step": 3763
    },
    {
      "epoch": 0.3394507823420661,
      "grad_norm": 1.8178276817103585,
      "learning_rate": 3.0768770572293852e-06,
      "loss": 0.8135,
      "step": 3764
    },
    {
      "epoch": 0.3395409658655364,
      "grad_norm": 1.9001166684319555,
      "learning_rate": 3.0763847175665437e-06,
      "loss": 0.8559,
      "step": 3765
    },
    {
      "epoch": 0.3396311493890066,
      "grad_norm": 3.019480400101428,
      "learning_rate": 3.0758922860606237e-06,
      "loss": 0.7147,
      "step": 3766
    },
    {
      "epoch": 0.3397213329124769,
      "grad_norm": 1.6455403401134083,
      "learning_rate": 3.0753997627536404e-06,
      "loss": 0.7128,
      "step": 3767
    },
    {
      "epoch": 0.33981151643594715,
      "grad_norm": 2.0237500748330963,
      "learning_rate": 3.0749071476876203e-06,
      "loss": 0.7219,
      "step": 3768
    },
    {
      "epoch": 0.33990169995941744,
      "grad_norm": 1.5760992664502698,
      "learning_rate": 3.0744144409045952e-06,
      "loss": 0.7377,
      "step": 3769
    },
    {
      "epoch": 0.33999188348288767,
      "grad_norm": 2.196798995590258,
      "learning_rate": 3.0739216424466056e-06,
      "loss": 0.7935,
      "step": 3770
    },
    {
      "epoch": 0.34008206700635796,
      "grad_norm": 2.008685326760605,
      "learning_rate": 3.0734287523557002e-06,
      "loss": 0.7545,
      "step": 3771
    },
    {
      "epoch": 0.3401722505298282,
      "grad_norm": 2.306720005513866,
      "learning_rate": 3.0729357706739348e-06,
      "loss": 0.7234,
      "step": 3772
    },
    {
      "epoch": 0.3402624340532985,
      "grad_norm": 1.5174224263054559,
      "learning_rate": 3.0724426974433737e-06,
      "loss": 0.7578,
      "step": 3773
    },
    {
      "epoch": 0.3403526175767687,
      "grad_norm": 1.4012916039476324,
      "learning_rate": 3.0719495327060874e-06,
      "loss": 0.7471,
      "step": 3774
    },
    {
      "epoch": 0.340442801100239,
      "grad_norm": 1.6285077163237067,
      "learning_rate": 3.071456276504157e-06,
      "loss": 0.7815,
      "step": 3775
    },
    {
      "epoch": 0.34053298462370923,
      "grad_norm": 1.9745372302481066,
      "learning_rate": 3.070962928879669e-06,
      "loss": 0.7709,
      "step": 3776
    },
    {
      "epoch": 0.3406231681471795,
      "grad_norm": 1.8143904462650267,
      "learning_rate": 3.0704694898747185e-06,
      "loss": 0.7873,
      "step": 3777
    },
    {
      "epoch": 0.34071335167064976,
      "grad_norm": 1.9700628474904383,
      "learning_rate": 3.069975959531408e-06,
      "loss": 0.9003,
      "step": 3778
    },
    {
      "epoch": 0.34080353519412004,
      "grad_norm": 0.6711072971455442,
      "learning_rate": 3.06948233789185e-06,
      "loss": 0.6486,
      "step": 3779
    },
    {
      "epoch": 0.3408937187175903,
      "grad_norm": 2.2240853433426144,
      "learning_rate": 3.0689886249981614e-06,
      "loss": 0.7928,
      "step": 3780
    },
    {
      "epoch": 0.34098390224106057,
      "grad_norm": 2.5564212806977444,
      "learning_rate": 3.0684948208924693e-06,
      "loss": 0.8004,
      "step": 3781
    },
    {
      "epoch": 0.3410740857645308,
      "grad_norm": 2.036659900676937,
      "learning_rate": 3.068000925616907e-06,
      "loss": 0.793,
      "step": 3782
    },
    {
      "epoch": 0.3411642692880011,
      "grad_norm": 2.168789435036373,
      "learning_rate": 3.067506939213617e-06,
      "loss": 0.7664,
      "step": 3783
    },
    {
      "epoch": 0.3412544528114713,
      "grad_norm": 1.3637662096047984,
      "learning_rate": 3.0670128617247493e-06,
      "loss": 0.7777,
      "step": 3784
    },
    {
      "epoch": 0.3413446363349416,
      "grad_norm": 1.520665311894419,
      "learning_rate": 3.06651869319246e-06,
      "loss": 0.687,
      "step": 3785
    },
    {
      "epoch": 0.34143481985841184,
      "grad_norm": 1.6996750492987236,
      "learning_rate": 3.0660244336589154e-06,
      "loss": 0.7717,
      "step": 3786
    },
    {
      "epoch": 0.34152500338188213,
      "grad_norm": 1.5898681938672934,
      "learning_rate": 3.065530083166288e-06,
      "loss": 0.7913,
      "step": 3787
    },
    {
      "epoch": 0.3416151869053524,
      "grad_norm": 1.6800545224924586,
      "learning_rate": 3.0650356417567586e-06,
      "loss": 0.7381,
      "step": 3788
    },
    {
      "epoch": 0.34170537042882265,
      "grad_norm": 1.6482346396347427,
      "learning_rate": 3.0645411094725156e-06,
      "loss": 0.8251,
      "step": 3789
    },
    {
      "epoch": 0.34179555395229294,
      "grad_norm": 2.115252071176265,
      "learning_rate": 3.0640464863557556e-06,
      "loss": 0.7712,
      "step": 3790
    },
    {
      "epoch": 0.3418857374757632,
      "grad_norm": 1.5223931278496619,
      "learning_rate": 3.063551772448682e-06,
      "loss": 0.8097,
      "step": 3791
    },
    {
      "epoch": 0.34197592099923346,
      "grad_norm": 2.1902179726247892,
      "learning_rate": 3.0630569677935075e-06,
      "loss": 0.7068,
      "step": 3792
    },
    {
      "epoch": 0.3420661045227037,
      "grad_norm": 2.3301878117519004,
      "learning_rate": 3.06256207243245e-06,
      "loss": 0.8589,
      "step": 3793
    },
    {
      "epoch": 0.342156288046174,
      "grad_norm": 0.6350948243040347,
      "learning_rate": 3.0620670864077385e-06,
      "loss": 0.5689,
      "step": 3794
    },
    {
      "epoch": 0.3422464715696442,
      "grad_norm": 0.6331674498686833,
      "learning_rate": 3.0615720097616063e-06,
      "loss": 0.5717,
      "step": 3795
    },
    {
      "epoch": 0.3423366550931145,
      "grad_norm": 1.5866454850847955,
      "learning_rate": 3.0610768425362967e-06,
      "loss": 0.7297,
      "step": 3796
    },
    {
      "epoch": 0.34242683861658474,
      "grad_norm": 1.9890988810314352,
      "learning_rate": 3.0605815847740603e-06,
      "loss": 0.7522,
      "step": 3797
    },
    {
      "epoch": 0.342517022140055,
      "grad_norm": 2.090564525054596,
      "learning_rate": 3.0600862365171553e-06,
      "loss": 0.743,
      "step": 3798
    },
    {
      "epoch": 0.34260720566352526,
      "grad_norm": 1.4781238031684918,
      "learning_rate": 3.0595907978078474e-06,
      "loss": 0.7549,
      "step": 3799
    },
    {
      "epoch": 0.34269738918699555,
      "grad_norm": 1.710776682920855,
      "learning_rate": 3.05909526868841e-06,
      "loss": 0.744,
      "step": 3800
    },
    {
      "epoch": 0.3427875727104658,
      "grad_norm": 1.7134776311192952,
      "learning_rate": 3.0585996492011243e-06,
      "loss": 0.7842,
      "step": 3801
    },
    {
      "epoch": 0.34287775623393607,
      "grad_norm": 1.9013900716273429,
      "learning_rate": 3.05810393938828e-06,
      "loss": 0.8192,
      "step": 3802
    },
    {
      "epoch": 0.3429679397574063,
      "grad_norm": 1.7625373917529439,
      "learning_rate": 3.0576081392921723e-06,
      "loss": 0.6871,
      "step": 3803
    },
    {
      "epoch": 0.3430581232808766,
      "grad_norm": 1.6356634775838994,
      "learning_rate": 3.057112248955107e-06,
      "loss": 0.8163,
      "step": 3804
    },
    {
      "epoch": 0.3431483068043468,
      "grad_norm": 4.042500020005339,
      "learning_rate": 3.0566162684193963e-06,
      "loss": 0.7948,
      "step": 3805
    },
    {
      "epoch": 0.3432384903278171,
      "grad_norm": 2.0113076984481797,
      "learning_rate": 3.056120197727359e-06,
      "loss": 0.7224,
      "step": 3806
    },
    {
      "epoch": 0.34332867385128735,
      "grad_norm": 1.669183579027665,
      "learning_rate": 3.0556240369213236e-06,
      "loss": 0.7607,
      "step": 3807
    },
    {
      "epoch": 0.34341885737475764,
      "grad_norm": 2.0550637716529145,
      "learning_rate": 3.055127786043624e-06,
      "loss": 0.6865,
      "step": 3808
    },
    {
      "epoch": 0.34350904089822787,
      "grad_norm": 1.476346358369199,
      "learning_rate": 3.054631445136604e-06,
      "loss": 0.8569,
      "step": 3809
    },
    {
      "epoch": 0.34359922442169816,
      "grad_norm": 11.305717331720514,
      "learning_rate": 3.0541350142426147e-06,
      "loss": 0.7866,
      "step": 3810
    },
    {
      "epoch": 0.3436894079451684,
      "grad_norm": 1.7669170007345396,
      "learning_rate": 3.053638493404012e-06,
      "loss": 0.7273,
      "step": 3811
    },
    {
      "epoch": 0.3437795914686387,
      "grad_norm": 1.5937562082207133,
      "learning_rate": 3.0531418826631643e-06,
      "loss": 0.7114,
      "step": 3812
    },
    {
      "epoch": 0.34386977499210897,
      "grad_norm": 1.7549808328348073,
      "learning_rate": 3.052645182062444e-06,
      "loss": 0.8522,
      "step": 3813
    },
    {
      "epoch": 0.3439599585155792,
      "grad_norm": 2.4896592856171047,
      "learning_rate": 3.0521483916442324e-06,
      "loss": 0.816,
      "step": 3814
    },
    {
      "epoch": 0.3440501420390495,
      "grad_norm": 1.795620555029047,
      "learning_rate": 3.0516515114509183e-06,
      "loss": 0.7663,
      "step": 3815
    },
    {
      "epoch": 0.3441403255625197,
      "grad_norm": 1.3956260764514616,
      "learning_rate": 3.0511545415249e-06,
      "loss": 0.7206,
      "step": 3816
    },
    {
      "epoch": 0.34423050908599,
      "grad_norm": 1.7077741948445007,
      "learning_rate": 3.050657481908579e-06,
      "loss": 0.8079,
      "step": 3817
    },
    {
      "epoch": 0.34432069260946024,
      "grad_norm": 1.6870307976489898,
      "learning_rate": 3.0501603326443677e-06,
      "loss": 0.8834,
      "step": 3818
    },
    {
      "epoch": 0.34441087613293053,
      "grad_norm": 1.3577350502181496,
      "learning_rate": 3.049663093774687e-06,
      "loss": 0.7548,
      "step": 3819
    },
    {
      "epoch": 0.34450105965640077,
      "grad_norm": 2.14693784219201,
      "learning_rate": 3.0491657653419643e-06,
      "loss": 0.6724,
      "step": 3820
    },
    {
      "epoch": 0.34459124317987105,
      "grad_norm": 1.5972182739135299,
      "learning_rate": 3.0486683473886325e-06,
      "loss": 0.819,
      "step": 3821
    },
    {
      "epoch": 0.3446814267033413,
      "grad_norm": 2.3690650468738887,
      "learning_rate": 3.0481708399571355e-06,
      "loss": 0.8415,
      "step": 3822
    },
    {
      "epoch": 0.3447716102268116,
      "grad_norm": 1.8931091004664578,
      "learning_rate": 3.047673243089922e-06,
      "loss": 0.7998,
      "step": 3823
    },
    {
      "epoch": 0.3448617937502818,
      "grad_norm": 1.5515711235736402,
      "learning_rate": 3.047175556829451e-06,
      "loss": 0.8053,
      "step": 3824
    },
    {
      "epoch": 0.3449519772737521,
      "grad_norm": 2.487600571789379,
      "learning_rate": 3.046677781218188e-06,
      "loss": 0.8208,
      "step": 3825
    },
    {
      "epoch": 0.34504216079722233,
      "grad_norm": 1.5870003732537814,
      "learning_rate": 3.0461799162986043e-06,
      "loss": 0.7521,
      "step": 3826
    },
    {
      "epoch": 0.3451323443206926,
      "grad_norm": 6.200382270103203,
      "learning_rate": 3.045681962113183e-06,
      "loss": 0.6987,
      "step": 3827
    },
    {
      "epoch": 0.34522252784416285,
      "grad_norm": 1.701764604600241,
      "learning_rate": 3.0451839187044095e-06,
      "loss": 0.7226,
      "step": 3828
    },
    {
      "epoch": 0.34531271136763314,
      "grad_norm": 4.250727647311248,
      "learning_rate": 3.0446857861147816e-06,
      "loss": 0.7902,
      "step": 3829
    },
    {
      "epoch": 0.3454028948911034,
      "grad_norm": 3.319632570547889,
      "learning_rate": 3.044187564386802e-06,
      "loss": 0.7456,
      "step": 3830
    },
    {
      "epoch": 0.34549307841457366,
      "grad_norm": 1.9374993847261497,
      "learning_rate": 3.0436892535629818e-06,
      "loss": 0.8367,
      "step": 3831
    },
    {
      "epoch": 0.3455832619380439,
      "grad_norm": 1.7933026835978887,
      "learning_rate": 3.0431908536858393e-06,
      "loss": 0.856,
      "step": 3832
    },
    {
      "epoch": 0.3456734454615142,
      "grad_norm": 1.9302950995168775,
      "learning_rate": 3.0426923647979016e-06,
      "loss": 0.762,
      "step": 3833
    },
    {
      "epoch": 0.3457636289849844,
      "grad_norm": 1.9884061942722895,
      "learning_rate": 3.0421937869417016e-06,
      "loss": 0.7148,
      "step": 3834
    },
    {
      "epoch": 0.3458538125084547,
      "grad_norm": 1.9787169765432595,
      "learning_rate": 3.041695120159782e-06,
      "loss": 0.8379,
      "step": 3835
    },
    {
      "epoch": 0.345943996031925,
      "grad_norm": 5.49650046247468,
      "learning_rate": 3.04119636449469e-06,
      "loss": 0.7068,
      "step": 3836
    },
    {
      "epoch": 0.3460341795553952,
      "grad_norm": 2.3050694036068275,
      "learning_rate": 3.040697519988983e-06,
      "loss": 0.8075,
      "step": 3837
    },
    {
      "epoch": 0.3461243630788655,
      "grad_norm": 0.6433766977077665,
      "learning_rate": 3.040198586685226e-06,
      "loss": 0.5914,
      "step": 3838
    },
    {
      "epoch": 0.34621454660233575,
      "grad_norm": 1.8566991962525836,
      "learning_rate": 3.039699564625989e-06,
      "loss": 0.7269,
      "step": 3839
    },
    {
      "epoch": 0.34630473012580604,
      "grad_norm": 0.6771407151971583,
      "learning_rate": 3.039200453853853e-06,
      "loss": 0.5853,
      "step": 3840
    },
    {
      "epoch": 0.34639491364927627,
      "grad_norm": 1.8275285709720426,
      "learning_rate": 3.038701254411404e-06,
      "loss": 0.8233,
      "step": 3841
    },
    {
      "epoch": 0.34648509717274656,
      "grad_norm": 1.8239000846701727,
      "learning_rate": 3.0382019663412367e-06,
      "loss": 0.7951,
      "step": 3842
    },
    {
      "epoch": 0.3465752806962168,
      "grad_norm": 2.4830221655382867,
      "learning_rate": 3.0377025896859532e-06,
      "loss": 0.8127,
      "step": 3843
    },
    {
      "epoch": 0.3466654642196871,
      "grad_norm": 1.591685435132854,
      "learning_rate": 3.0372031244881627e-06,
      "loss": 0.8778,
      "step": 3844
    },
    {
      "epoch": 0.3467556477431573,
      "grad_norm": 1.5607025488979234,
      "learning_rate": 3.0367035707904826e-06,
      "loss": 0.7336,
      "step": 3845
    },
    {
      "epoch": 0.3468458312666276,
      "grad_norm": 1.6888355339068835,
      "learning_rate": 3.036203928635537e-06,
      "loss": 0.689,
      "step": 3846
    },
    {
      "epoch": 0.34693601479009784,
      "grad_norm": 1.8448922449915288,
      "learning_rate": 3.035704198065959e-06,
      "loss": 0.8429,
      "step": 3847
    },
    {
      "epoch": 0.3470261983135681,
      "grad_norm": 2.858135684402352,
      "learning_rate": 3.0352043791243886e-06,
      "loss": 0.7545,
      "step": 3848
    },
    {
      "epoch": 0.34711638183703836,
      "grad_norm": 0.6966976555939377,
      "learning_rate": 3.034704471853472e-06,
      "loss": 0.5896,
      "step": 3849
    },
    {
      "epoch": 0.34720656536050865,
      "grad_norm": 1.42371467379048,
      "learning_rate": 3.0342044762958646e-06,
      "loss": 0.8219,
      "step": 3850
    },
    {
      "epoch": 0.3472967488839789,
      "grad_norm": 1.9510916539713492,
      "learning_rate": 3.0337043924942286e-06,
      "loss": 0.8107,
      "step": 3851
    },
    {
      "epoch": 0.34738693240744917,
      "grad_norm": 1.5924388503666262,
      "learning_rate": 3.0332042204912343e-06,
      "loss": 0.763,
      "step": 3852
    },
    {
      "epoch": 0.3474771159309194,
      "grad_norm": 0.6788640569957899,
      "learning_rate": 3.0327039603295587e-06,
      "loss": 0.5935,
      "step": 3853
    },
    {
      "epoch": 0.3475672994543897,
      "grad_norm": 2.3076394007804653,
      "learning_rate": 3.032203612051887e-06,
      "loss": 0.733,
      "step": 3854
    },
    {
      "epoch": 0.3476574829778599,
      "grad_norm": 2.3474526213881006,
      "learning_rate": 3.0317031757009116e-06,
      "loss": 0.757,
      "step": 3855
    },
    {
      "epoch": 0.3477476665013302,
      "grad_norm": 1.9114316960972513,
      "learning_rate": 3.0312026513193326e-06,
      "loss": 0.7351,
      "step": 3856
    },
    {
      "epoch": 0.34783785002480044,
      "grad_norm": 1.6362854012899117,
      "learning_rate": 3.0307020389498573e-06,
      "loss": 0.7473,
      "step": 3857
    },
    {
      "epoch": 0.34792803354827073,
      "grad_norm": 1.5870570849011272,
      "learning_rate": 3.0302013386352004e-06,
      "loss": 0.703,
      "step": 3858
    },
    {
      "epoch": 0.348018217071741,
      "grad_norm": 1.4050122641981821,
      "learning_rate": 3.0297005504180854e-06,
      "loss": 0.7662,
      "step": 3859
    },
    {
      "epoch": 0.34810840059521125,
      "grad_norm": 0.6053607475183085,
      "learning_rate": 3.0291996743412417e-06,
      "loss": 0.5131,
      "step": 3860
    },
    {
      "epoch": 0.34819858411868154,
      "grad_norm": 1.5243840333765566,
      "learning_rate": 3.0286987104474063e-06,
      "loss": 0.8667,
      "step": 3861
    },
    {
      "epoch": 0.3482887676421518,
      "grad_norm": 1.51316311069349,
      "learning_rate": 3.028197658779325e-06,
      "loss": 0.7423,
      "step": 3862
    },
    {
      "epoch": 0.34837895116562206,
      "grad_norm": 1.8913508511497212,
      "learning_rate": 3.0276965193797503e-06,
      "loss": 0.7893,
      "step": 3863
    },
    {
      "epoch": 0.3484691346890923,
      "grad_norm": 1.4541443561638687,
      "learning_rate": 3.0271952922914423e-06,
      "loss": 0.7352,
      "step": 3864
    },
    {
      "epoch": 0.3485593182125626,
      "grad_norm": 1.786378471287323,
      "learning_rate": 3.0266939775571675e-06,
      "loss": 0.7518,
      "step": 3865
    },
    {
      "epoch": 0.3486495017360328,
      "grad_norm": 1.6819174225275693,
      "learning_rate": 3.026192575219701e-06,
      "loss": 0.7601,
      "step": 3866
    },
    {
      "epoch": 0.3487396852595031,
      "grad_norm": 1.4383976662851312,
      "learning_rate": 3.025691085321826e-06,
      "loss": 0.6671,
      "step": 3867
    },
    {
      "epoch": 0.34882986878297334,
      "grad_norm": 2.269992065919967,
      "learning_rate": 3.025189507906332e-06,
      "loss": 0.7542,
      "step": 3868
    },
    {
      "epoch": 0.34892005230644363,
      "grad_norm": 1.8981544063778533,
      "learning_rate": 3.0246878430160166e-06,
      "loss": 0.7901,
      "step": 3869
    },
    {
      "epoch": 0.34901023582991386,
      "grad_norm": 2.0061051646208767,
      "learning_rate": 3.024186090693684e-06,
      "loss": 0.7073,
      "step": 3870
    },
    {
      "epoch": 0.34910041935338415,
      "grad_norm": 2.366387864825721,
      "learning_rate": 3.023684250982147e-06,
      "loss": 0.7976,
      "step": 3871
    },
    {
      "epoch": 0.3491906028768544,
      "grad_norm": 1.7574278007860198,
      "learning_rate": 3.0231823239242252e-06,
      "loss": 0.6909,
      "step": 3872
    },
    {
      "epoch": 0.34928078640032467,
      "grad_norm": 2.2333257385618213,
      "learning_rate": 3.0226803095627457e-06,
      "loss": 0.7259,
      "step": 3873
    },
    {
      "epoch": 0.3493709699237949,
      "grad_norm": 2.2151184449915533,
      "learning_rate": 3.022178207940543e-06,
      "loss": 0.7548,
      "step": 3874
    },
    {
      "epoch": 0.3494611534472652,
      "grad_norm": 1.647978877047933,
      "learning_rate": 3.02167601910046e-06,
      "loss": 0.7806,
      "step": 3875
    },
    {
      "epoch": 0.3495513369707354,
      "grad_norm": 1.9058930735141117,
      "learning_rate": 3.021173743085345e-06,
      "loss": 0.7944,
      "step": 3876
    },
    {
      "epoch": 0.3496415204942057,
      "grad_norm": 0.6497504030191918,
      "learning_rate": 3.0206713799380557e-06,
      "loss": 0.5796,
      "step": 3877
    },
    {
      "epoch": 0.34973170401767595,
      "grad_norm": 1.6975755485386257,
      "learning_rate": 3.0201689297014565e-06,
      "loss": 0.7782,
      "step": 3878
    },
    {
      "epoch": 0.34982188754114624,
      "grad_norm": 1.4164843815980313,
      "learning_rate": 3.0196663924184187e-06,
      "loss": 0.8161,
      "step": 3879
    },
    {
      "epoch": 0.34991207106461647,
      "grad_norm": 2.2530310665448905,
      "learning_rate": 3.019163768131822e-06,
      "loss": 0.8096,
      "step": 3880
    },
    {
      "epoch": 0.35000225458808676,
      "grad_norm": 1.7288300826251966,
      "learning_rate": 3.0186610568845533e-06,
      "loss": 0.691,
      "step": 3881
    },
    {
      "epoch": 0.350092438111557,
      "grad_norm": 7.858194736567825,
      "learning_rate": 3.018158258719507e-06,
      "loss": 0.7078,
      "step": 3882
    },
    {
      "epoch": 0.3501826216350273,
      "grad_norm": 1.4293048117167846,
      "learning_rate": 3.0176553736795827e-06,
      "loss": 0.7964,
      "step": 3883
    },
    {
      "epoch": 0.35027280515849757,
      "grad_norm": 0.5320081630184491,
      "learning_rate": 3.017152401807691e-06,
      "loss": 0.5174,
      "step": 3884
    },
    {
      "epoch": 0.3503629886819678,
      "grad_norm": 3.019882280442722,
      "learning_rate": 3.0166493431467476e-06,
      "loss": 0.7559,
      "step": 3885
    },
    {
      "epoch": 0.3504531722054381,
      "grad_norm": 1.5331698468721298,
      "learning_rate": 3.016146197739677e-06,
      "loss": 0.7442,
      "step": 3886
    },
    {
      "epoch": 0.3505433557289083,
      "grad_norm": 2.328654286684855,
      "learning_rate": 3.0156429656294097e-06,
      "loss": 0.7046,
      "step": 3887
    },
    {
      "epoch": 0.3506335392523786,
      "grad_norm": 1.933880224102865,
      "learning_rate": 3.0151396468588844e-06,
      "loss": 0.7107,
      "step": 3888
    },
    {
      "epoch": 0.35072372277584885,
      "grad_norm": 1.904236464887156,
      "learning_rate": 3.014636241471047e-06,
      "loss": 0.724,
      "step": 3889
    },
    {
      "epoch": 0.35081390629931913,
      "grad_norm": 3.4565398993157195,
      "learning_rate": 3.0141327495088514e-06,
      "loss": 0.7661,
      "step": 3890
    },
    {
      "epoch": 0.35090408982278937,
      "grad_norm": 2.1733606320794423,
      "learning_rate": 3.0136291710152566e-06,
      "loss": 0.7279,
      "step": 3891
    },
    {
      "epoch": 0.35099427334625966,
      "grad_norm": 3.063857264358891,
      "learning_rate": 3.0131255060332325e-06,
      "loss": 0.8063,
      "step": 3892
    },
    {
      "epoch": 0.3510844568697299,
      "grad_norm": 1.5684805285260888,
      "learning_rate": 3.012621754605754e-06,
      "loss": 0.7911,
      "step": 3893
    },
    {
      "epoch": 0.3511746403932002,
      "grad_norm": 1.7387811863376577,
      "learning_rate": 3.0121179167758035e-06,
      "loss": 0.791,
      "step": 3894
    },
    {
      "epoch": 0.3512648239166704,
      "grad_norm": 2.021038739753756,
      "learning_rate": 3.0116139925863717e-06,
      "loss": 0.8431,
      "step": 3895
    },
    {
      "epoch": 0.3513550074401407,
      "grad_norm": 1.6462940182952384,
      "learning_rate": 3.011109982080456e-06,
      "loss": 0.8078,
      "step": 3896
    },
    {
      "epoch": 0.35144519096361093,
      "grad_norm": 1.8043077077497052,
      "learning_rate": 3.0106058853010614e-06,
      "loss": 0.7147,
      "step": 3897
    },
    {
      "epoch": 0.3515353744870812,
      "grad_norm": 1.8306459179142545,
      "learning_rate": 3.010101702291201e-06,
      "loss": 0.7525,
      "step": 3898
    },
    {
      "epoch": 0.35162555801055145,
      "grad_norm": 1.7068084641251426,
      "learning_rate": 3.009597433093893e-06,
      "loss": 0.7629,
      "step": 3899
    },
    {
      "epoch": 0.35171574153402174,
      "grad_norm": 1.7838544710999018,
      "learning_rate": 3.009093077752165e-06,
      "loss": 0.7779,
      "step": 3900
    },
    {
      "epoch": 0.351805925057492,
      "grad_norm": 1.7368450275043685,
      "learning_rate": 3.008588636309052e-06,
      "loss": 0.7961,
      "step": 3901
    },
    {
      "epoch": 0.35189610858096226,
      "grad_norm": 2.0133576171033662,
      "learning_rate": 3.0080841088075947e-06,
      "loss": 0.7354,
      "step": 3902
    },
    {
      "epoch": 0.3519862921044325,
      "grad_norm": 1.8627046536650083,
      "learning_rate": 3.0075794952908436e-06,
      "loss": 0.8148,
      "step": 3903
    },
    {
      "epoch": 0.3520764756279028,
      "grad_norm": 2.4131942224540577,
      "learning_rate": 3.0070747958018528e-06,
      "loss": 0.7848,
      "step": 3904
    },
    {
      "epoch": 0.352166659151373,
      "grad_norm": 1.4286032979680146,
      "learning_rate": 3.0065700103836894e-06,
      "loss": 0.7835,
      "step": 3905
    },
    {
      "epoch": 0.3522568426748433,
      "grad_norm": 2.1722185700178,
      "learning_rate": 3.0060651390794214e-06,
      "loss": 0.7874,
      "step": 3906
    },
    {
      "epoch": 0.3523470261983136,
      "grad_norm": 1.7211910599512064,
      "learning_rate": 3.005560181932128e-06,
      "loss": 0.749,
      "step": 3907
    },
    {
      "epoch": 0.35243720972178383,
      "grad_norm": 1.6080483459160502,
      "learning_rate": 3.005055138984896e-06,
      "loss": 0.7257,
      "step": 3908
    },
    {
      "epoch": 0.3525273932452541,
      "grad_norm": 3.2005055207066477,
      "learning_rate": 3.0045500102808174e-06,
      "loss": 0.7432,
      "step": 3909
    },
    {
      "epoch": 0.35261757676872435,
      "grad_norm": 2.2276685677911794,
      "learning_rate": 3.0040447958629927e-06,
      "loss": 0.8517,
      "step": 3910
    },
    {
      "epoch": 0.35270776029219464,
      "grad_norm": 1.972209555028553,
      "learning_rate": 3.00353949577453e-06,
      "loss": 0.7868,
      "step": 3911
    },
    {
      "epoch": 0.35279794381566487,
      "grad_norm": 2.3357756185415885,
      "learning_rate": 3.003034110058544e-06,
      "loss": 0.6054,
      "step": 3912
    },
    {
      "epoch": 0.35288812733913516,
      "grad_norm": 1.7170346109571029,
      "learning_rate": 3.002528638758157e-06,
      "loss": 0.7454,
      "step": 3913
    },
    {
      "epoch": 0.3529783108626054,
      "grad_norm": 2.6747453229652742,
      "learning_rate": 3.0020230819164985e-06,
      "loss": 0.7591,
      "step": 3914
    },
    {
      "epoch": 0.3530684943860757,
      "grad_norm": 1.6466845552474088,
      "learning_rate": 3.0015174395767064e-06,
      "loss": 0.7713,
      "step": 3915
    },
    {
      "epoch": 0.3531586779095459,
      "grad_norm": 1.6724373415398717,
      "learning_rate": 3.001011711781923e-06,
      "loss": 0.7839,
      "step": 3916
    },
    {
      "epoch": 0.3532488614330162,
      "grad_norm": 1.6084131866686753,
      "learning_rate": 3.0005058985753017e-06,
      "loss": 0.7214,
      "step": 3917
    },
    {
      "epoch": 0.35333904495648644,
      "grad_norm": 1.4927057295051362,
      "learning_rate": 3e-06,
      "loss": 0.7543,
      "step": 3918
    },
    {
      "epoch": 0.3534292284799567,
      "grad_norm": 0.6546599789755019,
      "learning_rate": 2.9994940160991843e-06,
      "loss": 0.6063,
      "step": 3919
    },
    {
      "epoch": 0.35351941200342696,
      "grad_norm": 0.627977313049508,
      "learning_rate": 2.9989879469160285e-06,
      "loss": 0.5633,
      "step": 3920
    },
    {
      "epoch": 0.35360959552689725,
      "grad_norm": 3.8476965267958563,
      "learning_rate": 2.9984817924937124e-06,
      "loss": 0.7833,
      "step": 3921
    },
    {
      "epoch": 0.3536997790503675,
      "grad_norm": 4.060877784210687,
      "learning_rate": 2.997975552875424e-06,
      "loss": 0.7079,
      "step": 3922
    },
    {
      "epoch": 0.35378996257383777,
      "grad_norm": 1.477511307984534,
      "learning_rate": 2.997469228104358e-06,
      "loss": 0.756,
      "step": 3923
    },
    {
      "epoch": 0.353880146097308,
      "grad_norm": 1.6365554473183048,
      "learning_rate": 2.996962818223718e-06,
      "loss": 0.6607,
      "step": 3924
    },
    {
      "epoch": 0.3539703296207783,
      "grad_norm": 2.3986845106733785,
      "learning_rate": 2.9964563232767135e-06,
      "loss": 0.7144,
      "step": 3925
    },
    {
      "epoch": 0.3540605131442485,
      "grad_norm": 1.94739699614059,
      "learning_rate": 2.9959497433065617e-06,
      "loss": 0.7633,
      "step": 3926
    },
    {
      "epoch": 0.3541506966677188,
      "grad_norm": 1.6184512742762511,
      "learning_rate": 2.9954430783564848e-06,
      "loss": 0.6859,
      "step": 3927
    },
    {
      "epoch": 0.35424088019118904,
      "grad_norm": 2.2362540286919694,
      "learning_rate": 2.994936328469716e-06,
      "loss": 0.7949,
      "step": 3928
    },
    {
      "epoch": 0.35433106371465933,
      "grad_norm": 1.661633505201773,
      "learning_rate": 2.994429493689494e-06,
      "loss": 0.7657,
      "step": 3929
    },
    {
      "epoch": 0.35442124723812957,
      "grad_norm": 1.466035160022973,
      "learning_rate": 2.9939225740590642e-06,
      "loss": 0.7903,
      "step": 3930
    },
    {
      "epoch": 0.35451143076159986,
      "grad_norm": 6.549997682061404,
      "learning_rate": 2.99341556962168e-06,
      "loss": 0.725,
      "step": 3931
    },
    {
      "epoch": 0.35460161428507014,
      "grad_norm": 1.6250652153427017,
      "learning_rate": 2.992908480420602e-06,
      "loss": 0.8109,
      "step": 3932
    },
    {
      "epoch": 0.3546917978085404,
      "grad_norm": 2.3144690919478244,
      "learning_rate": 2.9924013064990974e-06,
      "loss": 0.755,
      "step": 3933
    },
    {
      "epoch": 0.35478198133201067,
      "grad_norm": 0.6605673418534416,
      "learning_rate": 2.991894047900441e-06,
      "loss": 0.6113,
      "step": 3934
    },
    {
      "epoch": 0.3548721648554809,
      "grad_norm": 1.6974309526875742,
      "learning_rate": 2.991386704667916e-06,
      "loss": 0.7816,
      "step": 3935
    },
    {
      "epoch": 0.3549623483789512,
      "grad_norm": 1.8932299362523868,
      "learning_rate": 2.9908792768448097e-06,
      "loss": 0.8282,
      "step": 3936
    },
    {
      "epoch": 0.3550525319024214,
      "grad_norm": 3.519170575390782,
      "learning_rate": 2.990371764474421e-06,
      "loss": 0.7631,
      "step": 3937
    },
    {
      "epoch": 0.3551427154258917,
      "grad_norm": 1.969837070253148,
      "learning_rate": 2.9898641676000518e-06,
      "loss": 0.7335,
      "step": 3938
    },
    {
      "epoch": 0.35523289894936194,
      "grad_norm": 1.5754317161219924,
      "learning_rate": 2.9893564862650138e-06,
      "loss": 0.7654,
      "step": 3939
    },
    {
      "epoch": 0.35532308247283223,
      "grad_norm": 4.084889381433103,
      "learning_rate": 2.9888487205126254e-06,
      "loss": 0.8185,
      "step": 3940
    },
    {
      "epoch": 0.35541326599630246,
      "grad_norm": 1.46546886076003,
      "learning_rate": 2.9883408703862115e-06,
      "loss": 0.7605,
      "step": 3941
    },
    {
      "epoch": 0.35550344951977275,
      "grad_norm": 1.8845475935458562,
      "learning_rate": 2.987832935929105e-06,
      "loss": 0.847,
      "step": 3942
    },
    {
      "epoch": 0.355593633043243,
      "grad_norm": 1.610256287096836,
      "learning_rate": 2.9873249171846454e-06,
      "loss": 0.8308,
      "step": 3943
    },
    {
      "epoch": 0.3556838165667133,
      "grad_norm": 1.8179076849661426,
      "learning_rate": 2.98681681419618e-06,
      "loss": 0.7597,
      "step": 3944
    },
    {
      "epoch": 0.3557740000901835,
      "grad_norm": 1.8068652204145312,
      "learning_rate": 2.9863086270070627e-06,
      "loss": 0.8199,
      "step": 3945
    },
    {
      "epoch": 0.3558641836136538,
      "grad_norm": 2.355891504512968,
      "learning_rate": 2.985800355660655e-06,
      "loss": 0.7732,
      "step": 3946
    },
    {
      "epoch": 0.35595436713712403,
      "grad_norm": 0.6746701494241366,
      "learning_rate": 2.9852920002003252e-06,
      "loss": 0.609,
      "step": 3947
    },
    {
      "epoch": 0.3560445506605943,
      "grad_norm": 1.880764238876888,
      "learning_rate": 2.9847835606694494e-06,
      "loss": 0.8056,
      "step": 3948
    },
    {
      "epoch": 0.35613473418406455,
      "grad_norm": 2.2398529127057967,
      "learning_rate": 2.98427503711141e-06,
      "loss": 0.694,
      "step": 3949
    },
    {
      "epoch": 0.35622491770753484,
      "grad_norm": 1.7408341424239222,
      "learning_rate": 2.9837664295695973e-06,
      "loss": 0.7994,
      "step": 3950
    },
    {
      "epoch": 0.35631510123100507,
      "grad_norm": 2.495903951147903,
      "learning_rate": 2.983257738087408e-06,
      "loss": 0.7582,
      "step": 3951
    },
    {
      "epoch": 0.35640528475447536,
      "grad_norm": 1.9626933591796678,
      "learning_rate": 2.982748962708247e-06,
      "loss": 0.799,
      "step": 3952
    },
    {
      "epoch": 0.3564954682779456,
      "grad_norm": 1.5867892829548498,
      "learning_rate": 2.982240103475526e-06,
      "loss": 0.8173,
      "step": 3953
    },
    {
      "epoch": 0.3565856518014159,
      "grad_norm": 2.1359972528947173,
      "learning_rate": 2.981731160432663e-06,
      "loss": 0.7684,
      "step": 3954
    },
    {
      "epoch": 0.35667583532488617,
      "grad_norm": 1.568291269536216,
      "learning_rate": 2.981222133623084e-06,
      "loss": 0.6706,
      "step": 3955
    },
    {
      "epoch": 0.3567660188483564,
      "grad_norm": 1.6146055240541803,
      "learning_rate": 2.980713023090222e-06,
      "loss": 0.7742,
      "step": 3956
    },
    {
      "epoch": 0.3568562023718267,
      "grad_norm": 2.935092994997301,
      "learning_rate": 2.980203828877518e-06,
      "loss": 0.7391,
      "step": 3957
    },
    {
      "epoch": 0.3569463858952969,
      "grad_norm": 2.1808988971740124,
      "learning_rate": 2.9796945510284182e-06,
      "loss": 0.8063,
      "step": 3958
    },
    {
      "epoch": 0.3570365694187672,
      "grad_norm": 3.3446381717335125,
      "learning_rate": 2.9791851895863774e-06,
      "loss": 0.8226,
      "step": 3959
    },
    {
      "epoch": 0.35712675294223745,
      "grad_norm": 2.435781019468845,
      "learning_rate": 2.978675744594857e-06,
      "loss": 0.7965,
      "step": 3960
    },
    {
      "epoch": 0.35721693646570774,
      "grad_norm": 1.3551782302191067,
      "learning_rate": 2.978166216097326e-06,
      "loss": 0.7235,
      "step": 3961
    },
    {
      "epoch": 0.35730711998917797,
      "grad_norm": 1.966049097641396,
      "learning_rate": 2.9776566041372596e-06,
      "loss": 0.7541,
      "step": 3962
    },
    {
      "epoch": 0.35739730351264826,
      "grad_norm": 1.7623873032055821,
      "learning_rate": 2.977146908758141e-06,
      "loss": 0.8239,
      "step": 3963
    },
    {
      "epoch": 0.3574874870361185,
      "grad_norm": 1.5324071287855134,
      "learning_rate": 2.9766371300034604e-06,
      "loss": 0.7827,
      "step": 3964
    },
    {
      "epoch": 0.3575776705595888,
      "grad_norm": 1.6983636919002978,
      "learning_rate": 2.9761272679167142e-06,
      "loss": 0.8174,
      "step": 3965
    },
    {
      "epoch": 0.357667854083059,
      "grad_norm": 1.9253630580619676,
      "learning_rate": 2.9756173225414072e-06,
      "loss": 0.6153,
      "step": 3966
    },
    {
      "epoch": 0.3577580376065293,
      "grad_norm": 1.84107304869384,
      "learning_rate": 2.975107293921051e-06,
      "loss": 0.9427,
      "step": 3967
    },
    {
      "epoch": 0.35784822112999953,
      "grad_norm": 2.3272802881435015,
      "learning_rate": 2.9745971820991643e-06,
      "loss": 0.8067,
      "step": 3968
    },
    {
      "epoch": 0.3579384046534698,
      "grad_norm": 1.6451469954002924,
      "learning_rate": 2.9740869871192715e-06,
      "loss": 0.8088,
      "step": 3969
    },
    {
      "epoch": 0.35802858817694005,
      "grad_norm": 1.9442630986011207,
      "learning_rate": 2.9735767090249065e-06,
      "loss": 0.8307,
      "step": 3970
    },
    {
      "epoch": 0.35811877170041034,
      "grad_norm": 1.5280280178007877,
      "learning_rate": 2.973066347859608e-06,
      "loss": 0.6808,
      "step": 3971
    },
    {
      "epoch": 0.3582089552238806,
      "grad_norm": 1.3314453357787743,
      "learning_rate": 2.972555903666923e-06,
      "loss": 0.7793,
      "step": 3972
    },
    {
      "epoch": 0.35829913874735086,
      "grad_norm": 2.0201601565483536,
      "learning_rate": 2.972045376490406e-06,
      "loss": 0.6837,
      "step": 3973
    },
    {
      "epoch": 0.3583893222708211,
      "grad_norm": 1.4330930745045387,
      "learning_rate": 2.9715347663736177e-06,
      "loss": 0.6934,
      "step": 3974
    },
    {
      "epoch": 0.3584795057942914,
      "grad_norm": 1.5130553337361419,
      "learning_rate": 2.9710240733601266e-06,
      "loss": 0.767,
      "step": 3975
    },
    {
      "epoch": 0.3585696893177616,
      "grad_norm": 1.85666356216025,
      "learning_rate": 2.970513297493507e-06,
      "loss": 0.8574,
      "step": 3976
    },
    {
      "epoch": 0.3586598728412319,
      "grad_norm": 1.6794850781917214,
      "learning_rate": 2.9700024388173416e-06,
      "loss": 0.8193,
      "step": 3977
    },
    {
      "epoch": 0.35875005636470214,
      "grad_norm": 2.1912324941907495,
      "learning_rate": 2.969491497375219e-06,
      "loss": 0.8192,
      "step": 3978
    },
    {
      "epoch": 0.35884023988817243,
      "grad_norm": 1.4052039176281472,
      "learning_rate": 2.9689804732107364e-06,
      "loss": 0.7908,
      "step": 3979
    },
    {
      "epoch": 0.3589304234116427,
      "grad_norm": 1.8955195100827236,
      "learning_rate": 2.9684693663674968e-06,
      "loss": 0.7244,
      "step": 3980
    },
    {
      "epoch": 0.35902060693511295,
      "grad_norm": 1.785563683289885,
      "learning_rate": 2.9679581768891115e-06,
      "loss": 0.8527,
      "step": 3981
    },
    {
      "epoch": 0.35911079045858324,
      "grad_norm": 1.6820731321259692,
      "learning_rate": 2.967446904819197e-06,
      "loss": 0.6903,
      "step": 3982
    },
    {
      "epoch": 0.3592009739820535,
      "grad_norm": 1.6485573964561866,
      "learning_rate": 2.966935550201378e-06,
      "loss": 0.7959,
      "step": 3983
    },
    {
      "epoch": 0.35929115750552376,
      "grad_norm": 1.5210343406607756,
      "learning_rate": 2.966424113079286e-06,
      "loss": 0.7671,
      "step": 3984
    },
    {
      "epoch": 0.359381341028994,
      "grad_norm": 1.752140371125257,
      "learning_rate": 2.9659125934965596e-06,
      "loss": 0.8176,
      "step": 3985
    },
    {
      "epoch": 0.3594715245524643,
      "grad_norm": 1.6694010318281733,
      "learning_rate": 2.9654009914968457e-06,
      "loss": 0.8301,
      "step": 3986
    },
    {
      "epoch": 0.3595617080759345,
      "grad_norm": 2.0279423706398747,
      "learning_rate": 2.9648893071237956e-06,
      "loss": 0.6661,
      "step": 3987
    },
    {
      "epoch": 0.3596518915994048,
      "grad_norm": 2.2196066908350027,
      "learning_rate": 2.964377540421069e-06,
      "loss": 0.7444,
      "step": 3988
    },
    {
      "epoch": 0.35974207512287504,
      "grad_norm": 1.9155356829731809,
      "learning_rate": 2.963865691432334e-06,
      "loss": 0.7944,
      "step": 3989
    },
    {
      "epoch": 0.3598322586463453,
      "grad_norm": 2.1977550453194947,
      "learning_rate": 2.963353760201263e-06,
      "loss": 0.6891,
      "step": 3990
    },
    {
      "epoch": 0.35992244216981556,
      "grad_norm": 1.826532110087057,
      "learning_rate": 2.962841746771537e-06,
      "loss": 0.7559,
      "step": 3991
    },
    {
      "epoch": 0.36001262569328585,
      "grad_norm": 1.9525157130704447,
      "learning_rate": 2.9623296511868445e-06,
      "loss": 0.7708,
      "step": 3992
    },
    {
      "epoch": 0.3601028092167561,
      "grad_norm": 1.4802049386694485,
      "learning_rate": 2.96181747349088e-06,
      "loss": 0.6972,
      "step": 3993
    },
    {
      "epoch": 0.36019299274022637,
      "grad_norm": 1.6011618508487908,
      "learning_rate": 2.961305213727345e-06,
      "loss": 0.7906,
      "step": 3994
    },
    {
      "epoch": 0.3602831762636966,
      "grad_norm": 0.6503728686592332,
      "learning_rate": 2.960792871939949e-06,
      "loss": 0.6033,
      "step": 3995
    },
    {
      "epoch": 0.3603733597871669,
      "grad_norm": 2.0153503230812384,
      "learning_rate": 2.9602804481724064e-06,
      "loss": 0.8643,
      "step": 3996
    },
    {
      "epoch": 0.3604635433106371,
      "grad_norm": 1.7750468073301309,
      "learning_rate": 2.9597679424684427e-06,
      "loss": 0.7681,
      "step": 3997
    },
    {
      "epoch": 0.3605537268341074,
      "grad_norm": 1.8010450323046567,
      "learning_rate": 2.9592553548717848e-06,
      "loss": 0.7813,
      "step": 3998
    },
    {
      "epoch": 0.36064391035757765,
      "grad_norm": 2.6694967828657274,
      "learning_rate": 2.958742685426171e-06,
      "loss": 0.8547,
      "step": 3999
    },
    {
      "epoch": 0.36073409388104793,
      "grad_norm": 2.0655965967577283,
      "learning_rate": 2.9582299341753446e-06,
      "loss": 0.841,
      "step": 4000
    },
    {
      "epoch": 0.36082427740451817,
      "grad_norm": 11.948538744340247,
      "learning_rate": 2.957717101163057e-06,
      "loss": 0.7101,
      "step": 4001
    },
    {
      "epoch": 0.36091446092798846,
      "grad_norm": 1.9704792360549022,
      "learning_rate": 2.9572041864330655e-06,
      "loss": 0.7696,
      "step": 4002
    },
    {
      "epoch": 0.36100464445145874,
      "grad_norm": 1.6453129231306585,
      "learning_rate": 2.9566911900291346e-06,
      "loss": 0.7922,
      "step": 4003
    },
    {
      "epoch": 0.361094827974929,
      "grad_norm": 1.6419427574996759,
      "learning_rate": 2.9561781119950368e-06,
      "loss": 0.6865,
      "step": 4004
    },
    {
      "epoch": 0.36118501149839927,
      "grad_norm": 1.5033006910401807,
      "learning_rate": 2.9556649523745493e-06,
      "loss": 0.7158,
      "step": 4005
    },
    {
      "epoch": 0.3612751950218695,
      "grad_norm": 1.8998835201698583,
      "learning_rate": 2.955151711211459e-06,
      "loss": 0.832,
      "step": 4006
    },
    {
      "epoch": 0.3613653785453398,
      "grad_norm": 1.4903430024199331,
      "learning_rate": 2.9546383885495583e-06,
      "loss": 0.7396,
      "step": 4007
    },
    {
      "epoch": 0.36145556206881,
      "grad_norm": 1.9815419917541084,
      "learning_rate": 2.9541249844326464e-06,
      "loss": 0.7222,
      "step": 4008
    },
    {
      "epoch": 0.3615457455922803,
      "grad_norm": 2.076301290761606,
      "learning_rate": 2.9536114989045295e-06,
      "loss": 0.7528,
      "step": 4009
    },
    {
      "epoch": 0.36163592911575054,
      "grad_norm": 2.0499236720692346,
      "learning_rate": 2.9530979320090216e-06,
      "loss": 0.8318,
      "step": 4010
    },
    {
      "epoch": 0.36172611263922083,
      "grad_norm": 2.537378592006156,
      "learning_rate": 2.9525842837899422e-06,
      "loss": 0.8526,
      "step": 4011
    },
    {
      "epoch": 0.36181629616269106,
      "grad_norm": 1.6071109586910675,
      "learning_rate": 2.95207055429112e-06,
      "loss": 0.8529,
      "step": 4012
    },
    {
      "epoch": 0.36190647968616135,
      "grad_norm": 1.578191548067974,
      "learning_rate": 2.951556743556388e-06,
      "loss": 0.764,
      "step": 4013
    },
    {
      "epoch": 0.3619966632096316,
      "grad_norm": 1.914772493677817,
      "learning_rate": 2.951042851629588e-06,
      "loss": 0.8126,
      "step": 4014
    },
    {
      "epoch": 0.3620868467331019,
      "grad_norm": 1.7851637291282056,
      "learning_rate": 2.950528878554568e-06,
      "loss": 0.7292,
      "step": 4015
    },
    {
      "epoch": 0.3621770302565721,
      "grad_norm": 1.425498453600368,
      "learning_rate": 2.950014824375183e-06,
      "loss": 0.8058,
      "step": 4016
    },
    {
      "epoch": 0.3622672137800424,
      "grad_norm": 0.6324994525605497,
      "learning_rate": 2.949500689135295e-06,
      "loss": 0.5278,
      "step": 4017
    },
    {
      "epoch": 0.36235739730351263,
      "grad_norm": 1.5589980745872563,
      "learning_rate": 2.9489864728787722e-06,
      "loss": 0.8525,
      "step": 4018
    },
    {
      "epoch": 0.3624475808269829,
      "grad_norm": 1.6030595435151282,
      "learning_rate": 2.9484721756494915e-06,
      "loss": 0.8314,
      "step": 4019
    },
    {
      "epoch": 0.36253776435045315,
      "grad_norm": 1.6318895569717837,
      "learning_rate": 2.9479577974913343e-06,
      "loss": 0.7735,
      "step": 4020
    },
    {
      "epoch": 0.36262794787392344,
      "grad_norm": 1.5975620829617971,
      "learning_rate": 2.9474433384481908e-06,
      "loss": 0.7689,
      "step": 4021
    },
    {
      "epoch": 0.3627181313973937,
      "grad_norm": 2.030786549443199,
      "learning_rate": 2.9469287985639577e-06,
      "loss": 0.7635,
      "step": 4022
    },
    {
      "epoch": 0.36280831492086396,
      "grad_norm": 1.5400013649922055,
      "learning_rate": 2.9464141778825384e-06,
      "loss": 0.8154,
      "step": 4023
    },
    {
      "epoch": 0.3628984984443342,
      "grad_norm": 1.6182229233363388,
      "learning_rate": 2.9458994764478427e-06,
      "loss": 0.8081,
      "step": 4024
    },
    {
      "epoch": 0.3629886819678045,
      "grad_norm": 1.657404083652743,
      "learning_rate": 2.9453846943037883e-06,
      "loss": 0.8099,
      "step": 4025
    },
    {
      "epoch": 0.36307886549127477,
      "grad_norm": 1.4354426547255357,
      "learning_rate": 2.9448698314942987e-06,
      "loss": 0.7192,
      "step": 4026
    },
    {
      "epoch": 0.363169049014745,
      "grad_norm": 0.7419732838728211,
      "learning_rate": 2.944354888063305e-06,
      "loss": 0.6234,
      "step": 4027
    },
    {
      "epoch": 0.3632592325382153,
      "grad_norm": 1.6448431376190364,
      "learning_rate": 2.9438398640547453e-06,
      "loss": 0.748,
      "step": 4028
    },
    {
      "epoch": 0.3633494160616855,
      "grad_norm": 1.7061061840441,
      "learning_rate": 2.943324759512564e-06,
      "loss": 0.7137,
      "step": 4029
    },
    {
      "epoch": 0.3634395995851558,
      "grad_norm": 1.4059612825497256,
      "learning_rate": 2.9428095744807134e-06,
      "loss": 0.7941,
      "step": 4030
    },
    {
      "epoch": 0.36352978310862605,
      "grad_norm": 1.49343851584739,
      "learning_rate": 2.942294309003151e-06,
      "loss": 0.7283,
      "step": 4031
    },
    {
      "epoch": 0.36361996663209634,
      "grad_norm": 1.6637193209530126,
      "learning_rate": 2.941778963123843e-06,
      "loss": 0.7893,
      "step": 4032
    },
    {
      "epoch": 0.36371015015556657,
      "grad_norm": 5.240623456833031,
      "learning_rate": 2.94126353688676e-06,
      "loss": 0.7739,
      "step": 4033
    },
    {
      "epoch": 0.36380033367903686,
      "grad_norm": 1.8169746412020285,
      "learning_rate": 2.9407480303358825e-06,
      "loss": 0.7221,
      "step": 4034
    },
    {
      "epoch": 0.3638905172025071,
      "grad_norm": 1.5780199979087872,
      "learning_rate": 2.940232443515195e-06,
      "loss": 0.7774,
      "step": 4035
    },
    {
      "epoch": 0.3639807007259774,
      "grad_norm": 1.5054253374461173,
      "learning_rate": 2.9397167764686916e-06,
      "loss": 0.7959,
      "step": 4036
    },
    {
      "epoch": 0.3640708842494476,
      "grad_norm": 1.9557974270533067,
      "learning_rate": 2.9392010292403714e-06,
      "loss": 0.7444,
      "step": 4037
    },
    {
      "epoch": 0.3641610677729179,
      "grad_norm": 4.356988506576021,
      "learning_rate": 2.9386852018742404e-06,
      "loss": 0.7659,
      "step": 4038
    },
    {
      "epoch": 0.36425125129638813,
      "grad_norm": 1.4545014947783237,
      "learning_rate": 2.938169294414312e-06,
      "loss": 0.7469,
      "step": 4039
    },
    {
      "epoch": 0.3643414348198584,
      "grad_norm": 1.4869421994969325,
      "learning_rate": 2.9376533069046067e-06,
      "loss": 0.8012,
      "step": 4040
    },
    {
      "epoch": 0.36443161834332866,
      "grad_norm": 0.7376508089821185,
      "learning_rate": 2.9371372393891514e-06,
      "loss": 0.6353,
      "step": 4041
    },
    {
      "epoch": 0.36452180186679894,
      "grad_norm": 1.3862600508137193,
      "learning_rate": 2.936621091911979e-06,
      "loss": 0.8417,
      "step": 4042
    },
    {
      "epoch": 0.3646119853902692,
      "grad_norm": 1.9430699425250213,
      "learning_rate": 2.936104864517131e-06,
      "loss": 0.7018,
      "step": 4043
    },
    {
      "epoch": 0.36470216891373947,
      "grad_norm": 2.7099475647192572,
      "learning_rate": 2.9355885572486535e-06,
      "loss": 0.7287,
      "step": 4044
    },
    {
      "epoch": 0.3647923524372097,
      "grad_norm": 1.6527388440772492,
      "learning_rate": 2.9350721701506026e-06,
      "loss": 0.8004,
      "step": 4045
    },
    {
      "epoch": 0.36488253596068,
      "grad_norm": 2.0918257112055945,
      "learning_rate": 2.9345557032670375e-06,
      "loss": 0.7682,
      "step": 4046
    },
    {
      "epoch": 0.3649727194841502,
      "grad_norm": 2.034698493422788,
      "learning_rate": 2.934039156642027e-06,
      "loss": 0.8131,
      "step": 4047
    },
    {
      "epoch": 0.3650629030076205,
      "grad_norm": 1.9719488405164718,
      "learning_rate": 2.9335225303196454e-06,
      "loss": 0.6302,
      "step": 4048
    },
    {
      "epoch": 0.36515308653109074,
      "grad_norm": 1.73739800462561,
      "learning_rate": 2.933005824343974e-06,
      "loss": 0.7365,
      "step": 4049
    },
    {
      "epoch": 0.36524327005456103,
      "grad_norm": 8.465988328247532,
      "learning_rate": 2.932489038759101e-06,
      "loss": 0.7396,
      "step": 4050
    },
    {
      "epoch": 0.3653334535780313,
      "grad_norm": 1.4402457840759626,
      "learning_rate": 2.9319721736091215e-06,
      "loss": 0.6643,
      "step": 4051
    },
    {
      "epoch": 0.36542363710150155,
      "grad_norm": 1.3338080494912201,
      "learning_rate": 2.9314552289381377e-06,
      "loss": 0.6738,
      "step": 4052
    },
    {
      "epoch": 0.36551382062497184,
      "grad_norm": 2.019743976420191,
      "learning_rate": 2.9309382047902574e-06,
      "loss": 0.7394,
      "step": 4053
    },
    {
      "epoch": 0.3656040041484421,
      "grad_norm": 1.8188759042479261,
      "learning_rate": 2.9304211012095963e-06,
      "loss": 0.8135,
      "step": 4054
    },
    {
      "epoch": 0.36569418767191236,
      "grad_norm": 1.8407210346032434,
      "learning_rate": 2.929903918240277e-06,
      "loss": 0.8283,
      "step": 4055
    },
    {
      "epoch": 0.3657843711953826,
      "grad_norm": 0.6444680616442345,
      "learning_rate": 2.9293866559264273e-06,
      "loss": 0.5794,
      "step": 4056
    },
    {
      "epoch": 0.3658745547188529,
      "grad_norm": 1.879552844925049,
      "learning_rate": 2.928869314312184e-06,
      "loss": 0.7735,
      "step": 4057
    },
    {
      "epoch": 0.3659647382423231,
      "grad_norm": 2.2506012643008986,
      "learning_rate": 2.9283518934416892e-06,
      "loss": 0.7701,
      "step": 4058
    },
    {
      "epoch": 0.3660549217657934,
      "grad_norm": 1.6853172879652372,
      "learning_rate": 2.927834393359092e-06,
      "loss": 0.7829,
      "step": 4059
    },
    {
      "epoch": 0.36614510528926364,
      "grad_norm": 1.616856195541758,
      "learning_rate": 2.927316814108548e-06,
      "loss": 0.8448,
      "step": 4060
    },
    {
      "epoch": 0.36623528881273393,
      "grad_norm": 1.8248595066077857,
      "learning_rate": 2.92679915573422e-06,
      "loss": 0.6783,
      "step": 4061
    },
    {
      "epoch": 0.36632547233620416,
      "grad_norm": 1.571788037664576,
      "learning_rate": 2.926281418280278e-06,
      "loss": 0.7147,
      "step": 4062
    },
    {
      "epoch": 0.36641565585967445,
      "grad_norm": 1.7690758529668578,
      "learning_rate": 2.925763601790899e-06,
      "loss": 0.7775,
      "step": 4063
    },
    {
      "epoch": 0.3665058393831447,
      "grad_norm": 1.7283629269152385,
      "learning_rate": 2.9252457063102635e-06,
      "loss": 0.7156,
      "step": 4064
    },
    {
      "epoch": 0.36659602290661497,
      "grad_norm": 1.7229720537661317,
      "learning_rate": 2.9247277318825626e-06,
      "loss": 0.7294,
      "step": 4065
    },
    {
      "epoch": 0.3666862064300852,
      "grad_norm": 1.4885087928140666,
      "learning_rate": 2.924209678551993e-06,
      "loss": 0.8652,
      "step": 4066
    },
    {
      "epoch": 0.3667763899535555,
      "grad_norm": 1.988481612686655,
      "learning_rate": 2.923691546362757e-06,
      "loss": 0.7803,
      "step": 4067
    },
    {
      "epoch": 0.3668665734770257,
      "grad_norm": 1.6185945296283357,
      "learning_rate": 2.9231733353590663e-06,
      "loss": 0.7283,
      "step": 4068
    },
    {
      "epoch": 0.366956757000496,
      "grad_norm": 1.7914374966397977,
      "learning_rate": 2.922655045585136e-06,
      "loss": 0.8466,
      "step": 4069
    },
    {
      "epoch": 0.36704694052396625,
      "grad_norm": 0.7030483415988716,
      "learning_rate": 2.92213667708519e-06,
      "loss": 0.621,
      "step": 4070
    },
    {
      "epoch": 0.36713712404743654,
      "grad_norm": 1.635029214370524,
      "learning_rate": 2.921618229903457e-06,
      "loss": 0.7795,
      "step": 4071
    },
    {
      "epoch": 0.36722730757090677,
      "grad_norm": 1.553435468440242,
      "learning_rate": 2.9210997040841752e-06,
      "loss": 0.692,
      "step": 4072
    },
    {
      "epoch": 0.36731749109437706,
      "grad_norm": 1.927821504272269,
      "learning_rate": 2.9205810996715885e-06,
      "loss": 0.7316,
      "step": 4073
    },
    {
      "epoch": 0.36740767461784735,
      "grad_norm": 2.811871267614285,
      "learning_rate": 2.9200624167099456e-06,
      "loss": 0.729,
      "step": 4074
    },
    {
      "epoch": 0.3674978581413176,
      "grad_norm": 1.6278776951149265,
      "learning_rate": 2.919543655243505e-06,
      "loss": 0.6825,
      "step": 4075
    },
    {
      "epoch": 0.36758804166478787,
      "grad_norm": 1.793242190584531,
      "learning_rate": 2.919024815316529e-06,
      "loss": 0.7962,
      "step": 4076
    },
    {
      "epoch": 0.3676782251882581,
      "grad_norm": 1.8231082125346278,
      "learning_rate": 2.9185058969732877e-06,
      "loss": 0.8945,
      "step": 4077
    },
    {
      "epoch": 0.3677684087117284,
      "grad_norm": 1.5032664337554114,
      "learning_rate": 2.917986900258059e-06,
      "loss": 0.8024,
      "step": 4078
    },
    {
      "epoch": 0.3678585922351986,
      "grad_norm": 1.7894364678274406,
      "learning_rate": 2.917467825215126e-06,
      "loss": 0.7665,
      "step": 4079
    },
    {
      "epoch": 0.3679487757586689,
      "grad_norm": 1.5387922330463475,
      "learning_rate": 2.9169486718887803e-06,
      "loss": 0.7176,
      "step": 4080
    },
    {
      "epoch": 0.36803895928213914,
      "grad_norm": 1.4223526372488366,
      "learning_rate": 2.9164294403233173e-06,
      "loss": 0.818,
      "step": 4081
    },
    {
      "epoch": 0.36812914280560943,
      "grad_norm": 1.6605292406135146,
      "learning_rate": 2.915910130563041e-06,
      "loss": 0.7108,
      "step": 4082
    },
    {
      "epoch": 0.36821932632907967,
      "grad_norm": 2.2682650213258184,
      "learning_rate": 2.915390742652262e-06,
      "loss": 0.7172,
      "step": 4083
    },
    {
      "epoch": 0.36830950985254995,
      "grad_norm": 2.1513781588391945,
      "learning_rate": 2.914871276635298e-06,
      "loss": 0.7042,
      "step": 4084
    },
    {
      "epoch": 0.3683996933760202,
      "grad_norm": 2.34829139686467,
      "learning_rate": 2.914351732556472e-06,
      "loss": 0.7483,
      "step": 4085
    },
    {
      "epoch": 0.3684898768994905,
      "grad_norm": 1.3756190553681436,
      "learning_rate": 2.9138321104601144e-06,
      "loss": 0.6927,
      "step": 4086
    },
    {
      "epoch": 0.3685800604229607,
      "grad_norm": 1.6621619461657073,
      "learning_rate": 2.9133124103905623e-06,
      "loss": 0.7015,
      "step": 4087
    },
    {
      "epoch": 0.368670243946431,
      "grad_norm": 2.8019382954514422,
      "learning_rate": 2.9127926323921596e-06,
      "loss": 0.7599,
      "step": 4088
    },
    {
      "epoch": 0.36876042746990123,
      "grad_norm": 1.8983273572328103,
      "learning_rate": 2.912272776509256e-06,
      "loss": 0.7884,
      "step": 4089
    },
    {
      "epoch": 0.3688506109933715,
      "grad_norm": 1.6817592175853318,
      "learning_rate": 2.911752842786209e-06,
      "loss": 0.7364,
      "step": 4090
    },
    {
      "epoch": 0.36894079451684175,
      "grad_norm": 1.7932776224542224,
      "learning_rate": 2.911232831267383e-06,
      "loss": 0.7146,
      "step": 4091
    },
    {
      "epoch": 0.36903097804031204,
      "grad_norm": 1.7083861955356945,
      "learning_rate": 2.910712741997146e-06,
      "loss": 0.7902,
      "step": 4092
    },
    {
      "epoch": 0.3691211615637823,
      "grad_norm": 2.525053280560953,
      "learning_rate": 2.910192575019877e-06,
      "loss": 0.7579,
      "step": 4093
    },
    {
      "epoch": 0.36921134508725256,
      "grad_norm": 1.4632564642375774,
      "learning_rate": 2.9096723303799583e-06,
      "loss": 0.7967,
      "step": 4094
    },
    {
      "epoch": 0.3693015286107228,
      "grad_norm": 2.228378786504619,
      "learning_rate": 2.9091520081217805e-06,
      "loss": 0.776,
      "step": 4095
    },
    {
      "epoch": 0.3693917121341931,
      "grad_norm": 1.754929887398134,
      "learning_rate": 2.908631608289741e-06,
      "loss": 0.7148,
      "step": 4096
    },
    {
      "epoch": 0.3694818956576633,
      "grad_norm": 1.8258764670619372,
      "learning_rate": 2.9081111309282423e-06,
      "loss": 0.7599,
      "step": 4097
    },
    {
      "epoch": 0.3695720791811336,
      "grad_norm": 0.6421856362717215,
      "learning_rate": 2.9075905760816942e-06,
      "loss": 0.6326,
      "step": 4098
    },
    {
      "epoch": 0.3696622627046039,
      "grad_norm": 3.1635369511932274,
      "learning_rate": 2.907069943794514e-06,
      "loss": 0.8068,
      "step": 4099
    },
    {
      "epoch": 0.3697524462280741,
      "grad_norm": 1.6090270471806742,
      "learning_rate": 2.906549234111125e-06,
      "loss": 0.736,
      "step": 4100
    },
    {
      "epoch": 0.3698426297515444,
      "grad_norm": 2.472224434364653,
      "learning_rate": 2.906028447075956e-06,
      "loss": 0.8226,
      "step": 4101
    },
    {
      "epoch": 0.36993281327501465,
      "grad_norm": 1.5311949583787738,
      "learning_rate": 2.905507582733445e-06,
      "loss": 0.8446,
      "step": 4102
    },
    {
      "epoch": 0.37002299679848494,
      "grad_norm": 1.656350834494106,
      "learning_rate": 2.904986641128033e-06,
      "loss": 0.8813,
      "step": 4103
    },
    {
      "epoch": 0.37011318032195517,
      "grad_norm": 1.3892296637927066,
      "learning_rate": 2.9044656223041716e-06,
      "loss": 0.7547,
      "step": 4104
    },
    {
      "epoch": 0.37020336384542546,
      "grad_norm": 1.3917113357882662,
      "learning_rate": 2.9039445263063157e-06,
      "loss": 0.7225,
      "step": 4105
    },
    {
      "epoch": 0.3702935473688957,
      "grad_norm": 2.2137766156139405,
      "learning_rate": 2.903423353178929e-06,
      "loss": 0.6683,
      "step": 4106
    },
    {
      "epoch": 0.370383730892366,
      "grad_norm": 1.4883747221662205,
      "learning_rate": 2.9029021029664802e-06,
      "loss": 0.7647,
      "step": 4107
    },
    {
      "epoch": 0.3704739144158362,
      "grad_norm": 1.520053018849917,
      "learning_rate": 2.9023807757134455e-06,
      "loss": 0.8207,
      "step": 4108
    },
    {
      "epoch": 0.3705640979393065,
      "grad_norm": 1.7089450330293035,
      "learning_rate": 2.901859371464307e-06,
      "loss": 0.8294,
      "step": 4109
    },
    {
      "epoch": 0.37065428146277674,
      "grad_norm": 2.7068366879429746,
      "learning_rate": 2.9013378902635535e-06,
      "loss": 0.8077,
      "step": 4110
    },
    {
      "epoch": 0.370744464986247,
      "grad_norm": 2.4532744732651333,
      "learning_rate": 2.9008163321556823e-06,
      "loss": 0.738,
      "step": 4111
    },
    {
      "epoch": 0.37083464850971726,
      "grad_norm": 1.689023636927516,
      "learning_rate": 2.900294697185194e-06,
      "loss": 0.7783,
      "step": 4112
    },
    {
      "epoch": 0.37092483203318755,
      "grad_norm": 1.5166680591003956,
      "learning_rate": 2.899772985396599e-06,
      "loss": 0.7841,
      "step": 4113
    },
    {
      "epoch": 0.3710150155566578,
      "grad_norm": 1.6485064162180667,
      "learning_rate": 2.8992511968344104e-06,
      "loss": 0.7584,
      "step": 4114
    },
    {
      "epoch": 0.37110519908012807,
      "grad_norm": 1.6433964889466577,
      "learning_rate": 2.8987293315431523e-06,
      "loss": 0.712,
      "step": 4115
    },
    {
      "epoch": 0.3711953826035983,
      "grad_norm": 1.4399775011901108,
      "learning_rate": 2.898207389567351e-06,
      "loss": 0.6962,
      "step": 4116
    },
    {
      "epoch": 0.3712855661270686,
      "grad_norm": 2.2927634880044363,
      "learning_rate": 2.897685370951543e-06,
      "loss": 0.7578,
      "step": 4117
    },
    {
      "epoch": 0.3713757496505388,
      "grad_norm": 1.7927712059485035,
      "learning_rate": 2.89716327574027e-06,
      "loss": 0.6577,
      "step": 4118
    },
    {
      "epoch": 0.3714659331740091,
      "grad_norm": 1.6265611485485725,
      "learning_rate": 2.8966411039780787e-06,
      "loss": 0.8515,
      "step": 4119
    },
    {
      "epoch": 0.37155611669747934,
      "grad_norm": 1.7536791546107526,
      "learning_rate": 2.8961188557095248e-06,
      "loss": 0.7713,
      "step": 4120
    },
    {
      "epoch": 0.37164630022094963,
      "grad_norm": 1.8217995263198201,
      "learning_rate": 2.895596530979168e-06,
      "loss": 0.7123,
      "step": 4121
    },
    {
      "epoch": 0.3717364837444199,
      "grad_norm": 1.6961436541510768,
      "learning_rate": 2.895074129831578e-06,
      "loss": 0.7791,
      "step": 4122
    },
    {
      "epoch": 0.37182666726789015,
      "grad_norm": 1.7313967040113531,
      "learning_rate": 2.8945516523113275e-06,
      "loss": 0.6792,
      "step": 4123
    },
    {
      "epoch": 0.37191685079136044,
      "grad_norm": 0.6769461101470094,
      "learning_rate": 2.894029098462998e-06,
      "loss": 0.5517,
      "step": 4124
    },
    {
      "epoch": 0.3720070343148307,
      "grad_norm": 1.5455857886721345,
      "learning_rate": 2.8935064683311756e-06,
      "loss": 0.7928,
      "step": 4125
    },
    {
      "epoch": 0.37209721783830096,
      "grad_norm": 0.6776881133459085,
      "learning_rate": 2.8929837619604544e-06,
      "loss": 0.6562,
      "step": 4126
    },
    {
      "epoch": 0.3721874013617712,
      "grad_norm": 1.9586536740667817,
      "learning_rate": 2.8924609793954346e-06,
      "loss": 0.8142,
      "step": 4127
    },
    {
      "epoch": 0.3722775848852415,
      "grad_norm": 3.104158405745265,
      "learning_rate": 2.891938120680724e-06,
      "loss": 0.7223,
      "step": 4128
    },
    {
      "epoch": 0.3723677684087117,
      "grad_norm": 1.5067713013468162,
      "learning_rate": 2.8914151858609343e-06,
      "loss": 0.7574,
      "step": 4129
    },
    {
      "epoch": 0.372457951932182,
      "grad_norm": 1.8360003521488824,
      "learning_rate": 2.8908921749806858e-06,
      "loss": 0.7893,
      "step": 4130
    },
    {
      "epoch": 0.37254813545565224,
      "grad_norm": 3.809888852317952,
      "learning_rate": 2.890369088084605e-06,
      "loss": 0.7778,
      "step": 4131
    },
    {
      "epoch": 0.37263831897912253,
      "grad_norm": 1.6562371163496912,
      "learning_rate": 2.889845925217323e-06,
      "loss": 0.7907,
      "step": 4132
    },
    {
      "epoch": 0.37272850250259276,
      "grad_norm": 1.718620156672184,
      "learning_rate": 2.8893226864234813e-06,
      "loss": 0.7772,
      "step": 4133
    },
    {
      "epoch": 0.37281868602606305,
      "grad_norm": 1.490451941850792,
      "learning_rate": 2.8887993717477236e-06,
      "loss": 0.7777,
      "step": 4134
    },
    {
      "epoch": 0.3729088695495333,
      "grad_norm": 1.7773107504663175,
      "learning_rate": 2.8882759812347035e-06,
      "loss": 0.6613,
      "step": 4135
    },
    {
      "epoch": 0.3729990530730036,
      "grad_norm": 1.5633708053170248,
      "learning_rate": 2.887752514929078e-06,
      "loss": 0.7743,
      "step": 4136
    },
    {
      "epoch": 0.3730892365964738,
      "grad_norm": 1.7622248901578428,
      "learning_rate": 2.887228972875513e-06,
      "loss": 0.7246,
      "step": 4137
    },
    {
      "epoch": 0.3731794201199441,
      "grad_norm": 1.8167115313854985,
      "learning_rate": 2.88670535511868e-06,
      "loss": 0.7681,
      "step": 4138
    },
    {
      "epoch": 0.3732696036434143,
      "grad_norm": 1.5812731775085413,
      "learning_rate": 2.886181661703257e-06,
      "loss": 0.7133,
      "step": 4139
    },
    {
      "epoch": 0.3733597871668846,
      "grad_norm": 1.735465179656747,
      "learning_rate": 2.8856578926739285e-06,
      "loss": 0.7336,
      "step": 4140
    },
    {
      "epoch": 0.37344997069035485,
      "grad_norm": 1.672605479724688,
      "learning_rate": 2.8851340480753846e-06,
      "loss": 0.7236,
      "step": 4141
    },
    {
      "epoch": 0.37354015421382514,
      "grad_norm": 2.254052116130021,
      "learning_rate": 2.8846101279523232e-06,
      "loss": 0.7711,
      "step": 4142
    },
    {
      "epoch": 0.37363033773729537,
      "grad_norm": 1.641953357442954,
      "learning_rate": 2.8840861323494487e-06,
      "loss": 0.7468,
      "step": 4143
    },
    {
      "epoch": 0.37372052126076566,
      "grad_norm": 2.164651398113332,
      "learning_rate": 2.88356206131147e-06,
      "loss": 0.8307,
      "step": 4144
    },
    {
      "epoch": 0.37381070478423595,
      "grad_norm": 1.5819537862014952,
      "learning_rate": 2.883037914883104e-06,
      "loss": 0.746,
      "step": 4145
    },
    {
      "epoch": 0.3739008883077062,
      "grad_norm": 2.5144903811067696,
      "learning_rate": 2.882513693109075e-06,
      "loss": 0.7382,
      "step": 4146
    },
    {
      "epoch": 0.37399107183117647,
      "grad_norm": 1.3859164643868493,
      "learning_rate": 2.8819893960341106e-06,
      "loss": 0.7451,
      "step": 4147
    },
    {
      "epoch": 0.3740812553546467,
      "grad_norm": 1.8190585553064715,
      "learning_rate": 2.881465023702948e-06,
      "loss": 0.8207,
      "step": 4148
    },
    {
      "epoch": 0.374171438878117,
      "grad_norm": 1.6837726683814689,
      "learning_rate": 2.8809405761603294e-06,
      "loss": 0.8521,
      "step": 4149
    },
    {
      "epoch": 0.3742616224015872,
      "grad_norm": 1.9720778419885105,
      "learning_rate": 2.880416053451003e-06,
      "loss": 0.6934,
      "step": 4150
    },
    {
      "epoch": 0.3743518059250575,
      "grad_norm": 1.77623424063986,
      "learning_rate": 2.879891455619725e-06,
      "loss": 0.8562,
      "step": 4151
    },
    {
      "epoch": 0.37444198944852775,
      "grad_norm": 1.5378075920015808,
      "learning_rate": 2.879366782711256e-06,
      "loss": 0.7709,
      "step": 4152
    },
    {
      "epoch": 0.37453217297199803,
      "grad_norm": 1.5777280704894618,
      "learning_rate": 2.8788420347703643e-06,
      "loss": 0.7882,
      "step": 4153
    },
    {
      "epoch": 0.37462235649546827,
      "grad_norm": 0.6380562804878637,
      "learning_rate": 2.8783172118418244e-06,
      "loss": 0.5779,
      "step": 4154
    },
    {
      "epoch": 0.37471254001893856,
      "grad_norm": 1.44688725641386,
      "learning_rate": 2.877792313970417e-06,
      "loss": 0.7732,
      "step": 4155
    },
    {
      "epoch": 0.3748027235424088,
      "grad_norm": 1.6286633353842728,
      "learning_rate": 2.8772673412009293e-06,
      "loss": 0.8208,
      "step": 4156
    },
    {
      "epoch": 0.3748929070658791,
      "grad_norm": 1.6618659337640387,
      "learning_rate": 2.8767422935781545e-06,
      "loss": 0.8186,
      "step": 4157
    },
    {
      "epoch": 0.3749830905893493,
      "grad_norm": 1.5312690733675385,
      "learning_rate": 2.8762171711468935e-06,
      "loss": 0.8047,
      "step": 4158
    },
    {
      "epoch": 0.3750732741128196,
      "grad_norm": 1.4987730730458513,
      "learning_rate": 2.875691973951952e-06,
      "loss": 0.7708,
      "step": 4159
    },
    {
      "epoch": 0.37516345763628983,
      "grad_norm": 0.6468324674914897,
      "learning_rate": 2.8751667020381425e-06,
      "loss": 0.6479,
      "step": 4160
    },
    {
      "epoch": 0.3752536411597601,
      "grad_norm": 1.9726966778228936,
      "learning_rate": 2.8746413554502837e-06,
      "loss": 0.8336,
      "step": 4161
    },
    {
      "epoch": 0.37534382468323035,
      "grad_norm": 1.5771323377419355,
      "learning_rate": 2.8741159342332027e-06,
      "loss": 0.7052,
      "step": 4162
    },
    {
      "epoch": 0.37543400820670064,
      "grad_norm": 2.4578755054776975,
      "learning_rate": 2.87359043843173e-06,
      "loss": 0.8311,
      "step": 4163
    },
    {
      "epoch": 0.3755241917301709,
      "grad_norm": 1.8333487582280314,
      "learning_rate": 2.873064868090704e-06,
      "loss": 0.8093,
      "step": 4164
    },
    {
      "epoch": 0.37561437525364116,
      "grad_norm": 1.9692232683301396,
      "learning_rate": 2.8725392232549697e-06,
      "loss": 0.8643,
      "step": 4165
    },
    {
      "epoch": 0.3757045587771114,
      "grad_norm": 1.7436771610046058,
      "learning_rate": 2.872013503969378e-06,
      "loss": 0.8387,
      "step": 4166
    },
    {
      "epoch": 0.3757947423005817,
      "grad_norm": 1.4022587402419573,
      "learning_rate": 2.8714877102787853e-06,
      "loss": 0.7772,
      "step": 4167
    },
    {
      "epoch": 0.3758849258240519,
      "grad_norm": 1.9304901183615717,
      "learning_rate": 2.8709618422280564e-06,
      "loss": 0.7767,
      "step": 4168
    },
    {
      "epoch": 0.3759751093475222,
      "grad_norm": 2.0176653330771357,
      "learning_rate": 2.8704358998620605e-06,
      "loss": 0.7476,
      "step": 4169
    },
    {
      "epoch": 0.3760652928709925,
      "grad_norm": 1.8103548048868836,
      "learning_rate": 2.8699098832256735e-06,
      "loss": 0.8028,
      "step": 4170
    },
    {
      "epoch": 0.37615547639446273,
      "grad_norm": 1.9034847151330738,
      "learning_rate": 2.86938379236378e-06,
      "loss": 0.6898,
      "step": 4171
    },
    {
      "epoch": 0.376245659917933,
      "grad_norm": 2.0398934377625606,
      "learning_rate": 2.868857627321266e-06,
      "loss": 0.6998,
      "step": 4172
    },
    {
      "epoch": 0.37633584344140325,
      "grad_norm": 2.2585958949909823,
      "learning_rate": 2.8683313881430296e-06,
      "loss": 0.7725,
      "step": 4173
    },
    {
      "epoch": 0.37642602696487354,
      "grad_norm": 2.9516145587488163,
      "learning_rate": 2.8678050748739706e-06,
      "loss": 0.7666,
      "step": 4174
    },
    {
      "epoch": 0.37651621048834377,
      "grad_norm": 1.5221134790080224,
      "learning_rate": 2.8672786875589976e-06,
      "loss": 0.7418,
      "step": 4175
    },
    {
      "epoch": 0.37660639401181406,
      "grad_norm": 1.5287448086342428,
      "learning_rate": 2.866752226243025e-06,
      "loss": 0.7216,
      "step": 4176
    },
    {
      "epoch": 0.3766965775352843,
      "grad_norm": 1.6416680789431517,
      "learning_rate": 2.8662256909709733e-06,
      "loss": 0.8154,
      "step": 4177
    },
    {
      "epoch": 0.3767867610587546,
      "grad_norm": 1.8163080004300338,
      "learning_rate": 2.865699081787769e-06,
      "loss": 0.811,
      "step": 4178
    },
    {
      "epoch": 0.3768769445822248,
      "grad_norm": 2.061799710770366,
      "learning_rate": 2.8651723987383465e-06,
      "loss": 0.7494,
      "step": 4179
    },
    {
      "epoch": 0.3769671281056951,
      "grad_norm": 1.7793432369954663,
      "learning_rate": 2.8646456418676437e-06,
      "loss": 0.7477,
      "step": 4180
    },
    {
      "epoch": 0.37705731162916534,
      "grad_norm": 0.6567900116127285,
      "learning_rate": 2.8641188112206067e-06,
      "loss": 0.5963,
      "step": 4181
    },
    {
      "epoch": 0.3771474951526356,
      "grad_norm": 2.3858713996489502,
      "learning_rate": 2.863591906842189e-06,
      "loss": 0.8026,
      "step": 4182
    },
    {
      "epoch": 0.37723767867610586,
      "grad_norm": 1.5025593857287642,
      "learning_rate": 2.863064928777347e-06,
      "loss": 0.8051,
      "step": 4183
    },
    {
      "epoch": 0.37732786219957615,
      "grad_norm": 1.9638821806636462,
      "learning_rate": 2.862537877071047e-06,
      "loss": 0.7451,
      "step": 4184
    },
    {
      "epoch": 0.3774180457230464,
      "grad_norm": 0.6518708530853708,
      "learning_rate": 2.8620107517682597e-06,
      "loss": 0.5608,
      "step": 4185
    },
    {
      "epoch": 0.37750822924651667,
      "grad_norm": 1.3907835366393464,
      "learning_rate": 2.8614835529139618e-06,
      "loss": 0.7014,
      "step": 4186
    },
    {
      "epoch": 0.3775984127699869,
      "grad_norm": 1.7648804031355287,
      "learning_rate": 2.8609562805531367e-06,
      "loss": 0.7329,
      "step": 4187
    },
    {
      "epoch": 0.3776885962934572,
      "grad_norm": 1.6955119072854021,
      "learning_rate": 2.8604289347307746e-06,
      "loss": 0.8062,
      "step": 4188
    },
    {
      "epoch": 0.3777787798169274,
      "grad_norm": 1.8226494584200246,
      "learning_rate": 2.859901515491871e-06,
      "loss": 0.7656,
      "step": 4189
    },
    {
      "epoch": 0.3778689633403977,
      "grad_norm": 1.3965467319037965,
      "learning_rate": 2.8593740228814298e-06,
      "loss": 0.6573,
      "step": 4190
    },
    {
      "epoch": 0.37795914686386795,
      "grad_norm": 1.703485713172168,
      "learning_rate": 2.8588464569444574e-06,
      "loss": 0.7605,
      "step": 4191
    },
    {
      "epoch": 0.37804933038733823,
      "grad_norm": 1.6114233370307505,
      "learning_rate": 2.8583188177259697e-06,
      "loss": 0.7587,
      "step": 4192
    },
    {
      "epoch": 0.3781395139108085,
      "grad_norm": 2.1025333927147196,
      "learning_rate": 2.857791105270988e-06,
      "loss": 0.7554,
      "step": 4193
    },
    {
      "epoch": 0.37822969743427876,
      "grad_norm": 1.5492795500412524,
      "learning_rate": 2.857263319624539e-06,
      "loss": 0.7994,
      "step": 4194
    },
    {
      "epoch": 0.37831988095774904,
      "grad_norm": 5.507341859878708,
      "learning_rate": 2.856735460831657e-06,
      "loss": 0.7537,
      "step": 4195
    },
    {
      "epoch": 0.3784100644812193,
      "grad_norm": 1.5551153773100619,
      "learning_rate": 2.856207528937382e-06,
      "loss": 0.7825,
      "step": 4196
    },
    {
      "epoch": 0.37850024800468957,
      "grad_norm": 1.4087451838603111,
      "learning_rate": 2.855679523986759e-06,
      "loss": 0.7552,
      "step": 4197
    },
    {
      "epoch": 0.3785904315281598,
      "grad_norm": 7.603088915669098,
      "learning_rate": 2.8551514460248406e-06,
      "loss": 0.8184,
      "step": 4198
    },
    {
      "epoch": 0.3786806150516301,
      "grad_norm": 1.6187274621992584,
      "learning_rate": 2.8546232950966868e-06,
      "loss": 0.7738,
      "step": 4199
    },
    {
      "epoch": 0.3787707985751003,
      "grad_norm": 2.0617623599558375,
      "learning_rate": 2.85409507124736e-06,
      "loss": 0.718,
      "step": 4200
    },
    {
      "epoch": 0.3788609820985706,
      "grad_norm": 4.193141496290121,
      "learning_rate": 2.8535667745219324e-06,
      "loss": 0.7644,
      "step": 4201
    },
    {
      "epoch": 0.37895116562204084,
      "grad_norm": 1.4914576957705892,
      "learning_rate": 2.853038404965481e-06,
      "loss": 0.7557,
      "step": 4202
    },
    {
      "epoch": 0.37904134914551113,
      "grad_norm": 2.9098157440265133,
      "learning_rate": 2.8525099626230894e-06,
      "loss": 0.8393,
      "step": 4203
    },
    {
      "epoch": 0.37913153266898136,
      "grad_norm": 2.1425838000936124,
      "learning_rate": 2.8519814475398472e-06,
      "loss": 0.6971,
      "step": 4204
    },
    {
      "epoch": 0.37922171619245165,
      "grad_norm": 1.3440489769055524,
      "learning_rate": 2.8514528597608502e-06,
      "loss": 0.7455,
      "step": 4205
    },
    {
      "epoch": 0.3793118997159219,
      "grad_norm": 1.2590583649333313,
      "learning_rate": 2.8509241993312004e-06,
      "loss": 0.7969,
      "step": 4206
    },
    {
      "epoch": 0.3794020832393922,
      "grad_norm": 1.759543554115483,
      "learning_rate": 2.850395466296006e-06,
      "loss": 0.8217,
      "step": 4207
    },
    {
      "epoch": 0.3794922667628624,
      "grad_norm": 1.7501180472749782,
      "learning_rate": 2.849866660700381e-06,
      "loss": 0.7948,
      "step": 4208
    },
    {
      "epoch": 0.3795824502863327,
      "grad_norm": 1.8899568370149142,
      "learning_rate": 2.8493377825894464e-06,
      "loss": 0.8281,
      "step": 4209
    },
    {
      "epoch": 0.37967263380980293,
      "grad_norm": 2.3493981016687044,
      "learning_rate": 2.848808832008329e-06,
      "loss": 0.8298,
      "step": 4210
    },
    {
      "epoch": 0.3797628173332732,
      "grad_norm": 1.6367576430622839,
      "learning_rate": 2.848279809002162e-06,
      "loss": 0.7986,
      "step": 4211
    },
    {
      "epoch": 0.37985300085674345,
      "grad_norm": 1.85753572678639,
      "learning_rate": 2.8477507136160842e-06,
      "loss": 0.7894,
      "step": 4212
    },
    {
      "epoch": 0.37994318438021374,
      "grad_norm": 0.6397162015706787,
      "learning_rate": 2.847221545895241e-06,
      "loss": 0.589,
      "step": 4213
    },
    {
      "epoch": 0.38003336790368397,
      "grad_norm": 1.493184420719415,
      "learning_rate": 2.846692305884785e-06,
      "loss": 0.7654,
      "step": 4214
    },
    {
      "epoch": 0.38012355142715426,
      "grad_norm": 1.3838090915347645,
      "learning_rate": 2.8461629936298718e-06,
      "loss": 0.816,
      "step": 4215
    },
    {
      "epoch": 0.3802137349506245,
      "grad_norm": 4.282827935787694,
      "learning_rate": 2.845633609175666e-06,
      "loss": 0.8241,
      "step": 4216
    },
    {
      "epoch": 0.3803039184740948,
      "grad_norm": 1.7032959178590097,
      "learning_rate": 2.8451041525673383e-06,
      "loss": 0.8076,
      "step": 4217
    },
    {
      "epoch": 0.38039410199756507,
      "grad_norm": 6.643560357675259,
      "learning_rate": 2.8445746238500647e-06,
      "loss": 0.8054,
      "step": 4218
    },
    {
      "epoch": 0.3804842855210353,
      "grad_norm": 1.897454843398442,
      "learning_rate": 2.844045023069027e-06,
      "loss": 0.749,
      "step": 4219
    },
    {
      "epoch": 0.3805744690445056,
      "grad_norm": 1.6589036951101979,
      "learning_rate": 2.8435153502694136e-06,
      "loss": 0.8145,
      "step": 4220
    },
    {
      "epoch": 0.3806646525679758,
      "grad_norm": 1.74485526656959,
      "learning_rate": 2.84298560549642e-06,
      "loss": 0.7061,
      "step": 4221
    },
    {
      "epoch": 0.3807548360914461,
      "grad_norm": 1.6780994093664607,
      "learning_rate": 2.8424557887952462e-06,
      "loss": 0.794,
      "step": 4222
    },
    {
      "epoch": 0.38084501961491635,
      "grad_norm": 1.672982606581927,
      "learning_rate": 2.841925900211099e-06,
      "loss": 0.6988,
      "step": 4223
    },
    {
      "epoch": 0.38093520313838664,
      "grad_norm": 1.575400011037972,
      "learning_rate": 2.841395939789192e-06,
      "loss": 0.8252,
      "step": 4224
    },
    {
      "epoch": 0.38102538666185687,
      "grad_norm": 1.4835076458267549,
      "learning_rate": 2.8408659075747435e-06,
      "loss": 0.7921,
      "step": 4225
    },
    {
      "epoch": 0.38111557018532716,
      "grad_norm": 1.7479777231870137,
      "learning_rate": 2.8403358036129796e-06,
      "loss": 0.7797,
      "step": 4226
    },
    {
      "epoch": 0.3812057537087974,
      "grad_norm": 1.591039409501817,
      "learning_rate": 2.839805627949132e-06,
      "loss": 0.8298,
      "step": 4227
    },
    {
      "epoch": 0.3812959372322677,
      "grad_norm": 3.0308343661368324,
      "learning_rate": 2.8392753806284367e-06,
      "loss": 0.8291,
      "step": 4228
    },
    {
      "epoch": 0.3813861207557379,
      "grad_norm": 4.620692747618641,
      "learning_rate": 2.838745061696139e-06,
      "loss": 0.7268,
      "step": 4229
    },
    {
      "epoch": 0.3814763042792082,
      "grad_norm": 1.6315275548068797,
      "learning_rate": 2.838214671197487e-06,
      "loss": 0.7132,
      "step": 4230
    },
    {
      "epoch": 0.38156648780267843,
      "grad_norm": 1.3790538157750005,
      "learning_rate": 2.8376842091777377e-06,
      "loss": 0.883,
      "step": 4231
    },
    {
      "epoch": 0.3816566713261487,
      "grad_norm": 1.5936201454068908,
      "learning_rate": 2.8371536756821524e-06,
      "loss": 0.6836,
      "step": 4232
    },
    {
      "epoch": 0.38174685484961896,
      "grad_norm": 1.5554795284463012,
      "learning_rate": 2.836623070756e-06,
      "loss": 0.7547,
      "step": 4233
    },
    {
      "epoch": 0.38183703837308924,
      "grad_norm": 1.6403930500098163,
      "learning_rate": 2.8360923944445542e-06,
      "loss": 0.8013,
      "step": 4234
    },
    {
      "epoch": 0.3819272218965595,
      "grad_norm": 1.5189888189480636,
      "learning_rate": 2.8355616467930947e-06,
      "loss": 0.7614,
      "step": 4235
    },
    {
      "epoch": 0.38201740542002977,
      "grad_norm": 1.967026698029724,
      "learning_rate": 2.8350308278469085e-06,
      "loss": 0.7533,
      "step": 4236
    },
    {
      "epoch": 0.3821075889435,
      "grad_norm": 4.324212133717838,
      "learning_rate": 2.8344999376512877e-06,
      "loss": 0.8089,
      "step": 4237
    },
    {
      "epoch": 0.3821977724669703,
      "grad_norm": 1.4350394876875459,
      "learning_rate": 2.8339689762515307e-06,
      "loss": 0.7671,
      "step": 4238
    },
    {
      "epoch": 0.3822879559904405,
      "grad_norm": 1.4551260511454978,
      "learning_rate": 2.8334379436929424e-06,
      "loss": 0.703,
      "step": 4239
    },
    {
      "epoch": 0.3823781395139108,
      "grad_norm": 2.245965200974957,
      "learning_rate": 2.832906840020833e-06,
      "loss": 0.7431,
      "step": 4240
    },
    {
      "epoch": 0.3824683230373811,
      "grad_norm": 1.573249082828426,
      "learning_rate": 2.83237566528052e-06,
      "loss": 0.7863,
      "step": 4241
    },
    {
      "epoch": 0.38255850656085133,
      "grad_norm": 1.5459455529923336,
      "learning_rate": 2.831844419517325e-06,
      "loss": 0.7576,
      "step": 4242
    },
    {
      "epoch": 0.3826486900843216,
      "grad_norm": 1.9347320596859114,
      "learning_rate": 2.8313131027765774e-06,
      "loss": 0.809,
      "step": 4243
    },
    {
      "epoch": 0.38273887360779185,
      "grad_norm": 1.5819219104046465,
      "learning_rate": 2.8307817151036124e-06,
      "loss": 0.7488,
      "step": 4244
    },
    {
      "epoch": 0.38282905713126214,
      "grad_norm": 3.002621459028877,
      "learning_rate": 2.8302502565437704e-06,
      "loss": 0.7501,
      "step": 4245
    },
    {
      "epoch": 0.3829192406547324,
      "grad_norm": 1.7483499104477893,
      "learning_rate": 2.829718727142398e-06,
      "loss": 0.7034,
      "step": 4246
    },
    {
      "epoch": 0.38300942417820266,
      "grad_norm": 1.4934032340946943,
      "learning_rate": 2.829187126944849e-06,
      "loss": 0.7873,
      "step": 4247
    },
    {
      "epoch": 0.3830996077016729,
      "grad_norm": 2.007081726318544,
      "learning_rate": 2.8286554559964826e-06,
      "loss": 0.7211,
      "step": 4248
    },
    {
      "epoch": 0.3831897912251432,
      "grad_norm": 1.733197061366027,
      "learning_rate": 2.8281237143426637e-06,
      "loss": 0.7938,
      "step": 4249
    },
    {
      "epoch": 0.3832799747486134,
      "grad_norm": 1.5660842983216015,
      "learning_rate": 2.8275919020287626e-06,
      "loss": 0.6784,
      "step": 4250
    },
    {
      "epoch": 0.3833701582720837,
      "grad_norm": 1.719704241718789,
      "learning_rate": 2.827060019100158e-06,
      "loss": 0.8025,
      "step": 4251
    },
    {
      "epoch": 0.38346034179555394,
      "grad_norm": 1.478794731695373,
      "learning_rate": 2.8265280656022315e-06,
      "loss": 0.7678,
      "step": 4252
    },
    {
      "epoch": 0.3835505253190242,
      "grad_norm": 1.6896827144847821,
      "learning_rate": 2.825996041580373e-06,
      "loss": 0.6642,
      "step": 4253
    },
    {
      "epoch": 0.38364070884249446,
      "grad_norm": 3.0032239121612543,
      "learning_rate": 2.825463947079978e-06,
      "loss": 0.7966,
      "step": 4254
    },
    {
      "epoch": 0.38373089236596475,
      "grad_norm": 1.546001534509329,
      "learning_rate": 2.8249317821464483e-06,
      "loss": 0.7751,
      "step": 4255
    },
    {
      "epoch": 0.383821075889435,
      "grad_norm": 1.5290853809646023,
      "learning_rate": 2.824399546825189e-06,
      "loss": 0.7075,
      "step": 4256
    },
    {
      "epoch": 0.38391125941290527,
      "grad_norm": 2.243622111551622,
      "learning_rate": 2.823867241161616e-06,
      "loss": 0.7326,
      "step": 4257
    },
    {
      "epoch": 0.3840014429363755,
      "grad_norm": 1.7607810491672977,
      "learning_rate": 2.8233348652011456e-06,
      "loss": 0.8256,
      "step": 4258
    },
    {
      "epoch": 0.3840916264598458,
      "grad_norm": 3.782437350907163,
      "learning_rate": 2.8228024189892057e-06,
      "loss": 0.6851,
      "step": 4259
    },
    {
      "epoch": 0.384181809983316,
      "grad_norm": 1.2947207563669736,
      "learning_rate": 2.822269902571226e-06,
      "loss": 0.7768,
      "step": 4260
    },
    {
      "epoch": 0.3842719935067863,
      "grad_norm": 1.5574782737086064,
      "learning_rate": 2.8217373159926446e-06,
      "loss": 0.7539,
      "step": 4261
    },
    {
      "epoch": 0.38436217703025655,
      "grad_norm": 1.4482435045088415,
      "learning_rate": 2.8212046592989046e-06,
      "loss": 0.7119,
      "step": 4262
    },
    {
      "epoch": 0.38445236055372684,
      "grad_norm": 1.5873688500965093,
      "learning_rate": 2.820671932535455e-06,
      "loss": 0.7856,
      "step": 4263
    },
    {
      "epoch": 0.3845425440771971,
      "grad_norm": 2.660948782340842,
      "learning_rate": 2.8201391357477506e-06,
      "loss": 0.7594,
      "step": 4264
    },
    {
      "epoch": 0.38463272760066736,
      "grad_norm": 1.3720602775016706,
      "learning_rate": 2.8196062689812525e-06,
      "loss": 0.7514,
      "step": 4265
    },
    {
      "epoch": 0.38472291112413765,
      "grad_norm": 3.9392116398857064,
      "learning_rate": 2.819073332281429e-06,
      "loss": 0.7186,
      "step": 4266
    },
    {
      "epoch": 0.3848130946476079,
      "grad_norm": 0.619319733426057,
      "learning_rate": 2.8185403256937524e-06,
      "loss": 0.55,
      "step": 4267
    },
    {
      "epoch": 0.38490327817107817,
      "grad_norm": 1.9864082432795924,
      "learning_rate": 2.8180072492637016e-06,
      "loss": 0.7379,
      "step": 4268
    },
    {
      "epoch": 0.3849934616945484,
      "grad_norm": 2.4250998407634476,
      "learning_rate": 2.817474103036762e-06,
      "loss": 0.7687,
      "step": 4269
    },
    {
      "epoch": 0.3850836452180187,
      "grad_norm": 1.938604347763924,
      "learning_rate": 2.816940887058425e-06,
      "loss": 0.7985,
      "step": 4270
    },
    {
      "epoch": 0.3851738287414889,
      "grad_norm": 1.6421026206107892,
      "learning_rate": 2.816407601374186e-06,
      "loss": 0.809,
      "step": 4271
    },
    {
      "epoch": 0.3852640122649592,
      "grad_norm": 1.7661920497752936,
      "learning_rate": 2.815874246029549e-06,
      "loss": 0.7079,
      "step": 4272
    },
    {
      "epoch": 0.38535419578842944,
      "grad_norm": 1.5013947360410804,
      "learning_rate": 2.815340821070023e-06,
      "loss": 0.7236,
      "step": 4273
    },
    {
      "epoch": 0.38544437931189973,
      "grad_norm": 1.7118174759056644,
      "learning_rate": 2.814807326541122e-06,
      "loss": 0.7488,
      "step": 4274
    },
    {
      "epoch": 0.38553456283536996,
      "grad_norm": 1.3853654732483731,
      "learning_rate": 2.8142737624883676e-06,
      "loss": 0.7019,
      "step": 4275
    },
    {
      "epoch": 0.38562474635884025,
      "grad_norm": 1.6035284861932841,
      "learning_rate": 2.8137401289572854e-06,
      "loss": 0.7715,
      "step": 4276
    },
    {
      "epoch": 0.3857149298823105,
      "grad_norm": 1.5037065645890841,
      "learning_rate": 2.8132064259934086e-06,
      "loss": 0.8142,
      "step": 4277
    },
    {
      "epoch": 0.3858051134057808,
      "grad_norm": 1.471343733038804,
      "learning_rate": 2.812672653642276e-06,
      "loss": 0.7664,
      "step": 4278
    },
    {
      "epoch": 0.385895296929251,
      "grad_norm": 1.561511604498044,
      "learning_rate": 2.812138811949431e-06,
      "loss": 0.7825,
      "step": 4279
    },
    {
      "epoch": 0.3859854804527213,
      "grad_norm": 1.4979499477178535,
      "learning_rate": 2.8116049009604247e-06,
      "loss": 0.833,
      "step": 4280
    },
    {
      "epoch": 0.38607566397619153,
      "grad_norm": 1.6394722658157497,
      "learning_rate": 2.8110709207208132e-06,
      "loss": 0.659,
      "step": 4281
    },
    {
      "epoch": 0.3861658474996618,
      "grad_norm": 3.6378996639329886,
      "learning_rate": 2.810536871276158e-06,
      "loss": 0.8099,
      "step": 4282
    },
    {
      "epoch": 0.38625603102313205,
      "grad_norm": 0.7330966130295101,
      "learning_rate": 2.8100027526720283e-06,
      "loss": 0.6308,
      "step": 4283
    },
    {
      "epoch": 0.38634621454660234,
      "grad_norm": 1.6508596261767916,
      "learning_rate": 2.8094685649539974e-06,
      "loss": 0.6985,
      "step": 4284
    },
    {
      "epoch": 0.3864363980700726,
      "grad_norm": 0.7209194861361948,
      "learning_rate": 2.8089343081676455e-06,
      "loss": 0.6023,
      "step": 4285
    },
    {
      "epoch": 0.38652658159354286,
      "grad_norm": 1.9656842900538918,
      "learning_rate": 2.8083999823585577e-06,
      "loss": 0.7577,
      "step": 4286
    },
    {
      "epoch": 0.3866167651170131,
      "grad_norm": 13.109445419395191,
      "learning_rate": 2.8078655875723254e-06,
      "loss": 0.7954,
      "step": 4287
    },
    {
      "epoch": 0.3867069486404834,
      "grad_norm": 2.2367492012831742,
      "learning_rate": 2.807331123854547e-06,
      "loss": 0.8585,
      "step": 4288
    },
    {
      "epoch": 0.38679713216395367,
      "grad_norm": 1.9224122699815867,
      "learning_rate": 2.806796591250826e-06,
      "loss": 0.6444,
      "step": 4289
    },
    {
      "epoch": 0.3868873156874239,
      "grad_norm": 2.1368424904540833,
      "learning_rate": 2.8062619898067707e-06,
      "loss": 0.8406,
      "step": 4290
    },
    {
      "epoch": 0.3869774992108942,
      "grad_norm": 1.6225264135952706,
      "learning_rate": 2.8057273195679963e-06,
      "loss": 0.7883,
      "step": 4291
    },
    {
      "epoch": 0.3870676827343644,
      "grad_norm": 1.7844432517124644,
      "learning_rate": 2.8051925805801253e-06,
      "loss": 0.7859,
      "step": 4292
    },
    {
      "epoch": 0.3871578662578347,
      "grad_norm": 1.452383775246741,
      "learning_rate": 2.804657772888783e-06,
      "loss": 0.7984,
      "step": 4293
    },
    {
      "epoch": 0.38724804978130495,
      "grad_norm": 1.9273558827711448,
      "learning_rate": 2.804122896539602e-06,
      "loss": 0.7711,
      "step": 4294
    },
    {
      "epoch": 0.38733823330477524,
      "grad_norm": 1.5342993443749542,
      "learning_rate": 2.8035879515782225e-06,
      "loss": 0.7242,
      "step": 4295
    },
    {
      "epoch": 0.38742841682824547,
      "grad_norm": 2.169464090507709,
      "learning_rate": 2.803052938050288e-06,
      "loss": 0.8207,
      "step": 4296
    },
    {
      "epoch": 0.38751860035171576,
      "grad_norm": 14.775805642097723,
      "learning_rate": 2.802517856001449e-06,
      "loss": 0.8177,
      "step": 4297
    },
    {
      "epoch": 0.387608783875186,
      "grad_norm": 6.5681938538466555,
      "learning_rate": 2.801982705477361e-06,
      "loss": 0.7379,
      "step": 4298
    },
    {
      "epoch": 0.3876989673986563,
      "grad_norm": 1.3765833581389877,
      "learning_rate": 2.8014474865236867e-06,
      "loss": 0.7568,
      "step": 4299
    },
    {
      "epoch": 0.3877891509221265,
      "grad_norm": 1.610377767799591,
      "learning_rate": 2.800912199186094e-06,
      "loss": 0.7593,
      "step": 4300
    },
    {
      "epoch": 0.3878793344455968,
      "grad_norm": 1.4658356017065723,
      "learning_rate": 2.800376843510256e-06,
      "loss": 0.7811,
      "step": 4301
    },
    {
      "epoch": 0.38796951796906703,
      "grad_norm": 1.5726190584411763,
      "learning_rate": 2.799841419541852e-06,
      "loss": 0.791,
      "step": 4302
    },
    {
      "epoch": 0.3880597014925373,
      "grad_norm": 0.6439585597880062,
      "learning_rate": 2.799305927326568e-06,
      "loss": 0.5852,
      "step": 4303
    },
    {
      "epoch": 0.38814988501600756,
      "grad_norm": 0.7254181330266334,
      "learning_rate": 2.7987703669100955e-06,
      "loss": 0.5946,
      "step": 4304
    },
    {
      "epoch": 0.38824006853947785,
      "grad_norm": 1.5102464865037588,
      "learning_rate": 2.79823473833813e-06,
      "loss": 0.7801,
      "step": 4305
    },
    {
      "epoch": 0.3883302520629481,
      "grad_norm": 1.6231197336234116,
      "learning_rate": 2.797699041656376e-06,
      "loss": 0.7193,
      "step": 4306
    },
    {
      "epoch": 0.38842043558641837,
      "grad_norm": 2.1687548062933617,
      "learning_rate": 2.7971632769105412e-06,
      "loss": 0.7856,
      "step": 4307
    },
    {
      "epoch": 0.3885106191098886,
      "grad_norm": 2.687320703247618,
      "learning_rate": 2.79662744414634e-06,
      "loss": 0.7376,
      "step": 4308
    },
    {
      "epoch": 0.3886008026333589,
      "grad_norm": 1.5794930146857764,
      "learning_rate": 2.7960915434094923e-06,
      "loss": 0.7153,
      "step": 4309
    },
    {
      "epoch": 0.3886909861568291,
      "grad_norm": 2.174786598873754,
      "learning_rate": 2.7955555747457256e-06,
      "loss": 0.7444,
      "step": 4310
    },
    {
      "epoch": 0.3887811696802994,
      "grad_norm": 1.544495341646203,
      "learning_rate": 2.79501953820077e-06,
      "loss": 0.7675,
      "step": 4311
    },
    {
      "epoch": 0.3888713532037697,
      "grad_norm": 1.6975661385956708,
      "learning_rate": 2.7944834338203637e-06,
      "loss": 0.7839,
      "step": 4312
    },
    {
      "epoch": 0.38896153672723993,
      "grad_norm": 8.758802509414503,
      "learning_rate": 2.79394726165025e-06,
      "loss": 0.7329,
      "step": 4313
    },
    {
      "epoch": 0.3890517202507102,
      "grad_norm": 7.8116254393304425,
      "learning_rate": 2.793411021736178e-06,
      "loss": 0.752,
      "step": 4314
    },
    {
      "epoch": 0.38914190377418045,
      "grad_norm": 1.5958390102169302,
      "learning_rate": 2.7928747141239027e-06,
      "loss": 0.8616,
      "step": 4315
    },
    {
      "epoch": 0.38923208729765074,
      "grad_norm": 0.6488620103856075,
      "learning_rate": 2.7923383388591856e-06,
      "loss": 0.5874,
      "step": 4316
    },
    {
      "epoch": 0.389322270821121,
      "grad_norm": 1.9384337605476132,
      "learning_rate": 2.7918018959877923e-06,
      "loss": 0.7365,
      "step": 4317
    },
    {
      "epoch": 0.38941245434459126,
      "grad_norm": 1.5875545582444819,
      "learning_rate": 2.791265385555495e-06,
      "loss": 0.795,
      "step": 4318
    },
    {
      "epoch": 0.3895026378680615,
      "grad_norm": 1.8280435527688343,
      "learning_rate": 2.790728807608072e-06,
      "loss": 0.8316,
      "step": 4319
    },
    {
      "epoch": 0.3895928213915318,
      "grad_norm": 2.0382008102060527,
      "learning_rate": 2.790192162191307e-06,
      "loss": 0.758,
      "step": 4320
    },
    {
      "epoch": 0.389683004915002,
      "grad_norm": 2.0729562829096255,
      "learning_rate": 2.78965544935099e-06,
      "loss": 0.7979,
      "step": 4321
    },
    {
      "epoch": 0.3897731884384723,
      "grad_norm": 2.062397694217822,
      "learning_rate": 2.789118669132916e-06,
      "loss": 0.7495,
      "step": 4322
    },
    {
      "epoch": 0.38986337196194254,
      "grad_norm": 2.190242791674182,
      "learning_rate": 2.7885818215828856e-06,
      "loss": 0.8401,
      "step": 4323
    },
    {
      "epoch": 0.38995355548541283,
      "grad_norm": 2.2635154637170216,
      "learning_rate": 2.7880449067467064e-06,
      "loss": 0.7263,
      "step": 4324
    },
    {
      "epoch": 0.39004373900888306,
      "grad_norm": 1.5889570683224654,
      "learning_rate": 2.78750792467019e-06,
      "loss": 0.725,
      "step": 4325
    },
    {
      "epoch": 0.39013392253235335,
      "grad_norm": 2.2980545025662766,
      "learning_rate": 2.786970875399156e-06,
      "loss": 0.6341,
      "step": 4326
    },
    {
      "epoch": 0.3902241060558236,
      "grad_norm": 1.3357907487899883,
      "learning_rate": 2.7864337589794267e-06,
      "loss": 0.699,
      "step": 4327
    },
    {
      "epoch": 0.39031428957929387,
      "grad_norm": 1.8454317085076275,
      "learning_rate": 2.7858965754568335e-06,
      "loss": 0.7816,
      "step": 4328
    },
    {
      "epoch": 0.3904044731027641,
      "grad_norm": 1.7846138628939254,
      "learning_rate": 2.785359324877211e-06,
      "loss": 0.706,
      "step": 4329
    },
    {
      "epoch": 0.3904946566262344,
      "grad_norm": 1.533492178912845,
      "learning_rate": 2.7848220072864e-06,
      "loss": 0.7955,
      "step": 4330
    },
    {
      "epoch": 0.3905848401497046,
      "grad_norm": 1.7342680477123247,
      "learning_rate": 2.784284622730248e-06,
      "loss": 0.7557,
      "step": 4331
    },
    {
      "epoch": 0.3906750236731749,
      "grad_norm": 1.8771750867676744,
      "learning_rate": 2.7837471712546073e-06,
      "loss": 0.7833,
      "step": 4332
    },
    {
      "epoch": 0.39076520719664515,
      "grad_norm": 1.383362870686178,
      "learning_rate": 2.783209652905337e-06,
      "loss": 0.7378,
      "step": 4333
    },
    {
      "epoch": 0.39085539072011544,
      "grad_norm": 1.6241877066180945,
      "learning_rate": 2.7826720677283e-06,
      "loss": 0.7261,
      "step": 4334
    },
    {
      "epoch": 0.39094557424358567,
      "grad_norm": 1.6445400935499301,
      "learning_rate": 2.782134415769367e-06,
      "loss": 0.7347,
      "step": 4335
    },
    {
      "epoch": 0.39103575776705596,
      "grad_norm": 1.5236133180546472,
      "learning_rate": 2.7815966970744126e-06,
      "loss": 0.8125,
      "step": 4336
    },
    {
      "epoch": 0.39112594129052625,
      "grad_norm": 1.8501171796649019,
      "learning_rate": 2.7810589116893184e-06,
      "loss": 0.8313,
      "step": 4337
    },
    {
      "epoch": 0.3912161248139965,
      "grad_norm": 4.395532654999505,
      "learning_rate": 2.780521059659972e-06,
      "loss": 0.7703,
      "step": 4338
    },
    {
      "epoch": 0.39130630833746677,
      "grad_norm": 3.1690718568243006,
      "learning_rate": 2.7799831410322637e-06,
      "loss": 0.7542,
      "step": 4339
    },
    {
      "epoch": 0.391396491860937,
      "grad_norm": 3.7343092397861284,
      "learning_rate": 2.779445155852094e-06,
      "loss": 0.7192,
      "step": 4340
    },
    {
      "epoch": 0.3914866753844073,
      "grad_norm": 0.6516770708086107,
      "learning_rate": 2.7789071041653655e-06,
      "loss": 0.6268,
      "step": 4341
    },
    {
      "epoch": 0.3915768589078775,
      "grad_norm": 1.5389535158559708,
      "learning_rate": 2.7783689860179875e-06,
      "loss": 0.8203,
      "step": 4342
    },
    {
      "epoch": 0.3916670424313478,
      "grad_norm": 4.816209824664548,
      "learning_rate": 2.7778308014558767e-06,
      "loss": 0.7036,
      "step": 4343
    },
    {
      "epoch": 0.39175722595481804,
      "grad_norm": 1.6820076465500917,
      "learning_rate": 2.7772925505249524e-06,
      "loss": 0.8423,
      "step": 4344
    },
    {
      "epoch": 0.39184740947828833,
      "grad_norm": 1.8385593604370205,
      "learning_rate": 2.7767542332711417e-06,
      "loss": 0.7357,
      "step": 4345
    },
    {
      "epoch": 0.39193759300175857,
      "grad_norm": 1.8822779866242005,
      "learning_rate": 2.776215849740377e-06,
      "loss": 0.7408,
      "step": 4346
    },
    {
      "epoch": 0.39202777652522885,
      "grad_norm": 1.461844096678707,
      "learning_rate": 2.775677399978596e-06,
      "loss": 0.7826,
      "step": 4347
    },
    {
      "epoch": 0.3921179600486991,
      "grad_norm": 0.5675844500636229,
      "learning_rate": 2.775138884031742e-06,
      "loss": 0.5729,
      "step": 4348
    },
    {
      "epoch": 0.3922081435721694,
      "grad_norm": 1.7342787707327383,
      "learning_rate": 2.774600301945764e-06,
      "loss": 0.727,
      "step": 4349
    },
    {
      "epoch": 0.3922983270956396,
      "grad_norm": 10.538369327765036,
      "learning_rate": 2.774061653766618e-06,
      "loss": 0.7957,
      "step": 4350
    },
    {
      "epoch": 0.3923885106191099,
      "grad_norm": 1.9393995878853507,
      "learning_rate": 2.773522939540263e-06,
      "loss": 0.7606,
      "step": 4351
    },
    {
      "epoch": 0.39247869414258013,
      "grad_norm": 1.8830910630939484,
      "learning_rate": 2.7729841593126663e-06,
      "loss": 0.6913,
      "step": 4352
    },
    {
      "epoch": 0.3925688776660504,
      "grad_norm": 1.4884309468114076,
      "learning_rate": 2.7724453131297988e-06,
      "loss": 0.8303,
      "step": 4353
    },
    {
      "epoch": 0.39265906118952065,
      "grad_norm": 1.5625461571551595,
      "learning_rate": 2.771906401037637e-06,
      "loss": 0.7926,
      "step": 4354
    },
    {
      "epoch": 0.39274924471299094,
      "grad_norm": 20.571216718400265,
      "learning_rate": 2.7713674230821664e-06,
      "loss": 0.7226,
      "step": 4355
    },
    {
      "epoch": 0.3928394282364612,
      "grad_norm": 1.3895571176704964,
      "learning_rate": 2.7708283793093724e-06,
      "loss": 0.7178,
      "step": 4356
    },
    {
      "epoch": 0.39292961175993146,
      "grad_norm": 1.5923165999942692,
      "learning_rate": 2.7702892697652514e-06,
      "loss": 0.7391,
      "step": 4357
    },
    {
      "epoch": 0.3930197952834017,
      "grad_norm": 1.7052069022192646,
      "learning_rate": 2.7697500944958024e-06,
      "loss": 0.7474,
      "step": 4358
    },
    {
      "epoch": 0.393109978806872,
      "grad_norm": 2.172641756022425,
      "learning_rate": 2.7692108535470312e-06,
      "loss": 0.8661,
      "step": 4359
    },
    {
      "epoch": 0.3932001623303423,
      "grad_norm": 2.2783904246739834,
      "learning_rate": 2.768671546964948e-06,
      "loss": 0.8008,
      "step": 4360
    },
    {
      "epoch": 0.3932903458538125,
      "grad_norm": 1.6706470008254848,
      "learning_rate": 2.7681321747955713e-06,
      "loss": 0.7936,
      "step": 4361
    },
    {
      "epoch": 0.3933805293772828,
      "grad_norm": 2.458552582441275,
      "learning_rate": 2.767592737084921e-06,
      "loss": 0.7798,
      "step": 4362
    },
    {
      "epoch": 0.39347071290075303,
      "grad_norm": 2.324270039481356,
      "learning_rate": 2.767053233879026e-06,
      "loss": 0.8619,
      "step": 4363
    },
    {
      "epoch": 0.3935608964242233,
      "grad_norm": 1.7335984837251237,
      "learning_rate": 2.76651366522392e-06,
      "loss": 0.8438,
      "step": 4364
    },
    {
      "epoch": 0.39365107994769355,
      "grad_norm": 1.5991977259787211,
      "learning_rate": 2.7659740311656413e-06,
      "loss": 0.7375,
      "step": 4365
    },
    {
      "epoch": 0.39374126347116384,
      "grad_norm": 2.55796552327345,
      "learning_rate": 2.7654343317502352e-06,
      "loss": 0.7879,
      "step": 4366
    },
    {
      "epoch": 0.39383144699463407,
      "grad_norm": 5.477815623450336,
      "learning_rate": 2.7648945670237502e-06,
      "loss": 0.7465,
      "step": 4367
    },
    {
      "epoch": 0.39392163051810436,
      "grad_norm": 2.0984863684581647,
      "learning_rate": 2.7643547370322446e-06,
      "loss": 0.8218,
      "step": 4368
    },
    {
      "epoch": 0.3940118140415746,
      "grad_norm": 1.8230039157877689,
      "learning_rate": 2.7638148418217775e-06,
      "loss": 0.8057,
      "step": 4369
    },
    {
      "epoch": 0.3941019975650449,
      "grad_norm": 1.6703094568037495,
      "learning_rate": 2.7632748814384163e-06,
      "loss": 0.8081,
      "step": 4370
    },
    {
      "epoch": 0.3941921810885151,
      "grad_norm": 1.2934024233323307,
      "learning_rate": 2.7627348559282335e-06,
      "loss": 0.7928,
      "step": 4371
    },
    {
      "epoch": 0.3942823646119854,
      "grad_norm": 1.8088940233151993,
      "learning_rate": 2.7621947653373075e-06,
      "loss": 0.7049,
      "step": 4372
    },
    {
      "epoch": 0.39437254813545564,
      "grad_norm": 0.6475567178665946,
      "learning_rate": 2.7616546097117213e-06,
      "loss": 0.5649,
      "step": 4373
    },
    {
      "epoch": 0.3944627316589259,
      "grad_norm": 1.5942120349820752,
      "learning_rate": 2.761114389097564e-06,
      "loss": 0.7609,
      "step": 4374
    },
    {
      "epoch": 0.39455291518239616,
      "grad_norm": 1.8949619708107843,
      "learning_rate": 2.7605741035409305e-06,
      "loss": 0.798,
      "step": 4375
    },
    {
      "epoch": 0.39464309870586645,
      "grad_norm": 3.9138502722060116,
      "learning_rate": 2.76003375308792e-06,
      "loss": 0.7475,
      "step": 4376
    },
    {
      "epoch": 0.3947332822293367,
      "grad_norm": 1.8452527662359077,
      "learning_rate": 2.75949333778464e-06,
      "loss": 0.7325,
      "step": 4377
    },
    {
      "epoch": 0.39482346575280697,
      "grad_norm": 1.5128987756061913,
      "learning_rate": 2.7589528576772e-06,
      "loss": 0.6474,
      "step": 4378
    },
    {
      "epoch": 0.3949136492762772,
      "grad_norm": 1.9877031906195208,
      "learning_rate": 2.758412312811717e-06,
      "loss": 0.7551,
      "step": 4379
    },
    {
      "epoch": 0.3950038327997475,
      "grad_norm": 3.105811142190096,
      "learning_rate": 2.7578717032343146e-06,
      "loss": 0.6308,
      "step": 4380
    },
    {
      "epoch": 0.3950940163232177,
      "grad_norm": 1.7721345122758032,
      "learning_rate": 2.757331028991119e-06,
      "loss": 0.793,
      "step": 4381
    },
    {
      "epoch": 0.395184199846688,
      "grad_norm": 3.3423916918854473,
      "learning_rate": 2.7567902901282642e-06,
      "loss": 0.7846,
      "step": 4382
    },
    {
      "epoch": 0.3952743833701583,
      "grad_norm": 1.838833087882517,
      "learning_rate": 2.7562494866918892e-06,
      "loss": 0.7561,
      "step": 4383
    },
    {
      "epoch": 0.39536456689362853,
      "grad_norm": 1.4541163190780784,
      "learning_rate": 2.7557086187281378e-06,
      "loss": 0.7348,
      "step": 4384
    },
    {
      "epoch": 0.3954547504170988,
      "grad_norm": 3.8619812107745277,
      "learning_rate": 2.75516768628316e-06,
      "loss": 0.8638,
      "step": 4385
    },
    {
      "epoch": 0.39554493394056905,
      "grad_norm": 1.6053154122666105,
      "learning_rate": 2.7546266894031114e-06,
      "loss": 0.8326,
      "step": 4386
    },
    {
      "epoch": 0.39563511746403934,
      "grad_norm": 2.5416994457528377,
      "learning_rate": 2.7540856281341526e-06,
      "loss": 0.7113,
      "step": 4387
    },
    {
      "epoch": 0.3957253009875096,
      "grad_norm": 1.6969198877850136,
      "learning_rate": 2.7535445025224506e-06,
      "loss": 0.7303,
      "step": 4388
    },
    {
      "epoch": 0.39581548451097986,
      "grad_norm": 2.101685474300506,
      "learning_rate": 2.753003312614176e-06,
      "loss": 0.7658,
      "step": 4389
    },
    {
      "epoch": 0.3959056680344501,
      "grad_norm": 1.664145812335489,
      "learning_rate": 2.7524620584555065e-06,
      "loss": 0.7707,
      "step": 4390
    },
    {
      "epoch": 0.3959958515579204,
      "grad_norm": 3.963892930749123,
      "learning_rate": 2.7519207400926253e-06,
      "loss": 0.8034,
      "step": 4391
    },
    {
      "epoch": 0.3960860350813906,
      "grad_norm": 1.563569427724961,
      "learning_rate": 2.751379357571721e-06,
      "loss": 0.7465,
      "step": 4392
    },
    {
      "epoch": 0.3961762186048609,
      "grad_norm": 1.524206192535807,
      "learning_rate": 2.7508379109389865e-06,
      "loss": 0.8247,
      "step": 4393
    },
    {
      "epoch": 0.39626640212833114,
      "grad_norm": 1.793986179341098,
      "learning_rate": 2.750296400240622e-06,
      "loss": 0.7783,
      "step": 4394
    },
    {
      "epoch": 0.39635658565180143,
      "grad_norm": 1.3655122356865321,
      "learning_rate": 2.7497548255228305e-06,
      "loss": 0.7964,
      "step": 4395
    },
    {
      "epoch": 0.39644676917527166,
      "grad_norm": 2.0867554850256576,
      "learning_rate": 2.749213186831824e-06,
      "loss": 0.7141,
      "step": 4396
    },
    {
      "epoch": 0.39653695269874195,
      "grad_norm": 1.7613526363532723,
      "learning_rate": 2.7486714842138173e-06,
      "loss": 0.814,
      "step": 4397
    },
    {
      "epoch": 0.3966271362222122,
      "grad_norm": 1.790205145007733,
      "learning_rate": 2.748129717715031e-06,
      "loss": 0.7152,
      "step": 4398
    },
    {
      "epoch": 0.3967173197456825,
      "grad_norm": 1.5488139629317852,
      "learning_rate": 2.747587887381692e-06,
      "loss": 0.7837,
      "step": 4399
    },
    {
      "epoch": 0.3968075032691527,
      "grad_norm": 2.3406965583486854,
      "learning_rate": 2.7470459932600328e-06,
      "loss": 0.7141,
      "step": 4400
    },
    {
      "epoch": 0.396897686792623,
      "grad_norm": 1.979309585497352,
      "learning_rate": 2.7465040353962897e-06,
      "loss": 0.7645,
      "step": 4401
    },
    {
      "epoch": 0.3969878703160932,
      "grad_norm": 1.5930315530098227,
      "learning_rate": 2.745962013836706e-06,
      "loss": 0.7588,
      "step": 4402
    },
    {
      "epoch": 0.3970780538395635,
      "grad_norm": 2.084190637307533,
      "learning_rate": 2.74541992862753e-06,
      "loss": 0.7413,
      "step": 4403
    },
    {
      "epoch": 0.39716823736303375,
      "grad_norm": 2.550738814791067,
      "learning_rate": 2.744877779815016e-06,
      "loss": 0.763,
      "step": 4404
    },
    {
      "epoch": 0.39725842088650404,
      "grad_norm": 1.5249726996557789,
      "learning_rate": 2.7443355674454234e-06,
      "loss": 0.7616,
      "step": 4405
    },
    {
      "epoch": 0.39734860440997427,
      "grad_norm": 2.184309239959838,
      "learning_rate": 2.743793291565015e-06,
      "loss": 0.8067,
      "step": 4406
    },
    {
      "epoch": 0.39743878793344456,
      "grad_norm": 1.9246621120219727,
      "learning_rate": 2.7432509522200617e-06,
      "loss": 0.6891,
      "step": 4407
    },
    {
      "epoch": 0.39752897145691485,
      "grad_norm": 1.7727150811040684,
      "learning_rate": 2.7427085494568383e-06,
      "loss": 0.7075,
      "step": 4408
    },
    {
      "epoch": 0.3976191549803851,
      "grad_norm": 0.6624994844758529,
      "learning_rate": 2.742166083321628e-06,
      "loss": 0.6066,
      "step": 4409
    },
    {
      "epoch": 0.39770933850385537,
      "grad_norm": 1.6754915569460387,
      "learning_rate": 2.7416235538607137e-06,
      "loss": 0.8031,
      "step": 4410
    },
    {
      "epoch": 0.3977995220273256,
      "grad_norm": 1.4149030853080393,
      "learning_rate": 2.7410809611203894e-06,
      "loss": 0.741,
      "step": 4411
    },
    {
      "epoch": 0.3978897055507959,
      "grad_norm": 1.6159177946483785,
      "learning_rate": 2.7405383051469507e-06,
      "loss": 0.723,
      "step": 4412
    },
    {
      "epoch": 0.3979798890742661,
      "grad_norm": 1.6404586707678546,
      "learning_rate": 2.7399955859867e-06,
      "loss": 0.5772,
      "step": 4413
    },
    {
      "epoch": 0.3980700725977364,
      "grad_norm": 0.6102157807951163,
      "learning_rate": 2.7394528036859465e-06,
      "loss": 0.5503,
      "step": 4414
    },
    {
      "epoch": 0.39816025612120665,
      "grad_norm": 1.5756825345712275,
      "learning_rate": 2.738909958291002e-06,
      "loss": 0.7629,
      "step": 4415
    },
    {
      "epoch": 0.39825043964467693,
      "grad_norm": 1.672624081512101,
      "learning_rate": 2.7383670498481863e-06,
      "loss": 0.6763,
      "step": 4416
    },
    {
      "epoch": 0.39834062316814717,
      "grad_norm": 0.5848958483928385,
      "learning_rate": 2.737824078403822e-06,
      "loss": 0.5104,
      "step": 4417
    },
    {
      "epoch": 0.39843080669161746,
      "grad_norm": 1.634204034269454,
      "learning_rate": 2.737281044004239e-06,
      "loss": 0.7449,
      "step": 4418
    },
    {
      "epoch": 0.3985209902150877,
      "grad_norm": 1.6522166967581051,
      "learning_rate": 2.736737946695772e-06,
      "loss": 0.7466,
      "step": 4419
    },
    {
      "epoch": 0.398611173738558,
      "grad_norm": 2.154485229115769,
      "learning_rate": 2.736194786524761e-06,
      "loss": 0.7578,
      "step": 4420
    },
    {
      "epoch": 0.3987013572620282,
      "grad_norm": 1.5610860912328708,
      "learning_rate": 2.7356515635375517e-06,
      "loss": 0.7615,
      "step": 4421
    },
    {
      "epoch": 0.3987915407854985,
      "grad_norm": 4.601928696366387,
      "learning_rate": 2.735108277780495e-06,
      "loss": 0.675,
      "step": 4422
    },
    {
      "epoch": 0.39888172430896873,
      "grad_norm": 1.6438848897030085,
      "learning_rate": 2.7345649292999456e-06,
      "loss": 0.7532,
      "step": 4423
    },
    {
      "epoch": 0.398971907832439,
      "grad_norm": 2.8911335265847686,
      "learning_rate": 2.734021518142267e-06,
      "loss": 0.7665,
      "step": 4424
    },
    {
      "epoch": 0.39906209135590925,
      "grad_norm": 4.274714472643972,
      "learning_rate": 2.733478044353825e-06,
      "loss": 0.5978,
      "step": 4425
    },
    {
      "epoch": 0.39915227487937954,
      "grad_norm": 2.3422660707575313,
      "learning_rate": 2.7329345079809917e-06,
      "loss": 0.7633,
      "step": 4426
    },
    {
      "epoch": 0.3992424584028498,
      "grad_norm": 3.26090640613919,
      "learning_rate": 2.7323909090701447e-06,
      "loss": 0.7043,
      "step": 4427
    },
    {
      "epoch": 0.39933264192632006,
      "grad_norm": 2.759889852288518,
      "learning_rate": 2.731847247667667e-06,
      "loss": 0.8152,
      "step": 4428
    },
    {
      "epoch": 0.3994228254497903,
      "grad_norm": 2.0256965428247296,
      "learning_rate": 2.731303523819947e-06,
      "loss": 0.7372,
      "step": 4429
    },
    {
      "epoch": 0.3995130089732606,
      "grad_norm": 2.1763489015248023,
      "learning_rate": 2.7307597375733783e-06,
      "loss": 0.8231,
      "step": 4430
    },
    {
      "epoch": 0.3996031924967309,
      "grad_norm": 1.8426593852366233,
      "learning_rate": 2.7302158889743587e-06,
      "loss": 0.7658,
      "step": 4431
    },
    {
      "epoch": 0.3996933760202011,
      "grad_norm": 1.6292384698496976,
      "learning_rate": 2.7296719780692937e-06,
      "loss": 0.7224,
      "step": 4432
    },
    {
      "epoch": 0.3997835595436714,
      "grad_norm": 1.486825546462053,
      "learning_rate": 2.7291280049045916e-06,
      "loss": 0.6719,
      "step": 4433
    },
    {
      "epoch": 0.39987374306714163,
      "grad_norm": 1.577447273594417,
      "learning_rate": 2.7285839695266683e-06,
      "loss": 0.7011,
      "step": 4434
    },
    {
      "epoch": 0.3999639265906119,
      "grad_norm": 4.2549757160430675,
      "learning_rate": 2.7280398719819423e-06,
      "loss": 0.7085,
      "step": 4435
    },
    {
      "epoch": 0.40005411011408215,
      "grad_norm": 2.0099679976615388,
      "learning_rate": 2.727495712316841e-06,
      "loss": 0.8239,
      "step": 4436
    },
    {
      "epoch": 0.40014429363755244,
      "grad_norm": 2.0059999827102555,
      "learning_rate": 2.7269514905777945e-06,
      "loss": 0.7113,
      "step": 4437
    },
    {
      "epoch": 0.4002344771610227,
      "grad_norm": 1.5418064810098682,
      "learning_rate": 2.7264072068112377e-06,
      "loss": 0.8832,
      "step": 4438
    },
    {
      "epoch": 0.40032466068449296,
      "grad_norm": 2.1676614629495408,
      "learning_rate": 2.7258628610636133e-06,
      "loss": 0.7248,
      "step": 4439
    },
    {
      "epoch": 0.4004148442079632,
      "grad_norm": 1.970489641616803,
      "learning_rate": 2.7253184533813667e-06,
      "loss": 0.8366,
      "step": 4440
    },
    {
      "epoch": 0.4005050277314335,
      "grad_norm": 1.7132505345407028,
      "learning_rate": 2.72477398381095e-06,
      "loss": 0.7735,
      "step": 4441
    },
    {
      "epoch": 0.4005952112549037,
      "grad_norm": 1.3459105311044015,
      "learning_rate": 2.724229452398821e-06,
      "loss": 0.7636,
      "step": 4442
    },
    {
      "epoch": 0.400685394778374,
      "grad_norm": 1.6654709420877323,
      "learning_rate": 2.7236848591914422e-06,
      "loss": 0.7803,
      "step": 4443
    },
    {
      "epoch": 0.40077557830184424,
      "grad_norm": 1.5192903067781818,
      "learning_rate": 2.7231402042352803e-06,
      "loss": 0.7488,
      "step": 4444
    },
    {
      "epoch": 0.4008657618253145,
      "grad_norm": 1.476670841501045,
      "learning_rate": 2.722595487576809e-06,
      "loss": 0.7896,
      "step": 4445
    },
    {
      "epoch": 0.40095594534878476,
      "grad_norm": 1.6392905484842684,
      "learning_rate": 2.722050709262506e-06,
      "loss": 0.8087,
      "step": 4446
    },
    {
      "epoch": 0.40104612887225505,
      "grad_norm": 1.3450815567475336,
      "learning_rate": 2.7215058693388557e-06,
      "loss": 0.74,
      "step": 4447
    },
    {
      "epoch": 0.4011363123957253,
      "grad_norm": 2.0557496549134475,
      "learning_rate": 2.720960967852346e-06,
      "loss": 0.6695,
      "step": 4448
    },
    {
      "epoch": 0.40122649591919557,
      "grad_norm": 1.5571165039487274,
      "learning_rate": 2.720416004849471e-06,
      "loss": 0.6914,
      "step": 4449
    },
    {
      "epoch": 0.4013166794426658,
      "grad_norm": 2.442105759163683,
      "learning_rate": 2.7198709803767304e-06,
      "loss": 0.713,
      "step": 4450
    },
    {
      "epoch": 0.4014068629661361,
      "grad_norm": 1.3547238108554565,
      "learning_rate": 2.7193258944806286e-06,
      "loss": 0.7289,
      "step": 4451
    },
    {
      "epoch": 0.4014970464896063,
      "grad_norm": 1.5609624545648615,
      "learning_rate": 2.718780747207675e-06,
      "loss": 0.7638,
      "step": 4452
    },
    {
      "epoch": 0.4015872300130766,
      "grad_norm": 2.609175897186174,
      "learning_rate": 2.7182355386043847e-06,
      "loss": 0.8234,
      "step": 4453
    },
    {
      "epoch": 0.40167741353654685,
      "grad_norm": 1.916312191844199,
      "learning_rate": 2.717690268717278e-06,
      "loss": 0.8257,
      "step": 4454
    },
    {
      "epoch": 0.40176759706001713,
      "grad_norm": 1.4892427637559345,
      "learning_rate": 2.7171449375928803e-06,
      "loss": 0.7877,
      "step": 4455
    },
    {
      "epoch": 0.4018577805834874,
      "grad_norm": 2.1045327450413245,
      "learning_rate": 2.716599545277722e-06,
      "loss": 0.7942,
      "step": 4456
    },
    {
      "epoch": 0.40194796410695766,
      "grad_norm": 1.5862351312083252,
      "learning_rate": 2.7160540918183394e-06,
      "loss": 0.7263,
      "step": 4457
    },
    {
      "epoch": 0.40203814763042794,
      "grad_norm": 2.8463223738551124,
      "learning_rate": 2.715508577261273e-06,
      "loss": 0.676,
      "step": 4458
    },
    {
      "epoch": 0.4021283311538982,
      "grad_norm": 2.346242672680656,
      "learning_rate": 2.7149630016530702e-06,
      "loss": 0.7381,
      "step": 4459
    },
    {
      "epoch": 0.40221851467736847,
      "grad_norm": 1.5288970151689596,
      "learning_rate": 2.7144173650402815e-06,
      "loss": 0.7795,
      "step": 4460
    },
    {
      "epoch": 0.4023086982008387,
      "grad_norm": 1.442792685713285,
      "learning_rate": 2.7138716674694636e-06,
      "loss": 0.7832,
      "step": 4461
    },
    {
      "epoch": 0.402398881724309,
      "grad_norm": 2.162954769936716,
      "learning_rate": 2.7133259089871795e-06,
      "loss": 0.7143,
      "step": 4462
    },
    {
      "epoch": 0.4024890652477792,
      "grad_norm": 1.9714834825102932,
      "learning_rate": 2.712780089639995e-06,
      "loss": 0.758,
      "step": 4463
    },
    {
      "epoch": 0.4025792487712495,
      "grad_norm": 1.5384261466503713,
      "learning_rate": 2.712234209474483e-06,
      "loss": 0.6356,
      "step": 4464
    },
    {
      "epoch": 0.40266943229471974,
      "grad_norm": 2.080813167883227,
      "learning_rate": 2.7116882685372218e-06,
      "loss": 0.805,
      "step": 4465
    },
    {
      "epoch": 0.40275961581819003,
      "grad_norm": 1.548013056564468,
      "learning_rate": 2.7111422668747927e-06,
      "loss": 0.7242,
      "step": 4466
    },
    {
      "epoch": 0.40284979934166026,
      "grad_norm": 1.4618962044093387,
      "learning_rate": 2.7105962045337846e-06,
      "loss": 0.7575,
      "step": 4467
    },
    {
      "epoch": 0.40293998286513055,
      "grad_norm": 1.5860421282451356,
      "learning_rate": 2.7100500815607898e-06,
      "loss": 0.7535,
      "step": 4468
    },
    {
      "epoch": 0.4030301663886008,
      "grad_norm": 1.6092181082980836,
      "learning_rate": 2.709503898002407e-06,
      "loss": 0.7782,
      "step": 4469
    },
    {
      "epoch": 0.4031203499120711,
      "grad_norm": 1.7129145656465479,
      "learning_rate": 2.708957653905239e-06,
      "loss": 0.7549,
      "step": 4470
    },
    {
      "epoch": 0.4032105334355413,
      "grad_norm": 1.4580594032598426,
      "learning_rate": 2.7084113493158956e-06,
      "loss": 0.7715,
      "step": 4471
    },
    {
      "epoch": 0.4033007169590116,
      "grad_norm": 1.3694977777441462,
      "learning_rate": 2.7078649842809888e-06,
      "loss": 0.7716,
      "step": 4472
    },
    {
      "epoch": 0.40339090048248183,
      "grad_norm": 1.50764503932647,
      "learning_rate": 2.707318558847139e-06,
      "loss": 0.8083,
      "step": 4473
    },
    {
      "epoch": 0.4034810840059521,
      "grad_norm": 0.7429217671682636,
      "learning_rate": 2.7067720730609697e-06,
      "loss": 0.5804,
      "step": 4474
    },
    {
      "epoch": 0.40357126752942235,
      "grad_norm": 1.6951004106605838,
      "learning_rate": 2.70622552696911e-06,
      "loss": 0.7311,
      "step": 4475
    },
    {
      "epoch": 0.40366145105289264,
      "grad_norm": 1.4424593420265468,
      "learning_rate": 2.7056789206181943e-06,
      "loss": 0.7143,
      "step": 4476
    },
    {
      "epoch": 0.40375163457636287,
      "grad_norm": 6.172781962138272,
      "learning_rate": 2.7051322540548615e-06,
      "loss": 0.8157,
      "step": 4477
    },
    {
      "epoch": 0.40384181809983316,
      "grad_norm": 1.7978505015952873,
      "learning_rate": 2.704585527325757e-06,
      "loss": 0.7341,
      "step": 4478
    },
    {
      "epoch": 0.40393200162330345,
      "grad_norm": 1.9122411041551566,
      "learning_rate": 2.7040387404775303e-06,
      "loss": 0.7216,
      "step": 4479
    },
    {
      "epoch": 0.4040221851467737,
      "grad_norm": 1.5684017113925086,
      "learning_rate": 2.703491893556837e-06,
      "loss": 0.7934,
      "step": 4480
    },
    {
      "epoch": 0.40411236867024397,
      "grad_norm": 1.4035488465536816,
      "learning_rate": 2.702944986610335e-06,
      "loss": 0.7363,
      "step": 4481
    },
    {
      "epoch": 0.4042025521937142,
      "grad_norm": 1.7794629552318189,
      "learning_rate": 2.7023980196846917e-06,
      "loss": 0.7751,
      "step": 4482
    },
    {
      "epoch": 0.4042927357171845,
      "grad_norm": 0.7164085646064666,
      "learning_rate": 2.7018509928265763e-06,
      "loss": 0.5981,
      "step": 4483
    },
    {
      "epoch": 0.4043829192406547,
      "grad_norm": 1.8286827736294815,
      "learning_rate": 2.7013039060826635e-06,
      "loss": 0.7882,
      "step": 4484
    },
    {
      "epoch": 0.404473102764125,
      "grad_norm": 1.9571962572535613,
      "learning_rate": 2.7007567594996347e-06,
      "loss": 0.8318,
      "step": 4485
    },
    {
      "epoch": 0.40456328628759525,
      "grad_norm": 1.7355829791929236,
      "learning_rate": 2.7002095531241757e-06,
      "loss": 0.6676,
      "step": 4486
    },
    {
      "epoch": 0.40465346981106554,
      "grad_norm": 2.0103273783672377,
      "learning_rate": 2.6996622870029767e-06,
      "loss": 0.7344,
      "step": 4487
    },
    {
      "epoch": 0.40474365333453577,
      "grad_norm": 1.8085788536694178,
      "learning_rate": 2.6991149611827335e-06,
      "loss": 0.8716,
      "step": 4488
    },
    {
      "epoch": 0.40483383685800606,
      "grad_norm": 1.740328220016273,
      "learning_rate": 2.6985675757101466e-06,
      "loss": 0.798,
      "step": 4489
    },
    {
      "epoch": 0.4049240203814763,
      "grad_norm": 2.75522498622734,
      "learning_rate": 2.698020130631922e-06,
      "loss": 0.7201,
      "step": 4490
    },
    {
      "epoch": 0.4050142039049466,
      "grad_norm": 1.7097110620184859,
      "learning_rate": 2.6974726259947713e-06,
      "loss": 0.6933,
      "step": 4491
    },
    {
      "epoch": 0.4051043874284168,
      "grad_norm": 3.5073956961971295,
      "learning_rate": 2.6969250618454106e-06,
      "loss": 0.7431,
      "step": 4492
    },
    {
      "epoch": 0.4051945709518871,
      "grad_norm": 1.467730391449207,
      "learning_rate": 2.696377438230561e-06,
      "loss": 0.8327,
      "step": 4493
    },
    {
      "epoch": 0.40528475447535733,
      "grad_norm": 2.1876323114979948,
      "learning_rate": 2.6958297551969484e-06,
      "loss": 0.6986,
      "step": 4494
    },
    {
      "epoch": 0.4053749379988276,
      "grad_norm": 2.3885664709982333,
      "learning_rate": 2.695282012791304e-06,
      "loss": 0.7291,
      "step": 4495
    },
    {
      "epoch": 0.40546512152229786,
      "grad_norm": 2.4502134113395915,
      "learning_rate": 2.6947342110603646e-06,
      "loss": 0.8149,
      "step": 4496
    },
    {
      "epoch": 0.40555530504576814,
      "grad_norm": 1.6950665836140868,
      "learning_rate": 2.6941863500508717e-06,
      "loss": 0.8166,
      "step": 4497
    },
    {
      "epoch": 0.4056454885692384,
      "grad_norm": 1.6761794250787632,
      "learning_rate": 2.693638429809572e-06,
      "loss": 0.8163,
      "step": 4498
    },
    {
      "epoch": 0.40573567209270867,
      "grad_norm": 1.8772946621576683,
      "learning_rate": 2.6930904503832167e-06,
      "loss": 0.7452,
      "step": 4499
    },
    {
      "epoch": 0.4058258556161789,
      "grad_norm": 1.551112449504842,
      "learning_rate": 2.692542411818562e-06,
      "loss": 0.7442,
      "step": 4500
    },
    {
      "epoch": 0.4059160391396492,
      "grad_norm": 1.7178535204204255,
      "learning_rate": 2.69199431416237e-06,
      "loss": 0.7756,
      "step": 4501
    },
    {
      "epoch": 0.4060062226631194,
      "grad_norm": 1.4949232657141753,
      "learning_rate": 2.691446157461408e-06,
      "loss": 0.7161,
      "step": 4502
    },
    {
      "epoch": 0.4060964061865897,
      "grad_norm": 1.8148268529308174,
      "learning_rate": 2.690897941762447e-06,
      "loss": 0.8221,
      "step": 4503
    },
    {
      "epoch": 0.40618658971006,
      "grad_norm": 1.7319197588298918,
      "learning_rate": 2.6903496671122642e-06,
      "loss": 0.7402,
      "step": 4504
    },
    {
      "epoch": 0.40627677323353023,
      "grad_norm": 0.6427196358460255,
      "learning_rate": 2.689801333557641e-06,
      "loss": 0.5293,
      "step": 4505
    },
    {
      "epoch": 0.4063669567570005,
      "grad_norm": 1.8862073375005446,
      "learning_rate": 2.689252941145365e-06,
      "loss": 0.7638,
      "step": 4506
    },
    {
      "epoch": 0.40645714028047075,
      "grad_norm": 1.381717097156213,
      "learning_rate": 2.6887044899222277e-06,
      "loss": 0.7572,
      "step": 4507
    },
    {
      "epoch": 0.40654732380394104,
      "grad_norm": 1.6036011908837204,
      "learning_rate": 2.688155979935025e-06,
      "loss": 0.8071,
      "step": 4508
    },
    {
      "epoch": 0.4066375073274113,
      "grad_norm": 1.6147360534632669,
      "learning_rate": 2.68760741123056e-06,
      "loss": 0.7192,
      "step": 4509
    },
    {
      "epoch": 0.40672769085088156,
      "grad_norm": 2.6590483342684523,
      "learning_rate": 2.6870587838556394e-06,
      "loss": 0.7802,
      "step": 4510
    },
    {
      "epoch": 0.4068178743743518,
      "grad_norm": 1.8082141848687605,
      "learning_rate": 2.686510097857075e-06,
      "loss": 0.7772,
      "step": 4511
    },
    {
      "epoch": 0.4069080578978221,
      "grad_norm": 1.4458378842123711,
      "learning_rate": 2.685961353281683e-06,
      "loss": 0.7544,
      "step": 4512
    },
    {
      "epoch": 0.4069982414212923,
      "grad_norm": 1.821447387031801,
      "learning_rate": 2.6854125501762863e-06,
      "loss": 0.7463,
      "step": 4513
    },
    {
      "epoch": 0.4070884249447626,
      "grad_norm": 1.9094289319402205,
      "learning_rate": 2.684863688587712e-06,
      "loss": 0.7659,
      "step": 4514
    },
    {
      "epoch": 0.40717860846823284,
      "grad_norm": 1.8381889012036219,
      "learning_rate": 2.6843147685627916e-06,
      "loss": 0.7328,
      "step": 4515
    },
    {
      "epoch": 0.4072687919917031,
      "grad_norm": 2.626737382956701,
      "learning_rate": 2.683765790148361e-06,
      "loss": 0.7427,
      "step": 4516
    },
    {
      "epoch": 0.40735897551517336,
      "grad_norm": 1.751151931924835,
      "learning_rate": 2.6832167533912637e-06,
      "loss": 0.8132,
      "step": 4517
    },
    {
      "epoch": 0.40744915903864365,
      "grad_norm": 2.655345897520648,
      "learning_rate": 2.682667658338345e-06,
      "loss": 0.8356,
      "step": 4518
    },
    {
      "epoch": 0.4075393425621139,
      "grad_norm": 1.6474769288504043,
      "learning_rate": 2.682118505036458e-06,
      "loss": 0.732,
      "step": 4519
    },
    {
      "epoch": 0.40762952608558417,
      "grad_norm": 0.6274306715703802,
      "learning_rate": 2.681569293532459e-06,
      "loss": 0.5496,
      "step": 4520
    },
    {
      "epoch": 0.4077197096090544,
      "grad_norm": 0.6567347643449629,
      "learning_rate": 2.6810200238732102e-06,
      "loss": 0.5783,
      "step": 4521
    },
    {
      "epoch": 0.4078098931325247,
      "grad_norm": 1.3736884624523475,
      "learning_rate": 2.6804706961055776e-06,
      "loss": 0.8038,
      "step": 4522
    },
    {
      "epoch": 0.4079000766559949,
      "grad_norm": 1.447677821303844,
      "learning_rate": 2.6799213102764326e-06,
      "loss": 0.795,
      "step": 4523
    },
    {
      "epoch": 0.4079902601794652,
      "grad_norm": 2.0797376757042345,
      "learning_rate": 2.679371866432653e-06,
      "loss": 0.8535,
      "step": 4524
    },
    {
      "epoch": 0.40808044370293545,
      "grad_norm": 1.6732010618240667,
      "learning_rate": 2.6788223646211194e-06,
      "loss": 0.7313,
      "step": 4525
    },
    {
      "epoch": 0.40817062722640574,
      "grad_norm": 1.5417765586796035,
      "learning_rate": 2.6782728048887183e-06,
      "loss": 0.6596,
      "step": 4526
    },
    {
      "epoch": 0.408260810749876,
      "grad_norm": 1.4628498800025038,
      "learning_rate": 2.6777231872823416e-06,
      "loss": 0.7466,
      "step": 4527
    },
    {
      "epoch": 0.40835099427334626,
      "grad_norm": 1.8672890735276102,
      "learning_rate": 2.6771735118488864e-06,
      "loss": 0.7613,
      "step": 4528
    },
    {
      "epoch": 0.40844117779681655,
      "grad_norm": 1.4799014491699678,
      "learning_rate": 2.6766237786352523e-06,
      "loss": 0.7974,
      "step": 4529
    },
    {
      "epoch": 0.4085313613202868,
      "grad_norm": 1.8815375479003695,
      "learning_rate": 2.676073987688347e-06,
      "loss": 0.6917,
      "step": 4530
    },
    {
      "epoch": 0.40862154484375707,
      "grad_norm": 1.5961141815595654,
      "learning_rate": 2.6755241390550818e-06,
      "loss": 0.7617,
      "step": 4531
    },
    {
      "epoch": 0.4087117283672273,
      "grad_norm": 1.765197634668103,
      "learning_rate": 2.6749742327823716e-06,
      "loss": 0.7614,
      "step": 4532
    },
    {
      "epoch": 0.4088019118906976,
      "grad_norm": 2.8748981623650125,
      "learning_rate": 2.674424268917138e-06,
      "loss": 0.7753,
      "step": 4533
    },
    {
      "epoch": 0.4088920954141678,
      "grad_norm": 2.0251843780596173,
      "learning_rate": 2.6738742475063074e-06,
      "loss": 0.8217,
      "step": 4534
    },
    {
      "epoch": 0.4089822789376381,
      "grad_norm": 1.8700121339618496,
      "learning_rate": 2.6733241685968104e-06,
      "loss": 0.8042,
      "step": 4535
    },
    {
      "epoch": 0.40907246246110834,
      "grad_norm": 1.9799996747873982,
      "learning_rate": 2.6727740322355826e-06,
      "loss": 0.7832,
      "step": 4536
    },
    {
      "epoch": 0.40916264598457863,
      "grad_norm": 2.3995992723339548,
      "learning_rate": 2.6722238384695644e-06,
      "loss": 0.746,
      "step": 4537
    },
    {
      "epoch": 0.40925282950804887,
      "grad_norm": 1.5293714717049505,
      "learning_rate": 2.671673587345702e-06,
      "loss": 0.7758,
      "step": 4538
    },
    {
      "epoch": 0.40934301303151915,
      "grad_norm": 0.6621824466139853,
      "learning_rate": 2.6711232789109455e-06,
      "loss": 0.5685,
      "step": 4539
    },
    {
      "epoch": 0.4094331965549894,
      "grad_norm": 1.9897924649504553,
      "learning_rate": 2.6705729132122497e-06,
      "loss": 0.7641,
      "step": 4540
    },
    {
      "epoch": 0.4095233800784597,
      "grad_norm": 1.853188617508479,
      "learning_rate": 2.670022490296576e-06,
      "loss": 0.8383,
      "step": 4541
    },
    {
      "epoch": 0.4096135636019299,
      "grad_norm": 2.6009340185811043,
      "learning_rate": 2.669472010210889e-06,
      "loss": 0.8202,
      "step": 4542
    },
    {
      "epoch": 0.4097037471254002,
      "grad_norm": 1.7650290639715904,
      "learning_rate": 2.668921473002159e-06,
      "loss": 0.7416,
      "step": 4543
    },
    {
      "epoch": 0.40979393064887043,
      "grad_norm": 1.613206986387539,
      "learning_rate": 2.6683708787173596e-06,
      "loss": 0.8142,
      "step": 4544
    },
    {
      "epoch": 0.4098841141723407,
      "grad_norm": 1.5535649986948334,
      "learning_rate": 2.6678202274034718e-06,
      "loss": 0.7792,
      "step": 4545
    },
    {
      "epoch": 0.40997429769581095,
      "grad_norm": 1.6472791606517931,
      "learning_rate": 2.66726951910748e-06,
      "loss": 0.8523,
      "step": 4546
    },
    {
      "epoch": 0.41006448121928124,
      "grad_norm": 1.5056165605801783,
      "learning_rate": 2.6667187538763737e-06,
      "loss": 0.7626,
      "step": 4547
    },
    {
      "epoch": 0.4101546647427515,
      "grad_norm": 1.796354865382512,
      "learning_rate": 2.6661679317571473e-06,
      "loss": 0.7428,
      "step": 4548
    },
    {
      "epoch": 0.41024484826622176,
      "grad_norm": 2.6118299619758285,
      "learning_rate": 2.665617052796799e-06,
      "loss": 0.7434,
      "step": 4549
    },
    {
      "epoch": 0.41033503178969205,
      "grad_norm": 1.9658478433165956,
      "learning_rate": 2.6650661170423346e-06,
      "loss": 0.6463,
      "step": 4550
    },
    {
      "epoch": 0.4104252153131623,
      "grad_norm": 1.8169939300486109,
      "learning_rate": 2.6645151245407614e-06,
      "loss": 0.8483,
      "step": 4551
    },
    {
      "epoch": 0.4105153988366326,
      "grad_norm": 1.530274372307468,
      "learning_rate": 2.6639640753390936e-06,
      "loss": 0.7567,
      "step": 4552
    },
    {
      "epoch": 0.4106055823601028,
      "grad_norm": 1.475091191075289,
      "learning_rate": 2.66341296948435e-06,
      "loss": 0.6706,
      "step": 4553
    },
    {
      "epoch": 0.4106957658835731,
      "grad_norm": 2.443683898500085,
      "learning_rate": 2.6628618070235534e-06,
      "loss": 0.7721,
      "step": 4554
    },
    {
      "epoch": 0.4107859494070433,
      "grad_norm": 2.431421967139311,
      "learning_rate": 2.662310588003733e-06,
      "loss": 0.7668,
      "step": 4555
    },
    {
      "epoch": 0.4108761329305136,
      "grad_norm": 2.3453565050809,
      "learning_rate": 2.6617593124719205e-06,
      "loss": 0.7373,
      "step": 4556
    },
    {
      "epoch": 0.41096631645398385,
      "grad_norm": 1.4085628986418166,
      "learning_rate": 2.661207980475155e-06,
      "loss": 0.7697,
      "step": 4557
    },
    {
      "epoch": 0.41105649997745414,
      "grad_norm": 2.0395745697716423,
      "learning_rate": 2.6606565920604793e-06,
      "loss": 0.722,
      "step": 4558
    },
    {
      "epoch": 0.41114668350092437,
      "grad_norm": 2.1598689352114078,
      "learning_rate": 2.66010514727494e-06,
      "loss": 0.8397,
      "step": 4559
    },
    {
      "epoch": 0.41123686702439466,
      "grad_norm": 1.7889307631586433,
      "learning_rate": 2.659553646165589e-06,
      "loss": 0.6586,
      "step": 4560
    },
    {
      "epoch": 0.4113270505478649,
      "grad_norm": 1.9621066069692612,
      "learning_rate": 2.659002088779485e-06,
      "loss": 0.7357,
      "step": 4561
    },
    {
      "epoch": 0.4114172340713352,
      "grad_norm": 0.644402739689033,
      "learning_rate": 2.6584504751636888e-06,
      "loss": 0.5276,
      "step": 4562
    },
    {
      "epoch": 0.4115074175948054,
      "grad_norm": 1.6188111194280546,
      "learning_rate": 2.657898805365268e-06,
      "loss": 0.7281,
      "step": 4563
    },
    {
      "epoch": 0.4115976011182757,
      "grad_norm": 1.7990768873611163,
      "learning_rate": 2.657347079431293e-06,
      "loss": 0.7303,
      "step": 4564
    },
    {
      "epoch": 0.41168778464174594,
      "grad_norm": 1.304087975091153,
      "learning_rate": 2.6567952974088403e-06,
      "loss": 0.7333,
      "step": 4565
    },
    {
      "epoch": 0.4117779681652162,
      "grad_norm": 1.5251896099345947,
      "learning_rate": 2.6562434593449917e-06,
      "loss": 0.8128,
      "step": 4566
    },
    {
      "epoch": 0.41186815168868646,
      "grad_norm": 1.9120571920348466,
      "learning_rate": 2.6556915652868325e-06,
      "loss": 0.7854,
      "step": 4567
    },
    {
      "epoch": 0.41195833521215675,
      "grad_norm": 1.8206985129834246,
      "learning_rate": 2.6551396152814534e-06,
      "loss": 0.7673,
      "step": 4568
    },
    {
      "epoch": 0.412048518735627,
      "grad_norm": 1.7944836833995406,
      "learning_rate": 2.65458760937595e-06,
      "loss": 0.7544,
      "step": 4569
    },
    {
      "epoch": 0.41213870225909727,
      "grad_norm": 1.4735727064086073,
      "learning_rate": 2.654035547617423e-06,
      "loss": 0.709,
      "step": 4570
    },
    {
      "epoch": 0.4122288857825675,
      "grad_norm": 1.9311359853328962,
      "learning_rate": 2.653483430052976e-06,
      "loss": 0.6663,
      "step": 4571
    },
    {
      "epoch": 0.4123190693060378,
      "grad_norm": 2.641201972001209,
      "learning_rate": 2.6529312567297197e-06,
      "loss": 0.7466,
      "step": 4572
    },
    {
      "epoch": 0.412409252829508,
      "grad_norm": 1.814080962909383,
      "learning_rate": 2.652379027694768e-06,
      "loss": 0.8379,
      "step": 4573
    },
    {
      "epoch": 0.4124994363529783,
      "grad_norm": 2.908577919834205,
      "learning_rate": 2.651826742995241e-06,
      "loss": 0.8188,
      "step": 4574
    },
    {
      "epoch": 0.4125896198764486,
      "grad_norm": 1.9220218757392729,
      "learning_rate": 2.651274402678262e-06,
      "loss": 0.7544,
      "step": 4575
    },
    {
      "epoch": 0.41267980339991883,
      "grad_norm": 1.5980265218613061,
      "learning_rate": 2.6507220067909597e-06,
      "loss": 0.7864,
      "step": 4576
    },
    {
      "epoch": 0.4127699869233891,
      "grad_norm": 1.6428427828877903,
      "learning_rate": 2.650169555380468e-06,
      "loss": 0.7203,
      "step": 4577
    },
    {
      "epoch": 0.41286017044685935,
      "grad_norm": 1.9658546350006068,
      "learning_rate": 2.6496170484939253e-06,
      "loss": 0.7546,
      "step": 4578
    },
    {
      "epoch": 0.41295035397032964,
      "grad_norm": 1.956001178163825,
      "learning_rate": 2.6490644861784735e-06,
      "loss": 0.7753,
      "step": 4579
    },
    {
      "epoch": 0.4130405374937999,
      "grad_norm": 2.0405230790253652,
      "learning_rate": 2.648511868481261e-06,
      "loss": 0.7803,
      "step": 4580
    },
    {
      "epoch": 0.41313072101727016,
      "grad_norm": 1.730351222333378,
      "learning_rate": 2.6479591954494397e-06,
      "loss": 0.7605,
      "step": 4581
    },
    {
      "epoch": 0.4132209045407404,
      "grad_norm": 2.2658657603757044,
      "learning_rate": 2.647406467130167e-06,
      "loss": 0.6598,
      "step": 4582
    },
    {
      "epoch": 0.4133110880642107,
      "grad_norm": 1.8223391535440243,
      "learning_rate": 2.646853683570605e-06,
      "loss": 0.7832,
      "step": 4583
    },
    {
      "epoch": 0.4134012715876809,
      "grad_norm": 1.5484342326756255,
      "learning_rate": 2.6463008448179196e-06,
      "loss": 0.6988,
      "step": 4584
    },
    {
      "epoch": 0.4134914551111512,
      "grad_norm": 1.8607189546593972,
      "learning_rate": 2.6457479509192828e-06,
      "loss": 0.709,
      "step": 4585
    },
    {
      "epoch": 0.41358163863462144,
      "grad_norm": 2.087828618298443,
      "learning_rate": 2.645195001921871e-06,
      "loss": 0.7542,
      "step": 4586
    },
    {
      "epoch": 0.41367182215809173,
      "grad_norm": 1.632853165261595,
      "learning_rate": 2.644641997872863e-06,
      "loss": 0.7364,
      "step": 4587
    },
    {
      "epoch": 0.41376200568156196,
      "grad_norm": 2.7392425808678573,
      "learning_rate": 2.644088938819445e-06,
      "loss": 0.7636,
      "step": 4588
    },
    {
      "epoch": 0.41385218920503225,
      "grad_norm": 1.8696401281411066,
      "learning_rate": 2.6435358248088077e-06,
      "loss": 0.8442,
      "step": 4589
    },
    {
      "epoch": 0.4139423727285025,
      "grad_norm": 2.9621895027817895,
      "learning_rate": 2.642982655888146e-06,
      "loss": 0.8013,
      "step": 4590
    },
    {
      "epoch": 0.41403255625197277,
      "grad_norm": 2.7215733392727155,
      "learning_rate": 2.6424294321046585e-06,
      "loss": 0.721,
      "step": 4591
    },
    {
      "epoch": 0.414122739775443,
      "grad_norm": 0.7416275321265258,
      "learning_rate": 2.641876153505549e-06,
      "loss": 0.5746,
      "step": 4592
    },
    {
      "epoch": 0.4142129232989133,
      "grad_norm": 1.8684496584003976,
      "learning_rate": 2.641322820138027e-06,
      "loss": 0.7661,
      "step": 4593
    },
    {
      "epoch": 0.4143031068223835,
      "grad_norm": 2.1505680132330194,
      "learning_rate": 2.640769432049306e-06,
      "loss": 0.7121,
      "step": 4594
    },
    {
      "epoch": 0.4143932903458538,
      "grad_norm": 1.6827106328113117,
      "learning_rate": 2.6402159892866038e-06,
      "loss": 0.679,
      "step": 4595
    },
    {
      "epoch": 0.41448347386932405,
      "grad_norm": 1.362415902578398,
      "learning_rate": 2.639662491897143e-06,
      "loss": 0.747,
      "step": 4596
    },
    {
      "epoch": 0.41457365739279434,
      "grad_norm": 1.6539257943967653,
      "learning_rate": 2.639108939928152e-06,
      "loss": 0.7714,
      "step": 4597
    },
    {
      "epoch": 0.4146638409162646,
      "grad_norm": 1.820494548551616,
      "learning_rate": 2.638555333426862e-06,
      "loss": 0.7714,
      "step": 4598
    },
    {
      "epoch": 0.41475402443973486,
      "grad_norm": 1.4230501956571107,
      "learning_rate": 2.6380016724405093e-06,
      "loss": 0.7916,
      "step": 4599
    },
    {
      "epoch": 0.41484420796320515,
      "grad_norm": 1.6887767341807371,
      "learning_rate": 2.637447957016336e-06,
      "loss": 0.7139,
      "step": 4600
    },
    {
      "epoch": 0.4149343914866754,
      "grad_norm": 1.9921025276762052,
      "learning_rate": 2.636894187201589e-06,
      "loss": 0.8215,
      "step": 4601
    },
    {
      "epoch": 0.41502457501014567,
      "grad_norm": 1.9001775859359202,
      "learning_rate": 2.6363403630435176e-06,
      "loss": 0.6676,
      "step": 4602
    },
    {
      "epoch": 0.4151147585336159,
      "grad_norm": 1.5156891012892346,
      "learning_rate": 2.635786484589378e-06,
      "loss": 0.7501,
      "step": 4603
    },
    {
      "epoch": 0.4152049420570862,
      "grad_norm": 1.9732965317829962,
      "learning_rate": 2.63523255188643e-06,
      "loss": 0.6644,
      "step": 4604
    },
    {
      "epoch": 0.4152951255805564,
      "grad_norm": 1.4254096396075182,
      "learning_rate": 2.6346785649819375e-06,
      "loss": 0.7818,
      "step": 4605
    },
    {
      "epoch": 0.4153853091040267,
      "grad_norm": 1.6117428887353709,
      "learning_rate": 2.6341245239231706e-06,
      "loss": 0.7843,
      "step": 4606
    },
    {
      "epoch": 0.41547549262749695,
      "grad_norm": 1.7698569429636166,
      "learning_rate": 2.6335704287574024e-06,
      "loss": 0.7595,
      "step": 4607
    },
    {
      "epoch": 0.41556567615096723,
      "grad_norm": 1.913642183203138,
      "learning_rate": 2.6330162795319124e-06,
      "loss": 0.8332,
      "step": 4608
    },
    {
      "epoch": 0.41565585967443747,
      "grad_norm": 1.930608614490511,
      "learning_rate": 2.632462076293983e-06,
      "loss": 0.812,
      "step": 4609
    },
    {
      "epoch": 0.41574604319790776,
      "grad_norm": 1.447083496495018,
      "learning_rate": 2.6319078190909017e-06,
      "loss": 0.7599,
      "step": 4610
    },
    {
      "epoch": 0.415836226721378,
      "grad_norm": 2.3992331114040724,
      "learning_rate": 2.6313535079699606e-06,
      "loss": 0.6758,
      "step": 4611
    },
    {
      "epoch": 0.4159264102448483,
      "grad_norm": 2.130704851430604,
      "learning_rate": 2.6307991429784572e-06,
      "loss": 0.7378,
      "step": 4612
    },
    {
      "epoch": 0.4160165937683185,
      "grad_norm": 2.162915197656209,
      "learning_rate": 2.6302447241636924e-06,
      "loss": 0.824,
      "step": 4613
    },
    {
      "epoch": 0.4161067772917888,
      "grad_norm": 2.1675529012417196,
      "learning_rate": 2.629690251572973e-06,
      "loss": 0.8373,
      "step": 4614
    },
    {
      "epoch": 0.41619696081525903,
      "grad_norm": 2.73036361732061,
      "learning_rate": 2.629135725253609e-06,
      "loss": 0.7367,
      "step": 4615
    },
    {
      "epoch": 0.4162871443387293,
      "grad_norm": 1.5686416466972863,
      "learning_rate": 2.6285811452529162e-06,
      "loss": 0.7467,
      "step": 4616
    },
    {
      "epoch": 0.41637732786219955,
      "grad_norm": 1.5977210145988325,
      "learning_rate": 2.6280265116182136e-06,
      "loss": 0.7777,
      "step": 4617
    },
    {
      "epoch": 0.41646751138566984,
      "grad_norm": 1.5611341992108034,
      "learning_rate": 2.6274718243968266e-06,
      "loss": 0.7646,
      "step": 4618
    },
    {
      "epoch": 0.4165576949091401,
      "grad_norm": 2.662971250710491,
      "learning_rate": 2.626917083636084e-06,
      "loss": 0.6801,
      "step": 4619
    },
    {
      "epoch": 0.41664787843261036,
      "grad_norm": 1.479494121285865,
      "learning_rate": 2.6263622893833183e-06,
      "loss": 0.6907,
      "step": 4620
    },
    {
      "epoch": 0.4167380619560806,
      "grad_norm": 1.8253185542887405,
      "learning_rate": 2.625807441685869e-06,
      "loss": 0.7061,
      "step": 4621
    },
    {
      "epoch": 0.4168282454795509,
      "grad_norm": 0.6216690470042266,
      "learning_rate": 2.625252540591078e-06,
      "loss": 0.5612,
      "step": 4622
    },
    {
      "epoch": 0.4169184290030212,
      "grad_norm": 1.839149813981733,
      "learning_rate": 2.6246975861462927e-06,
      "loss": 0.8372,
      "step": 4623
    },
    {
      "epoch": 0.4170086125264914,
      "grad_norm": 1.8331016553087498,
      "learning_rate": 2.624142578398864e-06,
      "loss": 0.8015,
      "step": 4624
    },
    {
      "epoch": 0.4170987960499617,
      "grad_norm": 1.4409252491026683,
      "learning_rate": 2.6235875173961498e-06,
      "loss": 0.7436,
      "step": 4625
    },
    {
      "epoch": 0.41718897957343193,
      "grad_norm": 1.66401829907094,
      "learning_rate": 2.62303240318551e-06,
      "loss": 0.6893,
      "step": 4626
    },
    {
      "epoch": 0.4172791630969022,
      "grad_norm": 7.551650153704507,
      "learning_rate": 2.62247723581431e-06,
      "loss": 0.8369,
      "step": 4627
    },
    {
      "epoch": 0.41736934662037245,
      "grad_norm": 1.7736363236478982,
      "learning_rate": 2.62192201532992e-06,
      "loss": 0.7626,
      "step": 4628
    },
    {
      "epoch": 0.41745953014384274,
      "grad_norm": 1.7331222271131859,
      "learning_rate": 2.6213667417797145e-06,
      "loss": 0.7775,
      "step": 4629
    },
    {
      "epoch": 0.41754971366731297,
      "grad_norm": 1.8611713355396313,
      "learning_rate": 2.6208114152110725e-06,
      "loss": 0.7881,
      "step": 4630
    },
    {
      "epoch": 0.41763989719078326,
      "grad_norm": 1.62415174205663,
      "learning_rate": 2.6202560356713774e-06,
      "loss": 0.7786,
      "step": 4631
    },
    {
      "epoch": 0.4177300807142535,
      "grad_norm": 0.6445611195434359,
      "learning_rate": 2.619700603208017e-06,
      "loss": 0.6392,
      "step": 4632
    },
    {
      "epoch": 0.4178202642377238,
      "grad_norm": 1.411042460680688,
      "learning_rate": 2.6191451178683842e-06,
      "loss": 0.6737,
      "step": 4633
    },
    {
      "epoch": 0.417910447761194,
      "grad_norm": 1.5904499369815692,
      "learning_rate": 2.6185895796998764e-06,
      "loss": 0.7924,
      "step": 4634
    },
    {
      "epoch": 0.4180006312846643,
      "grad_norm": 1.8815364708246,
      "learning_rate": 2.6180339887498946e-06,
      "loss": 0.8336,
      "step": 4635
    },
    {
      "epoch": 0.41809081480813454,
      "grad_norm": 2.3231611137466994,
      "learning_rate": 2.617478345065846e-06,
      "loss": 0.7008,
      "step": 4636
    },
    {
      "epoch": 0.4181809983316048,
      "grad_norm": 1.5660106130754328,
      "learning_rate": 2.616922648695139e-06,
      "loss": 0.7798,
      "step": 4637
    },
    {
      "epoch": 0.41827118185507506,
      "grad_norm": 1.818785194502153,
      "learning_rate": 2.61636689968519e-06,
      "loss": 0.6614,
      "step": 4638
    },
    {
      "epoch": 0.41836136537854535,
      "grad_norm": 1.9293806318541327,
      "learning_rate": 2.6158110980834186e-06,
      "loss": 0.8229,
      "step": 4639
    },
    {
      "epoch": 0.4184515489020156,
      "grad_norm": 1.87486800682869,
      "learning_rate": 2.615255243937249e-06,
      "loss": 0.8116,
      "step": 4640
    },
    {
      "epoch": 0.41854173242548587,
      "grad_norm": 1.8585740616322024,
      "learning_rate": 2.61469933729411e-06,
      "loss": 0.7854,
      "step": 4641
    },
    {
      "epoch": 0.4186319159489561,
      "grad_norm": 2.465719270052306,
      "learning_rate": 2.614143378201433e-06,
      "loss": 0.7434,
      "step": 4642
    },
    {
      "epoch": 0.4187220994724264,
      "grad_norm": 1.6276444412334798,
      "learning_rate": 2.6135873667066567e-06,
      "loss": 0.7467,
      "step": 4643
    },
    {
      "epoch": 0.4188122829958966,
      "grad_norm": 1.730707638614589,
      "learning_rate": 2.613031302857224e-06,
      "loss": 0.7193,
      "step": 4644
    },
    {
      "epoch": 0.4189024665193669,
      "grad_norm": 1.8444736159876245,
      "learning_rate": 2.6124751867005792e-06,
      "loss": 0.7861,
      "step": 4645
    },
    {
      "epoch": 0.4189926500428372,
      "grad_norm": 1.9458603910842567,
      "learning_rate": 2.611919018284175e-06,
      "loss": 0.8105,
      "step": 4646
    },
    {
      "epoch": 0.41908283356630743,
      "grad_norm": 2.5699930195398193,
      "learning_rate": 2.611362797655466e-06,
      "loss": 0.8129,
      "step": 4647
    },
    {
      "epoch": 0.4191730170897777,
      "grad_norm": 2.063981304549779,
      "learning_rate": 2.6108065248619124e-06,
      "loss": 0.8218,
      "step": 4648
    },
    {
      "epoch": 0.41926320061324795,
      "grad_norm": 2.1911266236898497,
      "learning_rate": 2.610250199950978e-06,
      "loss": 0.695,
      "step": 4649
    },
    {
      "epoch": 0.41935338413671824,
      "grad_norm": 81.05570217738385,
      "learning_rate": 2.609693822970131e-06,
      "loss": 0.7367,
      "step": 4650
    },
    {
      "epoch": 0.4194435676601885,
      "grad_norm": 3.4384768398537235,
      "learning_rate": 2.609137393966846e-06,
      "loss": 0.7285,
      "step": 4651
    },
    {
      "epoch": 0.41953375118365877,
      "grad_norm": 2.2984649967750537,
      "learning_rate": 2.6085809129886e-06,
      "loss": 0.7152,
      "step": 4652
    },
    {
      "epoch": 0.419623934707129,
      "grad_norm": 2.7169382482256808,
      "learning_rate": 2.608024380082874e-06,
      "loss": 0.7758,
      "step": 4653
    },
    {
      "epoch": 0.4197141182305993,
      "grad_norm": 2.095722237506045,
      "learning_rate": 2.6074677952971554e-06,
      "loss": 0.7639,
      "step": 4654
    },
    {
      "epoch": 0.4198043017540695,
      "grad_norm": 1.794067710891489,
      "learning_rate": 2.606911158678935e-06,
      "loss": 0.6877,
      "step": 4655
    },
    {
      "epoch": 0.4198944852775398,
      "grad_norm": 2.0973573041897247,
      "learning_rate": 2.606354470275708e-06,
      "loss": 0.7502,
      "step": 4656
    },
    {
      "epoch": 0.41998466880101004,
      "grad_norm": 1.6263370148750715,
      "learning_rate": 2.6057977301349744e-06,
      "loss": 0.7334,
      "step": 4657
    },
    {
      "epoch": 0.42007485232448033,
      "grad_norm": 1.624175889778601,
      "learning_rate": 2.6052409383042383e-06,
      "loss": 0.7287,
      "step": 4658
    },
    {
      "epoch": 0.42016503584795056,
      "grad_norm": 2.7918317281026757,
      "learning_rate": 2.6046840948310074e-06,
      "loss": 0.7366,
      "step": 4659
    },
    {
      "epoch": 0.42025521937142085,
      "grad_norm": 1.6022451295080753,
      "learning_rate": 2.6041271997627962e-06,
      "loss": 0.7642,
      "step": 4660
    },
    {
      "epoch": 0.4203454028948911,
      "grad_norm": 2.0401191166153136,
      "learning_rate": 2.6035702531471202e-06,
      "loss": 0.7506,
      "step": 4661
    },
    {
      "epoch": 0.4204355864183614,
      "grad_norm": 1.572800141454052,
      "learning_rate": 2.6030132550315035e-06,
      "loss": 0.7555,
      "step": 4662
    },
    {
      "epoch": 0.4205257699418316,
      "grad_norm": 0.6451676434734771,
      "learning_rate": 2.60245620546347e-06,
      "loss": 0.6144,
      "step": 4663
    },
    {
      "epoch": 0.4206159534653019,
      "grad_norm": 1.859833220110991,
      "learning_rate": 2.6018991044905517e-06,
      "loss": 0.7357,
      "step": 4664
    },
    {
      "epoch": 0.42070613698877213,
      "grad_norm": 1.6409497620809674,
      "learning_rate": 2.6013419521602825e-06,
      "loss": 0.7925,
      "step": 4665
    },
    {
      "epoch": 0.4207963205122424,
      "grad_norm": 1.6890809459543625,
      "learning_rate": 2.600784748520202e-06,
      "loss": 0.8896,
      "step": 4666
    },
    {
      "epoch": 0.42088650403571265,
      "grad_norm": 1.596707461054317,
      "learning_rate": 2.6002274936178544e-06,
      "loss": 0.6974,
      "step": 4667
    },
    {
      "epoch": 0.42097668755918294,
      "grad_norm": 1.629709754565136,
      "learning_rate": 2.5996701875007873e-06,
      "loss": 0.7692,
      "step": 4668
    },
    {
      "epoch": 0.4210668710826532,
      "grad_norm": 1.7896959940657315,
      "learning_rate": 2.5991128302165533e-06,
      "loss": 0.7524,
      "step": 4669
    },
    {
      "epoch": 0.42115705460612346,
      "grad_norm": 2.7687137403331215,
      "learning_rate": 2.5985554218127094e-06,
      "loss": 0.8216,
      "step": 4670
    },
    {
      "epoch": 0.42124723812959375,
      "grad_norm": 5.336312276655973,
      "learning_rate": 2.597997962336816e-06,
      "loss": 0.694,
      "step": 4671
    },
    {
      "epoch": 0.421337421653064,
      "grad_norm": 1.7140681917831517,
      "learning_rate": 2.5974404518364393e-06,
      "loss": 0.7531,
      "step": 4672
    },
    {
      "epoch": 0.42142760517653427,
      "grad_norm": 1.734939689467469,
      "learning_rate": 2.596882890359149e-06,
      "loss": 0.885,
      "step": 4673
    },
    {
      "epoch": 0.4215177887000045,
      "grad_norm": 1.9550011130912714,
      "learning_rate": 2.5963252779525196e-06,
      "loss": 0.6817,
      "step": 4674
    },
    {
      "epoch": 0.4216079722234748,
      "grad_norm": 2.5596137190496284,
      "learning_rate": 2.595767614664129e-06,
      "loss": 0.7586,
      "step": 4675
    },
    {
      "epoch": 0.421698155746945,
      "grad_norm": 1.532797440232268,
      "learning_rate": 2.5952099005415607e-06,
      "loss": 0.7386,
      "step": 4676
    },
    {
      "epoch": 0.4217883392704153,
      "grad_norm": 2.079936793724139,
      "learning_rate": 2.594652135632402e-06,
      "loss": 0.7217,
      "step": 4677
    },
    {
      "epoch": 0.42187852279388555,
      "grad_norm": 1.6396607244598072,
      "learning_rate": 2.594094319984244e-06,
      "loss": 0.7155,
      "step": 4678
    },
    {
      "epoch": 0.42196870631735584,
      "grad_norm": 1.758340713822964,
      "learning_rate": 2.5935364536446825e-06,
      "loss": 0.8165,
      "step": 4679
    },
    {
      "epoch": 0.42205888984082607,
      "grad_norm": 3.537792390527477,
      "learning_rate": 2.5929785366613185e-06,
      "loss": 0.758,
      "step": 4680
    },
    {
      "epoch": 0.42214907336429636,
      "grad_norm": 2.2850680864069193,
      "learning_rate": 2.592420569081756e-06,
      "loss": 0.7974,
      "step": 4681
    },
    {
      "epoch": 0.4222392568877666,
      "grad_norm": 1.6908416863027236,
      "learning_rate": 2.5918625509536037e-06,
      "loss": 0.7223,
      "step": 4682
    },
    {
      "epoch": 0.4223294404112369,
      "grad_norm": 0.633040963834648,
      "learning_rate": 2.591304482324475e-06,
      "loss": 0.579,
      "step": 4683
    },
    {
      "epoch": 0.4224196239347071,
      "grad_norm": 2.3129475134779343,
      "learning_rate": 2.5907463632419878e-06,
      "loss": 0.7724,
      "step": 4684
    },
    {
      "epoch": 0.4225098074581774,
      "grad_norm": 3.4847340762289525,
      "learning_rate": 2.5901881937537632e-06,
      "loss": 0.7284,
      "step": 4685
    },
    {
      "epoch": 0.42259999098164763,
      "grad_norm": 0.8033686450132945,
      "learning_rate": 2.589629973907428e-06,
      "loss": 0.6121,
      "step": 4686
    },
    {
      "epoch": 0.4226901745051179,
      "grad_norm": 2.027543780210471,
      "learning_rate": 2.589071703750612e-06,
      "loss": 0.7095,
      "step": 4687
    },
    {
      "epoch": 0.42278035802858815,
      "grad_norm": 1.6337676106724521,
      "learning_rate": 2.5885133833309504e-06,
      "loss": 0.7973,
      "step": 4688
    },
    {
      "epoch": 0.42287054155205844,
      "grad_norm": 1.5030070205037207,
      "learning_rate": 2.5879550126960814e-06,
      "loss": 0.84,
      "step": 4689
    },
    {
      "epoch": 0.4229607250755287,
      "grad_norm": 1.605487312896411,
      "learning_rate": 2.5873965918936494e-06,
      "loss": 0.8141,
      "step": 4690
    },
    {
      "epoch": 0.42305090859899896,
      "grad_norm": 1.7439374816083002,
      "learning_rate": 2.586838120971301e-06,
      "loss": 0.7859,
      "step": 4691
    },
    {
      "epoch": 0.4231410921224692,
      "grad_norm": 1.6011543312051577,
      "learning_rate": 2.586279599976689e-06,
      "loss": 0.8195,
      "step": 4692
    },
    {
      "epoch": 0.4232312756459395,
      "grad_norm": 1.602518085014666,
      "learning_rate": 2.585721028957468e-06,
      "loss": 0.7613,
      "step": 4693
    },
    {
      "epoch": 0.4233214591694098,
      "grad_norm": 1.6803869875743007,
      "learning_rate": 2.585162407961299e-06,
      "loss": 0.7724,
      "step": 4694
    },
    {
      "epoch": 0.42341164269288,
      "grad_norm": 2.489462866854615,
      "learning_rate": 2.584603737035847e-06,
      "loss": 0.7029,
      "step": 4695
    },
    {
      "epoch": 0.4235018262163503,
      "grad_norm": 2.096766448302435,
      "learning_rate": 2.5840450162287806e-06,
      "loss": 0.7599,
      "step": 4696
    },
    {
      "epoch": 0.42359200973982053,
      "grad_norm": 1.8946002750996942,
      "learning_rate": 2.583486245587774e-06,
      "loss": 0.6732,
      "step": 4697
    },
    {
      "epoch": 0.4236821932632908,
      "grad_norm": 1.7254727061436255,
      "learning_rate": 2.5829274251605023e-06,
      "loss": 0.8663,
      "step": 4698
    },
    {
      "epoch": 0.42377237678676105,
      "grad_norm": 3.2038175997112006,
      "learning_rate": 2.582368554994649e-06,
      "loss": 0.7328,
      "step": 4699
    },
    {
      "epoch": 0.42386256031023134,
      "grad_norm": 0.747932445249884,
      "learning_rate": 2.5818096351378994e-06,
      "loss": 0.6144,
      "step": 4700
    },
    {
      "epoch": 0.4239527438337016,
      "grad_norm": 2.3896281838192057,
      "learning_rate": 2.5812506656379435e-06,
      "loss": 0.8176,
      "step": 4701
    },
    {
      "epoch": 0.42404292735717186,
      "grad_norm": 1.7060467916916242,
      "learning_rate": 2.580691646542476e-06,
      "loss": 0.7124,
      "step": 4702
    },
    {
      "epoch": 0.4241331108806421,
      "grad_norm": 1.6060667386136933,
      "learning_rate": 2.5801325778991958e-06,
      "loss": 0.7952,
      "step": 4703
    },
    {
      "epoch": 0.4242232944041124,
      "grad_norm": 2.752697488627783,
      "learning_rate": 2.5795734597558043e-06,
      "loss": 0.6201,
      "step": 4704
    },
    {
      "epoch": 0.4243134779275826,
      "grad_norm": 2.267097678531449,
      "learning_rate": 2.579014292160011e-06,
      "loss": 0.7606,
      "step": 4705
    },
    {
      "epoch": 0.4244036614510529,
      "grad_norm": 1.3544414045939357,
      "learning_rate": 2.5784550751595236e-06,
      "loss": 0.7414,
      "step": 4706
    },
    {
      "epoch": 0.42449384497452314,
      "grad_norm": 6.521935058083258,
      "learning_rate": 2.577895808802061e-06,
      "loss": 0.8198,
      "step": 4707
    },
    {
      "epoch": 0.4245840284979934,
      "grad_norm": 1.6406812385729364,
      "learning_rate": 2.577336493135341e-06,
      "loss": 0.6786,
      "step": 4708
    },
    {
      "epoch": 0.42467421202146366,
      "grad_norm": 1.5253252120407577,
      "learning_rate": 2.576777128207088e-06,
      "loss": 0.8532,
      "step": 4709
    },
    {
      "epoch": 0.42476439554493395,
      "grad_norm": 1.3849215614747585,
      "learning_rate": 2.5762177140650306e-06,
      "loss": 0.719,
      "step": 4710
    },
    {
      "epoch": 0.4248545790684042,
      "grad_norm": 1.587513483545908,
      "learning_rate": 2.5756582507569003e-06,
      "loss": 0.768,
      "step": 4711
    },
    {
      "epoch": 0.42494476259187447,
      "grad_norm": 1.598720802118056,
      "learning_rate": 2.5750987383304335e-06,
      "loss": 0.7815,
      "step": 4712
    },
    {
      "epoch": 0.4250349461153447,
      "grad_norm": 2.6920105985061578,
      "learning_rate": 2.574539176833372e-06,
      "loss": 0.7452,
      "step": 4713
    },
    {
      "epoch": 0.425125129638815,
      "grad_norm": 2.241303911129836,
      "learning_rate": 2.5739795663134594e-06,
      "loss": 0.7122,
      "step": 4714
    },
    {
      "epoch": 0.4252153131622852,
      "grad_norm": 1.5339979310550191,
      "learning_rate": 2.5734199068184454e-06,
      "loss": 0.8453,
      "step": 4715
    },
    {
      "epoch": 0.4253054966857555,
      "grad_norm": 1.9211771248452565,
      "learning_rate": 2.572860198396083e-06,
      "loss": 0.6868,
      "step": 4716
    },
    {
      "epoch": 0.4253956802092258,
      "grad_norm": 1.4911409872520154,
      "learning_rate": 2.57230044109413e-06,
      "loss": 0.7837,
      "step": 4717
    },
    {
      "epoch": 0.42548586373269603,
      "grad_norm": 2.443858631578127,
      "learning_rate": 2.5717406349603483e-06,
      "loss": 0.7421,
      "step": 4718
    },
    {
      "epoch": 0.4255760472561663,
      "grad_norm": 1.50988792723522,
      "learning_rate": 2.5711807800425026e-06,
      "loss": 0.7773,
      "step": 4719
    },
    {
      "epoch": 0.42566623077963656,
      "grad_norm": 2.3891868504058746,
      "learning_rate": 2.5706208763883633e-06,
      "loss": 0.5986,
      "step": 4720
    },
    {
      "epoch": 0.42575641430310684,
      "grad_norm": 2.1954395623955008,
      "learning_rate": 2.570060924045704e-06,
      "loss": 0.7129,
      "step": 4721
    },
    {
      "epoch": 0.4258465978265771,
      "grad_norm": 1.7485773570942396,
      "learning_rate": 2.569500923062304e-06,
      "loss": 0.7932,
      "step": 4722
    },
    {
      "epoch": 0.42593678135004737,
      "grad_norm": 1.6482396300822972,
      "learning_rate": 2.5689408734859445e-06,
      "loss": 0.6816,
      "step": 4723
    },
    {
      "epoch": 0.4260269648735176,
      "grad_norm": 1.6205683948474807,
      "learning_rate": 2.5683807753644127e-06,
      "loss": 0.7099,
      "step": 4724
    },
    {
      "epoch": 0.4261171483969879,
      "grad_norm": 1.6284350255619064,
      "learning_rate": 2.5678206287454996e-06,
      "loss": 0.7331,
      "step": 4725
    },
    {
      "epoch": 0.4262073319204581,
      "grad_norm": 1.9078236947445857,
      "learning_rate": 2.567260433676999e-06,
      "loss": 0.7986,
      "step": 4726
    },
    {
      "epoch": 0.4262975154439284,
      "grad_norm": 3.4831711951915607,
      "learning_rate": 2.5667001902067107e-06,
      "loss": 0.7341,
      "step": 4727
    },
    {
      "epoch": 0.42638769896739864,
      "grad_norm": 1.7853001509415793,
      "learning_rate": 2.566139898382437e-06,
      "loss": 0.8143,
      "step": 4728
    },
    {
      "epoch": 0.42647788249086893,
      "grad_norm": 3.173313359697244,
      "learning_rate": 2.5655795582519853e-06,
      "loss": 0.7766,
      "step": 4729
    },
    {
      "epoch": 0.42656806601433916,
      "grad_norm": 1.5716655462725555,
      "learning_rate": 2.565019169863168e-06,
      "loss": 0.827,
      "step": 4730
    },
    {
      "epoch": 0.42665824953780945,
      "grad_norm": 0.5874292331036345,
      "learning_rate": 2.5644587332637994e-06,
      "loss": 0.5483,
      "step": 4731
    },
    {
      "epoch": 0.4267484330612797,
      "grad_norm": 1.8953857382565953,
      "learning_rate": 2.5638982485016994e-06,
      "loss": 0.7892,
      "step": 4732
    },
    {
      "epoch": 0.42683861658475,
      "grad_norm": 0.7155511826336495,
      "learning_rate": 2.5633377156246917e-06,
      "loss": 0.553,
      "step": 4733
    },
    {
      "epoch": 0.4269288001082202,
      "grad_norm": 2.0102948825561744,
      "learning_rate": 2.562777134680603e-06,
      "loss": 0.7447,
      "step": 4734
    },
    {
      "epoch": 0.4270189836316905,
      "grad_norm": 1.9928727112243523,
      "learning_rate": 2.562216505717267e-06,
      "loss": 0.7742,
      "step": 4735
    },
    {
      "epoch": 0.42710916715516073,
      "grad_norm": 2.1850917093335562,
      "learning_rate": 2.561655828782518e-06,
      "loss": 0.783,
      "step": 4736
    },
    {
      "epoch": 0.427199350678631,
      "grad_norm": 2.250563656882791,
      "learning_rate": 2.561095103924197e-06,
      "loss": 0.7406,
      "step": 4737
    },
    {
      "epoch": 0.42728953420210125,
      "grad_norm": 1.6857469424508957,
      "learning_rate": 2.560534331190148e-06,
      "loss": 0.7114,
      "step": 4738
    },
    {
      "epoch": 0.42737971772557154,
      "grad_norm": 0.578678484755721,
      "learning_rate": 2.559973510628218e-06,
      "loss": 0.5494,
      "step": 4739
    },
    {
      "epoch": 0.4274699012490418,
      "grad_norm": 1.7881470049477786,
      "learning_rate": 2.5594126422862615e-06,
      "loss": 0.7081,
      "step": 4740
    },
    {
      "epoch": 0.42756008477251206,
      "grad_norm": 2.794236627649639,
      "learning_rate": 2.558851726212134e-06,
      "loss": 0.7113,
      "step": 4741
    },
    {
      "epoch": 0.42765026829598235,
      "grad_norm": 3.1141729860664444,
      "learning_rate": 2.5582907624536953e-06,
      "loss": 0.7325,
      "step": 4742
    },
    {
      "epoch": 0.4277404518194526,
      "grad_norm": 1.7682934067245812,
      "learning_rate": 2.557729751058811e-06,
      "loss": 0.7689,
      "step": 4743
    },
    {
      "epoch": 0.42783063534292287,
      "grad_norm": 1.5678668359195533,
      "learning_rate": 2.557168692075348e-06,
      "loss": 0.7309,
      "step": 4744
    },
    {
      "epoch": 0.4279208188663931,
      "grad_norm": 1.3957079456245027,
      "learning_rate": 2.556607585551181e-06,
      "loss": 0.7018,
      "step": 4745
    },
    {
      "epoch": 0.4280110023898634,
      "grad_norm": 0.7220299971337872,
      "learning_rate": 2.5560464315341844e-06,
      "loss": 0.678,
      "step": 4746
    },
    {
      "epoch": 0.4281011859133336,
      "grad_norm": 1.7126879512364612,
      "learning_rate": 2.555485230072242e-06,
      "loss": 0.7509,
      "step": 4747
    },
    {
      "epoch": 0.4281913694368039,
      "grad_norm": 1.8624611895313083,
      "learning_rate": 2.5549239812132354e-06,
      "loss": 0.807,
      "step": 4748
    },
    {
      "epoch": 0.42828155296027415,
      "grad_norm": 1.6493445337162451,
      "learning_rate": 2.5543626850050556e-06,
      "loss": 0.8195,
      "step": 4749
    },
    {
      "epoch": 0.42837173648374444,
      "grad_norm": 1.4657324777981036,
      "learning_rate": 2.5538013414955944e-06,
      "loss": 0.8119,
      "step": 4750
    },
    {
      "epoch": 0.42846192000721467,
      "grad_norm": 1.694533138177364,
      "learning_rate": 2.5532399507327494e-06,
      "loss": 0.7131,
      "step": 4751
    },
    {
      "epoch": 0.42855210353068496,
      "grad_norm": 0.6368757417129824,
      "learning_rate": 2.552678512764421e-06,
      "loss": 0.587,
      "step": 4752
    },
    {
      "epoch": 0.4286422870541552,
      "grad_norm": 1.993148093414256,
      "learning_rate": 2.5521170276385147e-06,
      "loss": 0.8138,
      "step": 4753
    },
    {
      "epoch": 0.4287324705776255,
      "grad_norm": 2.3256122931379566,
      "learning_rate": 2.5515554954029394e-06,
      "loss": 0.8143,
      "step": 4754
    },
    {
      "epoch": 0.4288226541010957,
      "grad_norm": 1.820675269394428,
      "learning_rate": 2.550993916105608e-06,
      "loss": 0.7913,
      "step": 4755
    },
    {
      "epoch": 0.428912837624566,
      "grad_norm": 1.473412600293735,
      "learning_rate": 2.550432289794437e-06,
      "loss": 0.7141,
      "step": 4756
    },
    {
      "epoch": 0.42900302114803623,
      "grad_norm": 1.8497701734114833,
      "learning_rate": 2.5498706165173483e-06,
      "loss": 0.768,
      "step": 4757
    },
    {
      "epoch": 0.4290932046715065,
      "grad_norm": 1.323311328841118,
      "learning_rate": 2.5493088963222668e-06,
      "loss": 0.7327,
      "step": 4758
    },
    {
      "epoch": 0.42918338819497676,
      "grad_norm": 1.8032336303145626,
      "learning_rate": 2.548747129257121e-06,
      "loss": 0.8086,
      "step": 4759
    },
    {
      "epoch": 0.42927357171844704,
      "grad_norm": 1.903779476987046,
      "learning_rate": 2.548185315369845e-06,
      "loss": 0.6778,
      "step": 4760
    },
    {
      "epoch": 0.4293637552419173,
      "grad_norm": 1.4974035998156832,
      "learning_rate": 2.5476234547083746e-06,
      "loss": 0.845,
      "step": 4761
    },
    {
      "epoch": 0.42945393876538757,
      "grad_norm": 1.5937947285677359,
      "learning_rate": 2.547061547320652e-06,
      "loss": 0.8314,
      "step": 4762
    },
    {
      "epoch": 0.4295441222888578,
      "grad_norm": 1.371695275145484,
      "learning_rate": 2.5464995932546217e-06,
      "loss": 0.7408,
      "step": 4763
    },
    {
      "epoch": 0.4296343058123281,
      "grad_norm": 2.078206340373873,
      "learning_rate": 2.545937592558232e-06,
      "loss": 0.7136,
      "step": 4764
    },
    {
      "epoch": 0.4297244893357984,
      "grad_norm": 1.5109669789985967,
      "learning_rate": 2.5453755452794374e-06,
      "loss": 0.7076,
      "step": 4765
    },
    {
      "epoch": 0.4298146728592686,
      "grad_norm": 1.9748774647424787,
      "learning_rate": 2.5448134514661938e-06,
      "loss": 0.7997,
      "step": 4766
    },
    {
      "epoch": 0.4299048563827389,
      "grad_norm": 1.4000405016217694,
      "learning_rate": 2.5442513111664623e-06,
      "loss": 0.7558,
      "step": 4767
    },
    {
      "epoch": 0.42999503990620913,
      "grad_norm": 2.012071182341733,
      "learning_rate": 2.5436891244282084e-06,
      "loss": 0.7591,
      "step": 4768
    },
    {
      "epoch": 0.4300852234296794,
      "grad_norm": 1.703439246068132,
      "learning_rate": 2.5431268912994004e-06,
      "loss": 0.8534,
      "step": 4769
    },
    {
      "epoch": 0.43017540695314965,
      "grad_norm": 1.6231701157990182,
      "learning_rate": 2.5425646118280108e-06,
      "loss": 0.7278,
      "step": 4770
    },
    {
      "epoch": 0.43026559047661994,
      "grad_norm": 0.5588498628799613,
      "learning_rate": 2.5420022860620172e-06,
      "loss": 0.5628,
      "step": 4771
    },
    {
      "epoch": 0.4303557740000902,
      "grad_norm": 2.1998905761555747,
      "learning_rate": 2.5414399140493995e-06,
      "loss": 0.6871,
      "step": 4772
    },
    {
      "epoch": 0.43044595752356046,
      "grad_norm": 2.1411158980350313,
      "learning_rate": 2.5408774958381436e-06,
      "loss": 0.6954,
      "step": 4773
    },
    {
      "epoch": 0.4305361410470307,
      "grad_norm": 2.0113987344070825,
      "learning_rate": 2.540315031476237e-06,
      "loss": 0.7937,
      "step": 4774
    },
    {
      "epoch": 0.430626324570501,
      "grad_norm": 1.5860715912755536,
      "learning_rate": 2.5397525210116737e-06,
      "loss": 0.7643,
      "step": 4775
    },
    {
      "epoch": 0.4307165080939712,
      "grad_norm": 1.5220411895311488,
      "learning_rate": 2.539189964492448e-06,
      "loss": 0.7123,
      "step": 4776
    },
    {
      "epoch": 0.4308066916174415,
      "grad_norm": 1.5886893615435616,
      "learning_rate": 2.5386273619665613e-06,
      "loss": 0.798,
      "step": 4777
    },
    {
      "epoch": 0.43089687514091174,
      "grad_norm": 0.6840446075288248,
      "learning_rate": 2.5380647134820186e-06,
      "loss": 0.5838,
      "step": 4778
    },
    {
      "epoch": 0.43098705866438203,
      "grad_norm": 1.4293651947098833,
      "learning_rate": 2.5375020190868277e-06,
      "loss": 0.8133,
      "step": 4779
    },
    {
      "epoch": 0.43107724218785226,
      "grad_norm": 1.8089411424558586,
      "learning_rate": 2.536939278829001e-06,
      "loss": 0.7511,
      "step": 4780
    },
    {
      "epoch": 0.43116742571132255,
      "grad_norm": 1.7719144625028482,
      "learning_rate": 2.5363764927565536e-06,
      "loss": 0.8153,
      "step": 4781
    },
    {
      "epoch": 0.4312576092347928,
      "grad_norm": 2.5982721272670353,
      "learning_rate": 2.5358136609175064e-06,
      "loss": 0.7177,
      "step": 4782
    },
    {
      "epoch": 0.43134779275826307,
      "grad_norm": 1.4496201116215166,
      "learning_rate": 2.535250783359884e-06,
      "loss": 0.7574,
      "step": 4783
    },
    {
      "epoch": 0.4314379762817333,
      "grad_norm": 1.3683118719926206,
      "learning_rate": 2.5346878601317124e-06,
      "loss": 0.6495,
      "step": 4784
    },
    {
      "epoch": 0.4315281598052036,
      "grad_norm": 0.5813197452990623,
      "learning_rate": 2.534124891281025e-06,
      "loss": 0.5476,
      "step": 4785
    },
    {
      "epoch": 0.4316183433286738,
      "grad_norm": 2.19438265728246,
      "learning_rate": 2.533561876855857e-06,
      "loss": 0.7731,
      "step": 4786
    },
    {
      "epoch": 0.4317085268521441,
      "grad_norm": 11.382800435391685,
      "learning_rate": 2.532998816904247e-06,
      "loss": 0.7098,
      "step": 4787
    },
    {
      "epoch": 0.4317987103756144,
      "grad_norm": 1.6383750382965683,
      "learning_rate": 2.53243571147424e-06,
      "loss": 0.6537,
      "step": 4788
    },
    {
      "epoch": 0.43188889389908464,
      "grad_norm": 2.0748309652579677,
      "learning_rate": 2.5318725606138815e-06,
      "loss": 0.6719,
      "step": 4789
    },
    {
      "epoch": 0.4319790774225549,
      "grad_norm": 1.6463955349558572,
      "learning_rate": 2.5313093643712235e-06,
      "loss": 0.7617,
      "step": 4790
    },
    {
      "epoch": 0.43206926094602516,
      "grad_norm": 1.5301002545639693,
      "learning_rate": 2.530746122794321e-06,
      "loss": 0.8024,
      "step": 4791
    },
    {
      "epoch": 0.43215944446949545,
      "grad_norm": 1.474306916803103,
      "learning_rate": 2.5301828359312323e-06,
      "loss": 0.6873,
      "step": 4792
    },
    {
      "epoch": 0.4322496279929657,
      "grad_norm": 1.7584450494913648,
      "learning_rate": 2.529619503830021e-06,
      "loss": 0.7121,
      "step": 4793
    },
    {
      "epoch": 0.43233981151643597,
      "grad_norm": 3.1113455964081016,
      "learning_rate": 2.529056126538753e-06,
      "loss": 0.6924,
      "step": 4794
    },
    {
      "epoch": 0.4324299950399062,
      "grad_norm": 2.8332670054432394,
      "learning_rate": 2.5284927041054995e-06,
      "loss": 0.7888,
      "step": 4795
    },
    {
      "epoch": 0.4325201785633765,
      "grad_norm": 0.6147225567081831,
      "learning_rate": 2.5279292365783348e-06,
      "loss": 0.5251,
      "step": 4796
    },
    {
      "epoch": 0.4326103620868467,
      "grad_norm": 1.6738056194701998,
      "learning_rate": 2.527365724005336e-06,
      "loss": 0.849,
      "step": 4797
    },
    {
      "epoch": 0.432700545610317,
      "grad_norm": 1.7003391825077987,
      "learning_rate": 2.526802166434586e-06,
      "loss": 0.8099,
      "step": 4798
    },
    {
      "epoch": 0.43279072913378724,
      "grad_norm": 1.4794362677654076,
      "learning_rate": 2.5262385639141708e-06,
      "loss": 0.7246,
      "step": 4799
    },
    {
      "epoch": 0.43288091265725753,
      "grad_norm": 1.6221803263838341,
      "learning_rate": 2.525674916492179e-06,
      "loss": 0.828,
      "step": 4800
    },
    {
      "epoch": 0.43297109618072777,
      "grad_norm": 1.594143239294674,
      "learning_rate": 2.5251112242167056e-06,
      "loss": 0.7621,
      "step": 4801
    },
    {
      "epoch": 0.43306127970419805,
      "grad_norm": 1.631701954170072,
      "learning_rate": 2.5245474871358464e-06,
      "loss": 0.7155,
      "step": 4802
    },
    {
      "epoch": 0.4331514632276683,
      "grad_norm": 1.866648193109017,
      "learning_rate": 2.5239837052977037e-06,
      "loss": 0.6741,
      "step": 4803
    },
    {
      "epoch": 0.4332416467511386,
      "grad_norm": 1.7490235738152515,
      "learning_rate": 2.523419878750381e-06,
      "loss": 0.8647,
      "step": 4804
    },
    {
      "epoch": 0.4333318302746088,
      "grad_norm": 2.0523009854476473,
      "learning_rate": 2.522856007541989e-06,
      "loss": 0.7118,
      "step": 4805
    },
    {
      "epoch": 0.4334220137980791,
      "grad_norm": 1.8123793726275765,
      "learning_rate": 2.5222920917206397e-06,
      "loss": 0.6651,
      "step": 4806
    },
    {
      "epoch": 0.43351219732154933,
      "grad_norm": 0.6078948359876132,
      "learning_rate": 2.5217281313344493e-06,
      "loss": 0.5626,
      "step": 4807
    },
    {
      "epoch": 0.4336023808450196,
      "grad_norm": 1.7215564356939614,
      "learning_rate": 2.5211641264315372e-06,
      "loss": 0.8007,
      "step": 4808
    },
    {
      "epoch": 0.43369256436848985,
      "grad_norm": 1.6571721083192255,
      "learning_rate": 2.5206000770600286e-06,
      "loss": 0.7352,
      "step": 4809
    },
    {
      "epoch": 0.43378274789196014,
      "grad_norm": 3.735241649730947,
      "learning_rate": 2.520035983268051e-06,
      "loss": 0.7134,
      "step": 4810
    },
    {
      "epoch": 0.4338729314154304,
      "grad_norm": 4.057817548264693,
      "learning_rate": 2.5194718451037357e-06,
      "loss": 0.7566,
      "step": 4811
    },
    {
      "epoch": 0.43396311493890066,
      "grad_norm": 3.0227258125557177,
      "learning_rate": 2.518907662615218e-06,
      "loss": 0.7825,
      "step": 4812
    },
    {
      "epoch": 0.43405329846237095,
      "grad_norm": 1.9216424444656577,
      "learning_rate": 2.5183434358506373e-06,
      "loss": 0.7205,
      "step": 4813
    },
    {
      "epoch": 0.4341434819858412,
      "grad_norm": 1.6875649898871372,
      "learning_rate": 2.5177791648581368e-06,
      "loss": 0.7363,
      "step": 4814
    },
    {
      "epoch": 0.4342336655093115,
      "grad_norm": 1.5410757306927205,
      "learning_rate": 2.517214849685863e-06,
      "loss": 0.7577,
      "step": 4815
    },
    {
      "epoch": 0.4343238490327817,
      "grad_norm": 1.5636737229781579,
      "learning_rate": 2.5166504903819663e-06,
      "loss": 0.7739,
      "step": 4816
    },
    {
      "epoch": 0.434414032556252,
      "grad_norm": 2.299099007643699,
      "learning_rate": 2.5160860869946014e-06,
      "loss": 0.823,
      "step": 4817
    },
    {
      "epoch": 0.4345042160797222,
      "grad_norm": 1.4793050011793654,
      "learning_rate": 2.5155216395719253e-06,
      "loss": 0.8399,
      "step": 4818
    },
    {
      "epoch": 0.4345943996031925,
      "grad_norm": 1.574677961214233,
      "learning_rate": 2.5149571481621e-06,
      "loss": 0.7164,
      "step": 4819
    },
    {
      "epoch": 0.43468458312666275,
      "grad_norm": 0.596887405626087,
      "learning_rate": 2.514392612813292e-06,
      "loss": 0.5455,
      "step": 4820
    },
    {
      "epoch": 0.43477476665013304,
      "grad_norm": 1.7034548518784716,
      "learning_rate": 2.5138280335736695e-06,
      "loss": 0.8143,
      "step": 4821
    },
    {
      "epoch": 0.43486495017360327,
      "grad_norm": 1.8018892731301388,
      "learning_rate": 2.5132634104914064e-06,
      "loss": 0.8382,
      "step": 4822
    },
    {
      "epoch": 0.43495513369707356,
      "grad_norm": 1.6202558075303841,
      "learning_rate": 2.5126987436146794e-06,
      "loss": 0.7686,
      "step": 4823
    },
    {
      "epoch": 0.4350453172205438,
      "grad_norm": 1.7199499623095176,
      "learning_rate": 2.5121340329916675e-06,
      "loss": 0.7392,
      "step": 4824
    },
    {
      "epoch": 0.4351355007440141,
      "grad_norm": 1.5407080978051007,
      "learning_rate": 2.5115692786705566e-06,
      "loss": 0.6751,
      "step": 4825
    },
    {
      "epoch": 0.4352256842674843,
      "grad_norm": 1.8663535071130506,
      "learning_rate": 2.511004480699534e-06,
      "loss": 0.7931,
      "step": 4826
    },
    {
      "epoch": 0.4353158677909546,
      "grad_norm": 0.7169196416097047,
      "learning_rate": 2.510439639126791e-06,
      "loss": 0.6195,
      "step": 4827
    },
    {
      "epoch": 0.43540605131442484,
      "grad_norm": 1.47843410816501,
      "learning_rate": 2.509874754000524e-06,
      "loss": 0.7555,
      "step": 4828
    },
    {
      "epoch": 0.4354962348378951,
      "grad_norm": 2.354636480212811,
      "learning_rate": 2.509309825368932e-06,
      "loss": 0.7487,
      "step": 4829
    },
    {
      "epoch": 0.43558641836136536,
      "grad_norm": 1.6213979412688317,
      "learning_rate": 2.5087448532802173e-06,
      "loss": 0.759,
      "step": 4830
    },
    {
      "epoch": 0.43567660188483565,
      "grad_norm": 3.826184960616023,
      "learning_rate": 2.508179837782586e-06,
      "loss": 0.7397,
      "step": 4831
    },
    {
      "epoch": 0.4357667854083059,
      "grad_norm": 1.6666694958980706,
      "learning_rate": 2.5076147789242493e-06,
      "loss": 0.7318,
      "step": 4832
    },
    {
      "epoch": 0.43585696893177617,
      "grad_norm": 3.8739137665264827,
      "learning_rate": 2.5070496767534202e-06,
      "loss": 0.7736,
      "step": 4833
    },
    {
      "epoch": 0.4359471524552464,
      "grad_norm": 1.885435943785977,
      "learning_rate": 2.506484531318317e-06,
      "loss": 0.7878,
      "step": 4834
    },
    {
      "epoch": 0.4360373359787167,
      "grad_norm": 0.9036706193644962,
      "learning_rate": 2.5059193426671613e-06,
      "loss": 0.5918,
      "step": 4835
    },
    {
      "epoch": 0.436127519502187,
      "grad_norm": 2.2231664625091923,
      "learning_rate": 2.5053541108481772e-06,
      "loss": 0.8894,
      "step": 4836
    },
    {
      "epoch": 0.4362177030256572,
      "grad_norm": 2.1657951387230185,
      "learning_rate": 2.5047888359095935e-06,
      "loss": 0.7409,
      "step": 4837
    },
    {
      "epoch": 0.4363078865491275,
      "grad_norm": 1.5848011775421127,
      "learning_rate": 2.5042235178996436e-06,
      "loss": 0.7404,
      "step": 4838
    },
    {
      "epoch": 0.43639807007259773,
      "grad_norm": 1.8229085068293005,
      "learning_rate": 2.5036581568665627e-06,
      "loss": 0.8195,
      "step": 4839
    },
    {
      "epoch": 0.436488253596068,
      "grad_norm": 1.6527319918718786,
      "learning_rate": 2.503092752858591e-06,
      "loss": 0.8227,
      "step": 4840
    },
    {
      "epoch": 0.43657843711953825,
      "grad_norm": 0.8324760358528752,
      "learning_rate": 2.502527305923971e-06,
      "loss": 0.6079,
      "step": 4841
    },
    {
      "epoch": 0.43666862064300854,
      "grad_norm": 2.0034299288082864,
      "learning_rate": 2.5019618161109506e-06,
      "loss": 0.7514,
      "step": 4842
    },
    {
      "epoch": 0.4367588041664788,
      "grad_norm": 1.6836513146562955,
      "learning_rate": 2.5013962834677804e-06,
      "loss": 0.7742,
      "step": 4843
    },
    {
      "epoch": 0.43684898768994906,
      "grad_norm": 2.076210229758021,
      "learning_rate": 2.500830708042715e-06,
      "loss": 0.7461,
      "step": 4844
    },
    {
      "epoch": 0.4369391712134193,
      "grad_norm": 1.5433925384689124,
      "learning_rate": 2.500265089884011e-06,
      "loss": 0.8184,
      "step": 4845
    },
    {
      "epoch": 0.4370293547368896,
      "grad_norm": 1.8176923895048276,
      "learning_rate": 2.499699429039932e-06,
      "loss": 0.7593,
      "step": 4846
    },
    {
      "epoch": 0.4371195382603598,
      "grad_norm": 1.9073729412513492,
      "learning_rate": 2.4991337255587425e-06,
      "loss": 0.7081,
      "step": 4847
    },
    {
      "epoch": 0.4372097217838301,
      "grad_norm": 1.697389727621772,
      "learning_rate": 2.4985679794887106e-06,
      "loss": 0.6266,
      "step": 4848
    },
    {
      "epoch": 0.43729990530730034,
      "grad_norm": 1.7357449319164686,
      "learning_rate": 2.49800219087811e-06,
      "loss": 0.7874,
      "step": 4849
    },
    {
      "epoch": 0.43739008883077063,
      "grad_norm": 1.8487986607624718,
      "learning_rate": 2.4974363597752163e-06,
      "loss": 0.7916,
      "step": 4850
    },
    {
      "epoch": 0.43748027235424086,
      "grad_norm": 2.122293206914568,
      "learning_rate": 2.4968704862283097e-06,
      "loss": 0.8053,
      "step": 4851
    },
    {
      "epoch": 0.43757045587771115,
      "grad_norm": 1.8313000486649795,
      "learning_rate": 2.4963045702856737e-06,
      "loss": 0.8264,
      "step": 4852
    },
    {
      "epoch": 0.4376606394011814,
      "grad_norm": 2.719483375292302,
      "learning_rate": 2.4957386119955946e-06,
      "loss": 0.6993,
      "step": 4853
    },
    {
      "epoch": 0.4377508229246517,
      "grad_norm": 0.662600332436045,
      "learning_rate": 2.495172611406364e-06,
      "loss": 0.5688,
      "step": 4854
    },
    {
      "epoch": 0.4378410064481219,
      "grad_norm": 2.0805654322083202,
      "learning_rate": 2.4946065685662757e-06,
      "loss": 0.7853,
      "step": 4855
    },
    {
      "epoch": 0.4379311899715922,
      "grad_norm": 1.8537115831567927,
      "learning_rate": 2.4940404835236283e-06,
      "loss": 0.758,
      "step": 4856
    },
    {
      "epoch": 0.4380213734950624,
      "grad_norm": 1.6872191725285934,
      "learning_rate": 2.4934743563267223e-06,
      "loss": 0.7501,
      "step": 4857
    },
    {
      "epoch": 0.4381115570185327,
      "grad_norm": 3.486166044683052,
      "learning_rate": 2.4929081870238635e-06,
      "loss": 0.7216,
      "step": 4858
    },
    {
      "epoch": 0.43820174054200295,
      "grad_norm": 1.4965800398895805,
      "learning_rate": 2.49234197566336e-06,
      "loss": 0.7938,
      "step": 4859
    },
    {
      "epoch": 0.43829192406547324,
      "grad_norm": 1.571487290969711,
      "learning_rate": 2.4917757222935247e-06,
      "loss": 0.7774,
      "step": 4860
    },
    {
      "epoch": 0.4383821075889435,
      "grad_norm": 1.7624155091940519,
      "learning_rate": 2.4912094269626725e-06,
      "loss": 0.7773,
      "step": 4861
    },
    {
      "epoch": 0.43847229111241376,
      "grad_norm": 1.7998272071129127,
      "learning_rate": 2.4906430897191245e-06,
      "loss": 0.7949,
      "step": 4862
    },
    {
      "epoch": 0.43856247463588405,
      "grad_norm": 1.6540186264068657,
      "learning_rate": 2.490076710611202e-06,
      "loss": 0.7943,
      "step": 4863
    },
    {
      "epoch": 0.4386526581593543,
      "grad_norm": 3.7429192132285425,
      "learning_rate": 2.4895102896872326e-06,
      "loss": 0.8059,
      "step": 4864
    },
    {
      "epoch": 0.43874284168282457,
      "grad_norm": 1.757587808447614,
      "learning_rate": 2.4889438269955457e-06,
      "loss": 0.6879,
      "step": 4865
    },
    {
      "epoch": 0.4388330252062948,
      "grad_norm": 1.842895374397899,
      "learning_rate": 2.4883773225844755e-06,
      "loss": 0.7411,
      "step": 4866
    },
    {
      "epoch": 0.4389232087297651,
      "grad_norm": 1.7881880043020808,
      "learning_rate": 2.48781077650236e-06,
      "loss": 0.7489,
      "step": 4867
    },
    {
      "epoch": 0.4390133922532353,
      "grad_norm": 2.2100102397629695,
      "learning_rate": 2.4872441887975386e-06,
      "loss": 0.7193,
      "step": 4868
    },
    {
      "epoch": 0.4391035757767056,
      "grad_norm": 1.3123564868758106,
      "learning_rate": 2.486677559518356e-06,
      "loss": 0.6958,
      "step": 4869
    },
    {
      "epoch": 0.43919375930017585,
      "grad_norm": 1.7049329068807142,
      "learning_rate": 2.4861108887131614e-06,
      "loss": 0.7446,
      "step": 4870
    },
    {
      "epoch": 0.43928394282364613,
      "grad_norm": 1.442302559574482,
      "learning_rate": 2.485544176430305e-06,
      "loss": 0.7255,
      "step": 4871
    },
    {
      "epoch": 0.43937412634711637,
      "grad_norm": 2.404385128927054,
      "learning_rate": 2.4849774227181425e-06,
      "loss": 0.8366,
      "step": 4872
    },
    {
      "epoch": 0.43946430987058666,
      "grad_norm": 1.4153511540379122,
      "learning_rate": 2.484410627625032e-06,
      "loss": 0.7521,
      "step": 4873
    },
    {
      "epoch": 0.4395544933940569,
      "grad_norm": 1.889353490083021,
      "learning_rate": 2.4838437911993356e-06,
      "loss": 0.7955,
      "step": 4874
    },
    {
      "epoch": 0.4396446769175272,
      "grad_norm": 1.7682038774961948,
      "learning_rate": 2.483276913489419e-06,
      "loss": 0.7926,
      "step": 4875
    },
    {
      "epoch": 0.4397348604409974,
      "grad_norm": 2.0037715636084807,
      "learning_rate": 2.4827099945436516e-06,
      "loss": 0.656,
      "step": 4876
    },
    {
      "epoch": 0.4398250439644677,
      "grad_norm": 2.1969019881390937,
      "learning_rate": 2.482143034410405e-06,
      "loss": 0.8365,
      "step": 4877
    },
    {
      "epoch": 0.43991522748793793,
      "grad_norm": 1.8168805559333856,
      "learning_rate": 2.4815760331380573e-06,
      "loss": 0.7862,
      "step": 4878
    },
    {
      "epoch": 0.4400054110114082,
      "grad_norm": 1.4628367598717584,
      "learning_rate": 2.481008990774987e-06,
      "loss": 0.7163,
      "step": 4879
    },
    {
      "epoch": 0.44009559453487845,
      "grad_norm": 1.8740249641734272,
      "learning_rate": 2.480441907369577e-06,
      "loss": 0.7538,
      "step": 4880
    },
    {
      "epoch": 0.44018577805834874,
      "grad_norm": 2.062579182347521,
      "learning_rate": 2.479874782970214e-06,
      "loss": 0.7009,
      "step": 4881
    },
    {
      "epoch": 0.440275961581819,
      "grad_norm": 2.241239553364429,
      "learning_rate": 2.4793076176252887e-06,
      "loss": 0.7836,
      "step": 4882
    },
    {
      "epoch": 0.44036614510528926,
      "grad_norm": 1.5839286822630638,
      "learning_rate": 2.478740411383195e-06,
      "loss": 0.6392,
      "step": 4883
    },
    {
      "epoch": 0.44045632862875955,
      "grad_norm": 1.6281891852263923,
      "learning_rate": 2.4781731642923296e-06,
      "loss": 0.7504,
      "step": 4884
    },
    {
      "epoch": 0.4405465121522298,
      "grad_norm": 1.5113764727784609,
      "learning_rate": 2.477605876401093e-06,
      "loss": 0.7403,
      "step": 4885
    },
    {
      "epoch": 0.4406366956757001,
      "grad_norm": 1.542682785382435,
      "learning_rate": 2.4770385477578894e-06,
      "loss": 0.7064,
      "step": 4886
    },
    {
      "epoch": 0.4407268791991703,
      "grad_norm": 1.3659292480107779,
      "learning_rate": 2.476471178411127e-06,
      "loss": 0.713,
      "step": 4887
    },
    {
      "epoch": 0.4408170627226406,
      "grad_norm": 4.107317391235275,
      "learning_rate": 2.475903768409216e-06,
      "loss": 0.7794,
      "step": 4888
    },
    {
      "epoch": 0.44090724624611083,
      "grad_norm": 4.035934447059555,
      "learning_rate": 2.475336317800572e-06,
      "loss": 0.7295,
      "step": 4889
    },
    {
      "epoch": 0.4409974297695811,
      "grad_norm": 1.8601194542294013,
      "learning_rate": 2.4747688266336118e-06,
      "loss": 0.7438,
      "step": 4890
    },
    {
      "epoch": 0.44108761329305135,
      "grad_norm": 2.2246561020388897,
      "learning_rate": 2.4742012949567574e-06,
      "loss": 0.8557,
      "step": 4891
    },
    {
      "epoch": 0.44117779681652164,
      "grad_norm": 1.6920797579659521,
      "learning_rate": 2.4736337228184338e-06,
      "loss": 0.7329,
      "step": 4892
    },
    {
      "epoch": 0.44126798033999187,
      "grad_norm": 1.7179016794085586,
      "learning_rate": 2.4730661102670692e-06,
      "loss": 0.8549,
      "step": 4893
    },
    {
      "epoch": 0.44135816386346216,
      "grad_norm": 1.6444700687290725,
      "learning_rate": 2.472498457351096e-06,
      "loss": 0.7567,
      "step": 4894
    },
    {
      "epoch": 0.4414483473869324,
      "grad_norm": 1.9860744258038654,
      "learning_rate": 2.4719307641189495e-06,
      "loss": 0.7446,
      "step": 4895
    },
    {
      "epoch": 0.4415385309104027,
      "grad_norm": 2.1014592595604333,
      "learning_rate": 2.4713630306190673e-06,
      "loss": 0.7197,
      "step": 4896
    },
    {
      "epoch": 0.4416287144338729,
      "grad_norm": 2.212260036063366,
      "learning_rate": 2.4707952568998923e-06,
      "loss": 0.8372,
      "step": 4897
    },
    {
      "epoch": 0.4417188979573432,
      "grad_norm": 3.066204244192166,
      "learning_rate": 2.4702274430098703e-06,
      "loss": 0.7304,
      "step": 4898
    },
    {
      "epoch": 0.44180908148081344,
      "grad_norm": 2.408591976158932,
      "learning_rate": 2.4696595889974497e-06,
      "loss": 0.7108,
      "step": 4899
    },
    {
      "epoch": 0.4418992650042837,
      "grad_norm": 2.6795391786860527,
      "learning_rate": 2.469091694911084e-06,
      "loss": 0.8101,
      "step": 4900
    },
    {
      "epoch": 0.44198944852775396,
      "grad_norm": 2.1020565090478,
      "learning_rate": 2.4685237607992276e-06,
      "loss": 0.7517,
      "step": 4901
    },
    {
      "epoch": 0.44207963205122425,
      "grad_norm": 1.9575649607318857,
      "learning_rate": 2.4679557867103416e-06,
      "loss": 0.8189,
      "step": 4902
    },
    {
      "epoch": 0.4421698155746945,
      "grad_norm": 1.5202567518624122,
      "learning_rate": 2.4673877726928865e-06,
      "loss": 0.7615,
      "step": 4903
    },
    {
      "epoch": 0.44225999909816477,
      "grad_norm": 4.790100084824099,
      "learning_rate": 2.46681971879533e-06,
      "loss": 0.6552,
      "step": 4904
    },
    {
      "epoch": 0.442350182621635,
      "grad_norm": 1.8403706373222475,
      "learning_rate": 2.4662516250661407e-06,
      "loss": 0.6869,
      "step": 4905
    },
    {
      "epoch": 0.4424403661451053,
      "grad_norm": 2.22295690021613,
      "learning_rate": 2.465683491553792e-06,
      "loss": 0.727,
      "step": 4906
    },
    {
      "epoch": 0.4425305496685755,
      "grad_norm": 1.734882589759389,
      "learning_rate": 2.4651153183067604e-06,
      "loss": 0.7631,
      "step": 4907
    },
    {
      "epoch": 0.4426207331920458,
      "grad_norm": 2.17166949061427,
      "learning_rate": 2.4645471053735245e-06,
      "loss": 0.6953,
      "step": 4908
    },
    {
      "epoch": 0.4427109167155161,
      "grad_norm": 2.0447347385943733,
      "learning_rate": 2.4639788528025684e-06,
      "loss": 0.6945,
      "step": 4909
    },
    {
      "epoch": 0.44280110023898633,
      "grad_norm": 1.3711475242220426,
      "learning_rate": 2.463410560642378e-06,
      "loss": 0.7597,
      "step": 4910
    },
    {
      "epoch": 0.4428912837624566,
      "grad_norm": 0.6348510449962333,
      "learning_rate": 2.4628422289414448e-06,
      "loss": 0.5883,
      "step": 4911
    },
    {
      "epoch": 0.44298146728592686,
      "grad_norm": 0.670764337766765,
      "learning_rate": 2.4622738577482592e-06,
      "loss": 0.564,
      "step": 4912
    },
    {
      "epoch": 0.44307165080939714,
      "grad_norm": 2.107955002545151,
      "learning_rate": 2.461705447111319e-06,
      "loss": 0.7404,
      "step": 4913
    },
    {
      "epoch": 0.4431618343328674,
      "grad_norm": 1.612989865790906,
      "learning_rate": 2.4611369970791246e-06,
      "loss": 0.7274,
      "step": 4914
    },
    {
      "epoch": 0.44325201785633767,
      "grad_norm": 1.9713319468957238,
      "learning_rate": 2.460568507700179e-06,
      "loss": 0.7584,
      "step": 4915
    },
    {
      "epoch": 0.4433422013798079,
      "grad_norm": 2.1392531628341804,
      "learning_rate": 2.4599999790229887e-06,
      "loss": 0.798,
      "step": 4916
    },
    {
      "epoch": 0.4434323849032782,
      "grad_norm": 2.032795479517763,
      "learning_rate": 2.459431411096064e-06,
      "loss": 0.7227,
      "step": 4917
    },
    {
      "epoch": 0.4435225684267484,
      "grad_norm": 1.854759193069581,
      "learning_rate": 2.458862803967918e-06,
      "loss": 0.726,
      "step": 4918
    },
    {
      "epoch": 0.4436127519502187,
      "grad_norm": 3.9377303283172918,
      "learning_rate": 2.4582941576870667e-06,
      "loss": 0.7692,
      "step": 4919
    },
    {
      "epoch": 0.44370293547368894,
      "grad_norm": 2.5967233000697716,
      "learning_rate": 2.4577254723020315e-06,
      "loss": 0.7347,
      "step": 4920
    },
    {
      "epoch": 0.44379311899715923,
      "grad_norm": 1.559876948663164,
      "learning_rate": 2.457156747861335e-06,
      "loss": 0.768,
      "step": 4921
    },
    {
      "epoch": 0.44388330252062946,
      "grad_norm": 2.916406093038437,
      "learning_rate": 2.456587984413504e-06,
      "loss": 0.8214,
      "step": 4922
    },
    {
      "epoch": 0.44397348604409975,
      "grad_norm": 1.957611667960579,
      "learning_rate": 2.4560191820070683e-06,
      "loss": 0.6909,
      "step": 4923
    },
    {
      "epoch": 0.44406366956757,
      "grad_norm": 2.7033939117561765,
      "learning_rate": 2.4554503406905617e-06,
      "loss": 0.7403,
      "step": 4924
    },
    {
      "epoch": 0.4441538530910403,
      "grad_norm": 1.676036966371396,
      "learning_rate": 2.454881460512521e-06,
      "loss": 0.7987,
      "step": 4925
    },
    {
      "epoch": 0.4442440366145105,
      "grad_norm": 1.337950366837342,
      "learning_rate": 2.4543125415214856e-06,
      "loss": 0.728,
      "step": 4926
    },
    {
      "epoch": 0.4443342201379808,
      "grad_norm": 2.120878935837625,
      "learning_rate": 2.4537435837659996e-06,
      "loss": 0.7813,
      "step": 4927
    },
    {
      "epoch": 0.44442440366145103,
      "grad_norm": 2.0427908883797823,
      "learning_rate": 2.4531745872946085e-06,
      "loss": 0.7065,
      "step": 4928
    },
    {
      "epoch": 0.4445145871849213,
      "grad_norm": 2.08112480084846,
      "learning_rate": 2.4526055521558632e-06,
      "loss": 0.8173,
      "step": 4929
    },
    {
      "epoch": 0.44460477070839155,
      "grad_norm": 2.441476366180646,
      "learning_rate": 2.4520364783983164e-06,
      "loss": 0.7773,
      "step": 4930
    },
    {
      "epoch": 0.44469495423186184,
      "grad_norm": 1.5227687052365593,
      "learning_rate": 2.451467366070525e-06,
      "loss": 0.7389,
      "step": 4931
    },
    {
      "epoch": 0.4447851377553321,
      "grad_norm": 2.4860000573708367,
      "learning_rate": 2.450898215221048e-06,
      "loss": 0.8317,
      "step": 4932
    },
    {
      "epoch": 0.44487532127880236,
      "grad_norm": 2.2438416443707,
      "learning_rate": 2.4503290258984498e-06,
      "loss": 0.9011,
      "step": 4933
    },
    {
      "epoch": 0.44496550480227265,
      "grad_norm": 3.0152483294752557,
      "learning_rate": 2.4497597981512952e-06,
      "loss": 0.6166,
      "step": 4934
    },
    {
      "epoch": 0.4450556883257429,
      "grad_norm": 0.6462180519692389,
      "learning_rate": 2.4491905320281555e-06,
      "loss": 0.5919,
      "step": 4935
    },
    {
      "epoch": 0.44514587184921317,
      "grad_norm": 2.1107888463760642,
      "learning_rate": 2.448621227577602e-06,
      "loss": 0.7343,
      "step": 4936
    },
    {
      "epoch": 0.4452360553726834,
      "grad_norm": 2.764162528221138,
      "learning_rate": 2.4480518848482123e-06,
      "loss": 0.7013,
      "step": 4937
    },
    {
      "epoch": 0.4453262388961537,
      "grad_norm": 1.7101881223786963,
      "learning_rate": 2.447482503888565e-06,
      "loss": 0.7669,
      "step": 4938
    },
    {
      "epoch": 0.4454164224196239,
      "grad_norm": 2.0700442734189997,
      "learning_rate": 2.4469130847472434e-06,
      "loss": 0.7366,
      "step": 4939
    },
    {
      "epoch": 0.4455066059430942,
      "grad_norm": 1.755168299308039,
      "learning_rate": 2.4463436274728326e-06,
      "loss": 0.7946,
      "step": 4940
    },
    {
      "epoch": 0.44559678946656445,
      "grad_norm": 2.157526716329892,
      "learning_rate": 2.4457741321139227e-06,
      "loss": 0.6027,
      "step": 4941
    },
    {
      "epoch": 0.44568697299003474,
      "grad_norm": 1.6543344180408257,
      "learning_rate": 2.4452045987191063e-06,
      "loss": 0.7771,
      "step": 4942
    },
    {
      "epoch": 0.44577715651350497,
      "grad_norm": 1.5725565955040322,
      "learning_rate": 2.4446350273369776e-06,
      "loss": 0.7309,
      "step": 4943
    },
    {
      "epoch": 0.44586734003697526,
      "grad_norm": 1.5482008675260766,
      "learning_rate": 2.4440654180161374e-06,
      "loss": 0.7893,
      "step": 4944
    },
    {
      "epoch": 0.4459575235604455,
      "grad_norm": 1.5819757145682625,
      "learning_rate": 2.4434957708051875e-06,
      "loss": 0.7125,
      "step": 4945
    },
    {
      "epoch": 0.4460477070839158,
      "grad_norm": 1.81031200280521,
      "learning_rate": 2.4429260857527324e-06,
      "loss": 0.6942,
      "step": 4946
    },
    {
      "epoch": 0.446137890607386,
      "grad_norm": 1.9967620387833818,
      "learning_rate": 2.4423563629073815e-06,
      "loss": 0.6626,
      "step": 4947
    },
    {
      "epoch": 0.4462280741308563,
      "grad_norm": 1.4502840881918282,
      "learning_rate": 2.4417866023177466e-06,
      "loss": 0.7554,
      "step": 4948
    },
    {
      "epoch": 0.44631825765432653,
      "grad_norm": 3.196446597572882,
      "learning_rate": 2.441216804032443e-06,
      "loss": 0.7893,
      "step": 4949
    },
    {
      "epoch": 0.4464084411777968,
      "grad_norm": 1.6726275737786267,
      "learning_rate": 2.440646968100089e-06,
      "loss": 0.77,
      "step": 4950
    },
    {
      "epoch": 0.44649862470126706,
      "grad_norm": 0.6779135431484611,
      "learning_rate": 2.4400770945693055e-06,
      "loss": 0.5746,
      "step": 4951
    },
    {
      "epoch": 0.44658880822473734,
      "grad_norm": 2.074190588578543,
      "learning_rate": 2.4395071834887177e-06,
      "loss": 0.7921,
      "step": 4952
    },
    {
      "epoch": 0.4466789917482076,
      "grad_norm": 0.6985695878306072,
      "learning_rate": 2.438937234906954e-06,
      "loss": 0.6356,
      "step": 4953
    },
    {
      "epoch": 0.44676917527167787,
      "grad_norm": 1.8176374303184017,
      "learning_rate": 2.4383672488726447e-06,
      "loss": 0.7477,
      "step": 4954
    },
    {
      "epoch": 0.44685935879514815,
      "grad_norm": 2.104604908342881,
      "learning_rate": 2.4377972254344256e-06,
      "loss": 0.681,
      "step": 4955
    },
    {
      "epoch": 0.4469495423186184,
      "grad_norm": 1.8874274713589667,
      "learning_rate": 2.437227164640932e-06,
      "loss": 0.6411,
      "step": 4956
    },
    {
      "epoch": 0.4470397258420887,
      "grad_norm": 3.576903288717885,
      "learning_rate": 2.436657066540807e-06,
      "loss": 0.7019,
      "step": 4957
    },
    {
      "epoch": 0.4471299093655589,
      "grad_norm": 11.704205818332598,
      "learning_rate": 2.4360869311826927e-06,
      "loss": 0.6335,
      "step": 4958
    },
    {
      "epoch": 0.4472200928890292,
      "grad_norm": 3.6239898852781907,
      "learning_rate": 2.4355167586152367e-06,
      "loss": 0.849,
      "step": 4959
    },
    {
      "epoch": 0.44731027641249943,
      "grad_norm": 2.110827249828698,
      "learning_rate": 2.4349465488870896e-06,
      "loss": 0.7222,
      "step": 4960
    },
    {
      "epoch": 0.4474004599359697,
      "grad_norm": 2.3084659960619662,
      "learning_rate": 2.434376302046905e-06,
      "loss": 0.8412,
      "step": 4961
    },
    {
      "epoch": 0.44749064345943995,
      "grad_norm": 1.6662071627743038,
      "learning_rate": 2.433806018143339e-06,
      "loss": 0.7764,
      "step": 4962
    },
    {
      "epoch": 0.44758082698291024,
      "grad_norm": 0.6751617723454089,
      "learning_rate": 2.433235697225051e-06,
      "loss": 0.6162,
      "step": 4963
    },
    {
      "epoch": 0.4476710105063805,
      "grad_norm": 1.8256739299607168,
      "learning_rate": 2.4326653393407048e-06,
      "loss": 0.7741,
      "step": 4964
    },
    {
      "epoch": 0.44776119402985076,
      "grad_norm": 2.73404286138484,
      "learning_rate": 2.432094944538966e-06,
      "loss": 0.7195,
      "step": 4965
    },
    {
      "epoch": 0.447851377553321,
      "grad_norm": 1.6408070054597066,
      "learning_rate": 2.4315245128685047e-06,
      "loss": 0.7235,
      "step": 4966
    },
    {
      "epoch": 0.4479415610767913,
      "grad_norm": 1.8312146414510166,
      "learning_rate": 2.4309540443779925e-06,
      "loss": 0.755,
      "step": 4967
    },
    {
      "epoch": 0.4480317446002615,
      "grad_norm": 1.5488470588671674,
      "learning_rate": 2.4303835391161047e-06,
      "loss": 0.6902,
      "step": 4968
    },
    {
      "epoch": 0.4481219281237318,
      "grad_norm": 1.676674984645245,
      "learning_rate": 2.42981299713152e-06,
      "loss": 0.7064,
      "step": 4969
    },
    {
      "epoch": 0.44821211164720204,
      "grad_norm": 2.706274766346151,
      "learning_rate": 2.4292424184729204e-06,
      "loss": 0.79,
      "step": 4970
    },
    {
      "epoch": 0.4483022951706723,
      "grad_norm": 3.2089134946809046,
      "learning_rate": 2.4286718031889913e-06,
      "loss": 0.7457,
      "step": 4971
    },
    {
      "epoch": 0.44839247869414256,
      "grad_norm": 2.0493848715146252,
      "learning_rate": 2.4281011513284202e-06,
      "loss": 0.727,
      "step": 4972
    },
    {
      "epoch": 0.44848266221761285,
      "grad_norm": 2.70848346196295,
      "learning_rate": 2.4275304629398985e-06,
      "loss": 0.6909,
      "step": 4973
    },
    {
      "epoch": 0.4485728457410831,
      "grad_norm": 1.433694279859272,
      "learning_rate": 2.4269597380721194e-06,
      "loss": 0.8325,
      "step": 4974
    },
    {
      "epoch": 0.44866302926455337,
      "grad_norm": 2.0224566936218147,
      "learning_rate": 2.426388976773782e-06,
      "loss": 0.7535,
      "step": 4975
    },
    {
      "epoch": 0.4487532127880236,
      "grad_norm": 1.7317373478073317,
      "learning_rate": 2.425818179093586e-06,
      "loss": 0.8632,
      "step": 4976
    },
    {
      "epoch": 0.4488433963114939,
      "grad_norm": 1.5377676692041171,
      "learning_rate": 2.4252473450802346e-06,
      "loss": 0.6641,
      "step": 4977
    },
    {
      "epoch": 0.4489335798349641,
      "grad_norm": 4.670661896353538,
      "learning_rate": 2.4246764747824355e-06,
      "loss": 0.7907,
      "step": 4978
    },
    {
      "epoch": 0.4490237633584344,
      "grad_norm": 1.5932706224132538,
      "learning_rate": 2.424105568248897e-06,
      "loss": 0.7652,
      "step": 4979
    },
    {
      "epoch": 0.4491139468819047,
      "grad_norm": 1.816482575156326,
      "learning_rate": 2.4235346255283337e-06,
      "loss": 0.6939,
      "step": 4980
    },
    {
      "epoch": 0.44920413040537494,
      "grad_norm": 1.830805647126603,
      "learning_rate": 2.42296364666946e-06,
      "loss": 0.7648,
      "step": 4981
    },
    {
      "epoch": 0.4492943139288452,
      "grad_norm": 5.7308481789006915,
      "learning_rate": 2.4223926317209965e-06,
      "loss": 0.7047,
      "step": 4982
    },
    {
      "epoch": 0.44938449745231546,
      "grad_norm": 1.5986320668045757,
      "learning_rate": 2.4218215807316647e-06,
      "loss": 0.7549,
      "step": 4983
    },
    {
      "epoch": 0.44947468097578575,
      "grad_norm": 2.058227738157063,
      "learning_rate": 2.4212504937501894e-06,
      "loss": 0.8427,
      "step": 4984
    },
    {
      "epoch": 0.449564864499256,
      "grad_norm": 1.4677424119828892,
      "learning_rate": 2.4206793708253e-06,
      "loss": 0.8108,
      "step": 4985
    },
    {
      "epoch": 0.44965504802272627,
      "grad_norm": 1.9340324135463576,
      "learning_rate": 2.420108212005726e-06,
      "loss": 0.7118,
      "step": 4986
    },
    {
      "epoch": 0.4497452315461965,
      "grad_norm": 2.0114354795253706,
      "learning_rate": 2.4195370173402034e-06,
      "loss": 0.7491,
      "step": 4987
    },
    {
      "epoch": 0.4498354150696668,
      "grad_norm": 2.560543057390127,
      "learning_rate": 2.4189657868774696e-06,
      "loss": 0.8722,
      "step": 4988
    },
    {
      "epoch": 0.449925598593137,
      "grad_norm": 1.809497846211577,
      "learning_rate": 2.418394520666264e-06,
      "loss": 0.6918,
      "step": 4989
    },
    {
      "epoch": 0.4500157821166073,
      "grad_norm": 1.9767975799326516,
      "learning_rate": 2.4178232187553307e-06,
      "loss": 0.74,
      "step": 4990
    },
    {
      "epoch": 0.45010596564007754,
      "grad_norm": 1.5057593880198739,
      "learning_rate": 2.417251881193417e-06,
      "loss": 0.7792,
      "step": 4991
    },
    {
      "epoch": 0.45019614916354783,
      "grad_norm": 2.9146758960334607,
      "learning_rate": 2.4166805080292723e-06,
      "loss": 0.8099,
      "step": 4992
    },
    {
      "epoch": 0.45028633268701806,
      "grad_norm": 3.7861626608879537,
      "learning_rate": 2.4161090993116485e-06,
      "loss": 0.702,
      "step": 4993
    },
    {
      "epoch": 0.45037651621048835,
      "grad_norm": 1.9782324811472343,
      "learning_rate": 2.4155376550893026e-06,
      "loss": 0.6347,
      "step": 4994
    },
    {
      "epoch": 0.4504666997339586,
      "grad_norm": 2.6063053555178777,
      "learning_rate": 2.4149661754109926e-06,
      "loss": 0.7627,
      "step": 4995
    },
    {
      "epoch": 0.4505568832574289,
      "grad_norm": 2.034700602596316,
      "learning_rate": 2.41439466032548e-06,
      "loss": 0.8098,
      "step": 4996
    },
    {
      "epoch": 0.4506470667808991,
      "grad_norm": 1.7246208492947142,
      "learning_rate": 2.41382310988153e-06,
      "loss": 0.7033,
      "step": 4997
    },
    {
      "epoch": 0.4507372503043694,
      "grad_norm": 2.612269640535931,
      "learning_rate": 2.413251524127911e-06,
      "loss": 0.8149,
      "step": 4998
    },
    {
      "epoch": 0.45082743382783963,
      "grad_norm": 1.8607822510398448,
      "learning_rate": 2.412679903113393e-06,
      "loss": 0.7274,
      "step": 4999
    },
    {
      "epoch": 0.4509176173513099,
      "grad_norm": 2.140006539878478,
      "learning_rate": 2.4121082468867505e-06,
      "loss": 0.7715,
      "step": 5000
    },
    {
      "epoch": 0.45100780087478015,
      "grad_norm": 2.5658991528954505,
      "learning_rate": 2.4115365554967597e-06,
      "loss": 0.6946,
      "step": 5001
    },
    {
      "epoch": 0.45109798439825044,
      "grad_norm": 1.9334555836644647,
      "learning_rate": 2.4109648289922006e-06,
      "loss": 0.7569,
      "step": 5002
    },
    {
      "epoch": 0.45118816792172073,
      "grad_norm": 3.6148998697218757,
      "learning_rate": 2.4103930674218565e-06,
      "loss": 0.7599,
      "step": 5003
    },
    {
      "epoch": 0.45127835144519096,
      "grad_norm": 1.5230615396471168,
      "learning_rate": 2.409821270834513e-06,
      "loss": 0.7287,
      "step": 5004
    },
    {
      "epoch": 0.45136853496866125,
      "grad_norm": 2.5578560967012756,
      "learning_rate": 2.409249439278959e-06,
      "loss": 0.8098,
      "step": 5005
    },
    {
      "epoch": 0.4514587184921315,
      "grad_norm": 0.6296396185226986,
      "learning_rate": 2.408677572803986e-06,
      "loss": 0.553,
      "step": 5006
    },
    {
      "epoch": 0.45154890201560177,
      "grad_norm": 1.7052113763886527,
      "learning_rate": 2.408105671458389e-06,
      "loss": 0.7627,
      "step": 5007
    },
    {
      "epoch": 0.451639085539072,
      "grad_norm": 3.9267496764946177,
      "learning_rate": 2.4075337352909663e-06,
      "loss": 0.6963,
      "step": 5008
    },
    {
      "epoch": 0.4517292690625423,
      "grad_norm": 1.7257983557845022,
      "learning_rate": 2.4069617643505177e-06,
      "loss": 0.8815,
      "step": 5009
    },
    {
      "epoch": 0.4518194525860125,
      "grad_norm": 1.5163605468403338,
      "learning_rate": 2.406389758685848e-06,
      "loss": 0.8425,
      "step": 5010
    },
    {
      "epoch": 0.4519096361094828,
      "grad_norm": 1.565210585274158,
      "learning_rate": 2.405817718345763e-06,
      "loss": 0.7929,
      "step": 5011
    },
    {
      "epoch": 0.45199981963295305,
      "grad_norm": 1.9633487299696997,
      "learning_rate": 2.4052456433790726e-06,
      "loss": 0.7105,
      "step": 5012
    },
    {
      "epoch": 0.45209000315642334,
      "grad_norm": 1.6737625305071395,
      "learning_rate": 2.4046735338345897e-06,
      "loss": 0.8115,
      "step": 5013
    },
    {
      "epoch": 0.45218018667989357,
      "grad_norm": 1.9328583523022833,
      "learning_rate": 2.404101389761129e-06,
      "loss": 0.7692,
      "step": 5014
    },
    {
      "epoch": 0.45227037020336386,
      "grad_norm": 0.6595097824394968,
      "learning_rate": 2.4035292112075097e-06,
      "loss": 0.5635,
      "step": 5015
    },
    {
      "epoch": 0.4523605537268341,
      "grad_norm": 1.613960400257467,
      "learning_rate": 2.4029569982225534e-06,
      "loss": 0.8177,
      "step": 5016
    },
    {
      "epoch": 0.4524507372503044,
      "grad_norm": 0.704443754514807,
      "learning_rate": 2.402384750855084e-06,
      "loss": 0.6361,
      "step": 5017
    },
    {
      "epoch": 0.4525409207737746,
      "grad_norm": 1.5767748505385197,
      "learning_rate": 2.4018124691539286e-06,
      "loss": 0.7605,
      "step": 5018
    },
    {
      "epoch": 0.4526311042972449,
      "grad_norm": 2.216003652480107,
      "learning_rate": 2.4012401531679178e-06,
      "loss": 0.7522,
      "step": 5019
    },
    {
      "epoch": 0.45272128782071513,
      "grad_norm": 1.9290554231519739,
      "learning_rate": 2.4006678029458847e-06,
      "loss": 0.7566,
      "step": 5020
    },
    {
      "epoch": 0.4528114713441854,
      "grad_norm": 1.831255522840068,
      "learning_rate": 2.400095418536666e-06,
      "loss": 0.6645,
      "step": 5021
    },
    {
      "epoch": 0.45290165486765566,
      "grad_norm": 1.588479021014904,
      "learning_rate": 2.3995229999890996e-06,
      "loss": 0.7352,
      "step": 5022
    },
    {
      "epoch": 0.45299183839112594,
      "grad_norm": 5.26115268304592,
      "learning_rate": 2.398950547352028e-06,
      "loss": 0.6986,
      "step": 5023
    },
    {
      "epoch": 0.4530820219145962,
      "grad_norm": 3.4338507962925138,
      "learning_rate": 2.398378060674295e-06,
      "loss": 0.6793,
      "step": 5024
    },
    {
      "epoch": 0.45317220543806647,
      "grad_norm": 1.6013484904889794,
      "learning_rate": 2.39780554000475e-06,
      "loss": 0.758,
      "step": 5025
    },
    {
      "epoch": 0.4532623889615367,
      "grad_norm": 1.6085038280874322,
      "learning_rate": 2.3972329853922434e-06,
      "loss": 0.7875,
      "step": 5026
    },
    {
      "epoch": 0.453352572485007,
      "grad_norm": 0.6064687285148824,
      "learning_rate": 2.3966603968856278e-06,
      "loss": 0.571,
      "step": 5027
    },
    {
      "epoch": 0.4534427560084773,
      "grad_norm": 1.718095273246056,
      "learning_rate": 2.39608777453376e-06,
      "loss": 0.8286,
      "step": 5028
    },
    {
      "epoch": 0.4535329395319475,
      "grad_norm": 8.527284699749735,
      "learning_rate": 2.3955151183854993e-06,
      "loss": 0.7836,
      "step": 5029
    },
    {
      "epoch": 0.4536231230554178,
      "grad_norm": 2.7910366391244414,
      "learning_rate": 2.3949424284897073e-06,
      "loss": 0.715,
      "step": 5030
    },
    {
      "epoch": 0.45371330657888803,
      "grad_norm": 1.6897402128051016,
      "learning_rate": 2.39436970489525e-06,
      "loss": 0.6264,
      "step": 5031
    },
    {
      "epoch": 0.4538034901023583,
      "grad_norm": 1.6031066149871327,
      "learning_rate": 2.3937969476509955e-06,
      "loss": 0.7597,
      "step": 5032
    },
    {
      "epoch": 0.45389367362582855,
      "grad_norm": 2.67127894841504,
      "learning_rate": 2.393224156805813e-06,
      "loss": 0.8257,
      "step": 5033
    },
    {
      "epoch": 0.45398385714929884,
      "grad_norm": 3.056605836712976,
      "learning_rate": 2.392651332408578e-06,
      "loss": 0.7362,
      "step": 5034
    },
    {
      "epoch": 0.4540740406727691,
      "grad_norm": 4.222952776721396,
      "learning_rate": 2.3920784745081655e-06,
      "loss": 0.7442,
      "step": 5035
    },
    {
      "epoch": 0.45416422419623936,
      "grad_norm": 1.7589727345013038,
      "learning_rate": 2.391505583153456e-06,
      "loss": 0.7934,
      "step": 5036
    },
    {
      "epoch": 0.4542544077197096,
      "grad_norm": 3.698548269285117,
      "learning_rate": 2.3909326583933315e-06,
      "loss": 0.7612,
      "step": 5037
    },
    {
      "epoch": 0.4543445912431799,
      "grad_norm": 2.362008046594362,
      "learning_rate": 2.3903597002766777e-06,
      "loss": 0.7259,
      "step": 5038
    },
    {
      "epoch": 0.4544347747666501,
      "grad_norm": 2.317712733803904,
      "learning_rate": 2.389786708852381e-06,
      "loss": 0.7198,
      "step": 5039
    },
    {
      "epoch": 0.4545249582901204,
      "grad_norm": 0.9481279989960829,
      "learning_rate": 2.389213684169333e-06,
      "loss": 0.6964,
      "step": 5040
    },
    {
      "epoch": 0.45461514181359064,
      "grad_norm": 1.7504919586781456,
      "learning_rate": 2.388640626276428e-06,
      "loss": 0.7185,
      "step": 5041
    },
    {
      "epoch": 0.45470532533706093,
      "grad_norm": 0.682914679976113,
      "learning_rate": 2.388067535222561e-06,
      "loss": 0.5353,
      "step": 5042
    },
    {
      "epoch": 0.45479550886053116,
      "grad_norm": 1.8246990608937514,
      "learning_rate": 2.3874944110566332e-06,
      "loss": 0.7753,
      "step": 5043
    },
    {
      "epoch": 0.45488569238400145,
      "grad_norm": 0.7695288488670106,
      "learning_rate": 2.3869212538275447e-06,
      "loss": 0.6782,
      "step": 5044
    },
    {
      "epoch": 0.4549758759074717,
      "grad_norm": 3.887964526324666,
      "learning_rate": 2.386348063584202e-06,
      "loss": 0.7683,
      "step": 5045
    },
    {
      "epoch": 0.45506605943094197,
      "grad_norm": 1.843655729308395,
      "learning_rate": 2.385774840375511e-06,
      "loss": 0.8101,
      "step": 5046
    },
    {
      "epoch": 0.4551562429544122,
      "grad_norm": 1.5841619013447588,
      "learning_rate": 2.385201584250385e-06,
      "loss": 0.7113,
      "step": 5047
    },
    {
      "epoch": 0.4552464264778825,
      "grad_norm": 0.6324232811820859,
      "learning_rate": 2.3846282952577346e-06,
      "loss": 0.6128,
      "step": 5048
    },
    {
      "epoch": 0.4553366100013527,
      "grad_norm": 1.6118578970490982,
      "learning_rate": 2.3840549734464785e-06,
      "loss": 0.8437,
      "step": 5049
    },
    {
      "epoch": 0.455426793524823,
      "grad_norm": 5.7320169283077185,
      "learning_rate": 2.3834816188655336e-06,
      "loss": 0.7011,
      "step": 5050
    },
    {
      "epoch": 0.4555169770482933,
      "grad_norm": 1.7733536755081063,
      "learning_rate": 2.3829082315638224e-06,
      "loss": 0.7017,
      "step": 5051
    },
    {
      "epoch": 0.45560716057176354,
      "grad_norm": 4.876484375938716,
      "learning_rate": 2.3823348115902695e-06,
      "loss": 0.6154,
      "step": 5052
    },
    {
      "epoch": 0.4556973440952338,
      "grad_norm": 1.7657237067524598,
      "learning_rate": 2.3817613589938026e-06,
      "loss": 0.7752,
      "step": 5053
    },
    {
      "epoch": 0.45578752761870406,
      "grad_norm": 1.8408529506390414,
      "learning_rate": 2.3811878738233517e-06,
      "loss": 0.7911,
      "step": 5054
    },
    {
      "epoch": 0.45587771114217435,
      "grad_norm": 1.9319813180063647,
      "learning_rate": 2.380614356127849e-06,
      "loss": 0.7023,
      "step": 5055
    },
    {
      "epoch": 0.4559678946656446,
      "grad_norm": 1.8597035478384438,
      "learning_rate": 2.3800408059562318e-06,
      "loss": 0.75,
      "step": 5056
    },
    {
      "epoch": 0.45605807818911487,
      "grad_norm": 2.2070687349900227,
      "learning_rate": 2.3794672233574365e-06,
      "loss": 0.7676,
      "step": 5057
    },
    {
      "epoch": 0.4561482617125851,
      "grad_norm": 1.8795107307300274,
      "learning_rate": 2.3788936083804058e-06,
      "loss": 0.7663,
      "step": 5058
    },
    {
      "epoch": 0.4562384452360554,
      "grad_norm": 1.9147033864080263,
      "learning_rate": 2.378319961074083e-06,
      "loss": 0.7557,
      "step": 5059
    },
    {
      "epoch": 0.4563286287595256,
      "grad_norm": 26.919084029010673,
      "learning_rate": 2.377746281487415e-06,
      "loss": 0.7919,
      "step": 5060
    },
    {
      "epoch": 0.4564188122829959,
      "grad_norm": 1.5423732719111822,
      "learning_rate": 2.377172569669352e-06,
      "loss": 0.845,
      "step": 5061
    },
    {
      "epoch": 0.45650899580646614,
      "grad_norm": 1.9191743437392212,
      "learning_rate": 2.376598825668845e-06,
      "loss": 0.6119,
      "step": 5062
    },
    {
      "epoch": 0.45659917932993643,
      "grad_norm": 1.8387733796148453,
      "learning_rate": 2.3760250495348495e-06,
      "loss": 0.7444,
      "step": 5063
    },
    {
      "epoch": 0.45668936285340667,
      "grad_norm": 2.139672505997673,
      "learning_rate": 2.3754512413163236e-06,
      "loss": 0.8017,
      "step": 5064
    },
    {
      "epoch": 0.45677954637687695,
      "grad_norm": 1.8062842765116576,
      "learning_rate": 2.3748774010622285e-06,
      "loss": 0.7418,
      "step": 5065
    },
    {
      "epoch": 0.4568697299003472,
      "grad_norm": 1.6049643510798939,
      "learning_rate": 2.3743035288215254e-06,
      "loss": 0.7663,
      "step": 5066
    },
    {
      "epoch": 0.4569599134238175,
      "grad_norm": 1.7619412233073066,
      "learning_rate": 2.3737296246431815e-06,
      "loss": 0.7672,
      "step": 5067
    },
    {
      "epoch": 0.4570500969472877,
      "grad_norm": 1.6066745211909186,
      "learning_rate": 2.3731556885761656e-06,
      "loss": 0.7309,
      "step": 5068
    },
    {
      "epoch": 0.457140280470758,
      "grad_norm": 0.6470807830856908,
      "learning_rate": 2.372581720669449e-06,
      "loss": 0.5797,
      "step": 5069
    },
    {
      "epoch": 0.45723046399422823,
      "grad_norm": 2.7765820366768357,
      "learning_rate": 2.3720077209720046e-06,
      "loss": 0.7854,
      "step": 5070
    },
    {
      "epoch": 0.4573206475176985,
      "grad_norm": 0.6914095258904562,
      "learning_rate": 2.3714336895328112e-06,
      "loss": 0.5735,
      "step": 5071
    },
    {
      "epoch": 0.45741083104116875,
      "grad_norm": 1.9127683663165156,
      "learning_rate": 2.370859626400847e-06,
      "loss": 0.7089,
      "step": 5072
    },
    {
      "epoch": 0.45750101456463904,
      "grad_norm": 1.561785954398152,
      "learning_rate": 2.3702855316250943e-06,
      "loss": 0.733,
      "step": 5073
    },
    {
      "epoch": 0.45759119808810933,
      "grad_norm": 2.093472704388518,
      "learning_rate": 2.369711405254539e-06,
      "loss": 0.7588,
      "step": 5074
    },
    {
      "epoch": 0.45768138161157956,
      "grad_norm": 2.558549673851034,
      "learning_rate": 2.3691372473381673e-06,
      "loss": 0.8116,
      "step": 5075
    },
    {
      "epoch": 0.45777156513504985,
      "grad_norm": 1.700846696820141,
      "learning_rate": 2.3685630579249708e-06,
      "loss": 0.8158,
      "step": 5076
    },
    {
      "epoch": 0.4578617486585201,
      "grad_norm": 2.0349399795866265,
      "learning_rate": 2.367988837063942e-06,
      "loss": 0.7321,
      "step": 5077
    },
    {
      "epoch": 0.4579519321819904,
      "grad_norm": 1.7640560910266454,
      "learning_rate": 2.367414584804076e-06,
      "loss": 0.8333,
      "step": 5078
    },
    {
      "epoch": 0.4580421157054606,
      "grad_norm": 1.7081783115792626,
      "learning_rate": 2.366840301194372e-06,
      "loss": 0.7923,
      "step": 5079
    },
    {
      "epoch": 0.4581322992289309,
      "grad_norm": 1.8043228375486493,
      "learning_rate": 2.3662659862838308e-06,
      "loss": 0.7443,
      "step": 5080
    },
    {
      "epoch": 0.45822248275240113,
      "grad_norm": 1.7340361118010286,
      "learning_rate": 2.365691640121456e-06,
      "loss": 0.8139,
      "step": 5081
    },
    {
      "epoch": 0.4583126662758714,
      "grad_norm": 2.0934960510202085,
      "learning_rate": 2.365117262756254e-06,
      "loss": 0.6871,
      "step": 5082
    },
    {
      "epoch": 0.45840284979934165,
      "grad_norm": 1.909112188604585,
      "learning_rate": 2.3645428542372342e-06,
      "loss": 0.7478,
      "step": 5083
    },
    {
      "epoch": 0.45849303332281194,
      "grad_norm": 0.6157330868326834,
      "learning_rate": 2.3639684146134083e-06,
      "loss": 0.5316,
      "step": 5084
    },
    {
      "epoch": 0.45858321684628217,
      "grad_norm": 1.9973069179063108,
      "learning_rate": 2.3633939439337897e-06,
      "loss": 0.722,
      "step": 5085
    },
    {
      "epoch": 0.45867340036975246,
      "grad_norm": 1.5769259741758317,
      "learning_rate": 2.362819442247396e-06,
      "loss": 0.7259,
      "step": 5086
    },
    {
      "epoch": 0.4587635838932227,
      "grad_norm": 0.6270223799153452,
      "learning_rate": 2.3622449096032477e-06,
      "loss": 0.5419,
      "step": 5087
    },
    {
      "epoch": 0.458853767416693,
      "grad_norm": 1.729958280269787,
      "learning_rate": 2.361670346050366e-06,
      "loss": 0.7946,
      "step": 5088
    },
    {
      "epoch": 0.4589439509401632,
      "grad_norm": 4.439574710307108,
      "learning_rate": 2.3610957516377757e-06,
      "loss": 0.6951,
      "step": 5089
    },
    {
      "epoch": 0.4590341344636335,
      "grad_norm": 1.6944608878638918,
      "learning_rate": 2.3605211264145048e-06,
      "loss": 0.7569,
      "step": 5090
    },
    {
      "epoch": 0.45912431798710374,
      "grad_norm": 1.3940055561556979,
      "learning_rate": 2.3599464704295836e-06,
      "loss": 0.7233,
      "step": 5091
    },
    {
      "epoch": 0.459214501510574,
      "grad_norm": 1.8418876164434317,
      "learning_rate": 2.359371783732045e-06,
      "loss": 0.6465,
      "step": 5092
    },
    {
      "epoch": 0.45930468503404426,
      "grad_norm": 1.899417012268757,
      "learning_rate": 2.358797066370924e-06,
      "loss": 0.7395,
      "step": 5093
    },
    {
      "epoch": 0.45939486855751455,
      "grad_norm": 1.7242729921714781,
      "learning_rate": 2.3582223183952594e-06,
      "loss": 0.7524,
      "step": 5094
    },
    {
      "epoch": 0.4594850520809848,
      "grad_norm": 1.7437496281011589,
      "learning_rate": 2.357647539854091e-06,
      "loss": 0.7864,
      "step": 5095
    },
    {
      "epoch": 0.45957523560445507,
      "grad_norm": 1.8146686733892492,
      "learning_rate": 2.3570727307964624e-06,
      "loss": 0.7627,
      "step": 5096
    },
    {
      "epoch": 0.4596654191279253,
      "grad_norm": 1.6574381758953434,
      "learning_rate": 2.35649789127142e-06,
      "loss": 0.8059,
      "step": 5097
    },
    {
      "epoch": 0.4597556026513956,
      "grad_norm": 1.8272312946789606,
      "learning_rate": 2.3559230213280115e-06,
      "loss": 0.7262,
      "step": 5098
    },
    {
      "epoch": 0.4598457861748659,
      "grad_norm": 4.683758272235489,
      "learning_rate": 2.3553481210152886e-06,
      "loss": 0.6711,
      "step": 5099
    },
    {
      "epoch": 0.4599359696983361,
      "grad_norm": 0.8371202177127867,
      "learning_rate": 2.3547731903823043e-06,
      "loss": 0.5973,
      "step": 5100
    },
    {
      "epoch": 0.4600261532218064,
      "grad_norm": 4.567309835950835,
      "learning_rate": 2.3541982294781155e-06,
      "loss": 0.7304,
      "step": 5101
    },
    {
      "epoch": 0.46011633674527663,
      "grad_norm": 7.309680101567774,
      "learning_rate": 2.3536232383517804e-06,
      "loss": 0.7503,
      "step": 5102
    },
    {
      "epoch": 0.4602065202687469,
      "grad_norm": 1.4499400521086596,
      "learning_rate": 2.3530482170523602e-06,
      "loss": 0.7497,
      "step": 5103
    },
    {
      "epoch": 0.46029670379221715,
      "grad_norm": 1.7588377950589995,
      "learning_rate": 2.3524731656289206e-06,
      "loss": 0.8493,
      "step": 5104
    },
    {
      "epoch": 0.46038688731568744,
      "grad_norm": 1.5341179129282116,
      "learning_rate": 2.351898084130526e-06,
      "loss": 0.8462,
      "step": 5105
    },
    {
      "epoch": 0.4604770708391577,
      "grad_norm": 1.8204540148560986,
      "learning_rate": 2.351322972606247e-06,
      "loss": 0.8234,
      "step": 5106
    },
    {
      "epoch": 0.46056725436262796,
      "grad_norm": 12.403274448346922,
      "learning_rate": 2.350747831105155e-06,
      "loss": 0.9059,
      "step": 5107
    },
    {
      "epoch": 0.4606574378860982,
      "grad_norm": 2.2875370502077117,
      "learning_rate": 2.350172659676323e-06,
      "loss": 0.8358,
      "step": 5108
    },
    {
      "epoch": 0.4607476214095685,
      "grad_norm": 1.6955019234162931,
      "learning_rate": 2.3495974583688306e-06,
      "loss": 0.7926,
      "step": 5109
    },
    {
      "epoch": 0.4608378049330387,
      "grad_norm": 1.6119953045835809,
      "learning_rate": 2.3490222272317543e-06,
      "loss": 0.8263,
      "step": 5110
    },
    {
      "epoch": 0.460927988456509,
      "grad_norm": 2.4617224031964224,
      "learning_rate": 2.348446966314177e-06,
      "loss": 0.8011,
      "step": 5111
    },
    {
      "epoch": 0.46101817197997924,
      "grad_norm": 1.9440945408873718,
      "learning_rate": 2.3478716756651837e-06,
      "loss": 0.7282,
      "step": 5112
    },
    {
      "epoch": 0.46110835550344953,
      "grad_norm": 1.8252085710251118,
      "learning_rate": 2.347296355333861e-06,
      "loss": 0.7499,
      "step": 5113
    },
    {
      "epoch": 0.46119853902691976,
      "grad_norm": 2.293364768459688,
      "learning_rate": 2.3467210053692972e-06,
      "loss": 0.7655,
      "step": 5114
    },
    {
      "epoch": 0.46128872255039005,
      "grad_norm": 2.072010308589846,
      "learning_rate": 2.3461456258205866e-06,
      "loss": 0.7583,
      "step": 5115
    },
    {
      "epoch": 0.4613789060738603,
      "grad_norm": 1.8179966024864354,
      "learning_rate": 2.345570216736822e-06,
      "loss": 0.6993,
      "step": 5116
    },
    {
      "epoch": 0.4614690895973306,
      "grad_norm": 1.8171771638794405,
      "learning_rate": 2.3449947781671013e-06,
      "loss": 0.7679,
      "step": 5117
    },
    {
      "epoch": 0.4615592731208008,
      "grad_norm": 2.248245933064978,
      "learning_rate": 2.3444193101605237e-06,
      "loss": 0.7866,
      "step": 5118
    },
    {
      "epoch": 0.4616494566442711,
      "grad_norm": 1.7505059192146613,
      "learning_rate": 2.3438438127661913e-06,
      "loss": 0.7757,
      "step": 5119
    },
    {
      "epoch": 0.4617396401677413,
      "grad_norm": 1.647939308419275,
      "learning_rate": 2.3432682860332096e-06,
      "loss": 0.5848,
      "step": 5120
    },
    {
      "epoch": 0.4618298236912116,
      "grad_norm": 1.8456502916958728,
      "learning_rate": 2.342692730010684e-06,
      "loss": 0.7069,
      "step": 5121
    },
    {
      "epoch": 0.4619200072146819,
      "grad_norm": 2.633023035704221,
      "learning_rate": 2.342117144747726e-06,
      "loss": 0.77,
      "step": 5122
    },
    {
      "epoch": 0.46201019073815214,
      "grad_norm": 1.8919678010035423,
      "learning_rate": 2.3415415302934457e-06,
      "loss": 0.759,
      "step": 5123
    },
    {
      "epoch": 0.4621003742616224,
      "grad_norm": 1.6511447605553695,
      "learning_rate": 2.340965886696959e-06,
      "loss": 0.7855,
      "step": 5124
    },
    {
      "epoch": 0.46219055778509266,
      "grad_norm": 1.8268297611759996,
      "learning_rate": 2.340390214007384e-06,
      "loss": 0.7059,
      "step": 5125
    },
    {
      "epoch": 0.46228074130856295,
      "grad_norm": 1.9566141319804102,
      "learning_rate": 2.339814512273838e-06,
      "loss": 0.6408,
      "step": 5126
    },
    {
      "epoch": 0.4623709248320332,
      "grad_norm": 1.7368929344478499,
      "learning_rate": 2.3392387815454447e-06,
      "loss": 0.863,
      "step": 5127
    },
    {
      "epoch": 0.46246110835550347,
      "grad_norm": 1.9165909240794374,
      "learning_rate": 2.3386630218713273e-06,
      "loss": 0.7522,
      "step": 5128
    },
    {
      "epoch": 0.4625512918789737,
      "grad_norm": 1.7565389809478626,
      "learning_rate": 2.3380872333006135e-06,
      "loss": 0.7883,
      "step": 5129
    },
    {
      "epoch": 0.462641475402444,
      "grad_norm": 12.901712206387387,
      "learning_rate": 2.3375114158824335e-06,
      "loss": 0.8051,
      "step": 5130
    },
    {
      "epoch": 0.4627316589259142,
      "grad_norm": 1.7018974850192272,
      "learning_rate": 2.3369355696659184e-06,
      "loss": 0.7919,
      "step": 5131
    },
    {
      "epoch": 0.4628218424493845,
      "grad_norm": 1.8583922153033974,
      "learning_rate": 2.336359694700202e-06,
      "loss": 0.7443,
      "step": 5132
    },
    {
      "epoch": 0.46291202597285475,
      "grad_norm": 1.6603041639070268,
      "learning_rate": 2.335783791034422e-06,
      "loss": 0.7605,
      "step": 5133
    },
    {
      "epoch": 0.46300220949632503,
      "grad_norm": 1.805276287521924,
      "learning_rate": 2.3352078587177173e-06,
      "loss": 0.7412,
      "step": 5134
    },
    {
      "epoch": 0.46309239301979527,
      "grad_norm": 3.3434416726821974,
      "learning_rate": 2.33463189779923e-06,
      "loss": 0.7693,
      "step": 5135
    },
    {
      "epoch": 0.46318257654326556,
      "grad_norm": 1.5130357944013653,
      "learning_rate": 2.334055908328104e-06,
      "loss": 0.828,
      "step": 5136
    },
    {
      "epoch": 0.4632727600667358,
      "grad_norm": 1.560712249357972,
      "learning_rate": 2.3334798903534866e-06,
      "loss": 0.6723,
      "step": 5137
    },
    {
      "epoch": 0.4633629435902061,
      "grad_norm": 1.4486433259464961,
      "learning_rate": 2.3329038439245252e-06,
      "loss": 0.653,
      "step": 5138
    },
    {
      "epoch": 0.4634531271136763,
      "grad_norm": 2.291352464348347,
      "learning_rate": 2.3323277690903724e-06,
      "loss": 0.7573,
      "step": 5139
    },
    {
      "epoch": 0.4635433106371466,
      "grad_norm": 1.5612656867870918,
      "learning_rate": 2.3317516659001827e-06,
      "loss": 0.7314,
      "step": 5140
    },
    {
      "epoch": 0.46363349416061683,
      "grad_norm": 1.53947342552152,
      "learning_rate": 2.331175534403111e-06,
      "loss": 0.768,
      "step": 5141
    },
    {
      "epoch": 0.4637236776840871,
      "grad_norm": 2.3106398446862735,
      "learning_rate": 2.3305993746483167e-06,
      "loss": 0.8431,
      "step": 5142
    },
    {
      "epoch": 0.46381386120755735,
      "grad_norm": 2.3868751786784554,
      "learning_rate": 2.3300231866849606e-06,
      "loss": 0.7356,
      "step": 5143
    },
    {
      "epoch": 0.46390404473102764,
      "grad_norm": 1.9637959228818291,
      "learning_rate": 2.3294469705622067e-06,
      "loss": 0.797,
      "step": 5144
    },
    {
      "epoch": 0.4639942282544979,
      "grad_norm": 1.926138761503037,
      "learning_rate": 2.3288707263292203e-06,
      "loss": 0.7357,
      "step": 5145
    },
    {
      "epoch": 0.46408441177796816,
      "grad_norm": 1.391063706708694,
      "learning_rate": 2.3282944540351707e-06,
      "loss": 0.7772,
      "step": 5146
    },
    {
      "epoch": 0.46417459530143845,
      "grad_norm": 2.1350517395325475,
      "learning_rate": 2.327718153729228e-06,
      "loss": 0.7317,
      "step": 5147
    },
    {
      "epoch": 0.4642647788249087,
      "grad_norm": 1.6712036210831216,
      "learning_rate": 2.327141825460566e-06,
      "loss": 0.8032,
      "step": 5148
    },
    {
      "epoch": 0.464354962348379,
      "grad_norm": 0.6713492310201739,
      "learning_rate": 2.326565469278358e-06,
      "loss": 0.5852,
      "step": 5149
    },
    {
      "epoch": 0.4644451458718492,
      "grad_norm": 0.7133006549715077,
      "learning_rate": 2.3259890852317846e-06,
      "loss": 0.5944,
      "step": 5150
    },
    {
      "epoch": 0.4645353293953195,
      "grad_norm": 1.8299783294728744,
      "learning_rate": 2.3254126733700246e-06,
      "loss": 0.7586,
      "step": 5151
    },
    {
      "epoch": 0.46462551291878973,
      "grad_norm": 2.1459212670935086,
      "learning_rate": 2.324836233742262e-06,
      "loss": 0.7868,
      "step": 5152
    },
    {
      "epoch": 0.46471569644226,
      "grad_norm": 1.945126199035883,
      "learning_rate": 2.3242597663976793e-06,
      "loss": 0.7951,
      "step": 5153
    },
    {
      "epoch": 0.46480587996573025,
      "grad_norm": 1.8791618569841881,
      "learning_rate": 2.3236832713854663e-06,
      "loss": 0.7374,
      "step": 5154
    },
    {
      "epoch": 0.46489606348920054,
      "grad_norm": 1.6229263426166205,
      "learning_rate": 2.323106748754812e-06,
      "loss": 0.7534,
      "step": 5155
    },
    {
      "epoch": 0.4649862470126708,
      "grad_norm": 2.363326546731449,
      "learning_rate": 2.3225301985549077e-06,
      "loss": 0.8557,
      "step": 5156
    },
    {
      "epoch": 0.46507643053614106,
      "grad_norm": 2.8089163094937937,
      "learning_rate": 2.321953620834948e-06,
      "loss": 0.6922,
      "step": 5157
    },
    {
      "epoch": 0.4651666140596113,
      "grad_norm": 1.506155576160175,
      "learning_rate": 2.3213770156441314e-06,
      "loss": 0.7207,
      "step": 5158
    },
    {
      "epoch": 0.4652567975830816,
      "grad_norm": 2.5179021728879554,
      "learning_rate": 2.3208003830316554e-06,
      "loss": 0.8114,
      "step": 5159
    },
    {
      "epoch": 0.4653469811065518,
      "grad_norm": 1.852317109031026,
      "learning_rate": 2.3202237230467215e-06,
      "loss": 0.8474,
      "step": 5160
    },
    {
      "epoch": 0.4654371646300221,
      "grad_norm": 2.0193843825041724,
      "learning_rate": 2.3196470357385338e-06,
      "loss": 0.7161,
      "step": 5161
    },
    {
      "epoch": 0.46552734815349234,
      "grad_norm": 1.4973971513383013,
      "learning_rate": 2.319070321156299e-06,
      "loss": 0.7524,
      "step": 5162
    },
    {
      "epoch": 0.4656175316769626,
      "grad_norm": 1.9992407311218332,
      "learning_rate": 2.318493579349224e-06,
      "loss": 0.8882,
      "step": 5163
    },
    {
      "epoch": 0.46570771520043286,
      "grad_norm": 1.7698173375893929,
      "learning_rate": 2.317916810366522e-06,
      "loss": 0.7468,
      "step": 5164
    },
    {
      "epoch": 0.46579789872390315,
      "grad_norm": 1.827624586688293,
      "learning_rate": 2.317340014257404e-06,
      "loss": 0.7002,
      "step": 5165
    },
    {
      "epoch": 0.4658880822473734,
      "grad_norm": 1.5307832123191991,
      "learning_rate": 2.316763191071086e-06,
      "loss": 0.6851,
      "step": 5166
    },
    {
      "epoch": 0.46597826577084367,
      "grad_norm": 1.5915360125830502,
      "learning_rate": 2.316186340856787e-06,
      "loss": 0.7164,
      "step": 5167
    },
    {
      "epoch": 0.4660684492943139,
      "grad_norm": 2.5061528783991704,
      "learning_rate": 2.315609463663725e-06,
      "loss": 0.7142,
      "step": 5168
    },
    {
      "epoch": 0.4661586328177842,
      "grad_norm": 1.939900971765873,
      "learning_rate": 2.315032559541123e-06,
      "loss": 0.7286,
      "step": 5169
    },
    {
      "epoch": 0.4662488163412545,
      "grad_norm": 1.8663979620767481,
      "learning_rate": 2.314455628538207e-06,
      "loss": 0.7828,
      "step": 5170
    },
    {
      "epoch": 0.4663389998647247,
      "grad_norm": 1.9649466455413538,
      "learning_rate": 2.3138786707042023e-06,
      "loss": 0.7279,
      "step": 5171
    },
    {
      "epoch": 0.466429183388195,
      "grad_norm": 1.6074471639727352,
      "learning_rate": 2.3133016860883387e-06,
      "loss": 0.7642,
      "step": 5172
    },
    {
      "epoch": 0.46651936691166523,
      "grad_norm": 1.8597975819691877,
      "learning_rate": 2.3127246747398475e-06,
      "loss": 0.7362,
      "step": 5173
    },
    {
      "epoch": 0.4666095504351355,
      "grad_norm": 6.990854828918541,
      "learning_rate": 2.312147636707963e-06,
      "loss": 0.7554,
      "step": 5174
    },
    {
      "epoch": 0.46669973395860576,
      "grad_norm": 1.8986734000369665,
      "learning_rate": 2.3115705720419214e-06,
      "loss": 0.7446,
      "step": 5175
    },
    {
      "epoch": 0.46678991748207604,
      "grad_norm": 1.6436549228106065,
      "learning_rate": 2.31099348079096e-06,
      "loss": 0.7791,
      "step": 5176
    },
    {
      "epoch": 0.4668801010055463,
      "grad_norm": 2.51964215175012,
      "learning_rate": 2.31041636300432e-06,
      "loss": 0.8006,
      "step": 5177
    },
    {
      "epoch": 0.46697028452901657,
      "grad_norm": 1.9421401910398286,
      "learning_rate": 2.3098392187312445e-06,
      "loss": 0.838,
      "step": 5178
    },
    {
      "epoch": 0.4670604680524868,
      "grad_norm": 3.982347997529258,
      "learning_rate": 2.309262048020978e-06,
      "loss": 0.7343,
      "step": 5179
    },
    {
      "epoch": 0.4671506515759571,
      "grad_norm": 1.8631578960413029,
      "learning_rate": 2.308684850922769e-06,
      "loss": 0.7877,
      "step": 5180
    },
    {
      "epoch": 0.4672408350994273,
      "grad_norm": 1.8611526326471124,
      "learning_rate": 2.3081076274858664e-06,
      "loss": 0.7562,
      "step": 5181
    },
    {
      "epoch": 0.4673310186228976,
      "grad_norm": 1.562209827181892,
      "learning_rate": 2.307530377759522e-06,
      "loss": 0.7863,
      "step": 5182
    },
    {
      "epoch": 0.46742120214636784,
      "grad_norm": 2.128515477144597,
      "learning_rate": 2.30695310179299e-06,
      "loss": 0.6847,
      "step": 5183
    },
    {
      "epoch": 0.46751138566983813,
      "grad_norm": 1.5004186840656464,
      "learning_rate": 2.3063757996355267e-06,
      "loss": 0.7531,
      "step": 5184
    },
    {
      "epoch": 0.46760156919330836,
      "grad_norm": 2.3742308123960916,
      "learning_rate": 2.3057984713363903e-06,
      "loss": 0.7873,
      "step": 5185
    },
    {
      "epoch": 0.46769175271677865,
      "grad_norm": 0.677430868211191,
      "learning_rate": 2.3052211169448436e-06,
      "loss": 0.5571,
      "step": 5186
    },
    {
      "epoch": 0.4677819362402489,
      "grad_norm": 1.7155369637002933,
      "learning_rate": 2.3046437365101474e-06,
      "loss": 0.7027,
      "step": 5187
    },
    {
      "epoch": 0.4678721197637192,
      "grad_norm": 1.644151004859096,
      "learning_rate": 2.3040663300815673e-06,
      "loss": 0.7318,
      "step": 5188
    },
    {
      "epoch": 0.4679623032871894,
      "grad_norm": 0.7233234521598183,
      "learning_rate": 2.3034888977083723e-06,
      "loss": 0.5657,
      "step": 5189
    },
    {
      "epoch": 0.4680524868106597,
      "grad_norm": 4.036990078592381,
      "learning_rate": 2.30291143943983e-06,
      "loss": 0.6716,
      "step": 5190
    },
    {
      "epoch": 0.46814267033412993,
      "grad_norm": 1.6047322981096228,
      "learning_rate": 2.3023339553252145e-06,
      "loss": 0.7172,
      "step": 5191
    },
    {
      "epoch": 0.4682328538576002,
      "grad_norm": 2.8546488742063048,
      "learning_rate": 2.301756445413799e-06,
      "loss": 0.7924,
      "step": 5192
    },
    {
      "epoch": 0.4683230373810705,
      "grad_norm": 2.264853721088779,
      "learning_rate": 2.3011789097548585e-06,
      "loss": 0.759,
      "step": 5193
    },
    {
      "epoch": 0.46841322090454074,
      "grad_norm": 1.609930664725658,
      "learning_rate": 2.3006013483976738e-06,
      "loss": 0.7555,
      "step": 5194
    },
    {
      "epoch": 0.468503404428011,
      "grad_norm": 3.3592563342050106,
      "learning_rate": 2.300023761391524e-06,
      "loss": 0.7533,
      "step": 5195
    },
    {
      "epoch": 0.46859358795148126,
      "grad_norm": 2.94381003672938,
      "learning_rate": 2.299446148785693e-06,
      "loss": 0.8099,
      "step": 5196
    },
    {
      "epoch": 0.46868377147495155,
      "grad_norm": 1.9242431193027139,
      "learning_rate": 2.2988685106294654e-06,
      "loss": 0.7371,
      "step": 5197
    },
    {
      "epoch": 0.4687739549984218,
      "grad_norm": 2.3484051060679056,
      "learning_rate": 2.2982908469721284e-06,
      "loss": 0.6569,
      "step": 5198
    },
    {
      "epoch": 0.46886413852189207,
      "grad_norm": 2.1864757864880833,
      "learning_rate": 2.2977131578629714e-06,
      "loss": 0.8263,
      "step": 5199
    },
    {
      "epoch": 0.4689543220453623,
      "grad_norm": 1.9783643267398996,
      "learning_rate": 2.297135443351286e-06,
      "loss": 0.7799,
      "step": 5200
    },
    {
      "epoch": 0.4690445055688326,
      "grad_norm": 2.2707190761927905,
      "learning_rate": 2.296557703486367e-06,
      "loss": 0.7664,
      "step": 5201
    },
    {
      "epoch": 0.4691346890923028,
      "grad_norm": 2.2551050551613896,
      "learning_rate": 2.295979938317509e-06,
      "loss": 0.7819,
      "step": 5202
    },
    {
      "epoch": 0.4692248726157731,
      "grad_norm": 2.174090127159397,
      "learning_rate": 2.295402147894011e-06,
      "loss": 0.763,
      "step": 5203
    },
    {
      "epoch": 0.46931505613924335,
      "grad_norm": 1.3941411773774603,
      "learning_rate": 2.2948243322651723e-06,
      "loss": 0.681,
      "step": 5204
    },
    {
      "epoch": 0.46940523966271364,
      "grad_norm": 1.7247503791097591,
      "learning_rate": 2.2942464914802962e-06,
      "loss": 0.7609,
      "step": 5205
    },
    {
      "epoch": 0.46949542318618387,
      "grad_norm": 1.736346456046902,
      "learning_rate": 2.293668625588687e-06,
      "loss": 0.7325,
      "step": 5206
    },
    {
      "epoch": 0.46958560670965416,
      "grad_norm": 2.2302699225906055,
      "learning_rate": 2.293090734639651e-06,
      "loss": 0.8022,
      "step": 5207
    },
    {
      "epoch": 0.4696757902331244,
      "grad_norm": 1.5474421684786048,
      "learning_rate": 2.2925128186824983e-06,
      "loss": 0.8735,
      "step": 5208
    },
    {
      "epoch": 0.4697659737565947,
      "grad_norm": 2.31657483675883,
      "learning_rate": 2.2919348777665384e-06,
      "loss": 0.8627,
      "step": 5209
    },
    {
      "epoch": 0.4698561572800649,
      "grad_norm": 1.6496019634682841,
      "learning_rate": 2.2913569119410856e-06,
      "loss": 0.7623,
      "step": 5210
    },
    {
      "epoch": 0.4699463408035352,
      "grad_norm": 2.785631650181108,
      "learning_rate": 2.290778921255454e-06,
      "loss": 0.7828,
      "step": 5211
    },
    {
      "epoch": 0.47003652432700543,
      "grad_norm": 1.5052285462532553,
      "learning_rate": 2.2902009057589613e-06,
      "loss": 0.795,
      "step": 5212
    },
    {
      "epoch": 0.4701267078504757,
      "grad_norm": 1.4391385566750818,
      "learning_rate": 2.2896228655009276e-06,
      "loss": 0.7138,
      "step": 5213
    },
    {
      "epoch": 0.47021689137394596,
      "grad_norm": 2.5223834313489397,
      "learning_rate": 2.289044800530674e-06,
      "loss": 0.7281,
      "step": 5214
    },
    {
      "epoch": 0.47030707489741624,
      "grad_norm": 2.5441806325603085,
      "learning_rate": 2.2884667108975245e-06,
      "loss": 0.7231,
      "step": 5215
    },
    {
      "epoch": 0.4703972584208865,
      "grad_norm": 1.8255100295272724,
      "learning_rate": 2.287888596650804e-06,
      "loss": 0.7997,
      "step": 5216
    },
    {
      "epoch": 0.47048744194435677,
      "grad_norm": 1.8651605561513407,
      "learning_rate": 2.287310457839841e-06,
      "loss": 0.7618,
      "step": 5217
    },
    {
      "epoch": 0.47057762546782705,
      "grad_norm": 1.8330227054078325,
      "learning_rate": 2.286732294513966e-06,
      "loss": 0.8308,
      "step": 5218
    },
    {
      "epoch": 0.4706678089912973,
      "grad_norm": 1.8626170384033558,
      "learning_rate": 2.2861541067225106e-06,
      "loss": 0.7706,
      "step": 5219
    },
    {
      "epoch": 0.4707579925147676,
      "grad_norm": 14.596749432464087,
      "learning_rate": 2.2855758945148095e-06,
      "loss": 0.7411,
      "step": 5220
    },
    {
      "epoch": 0.4708481760382378,
      "grad_norm": 1.6547587176766305,
      "learning_rate": 2.2849976579401977e-06,
      "loss": 0.8156,
      "step": 5221
    },
    {
      "epoch": 0.4709383595617081,
      "grad_norm": 1.9094706984292245,
      "learning_rate": 2.284419397048014e-06,
      "loss": 0.7541,
      "step": 5222
    },
    {
      "epoch": 0.47102854308517833,
      "grad_norm": 1.6588841489979802,
      "learning_rate": 2.2838411118875997e-06,
      "loss": 0.7628,
      "step": 5223
    },
    {
      "epoch": 0.4711187266086486,
      "grad_norm": 1.420286863660102,
      "learning_rate": 2.283262802508296e-06,
      "loss": 0.68,
      "step": 5224
    },
    {
      "epoch": 0.47120891013211885,
      "grad_norm": 1.3769575403020966,
      "learning_rate": 2.2826844689594492e-06,
      "loss": 0.7338,
      "step": 5225
    },
    {
      "epoch": 0.47129909365558914,
      "grad_norm": 1.6460877797254316,
      "learning_rate": 2.282106111290404e-06,
      "loss": 0.8369,
      "step": 5226
    },
    {
      "epoch": 0.4713892771790594,
      "grad_norm": 0.6912644451004795,
      "learning_rate": 2.2815277295505098e-06,
      "loss": 0.6011,
      "step": 5227
    },
    {
      "epoch": 0.47147946070252966,
      "grad_norm": 1.7454210730841808,
      "learning_rate": 2.2809493237891174e-06,
      "loss": 0.7078,
      "step": 5228
    },
    {
      "epoch": 0.4715696442259999,
      "grad_norm": 1.4779956462625639,
      "learning_rate": 2.2803708940555796e-06,
      "loss": 0.7903,
      "step": 5229
    },
    {
      "epoch": 0.4716598277494702,
      "grad_norm": 1.9228861560102382,
      "learning_rate": 2.2797924403992514e-06,
      "loss": 0.8252,
      "step": 5230
    },
    {
      "epoch": 0.4717500112729404,
      "grad_norm": 1.852143079940415,
      "learning_rate": 2.2792139628694892e-06,
      "loss": 0.7771,
      "step": 5231
    },
    {
      "epoch": 0.4718401947964107,
      "grad_norm": 8.779559098211275,
      "learning_rate": 2.2786354615156524e-06,
      "loss": 0.7186,
      "step": 5232
    },
    {
      "epoch": 0.47193037831988094,
      "grad_norm": 1.9056982899225536,
      "learning_rate": 2.2780569363871016e-06,
      "loss": 0.8282,
      "step": 5233
    },
    {
      "epoch": 0.4720205618433512,
      "grad_norm": 0.6161290520913814,
      "learning_rate": 2.277478387533199e-06,
      "loss": 0.5734,
      "step": 5234
    },
    {
      "epoch": 0.47211074536682146,
      "grad_norm": 2.932042185241815,
      "learning_rate": 2.276899815003311e-06,
      "loss": 0.7969,
      "step": 5235
    },
    {
      "epoch": 0.47220092889029175,
      "grad_norm": 1.8285004971894445,
      "learning_rate": 2.2763212188468045e-06,
      "loss": 0.7529,
      "step": 5236
    },
    {
      "epoch": 0.472291112413762,
      "grad_norm": 1.2781977700407547,
      "learning_rate": 2.2757425991130473e-06,
      "loss": 0.7726,
      "step": 5237
    },
    {
      "epoch": 0.47238129593723227,
      "grad_norm": 0.7746012246246073,
      "learning_rate": 2.2751639558514117e-06,
      "loss": 0.5859,
      "step": 5238
    },
    {
      "epoch": 0.4724714794607025,
      "grad_norm": 1.3927085355284416,
      "learning_rate": 2.2745852891112697e-06,
      "loss": 0.7991,
      "step": 5239
    },
    {
      "epoch": 0.4725616629841728,
      "grad_norm": 1.798299805037246,
      "learning_rate": 2.274006598941997e-06,
      "loss": 0.7656,
      "step": 5240
    },
    {
      "epoch": 0.4726518465076431,
      "grad_norm": 1.5771305236743864,
      "learning_rate": 2.27342788539297e-06,
      "loss": 0.7402,
      "step": 5241
    },
    {
      "epoch": 0.4727420300311133,
      "grad_norm": 1.5372967345311055,
      "learning_rate": 2.2728491485135684e-06,
      "loss": 0.7156,
      "step": 5242
    },
    {
      "epoch": 0.4728322135545836,
      "grad_norm": 2.004543745873328,
      "learning_rate": 2.272270388353173e-06,
      "loss": 0.8148,
      "step": 5243
    },
    {
      "epoch": 0.47292239707805384,
      "grad_norm": 1.8590449192474574,
      "learning_rate": 2.2716916049611666e-06,
      "loss": 0.7344,
      "step": 5244
    },
    {
      "epoch": 0.4730125806015241,
      "grad_norm": 2.198052593445445,
      "learning_rate": 2.2711127983869346e-06,
      "loss": 0.7392,
      "step": 5245
    },
    {
      "epoch": 0.47310276412499436,
      "grad_norm": 1.4909679762012795,
      "learning_rate": 2.270533968679864e-06,
      "loss": 0.7746,
      "step": 5246
    },
    {
      "epoch": 0.47319294764846465,
      "grad_norm": 1.5177751999863094,
      "learning_rate": 2.269955115889343e-06,
      "loss": 0.7655,
      "step": 5247
    },
    {
      "epoch": 0.4732831311719349,
      "grad_norm": 1.6178857153008857,
      "learning_rate": 2.269376240064763e-06,
      "loss": 0.7243,
      "step": 5248
    },
    {
      "epoch": 0.47337331469540517,
      "grad_norm": 1.6051219556351966,
      "learning_rate": 2.268797341255517e-06,
      "loss": 0.7789,
      "step": 5249
    },
    {
      "epoch": 0.4734634982188754,
      "grad_norm": 1.8883847922810282,
      "learning_rate": 2.268218419511e-06,
      "loss": 0.6752,
      "step": 5250
    },
    {
      "epoch": 0.4735536817423457,
      "grad_norm": 3.83119632751805,
      "learning_rate": 2.267639474880608e-06,
      "loss": 0.6814,
      "step": 5251
    },
    {
      "epoch": 0.4736438652658159,
      "grad_norm": 1.7112306665926453,
      "learning_rate": 2.2670605074137407e-06,
      "loss": 0.7366,
      "step": 5252
    },
    {
      "epoch": 0.4737340487892862,
      "grad_norm": 1.5788514382098708,
      "learning_rate": 2.2664815171597983e-06,
      "loss": 0.7428,
      "step": 5253
    },
    {
      "epoch": 0.47382423231275644,
      "grad_norm": 1.684442293029581,
      "learning_rate": 2.265902504168183e-06,
      "loss": 0.7069,
      "step": 5254
    },
    {
      "epoch": 0.47391441583622673,
      "grad_norm": 1.7581974540505425,
      "learning_rate": 2.2653234684883007e-06,
      "loss": 0.718,
      "step": 5255
    },
    {
      "epoch": 0.47400459935969697,
      "grad_norm": 2.0644564741354694,
      "learning_rate": 2.264744410169556e-06,
      "loss": 0.6741,
      "step": 5256
    },
    {
      "epoch": 0.47409478288316725,
      "grad_norm": 1.787109041474525,
      "learning_rate": 2.264165329261359e-06,
      "loss": 0.7366,
      "step": 5257
    },
    {
      "epoch": 0.4741849664066375,
      "grad_norm": 0.5790731799421794,
      "learning_rate": 2.26358622581312e-06,
      "loss": 0.5756,
      "step": 5258
    },
    {
      "epoch": 0.4742751499301078,
      "grad_norm": 1.4616951023556062,
      "learning_rate": 2.2630070998742504e-06,
      "loss": 0.8258,
      "step": 5259
    },
    {
      "epoch": 0.474365333453578,
      "grad_norm": 1.3539962905906229,
      "learning_rate": 2.262427951494165e-06,
      "loss": 0.7839,
      "step": 5260
    },
    {
      "epoch": 0.4744555169770483,
      "grad_norm": 0.6578086553900719,
      "learning_rate": 2.2618487807222794e-06,
      "loss": 0.6162,
      "step": 5261
    },
    {
      "epoch": 0.47454570050051853,
      "grad_norm": 1.6101616723766579,
      "learning_rate": 2.261269587608012e-06,
      "loss": 0.8364,
      "step": 5262
    },
    {
      "epoch": 0.4746358840239888,
      "grad_norm": 2.3561925582562284,
      "learning_rate": 2.260690372200783e-06,
      "loss": 0.7602,
      "step": 5263
    },
    {
      "epoch": 0.47472606754745905,
      "grad_norm": 1.4402768225295257,
      "learning_rate": 2.2601111345500138e-06,
      "loss": 0.8479,
      "step": 5264
    },
    {
      "epoch": 0.47481625107092934,
      "grad_norm": 1.4860129705250642,
      "learning_rate": 2.2595318747051286e-06,
      "loss": 0.7524,
      "step": 5265
    },
    {
      "epoch": 0.47490643459439963,
      "grad_norm": 1.7740472761948702,
      "learning_rate": 2.258952592715553e-06,
      "loss": 0.7865,
      "step": 5266
    },
    {
      "epoch": 0.47499661811786986,
      "grad_norm": 0.7048476415724183,
      "learning_rate": 2.2583732886307142e-06,
      "loss": 0.6175,
      "step": 5267
    },
    {
      "epoch": 0.47508680164134015,
      "grad_norm": 1.426705758296014,
      "learning_rate": 2.2577939625000414e-06,
      "loss": 0.7082,
      "step": 5268
    },
    {
      "epoch": 0.4751769851648104,
      "grad_norm": 0.7084533131655693,
      "learning_rate": 2.257214614372967e-06,
      "loss": 0.5775,
      "step": 5269
    },
    {
      "epoch": 0.4752671686882807,
      "grad_norm": 1.604822404608261,
      "learning_rate": 2.2566352442989227e-06,
      "loss": 0.7833,
      "step": 5270
    },
    {
      "epoch": 0.4753573522117509,
      "grad_norm": 3.533021735600917,
      "learning_rate": 2.256055852327344e-06,
      "loss": 0.7641,
      "step": 5271
    },
    {
      "epoch": 0.4754475357352212,
      "grad_norm": 1.533116429301463,
      "learning_rate": 2.2554764385076685e-06,
      "loss": 0.7097,
      "step": 5272
    },
    {
      "epoch": 0.4755377192586914,
      "grad_norm": 1.7140291055992336,
      "learning_rate": 2.2548970028893348e-06,
      "loss": 0.7728,
      "step": 5273
    },
    {
      "epoch": 0.4756279027821617,
      "grad_norm": 1.7983758380570027,
      "learning_rate": 2.254317545521783e-06,
      "loss": 0.7731,
      "step": 5274
    },
    {
      "epoch": 0.47571808630563195,
      "grad_norm": 1.7798377344557093,
      "learning_rate": 2.253738066454457e-06,
      "loss": 0.7954,
      "step": 5275
    },
    {
      "epoch": 0.47580826982910224,
      "grad_norm": 1.6179414917230592,
      "learning_rate": 2.2531585657367986e-06,
      "loss": 0.737,
      "step": 5276
    },
    {
      "epoch": 0.47589845335257247,
      "grad_norm": 1.5988034184697768,
      "learning_rate": 2.252579043418256e-06,
      "loss": 0.7128,
      "step": 5277
    },
    {
      "epoch": 0.47598863687604276,
      "grad_norm": 1.5727250275067146,
      "learning_rate": 2.251999499548277e-06,
      "loss": 0.7373,
      "step": 5278
    },
    {
      "epoch": 0.476078820399513,
      "grad_norm": 2.6483903278662213,
      "learning_rate": 2.251419934176311e-06,
      "loss": 0.7194,
      "step": 5279
    },
    {
      "epoch": 0.4761690039229833,
      "grad_norm": 1.4285846062461156,
      "learning_rate": 2.25084034735181e-06,
      "loss": 0.8117,
      "step": 5280
    },
    {
      "epoch": 0.4762591874464535,
      "grad_norm": 1.4858063071305627,
      "learning_rate": 2.2502607391242274e-06,
      "loss": 0.8086,
      "step": 5281
    },
    {
      "epoch": 0.4763493709699238,
      "grad_norm": 0.6663599550505543,
      "learning_rate": 2.2496811095430182e-06,
      "loss": 0.5831,
      "step": 5282
    },
    {
      "epoch": 0.47643955449339404,
      "grad_norm": 2.5311970646350495,
      "learning_rate": 2.249101458657641e-06,
      "loss": 0.757,
      "step": 5283
    },
    {
      "epoch": 0.4765297380168643,
      "grad_norm": 3.5909612035097656,
      "learning_rate": 2.2485217865175526e-06,
      "loss": 0.8282,
      "step": 5284
    },
    {
      "epoch": 0.47661992154033456,
      "grad_norm": 3.0297663882501906,
      "learning_rate": 2.2479420931722156e-06,
      "loss": 0.8977,
      "step": 5285
    },
    {
      "epoch": 0.47671010506380485,
      "grad_norm": 0.5860969326550182,
      "learning_rate": 2.2473623786710923e-06,
      "loss": 0.6191,
      "step": 5286
    },
    {
      "epoch": 0.4768002885872751,
      "grad_norm": 1.9316683961050318,
      "learning_rate": 2.2467826430636465e-06,
      "loss": 0.6737,
      "step": 5287
    },
    {
      "epoch": 0.47689047211074537,
      "grad_norm": 1.630140756134697,
      "learning_rate": 2.246202886399345e-06,
      "loss": 0.7827,
      "step": 5288
    },
    {
      "epoch": 0.47698065563421566,
      "grad_norm": 2.0048659259365564,
      "learning_rate": 2.2456231087276556e-06,
      "loss": 0.7579,
      "step": 5289
    },
    {
      "epoch": 0.4770708391576859,
      "grad_norm": 1.5242317672322239,
      "learning_rate": 2.245043310098048e-06,
      "loss": 0.6863,
      "step": 5290
    },
    {
      "epoch": 0.4771610226811562,
      "grad_norm": 1.785189037731411,
      "learning_rate": 2.244463490559995e-06,
      "loss": 0.8028,
      "step": 5291
    },
    {
      "epoch": 0.4772512062046264,
      "grad_norm": 1.7029549750056114,
      "learning_rate": 2.2438836501629683e-06,
      "loss": 0.7749,
      "step": 5292
    },
    {
      "epoch": 0.4773413897280967,
      "grad_norm": 1.4565883914567166,
      "learning_rate": 2.2433037889564437e-06,
      "loss": 0.7661,
      "step": 5293
    },
    {
      "epoch": 0.47743157325156693,
      "grad_norm": 2.1005811704530983,
      "learning_rate": 2.242723906989899e-06,
      "loss": 0.7015,
      "step": 5294
    },
    {
      "epoch": 0.4775217567750372,
      "grad_norm": 1.3146733047157853,
      "learning_rate": 2.2421440043128114e-06,
      "loss": 0.6912,
      "step": 5295
    },
    {
      "epoch": 0.47761194029850745,
      "grad_norm": 2.411566965994482,
      "learning_rate": 2.241564080974662e-06,
      "loss": 0.6677,
      "step": 5296
    },
    {
      "epoch": 0.47770212382197774,
      "grad_norm": 2.111362227624098,
      "learning_rate": 2.2409841370249343e-06,
      "loss": 0.7177,
      "step": 5297
    },
    {
      "epoch": 0.477792307345448,
      "grad_norm": 1.7653584236847506,
      "learning_rate": 2.2404041725131106e-06,
      "loss": 0.848,
      "step": 5298
    },
    {
      "epoch": 0.47788249086891826,
      "grad_norm": 2.173995485401116,
      "learning_rate": 2.239824187488677e-06,
      "loss": 0.7404,
      "step": 5299
    },
    {
      "epoch": 0.4779726743923885,
      "grad_norm": 0.7344004038211236,
      "learning_rate": 2.239244182001122e-06,
      "loss": 0.6189,
      "step": 5300
    },
    {
      "epoch": 0.4780628579158588,
      "grad_norm": 1.5307621081289378,
      "learning_rate": 2.2386641560999336e-06,
      "loss": 0.776,
      "step": 5301
    },
    {
      "epoch": 0.478153041439329,
      "grad_norm": 1.8829789464039974,
      "learning_rate": 2.238084109834604e-06,
      "loss": 0.7918,
      "step": 5302
    },
    {
      "epoch": 0.4782432249627993,
      "grad_norm": 3.1384506669870293,
      "learning_rate": 2.237504043254625e-06,
      "loss": 0.7663,
      "step": 5303
    },
    {
      "epoch": 0.47833340848626954,
      "grad_norm": 1.4636986083823473,
      "learning_rate": 2.2369239564094915e-06,
      "loss": 0.7311,
      "step": 5304
    },
    {
      "epoch": 0.47842359200973983,
      "grad_norm": 12.43692950756624,
      "learning_rate": 2.2363438493486995e-06,
      "loss": 0.6863,
      "step": 5305
    },
    {
      "epoch": 0.47851377553321006,
      "grad_norm": 2.07524299118022,
      "learning_rate": 2.235763722121747e-06,
      "loss": 0.6427,
      "step": 5306
    },
    {
      "epoch": 0.47860395905668035,
      "grad_norm": 1.5396320818804543,
      "learning_rate": 2.2351835747781346e-06,
      "loss": 0.7764,
      "step": 5307
    },
    {
      "epoch": 0.4786941425801506,
      "grad_norm": 1.5153844583504568,
      "learning_rate": 2.234603407367362e-06,
      "loss": 0.7975,
      "step": 5308
    },
    {
      "epoch": 0.47878432610362087,
      "grad_norm": 1.5970012189398253,
      "learning_rate": 2.2340232199389337e-06,
      "loss": 0.7746,
      "step": 5309
    },
    {
      "epoch": 0.4788745096270911,
      "grad_norm": 1.686009631942803,
      "learning_rate": 2.2334430125423538e-06,
      "loss": 0.6422,
      "step": 5310
    },
    {
      "epoch": 0.4789646931505614,
      "grad_norm": 1.7909941926794932,
      "learning_rate": 2.232862785227128e-06,
      "loss": 0.8056,
      "step": 5311
    },
    {
      "epoch": 0.4790548766740317,
      "grad_norm": 1.71883406000052,
      "learning_rate": 2.232282538042766e-06,
      "loss": 0.7926,
      "step": 5312
    },
    {
      "epoch": 0.4791450601975019,
      "grad_norm": 1.7153707402718545,
      "learning_rate": 2.231702271038777e-06,
      "loss": 0.8058,
      "step": 5313
    },
    {
      "epoch": 0.4792352437209722,
      "grad_norm": 1.7694877299163345,
      "learning_rate": 2.231121984264673e-06,
      "loss": 0.7373,
      "step": 5314
    },
    {
      "epoch": 0.47932542724444244,
      "grad_norm": 2.3723001694609933,
      "learning_rate": 2.2305416777699665e-06,
      "loss": 0.7931,
      "step": 5315
    },
    {
      "epoch": 0.4794156107679127,
      "grad_norm": 1.717543577393768,
      "learning_rate": 2.229961351604173e-06,
      "loss": 0.7655,
      "step": 5316
    },
    {
      "epoch": 0.47950579429138296,
      "grad_norm": 1.697517543134644,
      "learning_rate": 2.2293810058168085e-06,
      "loss": 0.7193,
      "step": 5317
    },
    {
      "epoch": 0.47959597781485325,
      "grad_norm": 1.549307249987717,
      "learning_rate": 2.2288006404573922e-06,
      "loss": 0.7841,
      "step": 5318
    },
    {
      "epoch": 0.4796861613383235,
      "grad_norm": 1.7790095644817592,
      "learning_rate": 2.228220255575444e-06,
      "loss": 0.7439,
      "step": 5319
    },
    {
      "epoch": 0.47977634486179377,
      "grad_norm": 6.30242145566559,
      "learning_rate": 2.2276398512204847e-06,
      "loss": 0.7066,
      "step": 5320
    },
    {
      "epoch": 0.479866528385264,
      "grad_norm": 1.6636060348348085,
      "learning_rate": 2.2270594274420382e-06,
      "loss": 0.7684,
      "step": 5321
    },
    {
      "epoch": 0.4799567119087343,
      "grad_norm": 0.6728958978981159,
      "learning_rate": 2.22647898428963e-06,
      "loss": 0.6082,
      "step": 5322
    },
    {
      "epoch": 0.4800468954322045,
      "grad_norm": 1.3385895400138617,
      "learning_rate": 2.225898521812785e-06,
      "loss": 0.8036,
      "step": 5323
    },
    {
      "epoch": 0.4801370789556748,
      "grad_norm": 1.570545539834516,
      "learning_rate": 2.2253180400610337e-06,
      "loss": 0.8076,
      "step": 5324
    },
    {
      "epoch": 0.48022726247914505,
      "grad_norm": 3.76107031441492,
      "learning_rate": 2.2247375390839037e-06,
      "loss": 0.745,
      "step": 5325
    },
    {
      "epoch": 0.48031744600261533,
      "grad_norm": 1.7542027008390746,
      "learning_rate": 2.224157018930928e-06,
      "loss": 0.7269,
      "step": 5326
    },
    {
      "epoch": 0.48040762952608557,
      "grad_norm": 1.9377924175479864,
      "learning_rate": 2.2235764796516395e-06,
      "loss": 0.7596,
      "step": 5327
    },
    {
      "epoch": 0.48049781304955586,
      "grad_norm": 1.5693100343761044,
      "learning_rate": 2.222995921295573e-06,
      "loss": 0.7583,
      "step": 5328
    },
    {
      "epoch": 0.4805879965730261,
      "grad_norm": 3.0041205400443256,
      "learning_rate": 2.222415343912265e-06,
      "loss": 0.7381,
      "step": 5329
    },
    {
      "epoch": 0.4806781800964964,
      "grad_norm": 1.9183128032806507,
      "learning_rate": 2.221834747551254e-06,
      "loss": 0.8292,
      "step": 5330
    },
    {
      "epoch": 0.4807683636199666,
      "grad_norm": 1.5827601131344478,
      "learning_rate": 2.221254132262078e-06,
      "loss": 0.7323,
      "step": 5331
    },
    {
      "epoch": 0.4808585471434369,
      "grad_norm": 1.7258949197451376,
      "learning_rate": 2.2206734980942802e-06,
      "loss": 0.8011,
      "step": 5332
    },
    {
      "epoch": 0.48094873066690713,
      "grad_norm": 4.019437768440412,
      "learning_rate": 2.2200928450974024e-06,
      "loss": 0.7336,
      "step": 5333
    },
    {
      "epoch": 0.4810389141903774,
      "grad_norm": 1.6150417374452541,
      "learning_rate": 2.21951217332099e-06,
      "loss": 0.7857,
      "step": 5334
    },
    {
      "epoch": 0.48112909771384765,
      "grad_norm": 2.0650019642448783,
      "learning_rate": 2.2189314828145883e-06,
      "loss": 0.6723,
      "step": 5335
    },
    {
      "epoch": 0.48121928123731794,
      "grad_norm": 1.540034495499926,
      "learning_rate": 2.2183507736277453e-06,
      "loss": 0.7458,
      "step": 5336
    },
    {
      "epoch": 0.48130946476078823,
      "grad_norm": 1.573791597405905,
      "learning_rate": 2.2177700458100107e-06,
      "loss": 0.7732,
      "step": 5337
    },
    {
      "epoch": 0.48139964828425846,
      "grad_norm": 2.1889059725555717,
      "learning_rate": 2.2171892994109346e-06,
      "loss": 0.7354,
      "step": 5338
    },
    {
      "epoch": 0.48148983180772875,
      "grad_norm": 1.618645568189062,
      "learning_rate": 2.21660853448007e-06,
      "loss": 0.8274,
      "step": 5339
    },
    {
      "epoch": 0.481580015331199,
      "grad_norm": 2.549367235250916,
      "learning_rate": 2.2160277510669703e-06,
      "loss": 0.7169,
      "step": 5340
    },
    {
      "epoch": 0.4816701988546693,
      "grad_norm": 1.37801788502009,
      "learning_rate": 2.215446949221193e-06,
      "loss": 0.7495,
      "step": 5341
    },
    {
      "epoch": 0.4817603823781395,
      "grad_norm": 1.666847441724027,
      "learning_rate": 2.2148661289922924e-06,
      "loss": 0.7711,
      "step": 5342
    },
    {
      "epoch": 0.4818505659016098,
      "grad_norm": 2.614956853156868,
      "learning_rate": 2.21428529042983e-06,
      "loss": 0.8252,
      "step": 5343
    },
    {
      "epoch": 0.48194074942508003,
      "grad_norm": 3.1919003201076657,
      "learning_rate": 2.2137044335833647e-06,
      "loss": 0.8097,
      "step": 5344
    },
    {
      "epoch": 0.4820309329485503,
      "grad_norm": 2.1394753814871303,
      "learning_rate": 2.213123558502459e-06,
      "loss": 0.7426,
      "step": 5345
    },
    {
      "epoch": 0.48212111647202055,
      "grad_norm": 1.7831005054373252,
      "learning_rate": 2.2125426652366763e-06,
      "loss": 0.7773,
      "step": 5346
    },
    {
      "epoch": 0.48221129999549084,
      "grad_norm": 1.6529738935842555,
      "learning_rate": 2.211961753835581e-06,
      "loss": 0.8535,
      "step": 5347
    },
    {
      "epoch": 0.48230148351896107,
      "grad_norm": 1.3217616057401396,
      "learning_rate": 2.21138082434874e-06,
      "loss": 0.7511,
      "step": 5348
    },
    {
      "epoch": 0.48239166704243136,
      "grad_norm": 3.9748453506040033,
      "learning_rate": 2.210799876825722e-06,
      "loss": 0.7921,
      "step": 5349
    },
    {
      "epoch": 0.4824818505659016,
      "grad_norm": 1.6328909093109083,
      "learning_rate": 2.210218911316096e-06,
      "loss": 0.7216,
      "step": 5350
    },
    {
      "epoch": 0.4825720340893719,
      "grad_norm": 1.7354648362061762,
      "learning_rate": 2.2096379278694336e-06,
      "loss": 0.7105,
      "step": 5351
    },
    {
      "epoch": 0.4826622176128421,
      "grad_norm": 1.5730483482094386,
      "learning_rate": 2.2090569265353074e-06,
      "loss": 0.7047,
      "step": 5352
    },
    {
      "epoch": 0.4827524011363124,
      "grad_norm": 2.7367048528363607,
      "learning_rate": 2.2084759073632912e-06,
      "loss": 0.761,
      "step": 5353
    },
    {
      "epoch": 0.48284258465978264,
      "grad_norm": 2.2501502516699485,
      "learning_rate": 2.2078948704029606e-06,
      "loss": 0.7203,
      "step": 5354
    },
    {
      "epoch": 0.4829327681832529,
      "grad_norm": 1.6325890055018808,
      "learning_rate": 2.2073138157038935e-06,
      "loss": 0.6847,
      "step": 5355
    },
    {
      "epoch": 0.48302295170672316,
      "grad_norm": 1.6437932258985275,
      "learning_rate": 2.2067327433156687e-06,
      "loss": 0.7505,
      "step": 5356
    },
    {
      "epoch": 0.48311313523019345,
      "grad_norm": 1.7084917677317337,
      "learning_rate": 2.2061516532878667e-06,
      "loss": 0.7756,
      "step": 5357
    },
    {
      "epoch": 0.4832033187536637,
      "grad_norm": 1.5102430923523693,
      "learning_rate": 2.2055705456700686e-06,
      "loss": 0.8048,
      "step": 5358
    },
    {
      "epoch": 0.48329350227713397,
      "grad_norm": 1.9438634170855336,
      "learning_rate": 2.204989420511858e-06,
      "loss": 0.6561,
      "step": 5359
    },
    {
      "epoch": 0.48338368580060426,
      "grad_norm": 1.665299148606266,
      "learning_rate": 2.20440827786282e-06,
      "loss": 0.7049,
      "step": 5360
    },
    {
      "epoch": 0.4834738693240745,
      "grad_norm": 1.6166132400310311,
      "learning_rate": 2.20382711777254e-06,
      "loss": 0.7628,
      "step": 5361
    },
    {
      "epoch": 0.4835640528475448,
      "grad_norm": 1.679896922695907,
      "learning_rate": 2.203245940290607e-06,
      "loss": 0.7674,
      "step": 5362
    },
    {
      "epoch": 0.483654236371015,
      "grad_norm": 1.9398864692227509,
      "learning_rate": 2.2026647454666097e-06,
      "loss": 0.7576,
      "step": 5363
    },
    {
      "epoch": 0.4837444198944853,
      "grad_norm": 1.8691450256487185,
      "learning_rate": 2.2020835333501384e-06,
      "loss": 0.7936,
      "step": 5364
    },
    {
      "epoch": 0.48383460341795553,
      "grad_norm": 1.59860290984079,
      "learning_rate": 2.2015023039907863e-06,
      "loss": 0.8177,
      "step": 5365
    },
    {
      "epoch": 0.4839247869414258,
      "grad_norm": 1.4297271889769398,
      "learning_rate": 2.2009210574381464e-06,
      "loss": 0.7394,
      "step": 5366
    },
    {
      "epoch": 0.48401497046489605,
      "grad_norm": 1.4620087312001904,
      "learning_rate": 2.2003397937418134e-06,
      "loss": 0.7049,
      "step": 5367
    },
    {
      "epoch": 0.48410515398836634,
      "grad_norm": 0.6786063353802954,
      "learning_rate": 2.1997585129513852e-06,
      "loss": 0.603,
      "step": 5368
    },
    {
      "epoch": 0.4841953375118366,
      "grad_norm": 1.9083950293727883,
      "learning_rate": 2.1991772151164595e-06,
      "loss": 0.7814,
      "step": 5369
    },
    {
      "epoch": 0.48428552103530687,
      "grad_norm": 1.4723013290808649,
      "learning_rate": 2.1985959002866346e-06,
      "loss": 0.8039,
      "step": 5370
    },
    {
      "epoch": 0.4843757045587771,
      "grad_norm": 2.623527568023062,
      "learning_rate": 2.198014568511513e-06,
      "loss": 0.8302,
      "step": 5371
    },
    {
      "epoch": 0.4844658880822474,
      "grad_norm": 0.6655347995802217,
      "learning_rate": 2.1974332198406965e-06,
      "loss": 0.5754,
      "step": 5372
    },
    {
      "epoch": 0.4845560716057176,
      "grad_norm": 1.54996536277709,
      "learning_rate": 2.196851854323789e-06,
      "loss": 0.7421,
      "step": 5373
    },
    {
      "epoch": 0.4846462551291879,
      "grad_norm": 1.7785171156888933,
      "learning_rate": 2.196270472010396e-06,
      "loss": 0.7431,
      "step": 5374
    },
    {
      "epoch": 0.48473643865265814,
      "grad_norm": 1.5744518930906133,
      "learning_rate": 2.195689072950124e-06,
      "loss": 0.7574,
      "step": 5375
    },
    {
      "epoch": 0.48482662217612843,
      "grad_norm": 0.5868952235595739,
      "learning_rate": 2.195107657192581e-06,
      "loss": 0.5633,
      "step": 5376
    },
    {
      "epoch": 0.48491680569959866,
      "grad_norm": 0.596459555718487,
      "learning_rate": 2.194526224787378e-06,
      "loss": 0.5175,
      "step": 5377
    },
    {
      "epoch": 0.48500698922306895,
      "grad_norm": 2.145895046593582,
      "learning_rate": 2.1939447757841236e-06,
      "loss": 0.7264,
      "step": 5378
    },
    {
      "epoch": 0.4850971727465392,
      "grad_norm": 2.257328073876629,
      "learning_rate": 2.193363310232432e-06,
      "loss": 0.8197,
      "step": 5379
    },
    {
      "epoch": 0.4851873562700095,
      "grad_norm": 1.324523080609585,
      "learning_rate": 2.192781828181917e-06,
      "loss": 0.8575,
      "step": 5380
    },
    {
      "epoch": 0.4852775397934797,
      "grad_norm": 1.3454647212909014,
      "learning_rate": 2.192200329682193e-06,
      "loss": 0.7805,
      "step": 5381
    },
    {
      "epoch": 0.48536772331695,
      "grad_norm": 1.4803833144203347,
      "learning_rate": 2.1916188147828767e-06,
      "loss": 0.7566,
      "step": 5382
    },
    {
      "epoch": 0.48545790684042023,
      "grad_norm": 1.4884414386396818,
      "learning_rate": 2.191037283533587e-06,
      "loss": 0.7718,
      "step": 5383
    },
    {
      "epoch": 0.4855480903638905,
      "grad_norm": 2.365041735968377,
      "learning_rate": 2.1904557359839428e-06,
      "loss": 0.7561,
      "step": 5384
    },
    {
      "epoch": 0.4856382738873608,
      "grad_norm": 1.9681943684597676,
      "learning_rate": 2.189874172183565e-06,
      "loss": 0.7339,
      "step": 5385
    },
    {
      "epoch": 0.48572845741083104,
      "grad_norm": 1.7328880052824904,
      "learning_rate": 2.1892925921820763e-06,
      "loss": 0.7272,
      "step": 5386
    },
    {
      "epoch": 0.4858186409343013,
      "grad_norm": 1.6981141457265185,
      "learning_rate": 2.1887109960290994e-06,
      "loss": 0.7559,
      "step": 5387
    },
    {
      "epoch": 0.48590882445777156,
      "grad_norm": 2.3557234036526022,
      "learning_rate": 2.18812938377426e-06,
      "loss": 0.8523,
      "step": 5388
    },
    {
      "epoch": 0.48599900798124185,
      "grad_norm": 1.6948657172245882,
      "learning_rate": 2.187547755467184e-06,
      "loss": 0.7939,
      "step": 5389
    },
    {
      "epoch": 0.4860891915047121,
      "grad_norm": 1.7736956034646982,
      "learning_rate": 2.1869661111574994e-06,
      "loss": 0.6525,
      "step": 5390
    },
    {
      "epoch": 0.48617937502818237,
      "grad_norm": 1.8084252033834023,
      "learning_rate": 2.1863844508948353e-06,
      "loss": 0.8056,
      "step": 5391
    },
    {
      "epoch": 0.4862695585516526,
      "grad_norm": 1.45525810613506,
      "learning_rate": 2.185802774728823e-06,
      "loss": 0.7736,
      "step": 5392
    },
    {
      "epoch": 0.4863597420751229,
      "grad_norm": 1.6898251515289586,
      "learning_rate": 2.1852210827090927e-06,
      "loss": 0.7102,
      "step": 5393
    },
    {
      "epoch": 0.4864499255985931,
      "grad_norm": 1.509758913486659,
      "learning_rate": 2.184639374885278e-06,
      "loss": 0.6983,
      "step": 5394
    },
    {
      "epoch": 0.4865401091220634,
      "grad_norm": 1.9501462270245158,
      "learning_rate": 2.184057651307014e-06,
      "loss": 0.8551,
      "step": 5395
    },
    {
      "epoch": 0.48663029264553365,
      "grad_norm": 1.9945103884426656,
      "learning_rate": 2.183475912023937e-06,
      "loss": 0.8318,
      "step": 5396
    },
    {
      "epoch": 0.48672047616900393,
      "grad_norm": 1.875598176270532,
      "learning_rate": 2.1828941570856826e-06,
      "loss": 0.7745,
      "step": 5397
    },
    {
      "epoch": 0.48681065969247417,
      "grad_norm": 0.6432893289906996,
      "learning_rate": 2.1823123865418903e-06,
      "loss": 0.5727,
      "step": 5398
    },
    {
      "epoch": 0.48690084321594446,
      "grad_norm": 2.136998466858874,
      "learning_rate": 2.1817306004422e-06,
      "loss": 0.7312,
      "step": 5399
    },
    {
      "epoch": 0.4869910267394147,
      "grad_norm": 1.681166312818634,
      "learning_rate": 2.1811487988362527e-06,
      "loss": 0.7833,
      "step": 5400
    },
    {
      "epoch": 0.487081210262885,
      "grad_norm": 1.7857545480276373,
      "learning_rate": 2.1805669817736917e-06,
      "loss": 0.8309,
      "step": 5401
    },
    {
      "epoch": 0.4871713937863552,
      "grad_norm": 1.7640534555289482,
      "learning_rate": 2.17998514930416e-06,
      "loss": 0.5781,
      "step": 5402
    },
    {
      "epoch": 0.4872615773098255,
      "grad_norm": 2.442748263967395,
      "learning_rate": 2.1794033014773025e-06,
      "loss": 0.7608,
      "step": 5403
    },
    {
      "epoch": 0.48735176083329573,
      "grad_norm": 1.589032315217901,
      "learning_rate": 2.178821438342766e-06,
      "loss": 0.824,
      "step": 5404
    },
    {
      "epoch": 0.487441944356766,
      "grad_norm": 1.7231494431116237,
      "learning_rate": 2.1782395599501996e-06,
      "loss": 0.7477,
      "step": 5405
    },
    {
      "epoch": 0.48753212788023625,
      "grad_norm": 0.6561488573152324,
      "learning_rate": 2.1776576663492498e-06,
      "loss": 0.5844,
      "step": 5406
    },
    {
      "epoch": 0.48762231140370654,
      "grad_norm": 2.202786899893581,
      "learning_rate": 2.177075757589569e-06,
      "loss": 0.8442,
      "step": 5407
    },
    {
      "epoch": 0.48771249492717683,
      "grad_norm": 1.5880635860598091,
      "learning_rate": 2.176493833720808e-06,
      "loss": 0.6841,
      "step": 5408
    },
    {
      "epoch": 0.48780267845064706,
      "grad_norm": 1.6179869514005185,
      "learning_rate": 2.1759118947926195e-06,
      "loss": 0.7245,
      "step": 5409
    },
    {
      "epoch": 0.48789286197411735,
      "grad_norm": 1.8044194274231964,
      "learning_rate": 2.1753299408546587e-06,
      "loss": 0.7687,
      "step": 5410
    },
    {
      "epoch": 0.4879830454975876,
      "grad_norm": 1.526250111097977,
      "learning_rate": 2.1747479719565803e-06,
      "loss": 0.7626,
      "step": 5411
    },
    {
      "epoch": 0.4880732290210579,
      "grad_norm": 1.5137663780983304,
      "learning_rate": 2.174165988148042e-06,
      "loss": 0.821,
      "step": 5412
    },
    {
      "epoch": 0.4881634125445281,
      "grad_norm": 1.4914880681163272,
      "learning_rate": 2.1735839894787003e-06,
      "loss": 0.7241,
      "step": 5413
    },
    {
      "epoch": 0.4882535960679984,
      "grad_norm": 0.7666466183390729,
      "learning_rate": 2.1730019759982163e-06,
      "loss": 0.6291,
      "step": 5414
    },
    {
      "epoch": 0.48834377959146863,
      "grad_norm": 1.5257501508177336,
      "learning_rate": 2.172419947756249e-06,
      "loss": 0.8084,
      "step": 5415
    },
    {
      "epoch": 0.4884339631149389,
      "grad_norm": 2.002145570018554,
      "learning_rate": 2.171837904802461e-06,
      "loss": 0.809,
      "step": 5416
    },
    {
      "epoch": 0.48852414663840915,
      "grad_norm": 2.0704085417729012,
      "learning_rate": 2.171255847186516e-06,
      "loss": 0.7538,
      "step": 5417
    },
    {
      "epoch": 0.48861433016187944,
      "grad_norm": 1.8524569511817124,
      "learning_rate": 2.1706737749580783e-06,
      "loss": 0.7436,
      "step": 5418
    },
    {
      "epoch": 0.4887045136853497,
      "grad_norm": 1.5739045314652569,
      "learning_rate": 2.1700916881668127e-06,
      "loss": 0.8291,
      "step": 5419
    },
    {
      "epoch": 0.48879469720881996,
      "grad_norm": 1.9131339798834637,
      "learning_rate": 2.1695095868623862e-06,
      "loss": 0.8484,
      "step": 5420
    },
    {
      "epoch": 0.4888848807322902,
      "grad_norm": 1.6032133200783087,
      "learning_rate": 2.168927471094467e-06,
      "loss": 0.6936,
      "step": 5421
    },
    {
      "epoch": 0.4889750642557605,
      "grad_norm": 1.940382382373495,
      "learning_rate": 2.168345340912725e-06,
      "loss": 0.7837,
      "step": 5422
    },
    {
      "epoch": 0.4890652477792307,
      "grad_norm": 1.9940163987460127,
      "learning_rate": 2.1677631963668298e-06,
      "loss": 0.7781,
      "step": 5423
    },
    {
      "epoch": 0.489155431302701,
      "grad_norm": 2.0502467448907957,
      "learning_rate": 2.167181037506453e-06,
      "loss": 0.7161,
      "step": 5424
    },
    {
      "epoch": 0.48924561482617124,
      "grad_norm": 1.5055932033628643,
      "learning_rate": 2.1665988643812693e-06,
      "loss": 0.6971,
      "step": 5425
    },
    {
      "epoch": 0.4893357983496415,
      "grad_norm": 1.7567910850783361,
      "learning_rate": 2.166016677040951e-06,
      "loss": 0.757,
      "step": 5426
    },
    {
      "epoch": 0.48942598187311176,
      "grad_norm": 1.7338986129159415,
      "learning_rate": 2.165434475535175e-06,
      "loss": 0.7493,
      "step": 5427
    },
    {
      "epoch": 0.48951616539658205,
      "grad_norm": 2.287745490737421,
      "learning_rate": 2.1648522599136173e-06,
      "loss": 0.8013,
      "step": 5428
    },
    {
      "epoch": 0.4896063489200523,
      "grad_norm": 1.382349486340728,
      "learning_rate": 2.164270030225956e-06,
      "loss": 0.7308,
      "step": 5429
    },
    {
      "epoch": 0.48969653244352257,
      "grad_norm": 1.358377331721259,
      "learning_rate": 2.16368778652187e-06,
      "loss": 0.6683,
      "step": 5430
    },
    {
      "epoch": 0.4897867159669928,
      "grad_norm": 1.6918970677740726,
      "learning_rate": 2.163105528851039e-06,
      "loss": 0.7993,
      "step": 5431
    },
    {
      "epoch": 0.4898768994904631,
      "grad_norm": 1.5663311338607457,
      "learning_rate": 2.1625232572631448e-06,
      "loss": 0.7619,
      "step": 5432
    },
    {
      "epoch": 0.4899670830139334,
      "grad_norm": 0.7454986595689274,
      "learning_rate": 2.161940971807871e-06,
      "loss": 0.5915,
      "step": 5433
    },
    {
      "epoch": 0.4900572665374036,
      "grad_norm": 2.3291476263133544,
      "learning_rate": 2.1613586725348994e-06,
      "loss": 0.7467,
      "step": 5434
    },
    {
      "epoch": 0.4901474500608739,
      "grad_norm": 1.5544757794701565,
      "learning_rate": 2.1607763594939176e-06,
      "loss": 0.8378,
      "step": 5435
    },
    {
      "epoch": 0.49023763358434413,
      "grad_norm": 1.5083620325101543,
      "learning_rate": 2.1601940327346093e-06,
      "loss": 0.6686,
      "step": 5436
    },
    {
      "epoch": 0.4903278171078144,
      "grad_norm": 1.4936912908149629,
      "learning_rate": 2.159611692306663e-06,
      "loss": 0.8317,
      "step": 5437
    },
    {
      "epoch": 0.49041800063128466,
      "grad_norm": 2.2283601698285382,
      "learning_rate": 2.1590293382597667e-06,
      "loss": 0.7183,
      "step": 5438
    },
    {
      "epoch": 0.49050818415475494,
      "grad_norm": 1.7981629109819701,
      "learning_rate": 2.1584469706436102e-06,
      "loss": 0.7008,
      "step": 5439
    },
    {
      "epoch": 0.4905983676782252,
      "grad_norm": 1.5037373237360963,
      "learning_rate": 2.1578645895078855e-06,
      "loss": 0.6719,
      "step": 5440
    },
    {
      "epoch": 0.49068855120169547,
      "grad_norm": 1.742450369260817,
      "learning_rate": 2.157282194902283e-06,
      "loss": 0.8169,
      "step": 5441
    },
    {
      "epoch": 0.4907787347251657,
      "grad_norm": 1.8081239291945135,
      "learning_rate": 2.1566997868764965e-06,
      "loss": 0.6965,
      "step": 5442
    },
    {
      "epoch": 0.490868918248636,
      "grad_norm": 1.7331210578023466,
      "learning_rate": 2.15611736548022e-06,
      "loss": 0.7214,
      "step": 5443
    },
    {
      "epoch": 0.4909591017721062,
      "grad_norm": 1.6212941241234158,
      "learning_rate": 2.155534930763149e-06,
      "loss": 0.7706,
      "step": 5444
    },
    {
      "epoch": 0.4910492852955765,
      "grad_norm": 2.317489396516478,
      "learning_rate": 2.1549524827749804e-06,
      "loss": 0.6421,
      "step": 5445
    },
    {
      "epoch": 0.49113946881904674,
      "grad_norm": 4.205524879745709,
      "learning_rate": 2.1543700215654115e-06,
      "loss": 0.6093,
      "step": 5446
    },
    {
      "epoch": 0.49122965234251703,
      "grad_norm": 1.5111164795711622,
      "learning_rate": 2.153787547184141e-06,
      "loss": 0.7482,
      "step": 5447
    },
    {
      "epoch": 0.49131983586598726,
      "grad_norm": 1.5048809905539344,
      "learning_rate": 2.1532050596808695e-06,
      "loss": 0.6848,
      "step": 5448
    },
    {
      "epoch": 0.49141001938945755,
      "grad_norm": 1.857792284882002,
      "learning_rate": 2.152622559105297e-06,
      "loss": 0.806,
      "step": 5449
    },
    {
      "epoch": 0.4915002029129278,
      "grad_norm": 1.7587136226489772,
      "learning_rate": 2.152040045507126e-06,
      "loss": 0.8048,
      "step": 5450
    },
    {
      "epoch": 0.4915903864363981,
      "grad_norm": 2.760449669598678,
      "learning_rate": 2.1514575189360607e-06,
      "loss": 0.7726,
      "step": 5451
    },
    {
      "epoch": 0.4916805699598683,
      "grad_norm": 1.6466023865533161,
      "learning_rate": 2.1508749794418043e-06,
      "loss": 0.7165,
      "step": 5452
    },
    {
      "epoch": 0.4917707534833386,
      "grad_norm": 1.619929766721983,
      "learning_rate": 2.1502924270740626e-06,
      "loss": 0.7272,
      "step": 5453
    },
    {
      "epoch": 0.49186093700680883,
      "grad_norm": 1.8859857434515679,
      "learning_rate": 2.1497098618825427e-06,
      "loss": 0.7632,
      "step": 5454
    },
    {
      "epoch": 0.4919511205302791,
      "grad_norm": 1.789929775267056,
      "learning_rate": 2.1491272839169516e-06,
      "loss": 0.8031,
      "step": 5455
    },
    {
      "epoch": 0.4920413040537494,
      "grad_norm": 1.5731439826192863,
      "learning_rate": 2.1485446932269986e-06,
      "loss": 0.8734,
      "step": 5456
    },
    {
      "epoch": 0.49213148757721964,
      "grad_norm": 1.6481865425177964,
      "learning_rate": 2.147962089862393e-06,
      "loss": 0.7953,
      "step": 5457
    },
    {
      "epoch": 0.49222167110068993,
      "grad_norm": 2.3003013123223495,
      "learning_rate": 2.1473794738728462e-06,
      "loss": 0.6641,
      "step": 5458
    },
    {
      "epoch": 0.49231185462416016,
      "grad_norm": 1.9044173764959025,
      "learning_rate": 2.14679684530807e-06,
      "loss": 0.7894,
      "step": 5459
    },
    {
      "epoch": 0.49240203814763045,
      "grad_norm": 1.6288834897287656,
      "learning_rate": 2.1462142042177774e-06,
      "loss": 0.738,
      "step": 5460
    },
    {
      "epoch": 0.4924922216711007,
      "grad_norm": 1.5247597458242272,
      "learning_rate": 2.145631550651683e-06,
      "loss": 0.753,
      "step": 5461
    },
    {
      "epoch": 0.49258240519457097,
      "grad_norm": 1.7055859765285033,
      "learning_rate": 2.1450488846595016e-06,
      "loss": 0.8558,
      "step": 5462
    },
    {
      "epoch": 0.4926725887180412,
      "grad_norm": 1.7056243476013748,
      "learning_rate": 2.14446620629095e-06,
      "loss": 0.7397,
      "step": 5463
    },
    {
      "epoch": 0.4927627722415115,
      "grad_norm": 1.517949082290843,
      "learning_rate": 2.1438835155957445e-06,
      "loss": 0.8237,
      "step": 5464
    },
    {
      "epoch": 0.4928529557649817,
      "grad_norm": 1.7904772765787749,
      "learning_rate": 2.143300812623604e-06,
      "loss": 0.7966,
      "step": 5465
    },
    {
      "epoch": 0.492943139288452,
      "grad_norm": 1.71062456169304,
      "learning_rate": 2.1427180974242485e-06,
      "loss": 0.7435,
      "step": 5466
    },
    {
      "epoch": 0.49303332281192225,
      "grad_norm": 45.17411789658294,
      "learning_rate": 2.142135370047398e-06,
      "loss": 0.7844,
      "step": 5467
    },
    {
      "epoch": 0.49312350633539254,
      "grad_norm": 1.9371602775776662,
      "learning_rate": 2.1415526305427735e-06,
      "loss": 0.7903,
      "step": 5468
    },
    {
      "epoch": 0.49321368985886277,
      "grad_norm": 1.3648215612519563,
      "learning_rate": 2.140969878960098e-06,
      "loss": 0.681,
      "step": 5469
    },
    {
      "epoch": 0.49330387338233306,
      "grad_norm": 2.467745516233764,
      "learning_rate": 2.1403871153490956e-06,
      "loss": 0.8174,
      "step": 5470
    },
    {
      "epoch": 0.4933940569058033,
      "grad_norm": 3.9427318631086945,
      "learning_rate": 2.13980433975949e-06,
      "loss": 0.8781,
      "step": 5471
    },
    {
      "epoch": 0.4934842404292736,
      "grad_norm": 3.8637292677373396,
      "learning_rate": 2.1392215522410076e-06,
      "loss": 0.8045,
      "step": 5472
    },
    {
      "epoch": 0.4935744239527438,
      "grad_norm": 2.1354761503044903,
      "learning_rate": 2.1386387528433743e-06,
      "loss": 0.8739,
      "step": 5473
    },
    {
      "epoch": 0.4936646074762141,
      "grad_norm": 2.0481805496202656,
      "learning_rate": 2.1380559416163186e-06,
      "loss": 0.7901,
      "step": 5474
    },
    {
      "epoch": 0.49375479099968433,
      "grad_norm": 1.4711694472048134,
      "learning_rate": 2.1374731186095685e-06,
      "loss": 0.8058,
      "step": 5475
    },
    {
      "epoch": 0.4938449745231546,
      "grad_norm": 1.6361552795783816,
      "learning_rate": 2.136890283872854e-06,
      "loss": 0.7129,
      "step": 5476
    },
    {
      "epoch": 0.49393515804662486,
      "grad_norm": 2.206862937788408,
      "learning_rate": 2.136307437455906e-06,
      "loss": 0.7402,
      "step": 5477
    },
    {
      "epoch": 0.49402534157009514,
      "grad_norm": 1.422756215437372,
      "learning_rate": 2.135724579408456e-06,
      "loss": 0.7047,
      "step": 5478
    },
    {
      "epoch": 0.49411552509356543,
      "grad_norm": 5.2957944414865175,
      "learning_rate": 2.1351417097802356e-06,
      "loss": 0.8546,
      "step": 5479
    },
    {
      "epoch": 0.49420570861703567,
      "grad_norm": 1.9453120097577195,
      "learning_rate": 2.1345588286209798e-06,
      "loss": 0.6427,
      "step": 5480
    },
    {
      "epoch": 0.49429589214050595,
      "grad_norm": 2.1329703359924235,
      "learning_rate": 2.1339759359804227e-06,
      "loss": 0.6058,
      "step": 5481
    },
    {
      "epoch": 0.4943860756639762,
      "grad_norm": 1.8780961223292731,
      "learning_rate": 2.1333930319082997e-06,
      "loss": 0.7358,
      "step": 5482
    },
    {
      "epoch": 0.4944762591874465,
      "grad_norm": 1.5028758296524825,
      "learning_rate": 2.132810116454348e-06,
      "loss": 0.7626,
      "step": 5483
    },
    {
      "epoch": 0.4945664427109167,
      "grad_norm": 3.0850185909360253,
      "learning_rate": 2.132227189668305e-06,
      "loss": 0.7513,
      "step": 5484
    },
    {
      "epoch": 0.494656626234387,
      "grad_norm": 1.7050684768036737,
      "learning_rate": 2.1316442515999096e-06,
      "loss": 0.701,
      "step": 5485
    },
    {
      "epoch": 0.49474680975785723,
      "grad_norm": 1.6594169137262615,
      "learning_rate": 2.1310613022989e-06,
      "loss": 0.7873,
      "step": 5486
    },
    {
      "epoch": 0.4948369932813275,
      "grad_norm": 1.6666074821772066,
      "learning_rate": 2.130478341815017e-06,
      "loss": 0.6064,
      "step": 5487
    },
    {
      "epoch": 0.49492717680479775,
      "grad_norm": 1.719026577977623,
      "learning_rate": 2.1298953701980033e-06,
      "loss": 0.6603,
      "step": 5488
    },
    {
      "epoch": 0.49501736032826804,
      "grad_norm": 2.3740841957015824,
      "learning_rate": 2.1293123874976003e-06,
      "loss": 0.7637,
      "step": 5489
    },
    {
      "epoch": 0.4951075438517383,
      "grad_norm": 1.7236436708448701,
      "learning_rate": 2.1287293937635513e-06,
      "loss": 0.7099,
      "step": 5490
    },
    {
      "epoch": 0.49519772737520856,
      "grad_norm": 1.719648577142021,
      "learning_rate": 2.1281463890456005e-06,
      "loss": 0.8115,
      "step": 5491
    },
    {
      "epoch": 0.4952879108986788,
      "grad_norm": 1.697896578929314,
      "learning_rate": 2.127563373393493e-06,
      "loss": 0.7496,
      "step": 5492
    },
    {
      "epoch": 0.4953780944221491,
      "grad_norm": 1.9086633632860945,
      "learning_rate": 2.1269803468569756e-06,
      "loss": 0.7512,
      "step": 5493
    },
    {
      "epoch": 0.4954682779456193,
      "grad_norm": 1.620633347177543,
      "learning_rate": 2.126397309485794e-06,
      "loss": 0.7941,
      "step": 5494
    },
    {
      "epoch": 0.4955584614690896,
      "grad_norm": 2.0040287924733047,
      "learning_rate": 2.1258142613296983e-06,
      "loss": 0.72,
      "step": 5495
    },
    {
      "epoch": 0.49564864499255984,
      "grad_norm": 1.4043661426762297,
      "learning_rate": 2.125231202438435e-06,
      "loss": 0.7282,
      "step": 5496
    },
    {
      "epoch": 0.49573882851603013,
      "grad_norm": 2.1263489648930416,
      "learning_rate": 2.1246481328617553e-06,
      "loss": 0.7904,
      "step": 5497
    },
    {
      "epoch": 0.49582901203950036,
      "grad_norm": 2.007939791076875,
      "learning_rate": 2.1240650526494096e-06,
      "loss": 0.7404,
      "step": 5498
    },
    {
      "epoch": 0.49591919556297065,
      "grad_norm": 1.7280446590904517,
      "learning_rate": 2.1234819618511493e-06,
      "loss": 0.7316,
      "step": 5499
    },
    {
      "epoch": 0.4960093790864409,
      "grad_norm": 0.6980849509162584,
      "learning_rate": 2.122898860516728e-06,
      "loss": 0.6191,
      "step": 5500
    },
    {
      "epoch": 0.49609956260991117,
      "grad_norm": 1.753839028478478,
      "learning_rate": 2.1223157486958976e-06,
      "loss": 0.7441,
      "step": 5501
    },
    {
      "epoch": 0.4961897461333814,
      "grad_norm": 1.432386508460318,
      "learning_rate": 2.1217326264384127e-06,
      "loss": 0.6459,
      "step": 5502
    },
    {
      "epoch": 0.4962799296568517,
      "grad_norm": 2.242340082698936,
      "learning_rate": 2.1211494937940296e-06,
      "loss": 0.7752,
      "step": 5503
    },
    {
      "epoch": 0.496370113180322,
      "grad_norm": 1.5492932461877917,
      "learning_rate": 2.1205663508125034e-06,
      "loss": 0.7696,
      "step": 5504
    },
    {
      "epoch": 0.4964602967037922,
      "grad_norm": 1.4543143706559079,
      "learning_rate": 2.1199831975435914e-06,
      "loss": 0.7186,
      "step": 5505
    },
    {
      "epoch": 0.4965504802272625,
      "grad_norm": 2.5843381978266406,
      "learning_rate": 2.1194000340370517e-06,
      "loss": 0.8033,
      "step": 5506
    },
    {
      "epoch": 0.49664066375073274,
      "grad_norm": 1.7923370777824346,
      "learning_rate": 2.1188168603426423e-06,
      "loss": 0.7035,
      "step": 5507
    },
    {
      "epoch": 0.496730847274203,
      "grad_norm": 2.492469317248322,
      "learning_rate": 2.118233676510123e-06,
      "loss": 0.7219,
      "step": 5508
    },
    {
      "epoch": 0.49682103079767326,
      "grad_norm": 1.6742755376500937,
      "learning_rate": 2.117650482589255e-06,
      "loss": 0.7665,
      "step": 5509
    },
    {
      "epoch": 0.49691121432114355,
      "grad_norm": 2.3823151616610216,
      "learning_rate": 2.1170672786297988e-06,
      "loss": 0.7956,
      "step": 5510
    },
    {
      "epoch": 0.4970013978446138,
      "grad_norm": 1.407248121595459,
      "learning_rate": 2.1164840646815174e-06,
      "loss": 0.7271,
      "step": 5511
    },
    {
      "epoch": 0.49709158136808407,
      "grad_norm": 2.9874212087033274,
      "learning_rate": 2.1159008407941726e-06,
      "loss": 0.8382,
      "step": 5512
    },
    {
      "epoch": 0.4971817648915543,
      "grad_norm": 1.7462629197884483,
      "learning_rate": 2.1153176070175293e-06,
      "loss": 0.7505,
      "step": 5513
    },
    {
      "epoch": 0.4972719484150246,
      "grad_norm": 1.478245820154235,
      "learning_rate": 2.114734363401352e-06,
      "loss": 0.6478,
      "step": 5514
    },
    {
      "epoch": 0.4973621319384948,
      "grad_norm": 2.2104461447439694,
      "learning_rate": 2.1141511099954056e-06,
      "loss": 0.7546,
      "step": 5515
    },
    {
      "epoch": 0.4974523154619651,
      "grad_norm": 1.626563933534018,
      "learning_rate": 2.1135678468494576e-06,
      "loss": 0.7388,
      "step": 5516
    },
    {
      "epoch": 0.49754249898543534,
      "grad_norm": 2.1298024382330207,
      "learning_rate": 2.112984574013275e-06,
      "loss": 0.794,
      "step": 5517
    },
    {
      "epoch": 0.49763268250890563,
      "grad_norm": 1.7205011636807122,
      "learning_rate": 2.112401291536625e-06,
      "loss": 0.7142,
      "step": 5518
    },
    {
      "epoch": 0.49772286603237587,
      "grad_norm": 1.5354883354089277,
      "learning_rate": 2.111817999469278e-06,
      "loss": 0.792,
      "step": 5519
    },
    {
      "epoch": 0.49781304955584615,
      "grad_norm": 1.841890205292061,
      "learning_rate": 2.1112346978610016e-06,
      "loss": 0.7125,
      "step": 5520
    },
    {
      "epoch": 0.4979032330793164,
      "grad_norm": 1.5715626157835294,
      "learning_rate": 2.1106513867615678e-06,
      "loss": 0.7064,
      "step": 5521
    },
    {
      "epoch": 0.4979934166027867,
      "grad_norm": 2.668833021529796,
      "learning_rate": 2.110068066220748e-06,
      "loss": 0.7572,
      "step": 5522
    },
    {
      "epoch": 0.4980836001262569,
      "grad_norm": 1.5640141589214085,
      "learning_rate": 2.109484736288313e-06,
      "loss": 0.8238,
      "step": 5523
    },
    {
      "epoch": 0.4981737836497272,
      "grad_norm": 1.8018669777650909,
      "learning_rate": 2.108901397014037e-06,
      "loss": 0.774,
      "step": 5524
    },
    {
      "epoch": 0.49826396717319743,
      "grad_norm": 1.4289834211814616,
      "learning_rate": 2.1083180484476934e-06,
      "loss": 0.7596,
      "step": 5525
    },
    {
      "epoch": 0.4983541506966677,
      "grad_norm": 1.9571155522324273,
      "learning_rate": 2.1077346906390567e-06,
      "loss": 0.7791,
      "step": 5526
    },
    {
      "epoch": 0.498444334220138,
      "grad_norm": 1.7699954200944,
      "learning_rate": 2.107151323637902e-06,
      "loss": 0.7742,
      "step": 5527
    },
    {
      "epoch": 0.49853451774360824,
      "grad_norm": 1.8450342651741587,
      "learning_rate": 2.106567947494006e-06,
      "loss": 0.7397,
      "step": 5528
    },
    {
      "epoch": 0.49862470126707853,
      "grad_norm": 3.160108115604137,
      "learning_rate": 2.1059845622571447e-06,
      "loss": 0.8242,
      "step": 5529
    },
    {
      "epoch": 0.49871488479054876,
      "grad_norm": 1.551723395818774,
      "learning_rate": 2.1054011679770956e-06,
      "loss": 0.7955,
      "step": 5530
    },
    {
      "epoch": 0.49880506831401905,
      "grad_norm": 1.6905946965066598,
      "learning_rate": 2.104817764703638e-06,
      "loss": 0.766,
      "step": 5531
    },
    {
      "epoch": 0.4988952518374893,
      "grad_norm": 1.953855759289754,
      "learning_rate": 2.1042343524865516e-06,
      "loss": 0.7622,
      "step": 5532
    },
    {
      "epoch": 0.4989854353609596,
      "grad_norm": 1.5379627771871014,
      "learning_rate": 2.103650931375615e-06,
      "loss": 0.8655,
      "step": 5533
    },
    {
      "epoch": 0.4990756188844298,
      "grad_norm": 1.8795404137246798,
      "learning_rate": 2.1030675014206094e-06,
      "loss": 0.6934,
      "step": 5534
    },
    {
      "epoch": 0.4991658024079001,
      "grad_norm": 2.899934037872453,
      "learning_rate": 2.1024840626713166e-06,
      "loss": 0.6912,
      "step": 5535
    },
    {
      "epoch": 0.4992559859313703,
      "grad_norm": 1.5670617461025658,
      "learning_rate": 2.1019006151775177e-06,
      "loss": 0.8582,
      "step": 5536
    },
    {
      "epoch": 0.4993461694548406,
      "grad_norm": 1.8576427693390845,
      "learning_rate": 2.101317158988997e-06,
      "loss": 0.7491,
      "step": 5537
    },
    {
      "epoch": 0.49943635297831085,
      "grad_norm": 3.185424877195253,
      "learning_rate": 2.1007336941555374e-06,
      "loss": 0.7318,
      "step": 5538
    },
    {
      "epoch": 0.49952653650178114,
      "grad_norm": 1.5276137793787257,
      "learning_rate": 2.1001502207269238e-06,
      "loss": 0.723,
      "step": 5539
    },
    {
      "epoch": 0.49961672002525137,
      "grad_norm": 2.127526744490642,
      "learning_rate": 2.0995667387529407e-06,
      "loss": 0.7147,
      "step": 5540
    },
    {
      "epoch": 0.49970690354872166,
      "grad_norm": 1.6652742370802556,
      "learning_rate": 2.098983248283375e-06,
      "loss": 0.7115,
      "step": 5541
    },
    {
      "epoch": 0.4997970870721919,
      "grad_norm": 1.7944015064095422,
      "learning_rate": 2.098399749368012e-06,
      "loss": 0.7608,
      "step": 5542
    },
    {
      "epoch": 0.4998872705956622,
      "grad_norm": 0.6311311638271007,
      "learning_rate": 2.09781624205664e-06,
      "loss": 0.5704,
      "step": 5543
    },
    {
      "epoch": 0.4999774541191324,
      "grad_norm": 1.5933513516773279,
      "learning_rate": 2.0972327263990477e-06,
      "loss": 0.8035,
      "step": 5544
    },
    {
      "epoch": 0.5000676376426026,
      "grad_norm": 1.7740752968016673,
      "learning_rate": 2.0966492024450226e-06,
      "loss": 0.6921,
      "step": 5545
    },
    {
      "epoch": 0.500157821166073,
      "grad_norm": 4.528819096391124,
      "learning_rate": 2.0960656702443545e-06,
      "loss": 0.9054,
      "step": 5546
    },
    {
      "epoch": 0.5002480046895432,
      "grad_norm": 1.6345286134844996,
      "learning_rate": 2.0954821298468343e-06,
      "loss": 0.7512,
      "step": 5547
    },
    {
      "epoch": 0.5003381882130135,
      "grad_norm": 1.5249268905124864,
      "learning_rate": 2.0948985813022513e-06,
      "loss": 0.8152,
      "step": 5548
    },
    {
      "epoch": 0.5004283717364837,
      "grad_norm": 4.355364742662315,
      "learning_rate": 2.094315024660399e-06,
      "loss": 0.7961,
      "step": 5549
    },
    {
      "epoch": 0.500518555259954,
      "grad_norm": 3.3162641910460713,
      "learning_rate": 2.0937314599710676e-06,
      "loss": 0.7824,
      "step": 5550
    },
    {
      "epoch": 0.5006087387834243,
      "grad_norm": 3.7782733645151185,
      "learning_rate": 2.0931478872840526e-06,
      "loss": 0.7472,
      "step": 5551
    },
    {
      "epoch": 0.5006989223068945,
      "grad_norm": 2.011174930288195,
      "learning_rate": 2.092564306649145e-06,
      "loss": 0.7393,
      "step": 5552
    },
    {
      "epoch": 0.5007891058303648,
      "grad_norm": 1.7282555335966232,
      "learning_rate": 2.091980718116141e-06,
      "loss": 0.7922,
      "step": 5553
    },
    {
      "epoch": 0.5008792893538351,
      "grad_norm": 1.718664271210707,
      "learning_rate": 2.091397121734835e-06,
      "loss": 0.7955,
      "step": 5554
    },
    {
      "epoch": 0.5009694728773053,
      "grad_norm": 1.6864420082035705,
      "learning_rate": 2.090813517555022e-06,
      "loss": 0.7856,
      "step": 5555
    },
    {
      "epoch": 0.5010596564007755,
      "grad_norm": 1.7743472819969959,
      "learning_rate": 2.0902299056265e-06,
      "loss": 0.7001,
      "step": 5556
    },
    {
      "epoch": 0.5011498399242459,
      "grad_norm": 1.7673667486267843,
      "learning_rate": 2.0896462859990643e-06,
      "loss": 0.7422,
      "step": 5557
    },
    {
      "epoch": 0.5012400234477161,
      "grad_norm": 1.5746207280129436,
      "learning_rate": 2.089062658722513e-06,
      "loss": 0.8185,
      "step": 5558
    },
    {
      "epoch": 0.5013302069711864,
      "grad_norm": 1.7927821774949693,
      "learning_rate": 2.0884790238466452e-06,
      "loss": 0.8078,
      "step": 5559
    },
    {
      "epoch": 0.5014203904946566,
      "grad_norm": 2.0058314424555252,
      "learning_rate": 2.087895381421259e-06,
      "loss": 0.7943,
      "step": 5560
    },
    {
      "epoch": 0.5015105740181269,
      "grad_norm": 2.0204608482186854,
      "learning_rate": 2.087311731496154e-06,
      "loss": 0.7429,
      "step": 5561
    },
    {
      "epoch": 0.5016007575415972,
      "grad_norm": 1.3860883542505071,
      "learning_rate": 2.08672807412113e-06,
      "loss": 0.7496,
      "step": 5562
    },
    {
      "epoch": 0.5016909410650674,
      "grad_norm": 2.3583657840971877,
      "learning_rate": 2.08614440934599e-06,
      "loss": 0.8489,
      "step": 5563
    },
    {
      "epoch": 0.5017811245885376,
      "grad_norm": 2.6729702851727066,
      "learning_rate": 2.0855607372205337e-06,
      "loss": 0.6681,
      "step": 5564
    },
    {
      "epoch": 0.501871308112008,
      "grad_norm": 3.7234747868382585,
      "learning_rate": 2.0849770577945623e-06,
      "loss": 0.7758,
      "step": 5565
    },
    {
      "epoch": 0.5019614916354782,
      "grad_norm": 1.2968065634445483,
      "learning_rate": 2.084393371117881e-06,
      "loss": 0.7887,
      "step": 5566
    },
    {
      "epoch": 0.5020516751589484,
      "grad_norm": 1.6171959770828168,
      "learning_rate": 2.0838096772402902e-06,
      "loss": 0.6896,
      "step": 5567
    },
    {
      "epoch": 0.5021418586824187,
      "grad_norm": 1.9195877358769955,
      "learning_rate": 2.0832259762115973e-06,
      "loss": 0.647,
      "step": 5568
    },
    {
      "epoch": 0.502232042205889,
      "grad_norm": 1.3308708528561728,
      "learning_rate": 2.082642268081605e-06,
      "loss": 0.7845,
      "step": 5569
    },
    {
      "epoch": 0.5023222257293593,
      "grad_norm": 1.9166826026364583,
      "learning_rate": 2.082058552900118e-06,
      "loss": 0.7115,
      "step": 5570
    },
    {
      "epoch": 0.5024124092528295,
      "grad_norm": 1.6263453709685334,
      "learning_rate": 2.081474830716944e-06,
      "loss": 0.7318,
      "step": 5571
    },
    {
      "epoch": 0.5025025927762997,
      "grad_norm": 2.117689798039126,
      "learning_rate": 2.080891101581887e-06,
      "loss": 0.7841,
      "step": 5572
    },
    {
      "epoch": 0.5025927762997701,
      "grad_norm": 12.618052771220475,
      "learning_rate": 2.080307365544755e-06,
      "loss": 0.7975,
      "step": 5573
    },
    {
      "epoch": 0.5026829598232403,
      "grad_norm": 1.740167995432766,
      "learning_rate": 2.0797236226553567e-06,
      "loss": 0.792,
      "step": 5574
    },
    {
      "epoch": 0.5027731433467105,
      "grad_norm": 1.7909839423485874,
      "learning_rate": 2.079139872963499e-06,
      "loss": 0.6959,
      "step": 5575
    },
    {
      "epoch": 0.5028633268701809,
      "grad_norm": 2.07489853116781,
      "learning_rate": 2.078556116518991e-06,
      "loss": 0.6636,
      "step": 5576
    },
    {
      "epoch": 0.5029535103936511,
      "grad_norm": 1.8762868915466122,
      "learning_rate": 2.077972353371642e-06,
      "loss": 0.7311,
      "step": 5577
    },
    {
      "epoch": 0.5030436939171213,
      "grad_norm": 1.465401911888578,
      "learning_rate": 2.077388583571262e-06,
      "loss": 0.7042,
      "step": 5578
    },
    {
      "epoch": 0.5031338774405916,
      "grad_norm": 0.6566010172175192,
      "learning_rate": 2.0768048071676608e-06,
      "loss": 0.5143,
      "step": 5579
    },
    {
      "epoch": 0.5032240609640619,
      "grad_norm": 1.7865208057865032,
      "learning_rate": 2.0762210242106505e-06,
      "loss": 0.7192,
      "step": 5580
    },
    {
      "epoch": 0.5033142444875321,
      "grad_norm": 2.021811048861234,
      "learning_rate": 2.0756372347500424e-06,
      "loss": 0.7668,
      "step": 5581
    },
    {
      "epoch": 0.5034044280110024,
      "grad_norm": 1.6604546491936505,
      "learning_rate": 2.0750534388356473e-06,
      "loss": 0.862,
      "step": 5582
    },
    {
      "epoch": 0.5034946115344726,
      "grad_norm": 3.1217290926297556,
      "learning_rate": 2.07446963651728e-06,
      "loss": 0.7017,
      "step": 5583
    },
    {
      "epoch": 0.503584795057943,
      "grad_norm": 1.7224725811330428,
      "learning_rate": 2.0738858278447516e-06,
      "loss": 0.6951,
      "step": 5584
    },
    {
      "epoch": 0.5036749785814132,
      "grad_norm": 1.760276322347735,
      "learning_rate": 2.073302012867878e-06,
      "loss": 0.8458,
      "step": 5585
    },
    {
      "epoch": 0.5037651621048834,
      "grad_norm": 1.4517735585488234,
      "learning_rate": 2.0727181916364725e-06,
      "loss": 0.7908,
      "step": 5586
    },
    {
      "epoch": 0.5038553456283537,
      "grad_norm": 1.4269330114935883,
      "learning_rate": 2.0721343642003493e-06,
      "loss": 0.7946,
      "step": 5587
    },
    {
      "epoch": 0.503945529151824,
      "grad_norm": 1.8073583878248984,
      "learning_rate": 2.0715505306093247e-06,
      "loss": 0.7866,
      "step": 5588
    },
    {
      "epoch": 0.5040357126752942,
      "grad_norm": 2.075634029912101,
      "learning_rate": 2.070966690913214e-06,
      "loss": 0.785,
      "step": 5589
    },
    {
      "epoch": 0.5041258961987645,
      "grad_norm": 2.3547513005169733,
      "learning_rate": 2.0703828451618346e-06,
      "loss": 0.6431,
      "step": 5590
    },
    {
      "epoch": 0.5042160797222347,
      "grad_norm": 1.3753779065389968,
      "learning_rate": 2.069798993405002e-06,
      "loss": 0.7516,
      "step": 5591
    },
    {
      "epoch": 0.504306263245705,
      "grad_norm": 1.6670021990948631,
      "learning_rate": 2.0692151356925345e-06,
      "loss": 0.7964,
      "step": 5592
    },
    {
      "epoch": 0.5043964467691753,
      "grad_norm": 1.7034846634774858,
      "learning_rate": 2.068631272074251e-06,
      "loss": 0.783,
      "step": 5593
    },
    {
      "epoch": 0.5044866302926455,
      "grad_norm": 1.8928685393594435,
      "learning_rate": 2.0680474025999676e-06,
      "loss": 0.8835,
      "step": 5594
    },
    {
      "epoch": 0.5045768138161157,
      "grad_norm": 0.8743635315105472,
      "learning_rate": 2.0674635273195055e-06,
      "loss": 0.632,
      "step": 5595
    },
    {
      "epoch": 0.5046669973395861,
      "grad_norm": 2.3617005665817645,
      "learning_rate": 2.066879646282682e-06,
      "loss": 0.7551,
      "step": 5596
    },
    {
      "epoch": 0.5047571808630563,
      "grad_norm": 1.7380110284172257,
      "learning_rate": 2.0662957595393194e-06,
      "loss": 0.7325,
      "step": 5597
    },
    {
      "epoch": 0.5048473643865266,
      "grad_norm": 1.6815745554493466,
      "learning_rate": 2.0657118671392373e-06,
      "loss": 0.786,
      "step": 5598
    },
    {
      "epoch": 0.5049375479099969,
      "grad_norm": 1.875204583768808,
      "learning_rate": 2.0651279691322558e-06,
      "loss": 0.8151,
      "step": 5599
    },
    {
      "epoch": 0.5050277314334671,
      "grad_norm": 1.610077047459184,
      "learning_rate": 2.0645440655681973e-06,
      "loss": 0.7108,
      "step": 5600
    },
    {
      "epoch": 0.5051179149569374,
      "grad_norm": 2.217501651802007,
      "learning_rate": 2.0639601564968826e-06,
      "loss": 0.8233,
      "step": 5601
    },
    {
      "epoch": 0.5052080984804076,
      "grad_norm": 0.6949159584415388,
      "learning_rate": 2.0633762419681355e-06,
      "loss": 0.6664,
      "step": 5602
    },
    {
      "epoch": 0.5052982820038779,
      "grad_norm": 1.4034698130612986,
      "learning_rate": 2.062792322031777e-06,
      "loss": 0.7734,
      "step": 5603
    },
    {
      "epoch": 0.5053884655273482,
      "grad_norm": 1.7955980490478671,
      "learning_rate": 2.062208396737632e-06,
      "loss": 0.7214,
      "step": 5604
    },
    {
      "epoch": 0.5054786490508184,
      "grad_norm": 1.4365802808104742,
      "learning_rate": 2.0616244661355235e-06,
      "loss": 0.7784,
      "step": 5605
    },
    {
      "epoch": 0.5055688325742886,
      "grad_norm": 2.2439217589366818,
      "learning_rate": 2.0610405302752752e-06,
      "loss": 0.7943,
      "step": 5606
    },
    {
      "epoch": 0.505659016097759,
      "grad_norm": 1.8538694536405085,
      "learning_rate": 2.060456589206713e-06,
      "loss": 0.7528,
      "step": 5607
    },
    {
      "epoch": 0.5057491996212292,
      "grad_norm": 2.4700131625237893,
      "learning_rate": 2.0598726429796614e-06,
      "loss": 0.7041,
      "step": 5608
    },
    {
      "epoch": 0.5058393831446995,
      "grad_norm": 1.5069300782221564,
      "learning_rate": 2.059288691643945e-06,
      "loss": 0.7147,
      "step": 5609
    },
    {
      "epoch": 0.5059295666681697,
      "grad_norm": 1.701913805409575,
      "learning_rate": 2.0587047352493913e-06,
      "loss": 0.8427,
      "step": 5610
    },
    {
      "epoch": 0.50601975019164,
      "grad_norm": 2.443489540834876,
      "learning_rate": 2.0581207738458248e-06,
      "loss": 0.7364,
      "step": 5611
    },
    {
      "epoch": 0.5061099337151103,
      "grad_norm": 2.755493399301873,
      "learning_rate": 2.0575368074830743e-06,
      "loss": 0.7174,
      "step": 5612
    },
    {
      "epoch": 0.5062001172385805,
      "grad_norm": 1.6814151838760252,
      "learning_rate": 2.0569528362109667e-06,
      "loss": 0.6728,
      "step": 5613
    },
    {
      "epoch": 0.5062903007620507,
      "grad_norm": 1.5377009996977422,
      "learning_rate": 2.056368860079327e-06,
      "loss": 0.7553,
      "step": 5614
    },
    {
      "epoch": 0.5063804842855211,
      "grad_norm": 5.229047290103025,
      "learning_rate": 2.0557848791379874e-06,
      "loss": 0.6953,
      "step": 5615
    },
    {
      "epoch": 0.5064706678089913,
      "grad_norm": 1.6626973171304946,
      "learning_rate": 2.0552008934367734e-06,
      "loss": 0.6638,
      "step": 5616
    },
    {
      "epoch": 0.5065608513324615,
      "grad_norm": 1.9940376217773823,
      "learning_rate": 2.0546169030255154e-06,
      "loss": 0.8347,
      "step": 5617
    },
    {
      "epoch": 0.5066510348559318,
      "grad_norm": 1.7458283566181685,
      "learning_rate": 2.054032907954041e-06,
      "loss": 0.8344,
      "step": 5618
    },
    {
      "epoch": 0.5067412183794021,
      "grad_norm": 1.47969981332427,
      "learning_rate": 2.053448908272182e-06,
      "loss": 0.724,
      "step": 5619
    },
    {
      "epoch": 0.5068314019028723,
      "grad_norm": 1.6567661992471645,
      "learning_rate": 2.0528649040297673e-06,
      "loss": 0.797,
      "step": 5620
    },
    {
      "epoch": 0.5069215854263426,
      "grad_norm": 1.7953475430806027,
      "learning_rate": 2.0522808952766266e-06,
      "loss": 0.7601,
      "step": 5621
    },
    {
      "epoch": 0.5070117689498129,
      "grad_norm": 1.5261562247995801,
      "learning_rate": 2.0516968820625925e-06,
      "loss": 0.8038,
      "step": 5622
    },
    {
      "epoch": 0.5071019524732832,
      "grad_norm": 2.8399977189713037,
      "learning_rate": 2.051112864437495e-06,
      "loss": 0.7482,
      "step": 5623
    },
    {
      "epoch": 0.5071921359967534,
      "grad_norm": 3.1059276697563805,
      "learning_rate": 2.050528842451166e-06,
      "loss": 0.7261,
      "step": 5624
    },
    {
      "epoch": 0.5072823195202236,
      "grad_norm": 1.5504571302109809,
      "learning_rate": 2.049944816153438e-06,
      "loss": 0.6416,
      "step": 5625
    },
    {
      "epoch": 0.507372503043694,
      "grad_norm": 1.4969625714196475,
      "learning_rate": 2.049360785594142e-06,
      "loss": 0.8022,
      "step": 5626
    },
    {
      "epoch": 0.5074626865671642,
      "grad_norm": 0.6153682986833532,
      "learning_rate": 2.048776750823113e-06,
      "loss": 0.5106,
      "step": 5627
    },
    {
      "epoch": 0.5075528700906344,
      "grad_norm": 0.6711138472473221,
      "learning_rate": 2.0481927118901817e-06,
      "loss": 0.564,
      "step": 5628
    },
    {
      "epoch": 0.5076430536141047,
      "grad_norm": 1.799556158234075,
      "learning_rate": 2.0476086688451824e-06,
      "loss": 0.7037,
      "step": 5629
    },
    {
      "epoch": 0.507733237137575,
      "grad_norm": 1.7821053576080272,
      "learning_rate": 2.04702462173795e-06,
      "loss": 0.7459,
      "step": 5630
    },
    {
      "epoch": 0.5078234206610452,
      "grad_norm": 2.101839068624891,
      "learning_rate": 2.0464405706183167e-06,
      "loss": 0.7379,
      "step": 5631
    },
    {
      "epoch": 0.5079136041845155,
      "grad_norm": 1.916735779165819,
      "learning_rate": 2.045856515536118e-06,
      "loss": 0.7735,
      "step": 5632
    },
    {
      "epoch": 0.5080037877079857,
      "grad_norm": 1.6097940436420917,
      "learning_rate": 2.045272456541188e-06,
      "loss": 0.7746,
      "step": 5633
    },
    {
      "epoch": 0.508093971231456,
      "grad_norm": 1.588536280221614,
      "learning_rate": 2.0446883936833635e-06,
      "loss": 0.8261,
      "step": 5634
    },
    {
      "epoch": 0.5081841547549263,
      "grad_norm": 2.2283258248628024,
      "learning_rate": 2.0441043270124782e-06,
      "loss": 0.7334,
      "step": 5635
    },
    {
      "epoch": 0.5082743382783965,
      "grad_norm": 1.9993479380992663,
      "learning_rate": 2.0435202565783683e-06,
      "loss": 0.786,
      "step": 5636
    },
    {
      "epoch": 0.5083645218018668,
      "grad_norm": 1.5314630632669535,
      "learning_rate": 2.042936182430871e-06,
      "loss": 0.72,
      "step": 5637
    },
    {
      "epoch": 0.5084547053253371,
      "grad_norm": 1.572506183678821,
      "learning_rate": 2.0423521046198206e-06,
      "loss": 0.7624,
      "step": 5638
    },
    {
      "epoch": 0.5085448888488073,
      "grad_norm": 1.5236665993601946,
      "learning_rate": 2.041768023195056e-06,
      "loss": 0.6751,
      "step": 5639
    },
    {
      "epoch": 0.5086350723722776,
      "grad_norm": 1.667741317788182,
      "learning_rate": 2.0411839382064126e-06,
      "loss": 0.828,
      "step": 5640
    },
    {
      "epoch": 0.5087252558957478,
      "grad_norm": 1.7200951600919434,
      "learning_rate": 2.040599849703729e-06,
      "loss": 0.6533,
      "step": 5641
    },
    {
      "epoch": 0.5088154394192181,
      "grad_norm": 2.0225350860457003,
      "learning_rate": 2.040015757736843e-06,
      "loss": 0.698,
      "step": 5642
    },
    {
      "epoch": 0.5089056229426884,
      "grad_norm": 1.5391229143242424,
      "learning_rate": 2.039431662355591e-06,
      "loss": 0.6835,
      "step": 5643
    },
    {
      "epoch": 0.5089958064661586,
      "grad_norm": 1.680360313332617,
      "learning_rate": 2.0388475636098126e-06,
      "loss": 0.7743,
      "step": 5644
    },
    {
      "epoch": 0.5090859899896288,
      "grad_norm": 1.474186271952582,
      "learning_rate": 2.038263461549346e-06,
      "loss": 0.7906,
      "step": 5645
    },
    {
      "epoch": 0.5091761735130992,
      "grad_norm": 1.8439544305776971,
      "learning_rate": 2.0376793562240297e-06,
      "loss": 0.7606,
      "step": 5646
    },
    {
      "epoch": 0.5092663570365694,
      "grad_norm": 1.5884207841082554,
      "learning_rate": 2.037095247683703e-06,
      "loss": 0.7753,
      "step": 5647
    },
    {
      "epoch": 0.5093565405600397,
      "grad_norm": 1.5549754733883603,
      "learning_rate": 2.0365111359782046e-06,
      "loss": 0.7136,
      "step": 5648
    },
    {
      "epoch": 0.50944672408351,
      "grad_norm": 1.7424881337882396,
      "learning_rate": 2.0359270211573757e-06,
      "loss": 0.7856,
      "step": 5649
    },
    {
      "epoch": 0.5095369076069802,
      "grad_norm": 2.0724005767571128,
      "learning_rate": 2.0353429032710545e-06,
      "loss": 0.8205,
      "step": 5650
    },
    {
      "epoch": 0.5096270911304505,
      "grad_norm": 2.8592510926128734,
      "learning_rate": 2.0347587823690825e-06,
      "loss": 0.7309,
      "step": 5651
    },
    {
      "epoch": 0.5097172746539207,
      "grad_norm": 1.6494762889988326,
      "learning_rate": 2.034174658501299e-06,
      "loss": 0.7565,
      "step": 5652
    },
    {
      "epoch": 0.509807458177391,
      "grad_norm": 2.3336424509285125,
      "learning_rate": 2.0335905317175453e-06,
      "loss": 0.7938,
      "step": 5653
    },
    {
      "epoch": 0.5098976417008613,
      "grad_norm": 2.3933209943674663,
      "learning_rate": 2.033006402067663e-06,
      "loss": 0.7631,
      "step": 5654
    },
    {
      "epoch": 0.5099878252243315,
      "grad_norm": 1.8396660137784757,
      "learning_rate": 2.0324222696014912e-06,
      "loss": 0.6597,
      "step": 5655
    },
    {
      "epoch": 0.5100780087478017,
      "grad_norm": 1.7485974004341112,
      "learning_rate": 2.0318381343688733e-06,
      "loss": 0.7997,
      "step": 5656
    },
    {
      "epoch": 0.5101681922712721,
      "grad_norm": 1.6641375027799143,
      "learning_rate": 2.0312539964196505e-06,
      "loss": 0.7399,
      "step": 5657
    },
    {
      "epoch": 0.5102583757947423,
      "grad_norm": 1.57320513406377,
      "learning_rate": 2.030669855803664e-06,
      "loss": 0.7367,
      "step": 5658
    },
    {
      "epoch": 0.5103485593182125,
      "grad_norm": 1.5185432261926803,
      "learning_rate": 2.0300857125707563e-06,
      "loss": 0.8079,
      "step": 5659
    },
    {
      "epoch": 0.5104387428416828,
      "grad_norm": 1.7775122929401486,
      "learning_rate": 2.0295015667707697e-06,
      "loss": 0.7838,
      "step": 5660
    },
    {
      "epoch": 0.5105289263651531,
      "grad_norm": 1.5049212312838085,
      "learning_rate": 2.0289174184535472e-06,
      "loss": 0.7627,
      "step": 5661
    },
    {
      "epoch": 0.5106191098886234,
      "grad_norm": 1.7663165063343875,
      "learning_rate": 2.02833326766893e-06,
      "loss": 0.7307,
      "step": 5662
    },
    {
      "epoch": 0.5107092934120936,
      "grad_norm": 1.7900615717778021,
      "learning_rate": 2.027749114466763e-06,
      "loss": 0.8203,
      "step": 5663
    },
    {
      "epoch": 0.5107994769355638,
      "grad_norm": 2.1341378701603464,
      "learning_rate": 2.027164958896889e-06,
      "loss": 0.8009,
      "step": 5664
    },
    {
      "epoch": 0.5108896604590342,
      "grad_norm": 1.8204301787882158,
      "learning_rate": 2.02658080100915e-06,
      "loss": 0.8777,
      "step": 5665
    },
    {
      "epoch": 0.5109798439825044,
      "grad_norm": 1.8217069990062265,
      "learning_rate": 2.0259966408533915e-06,
      "loss": 0.8322,
      "step": 5666
    },
    {
      "epoch": 0.5110700275059746,
      "grad_norm": 1.653120431010345,
      "learning_rate": 2.025412478479455e-06,
      "loss": 0.8432,
      "step": 5667
    },
    {
      "epoch": 0.5111602110294449,
      "grad_norm": 2.2115535888664835,
      "learning_rate": 2.0248283139371862e-06,
      "loss": 0.8064,
      "step": 5668
    },
    {
      "epoch": 0.5112503945529152,
      "grad_norm": 5.339922510220643,
      "learning_rate": 2.024244147276429e-06,
      "loss": 0.7307,
      "step": 5669
    },
    {
      "epoch": 0.5113405780763854,
      "grad_norm": 1.4118606960455224,
      "learning_rate": 2.023659978547027e-06,
      "loss": 0.8279,
      "step": 5670
    },
    {
      "epoch": 0.5114307615998557,
      "grad_norm": 1.7204771207370753,
      "learning_rate": 2.023075807798826e-06,
      "loss": 0.76,
      "step": 5671
    },
    {
      "epoch": 0.511520945123326,
      "grad_norm": 1.4928049936918366,
      "learning_rate": 2.0224916350816696e-06,
      "loss": 0.6804,
      "step": 5672
    },
    {
      "epoch": 0.5116111286467963,
      "grad_norm": 1.8157361835309649,
      "learning_rate": 2.0219074604454026e-06,
      "loss": 0.7792,
      "step": 5673
    },
    {
      "epoch": 0.5117013121702665,
      "grad_norm": 1.9379593089411276,
      "learning_rate": 2.02132328393987e-06,
      "loss": 0.7547,
      "step": 5674
    },
    {
      "epoch": 0.5117914956937367,
      "grad_norm": 1.7748114015559906,
      "learning_rate": 2.0207391056149174e-06,
      "loss": 0.7183,
      "step": 5675
    },
    {
      "epoch": 0.5118816792172071,
      "grad_norm": 2.1390430703127117,
      "learning_rate": 2.020154925520391e-06,
      "loss": 0.7325,
      "step": 5676
    },
    {
      "epoch": 0.5119718627406773,
      "grad_norm": 1.5094676050562037,
      "learning_rate": 2.0195707437061332e-06,
      "loss": 0.856,
      "step": 5677
    },
    {
      "epoch": 0.5120620462641475,
      "grad_norm": 14.551334406432026,
      "learning_rate": 2.0189865602219934e-06,
      "loss": 0.7603,
      "step": 5678
    },
    {
      "epoch": 0.5121522297876178,
      "grad_norm": 1.865602466951967,
      "learning_rate": 2.0184023751178154e-06,
      "loss": 0.8342,
      "step": 5679
    },
    {
      "epoch": 0.5122424133110881,
      "grad_norm": 1.5607643405610496,
      "learning_rate": 2.017818188443444e-06,
      "loss": 0.7129,
      "step": 5680
    },
    {
      "epoch": 0.5123325968345583,
      "grad_norm": 3.257383762021831,
      "learning_rate": 2.017234000248728e-06,
      "loss": 0.7727,
      "step": 5681
    },
    {
      "epoch": 0.5124227803580286,
      "grad_norm": 1.5331874190576922,
      "learning_rate": 2.0166498105835108e-06,
      "loss": 0.696,
      "step": 5682
    },
    {
      "epoch": 0.5125129638814988,
      "grad_norm": 1.3884687530001751,
      "learning_rate": 2.0160656194976407e-06,
      "loss": 0.6937,
      "step": 5683
    },
    {
      "epoch": 0.5126031474049692,
      "grad_norm": 1.6284020098543333,
      "learning_rate": 2.0154814270409634e-06,
      "loss": 0.7147,
      "step": 5684
    },
    {
      "epoch": 0.5126933309284394,
      "grad_norm": 1.7324196042429805,
      "learning_rate": 2.0148972332633247e-06,
      "loss": 0.7272,
      "step": 5685
    },
    {
      "epoch": 0.5127835144519096,
      "grad_norm": 2.1257345388622526,
      "learning_rate": 2.0143130382145733e-06,
      "loss": 0.8247,
      "step": 5686
    },
    {
      "epoch": 0.5128736979753798,
      "grad_norm": 2.1740300306686815,
      "learning_rate": 2.0137288419445533e-06,
      "loss": 0.7329,
      "step": 5687
    },
    {
      "epoch": 0.5129638814988502,
      "grad_norm": 2.0541222088421716,
      "learning_rate": 2.0131446445031134e-06,
      "loss": 0.8046,
      "step": 5688
    },
    {
      "epoch": 0.5130540650223204,
      "grad_norm": 1.8252984390801341,
      "learning_rate": 2.0125604459400994e-06,
      "loss": 0.8407,
      "step": 5689
    },
    {
      "epoch": 0.5131442485457907,
      "grad_norm": 1.726289131997183,
      "learning_rate": 2.0119762463053596e-06,
      "loss": 0.8198,
      "step": 5690
    },
    {
      "epoch": 0.5132344320692609,
      "grad_norm": 1.7436928852530986,
      "learning_rate": 2.0113920456487406e-06,
      "loss": 0.7521,
      "step": 5691
    },
    {
      "epoch": 0.5133246155927312,
      "grad_norm": 1.9214502501011812,
      "learning_rate": 2.010807844020088e-06,
      "loss": 0.7549,
      "step": 5692
    },
    {
      "epoch": 0.5134147991162015,
      "grad_norm": 1.5173693971191862,
      "learning_rate": 2.0102236414692524e-06,
      "loss": 0.8119,
      "step": 5693
    },
    {
      "epoch": 0.5135049826396717,
      "grad_norm": 1.4907792721295328,
      "learning_rate": 2.0096394380460777e-06,
      "loss": 0.7471,
      "step": 5694
    },
    {
      "epoch": 0.513595166163142,
      "grad_norm": 1.6243624536925823,
      "learning_rate": 2.0090552338004136e-06,
      "loss": 0.7919,
      "step": 5695
    },
    {
      "epoch": 0.5136853496866123,
      "grad_norm": 8.211516355368977,
      "learning_rate": 2.0084710287821077e-06,
      "loss": 0.7594,
      "step": 5696
    },
    {
      "epoch": 0.5137755332100825,
      "grad_norm": 1.6538339663041137,
      "learning_rate": 2.007886823041006e-06,
      "loss": 0.7987,
      "step": 5697
    },
    {
      "epoch": 0.5138657167335527,
      "grad_norm": 1.5896765318713948,
      "learning_rate": 2.0073026166269577e-06,
      "loss": 0.7749,
      "step": 5698
    },
    {
      "epoch": 0.5139559002570231,
      "grad_norm": 1.351088159487167,
      "learning_rate": 2.0067184095898093e-06,
      "loss": 0.8192,
      "step": 5699
    },
    {
      "epoch": 0.5140460837804933,
      "grad_norm": 1.6834153083002799,
      "learning_rate": 2.0061342019794094e-06,
      "loss": 0.7697,
      "step": 5700
    },
    {
      "epoch": 0.5141362673039636,
      "grad_norm": 1.460513946006923,
      "learning_rate": 2.0055499938456058e-06,
      "loss": 0.833,
      "step": 5701
    },
    {
      "epoch": 0.5142264508274338,
      "grad_norm": 2.392454357449276,
      "learning_rate": 2.0049657852382464e-06,
      "loss": 0.7308,
      "step": 5702
    },
    {
      "epoch": 0.5143166343509041,
      "grad_norm": 1.8760042044545704,
      "learning_rate": 2.0043815762071782e-06,
      "loss": 0.6936,
      "step": 5703
    },
    {
      "epoch": 0.5144068178743744,
      "grad_norm": 1.737496462482172,
      "learning_rate": 2.0037973668022492e-06,
      "loss": 0.7284,
      "step": 5704
    },
    {
      "epoch": 0.5144970013978446,
      "grad_norm": 1.7937674810641002,
      "learning_rate": 2.003213157073309e-06,
      "loss": 0.6436,
      "step": 5705
    },
    {
      "epoch": 0.5145871849213148,
      "grad_norm": 1.5755856923606402,
      "learning_rate": 2.002628947070204e-06,
      "loss": 0.7593,
      "step": 5706
    },
    {
      "epoch": 0.5146773684447852,
      "grad_norm": 1.6831286998414894,
      "learning_rate": 2.002044736842783e-06,
      "loss": 0.8257,
      "step": 5707
    },
    {
      "epoch": 0.5147675519682554,
      "grad_norm": 1.5507983151513038,
      "learning_rate": 2.001460526440894e-06,
      "loss": 0.7397,
      "step": 5708
    },
    {
      "epoch": 0.5148577354917256,
      "grad_norm": 0.6152722786020652,
      "learning_rate": 2.0008763159143843e-06,
      "loss": 0.5807,
      "step": 5709
    },
    {
      "epoch": 0.5149479190151959,
      "grad_norm": 1.7405826720975859,
      "learning_rate": 2.000292105313103e-06,
      "loss": 0.7906,
      "step": 5710
    },
    {
      "epoch": 0.5150381025386662,
      "grad_norm": 1.7066260234394104,
      "learning_rate": 1.999707894686897e-06,
      "loss": 0.7987,
      "step": 5711
    },
    {
      "epoch": 0.5151282860621365,
      "grad_norm": 1.4305701527678942,
      "learning_rate": 1.9991236840856155e-06,
      "loss": 0.7484,
      "step": 5712
    },
    {
      "epoch": 0.5152184695856067,
      "grad_norm": 2.096792032388382,
      "learning_rate": 1.9985394735591065e-06,
      "loss": 0.7334,
      "step": 5713
    },
    {
      "epoch": 0.5153086531090769,
      "grad_norm": 5.567655081950773,
      "learning_rate": 1.997955263157217e-06,
      "loss": 0.8084,
      "step": 5714
    },
    {
      "epoch": 0.5153988366325473,
      "grad_norm": 1.7084342027363377,
      "learning_rate": 1.997371052929796e-06,
      "loss": 0.795,
      "step": 5715
    },
    {
      "epoch": 0.5154890201560175,
      "grad_norm": 1.693340508202999,
      "learning_rate": 1.996786842926691e-06,
      "loss": 0.716,
      "step": 5716
    },
    {
      "epoch": 0.5155792036794877,
      "grad_norm": 1.619265341010666,
      "learning_rate": 1.9962026331977506e-06,
      "loss": 0.7707,
      "step": 5717
    },
    {
      "epoch": 0.5156693872029581,
      "grad_norm": 2.0053012446556817,
      "learning_rate": 1.9956184237928224e-06,
      "loss": 0.7763,
      "step": 5718
    },
    {
      "epoch": 0.5157595707264283,
      "grad_norm": 3.7913842707401115,
      "learning_rate": 1.995034214761754e-06,
      "loss": 0.752,
      "step": 5719
    },
    {
      "epoch": 0.5158497542498985,
      "grad_norm": 1.2802327933068613,
      "learning_rate": 1.9944500061543945e-06,
      "loss": 0.735,
      "step": 5720
    },
    {
      "epoch": 0.5159399377733688,
      "grad_norm": 1.6972711034056436,
      "learning_rate": 1.99386579802059e-06,
      "loss": 0.7798,
      "step": 5721
    },
    {
      "epoch": 0.5160301212968391,
      "grad_norm": 1.864240994984549,
      "learning_rate": 1.993281590410191e-06,
      "loss": 0.7381,
      "step": 5722
    },
    {
      "epoch": 0.5161203048203094,
      "grad_norm": 1.5359387138726686,
      "learning_rate": 1.992697383373043e-06,
      "loss": 0.7089,
      "step": 5723
    },
    {
      "epoch": 0.5162104883437796,
      "grad_norm": 0.7803907246579572,
      "learning_rate": 1.9921131769589937e-06,
      "loss": 0.6222,
      "step": 5724
    },
    {
      "epoch": 0.5163006718672498,
      "grad_norm": 1.9579245837399875,
      "learning_rate": 1.991528971217893e-06,
      "loss": 0.7657,
      "step": 5725
    },
    {
      "epoch": 0.5163908553907202,
      "grad_norm": 1.8381647762909485,
      "learning_rate": 1.9909447661995858e-06,
      "loss": 0.7575,
      "step": 5726
    },
    {
      "epoch": 0.5164810389141904,
      "grad_norm": 1.8221299426560071,
      "learning_rate": 1.990360561953922e-06,
      "loss": 0.7789,
      "step": 5727
    },
    {
      "epoch": 0.5165712224376606,
      "grad_norm": 1.370356565351992,
      "learning_rate": 1.9897763585307483e-06,
      "loss": 0.7823,
      "step": 5728
    },
    {
      "epoch": 0.5166614059611309,
      "grad_norm": 1.5344163504504054,
      "learning_rate": 1.989192155979912e-06,
      "loss": 0.8001,
      "step": 5729
    },
    {
      "epoch": 0.5167515894846012,
      "grad_norm": 1.6375249729727406,
      "learning_rate": 1.98860795435126e-06,
      "loss": 0.7927,
      "step": 5730
    },
    {
      "epoch": 0.5168417730080714,
      "grad_norm": 1.617819105270781,
      "learning_rate": 1.9880237536946406e-06,
      "loss": 0.7674,
      "step": 5731
    },
    {
      "epoch": 0.5169319565315417,
      "grad_norm": 1.830591542988925,
      "learning_rate": 1.987439554059901e-06,
      "loss": 0.7313,
      "step": 5732
    },
    {
      "epoch": 0.5170221400550119,
      "grad_norm": 1.6075209521030185,
      "learning_rate": 1.9868553554968864e-06,
      "loss": 0.7327,
      "step": 5733
    },
    {
      "epoch": 0.5171123235784822,
      "grad_norm": 2.4932172316108887,
      "learning_rate": 1.986271158055447e-06,
      "loss": 0.712,
      "step": 5734
    },
    {
      "epoch": 0.5172025071019525,
      "grad_norm": 1.599263918955766,
      "learning_rate": 1.9856869617854273e-06,
      "loss": 0.7514,
      "step": 5735
    },
    {
      "epoch": 0.5172926906254227,
      "grad_norm": 1.5885206711012196,
      "learning_rate": 1.9851027667366746e-06,
      "loss": 0.738,
      "step": 5736
    },
    {
      "epoch": 0.517382874148893,
      "grad_norm": 1.511190159372547,
      "learning_rate": 1.984518572959037e-06,
      "loss": 0.8594,
      "step": 5737
    },
    {
      "epoch": 0.5174730576723633,
      "grad_norm": 1.6920181823727487,
      "learning_rate": 1.9839343805023587e-06,
      "loss": 0.68,
      "step": 5738
    },
    {
      "epoch": 0.5175632411958335,
      "grad_norm": 2.5644226304832,
      "learning_rate": 1.9833501894164886e-06,
      "loss": 0.8532,
      "step": 5739
    },
    {
      "epoch": 0.5176534247193038,
      "grad_norm": 1.5579037776365219,
      "learning_rate": 1.982765999751273e-06,
      "loss": 0.8001,
      "step": 5740
    },
    {
      "epoch": 0.5177436082427741,
      "grad_norm": 1.8026283042839868,
      "learning_rate": 1.9821818115565553e-06,
      "loss": 0.7948,
      "step": 5741
    },
    {
      "epoch": 0.5178337917662443,
      "grad_norm": 2.4923379786871007,
      "learning_rate": 1.9815976248821853e-06,
      "loss": 0.666,
      "step": 5742
    },
    {
      "epoch": 0.5179239752897146,
      "grad_norm": 6.664568793505634,
      "learning_rate": 1.981013439778007e-06,
      "loss": 0.7736,
      "step": 5743
    },
    {
      "epoch": 0.5180141588131848,
      "grad_norm": 1.6828222077921156,
      "learning_rate": 1.9804292562938666e-06,
      "loss": 0.7731,
      "step": 5744
    },
    {
      "epoch": 0.5181043423366551,
      "grad_norm": 1.6497211249527468,
      "learning_rate": 1.97984507447961e-06,
      "loss": 0.7955,
      "step": 5745
    },
    {
      "epoch": 0.5181945258601254,
      "grad_norm": 1.6948271025751538,
      "learning_rate": 1.9792608943850824e-06,
      "loss": 0.7009,
      "step": 5746
    },
    {
      "epoch": 0.5182847093835956,
      "grad_norm": 1.626598452299895,
      "learning_rate": 1.9786767160601305e-06,
      "loss": 0.741,
      "step": 5747
    },
    {
      "epoch": 0.5183748929070658,
      "grad_norm": 2.0043115633521302,
      "learning_rate": 1.9780925395545977e-06,
      "loss": 0.6939,
      "step": 5748
    },
    {
      "epoch": 0.5184650764305362,
      "grad_norm": 0.6360342406751408,
      "learning_rate": 1.9775083649183306e-06,
      "loss": 0.5669,
      "step": 5749
    },
    {
      "epoch": 0.5185552599540064,
      "grad_norm": 1.9338117996726942,
      "learning_rate": 1.976924192201174e-06,
      "loss": 0.8018,
      "step": 5750
    },
    {
      "epoch": 0.5186454434774767,
      "grad_norm": 1.5212085557117745,
      "learning_rate": 1.9763400214529723e-06,
      "loss": 0.6905,
      "step": 5751
    },
    {
      "epoch": 0.5187356270009469,
      "grad_norm": 1.5559154625853027,
      "learning_rate": 1.9757558527235713e-06,
      "loss": 0.6348,
      "step": 5752
    },
    {
      "epoch": 0.5188258105244172,
      "grad_norm": 1.5695410964072134,
      "learning_rate": 1.9751716860628136e-06,
      "loss": 0.7499,
      "step": 5753
    },
    {
      "epoch": 0.5189159940478875,
      "grad_norm": 2.3191287942038117,
      "learning_rate": 1.974587521520545e-06,
      "loss": 0.7879,
      "step": 5754
    },
    {
      "epoch": 0.5190061775713577,
      "grad_norm": 1.704646489588009,
      "learning_rate": 1.9740033591466088e-06,
      "loss": 0.7954,
      "step": 5755
    },
    {
      "epoch": 0.5190963610948279,
      "grad_norm": 1.715535295986685,
      "learning_rate": 1.97341919899085e-06,
      "loss": 0.8776,
      "step": 5756
    },
    {
      "epoch": 0.5191865446182983,
      "grad_norm": 1.617934565825393,
      "learning_rate": 1.9728350411031114e-06,
      "loss": 0.7847,
      "step": 5757
    },
    {
      "epoch": 0.5192767281417685,
      "grad_norm": 1.7136224729583471,
      "learning_rate": 1.9722508855332367e-06,
      "loss": 0.8372,
      "step": 5758
    },
    {
      "epoch": 0.5193669116652387,
      "grad_norm": 1.40091060417748,
      "learning_rate": 1.97166673233107e-06,
      "loss": 0.6913,
      "step": 5759
    },
    {
      "epoch": 0.519457095188709,
      "grad_norm": 5.734005450988244,
      "learning_rate": 1.971082581546453e-06,
      "loss": 0.8027,
      "step": 5760
    },
    {
      "epoch": 0.5195472787121793,
      "grad_norm": 1.7076310908652443,
      "learning_rate": 1.9704984332292306e-06,
      "loss": 0.7861,
      "step": 5761
    },
    {
      "epoch": 0.5196374622356495,
      "grad_norm": 1.6792871752234986,
      "learning_rate": 1.9699142874292444e-06,
      "loss": 0.8514,
      "step": 5762
    },
    {
      "epoch": 0.5197276457591198,
      "grad_norm": 1.5258353118772543,
      "learning_rate": 1.969330144196336e-06,
      "loss": 0.8542,
      "step": 5763
    },
    {
      "epoch": 0.51981782928259,
      "grad_norm": 2.126608408166814,
      "learning_rate": 1.9687460035803497e-06,
      "loss": 0.7083,
      "step": 5764
    },
    {
      "epoch": 0.5199080128060604,
      "grad_norm": 1.8100954087544516,
      "learning_rate": 1.9681618656311265e-06,
      "loss": 0.744,
      "step": 5765
    },
    {
      "epoch": 0.5199981963295306,
      "grad_norm": 1.5300922298682116,
      "learning_rate": 1.9675777303985086e-06,
      "loss": 0.6743,
      "step": 5766
    },
    {
      "epoch": 0.5200883798530008,
      "grad_norm": 2.8866957894574865,
      "learning_rate": 1.9669935979323376e-06,
      "loss": 0.7928,
      "step": 5767
    },
    {
      "epoch": 0.5201785633764712,
      "grad_norm": 1.6061759931681407,
      "learning_rate": 1.9664094682824545e-06,
      "loss": 0.8197,
      "step": 5768
    },
    {
      "epoch": 0.5202687468999414,
      "grad_norm": 9.229587611350029,
      "learning_rate": 1.965825341498701e-06,
      "loss": 0.7213,
      "step": 5769
    },
    {
      "epoch": 0.5203589304234116,
      "grad_norm": 2.7300756168722384,
      "learning_rate": 1.9652412176309177e-06,
      "loss": 0.8373,
      "step": 5770
    },
    {
      "epoch": 0.5204491139468819,
      "grad_norm": 1.5593188900685717,
      "learning_rate": 1.9646570967289453e-06,
      "loss": 0.631,
      "step": 5771
    },
    {
      "epoch": 0.5205392974703522,
      "grad_norm": 1.4999996821085275,
      "learning_rate": 1.9640729788426246e-06,
      "loss": 0.7977,
      "step": 5772
    },
    {
      "epoch": 0.5206294809938224,
      "grad_norm": 1.4494613074316034,
      "learning_rate": 1.963488864021795e-06,
      "loss": 0.7935,
      "step": 5773
    },
    {
      "epoch": 0.5207196645172927,
      "grad_norm": 2.0553214252006113,
      "learning_rate": 1.962904752316298e-06,
      "loss": 0.6552,
      "step": 5774
    },
    {
      "epoch": 0.5208098480407629,
      "grad_norm": 8.089018038649057,
      "learning_rate": 1.9623206437759706e-06,
      "loss": 0.6927,
      "step": 5775
    },
    {
      "epoch": 0.5209000315642333,
      "grad_norm": 1.443397862149675,
      "learning_rate": 1.9617365384506545e-06,
      "loss": 0.7205,
      "step": 5776
    },
    {
      "epoch": 0.5209902150877035,
      "grad_norm": 1.3604879536822672,
      "learning_rate": 1.9611524363901872e-06,
      "loss": 0.7489,
      "step": 5777
    },
    {
      "epoch": 0.5210803986111737,
      "grad_norm": 2.7995059463048113,
      "learning_rate": 1.960568337644409e-06,
      "loss": 0.7505,
      "step": 5778
    },
    {
      "epoch": 0.521170582134644,
      "grad_norm": 1.428803551071641,
      "learning_rate": 1.9599842422631576e-06,
      "loss": 0.7463,
      "step": 5779
    },
    {
      "epoch": 0.5212607656581143,
      "grad_norm": 1.6059934774910054,
      "learning_rate": 1.9594001502962703e-06,
      "loss": 0.7565,
      "step": 5780
    },
    {
      "epoch": 0.5213509491815845,
      "grad_norm": 1.3242154191687,
      "learning_rate": 1.9588160617935868e-06,
      "loss": 0.8207,
      "step": 5781
    },
    {
      "epoch": 0.5214411327050548,
      "grad_norm": 1.5767023453630724,
      "learning_rate": 1.958231976804944e-06,
      "loss": 0.8477,
      "step": 5782
    },
    {
      "epoch": 0.521531316228525,
      "grad_norm": 2.162242688873559,
      "learning_rate": 1.957647895380179e-06,
      "loss": 0.6913,
      "step": 5783
    },
    {
      "epoch": 0.5216214997519953,
      "grad_norm": 1.3129410683799545,
      "learning_rate": 1.9570638175691297e-06,
      "loss": 0.6879,
      "step": 5784
    },
    {
      "epoch": 0.5217116832754656,
      "grad_norm": 2.707823534203977,
      "learning_rate": 1.956479743421632e-06,
      "loss": 0.8019,
      "step": 5785
    },
    {
      "epoch": 0.5218018667989358,
      "grad_norm": 1.5484566357274172,
      "learning_rate": 1.955895672987522e-06,
      "loss": 0.7013,
      "step": 5786
    },
    {
      "epoch": 0.521892050322406,
      "grad_norm": 1.6485383784807974,
      "learning_rate": 1.9553116063166367e-06,
      "loss": 0.6675,
      "step": 5787
    },
    {
      "epoch": 0.5219822338458764,
      "grad_norm": 2.5235762897506486,
      "learning_rate": 1.954727543458812e-06,
      "loss": 0.6662,
      "step": 5788
    },
    {
      "epoch": 0.5220724173693466,
      "grad_norm": 2.956706786401376,
      "learning_rate": 1.954143484463883e-06,
      "loss": 0.7902,
      "step": 5789
    },
    {
      "epoch": 0.5221626008928169,
      "grad_norm": 1.628119848320072,
      "learning_rate": 1.9535594293816836e-06,
      "loss": 0.7448,
      "step": 5790
    },
    {
      "epoch": 0.5222527844162872,
      "grad_norm": 1.9840150041012854,
      "learning_rate": 1.952975378262051e-06,
      "loss": 0.7103,
      "step": 5791
    },
    {
      "epoch": 0.5223429679397574,
      "grad_norm": 2.0744082836619793,
      "learning_rate": 1.952391331154817e-06,
      "loss": 0.7161,
      "step": 5792
    },
    {
      "epoch": 0.5224331514632277,
      "grad_norm": 1.5689377721745157,
      "learning_rate": 1.9518072881098185e-06,
      "loss": 0.8131,
      "step": 5793
    },
    {
      "epoch": 0.5225233349866979,
      "grad_norm": 1.507735966929117,
      "learning_rate": 1.9512232491768867e-06,
      "loss": 0.7495,
      "step": 5794
    },
    {
      "epoch": 0.5226135185101682,
      "grad_norm": 1.4567389719205057,
      "learning_rate": 1.9506392144058573e-06,
      "loss": 0.7469,
      "step": 5795
    },
    {
      "epoch": 0.5227037020336385,
      "grad_norm": 2.0611741399367602,
      "learning_rate": 1.9500551838465623e-06,
      "loss": 0.7546,
      "step": 5796
    },
    {
      "epoch": 0.5227938855571087,
      "grad_norm": 1.4691628220901487,
      "learning_rate": 1.9494711575488337e-06,
      "loss": 0.7712,
      "step": 5797
    },
    {
      "epoch": 0.5228840690805789,
      "grad_norm": 1.8126424536136005,
      "learning_rate": 1.948887135562505e-06,
      "loss": 0.7031,
      "step": 5798
    },
    {
      "epoch": 0.5229742526040493,
      "grad_norm": 1.4611362587263863,
      "learning_rate": 1.9483031179374074e-06,
      "loss": 0.8054,
      "step": 5799
    },
    {
      "epoch": 0.5230644361275195,
      "grad_norm": 1.9466184346002524,
      "learning_rate": 1.9477191047233736e-06,
      "loss": 0.8165,
      "step": 5800
    },
    {
      "epoch": 0.5231546196509897,
      "grad_norm": 1.8637043534540934,
      "learning_rate": 1.9471350959702334e-06,
      "loss": 0.6406,
      "step": 5801
    },
    {
      "epoch": 0.52324480317446,
      "grad_norm": 3.4068866099844026,
      "learning_rate": 1.9465510917278184e-06,
      "loss": 0.7631,
      "step": 5802
    },
    {
      "epoch": 0.5233349866979303,
      "grad_norm": 1.791363498262456,
      "learning_rate": 1.9459670920459593e-06,
      "loss": 0.7524,
      "step": 5803
    },
    {
      "epoch": 0.5234251702214006,
      "grad_norm": 2.0380077922848026,
      "learning_rate": 1.945383096974485e-06,
      "loss": 0.6696,
      "step": 5804
    },
    {
      "epoch": 0.5235153537448708,
      "grad_norm": 2.4563978194390605,
      "learning_rate": 1.944799106563227e-06,
      "loss": 0.6829,
      "step": 5805
    },
    {
      "epoch": 0.523605537268341,
      "grad_norm": 3.5333278248102205,
      "learning_rate": 1.9442151208620133e-06,
      "loss": 0.7941,
      "step": 5806
    },
    {
      "epoch": 0.5236957207918114,
      "grad_norm": 1.8733566712162242,
      "learning_rate": 1.943631139920672e-06,
      "loss": 0.8017,
      "step": 5807
    },
    {
      "epoch": 0.5237859043152816,
      "grad_norm": 9.38660699124585,
      "learning_rate": 1.943047163789034e-06,
      "loss": 0.7108,
      "step": 5808
    },
    {
      "epoch": 0.5238760878387518,
      "grad_norm": 1.5128158018962252,
      "learning_rate": 1.942463192516925e-06,
      "loss": 0.7165,
      "step": 5809
    },
    {
      "epoch": 0.5239662713622221,
      "grad_norm": 1.7049575885408834,
      "learning_rate": 1.9418792261541746e-06,
      "loss": 0.7787,
      "step": 5810
    },
    {
      "epoch": 0.5240564548856924,
      "grad_norm": 1.6546907102501467,
      "learning_rate": 1.9412952647506094e-06,
      "loss": 0.7178,
      "step": 5811
    },
    {
      "epoch": 0.5241466384091626,
      "grad_norm": 1.568766805855145,
      "learning_rate": 1.9407113083560552e-06,
      "loss": 0.7301,
      "step": 5812
    },
    {
      "epoch": 0.5242368219326329,
      "grad_norm": 1.723998431366563,
      "learning_rate": 1.940127357020339e-06,
      "loss": 0.7373,
      "step": 5813
    },
    {
      "epoch": 0.5243270054561032,
      "grad_norm": 2.9664958226020035,
      "learning_rate": 1.939543410793287e-06,
      "loss": 0.7668,
      "step": 5814
    },
    {
      "epoch": 0.5244171889795735,
      "grad_norm": 2.1055679439469883,
      "learning_rate": 1.9389594697247246e-06,
      "loss": 0.7918,
      "step": 5815
    },
    {
      "epoch": 0.5245073725030437,
      "grad_norm": 2.9201171809682465,
      "learning_rate": 1.9383755338644763e-06,
      "loss": 0.8293,
      "step": 5816
    },
    {
      "epoch": 0.5245975560265139,
      "grad_norm": 1.970775349619589,
      "learning_rate": 1.937791603262368e-06,
      "loss": 0.7908,
      "step": 5817
    },
    {
      "epoch": 0.5246877395499843,
      "grad_norm": 2.672178463127051,
      "learning_rate": 1.9372076779682235e-06,
      "loss": 0.7079,
      "step": 5818
    },
    {
      "epoch": 0.5247779230734545,
      "grad_norm": 2.110576372295259,
      "learning_rate": 1.9366237580318648e-06,
      "loss": 0.7083,
      "step": 5819
    },
    {
      "epoch": 0.5248681065969247,
      "grad_norm": 1.3470007145178131,
      "learning_rate": 1.9360398435031176e-06,
      "loss": 0.7601,
      "step": 5820
    },
    {
      "epoch": 0.524958290120395,
      "grad_norm": 3.4853804296664594,
      "learning_rate": 1.9354559344318025e-06,
      "loss": 0.7708,
      "step": 5821
    },
    {
      "epoch": 0.5250484736438653,
      "grad_norm": 1.9653581131809705,
      "learning_rate": 1.934872030867744e-06,
      "loss": 0.8097,
      "step": 5822
    },
    {
      "epoch": 0.5251386571673355,
      "grad_norm": 1.8186303089517168,
      "learning_rate": 1.934288132860763e-06,
      "loss": 0.7693,
      "step": 5823
    },
    {
      "epoch": 0.5252288406908058,
      "grad_norm": 1.527837492999186,
      "learning_rate": 1.93370424046068e-06,
      "loss": 0.7568,
      "step": 5824
    },
    {
      "epoch": 0.525319024214276,
      "grad_norm": 4.369689988721646,
      "learning_rate": 1.9331203537173177e-06,
      "loss": 0.819,
      "step": 5825
    },
    {
      "epoch": 0.5254092077377464,
      "grad_norm": 1.3366689631489141,
      "learning_rate": 1.9325364726804947e-06,
      "loss": 0.8703,
      "step": 5826
    },
    {
      "epoch": 0.5254993912612166,
      "grad_norm": 1.8815325426606861,
      "learning_rate": 1.9319525974000327e-06,
      "loss": 0.6847,
      "step": 5827
    },
    {
      "epoch": 0.5255895747846868,
      "grad_norm": 12.782328590453929,
      "learning_rate": 1.93136872792575e-06,
      "loss": 0.8141,
      "step": 5828
    },
    {
      "epoch": 0.525679758308157,
      "grad_norm": 1.5174791458431471,
      "learning_rate": 1.9307848643074653e-06,
      "loss": 0.7107,
      "step": 5829
    },
    {
      "epoch": 0.5257699418316274,
      "grad_norm": 1.7289992868081117,
      "learning_rate": 1.9302010065949985e-06,
      "loss": 0.7931,
      "step": 5830
    },
    {
      "epoch": 0.5258601253550976,
      "grad_norm": 3.5088844615475367,
      "learning_rate": 1.9296171548381657e-06,
      "loss": 0.7167,
      "step": 5831
    },
    {
      "epoch": 0.5259503088785679,
      "grad_norm": 1.581665072036652,
      "learning_rate": 1.9290333090867862e-06,
      "loss": 0.7449,
      "step": 5832
    },
    {
      "epoch": 0.5260404924020381,
      "grad_norm": 2.1622872352671076,
      "learning_rate": 1.928449469390676e-06,
      "loss": 0.7606,
      "step": 5833
    },
    {
      "epoch": 0.5261306759255084,
      "grad_norm": 1.5939060209610871,
      "learning_rate": 1.927865635799651e-06,
      "loss": 0.7272,
      "step": 5834
    },
    {
      "epoch": 0.5262208594489787,
      "grad_norm": 1.5013911630803631,
      "learning_rate": 1.927281808363528e-06,
      "loss": 0.6871,
      "step": 5835
    },
    {
      "epoch": 0.5263110429724489,
      "grad_norm": 1.5797493761208992,
      "learning_rate": 1.9266979871321216e-06,
      "loss": 0.8295,
      "step": 5836
    },
    {
      "epoch": 0.5264012264959192,
      "grad_norm": 1.5951001301490186,
      "learning_rate": 1.9261141721552482e-06,
      "loss": 0.7066,
      "step": 5837
    },
    {
      "epoch": 0.5264914100193895,
      "grad_norm": 1.682591327948958,
      "learning_rate": 1.9255303634827204e-06,
      "loss": 0.8234,
      "step": 5838
    },
    {
      "epoch": 0.5265815935428597,
      "grad_norm": 1.670247221367861,
      "learning_rate": 1.924946561164352e-06,
      "loss": 0.8184,
      "step": 5839
    },
    {
      "epoch": 0.52667177706633,
      "grad_norm": 1.7417111280080644,
      "learning_rate": 1.9243627652499582e-06,
      "loss": 0.7091,
      "step": 5840
    },
    {
      "epoch": 0.5267619605898003,
      "grad_norm": 2.043612493619661,
      "learning_rate": 1.9237789757893493e-06,
      "loss": 0.8176,
      "step": 5841
    },
    {
      "epoch": 0.5268521441132705,
      "grad_norm": 35.246043314618134,
      "learning_rate": 1.9231951928323395e-06,
      "loss": 0.7997,
      "step": 5842
    },
    {
      "epoch": 0.5269423276367408,
      "grad_norm": 1.673727061404359,
      "learning_rate": 1.922611416428738e-06,
      "loss": 0.7113,
      "step": 5843
    },
    {
      "epoch": 0.527032511160211,
      "grad_norm": 1.5082252214300553,
      "learning_rate": 1.922027646628358e-06,
      "loss": 0.7794,
      "step": 5844
    },
    {
      "epoch": 0.5271226946836813,
      "grad_norm": 0.6695421438321302,
      "learning_rate": 1.9214438834810092e-06,
      "loss": 0.6355,
      "step": 5845
    },
    {
      "epoch": 0.5272128782071516,
      "grad_norm": 3.361433525139699,
      "learning_rate": 1.9208601270365008e-06,
      "loss": 0.6846,
      "step": 5846
    },
    {
      "epoch": 0.5273030617306218,
      "grad_norm": 1.5493177142830517,
      "learning_rate": 1.9202763773446435e-06,
      "loss": 0.8344,
      "step": 5847
    },
    {
      "epoch": 0.527393245254092,
      "grad_norm": 1.5963436041311914,
      "learning_rate": 1.9196926344552444e-06,
      "loss": 0.6519,
      "step": 5848
    },
    {
      "epoch": 0.5274834287775624,
      "grad_norm": 1.7580908660884005,
      "learning_rate": 1.919108898418113e-06,
      "loss": 0.783,
      "step": 5849
    },
    {
      "epoch": 0.5275736123010326,
      "grad_norm": 1.390243906672816,
      "learning_rate": 1.918525169283057e-06,
      "loss": 0.6549,
      "step": 5850
    },
    {
      "epoch": 0.5276637958245028,
      "grad_norm": 1.8130888804401881,
      "learning_rate": 1.9179414470998817e-06,
      "loss": 0.7836,
      "step": 5851
    },
    {
      "epoch": 0.5277539793479731,
      "grad_norm": 1.5768360879748546,
      "learning_rate": 1.917357731918395e-06,
      "loss": 0.7171,
      "step": 5852
    },
    {
      "epoch": 0.5278441628714434,
      "grad_norm": 1.5751897122401222,
      "learning_rate": 1.9167740237884025e-06,
      "loss": 0.8878,
      "step": 5853
    },
    {
      "epoch": 0.5279343463949137,
      "grad_norm": 1.58128275177702,
      "learning_rate": 1.916190322759709e-06,
      "loss": 0.7933,
      "step": 5854
    },
    {
      "epoch": 0.5280245299183839,
      "grad_norm": 1.6153038960155932,
      "learning_rate": 1.91560662888212e-06,
      "loss": 0.86,
      "step": 5855
    },
    {
      "epoch": 0.5281147134418541,
      "grad_norm": 2.2988102862784925,
      "learning_rate": 1.915022942205438e-06,
      "loss": 0.8027,
      "step": 5856
    },
    {
      "epoch": 0.5282048969653245,
      "grad_norm": 2.212531280296458,
      "learning_rate": 1.914439262779468e-06,
      "loss": 0.7501,
      "step": 5857
    },
    {
      "epoch": 0.5282950804887947,
      "grad_norm": 1.9278956445334217,
      "learning_rate": 1.9138555906540103e-06,
      "loss": 0.6915,
      "step": 5858
    },
    {
      "epoch": 0.5283852640122649,
      "grad_norm": 2.7196419666769374,
      "learning_rate": 1.91327192587887e-06,
      "loss": 0.6654,
      "step": 5859
    },
    {
      "epoch": 0.5284754475357353,
      "grad_norm": 21.927741942553475,
      "learning_rate": 1.912688268503846e-06,
      "loss": 0.7248,
      "step": 5860
    },
    {
      "epoch": 0.5285656310592055,
      "grad_norm": 1.788655130574981,
      "learning_rate": 1.912104618578741e-06,
      "loss": 0.7239,
      "step": 5861
    },
    {
      "epoch": 0.5286558145826757,
      "grad_norm": 1.748596650517369,
      "learning_rate": 1.9115209761533554e-06,
      "loss": 0.7314,
      "step": 5862
    },
    {
      "epoch": 0.528745998106146,
      "grad_norm": 1.7859405197158076,
      "learning_rate": 1.9109373412774863e-06,
      "loss": 0.6695,
      "step": 5863
    },
    {
      "epoch": 0.5288361816296163,
      "grad_norm": 1.4292833768087851,
      "learning_rate": 1.910353714000936e-06,
      "loss": 0.7564,
      "step": 5864
    },
    {
      "epoch": 0.5289263651530866,
      "grad_norm": 1.5927185760162041,
      "learning_rate": 1.9097700943734997e-06,
      "loss": 0.7168,
      "step": 5865
    },
    {
      "epoch": 0.5290165486765568,
      "grad_norm": 1.525467835253842,
      "learning_rate": 1.909186482444977e-06,
      "loss": 0.655,
      "step": 5866
    },
    {
      "epoch": 0.529106732200027,
      "grad_norm": 2.2186572633087045,
      "learning_rate": 1.9086028782651652e-06,
      "loss": 0.6555,
      "step": 5867
    },
    {
      "epoch": 0.5291969157234974,
      "grad_norm": 1.475442209066669,
      "learning_rate": 1.908019281883859e-06,
      "loss": 0.747,
      "step": 5868
    },
    {
      "epoch": 0.5292870992469676,
      "grad_norm": 1.6403594029109547,
      "learning_rate": 1.9074356933508545e-06,
      "loss": 0.7465,
      "step": 5869
    },
    {
      "epoch": 0.5293772827704378,
      "grad_norm": 2.842090331980875,
      "learning_rate": 1.9068521127159477e-06,
      "loss": 0.7904,
      "step": 5870
    },
    {
      "epoch": 0.5294674662939081,
      "grad_norm": 1.9684201524318008,
      "learning_rate": 1.9062685400289322e-06,
      "loss": 0.7054,
      "step": 5871
    },
    {
      "epoch": 0.5295576498173784,
      "grad_norm": 1.660435336720774,
      "learning_rate": 1.9056849753396018e-06,
      "loss": 0.7287,
      "step": 5872
    },
    {
      "epoch": 0.5296478333408486,
      "grad_norm": 2.140010327822283,
      "learning_rate": 1.9051014186977485e-06,
      "loss": 0.7903,
      "step": 5873
    },
    {
      "epoch": 0.5297380168643189,
      "grad_norm": 1.5433569311201307,
      "learning_rate": 1.9045178701531664e-06,
      "loss": 0.7569,
      "step": 5874
    },
    {
      "epoch": 0.5298282003877891,
      "grad_norm": 1.3837560247274603,
      "learning_rate": 1.903934329755645e-06,
      "loss": 0.7481,
      "step": 5875
    },
    {
      "epoch": 0.5299183839112594,
      "grad_norm": 2.1015598906887263,
      "learning_rate": 1.9033507975549775e-06,
      "loss": 0.7899,
      "step": 5876
    },
    {
      "epoch": 0.5300085674347297,
      "grad_norm": 1.790181505495065,
      "learning_rate": 1.9027672736009525e-06,
      "loss": 0.746,
      "step": 5877
    },
    {
      "epoch": 0.5300987509581999,
      "grad_norm": 1.4657864804973102,
      "learning_rate": 1.9021837579433593e-06,
      "loss": 0.7021,
      "step": 5878
    },
    {
      "epoch": 0.5301889344816701,
      "grad_norm": 2.539963219143547,
      "learning_rate": 1.901600250631988e-06,
      "loss": 0.7285,
      "step": 5879
    },
    {
      "epoch": 0.5302791180051405,
      "grad_norm": 3.0806533576277744,
      "learning_rate": 1.901016751716625e-06,
      "loss": 0.7442,
      "step": 5880
    },
    {
      "epoch": 0.5303693015286107,
      "grad_norm": 7.103232343729431,
      "learning_rate": 1.9004332612470593e-06,
      "loss": 0.7177,
      "step": 5881
    },
    {
      "epoch": 0.530459485052081,
      "grad_norm": 1.446844495282846,
      "learning_rate": 1.8998497792730763e-06,
      "loss": 0.7811,
      "step": 5882
    },
    {
      "epoch": 0.5305496685755512,
      "grad_norm": 1.5265531322929775,
      "learning_rate": 1.8992663058444629e-06,
      "loss": 0.8027,
      "step": 5883
    },
    {
      "epoch": 0.5306398520990215,
      "grad_norm": 1.611814039541204,
      "learning_rate": 1.8986828410110032e-06,
      "loss": 0.7887,
      "step": 5884
    },
    {
      "epoch": 0.5307300356224918,
      "grad_norm": 2.0390574721935737,
      "learning_rate": 1.8980993848224823e-06,
      "loss": 0.7515,
      "step": 5885
    },
    {
      "epoch": 0.530820219145962,
      "grad_norm": 1.9546629077026076,
      "learning_rate": 1.8975159373286843e-06,
      "loss": 0.7124,
      "step": 5886
    },
    {
      "epoch": 0.5309104026694323,
      "grad_norm": 1.7468628693570594,
      "learning_rate": 1.8969324985793904e-06,
      "loss": 0.7914,
      "step": 5887
    },
    {
      "epoch": 0.5310005861929026,
      "grad_norm": 3.20708269456683,
      "learning_rate": 1.8963490686243851e-06,
      "loss": 0.7443,
      "step": 5888
    },
    {
      "epoch": 0.5310907697163728,
      "grad_norm": 2.1243140291519764,
      "learning_rate": 1.8957656475134486e-06,
      "loss": 0.6898,
      "step": 5889
    },
    {
      "epoch": 0.531180953239843,
      "grad_norm": 2.0078323303905807,
      "learning_rate": 1.895182235296361e-06,
      "loss": 0.7341,
      "step": 5890
    },
    {
      "epoch": 0.5312711367633134,
      "grad_norm": 3.4891230644986297,
      "learning_rate": 1.8945988320229042e-06,
      "loss": 0.6786,
      "step": 5891
    },
    {
      "epoch": 0.5313613202867836,
      "grad_norm": 1.8542777324658386,
      "learning_rate": 1.8940154377428553e-06,
      "loss": 0.8346,
      "step": 5892
    },
    {
      "epoch": 0.5314515038102539,
      "grad_norm": 1.5505312470673798,
      "learning_rate": 1.8934320525059944e-06,
      "loss": 0.7803,
      "step": 5893
    },
    {
      "epoch": 0.5315416873337241,
      "grad_norm": 2.0409221719035684,
      "learning_rate": 1.8928486763620984e-06,
      "loss": 0.6442,
      "step": 5894
    },
    {
      "epoch": 0.5316318708571944,
      "grad_norm": 1.6512589737838177,
      "learning_rate": 1.892265309360943e-06,
      "loss": 0.767,
      "step": 5895
    },
    {
      "epoch": 0.5317220543806647,
      "grad_norm": 1.7134106324096734,
      "learning_rate": 1.8916819515523067e-06,
      "loss": 0.7389,
      "step": 5896
    },
    {
      "epoch": 0.5318122379041349,
      "grad_norm": 8.03638955806065,
      "learning_rate": 1.891098602985963e-06,
      "loss": 0.6696,
      "step": 5897
    },
    {
      "epoch": 0.5319024214276051,
      "grad_norm": 1.9443209162332549,
      "learning_rate": 1.8905152637116872e-06,
      "loss": 0.7761,
      "step": 5898
    },
    {
      "epoch": 0.5319926049510755,
      "grad_norm": 1.9355181740981933,
      "learning_rate": 1.8899319337792527e-06,
      "loss": 0.7906,
      "step": 5899
    },
    {
      "epoch": 0.5320827884745457,
      "grad_norm": 1.3015040316964426,
      "learning_rate": 1.8893486132384325e-06,
      "loss": 0.7909,
      "step": 5900
    },
    {
      "epoch": 0.5321729719980159,
      "grad_norm": 1.9135160327814846,
      "learning_rate": 1.888765302138999e-06,
      "loss": 0.7235,
      "step": 5901
    },
    {
      "epoch": 0.5322631555214862,
      "grad_norm": 2.1612939856373763,
      "learning_rate": 1.8881820005307224e-06,
      "loss": 0.6986,
      "step": 5902
    },
    {
      "epoch": 0.5323533390449565,
      "grad_norm": 1.6046635077149176,
      "learning_rate": 1.8875987084633748e-06,
      "loss": 0.6863,
      "step": 5903
    },
    {
      "epoch": 0.5324435225684268,
      "grad_norm": 3.137028243020736,
      "learning_rate": 1.8870154259867246e-06,
      "loss": 0.7887,
      "step": 5904
    },
    {
      "epoch": 0.532533706091897,
      "grad_norm": 1.7355776904055875,
      "learning_rate": 1.886432153150542e-06,
      "loss": 0.693,
      "step": 5905
    },
    {
      "epoch": 0.5326238896153672,
      "grad_norm": 1.580087932843865,
      "learning_rate": 1.8858488900045944e-06,
      "loss": 0.7313,
      "step": 5906
    },
    {
      "epoch": 0.5327140731388376,
      "grad_norm": 2.41562342871927,
      "learning_rate": 1.885265636598648e-06,
      "loss": 0.7033,
      "step": 5907
    },
    {
      "epoch": 0.5328042566623078,
      "grad_norm": 1.5036875379976125,
      "learning_rate": 1.884682392982471e-06,
      "loss": 0.7839,
      "step": 5908
    },
    {
      "epoch": 0.532894440185778,
      "grad_norm": 2.40809016028517,
      "learning_rate": 1.8840991592058274e-06,
      "loss": 0.805,
      "step": 5909
    },
    {
      "epoch": 0.5329846237092484,
      "grad_norm": 1.572976808050828,
      "learning_rate": 1.8835159353184828e-06,
      "loss": 0.7691,
      "step": 5910
    },
    {
      "epoch": 0.5330748072327186,
      "grad_norm": 1.6769228872313893,
      "learning_rate": 1.8829327213702013e-06,
      "loss": 0.8209,
      "step": 5911
    },
    {
      "epoch": 0.5331649907561888,
      "grad_norm": 1.694608832377649,
      "learning_rate": 1.8823495174107452e-06,
      "loss": 0.826,
      "step": 5912
    },
    {
      "epoch": 0.5332551742796591,
      "grad_norm": 4.3309497635613585,
      "learning_rate": 1.8817663234898773e-06,
      "loss": 0.7284,
      "step": 5913
    },
    {
      "epoch": 0.5333453578031294,
      "grad_norm": 1.9641141655559446,
      "learning_rate": 1.881183139657358e-06,
      "loss": 0.7631,
      "step": 5914
    },
    {
      "epoch": 0.5334355413265996,
      "grad_norm": 1.7193701578929725,
      "learning_rate": 1.8805999659629488e-06,
      "loss": 0.769,
      "step": 5915
    },
    {
      "epoch": 0.5335257248500699,
      "grad_norm": 0.6198778787058753,
      "learning_rate": 1.880016802456409e-06,
      "loss": 0.5363,
      "step": 5916
    },
    {
      "epoch": 0.5336159083735401,
      "grad_norm": 1.92277623576824,
      "learning_rate": 1.8794336491874964e-06,
      "loss": 0.695,
      "step": 5917
    },
    {
      "epoch": 0.5337060918970105,
      "grad_norm": 2.0703220583137254,
      "learning_rate": 1.8788505062059708e-06,
      "loss": 0.7121,
      "step": 5918
    },
    {
      "epoch": 0.5337962754204807,
      "grad_norm": 1.6029142304413444,
      "learning_rate": 1.8782673735615869e-06,
      "loss": 0.7387,
      "step": 5919
    },
    {
      "epoch": 0.5338864589439509,
      "grad_norm": 1.6892530025078936,
      "learning_rate": 1.8776842513041026e-06,
      "loss": 0.7088,
      "step": 5920
    },
    {
      "epoch": 0.5339766424674212,
      "grad_norm": 2.1255019100075634,
      "learning_rate": 1.8771011394832727e-06,
      "loss": 0.7205,
      "step": 5921
    },
    {
      "epoch": 0.5340668259908915,
      "grad_norm": 9.847535002104626,
      "learning_rate": 1.8765180381488501e-06,
      "loss": 0.7333,
      "step": 5922
    },
    {
      "epoch": 0.5341570095143617,
      "grad_norm": 0.7320563052519546,
      "learning_rate": 1.8759349473505905e-06,
      "loss": 0.6678,
      "step": 5923
    },
    {
      "epoch": 0.534247193037832,
      "grad_norm": 1.8588513189842477,
      "learning_rate": 1.8753518671382447e-06,
      "loss": 0.676,
      "step": 5924
    },
    {
      "epoch": 0.5343373765613022,
      "grad_norm": 1.7058275810757202,
      "learning_rate": 1.8747687975615649e-06,
      "loss": 0.7318,
      "step": 5925
    },
    {
      "epoch": 0.5344275600847725,
      "grad_norm": 1.6091206914470226,
      "learning_rate": 1.874185738670302e-06,
      "loss": 0.6433,
      "step": 5926
    },
    {
      "epoch": 0.5345177436082428,
      "grad_norm": 2.1330576326119717,
      "learning_rate": 1.8736026905142057e-06,
      "loss": 0.739,
      "step": 5927
    },
    {
      "epoch": 0.534607927131713,
      "grad_norm": 1.5577758325606055,
      "learning_rate": 1.873019653143025e-06,
      "loss": 0.7284,
      "step": 5928
    },
    {
      "epoch": 0.5346981106551832,
      "grad_norm": 1.4655555724920915,
      "learning_rate": 1.8724366266065069e-06,
      "loss": 0.7613,
      "step": 5929
    },
    {
      "epoch": 0.5347882941786536,
      "grad_norm": 1.7653110191216101,
      "learning_rate": 1.8718536109543998e-06,
      "loss": 0.826,
      "step": 5930
    },
    {
      "epoch": 0.5348784777021238,
      "grad_norm": 2.080595455401084,
      "learning_rate": 1.8712706062364485e-06,
      "loss": 0.7672,
      "step": 5931
    },
    {
      "epoch": 0.534968661225594,
      "grad_norm": 1.7304750877247093,
      "learning_rate": 1.8706876125024e-06,
      "loss": 0.6389,
      "step": 5932
    },
    {
      "epoch": 0.5350588447490644,
      "grad_norm": 1.7900861451928238,
      "learning_rate": 1.870104629801997e-06,
      "loss": 0.7946,
      "step": 5933
    },
    {
      "epoch": 0.5351490282725346,
      "grad_norm": 1.7791354695547246,
      "learning_rate": 1.8695216581849823e-06,
      "loss": 0.7621,
      "step": 5934
    },
    {
      "epoch": 0.5352392117960049,
      "grad_norm": 1.6967179761421651,
      "learning_rate": 1.8689386977011003e-06,
      "loss": 0.7206,
      "step": 5935
    },
    {
      "epoch": 0.5353293953194751,
      "grad_norm": 0.6778095434536786,
      "learning_rate": 1.8683557484000903e-06,
      "loss": 0.576,
      "step": 5936
    },
    {
      "epoch": 0.5354195788429454,
      "grad_norm": 1.692647711817616,
      "learning_rate": 1.8677728103316947e-06,
      "loss": 0.7811,
      "step": 5937
    },
    {
      "epoch": 0.5355097623664157,
      "grad_norm": 1.4298001437811567,
      "learning_rate": 1.8671898835456518e-06,
      "loss": 0.7072,
      "step": 5938
    },
    {
      "epoch": 0.5355999458898859,
      "grad_norm": 2.7771539422054508,
      "learning_rate": 1.8666069680917003e-06,
      "loss": 0.8547,
      "step": 5939
    },
    {
      "epoch": 0.5356901294133561,
      "grad_norm": 2.8036510709604636,
      "learning_rate": 1.8660240640195775e-06,
      "loss": 0.7123,
      "step": 5940
    },
    {
      "epoch": 0.5357803129368265,
      "grad_norm": 1.4961108811794563,
      "learning_rate": 1.8654411713790203e-06,
      "loss": 0.7841,
      "step": 5941
    },
    {
      "epoch": 0.5358704964602967,
      "grad_norm": 1.8703644352508537,
      "learning_rate": 1.8648582902197648e-06,
      "loss": 0.773,
      "step": 5942
    },
    {
      "epoch": 0.535960679983767,
      "grad_norm": 1.786239862753611,
      "learning_rate": 1.8642754205915452e-06,
      "loss": 0.7581,
      "step": 5943
    },
    {
      "epoch": 0.5360508635072372,
      "grad_norm": 1.5001644203353715,
      "learning_rate": 1.8636925625440943e-06,
      "loss": 0.7007,
      "step": 5944
    },
    {
      "epoch": 0.5361410470307075,
      "grad_norm": 2.0795126276130422,
      "learning_rate": 1.863109716127146e-06,
      "loss": 0.6577,
      "step": 5945
    },
    {
      "epoch": 0.5362312305541778,
      "grad_norm": 1.4673300026944351,
      "learning_rate": 1.8625268813904311e-06,
      "loss": 0.795,
      "step": 5946
    },
    {
      "epoch": 0.536321414077648,
      "grad_norm": 1.6361324744277137,
      "learning_rate": 1.8619440583836814e-06,
      "loss": 0.7753,
      "step": 5947
    },
    {
      "epoch": 0.5364115976011182,
      "grad_norm": 1.6833881155233785,
      "learning_rate": 1.8613612471566249e-06,
      "loss": 0.7602,
      "step": 5948
    },
    {
      "epoch": 0.5365017811245886,
      "grad_norm": 1.8667932836414634,
      "learning_rate": 1.8607784477589922e-06,
      "loss": 0.698,
      "step": 5949
    },
    {
      "epoch": 0.5365919646480588,
      "grad_norm": 1.5624761961077425,
      "learning_rate": 1.8601956602405103e-06,
      "loss": 0.7664,
      "step": 5950
    },
    {
      "epoch": 0.536682148171529,
      "grad_norm": 1.5820185154531834,
      "learning_rate": 1.8596128846509043e-06,
      "loss": 0.794,
      "step": 5951
    },
    {
      "epoch": 0.5367723316949993,
      "grad_norm": 1.849135663479041,
      "learning_rate": 1.859030121039902e-06,
      "loss": 0.7085,
      "step": 5952
    },
    {
      "epoch": 0.5368625152184696,
      "grad_norm": 1.6039793805202593,
      "learning_rate": 1.8584473694572268e-06,
      "loss": 0.7055,
      "step": 5953
    },
    {
      "epoch": 0.5369526987419398,
      "grad_norm": 1.7359488280001396,
      "learning_rate": 1.8578646299526026e-06,
      "loss": 0.7701,
      "step": 5954
    },
    {
      "epoch": 0.5370428822654101,
      "grad_norm": 1.7011768503039688,
      "learning_rate": 1.8572819025757518e-06,
      "loss": 0.7211,
      "step": 5955
    },
    {
      "epoch": 0.5371330657888804,
      "grad_norm": 1.8232463720158194,
      "learning_rate": 1.8566991873763959e-06,
      "loss": 0.7664,
      "step": 5956
    },
    {
      "epoch": 0.5372232493123507,
      "grad_norm": 1.5991438303488597,
      "learning_rate": 1.856116484404256e-06,
      "loss": 0.6975,
      "step": 5957
    },
    {
      "epoch": 0.5373134328358209,
      "grad_norm": 1.808812632817523,
      "learning_rate": 1.8555337937090502e-06,
      "loss": 0.7228,
      "step": 5958
    },
    {
      "epoch": 0.5374036163592911,
      "grad_norm": 1.8207051913682022,
      "learning_rate": 1.8549511153404984e-06,
      "loss": 0.7358,
      "step": 5959
    },
    {
      "epoch": 0.5374937998827615,
      "grad_norm": 2.5686097229758484,
      "learning_rate": 1.854368449348317e-06,
      "loss": 0.8599,
      "step": 5960
    },
    {
      "epoch": 0.5375839834062317,
      "grad_norm": 2.9496643134037437,
      "learning_rate": 1.853785795782222e-06,
      "loss": 0.7122,
      "step": 5961
    },
    {
      "epoch": 0.5376741669297019,
      "grad_norm": 2.0845965815271894,
      "learning_rate": 1.85320315469193e-06,
      "loss": 0.6215,
      "step": 5962
    },
    {
      "epoch": 0.5377643504531722,
      "grad_norm": 0.6523635998292804,
      "learning_rate": 1.8526205261271534e-06,
      "loss": 0.5821,
      "step": 5963
    },
    {
      "epoch": 0.5378545339766425,
      "grad_norm": 3.1785109684124673,
      "learning_rate": 1.852037910137607e-06,
      "loss": 0.6678,
      "step": 5964
    },
    {
      "epoch": 0.5379447175001127,
      "grad_norm": 11.666779290518523,
      "learning_rate": 1.851455306773002e-06,
      "loss": 0.8357,
      "step": 5965
    },
    {
      "epoch": 0.538034901023583,
      "grad_norm": 1.3610271192227008,
      "learning_rate": 1.8508727160830483e-06,
      "loss": 0.7254,
      "step": 5966
    },
    {
      "epoch": 0.5381250845470532,
      "grad_norm": 2.18985131460102,
      "learning_rate": 1.8502901381174575e-06,
      "loss": 0.8199,
      "step": 5967
    },
    {
      "epoch": 0.5382152680705236,
      "grad_norm": 2.0092857091829197,
      "learning_rate": 1.8497075729259372e-06,
      "loss": 0.6585,
      "step": 5968
    },
    {
      "epoch": 0.5383054515939938,
      "grad_norm": 1.5977419804559967,
      "learning_rate": 1.8491250205581963e-06,
      "loss": 0.6424,
      "step": 5969
    },
    {
      "epoch": 0.538395635117464,
      "grad_norm": 1.4128448976726924,
      "learning_rate": 1.8485424810639393e-06,
      "loss": 0.7559,
      "step": 5970
    },
    {
      "epoch": 0.5384858186409343,
      "grad_norm": 2.0164516439431024,
      "learning_rate": 1.847959954492874e-06,
      "loss": 0.8107,
      "step": 5971
    },
    {
      "epoch": 0.5385760021644046,
      "grad_norm": 2.393399492303937,
      "learning_rate": 1.8473774408947035e-06,
      "loss": 0.8425,
      "step": 5972
    },
    {
      "epoch": 0.5386661856878748,
      "grad_norm": 0.6574421000642023,
      "learning_rate": 1.8467949403191308e-06,
      "loss": 0.5532,
      "step": 5973
    },
    {
      "epoch": 0.5387563692113451,
      "grad_norm": 2.1889013978543685,
      "learning_rate": 1.8462124528158592e-06,
      "loss": 0.8077,
      "step": 5974
    },
    {
      "epoch": 0.5388465527348153,
      "grad_norm": 1.6935117096452468,
      "learning_rate": 1.8456299784345881e-06,
      "loss": 0.7364,
      "step": 5975
    },
    {
      "epoch": 0.5389367362582856,
      "grad_norm": 1.7937767186318208,
      "learning_rate": 1.8450475172250194e-06,
      "loss": 0.7744,
      "step": 5976
    },
    {
      "epoch": 0.5390269197817559,
      "grad_norm": 1.7455920836462346,
      "learning_rate": 1.844465069236851e-06,
      "loss": 0.7351,
      "step": 5977
    },
    {
      "epoch": 0.5391171033052261,
      "grad_norm": 1.539507264291413,
      "learning_rate": 1.8438826345197796e-06,
      "loss": 0.6593,
      "step": 5978
    },
    {
      "epoch": 0.5392072868286965,
      "grad_norm": 4.554216782601558,
      "learning_rate": 1.8433002131235036e-06,
      "loss": 0.7452,
      "step": 5979
    },
    {
      "epoch": 0.5392974703521667,
      "grad_norm": 1.826546076822319,
      "learning_rate": 1.8427178050977167e-06,
      "loss": 0.773,
      "step": 5980
    },
    {
      "epoch": 0.5393876538756369,
      "grad_norm": 1.4470104244782012,
      "learning_rate": 1.8421354104921143e-06,
      "loss": 0.8296,
      "step": 5981
    },
    {
      "epoch": 0.5394778373991072,
      "grad_norm": 0.6047861496843124,
      "learning_rate": 1.8415530293563894e-06,
      "loss": 0.549,
      "step": 5982
    },
    {
      "epoch": 0.5395680209225775,
      "grad_norm": 1.6577558329799882,
      "learning_rate": 1.8409706617402333e-06,
      "loss": 0.7331,
      "step": 5983
    },
    {
      "epoch": 0.5396582044460477,
      "grad_norm": 1.6559423484810454,
      "learning_rate": 1.8403883076933378e-06,
      "loss": 0.7722,
      "step": 5984
    },
    {
      "epoch": 0.539748387969518,
      "grad_norm": 1.4662177800559664,
      "learning_rate": 1.839805967265391e-06,
      "loss": 0.745,
      "step": 5985
    },
    {
      "epoch": 0.5398385714929882,
      "grad_norm": 2.0860464231934097,
      "learning_rate": 1.839223640506083e-06,
      "loss": 0.8086,
      "step": 5986
    },
    {
      "epoch": 0.5399287550164585,
      "grad_norm": 1.9553249938770532,
      "learning_rate": 1.8386413274650998e-06,
      "loss": 0.7439,
      "step": 5987
    },
    {
      "epoch": 0.5400189385399288,
      "grad_norm": 1.6763559347497485,
      "learning_rate": 1.8380590281921294e-06,
      "loss": 0.7993,
      "step": 5988
    },
    {
      "epoch": 0.540109122063399,
      "grad_norm": 1.4734860619544963,
      "learning_rate": 1.8374767427368552e-06,
      "loss": 0.6595,
      "step": 5989
    },
    {
      "epoch": 0.5401993055868692,
      "grad_norm": 2.0293582029502204,
      "learning_rate": 1.8368944711489608e-06,
      "loss": 0.783,
      "step": 5990
    },
    {
      "epoch": 0.5402894891103396,
      "grad_norm": 2.1946024442146865,
      "learning_rate": 1.8363122134781304e-06,
      "loss": 0.7811,
      "step": 5991
    },
    {
      "epoch": 0.5403796726338098,
      "grad_norm": 1.7206684156677705,
      "learning_rate": 1.835729969774044e-06,
      "loss": 0.78,
      "step": 5992
    },
    {
      "epoch": 0.54046985615728,
      "grad_norm": 1.5612915706674027,
      "learning_rate": 1.8351477400863823e-06,
      "loss": 0.7601,
      "step": 5993
    },
    {
      "epoch": 0.5405600396807503,
      "grad_norm": 1.751137227709752,
      "learning_rate": 1.8345655244648249e-06,
      "loss": 0.7308,
      "step": 5994
    },
    {
      "epoch": 0.5406502232042206,
      "grad_norm": 1.3389545748827882,
      "learning_rate": 1.8339833229590486e-06,
      "loss": 0.7421,
      "step": 5995
    },
    {
      "epoch": 0.5407404067276909,
      "grad_norm": 2.133374709483079,
      "learning_rate": 1.833401135618731e-06,
      "loss": 0.8259,
      "step": 5996
    },
    {
      "epoch": 0.5408305902511611,
      "grad_norm": 1.4879725178570016,
      "learning_rate": 1.8328189624935466e-06,
      "loss": 0.682,
      "step": 5997
    },
    {
      "epoch": 0.5409207737746313,
      "grad_norm": 1.7897881115008627,
      "learning_rate": 1.832236803633171e-06,
      "loss": 0.7861,
      "step": 5998
    },
    {
      "epoch": 0.5410109572981017,
      "grad_norm": 1.6670172163744204,
      "learning_rate": 1.831654659087276e-06,
      "loss": 0.7165,
      "step": 5999
    },
    {
      "epoch": 0.5411011408215719,
      "grad_norm": 1.6870141212898957,
      "learning_rate": 1.831072528905533e-06,
      "loss": 0.7701,
      "step": 6000
    },
    {
      "epoch": 0.5411913243450421,
      "grad_norm": 1.4717046359655395,
      "learning_rate": 1.8304904131376142e-06,
      "loss": 0.8087,
      "step": 6001
    },
    {
      "epoch": 0.5412815078685124,
      "grad_norm": 1.6650724256646436,
      "learning_rate": 1.8299083118331874e-06,
      "loss": 0.7701,
      "step": 6002
    },
    {
      "epoch": 0.5413716913919827,
      "grad_norm": 1.921468738322449,
      "learning_rate": 1.8293262250419217e-06,
      "loss": 0.8856,
      "step": 6003
    },
    {
      "epoch": 0.541461874915453,
      "grad_norm": 1.719676374995248,
      "learning_rate": 1.828744152813484e-06,
      "loss": 0.8672,
      "step": 6004
    },
    {
      "epoch": 0.5415520584389232,
      "grad_norm": 1.6763879349158115,
      "learning_rate": 1.8281620951975382e-06,
      "loss": 0.7042,
      "step": 6005
    },
    {
      "epoch": 0.5416422419623935,
      "grad_norm": 4.1511120696501225,
      "learning_rate": 1.827580052243751e-06,
      "loss": 0.7129,
      "step": 6006
    },
    {
      "epoch": 0.5417324254858638,
      "grad_norm": 1.5269181617135874,
      "learning_rate": 1.826998024001784e-06,
      "loss": 0.7872,
      "step": 6007
    },
    {
      "epoch": 0.541822609009334,
      "grad_norm": 5.187885683728164,
      "learning_rate": 1.8264160105212995e-06,
      "loss": 0.7777,
      "step": 6008
    },
    {
      "epoch": 0.5419127925328042,
      "grad_norm": 2.131721972338864,
      "learning_rate": 1.8258340118519582e-06,
      "loss": 0.7223,
      "step": 6009
    },
    {
      "epoch": 0.5420029760562746,
      "grad_norm": 1.8507907079774628,
      "learning_rate": 1.82525202804342e-06,
      "loss": 0.7383,
      "step": 6010
    },
    {
      "epoch": 0.5420931595797448,
      "grad_norm": 1.8144856461319612,
      "learning_rate": 1.8246700591453415e-06,
      "loss": 0.7362,
      "step": 6011
    },
    {
      "epoch": 0.542183343103215,
      "grad_norm": 1.5580699681105319,
      "learning_rate": 1.8240881052073801e-06,
      "loss": 0.7576,
      "step": 6012
    },
    {
      "epoch": 0.5422735266266853,
      "grad_norm": 2.5085158267967524,
      "learning_rate": 1.8235061662791923e-06,
      "loss": 0.7206,
      "step": 6013
    },
    {
      "epoch": 0.5423637101501556,
      "grad_norm": 1.7898949431379214,
      "learning_rate": 1.8229242424104309e-06,
      "loss": 0.8229,
      "step": 6014
    },
    {
      "epoch": 0.5424538936736258,
      "grad_norm": 0.6202798464850159,
      "learning_rate": 1.8223423336507503e-06,
      "loss": 0.5598,
      "step": 6015
    },
    {
      "epoch": 0.5425440771970961,
      "grad_norm": 1.8583916379851306,
      "learning_rate": 1.8217604400498012e-06,
      "loss": 0.7368,
      "step": 6016
    },
    {
      "epoch": 0.5426342607205663,
      "grad_norm": 1.8035447767519572,
      "learning_rate": 1.8211785616572333e-06,
      "loss": 0.7994,
      "step": 6017
    },
    {
      "epoch": 0.5427244442440367,
      "grad_norm": 2.0454132451598483,
      "learning_rate": 1.8205966985226975e-06,
      "loss": 0.8221,
      "step": 6018
    },
    {
      "epoch": 0.5428146277675069,
      "grad_norm": 1.3721693552998564,
      "learning_rate": 1.8200148506958397e-06,
      "loss": 0.7618,
      "step": 6019
    },
    {
      "epoch": 0.5429048112909771,
      "grad_norm": 0.6545130586042683,
      "learning_rate": 1.819433018226308e-06,
      "loss": 0.6068,
      "step": 6020
    },
    {
      "epoch": 0.5429949948144474,
      "grad_norm": 0.5999981244375796,
      "learning_rate": 1.8188512011637471e-06,
      "loss": 0.551,
      "step": 6021
    },
    {
      "epoch": 0.5430851783379177,
      "grad_norm": 1.7203174726118102,
      "learning_rate": 1.8182693995578e-06,
      "loss": 0.7303,
      "step": 6022
    },
    {
      "epoch": 0.5431753618613879,
      "grad_norm": 1.4696158433949984,
      "learning_rate": 1.8176876134581098e-06,
      "loss": 0.6342,
      "step": 6023
    },
    {
      "epoch": 0.5432655453848582,
      "grad_norm": 2.0345380726512303,
      "learning_rate": 1.8171058429143176e-06,
      "loss": 0.6997,
      "step": 6024
    },
    {
      "epoch": 0.5433557289083284,
      "grad_norm": 1.4670297816191296,
      "learning_rate": 1.8165240879760637e-06,
      "loss": 0.7911,
      "step": 6025
    },
    {
      "epoch": 0.5434459124317987,
      "grad_norm": 1.4696659721989476,
      "learning_rate": 1.8159423486929862e-06,
      "loss": 0.7829,
      "step": 6026
    },
    {
      "epoch": 0.543536095955269,
      "grad_norm": 8.198224912969442,
      "learning_rate": 1.815360625114722e-06,
      "loss": 0.7523,
      "step": 6027
    },
    {
      "epoch": 0.5436262794787392,
      "grad_norm": 1.5260361637714972,
      "learning_rate": 1.814778917290908e-06,
      "loss": 0.7369,
      "step": 6028
    },
    {
      "epoch": 0.5437164630022095,
      "grad_norm": 2.026352597839044,
      "learning_rate": 1.8141972252711773e-06,
      "loss": 0.7697,
      "step": 6029
    },
    {
      "epoch": 0.5438066465256798,
      "grad_norm": 1.897569685650254,
      "learning_rate": 1.8136155491051645e-06,
      "loss": 0.7484,
      "step": 6030
    },
    {
      "epoch": 0.54389683004915,
      "grad_norm": 0.672190902431785,
      "learning_rate": 1.8130338888424998e-06,
      "loss": 0.5944,
      "step": 6031
    },
    {
      "epoch": 0.5439870135726202,
      "grad_norm": 4.45765535897948,
      "learning_rate": 1.812452244532816e-06,
      "loss": 0.7737,
      "step": 6032
    },
    {
      "epoch": 0.5440771970960906,
      "grad_norm": 1.6157276733357433,
      "learning_rate": 1.8118706162257405e-06,
      "loss": 0.6897,
      "step": 6033
    },
    {
      "epoch": 0.5441673806195608,
      "grad_norm": 1.5537333411921836,
      "learning_rate": 1.8112890039709002e-06,
      "loss": 0.8324,
      "step": 6034
    },
    {
      "epoch": 0.5442575641430311,
      "grad_norm": 1.7048730541189894,
      "learning_rate": 1.8107074078179238e-06,
      "loss": 0.7591,
      "step": 6035
    },
    {
      "epoch": 0.5443477476665013,
      "grad_norm": 1.4624930683199948,
      "learning_rate": 1.8101258278164348e-06,
      "loss": 0.7045,
      "step": 6036
    },
    {
      "epoch": 0.5444379311899716,
      "grad_norm": 1.5916498596076303,
      "learning_rate": 1.8095442640160575e-06,
      "loss": 0.86,
      "step": 6037
    },
    {
      "epoch": 0.5445281147134419,
      "grad_norm": 2.1819016890367307,
      "learning_rate": 1.8089627164664132e-06,
      "loss": 0.7174,
      "step": 6038
    },
    {
      "epoch": 0.5446182982369121,
      "grad_norm": 1.3265066272125743,
      "learning_rate": 1.8083811852171233e-06,
      "loss": 0.6729,
      "step": 6039
    },
    {
      "epoch": 0.5447084817603823,
      "grad_norm": 1.613747443943969,
      "learning_rate": 1.8077996703178078e-06,
      "loss": 0.7524,
      "step": 6040
    },
    {
      "epoch": 0.5447986652838527,
      "grad_norm": 1.5433198553503351,
      "learning_rate": 1.8072181718180833e-06,
      "loss": 0.7013,
      "step": 6041
    },
    {
      "epoch": 0.5448888488073229,
      "grad_norm": 1.5826161751431416,
      "learning_rate": 1.806636689767568e-06,
      "loss": 0.8437,
      "step": 6042
    },
    {
      "epoch": 0.5449790323307931,
      "grad_norm": 1.6855417828355874,
      "learning_rate": 1.8060552242158769e-06,
      "loss": 0.6902,
      "step": 6043
    },
    {
      "epoch": 0.5450692158542634,
      "grad_norm": 1.5491712077161293,
      "learning_rate": 1.8054737752126224e-06,
      "loss": 0.7273,
      "step": 6044
    },
    {
      "epoch": 0.5451593993777337,
      "grad_norm": 1.5142897396985247,
      "learning_rate": 1.804892342807419e-06,
      "loss": 0.7401,
      "step": 6045
    },
    {
      "epoch": 0.545249582901204,
      "grad_norm": 2.3236603394643254,
      "learning_rate": 1.8043109270498756e-06,
      "loss": 0.8255,
      "step": 6046
    },
    {
      "epoch": 0.5453397664246742,
      "grad_norm": 1.37275482153823,
      "learning_rate": 1.803729527989604e-06,
      "loss": 0.7631,
      "step": 6047
    },
    {
      "epoch": 0.5454299499481444,
      "grad_norm": 2.714433685729385,
      "learning_rate": 1.8031481456762112e-06,
      "loss": 0.7489,
      "step": 6048
    },
    {
      "epoch": 0.5455201334716148,
      "grad_norm": 1.8843733147001558,
      "learning_rate": 1.8025667801593033e-06,
      "loss": 0.6831,
      "step": 6049
    },
    {
      "epoch": 0.545610316995085,
      "grad_norm": 2.1249325685460985,
      "learning_rate": 1.8019854314884871e-06,
      "loss": 0.6923,
      "step": 6050
    },
    {
      "epoch": 0.5457005005185552,
      "grad_norm": 1.777475943208085,
      "learning_rate": 1.8014040997133652e-06,
      "loss": 0.8067,
      "step": 6051
    },
    {
      "epoch": 0.5457906840420256,
      "grad_norm": 0.7577321511371772,
      "learning_rate": 1.8008227848835414e-06,
      "loss": 0.6148,
      "step": 6052
    },
    {
      "epoch": 0.5458808675654958,
      "grad_norm": 1.9892616477933083,
      "learning_rate": 1.8002414870486144e-06,
      "loss": 0.7067,
      "step": 6053
    },
    {
      "epoch": 0.545971051088966,
      "grad_norm": 1.3395605204979661,
      "learning_rate": 1.7996602062581864e-06,
      "loss": 0.7082,
      "step": 6054
    },
    {
      "epoch": 0.5460612346124363,
      "grad_norm": 1.6287204261372557,
      "learning_rate": 1.7990789425618544e-06,
      "loss": 0.8031,
      "step": 6055
    },
    {
      "epoch": 0.5461514181359066,
      "grad_norm": 1.379731792978853,
      "learning_rate": 1.7984976960092137e-06,
      "loss": 0.7357,
      "step": 6056
    },
    {
      "epoch": 0.5462416016593769,
      "grad_norm": 2.10895415452423,
      "learning_rate": 1.7979164666498617e-06,
      "loss": 0.7228,
      "step": 6057
    },
    {
      "epoch": 0.5463317851828471,
      "grad_norm": 0.617789807556522,
      "learning_rate": 1.7973352545333901e-06,
      "loss": 0.5543,
      "step": 6058
    },
    {
      "epoch": 0.5464219687063173,
      "grad_norm": 1.674995931933573,
      "learning_rate": 1.796754059709393e-06,
      "loss": 0.6756,
      "step": 6059
    },
    {
      "epoch": 0.5465121522297877,
      "grad_norm": 1.914161337032823,
      "learning_rate": 1.7961728822274603e-06,
      "loss": 0.8131,
      "step": 6060
    },
    {
      "epoch": 0.5466023357532579,
      "grad_norm": 1.6117055370031437,
      "learning_rate": 1.7955917221371802e-06,
      "loss": 0.7117,
      "step": 6061
    },
    {
      "epoch": 0.5466925192767281,
      "grad_norm": 1.831625562049854,
      "learning_rate": 1.7950105794881422e-06,
      "loss": 0.6692,
      "step": 6062
    },
    {
      "epoch": 0.5467827028001984,
      "grad_norm": 1.3909163437731855,
      "learning_rate": 1.7944294543299317e-06,
      "loss": 0.7347,
      "step": 6063
    },
    {
      "epoch": 0.5468728863236687,
      "grad_norm": 2.3540687765738895,
      "learning_rate": 1.7938483467121333e-06,
      "loss": 0.8097,
      "step": 6064
    },
    {
      "epoch": 0.5469630698471389,
      "grad_norm": 1.7205486249895319,
      "learning_rate": 1.7932672566843313e-06,
      "loss": 0.7342,
      "step": 6065
    },
    {
      "epoch": 0.5470532533706092,
      "grad_norm": 17.366491635856715,
      "learning_rate": 1.7926861842961065e-06,
      "loss": 0.6776,
      "step": 6066
    },
    {
      "epoch": 0.5471434368940794,
      "grad_norm": 1.9459272891552128,
      "learning_rate": 1.7921051295970399e-06,
      "loss": 0.6754,
      "step": 6067
    },
    {
      "epoch": 0.5472336204175497,
      "grad_norm": 1.5653718877010694,
      "learning_rate": 1.7915240926367092e-06,
      "loss": 0.7217,
      "step": 6068
    },
    {
      "epoch": 0.54732380394102,
      "grad_norm": 1.6076428619548055,
      "learning_rate": 1.7909430734646932e-06,
      "loss": 0.794,
      "step": 6069
    },
    {
      "epoch": 0.5474139874644902,
      "grad_norm": 0.6824228121877173,
      "learning_rate": 1.790362072130567e-06,
      "loss": 0.5872,
      "step": 6070
    },
    {
      "epoch": 0.5475041709879604,
      "grad_norm": 1.5784462328004458,
      "learning_rate": 1.7897810886839037e-06,
      "loss": 0.7406,
      "step": 6071
    },
    {
      "epoch": 0.5475943545114308,
      "grad_norm": 1.474863437214445,
      "learning_rate": 1.7892001231742782e-06,
      "loss": 0.7674,
      "step": 6072
    },
    {
      "epoch": 0.547684538034901,
      "grad_norm": 1.614722469446552,
      "learning_rate": 1.7886191756512594e-06,
      "loss": 0.761,
      "step": 6073
    },
    {
      "epoch": 0.5477747215583713,
      "grad_norm": 2.4778525665734845,
      "learning_rate": 1.7880382461644192e-06,
      "loss": 0.832,
      "step": 6074
    },
    {
      "epoch": 0.5478649050818416,
      "grad_norm": 2.570726633768676,
      "learning_rate": 1.7874573347633235e-06,
      "loss": 0.7281,
      "step": 6075
    },
    {
      "epoch": 0.5479550886053118,
      "grad_norm": 1.730676022897145,
      "learning_rate": 1.7868764414975408e-06,
      "loss": 0.8256,
      "step": 6076
    },
    {
      "epoch": 0.5480452721287821,
      "grad_norm": 2.052283094981543,
      "learning_rate": 1.7862955664166353e-06,
      "loss": 0.6936,
      "step": 6077
    },
    {
      "epoch": 0.5481354556522523,
      "grad_norm": 1.3198540247050623,
      "learning_rate": 1.78571470957017e-06,
      "loss": 0.8338,
      "step": 6078
    },
    {
      "epoch": 0.5482256391757226,
      "grad_norm": 1.696045185449859,
      "learning_rate": 1.7851338710077074e-06,
      "loss": 0.823,
      "step": 6079
    },
    {
      "epoch": 0.5483158226991929,
      "grad_norm": 2.0711996229142677,
      "learning_rate": 1.7845530507788076e-06,
      "loss": 0.8596,
      "step": 6080
    },
    {
      "epoch": 0.5484060062226631,
      "grad_norm": 1.705157126293029,
      "learning_rate": 1.7839722489330298e-06,
      "loss": 0.7238,
      "step": 6081
    },
    {
      "epoch": 0.5484961897461333,
      "grad_norm": 1.5396646009305939,
      "learning_rate": 1.7833914655199308e-06,
      "loss": 0.7231,
      "step": 6082
    },
    {
      "epoch": 0.5485863732696037,
      "grad_norm": 2.120720874475305,
      "learning_rate": 1.7828107005890658e-06,
      "loss": 0.7988,
      "step": 6083
    },
    {
      "epoch": 0.5486765567930739,
      "grad_norm": 1.9352573212836346,
      "learning_rate": 1.7822299541899898e-06,
      "loss": 0.6738,
      "step": 6084
    },
    {
      "epoch": 0.5487667403165442,
      "grad_norm": 3.522211034765704,
      "learning_rate": 1.7816492263722545e-06,
      "loss": 0.7493,
      "step": 6085
    },
    {
      "epoch": 0.5488569238400144,
      "grad_norm": 2.0119734223280736,
      "learning_rate": 1.781068517185412e-06,
      "loss": 0.7799,
      "step": 6086
    },
    {
      "epoch": 0.5489471073634847,
      "grad_norm": 1.6794603770863799,
      "learning_rate": 1.7804878266790104e-06,
      "loss": 0.8182,
      "step": 6087
    },
    {
      "epoch": 0.549037290886955,
      "grad_norm": 2.174672582251484,
      "learning_rate": 1.779907154902597e-06,
      "loss": 0.6831,
      "step": 6088
    },
    {
      "epoch": 0.5491274744104252,
      "grad_norm": 2.4626030992332133,
      "learning_rate": 1.7793265019057198e-06,
      "loss": 0.6816,
      "step": 6089
    },
    {
      "epoch": 0.5492176579338954,
      "grad_norm": 0.726395495510104,
      "learning_rate": 1.7787458677379212e-06,
      "loss": 0.5946,
      "step": 6090
    },
    {
      "epoch": 0.5493078414573658,
      "grad_norm": 1.6430608922177163,
      "learning_rate": 1.7781652524487463e-06,
      "loss": 0.8101,
      "step": 6091
    },
    {
      "epoch": 0.549398024980836,
      "grad_norm": 3.722057378417864,
      "learning_rate": 1.777584656087735e-06,
      "loss": 0.7026,
      "step": 6092
    },
    {
      "epoch": 0.5494882085043062,
      "grad_norm": 1.8530976574892482,
      "learning_rate": 1.777004078704427e-06,
      "loss": 0.6757,
      "step": 6093
    },
    {
      "epoch": 0.5495783920277765,
      "grad_norm": 1.6853946692214377,
      "learning_rate": 1.7764235203483603e-06,
      "loss": 0.7717,
      "step": 6094
    },
    {
      "epoch": 0.5496685755512468,
      "grad_norm": 2.929327940175239,
      "learning_rate": 1.775842981069072e-06,
      "loss": 0.7224,
      "step": 6095
    },
    {
      "epoch": 0.549758759074717,
      "grad_norm": 1.4756285928232482,
      "learning_rate": 1.7752624609160966e-06,
      "loss": 0.761,
      "step": 6096
    },
    {
      "epoch": 0.5498489425981873,
      "grad_norm": 2.171735031777194,
      "learning_rate": 1.7746819599389665e-06,
      "loss": 0.791,
      "step": 6097
    },
    {
      "epoch": 0.5499391261216575,
      "grad_norm": 2.984507133394837,
      "learning_rate": 1.774101478187215e-06,
      "loss": 0.7545,
      "step": 6098
    },
    {
      "epoch": 0.5500293096451279,
      "grad_norm": 1.7339919758471336,
      "learning_rate": 1.773521015710371e-06,
      "loss": 0.8219,
      "step": 6099
    },
    {
      "epoch": 0.5501194931685981,
      "grad_norm": 2.5979861860194333,
      "learning_rate": 1.7729405725579614e-06,
      "loss": 0.8342,
      "step": 6100
    },
    {
      "epoch": 0.5502096766920683,
      "grad_norm": 1.5337776808351025,
      "learning_rate": 1.7723601487795151e-06,
      "loss": 0.7294,
      "step": 6101
    },
    {
      "epoch": 0.5502998602155387,
      "grad_norm": 1.6130838712158715,
      "learning_rate": 1.7717797444245557e-06,
      "loss": 0.7012,
      "step": 6102
    },
    {
      "epoch": 0.5503900437390089,
      "grad_norm": 1.2999099700270713,
      "learning_rate": 1.7711993595426076e-06,
      "loss": 0.6911,
      "step": 6103
    },
    {
      "epoch": 0.5504802272624791,
      "grad_norm": 1.9838228433716194,
      "learning_rate": 1.7706189941831915e-06,
      "loss": 0.8624,
      "step": 6104
    },
    {
      "epoch": 0.5505704107859494,
      "grad_norm": 1.8445286803690144,
      "learning_rate": 1.770038648395827e-06,
      "loss": 0.7126,
      "step": 6105
    },
    {
      "epoch": 0.5506605943094197,
      "grad_norm": 0.6103100815982435,
      "learning_rate": 1.7694583222300336e-06,
      "loss": 0.5949,
      "step": 6106
    },
    {
      "epoch": 0.55075077783289,
      "grad_norm": 1.6148849533722747,
      "learning_rate": 1.7688780157353272e-06,
      "loss": 0.7354,
      "step": 6107
    },
    {
      "epoch": 0.5508409613563602,
      "grad_norm": 1.8137710323994016,
      "learning_rate": 1.768297728961223e-06,
      "loss": 0.6963,
      "step": 6108
    },
    {
      "epoch": 0.5509311448798304,
      "grad_norm": 2.194298995572439,
      "learning_rate": 1.7677174619572342e-06,
      "loss": 0.7588,
      "step": 6109
    },
    {
      "epoch": 0.5510213284033008,
      "grad_norm": 2.734150904600625,
      "learning_rate": 1.7671372147728717e-06,
      "loss": 0.774,
      "step": 6110
    },
    {
      "epoch": 0.551111511926771,
      "grad_norm": 1.7978424121762662,
      "learning_rate": 1.7665569874576471e-06,
      "loss": 0.8208,
      "step": 6111
    },
    {
      "epoch": 0.5512016954502412,
      "grad_norm": 3.0656794956352655,
      "learning_rate": 1.7659767800610664e-06,
      "loss": 0.7582,
      "step": 6112
    },
    {
      "epoch": 0.5512918789737115,
      "grad_norm": 3.8475747036885286,
      "learning_rate": 1.7653965926326379e-06,
      "loss": 0.6626,
      "step": 6113
    },
    {
      "epoch": 0.5513820624971818,
      "grad_norm": 1.7698568082529615,
      "learning_rate": 1.764816425221866e-06,
      "loss": 0.7108,
      "step": 6114
    },
    {
      "epoch": 0.551472246020652,
      "grad_norm": 1.4877907252506057,
      "learning_rate": 1.7642362778782524e-06,
      "loss": 0.67,
      "step": 6115
    },
    {
      "epoch": 0.5515624295441223,
      "grad_norm": 1.89304241394472,
      "learning_rate": 1.7636561506513005e-06,
      "loss": 0.7337,
      "step": 6116
    },
    {
      "epoch": 0.5516526130675925,
      "grad_norm": 1.412726514090141,
      "learning_rate": 1.7630760435905083e-06,
      "loss": 0.7278,
      "step": 6117
    },
    {
      "epoch": 0.5517427965910628,
      "grad_norm": 1.726847560359443,
      "learning_rate": 1.762495956745375e-06,
      "loss": 0.7094,
      "step": 6118
    },
    {
      "epoch": 0.5518329801145331,
      "grad_norm": 1.7223955506435178,
      "learning_rate": 1.7619158901653962e-06,
      "loss": 0.8038,
      "step": 6119
    },
    {
      "epoch": 0.5519231636380033,
      "grad_norm": 1.6400553122938941,
      "learning_rate": 1.761335843900066e-06,
      "loss": 0.7519,
      "step": 6120
    },
    {
      "epoch": 0.5520133471614735,
      "grad_norm": 1.4729896390645054,
      "learning_rate": 1.7607558179988785e-06,
      "loss": 0.7375,
      "step": 6121
    },
    {
      "epoch": 0.5521035306849439,
      "grad_norm": 1.6295975423473166,
      "learning_rate": 1.760175812511323e-06,
      "loss": 0.6814,
      "step": 6122
    },
    {
      "epoch": 0.5521937142084141,
      "grad_norm": 1.5972645472940161,
      "learning_rate": 1.75959582748689e-06,
      "loss": 0.7142,
      "step": 6123
    },
    {
      "epoch": 0.5522838977318844,
      "grad_norm": 1.658479665245724,
      "learning_rate": 1.7590158629750657e-06,
      "loss": 0.7879,
      "step": 6124
    },
    {
      "epoch": 0.5523740812553547,
      "grad_norm": 1.6054903571909365,
      "learning_rate": 1.7584359190253376e-06,
      "loss": 0.8048,
      "step": 6125
    },
    {
      "epoch": 0.5524642647788249,
      "grad_norm": 2.493405609443262,
      "learning_rate": 1.7578559956871892e-06,
      "loss": 0.8177,
      "step": 6126
    },
    {
      "epoch": 0.5525544483022952,
      "grad_norm": 1.4657633018716383,
      "learning_rate": 1.7572760930101012e-06,
      "loss": 0.7501,
      "step": 6127
    },
    {
      "epoch": 0.5526446318257654,
      "grad_norm": 0.6863415885564379,
      "learning_rate": 1.7566962110435563e-06,
      "loss": 0.6342,
      "step": 6128
    },
    {
      "epoch": 0.5527348153492357,
      "grad_norm": 1.559700709023143,
      "learning_rate": 1.7561163498370313e-06,
      "loss": 0.7702,
      "step": 6129
    },
    {
      "epoch": 0.552824998872706,
      "grad_norm": 1.5429221617027102,
      "learning_rate": 1.755536509440005e-06,
      "loss": 0.7175,
      "step": 6130
    },
    {
      "epoch": 0.5529151823961762,
      "grad_norm": 3.125153499648026,
      "learning_rate": 1.7549566899019519e-06,
      "loss": 0.7531,
      "step": 6131
    },
    {
      "epoch": 0.5530053659196464,
      "grad_norm": 1.618078087944056,
      "learning_rate": 1.754376891272344e-06,
      "loss": 0.7682,
      "step": 6132
    },
    {
      "epoch": 0.5530955494431168,
      "grad_norm": 1.709786642048687,
      "learning_rate": 1.753797113600655e-06,
      "loss": 0.703,
      "step": 6133
    },
    {
      "epoch": 0.553185732966587,
      "grad_norm": 1.8543190054158523,
      "learning_rate": 1.7532173569363535e-06,
      "loss": 0.8901,
      "step": 6134
    },
    {
      "epoch": 0.5532759164900573,
      "grad_norm": 1.9803557655263837,
      "learning_rate": 1.7526376213289077e-06,
      "loss": 0.5705,
      "step": 6135
    },
    {
      "epoch": 0.5533661000135275,
      "grad_norm": 1.9409924685977213,
      "learning_rate": 1.7520579068277844e-06,
      "loss": 0.7378,
      "step": 6136
    },
    {
      "epoch": 0.5534562835369978,
      "grad_norm": 1.6021702055409697,
      "learning_rate": 1.7514782134824472e-06,
      "loss": 0.7808,
      "step": 6137
    },
    {
      "epoch": 0.5535464670604681,
      "grad_norm": 1.6871177805615887,
      "learning_rate": 1.7508985413423599e-06,
      "loss": 0.7474,
      "step": 6138
    },
    {
      "epoch": 0.5536366505839383,
      "grad_norm": 1.5773766886883989,
      "learning_rate": 1.7503188904569814e-06,
      "loss": 0.8261,
      "step": 6139
    },
    {
      "epoch": 0.5537268341074085,
      "grad_norm": 2.8335732938092777,
      "learning_rate": 1.7497392608757728e-06,
      "loss": 0.7229,
      "step": 6140
    },
    {
      "epoch": 0.5538170176308789,
      "grad_norm": 1.6642938730635128,
      "learning_rate": 1.7491596526481897e-06,
      "loss": 0.7708,
      "step": 6141
    },
    {
      "epoch": 0.5539072011543491,
      "grad_norm": 1.646690708685022,
      "learning_rate": 1.7485800658236888e-06,
      "loss": 0.6819,
      "step": 6142
    },
    {
      "epoch": 0.5539973846778193,
      "grad_norm": 2.603810298695786,
      "learning_rate": 1.7480005004517232e-06,
      "loss": 0.8264,
      "step": 6143
    },
    {
      "epoch": 0.5540875682012896,
      "grad_norm": 1.6757403459869846,
      "learning_rate": 1.7474209565817435e-06,
      "loss": 0.8185,
      "step": 6144
    },
    {
      "epoch": 0.5541777517247599,
      "grad_norm": 1.3923724906796722,
      "learning_rate": 1.7468414342632014e-06,
      "loss": 0.7546,
      "step": 6145
    },
    {
      "epoch": 0.5542679352482301,
      "grad_norm": 1.7569347309284644,
      "learning_rate": 1.746261933545543e-06,
      "loss": 0.807,
      "step": 6146
    },
    {
      "epoch": 0.5543581187717004,
      "grad_norm": 1.4004868137837911,
      "learning_rate": 1.7456824544782165e-06,
      "loss": 0.7482,
      "step": 6147
    },
    {
      "epoch": 0.5544483022951707,
      "grad_norm": 1.8707991907732577,
      "learning_rate": 1.7451029971106653e-06,
      "loss": 0.7582,
      "step": 6148
    },
    {
      "epoch": 0.554538485818641,
      "grad_norm": 1.4152309585096252,
      "learning_rate": 1.7445235614923313e-06,
      "loss": 0.7817,
      "step": 6149
    },
    {
      "epoch": 0.5546286693421112,
      "grad_norm": 1.7937155770112407,
      "learning_rate": 1.7439441476726556e-06,
      "loss": 0.7415,
      "step": 6150
    },
    {
      "epoch": 0.5547188528655814,
      "grad_norm": 3.7768087235007193,
      "learning_rate": 1.7433647557010776e-06,
      "loss": 0.761,
      "step": 6151
    },
    {
      "epoch": 0.5548090363890518,
      "grad_norm": 2.3237669433501362,
      "learning_rate": 1.7427853856270338e-06,
      "loss": 0.7169,
      "step": 6152
    },
    {
      "epoch": 0.554899219912522,
      "grad_norm": 2.2254038154970455,
      "learning_rate": 1.7422060374999593e-06,
      "loss": 0.7172,
      "step": 6153
    },
    {
      "epoch": 0.5549894034359922,
      "grad_norm": 1.502633802451149,
      "learning_rate": 1.7416267113692862e-06,
      "loss": 0.7136,
      "step": 6154
    },
    {
      "epoch": 0.5550795869594625,
      "grad_norm": 1.5000328219319632,
      "learning_rate": 1.7410474072844475e-06,
      "loss": 0.784,
      "step": 6155
    },
    {
      "epoch": 0.5551697704829328,
      "grad_norm": 1.4855798750248117,
      "learning_rate": 1.740468125294871e-06,
      "loss": 0.7363,
      "step": 6156
    },
    {
      "epoch": 0.555259954006403,
      "grad_norm": 2.251271206651575,
      "learning_rate": 1.739888865449986e-06,
      "loss": 0.6694,
      "step": 6157
    },
    {
      "epoch": 0.5553501375298733,
      "grad_norm": 6.476842858075463,
      "learning_rate": 1.7393096277992174e-06,
      "loss": 0.7773,
      "step": 6158
    },
    {
      "epoch": 0.5554403210533435,
      "grad_norm": 1.5893667944344931,
      "learning_rate": 1.738730412391988e-06,
      "loss": 0.7663,
      "step": 6159
    },
    {
      "epoch": 0.5555305045768139,
      "grad_norm": 3.9261855794970875,
      "learning_rate": 1.738151219277721e-06,
      "loss": 0.8044,
      "step": 6160
    },
    {
      "epoch": 0.5556206881002841,
      "grad_norm": 2.926956898050334,
      "learning_rate": 1.7375720485058349e-06,
      "loss": 0.713,
      "step": 6161
    },
    {
      "epoch": 0.5557108716237543,
      "grad_norm": 2.050157085075758,
      "learning_rate": 1.7369929001257498e-06,
      "loss": 0.7738,
      "step": 6162
    },
    {
      "epoch": 0.5558010551472246,
      "grad_norm": 1.961153360469456,
      "learning_rate": 1.73641377418688e-06,
      "loss": 0.6678,
      "step": 6163
    },
    {
      "epoch": 0.5558912386706949,
      "grad_norm": 1.8013000402971628,
      "learning_rate": 1.7358346707386408e-06,
      "loss": 0.6878,
      "step": 6164
    },
    {
      "epoch": 0.5559814221941651,
      "grad_norm": 1.8785542180124073,
      "learning_rate": 1.7352555898304439e-06,
      "loss": 0.8052,
      "step": 6165
    },
    {
      "epoch": 0.5560716057176354,
      "grad_norm": 1.7650704652374132,
      "learning_rate": 1.7346765315116996e-06,
      "loss": 0.6794,
      "step": 6166
    },
    {
      "epoch": 0.5561617892411056,
      "grad_norm": 1.5678659995586426,
      "learning_rate": 1.734097495831817e-06,
      "loss": 0.7146,
      "step": 6167
    },
    {
      "epoch": 0.5562519727645759,
      "grad_norm": 1.8206116262744194,
      "learning_rate": 1.7335184828402015e-06,
      "loss": 0.7641,
      "step": 6168
    },
    {
      "epoch": 0.5563421562880462,
      "grad_norm": 1.4093627810741503,
      "learning_rate": 1.7329394925862595e-06,
      "loss": 0.7327,
      "step": 6169
    },
    {
      "epoch": 0.5564323398115164,
      "grad_norm": 2.1686785111751266,
      "learning_rate": 1.7323605251193922e-06,
      "loss": 0.7994,
      "step": 6170
    },
    {
      "epoch": 0.5565225233349868,
      "grad_norm": 2.0228507468905255,
      "learning_rate": 1.7317815804890001e-06,
      "loss": 0.7738,
      "step": 6171
    },
    {
      "epoch": 0.556612706858457,
      "grad_norm": 1.7175117887622644,
      "learning_rate": 1.731202658744483e-06,
      "loss": 0.709,
      "step": 6172
    },
    {
      "epoch": 0.5567028903819272,
      "grad_norm": 1.4656775784364289,
      "learning_rate": 1.7306237599352365e-06,
      "loss": 0.8147,
      "step": 6173
    },
    {
      "epoch": 0.5567930739053975,
      "grad_norm": 1.419687789658339,
      "learning_rate": 1.730044884110657e-06,
      "loss": 0.7409,
      "step": 6174
    },
    {
      "epoch": 0.5568832574288678,
      "grad_norm": 1.670462394818588,
      "learning_rate": 1.7294660313201366e-06,
      "loss": 0.714,
      "step": 6175
    },
    {
      "epoch": 0.556973440952338,
      "grad_norm": 2.436936337759859,
      "learning_rate": 1.7288872016130652e-06,
      "loss": 0.7351,
      "step": 6176
    },
    {
      "epoch": 0.5570636244758083,
      "grad_norm": 1.9525650442428035,
      "learning_rate": 1.7283083950388334e-06,
      "loss": 0.666,
      "step": 6177
    },
    {
      "epoch": 0.5571538079992785,
      "grad_norm": 1.5656136578969464,
      "learning_rate": 1.727729611646827e-06,
      "loss": 0.658,
      "step": 6178
    },
    {
      "epoch": 0.5572439915227488,
      "grad_norm": 1.5842849314593233,
      "learning_rate": 1.7271508514864318e-06,
      "loss": 0.8071,
      "step": 6179
    },
    {
      "epoch": 0.5573341750462191,
      "grad_norm": 1.9457387801837789,
      "learning_rate": 1.7265721146070302e-06,
      "loss": 0.7561,
      "step": 6180
    },
    {
      "epoch": 0.5574243585696893,
      "grad_norm": 1.6756252402931846,
      "learning_rate": 1.7259934010580035e-06,
      "loss": 0.752,
      "step": 6181
    },
    {
      "epoch": 0.5575145420931595,
      "grad_norm": 1.6295423843569616,
      "learning_rate": 1.725414710888731e-06,
      "loss": 0.7031,
      "step": 6182
    },
    {
      "epoch": 0.5576047256166299,
      "grad_norm": 1.5093671462839962,
      "learning_rate": 1.7248360441485885e-06,
      "loss": 0.728,
      "step": 6183
    },
    {
      "epoch": 0.5576949091401001,
      "grad_norm": 1.6067929341070408,
      "learning_rate": 1.7242574008869528e-06,
      "loss": 0.8271,
      "step": 6184
    },
    {
      "epoch": 0.5577850926635703,
      "grad_norm": 0.6808862091046614,
      "learning_rate": 1.7236787811531951e-06,
      "loss": 0.6287,
      "step": 6185
    },
    {
      "epoch": 0.5578752761870406,
      "grad_norm": 1.579398896882299,
      "learning_rate": 1.7231001849966887e-06,
      "loss": 0.7211,
      "step": 6186
    },
    {
      "epoch": 0.5579654597105109,
      "grad_norm": 0.6473598487614537,
      "learning_rate": 1.722521612466801e-06,
      "loss": 0.6076,
      "step": 6187
    },
    {
      "epoch": 0.5580556432339812,
      "grad_norm": 1.9033324631234143,
      "learning_rate": 1.7219430636128984e-06,
      "loss": 0.694,
      "step": 6188
    },
    {
      "epoch": 0.5581458267574514,
      "grad_norm": 1.6817134969874348,
      "learning_rate": 1.7213645384843479e-06,
      "loss": 0.6963,
      "step": 6189
    },
    {
      "epoch": 0.5582360102809216,
      "grad_norm": 1.712295202553361,
      "learning_rate": 1.7207860371305108e-06,
      "loss": 0.7444,
      "step": 6190
    },
    {
      "epoch": 0.558326193804392,
      "grad_norm": 2.6211812534876815,
      "learning_rate": 1.7202075596007487e-06,
      "loss": 0.7197,
      "step": 6191
    },
    {
      "epoch": 0.5584163773278622,
      "grad_norm": 1.664742487562638,
      "learning_rate": 1.7196291059444206e-06,
      "loss": 0.7076,
      "step": 6192
    },
    {
      "epoch": 0.5585065608513324,
      "grad_norm": 2.628390529431941,
      "learning_rate": 1.7190506762108828e-06,
      "loss": 0.7399,
      "step": 6193
    },
    {
      "epoch": 0.5585967443748028,
      "grad_norm": 1.6994357266506797,
      "learning_rate": 1.7184722704494907e-06,
      "loss": 0.7529,
      "step": 6194
    },
    {
      "epoch": 0.558686927898273,
      "grad_norm": 1.5550130378691733,
      "learning_rate": 1.717893888709596e-06,
      "loss": 0.7656,
      "step": 6195
    },
    {
      "epoch": 0.5587771114217432,
      "grad_norm": 1.6603388427850256,
      "learning_rate": 1.7173155310405515e-06,
      "loss": 0.7786,
      "step": 6196
    },
    {
      "epoch": 0.5588672949452135,
      "grad_norm": 1.4071873401944186,
      "learning_rate": 1.7167371974917043e-06,
      "loss": 0.8269,
      "step": 6197
    },
    {
      "epoch": 0.5589574784686838,
      "grad_norm": 2.1630377701677146,
      "learning_rate": 1.7161588881124003e-06,
      "loss": 0.7727,
      "step": 6198
    },
    {
      "epoch": 0.559047661992154,
      "grad_norm": 1.9849740273148109,
      "learning_rate": 1.7155806029519861e-06,
      "loss": 0.7368,
      "step": 6199
    },
    {
      "epoch": 0.5591378455156243,
      "grad_norm": 1.7790856178399224,
      "learning_rate": 1.7150023420598023e-06,
      "loss": 0.7921,
      "step": 6200
    },
    {
      "epoch": 0.5592280290390945,
      "grad_norm": 1.4938415626969674,
      "learning_rate": 1.714424105485191e-06,
      "loss": 0.8057,
      "step": 6201
    },
    {
      "epoch": 0.5593182125625649,
      "grad_norm": 3.486018866491778,
      "learning_rate": 1.7138458932774896e-06,
      "loss": 0.7129,
      "step": 6202
    },
    {
      "epoch": 0.5594083960860351,
      "grad_norm": 0.655209829594572,
      "learning_rate": 1.7132677054860335e-06,
      "loss": 0.6073,
      "step": 6203
    },
    {
      "epoch": 0.5594985796095053,
      "grad_norm": 1.8474322257818883,
      "learning_rate": 1.7126895421601586e-06,
      "loss": 0.7742,
      "step": 6204
    },
    {
      "epoch": 0.5595887631329756,
      "grad_norm": 2.208426731461758,
      "learning_rate": 1.712111403349196e-06,
      "loss": 0.6907,
      "step": 6205
    },
    {
      "epoch": 0.5596789466564459,
      "grad_norm": 1.7983383854019337,
      "learning_rate": 1.7115332891024757e-06,
      "loss": 0.787,
      "step": 6206
    },
    {
      "epoch": 0.5597691301799161,
      "grad_norm": 0.7180063298034053,
      "learning_rate": 1.7109551994693257e-06,
      "loss": 0.6571,
      "step": 6207
    },
    {
      "epoch": 0.5598593137033864,
      "grad_norm": 2.285399960294833,
      "learning_rate": 1.7103771344990725e-06,
      "loss": 0.6962,
      "step": 6208
    },
    {
      "epoch": 0.5599494972268566,
      "grad_norm": 1.7267016186484392,
      "learning_rate": 1.709799094241039e-06,
      "loss": 0.6715,
      "step": 6209
    },
    {
      "epoch": 0.560039680750327,
      "grad_norm": 20.591530015966214,
      "learning_rate": 1.709221078744546e-06,
      "loss": 0.784,
      "step": 6210
    },
    {
      "epoch": 0.5601298642737972,
      "grad_norm": 1.5126154800341172,
      "learning_rate": 1.7086430880589148e-06,
      "loss": 0.7628,
      "step": 6211
    },
    {
      "epoch": 0.5602200477972674,
      "grad_norm": 1.870089076057482,
      "learning_rate": 1.7080651222334612e-06,
      "loss": 0.7564,
      "step": 6212
    },
    {
      "epoch": 0.5603102313207377,
      "grad_norm": 2.048309522365715,
      "learning_rate": 1.7074871813175018e-06,
      "loss": 0.8104,
      "step": 6213
    },
    {
      "epoch": 0.560400414844208,
      "grad_norm": 1.7379191160706835,
      "learning_rate": 1.706909265360349e-06,
      "loss": 0.6701,
      "step": 6214
    },
    {
      "epoch": 0.5604905983676782,
      "grad_norm": 0.7050425977553285,
      "learning_rate": 1.7063313744113128e-06,
      "loss": 0.5964,
      "step": 6215
    },
    {
      "epoch": 0.5605807818911485,
      "grad_norm": 0.6420772652790047,
      "learning_rate": 1.7057535085197042e-06,
      "loss": 0.5699,
      "step": 6216
    },
    {
      "epoch": 0.5606709654146187,
      "grad_norm": 3.641774925060219,
      "learning_rate": 1.705175667734828e-06,
      "loss": 0.6727,
      "step": 6217
    },
    {
      "epoch": 0.560761148938089,
      "grad_norm": 3.7473358863487523,
      "learning_rate": 1.7045978521059894e-06,
      "loss": 0.7202,
      "step": 6218
    },
    {
      "epoch": 0.5608513324615593,
      "grad_norm": 1.3822082233708155,
      "learning_rate": 1.7040200616824914e-06,
      "loss": 0.7516,
      "step": 6219
    },
    {
      "epoch": 0.5609415159850295,
      "grad_norm": 1.7495657518300953,
      "learning_rate": 1.7034422965136333e-06,
      "loss": 0.696,
      "step": 6220
    },
    {
      "epoch": 0.5610316995084998,
      "grad_norm": 1.9343339216218367,
      "learning_rate": 1.7028645566487137e-06,
      "loss": 0.7908,
      "step": 6221
    },
    {
      "epoch": 0.5611218830319701,
      "grad_norm": 1.7057804789117292,
      "learning_rate": 1.7022868421370284e-06,
      "loss": 0.7244,
      "step": 6222
    },
    {
      "epoch": 0.5612120665554403,
      "grad_norm": 2.0630931579295924,
      "learning_rate": 1.701709153027872e-06,
      "loss": 0.7966,
      "step": 6223
    },
    {
      "epoch": 0.5613022500789105,
      "grad_norm": 2.2312846119651084,
      "learning_rate": 1.7011314893705353e-06,
      "loss": 0.7139,
      "step": 6224
    },
    {
      "epoch": 0.5613924336023809,
      "grad_norm": 1.47976684035347,
      "learning_rate": 1.700553851214307e-06,
      "loss": 0.7547,
      "step": 6225
    },
    {
      "epoch": 0.5614826171258511,
      "grad_norm": 1.5019067724885824,
      "learning_rate": 1.699976238608476e-06,
      "loss": 0.6198,
      "step": 6226
    },
    {
      "epoch": 0.5615728006493214,
      "grad_norm": 3.85348914437493,
      "learning_rate": 1.699398651602326e-06,
      "loss": 0.6998,
      "step": 6227
    },
    {
      "epoch": 0.5616629841727916,
      "grad_norm": 1.4224062754991436,
      "learning_rate": 1.6988210902451413e-06,
      "loss": 0.7908,
      "step": 6228
    },
    {
      "epoch": 0.5617531676962619,
      "grad_norm": 1.7845361079023043,
      "learning_rate": 1.6982435545862011e-06,
      "loss": 0.7331,
      "step": 6229
    },
    {
      "epoch": 0.5618433512197322,
      "grad_norm": 1.643587688946566,
      "learning_rate": 1.6976660446747853e-06,
      "loss": 0.7597,
      "step": 6230
    },
    {
      "epoch": 0.5619335347432024,
      "grad_norm": 2.072998722567351,
      "learning_rate": 1.6970885605601696e-06,
      "loss": 0.6799,
      "step": 6231
    },
    {
      "epoch": 0.5620237182666726,
      "grad_norm": 1.7010659886723951,
      "learning_rate": 1.6965111022916282e-06,
      "loss": 0.8592,
      "step": 6232
    },
    {
      "epoch": 0.562113901790143,
      "grad_norm": 2.071811924775472,
      "learning_rate": 1.6959336699184323e-06,
      "loss": 0.7256,
      "step": 6233
    },
    {
      "epoch": 0.5622040853136132,
      "grad_norm": 2.322559642867699,
      "learning_rate": 1.6953562634898529e-06,
      "loss": 0.7426,
      "step": 6234
    },
    {
      "epoch": 0.5622942688370834,
      "grad_norm": 2.4274072280898125,
      "learning_rate": 1.6947788830551569e-06,
      "loss": 0.722,
      "step": 6235
    },
    {
      "epoch": 0.5623844523605537,
      "grad_norm": 0.7703284766840419,
      "learning_rate": 1.6942015286636093e-06,
      "loss": 0.6306,
      "step": 6236
    },
    {
      "epoch": 0.562474635884024,
      "grad_norm": 1.6691095886273184,
      "learning_rate": 1.6936242003644735e-06,
      "loss": 0.7362,
      "step": 6237
    },
    {
      "epoch": 0.5625648194074943,
      "grad_norm": 2.3260188464764893,
      "learning_rate": 1.6930468982070106e-06,
      "loss": 0.8203,
      "step": 6238
    },
    {
      "epoch": 0.5626550029309645,
      "grad_norm": 0.5958387877030132,
      "learning_rate": 1.692469622240478e-06,
      "loss": 0.5184,
      "step": 6239
    },
    {
      "epoch": 0.5627451864544347,
      "grad_norm": 1.6286017775994228,
      "learning_rate": 1.6918923725141339e-06,
      "loss": 0.7058,
      "step": 6240
    },
    {
      "epoch": 0.5628353699779051,
      "grad_norm": 1.6916767282922747,
      "learning_rate": 1.6913151490772312e-06,
      "loss": 0.8018,
      "step": 6241
    },
    {
      "epoch": 0.5629255535013753,
      "grad_norm": 1.5652178206304985,
      "learning_rate": 1.6907379519790215e-06,
      "loss": 0.7881,
      "step": 6242
    },
    {
      "epoch": 0.5630157370248455,
      "grad_norm": 1.965333790274834,
      "learning_rate": 1.6901607812687558e-06,
      "loss": 0.6615,
      "step": 6243
    },
    {
      "epoch": 0.5631059205483159,
      "grad_norm": 1.4171849125646083,
      "learning_rate": 1.6895836369956794e-06,
      "loss": 0.7061,
      "step": 6244
    },
    {
      "epoch": 0.5631961040717861,
      "grad_norm": 0.6691290164615314,
      "learning_rate": 1.6890065192090402e-06,
      "loss": 0.543,
      "step": 6245
    },
    {
      "epoch": 0.5632862875952563,
      "grad_norm": 2.67770054056396,
      "learning_rate": 1.6884294279580793e-06,
      "loss": 0.6506,
      "step": 6246
    },
    {
      "epoch": 0.5633764711187266,
      "grad_norm": 8.855871944394172,
      "learning_rate": 1.6878523632920371e-06,
      "loss": 0.775,
      "step": 6247
    },
    {
      "epoch": 0.5634666546421969,
      "grad_norm": 1.6586604302146988,
      "learning_rate": 1.6872753252601525e-06,
      "loss": 0.8171,
      "step": 6248
    },
    {
      "epoch": 0.5635568381656672,
      "grad_norm": 3.6147291756852,
      "learning_rate": 1.6866983139116616e-06,
      "loss": 0.6894,
      "step": 6249
    },
    {
      "epoch": 0.5636470216891374,
      "grad_norm": 1.676593290332436,
      "learning_rate": 1.6861213292957981e-06,
      "loss": 0.7185,
      "step": 6250
    },
    {
      "epoch": 0.5637372052126076,
      "grad_norm": 2.118632594093677,
      "learning_rate": 1.685544371461793e-06,
      "loss": 0.783,
      "step": 6251
    },
    {
      "epoch": 0.563827388736078,
      "grad_norm": 2.4882044997423023,
      "learning_rate": 1.6849674404588767e-06,
      "loss": 0.7719,
      "step": 6252
    },
    {
      "epoch": 0.5639175722595482,
      "grad_norm": 2.1712402953181917,
      "learning_rate": 1.6843905363362758e-06,
      "loss": 0.7859,
      "step": 6253
    },
    {
      "epoch": 0.5640077557830184,
      "grad_norm": 1.9004904239556801,
      "learning_rate": 1.6838136591432136e-06,
      "loss": 0.6929,
      "step": 6254
    },
    {
      "epoch": 0.5640979393064887,
      "grad_norm": 2.112284709185168,
      "learning_rate": 1.6832368089289139e-06,
      "loss": 0.7733,
      "step": 6255
    },
    {
      "epoch": 0.564188122829959,
      "grad_norm": 1.5598237578776804,
      "learning_rate": 1.682659985742596e-06,
      "loss": 0.7647,
      "step": 6256
    },
    {
      "epoch": 0.5642783063534292,
      "grad_norm": 2.5349493427218652,
      "learning_rate": 1.6820831896334782e-06,
      "loss": 0.7892,
      "step": 6257
    },
    {
      "epoch": 0.5643684898768995,
      "grad_norm": 1.6231568961375114,
      "learning_rate": 1.681506420650776e-06,
      "loss": 0.7684,
      "step": 6258
    },
    {
      "epoch": 0.5644586734003697,
      "grad_norm": 1.4858100780322423,
      "learning_rate": 1.680929678843701e-06,
      "loss": 0.7355,
      "step": 6259
    },
    {
      "epoch": 0.56454885692384,
      "grad_norm": 1.632868204581161,
      "learning_rate": 1.6803529642614662e-06,
      "loss": 0.7966,
      "step": 6260
    },
    {
      "epoch": 0.5646390404473103,
      "grad_norm": 1.7200898236839661,
      "learning_rate": 1.6797762769532785e-06,
      "loss": 0.7603,
      "step": 6261
    },
    {
      "epoch": 0.5647292239707805,
      "grad_norm": 1.664684770315835,
      "learning_rate": 1.679199616968345e-06,
      "loss": 0.784,
      "step": 6262
    },
    {
      "epoch": 0.5648194074942507,
      "grad_norm": 1.536215458009033,
      "learning_rate": 1.6786229843558689e-06,
      "loss": 0.7773,
      "step": 6263
    },
    {
      "epoch": 0.5649095910177211,
      "grad_norm": 2.046520988776128,
      "learning_rate": 1.6780463791650514e-06,
      "loss": 0.7304,
      "step": 6264
    },
    {
      "epoch": 0.5649997745411913,
      "grad_norm": 1.7457394507148265,
      "learning_rate": 1.6774698014450928e-06,
      "loss": 0.6942,
      "step": 6265
    },
    {
      "epoch": 0.5650899580646616,
      "grad_norm": 0.6543944556505632,
      "learning_rate": 1.6768932512451883e-06,
      "loss": 0.5787,
      "step": 6266
    },
    {
      "epoch": 0.5651801415881319,
      "grad_norm": 3.636267678121775,
      "learning_rate": 1.676316728614534e-06,
      "loss": 0.7031,
      "step": 6267
    },
    {
      "epoch": 0.5652703251116021,
      "grad_norm": 0.5712201720866509,
      "learning_rate": 1.675740233602321e-06,
      "loss": 0.5885,
      "step": 6268
    },
    {
      "epoch": 0.5653605086350724,
      "grad_norm": 1.5584262385017875,
      "learning_rate": 1.6751637662577385e-06,
      "loss": 0.7958,
      "step": 6269
    },
    {
      "epoch": 0.5654506921585426,
      "grad_norm": 0.7936242311816112,
      "learning_rate": 1.6745873266299753e-06,
      "loss": 0.5322,
      "step": 6270
    },
    {
      "epoch": 0.565540875682013,
      "grad_norm": 1.434584023531525,
      "learning_rate": 1.6740109147682148e-06,
      "loss": 0.7099,
      "step": 6271
    },
    {
      "epoch": 0.5656310592054832,
      "grad_norm": 1.5693842483424232,
      "learning_rate": 1.6734345307216418e-06,
      "loss": 0.7647,
      "step": 6272
    },
    {
      "epoch": 0.5657212427289534,
      "grad_norm": 2.158710016301115,
      "learning_rate": 1.6728581745394346e-06,
      "loss": 0.7253,
      "step": 6273
    },
    {
      "epoch": 0.5658114262524236,
      "grad_norm": 1.9807841935707267,
      "learning_rate": 1.672281846270772e-06,
      "loss": 0.8269,
      "step": 6274
    },
    {
      "epoch": 0.565901609775894,
      "grad_norm": 1.7764531513739643,
      "learning_rate": 1.6717055459648295e-06,
      "loss": 0.7462,
      "step": 6275
    },
    {
      "epoch": 0.5659917932993642,
      "grad_norm": 1.7955361727430568,
      "learning_rate": 1.6711292736707793e-06,
      "loss": 0.7042,
      "step": 6276
    },
    {
      "epoch": 0.5660819768228345,
      "grad_norm": 2.161329947339339,
      "learning_rate": 1.6705530294377938e-06,
      "loss": 0.8202,
      "step": 6277
    },
    {
      "epoch": 0.5661721603463047,
      "grad_norm": 2.6776285074064208,
      "learning_rate": 1.6699768133150395e-06,
      "loss": 0.7389,
      "step": 6278
    },
    {
      "epoch": 0.566262343869775,
      "grad_norm": 2.0274975668707946,
      "learning_rate": 1.6694006253516837e-06,
      "loss": 0.7371,
      "step": 6279
    },
    {
      "epoch": 0.5663525273932453,
      "grad_norm": 1.7870401337764585,
      "learning_rate": 1.6688244655968896e-06,
      "loss": 0.7468,
      "step": 6280
    },
    {
      "epoch": 0.5664427109167155,
      "grad_norm": 4.117068670827294,
      "learning_rate": 1.6682483340998175e-06,
      "loss": 0.6763,
      "step": 6281
    },
    {
      "epoch": 0.5665328944401857,
      "grad_norm": 1.6248342722966578,
      "learning_rate": 1.6676722309096276e-06,
      "loss": 0.7109,
      "step": 6282
    },
    {
      "epoch": 0.5666230779636561,
      "grad_norm": 0.7356486942361342,
      "learning_rate": 1.6670961560754744e-06,
      "loss": 0.6477,
      "step": 6283
    },
    {
      "epoch": 0.5667132614871263,
      "grad_norm": 1.6172932023596345,
      "learning_rate": 1.6665201096465138e-06,
      "loss": 0.7443,
      "step": 6284
    },
    {
      "epoch": 0.5668034450105965,
      "grad_norm": 1.6999242260819056,
      "learning_rate": 1.6659440916718961e-06,
      "loss": 0.7917,
      "step": 6285
    },
    {
      "epoch": 0.5668936285340668,
      "grad_norm": 1.705518947392063,
      "learning_rate": 1.6653681022007696e-06,
      "loss": 0.7585,
      "step": 6286
    },
    {
      "epoch": 0.5669838120575371,
      "grad_norm": 2.3748876645475194,
      "learning_rate": 1.6647921412822825e-06,
      "loss": 0.678,
      "step": 6287
    },
    {
      "epoch": 0.5670739955810074,
      "grad_norm": 1.9254057171803383,
      "learning_rate": 1.6642162089655782e-06,
      "loss": 0.749,
      "step": 6288
    },
    {
      "epoch": 0.5671641791044776,
      "grad_norm": 2.365152119750556,
      "learning_rate": 1.663640305299798e-06,
      "loss": 0.7131,
      "step": 6289
    },
    {
      "epoch": 0.5672543626279479,
      "grad_norm": 0.6597982033816389,
      "learning_rate": 1.6630644303340824e-06,
      "loss": 0.6105,
      "step": 6290
    },
    {
      "epoch": 0.5673445461514182,
      "grad_norm": 2.8213102736687503,
      "learning_rate": 1.662488584117567e-06,
      "loss": 0.7562,
      "step": 6291
    },
    {
      "epoch": 0.5674347296748884,
      "grad_norm": 1.5669378962330462,
      "learning_rate": 1.6619127666993867e-06,
      "loss": 0.8063,
      "step": 6292
    },
    {
      "epoch": 0.5675249131983586,
      "grad_norm": 2.1387336338828313,
      "learning_rate": 1.6613369781286727e-06,
      "loss": 0.7504,
      "step": 6293
    },
    {
      "epoch": 0.567615096721829,
      "grad_norm": 2.8204421755917872,
      "learning_rate": 1.6607612184545562e-06,
      "loss": 0.6838,
      "step": 6294
    },
    {
      "epoch": 0.5677052802452992,
      "grad_norm": 2.129623880086807,
      "learning_rate": 1.6601854877261617e-06,
      "loss": 0.822,
      "step": 6295
    },
    {
      "epoch": 0.5677954637687694,
      "grad_norm": 1.957405587522351,
      "learning_rate": 1.6596097859926163e-06,
      "loss": 0.7473,
      "step": 6296
    },
    {
      "epoch": 0.5678856472922397,
      "grad_norm": 2.379243372689994,
      "learning_rate": 1.6590341133030407e-06,
      "loss": 0.8406,
      "step": 6297
    },
    {
      "epoch": 0.56797583081571,
      "grad_norm": 2.663726337959671,
      "learning_rate": 1.658458469706554e-06,
      "loss": 0.8239,
      "step": 6298
    },
    {
      "epoch": 0.5680660143391802,
      "grad_norm": 1.5623023861852503,
      "learning_rate": 1.6578828552522746e-06,
      "loss": 0.7017,
      "step": 6299
    },
    {
      "epoch": 0.5681561978626505,
      "grad_norm": 1.8852200765669413,
      "learning_rate": 1.6573072699893156e-06,
      "loss": 0.795,
      "step": 6300
    },
    {
      "epoch": 0.5682463813861207,
      "grad_norm": 1.9827938238275904,
      "learning_rate": 1.6567317139667906e-06,
      "loss": 0.722,
      "step": 6301
    },
    {
      "epoch": 0.5683365649095911,
      "grad_norm": 1.8378846576849195,
      "learning_rate": 1.6561561872338087e-06,
      "loss": 0.7665,
      "step": 6302
    },
    {
      "epoch": 0.5684267484330613,
      "grad_norm": 1.7719708398266731,
      "learning_rate": 1.6555806898394764e-06,
      "loss": 0.6612,
      "step": 6303
    },
    {
      "epoch": 0.5685169319565315,
      "grad_norm": 1.7767061200631078,
      "learning_rate": 1.6550052218328987e-06,
      "loss": 0.7243,
      "step": 6304
    },
    {
      "epoch": 0.5686071154800018,
      "grad_norm": 1.5797828803995475,
      "learning_rate": 1.6544297832631777e-06,
      "loss": 0.7223,
      "step": 6305
    },
    {
      "epoch": 0.5686972990034721,
      "grad_norm": 2.013007660329601,
      "learning_rate": 1.6538543741794135e-06,
      "loss": 0.7514,
      "step": 6306
    },
    {
      "epoch": 0.5687874825269423,
      "grad_norm": 2.269717919556677,
      "learning_rate": 1.6532789946307028e-06,
      "loss": 0.7175,
      "step": 6307
    },
    {
      "epoch": 0.5688776660504126,
      "grad_norm": 0.6403318059867636,
      "learning_rate": 1.6527036446661393e-06,
      "loss": 0.6134,
      "step": 6308
    },
    {
      "epoch": 0.5689678495738828,
      "grad_norm": 1.6527511058475348,
      "learning_rate": 1.6521283243348165e-06,
      "loss": 0.7796,
      "step": 6309
    },
    {
      "epoch": 0.5690580330973531,
      "grad_norm": 1.9036171663546322,
      "learning_rate": 1.6515530336858227e-06,
      "loss": 0.7414,
      "step": 6310
    },
    {
      "epoch": 0.5691482166208234,
      "grad_norm": 2.2712497737634574,
      "learning_rate": 1.6509777727682457e-06,
      "loss": 0.7993,
      "step": 6311
    },
    {
      "epoch": 0.5692384001442936,
      "grad_norm": 1.7821307096441366,
      "learning_rate": 1.65040254163117e-06,
      "loss": 0.7897,
      "step": 6312
    },
    {
      "epoch": 0.569328583667764,
      "grad_norm": 2.1532394843815235,
      "learning_rate": 1.649827340323676e-06,
      "loss": 0.6554,
      "step": 6313
    },
    {
      "epoch": 0.5694187671912342,
      "grad_norm": 2.202849350587338,
      "learning_rate": 1.6492521688948454e-06,
      "loss": 0.7127,
      "step": 6314
    },
    {
      "epoch": 0.5695089507147044,
      "grad_norm": 1.9380555125613113,
      "learning_rate": 1.6486770273937526e-06,
      "loss": 0.7698,
      "step": 6315
    },
    {
      "epoch": 0.5695991342381747,
      "grad_norm": 1.4098460524636507,
      "learning_rate": 1.6481019158694738e-06,
      "loss": 0.7283,
      "step": 6316
    },
    {
      "epoch": 0.569689317761645,
      "grad_norm": 0.754646056875564,
      "learning_rate": 1.6475268343710792e-06,
      "loss": 0.653,
      "step": 6317
    },
    {
      "epoch": 0.5697795012851152,
      "grad_norm": 1.8186112341100553,
      "learning_rate": 1.6469517829476396e-06,
      "loss": 0.6789,
      "step": 6318
    },
    {
      "epoch": 0.5698696848085855,
      "grad_norm": 2.1673216807790334,
      "learning_rate": 1.64637676164822e-06,
      "loss": 0.7336,
      "step": 6319
    },
    {
      "epoch": 0.5699598683320557,
      "grad_norm": 1.4202485056477967,
      "learning_rate": 1.6458017705218848e-06,
      "loss": 0.7832,
      "step": 6320
    },
    {
      "epoch": 0.570050051855526,
      "grad_norm": 1.6237933740850194,
      "learning_rate": 1.645226809617696e-06,
      "loss": 0.7753,
      "step": 6321
    },
    {
      "epoch": 0.5701402353789963,
      "grad_norm": 2.316440112427712,
      "learning_rate": 1.6446518789847112e-06,
      "loss": 0.7815,
      "step": 6322
    },
    {
      "epoch": 0.5702304189024665,
      "grad_norm": 2.049429195300558,
      "learning_rate": 1.6440769786719883e-06,
      "loss": 0.7735,
      "step": 6323
    },
    {
      "epoch": 0.5703206024259367,
      "grad_norm": 1.4284822146624059,
      "learning_rate": 1.6435021087285803e-06,
      "loss": 0.7882,
      "step": 6324
    },
    {
      "epoch": 0.5704107859494071,
      "grad_norm": 1.6980263922249934,
      "learning_rate": 1.642927269203537e-06,
      "loss": 0.7796,
      "step": 6325
    },
    {
      "epoch": 0.5705009694728773,
      "grad_norm": 2.247738867637841,
      "learning_rate": 1.642352460145909e-06,
      "loss": 0.8022,
      "step": 6326
    },
    {
      "epoch": 0.5705911529963476,
      "grad_norm": 1.7317366594274606,
      "learning_rate": 1.6417776816047402e-06,
      "loss": 0.7535,
      "step": 6327
    },
    {
      "epoch": 0.5706813365198178,
      "grad_norm": 1.61677464914356,
      "learning_rate": 1.6412029336290755e-06,
      "loss": 0.7363,
      "step": 6328
    },
    {
      "epoch": 0.5707715200432881,
      "grad_norm": 2.953739324381711,
      "learning_rate": 1.6406282162679551e-06,
      "loss": 0.7493,
      "step": 6329
    },
    {
      "epoch": 0.5708617035667584,
      "grad_norm": 1.8516732355797296,
      "learning_rate": 1.6400535295704162e-06,
      "loss": 0.7232,
      "step": 6330
    },
    {
      "epoch": 0.5709518870902286,
      "grad_norm": 1.5737503907413013,
      "learning_rate": 1.6394788735854955e-06,
      "loss": 0.6883,
      "step": 6331
    },
    {
      "epoch": 0.5710420706136988,
      "grad_norm": 1.8476811542405243,
      "learning_rate": 1.6389042483622246e-06,
      "loss": 0.8155,
      "step": 6332
    },
    {
      "epoch": 0.5711322541371692,
      "grad_norm": 1.5223876465681596,
      "learning_rate": 1.638329653949635e-06,
      "loss": 0.7439,
      "step": 6333
    },
    {
      "epoch": 0.5712224376606394,
      "grad_norm": 1.4666369232860919,
      "learning_rate": 1.637755090396753e-06,
      "loss": 0.7155,
      "step": 6334
    },
    {
      "epoch": 0.5713126211841096,
      "grad_norm": 0.6475832954427672,
      "learning_rate": 1.6371805577526039e-06,
      "loss": 0.6107,
      "step": 6335
    },
    {
      "epoch": 0.5714028047075799,
      "grad_norm": 1.6437539191253703,
      "learning_rate": 1.636606056066211e-06,
      "loss": 0.6984,
      "step": 6336
    },
    {
      "epoch": 0.5714929882310502,
      "grad_norm": 1.949129947328989,
      "learning_rate": 1.636031585386592e-06,
      "loss": 0.6818,
      "step": 6337
    },
    {
      "epoch": 0.5715831717545204,
      "grad_norm": 1.428259464669914,
      "learning_rate": 1.635457145762766e-06,
      "loss": 0.8101,
      "step": 6338
    },
    {
      "epoch": 0.5716733552779907,
      "grad_norm": 6.416277522717916,
      "learning_rate": 1.6348827372437456e-06,
      "loss": 0.7811,
      "step": 6339
    },
    {
      "epoch": 0.571763538801461,
      "grad_norm": 1.5270841333960519,
      "learning_rate": 1.634308359878544e-06,
      "loss": 0.7546,
      "step": 6340
    },
    {
      "epoch": 0.5718537223249313,
      "grad_norm": 1.5804826103202332,
      "learning_rate": 1.6337340137161695e-06,
      "loss": 0.8246,
      "step": 6341
    },
    {
      "epoch": 0.5719439058484015,
      "grad_norm": 1.7057430897695318,
      "learning_rate": 1.6331596988056277e-06,
      "loss": 0.756,
      "step": 6342
    },
    {
      "epoch": 0.5720340893718717,
      "grad_norm": 1.444361193205913,
      "learning_rate": 1.632585415195924e-06,
      "loss": 0.7767,
      "step": 6343
    },
    {
      "epoch": 0.5721242728953421,
      "grad_norm": 1.72125969501056,
      "learning_rate": 1.6320111629360583e-06,
      "loss": 0.7801,
      "step": 6344
    },
    {
      "epoch": 0.5722144564188123,
      "grad_norm": 1.9608883794582332,
      "learning_rate": 1.631436942075029e-06,
      "loss": 0.7323,
      "step": 6345
    },
    {
      "epoch": 0.5723046399422825,
      "grad_norm": 1.9503555193842987,
      "learning_rate": 1.630862752661833e-06,
      "loss": 0.7269,
      "step": 6346
    },
    {
      "epoch": 0.5723948234657528,
      "grad_norm": 2.1889114186159664,
      "learning_rate": 1.6302885947454612e-06,
      "loss": 0.6768,
      "step": 6347
    },
    {
      "epoch": 0.5724850069892231,
      "grad_norm": 2.454424179979356,
      "learning_rate": 1.6297144683749057e-06,
      "loss": 0.7644,
      "step": 6348
    },
    {
      "epoch": 0.5725751905126933,
      "grad_norm": 2.0982041309424018,
      "learning_rate": 1.629140373599153e-06,
      "loss": 0.7566,
      "step": 6349
    },
    {
      "epoch": 0.5726653740361636,
      "grad_norm": 1.6850667293930304,
      "learning_rate": 1.628566310467189e-06,
      "loss": 0.7632,
      "step": 6350
    },
    {
      "epoch": 0.5727555575596338,
      "grad_norm": 1.697155982932819,
      "learning_rate": 1.6279922790279957e-06,
      "loss": 0.7124,
      "step": 6351
    },
    {
      "epoch": 0.5728457410831042,
      "grad_norm": 1.7921441868650754,
      "learning_rate": 1.6274182793305512e-06,
      "loss": 0.7695,
      "step": 6352
    },
    {
      "epoch": 0.5729359246065744,
      "grad_norm": 2.2525568315828344,
      "learning_rate": 1.626844311423835e-06,
      "loss": 0.7835,
      "step": 6353
    },
    {
      "epoch": 0.5730261081300446,
      "grad_norm": 2.0597755470955668,
      "learning_rate": 1.6262703753568181e-06,
      "loss": 0.83,
      "step": 6354
    },
    {
      "epoch": 0.5731162916535149,
      "grad_norm": 1.9368019076918197,
      "learning_rate": 1.6256964711784747e-06,
      "loss": 0.8242,
      "step": 6355
    },
    {
      "epoch": 0.5732064751769852,
      "grad_norm": 1.6031752491115916,
      "learning_rate": 1.6251225989377723e-06,
      "loss": 0.765,
      "step": 6356
    },
    {
      "epoch": 0.5732966587004554,
      "grad_norm": 1.9796291881665595,
      "learning_rate": 1.624548758683676e-06,
      "loss": 0.7589,
      "step": 6357
    },
    {
      "epoch": 0.5733868422239257,
      "grad_norm": 1.618190435955491,
      "learning_rate": 1.6239749504651505e-06,
      "loss": 0.8433,
      "step": 6358
    },
    {
      "epoch": 0.5734770257473959,
      "grad_norm": 1.5330742848682168,
      "learning_rate": 1.6234011743311552e-06,
      "loss": 0.7316,
      "step": 6359
    },
    {
      "epoch": 0.5735672092708662,
      "grad_norm": 1.772451656096567,
      "learning_rate": 1.6228274303306483e-06,
      "loss": 0.6837,
      "step": 6360
    },
    {
      "epoch": 0.5736573927943365,
      "grad_norm": 1.5978187534352404,
      "learning_rate": 1.6222537185125847e-06,
      "loss": 0.8356,
      "step": 6361
    },
    {
      "epoch": 0.5737475763178067,
      "grad_norm": 1.699440145867811,
      "learning_rate": 1.6216800389259172e-06,
      "loss": 0.7699,
      "step": 6362
    },
    {
      "epoch": 0.573837759841277,
      "grad_norm": 1.4489954856133545,
      "learning_rate": 1.6211063916195949e-06,
      "loss": 0.7579,
      "step": 6363
    },
    {
      "epoch": 0.5739279433647473,
      "grad_norm": 1.8517019484918689,
      "learning_rate": 1.6205327766425633e-06,
      "loss": 0.7647,
      "step": 6364
    },
    {
      "epoch": 0.5740181268882175,
      "grad_norm": 1.4474820735789855,
      "learning_rate": 1.6199591940437689e-06,
      "loss": 0.7648,
      "step": 6365
    },
    {
      "epoch": 0.5741083104116878,
      "grad_norm": 1.5686728044339282,
      "learning_rate": 1.6193856438721505e-06,
      "loss": 0.7118,
      "step": 6366
    },
    {
      "epoch": 0.5741984939351581,
      "grad_norm": 1.7941908982349348,
      "learning_rate": 1.6188121261766483e-06,
      "loss": 0.7636,
      "step": 6367
    },
    {
      "epoch": 0.5742886774586283,
      "grad_norm": 1.7346158934313818,
      "learning_rate": 1.6182386410061976e-06,
      "loss": 0.6882,
      "step": 6368
    },
    {
      "epoch": 0.5743788609820986,
      "grad_norm": 9.921571645830102,
      "learning_rate": 1.61766518840973e-06,
      "loss": 0.663,
      "step": 6369
    },
    {
      "epoch": 0.5744690445055688,
      "grad_norm": 1.7938840437193233,
      "learning_rate": 1.6170917684361779e-06,
      "loss": 0.6272,
      "step": 6370
    },
    {
      "epoch": 0.5745592280290391,
      "grad_norm": 1.71039666811937,
      "learning_rate": 1.6165183811344662e-06,
      "loss": 0.8308,
      "step": 6371
    },
    {
      "epoch": 0.5746494115525094,
      "grad_norm": 4.01729492613619,
      "learning_rate": 1.6159450265535218e-06,
      "loss": 0.7517,
      "step": 6372
    },
    {
      "epoch": 0.5747395950759796,
      "grad_norm": 1.486918468761943,
      "learning_rate": 1.6153717047422652e-06,
      "loss": 0.8411,
      "step": 6373
    },
    {
      "epoch": 0.5748297785994498,
      "grad_norm": 2.3385284242205504,
      "learning_rate": 1.6147984157496155e-06,
      "loss": 0.6445,
      "step": 6374
    },
    {
      "epoch": 0.5749199621229202,
      "grad_norm": 1.4846766918176781,
      "learning_rate": 1.6142251596244886e-06,
      "loss": 0.6792,
      "step": 6375
    },
    {
      "epoch": 0.5750101456463904,
      "grad_norm": 2.2625291390413045,
      "learning_rate": 1.6136519364157983e-06,
      "loss": 0.8115,
      "step": 6376
    },
    {
      "epoch": 0.5751003291698606,
      "grad_norm": 1.5885217217199066,
      "learning_rate": 1.6130787461724555e-06,
      "loss": 0.6939,
      "step": 6377
    },
    {
      "epoch": 0.5751905126933309,
      "grad_norm": 1.5075949081986721,
      "learning_rate": 1.6125055889433679e-06,
      "loss": 0.8289,
      "step": 6378
    },
    {
      "epoch": 0.5752806962168012,
      "grad_norm": 3.1028018656400596,
      "learning_rate": 1.6119324647774386e-06,
      "loss": 0.7772,
      "step": 6379
    },
    {
      "epoch": 0.5753708797402715,
      "grad_norm": 2.1281480471445557,
      "learning_rate": 1.6113593737235724e-06,
      "loss": 0.7665,
      "step": 6380
    },
    {
      "epoch": 0.5754610632637417,
      "grad_norm": 1.579580938440894,
      "learning_rate": 1.6107863158306665e-06,
      "loss": 0.7056,
      "step": 6381
    },
    {
      "epoch": 0.5755512467872119,
      "grad_norm": 2.096867759453315,
      "learning_rate": 1.610213291147619e-06,
      "loss": 0.6492,
      "step": 6382
    },
    {
      "epoch": 0.5756414303106823,
      "grad_norm": 1.4177953861070556,
      "learning_rate": 1.609640299723322e-06,
      "loss": 0.7753,
      "step": 6383
    },
    {
      "epoch": 0.5757316138341525,
      "grad_norm": 1.5043865279170836,
      "learning_rate": 1.609067341606668e-06,
      "loss": 0.7438,
      "step": 6384
    },
    {
      "epoch": 0.5758217973576227,
      "grad_norm": 1.8269085871936543,
      "learning_rate": 1.6084944168465438e-06,
      "loss": 0.7459,
      "step": 6385
    },
    {
      "epoch": 0.5759119808810931,
      "grad_norm": 1.8809350494657804,
      "learning_rate": 1.6079215254918339e-06,
      "loss": 0.7767,
      "step": 6386
    },
    {
      "epoch": 0.5760021644045633,
      "grad_norm": 1.931749115183556,
      "learning_rate": 1.6073486675914222e-06,
      "loss": 0.6621,
      "step": 6387
    },
    {
      "epoch": 0.5760923479280335,
      "grad_norm": 4.274691047435521,
      "learning_rate": 1.606775843194187e-06,
      "loss": 0.7847,
      "step": 6388
    },
    {
      "epoch": 0.5761825314515038,
      "grad_norm": 2.0407073303719088,
      "learning_rate": 1.6062030523490053e-06,
      "loss": 0.7412,
      "step": 6389
    },
    {
      "epoch": 0.5762727149749741,
      "grad_norm": 1.6532365267366256,
      "learning_rate": 1.60563029510475e-06,
      "loss": 0.6818,
      "step": 6390
    },
    {
      "epoch": 0.5763628984984444,
      "grad_norm": 0.6572564899041455,
      "learning_rate": 1.6050575715102927e-06,
      "loss": 0.5695,
      "step": 6391
    },
    {
      "epoch": 0.5764530820219146,
      "grad_norm": 2.3267290679557355,
      "learning_rate": 1.6044848816145014e-06,
      "loss": 0.808,
      "step": 6392
    },
    {
      "epoch": 0.5765432655453848,
      "grad_norm": 2.126493490137845,
      "learning_rate": 1.60391222546624e-06,
      "loss": 0.7811,
      "step": 6393
    },
    {
      "epoch": 0.5766334490688552,
      "grad_norm": 1.7816742508919816,
      "learning_rate": 1.6033396031143725e-06,
      "loss": 0.7508,
      "step": 6394
    },
    {
      "epoch": 0.5767236325923254,
      "grad_norm": 2.0165851287541647,
      "learning_rate": 1.602767014607757e-06,
      "loss": 0.7563,
      "step": 6395
    },
    {
      "epoch": 0.5768138161157956,
      "grad_norm": 3.8659610537458327,
      "learning_rate": 1.6021944599952493e-06,
      "loss": 0.7645,
      "step": 6396
    },
    {
      "epoch": 0.5769039996392659,
      "grad_norm": 2.2711691535161043,
      "learning_rate": 1.6016219393257048e-06,
      "loss": 0.7694,
      "step": 6397
    },
    {
      "epoch": 0.5769941831627362,
      "grad_norm": 1.9601294079702978,
      "learning_rate": 1.6010494526479722e-06,
      "loss": 0.7079,
      "step": 6398
    },
    {
      "epoch": 0.5770843666862064,
      "grad_norm": 2.205407657783866,
      "learning_rate": 1.6004770000109006e-06,
      "loss": 0.7013,
      "step": 6399
    },
    {
      "epoch": 0.5771745502096767,
      "grad_norm": 1.5991658211779218,
      "learning_rate": 1.5999045814633348e-06,
      "loss": 0.7229,
      "step": 6400
    },
    {
      "epoch": 0.5772647337331469,
      "grad_norm": 1.4319136334539402,
      "learning_rate": 1.5993321970541151e-06,
      "loss": 0.7652,
      "step": 6401
    },
    {
      "epoch": 0.5773549172566173,
      "grad_norm": 3.3304416670605224,
      "learning_rate": 1.5987598468320825e-06,
      "loss": 0.7548,
      "step": 6402
    },
    {
      "epoch": 0.5774451007800875,
      "grad_norm": 1.5162963953176731,
      "learning_rate": 1.5981875308460717e-06,
      "loss": 0.7438,
      "step": 6403
    },
    {
      "epoch": 0.5775352843035577,
      "grad_norm": 2.7553861064322764,
      "learning_rate": 1.5976152491449169e-06,
      "loss": 0.742,
      "step": 6404
    },
    {
      "epoch": 0.577625467827028,
      "grad_norm": 1.814452796500744,
      "learning_rate": 1.5970430017774468e-06,
      "loss": 0.818,
      "step": 6405
    },
    {
      "epoch": 0.5777156513504983,
      "grad_norm": 8.108271823295592,
      "learning_rate": 1.5964707887924904e-06,
      "loss": 0.7675,
      "step": 6406
    },
    {
      "epoch": 0.5778058348739685,
      "grad_norm": 1.3304807078909973,
      "learning_rate": 1.5958986102388714e-06,
      "loss": 0.6418,
      "step": 6407
    },
    {
      "epoch": 0.5778960183974388,
      "grad_norm": 1.6746344921950405,
      "learning_rate": 1.5953264661654104e-06,
      "loss": 0.7093,
      "step": 6408
    },
    {
      "epoch": 0.5779862019209091,
      "grad_norm": 2.016053146903051,
      "learning_rate": 1.5947543566209276e-06,
      "loss": 0.786,
      "step": 6409
    },
    {
      "epoch": 0.5780763854443793,
      "grad_norm": 2.6221692398702974,
      "learning_rate": 1.5941822816542367e-06,
      "loss": 0.7356,
      "step": 6410
    },
    {
      "epoch": 0.5781665689678496,
      "grad_norm": 2.0790595914475465,
      "learning_rate": 1.5936102413141519e-06,
      "loss": 0.715,
      "step": 6411
    },
    {
      "epoch": 0.5782567524913198,
      "grad_norm": 1.892251127198607,
      "learning_rate": 1.5930382356494823e-06,
      "loss": 0.671,
      "step": 6412
    },
    {
      "epoch": 0.5783469360147901,
      "grad_norm": 1.5802102994393188,
      "learning_rate": 1.5924662647090335e-06,
      "loss": 0.801,
      "step": 6413
    },
    {
      "epoch": 0.5784371195382604,
      "grad_norm": 2.4616842438783326,
      "learning_rate": 1.5918943285416108e-06,
      "loss": 0.7727,
      "step": 6414
    },
    {
      "epoch": 0.5785273030617306,
      "grad_norm": 1.4619636399826597,
      "learning_rate": 1.5913224271960139e-06,
      "loss": 0.6922,
      "step": 6415
    },
    {
      "epoch": 0.5786174865852008,
      "grad_norm": 1.6340350816899125,
      "learning_rate": 1.590750560721041e-06,
      "loss": 0.7689,
      "step": 6416
    },
    {
      "epoch": 0.5787076701086712,
      "grad_norm": 1.8412692304281602,
      "learning_rate": 1.5901787291654874e-06,
      "loss": 0.7217,
      "step": 6417
    },
    {
      "epoch": 0.5787978536321414,
      "grad_norm": 2.5087018203860736,
      "learning_rate": 1.5896069325781435e-06,
      "loss": 0.6262,
      "step": 6418
    },
    {
      "epoch": 0.5788880371556117,
      "grad_norm": 1.8475587589278561,
      "learning_rate": 1.5890351710077998e-06,
      "loss": 0.7954,
      "step": 6419
    },
    {
      "epoch": 0.5789782206790819,
      "grad_norm": 1.6100198879553682,
      "learning_rate": 1.5884634445032406e-06,
      "loss": 0.8089,
      "step": 6420
    },
    {
      "epoch": 0.5790684042025522,
      "grad_norm": 0.6011001990348906,
      "learning_rate": 1.5878917531132501e-06,
      "loss": 0.5981,
      "step": 6421
    },
    {
      "epoch": 0.5791585877260225,
      "grad_norm": 1.807817791112544,
      "learning_rate": 1.5873200968866077e-06,
      "loss": 0.6319,
      "step": 6422
    },
    {
      "epoch": 0.5792487712494927,
      "grad_norm": 2.2973694561055433,
      "learning_rate": 1.586748475872089e-06,
      "loss": 0.7312,
      "step": 6423
    },
    {
      "epoch": 0.5793389547729629,
      "grad_norm": 1.861707938831617,
      "learning_rate": 1.58617689011847e-06,
      "loss": 0.6201,
      "step": 6424
    },
    {
      "epoch": 0.5794291382964333,
      "grad_norm": 1.7471239434630907,
      "learning_rate": 1.5856053396745198e-06,
      "loss": 0.7885,
      "step": 6425
    },
    {
      "epoch": 0.5795193218199035,
      "grad_norm": 2.4296061511389246,
      "learning_rate": 1.5850338245890078e-06,
      "loss": 0.8162,
      "step": 6426
    },
    {
      "epoch": 0.5796095053433737,
      "grad_norm": 1.8205684105419875,
      "learning_rate": 1.5844623449106974e-06,
      "loss": 0.7393,
      "step": 6427
    },
    {
      "epoch": 0.579699688866844,
      "grad_norm": 2.047917340868053,
      "learning_rate": 1.583890900688351e-06,
      "loss": 0.7924,
      "step": 6428
    },
    {
      "epoch": 0.5797898723903143,
      "grad_norm": 2.2620830341347578,
      "learning_rate": 1.583319491970728e-06,
      "loss": 0.8281,
      "step": 6429
    },
    {
      "epoch": 0.5798800559137846,
      "grad_norm": 1.4358234580230442,
      "learning_rate": 1.5827481188065828e-06,
      "loss": 0.7657,
      "step": 6430
    },
    {
      "epoch": 0.5799702394372548,
      "grad_norm": 2.0438599256531544,
      "learning_rate": 1.5821767812446689e-06,
      "loss": 0.7263,
      "step": 6431
    },
    {
      "epoch": 0.5800604229607251,
      "grad_norm": 1.7069830636702235,
      "learning_rate": 1.581605479333736e-06,
      "loss": 0.8658,
      "step": 6432
    },
    {
      "epoch": 0.5801506064841954,
      "grad_norm": 2.134003359161851,
      "learning_rate": 1.5810342131225308e-06,
      "loss": 0.7608,
      "step": 6433
    },
    {
      "epoch": 0.5802407900076656,
      "grad_norm": 2.940693681585768,
      "learning_rate": 1.580462982659797e-06,
      "loss": 0.7997,
      "step": 6434
    },
    {
      "epoch": 0.5803309735311358,
      "grad_norm": 2.9019622346614655,
      "learning_rate": 1.5798917879942736e-06,
      "loss": 0.8304,
      "step": 6435
    },
    {
      "epoch": 0.5804211570546062,
      "grad_norm": 1.4009202623296015,
      "learning_rate": 1.5793206291747006e-06,
      "loss": 0.6963,
      "step": 6436
    },
    {
      "epoch": 0.5805113405780764,
      "grad_norm": 1.3859399462208453,
      "learning_rate": 1.57874950624981e-06,
      "loss": 0.6882,
      "step": 6437
    },
    {
      "epoch": 0.5806015241015466,
      "grad_norm": 2.0142805953574334,
      "learning_rate": 1.5781784192683351e-06,
      "loss": 0.7535,
      "step": 6438
    },
    {
      "epoch": 0.5806917076250169,
      "grad_norm": 1.3505560153782732,
      "learning_rate": 1.5776073682790033e-06,
      "loss": 0.7131,
      "step": 6439
    },
    {
      "epoch": 0.5807818911484872,
      "grad_norm": 2.450896982531484,
      "learning_rate": 1.5770363533305393e-06,
      "loss": 0.7742,
      "step": 6440
    },
    {
      "epoch": 0.5808720746719575,
      "grad_norm": 2.005336080337902,
      "learning_rate": 1.5764653744716665e-06,
      "loss": 0.7594,
      "step": 6441
    },
    {
      "epoch": 0.5809622581954277,
      "grad_norm": 1.7064550199753215,
      "learning_rate": 1.575894431751103e-06,
      "loss": 0.7835,
      "step": 6442
    },
    {
      "epoch": 0.5810524417188979,
      "grad_norm": 2.973898829696057,
      "learning_rate": 1.575323525217565e-06,
      "loss": 0.7817,
      "step": 6443
    },
    {
      "epoch": 0.5811426252423683,
      "grad_norm": 1.6231536646483227,
      "learning_rate": 1.574752654919766e-06,
      "loss": 0.8368,
      "step": 6444
    },
    {
      "epoch": 0.5812328087658385,
      "grad_norm": 1.5957643830997197,
      "learning_rate": 1.5741818209064146e-06,
      "loss": 0.7218,
      "step": 6445
    },
    {
      "epoch": 0.5813229922893087,
      "grad_norm": 1.8334830902085486,
      "learning_rate": 1.5736110232262183e-06,
      "loss": 0.761,
      "step": 6446
    },
    {
      "epoch": 0.581413175812779,
      "grad_norm": 1.7046029913468181,
      "learning_rate": 1.5730402619278804e-06,
      "loss": 0.8011,
      "step": 6447
    },
    {
      "epoch": 0.5815033593362493,
      "grad_norm": 0.6598917865905084,
      "learning_rate": 1.5724695370601024e-06,
      "loss": 0.5564,
      "step": 6448
    },
    {
      "epoch": 0.5815935428597195,
      "grad_norm": 1.670333650917191,
      "learning_rate": 1.5718988486715798e-06,
      "loss": 0.6776,
      "step": 6449
    },
    {
      "epoch": 0.5816837263831898,
      "grad_norm": 1.622309952407461,
      "learning_rate": 1.5713281968110087e-06,
      "loss": 0.7374,
      "step": 6450
    },
    {
      "epoch": 0.58177390990666,
      "grad_norm": 1.6398034354784221,
      "learning_rate": 1.5707575815270796e-06,
      "loss": 0.7364,
      "step": 6451
    },
    {
      "epoch": 0.5818640934301303,
      "grad_norm": 2.0648190725282958,
      "learning_rate": 1.57018700286848e-06,
      "loss": 0.7587,
      "step": 6452
    },
    {
      "epoch": 0.5819542769536006,
      "grad_norm": 2.4742622162084773,
      "learning_rate": 1.5696164608838956e-06,
      "loss": 0.7701,
      "step": 6453
    },
    {
      "epoch": 0.5820444604770708,
      "grad_norm": 2.1255605743613946,
      "learning_rate": 1.5690459556220073e-06,
      "loss": 0.7743,
      "step": 6454
    },
    {
      "epoch": 0.582134644000541,
      "grad_norm": 1.847662121237203,
      "learning_rate": 1.5684754871314949e-06,
      "loss": 0.8122,
      "step": 6455
    },
    {
      "epoch": 0.5822248275240114,
      "grad_norm": 1.6313138956585955,
      "learning_rate": 1.5679050554610335e-06,
      "loss": 0.7548,
      "step": 6456
    },
    {
      "epoch": 0.5823150110474816,
      "grad_norm": 1.4166662365781841,
      "learning_rate": 1.567334660659295e-06,
      "loss": 0.7577,
      "step": 6457
    },
    {
      "epoch": 0.5824051945709519,
      "grad_norm": 1.622168641896608,
      "learning_rate": 1.5667643027749488e-06,
      "loss": 0.7701,
      "step": 6458
    },
    {
      "epoch": 0.5824953780944222,
      "grad_norm": 1.9004545446076828,
      "learning_rate": 1.5661939818566614e-06,
      "loss": 0.829,
      "step": 6459
    },
    {
      "epoch": 0.5825855616178924,
      "grad_norm": 3.49921531055347,
      "learning_rate": 1.5656236979530956e-06,
      "loss": 0.6199,
      "step": 6460
    },
    {
      "epoch": 0.5826757451413627,
      "grad_norm": 1.5585555834387463,
      "learning_rate": 1.5650534511129106e-06,
      "loss": 0.6966,
      "step": 6461
    },
    {
      "epoch": 0.5827659286648329,
      "grad_norm": 1.8945095414953108,
      "learning_rate": 1.5644832413847635e-06,
      "loss": 0.7394,
      "step": 6462
    },
    {
      "epoch": 0.5828561121883032,
      "grad_norm": 1.314496383975436,
      "learning_rate": 1.5639130688173082e-06,
      "loss": 0.7969,
      "step": 6463
    },
    {
      "epoch": 0.5829462957117735,
      "grad_norm": 3.0640010658341987,
      "learning_rate": 1.5633429334591932e-06,
      "loss": 0.7286,
      "step": 6464
    },
    {
      "epoch": 0.5830364792352437,
      "grad_norm": 1.8715564259271151,
      "learning_rate": 1.562772835359068e-06,
      "loss": 0.7845,
      "step": 6465
    },
    {
      "epoch": 0.5831266627587139,
      "grad_norm": 2.0342790759745215,
      "learning_rate": 1.5622027745655753e-06,
      "loss": 0.7209,
      "step": 6466
    },
    {
      "epoch": 0.5832168462821843,
      "grad_norm": 1.6533215382498778,
      "learning_rate": 1.561632751127355e-06,
      "loss": 0.7099,
      "step": 6467
    },
    {
      "epoch": 0.5833070298056545,
      "grad_norm": 1.8188089210430736,
      "learning_rate": 1.561062765093046e-06,
      "loss": 0.7967,
      "step": 6468
    },
    {
      "epoch": 0.5833972133291248,
      "grad_norm": 1.9064027068461076,
      "learning_rate": 1.5604928165112817e-06,
      "loss": 0.749,
      "step": 6469
    },
    {
      "epoch": 0.583487396852595,
      "grad_norm": 1.8098166103085165,
      "learning_rate": 1.5599229054306945e-06,
      "loss": 0.7467,
      "step": 6470
    },
    {
      "epoch": 0.5835775803760653,
      "grad_norm": 1.6815330125356784,
      "learning_rate": 1.5593530318999111e-06,
      "loss": 0.7474,
      "step": 6471
    },
    {
      "epoch": 0.5836677638995356,
      "grad_norm": 1.7837605769719973,
      "learning_rate": 1.5587831959675572e-06,
      "loss": 0.784,
      "step": 6472
    },
    {
      "epoch": 0.5837579474230058,
      "grad_norm": 1.5113952448284624,
      "learning_rate": 1.5582133976822534e-06,
      "loss": 0.7403,
      "step": 6473
    },
    {
      "epoch": 0.583848130946476,
      "grad_norm": 1.5956461229730454,
      "learning_rate": 1.5576436370926185e-06,
      "loss": 0.7073,
      "step": 6474
    },
    {
      "epoch": 0.5839383144699464,
      "grad_norm": 1.5682335752055487,
      "learning_rate": 1.5570739142472679e-06,
      "loss": 0.7353,
      "step": 6475
    },
    {
      "epoch": 0.5840284979934166,
      "grad_norm": 3.4266207459139864,
      "learning_rate": 1.5565042291948127e-06,
      "loss": 0.8165,
      "step": 6476
    },
    {
      "epoch": 0.5841186815168868,
      "grad_norm": 1.636524780724513,
      "learning_rate": 1.5559345819838624e-06,
      "loss": 0.7917,
      "step": 6477
    },
    {
      "epoch": 0.5842088650403571,
      "grad_norm": 1.330068579820169,
      "learning_rate": 1.5553649726630226e-06,
      "loss": 0.7312,
      "step": 6478
    },
    {
      "epoch": 0.5842990485638274,
      "grad_norm": 1.5291733965955165,
      "learning_rate": 1.5547954012808942e-06,
      "loss": 0.6837,
      "step": 6479
    },
    {
      "epoch": 0.5843892320872976,
      "grad_norm": 2.4342782663657387,
      "learning_rate": 1.5542258678860776e-06,
      "loss": 0.7311,
      "step": 6480
    },
    {
      "epoch": 0.5844794156107679,
      "grad_norm": 1.7355368219600034,
      "learning_rate": 1.553656372527167e-06,
      "loss": 0.8708,
      "step": 6481
    },
    {
      "epoch": 0.5845695991342382,
      "grad_norm": 2.6877632012229777,
      "learning_rate": 1.5530869152527568e-06,
      "loss": 0.8562,
      "step": 6482
    },
    {
      "epoch": 0.5846597826577085,
      "grad_norm": 1.5499242917921752,
      "learning_rate": 1.5525174961114353e-06,
      "loss": 0.7728,
      "step": 6483
    },
    {
      "epoch": 0.5847499661811787,
      "grad_norm": 1.720170283870604,
      "learning_rate": 1.5519481151517875e-06,
      "loss": 0.7915,
      "step": 6484
    },
    {
      "epoch": 0.5848401497046489,
      "grad_norm": 2.2162508792090136,
      "learning_rate": 1.551378772422398e-06,
      "loss": 0.7612,
      "step": 6485
    },
    {
      "epoch": 0.5849303332281193,
      "grad_norm": 1.4621960120249908,
      "learning_rate": 1.5508094679718447e-06,
      "loss": 0.7646,
      "step": 6486
    },
    {
      "epoch": 0.5850205167515895,
      "grad_norm": 1.761205289664416,
      "learning_rate": 1.5502402018487048e-06,
      "loss": 0.6583,
      "step": 6487
    },
    {
      "epoch": 0.5851107002750597,
      "grad_norm": 2.1653259114849197,
      "learning_rate": 1.54967097410155e-06,
      "loss": 0.759,
      "step": 6488
    },
    {
      "epoch": 0.58520088379853,
      "grad_norm": 1.5498299936152666,
      "learning_rate": 1.5491017847789519e-06,
      "loss": 0.6775,
      "step": 6489
    },
    {
      "epoch": 0.5852910673220003,
      "grad_norm": 1.7496986129634025,
      "learning_rate": 1.5485326339294755e-06,
      "loss": 0.6989,
      "step": 6490
    },
    {
      "epoch": 0.5853812508454705,
      "grad_norm": 1.5938830600618796,
      "learning_rate": 1.5479635216016832e-06,
      "loss": 0.7245,
      "step": 6491
    },
    {
      "epoch": 0.5854714343689408,
      "grad_norm": 2.1300819440311103,
      "learning_rate": 1.547394447844137e-06,
      "loss": 0.678,
      "step": 6492
    },
    {
      "epoch": 0.585561617892411,
      "grad_norm": 1.6130921481606815,
      "learning_rate": 1.546825412705391e-06,
      "loss": 0.6713,
      "step": 6493
    },
    {
      "epoch": 0.5856518014158814,
      "grad_norm": 2.9275915354060498,
      "learning_rate": 1.5462564162340007e-06,
      "loss": 0.78,
      "step": 6494
    },
    {
      "epoch": 0.5857419849393516,
      "grad_norm": 1.9512448225618595,
      "learning_rate": 1.5456874584785144e-06,
      "loss": 0.7625,
      "step": 6495
    },
    {
      "epoch": 0.5858321684628218,
      "grad_norm": 1.81897139365177,
      "learning_rate": 1.5451185394874785e-06,
      "loss": 0.7301,
      "step": 6496
    },
    {
      "epoch": 0.5859223519862921,
      "grad_norm": 1.728083359283895,
      "learning_rate": 1.5445496593094381e-06,
      "loss": 0.7646,
      "step": 6497
    },
    {
      "epoch": 0.5860125355097624,
      "grad_norm": 1.8607492577498654,
      "learning_rate": 1.5439808179929316e-06,
      "loss": 0.7694,
      "step": 6498
    },
    {
      "epoch": 0.5861027190332326,
      "grad_norm": 1.6411615856612791,
      "learning_rate": 1.543412015586496e-06,
      "loss": 0.7246,
      "step": 6499
    },
    {
      "epoch": 0.5861929025567029,
      "grad_norm": 1.766819845012738,
      "learning_rate": 1.5428432521386655e-06,
      "loss": 0.7208,
      "step": 6500
    },
    {
      "epoch": 0.5862830860801731,
      "grad_norm": 1.6949161470540168,
      "learning_rate": 1.5422745276979688e-06,
      "loss": 0.7868,
      "step": 6501
    },
    {
      "epoch": 0.5863732696036434,
      "grad_norm": 1.509071733482426,
      "learning_rate": 1.5417058423129336e-06,
      "loss": 0.7442,
      "step": 6502
    },
    {
      "epoch": 0.5864634531271137,
      "grad_norm": 1.3741373910941244,
      "learning_rate": 1.5411371960320822e-06,
      "loss": 0.7862,
      "step": 6503
    },
    {
      "epoch": 0.5865536366505839,
      "grad_norm": 1.4557636054463021,
      "learning_rate": 1.5405685889039363e-06,
      "loss": 0.7851,
      "step": 6504
    },
    {
      "epoch": 0.5866438201740543,
      "grad_norm": 2.0387095880391577,
      "learning_rate": 1.5400000209770118e-06,
      "loss": 0.7525,
      "step": 6505
    },
    {
      "epoch": 0.5867340036975245,
      "grad_norm": 1.9248573919209673,
      "learning_rate": 1.5394314922998208e-06,
      "loss": 0.7681,
      "step": 6506
    },
    {
      "epoch": 0.5868241872209947,
      "grad_norm": 2.3286306356982007,
      "learning_rate": 1.5388630029208756e-06,
      "loss": 0.8031,
      "step": 6507
    },
    {
      "epoch": 0.586914370744465,
      "grad_norm": 2.183064896580467,
      "learning_rate": 1.5382945528886806e-06,
      "loss": 0.7762,
      "step": 6508
    },
    {
      "epoch": 0.5870045542679353,
      "grad_norm": 2.4701857435989787,
      "learning_rate": 1.5377261422517412e-06,
      "loss": 0.745,
      "step": 6509
    },
    {
      "epoch": 0.5870947377914055,
      "grad_norm": 2.1946187399470567,
      "learning_rate": 1.5371577710585553e-06,
      "loss": 0.7128,
      "step": 6510
    },
    {
      "epoch": 0.5871849213148758,
      "grad_norm": 0.7276551838826665,
      "learning_rate": 1.536589439357621e-06,
      "loss": 0.5653,
      "step": 6511
    },
    {
      "epoch": 0.587275104838346,
      "grad_norm": 1.6034194229763412,
      "learning_rate": 1.5360211471974315e-06,
      "loss": 0.7576,
      "step": 6512
    },
    {
      "epoch": 0.5873652883618163,
      "grad_norm": 1.6234040125604705,
      "learning_rate": 1.5354528946264753e-06,
      "loss": 0.8094,
      "step": 6513
    },
    {
      "epoch": 0.5874554718852866,
      "grad_norm": 2.302795668183414,
      "learning_rate": 1.5348846816932399e-06,
      "loss": 0.6901,
      "step": 6514
    },
    {
      "epoch": 0.5875456554087568,
      "grad_norm": 0.6535260570762879,
      "learning_rate": 1.5343165084462077e-06,
      "loss": 0.5756,
      "step": 6515
    },
    {
      "epoch": 0.587635838932227,
      "grad_norm": 1.8038293028256027,
      "learning_rate": 1.5337483749338595e-06,
      "loss": 0.7073,
      "step": 6516
    },
    {
      "epoch": 0.5877260224556974,
      "grad_norm": 1.526095531498125,
      "learning_rate": 1.5331802812046708e-06,
      "loss": 0.7259,
      "step": 6517
    },
    {
      "epoch": 0.5878162059791676,
      "grad_norm": 15.006699655350554,
      "learning_rate": 1.5326122273071133e-06,
      "loss": 0.6056,
      "step": 6518
    },
    {
      "epoch": 0.5879063895026378,
      "grad_norm": 1.909634946119652,
      "learning_rate": 1.532044213289659e-06,
      "loss": 0.7904,
      "step": 6519
    },
    {
      "epoch": 0.5879965730261081,
      "grad_norm": 2.0201849405212116,
      "learning_rate": 1.5314762392007718e-06,
      "loss": 0.6411,
      "step": 6520
    },
    {
      "epoch": 0.5880867565495784,
      "grad_norm": 1.4893015970766477,
      "learning_rate": 1.530908305088916e-06,
      "loss": 0.8058,
      "step": 6521
    },
    {
      "epoch": 0.5881769400730487,
      "grad_norm": 1.8084193365972285,
      "learning_rate": 1.5303404110025501e-06,
      "loss": 0.7504,
      "step": 6522
    },
    {
      "epoch": 0.5882671235965189,
      "grad_norm": 9.718481127586044,
      "learning_rate": 1.5297725569901293e-06,
      "loss": 0.7105,
      "step": 6523
    },
    {
      "epoch": 0.5883573071199891,
      "grad_norm": 2.967204123114475,
      "learning_rate": 1.5292047431001077e-06,
      "loss": 0.6499,
      "step": 6524
    },
    {
      "epoch": 0.5884474906434595,
      "grad_norm": 0.6857281958603906,
      "learning_rate": 1.5286369693809321e-06,
      "loss": 0.5887,
      "step": 6525
    },
    {
      "epoch": 0.5885376741669297,
      "grad_norm": 2.1728198514188817,
      "learning_rate": 1.5280692358810506e-06,
      "loss": 0.762,
      "step": 6526
    },
    {
      "epoch": 0.5886278576903999,
      "grad_norm": 2.1826538902115393,
      "learning_rate": 1.527501542648904e-06,
      "loss": 0.7,
      "step": 6527
    },
    {
      "epoch": 0.5887180412138703,
      "grad_norm": 1.970770026622399,
      "learning_rate": 1.5269338897329308e-06,
      "loss": 0.6599,
      "step": 6528
    },
    {
      "epoch": 0.5888082247373405,
      "grad_norm": 1.5555496509000943,
      "learning_rate": 1.5263662771815662e-06,
      "loss": 0.781,
      "step": 6529
    },
    {
      "epoch": 0.5888984082608107,
      "grad_norm": 1.81933305446296,
      "learning_rate": 1.5257987050432429e-06,
      "loss": 0.8273,
      "step": 6530
    },
    {
      "epoch": 0.588988591784281,
      "grad_norm": 1.8511576793453977,
      "learning_rate": 1.5252311733663887e-06,
      "loss": 0.6626,
      "step": 6531
    },
    {
      "epoch": 0.5890787753077513,
      "grad_norm": 2.0226337500860865,
      "learning_rate": 1.5246636821994281e-06,
      "loss": 0.6961,
      "step": 6532
    },
    {
      "epoch": 0.5891689588312216,
      "grad_norm": 1.5098272905673413,
      "learning_rate": 1.524096231590784e-06,
      "loss": 0.7994,
      "step": 6533
    },
    {
      "epoch": 0.5892591423546918,
      "grad_norm": 1.8458711093570452,
      "learning_rate": 1.5235288215888736e-06,
      "loss": 0.7965,
      "step": 6534
    },
    {
      "epoch": 0.589349325878162,
      "grad_norm": 1.700309876641674,
      "learning_rate": 1.5229614522421102e-06,
      "loss": 0.7567,
      "step": 6535
    },
    {
      "epoch": 0.5894395094016324,
      "grad_norm": 4.0188326956159255,
      "learning_rate": 1.5223941235989071e-06,
      "loss": 0.5904,
      "step": 6536
    },
    {
      "epoch": 0.5895296929251026,
      "grad_norm": 1.6233683978033366,
      "learning_rate": 1.52182683570767e-06,
      "loss": 0.7235,
      "step": 6537
    },
    {
      "epoch": 0.5896198764485728,
      "grad_norm": 1.7092669967201322,
      "learning_rate": 1.5212595886168046e-06,
      "loss": 0.7786,
      "step": 6538
    },
    {
      "epoch": 0.5897100599720431,
      "grad_norm": 1.727623591880702,
      "learning_rate": 1.520692382374711e-06,
      "loss": 0.7666,
      "step": 6539
    },
    {
      "epoch": 0.5898002434955134,
      "grad_norm": 2.8542543012618853,
      "learning_rate": 1.5201252170297854e-06,
      "loss": 0.778,
      "step": 6540
    },
    {
      "epoch": 0.5898904270189836,
      "grad_norm": 1.8151537442836523,
      "learning_rate": 1.5195580926304232e-06,
      "loss": 0.6388,
      "step": 6541
    },
    {
      "epoch": 0.5899806105424539,
      "grad_norm": 2.8947636849468408,
      "learning_rate": 1.5189910092250131e-06,
      "loss": 0.7729,
      "step": 6542
    },
    {
      "epoch": 0.5900707940659241,
      "grad_norm": 2.634892488200883,
      "learning_rate": 1.5184239668619427e-06,
      "loss": 0.7618,
      "step": 6543
    },
    {
      "epoch": 0.5901609775893945,
      "grad_norm": 1.6976721027755861,
      "learning_rate": 1.5178569655895946e-06,
      "loss": 0.755,
      "step": 6544
    },
    {
      "epoch": 0.5902511611128647,
      "grad_norm": 1.5275794430316707,
      "learning_rate": 1.5172900054563487e-06,
      "loss": 0.7826,
      "step": 6545
    },
    {
      "epoch": 0.5903413446363349,
      "grad_norm": 1.59162649167206,
      "learning_rate": 1.5167230865105814e-06,
      "loss": 0.77,
      "step": 6546
    },
    {
      "epoch": 0.5904315281598052,
      "grad_norm": 1.866405243392296,
      "learning_rate": 1.5161562088006644e-06,
      "loss": 0.6706,
      "step": 6547
    },
    {
      "epoch": 0.5905217116832755,
      "grad_norm": 2.1241360198252357,
      "learning_rate": 1.5155893723749685e-06,
      "loss": 0.7834,
      "step": 6548
    },
    {
      "epoch": 0.5906118952067457,
      "grad_norm": 3.169597691419474,
      "learning_rate": 1.5150225772818582e-06,
      "loss": 0.7611,
      "step": 6549
    },
    {
      "epoch": 0.590702078730216,
      "grad_norm": 1.7217247622210388,
      "learning_rate": 1.5144558235696949e-06,
      "loss": 0.8024,
      "step": 6550
    },
    {
      "epoch": 0.5907922622536863,
      "grad_norm": 1.6117898545364924,
      "learning_rate": 1.5138891112868388e-06,
      "loss": 0.6697,
      "step": 6551
    },
    {
      "epoch": 0.5908824457771565,
      "grad_norm": 0.7005524728618798,
      "learning_rate": 1.5133224404816433e-06,
      "loss": 0.5796,
      "step": 6552
    },
    {
      "epoch": 0.5909726293006268,
      "grad_norm": 3.023214800220369,
      "learning_rate": 1.5127558112024617e-06,
      "loss": 0.6768,
      "step": 6553
    },
    {
      "epoch": 0.591062812824097,
      "grad_norm": 0.6812261944732447,
      "learning_rate": 1.5121892234976404e-06,
      "loss": 0.5929,
      "step": 6554
    },
    {
      "epoch": 0.5911529963475673,
      "grad_norm": 1.6525250418838877,
      "learning_rate": 1.5116226774155243e-06,
      "loss": 0.7356,
      "step": 6555
    },
    {
      "epoch": 0.5912431798710376,
      "grad_norm": 1.8455953899608724,
      "learning_rate": 1.5110561730044547e-06,
      "loss": 0.7263,
      "step": 6556
    },
    {
      "epoch": 0.5913333633945078,
      "grad_norm": 1.8143648222592812,
      "learning_rate": 1.510489710312768e-06,
      "loss": 0.7532,
      "step": 6557
    },
    {
      "epoch": 0.591423546917978,
      "grad_norm": 2.4333880048210226,
      "learning_rate": 1.5099232893887987e-06,
      "loss": 0.7443,
      "step": 6558
    },
    {
      "epoch": 0.5915137304414484,
      "grad_norm": 1.552384477454498,
      "learning_rate": 1.5093569102808758e-06,
      "loss": 0.7237,
      "step": 6559
    },
    {
      "epoch": 0.5916039139649186,
      "grad_norm": 1.8562554227704815,
      "learning_rate": 1.5087905730373275e-06,
      "loss": 0.7602,
      "step": 6560
    },
    {
      "epoch": 0.5916940974883889,
      "grad_norm": 0.616512653948017,
      "learning_rate": 1.508224277706476e-06,
      "loss": 0.5905,
      "step": 6561
    },
    {
      "epoch": 0.5917842810118591,
      "grad_norm": 4.507771880232933,
      "learning_rate": 1.5076580243366399e-06,
      "loss": 0.7056,
      "step": 6562
    },
    {
      "epoch": 0.5918744645353294,
      "grad_norm": 3.3673639450398722,
      "learning_rate": 1.507091812976137e-06,
      "loss": 0.6612,
      "step": 6563
    },
    {
      "epoch": 0.5919646480587997,
      "grad_norm": 1.6770454386661173,
      "learning_rate": 1.5065256436732773e-06,
      "loss": 0.7018,
      "step": 6564
    },
    {
      "epoch": 0.5920548315822699,
      "grad_norm": 1.946705086091616,
      "learning_rate": 1.5059595164763717e-06,
      "loss": 0.6817,
      "step": 6565
    },
    {
      "epoch": 0.5921450151057401,
      "grad_norm": 0.5747369516989698,
      "learning_rate": 1.5053934314337243e-06,
      "loss": 0.586,
      "step": 6566
    },
    {
      "epoch": 0.5922351986292105,
      "grad_norm": 1.8454934621740067,
      "learning_rate": 1.5048273885936356e-06,
      "loss": 0.7409,
      "step": 6567
    },
    {
      "epoch": 0.5923253821526807,
      "grad_norm": 1.443262904844436,
      "learning_rate": 1.5042613880044053e-06,
      "loss": 0.6698,
      "step": 6568
    },
    {
      "epoch": 0.592415565676151,
      "grad_norm": 1.6994508782044304,
      "learning_rate": 1.5036954297143265e-06,
      "loss": 0.7768,
      "step": 6569
    },
    {
      "epoch": 0.5925057491996212,
      "grad_norm": 1.9379247384431928,
      "learning_rate": 1.50312951377169e-06,
      "loss": 0.7064,
      "step": 6570
    },
    {
      "epoch": 0.5925959327230915,
      "grad_norm": 0.5868589911750479,
      "learning_rate": 1.502563640224784e-06,
      "loss": 0.5885,
      "step": 6571
    },
    {
      "epoch": 0.5926861162465618,
      "grad_norm": 1.8776161380409109,
      "learning_rate": 1.5019978091218903e-06,
      "loss": 0.7453,
      "step": 6572
    },
    {
      "epoch": 0.592776299770032,
      "grad_norm": 1.3690009161484233,
      "learning_rate": 1.50143202051129e-06,
      "loss": 0.8093,
      "step": 6573
    },
    {
      "epoch": 0.5928664832935022,
      "grad_norm": 1.8870406419211232,
      "learning_rate": 1.500866274441258e-06,
      "loss": 0.8082,
      "step": 6574
    },
    {
      "epoch": 0.5929566668169726,
      "grad_norm": 2.5195974888337433,
      "learning_rate": 1.5003005709600682e-06,
      "loss": 0.7319,
      "step": 6575
    },
    {
      "epoch": 0.5930468503404428,
      "grad_norm": 1.9751437678099983,
      "learning_rate": 1.4997349101159885e-06,
      "loss": 0.8253,
      "step": 6576
    },
    {
      "epoch": 0.593137033863913,
      "grad_norm": 2.166563276122661,
      "learning_rate": 1.4991692919572854e-06,
      "loss": 0.7085,
      "step": 6577
    },
    {
      "epoch": 0.5932272173873834,
      "grad_norm": 2.1233213189392495,
      "learning_rate": 1.4986037165322199e-06,
      "loss": 0.7637,
      "step": 6578
    },
    {
      "epoch": 0.5933174009108536,
      "grad_norm": 0.7640100301189009,
      "learning_rate": 1.498038183889049e-06,
      "loss": 0.6156,
      "step": 6579
    },
    {
      "epoch": 0.5934075844343238,
      "grad_norm": 2.4696331134702834,
      "learning_rate": 1.4974726940760292e-06,
      "loss": 0.6691,
      "step": 6580
    },
    {
      "epoch": 0.5934977679577941,
      "grad_norm": 2.1867775950362645,
      "learning_rate": 1.496907247141409e-06,
      "loss": 0.8242,
      "step": 6581
    },
    {
      "epoch": 0.5935879514812644,
      "grad_norm": 0.7464137564431993,
      "learning_rate": 1.4963418431334372e-06,
      "loss": 0.6576,
      "step": 6582
    },
    {
      "epoch": 0.5936781350047347,
      "grad_norm": 2.373610742711673,
      "learning_rate": 1.4957764821003566e-06,
      "loss": 0.7326,
      "step": 6583
    },
    {
      "epoch": 0.5937683185282049,
      "grad_norm": 1.626726480476083,
      "learning_rate": 1.4952111640904063e-06,
      "loss": 0.7558,
      "step": 6584
    },
    {
      "epoch": 0.5938585020516751,
      "grad_norm": 1.5156525284931521,
      "learning_rate": 1.494645889151823e-06,
      "loss": 0.7343,
      "step": 6585
    },
    {
      "epoch": 0.5939486855751455,
      "grad_norm": 0.6510335820013972,
      "learning_rate": 1.494080657332839e-06,
      "loss": 0.5624,
      "step": 6586
    },
    {
      "epoch": 0.5940388690986157,
      "grad_norm": 1.98425352670924,
      "learning_rate": 1.4935154686816832e-06,
      "loss": 0.7037,
      "step": 6587
    },
    {
      "epoch": 0.5941290526220859,
      "grad_norm": 1.959570419568546,
      "learning_rate": 1.4929503232465802e-06,
      "loss": 0.8232,
      "step": 6588
    },
    {
      "epoch": 0.5942192361455562,
      "grad_norm": 1.658442000402221,
      "learning_rate": 1.492385221075751e-06,
      "loss": 0.6644,
      "step": 6589
    },
    {
      "epoch": 0.5943094196690265,
      "grad_norm": 3.2342588247176,
      "learning_rate": 1.4918201622174142e-06,
      "loss": 0.7814,
      "step": 6590
    },
    {
      "epoch": 0.5943996031924967,
      "grad_norm": 1.7810797359003234,
      "learning_rate": 1.4912551467197827e-06,
      "loss": 0.7687,
      "step": 6591
    },
    {
      "epoch": 0.594489786715967,
      "grad_norm": 2.3088097898861273,
      "learning_rate": 1.4906901746310678e-06,
      "loss": 0.8384,
      "step": 6592
    },
    {
      "epoch": 0.5945799702394372,
      "grad_norm": 1.6361948417534704,
      "learning_rate": 1.4901252459994757e-06,
      "loss": 0.7138,
      "step": 6593
    },
    {
      "epoch": 0.5946701537629075,
      "grad_norm": 1.6594706477241346,
      "learning_rate": 1.489560360873208e-06,
      "loss": 0.7335,
      "step": 6594
    },
    {
      "epoch": 0.5947603372863778,
      "grad_norm": 1.7163298473968538,
      "learning_rate": 1.4889955193004659e-06,
      "loss": 0.728,
      "step": 6595
    },
    {
      "epoch": 0.594850520809848,
      "grad_norm": 0.6444066707549958,
      "learning_rate": 1.4884307213294428e-06,
      "loss": 0.5746,
      "step": 6596
    },
    {
      "epoch": 0.5949407043333182,
      "grad_norm": 2.3376259713729555,
      "learning_rate": 1.4878659670083321e-06,
      "loss": 0.737,
      "step": 6597
    },
    {
      "epoch": 0.5950308878567886,
      "grad_norm": 1.643655212918408,
      "learning_rate": 1.4873012563853208e-06,
      "loss": 0.6714,
      "step": 6598
    },
    {
      "epoch": 0.5951210713802588,
      "grad_norm": 2.214361444400013,
      "learning_rate": 1.4867365895085935e-06,
      "loss": 0.753,
      "step": 6599
    },
    {
      "epoch": 0.5952112549037291,
      "grad_norm": 2.3889858635236254,
      "learning_rate": 1.4861719664263301e-06,
      "loss": 0.7268,
      "step": 6600
    },
    {
      "epoch": 0.5953014384271994,
      "grad_norm": 2.6065721820413206,
      "learning_rate": 1.485607387186708e-06,
      "loss": 0.852,
      "step": 6601
    },
    {
      "epoch": 0.5953916219506696,
      "grad_norm": 1.8768440715059278,
      "learning_rate": 1.4850428518379001e-06,
      "loss": 0.7435,
      "step": 6602
    },
    {
      "epoch": 0.5954818054741399,
      "grad_norm": 1.3705413227135181,
      "learning_rate": 1.4844783604280746e-06,
      "loss": 0.7448,
      "step": 6603
    },
    {
      "epoch": 0.5955719889976101,
      "grad_norm": 3.045776325602086,
      "learning_rate": 1.483913913005399e-06,
      "loss": 0.6923,
      "step": 6604
    },
    {
      "epoch": 0.5956621725210804,
      "grad_norm": 2.621691480722469,
      "learning_rate": 1.483349509618034e-06,
      "loss": 0.81,
      "step": 6605
    },
    {
      "epoch": 0.5957523560445507,
      "grad_norm": 1.6530049761808603,
      "learning_rate": 1.4827851503141367e-06,
      "loss": 0.6819,
      "step": 6606
    },
    {
      "epoch": 0.5958425395680209,
      "grad_norm": 1.5720601391827462,
      "learning_rate": 1.482220835141863e-06,
      "loss": 0.6756,
      "step": 6607
    },
    {
      "epoch": 0.5959327230914911,
      "grad_norm": 1.6877540114632965,
      "learning_rate": 1.481656564149362e-06,
      "loss": 0.828,
      "step": 6608
    },
    {
      "epoch": 0.5960229066149615,
      "grad_norm": 2.0172231090425456,
      "learning_rate": 1.4810923373847818e-06,
      "loss": 0.7464,
      "step": 6609
    },
    {
      "epoch": 0.5961130901384317,
      "grad_norm": 1.6856362682697148,
      "learning_rate": 1.4805281548962647e-06,
      "loss": 0.7554,
      "step": 6610
    },
    {
      "epoch": 0.596203273661902,
      "grad_norm": 2.1873700784520715,
      "learning_rate": 1.4799640167319488e-06,
      "loss": 0.6938,
      "step": 6611
    },
    {
      "epoch": 0.5962934571853722,
      "grad_norm": 1.5697807056809632,
      "learning_rate": 1.4793999229399714e-06,
      "loss": 0.7371,
      "step": 6612
    },
    {
      "epoch": 0.5963836407088425,
      "grad_norm": 1.6073468956777948,
      "learning_rate": 1.4788358735684626e-06,
      "loss": 0.8142,
      "step": 6613
    },
    {
      "epoch": 0.5964738242323128,
      "grad_norm": 1.5243185772215304,
      "learning_rate": 1.4782718686655514e-06,
      "loss": 0.8168,
      "step": 6614
    },
    {
      "epoch": 0.596564007755783,
      "grad_norm": 1.8697798222952093,
      "learning_rate": 1.4777079082793605e-06,
      "loss": 0.7997,
      "step": 6615
    },
    {
      "epoch": 0.5966541912792532,
      "grad_norm": 1.9464314012019435,
      "learning_rate": 1.4771439924580108e-06,
      "loss": 0.7203,
      "step": 6616
    },
    {
      "epoch": 0.5967443748027236,
      "grad_norm": 1.31386445284633,
      "learning_rate": 1.4765801212496189e-06,
      "loss": 0.7903,
      "step": 6617
    },
    {
      "epoch": 0.5968345583261938,
      "grad_norm": 1.8825202176856208,
      "learning_rate": 1.4760162947022968e-06,
      "loss": 0.6694,
      "step": 6618
    },
    {
      "epoch": 0.596924741849664,
      "grad_norm": 1.6917309878794449,
      "learning_rate": 1.475452512864154e-06,
      "loss": 0.7373,
      "step": 6619
    },
    {
      "epoch": 0.5970149253731343,
      "grad_norm": 1.4148814322194563,
      "learning_rate": 1.4748887757832945e-06,
      "loss": 0.7561,
      "step": 6620
    },
    {
      "epoch": 0.5971051088966046,
      "grad_norm": 1.58173991216943,
      "learning_rate": 1.4743250835078209e-06,
      "loss": 0.7834,
      "step": 6621
    },
    {
      "epoch": 0.5971952924200749,
      "grad_norm": 1.791308662960665,
      "learning_rate": 1.4737614360858297e-06,
      "loss": 0.7102,
      "step": 6622
    },
    {
      "epoch": 0.5972854759435451,
      "grad_norm": 2.329088664469004,
      "learning_rate": 1.4731978335654133e-06,
      "loss": 0.703,
      "step": 6623
    },
    {
      "epoch": 0.5973756594670154,
      "grad_norm": 2.167885058737933,
      "learning_rate": 1.4726342759946638e-06,
      "loss": 0.6882,
      "step": 6624
    },
    {
      "epoch": 0.5974658429904857,
      "grad_norm": 1.6392605148287702,
      "learning_rate": 1.4720707634216653e-06,
      "loss": 0.7496,
      "step": 6625
    },
    {
      "epoch": 0.5975560265139559,
      "grad_norm": 1.969666827633672,
      "learning_rate": 1.4715072958945e-06,
      "loss": 0.7569,
      "step": 6626
    },
    {
      "epoch": 0.5976462100374261,
      "grad_norm": 1.9832399982107385,
      "learning_rate": 1.470943873461247e-06,
      "loss": 0.7484,
      "step": 6627
    },
    {
      "epoch": 0.5977363935608965,
      "grad_norm": 2.8994834637197497,
      "learning_rate": 1.470380496169979e-06,
      "loss": 0.7116,
      "step": 6628
    },
    {
      "epoch": 0.5978265770843667,
      "grad_norm": 1.9224746512935698,
      "learning_rate": 1.4698171640687682e-06,
      "loss": 0.7395,
      "step": 6629
    },
    {
      "epoch": 0.5979167606078369,
      "grad_norm": 1.899038086383192,
      "learning_rate": 1.4692538772056792e-06,
      "loss": 0.6841,
      "step": 6630
    },
    {
      "epoch": 0.5980069441313072,
      "grad_norm": 1.4024094165101781,
      "learning_rate": 1.4686906356287772e-06,
      "loss": 0.6625,
      "step": 6631
    },
    {
      "epoch": 0.5980971276547775,
      "grad_norm": 1.5615824484872587,
      "learning_rate": 1.4681274393861194e-06,
      "loss": 0.78,
      "step": 6632
    },
    {
      "epoch": 0.5981873111782477,
      "grad_norm": 1.8855062502097621,
      "learning_rate": 1.4675642885257603e-06,
      "loss": 0.8486,
      "step": 6633
    },
    {
      "epoch": 0.598277494701718,
      "grad_norm": 2.7662398522316285,
      "learning_rate": 1.4670011830957529e-06,
      "loss": 0.8355,
      "step": 6634
    },
    {
      "epoch": 0.5983676782251882,
      "grad_norm": 2.2837917851476117,
      "learning_rate": 1.4664381231441427e-06,
      "loss": 0.7161,
      "step": 6635
    },
    {
      "epoch": 0.5984578617486586,
      "grad_norm": 1.7189893382497983,
      "learning_rate": 1.4658751087189746e-06,
      "loss": 0.7611,
      "step": 6636
    },
    {
      "epoch": 0.5985480452721288,
      "grad_norm": 1.6090390493610696,
      "learning_rate": 1.4653121398682874e-06,
      "loss": 0.7989,
      "step": 6637
    },
    {
      "epoch": 0.598638228795599,
      "grad_norm": 12.142294593410426,
      "learning_rate": 1.4647492166401159e-06,
      "loss": 0.6867,
      "step": 6638
    },
    {
      "epoch": 0.5987284123190693,
      "grad_norm": 1.7118382282075058,
      "learning_rate": 1.4641863390824934e-06,
      "loss": 0.7374,
      "step": 6639
    },
    {
      "epoch": 0.5988185958425396,
      "grad_norm": 2.0124516309529152,
      "learning_rate": 1.4636235072434465e-06,
      "loss": 0.7398,
      "step": 6640
    },
    {
      "epoch": 0.5989087793660098,
      "grad_norm": 1.6757613316488162,
      "learning_rate": 1.4630607211709994e-06,
      "loss": 0.7752,
      "step": 6641
    },
    {
      "epoch": 0.5989989628894801,
      "grad_norm": 2.831307491030363,
      "learning_rate": 1.4624979809131723e-06,
      "loss": 0.7554,
      "step": 6642
    },
    {
      "epoch": 0.5990891464129503,
      "grad_norm": 1.473650447469448,
      "learning_rate": 1.4619352865179814e-06,
      "loss": 0.658,
      "step": 6643
    },
    {
      "epoch": 0.5991793299364206,
      "grad_norm": 2.232884156430059,
      "learning_rate": 1.4613726380334391e-06,
      "loss": 0.7195,
      "step": 6644
    },
    {
      "epoch": 0.5992695134598909,
      "grad_norm": 2.0980651570576345,
      "learning_rate": 1.4608100355075522e-06,
      "loss": 0.6032,
      "step": 6645
    },
    {
      "epoch": 0.5993596969833611,
      "grad_norm": 0.6221361829056511,
      "learning_rate": 1.460247478988327e-06,
      "loss": 0.5561,
      "step": 6646
    },
    {
      "epoch": 0.5994498805068315,
      "grad_norm": 1.5361363822570568,
      "learning_rate": 1.4596849685237623e-06,
      "loss": 0.7654,
      "step": 6647
    },
    {
      "epoch": 0.5995400640303017,
      "grad_norm": 2.0554158475441047,
      "learning_rate": 1.459122504161856e-06,
      "loss": 0.6572,
      "step": 6648
    },
    {
      "epoch": 0.5996302475537719,
      "grad_norm": 1.9181466262905844,
      "learning_rate": 1.4585600859506001e-06,
      "loss": 0.72,
      "step": 6649
    },
    {
      "epoch": 0.5997204310772422,
      "grad_norm": 1.6401663729549787,
      "learning_rate": 1.4579977139379826e-06,
      "loss": 0.8087,
      "step": 6650
    },
    {
      "epoch": 0.5998106146007125,
      "grad_norm": 1.9621922707290842,
      "learning_rate": 1.4574353881719895e-06,
      "loss": 0.8066,
      "step": 6651
    },
    {
      "epoch": 0.5999007981241827,
      "grad_norm": 1.7224381150864998,
      "learning_rate": 1.4568731087005998e-06,
      "loss": 0.753,
      "step": 6652
    },
    {
      "epoch": 0.599990981647653,
      "grad_norm": 1.2657984391041157,
      "learning_rate": 1.4563108755717916e-06,
      "loss": 0.7718,
      "step": 6653
    },
    {
      "epoch": 0.6000811651711232,
      "grad_norm": 1.892683059727988,
      "learning_rate": 1.455748688833538e-06,
      "loss": 0.7771,
      "step": 6654
    },
    {
      "epoch": 0.6001713486945935,
      "grad_norm": 1.741291517103679,
      "learning_rate": 1.4551865485338065e-06,
      "loss": 0.6619,
      "step": 6655
    },
    {
      "epoch": 0.6002615322180638,
      "grad_norm": 1.5403822454201475,
      "learning_rate": 1.4546244547205629e-06,
      "loss": 0.8179,
      "step": 6656
    },
    {
      "epoch": 0.600351715741534,
      "grad_norm": 1.8130623175362444,
      "learning_rate": 1.4540624074417678e-06,
      "loss": 0.7751,
      "step": 6657
    },
    {
      "epoch": 0.6004418992650042,
      "grad_norm": 1.4328431703369529,
      "learning_rate": 1.453500406745379e-06,
      "loss": 0.8083,
      "step": 6658
    },
    {
      "epoch": 0.6005320827884746,
      "grad_norm": 3.2083339113177156,
      "learning_rate": 1.4529384526793486e-06,
      "loss": 0.8275,
      "step": 6659
    },
    {
      "epoch": 0.6006222663119448,
      "grad_norm": 1.636036733042389,
      "learning_rate": 1.4523765452916252e-06,
      "loss": 0.8052,
      "step": 6660
    },
    {
      "epoch": 0.600712449835415,
      "grad_norm": 1.9960933916722403,
      "learning_rate": 1.4518146846301554e-06,
      "loss": 0.7801,
      "step": 6661
    },
    {
      "epoch": 0.6008026333588853,
      "grad_norm": 2.373700338429722,
      "learning_rate": 1.4512528707428787e-06,
      "loss": 0.7255,
      "step": 6662
    },
    {
      "epoch": 0.6008928168823556,
      "grad_norm": 2.5766911073769965,
      "learning_rate": 1.4506911036777335e-06,
      "loss": 0.7309,
      "step": 6663
    },
    {
      "epoch": 0.6009830004058259,
      "grad_norm": 0.67243331067996,
      "learning_rate": 1.450129383482651e-06,
      "loss": 0.5488,
      "step": 6664
    },
    {
      "epoch": 0.6010731839292961,
      "grad_norm": 1.685577498386537,
      "learning_rate": 1.4495677102055629e-06,
      "loss": 0.7052,
      "step": 6665
    },
    {
      "epoch": 0.6011633674527663,
      "grad_norm": 1.8949680100209305,
      "learning_rate": 1.4490060838943924e-06,
      "loss": 0.7857,
      "step": 6666
    },
    {
      "epoch": 0.6012535509762367,
      "grad_norm": 2.5295152724544523,
      "learning_rate": 1.4484445045970609e-06,
      "loss": 0.7531,
      "step": 6667
    },
    {
      "epoch": 0.6013437344997069,
      "grad_norm": 2.740630400596874,
      "learning_rate": 1.447882972361485e-06,
      "loss": 0.7451,
      "step": 6668
    },
    {
      "epoch": 0.6014339180231771,
      "grad_norm": 1.5601831425243522,
      "learning_rate": 1.4473214872355785e-06,
      "loss": 0.7376,
      "step": 6669
    },
    {
      "epoch": 0.6015241015466475,
      "grad_norm": 2.0933876293373372,
      "learning_rate": 1.4467600492672508e-06,
      "loss": 0.7828,
      "step": 6670
    },
    {
      "epoch": 0.6016142850701177,
      "grad_norm": 1.6742675631728197,
      "learning_rate": 1.4461986585044054e-06,
      "loss": 0.7268,
      "step": 6671
    },
    {
      "epoch": 0.601704468593588,
      "grad_norm": 1.807269936320095,
      "learning_rate": 1.4456373149949446e-06,
      "loss": 0.7255,
      "step": 6672
    },
    {
      "epoch": 0.6017946521170582,
      "grad_norm": 4.307109393776301,
      "learning_rate": 1.4450760187867648e-06,
      "loss": 0.7582,
      "step": 6673
    },
    {
      "epoch": 0.6018848356405285,
      "grad_norm": 1.851546661695511,
      "learning_rate": 1.4445147699277581e-06,
      "loss": 0.7446,
      "step": 6674
    },
    {
      "epoch": 0.6019750191639988,
      "grad_norm": 1.6879875750287792,
      "learning_rate": 1.4439535684658154e-06,
      "loss": 0.7963,
      "step": 6675
    },
    {
      "epoch": 0.602065202687469,
      "grad_norm": 0.767904935185912,
      "learning_rate": 1.44339241444882e-06,
      "loss": 0.6179,
      "step": 6676
    },
    {
      "epoch": 0.6021553862109392,
      "grad_norm": 2.11274867653842,
      "learning_rate": 1.4428313079246518e-06,
      "loss": 0.7008,
      "step": 6677
    },
    {
      "epoch": 0.6022455697344096,
      "grad_norm": 0.6639165381291842,
      "learning_rate": 1.4422702489411896e-06,
      "loss": 0.5958,
      "step": 6678
    },
    {
      "epoch": 0.6023357532578798,
      "grad_norm": 1.9964479255371812,
      "learning_rate": 1.4417092375463043e-06,
      "loss": 0.7019,
      "step": 6679
    },
    {
      "epoch": 0.60242593678135,
      "grad_norm": 1.6759969221282593,
      "learning_rate": 1.441148273787866e-06,
      "loss": 0.7875,
      "step": 6680
    },
    {
      "epoch": 0.6025161203048203,
      "grad_norm": 0.8160537615858412,
      "learning_rate": 1.4405873577137383e-06,
      "loss": 0.6022,
      "step": 6681
    },
    {
      "epoch": 0.6026063038282906,
      "grad_norm": 3.152256015469008,
      "learning_rate": 1.4400264893717816e-06,
      "loss": 0.7586,
      "step": 6682
    },
    {
      "epoch": 0.6026964873517608,
      "grad_norm": 1.6160499884456017,
      "learning_rate": 1.4394656688098526e-06,
      "loss": 0.7972,
      "step": 6683
    },
    {
      "epoch": 0.6027866708752311,
      "grad_norm": 1.4890471963005971,
      "learning_rate": 1.4389048960758032e-06,
      "loss": 0.7717,
      "step": 6684
    },
    {
      "epoch": 0.6028768543987013,
      "grad_norm": 2.787452211526319,
      "learning_rate": 1.4383441712174826e-06,
      "loss": 0.7859,
      "step": 6685
    },
    {
      "epoch": 0.6029670379221717,
      "grad_norm": 3.1507971466442832,
      "learning_rate": 1.4377834942827333e-06,
      "loss": 0.8136,
      "step": 6686
    },
    {
      "epoch": 0.6030572214456419,
      "grad_norm": 2.3891250791767717,
      "learning_rate": 1.437222865319397e-06,
      "loss": 0.6905,
      "step": 6687
    },
    {
      "epoch": 0.6031474049691121,
      "grad_norm": 1.5294730327382484,
      "learning_rate": 1.4366622843753092e-06,
      "loss": 0.6801,
      "step": 6688
    },
    {
      "epoch": 0.6032375884925824,
      "grad_norm": 1.3100022208944815,
      "learning_rate": 1.4361017514983006e-06,
      "loss": 0.7684,
      "step": 6689
    },
    {
      "epoch": 0.6033277720160527,
      "grad_norm": 1.6559440042224371,
      "learning_rate": 1.4355412667362006e-06,
      "loss": 0.8053,
      "step": 6690
    },
    {
      "epoch": 0.6034179555395229,
      "grad_norm": 2.1358668775401237,
      "learning_rate": 1.4349808301368311e-06,
      "loss": 0.7105,
      "step": 6691
    },
    {
      "epoch": 0.6035081390629932,
      "grad_norm": 3.1036239422668794,
      "learning_rate": 1.4344204417480139e-06,
      "loss": 0.7927,
      "step": 6692
    },
    {
      "epoch": 0.6035983225864634,
      "grad_norm": 1.811526925219857,
      "learning_rate": 1.4338601016175628e-06,
      "loss": 0.7025,
      "step": 6693
    },
    {
      "epoch": 0.6036885061099337,
      "grad_norm": 1.57181276232325,
      "learning_rate": 1.433299809793289e-06,
      "loss": 0.7486,
      "step": 6694
    },
    {
      "epoch": 0.603778689633404,
      "grad_norm": 3.435939087635348,
      "learning_rate": 1.432739566323001e-06,
      "loss": 0.6337,
      "step": 6695
    },
    {
      "epoch": 0.6038688731568742,
      "grad_norm": 2.1121392116904953,
      "learning_rate": 1.4321793712545004e-06,
      "loss": 0.7146,
      "step": 6696
    },
    {
      "epoch": 0.6039590566803446,
      "grad_norm": 13.754515859999172,
      "learning_rate": 1.4316192246355873e-06,
      "loss": 0.7974,
      "step": 6697
    },
    {
      "epoch": 0.6040492402038148,
      "grad_norm": 0.6113445779457174,
      "learning_rate": 1.4310591265140555e-06,
      "loss": 0.5798,
      "step": 6698
    },
    {
      "epoch": 0.604139423727285,
      "grad_norm": 1.4504624451460482,
      "learning_rate": 1.4304990769376963e-06,
      "loss": 0.7562,
      "step": 6699
    },
    {
      "epoch": 0.6042296072507553,
      "grad_norm": 1.4637492655972952,
      "learning_rate": 1.4299390759542962e-06,
      "loss": 0.7319,
      "step": 6700
    },
    {
      "epoch": 0.6043197907742256,
      "grad_norm": 1.624641672387997,
      "learning_rate": 1.4293791236116368e-06,
      "loss": 0.735,
      "step": 6701
    },
    {
      "epoch": 0.6044099742976958,
      "grad_norm": 1.9460243240279018,
      "learning_rate": 1.4288192199574978e-06,
      "loss": 0.7154,
      "step": 6702
    },
    {
      "epoch": 0.6045001578211661,
      "grad_norm": 0.666889657536086,
      "learning_rate": 1.4282593650396524e-06,
      "loss": 0.5523,
      "step": 6703
    },
    {
      "epoch": 0.6045903413446363,
      "grad_norm": 2.16165214190558,
      "learning_rate": 1.4276995589058695e-06,
      "loss": 0.8511,
      "step": 6704
    },
    {
      "epoch": 0.6046805248681066,
      "grad_norm": 0.5936577875910899,
      "learning_rate": 1.4271398016039168e-06,
      "loss": 0.5484,
      "step": 6705
    },
    {
      "epoch": 0.6047707083915769,
      "grad_norm": 3.269650255450688,
      "learning_rate": 1.4265800931815542e-06,
      "loss": 0.7897,
      "step": 6706
    },
    {
      "epoch": 0.6048608919150471,
      "grad_norm": 1.752491267641,
      "learning_rate": 1.4260204336865406e-06,
      "loss": 0.7598,
      "step": 6707
    },
    {
      "epoch": 0.6049510754385173,
      "grad_norm": 2.488200858597599,
      "learning_rate": 1.4254608231666281e-06,
      "loss": 0.7495,
      "step": 6708
    },
    {
      "epoch": 0.6050412589619877,
      "grad_norm": 2.2612593057334474,
      "learning_rate": 1.4249012616695661e-06,
      "loss": 0.759,
      "step": 6709
    },
    {
      "epoch": 0.6051314424854579,
      "grad_norm": 1.6969088584374188,
      "learning_rate": 1.4243417492431e-06,
      "loss": 0.7561,
      "step": 6710
    },
    {
      "epoch": 0.6052216260089281,
      "grad_norm": 2.099549758419608,
      "learning_rate": 1.4237822859349696e-06,
      "loss": 0.7337,
      "step": 6711
    },
    {
      "epoch": 0.6053118095323984,
      "grad_norm": 1.445287673324567,
      "learning_rate": 1.423222871792912e-06,
      "loss": 0.7216,
      "step": 6712
    },
    {
      "epoch": 0.6054019930558687,
      "grad_norm": 1.7287272005929173,
      "learning_rate": 1.4226635068646586e-06,
      "loss": 0.8031,
      "step": 6713
    },
    {
      "epoch": 0.605492176579339,
      "grad_norm": 0.6253552380950097,
      "learning_rate": 1.4221041911979393e-06,
      "loss": 0.5724,
      "step": 6714
    },
    {
      "epoch": 0.6055823601028092,
      "grad_norm": 1.7137058106004206,
      "learning_rate": 1.4215449248404765e-06,
      "loss": 0.7697,
      "step": 6715
    },
    {
      "epoch": 0.6056725436262794,
      "grad_norm": 17.85842318931304,
      "learning_rate": 1.4209857078399896e-06,
      "loss": 0.7715,
      "step": 6716
    },
    {
      "epoch": 0.6057627271497498,
      "grad_norm": 2.091243211150337,
      "learning_rate": 1.4204265402441955e-06,
      "loss": 0.7171,
      "step": 6717
    },
    {
      "epoch": 0.60585291067322,
      "grad_norm": 1.6802866732013866,
      "learning_rate": 1.419867422100804e-06,
      "loss": 0.6233,
      "step": 6718
    },
    {
      "epoch": 0.6059430941966902,
      "grad_norm": 1.7476569566652782,
      "learning_rate": 1.4193083534575236e-06,
      "loss": 0.7383,
      "step": 6719
    },
    {
      "epoch": 0.6060332777201606,
      "grad_norm": 1.9008515005931235,
      "learning_rate": 1.4187493343620567e-06,
      "loss": 0.7746,
      "step": 6720
    },
    {
      "epoch": 0.6061234612436308,
      "grad_norm": 1.6437660303530806,
      "learning_rate": 1.4181903648621006e-06,
      "loss": 0.7814,
      "step": 6721
    },
    {
      "epoch": 0.606213644767101,
      "grad_norm": 1.5452228547140348,
      "learning_rate": 1.4176314450053512e-06,
      "loss": 0.7778,
      "step": 6722
    },
    {
      "epoch": 0.6063038282905713,
      "grad_norm": 2.0085964231054145,
      "learning_rate": 1.4170725748394977e-06,
      "loss": 0.7365,
      "step": 6723
    },
    {
      "epoch": 0.6063940118140416,
      "grad_norm": 1.4391919006420857,
      "learning_rate": 1.4165137544122266e-06,
      "loss": 0.6624,
      "step": 6724
    },
    {
      "epoch": 0.6064841953375119,
      "grad_norm": 1.7679649309158376,
      "learning_rate": 1.4159549837712194e-06,
      "loss": 0.8148,
      "step": 6725
    },
    {
      "epoch": 0.6065743788609821,
      "grad_norm": 1.514471972261899,
      "learning_rate": 1.415396262964153e-06,
      "loss": 0.6663,
      "step": 6726
    },
    {
      "epoch": 0.6066645623844523,
      "grad_norm": 1.8087177275314001,
      "learning_rate": 1.4148375920387016e-06,
      "loss": 0.826,
      "step": 6727
    },
    {
      "epoch": 0.6067547459079227,
      "grad_norm": 0.6496008582827107,
      "learning_rate": 1.4142789710425325e-06,
      "loss": 0.576,
      "step": 6728
    },
    {
      "epoch": 0.6068449294313929,
      "grad_norm": 20.860361048791496,
      "learning_rate": 1.4137204000233118e-06,
      "loss": 0.6911,
      "step": 6729
    },
    {
      "epoch": 0.6069351129548631,
      "grad_norm": 1.773772691147383,
      "learning_rate": 1.4131618790286987e-06,
      "loss": 0.7429,
      "step": 6730
    },
    {
      "epoch": 0.6070252964783334,
      "grad_norm": 1.866458638838371,
      "learning_rate": 1.4126034081063506e-06,
      "loss": 0.7354,
      "step": 6731
    },
    {
      "epoch": 0.6071154800018037,
      "grad_norm": 1.7291304879444436,
      "learning_rate": 1.4120449873039186e-06,
      "loss": 0.7333,
      "step": 6732
    },
    {
      "epoch": 0.6072056635252739,
      "grad_norm": 2.716195846979814,
      "learning_rate": 1.4114866166690494e-06,
      "loss": 0.7861,
      "step": 6733
    },
    {
      "epoch": 0.6072958470487442,
      "grad_norm": 1.583455214912331,
      "learning_rate": 1.4109282962493877e-06,
      "loss": 0.702,
      "step": 6734
    },
    {
      "epoch": 0.6073860305722144,
      "grad_norm": 1.7712116136611238,
      "learning_rate": 1.4103700260925716e-06,
      "loss": 0.7387,
      "step": 6735
    },
    {
      "epoch": 0.6074762140956848,
      "grad_norm": 1.9996100880106844,
      "learning_rate": 1.4098118062462364e-06,
      "loss": 0.7863,
      "step": 6736
    },
    {
      "epoch": 0.607566397619155,
      "grad_norm": 1.48281923378962,
      "learning_rate": 1.4092536367580123e-06,
      "loss": 0.7467,
      "step": 6737
    },
    {
      "epoch": 0.6076565811426252,
      "grad_norm": 1.7590592095850348,
      "learning_rate": 1.4086955176755248e-06,
      "loss": 0.6683,
      "step": 6738
    },
    {
      "epoch": 0.6077467646660955,
      "grad_norm": 1.8414770437234413,
      "learning_rate": 1.4081374490463964e-06,
      "loss": 0.6902,
      "step": 6739
    },
    {
      "epoch": 0.6078369481895658,
      "grad_norm": 1.706679947448754,
      "learning_rate": 1.4075794309182443e-06,
      "loss": 0.7611,
      "step": 6740
    },
    {
      "epoch": 0.607927131713036,
      "grad_norm": 2.0549375432679646,
      "learning_rate": 1.407021463338682e-06,
      "loss": 0.7676,
      "step": 6741
    },
    {
      "epoch": 0.6080173152365063,
      "grad_norm": 0.5906742135994495,
      "learning_rate": 1.4064635463553177e-06,
      "loss": 0.5213,
      "step": 6742
    },
    {
      "epoch": 0.6081074987599766,
      "grad_norm": 1.7127265808125791,
      "learning_rate": 1.4059056800157563e-06,
      "loss": 0.78,
      "step": 6743
    },
    {
      "epoch": 0.6081976822834468,
      "grad_norm": 1.5470990876000887,
      "learning_rate": 1.4053478643675982e-06,
      "loss": 0.6812,
      "step": 6744
    },
    {
      "epoch": 0.6082878658069171,
      "grad_norm": 2.1714420024198313,
      "learning_rate": 1.4047900994584389e-06,
      "loss": 0.6494,
      "step": 6745
    },
    {
      "epoch": 0.6083780493303873,
      "grad_norm": 2.342001198961324,
      "learning_rate": 1.404232385335871e-06,
      "loss": 0.6634,
      "step": 6746
    },
    {
      "epoch": 0.6084682328538576,
      "grad_norm": 7.092474869047222,
      "learning_rate": 1.4036747220474806e-06,
      "loss": 0.7487,
      "step": 6747
    },
    {
      "epoch": 0.6085584163773279,
      "grad_norm": 2.1748816447009323,
      "learning_rate": 1.4031171096408506e-06,
      "loss": 0.7049,
      "step": 6748
    },
    {
      "epoch": 0.6086485999007981,
      "grad_norm": 2.3424541705848343,
      "learning_rate": 1.4025595481635607e-06,
      "loss": 0.7753,
      "step": 6749
    },
    {
      "epoch": 0.6087387834242683,
      "grad_norm": 1.8541157176129317,
      "learning_rate": 1.4020020376631836e-06,
      "loss": 0.7711,
      "step": 6750
    },
    {
      "epoch": 0.6088289669477387,
      "grad_norm": 2.3369400781760743,
      "learning_rate": 1.4014445781872908e-06,
      "loss": 0.7626,
      "step": 6751
    },
    {
      "epoch": 0.6089191504712089,
      "grad_norm": 1.7547961316064544,
      "learning_rate": 1.4008871697834465e-06,
      "loss": 0.8127,
      "step": 6752
    },
    {
      "epoch": 0.6090093339946792,
      "grad_norm": 2.488289106946485,
      "learning_rate": 1.400329812499213e-06,
      "loss": 0.7832,
      "step": 6753
    },
    {
      "epoch": 0.6090995175181494,
      "grad_norm": 1.7105462310573762,
      "learning_rate": 1.3997725063821458e-06,
      "loss": 0.7637,
      "step": 6754
    },
    {
      "epoch": 0.6091897010416197,
      "grad_norm": 1.6924942428964267,
      "learning_rate": 1.3992152514797978e-06,
      "loss": 0.7131,
      "step": 6755
    },
    {
      "epoch": 0.60927988456509,
      "grad_norm": 0.6495673665532686,
      "learning_rate": 1.398658047839718e-06,
      "loss": 0.596,
      "step": 6756
    },
    {
      "epoch": 0.6093700680885602,
      "grad_norm": 0.6032795367133832,
      "learning_rate": 1.3981008955094481e-06,
      "loss": 0.5656,
      "step": 6757
    },
    {
      "epoch": 0.6094602516120304,
      "grad_norm": 8.502887908422828,
      "learning_rate": 1.39754379453653e-06,
      "loss": 0.7223,
      "step": 6758
    },
    {
      "epoch": 0.6095504351355008,
      "grad_norm": 3.419823854829234,
      "learning_rate": 1.3969867449684972e-06,
      "loss": 0.7582,
      "step": 6759
    },
    {
      "epoch": 0.609640618658971,
      "grad_norm": 2.16689822597946,
      "learning_rate": 1.396429746852879e-06,
      "loss": 0.7402,
      "step": 6760
    },
    {
      "epoch": 0.6097308021824412,
      "grad_norm": 1.6777769356459524,
      "learning_rate": 1.395872800237204e-06,
      "loss": 0.7259,
      "step": 6761
    },
    {
      "epoch": 0.6098209857059115,
      "grad_norm": 0.7000383656069757,
      "learning_rate": 1.3953159051689918e-06,
      "loss": 0.5975,
      "step": 6762
    },
    {
      "epoch": 0.6099111692293818,
      "grad_norm": 3.1044636185379404,
      "learning_rate": 1.3947590616957618e-06,
      "loss": 0.7924,
      "step": 6763
    },
    {
      "epoch": 0.610001352752852,
      "grad_norm": 5.162868067295333,
      "learning_rate": 1.3942022698650258e-06,
      "loss": 0.783,
      "step": 6764
    },
    {
      "epoch": 0.6100915362763223,
      "grad_norm": 4.37963817512862,
      "learning_rate": 1.3936455297242917e-06,
      "loss": 0.7823,
      "step": 6765
    },
    {
      "epoch": 0.6101817197997926,
      "grad_norm": 9.52873421908026,
      "learning_rate": 1.3930888413210652e-06,
      "loss": 0.7704,
      "step": 6766
    },
    {
      "epoch": 0.6102719033232629,
      "grad_norm": 2.7039773291873277,
      "learning_rate": 1.392532204702845e-06,
      "loss": 0.7114,
      "step": 6767
    },
    {
      "epoch": 0.6103620868467331,
      "grad_norm": 2.4281496535219325,
      "learning_rate": 1.3919756199171266e-06,
      "loss": 0.681,
      "step": 6768
    },
    {
      "epoch": 0.6104522703702033,
      "grad_norm": 1.9948377745024861,
      "learning_rate": 1.3914190870114009e-06,
      "loss": 0.8246,
      "step": 6769
    },
    {
      "epoch": 0.6105424538936737,
      "grad_norm": 1.6998099164719067,
      "learning_rate": 1.3908626060331541e-06,
      "loss": 0.7999,
      "step": 6770
    },
    {
      "epoch": 0.6106326374171439,
      "grad_norm": 2.6021299001183325,
      "learning_rate": 1.3903061770298693e-06,
      "loss": 0.7855,
      "step": 6771
    },
    {
      "epoch": 0.6107228209406141,
      "grad_norm": 1.7021539001902446,
      "learning_rate": 1.3897498000490223e-06,
      "loss": 0.7218,
      "step": 6772
    },
    {
      "epoch": 0.6108130044640844,
      "grad_norm": 1.6499277041231255,
      "learning_rate": 1.3891934751380879e-06,
      "loss": 0.8177,
      "step": 6773
    },
    {
      "epoch": 0.6109031879875547,
      "grad_norm": 1.735628929147473,
      "learning_rate": 1.3886372023445334e-06,
      "loss": 0.7694,
      "step": 6774
    },
    {
      "epoch": 0.610993371511025,
      "grad_norm": 1.4705179004117965,
      "learning_rate": 1.3880809817158246e-06,
      "loss": 0.8187,
      "step": 6775
    },
    {
      "epoch": 0.6110835550344952,
      "grad_norm": 1.8495313231163724,
      "learning_rate": 1.3875248132994206e-06,
      "loss": 0.7404,
      "step": 6776
    },
    {
      "epoch": 0.6111737385579654,
      "grad_norm": 1.8079637126277144,
      "learning_rate": 1.386968697142776e-06,
      "loss": 0.6694,
      "step": 6777
    },
    {
      "epoch": 0.6112639220814358,
      "grad_norm": 1.5001939807076907,
      "learning_rate": 1.386412633293343e-06,
      "loss": 0.7994,
      "step": 6778
    },
    {
      "epoch": 0.611354105604906,
      "grad_norm": 1.8786997850195932,
      "learning_rate": 1.3858566217985672e-06,
      "loss": 0.7248,
      "step": 6779
    },
    {
      "epoch": 0.6114442891283762,
      "grad_norm": 8.882383472076599,
      "learning_rate": 1.3853006627058905e-06,
      "loss": 0.836,
      "step": 6780
    },
    {
      "epoch": 0.6115344726518465,
      "grad_norm": 2.318687409329912,
      "learning_rate": 1.3847447560627512e-06,
      "loss": 0.6364,
      "step": 6781
    },
    {
      "epoch": 0.6116246561753168,
      "grad_norm": 1.695124391588167,
      "learning_rate": 1.3841889019165812e-06,
      "loss": 0.7917,
      "step": 6782
    },
    {
      "epoch": 0.611714839698787,
      "grad_norm": 2.269491434979124,
      "learning_rate": 1.3836331003148106e-06,
      "loss": 0.7125,
      "step": 6783
    },
    {
      "epoch": 0.6118050232222573,
      "grad_norm": 1.4142909418700296,
      "learning_rate": 1.3830773513048612e-06,
      "loss": 0.7874,
      "step": 6784
    },
    {
      "epoch": 0.6118952067457275,
      "grad_norm": 1.613584845473919,
      "learning_rate": 1.382521654934155e-06,
      "loss": 0.7664,
      "step": 6785
    },
    {
      "epoch": 0.6119853902691978,
      "grad_norm": 1.578694684531231,
      "learning_rate": 1.3819660112501057e-06,
      "loss": 0.6924,
      "step": 6786
    },
    {
      "epoch": 0.6120755737926681,
      "grad_norm": 1.6228940228444233,
      "learning_rate": 1.3814104203001234e-06,
      "loss": 0.7132,
      "step": 6787
    },
    {
      "epoch": 0.6121657573161383,
      "grad_norm": 1.6045773298411596,
      "learning_rate": 1.3808548821316156e-06,
      "loss": 0.7881,
      "step": 6788
    },
    {
      "epoch": 0.6122559408396087,
      "grad_norm": 1.5628217747288558,
      "learning_rate": 1.3802993967919824e-06,
      "loss": 0.685,
      "step": 6789
    },
    {
      "epoch": 0.6123461243630789,
      "grad_norm": 2.329854949725728,
      "learning_rate": 1.3797439643286227e-06,
      "loss": 0.7118,
      "step": 6790
    },
    {
      "epoch": 0.6124363078865491,
      "grad_norm": 2.0746410105294846,
      "learning_rate": 1.3791885847889277e-06,
      "loss": 0.7358,
      "step": 6791
    },
    {
      "epoch": 0.6125264914100194,
      "grad_norm": 1.5678602210528876,
      "learning_rate": 1.3786332582202853e-06,
      "loss": 0.7902,
      "step": 6792
    },
    {
      "epoch": 0.6126166749334897,
      "grad_norm": 1.6612860930162274,
      "learning_rate": 1.3780779846700799e-06,
      "loss": 0.7505,
      "step": 6793
    },
    {
      "epoch": 0.6127068584569599,
      "grad_norm": 1.497774221184643,
      "learning_rate": 1.3775227641856899e-06,
      "loss": 0.7633,
      "step": 6794
    },
    {
      "epoch": 0.6127970419804302,
      "grad_norm": 1.6949937230671925,
      "learning_rate": 1.37696759681449e-06,
      "loss": 0.7513,
      "step": 6795
    },
    {
      "epoch": 0.6128872255039004,
      "grad_norm": 0.8304401640399627,
      "learning_rate": 1.37641248260385e-06,
      "loss": 0.6073,
      "step": 6796
    },
    {
      "epoch": 0.6129774090273707,
      "grad_norm": 2.475249799972848,
      "learning_rate": 1.375857421601136e-06,
      "loss": 0.8008,
      "step": 6797
    },
    {
      "epoch": 0.613067592550841,
      "grad_norm": 1.671052819705325,
      "learning_rate": 1.3753024138537082e-06,
      "loss": 0.7603,
      "step": 6798
    },
    {
      "epoch": 0.6131577760743112,
      "grad_norm": 1.5926316769475446,
      "learning_rate": 1.3747474594089221e-06,
      "loss": 0.8187,
      "step": 6799
    },
    {
      "epoch": 0.6132479595977814,
      "grad_norm": 1.775575265211368,
      "learning_rate": 1.374192558314131e-06,
      "loss": 0.733,
      "step": 6800
    },
    {
      "epoch": 0.6133381431212518,
      "grad_norm": 1.719625492797701,
      "learning_rate": 1.373637710616681e-06,
      "loss": 0.8161,
      "step": 6801
    },
    {
      "epoch": 0.613428326644722,
      "grad_norm": 3.9793724819035785,
      "learning_rate": 1.373082916363916e-06,
      "loss": 0.8647,
      "step": 6802
    },
    {
      "epoch": 0.6135185101681923,
      "grad_norm": 1.9896753965584546,
      "learning_rate": 1.3725281756031732e-06,
      "loss": 0.7957,
      "step": 6803
    },
    {
      "epoch": 0.6136086936916625,
      "grad_norm": 1.8222353361337598,
      "learning_rate": 1.3719734883817858e-06,
      "loss": 0.7536,
      "step": 6804
    },
    {
      "epoch": 0.6136988772151328,
      "grad_norm": 1.6613702621157054,
      "learning_rate": 1.371418854747084e-06,
      "loss": 0.8682,
      "step": 6805
    },
    {
      "epoch": 0.6137890607386031,
      "grad_norm": 1.9482057190352462,
      "learning_rate": 1.3708642747463905e-06,
      "loss": 0.7876,
      "step": 6806
    },
    {
      "epoch": 0.6138792442620733,
      "grad_norm": 3.093002999798528,
      "learning_rate": 1.370309748427027e-06,
      "loss": 0.7811,
      "step": 6807
    },
    {
      "epoch": 0.6139694277855435,
      "grad_norm": 1.618384024588633,
      "learning_rate": 1.3697552758363079e-06,
      "loss": 0.8281,
      "step": 6808
    },
    {
      "epoch": 0.6140596113090139,
      "grad_norm": 4.328825415390266,
      "learning_rate": 1.3692008570215432e-06,
      "loss": 0.7576,
      "step": 6809
    },
    {
      "epoch": 0.6141497948324841,
      "grad_norm": 1.8284190666848636,
      "learning_rate": 1.3686464920300398e-06,
      "loss": 0.7022,
      "step": 6810
    },
    {
      "epoch": 0.6142399783559543,
      "grad_norm": 2.2461662479807933,
      "learning_rate": 1.3680921809090985e-06,
      "loss": 0.7339,
      "step": 6811
    },
    {
      "epoch": 0.6143301618794246,
      "grad_norm": 2.1711816572942366,
      "learning_rate": 1.3675379237060175e-06,
      "loss": 0.7979,
      "step": 6812
    },
    {
      "epoch": 0.6144203454028949,
      "grad_norm": 1.7683153164251655,
      "learning_rate": 1.366983720468088e-06,
      "loss": 0.7448,
      "step": 6813
    },
    {
      "epoch": 0.6145105289263652,
      "grad_norm": 2.9787933720397772,
      "learning_rate": 1.3664295712425972e-06,
      "loss": 0.7566,
      "step": 6814
    },
    {
      "epoch": 0.6146007124498354,
      "grad_norm": 1.7486297828957407,
      "learning_rate": 1.3658754760768296e-06,
      "loss": 0.6912,
      "step": 6815
    },
    {
      "epoch": 0.6146908959733057,
      "grad_norm": 1.9833611128435031,
      "learning_rate": 1.3653214350180621e-06,
      "loss": 0.785,
      "step": 6816
    },
    {
      "epoch": 0.614781079496776,
      "grad_norm": 1.5443198939780978,
      "learning_rate": 1.3647674481135703e-06,
      "loss": 0.8189,
      "step": 6817
    },
    {
      "epoch": 0.6148712630202462,
      "grad_norm": 5.202166128155343,
      "learning_rate": 1.3642135154106217e-06,
      "loss": 0.652,
      "step": 6818
    },
    {
      "epoch": 0.6149614465437164,
      "grad_norm": 2.0783600638393143,
      "learning_rate": 1.363659636956482e-06,
      "loss": 0.7019,
      "step": 6819
    },
    {
      "epoch": 0.6150516300671868,
      "grad_norm": 2.0113470530401854,
      "learning_rate": 1.3631058127984112e-06,
      "loss": 0.7218,
      "step": 6820
    },
    {
      "epoch": 0.615141813590657,
      "grad_norm": 5.819236215504225,
      "learning_rate": 1.3625520429836632e-06,
      "loss": 0.7331,
      "step": 6821
    },
    {
      "epoch": 0.6152319971141272,
      "grad_norm": 1.755721955655647,
      "learning_rate": 1.361998327559491e-06,
      "loss": 0.6626,
      "step": 6822
    },
    {
      "epoch": 0.6153221806375975,
      "grad_norm": 0.6558529242902088,
      "learning_rate": 1.3614446665731385e-06,
      "loss": 0.5813,
      "step": 6823
    },
    {
      "epoch": 0.6154123641610678,
      "grad_norm": 1.7364255450867339,
      "learning_rate": 1.3608910600718484e-06,
      "loss": 0.7655,
      "step": 6824
    },
    {
      "epoch": 0.615502547684538,
      "grad_norm": 1.6665326620907577,
      "learning_rate": 1.360337508102857e-06,
      "loss": 0.7587,
      "step": 6825
    },
    {
      "epoch": 0.6155927312080083,
      "grad_norm": 2.49146004227289,
      "learning_rate": 1.3597840107133962e-06,
      "loss": 0.8535,
      "step": 6826
    },
    {
      "epoch": 0.6156829147314785,
      "grad_norm": 1.905363783656887,
      "learning_rate": 1.3592305679506944e-06,
      "loss": 0.7595,
      "step": 6827
    },
    {
      "epoch": 0.6157730982549489,
      "grad_norm": 1.4324269548736452,
      "learning_rate": 1.3586771798619726e-06,
      "loss": 0.6588,
      "step": 6828
    },
    {
      "epoch": 0.6158632817784191,
      "grad_norm": 1.5274982813249525,
      "learning_rate": 1.358123846494451e-06,
      "loss": 0.8134,
      "step": 6829
    },
    {
      "epoch": 0.6159534653018893,
      "grad_norm": 2.316751438300552,
      "learning_rate": 1.3575705678953422e-06,
      "loss": 0.761,
      "step": 6830
    },
    {
      "epoch": 0.6160436488253596,
      "grad_norm": 1.9160413827827125,
      "learning_rate": 1.357017344111854e-06,
      "loss": 0.7945,
      "step": 6831
    },
    {
      "epoch": 0.6161338323488299,
      "grad_norm": 1.725054397278163,
      "learning_rate": 1.356464175191192e-06,
      "loss": 0.7657,
      "step": 6832
    },
    {
      "epoch": 0.6162240158723001,
      "grad_norm": 1.8160577911516507,
      "learning_rate": 1.3559110611805542e-06,
      "loss": 0.7822,
      "step": 6833
    },
    {
      "epoch": 0.6163141993957704,
      "grad_norm": 2.0895133354453392,
      "learning_rate": 1.3553580021271372e-06,
      "loss": 0.7936,
      "step": 6834
    },
    {
      "epoch": 0.6164043829192406,
      "grad_norm": 1.4132603430269124,
      "learning_rate": 1.3548049980781297e-06,
      "loss": 0.6979,
      "step": 6835
    },
    {
      "epoch": 0.616494566442711,
      "grad_norm": 1.6878634873733291,
      "learning_rate": 1.3542520490807166e-06,
      "loss": 0.7482,
      "step": 6836
    },
    {
      "epoch": 0.6165847499661812,
      "grad_norm": 2.123686328312925,
      "learning_rate": 1.3536991551820802e-06,
      "loss": 0.7424,
      "step": 6837
    },
    {
      "epoch": 0.6166749334896514,
      "grad_norm": 1.5040791994202507,
      "learning_rate": 1.3531463164293952e-06,
      "loss": 0.7245,
      "step": 6838
    },
    {
      "epoch": 0.6167651170131218,
      "grad_norm": 2.1527073824731597,
      "learning_rate": 1.3525935328698332e-06,
      "loss": 0.6612,
      "step": 6839
    },
    {
      "epoch": 0.616855300536592,
      "grad_norm": 1.9586666986782695,
      "learning_rate": 1.3520408045505605e-06,
      "loss": 0.6926,
      "step": 6840
    },
    {
      "epoch": 0.6169454840600622,
      "grad_norm": 1.722423788635323,
      "learning_rate": 1.3514881315187396e-06,
      "loss": 0.7207,
      "step": 6841
    },
    {
      "epoch": 0.6170356675835325,
      "grad_norm": 1.3814722236246015,
      "learning_rate": 1.3509355138215273e-06,
      "loss": 0.8137,
      "step": 6842
    },
    {
      "epoch": 0.6171258511070028,
      "grad_norm": 2.134441829322132,
      "learning_rate": 1.350382951506075e-06,
      "loss": 0.835,
      "step": 6843
    },
    {
      "epoch": 0.617216034630473,
      "grad_norm": 0.6408604445094525,
      "learning_rate": 1.349830444619532e-06,
      "loss": 0.5677,
      "step": 6844
    },
    {
      "epoch": 0.6173062181539433,
      "grad_norm": 2.176448589556431,
      "learning_rate": 1.3492779932090397e-06,
      "loss": 0.7697,
      "step": 6845
    },
    {
      "epoch": 0.6173964016774135,
      "grad_norm": 1.7334991855624926,
      "learning_rate": 1.3487255973217377e-06,
      "loss": 0.7793,
      "step": 6846
    },
    {
      "epoch": 0.6174865852008838,
      "grad_norm": 1.6094284048756353,
      "learning_rate": 1.3481732570047592e-06,
      "loss": 0.8206,
      "step": 6847
    },
    {
      "epoch": 0.6175767687243541,
      "grad_norm": 4.666576838764133,
      "learning_rate": 1.3476209723052314e-06,
      "loss": 0.6956,
      "step": 6848
    },
    {
      "epoch": 0.6176669522478243,
      "grad_norm": 1.851117945873826,
      "learning_rate": 1.3470687432702806e-06,
      "loss": 0.777,
      "step": 6849
    },
    {
      "epoch": 0.6177571357712945,
      "grad_norm": 2.0558614531862953,
      "learning_rate": 1.346516569947024e-06,
      "loss": 0.6886,
      "step": 6850
    },
    {
      "epoch": 0.6178473192947649,
      "grad_norm": 1.701773080677411,
      "learning_rate": 1.3459644523825774e-06,
      "loss": 0.7396,
      "step": 6851
    },
    {
      "epoch": 0.6179375028182351,
      "grad_norm": 1.9289131000655189,
      "learning_rate": 1.34541239062405e-06,
      "loss": 0.799,
      "step": 6852
    },
    {
      "epoch": 0.6180276863417054,
      "grad_norm": 2.6730623339839132,
      "learning_rate": 1.3448603847185464e-06,
      "loss": 0.7516,
      "step": 6853
    },
    {
      "epoch": 0.6181178698651756,
      "grad_norm": 1.443214089149545,
      "learning_rate": 1.344308434713168e-06,
      "loss": 0.695,
      "step": 6854
    },
    {
      "epoch": 0.6182080533886459,
      "grad_norm": 2.2147572011032874,
      "learning_rate": 1.3437565406550083e-06,
      "loss": 0.6898,
      "step": 6855
    },
    {
      "epoch": 0.6182982369121162,
      "grad_norm": 2.1632845475270552,
      "learning_rate": 1.34320470259116e-06,
      "loss": 0.7397,
      "step": 6856
    },
    {
      "epoch": 0.6183884204355864,
      "grad_norm": 2.186377973262577,
      "learning_rate": 1.3426529205687078e-06,
      "loss": 0.6581,
      "step": 6857
    },
    {
      "epoch": 0.6184786039590566,
      "grad_norm": 1.7206623882273517,
      "learning_rate": 1.3421011946347323e-06,
      "loss": 0.8096,
      "step": 6858
    },
    {
      "epoch": 0.618568787482527,
      "grad_norm": 1.7212015181440554,
      "learning_rate": 1.3415495248363113e-06,
      "loss": 0.7164,
      "step": 6859
    },
    {
      "epoch": 0.6186589710059972,
      "grad_norm": 2.109070819539324,
      "learning_rate": 1.3409979112205148e-06,
      "loss": 0.7752,
      "step": 6860
    },
    {
      "epoch": 0.6187491545294674,
      "grad_norm": 6.792563849167672,
      "learning_rate": 1.3404463538344107e-06,
      "loss": 0.7786,
      "step": 6861
    },
    {
      "epoch": 0.6188393380529378,
      "grad_norm": 2.285360108762116,
      "learning_rate": 1.3398948527250602e-06,
      "loss": 0.8406,
      "step": 6862
    },
    {
      "epoch": 0.618929521576408,
      "grad_norm": 0.785658376890602,
      "learning_rate": 1.3393434079395207e-06,
      "loss": 0.5923,
      "step": 6863
    },
    {
      "epoch": 0.6190197050998782,
      "grad_norm": 2.0940516240370197,
      "learning_rate": 1.3387920195248449e-06,
      "loss": 0.7423,
      "step": 6864
    },
    {
      "epoch": 0.6191098886233485,
      "grad_norm": 1.5459748548656382,
      "learning_rate": 1.3382406875280791e-06,
      "loss": 0.7679,
      "step": 6865
    },
    {
      "epoch": 0.6192000721468188,
      "grad_norm": 1.581432615448018,
      "learning_rate": 1.3376894119962672e-06,
      "loss": 0.7755,
      "step": 6866
    },
    {
      "epoch": 0.6192902556702891,
      "grad_norm": 2.242198026944746,
      "learning_rate": 1.3371381929764464e-06,
      "loss": 0.7786,
      "step": 6867
    },
    {
      "epoch": 0.6193804391937593,
      "grad_norm": 0.6362661145311125,
      "learning_rate": 1.3365870305156506e-06,
      "loss": 0.57,
      "step": 6868
    },
    {
      "epoch": 0.6194706227172295,
      "grad_norm": 2.4686183652912757,
      "learning_rate": 1.3360359246609073e-06,
      "loss": 0.7646,
      "step": 6869
    },
    {
      "epoch": 0.6195608062406999,
      "grad_norm": 1.7359786996149946,
      "learning_rate": 1.3354848754592387e-06,
      "loss": 0.7538,
      "step": 6870
    },
    {
      "epoch": 0.6196509897641701,
      "grad_norm": 1.6659600985398053,
      "learning_rate": 1.334933882957666e-06,
      "loss": 0.8407,
      "step": 6871
    },
    {
      "epoch": 0.6197411732876403,
      "grad_norm": 1.4494855691665813,
      "learning_rate": 1.3343829472032004e-06,
      "loss": 0.7181,
      "step": 6872
    },
    {
      "epoch": 0.6198313568111106,
      "grad_norm": 2.2215190317942963,
      "learning_rate": 1.3338320682428527e-06,
      "loss": 0.7423,
      "step": 6873
    },
    {
      "epoch": 0.6199215403345809,
      "grad_norm": 1.487506895289748,
      "learning_rate": 1.3332812461236263e-06,
      "loss": 0.7144,
      "step": 6874
    },
    {
      "epoch": 0.6200117238580511,
      "grad_norm": 1.4214624498085837,
      "learning_rate": 1.3327304808925192e-06,
      "loss": 0.7202,
      "step": 6875
    },
    {
      "epoch": 0.6201019073815214,
      "grad_norm": 2.2689931103470204,
      "learning_rate": 1.332179772596528e-06,
      "loss": 0.6169,
      "step": 6876
    },
    {
      "epoch": 0.6201920909049916,
      "grad_norm": 1.9891547961520766,
      "learning_rate": 1.3316291212826402e-06,
      "loss": 0.6774,
      "step": 6877
    },
    {
      "epoch": 0.620282274428462,
      "grad_norm": 2.484630835453985,
      "learning_rate": 1.3310785269978413e-06,
      "loss": 0.8064,
      "step": 6878
    },
    {
      "epoch": 0.6203724579519322,
      "grad_norm": 1.7143143725270413,
      "learning_rate": 1.3305279897891111e-06,
      "loss": 0.799,
      "step": 6879
    },
    {
      "epoch": 0.6204626414754024,
      "grad_norm": 1.935555189559431,
      "learning_rate": 1.329977509703424e-06,
      "loss": 0.617,
      "step": 6880
    },
    {
      "epoch": 0.6205528249988727,
      "grad_norm": 1.815757061198372,
      "learning_rate": 1.32942708678775e-06,
      "loss": 0.8414,
      "step": 6881
    },
    {
      "epoch": 0.620643008522343,
      "grad_norm": 1.7175476723930398,
      "learning_rate": 1.3288767210890548e-06,
      "loss": 0.7655,
      "step": 6882
    },
    {
      "epoch": 0.6207331920458132,
      "grad_norm": 1.7044818617778004,
      "learning_rate": 1.3283264126542986e-06,
      "loss": 0.7162,
      "step": 6883
    },
    {
      "epoch": 0.6208233755692835,
      "grad_norm": 1.5726298982370974,
      "learning_rate": 1.3277761615304356e-06,
      "loss": 0.7544,
      "step": 6884
    },
    {
      "epoch": 0.6209135590927538,
      "grad_norm": 1.4694247217823815,
      "learning_rate": 1.3272259677644178e-06,
      "loss": 0.8529,
      "step": 6885
    },
    {
      "epoch": 0.621003742616224,
      "grad_norm": 1.771984832965705,
      "learning_rate": 1.32667583140319e-06,
      "loss": 0.6272,
      "step": 6886
    },
    {
      "epoch": 0.6210939261396943,
      "grad_norm": 2.5441763218349656,
      "learning_rate": 1.3261257524936924e-06,
      "loss": 0.7986,
      "step": 6887
    },
    {
      "epoch": 0.6211841096631645,
      "grad_norm": 1.4537822918573193,
      "learning_rate": 1.3255757310828619e-06,
      "loss": 0.787,
      "step": 6888
    },
    {
      "epoch": 0.6212742931866349,
      "grad_norm": 3.1243296094887345,
      "learning_rate": 1.3250257672176282e-06,
      "loss": 0.7762,
      "step": 6889
    },
    {
      "epoch": 0.6213644767101051,
      "grad_norm": 1.7028131461951728,
      "learning_rate": 1.3244758609449183e-06,
      "loss": 0.6593,
      "step": 6890
    },
    {
      "epoch": 0.6214546602335753,
      "grad_norm": 2.0340583754108086,
      "learning_rate": 1.323926012311653e-06,
      "loss": 0.7691,
      "step": 6891
    },
    {
      "epoch": 0.6215448437570456,
      "grad_norm": 2.865548561932081,
      "learning_rate": 1.3233762213647476e-06,
      "loss": 0.7831,
      "step": 6892
    },
    {
      "epoch": 0.6216350272805159,
      "grad_norm": 1.6819620035803535,
      "learning_rate": 1.3228264881511137e-06,
      "loss": 0.7301,
      "step": 6893
    },
    {
      "epoch": 0.6217252108039861,
      "grad_norm": 2.376156575525092,
      "learning_rate": 1.322276812717658e-06,
      "loss": 0.8386,
      "step": 6894
    },
    {
      "epoch": 0.6218153943274564,
      "grad_norm": 1.5362006364655936,
      "learning_rate": 1.3217271951112818e-06,
      "loss": 0.6562,
      "step": 6895
    },
    {
      "epoch": 0.6219055778509266,
      "grad_norm": 11.431560370735292,
      "learning_rate": 1.321177635378881e-06,
      "loss": 0.7341,
      "step": 6896
    },
    {
      "epoch": 0.6219957613743969,
      "grad_norm": 2.1109833413771937,
      "learning_rate": 1.3206281335673475e-06,
      "loss": 0.6528,
      "step": 6897
    },
    {
      "epoch": 0.6220859448978672,
      "grad_norm": 1.9980459203932226,
      "learning_rate": 1.3200786897235677e-06,
      "loss": 0.728,
      "step": 6898
    },
    {
      "epoch": 0.6221761284213374,
      "grad_norm": 1.7318021231288827,
      "learning_rate": 1.3195293038944227e-06,
      "loss": 0.7725,
      "step": 6899
    },
    {
      "epoch": 0.6222663119448076,
      "grad_norm": 3.4737279416180433,
      "learning_rate": 1.3189799761267902e-06,
      "loss": 0.7478,
      "step": 6900
    },
    {
      "epoch": 0.622356495468278,
      "grad_norm": 2.3873512940756045,
      "learning_rate": 1.3184307064675412e-06,
      "loss": 0.7479,
      "step": 6901
    },
    {
      "epoch": 0.6224466789917482,
      "grad_norm": 0.7069838070626991,
      "learning_rate": 1.3178814949635416e-06,
      "loss": 0.6249,
      "step": 6902
    },
    {
      "epoch": 0.6225368625152184,
      "grad_norm": 1.8758556956524841,
      "learning_rate": 1.3173323416616549e-06,
      "loss": 0.7078,
      "step": 6903
    },
    {
      "epoch": 0.6226270460386887,
      "grad_norm": 1.5829282878312914,
      "learning_rate": 1.3167832466087361e-06,
      "loss": 0.8033,
      "step": 6904
    },
    {
      "epoch": 0.622717229562159,
      "grad_norm": 1.3307213929229835,
      "learning_rate": 1.3162342098516388e-06,
      "loss": 0.6988,
      "step": 6905
    },
    {
      "epoch": 0.6228074130856293,
      "grad_norm": 1.5254432971960004,
      "learning_rate": 1.3156852314372086e-06,
      "loss": 0.6712,
      "step": 6906
    },
    {
      "epoch": 0.6228975966090995,
      "grad_norm": 0.6836749873255576,
      "learning_rate": 1.3151363114122882e-06,
      "loss": 0.5619,
      "step": 6907
    },
    {
      "epoch": 0.6229877801325698,
      "grad_norm": 1.863839503696974,
      "learning_rate": 1.3145874498237133e-06,
      "loss": 0.767,
      "step": 6908
    },
    {
      "epoch": 0.6230779636560401,
      "grad_norm": 1.655283952077593,
      "learning_rate": 1.3140386467183166e-06,
      "loss": 0.8005,
      "step": 6909
    },
    {
      "epoch": 0.6231681471795103,
      "grad_norm": 5.523614818527098,
      "learning_rate": 1.3134899021429258e-06,
      "loss": 0.796,
      "step": 6910
    },
    {
      "epoch": 0.6232583307029805,
      "grad_norm": 4.376454356831196,
      "learning_rate": 1.3129412161443604e-06,
      "loss": 0.7323,
      "step": 6911
    },
    {
      "epoch": 0.6233485142264509,
      "grad_norm": 1.797457326984686,
      "learning_rate": 1.3123925887694402e-06,
      "loss": 0.7636,
      "step": 6912
    },
    {
      "epoch": 0.6234386977499211,
      "grad_norm": 2.1443681498556817,
      "learning_rate": 1.3118440200649756e-06,
      "loss": 0.7618,
      "step": 6913
    },
    {
      "epoch": 0.6235288812733913,
      "grad_norm": 6.907174027020804,
      "learning_rate": 1.3112955100777727e-06,
      "loss": 0.7141,
      "step": 6914
    },
    {
      "epoch": 0.6236190647968616,
      "grad_norm": 1.3919557354948564,
      "learning_rate": 1.3107470588546353e-06,
      "loss": 0.7859,
      "step": 6915
    },
    {
      "epoch": 0.6237092483203319,
      "grad_norm": 2.4779961224172333,
      "learning_rate": 1.3101986664423583e-06,
      "loss": 0.7227,
      "step": 6916
    },
    {
      "epoch": 0.6237994318438022,
      "grad_norm": 2.070026421124927,
      "learning_rate": 1.3096503328877358e-06,
      "loss": 0.728,
      "step": 6917
    },
    {
      "epoch": 0.6238896153672724,
      "grad_norm": 1.9359027061999563,
      "learning_rate": 1.309102058237553e-06,
      "loss": 0.7713,
      "step": 6918
    },
    {
      "epoch": 0.6239797988907426,
      "grad_norm": 1.4332972747495774,
      "learning_rate": 1.3085538425385917e-06,
      "loss": 0.8037,
      "step": 6919
    },
    {
      "epoch": 0.624069982414213,
      "grad_norm": 1.8787389986143768,
      "learning_rate": 1.3080056858376298e-06,
      "loss": 0.7831,
      "step": 6920
    },
    {
      "epoch": 0.6241601659376832,
      "grad_norm": 2.052481856271745,
      "learning_rate": 1.3074575881814383e-06,
      "loss": 0.7389,
      "step": 6921
    },
    {
      "epoch": 0.6242503494611534,
      "grad_norm": 1.9574930403818325,
      "learning_rate": 1.3069095496167841e-06,
      "loss": 0.6787,
      "step": 6922
    },
    {
      "epoch": 0.6243405329846237,
      "grad_norm": 1.482121574130864,
      "learning_rate": 1.3063615701904285e-06,
      "loss": 0.6614,
      "step": 6923
    },
    {
      "epoch": 0.624430716508094,
      "grad_norm": 1.8214173463534293,
      "learning_rate": 1.3058136499491283e-06,
      "loss": 0.7849,
      "step": 6924
    },
    {
      "epoch": 0.6245209000315642,
      "grad_norm": 2.6457755515529953,
      "learning_rate": 1.3052657889396357e-06,
      "loss": 0.7371,
      "step": 6925
    },
    {
      "epoch": 0.6246110835550345,
      "grad_norm": 1.5761668832692797,
      "learning_rate": 1.304717987208696e-06,
      "loss": 0.7368,
      "step": 6926
    },
    {
      "epoch": 0.6247012670785047,
      "grad_norm": 2.9996355153239,
      "learning_rate": 1.304170244803052e-06,
      "loss": 0.7393,
      "step": 6927
    },
    {
      "epoch": 0.624791450601975,
      "grad_norm": 2.6253476593856293,
      "learning_rate": 1.3036225617694387e-06,
      "loss": 0.7236,
      "step": 6928
    },
    {
      "epoch": 0.6248816341254453,
      "grad_norm": 3.072836578532399,
      "learning_rate": 1.3030749381545892e-06,
      "loss": 0.781,
      "step": 6929
    },
    {
      "epoch": 0.6249718176489155,
      "grad_norm": 1.792012477001112,
      "learning_rate": 1.3025273740052285e-06,
      "loss": 0.692,
      "step": 6930
    },
    {
      "epoch": 0.6250620011723858,
      "grad_norm": 1.4710260976347378,
      "learning_rate": 1.3019798693680774e-06,
      "loss": 0.8006,
      "step": 6931
    },
    {
      "epoch": 0.6251521846958561,
      "grad_norm": 1.6112758188716763,
      "learning_rate": 1.3014324242898536e-06,
      "loss": 0.8694,
      "step": 6932
    },
    {
      "epoch": 0.6252423682193263,
      "grad_norm": 2.0294601771246135,
      "learning_rate": 1.3008850388172668e-06,
      "loss": 0.8109,
      "step": 6933
    },
    {
      "epoch": 0.6253325517427966,
      "grad_norm": 1.6852929552820457,
      "learning_rate": 1.3003377129970233e-06,
      "loss": 0.7106,
      "step": 6934
    },
    {
      "epoch": 0.6254227352662669,
      "grad_norm": 1.390881889703999,
      "learning_rate": 1.2997904468758243e-06,
      "loss": 0.6711,
      "step": 6935
    },
    {
      "epoch": 0.6255129187897371,
      "grad_norm": 1.620223508062919,
      "learning_rate": 1.299243240500365e-06,
      "loss": 0.772,
      "step": 6936
    },
    {
      "epoch": 0.6256031023132074,
      "grad_norm": 1.818724500339726,
      "learning_rate": 1.2986960939173368e-06,
      "loss": 0.7697,
      "step": 6937
    },
    {
      "epoch": 0.6256932858366776,
      "grad_norm": 1.7365725230486682,
      "learning_rate": 1.298149007173424e-06,
      "loss": 0.798,
      "step": 6938
    },
    {
      "epoch": 0.625783469360148,
      "grad_norm": 1.9514663367176706,
      "learning_rate": 1.2976019803153087e-06,
      "loss": 0.7846,
      "step": 6939
    },
    {
      "epoch": 0.6258736528836182,
      "grad_norm": 1.6481137070686636,
      "learning_rate": 1.2970550133896652e-06,
      "loss": 0.7001,
      "step": 6940
    },
    {
      "epoch": 0.6259638364070884,
      "grad_norm": 1.5796699892728086,
      "learning_rate": 1.2965081064431634e-06,
      "loss": 0.8351,
      "step": 6941
    },
    {
      "epoch": 0.6260540199305586,
      "grad_norm": 2.12807108044762,
      "learning_rate": 1.2959612595224698e-06,
      "loss": 0.7779,
      "step": 6942
    },
    {
      "epoch": 0.626144203454029,
      "grad_norm": 2.0882418473590465,
      "learning_rate": 1.2954144726742424e-06,
      "loss": 0.7637,
      "step": 6943
    },
    {
      "epoch": 0.6262343869774992,
      "grad_norm": 1.4925271491515957,
      "learning_rate": 1.2948677459451385e-06,
      "loss": 0.7496,
      "step": 6944
    },
    {
      "epoch": 0.6263245705009695,
      "grad_norm": 1.8620513759740809,
      "learning_rate": 1.2943210793818064e-06,
      "loss": 0.8002,
      "step": 6945
    },
    {
      "epoch": 0.6264147540244397,
      "grad_norm": 1.8408651250347854,
      "learning_rate": 1.2937744730308899e-06,
      "loss": 0.7345,
      "step": 6946
    },
    {
      "epoch": 0.62650493754791,
      "grad_norm": 4.498975849169036,
      "learning_rate": 1.2932279269390305e-06,
      "loss": 0.7544,
      "step": 6947
    },
    {
      "epoch": 0.6265951210713803,
      "grad_norm": 1.723372814247636,
      "learning_rate": 1.292681441152861e-06,
      "loss": 0.6576,
      "step": 6948
    },
    {
      "epoch": 0.6266853045948505,
      "grad_norm": 1.7336391228148658,
      "learning_rate": 1.292135015719011e-06,
      "loss": 0.762,
      "step": 6949
    },
    {
      "epoch": 0.6267754881183207,
      "grad_norm": 2.0884089883595918,
      "learning_rate": 1.2915886506841046e-06,
      "loss": 0.7139,
      "step": 6950
    },
    {
      "epoch": 0.6268656716417911,
      "grad_norm": 1.5145177040765498,
      "learning_rate": 1.2910423460947613e-06,
      "loss": 0.6908,
      "step": 6951
    },
    {
      "epoch": 0.6269558551652613,
      "grad_norm": 1.785055746175036,
      "learning_rate": 1.290496101997594e-06,
      "loss": 0.7231,
      "step": 6952
    },
    {
      "epoch": 0.6270460386887315,
      "grad_norm": 0.7374868812040796,
      "learning_rate": 1.2899499184392105e-06,
      "loss": 0.5818,
      "step": 6953
    },
    {
      "epoch": 0.6271362222122018,
      "grad_norm": 1.6734154280374318,
      "learning_rate": 1.289403795466216e-06,
      "loss": 0.7276,
      "step": 6954
    },
    {
      "epoch": 0.6272264057356721,
      "grad_norm": 1.9229263283379028,
      "learning_rate": 1.288857733125207e-06,
      "loss": 0.7661,
      "step": 6955
    },
    {
      "epoch": 0.6273165892591424,
      "grad_norm": 3.1669575574347792,
      "learning_rate": 1.2883117314627785e-06,
      "loss": 0.7482,
      "step": 6956
    },
    {
      "epoch": 0.6274067727826126,
      "grad_norm": 1.6772819859469308,
      "learning_rate": 1.2877657905255168e-06,
      "loss": 0.6952,
      "step": 6957
    },
    {
      "epoch": 0.6274969563060829,
      "grad_norm": 1.6308994311136227,
      "learning_rate": 1.2872199103600046e-06,
      "loss": 0.8126,
      "step": 6958
    },
    {
      "epoch": 0.6275871398295532,
      "grad_norm": 1.5457519153359145,
      "learning_rate": 1.286674091012821e-06,
      "loss": 0.8223,
      "step": 6959
    },
    {
      "epoch": 0.6276773233530234,
      "grad_norm": 1.5540929069436213,
      "learning_rate": 1.2861283325305356e-06,
      "loss": 0.8083,
      "step": 6960
    },
    {
      "epoch": 0.6277675068764936,
      "grad_norm": 1.4331470593943623,
      "learning_rate": 1.2855826349597185e-06,
      "loss": 0.7521,
      "step": 6961
    },
    {
      "epoch": 0.627857690399964,
      "grad_norm": 1.355082871975011,
      "learning_rate": 1.2850369983469302e-06,
      "loss": 0.6071,
      "step": 6962
    },
    {
      "epoch": 0.6279478739234342,
      "grad_norm": 3.685090263638011,
      "learning_rate": 1.2844914227387266e-06,
      "loss": 0.7791,
      "step": 6963
    },
    {
      "epoch": 0.6280380574469044,
      "grad_norm": 8.672398456892832,
      "learning_rate": 1.2839459081816606e-06,
      "loss": 0.7744,
      "step": 6964
    },
    {
      "epoch": 0.6281282409703747,
      "grad_norm": 1.8052702124002553,
      "learning_rate": 1.283400454722278e-06,
      "loss": 0.8182,
      "step": 6965
    },
    {
      "epoch": 0.628218424493845,
      "grad_norm": 2.0710377700360203,
      "learning_rate": 1.28285506240712e-06,
      "loss": 0.8115,
      "step": 6966
    },
    {
      "epoch": 0.6283086080173153,
      "grad_norm": 1.590420854879609,
      "learning_rate": 1.2823097312827225e-06,
      "loss": 0.7772,
      "step": 6967
    },
    {
      "epoch": 0.6283987915407855,
      "grad_norm": 3.083559921461817,
      "learning_rate": 1.2817644613956153e-06,
      "loss": 0.7632,
      "step": 6968
    },
    {
      "epoch": 0.6284889750642557,
      "grad_norm": 3.2399370955023956,
      "learning_rate": 1.2812192527923253e-06,
      "loss": 0.8175,
      "step": 6969
    },
    {
      "epoch": 0.6285791585877261,
      "grad_norm": 1.79843821322562,
      "learning_rate": 1.2806741055193712e-06,
      "loss": 0.7888,
      "step": 6970
    },
    {
      "epoch": 0.6286693421111963,
      "grad_norm": 1.873621115702179,
      "learning_rate": 1.2801290196232695e-06,
      "loss": 0.7109,
      "step": 6971
    },
    {
      "epoch": 0.6287595256346665,
      "grad_norm": 1.7901840359366918,
      "learning_rate": 1.2795839951505282e-06,
      "loss": 0.7434,
      "step": 6972
    },
    {
      "epoch": 0.6288497091581368,
      "grad_norm": 6.927886075041217,
      "learning_rate": 1.2790390321476538e-06,
      "loss": 0.7715,
      "step": 6973
    },
    {
      "epoch": 0.6289398926816071,
      "grad_norm": 1.552975656248596,
      "learning_rate": 1.2784941306611446e-06,
      "loss": 0.7216,
      "step": 6974
    },
    {
      "epoch": 0.6290300762050773,
      "grad_norm": 1.7941887056538874,
      "learning_rate": 1.2779492907374935e-06,
      "loss": 0.8136,
      "step": 6975
    },
    {
      "epoch": 0.6291202597285476,
      "grad_norm": 1.8442652354190923,
      "learning_rate": 1.2774045124231911e-06,
      "loss": 0.7426,
      "step": 6976
    },
    {
      "epoch": 0.6292104432520178,
      "grad_norm": 1.8942270565629626,
      "learning_rate": 1.2768597957647197e-06,
      "loss": 0.6352,
      "step": 6977
    },
    {
      "epoch": 0.6293006267754881,
      "grad_norm": 1.8481778140579705,
      "learning_rate": 1.2763151408085582e-06,
      "loss": 0.7631,
      "step": 6978
    },
    {
      "epoch": 0.6293908102989584,
      "grad_norm": 2.2465703044038916,
      "learning_rate": 1.2757705476011788e-06,
      "loss": 0.7191,
      "step": 6979
    },
    {
      "epoch": 0.6294809938224286,
      "grad_norm": 1.6273074273768313,
      "learning_rate": 1.27522601618905e-06,
      "loss": 0.6971,
      "step": 6980
    },
    {
      "epoch": 0.629571177345899,
      "grad_norm": 0.5689746109254185,
      "learning_rate": 1.2746815466186337e-06,
      "loss": 0.6046,
      "step": 6981
    },
    {
      "epoch": 0.6296613608693692,
      "grad_norm": 2.076220794417205,
      "learning_rate": 1.274137138936387e-06,
      "loss": 0.7528,
      "step": 6982
    },
    {
      "epoch": 0.6297515443928394,
      "grad_norm": 1.8769298794167337,
      "learning_rate": 1.2735927931887625e-06,
      "loss": 0.7521,
      "step": 6983
    },
    {
      "epoch": 0.6298417279163097,
      "grad_norm": 0.6938080746167903,
      "learning_rate": 1.2730485094222061e-06,
      "loss": 0.5864,
      "step": 6984
    },
    {
      "epoch": 0.62993191143978,
      "grad_norm": 2.0156815395261773,
      "learning_rate": 1.2725042876831586e-06,
      "loss": 0.7889,
      "step": 6985
    },
    {
      "epoch": 0.6300220949632502,
      "grad_norm": 2.090528942689718,
      "learning_rate": 1.2719601280180573e-06,
      "loss": 0.6493,
      "step": 6986
    },
    {
      "epoch": 0.6301122784867205,
      "grad_norm": 1.5229258925353732,
      "learning_rate": 1.2714160304733317e-06,
      "loss": 0.8168,
      "step": 6987
    },
    {
      "epoch": 0.6302024620101907,
      "grad_norm": 2.219203419091548,
      "learning_rate": 1.2708719950954082e-06,
      "loss": 0.7645,
      "step": 6988
    },
    {
      "epoch": 0.630292645533661,
      "grad_norm": 2.3129830757969976,
      "learning_rate": 1.2703280219307065e-06,
      "loss": 0.7754,
      "step": 6989
    },
    {
      "epoch": 0.6303828290571313,
      "grad_norm": 1.6666035481263075,
      "learning_rate": 1.2697841110256411e-06,
      "loss": 0.6816,
      "step": 6990
    },
    {
      "epoch": 0.6304730125806015,
      "grad_norm": 1.5141347263615528,
      "learning_rate": 1.2692402624266221e-06,
      "loss": 0.7958,
      "step": 6991
    },
    {
      "epoch": 0.6305631961040717,
      "grad_norm": 1.8308193859228303,
      "learning_rate": 1.2686964761800529e-06,
      "loss": 0.6654,
      "step": 6992
    },
    {
      "epoch": 0.6306533796275421,
      "grad_norm": 0.6368682311412425,
      "learning_rate": 1.268152752332333e-06,
      "loss": 0.5844,
      "step": 6993
    },
    {
      "epoch": 0.6307435631510123,
      "grad_norm": 1.9018280771655505,
      "learning_rate": 1.2676090909298549e-06,
      "loss": 0.6633,
      "step": 6994
    },
    {
      "epoch": 0.6308337466744826,
      "grad_norm": 1.9033315236449895,
      "learning_rate": 1.2670654920190086e-06,
      "loss": 0.803,
      "step": 6995
    },
    {
      "epoch": 0.6309239301979528,
      "grad_norm": 1.3780174092271318,
      "learning_rate": 1.2665219556461754e-06,
      "loss": 0.8014,
      "step": 6996
    },
    {
      "epoch": 0.6310141137214231,
      "grad_norm": 1.745213297293115,
      "learning_rate": 1.2659784818577329e-06,
      "loss": 0.8512,
      "step": 6997
    },
    {
      "epoch": 0.6311042972448934,
      "grad_norm": 1.7907527618365604,
      "learning_rate": 1.2654350707000542e-06,
      "loss": 0.6348,
      "step": 6998
    },
    {
      "epoch": 0.6311944807683636,
      "grad_norm": 1.60237934402221,
      "learning_rate": 1.264891722219505e-06,
      "loss": 0.7365,
      "step": 6999
    },
    {
      "epoch": 0.6312846642918338,
      "grad_norm": 2.2329444840906283,
      "learning_rate": 1.2643484364624483e-06,
      "loss": 0.7483,
      "step": 7000
    },
    {
      "epoch": 0.6313748478153042,
      "grad_norm": 1.7349106932199987,
      "learning_rate": 1.2638052134752393e-06,
      "loss": 0.8028,
      "step": 7001
    },
    {
      "epoch": 0.6314650313387744,
      "grad_norm": 2.091622823531812,
      "learning_rate": 1.2632620533042277e-06,
      "loss": 0.8236,
      "step": 7002
    },
    {
      "epoch": 0.6315552148622446,
      "grad_norm": 2.149130858465315,
      "learning_rate": 1.2627189559957612e-06,
      "loss": 0.7173,
      "step": 7003
    },
    {
      "epoch": 0.631645398385715,
      "grad_norm": 2.2096014400485773,
      "learning_rate": 1.262175921596178e-06,
      "loss": 0.8118,
      "step": 7004
    },
    {
      "epoch": 0.6317355819091852,
      "grad_norm": 1.8134152962355572,
      "learning_rate": 1.2616329501518137e-06,
      "loss": 0.6265,
      "step": 7005
    },
    {
      "epoch": 0.6318257654326555,
      "grad_norm": 1.7771959180449752,
      "learning_rate": 1.2610900417089978e-06,
      "loss": 0.6735,
      "step": 7006
    },
    {
      "epoch": 0.6319159489561257,
      "grad_norm": 1.7875795173297824,
      "learning_rate": 1.2605471963140535e-06,
      "loss": 0.7646,
      "step": 7007
    },
    {
      "epoch": 0.632006132479596,
      "grad_norm": 2.230213905647036,
      "learning_rate": 1.2600044140133e-06,
      "loss": 0.8218,
      "step": 7008
    },
    {
      "epoch": 0.6320963160030663,
      "grad_norm": 1.6734897979343295,
      "learning_rate": 1.2594616948530493e-06,
      "loss": 0.7025,
      "step": 7009
    },
    {
      "epoch": 0.6321864995265365,
      "grad_norm": 1.9068668180507093,
      "learning_rate": 1.258919038879611e-06,
      "loss": 0.7736,
      "step": 7010
    },
    {
      "epoch": 0.6322766830500067,
      "grad_norm": 1.7946278992377256,
      "learning_rate": 1.2583764461392859e-06,
      "loss": 0.7538,
      "step": 7011
    },
    {
      "epoch": 0.6323668665734771,
      "grad_norm": 1.7452937604125236,
      "learning_rate": 1.2578339166783724e-06,
      "loss": 0.7124,
      "step": 7012
    },
    {
      "epoch": 0.6324570500969473,
      "grad_norm": 1.6402992288405454,
      "learning_rate": 1.2572914505431613e-06,
      "loss": 0.6872,
      "step": 7013
    },
    {
      "epoch": 0.6325472336204175,
      "grad_norm": 1.5200054929031706,
      "learning_rate": 1.2567490477799383e-06,
      "loss": 0.652,
      "step": 7014
    },
    {
      "epoch": 0.6326374171438878,
      "grad_norm": 9.969867703431465,
      "learning_rate": 1.2562067084349852e-06,
      "loss": 0.8455,
      "step": 7015
    },
    {
      "epoch": 0.6327276006673581,
      "grad_norm": 2.5714122889018363,
      "learning_rate": 1.2556644325545764e-06,
      "loss": 0.735,
      "step": 7016
    },
    {
      "epoch": 0.6328177841908283,
      "grad_norm": 2.075258730648615,
      "learning_rate": 1.255122220184983e-06,
      "loss": 0.7298,
      "step": 7017
    },
    {
      "epoch": 0.6329079677142986,
      "grad_norm": 2.639720874826933,
      "learning_rate": 1.2545800713724694e-06,
      "loss": 0.6619,
      "step": 7018
    },
    {
      "epoch": 0.6329981512377688,
      "grad_norm": 1.4202475823568332,
      "learning_rate": 1.254037986163294e-06,
      "loss": 0.6217,
      "step": 7019
    },
    {
      "epoch": 0.6330883347612392,
      "grad_norm": 1.74924398168074,
      "learning_rate": 1.2534959646037104e-06,
      "loss": 0.7127,
      "step": 7020
    },
    {
      "epoch": 0.6331785182847094,
      "grad_norm": 1.4360105842529671,
      "learning_rate": 1.2529540067399675e-06,
      "loss": 0.7565,
      "step": 7021
    },
    {
      "epoch": 0.6332687018081796,
      "grad_norm": 1.7039168030885588,
      "learning_rate": 1.252412112618308e-06,
      "loss": 0.6907,
      "step": 7022
    },
    {
      "epoch": 0.6333588853316499,
      "grad_norm": 1.9622977960495342,
      "learning_rate": 1.2518702822849696e-06,
      "loss": 0.6388,
      "step": 7023
    },
    {
      "epoch": 0.6334490688551202,
      "grad_norm": 2.094762301195709,
      "learning_rate": 1.2513285157861831e-06,
      "loss": 0.6967,
      "step": 7024
    },
    {
      "epoch": 0.6335392523785904,
      "grad_norm": 1.5485413178747551,
      "learning_rate": 1.2507868131681764e-06,
      "loss": 0.7458,
      "step": 7025
    },
    {
      "epoch": 0.6336294359020607,
      "grad_norm": 1.6528679487419415,
      "learning_rate": 1.250245174477169e-06,
      "loss": 0.7327,
      "step": 7026
    },
    {
      "epoch": 0.6337196194255309,
      "grad_norm": 0.652347975848553,
      "learning_rate": 1.2497035997593783e-06,
      "loss": 0.5202,
      "step": 7027
    },
    {
      "epoch": 0.6338098029490012,
      "grad_norm": 2.5663426128608964,
      "learning_rate": 1.2491620890610135e-06,
      "loss": 0.7499,
      "step": 7028
    },
    {
      "epoch": 0.6338999864724715,
      "grad_norm": 1.492187020666235,
      "learning_rate": 1.2486206424282788e-06,
      "loss": 0.7336,
      "step": 7029
    },
    {
      "epoch": 0.6339901699959417,
      "grad_norm": 1.6548370236135914,
      "learning_rate": 1.2480792599073743e-06,
      "loss": 0.807,
      "step": 7030
    },
    {
      "epoch": 0.634080353519412,
      "grad_norm": 2.808588486880065,
      "learning_rate": 1.247537941544493e-06,
      "loss": 0.7898,
      "step": 7031
    },
    {
      "epoch": 0.6341705370428823,
      "grad_norm": 2.9121745439346216,
      "learning_rate": 1.2469966873858242e-06,
      "loss": 0.7843,
      "step": 7032
    },
    {
      "epoch": 0.6342607205663525,
      "grad_norm": 2.0359528313364734,
      "learning_rate": 1.2464554974775496e-06,
      "loss": 0.6762,
      "step": 7033
    },
    {
      "epoch": 0.6343509040898228,
      "grad_norm": 1.843310934256047,
      "learning_rate": 1.2459143718658474e-06,
      "loss": 0.7497,
      "step": 7034
    },
    {
      "epoch": 0.6344410876132931,
      "grad_norm": 1.752294398721299,
      "learning_rate": 1.2453733105968886e-06,
      "loss": 0.6856,
      "step": 7035
    },
    {
      "epoch": 0.6345312711367633,
      "grad_norm": 0.610389403123852,
      "learning_rate": 1.2448323137168399e-06,
      "loss": 0.5861,
      "step": 7036
    },
    {
      "epoch": 0.6346214546602336,
      "grad_norm": 1.5326532434758158,
      "learning_rate": 1.2442913812718625e-06,
      "loss": 0.6721,
      "step": 7037
    },
    {
      "epoch": 0.6347116381837038,
      "grad_norm": 2.252646161755552,
      "learning_rate": 1.2437505133081108e-06,
      "loss": 0.736,
      "step": 7038
    },
    {
      "epoch": 0.6348018217071741,
      "grad_norm": 1.6353139014869544,
      "learning_rate": 1.2432097098717358e-06,
      "loss": 0.7668,
      "step": 7039
    },
    {
      "epoch": 0.6348920052306444,
      "grad_norm": 1.9795452425094313,
      "learning_rate": 1.2426689710088813e-06,
      "loss": 0.7536,
      "step": 7040
    },
    {
      "epoch": 0.6349821887541146,
      "grad_norm": 1.7274402441352654,
      "learning_rate": 1.2421282967656855e-06,
      "loss": 0.8046,
      "step": 7041
    },
    {
      "epoch": 0.6350723722775848,
      "grad_norm": 1.714554399483388,
      "learning_rate": 1.2415876871882827e-06,
      "loss": 0.779,
      "step": 7042
    },
    {
      "epoch": 0.6351625558010552,
      "grad_norm": 0.6730348423232592,
      "learning_rate": 1.2410471423227998e-06,
      "loss": 0.6497,
      "step": 7043
    },
    {
      "epoch": 0.6352527393245254,
      "grad_norm": 2.271609275841117,
      "learning_rate": 1.24050666221536e-06,
      "loss": 0.7413,
      "step": 7044
    },
    {
      "epoch": 0.6353429228479957,
      "grad_norm": 1.585089771724895,
      "learning_rate": 1.23996624691208e-06,
      "loss": 0.8165,
      "step": 7045
    },
    {
      "epoch": 0.6354331063714659,
      "grad_norm": 1.63675858988561,
      "learning_rate": 1.2394258964590693e-06,
      "loss": 0.7342,
      "step": 7046
    },
    {
      "epoch": 0.6355232898949362,
      "grad_norm": 1.9180731657550119,
      "learning_rate": 1.238885610902436e-06,
      "loss": 0.698,
      "step": 7047
    },
    {
      "epoch": 0.6356134734184065,
      "grad_norm": 2.7042207646848717,
      "learning_rate": 1.2383453902882787e-06,
      "loss": 0.7615,
      "step": 7048
    },
    {
      "epoch": 0.6357036569418767,
      "grad_norm": 1.5108189628045283,
      "learning_rate": 1.2378052346626927e-06,
      "loss": 0.813,
      "step": 7049
    },
    {
      "epoch": 0.6357938404653469,
      "grad_norm": 2.1116919331994772,
      "learning_rate": 1.2372651440717665e-06,
      "loss": 0.6754,
      "step": 7050
    },
    {
      "epoch": 0.6358840239888173,
      "grad_norm": 1.8079170295742515,
      "learning_rate": 1.236725118561584e-06,
      "loss": 0.7845,
      "step": 7051
    },
    {
      "epoch": 0.6359742075122875,
      "grad_norm": 1.356919054171047,
      "learning_rate": 1.2361851581782232e-06,
      "loss": 0.7499,
      "step": 7052
    },
    {
      "epoch": 0.6360643910357577,
      "grad_norm": 1.7947827307031616,
      "learning_rate": 1.2356452629677554e-06,
      "loss": 0.8114,
      "step": 7053
    },
    {
      "epoch": 0.6361545745592281,
      "grad_norm": 1.5082912179255334,
      "learning_rate": 1.2351054329762494e-06,
      "loss": 0.6842,
      "step": 7054
    },
    {
      "epoch": 0.6362447580826983,
      "grad_norm": 1.5869493685976348,
      "learning_rate": 1.2345656682497648e-06,
      "loss": 0.7766,
      "step": 7055
    },
    {
      "epoch": 0.6363349416061685,
      "grad_norm": 1.339388845196829,
      "learning_rate": 1.2340259688343583e-06,
      "loss": 0.7265,
      "step": 7056
    },
    {
      "epoch": 0.6364251251296388,
      "grad_norm": 1.4901349395735333,
      "learning_rate": 1.2334863347760803e-06,
      "loss": 0.6455,
      "step": 7057
    },
    {
      "epoch": 0.6365153086531091,
      "grad_norm": 0.6446439008813764,
      "learning_rate": 1.2329467661209734e-06,
      "loss": 0.5451,
      "step": 7058
    },
    {
      "epoch": 0.6366054921765794,
      "grad_norm": 1.671470165259147,
      "learning_rate": 1.2324072629150788e-06,
      "loss": 0.7925,
      "step": 7059
    },
    {
      "epoch": 0.6366956757000496,
      "grad_norm": 1.9327214903440357,
      "learning_rate": 1.2318678252044287e-06,
      "loss": 0.7956,
      "step": 7060
    },
    {
      "epoch": 0.6367858592235198,
      "grad_norm": 2.128473248325064,
      "learning_rate": 1.2313284530350512e-06,
      "loss": 0.8073,
      "step": 7061
    },
    {
      "epoch": 0.6368760427469902,
      "grad_norm": 1.7507674714158632,
      "learning_rate": 1.230789146452969e-06,
      "loss": 0.6833,
      "step": 7062
    },
    {
      "epoch": 0.6369662262704604,
      "grad_norm": 2.0842742575301907,
      "learning_rate": 1.2302499055041974e-06,
      "loss": 0.7531,
      "step": 7063
    },
    {
      "epoch": 0.6370564097939306,
      "grad_norm": 0.6695767728804219,
      "learning_rate": 1.2297107302347488e-06,
      "loss": 0.5825,
      "step": 7064
    },
    {
      "epoch": 0.6371465933174009,
      "grad_norm": 1.8873876171886605,
      "learning_rate": 1.2291716206906275e-06,
      "loss": 0.8172,
      "step": 7065
    },
    {
      "epoch": 0.6372367768408712,
      "grad_norm": 2.074470576572535,
      "learning_rate": 1.2286325769178345e-06,
      "loss": 0.8185,
      "step": 7066
    },
    {
      "epoch": 0.6373269603643414,
      "grad_norm": 1.7005500576746415,
      "learning_rate": 1.2280935989623633e-06,
      "loss": 0.7808,
      "step": 7067
    },
    {
      "epoch": 0.6374171438878117,
      "grad_norm": 1.8544909018722506,
      "learning_rate": 1.2275546868702017e-06,
      "loss": 0.6867,
      "step": 7068
    },
    {
      "epoch": 0.6375073274112819,
      "grad_norm": 2.0538281866643358,
      "learning_rate": 1.2270158406873341e-06,
      "loss": 0.6873,
      "step": 7069
    },
    {
      "epoch": 0.6375975109347523,
      "grad_norm": 1.8238308675928794,
      "learning_rate": 1.2264770604597363e-06,
      "loss": 0.7897,
      "step": 7070
    },
    {
      "epoch": 0.6376876944582225,
      "grad_norm": 1.6743441735449032,
      "learning_rate": 1.2259383462333819e-06,
      "loss": 0.8519,
      "step": 7071
    },
    {
      "epoch": 0.6377778779816927,
      "grad_norm": 1.9173615964951622,
      "learning_rate": 1.2253996980542359e-06,
      "loss": 0.7008,
      "step": 7072
    },
    {
      "epoch": 0.637868061505163,
      "grad_norm": 1.8528797598507902,
      "learning_rate": 1.2248611159682578e-06,
      "loss": 0.733,
      "step": 7073
    },
    {
      "epoch": 0.6379582450286333,
      "grad_norm": 0.7354733390332661,
      "learning_rate": 1.2243226000214044e-06,
      "loss": 0.5647,
      "step": 7074
    },
    {
      "epoch": 0.6380484285521035,
      "grad_norm": 3.0586882244149707,
      "learning_rate": 1.2237841502596232e-06,
      "loss": 0.7498,
      "step": 7075
    },
    {
      "epoch": 0.6381386120755738,
      "grad_norm": 1.5375116735495928,
      "learning_rate": 1.2232457667288583e-06,
      "loss": 0.751,
      "step": 7076
    },
    {
      "epoch": 0.6382287955990441,
      "grad_norm": 2.5111338644862946,
      "learning_rate": 1.2227074494750476e-06,
      "loss": 0.7566,
      "step": 7077
    },
    {
      "epoch": 0.6383189791225143,
      "grad_norm": 2.5923163806775777,
      "learning_rate": 1.2221691985441238e-06,
      "loss": 0.7868,
      "step": 7078
    },
    {
      "epoch": 0.6384091626459846,
      "grad_norm": 3.6220154150008677,
      "learning_rate": 1.2216310139820128e-06,
      "loss": 0.6962,
      "step": 7079
    },
    {
      "epoch": 0.6384993461694548,
      "grad_norm": 1.9703035431551656,
      "learning_rate": 1.2210928958346347e-06,
      "loss": 0.7888,
      "step": 7080
    },
    {
      "epoch": 0.6385895296929252,
      "grad_norm": 2.4355223041876575,
      "learning_rate": 1.2205548441479065e-06,
      "loss": 0.6779,
      "step": 7081
    },
    {
      "epoch": 0.6386797132163954,
      "grad_norm": 1.650837529636206,
      "learning_rate": 1.2200168589677357e-06,
      "loss": 0.6807,
      "step": 7082
    },
    {
      "epoch": 0.6387698967398656,
      "grad_norm": 1.922734262309025,
      "learning_rate": 1.2194789403400284e-06,
      "loss": 0.7164,
      "step": 7083
    },
    {
      "epoch": 0.6388600802633358,
      "grad_norm": 1.9321058926044443,
      "learning_rate": 1.2189410883106816e-06,
      "loss": 0.7267,
      "step": 7084
    },
    {
      "epoch": 0.6389502637868062,
      "grad_norm": 2.347397775750083,
      "learning_rate": 1.2184033029255872e-06,
      "loss": 0.8162,
      "step": 7085
    },
    {
      "epoch": 0.6390404473102764,
      "grad_norm": 0.6458739657592171,
      "learning_rate": 1.2178655842306334e-06,
      "loss": 0.5815,
      "step": 7086
    },
    {
      "epoch": 0.6391306308337467,
      "grad_norm": 1.8230931732465054,
      "learning_rate": 1.2173279322716999e-06,
      "loss": 0.7443,
      "step": 7087
    },
    {
      "epoch": 0.6392208143572169,
      "grad_norm": 1.7957907681240703,
      "learning_rate": 1.216790347094663e-06,
      "loss": 0.7456,
      "step": 7088
    },
    {
      "epoch": 0.6393109978806872,
      "grad_norm": 1.6486511092068314,
      "learning_rate": 1.2162528287453927e-06,
      "loss": 0.7959,
      "step": 7089
    },
    {
      "epoch": 0.6394011814041575,
      "grad_norm": 2.2059297496715167,
      "learning_rate": 1.215715377269752e-06,
      "loss": 0.7542,
      "step": 7090
    },
    {
      "epoch": 0.6394913649276277,
      "grad_norm": 1.4748225380129576,
      "learning_rate": 1.2151779927136003e-06,
      "loss": 0.7286,
      "step": 7091
    },
    {
      "epoch": 0.6395815484510979,
      "grad_norm": 3.1014391973089674,
      "learning_rate": 1.2146406751227893e-06,
      "loss": 0.824,
      "step": 7092
    },
    {
      "epoch": 0.6396717319745683,
      "grad_norm": 1.6050839299393842,
      "learning_rate": 1.214103424543167e-06,
      "loss": 0.78,
      "step": 7093
    },
    {
      "epoch": 0.6397619154980385,
      "grad_norm": 1.7003510309083703,
      "learning_rate": 1.2135662410205735e-06,
      "loss": 0.6666,
      "step": 7094
    },
    {
      "epoch": 0.6398520990215087,
      "grad_norm": 2.4817784013395943,
      "learning_rate": 1.2130291246008444e-06,
      "loss": 0.6659,
      "step": 7095
    },
    {
      "epoch": 0.639942282544979,
      "grad_norm": 2.021100790146465,
      "learning_rate": 1.21249207532981e-06,
      "loss": 0.6703,
      "step": 7096
    },
    {
      "epoch": 0.6400324660684493,
      "grad_norm": 15.691788896411119,
      "learning_rate": 1.2119550932532936e-06,
      "loss": 0.6697,
      "step": 7097
    },
    {
      "epoch": 0.6401226495919196,
      "grad_norm": 2.5113811829682473,
      "learning_rate": 1.2114181784171144e-06,
      "loss": 0.7262,
      "step": 7098
    },
    {
      "epoch": 0.6402128331153898,
      "grad_norm": 1.3931419654247863,
      "learning_rate": 1.2108813308670837e-06,
      "loss": 0.6489,
      "step": 7099
    },
    {
      "epoch": 0.6403030166388601,
      "grad_norm": 2.787104156852815,
      "learning_rate": 1.2103445506490099e-06,
      "loss": 0.7481,
      "step": 7100
    },
    {
      "epoch": 0.6403932001623304,
      "grad_norm": 0.710951333382972,
      "learning_rate": 1.209807837808693e-06,
      "loss": 0.5925,
      "step": 7101
    },
    {
      "epoch": 0.6404833836858006,
      "grad_norm": 2.428303707817435,
      "learning_rate": 1.2092711923919282e-06,
      "loss": 0.79,
      "step": 7102
    },
    {
      "epoch": 0.6405735672092708,
      "grad_norm": 1.8919520489138735,
      "learning_rate": 1.2087346144445053e-06,
      "loss": 0.7099,
      "step": 7103
    },
    {
      "epoch": 0.6406637507327412,
      "grad_norm": 2.0013768701885906,
      "learning_rate": 1.2081981040122081e-06,
      "loss": 0.7567,
      "step": 7104
    },
    {
      "epoch": 0.6407539342562114,
      "grad_norm": 6.648650079683851,
      "learning_rate": 1.2076616611408148e-06,
      "loss": 0.7013,
      "step": 7105
    },
    {
      "epoch": 0.6408441177796816,
      "grad_norm": 1.7747526291955242,
      "learning_rate": 1.2071252858760972e-06,
      "loss": 0.7787,
      "step": 7106
    },
    {
      "epoch": 0.6409343013031519,
      "grad_norm": 2.063238185215038,
      "learning_rate": 1.2065889782638218e-06,
      "loss": 0.7341,
      "step": 7107
    },
    {
      "epoch": 0.6410244848266222,
      "grad_norm": 1.57913221189141,
      "learning_rate": 1.2060527383497506e-06,
      "loss": 0.7295,
      "step": 7108
    },
    {
      "epoch": 0.6411146683500925,
      "grad_norm": 3.187306716145028,
      "learning_rate": 1.2055165661796363e-06,
      "loss": 0.8616,
      "step": 7109
    },
    {
      "epoch": 0.6412048518735627,
      "grad_norm": 1.9440685415978887,
      "learning_rate": 1.2049804617992303e-06,
      "loss": 0.8198,
      "step": 7110
    },
    {
      "epoch": 0.6412950353970329,
      "grad_norm": 0.6290057322797372,
      "learning_rate": 1.204444425254275e-06,
      "loss": 0.5927,
      "step": 7111
    },
    {
      "epoch": 0.6413852189205033,
      "grad_norm": 1.7793623308186974,
      "learning_rate": 1.203908456590507e-06,
      "loss": 0.7351,
      "step": 7112
    },
    {
      "epoch": 0.6414754024439735,
      "grad_norm": 1.7960453274153447,
      "learning_rate": 1.20337255585366e-06,
      "loss": 0.7245,
      "step": 7113
    },
    {
      "epoch": 0.6415655859674437,
      "grad_norm": 0.6159831984941037,
      "learning_rate": 1.2028367230894582e-06,
      "loss": 0.5384,
      "step": 7114
    },
    {
      "epoch": 0.641655769490914,
      "grad_norm": 1.7785622915491754,
      "learning_rate": 1.2023009583436237e-06,
      "loss": 0.6681,
      "step": 7115
    },
    {
      "epoch": 0.6417459530143843,
      "grad_norm": 2.0093247474093774,
      "learning_rate": 1.2017652616618698e-06,
      "loss": 0.811,
      "step": 7116
    },
    {
      "epoch": 0.6418361365378545,
      "grad_norm": 2.062060338178142,
      "learning_rate": 1.2012296330899048e-06,
      "loss": 0.7409,
      "step": 7117
    },
    {
      "epoch": 0.6419263200613248,
      "grad_norm": 1.934718257787105,
      "learning_rate": 1.200694072673432e-06,
      "loss": 0.7849,
      "step": 7118
    },
    {
      "epoch": 0.642016503584795,
      "grad_norm": 2.4574768923723758,
      "learning_rate": 1.200158580458148e-06,
      "loss": 0.7758,
      "step": 7119
    },
    {
      "epoch": 0.6421066871082654,
      "grad_norm": 1.5752743194499663,
      "learning_rate": 1.1996231564897448e-06,
      "loss": 0.7233,
      "step": 7120
    },
    {
      "epoch": 0.6421968706317356,
      "grad_norm": 1.7694033140708334,
      "learning_rate": 1.1990878008139062e-06,
      "loss": 0.7069,
      "step": 7121
    },
    {
      "epoch": 0.6422870541552058,
      "grad_norm": 1.5442620760158479,
      "learning_rate": 1.1985525134763132e-06,
      "loss": 0.7793,
      "step": 7122
    },
    {
      "epoch": 0.6423772376786762,
      "grad_norm": 3.593301496299823,
      "learning_rate": 1.1980172945226393e-06,
      "loss": 0.7912,
      "step": 7123
    },
    {
      "epoch": 0.6424674212021464,
      "grad_norm": 2.152296117681309,
      "learning_rate": 1.197482143998551e-06,
      "loss": 0.755,
      "step": 7124
    },
    {
      "epoch": 0.6425576047256166,
      "grad_norm": 1.6514866547144949,
      "learning_rate": 1.196947061949712e-06,
      "loss": 0.7344,
      "step": 7125
    },
    {
      "epoch": 0.6426477882490869,
      "grad_norm": 1.588256268321235,
      "learning_rate": 1.1964120484217768e-06,
      "loss": 0.8046,
      "step": 7126
    },
    {
      "epoch": 0.6427379717725572,
      "grad_norm": 0.7029326493692918,
      "learning_rate": 1.1958771034603975e-06,
      "loss": 0.5627,
      "step": 7127
    },
    {
      "epoch": 0.6428281552960274,
      "grad_norm": 1.69025041587135,
      "learning_rate": 1.1953422271112175e-06,
      "loss": 0.6912,
      "step": 7128
    },
    {
      "epoch": 0.6429183388194977,
      "grad_norm": 2.059055110184607,
      "learning_rate": 1.1948074194198748e-06,
      "loss": 0.734,
      "step": 7129
    },
    {
      "epoch": 0.6430085223429679,
      "grad_norm": 1.7579125269934295,
      "learning_rate": 1.1942726804320033e-06,
      "loss": 0.6641,
      "step": 7130
    },
    {
      "epoch": 0.6430987058664382,
      "grad_norm": 2.2996213933061393,
      "learning_rate": 1.1937380101932295e-06,
      "loss": 0.6889,
      "step": 7131
    },
    {
      "epoch": 0.6431888893899085,
      "grad_norm": 1.5391958730777129,
      "learning_rate": 1.1932034087491745e-06,
      "loss": 0.7422,
      "step": 7132
    },
    {
      "epoch": 0.6432790729133787,
      "grad_norm": 1.8936711512098823,
      "learning_rate": 1.1926688761454531e-06,
      "loss": 0.6664,
      "step": 7133
    },
    {
      "epoch": 0.643369256436849,
      "grad_norm": 1.2929894944108635,
      "learning_rate": 1.1921344124276746e-06,
      "loss": 0.7666,
      "step": 7134
    },
    {
      "epoch": 0.6434594399603193,
      "grad_norm": 1.6635690593275585,
      "learning_rate": 1.1916000176414431e-06,
      "loss": 0.7698,
      "step": 7135
    },
    {
      "epoch": 0.6435496234837895,
      "grad_norm": 1.756430391977659,
      "learning_rate": 1.1910656918323546e-06,
      "loss": 0.7818,
      "step": 7136
    },
    {
      "epoch": 0.6436398070072598,
      "grad_norm": 2.520954623076125,
      "learning_rate": 1.1905314350460024e-06,
      "loss": 0.7777,
      "step": 7137
    },
    {
      "epoch": 0.64372999053073,
      "grad_norm": 1.5887493143747984,
      "learning_rate": 1.1899972473279717e-06,
      "loss": 0.803,
      "step": 7138
    },
    {
      "epoch": 0.6438201740542003,
      "grad_norm": 1.471678715473302,
      "learning_rate": 1.1894631287238414e-06,
      "loss": 0.7424,
      "step": 7139
    },
    {
      "epoch": 0.6439103575776706,
      "grad_norm": 1.5140861485763182,
      "learning_rate": 1.188929079279187e-06,
      "loss": 0.7576,
      "step": 7140
    },
    {
      "epoch": 0.6440005411011408,
      "grad_norm": 2.1023762201802034,
      "learning_rate": 1.1883950990395751e-06,
      "loss": 0.687,
      "step": 7141
    },
    {
      "epoch": 0.644090724624611,
      "grad_norm": 2.307828153377643,
      "learning_rate": 1.187861188050569e-06,
      "loss": 0.6924,
      "step": 7142
    },
    {
      "epoch": 0.6441809081480814,
      "grad_norm": 1.6136002860145144,
      "learning_rate": 1.187327346357724e-06,
      "loss": 0.7372,
      "step": 7143
    },
    {
      "epoch": 0.6442710916715516,
      "grad_norm": 1.744385090234618,
      "learning_rate": 1.1867935740065912e-06,
      "loss": 0.774,
      "step": 7144
    },
    {
      "epoch": 0.6443612751950218,
      "grad_norm": 1.998565338078352,
      "learning_rate": 1.1862598710427148e-06,
      "loss": 0.6413,
      "step": 7145
    },
    {
      "epoch": 0.6444514587184921,
      "grad_norm": 2.207082022023857,
      "learning_rate": 1.1857262375116328e-06,
      "loss": 0.7152,
      "step": 7146
    },
    {
      "epoch": 0.6445416422419624,
      "grad_norm": 1.6384177482617839,
      "learning_rate": 1.1851926734588783e-06,
      "loss": 0.7527,
      "step": 7147
    },
    {
      "epoch": 0.6446318257654327,
      "grad_norm": 1.9319919309151412,
      "learning_rate": 1.184659178929977e-06,
      "loss": 0.7712,
      "step": 7148
    },
    {
      "epoch": 0.6447220092889029,
      "grad_norm": 1.816647552541334,
      "learning_rate": 1.1841257539704513e-06,
      "loss": 0.7439,
      "step": 7149
    },
    {
      "epoch": 0.6448121928123732,
      "grad_norm": 1.5262820561172301,
      "learning_rate": 1.1835923986258146e-06,
      "loss": 0.7695,
      "step": 7150
    },
    {
      "epoch": 0.6449023763358435,
      "grad_norm": 1.5546088270668257,
      "learning_rate": 1.1830591129415754e-06,
      "loss": 0.7081,
      "step": 7151
    },
    {
      "epoch": 0.6449925598593137,
      "grad_norm": 1.7108605772519636,
      "learning_rate": 1.182525896963238e-06,
      "loss": 0.6948,
      "step": 7152
    },
    {
      "epoch": 0.6450827433827839,
      "grad_norm": 3.5131664580562814,
      "learning_rate": 1.181992750736298e-06,
      "loss": 0.6842,
      "step": 7153
    },
    {
      "epoch": 0.6451729269062543,
      "grad_norm": 2.1002321251234672,
      "learning_rate": 1.1814596743062474e-06,
      "loss": 0.7433,
      "step": 7154
    },
    {
      "epoch": 0.6452631104297245,
      "grad_norm": 4.413315755419189,
      "learning_rate": 1.1809266677185711e-06,
      "loss": 0.7753,
      "step": 7155
    },
    {
      "epoch": 0.6453532939531947,
      "grad_norm": 0.6538390422451557,
      "learning_rate": 1.180393731018747e-06,
      "loss": 0.6128,
      "step": 7156
    },
    {
      "epoch": 0.645443477476665,
      "grad_norm": 0.5638548640565091,
      "learning_rate": 1.1798608642522498e-06,
      "loss": 0.5514,
      "step": 7157
    },
    {
      "epoch": 0.6455336610001353,
      "grad_norm": 1.994467713135571,
      "learning_rate": 1.1793280674645454e-06,
      "loss": 0.7305,
      "step": 7158
    },
    {
      "epoch": 0.6456238445236056,
      "grad_norm": 1.6851617366469775,
      "learning_rate": 1.1787953407010954e-06,
      "loss": 0.8235,
      "step": 7159
    },
    {
      "epoch": 0.6457140280470758,
      "grad_norm": 2.055237091012776,
      "learning_rate": 1.1782626840073554e-06,
      "loss": 0.7509,
      "step": 7160
    },
    {
      "epoch": 0.645804211570546,
      "grad_norm": 2.7725546707803295,
      "learning_rate": 1.1777300974287738e-06,
      "loss": 0.6548,
      "step": 7161
    },
    {
      "epoch": 0.6458943950940164,
      "grad_norm": 1.7004307425577778,
      "learning_rate": 1.1771975810107947e-06,
      "loss": 0.781,
      "step": 7162
    },
    {
      "epoch": 0.6459845786174866,
      "grad_norm": 2.3611797983488287,
      "learning_rate": 1.1766651347988542e-06,
      "loss": 0.8166,
      "step": 7163
    },
    {
      "epoch": 0.6460747621409568,
      "grad_norm": 1.9286521574310396,
      "learning_rate": 1.1761327588383848e-06,
      "loss": 0.8599,
      "step": 7164
    },
    {
      "epoch": 0.6461649456644271,
      "grad_norm": 2.045084046033457,
      "learning_rate": 1.1756004531748105e-06,
      "loss": 0.7658,
      "step": 7165
    },
    {
      "epoch": 0.6462551291878974,
      "grad_norm": 1.6442389511611823,
      "learning_rate": 1.1750682178535521e-06,
      "loss": 0.7861,
      "step": 7166
    },
    {
      "epoch": 0.6463453127113676,
      "grad_norm": 1.733237297197532,
      "learning_rate": 1.1745360529200218e-06,
      "loss": 0.6987,
      "step": 7167
    },
    {
      "epoch": 0.6464354962348379,
      "grad_norm": 1.567242634526998,
      "learning_rate": 1.1740039584196261e-06,
      "loss": 0.7751,
      "step": 7168
    },
    {
      "epoch": 0.6465256797583081,
      "grad_norm": 8.174875091625063,
      "learning_rate": 1.1734719343977683e-06,
      "loss": 0.7819,
      "step": 7169
    },
    {
      "epoch": 0.6466158632817784,
      "grad_norm": 1.9293892201306473,
      "learning_rate": 1.1729399808998416e-06,
      "loss": 0.6913,
      "step": 7170
    },
    {
      "epoch": 0.6467060468052487,
      "grad_norm": 1.834073683054153,
      "learning_rate": 1.1724080979712368e-06,
      "loss": 0.6377,
      "step": 7171
    },
    {
      "epoch": 0.6467962303287189,
      "grad_norm": 1.528820134048986,
      "learning_rate": 1.1718762856573365e-06,
      "loss": 0.8173,
      "step": 7172
    },
    {
      "epoch": 0.6468864138521893,
      "grad_norm": 1.3897503448886652,
      "learning_rate": 1.1713445440035172e-06,
      "loss": 0.7362,
      "step": 7173
    },
    {
      "epoch": 0.6469765973756595,
      "grad_norm": 1.7479408956613327,
      "learning_rate": 1.1708128730551506e-06,
      "loss": 0.7863,
      "step": 7174
    },
    {
      "epoch": 0.6470667808991297,
      "grad_norm": 1.6230088285721957,
      "learning_rate": 1.1702812728576019e-06,
      "loss": 0.7405,
      "step": 7175
    },
    {
      "epoch": 0.6471569644226,
      "grad_norm": 1.663546415006501,
      "learning_rate": 1.1697497434562303e-06,
      "loss": 0.7757,
      "step": 7176
    },
    {
      "epoch": 0.6472471479460703,
      "grad_norm": 2.4514872921749093,
      "learning_rate": 1.1692182848963885e-06,
      "loss": 0.7856,
      "step": 7177
    },
    {
      "epoch": 0.6473373314695405,
      "grad_norm": 2.174181913666237,
      "learning_rate": 1.1686868972234227e-06,
      "loss": 0.714,
      "step": 7178
    },
    {
      "epoch": 0.6474275149930108,
      "grad_norm": 2.190783733543086,
      "learning_rate": 1.1681555804826755e-06,
      "loss": 0.7533,
      "step": 7179
    },
    {
      "epoch": 0.647517698516481,
      "grad_norm": 1.5495811450214874,
      "learning_rate": 1.1676243347194806e-06,
      "loss": 0.6894,
      "step": 7180
    },
    {
      "epoch": 0.6476078820399513,
      "grad_norm": 1.3605435928011613,
      "learning_rate": 1.167093159979167e-06,
      "loss": 0.7712,
      "step": 7181
    },
    {
      "epoch": 0.6476980655634216,
      "grad_norm": 1.5162269733941625,
      "learning_rate": 1.1665620563070575e-06,
      "loss": 0.7773,
      "step": 7182
    },
    {
      "epoch": 0.6477882490868918,
      "grad_norm": 1.8236970666076024,
      "learning_rate": 1.1660310237484691e-06,
      "loss": 0.8652,
      "step": 7183
    },
    {
      "epoch": 0.647878432610362,
      "grad_norm": 1.4612176799871006,
      "learning_rate": 1.165500062348713e-06,
      "loss": 0.691,
      "step": 7184
    },
    {
      "epoch": 0.6479686161338324,
      "grad_norm": 1.694761406414015,
      "learning_rate": 1.164969172153091e-06,
      "loss": 0.7957,
      "step": 7185
    },
    {
      "epoch": 0.6480587996573026,
      "grad_norm": 2.044225477618535,
      "learning_rate": 1.1644383532069055e-06,
      "loss": 0.7663,
      "step": 7186
    },
    {
      "epoch": 0.6481489831807729,
      "grad_norm": 2.363442659777192,
      "learning_rate": 1.1639076055554454e-06,
      "loss": 0.7887,
      "step": 7187
    },
    {
      "epoch": 0.6482391667042431,
      "grad_norm": 2.218261987449769,
      "learning_rate": 1.163376929244e-06,
      "loss": 0.737,
      "step": 7188
    },
    {
      "epoch": 0.6483293502277134,
      "grad_norm": 2.8507580602261453,
      "learning_rate": 1.1628463243178472e-06,
      "loss": 0.7097,
      "step": 7189
    },
    {
      "epoch": 0.6484195337511837,
      "grad_norm": 2.022019408550746,
      "learning_rate": 1.1623157908222623e-06,
      "loss": 0.8115,
      "step": 7190
    },
    {
      "epoch": 0.6485097172746539,
      "grad_norm": 1.8667664631778809,
      "learning_rate": 1.1617853288025129e-06,
      "loss": 0.7884,
      "step": 7191
    },
    {
      "epoch": 0.6485999007981241,
      "grad_norm": 1.9213095973046792,
      "learning_rate": 1.1612549383038612e-06,
      "loss": 0.7252,
      "step": 7192
    },
    {
      "epoch": 0.6486900843215945,
      "grad_norm": 1.3766610776004224,
      "learning_rate": 1.1607246193715629e-06,
      "loss": 0.7291,
      "step": 7193
    },
    {
      "epoch": 0.6487802678450647,
      "grad_norm": 1.8478177989959217,
      "learning_rate": 1.1601943720508684e-06,
      "loss": 0.7678,
      "step": 7194
    },
    {
      "epoch": 0.6488704513685349,
      "grad_norm": 1.551978967558523,
      "learning_rate": 1.1596641963870202e-06,
      "loss": 0.8003,
      "step": 7195
    },
    {
      "epoch": 0.6489606348920053,
      "grad_norm": 1.8040407020815372,
      "learning_rate": 1.1591340924252561e-06,
      "loss": 0.7616,
      "step": 7196
    },
    {
      "epoch": 0.6490508184154755,
      "grad_norm": 0.7660273837481122,
      "learning_rate": 1.158604060210808e-06,
      "loss": 0.6237,
      "step": 7197
    },
    {
      "epoch": 0.6491410019389457,
      "grad_norm": 2.0933124597484687,
      "learning_rate": 1.1580740997889008e-06,
      "loss": 0.8142,
      "step": 7198
    },
    {
      "epoch": 0.649231185462416,
      "grad_norm": 1.505101350812513,
      "learning_rate": 1.1575442112047544e-06,
      "loss": 0.7459,
      "step": 7199
    },
    {
      "epoch": 0.6493213689858863,
      "grad_norm": 2.1334490262887034,
      "learning_rate": 1.1570143945035797e-06,
      "loss": 0.6887,
      "step": 7200
    },
    {
      "epoch": 0.6494115525093566,
      "grad_norm": 2.7837843759826897,
      "learning_rate": 1.1564846497305864e-06,
      "loss": 0.7069,
      "step": 7201
    },
    {
      "epoch": 0.6495017360328268,
      "grad_norm": 1.6720985191879345,
      "learning_rate": 1.1559549769309726e-06,
      "loss": 0.7269,
      "step": 7202
    },
    {
      "epoch": 0.649591919556297,
      "grad_norm": 1.5115936779555237,
      "learning_rate": 1.1554253761499358e-06,
      "loss": 0.7419,
      "step": 7203
    },
    {
      "epoch": 0.6496821030797674,
      "grad_norm": 1.92183803119614,
      "learning_rate": 1.1548958474326617e-06,
      "loss": 0.7879,
      "step": 7204
    },
    {
      "epoch": 0.6497722866032376,
      "grad_norm": 1.7161168128315751,
      "learning_rate": 1.154366390824334e-06,
      "loss": 0.6797,
      "step": 7205
    },
    {
      "epoch": 0.6498624701267078,
      "grad_norm": 0.7288248668838676,
      "learning_rate": 1.1538370063701287e-06,
      "loss": 0.5591,
      "step": 7206
    },
    {
      "epoch": 0.6499526536501781,
      "grad_norm": 1.435018055366944,
      "learning_rate": 1.1533076941152153e-06,
      "loss": 0.7048,
      "step": 7207
    },
    {
      "epoch": 0.6500428371736484,
      "grad_norm": 2.228757826726653,
      "learning_rate": 1.1527784541047583e-06,
      "loss": 0.6751,
      "step": 7208
    },
    {
      "epoch": 0.6501330206971186,
      "grad_norm": 2.383704446918449,
      "learning_rate": 1.1522492863839152e-06,
      "loss": 0.7064,
      "step": 7209
    },
    {
      "epoch": 0.6502232042205889,
      "grad_norm": 1.842859990887643,
      "learning_rate": 1.1517201909978382e-06,
      "loss": 0.7124,
      "step": 7210
    },
    {
      "epoch": 0.6503133877440591,
      "grad_norm": 0.6004379462292783,
      "learning_rate": 1.151191167991671e-06,
      "loss": 0.5787,
      "step": 7211
    },
    {
      "epoch": 0.6504035712675295,
      "grad_norm": 2.3858514136951614,
      "learning_rate": 1.1506622174105536e-06,
      "loss": 0.7569,
      "step": 7212
    },
    {
      "epoch": 0.6504937547909997,
      "grad_norm": 2.173566530018969,
      "learning_rate": 1.1501333392996194e-06,
      "loss": 0.7801,
      "step": 7213
    },
    {
      "epoch": 0.6505839383144699,
      "grad_norm": 0.655680113804178,
      "learning_rate": 1.1496045337039943e-06,
      "loss": 0.5847,
      "step": 7214
    },
    {
      "epoch": 0.6506741218379402,
      "grad_norm": 2.0031359643978273,
      "learning_rate": 1.1490758006687995e-06,
      "loss": 0.7064,
      "step": 7215
    },
    {
      "epoch": 0.6507643053614105,
      "grad_norm": 1.5168074095402186,
      "learning_rate": 1.1485471402391502e-06,
      "loss": 0.7303,
      "step": 7216
    },
    {
      "epoch": 0.6508544888848807,
      "grad_norm": 2.154830465352581,
      "learning_rate": 1.1480185524601522e-06,
      "loss": 0.7233,
      "step": 7217
    },
    {
      "epoch": 0.650944672408351,
      "grad_norm": 1.9864866780454387,
      "learning_rate": 1.1474900373769108e-06,
      "loss": 0.7546,
      "step": 7218
    },
    {
      "epoch": 0.6510348559318213,
      "grad_norm": 0.5702648142905999,
      "learning_rate": 1.1469615950345184e-06,
      "loss": 0.525,
      "step": 7219
    },
    {
      "epoch": 0.6511250394552915,
      "grad_norm": 0.7017425297623882,
      "learning_rate": 1.1464332254780678e-06,
      "loss": 0.5629,
      "step": 7220
    },
    {
      "epoch": 0.6512152229787618,
      "grad_norm": 3.518582062584321,
      "learning_rate": 1.1459049287526404e-06,
      "loss": 0.7337,
      "step": 7221
    },
    {
      "epoch": 0.651305406502232,
      "grad_norm": 2.900656290442424,
      "learning_rate": 1.1453767049033137e-06,
      "loss": 0.709,
      "step": 7222
    },
    {
      "epoch": 0.6513955900257024,
      "grad_norm": 3.5043532682819603,
      "learning_rate": 1.1448485539751586e-06,
      "loss": 0.7886,
      "step": 7223
    },
    {
      "epoch": 0.6514857735491726,
      "grad_norm": 1.5188024974901926,
      "learning_rate": 1.1443204760132408e-06,
      "loss": 0.723,
      "step": 7224
    },
    {
      "epoch": 0.6515759570726428,
      "grad_norm": 1.4242059235196647,
      "learning_rate": 1.1437924710626185e-06,
      "loss": 0.7278,
      "step": 7225
    },
    {
      "epoch": 0.651666140596113,
      "grad_norm": 0.651444369065691,
      "learning_rate": 1.1432645391683429e-06,
      "loss": 0.547,
      "step": 7226
    },
    {
      "epoch": 0.6517563241195834,
      "grad_norm": 2.2728867535556025,
      "learning_rate": 1.1427366803754609e-06,
      "loss": 0.7847,
      "step": 7227
    },
    {
      "epoch": 0.6518465076430536,
      "grad_norm": 1.469688440379759,
      "learning_rate": 1.142208894729012e-06,
      "loss": 0.7615,
      "step": 7228
    },
    {
      "epoch": 0.6519366911665239,
      "grad_norm": 1.5037060889272316,
      "learning_rate": 1.1416811822740301e-06,
      "loss": 0.7897,
      "step": 7229
    },
    {
      "epoch": 0.6520268746899941,
      "grad_norm": 1.4681874273221476,
      "learning_rate": 1.1411535430555428e-06,
      "loss": 0.7985,
      "step": 7230
    },
    {
      "epoch": 0.6521170582134644,
      "grad_norm": 1.542009813764838,
      "learning_rate": 1.1406259771185705e-06,
      "loss": 0.8058,
      "step": 7231
    },
    {
      "epoch": 0.6522072417369347,
      "grad_norm": 1.5518185775891635,
      "learning_rate": 1.1400984845081282e-06,
      "loss": 0.7092,
      "step": 7232
    },
    {
      "epoch": 0.6522974252604049,
      "grad_norm": 6.437114222550362,
      "learning_rate": 1.139571065269226e-06,
      "loss": 0.8329,
      "step": 7233
    },
    {
      "epoch": 0.6523876087838751,
      "grad_norm": 1.5373023952873666,
      "learning_rate": 1.139043719446863e-06,
      "loss": 0.7975,
      "step": 7234
    },
    {
      "epoch": 0.6524777923073455,
      "grad_norm": 2.2454759892649663,
      "learning_rate": 1.1385164470860385e-06,
      "loss": 0.8209,
      "step": 7235
    },
    {
      "epoch": 0.6525679758308157,
      "grad_norm": 1.9757612813390748,
      "learning_rate": 1.1379892482317403e-06,
      "loss": 0.7497,
      "step": 7236
    },
    {
      "epoch": 0.652658159354286,
      "grad_norm": 2.0737204087751926,
      "learning_rate": 1.1374621229289524e-06,
      "loss": 0.7382,
      "step": 7237
    },
    {
      "epoch": 0.6527483428777562,
      "grad_norm": 3.0185175646273565,
      "learning_rate": 1.1369350712226525e-06,
      "loss": 0.7366,
      "step": 7238
    },
    {
      "epoch": 0.6528385264012265,
      "grad_norm": 1.5072860984813636,
      "learning_rate": 1.136408093157811e-06,
      "loss": 0.6995,
      "step": 7239
    },
    {
      "epoch": 0.6529287099246968,
      "grad_norm": 3.591009139718323,
      "learning_rate": 1.1358811887793935e-06,
      "loss": 0.7234,
      "step": 7240
    },
    {
      "epoch": 0.653018893448167,
      "grad_norm": 1.6827571055752923,
      "learning_rate": 1.135354358132356e-06,
      "loss": 0.789,
      "step": 7241
    },
    {
      "epoch": 0.6531090769716373,
      "grad_norm": 1.664227616262595,
      "learning_rate": 1.1348276012616542e-06,
      "loss": 0.7893,
      "step": 7242
    },
    {
      "epoch": 0.6531992604951076,
      "grad_norm": 3.7452904692229865,
      "learning_rate": 1.134300918212231e-06,
      "loss": 0.7057,
      "step": 7243
    },
    {
      "epoch": 0.6532894440185778,
      "grad_norm": 1.858166790956427,
      "learning_rate": 1.133774309029027e-06,
      "loss": 0.7591,
      "step": 7244
    },
    {
      "epoch": 0.653379627542048,
      "grad_norm": 5.089972470252153,
      "learning_rate": 1.133247773756975e-06,
      "loss": 0.7937,
      "step": 7245
    },
    {
      "epoch": 0.6534698110655184,
      "grad_norm": 0.6555317626782362,
      "learning_rate": 1.1327213124410024e-06,
      "loss": 0.6177,
      "step": 7246
    },
    {
      "epoch": 0.6535599945889886,
      "grad_norm": 1.6455080298358535,
      "learning_rate": 1.1321949251260292e-06,
      "loss": 0.7921,
      "step": 7247
    },
    {
      "epoch": 0.6536501781124588,
      "grad_norm": 6.156424118135728,
      "learning_rate": 1.1316686118569712e-06,
      "loss": 0.6367,
      "step": 7248
    },
    {
      "epoch": 0.6537403616359291,
      "grad_norm": 1.6044814886663992,
      "learning_rate": 1.1311423726787335e-06,
      "loss": 0.66,
      "step": 7249
    },
    {
      "epoch": 0.6538305451593994,
      "grad_norm": 2.788221558380568,
      "learning_rate": 1.130616207636221e-06,
      "loss": 0.7189,
      "step": 7250
    },
    {
      "epoch": 0.6539207286828697,
      "grad_norm": 2.017634846129114,
      "learning_rate": 1.1300901167743263e-06,
      "loss": 0.756,
      "step": 7251
    },
    {
      "epoch": 0.6540109122063399,
      "grad_norm": 1.4195614951674378,
      "learning_rate": 1.12956410013794e-06,
      "loss": 0.7474,
      "step": 7252
    },
    {
      "epoch": 0.6541010957298101,
      "grad_norm": 3.857583807914354,
      "learning_rate": 1.1290381577719436e-06,
      "loss": 0.7369,
      "step": 7253
    },
    {
      "epoch": 0.6541912792532805,
      "grad_norm": 1.8111529112726719,
      "learning_rate": 1.1285122897212143e-06,
      "loss": 0.6545,
      "step": 7254
    },
    {
      "epoch": 0.6542814627767507,
      "grad_norm": 0.7441724635244085,
      "learning_rate": 1.1279864960306228e-06,
      "loss": 0.6183,
      "step": 7255
    },
    {
      "epoch": 0.6543716463002209,
      "grad_norm": 1.6317227025812004,
      "learning_rate": 1.1274607767450297e-06,
      "loss": 0.8242,
      "step": 7256
    },
    {
      "epoch": 0.6544618298236912,
      "grad_norm": 3.0056193493246393,
      "learning_rate": 1.126935131909296e-06,
      "loss": 0.724,
      "step": 7257
    },
    {
      "epoch": 0.6545520133471615,
      "grad_norm": 1.4126213696754417,
      "learning_rate": 1.1264095615682693e-06,
      "loss": 0.694,
      "step": 7258
    },
    {
      "epoch": 0.6546421968706317,
      "grad_norm": 3.0015382002204825,
      "learning_rate": 1.1258840657667973e-06,
      "loss": 0.6647,
      "step": 7259
    },
    {
      "epoch": 0.654732380394102,
      "grad_norm": 1.4701260552887023,
      "learning_rate": 1.125358644549716e-06,
      "loss": 0.6798,
      "step": 7260
    },
    {
      "epoch": 0.6548225639175722,
      "grad_norm": 1.4314922344331567,
      "learning_rate": 1.1248332979618578e-06,
      "loss": 0.6108,
      "step": 7261
    },
    {
      "epoch": 0.6549127474410426,
      "grad_norm": 1.491029299882437,
      "learning_rate": 1.1243080260480482e-06,
      "loss": 0.6387,
      "step": 7262
    },
    {
      "epoch": 0.6550029309645128,
      "grad_norm": 1.9965835836231431,
      "learning_rate": 1.1237828288531063e-06,
      "loss": 0.8052,
      "step": 7263
    },
    {
      "epoch": 0.655093114487983,
      "grad_norm": 2.1752695683891483,
      "learning_rate": 1.1232577064218449e-06,
      "loss": 0.6668,
      "step": 7264
    },
    {
      "epoch": 0.6551832980114533,
      "grad_norm": 2.1892073371670855,
      "learning_rate": 1.1227326587990711e-06,
      "loss": 0.8235,
      "step": 7265
    },
    {
      "epoch": 0.6552734815349236,
      "grad_norm": 1.4263670671513369,
      "learning_rate": 1.1222076860295832e-06,
      "loss": 0.7314,
      "step": 7266
    },
    {
      "epoch": 0.6553636650583938,
      "grad_norm": 1.6729027272075583,
      "learning_rate": 1.1216827881581756e-06,
      "loss": 0.7754,
      "step": 7267
    },
    {
      "epoch": 0.6554538485818641,
      "grad_norm": 2.7310662004792463,
      "learning_rate": 1.1211579652296355e-06,
      "loss": 0.7404,
      "step": 7268
    },
    {
      "epoch": 0.6555440321053344,
      "grad_norm": 2.307233845579697,
      "learning_rate": 1.1206332172887438e-06,
      "loss": 0.8316,
      "step": 7269
    },
    {
      "epoch": 0.6556342156288046,
      "grad_norm": 1.6713488454982666,
      "learning_rate": 1.1201085443802756e-06,
      "loss": 0.7585,
      "step": 7270
    },
    {
      "epoch": 0.6557243991522749,
      "grad_norm": 2.131021047694976,
      "learning_rate": 1.1195839465489964e-06,
      "loss": 0.7152,
      "step": 7271
    },
    {
      "epoch": 0.6558145826757451,
      "grad_norm": 1.7118902471336184,
      "learning_rate": 1.1190594238396708e-06,
      "loss": 0.7928,
      "step": 7272
    },
    {
      "epoch": 0.6559047661992154,
      "grad_norm": 1.5931041381236244,
      "learning_rate": 1.1185349762970515e-06,
      "loss": 0.7217,
      "step": 7273
    },
    {
      "epoch": 0.6559949497226857,
      "grad_norm": 2.8979829087263473,
      "learning_rate": 1.1180106039658896e-06,
      "loss": 0.7518,
      "step": 7274
    },
    {
      "epoch": 0.6560851332461559,
      "grad_norm": 0.5844929244654967,
      "learning_rate": 1.117486306890925e-06,
      "loss": 0.5584,
      "step": 7275
    },
    {
      "epoch": 0.6561753167696261,
      "grad_norm": 2.501243091518426,
      "learning_rate": 1.116962085116896e-06,
      "loss": 0.7015,
      "step": 7276
    },
    {
      "epoch": 0.6562655002930965,
      "grad_norm": 0.6653814344766861,
      "learning_rate": 1.1164379386885302e-06,
      "loss": 0.5178,
      "step": 7277
    },
    {
      "epoch": 0.6563556838165667,
      "grad_norm": 2.2407803788837453,
      "learning_rate": 1.1159138676505516e-06,
      "loss": 0.7958,
      "step": 7278
    },
    {
      "epoch": 0.656445867340037,
      "grad_norm": 1.9361217272574003,
      "learning_rate": 1.1153898720476761e-06,
      "loss": 0.8045,
      "step": 7279
    },
    {
      "epoch": 0.6565360508635072,
      "grad_norm": 1.6751973306368644,
      "learning_rate": 1.114865951924615e-06,
      "loss": 0.7337,
      "step": 7280
    },
    {
      "epoch": 0.6566262343869775,
      "grad_norm": 2.035880108344241,
      "learning_rate": 1.1143421073260721e-06,
      "loss": 0.6935,
      "step": 7281
    },
    {
      "epoch": 0.6567164179104478,
      "grad_norm": 2.131308336102794,
      "learning_rate": 1.1138183382967432e-06,
      "loss": 0.6883,
      "step": 7282
    },
    {
      "epoch": 0.656806601433918,
      "grad_norm": 2.491451238379727,
      "learning_rate": 1.11329464488132e-06,
      "loss": 0.6602,
      "step": 7283
    },
    {
      "epoch": 0.6568967849573882,
      "grad_norm": 1.5044110132272233,
      "learning_rate": 1.112771027124487e-06,
      "loss": 0.7229,
      "step": 7284
    },
    {
      "epoch": 0.6569869684808586,
      "grad_norm": 1.7606069097107508,
      "learning_rate": 1.112247485070922e-06,
      "loss": 0.7587,
      "step": 7285
    },
    {
      "epoch": 0.6570771520043288,
      "grad_norm": 1.746055244174837,
      "learning_rate": 1.1117240187652968e-06,
      "loss": 0.7513,
      "step": 7286
    },
    {
      "epoch": 0.657167335527799,
      "grad_norm": 1.9556135879139736,
      "learning_rate": 1.1112006282522767e-06,
      "loss": 0.8273,
      "step": 7287
    },
    {
      "epoch": 0.6572575190512693,
      "grad_norm": 1.7560085279957147,
      "learning_rate": 1.1106773135765183e-06,
      "loss": 0.7302,
      "step": 7288
    },
    {
      "epoch": 0.6573477025747396,
      "grad_norm": 1.6512189061993523,
      "learning_rate": 1.110154074782677e-06,
      "loss": 0.726,
      "step": 7289
    },
    {
      "epoch": 0.6574378860982099,
      "grad_norm": 1.7803813087647822,
      "learning_rate": 1.1096309119153948e-06,
      "loss": 0.7438,
      "step": 7290
    },
    {
      "epoch": 0.6575280696216801,
      "grad_norm": 0.6324729950609383,
      "learning_rate": 1.1091078250193145e-06,
      "loss": 0.6199,
      "step": 7291
    },
    {
      "epoch": 0.6576182531451504,
      "grad_norm": 2.708000416957085,
      "learning_rate": 1.108584814139066e-06,
      "loss": 0.754,
      "step": 7292
    },
    {
      "epoch": 0.6577084366686207,
      "grad_norm": 1.6239731552038144,
      "learning_rate": 1.108061879319276e-06,
      "loss": 0.7462,
      "step": 7293
    },
    {
      "epoch": 0.6577986201920909,
      "grad_norm": 2.0103040146437365,
      "learning_rate": 1.1075390206045648e-06,
      "loss": 0.8049,
      "step": 7294
    },
    {
      "epoch": 0.6578888037155611,
      "grad_norm": 1.8065587370588663,
      "learning_rate": 1.1070162380395454e-06,
      "loss": 0.7915,
      "step": 7295
    },
    {
      "epoch": 0.6579789872390315,
      "grad_norm": 1.5475833696577577,
      "learning_rate": 1.1064935316688253e-06,
      "loss": 0.6909,
      "step": 7296
    },
    {
      "epoch": 0.6580691707625017,
      "grad_norm": 1.6703082435203966,
      "learning_rate": 1.105970901537002e-06,
      "loss": 0.8345,
      "step": 7297
    },
    {
      "epoch": 0.6581593542859719,
      "grad_norm": 1.7116363352335673,
      "learning_rate": 1.1054483476886727e-06,
      "loss": 0.7935,
      "step": 7298
    },
    {
      "epoch": 0.6582495378094422,
      "grad_norm": 1.413316266336198,
      "learning_rate": 1.1049258701684222e-06,
      "loss": 0.6678,
      "step": 7299
    },
    {
      "epoch": 0.6583397213329125,
      "grad_norm": 1.6524687014110333,
      "learning_rate": 1.1044034690208315e-06,
      "loss": 0.6944,
      "step": 7300
    },
    {
      "epoch": 0.6584299048563828,
      "grad_norm": 2.3873038565868225,
      "learning_rate": 1.1038811442904755e-06,
      "loss": 0.782,
      "step": 7301
    },
    {
      "epoch": 0.658520088379853,
      "grad_norm": 1.7191897436612642,
      "learning_rate": 1.103358896021921e-06,
      "loss": 0.6794,
      "step": 7302
    },
    {
      "epoch": 0.6586102719033232,
      "grad_norm": 1.7622131195343704,
      "learning_rate": 1.1028367242597298e-06,
      "loss": 0.731,
      "step": 7303
    },
    {
      "epoch": 0.6587004554267936,
      "grad_norm": 2.046658802990712,
      "learning_rate": 1.102314629048457e-06,
      "loss": 0.7862,
      "step": 7304
    },
    {
      "epoch": 0.6587906389502638,
      "grad_norm": 1.5360054598701316,
      "learning_rate": 1.1017926104326484e-06,
      "loss": 0.7965,
      "step": 7305
    },
    {
      "epoch": 0.658880822473734,
      "grad_norm": 1.891380915576502,
      "learning_rate": 1.1012706684568483e-06,
      "loss": 0.8012,
      "step": 7306
    },
    {
      "epoch": 0.6589710059972043,
      "grad_norm": 1.7089292680847414,
      "learning_rate": 1.1007488031655894e-06,
      "loss": 0.7685,
      "step": 7307
    },
    {
      "epoch": 0.6590611895206746,
      "grad_norm": 2.4278874736458196,
      "learning_rate": 1.1002270146034013e-06,
      "loss": 0.7037,
      "step": 7308
    },
    {
      "epoch": 0.6591513730441448,
      "grad_norm": 1.5422581062270817,
      "learning_rate": 1.0997053028148052e-06,
      "loss": 0.7765,
      "step": 7309
    },
    {
      "epoch": 0.6592415565676151,
      "grad_norm": 1.4273251052582463,
      "learning_rate": 1.0991836678443173e-06,
      "loss": 0.7512,
      "step": 7310
    },
    {
      "epoch": 0.6593317400910853,
      "grad_norm": 2.8191618963102174,
      "learning_rate": 1.0986621097364465e-06,
      "loss": 0.691,
      "step": 7311
    },
    {
      "epoch": 0.6594219236145556,
      "grad_norm": 3.0695161441888583,
      "learning_rate": 1.0981406285356932e-06,
      "loss": 0.6632,
      "step": 7312
    },
    {
      "epoch": 0.6595121071380259,
      "grad_norm": 3.576602928563074,
      "learning_rate": 1.0976192242865554e-06,
      "loss": 0.6753,
      "step": 7313
    },
    {
      "epoch": 0.6596022906614961,
      "grad_norm": 1.7390737719997382,
      "learning_rate": 1.0970978970335202e-06,
      "loss": 0.8145,
      "step": 7314
    },
    {
      "epoch": 0.6596924741849665,
      "grad_norm": 2.5007574841205145,
      "learning_rate": 1.0965766468210714e-06,
      "loss": 0.7688,
      "step": 7315
    },
    {
      "epoch": 0.6597826577084367,
      "grad_norm": 1.6700341616351573,
      "learning_rate": 1.0960554736936843e-06,
      "loss": 0.7592,
      "step": 7316
    },
    {
      "epoch": 0.6598728412319069,
      "grad_norm": 3.0554719585966597,
      "learning_rate": 1.0955343776958283e-06,
      "loss": 0.6604,
      "step": 7317
    },
    {
      "epoch": 0.6599630247553772,
      "grad_norm": 1.830511898715857,
      "learning_rate": 1.0950133588719665e-06,
      "loss": 0.7653,
      "step": 7318
    },
    {
      "epoch": 0.6600532082788475,
      "grad_norm": 1.77670135626407,
      "learning_rate": 1.0944924172665551e-06,
      "loss": 0.7697,
      "step": 7319
    },
    {
      "epoch": 0.6601433918023177,
      "grad_norm": 2.8272097460699137,
      "learning_rate": 1.0939715529240437e-06,
      "loss": 0.7863,
      "step": 7320
    },
    {
      "epoch": 0.660233575325788,
      "grad_norm": 1.7205822281902814,
      "learning_rate": 1.0934507658888755e-06,
      "loss": 0.7801,
      "step": 7321
    },
    {
      "epoch": 0.6603237588492582,
      "grad_norm": 1.417999519455671,
      "learning_rate": 1.092930056205486e-06,
      "loss": 0.7233,
      "step": 7322
    },
    {
      "epoch": 0.6604139423727285,
      "grad_norm": 2.044278193823002,
      "learning_rate": 1.092409423918306e-06,
      "loss": 0.7268,
      "step": 7323
    },
    {
      "epoch": 0.6605041258961988,
      "grad_norm": 1.4026458329864275,
      "learning_rate": 1.0918888690717581e-06,
      "loss": 0.6772,
      "step": 7324
    },
    {
      "epoch": 0.660594309419669,
      "grad_norm": 1.705312950857032,
      "learning_rate": 1.091368391710259e-06,
      "loss": 0.8127,
      "step": 7325
    },
    {
      "epoch": 0.6606844929431392,
      "grad_norm": 1.6308298439504452,
      "learning_rate": 1.0908479918782198e-06,
      "loss": 0.6966,
      "step": 7326
    },
    {
      "epoch": 0.6607746764666096,
      "grad_norm": 1.506008591043556,
      "learning_rate": 1.0903276696200413e-06,
      "loss": 0.7728,
      "step": 7327
    },
    {
      "epoch": 0.6608648599900798,
      "grad_norm": 2.771437146548571,
      "learning_rate": 1.0898074249801234e-06,
      "loss": 0.755,
      "step": 7328
    },
    {
      "epoch": 0.6609550435135501,
      "grad_norm": 1.6772470177171763,
      "learning_rate": 1.0892872580028533e-06,
      "loss": 0.7661,
      "step": 7329
    },
    {
      "epoch": 0.6610452270370203,
      "grad_norm": 1.6086367052520756,
      "learning_rate": 1.0887671687326178e-06,
      "loss": 0.7789,
      "step": 7330
    },
    {
      "epoch": 0.6611354105604906,
      "grad_norm": 3.975113340477365,
      "learning_rate": 1.0882471572137908e-06,
      "loss": 0.7138,
      "step": 7331
    },
    {
      "epoch": 0.6612255940839609,
      "grad_norm": 1.6135367498288025,
      "learning_rate": 1.087727223490744e-06,
      "loss": 0.7962,
      "step": 7332
    },
    {
      "epoch": 0.6613157776074311,
      "grad_norm": 1.7281871073980617,
      "learning_rate": 1.0872073676078405e-06,
      "loss": 0.8068,
      "step": 7333
    },
    {
      "epoch": 0.6614059611309013,
      "grad_norm": 2.369911162888371,
      "learning_rate": 1.0866875896094375e-06,
      "loss": 0.698,
      "step": 7334
    },
    {
      "epoch": 0.6614961446543717,
      "grad_norm": 1.5626677613797004,
      "learning_rate": 1.0861678895398854e-06,
      "loss": 0.7698,
      "step": 7335
    },
    {
      "epoch": 0.6615863281778419,
      "grad_norm": 2.146613550160841,
      "learning_rate": 1.0856482674435286e-06,
      "loss": 0.7147,
      "step": 7336
    },
    {
      "epoch": 0.6616765117013121,
      "grad_norm": 0.7533045527001712,
      "learning_rate": 1.0851287233647024e-06,
      "loss": 0.5748,
      "step": 7337
    },
    {
      "epoch": 0.6617666952247825,
      "grad_norm": 1.6843740686191113,
      "learning_rate": 1.084609257347738e-06,
      "loss": 0.8061,
      "step": 7338
    },
    {
      "epoch": 0.6618568787482527,
      "grad_norm": 8.447604987764098,
      "learning_rate": 1.0840898694369594e-06,
      "loss": 0.7245,
      "step": 7339
    },
    {
      "epoch": 0.661947062271723,
      "grad_norm": 1.4779257964734196,
      "learning_rate": 1.083570559676683e-06,
      "loss": 0.7401,
      "step": 7340
    },
    {
      "epoch": 0.6620372457951932,
      "grad_norm": 1.6385503094108331,
      "learning_rate": 1.08305132811122e-06,
      "loss": 0.7276,
      "step": 7341
    },
    {
      "epoch": 0.6621274293186635,
      "grad_norm": 1.9680926724223207,
      "learning_rate": 1.0825321747848735e-06,
      "loss": 0.7515,
      "step": 7342
    },
    {
      "epoch": 0.6622176128421338,
      "grad_norm": 0.7312517916013501,
      "learning_rate": 1.0820130997419417e-06,
      "loss": 0.6447,
      "step": 7343
    },
    {
      "epoch": 0.662307796365604,
      "grad_norm": 1.809611287607797,
      "learning_rate": 1.0814941030267123e-06,
      "loss": 0.7431,
      "step": 7344
    },
    {
      "epoch": 0.6623979798890742,
      "grad_norm": 0.6044983254954556,
      "learning_rate": 1.080975184683472e-06,
      "loss": 0.5631,
      "step": 7345
    },
    {
      "epoch": 0.6624881634125446,
      "grad_norm": 1.8090990979500303,
      "learning_rate": 1.0804563447564948e-06,
      "loss": 0.8084,
      "step": 7346
    },
    {
      "epoch": 0.6625783469360148,
      "grad_norm": 1.7104149982972396,
      "learning_rate": 1.0799375832900545e-06,
      "loss": 0.6938,
      "step": 7347
    },
    {
      "epoch": 0.662668530459485,
      "grad_norm": 2.010150186801607,
      "learning_rate": 1.0794189003284118e-06,
      "loss": 0.783,
      "step": 7348
    },
    {
      "epoch": 0.6627587139829553,
      "grad_norm": 2.1040282125213277,
      "learning_rate": 1.0789002959158242e-06,
      "loss": 0.768,
      "step": 7349
    },
    {
      "epoch": 0.6628488975064256,
      "grad_norm": 1.9926419207335087,
      "learning_rate": 1.0783817700965428e-06,
      "loss": 0.7047,
      "step": 7350
    },
    {
      "epoch": 0.6629390810298958,
      "grad_norm": 1.4730462081612288,
      "learning_rate": 1.0778633229148102e-06,
      "loss": 0.7961,
      "step": 7351
    },
    {
      "epoch": 0.6630292645533661,
      "grad_norm": 2.0348434354273204,
      "learning_rate": 1.0773449544148645e-06,
      "loss": 0.7514,
      "step": 7352
    },
    {
      "epoch": 0.6631194480768363,
      "grad_norm": 1.4918395430622768,
      "learning_rate": 1.076826664640934e-06,
      "loss": 0.7897,
      "step": 7353
    },
    {
      "epoch": 0.6632096316003067,
      "grad_norm": 0.7456170201733913,
      "learning_rate": 1.0763084536372424e-06,
      "loss": 0.604,
      "step": 7354
    },
    {
      "epoch": 0.6632998151237769,
      "grad_norm": 1.8760941174207264,
      "learning_rate": 1.0757903214480068e-06,
      "loss": 0.7585,
      "step": 7355
    },
    {
      "epoch": 0.6633899986472471,
      "grad_norm": 2.080031588754672,
      "learning_rate": 1.0752722681174376e-06,
      "loss": 0.6999,
      "step": 7356
    },
    {
      "epoch": 0.6634801821707174,
      "grad_norm": 2.2136805472095546,
      "learning_rate": 1.074754293689737e-06,
      "loss": 0.673,
      "step": 7357
    },
    {
      "epoch": 0.6635703656941877,
      "grad_norm": 1.7429176476962418,
      "learning_rate": 1.0742363982091023e-06,
      "loss": 0.7546,
      "step": 7358
    },
    {
      "epoch": 0.6636605492176579,
      "grad_norm": 1.7818788455746357,
      "learning_rate": 1.0737185817197215e-06,
      "loss": 0.7076,
      "step": 7359
    },
    {
      "epoch": 0.6637507327411282,
      "grad_norm": 1.7283797560919043,
      "learning_rate": 1.0732008442657803e-06,
      "loss": 0.7108,
      "step": 7360
    },
    {
      "epoch": 0.6638409162645985,
      "grad_norm": 1.7865921355947798,
      "learning_rate": 1.0726831858914516e-06,
      "loss": 0.7352,
      "step": 7361
    },
    {
      "epoch": 0.6639310997880687,
      "grad_norm": 1.7705842310409339,
      "learning_rate": 1.0721656066409084e-06,
      "loss": 0.7691,
      "step": 7362
    },
    {
      "epoch": 0.664021283311539,
      "grad_norm": 2.6812281503209467,
      "learning_rate": 1.0716481065583108e-06,
      "loss": 0.7567,
      "step": 7363
    },
    {
      "epoch": 0.6641114668350092,
      "grad_norm": 0.6493951780398657,
      "learning_rate": 1.071130685687816e-06,
      "loss": 0.5749,
      "step": 7364
    },
    {
      "epoch": 0.6642016503584796,
      "grad_norm": 1.9003945769124826,
      "learning_rate": 1.0706133440735723e-06,
      "loss": 0.6878,
      "step": 7365
    },
    {
      "epoch": 0.6642918338819498,
      "grad_norm": 3.1531424757262405,
      "learning_rate": 1.070096081759723e-06,
      "loss": 0.7009,
      "step": 7366
    },
    {
      "epoch": 0.66438201740542,
      "grad_norm": 1.8111870054843728,
      "learning_rate": 1.069578898790404e-06,
      "loss": 0.8463,
      "step": 7367
    },
    {
      "epoch": 0.6644722009288903,
      "grad_norm": 1.7634237966185873,
      "learning_rate": 1.0690617952097424e-06,
      "loss": 0.7333,
      "step": 7368
    },
    {
      "epoch": 0.6645623844523606,
      "grad_norm": 1.7863136729697071,
      "learning_rate": 1.068544771061863e-06,
      "loss": 0.7962,
      "step": 7369
    },
    {
      "epoch": 0.6646525679758308,
      "grad_norm": 2.3953602088710833,
      "learning_rate": 1.0680278263908787e-06,
      "loss": 0.7905,
      "step": 7370
    },
    {
      "epoch": 0.6647427514993011,
      "grad_norm": 0.7907541478639516,
      "learning_rate": 1.0675109612408991e-06,
      "loss": 0.5847,
      "step": 7371
    },
    {
      "epoch": 0.6648329350227713,
      "grad_norm": 3.4654640924026934,
      "learning_rate": 1.0669941756560264e-06,
      "loss": 0.721,
      "step": 7372
    },
    {
      "epoch": 0.6649231185462416,
      "grad_norm": 2.4786465417121204,
      "learning_rate": 1.0664774696803548e-06,
      "loss": 0.6982,
      "step": 7373
    },
    {
      "epoch": 0.6650133020697119,
      "grad_norm": 2.2542684121946857,
      "learning_rate": 1.065960843357973e-06,
      "loss": 0.7205,
      "step": 7374
    },
    {
      "epoch": 0.6651034855931821,
      "grad_norm": 1.6192548134069502,
      "learning_rate": 1.065444296732963e-06,
      "loss": 0.766,
      "step": 7375
    },
    {
      "epoch": 0.6651936691166523,
      "grad_norm": 0.6720577479382237,
      "learning_rate": 1.064927829849397e-06,
      "loss": 0.5871,
      "step": 7376
    },
    {
      "epoch": 0.6652838526401227,
      "grad_norm": 3.1714038217037865,
      "learning_rate": 1.0644114427513465e-06,
      "loss": 0.7477,
      "step": 7377
    },
    {
      "epoch": 0.6653740361635929,
      "grad_norm": 2.496178853907913,
      "learning_rate": 1.0638951354828693e-06,
      "loss": 0.6786,
      "step": 7378
    },
    {
      "epoch": 0.6654642196870632,
      "grad_norm": 1.531673450697084,
      "learning_rate": 1.063378908088021e-06,
      "loss": 0.8032,
      "step": 7379
    },
    {
      "epoch": 0.6655544032105334,
      "grad_norm": 1.750440133468767,
      "learning_rate": 1.0628627606108486e-06,
      "loss": 0.7392,
      "step": 7380
    },
    {
      "epoch": 0.6656445867340037,
      "grad_norm": 2.204091286286689,
      "learning_rate": 1.062346693095393e-06,
      "loss": 0.8148,
      "step": 7381
    },
    {
      "epoch": 0.665734770257474,
      "grad_norm": 1.54465857575764,
      "learning_rate": 1.0618307055856882e-06,
      "loss": 0.686,
      "step": 7382
    },
    {
      "epoch": 0.6658249537809442,
      "grad_norm": 1.731636979133456,
      "learning_rate": 1.061314798125759e-06,
      "loss": 0.7985,
      "step": 7383
    },
    {
      "epoch": 0.6659151373044144,
      "grad_norm": 1.4495550623972546,
      "learning_rate": 1.0607989707596293e-06,
      "loss": 0.7079,
      "step": 7384
    },
    {
      "epoch": 0.6660053208278848,
      "grad_norm": 46.359261775224056,
      "learning_rate": 1.0602832235313078e-06,
      "loss": 0.8729,
      "step": 7385
    },
    {
      "epoch": 0.666095504351355,
      "grad_norm": 1.463836160627453,
      "learning_rate": 1.0597675564848053e-06,
      "loss": 0.7361,
      "step": 7386
    },
    {
      "epoch": 0.6661856878748252,
      "grad_norm": 1.8377917077649906,
      "learning_rate": 1.059251969664118e-06,
      "loss": 0.8165,
      "step": 7387
    },
    {
      "epoch": 0.6662758713982956,
      "grad_norm": 13.113978235871006,
      "learning_rate": 1.0587364631132402e-06,
      "loss": 0.7576,
      "step": 7388
    },
    {
      "epoch": 0.6663660549217658,
      "grad_norm": 1.4633887630788742,
      "learning_rate": 1.0582210368761573e-06,
      "loss": 0.7848,
      "step": 7389
    },
    {
      "epoch": 0.666456238445236,
      "grad_norm": 8.438867528789455,
      "learning_rate": 1.0577056909968485e-06,
      "loss": 0.7509,
      "step": 7390
    },
    {
      "epoch": 0.6665464219687063,
      "grad_norm": 1.7443784613525426,
      "learning_rate": 1.0571904255192857e-06,
      "loss": 0.7665,
      "step": 7391
    },
    {
      "epoch": 0.6666366054921766,
      "grad_norm": 2.0164828581891308,
      "learning_rate": 1.0566752404874354e-06,
      "loss": 0.7821,
      "step": 7392
    },
    {
      "epoch": 0.6667267890156469,
      "grad_norm": 1.8120099424632563,
      "learning_rate": 1.0561601359452543e-06,
      "loss": 0.7463,
      "step": 7393
    },
    {
      "epoch": 0.6668169725391171,
      "grad_norm": 2.241118172780649,
      "learning_rate": 1.0556451119366947e-06,
      "loss": 0.6844,
      "step": 7394
    },
    {
      "epoch": 0.6669071560625873,
      "grad_norm": 1.6174826974547678,
      "learning_rate": 1.0551301685057011e-06,
      "loss": 0.6896,
      "step": 7395
    },
    {
      "epoch": 0.6669973395860577,
      "grad_norm": 1.965393656836619,
      "learning_rate": 1.0546153056962117e-06,
      "loss": 0.8349,
      "step": 7396
    },
    {
      "epoch": 0.6670875231095279,
      "grad_norm": 1.8943070426489006,
      "learning_rate": 1.0541005235521578e-06,
      "loss": 0.7453,
      "step": 7397
    },
    {
      "epoch": 0.6671777066329981,
      "grad_norm": 2.1615660000991004,
      "learning_rate": 1.0535858221174614e-06,
      "loss": 0.7654,
      "step": 7398
    },
    {
      "epoch": 0.6672678901564684,
      "grad_norm": 1.4935077199413922,
      "learning_rate": 1.0530712014360426e-06,
      "loss": 0.7875,
      "step": 7399
    },
    {
      "epoch": 0.6673580736799387,
      "grad_norm": 1.6693774668833488,
      "learning_rate": 1.0525566615518088e-06,
      "loss": 0.7992,
      "step": 7400
    },
    {
      "epoch": 0.667448257203409,
      "grad_norm": 1.639330253150366,
      "learning_rate": 1.0520422025086662e-06,
      "loss": 0.7641,
      "step": 7401
    },
    {
      "epoch": 0.6675384407268792,
      "grad_norm": 1.834981336576674,
      "learning_rate": 1.0515278243505092e-06,
      "loss": 0.8013,
      "step": 7402
    },
    {
      "epoch": 0.6676286242503494,
      "grad_norm": 1.8746776303684736,
      "learning_rate": 1.0510135271212278e-06,
      "loss": 0.7787,
      "step": 7403
    },
    {
      "epoch": 0.6677188077738198,
      "grad_norm": 1.3605395623272607,
      "learning_rate": 1.0504993108647052e-06,
      "loss": 0.8124,
      "step": 7404
    },
    {
      "epoch": 0.66780899129729,
      "grad_norm": 1.5094417012326664,
      "learning_rate": 1.0499851756248168e-06,
      "loss": 0.725,
      "step": 7405
    },
    {
      "epoch": 0.6678991748207602,
      "grad_norm": 1.6424339857146222,
      "learning_rate": 1.0494711214454316e-06,
      "loss": 0.7328,
      "step": 7406
    },
    {
      "epoch": 0.6679893583442305,
      "grad_norm": 1.8937583998692111,
      "learning_rate": 1.0489571483704111e-06,
      "loss": 0.6646,
      "step": 7407
    },
    {
      "epoch": 0.6680795418677008,
      "grad_norm": 1.7972970218150488,
      "learning_rate": 1.048443256443612e-06,
      "loss": 0.6678,
      "step": 7408
    },
    {
      "epoch": 0.668169725391171,
      "grad_norm": 2.111868620651468,
      "learning_rate": 1.0479294457088801e-06,
      "loss": 0.7672,
      "step": 7409
    },
    {
      "epoch": 0.6682599089146413,
      "grad_norm": 4.2317262367441755,
      "learning_rate": 1.0474157162100574e-06,
      "loss": 0.8703,
      "step": 7410
    },
    {
      "epoch": 0.6683500924381116,
      "grad_norm": 1.7422535289346046,
      "learning_rate": 1.0469020679909786e-06,
      "loss": 0.6891,
      "step": 7411
    },
    {
      "epoch": 0.6684402759615818,
      "grad_norm": 1.7366833832949706,
      "learning_rate": 1.0463885010954705e-06,
      "loss": 0.6726,
      "step": 7412
    },
    {
      "epoch": 0.6685304594850521,
      "grad_norm": 1.8249533111661793,
      "learning_rate": 1.0458750155673536e-06,
      "loss": 0.8238,
      "step": 7413
    },
    {
      "epoch": 0.6686206430085223,
      "grad_norm": 1.7870149848066361,
      "learning_rate": 1.0453616114504421e-06,
      "loss": 0.7037,
      "step": 7414
    },
    {
      "epoch": 0.6687108265319927,
      "grad_norm": 1.5373406242325647,
      "learning_rate": 1.0448482887885406e-06,
      "loss": 0.7792,
      "step": 7415
    },
    {
      "epoch": 0.6688010100554629,
      "grad_norm": 1.933656202136174,
      "learning_rate": 1.044335047625451e-06,
      "loss": 0.7578,
      "step": 7416
    },
    {
      "epoch": 0.6688911935789331,
      "grad_norm": 2.540698747866596,
      "learning_rate": 1.0438218880049637e-06,
      "loss": 0.7634,
      "step": 7417
    },
    {
      "epoch": 0.6689813771024034,
      "grad_norm": 1.9212747892668014,
      "learning_rate": 1.0433088099708653e-06,
      "loss": 0.8091,
      "step": 7418
    },
    {
      "epoch": 0.6690715606258737,
      "grad_norm": 0.5977219439535025,
      "learning_rate": 1.0427958135669346e-06,
      "loss": 0.5126,
      "step": 7419
    },
    {
      "epoch": 0.6691617441493439,
      "grad_norm": 1.7808583816220975,
      "learning_rate": 1.0422828988369428e-06,
      "loss": 0.6865,
      "step": 7420
    },
    {
      "epoch": 0.6692519276728142,
      "grad_norm": 10.123615311493106,
      "learning_rate": 1.041770065824655e-06,
      "loss": 0.8197,
      "step": 7421
    },
    {
      "epoch": 0.6693421111962844,
      "grad_norm": 1.6749335859426564,
      "learning_rate": 1.0412573145738287e-06,
      "loss": 0.6061,
      "step": 7422
    },
    {
      "epoch": 0.6694322947197547,
      "grad_norm": 1.989693909879353,
      "learning_rate": 1.040744645128216e-06,
      "loss": 0.7242,
      "step": 7423
    },
    {
      "epoch": 0.669522478243225,
      "grad_norm": 2.8137664486689418,
      "learning_rate": 1.040232057531558e-06,
      "loss": 0.8086,
      "step": 7424
    },
    {
      "epoch": 0.6696126617666952,
      "grad_norm": 2.619943835551925,
      "learning_rate": 1.0397195518275932e-06,
      "loss": 0.7268,
      "step": 7425
    },
    {
      "epoch": 0.6697028452901654,
      "grad_norm": 2.7426393142131236,
      "learning_rate": 1.0392071280600512e-06,
      "loss": 0.7192,
      "step": 7426
    },
    {
      "epoch": 0.6697930288136358,
      "grad_norm": 1.7799655231608722,
      "learning_rate": 1.0386947862726549e-06,
      "loss": 0.7479,
      "step": 7427
    },
    {
      "epoch": 0.669883212337106,
      "grad_norm": 1.9343493902085704,
      "learning_rate": 1.0381825265091197e-06,
      "loss": 0.6831,
      "step": 7428
    },
    {
      "epoch": 0.6699733958605762,
      "grad_norm": 1.6855475822438406,
      "learning_rate": 1.037670348813155e-06,
      "loss": 0.6714,
      "step": 7429
    },
    {
      "epoch": 0.6700635793840465,
      "grad_norm": 1.7524496372185847,
      "learning_rate": 1.0371582532284624e-06,
      "loss": 0.7805,
      "step": 7430
    },
    {
      "epoch": 0.6701537629075168,
      "grad_norm": 1.577911891334036,
      "learning_rate": 1.0366462397987375e-06,
      "loss": 0.6831,
      "step": 7431
    },
    {
      "epoch": 0.6702439464309871,
      "grad_norm": 2.044704655256544,
      "learning_rate": 1.0361343085676665e-06,
      "loss": 0.7569,
      "step": 7432
    },
    {
      "epoch": 0.6703341299544573,
      "grad_norm": 1.8669947440916759,
      "learning_rate": 1.0356224595789309e-06,
      "loss": 0.8649,
      "step": 7433
    },
    {
      "epoch": 0.6704243134779276,
      "grad_norm": 2.858031243764146,
      "learning_rate": 1.0351106928762046e-06,
      "loss": 0.8108,
      "step": 7434
    },
    {
      "epoch": 0.6705144970013979,
      "grad_norm": 1.8893655412432768,
      "learning_rate": 1.034599008503154e-06,
      "loss": 0.8238,
      "step": 7435
    },
    {
      "epoch": 0.6706046805248681,
      "grad_norm": 1.4495401771453909,
      "learning_rate": 1.0340874065034406e-06,
      "loss": 0.7406,
      "step": 7436
    },
    {
      "epoch": 0.6706948640483383,
      "grad_norm": 2.0896628043014864,
      "learning_rate": 1.0335758869207137e-06,
      "loss": 0.6758,
      "step": 7437
    },
    {
      "epoch": 0.6707850475718087,
      "grad_norm": 1.8890417736800131,
      "learning_rate": 1.0330644497986227e-06,
      "loss": 0.7738,
      "step": 7438
    },
    {
      "epoch": 0.6708752310952789,
      "grad_norm": 1.8898256992520412,
      "learning_rate": 1.0325530951808029e-06,
      "loss": 0.7936,
      "step": 7439
    },
    {
      "epoch": 0.6709654146187491,
      "grad_norm": 1.743146281859597,
      "learning_rate": 1.0320418231108887e-06,
      "loss": 0.7572,
      "step": 7440
    },
    {
      "epoch": 0.6710555981422194,
      "grad_norm": 2.0490722510806996,
      "learning_rate": 1.0315306336325028e-06,
      "loss": 0.7942,
      "step": 7441
    },
    {
      "epoch": 0.6711457816656897,
      "grad_norm": 2.511653157708135,
      "learning_rate": 1.0310195267892635e-06,
      "loss": 0.7076,
      "step": 7442
    },
    {
      "epoch": 0.67123596518916,
      "grad_norm": 3.2344043380217578,
      "learning_rate": 1.030508502624781e-06,
      "loss": 0.697,
      "step": 7443
    },
    {
      "epoch": 0.6713261487126302,
      "grad_norm": 1.8462756686792685,
      "learning_rate": 1.0299975611826587e-06,
      "loss": 0.6988,
      "step": 7444
    },
    {
      "epoch": 0.6714163322361004,
      "grad_norm": 2.077568266443584,
      "learning_rate": 1.0294867025064928e-06,
      "loss": 0.7569,
      "step": 7445
    },
    {
      "epoch": 0.6715065157595708,
      "grad_norm": 1.6339705161816276,
      "learning_rate": 1.028975926639874e-06,
      "loss": 0.7553,
      "step": 7446
    },
    {
      "epoch": 0.671596699283041,
      "grad_norm": 2.8589839146719638,
      "learning_rate": 1.0284652336263823e-06,
      "loss": 0.6786,
      "step": 7447
    },
    {
      "epoch": 0.6716868828065112,
      "grad_norm": 1.9983247655070286,
      "learning_rate": 1.0279546235095938e-06,
      "loss": 0.6809,
      "step": 7448
    },
    {
      "epoch": 0.6717770663299815,
      "grad_norm": 0.739149319991025,
      "learning_rate": 1.0274440963330768e-06,
      "loss": 0.6089,
      "step": 7449
    },
    {
      "epoch": 0.6718672498534518,
      "grad_norm": 1.6318299472043702,
      "learning_rate": 1.0269336521403919e-06,
      "loss": 0.7502,
      "step": 7450
    },
    {
      "epoch": 0.671957433376922,
      "grad_norm": 1.893934143983905,
      "learning_rate": 1.0264232909750936e-06,
      "loss": 0.8777,
      "step": 7451
    },
    {
      "epoch": 0.6720476169003923,
      "grad_norm": 3.971843685143753,
      "learning_rate": 1.025913012880728e-06,
      "loss": 0.8089,
      "step": 7452
    },
    {
      "epoch": 0.6721378004238625,
      "grad_norm": 1.5346214391759192,
      "learning_rate": 1.0254028179008362e-06,
      "loss": 0.6612,
      "step": 7453
    },
    {
      "epoch": 0.6722279839473329,
      "grad_norm": 2.492371077039461,
      "learning_rate": 1.0248927060789483e-06,
      "loss": 0.6418,
      "step": 7454
    },
    {
      "epoch": 0.6723181674708031,
      "grad_norm": 1.9932361909337037,
      "learning_rate": 1.0243826774585928e-06,
      "loss": 0.7069,
      "step": 7455
    },
    {
      "epoch": 0.6724083509942733,
      "grad_norm": 2.6817388662414925,
      "learning_rate": 1.0238727320832854e-06,
      "loss": 0.7372,
      "step": 7456
    },
    {
      "epoch": 0.6724985345177437,
      "grad_norm": 1.8302847339516402,
      "learning_rate": 1.0233628699965403e-06,
      "loss": 0.756,
      "step": 7457
    },
    {
      "epoch": 0.6725887180412139,
      "grad_norm": 1.5328266823644268,
      "learning_rate": 1.0228530912418594e-06,
      "loss": 0.7729,
      "step": 7458
    },
    {
      "epoch": 0.6726789015646841,
      "grad_norm": 1.820015813528256,
      "learning_rate": 1.0223433958627404e-06,
      "loss": 0.7231,
      "step": 7459
    },
    {
      "epoch": 0.6727690850881544,
      "grad_norm": 2.13954781487639,
      "learning_rate": 1.021833783902674e-06,
      "loss": 0.7686,
      "step": 7460
    },
    {
      "epoch": 0.6728592686116247,
      "grad_norm": 1.9698926243831467,
      "learning_rate": 1.0213242554051427e-06,
      "loss": 0.7597,
      "step": 7461
    },
    {
      "epoch": 0.6729494521350949,
      "grad_norm": 1.568731166528974,
      "learning_rate": 1.0208148104136229e-06,
      "loss": 0.6892,
      "step": 7462
    },
    {
      "epoch": 0.6730396356585652,
      "grad_norm": 1.8564067196434035,
      "learning_rate": 1.020305448971582e-06,
      "loss": 0.7248,
      "step": 7463
    },
    {
      "epoch": 0.6731298191820354,
      "grad_norm": 2.24216102292049,
      "learning_rate": 1.0197961711224824e-06,
      "loss": 0.715,
      "step": 7464
    },
    {
      "epoch": 0.6732200027055057,
      "grad_norm": 2.0274061954120266,
      "learning_rate": 1.0192869769097777e-06,
      "loss": 0.7592,
      "step": 7465
    },
    {
      "epoch": 0.673310186228976,
      "grad_norm": 1.5409051543697243,
      "learning_rate": 1.018777866376916e-06,
      "loss": 0.7621,
      "step": 7466
    },
    {
      "epoch": 0.6734003697524462,
      "grad_norm": 1.7551507761165324,
      "learning_rate": 1.0182688395673374e-06,
      "loss": 0.6211,
      "step": 7467
    },
    {
      "epoch": 0.6734905532759164,
      "grad_norm": 2.2472481323601086,
      "learning_rate": 1.017759896524475e-06,
      "loss": 0.7169,
      "step": 7468
    },
    {
      "epoch": 0.6735807367993868,
      "grad_norm": 3.967618405474801,
      "learning_rate": 1.0172510372917528e-06,
      "loss": 0.7889,
      "step": 7469
    },
    {
      "epoch": 0.673670920322857,
      "grad_norm": 3.519029791161414,
      "learning_rate": 1.0167422619125925e-06,
      "loss": 0.6539,
      "step": 7470
    },
    {
      "epoch": 0.6737611038463273,
      "grad_norm": 1.7738616704971557,
      "learning_rate": 1.0162335704304026e-06,
      "loss": 0.75,
      "step": 7471
    },
    {
      "epoch": 0.6738512873697975,
      "grad_norm": 0.7258981928732334,
      "learning_rate": 1.0157249628885903e-06,
      "loss": 0.5934,
      "step": 7472
    },
    {
      "epoch": 0.6739414708932678,
      "grad_norm": 1.3019590343910061,
      "learning_rate": 1.0152164393305506e-06,
      "loss": 0.6562,
      "step": 7473
    },
    {
      "epoch": 0.6740316544167381,
      "grad_norm": 1.811185689117454,
      "learning_rate": 1.0147079997996746e-06,
      "loss": 0.7418,
      "step": 7474
    },
    {
      "epoch": 0.6741218379402083,
      "grad_norm": 1.6767990471596959,
      "learning_rate": 1.0141996443393446e-06,
      "loss": 0.7821,
      "step": 7475
    },
    {
      "epoch": 0.6742120214636785,
      "grad_norm": 1.9951306313217207,
      "learning_rate": 1.0136913729929369e-06,
      "loss": 0.7377,
      "step": 7476
    },
    {
      "epoch": 0.6743022049871489,
      "grad_norm": 1.5214581273058887,
      "learning_rate": 1.0131831858038203e-06,
      "loss": 0.7099,
      "step": 7477
    },
    {
      "epoch": 0.6743923885106191,
      "grad_norm": 1.463976712802513,
      "learning_rate": 1.0126750828153538e-06,
      "loss": 0.8133,
      "step": 7478
    },
    {
      "epoch": 0.6744825720340893,
      "grad_norm": 1.7590222074441637,
      "learning_rate": 1.012167064070895e-06,
      "loss": 0.7943,
      "step": 7479
    },
    {
      "epoch": 0.6745727555575597,
      "grad_norm": 3.70009273206647,
      "learning_rate": 1.0116591296137885e-06,
      "loss": 0.7716,
      "step": 7480
    },
    {
      "epoch": 0.6746629390810299,
      "grad_norm": 1.7810300724718469,
      "learning_rate": 1.0111512794873746e-06,
      "loss": 0.7742,
      "step": 7481
    },
    {
      "epoch": 0.6747531226045002,
      "grad_norm": 1.5100291664894032,
      "learning_rate": 1.010643513734986e-06,
      "loss": 0.7486,
      "step": 7482
    },
    {
      "epoch": 0.6748433061279704,
      "grad_norm": 1.8052561470944544,
      "learning_rate": 1.010135832399948e-06,
      "loss": 0.7894,
      "step": 7483
    },
    {
      "epoch": 0.6749334896514407,
      "grad_norm": 1.3543621460111446,
      "learning_rate": 1.0096282355255792e-06,
      "loss": 0.7562,
      "step": 7484
    },
    {
      "epoch": 0.675023673174911,
      "grad_norm": 9.626571725111226,
      "learning_rate": 1.0091207231551905e-06,
      "loss": 0.8054,
      "step": 7485
    },
    {
      "epoch": 0.6751138566983812,
      "grad_norm": 1.5508190698119753,
      "learning_rate": 1.0086132953320842e-06,
      "loss": 0.7157,
      "step": 7486
    },
    {
      "epoch": 0.6752040402218514,
      "grad_norm": 1.9920288621159714,
      "learning_rate": 1.0081059520995591e-06,
      "loss": 0.7778,
      "step": 7487
    },
    {
      "epoch": 0.6752942237453218,
      "grad_norm": 1.8774130234972588,
      "learning_rate": 1.0075986935009028e-06,
      "loss": 0.8173,
      "step": 7488
    },
    {
      "epoch": 0.675384407268792,
      "grad_norm": 1.738718659305654,
      "learning_rate": 1.0070915195793982e-06,
      "loss": 0.7621,
      "step": 7489
    },
    {
      "epoch": 0.6754745907922622,
      "grad_norm": 2.0298264430226407,
      "learning_rate": 1.0065844303783197e-06,
      "loss": 0.7338,
      "step": 7490
    },
    {
      "epoch": 0.6755647743157325,
      "grad_norm": 1.6615681467467922,
      "learning_rate": 1.0060774259409356e-06,
      "loss": 0.6668,
      "step": 7491
    },
    {
      "epoch": 0.6756549578392028,
      "grad_norm": 1.5569949255427564,
      "learning_rate": 1.0055705063105065e-06,
      "loss": 0.7828,
      "step": 7492
    },
    {
      "epoch": 0.675745141362673,
      "grad_norm": 2.1449534516519044,
      "learning_rate": 1.0050636715302837e-06,
      "loss": 0.7455,
      "step": 7493
    },
    {
      "epoch": 0.6758353248861433,
      "grad_norm": 1.8980758385294314,
      "learning_rate": 1.0045569216435157e-06,
      "loss": 0.7676,
      "step": 7494
    },
    {
      "epoch": 0.6759255084096135,
      "grad_norm": 1.5652313011525767,
      "learning_rate": 1.0040502566934384e-06,
      "loss": 0.8343,
      "step": 7495
    },
    {
      "epoch": 0.6760156919330839,
      "grad_norm": 2.3869016486730725,
      "learning_rate": 1.0035436767232866e-06,
      "loss": 0.688,
      "step": 7496
    },
    {
      "epoch": 0.6761058754565541,
      "grad_norm": 3.301587104615414,
      "learning_rate": 1.0030371817762816e-06,
      "loss": 0.7639,
      "step": 7497
    },
    {
      "epoch": 0.6761960589800243,
      "grad_norm": 1.6084963427671928,
      "learning_rate": 1.0025307718956417e-06,
      "loss": 0.7574,
      "step": 7498
    },
    {
      "epoch": 0.6762862425034946,
      "grad_norm": 1.9274866317898964,
      "learning_rate": 1.0020244471245765e-06,
      "loss": 0.7727,
      "step": 7499
    },
    {
      "epoch": 0.6763764260269649,
      "grad_norm": 1.8583659150670078,
      "learning_rate": 1.001518207506288e-06,
      "loss": 0.7038,
      "step": 7500
    },
    {
      "epoch": 0.6764666095504351,
      "grad_norm": 1.5213871387363822,
      "learning_rate": 1.0010120530839717e-06,
      "loss": 0.8131,
      "step": 7501
    },
    {
      "epoch": 0.6765567930739054,
      "grad_norm": 1.2207205076887362,
      "learning_rate": 1.0005059839008161e-06,
      "loss": 0.7755,
      "step": 7502
    },
    {
      "epoch": 0.6766469765973756,
      "grad_norm": 1.86450602905739,
      "learning_rate": 1.0000000000000004e-06,
      "loss": 0.7044,
      "step": 7503
    },
    {
      "epoch": 0.676737160120846,
      "grad_norm": 1.713455924659141,
      "learning_rate": 9.994941014246985e-07,
      "loss": 0.7944,
      "step": 7504
    },
    {
      "epoch": 0.6768273436443162,
      "grad_norm": 1.6489211543557383,
      "learning_rate": 9.989882882180766e-07,
      "loss": 0.6411,
      "step": 7505
    },
    {
      "epoch": 0.6769175271677864,
      "grad_norm": 1.8212919424407525,
      "learning_rate": 9.984825604232938e-07,
      "loss": 0.769,
      "step": 7506
    },
    {
      "epoch": 0.6770077106912568,
      "grad_norm": 1.8161118135437622,
      "learning_rate": 9.97976918083502e-07,
      "loss": 0.6867,
      "step": 7507
    },
    {
      "epoch": 0.677097894214727,
      "grad_norm": 3.1127616802961824,
      "learning_rate": 9.974713612418427e-07,
      "loss": 0.8202,
      "step": 7508
    },
    {
      "epoch": 0.6771880777381972,
      "grad_norm": 2.30758164580649,
      "learning_rate": 9.969658899414563e-07,
      "loss": 0.766,
      "step": 7509
    },
    {
      "epoch": 0.6772782612616675,
      "grad_norm": 2.0632817491894038,
      "learning_rate": 9.964605042254696e-07,
      "loss": 0.6746,
      "step": 7510
    },
    {
      "epoch": 0.6773684447851378,
      "grad_norm": 4.3198809278820525,
      "learning_rate": 9.959552041370076e-07,
      "loss": 0.6925,
      "step": 7511
    },
    {
      "epoch": 0.677458628308608,
      "grad_norm": 3.3585896838266267,
      "learning_rate": 9.954499897191824e-07,
      "loss": 0.6708,
      "step": 7512
    },
    {
      "epoch": 0.6775488118320783,
      "grad_norm": 2.225427170798743,
      "learning_rate": 9.949448610151043e-07,
      "loss": 0.7048,
      "step": 7513
    },
    {
      "epoch": 0.6776389953555485,
      "grad_norm": 1.8625481080876263,
      "learning_rate": 9.944398180678719e-07,
      "loss": 0.734,
      "step": 7514
    },
    {
      "epoch": 0.6777291788790188,
      "grad_norm": 1.7675608597223083,
      "learning_rate": 9.939348609205789e-07,
      "loss": 0.7011,
      "step": 7515
    },
    {
      "epoch": 0.6778193624024891,
      "grad_norm": 1.4113788746594882,
      "learning_rate": 9.93429989616311e-07,
      "loss": 0.7582,
      "step": 7516
    },
    {
      "epoch": 0.6779095459259593,
      "grad_norm": 1.6369283537382497,
      "learning_rate": 9.929252041981464e-07,
      "loss": 0.6656,
      "step": 7517
    },
    {
      "epoch": 0.6779997294494295,
      "grad_norm": 1.8592397175789612,
      "learning_rate": 9.924205047091572e-07,
      "loss": 0.7249,
      "step": 7518
    },
    {
      "epoch": 0.6780899129728999,
      "grad_norm": 1.7072593135068581,
      "learning_rate": 9.919158911924056e-07,
      "loss": 0.8724,
      "step": 7519
    },
    {
      "epoch": 0.6781800964963701,
      "grad_norm": 1.7293998913340194,
      "learning_rate": 9.914113636909483e-07,
      "loss": 0.7858,
      "step": 7520
    },
    {
      "epoch": 0.6782702800198404,
      "grad_norm": 0.7678688023084371,
      "learning_rate": 9.90906922247835e-07,
      "loss": 0.6279,
      "step": 7521
    },
    {
      "epoch": 0.6783604635433106,
      "grad_norm": 1.6827193465523222,
      "learning_rate": 9.904025669061072e-07,
      "loss": 0.7796,
      "step": 7522
    },
    {
      "epoch": 0.6784506470667809,
      "grad_norm": 2.127541985103106,
      "learning_rate": 9.89898297708799e-07,
      "loss": 0.7456,
      "step": 7523
    },
    {
      "epoch": 0.6785408305902512,
      "grad_norm": 4.305368469986456,
      "learning_rate": 9.893941146989388e-07,
      "loss": 0.7079,
      "step": 7524
    },
    {
      "epoch": 0.6786310141137214,
      "grad_norm": 1.8896492576104487,
      "learning_rate": 9.888900179195437e-07,
      "loss": 0.7117,
      "step": 7525
    },
    {
      "epoch": 0.6787211976371916,
      "grad_norm": 1.8166612671704792,
      "learning_rate": 9.883860074136285e-07,
      "loss": 0.8196,
      "step": 7526
    },
    {
      "epoch": 0.678811381160662,
      "grad_norm": 3.5836603917340843,
      "learning_rate": 9.87882083224196e-07,
      "loss": 0.7458,
      "step": 7527
    },
    {
      "epoch": 0.6789015646841322,
      "grad_norm": 0.6865507421584949,
      "learning_rate": 9.873782453942462e-07,
      "loss": 0.6045,
      "step": 7528
    },
    {
      "epoch": 0.6789917482076024,
      "grad_norm": 1.7513936488004265,
      "learning_rate": 9.868744939667676e-07,
      "loss": 0.6771,
      "step": 7529
    },
    {
      "epoch": 0.6790819317310728,
      "grad_norm": 1.4955743829409012,
      "learning_rate": 9.863708289847432e-07,
      "loss": 0.6325,
      "step": 7530
    },
    {
      "epoch": 0.679172115254543,
      "grad_norm": 2.0068016030746,
      "learning_rate": 9.85867250491149e-07,
      "loss": 0.6638,
      "step": 7531
    },
    {
      "epoch": 0.6792622987780133,
      "grad_norm": 2.0275995170356382,
      "learning_rate": 9.853637585289528e-07,
      "loss": 0.857,
      "step": 7532
    },
    {
      "epoch": 0.6793524823014835,
      "grad_norm": 1.676071888722489,
      "learning_rate": 9.848603531411159e-07,
      "loss": 0.7393,
      "step": 7533
    },
    {
      "epoch": 0.6794426658249538,
      "grad_norm": 2.023726512638336,
      "learning_rate": 9.843570343705899e-07,
      "loss": 0.761,
      "step": 7534
    },
    {
      "epoch": 0.6795328493484241,
      "grad_norm": 1.436832687730596,
      "learning_rate": 9.83853802260323e-07,
      "loss": 0.8227,
      "step": 7535
    },
    {
      "epoch": 0.6796230328718943,
      "grad_norm": 1.7751193369118372,
      "learning_rate": 9.833506568532524e-07,
      "loss": 0.6984,
      "step": 7536
    },
    {
      "epoch": 0.6797132163953645,
      "grad_norm": 1.760445483571889,
      "learning_rate": 9.828475981923093e-07,
      "loss": 0.5987,
      "step": 7537
    },
    {
      "epoch": 0.6798033999188349,
      "grad_norm": 2.3592881950931943,
      "learning_rate": 9.823446263204175e-07,
      "loss": 0.7288,
      "step": 7538
    },
    {
      "epoch": 0.6798935834423051,
      "grad_norm": 1.7229149033869153,
      "learning_rate": 9.818417412804937e-07,
      "loss": 0.7614,
      "step": 7539
    },
    {
      "epoch": 0.6799837669657753,
      "grad_norm": 2.591422541937736,
      "learning_rate": 9.813389431154463e-07,
      "loss": 0.8303,
      "step": 7540
    },
    {
      "epoch": 0.6800739504892456,
      "grad_norm": 7.833954427999978,
      "learning_rate": 9.808362318681783e-07,
      "loss": 0.7799,
      "step": 7541
    },
    {
      "epoch": 0.6801641340127159,
      "grad_norm": 2.9097061116317047,
      "learning_rate": 9.803336075815807e-07,
      "loss": 0.7724,
      "step": 7542
    },
    {
      "epoch": 0.6802543175361861,
      "grad_norm": 3.006156961186985,
      "learning_rate": 9.79831070298544e-07,
      "loss": 0.6797,
      "step": 7543
    },
    {
      "epoch": 0.6803445010596564,
      "grad_norm": 1.7763498063204424,
      "learning_rate": 9.793286200619443e-07,
      "loss": 0.7243,
      "step": 7544
    },
    {
      "epoch": 0.6804346845831266,
      "grad_norm": 1.6648809005480372,
      "learning_rate": 9.78826256914655e-07,
      "loss": 0.8114,
      "step": 7545
    },
    {
      "epoch": 0.680524868106597,
      "grad_norm": 1.7264062456912461,
      "learning_rate": 9.7832398089954e-07,
      "loss": 0.8008,
      "step": 7546
    },
    {
      "epoch": 0.6806150516300672,
      "grad_norm": 1.562183271731633,
      "learning_rate": 9.778217920594565e-07,
      "loss": 0.8538,
      "step": 7547
    },
    {
      "epoch": 0.6807052351535374,
      "grad_norm": 1.9270926879105899,
      "learning_rate": 9.773196904372547e-07,
      "loss": 0.7941,
      "step": 7548
    },
    {
      "epoch": 0.6807954186770077,
      "grad_norm": 5.2100505388558025,
      "learning_rate": 9.768176760757742e-07,
      "loss": 0.8379,
      "step": 7549
    },
    {
      "epoch": 0.680885602200478,
      "grad_norm": 1.9853647722291725,
      "learning_rate": 9.76315749017853e-07,
      "loss": 0.7286,
      "step": 7550
    },
    {
      "epoch": 0.6809757857239482,
      "grad_norm": 1.5398055088785498,
      "learning_rate": 9.758139093063161e-07,
      "loss": 0.7531,
      "step": 7551
    },
    {
      "epoch": 0.6810659692474185,
      "grad_norm": 2.5986647405028442,
      "learning_rate": 9.753121569839834e-07,
      "loss": 0.822,
      "step": 7552
    },
    {
      "epoch": 0.6811561527708888,
      "grad_norm": 0.747112557207858,
      "learning_rate": 9.748104920936678e-07,
      "loss": 0.5485,
      "step": 7553
    },
    {
      "epoch": 0.681246336294359,
      "grad_norm": 1.7144954388486322,
      "learning_rate": 9.743089146781738e-07,
      "loss": 0.7313,
      "step": 7554
    },
    {
      "epoch": 0.6813365198178293,
      "grad_norm": 2.7421605470207626,
      "learning_rate": 9.738074247802988e-07,
      "loss": 0.6777,
      "step": 7555
    },
    {
      "epoch": 0.6814267033412995,
      "grad_norm": 1.675882901040851,
      "learning_rate": 9.733060224428325e-07,
      "loss": 0.7683,
      "step": 7556
    },
    {
      "epoch": 0.6815168868647699,
      "grad_norm": 2.3294465071478,
      "learning_rate": 9.728047077085577e-07,
      "loss": 0.7614,
      "step": 7557
    },
    {
      "epoch": 0.6816070703882401,
      "grad_norm": 3.756335945859899,
      "learning_rate": 9.723034806202497e-07,
      "loss": 0.6527,
      "step": 7558
    },
    {
      "epoch": 0.6816972539117103,
      "grad_norm": 2.230063166044482,
      "learning_rate": 9.718023412206748e-07,
      "loss": 0.7698,
      "step": 7559
    },
    {
      "epoch": 0.6817874374351806,
      "grad_norm": 1.3463264870765086,
      "learning_rate": 9.713012895525935e-07,
      "loss": 0.7755,
      "step": 7560
    },
    {
      "epoch": 0.6818776209586509,
      "grad_norm": 2.062312146503996,
      "learning_rate": 9.708003256587584e-07,
      "loss": 0.739,
      "step": 7561
    },
    {
      "epoch": 0.6819678044821211,
      "grad_norm": 1.594924568736436,
      "learning_rate": 9.702994495819147e-07,
      "loss": 0.798,
      "step": 7562
    },
    {
      "epoch": 0.6820579880055914,
      "grad_norm": 1.931857105617773,
      "learning_rate": 9.697986613647999e-07,
      "loss": 0.6583,
      "step": 7563
    },
    {
      "epoch": 0.6821481715290616,
      "grad_norm": 0.6997212783204223,
      "learning_rate": 9.692979610501425e-07,
      "loss": 0.6134,
      "step": 7564
    },
    {
      "epoch": 0.6822383550525319,
      "grad_norm": 1.860307603852365,
      "learning_rate": 9.68797348680668e-07,
      "loss": 0.824,
      "step": 7565
    },
    {
      "epoch": 0.6823285385760022,
      "grad_norm": 1.5610086571833974,
      "learning_rate": 9.682968242990878e-07,
      "loss": 0.7201,
      "step": 7566
    },
    {
      "epoch": 0.6824187220994724,
      "grad_norm": 1.5337863080267033,
      "learning_rate": 9.677963879481132e-07,
      "loss": 0.8014,
      "step": 7567
    },
    {
      "epoch": 0.6825089056229426,
      "grad_norm": 2.2291908144014316,
      "learning_rate": 9.672960396704416e-07,
      "loss": 0.747,
      "step": 7568
    },
    {
      "epoch": 0.682599089146413,
      "grad_norm": 1.7964741135262896,
      "learning_rate": 9.667957795087657e-07,
      "loss": 0.7524,
      "step": 7569
    },
    {
      "epoch": 0.6826892726698832,
      "grad_norm": 2.0948486861617996,
      "learning_rate": 9.662956075057712e-07,
      "loss": 0.8158,
      "step": 7570
    },
    {
      "epoch": 0.6827794561933535,
      "grad_norm": 2.5459828130497035,
      "learning_rate": 9.657955237041354e-07,
      "loss": 0.8161,
      "step": 7571
    },
    {
      "epoch": 0.6828696397168237,
      "grad_norm": 1.768071867548007,
      "learning_rate": 9.652955281465278e-07,
      "loss": 0.736,
      "step": 7572
    },
    {
      "epoch": 0.682959823240294,
      "grad_norm": 1.8188315330661218,
      "learning_rate": 9.64795620875612e-07,
      "loss": 0.7556,
      "step": 7573
    },
    {
      "epoch": 0.6830500067637643,
      "grad_norm": 1.5531508936728735,
      "learning_rate": 9.64295801934041e-07,
      "loss": 0.6964,
      "step": 7574
    },
    {
      "epoch": 0.6831401902872345,
      "grad_norm": 0.6662645965322753,
      "learning_rate": 9.63796071364463e-07,
      "loss": 0.5883,
      "step": 7575
    },
    {
      "epoch": 0.6832303738107048,
      "grad_norm": 1.9973899380341178,
      "learning_rate": 9.632964292095179e-07,
      "loss": 0.8345,
      "step": 7576
    },
    {
      "epoch": 0.6833205573341751,
      "grad_norm": 1.41562081982667,
      "learning_rate": 9.627968755118374e-07,
      "loss": 0.7399,
      "step": 7577
    },
    {
      "epoch": 0.6834107408576453,
      "grad_norm": 0.6668812560384196,
      "learning_rate": 9.622974103140468e-07,
      "loss": 0.5684,
      "step": 7578
    },
    {
      "epoch": 0.6835009243811155,
      "grad_norm": 2.322551225275938,
      "learning_rate": 9.617980336587632e-07,
      "loss": 0.8145,
      "step": 7579
    },
    {
      "epoch": 0.6835911079045859,
      "grad_norm": 2.863869724976285,
      "learning_rate": 9.612987455885964e-07,
      "loss": 0.6328,
      "step": 7580
    },
    {
      "epoch": 0.6836812914280561,
      "grad_norm": 1.7213512503348696,
      "learning_rate": 9.607995461461467e-07,
      "loss": 0.7559,
      "step": 7581
    },
    {
      "epoch": 0.6837714749515263,
      "grad_norm": 1.9017586875647892,
      "learning_rate": 9.603004353740111e-07,
      "loss": 0.6483,
      "step": 7582
    },
    {
      "epoch": 0.6838616584749966,
      "grad_norm": 0.6945301969882804,
      "learning_rate": 9.598014133147738e-07,
      "loss": 0.6227,
      "step": 7583
    },
    {
      "epoch": 0.6839518419984669,
      "grad_norm": 1.5776379277405612,
      "learning_rate": 9.59302480011017e-07,
      "loss": 0.713,
      "step": 7584
    },
    {
      "epoch": 0.6840420255219372,
      "grad_norm": 0.6744665466266948,
      "learning_rate": 9.588036355053102e-07,
      "loss": 0.5987,
      "step": 7585
    },
    {
      "epoch": 0.6841322090454074,
      "grad_norm": 2.0150046170089495,
      "learning_rate": 9.583048798402182e-07,
      "loss": 0.6224,
      "step": 7586
    },
    {
      "epoch": 0.6842223925688776,
      "grad_norm": 2.326164700411375,
      "learning_rate": 9.57806213058298e-07,
      "loss": 0.7469,
      "step": 7587
    },
    {
      "epoch": 0.684312576092348,
      "grad_norm": 0.6958657538880862,
      "learning_rate": 9.57307635202098e-07,
      "loss": 0.6164,
      "step": 7588
    },
    {
      "epoch": 0.6844027596158182,
      "grad_norm": 1.6039352332616703,
      "learning_rate": 9.568091463141607e-07,
      "loss": 0.8549,
      "step": 7589
    },
    {
      "epoch": 0.6844929431392884,
      "grad_norm": 1.5426636985831583,
      "learning_rate": 9.563107464370187e-07,
      "loss": 0.718,
      "step": 7590
    },
    {
      "epoch": 0.6845831266627587,
      "grad_norm": 1.6583668119159825,
      "learning_rate": 9.558124356131982e-07,
      "loss": 0.7665,
      "step": 7591
    },
    {
      "epoch": 0.684673310186229,
      "grad_norm": 1.8396069805602568,
      "learning_rate": 9.553142138852187e-07,
      "loss": 0.6875,
      "step": 7592
    },
    {
      "epoch": 0.6847634937096992,
      "grad_norm": 1.920320890453312,
      "learning_rate": 9.548160812955905e-07,
      "loss": 0.6582,
      "step": 7593
    },
    {
      "epoch": 0.6848536772331695,
      "grad_norm": 1.96163952180182,
      "learning_rate": 9.543180378868175e-07,
      "loss": 0.5754,
      "step": 7594
    },
    {
      "epoch": 0.6849438607566397,
      "grad_norm": 1.8695252918574092,
      "learning_rate": 9.538200837013962e-07,
      "loss": 0.7892,
      "step": 7595
    },
    {
      "epoch": 0.68503404428011,
      "grad_norm": 0.6615449192916768,
      "learning_rate": 9.533222187818122e-07,
      "loss": 0.6155,
      "step": 7596
    },
    {
      "epoch": 0.6851242278035803,
      "grad_norm": 4.326083696900453,
      "learning_rate": 9.528244431705492e-07,
      "loss": 0.7599,
      "step": 7597
    },
    {
      "epoch": 0.6852144113270505,
      "grad_norm": 4.6300135064803465,
      "learning_rate": 9.523267569100774e-07,
      "loss": 0.758,
      "step": 7598
    },
    {
      "epoch": 0.6853045948505209,
      "grad_norm": 2.6118581686141655,
      "learning_rate": 9.518291600428652e-07,
      "loss": 0.7078,
      "step": 7599
    },
    {
      "epoch": 0.6853947783739911,
      "grad_norm": 2.248480071512426,
      "learning_rate": 9.513316526113677e-07,
      "loss": 0.7776,
      "step": 7600
    },
    {
      "epoch": 0.6854849618974613,
      "grad_norm": 1.8230536127875274,
      "learning_rate": 9.50834234658036e-07,
      "loss": 0.7746,
      "step": 7601
    },
    {
      "epoch": 0.6855751454209316,
      "grad_norm": 1.9413807673480068,
      "learning_rate": 9.503369062253123e-07,
      "loss": 0.7674,
      "step": 7602
    },
    {
      "epoch": 0.6856653289444019,
      "grad_norm": 0.6458212687278152,
      "learning_rate": 9.498396673556317e-07,
      "loss": 0.5908,
      "step": 7603
    },
    {
      "epoch": 0.6857555124678721,
      "grad_norm": 1.4405160858622612,
      "learning_rate": 9.493425180914219e-07,
      "loss": 0.7859,
      "step": 7604
    },
    {
      "epoch": 0.6858456959913424,
      "grad_norm": 3.06252319949958,
      "learning_rate": 9.488454584751e-07,
      "loss": 0.6811,
      "step": 7605
    },
    {
      "epoch": 0.6859358795148126,
      "grad_norm": 1.492092988381479,
      "learning_rate": 9.483484885490813e-07,
      "loss": 0.7551,
      "step": 7606
    },
    {
      "epoch": 0.686026063038283,
      "grad_norm": 2.3074083722108405,
      "learning_rate": 9.478516083557675e-07,
      "loss": 0.7635,
      "step": 7607
    },
    {
      "epoch": 0.6861162465617532,
      "grad_norm": 5.250366198165967,
      "learning_rate": 9.473548179375561e-07,
      "loss": 0.7674,
      "step": 7608
    },
    {
      "epoch": 0.6862064300852234,
      "grad_norm": 0.681249446168727,
      "learning_rate": 9.468581173368358e-07,
      "loss": 0.6171,
      "step": 7609
    },
    {
      "epoch": 0.6862966136086937,
      "grad_norm": 1.628320236361624,
      "learning_rate": 9.463615065959878e-07,
      "loss": 0.7587,
      "step": 7610
    },
    {
      "epoch": 0.686386797132164,
      "grad_norm": 4.435818205731258,
      "learning_rate": 9.458649857573857e-07,
      "loss": 0.8015,
      "step": 7611
    },
    {
      "epoch": 0.6864769806556342,
      "grad_norm": 1.5362746651565136,
      "learning_rate": 9.453685548633963e-07,
      "loss": 0.7578,
      "step": 7612
    },
    {
      "epoch": 0.6865671641791045,
      "grad_norm": 2.0899605691648975,
      "learning_rate": 9.448722139563756e-07,
      "loss": 0.7018,
      "step": 7613
    },
    {
      "epoch": 0.6866573477025747,
      "grad_norm": 2.3700533097142458,
      "learning_rate": 9.443759630786769e-07,
      "loss": 0.652,
      "step": 7614
    },
    {
      "epoch": 0.686747531226045,
      "grad_norm": 1.6829193959690576,
      "learning_rate": 9.438798022726408e-07,
      "loss": 0.8233,
      "step": 7615
    },
    {
      "epoch": 0.6868377147495153,
      "grad_norm": 2.1985111487225004,
      "learning_rate": 9.433837315806037e-07,
      "loss": 0.708,
      "step": 7616
    },
    {
      "epoch": 0.6869278982729855,
      "grad_norm": 1.8127579998681058,
      "learning_rate": 9.428877510448925e-07,
      "loss": 0.7087,
      "step": 7617
    },
    {
      "epoch": 0.6870180817964557,
      "grad_norm": 1.9789839074597086,
      "learning_rate": 9.423918607078272e-07,
      "loss": 0.6811,
      "step": 7618
    },
    {
      "epoch": 0.6871082653199261,
      "grad_norm": 1.5653348002765373,
      "learning_rate": 9.418960606117208e-07,
      "loss": 0.7593,
      "step": 7619
    },
    {
      "epoch": 0.6871984488433963,
      "grad_norm": 1.92157234351912,
      "learning_rate": 9.414003507988752e-07,
      "loss": 0.6937,
      "step": 7620
    },
    {
      "epoch": 0.6872886323668665,
      "grad_norm": 1.6408895370107974,
      "learning_rate": 9.409047313115904e-07,
      "loss": 0.6848,
      "step": 7621
    },
    {
      "epoch": 0.6873788158903368,
      "grad_norm": 1.994492039367901,
      "learning_rate": 9.404092021921521e-07,
      "loss": 0.6937,
      "step": 7622
    },
    {
      "epoch": 0.6874689994138071,
      "grad_norm": 1.8214878986217393,
      "learning_rate": 9.399137634828447e-07,
      "loss": 0.7307,
      "step": 7623
    },
    {
      "epoch": 0.6875591829372774,
      "grad_norm": 1.5999101255924173,
      "learning_rate": 9.394184152259396e-07,
      "loss": 0.7761,
      "step": 7624
    },
    {
      "epoch": 0.6876493664607476,
      "grad_norm": 1.5964540467952884,
      "learning_rate": 9.389231574637033e-07,
      "loss": 0.8563,
      "step": 7625
    },
    {
      "epoch": 0.6877395499842179,
      "grad_norm": 1.7255008246907044,
      "learning_rate": 9.384279902383938e-07,
      "loss": 0.8357,
      "step": 7626
    },
    {
      "epoch": 0.6878297335076882,
      "grad_norm": 6.028921043315562,
      "learning_rate": 9.379329135922615e-07,
      "loss": 0.7717,
      "step": 7627
    },
    {
      "epoch": 0.6879199170311584,
      "grad_norm": 1.6380905920881095,
      "learning_rate": 9.374379275675495e-07,
      "loss": 0.7098,
      "step": 7628
    },
    {
      "epoch": 0.6880101005546286,
      "grad_norm": 2.6619012293673157,
      "learning_rate": 9.369430322064931e-07,
      "loss": 0.7453,
      "step": 7629
    },
    {
      "epoch": 0.688100284078099,
      "grad_norm": 12.818785377129998,
      "learning_rate": 9.364482275513179e-07,
      "loss": 0.8324,
      "step": 7630
    },
    {
      "epoch": 0.6881904676015692,
      "grad_norm": 2.330854927520319,
      "learning_rate": 9.359535136442444e-07,
      "loss": 0.8154,
      "step": 7631
    },
    {
      "epoch": 0.6882806511250394,
      "grad_norm": 1.6784336822017474,
      "learning_rate": 9.354588905274843e-07,
      "loss": 0.6863,
      "step": 7632
    },
    {
      "epoch": 0.6883708346485097,
      "grad_norm": 1.5858589491159323,
      "learning_rate": 9.349643582432414e-07,
      "loss": 0.7849,
      "step": 7633
    },
    {
      "epoch": 0.68846101817198,
      "grad_norm": 1.6279234631470314,
      "learning_rate": 9.344699168337127e-07,
      "loss": 0.7457,
      "step": 7634
    },
    {
      "epoch": 0.6885512016954503,
      "grad_norm": 1.5630353392000242,
      "learning_rate": 9.339755663410845e-07,
      "loss": 0.7189,
      "step": 7635
    },
    {
      "epoch": 0.6886413852189205,
      "grad_norm": 1.7298986731606785,
      "learning_rate": 9.334813068075405e-07,
      "loss": 0.7853,
      "step": 7636
    },
    {
      "epoch": 0.6887315687423907,
      "grad_norm": 1.3895087315990946,
      "learning_rate": 9.329871382752506e-07,
      "loss": 0.757,
      "step": 7637
    },
    {
      "epoch": 0.6888217522658611,
      "grad_norm": 1.6201973884326968,
      "learning_rate": 9.32493060786383e-07,
      "loss": 0.7422,
      "step": 7638
    },
    {
      "epoch": 0.6889119357893313,
      "grad_norm": 1.510888238713308,
      "learning_rate": 9.31999074383093e-07,
      "loss": 0.8097,
      "step": 7639
    },
    {
      "epoch": 0.6890021193128015,
      "grad_norm": 2.7477233305837627,
      "learning_rate": 9.315051791075308e-07,
      "loss": 0.7486,
      "step": 7640
    },
    {
      "epoch": 0.6890923028362718,
      "grad_norm": 1.5001215090809965,
      "learning_rate": 9.310113750018382e-07,
      "loss": 0.7438,
      "step": 7641
    },
    {
      "epoch": 0.6891824863597421,
      "grad_norm": 1.916506269903001,
      "learning_rate": 9.305176621081496e-07,
      "loss": 0.8205,
      "step": 7642
    },
    {
      "epoch": 0.6892726698832123,
      "grad_norm": 1.5677929303060099,
      "learning_rate": 9.300240404685911e-07,
      "loss": 0.7851,
      "step": 7643
    },
    {
      "epoch": 0.6893628534066826,
      "grad_norm": 1.8322445352842118,
      "learning_rate": 9.295305101252812e-07,
      "loss": 0.754,
      "step": 7644
    },
    {
      "epoch": 0.6894530369301528,
      "grad_norm": 1.9913120515378124,
      "learning_rate": 9.290370711203314e-07,
      "loss": 0.7358,
      "step": 7645
    },
    {
      "epoch": 0.6895432204536232,
      "grad_norm": 2.8011276903392766,
      "learning_rate": 9.285437234958433e-07,
      "loss": 0.7457,
      "step": 7646
    },
    {
      "epoch": 0.6896334039770934,
      "grad_norm": 3.2203144604687597,
      "learning_rate": 9.280504672939124e-07,
      "loss": 0.744,
      "step": 7647
    },
    {
      "epoch": 0.6897235875005636,
      "grad_norm": 3.094556606991068,
      "learning_rate": 9.275573025566266e-07,
      "loss": 0.7908,
      "step": 7648
    },
    {
      "epoch": 0.689813771024034,
      "grad_norm": 3.0439316838301735,
      "learning_rate": 9.27064229326065e-07,
      "loss": 0.7825,
      "step": 7649
    },
    {
      "epoch": 0.6899039545475042,
      "grad_norm": 1.7341832922756135,
      "learning_rate": 9.265712476442995e-07,
      "loss": 0.7183,
      "step": 7650
    },
    {
      "epoch": 0.6899941380709744,
      "grad_norm": 2.000765058101152,
      "learning_rate": 9.260783575533949e-07,
      "loss": 0.6821,
      "step": 7651
    },
    {
      "epoch": 0.6900843215944447,
      "grad_norm": 1.5725915417005771,
      "learning_rate": 9.255855590954045e-07,
      "loss": 0.7011,
      "step": 7652
    },
    {
      "epoch": 0.690174505117915,
      "grad_norm": 3.674729556558247,
      "learning_rate": 9.250928523123802e-07,
      "loss": 0.8178,
      "step": 7653
    },
    {
      "epoch": 0.6902646886413852,
      "grad_norm": 1.4580168880274726,
      "learning_rate": 9.24600237246359e-07,
      "loss": 0.7593,
      "step": 7654
    },
    {
      "epoch": 0.6903548721648555,
      "grad_norm": 1.9629096945979783,
      "learning_rate": 9.241077139393769e-07,
      "loss": 0.7611,
      "step": 7655
    },
    {
      "epoch": 0.6904450556883257,
      "grad_norm": 5.453635292989031,
      "learning_rate": 9.236152824334564e-07,
      "loss": 0.7816,
      "step": 7656
    },
    {
      "epoch": 0.690535239211796,
      "grad_norm": 11.321063262285564,
      "learning_rate": 9.231229427706151e-07,
      "loss": 0.6498,
      "step": 7657
    },
    {
      "epoch": 0.6906254227352663,
      "grad_norm": 1.7419626407318047,
      "learning_rate": 9.226306949928622e-07,
      "loss": 0.7568,
      "step": 7658
    },
    {
      "epoch": 0.6907156062587365,
      "grad_norm": 1.6382547604698197,
      "learning_rate": 9.221385391421988e-07,
      "loss": 0.827,
      "step": 7659
    },
    {
      "epoch": 0.6908057897822067,
      "grad_norm": 2.673736102552709,
      "learning_rate": 9.216464752606192e-07,
      "loss": 0.7673,
      "step": 7660
    },
    {
      "epoch": 0.6908959733056771,
      "grad_norm": 1.659135715739429,
      "learning_rate": 9.211545033901078e-07,
      "loss": 0.8289,
      "step": 7661
    },
    {
      "epoch": 0.6909861568291473,
      "grad_norm": 1.7601540123137547,
      "learning_rate": 9.206626235726426e-07,
      "loss": 0.7836,
      "step": 7662
    },
    {
      "epoch": 0.6910763403526176,
      "grad_norm": 2.5802486112117955,
      "learning_rate": 9.20170835850194e-07,
      "loss": 0.689,
      "step": 7663
    },
    {
      "epoch": 0.6911665238760878,
      "grad_norm": 0.943544989543737,
      "learning_rate": 9.196791402647237e-07,
      "loss": 0.573,
      "step": 7664
    },
    {
      "epoch": 0.6912567073995581,
      "grad_norm": 1.77463017492849,
      "learning_rate": 9.191875368581861e-07,
      "loss": 0.676,
      "step": 7665
    },
    {
      "epoch": 0.6913468909230284,
      "grad_norm": 1.7972673070808936,
      "learning_rate": 9.186960256725271e-07,
      "loss": 0.7652,
      "step": 7666
    },
    {
      "epoch": 0.6914370744464986,
      "grad_norm": 1.711442916283693,
      "learning_rate": 9.182046067496856e-07,
      "loss": 0.8182,
      "step": 7667
    },
    {
      "epoch": 0.6915272579699688,
      "grad_norm": 2.43275449434978,
      "learning_rate": 9.177132801315927e-07,
      "loss": 0.8366,
      "step": 7668
    },
    {
      "epoch": 0.6916174414934392,
      "grad_norm": 2.1977925800411726,
      "learning_rate": 9.172220458601692e-07,
      "loss": 0.7258,
      "step": 7669
    },
    {
      "epoch": 0.6917076250169094,
      "grad_norm": 2.6869699709555546,
      "learning_rate": 9.167309039773324e-07,
      "loss": 0.7177,
      "step": 7670
    },
    {
      "epoch": 0.6917978085403796,
      "grad_norm": 1.3032652360942794,
      "learning_rate": 9.162398545249872e-07,
      "loss": 0.6477,
      "step": 7671
    },
    {
      "epoch": 0.69188799206385,
      "grad_norm": 1.302248016115707,
      "learning_rate": 9.157488975450334e-07,
      "loss": 0.8068,
      "step": 7672
    },
    {
      "epoch": 0.6919781755873202,
      "grad_norm": 1.5770530459251728,
      "learning_rate": 9.15258033079362e-07,
      "loss": 0.5798,
      "step": 7673
    },
    {
      "epoch": 0.6920683591107905,
      "grad_norm": 1.6502883427994888,
      "learning_rate": 9.147672611698567e-07,
      "loss": 0.7409,
      "step": 7674
    },
    {
      "epoch": 0.6921585426342607,
      "grad_norm": 1.899131867694056,
      "learning_rate": 9.142765818583933e-07,
      "loss": 0.7467,
      "step": 7675
    },
    {
      "epoch": 0.692248726157731,
      "grad_norm": 1.5361086000157933,
      "learning_rate": 9.13785995186837e-07,
      "loss": 0.7298,
      "step": 7676
    },
    {
      "epoch": 0.6923389096812013,
      "grad_norm": 2.8403890681724837,
      "learning_rate": 9.132955011970502e-07,
      "loss": 0.7173,
      "step": 7677
    },
    {
      "epoch": 0.6924290932046715,
      "grad_norm": 1.2708099967606572,
      "learning_rate": 9.128050999308827e-07,
      "loss": 0.7392,
      "step": 7678
    },
    {
      "epoch": 0.6925192767281417,
      "grad_norm": 3.0077362447212783,
      "learning_rate": 9.123147914301789e-07,
      "loss": 0.8432,
      "step": 7679
    },
    {
      "epoch": 0.6926094602516121,
      "grad_norm": 2.3137509854323373,
      "learning_rate": 9.118245757367745e-07,
      "loss": 0.8006,
      "step": 7680
    },
    {
      "epoch": 0.6926996437750823,
      "grad_norm": 0.6194301133025718,
      "learning_rate": 9.113344528924973e-07,
      "loss": 0.5605,
      "step": 7681
    },
    {
      "epoch": 0.6927898272985525,
      "grad_norm": 1.4153365828742877,
      "learning_rate": 9.108444229391676e-07,
      "loss": 0.7324,
      "step": 7682
    },
    {
      "epoch": 0.6928800108220228,
      "grad_norm": 1.844930044879387,
      "learning_rate": 9.103544859185979e-07,
      "loss": 0.7277,
      "step": 7683
    },
    {
      "epoch": 0.6929701943454931,
      "grad_norm": 2.3239698773662627,
      "learning_rate": 9.098646418725902e-07,
      "loss": 0.8383,
      "step": 7684
    },
    {
      "epoch": 0.6930603778689634,
      "grad_norm": 0.6421167868411412,
      "learning_rate": 9.093748908429437e-07,
      "loss": 0.6031,
      "step": 7685
    },
    {
      "epoch": 0.6931505613924336,
      "grad_norm": 1.6224450052401427,
      "learning_rate": 9.088852328714444e-07,
      "loss": 0.7398,
      "step": 7686
    },
    {
      "epoch": 0.6932407449159038,
      "grad_norm": 1.4756024181119691,
      "learning_rate": 9.083956679998735e-07,
      "loss": 0.6602,
      "step": 7687
    },
    {
      "epoch": 0.6933309284393742,
      "grad_norm": 1.3548476585390756,
      "learning_rate": 9.079061962700032e-07,
      "loss": 0.7171,
      "step": 7688
    },
    {
      "epoch": 0.6934211119628444,
      "grad_norm": 1.4796215038627254,
      "learning_rate": 9.074168177235979e-07,
      "loss": 0.772,
      "step": 7689
    },
    {
      "epoch": 0.6935112954863146,
      "grad_norm": 1.6670693467378273,
      "learning_rate": 9.069275324024151e-07,
      "loss": 0.7625,
      "step": 7690
    },
    {
      "epoch": 0.6936014790097849,
      "grad_norm": 14.924092296812864,
      "learning_rate": 9.064383403482005e-07,
      "loss": 0.6801,
      "step": 7691
    },
    {
      "epoch": 0.6936916625332552,
      "grad_norm": 1.4385363947693253,
      "learning_rate": 9.059492416026983e-07,
      "loss": 0.7123,
      "step": 7692
    },
    {
      "epoch": 0.6937818460567254,
      "grad_norm": 1.5325257572666735,
      "learning_rate": 9.054602362076378e-07,
      "loss": 0.7459,
      "step": 7693
    },
    {
      "epoch": 0.6938720295801957,
      "grad_norm": 4.277062811835561,
      "learning_rate": 9.049713242047468e-07,
      "loss": 0.7407,
      "step": 7694
    },
    {
      "epoch": 0.693962213103666,
      "grad_norm": 2.608912261318088,
      "learning_rate": 9.044825056357395e-07,
      "loss": 0.7214,
      "step": 7695
    },
    {
      "epoch": 0.6940523966271362,
      "grad_norm": 1.5265025288135763,
      "learning_rate": 9.039937805423255e-07,
      "loss": 0.7062,
      "step": 7696
    },
    {
      "epoch": 0.6941425801506065,
      "grad_norm": 0.5980486859195134,
      "learning_rate": 9.035051489662051e-07,
      "loss": 0.5174,
      "step": 7697
    },
    {
      "epoch": 0.6942327636740767,
      "grad_norm": 1.8052520529472913,
      "learning_rate": 9.030166109490718e-07,
      "loss": 0.7014,
      "step": 7698
    },
    {
      "epoch": 0.6943229471975471,
      "grad_norm": 5.681680573617721,
      "learning_rate": 9.025281665326099e-07,
      "loss": 0.7621,
      "step": 7699
    },
    {
      "epoch": 0.6944131307210173,
      "grad_norm": 2.478716181583522,
      "learning_rate": 9.020398157584967e-07,
      "loss": 0.7619,
      "step": 7700
    },
    {
      "epoch": 0.6945033142444875,
      "grad_norm": 1.7341746309028163,
      "learning_rate": 9.015515586684002e-07,
      "loss": 0.6526,
      "step": 7701
    },
    {
      "epoch": 0.6945934977679578,
      "grad_norm": 0.6599069158340537,
      "learning_rate": 9.010633953039812e-07,
      "loss": 0.5937,
      "step": 7702
    },
    {
      "epoch": 0.6946836812914281,
      "grad_norm": 1.476526492048135,
      "learning_rate": 9.005753257068929e-07,
      "loss": 0.7617,
      "step": 7703
    },
    {
      "epoch": 0.6947738648148983,
      "grad_norm": 1.7979385545561541,
      "learning_rate": 9.000873499187797e-07,
      "loss": 0.7647,
      "step": 7704
    },
    {
      "epoch": 0.6948640483383686,
      "grad_norm": 1.5538580133729294,
      "learning_rate": 8.995994679812797e-07,
      "loss": 0.7814,
      "step": 7705
    },
    {
      "epoch": 0.6949542318618388,
      "grad_norm": 9.819264406828596,
      "learning_rate": 8.991116799360192e-07,
      "loss": 0.7738,
      "step": 7706
    },
    {
      "epoch": 0.6950444153853091,
      "grad_norm": 1.673337849019456,
      "learning_rate": 8.986239858246217e-07,
      "loss": 0.746,
      "step": 7707
    },
    {
      "epoch": 0.6951345989087794,
      "grad_norm": 1.9109610205986187,
      "learning_rate": 8.981363856886972e-07,
      "loss": 0.8345,
      "step": 7708
    },
    {
      "epoch": 0.6952247824322496,
      "grad_norm": 1.7431387592263665,
      "learning_rate": 8.976488795698533e-07,
      "loss": 0.7345,
      "step": 7709
    },
    {
      "epoch": 0.6953149659557198,
      "grad_norm": 1.452888141319604,
      "learning_rate": 8.971614675096841e-07,
      "loss": 0.7634,
      "step": 7710
    },
    {
      "epoch": 0.6954051494791902,
      "grad_norm": 2.45701931463839,
      "learning_rate": 8.966741495497807e-07,
      "loss": 0.817,
      "step": 7711
    },
    {
      "epoch": 0.6954953330026604,
      "grad_norm": 1.572056499339473,
      "learning_rate": 8.961869257317218e-07,
      "loss": 0.7163,
      "step": 7712
    },
    {
      "epoch": 0.6955855165261307,
      "grad_norm": 1.7493154004115974,
      "learning_rate": 8.956997960970809e-07,
      "loss": 0.7351,
      "step": 7713
    },
    {
      "epoch": 0.6956757000496009,
      "grad_norm": 9.105234137367862,
      "learning_rate": 8.952127606874224e-07,
      "loss": 0.6623,
      "step": 7714
    },
    {
      "epoch": 0.6957658835730712,
      "grad_norm": 1.590419205878515,
      "learning_rate": 8.947258195443028e-07,
      "loss": 0.7525,
      "step": 7715
    },
    {
      "epoch": 0.6958560670965415,
      "grad_norm": 2.089625381041602,
      "learning_rate": 8.942389727092716e-07,
      "loss": 0.7366,
      "step": 7716
    },
    {
      "epoch": 0.6959462506200117,
      "grad_norm": 0.6774862973210491,
      "learning_rate": 8.937522202238677e-07,
      "loss": 0.5648,
      "step": 7717
    },
    {
      "epoch": 0.696036434143482,
      "grad_norm": 2.302586105012382,
      "learning_rate": 8.932655621296239e-07,
      "loss": 0.7675,
      "step": 7718
    },
    {
      "epoch": 0.6961266176669523,
      "grad_norm": 1.4292483463060928,
      "learning_rate": 8.927789984680649e-07,
      "loss": 0.7483,
      "step": 7719
    },
    {
      "epoch": 0.6962168011904225,
      "grad_norm": 1.7516891637870375,
      "learning_rate": 8.922925292807068e-07,
      "loss": 0.6221,
      "step": 7720
    },
    {
      "epoch": 0.6963069847138927,
      "grad_norm": 1.5968854392470055,
      "learning_rate": 8.91806154609058e-07,
      "loss": 0.8338,
      "step": 7721
    },
    {
      "epoch": 0.6963971682373631,
      "grad_norm": 1.6777220826592394,
      "learning_rate": 8.913198744946195e-07,
      "loss": 0.7212,
      "step": 7722
    },
    {
      "epoch": 0.6964873517608333,
      "grad_norm": 1.6494246867025733,
      "learning_rate": 8.908336889788807e-07,
      "loss": 0.774,
      "step": 7723
    },
    {
      "epoch": 0.6965775352843036,
      "grad_norm": 1.727338935641753,
      "learning_rate": 8.903475981033293e-07,
      "loss": 0.6291,
      "step": 7724
    },
    {
      "epoch": 0.6966677188077738,
      "grad_norm": 0.7413665648311599,
      "learning_rate": 8.898616019094376e-07,
      "loss": 0.6325,
      "step": 7725
    },
    {
      "epoch": 0.6967579023312441,
      "grad_norm": 1.9793827608159336,
      "learning_rate": 8.89375700438677e-07,
      "loss": 0.791,
      "step": 7726
    },
    {
      "epoch": 0.6968480858547144,
      "grad_norm": 1.903898508512453,
      "learning_rate": 8.888898937325047e-07,
      "loss": 0.7146,
      "step": 7727
    },
    {
      "epoch": 0.6969382693781846,
      "grad_norm": 2.0406105917265673,
      "learning_rate": 8.884041818323733e-07,
      "loss": 0.7174,
      "step": 7728
    },
    {
      "epoch": 0.6970284529016548,
      "grad_norm": 2.9132630425127433,
      "learning_rate": 8.879185647797262e-07,
      "loss": 0.7867,
      "step": 7729
    },
    {
      "epoch": 0.6971186364251252,
      "grad_norm": 1.689535890878188,
      "learning_rate": 8.874330426159993e-07,
      "loss": 0.7325,
      "step": 7730
    },
    {
      "epoch": 0.6972088199485954,
      "grad_norm": 1.4513281921560837,
      "learning_rate": 8.869476153826205e-07,
      "loss": 0.773,
      "step": 7731
    },
    {
      "epoch": 0.6972990034720656,
      "grad_norm": 2.5893740220819605,
      "learning_rate": 8.864622831210071e-07,
      "loss": 0.7502,
      "step": 7732
    },
    {
      "epoch": 0.6973891869955359,
      "grad_norm": 2.1524856437532214,
      "learning_rate": 8.85977045872573e-07,
      "loss": 0.7569,
      "step": 7733
    },
    {
      "epoch": 0.6974793705190062,
      "grad_norm": 1.7399612366534958,
      "learning_rate": 8.854919036787194e-07,
      "loss": 0.7307,
      "step": 7734
    },
    {
      "epoch": 0.6975695540424764,
      "grad_norm": 1.694201619653487,
      "learning_rate": 8.850068565808417e-07,
      "loss": 0.7158,
      "step": 7735
    },
    {
      "epoch": 0.6976597375659467,
      "grad_norm": 2.0321840926154544,
      "learning_rate": 8.845219046203271e-07,
      "loss": 0.7957,
      "step": 7736
    },
    {
      "epoch": 0.6977499210894169,
      "grad_norm": 2.4798917805224945,
      "learning_rate": 8.840370478385544e-07,
      "loss": 0.7772,
      "step": 7737
    },
    {
      "epoch": 0.6978401046128873,
      "grad_norm": 1.6996037891026987,
      "learning_rate": 8.83552286276894e-07,
      "loss": 0.7664,
      "step": 7738
    },
    {
      "epoch": 0.6979302881363575,
      "grad_norm": 1.9316641378948434,
      "learning_rate": 8.830676199767095e-07,
      "loss": 0.7713,
      "step": 7739
    },
    {
      "epoch": 0.6980204716598277,
      "grad_norm": 1.5594556523872987,
      "learning_rate": 8.825830489793527e-07,
      "loss": 0.7356,
      "step": 7740
    },
    {
      "epoch": 0.698110655183298,
      "grad_norm": 1.3750551819565635,
      "learning_rate": 8.820985733261732e-07,
      "loss": 0.708,
      "step": 7741
    },
    {
      "epoch": 0.6982008387067683,
      "grad_norm": 1.7337355938610641,
      "learning_rate": 8.816141930585066e-07,
      "loss": 0.6627,
      "step": 7742
    },
    {
      "epoch": 0.6982910222302385,
      "grad_norm": 0.6925422043798745,
      "learning_rate": 8.811299082176837e-07,
      "loss": 0.5901,
      "step": 7743
    },
    {
      "epoch": 0.6983812057537088,
      "grad_norm": 2.483939366563507,
      "learning_rate": 8.806457188450265e-07,
      "loss": 0.7115,
      "step": 7744
    },
    {
      "epoch": 0.6984713892771791,
      "grad_norm": 1.800349980344737,
      "learning_rate": 8.801616249818487e-07,
      "loss": 0.7873,
      "step": 7745
    },
    {
      "epoch": 0.6985615728006493,
      "grad_norm": 1.8390823376152026,
      "learning_rate": 8.796776266694564e-07,
      "loss": 0.73,
      "step": 7746
    },
    {
      "epoch": 0.6986517563241196,
      "grad_norm": 1.7736156222927417,
      "learning_rate": 8.79193723949145e-07,
      "loss": 0.6959,
      "step": 7747
    },
    {
      "epoch": 0.6987419398475898,
      "grad_norm": 1.685769571420678,
      "learning_rate": 8.787099168622063e-07,
      "loss": 0.6441,
      "step": 7748
    },
    {
      "epoch": 0.6988321233710602,
      "grad_norm": 1.6639565447534255,
      "learning_rate": 8.782262054499199e-07,
      "loss": 0.6913,
      "step": 7749
    },
    {
      "epoch": 0.6989223068945304,
      "grad_norm": 1.310776123371528,
      "learning_rate": 8.777425897535588e-07,
      "loss": 0.781,
      "step": 7750
    },
    {
      "epoch": 0.6990124904180006,
      "grad_norm": 8.034225684447307,
      "learning_rate": 8.77259069814388e-07,
      "loss": 0.7559,
      "step": 7751
    },
    {
      "epoch": 0.6991026739414709,
      "grad_norm": 2.8578942639150005,
      "learning_rate": 8.767756456736641e-07,
      "loss": 0.7494,
      "step": 7752
    },
    {
      "epoch": 0.6991928574649412,
      "grad_norm": 2.2645213628979612,
      "learning_rate": 8.762923173726358e-07,
      "loss": 0.6753,
      "step": 7753
    },
    {
      "epoch": 0.6992830409884114,
      "grad_norm": 1.850431201273301,
      "learning_rate": 8.758090849525428e-07,
      "loss": 0.8467,
      "step": 7754
    },
    {
      "epoch": 0.6993732245118817,
      "grad_norm": 1.783863693174741,
      "learning_rate": 8.753259484546174e-07,
      "loss": 0.6921,
      "step": 7755
    },
    {
      "epoch": 0.6994634080353519,
      "grad_norm": 1.7706128450643674,
      "learning_rate": 8.748429079200841e-07,
      "loss": 0.7095,
      "step": 7756
    },
    {
      "epoch": 0.6995535915588222,
      "grad_norm": 1.9089507060773026,
      "learning_rate": 8.743599633901575e-07,
      "loss": 0.8113,
      "step": 7757
    },
    {
      "epoch": 0.6996437750822925,
      "grad_norm": 1.9139105288840548,
      "learning_rate": 8.738771149060453e-07,
      "loss": 0.7632,
      "step": 7758
    },
    {
      "epoch": 0.6997339586057627,
      "grad_norm": 2.117011594327045,
      "learning_rate": 8.73394362508947e-07,
      "loss": 0.7181,
      "step": 7759
    },
    {
      "epoch": 0.6998241421292329,
      "grad_norm": 1.409846179295886,
      "learning_rate": 8.72911706240054e-07,
      "loss": 0.7121,
      "step": 7760
    },
    {
      "epoch": 0.6999143256527033,
      "grad_norm": 3.6176214431592837,
      "learning_rate": 8.724291461405493e-07,
      "loss": 0.6288,
      "step": 7761
    },
    {
      "epoch": 0.7000045091761735,
      "grad_norm": 2.048322908060947,
      "learning_rate": 8.71946682251606e-07,
      "loss": 0.7187,
      "step": 7762
    },
    {
      "epoch": 0.7000946926996438,
      "grad_norm": 1.569615982981236,
      "learning_rate": 8.714643146143932e-07,
      "loss": 0.7472,
      "step": 7763
    },
    {
      "epoch": 0.700184876223114,
      "grad_norm": 1.4763676751998054,
      "learning_rate": 8.709820432700663e-07,
      "loss": 0.7058,
      "step": 7764
    },
    {
      "epoch": 0.7002750597465843,
      "grad_norm": 1.4410524011491275,
      "learning_rate": 8.704998682597784e-07,
      "loss": 0.6637,
      "step": 7765
    },
    {
      "epoch": 0.7003652432700546,
      "grad_norm": 2.1417694373897977,
      "learning_rate": 8.700177896246688e-07,
      "loss": 0.7695,
      "step": 7766
    },
    {
      "epoch": 0.7004554267935248,
      "grad_norm": 1.9294196186152108,
      "learning_rate": 8.695358074058721e-07,
      "loss": 0.6788,
      "step": 7767
    },
    {
      "epoch": 0.7005456103169951,
      "grad_norm": 2.0506598736402872,
      "learning_rate": 8.690539216445136e-07,
      "loss": 0.789,
      "step": 7768
    },
    {
      "epoch": 0.7006357938404654,
      "grad_norm": 10.915983300016999,
      "learning_rate": 8.685721323817106e-07,
      "loss": 0.6517,
      "step": 7769
    },
    {
      "epoch": 0.7007259773639356,
      "grad_norm": 1.6053048674387322,
      "learning_rate": 8.680904396585718e-07,
      "loss": 0.7835,
      "step": 7770
    },
    {
      "epoch": 0.7008161608874058,
      "grad_norm": 1.7094423911916719,
      "learning_rate": 8.676088435161988e-07,
      "loss": 0.688,
      "step": 7771
    },
    {
      "epoch": 0.7009063444108762,
      "grad_norm": 1.8494049249299032,
      "learning_rate": 8.671273439956824e-07,
      "loss": 0.7687,
      "step": 7772
    },
    {
      "epoch": 0.7009965279343464,
      "grad_norm": 1.4691355584728427,
      "learning_rate": 8.666459411381075e-07,
      "loss": 0.752,
      "step": 7773
    },
    {
      "epoch": 0.7010867114578166,
      "grad_norm": 1.3575042480065074,
      "learning_rate": 8.661646349845501e-07,
      "loss": 0.7739,
      "step": 7774
    },
    {
      "epoch": 0.7011768949812869,
      "grad_norm": 2.4838641139224755,
      "learning_rate": 8.656834255760783e-07,
      "loss": 0.8087,
      "step": 7775
    },
    {
      "epoch": 0.7012670785047572,
      "grad_norm": 1.6484294005280877,
      "learning_rate": 8.652023129537509e-07,
      "loss": 0.7998,
      "step": 7776
    },
    {
      "epoch": 0.7013572620282275,
      "grad_norm": 3.8040101625324807,
      "learning_rate": 8.647212971586195e-07,
      "loss": 0.7779,
      "step": 7777
    },
    {
      "epoch": 0.7014474455516977,
      "grad_norm": 1.7147611628568247,
      "learning_rate": 8.642403782317275e-07,
      "loss": 0.7051,
      "step": 7778
    },
    {
      "epoch": 0.7015376290751679,
      "grad_norm": 1.7592526788012184,
      "learning_rate": 8.637595562141075e-07,
      "loss": 0.7538,
      "step": 7779
    },
    {
      "epoch": 0.7016278125986383,
      "grad_norm": 1.5066811067316808,
      "learning_rate": 8.632788311467889e-07,
      "loss": 0.7145,
      "step": 7780
    },
    {
      "epoch": 0.7017179961221085,
      "grad_norm": 1.495093347320748,
      "learning_rate": 8.627982030707867e-07,
      "loss": 0.7422,
      "step": 7781
    },
    {
      "epoch": 0.7018081796455787,
      "grad_norm": 1.9442199335214994,
      "learning_rate": 8.623176720271139e-07,
      "loss": 0.8144,
      "step": 7782
    },
    {
      "epoch": 0.701898363169049,
      "grad_norm": 2.8446392303900714,
      "learning_rate": 8.618372380567696e-07,
      "loss": 0.8189,
      "step": 7783
    },
    {
      "epoch": 0.7019885466925193,
      "grad_norm": 1.9518370387147725,
      "learning_rate": 8.613569012007478e-07,
      "loss": 0.7063,
      "step": 7784
    },
    {
      "epoch": 0.7020787302159895,
      "grad_norm": 2.143916140866378,
      "learning_rate": 8.608766615000338e-07,
      "loss": 0.7874,
      "step": 7785
    },
    {
      "epoch": 0.7021689137394598,
      "grad_norm": 1.8109406307479745,
      "learning_rate": 8.603965189956039e-07,
      "loss": 0.7716,
      "step": 7786
    },
    {
      "epoch": 0.70225909726293,
      "grad_norm": 1.739966649141446,
      "learning_rate": 8.599164737284276e-07,
      "loss": 0.754,
      "step": 7787
    },
    {
      "epoch": 0.7023492807864004,
      "grad_norm": 1.7617418917808634,
      "learning_rate": 8.594365257394634e-07,
      "loss": 0.7076,
      "step": 7788
    },
    {
      "epoch": 0.7024394643098706,
      "grad_norm": 1.7084190610759646,
      "learning_rate": 8.589566750696637e-07,
      "loss": 0.6947,
      "step": 7789
    },
    {
      "epoch": 0.7025296478333408,
      "grad_norm": 2.0565099719969986,
      "learning_rate": 8.584769217599721e-07,
      "loss": 0.6828,
      "step": 7790
    },
    {
      "epoch": 0.7026198313568112,
      "grad_norm": 2.2902894361065713,
      "learning_rate": 8.579972658513239e-07,
      "loss": 0.6945,
      "step": 7791
    },
    {
      "epoch": 0.7027100148802814,
      "grad_norm": 2.14078927627951,
      "learning_rate": 8.57517707384646e-07,
      "loss": 0.7529,
      "step": 7792
    },
    {
      "epoch": 0.7028001984037516,
      "grad_norm": 3.2714990962821764,
      "learning_rate": 8.570382464008574e-07,
      "loss": 0.7881,
      "step": 7793
    },
    {
      "epoch": 0.7028903819272219,
      "grad_norm": 1.7087019855978214,
      "learning_rate": 8.565588829408665e-07,
      "loss": 0.7127,
      "step": 7794
    },
    {
      "epoch": 0.7029805654506922,
      "grad_norm": 2.079324935159034,
      "learning_rate": 8.560796170455782e-07,
      "loss": 0.7246,
      "step": 7795
    },
    {
      "epoch": 0.7030707489741624,
      "grad_norm": 1.8936846227660213,
      "learning_rate": 8.556004487558828e-07,
      "loss": 0.7857,
      "step": 7796
    },
    {
      "epoch": 0.7031609324976327,
      "grad_norm": 1.5546022324685789,
      "learning_rate": 8.55121378112669e-07,
      "loss": 0.6633,
      "step": 7797
    },
    {
      "epoch": 0.7032511160211029,
      "grad_norm": 2.0989160146935513,
      "learning_rate": 8.546424051568111e-07,
      "loss": 0.7268,
      "step": 7798
    },
    {
      "epoch": 0.7033412995445733,
      "grad_norm": 1.6778587144244501,
      "learning_rate": 8.541635299291785e-07,
      "loss": 0.7969,
      "step": 7799
    },
    {
      "epoch": 0.7034314830680435,
      "grad_norm": 0.5849091979821394,
      "learning_rate": 8.536847524706317e-07,
      "loss": 0.6002,
      "step": 7800
    },
    {
      "epoch": 0.7035216665915137,
      "grad_norm": 1.6044841633814877,
      "learning_rate": 8.532060728220225e-07,
      "loss": 0.6787,
      "step": 7801
    },
    {
      "epoch": 0.703611850114984,
      "grad_norm": 3.464138918366258,
      "learning_rate": 8.527274910241955e-07,
      "loss": 0.7563,
      "step": 7802
    },
    {
      "epoch": 0.7037020336384543,
      "grad_norm": 1.7408930327297139,
      "learning_rate": 8.522490071179833e-07,
      "loss": 0.7421,
      "step": 7803
    },
    {
      "epoch": 0.7037922171619245,
      "grad_norm": 3.6868575231484186,
      "learning_rate": 8.517706211442159e-07,
      "loss": 0.6296,
      "step": 7804
    },
    {
      "epoch": 0.7038824006853948,
      "grad_norm": 3.6856882849543986,
      "learning_rate": 8.512923331437097e-07,
      "loss": 0.826,
      "step": 7805
    },
    {
      "epoch": 0.703972584208865,
      "grad_norm": 1.5963129864889058,
      "learning_rate": 8.508141431572755e-07,
      "loss": 0.7168,
      "step": 7806
    },
    {
      "epoch": 0.7040627677323353,
      "grad_norm": 2.1530122636090527,
      "learning_rate": 8.503360512257152e-07,
      "loss": 0.7347,
      "step": 7807
    },
    {
      "epoch": 0.7041529512558056,
      "grad_norm": 2.973967615080789,
      "learning_rate": 8.498580573898219e-07,
      "loss": 0.715,
      "step": 7808
    },
    {
      "epoch": 0.7042431347792758,
      "grad_norm": 1.9980881016450263,
      "learning_rate": 8.493801616903813e-07,
      "loss": 0.7579,
      "step": 7809
    },
    {
      "epoch": 0.704333318302746,
      "grad_norm": 2.070220484578115,
      "learning_rate": 8.489023641681705e-07,
      "loss": 0.769,
      "step": 7810
    },
    {
      "epoch": 0.7044235018262164,
      "grad_norm": 1.5601066570012072,
      "learning_rate": 8.484246648639555e-07,
      "loss": 0.7953,
      "step": 7811
    },
    {
      "epoch": 0.7045136853496866,
      "grad_norm": 1.5264239651498095,
      "learning_rate": 8.479470638184994e-07,
      "loss": 0.6264,
      "step": 7812
    },
    {
      "epoch": 0.7046038688731568,
      "grad_norm": 4.296077251940271,
      "learning_rate": 8.474695610725513e-07,
      "loss": 0.6684,
      "step": 7813
    },
    {
      "epoch": 0.7046940523966272,
      "grad_norm": 1.4244371741577648,
      "learning_rate": 8.469921566668552e-07,
      "loss": 0.8204,
      "step": 7814
    },
    {
      "epoch": 0.7047842359200974,
      "grad_norm": 1.7895586416948788,
      "learning_rate": 8.46514850642146e-07,
      "loss": 0.7973,
      "step": 7815
    },
    {
      "epoch": 0.7048744194435677,
      "grad_norm": 1.9158625643661267,
      "learning_rate": 8.460376430391499e-07,
      "loss": 0.735,
      "step": 7816
    },
    {
      "epoch": 0.7049646029670379,
      "grad_norm": 1.2970434332246563,
      "learning_rate": 8.455605338985858e-07,
      "loss": 0.7762,
      "step": 7817
    },
    {
      "epoch": 0.7050547864905082,
      "grad_norm": 1.6055933400466929,
      "learning_rate": 8.45083523261161e-07,
      "loss": 0.6062,
      "step": 7818
    },
    {
      "epoch": 0.7051449700139785,
      "grad_norm": 1.5124569059175506,
      "learning_rate": 8.446066111675796e-07,
      "loss": 0.7524,
      "step": 7819
    },
    {
      "epoch": 0.7052351535374487,
      "grad_norm": 2.20240111667038,
      "learning_rate": 8.441297976585314e-07,
      "loss": 0.6433,
      "step": 7820
    },
    {
      "epoch": 0.7053253370609189,
      "grad_norm": 2.031498937758164,
      "learning_rate": 8.436530827747037e-07,
      "loss": 0.8302,
      "step": 7821
    },
    {
      "epoch": 0.7054155205843893,
      "grad_norm": 1.8986935540976133,
      "learning_rate": 8.431764665567704e-07,
      "loss": 0.8806,
      "step": 7822
    },
    {
      "epoch": 0.7055057041078595,
      "grad_norm": 2.434258481960087,
      "learning_rate": 8.426999490453996e-07,
      "loss": 0.6881,
      "step": 7823
    },
    {
      "epoch": 0.7055958876313297,
      "grad_norm": 1.9815611826566375,
      "learning_rate": 8.422235302812504e-07,
      "loss": 0.756,
      "step": 7824
    },
    {
      "epoch": 0.7056860711548,
      "grad_norm": 1.9858758612190455,
      "learning_rate": 8.417472103049734e-07,
      "loss": 0.8177,
      "step": 7825
    },
    {
      "epoch": 0.7057762546782703,
      "grad_norm": 1.875936528602309,
      "learning_rate": 8.412709891572112e-07,
      "loss": 0.7833,
      "step": 7826
    },
    {
      "epoch": 0.7058664382017406,
      "grad_norm": 1.9580033071565996,
      "learning_rate": 8.407948668785978e-07,
      "loss": 0.7837,
      "step": 7827
    },
    {
      "epoch": 0.7059566217252108,
      "grad_norm": 2.227935260756705,
      "learning_rate": 8.403188435097576e-07,
      "loss": 0.7397,
      "step": 7828
    },
    {
      "epoch": 0.706046805248681,
      "grad_norm": 1.571036935591832,
      "learning_rate": 8.398429190913081e-07,
      "loss": 0.8184,
      "step": 7829
    },
    {
      "epoch": 0.7061369887721514,
      "grad_norm": 1.9047894745489053,
      "learning_rate": 8.393670936638578e-07,
      "loss": 0.7469,
      "step": 7830
    },
    {
      "epoch": 0.7062271722956216,
      "grad_norm": 11.781278493831591,
      "learning_rate": 8.388913672680067e-07,
      "loss": 0.7428,
      "step": 7831
    },
    {
      "epoch": 0.7063173558190918,
      "grad_norm": 2.105999202115885,
      "learning_rate": 8.384157399443472e-07,
      "loss": 0.6981,
      "step": 7832
    },
    {
      "epoch": 0.7064075393425621,
      "grad_norm": 1.6710645904065675,
      "learning_rate": 8.379402117334601e-07,
      "loss": 0.7758,
      "step": 7833
    },
    {
      "epoch": 0.7064977228660324,
      "grad_norm": 1.8283877061718066,
      "learning_rate": 8.374647826759232e-07,
      "loss": 0.7462,
      "step": 7834
    },
    {
      "epoch": 0.7065879063895026,
      "grad_norm": 1.552598095557487,
      "learning_rate": 8.369894528122998e-07,
      "loss": 0.7412,
      "step": 7835
    },
    {
      "epoch": 0.7066780899129729,
      "grad_norm": 1.7695084795634144,
      "learning_rate": 8.365142221831505e-07,
      "loss": 0.6081,
      "step": 7836
    },
    {
      "epoch": 0.7067682734364432,
      "grad_norm": 1.9703500089168293,
      "learning_rate": 8.360390908290222e-07,
      "loss": 0.711,
      "step": 7837
    },
    {
      "epoch": 0.7068584569599135,
      "grad_norm": 1.6435263274857492,
      "learning_rate": 8.355640587904569e-07,
      "loss": 0.7278,
      "step": 7838
    },
    {
      "epoch": 0.7069486404833837,
      "grad_norm": 4.6054277491018425,
      "learning_rate": 8.350891261079866e-07,
      "loss": 0.7545,
      "step": 7839
    },
    {
      "epoch": 0.7070388240068539,
      "grad_norm": 1.684942497381564,
      "learning_rate": 8.346142928221356e-07,
      "loss": 0.799,
      "step": 7840
    },
    {
      "epoch": 0.7071290075303243,
      "grad_norm": 1.8557672200319382,
      "learning_rate": 8.341395589734189e-07,
      "loss": 0.5938,
      "step": 7841
    },
    {
      "epoch": 0.7072191910537945,
      "grad_norm": 2.2635865610945545,
      "learning_rate": 8.336649246023433e-07,
      "loss": 0.7072,
      "step": 7842
    },
    {
      "epoch": 0.7073093745772647,
      "grad_norm": 1.6387281078561085,
      "learning_rate": 8.331903897494084e-07,
      "loss": 0.6525,
      "step": 7843
    },
    {
      "epoch": 0.707399558100735,
      "grad_norm": 2.0408907473385294,
      "learning_rate": 8.327159544551024e-07,
      "loss": 0.7613,
      "step": 7844
    },
    {
      "epoch": 0.7074897416242053,
      "grad_norm": 5.322825909442413,
      "learning_rate": 8.322416187599073e-07,
      "loss": 0.7077,
      "step": 7845
    },
    {
      "epoch": 0.7075799251476755,
      "grad_norm": 1.7746569100423908,
      "learning_rate": 8.317673827042963e-07,
      "loss": 0.7745,
      "step": 7846
    },
    {
      "epoch": 0.7076701086711458,
      "grad_norm": 1.881423247714941,
      "learning_rate": 8.312932463287339e-07,
      "loss": 0.7379,
      "step": 7847
    },
    {
      "epoch": 0.707760292194616,
      "grad_norm": 2.2071709242652955,
      "learning_rate": 8.308192096736759e-07,
      "loss": 0.7105,
      "step": 7848
    },
    {
      "epoch": 0.7078504757180863,
      "grad_norm": 3.0822080844291184,
      "learning_rate": 8.303452727795703e-07,
      "loss": 0.7256,
      "step": 7849
    },
    {
      "epoch": 0.7079406592415566,
      "grad_norm": 1.9381276621426602,
      "learning_rate": 8.298714356868542e-07,
      "loss": 0.8233,
      "step": 7850
    },
    {
      "epoch": 0.7080308427650268,
      "grad_norm": 4.295956045199387,
      "learning_rate": 8.293976984359605e-07,
      "loss": 0.7566,
      "step": 7851
    },
    {
      "epoch": 0.708121026288497,
      "grad_norm": 2.501252909462765,
      "learning_rate": 8.289240610673092e-07,
      "loss": 0.7476,
      "step": 7852
    },
    {
      "epoch": 0.7082112098119674,
      "grad_norm": 1.5486482417165155,
      "learning_rate": 8.284505236213144e-07,
      "loss": 0.7796,
      "step": 7853
    },
    {
      "epoch": 0.7083013933354376,
      "grad_norm": 1.8333207332293693,
      "learning_rate": 8.279770861383806e-07,
      "loss": 0.7384,
      "step": 7854
    },
    {
      "epoch": 0.7083915768589079,
      "grad_norm": 1.5316114193759358,
      "learning_rate": 8.275037486589042e-07,
      "loss": 0.7425,
      "step": 7855
    },
    {
      "epoch": 0.7084817603823781,
      "grad_norm": 0.6534671817936035,
      "learning_rate": 8.270305112232739e-07,
      "loss": 0.5594,
      "step": 7856
    },
    {
      "epoch": 0.7085719439058484,
      "grad_norm": 1.63010272901749,
      "learning_rate": 8.265573738718665e-07,
      "loss": 0.7439,
      "step": 7857
    },
    {
      "epoch": 0.7086621274293187,
      "grad_norm": 1.8918432929082354,
      "learning_rate": 8.260843366450559e-07,
      "loss": 0.705,
      "step": 7858
    },
    {
      "epoch": 0.7087523109527889,
      "grad_norm": 1.5773438134838165,
      "learning_rate": 8.256113995832017e-07,
      "loss": 0.747,
      "step": 7859
    },
    {
      "epoch": 0.7088424944762591,
      "grad_norm": 1.9575544864676497,
      "learning_rate": 8.251385627266583e-07,
      "loss": 0.7876,
      "step": 7860
    },
    {
      "epoch": 0.7089326779997295,
      "grad_norm": 1.8425297982900222,
      "learning_rate": 8.24665826115771e-07,
      "loss": 0.757,
      "step": 7861
    },
    {
      "epoch": 0.7090228615231997,
      "grad_norm": 3.0373126878793095,
      "learning_rate": 8.241931897908763e-07,
      "loss": 0.7649,
      "step": 7862
    },
    {
      "epoch": 0.7091130450466699,
      "grad_norm": 1.6775354127514437,
      "learning_rate": 8.237206537923016e-07,
      "loss": 0.7727,
      "step": 7863
    },
    {
      "epoch": 0.7092032285701403,
      "grad_norm": 2.243814655530626,
      "learning_rate": 8.232482181603671e-07,
      "loss": 0.739,
      "step": 7864
    },
    {
      "epoch": 0.7092934120936105,
      "grad_norm": 2.555723674284467,
      "learning_rate": 8.227758829353828e-07,
      "loss": 0.7256,
      "step": 7865
    },
    {
      "epoch": 0.7093835956170808,
      "grad_norm": 2.3033495106476316,
      "learning_rate": 8.223036481576522e-07,
      "loss": 0.7198,
      "step": 7866
    },
    {
      "epoch": 0.709473779140551,
      "grad_norm": 1.9225285977145181,
      "learning_rate": 8.218315138674672e-07,
      "loss": 0.759,
      "step": 7867
    },
    {
      "epoch": 0.7095639626640213,
      "grad_norm": 0.6219417373272137,
      "learning_rate": 8.21359480105114e-07,
      "loss": 0.5766,
      "step": 7868
    },
    {
      "epoch": 0.7096541461874916,
      "grad_norm": 1.6740162052705532,
      "learning_rate": 8.208875469108689e-07,
      "loss": 0.7779,
      "step": 7869
    },
    {
      "epoch": 0.7097443297109618,
      "grad_norm": 2.1355390063961264,
      "learning_rate": 8.204157143249997e-07,
      "loss": 0.7367,
      "step": 7870
    },
    {
      "epoch": 0.709834513234432,
      "grad_norm": 1.4986881241421859,
      "learning_rate": 8.199439823877668e-07,
      "loss": 0.7162,
      "step": 7871
    },
    {
      "epoch": 0.7099246967579024,
      "grad_norm": 1.5665207818604292,
      "learning_rate": 8.194723511394186e-07,
      "loss": 0.734,
      "step": 7872
    },
    {
      "epoch": 0.7100148802813726,
      "grad_norm": 1.6725611481942455,
      "learning_rate": 8.190008206202002e-07,
      "loss": 0.6939,
      "step": 7873
    },
    {
      "epoch": 0.7101050638048428,
      "grad_norm": 2.6101214774497907,
      "learning_rate": 8.185293908703423e-07,
      "loss": 0.7953,
      "step": 7874
    },
    {
      "epoch": 0.7101952473283131,
      "grad_norm": 1.6241346402612855,
      "learning_rate": 8.180580619300727e-07,
      "loss": 0.7089,
      "step": 7875
    },
    {
      "epoch": 0.7102854308517834,
      "grad_norm": 1.424644455975944,
      "learning_rate": 8.175868338396057e-07,
      "loss": 0.7025,
      "step": 7876
    },
    {
      "epoch": 0.7103756143752537,
      "grad_norm": 2.6351103673763636,
      "learning_rate": 8.171157066391499e-07,
      "loss": 0.7135,
      "step": 7877
    },
    {
      "epoch": 0.7104657978987239,
      "grad_norm": 2.014596484462519,
      "learning_rate": 8.166446803689045e-07,
      "loss": 0.7751,
      "step": 7878
    },
    {
      "epoch": 0.7105559814221941,
      "grad_norm": 0.6679782978290121,
      "learning_rate": 8.161737550690595e-07,
      "loss": 0.5842,
      "step": 7879
    },
    {
      "epoch": 0.7106461649456645,
      "grad_norm": 1.765533714972913,
      "learning_rate": 8.157029307797976e-07,
      "loss": 0.7747,
      "step": 7880
    },
    {
      "epoch": 0.7107363484691347,
      "grad_norm": 1.6360365144483409,
      "learning_rate": 8.152322075412925e-07,
      "loss": 0.8022,
      "step": 7881
    },
    {
      "epoch": 0.7108265319926049,
      "grad_norm": 1.6720885381071808,
      "learning_rate": 8.147615853937073e-07,
      "loss": 0.6934,
      "step": 7882
    },
    {
      "epoch": 0.7109167155160752,
      "grad_norm": 1.7516339031437762,
      "learning_rate": 8.142910643771992e-07,
      "loss": 0.6661,
      "step": 7883
    },
    {
      "epoch": 0.7110068990395455,
      "grad_norm": 7.591536587864974,
      "learning_rate": 8.138206445319152e-07,
      "loss": 0.7113,
      "step": 7884
    },
    {
      "epoch": 0.7110970825630157,
      "grad_norm": 2.8955881760559357,
      "learning_rate": 8.133503258979944e-07,
      "loss": 0.649,
      "step": 7885
    },
    {
      "epoch": 0.711187266086486,
      "grad_norm": 3.4334773715523323,
      "learning_rate": 8.12880108515567e-07,
      "loss": 0.6705,
      "step": 7886
    },
    {
      "epoch": 0.7112774496099563,
      "grad_norm": 4.327838323580689,
      "learning_rate": 8.124099924247543e-07,
      "loss": 0.718,
      "step": 7887
    },
    {
      "epoch": 0.7113676331334265,
      "grad_norm": 1.9298524979078142,
      "learning_rate": 8.119399776656701e-07,
      "loss": 0.65,
      "step": 7888
    },
    {
      "epoch": 0.7114578166568968,
      "grad_norm": 1.5403925381787058,
      "learning_rate": 8.114700642784167e-07,
      "loss": 0.7444,
      "step": 7889
    },
    {
      "epoch": 0.711548000180367,
      "grad_norm": 1.8735431733604857,
      "learning_rate": 8.110002523030921e-07,
      "loss": 0.8004,
      "step": 7890
    },
    {
      "epoch": 0.7116381837038374,
      "grad_norm": 2.3613934503926646,
      "learning_rate": 8.105305417797808e-07,
      "loss": 0.767,
      "step": 7891
    },
    {
      "epoch": 0.7117283672273076,
      "grad_norm": 3.1220167034320605,
      "learning_rate": 8.100609327485635e-07,
      "loss": 0.7852,
      "step": 7892
    },
    {
      "epoch": 0.7118185507507778,
      "grad_norm": 1.9746892443161892,
      "learning_rate": 8.095914252495082e-07,
      "loss": 0.7982,
      "step": 7893
    },
    {
      "epoch": 0.7119087342742481,
      "grad_norm": 2.146123687457565,
      "learning_rate": 8.091220193226762e-07,
      "loss": 0.7165,
      "step": 7894
    },
    {
      "epoch": 0.7119989177977184,
      "grad_norm": 1.6858167377031577,
      "learning_rate": 8.0865271500812e-07,
      "loss": 0.7286,
      "step": 7895
    },
    {
      "epoch": 0.7120891013211886,
      "grad_norm": 1.899704937861901,
      "learning_rate": 8.081835123458831e-07,
      "loss": 0.7847,
      "step": 7896
    },
    {
      "epoch": 0.7121792848446589,
      "grad_norm": 1.575304059545376,
      "learning_rate": 8.077144113760013e-07,
      "loss": 0.7286,
      "step": 7897
    },
    {
      "epoch": 0.7122694683681291,
      "grad_norm": 1.4216310061193145,
      "learning_rate": 8.072454121384995e-07,
      "loss": 0.7211,
      "step": 7898
    },
    {
      "epoch": 0.7123596518915994,
      "grad_norm": 1.4311765826856016,
      "learning_rate": 8.067765146733958e-07,
      "loss": 0.7862,
      "step": 7899
    },
    {
      "epoch": 0.7124498354150697,
      "grad_norm": 2.347204180716475,
      "learning_rate": 8.063077190206993e-07,
      "loss": 0.8255,
      "step": 7900
    },
    {
      "epoch": 0.7125400189385399,
      "grad_norm": 0.6074617624218976,
      "learning_rate": 8.058390252204101e-07,
      "loss": 0.4937,
      "step": 7901
    },
    {
      "epoch": 0.7126302024620101,
      "grad_norm": 2.1119404204013312,
      "learning_rate": 8.0537043331252e-07,
      "loss": 0.7378,
      "step": 7902
    },
    {
      "epoch": 0.7127203859854805,
      "grad_norm": 2.804394039857119,
      "learning_rate": 8.049019433370121e-07,
      "loss": 0.6827,
      "step": 7903
    },
    {
      "epoch": 0.7128105695089507,
      "grad_norm": 1.5077910742942184,
      "learning_rate": 8.044335553338588e-07,
      "loss": 0.7489,
      "step": 7904
    },
    {
      "epoch": 0.712900753032421,
      "grad_norm": 1.5204589679102671,
      "learning_rate": 8.039652693430281e-07,
      "loss": 0.7533,
      "step": 7905
    },
    {
      "epoch": 0.7129909365558912,
      "grad_norm": 2.9842615555377097,
      "learning_rate": 8.034970854044742e-07,
      "loss": 0.7189,
      "step": 7906
    },
    {
      "epoch": 0.7130811200793615,
      "grad_norm": 1.752634313663238,
      "learning_rate": 8.03029003558148e-07,
      "loss": 0.8365,
      "step": 7907
    },
    {
      "epoch": 0.7131713036028318,
      "grad_norm": 2.0623727239005123,
      "learning_rate": 8.025610238439864e-07,
      "loss": 0.8118,
      "step": 7908
    },
    {
      "epoch": 0.713261487126302,
      "grad_norm": 2.0277657777033453,
      "learning_rate": 8.020931463019207e-07,
      "loss": 0.8253,
      "step": 7909
    },
    {
      "epoch": 0.7133516706497723,
      "grad_norm": 1.7388021652985703,
      "learning_rate": 8.016253709718732e-07,
      "loss": 0.6882,
      "step": 7910
    },
    {
      "epoch": 0.7134418541732426,
      "grad_norm": 2.1061582550986095,
      "learning_rate": 8.011576978937567e-07,
      "loss": 0.6518,
      "step": 7911
    },
    {
      "epoch": 0.7135320376967128,
      "grad_norm": 1.617710121220937,
      "learning_rate": 8.006901271074764e-07,
      "loss": 0.7446,
      "step": 7912
    },
    {
      "epoch": 0.713622221220183,
      "grad_norm": 1.7988575488737668,
      "learning_rate": 8.002226586529261e-07,
      "loss": 0.7876,
      "step": 7913
    },
    {
      "epoch": 0.7137124047436534,
      "grad_norm": 4.998746333310499,
      "learning_rate": 7.997552925699956e-07,
      "loss": 0.8216,
      "step": 7914
    },
    {
      "epoch": 0.7138025882671236,
      "grad_norm": 1.7752391385454067,
      "learning_rate": 7.992880288985606e-07,
      "loss": 0.8383,
      "step": 7915
    },
    {
      "epoch": 0.7138927717905938,
      "grad_norm": 1.7705881360407911,
      "learning_rate": 7.988208676784918e-07,
      "loss": 0.7526,
      "step": 7916
    },
    {
      "epoch": 0.7139829553140641,
      "grad_norm": 3.2367969235558816,
      "learning_rate": 7.983538089496497e-07,
      "loss": 0.7393,
      "step": 7917
    },
    {
      "epoch": 0.7140731388375344,
      "grad_norm": 1.7696562127121742,
      "learning_rate": 7.978868527518864e-07,
      "loss": 0.7069,
      "step": 7918
    },
    {
      "epoch": 0.7141633223610047,
      "grad_norm": 2.061313866064834,
      "learning_rate": 7.974199991250455e-07,
      "loss": 0.7277,
      "step": 7919
    },
    {
      "epoch": 0.7142535058844749,
      "grad_norm": 1.8677021418687099,
      "learning_rate": 7.969532481089616e-07,
      "loss": 0.8454,
      "step": 7920
    },
    {
      "epoch": 0.7143436894079451,
      "grad_norm": 1.9852278070865477,
      "learning_rate": 7.964865997434589e-07,
      "loss": 0.7564,
      "step": 7921
    },
    {
      "epoch": 0.7144338729314155,
      "grad_norm": 1.6909022472332593,
      "learning_rate": 7.96020054068357e-07,
      "loss": 0.8185,
      "step": 7922
    },
    {
      "epoch": 0.7145240564548857,
      "grad_norm": 1.228780987679653,
      "learning_rate": 7.95553611123462e-07,
      "loss": 0.7013,
      "step": 7923
    },
    {
      "epoch": 0.7146142399783559,
      "grad_norm": 1.815314311300626,
      "learning_rate": 7.950872709485741e-07,
      "loss": 0.9321,
      "step": 7924
    },
    {
      "epoch": 0.7147044235018262,
      "grad_norm": 2.2871003057102954,
      "learning_rate": 7.946210335834842e-07,
      "loss": 0.7975,
      "step": 7925
    },
    {
      "epoch": 0.7147946070252965,
      "grad_norm": 1.5331430216871875,
      "learning_rate": 7.94154899067974e-07,
      "loss": 0.6691,
      "step": 7926
    },
    {
      "epoch": 0.7148847905487667,
      "grad_norm": 1.7078299672454176,
      "learning_rate": 7.936888674418177e-07,
      "loss": 0.6723,
      "step": 7927
    },
    {
      "epoch": 0.714974974072237,
      "grad_norm": 2.574484827301781,
      "learning_rate": 7.932229387447771e-07,
      "loss": 0.7504,
      "step": 7928
    },
    {
      "epoch": 0.7150651575957072,
      "grad_norm": 2.5869490843231473,
      "learning_rate": 7.927571130166109e-07,
      "loss": 0.7392,
      "step": 7929
    },
    {
      "epoch": 0.7151553411191776,
      "grad_norm": 0.8534483370237816,
      "learning_rate": 7.922913902970632e-07,
      "loss": 0.568,
      "step": 7930
    },
    {
      "epoch": 0.7152455246426478,
      "grad_norm": 1.809932205237238,
      "learning_rate": 7.918257706258744e-07,
      "loss": 0.7414,
      "step": 7931
    },
    {
      "epoch": 0.715335708166118,
      "grad_norm": 1.9997007622974268,
      "learning_rate": 7.913602540427724e-07,
      "loss": 0.7442,
      "step": 7932
    },
    {
      "epoch": 0.7154258916895884,
      "grad_norm": 1.7808148705907811,
      "learning_rate": 7.908948405874775e-07,
      "loss": 0.7387,
      "step": 7933
    },
    {
      "epoch": 0.7155160752130586,
      "grad_norm": 1.6622104995034244,
      "learning_rate": 7.904295302997019e-07,
      "loss": 0.6759,
      "step": 7934
    },
    {
      "epoch": 0.7156062587365288,
      "grad_norm": 1.5176792970217385,
      "learning_rate": 7.899643232191484e-07,
      "loss": 0.7576,
      "step": 7935
    },
    {
      "epoch": 0.7156964422599991,
      "grad_norm": 1.4821719234065487,
      "learning_rate": 7.894992193855108e-07,
      "loss": 0.7314,
      "step": 7936
    },
    {
      "epoch": 0.7157866257834694,
      "grad_norm": 1.9387296650728276,
      "learning_rate": 7.890342188384751e-07,
      "loss": 0.6966,
      "step": 7937
    },
    {
      "epoch": 0.7158768093069396,
      "grad_norm": 1.864915111440472,
      "learning_rate": 7.885693216177165e-07,
      "loss": 0.6865,
      "step": 7938
    },
    {
      "epoch": 0.7159669928304099,
      "grad_norm": 1.5512569284092246,
      "learning_rate": 7.88104527762903e-07,
      "loss": 0.7407,
      "step": 7939
    },
    {
      "epoch": 0.7160571763538801,
      "grad_norm": 4.016642994369038,
      "learning_rate": 7.876398373136936e-07,
      "loss": 0.7034,
      "step": 7940
    },
    {
      "epoch": 0.7161473598773505,
      "grad_norm": 1.8957071017764606,
      "learning_rate": 7.87175250309738e-07,
      "loss": 0.7399,
      "step": 7941
    },
    {
      "epoch": 0.7162375434008207,
      "grad_norm": 1.6745246497384825,
      "learning_rate": 7.867107667906785e-07,
      "loss": 0.8212,
      "step": 7942
    },
    {
      "epoch": 0.7163277269242909,
      "grad_norm": 1.8070546268677803,
      "learning_rate": 7.862463867961446e-07,
      "loss": 0.7417,
      "step": 7943
    },
    {
      "epoch": 0.7164179104477612,
      "grad_norm": 1.4052524419280468,
      "learning_rate": 7.857821103657632e-07,
      "loss": 0.8089,
      "step": 7944
    },
    {
      "epoch": 0.7165080939712315,
      "grad_norm": 1.729492049645322,
      "learning_rate": 7.853179375391459e-07,
      "loss": 0.7731,
      "step": 7945
    },
    {
      "epoch": 0.7165982774947017,
      "grad_norm": 1.6267341017555177,
      "learning_rate": 7.848538683559012e-07,
      "loss": 0.7307,
      "step": 7946
    },
    {
      "epoch": 0.716688461018172,
      "grad_norm": 1.827962754051678,
      "learning_rate": 7.843899028556238e-07,
      "loss": 0.7844,
      "step": 7947
    },
    {
      "epoch": 0.7167786445416422,
      "grad_norm": 1.934811480306273,
      "learning_rate": 7.839260410779029e-07,
      "loss": 0.7385,
      "step": 7948
    },
    {
      "epoch": 0.7168688280651125,
      "grad_norm": 1.4721956620529602,
      "learning_rate": 7.834622830623175e-07,
      "loss": 0.7623,
      "step": 7949
    },
    {
      "epoch": 0.7169590115885828,
      "grad_norm": 1.7057025548203386,
      "learning_rate": 7.82998628848438e-07,
      "loss": 0.7177,
      "step": 7950
    },
    {
      "epoch": 0.717049195112053,
      "grad_norm": 2.1217775033116855,
      "learning_rate": 7.825350784758261e-07,
      "loss": 0.7346,
      "step": 7951
    },
    {
      "epoch": 0.7171393786355232,
      "grad_norm": 1.4030899995036503,
      "learning_rate": 7.820716319840342e-07,
      "loss": 0.7068,
      "step": 7952
    },
    {
      "epoch": 0.7172295621589936,
      "grad_norm": 2.021605026404016,
      "learning_rate": 7.816082894126069e-07,
      "loss": 0.8234,
      "step": 7953
    },
    {
      "epoch": 0.7173197456824638,
      "grad_norm": 1.9673525376661773,
      "learning_rate": 7.811450508010778e-07,
      "loss": 0.718,
      "step": 7954
    },
    {
      "epoch": 0.717409929205934,
      "grad_norm": 1.5710845112351122,
      "learning_rate": 7.806819161889737e-07,
      "loss": 0.7176,
      "step": 7955
    },
    {
      "epoch": 0.7175001127294043,
      "grad_norm": 1.6191208197437312,
      "learning_rate": 7.802188856158119e-07,
      "loss": 0.7756,
      "step": 7956
    },
    {
      "epoch": 0.7175902962528746,
      "grad_norm": 1.9619722106124373,
      "learning_rate": 7.797559591211002e-07,
      "loss": 0.7286,
      "step": 7957
    },
    {
      "epoch": 0.7176804797763449,
      "grad_norm": 1.6150808572752973,
      "learning_rate": 7.79293136744339e-07,
      "loss": 0.7089,
      "step": 7958
    },
    {
      "epoch": 0.7177706632998151,
      "grad_norm": 1.8182422833792298,
      "learning_rate": 7.788304185250185e-07,
      "loss": 0.7421,
      "step": 7959
    },
    {
      "epoch": 0.7178608468232854,
      "grad_norm": 1.6695094262902639,
      "learning_rate": 7.78367804502619e-07,
      "loss": 0.7711,
      "step": 7960
    },
    {
      "epoch": 0.7179510303467557,
      "grad_norm": 2.422867799125614,
      "learning_rate": 7.779052947166156e-07,
      "loss": 0.755,
      "step": 7961
    },
    {
      "epoch": 0.7180412138702259,
      "grad_norm": 3.4816754065793933,
      "learning_rate": 7.774428892064697e-07,
      "loss": 0.7001,
      "step": 7962
    },
    {
      "epoch": 0.7181313973936961,
      "grad_norm": 1.488469870321649,
      "learning_rate": 7.769805880116391e-07,
      "loss": 0.7371,
      "step": 7963
    },
    {
      "epoch": 0.7182215809171665,
      "grad_norm": 2.3358773826594725,
      "learning_rate": 7.765183911715678e-07,
      "loss": 0.725,
      "step": 7964
    },
    {
      "epoch": 0.7183117644406367,
      "grad_norm": 1.7273071892508285,
      "learning_rate": 7.760562987256933e-07,
      "loss": 0.6453,
      "step": 7965
    },
    {
      "epoch": 0.718401947964107,
      "grad_norm": 2.120064220355458,
      "learning_rate": 7.755943107134444e-07,
      "loss": 0.7499,
      "step": 7966
    },
    {
      "epoch": 0.7184921314875772,
      "grad_norm": 0.6369113279843696,
      "learning_rate": 7.751324271742401e-07,
      "loss": 0.5687,
      "step": 7967
    },
    {
      "epoch": 0.7185823150110475,
      "grad_norm": 1.753916513119854,
      "learning_rate": 7.746706481474916e-07,
      "loss": 0.7591,
      "step": 7968
    },
    {
      "epoch": 0.7186724985345178,
      "grad_norm": 0.5978248707378214,
      "learning_rate": 7.742089736725992e-07,
      "loss": 0.5371,
      "step": 7969
    },
    {
      "epoch": 0.718762682057988,
      "grad_norm": 1.8729142986391547,
      "learning_rate": 7.737474037889559e-07,
      "loss": 0.6639,
      "step": 7970
    },
    {
      "epoch": 0.7188528655814582,
      "grad_norm": 1.5219508032748994,
      "learning_rate": 7.732859385359458e-07,
      "loss": 0.712,
      "step": 7971
    },
    {
      "epoch": 0.7189430491049286,
      "grad_norm": 1.673536170593656,
      "learning_rate": 7.728245779529434e-07,
      "loss": 0.6644,
      "step": 7972
    },
    {
      "epoch": 0.7190332326283988,
      "grad_norm": 3.1694880180992957,
      "learning_rate": 7.723633220793146e-07,
      "loss": 0.7714,
      "step": 7973
    },
    {
      "epoch": 0.719123416151869,
      "grad_norm": 1.8352034379617603,
      "learning_rate": 7.719021709544162e-07,
      "loss": 0.6345,
      "step": 7974
    },
    {
      "epoch": 0.7192135996753393,
      "grad_norm": 1.8139421710603598,
      "learning_rate": 7.714411246175964e-07,
      "loss": 0.8337,
      "step": 7975
    },
    {
      "epoch": 0.7193037831988096,
      "grad_norm": 1.649883413530792,
      "learning_rate": 7.709801831081946e-07,
      "loss": 0.664,
      "step": 7976
    },
    {
      "epoch": 0.7193939667222798,
      "grad_norm": 3.3263580933173786,
      "learning_rate": 7.705193464655391e-07,
      "loss": 0.7041,
      "step": 7977
    },
    {
      "epoch": 0.7194841502457501,
      "grad_norm": 1.3101741073019253,
      "learning_rate": 7.700586147289534e-07,
      "loss": 0.7177,
      "step": 7978
    },
    {
      "epoch": 0.7195743337692203,
      "grad_norm": 2.345049891805322,
      "learning_rate": 7.695979879377481e-07,
      "loss": 0.6999,
      "step": 7979
    },
    {
      "epoch": 0.7196645172926907,
      "grad_norm": 2.3482415458261414,
      "learning_rate": 7.691374661312266e-07,
      "loss": 0.7653,
      "step": 7980
    },
    {
      "epoch": 0.7197547008161609,
      "grad_norm": 1.6709787693923612,
      "learning_rate": 7.686770493486834e-07,
      "loss": 0.7194,
      "step": 7981
    },
    {
      "epoch": 0.7198448843396311,
      "grad_norm": 1.785784254065729,
      "learning_rate": 7.68216737629404e-07,
      "loss": 0.6901,
      "step": 7982
    },
    {
      "epoch": 0.7199350678631015,
      "grad_norm": 1.3974271992810017,
      "learning_rate": 7.67756531012665e-07,
      "loss": 0.8091,
      "step": 7983
    },
    {
      "epoch": 0.7200252513865717,
      "grad_norm": 1.4858988920344625,
      "learning_rate": 7.67296429537732e-07,
      "loss": 0.8007,
      "step": 7984
    },
    {
      "epoch": 0.7201154349100419,
      "grad_norm": 1.7105371015417152,
      "learning_rate": 7.668364332438661e-07,
      "loss": 0.8097,
      "step": 7985
    },
    {
      "epoch": 0.7202056184335122,
      "grad_norm": 1.665819914615712,
      "learning_rate": 7.663765421703145e-07,
      "loss": 0.7554,
      "step": 7986
    },
    {
      "epoch": 0.7202958019569825,
      "grad_norm": 1.7960338448137074,
      "learning_rate": 7.659167563563187e-07,
      "loss": 0.679,
      "step": 7987
    },
    {
      "epoch": 0.7203859854804527,
      "grad_norm": 1.939288882458956,
      "learning_rate": 7.654570758411096e-07,
      "loss": 0.7711,
      "step": 7988
    },
    {
      "epoch": 0.720476169003923,
      "grad_norm": 2.2752917092233576,
      "learning_rate": 7.649975006639103e-07,
      "loss": 0.8012,
      "step": 7989
    },
    {
      "epoch": 0.7205663525273932,
      "grad_norm": 0.6201013277684013,
      "learning_rate": 7.645380308639337e-07,
      "loss": 0.5516,
      "step": 7990
    },
    {
      "epoch": 0.7206565360508635,
      "grad_norm": 1.9322170095767655,
      "learning_rate": 7.640786664803853e-07,
      "loss": 0.7876,
      "step": 7991
    },
    {
      "epoch": 0.7207467195743338,
      "grad_norm": 1.6239036382828331,
      "learning_rate": 7.636194075524587e-07,
      "loss": 0.6627,
      "step": 7992
    },
    {
      "epoch": 0.720836903097804,
      "grad_norm": 1.6663316469442786,
      "learning_rate": 7.631602541193429e-07,
      "loss": 0.6971,
      "step": 7993
    },
    {
      "epoch": 0.7209270866212742,
      "grad_norm": 1.8880934988485838,
      "learning_rate": 7.627012062202132e-07,
      "loss": 0.7943,
      "step": 7994
    },
    {
      "epoch": 0.7210172701447446,
      "grad_norm": 1.8947424702044884,
      "learning_rate": 7.622422638942391e-07,
      "loss": 0.742,
      "step": 7995
    },
    {
      "epoch": 0.7211074536682148,
      "grad_norm": 22.239373916485725,
      "learning_rate": 7.617834271805801e-07,
      "loss": 0.8065,
      "step": 7996
    },
    {
      "epoch": 0.7211976371916851,
      "grad_norm": 0.6480641956555491,
      "learning_rate": 7.613246961183863e-07,
      "loss": 0.5653,
      "step": 7997
    },
    {
      "epoch": 0.7212878207151553,
      "grad_norm": 1.5150731448654888,
      "learning_rate": 7.608660707468002e-07,
      "loss": 0.7002,
      "step": 7998
    },
    {
      "epoch": 0.7213780042386256,
      "grad_norm": 1.7404935667318513,
      "learning_rate": 7.604075511049522e-07,
      "loss": 0.6366,
      "step": 7999
    },
    {
      "epoch": 0.7214681877620959,
      "grad_norm": 1.8831615203648457,
      "learning_rate": 7.599491372319682e-07,
      "loss": 0.6854,
      "step": 8000
    },
    {
      "epoch": 0.7215583712855661,
      "grad_norm": 2.8247192277043305,
      "learning_rate": 7.594908291669601e-07,
      "loss": 0.7422,
      "step": 8001
    },
    {
      "epoch": 0.7216485548090363,
      "grad_norm": 1.8009358834096076,
      "learning_rate": 7.590326269490359e-07,
      "loss": 0.7739,
      "step": 8002
    },
    {
      "epoch": 0.7217387383325067,
      "grad_norm": 1.572323702100885,
      "learning_rate": 7.585745306172899e-07,
      "loss": 0.7002,
      "step": 8003
    },
    {
      "epoch": 0.7218289218559769,
      "grad_norm": 2.2571559070085745,
      "learning_rate": 7.5811654021081e-07,
      "loss": 0.7231,
      "step": 8004
    },
    {
      "epoch": 0.7219191053794471,
      "grad_norm": 1.3802016875868772,
      "learning_rate": 7.576586557686748e-07,
      "loss": 0.7714,
      "step": 8005
    },
    {
      "epoch": 0.7220092889029175,
      "grad_norm": 3.7195554229713617,
      "learning_rate": 7.572008773299531e-07,
      "loss": 0.7098,
      "step": 8006
    },
    {
      "epoch": 0.7220994724263877,
      "grad_norm": 1.9530916134841785,
      "learning_rate": 7.567432049337055e-07,
      "loss": 0.7789,
      "step": 8007
    },
    {
      "epoch": 0.722189655949858,
      "grad_norm": 1.6688328811694169,
      "learning_rate": 7.562856386189834e-07,
      "loss": 0.6779,
      "step": 8008
    },
    {
      "epoch": 0.7222798394733282,
      "grad_norm": 2.82759240968295,
      "learning_rate": 7.558281784248275e-07,
      "loss": 0.7591,
      "step": 8009
    },
    {
      "epoch": 0.7223700229967985,
      "grad_norm": 1.7695278816106876,
      "learning_rate": 7.553708243902721e-07,
      "loss": 0.7345,
      "step": 8010
    },
    {
      "epoch": 0.7224602065202688,
      "grad_norm": 2.485091483966999,
      "learning_rate": 7.549135765543404e-07,
      "loss": 0.6319,
      "step": 8011
    },
    {
      "epoch": 0.722550390043739,
      "grad_norm": 2.4116502085424902,
      "learning_rate": 7.544564349560481e-07,
      "loss": 0.7644,
      "step": 8012
    },
    {
      "epoch": 0.7226405735672092,
      "grad_norm": 1.6259620459656043,
      "learning_rate": 7.539993996344009e-07,
      "loss": 0.723,
      "step": 8013
    },
    {
      "epoch": 0.7227307570906796,
      "grad_norm": 1.5695863629581845,
      "learning_rate": 7.535424706283941e-07,
      "loss": 0.7185,
      "step": 8014
    },
    {
      "epoch": 0.7228209406141498,
      "grad_norm": 1.6075839845066981,
      "learning_rate": 7.530856479770181e-07,
      "loss": 0.7039,
      "step": 8015
    },
    {
      "epoch": 0.72291112413762,
      "grad_norm": 1.6499433825633754,
      "learning_rate": 7.526289317192484e-07,
      "loss": 0.8223,
      "step": 8016
    },
    {
      "epoch": 0.7230013076610903,
      "grad_norm": 2.475513416559811,
      "learning_rate": 7.521723218940579e-07,
      "loss": 0.74,
      "step": 8017
    },
    {
      "epoch": 0.7230914911845606,
      "grad_norm": 1.8595909826674233,
      "learning_rate": 7.517158185404038e-07,
      "loss": 0.8045,
      "step": 8018
    },
    {
      "epoch": 0.7231816747080309,
      "grad_norm": 2.280570987345124,
      "learning_rate": 7.512594216972403e-07,
      "loss": 0.768,
      "step": 8019
    },
    {
      "epoch": 0.7232718582315011,
      "grad_norm": 11.085812590139318,
      "learning_rate": 7.508031314035078e-07,
      "loss": 0.778,
      "step": 8020
    },
    {
      "epoch": 0.7233620417549713,
      "grad_norm": 2.0585815892593256,
      "learning_rate": 7.503469476981401e-07,
      "loss": 0.7647,
      "step": 8021
    },
    {
      "epoch": 0.7234522252784417,
      "grad_norm": 1.7001620776386228,
      "learning_rate": 7.498908706200613e-07,
      "loss": 0.6162,
      "step": 8022
    },
    {
      "epoch": 0.7235424088019119,
      "grad_norm": 1.3934462143530773,
      "learning_rate": 7.494349002081866e-07,
      "loss": 0.7866,
      "step": 8023
    },
    {
      "epoch": 0.7236325923253821,
      "grad_norm": 1.6089789634737688,
      "learning_rate": 7.489790365014224e-07,
      "loss": 0.7746,
      "step": 8024
    },
    {
      "epoch": 0.7237227758488524,
      "grad_norm": 1.6301057273416384,
      "learning_rate": 7.485232795386642e-07,
      "loss": 0.7579,
      "step": 8025
    },
    {
      "epoch": 0.7238129593723227,
      "grad_norm": 1.482380621415241,
      "learning_rate": 7.480676293588002e-07,
      "loss": 0.7602,
      "step": 8026
    },
    {
      "epoch": 0.7239031428957929,
      "grad_norm": 0.8170034241365282,
      "learning_rate": 7.476120860007093e-07,
      "loss": 0.5967,
      "step": 8027
    },
    {
      "epoch": 0.7239933264192632,
      "grad_norm": 2.0227715416292678,
      "learning_rate": 7.471566495032608e-07,
      "loss": 0.7842,
      "step": 8028
    },
    {
      "epoch": 0.7240835099427335,
      "grad_norm": 1.6466500955718262,
      "learning_rate": 7.467013199053152e-07,
      "loss": 0.6757,
      "step": 8029
    },
    {
      "epoch": 0.7241736934662037,
      "grad_norm": 2.062958753024623,
      "learning_rate": 7.46246097245724e-07,
      "loss": 0.7434,
      "step": 8030
    },
    {
      "epoch": 0.724263876989674,
      "grad_norm": 2.7950985344184476,
      "learning_rate": 7.457909815633276e-07,
      "loss": 0.7484,
      "step": 8031
    },
    {
      "epoch": 0.7243540605131442,
      "grad_norm": 1.8972679271687958,
      "learning_rate": 7.453359728969618e-07,
      "loss": 0.7331,
      "step": 8032
    },
    {
      "epoch": 0.7244442440366146,
      "grad_norm": 1.8324588366122765,
      "learning_rate": 7.448810712854475e-07,
      "loss": 0.6657,
      "step": 8033
    },
    {
      "epoch": 0.7245344275600848,
      "grad_norm": 2.3121793369974544,
      "learning_rate": 7.444262767676022e-07,
      "loss": 0.7833,
      "step": 8034
    },
    {
      "epoch": 0.724624611083555,
      "grad_norm": 1.6234928624660316,
      "learning_rate": 7.439715893822296e-07,
      "loss": 0.8159,
      "step": 8035
    },
    {
      "epoch": 0.7247147946070253,
      "grad_norm": 1.9929916853054188,
      "learning_rate": 7.435170091681264e-07,
      "loss": 0.7669,
      "step": 8036
    },
    {
      "epoch": 0.7248049781304956,
      "grad_norm": 3.730002425042149,
      "learning_rate": 7.430625361640803e-07,
      "loss": 0.7054,
      "step": 8037
    },
    {
      "epoch": 0.7248951616539658,
      "grad_norm": 3.131679111095293,
      "learning_rate": 7.426081704088694e-07,
      "loss": 0.7833,
      "step": 8038
    },
    {
      "epoch": 0.7249853451774361,
      "grad_norm": 3.15809350893954,
      "learning_rate": 7.42153911941263e-07,
      "loss": 0.8118,
      "step": 8039
    },
    {
      "epoch": 0.7250755287009063,
      "grad_norm": 1.8232740288085627,
      "learning_rate": 7.416997608000192e-07,
      "loss": 0.7826,
      "step": 8040
    },
    {
      "epoch": 0.7251657122243766,
      "grad_norm": 1.9204231298625785,
      "learning_rate": 7.412457170238918e-07,
      "loss": 0.7277,
      "step": 8041
    },
    {
      "epoch": 0.7252558957478469,
      "grad_norm": 1.8414741953508185,
      "learning_rate": 7.407917806516193e-07,
      "loss": 0.6913,
      "step": 8042
    },
    {
      "epoch": 0.7253460792713171,
      "grad_norm": 1.980844977342883,
      "learning_rate": 7.403379517219354e-07,
      "loss": 0.7348,
      "step": 8043
    },
    {
      "epoch": 0.7254362627947873,
      "grad_norm": 1.5793862165891324,
      "learning_rate": 7.398842302735636e-07,
      "loss": 0.7742,
      "step": 8044
    },
    {
      "epoch": 0.7255264463182577,
      "grad_norm": 0.6672174344282324,
      "learning_rate": 7.394306163452171e-07,
      "loss": 0.5462,
      "step": 8045
    },
    {
      "epoch": 0.7256166298417279,
      "grad_norm": 1.367194170254376,
      "learning_rate": 7.38977109975601e-07,
      "loss": 0.7219,
      "step": 8046
    },
    {
      "epoch": 0.7257068133651982,
      "grad_norm": 1.9781570335554524,
      "learning_rate": 7.385237112034119e-07,
      "loss": 0.658,
      "step": 8047
    },
    {
      "epoch": 0.7257969968886684,
      "grad_norm": 1.604979057531894,
      "learning_rate": 7.380704200673342e-07,
      "loss": 0.8066,
      "step": 8048
    },
    {
      "epoch": 0.7258871804121387,
      "grad_norm": 1.8266403821228212,
      "learning_rate": 7.376172366060478e-07,
      "loss": 0.767,
      "step": 8049
    },
    {
      "epoch": 0.725977363935609,
      "grad_norm": 0.6420271346071036,
      "learning_rate": 7.371641608582187e-07,
      "loss": 0.5126,
      "step": 8050
    },
    {
      "epoch": 0.7260675474590792,
      "grad_norm": 2.276859709661904,
      "learning_rate": 7.367111928625067e-07,
      "loss": 0.6158,
      "step": 8051
    },
    {
      "epoch": 0.7261577309825495,
      "grad_norm": 1.8446262023346958,
      "learning_rate": 7.362583326575613e-07,
      "loss": 0.6576,
      "step": 8052
    },
    {
      "epoch": 0.7262479145060198,
      "grad_norm": 1.8980981342925305,
      "learning_rate": 7.358055802820234e-07,
      "loss": 0.726,
      "step": 8053
    },
    {
      "epoch": 0.72633809802949,
      "grad_norm": 6.280188019968997,
      "learning_rate": 7.353529357745245e-07,
      "loss": 0.7141,
      "step": 8054
    },
    {
      "epoch": 0.7264282815529602,
      "grad_norm": 2.317106967588952,
      "learning_rate": 7.349003991736851e-07,
      "loss": 0.7845,
      "step": 8055
    },
    {
      "epoch": 0.7265184650764306,
      "grad_norm": 1.8749190630928054,
      "learning_rate": 7.344479705181206e-07,
      "loss": 0.5949,
      "step": 8056
    },
    {
      "epoch": 0.7266086485999008,
      "grad_norm": 1.5650459147533835,
      "learning_rate": 7.339956498464322e-07,
      "loss": 0.6931,
      "step": 8057
    },
    {
      "epoch": 0.726698832123371,
      "grad_norm": 1.8390661325475555,
      "learning_rate": 7.335434371972169e-07,
      "loss": 0.79,
      "step": 8058
    },
    {
      "epoch": 0.7267890156468413,
      "grad_norm": 1.5003120574600912,
      "learning_rate": 7.33091332609058e-07,
      "loss": 0.731,
      "step": 8059
    },
    {
      "epoch": 0.7268791991703116,
      "grad_norm": 1.7089409871724475,
      "learning_rate": 7.326393361205323e-07,
      "loss": 0.7605,
      "step": 8060
    },
    {
      "epoch": 0.7269693826937819,
      "grad_norm": 1.748450887865725,
      "learning_rate": 7.321874477702068e-07,
      "loss": 0.8216,
      "step": 8061
    },
    {
      "epoch": 0.7270595662172521,
      "grad_norm": 0.749837619369319,
      "learning_rate": 7.317356675966386e-07,
      "loss": 0.5919,
      "step": 8062
    },
    {
      "epoch": 0.7271497497407223,
      "grad_norm": 1.4330047312070764,
      "learning_rate": 7.312839956383765e-07,
      "loss": 0.7775,
      "step": 8063
    },
    {
      "epoch": 0.7272399332641927,
      "grad_norm": 1.779095199662365,
      "learning_rate": 7.308324319339603e-07,
      "loss": 0.7199,
      "step": 8064
    },
    {
      "epoch": 0.7273301167876629,
      "grad_norm": 2.0360633746777985,
      "learning_rate": 7.303809765219182e-07,
      "loss": 0.7434,
      "step": 8065
    },
    {
      "epoch": 0.7274203003111331,
      "grad_norm": 2.2142868283704074,
      "learning_rate": 7.299296294407719e-07,
      "loss": 0.7876,
      "step": 8066
    },
    {
      "epoch": 0.7275104838346034,
      "grad_norm": 1.6323540148809852,
      "learning_rate": 7.294783907290327e-07,
      "loss": 0.7854,
      "step": 8067
    },
    {
      "epoch": 0.7276006673580737,
      "grad_norm": 2.0850393812087593,
      "learning_rate": 7.290272604252028e-07,
      "loss": 0.6114,
      "step": 8068
    },
    {
      "epoch": 0.727690850881544,
      "grad_norm": 2.481935274827903,
      "learning_rate": 7.285762385677758e-07,
      "loss": 0.6491,
      "step": 8069
    },
    {
      "epoch": 0.7277810344050142,
      "grad_norm": 1.7872450480006326,
      "learning_rate": 7.281253251952335e-07,
      "loss": 0.7798,
      "step": 8070
    },
    {
      "epoch": 0.7278712179284844,
      "grad_norm": 1.5814126394712094,
      "learning_rate": 7.276745203460526e-07,
      "loss": 0.7197,
      "step": 8071
    },
    {
      "epoch": 0.7279614014519548,
      "grad_norm": 0.7859698961986777,
      "learning_rate": 7.272238240586959e-07,
      "loss": 0.632,
      "step": 8072
    },
    {
      "epoch": 0.728051584975425,
      "grad_norm": 1.2998249559663617,
      "learning_rate": 7.267732363716219e-07,
      "loss": 0.6771,
      "step": 8073
    },
    {
      "epoch": 0.7281417684988952,
      "grad_norm": 2.195584317684755,
      "learning_rate": 7.263227573232753e-07,
      "loss": 0.6964,
      "step": 8074
    },
    {
      "epoch": 0.7282319520223655,
      "grad_norm": 1.7591215557540931,
      "learning_rate": 7.258723869520937e-07,
      "loss": 0.8309,
      "step": 8075
    },
    {
      "epoch": 0.7283221355458358,
      "grad_norm": 2.8245141176967263,
      "learning_rate": 7.254221252965059e-07,
      "loss": 0.7901,
      "step": 8076
    },
    {
      "epoch": 0.728412319069306,
      "grad_norm": 2.1895172626993484,
      "learning_rate": 7.249719723949301e-07,
      "loss": 0.6581,
      "step": 8077
    },
    {
      "epoch": 0.7285025025927763,
      "grad_norm": 1.4320460808335098,
      "learning_rate": 7.245219282857761e-07,
      "loss": 0.7648,
      "step": 8078
    },
    {
      "epoch": 0.7285926861162466,
      "grad_norm": 1.5466715360981373,
      "learning_rate": 7.240719930074442e-07,
      "loss": 0.783,
      "step": 8079
    },
    {
      "epoch": 0.7286828696397168,
      "grad_norm": 2.4094464757784335,
      "learning_rate": 7.236221665983257e-07,
      "loss": 0.7687,
      "step": 8080
    },
    {
      "epoch": 0.7287730531631871,
      "grad_norm": 0.7308814367066658,
      "learning_rate": 7.231724490968012e-07,
      "loss": 0.5573,
      "step": 8081
    },
    {
      "epoch": 0.7288632366866573,
      "grad_norm": 2.068616409828846,
      "learning_rate": 7.227228405412438e-07,
      "loss": 0.7123,
      "step": 8082
    },
    {
      "epoch": 0.7289534202101277,
      "grad_norm": 2.2958143211542166,
      "learning_rate": 7.222733409700165e-07,
      "loss": 0.701,
      "step": 8083
    },
    {
      "epoch": 0.7290436037335979,
      "grad_norm": 1.6140124716672761,
      "learning_rate": 7.21823950421473e-07,
      "loss": 0.7521,
      "step": 8084
    },
    {
      "epoch": 0.7291337872570681,
      "grad_norm": 1.6317619338910545,
      "learning_rate": 7.213746689339577e-07,
      "loss": 0.7162,
      "step": 8085
    },
    {
      "epoch": 0.7292239707805384,
      "grad_norm": 3.2583166962676096,
      "learning_rate": 7.20925496545807e-07,
      "loss": 0.7434,
      "step": 8086
    },
    {
      "epoch": 0.7293141543040087,
      "grad_norm": 2.417941535463771,
      "learning_rate": 7.20476433295344e-07,
      "loss": 0.6529,
      "step": 8087
    },
    {
      "epoch": 0.7294043378274789,
      "grad_norm": 1.4641697668231992,
      "learning_rate": 7.200274792208882e-07,
      "loss": 0.7826,
      "step": 8088
    },
    {
      "epoch": 0.7294945213509492,
      "grad_norm": 1.5834821664560592,
      "learning_rate": 7.195786343607444e-07,
      "loss": 0.6315,
      "step": 8089
    },
    {
      "epoch": 0.7295847048744194,
      "grad_norm": 2.1398680525058102,
      "learning_rate": 7.191298987532131e-07,
      "loss": 0.7379,
      "step": 8090
    },
    {
      "epoch": 0.7296748883978897,
      "grad_norm": 2.178043087003279,
      "learning_rate": 7.186812724365805e-07,
      "loss": 0.7507,
      "step": 8091
    },
    {
      "epoch": 0.72976507192136,
      "grad_norm": 1.8934698218637112,
      "learning_rate": 7.182327554491272e-07,
      "loss": 0.7547,
      "step": 8092
    },
    {
      "epoch": 0.7298552554448302,
      "grad_norm": 0.6850069751175543,
      "learning_rate": 7.177843478291225e-07,
      "loss": 0.5871,
      "step": 8093
    },
    {
      "epoch": 0.7299454389683004,
      "grad_norm": 1.9546044811545977,
      "learning_rate": 7.173360496148276e-07,
      "loss": 0.7751,
      "step": 8094
    },
    {
      "epoch": 0.7300356224917708,
      "grad_norm": 2.551294062008263,
      "learning_rate": 7.168878608444939e-07,
      "loss": 0.7578,
      "step": 8095
    },
    {
      "epoch": 0.730125806015241,
      "grad_norm": 2.3960628123193755,
      "learning_rate": 7.164397815563623e-07,
      "loss": 0.7435,
      "step": 8096
    },
    {
      "epoch": 0.7302159895387113,
      "grad_norm": 1.88451754661887,
      "learning_rate": 7.159918117886661e-07,
      "loss": 0.7491,
      "step": 8097
    },
    {
      "epoch": 0.7303061730621815,
      "grad_norm": 4.750083521058472,
      "learning_rate": 7.155439515796284e-07,
      "loss": 0.7612,
      "step": 8098
    },
    {
      "epoch": 0.7303963565856518,
      "grad_norm": 1.874559668652807,
      "learning_rate": 7.150962009674633e-07,
      "loss": 0.7424,
      "step": 8099
    },
    {
      "epoch": 0.7304865401091221,
      "grad_norm": 1.668344352652749,
      "learning_rate": 7.146485599903751e-07,
      "loss": 0.7894,
      "step": 8100
    },
    {
      "epoch": 0.7305767236325923,
      "grad_norm": 1.5882386298965296,
      "learning_rate": 7.142010286865592e-07,
      "loss": 0.8164,
      "step": 8101
    },
    {
      "epoch": 0.7306669071560626,
      "grad_norm": 0.5439870876522382,
      "learning_rate": 7.137536070942012e-07,
      "loss": 0.5479,
      "step": 8102
    },
    {
      "epoch": 0.7307570906795329,
      "grad_norm": 1.671060595511198,
      "learning_rate": 7.133062952514786e-07,
      "loss": 0.7434,
      "step": 8103
    },
    {
      "epoch": 0.7308472742030031,
      "grad_norm": 0.6431367295495856,
      "learning_rate": 7.128590931965562e-07,
      "loss": 0.604,
      "step": 8104
    },
    {
      "epoch": 0.7309374577264733,
      "grad_norm": 1.4754389772377468,
      "learning_rate": 7.124120009675945e-07,
      "loss": 0.8057,
      "step": 8105
    },
    {
      "epoch": 0.7310276412499437,
      "grad_norm": 2.3055143263269753,
      "learning_rate": 7.119650186027399e-07,
      "loss": 0.7701,
      "step": 8106
    },
    {
      "epoch": 0.7311178247734139,
      "grad_norm": 5.552965200311374,
      "learning_rate": 7.11518146140132e-07,
      "loss": 0.8643,
      "step": 8107
    },
    {
      "epoch": 0.7312080082968841,
      "grad_norm": 1.9938934800709351,
      "learning_rate": 7.110713836179007e-07,
      "loss": 0.7078,
      "step": 8108
    },
    {
      "epoch": 0.7312981918203544,
      "grad_norm": 1.9032727740062798,
      "learning_rate": 7.106247310741659e-07,
      "loss": 0.6738,
      "step": 8109
    },
    {
      "epoch": 0.7313883753438247,
      "grad_norm": 3.709784077532883,
      "learning_rate": 7.101781885470393e-07,
      "loss": 0.8127,
      "step": 8110
    },
    {
      "epoch": 0.731478558867295,
      "grad_norm": 1.7863800728637598,
      "learning_rate": 7.097317560746203e-07,
      "loss": 0.7811,
      "step": 8111
    },
    {
      "epoch": 0.7315687423907652,
      "grad_norm": 1.6255957905082192,
      "learning_rate": 7.092854336950036e-07,
      "loss": 0.8005,
      "step": 8112
    },
    {
      "epoch": 0.7316589259142354,
      "grad_norm": 2.0985317001729706,
      "learning_rate": 7.0883922144627e-07,
      "loss": 0.7971,
      "step": 8113
    },
    {
      "epoch": 0.7317491094377058,
      "grad_norm": 1.561229723515413,
      "learning_rate": 7.083931193664934e-07,
      "loss": 0.7786,
      "step": 8114
    },
    {
      "epoch": 0.731839292961176,
      "grad_norm": 1.572377000737845,
      "learning_rate": 7.079471274937378e-07,
      "loss": 0.7908,
      "step": 8115
    },
    {
      "epoch": 0.7319294764846462,
      "grad_norm": 2.548578733907832,
      "learning_rate": 7.075012458660574e-07,
      "loss": 0.7298,
      "step": 8116
    },
    {
      "epoch": 0.7320196600081165,
      "grad_norm": 0.6619014807456756,
      "learning_rate": 7.070554745214976e-07,
      "loss": 0.5764,
      "step": 8117
    },
    {
      "epoch": 0.7321098435315868,
      "grad_norm": 2.18739242289189,
      "learning_rate": 7.066098134980947e-07,
      "loss": 0.6963,
      "step": 8118
    },
    {
      "epoch": 0.732200027055057,
      "grad_norm": 2.5182519311411813,
      "learning_rate": 7.061642628338727e-07,
      "loss": 0.7541,
      "step": 8119
    },
    {
      "epoch": 0.7322902105785273,
      "grad_norm": 1.6259967974529188,
      "learning_rate": 7.057188225668513e-07,
      "loss": 0.7437,
      "step": 8120
    },
    {
      "epoch": 0.7323803941019975,
      "grad_norm": 1.6448221198611408,
      "learning_rate": 7.052734927350358e-07,
      "loss": 0.8909,
      "step": 8121
    },
    {
      "epoch": 0.7324705776254679,
      "grad_norm": 2.4146626044473893,
      "learning_rate": 7.048282733764252e-07,
      "loss": 0.77,
      "step": 8122
    },
    {
      "epoch": 0.7325607611489381,
      "grad_norm": 1.8436857309293178,
      "learning_rate": 7.043831645290077e-07,
      "loss": 0.8237,
      "step": 8123
    },
    {
      "epoch": 0.7326509446724083,
      "grad_norm": 1.66624057409392,
      "learning_rate": 7.039381662307624e-07,
      "loss": 0.7143,
      "step": 8124
    },
    {
      "epoch": 0.7327411281958787,
      "grad_norm": 1.6623804608728696,
      "learning_rate": 7.034932785196601e-07,
      "loss": 0.7209,
      "step": 8125
    },
    {
      "epoch": 0.7328313117193489,
      "grad_norm": 1.5217262242344471,
      "learning_rate": 7.030485014336585e-07,
      "loss": 0.7282,
      "step": 8126
    },
    {
      "epoch": 0.7329214952428191,
      "grad_norm": 1.8868108059557305,
      "learning_rate": 7.026038350107118e-07,
      "loss": 0.7032,
      "step": 8127
    },
    {
      "epoch": 0.7330116787662894,
      "grad_norm": 1.5761238478442252,
      "learning_rate": 7.021592792887579e-07,
      "loss": 0.8772,
      "step": 8128
    },
    {
      "epoch": 0.7331018622897597,
      "grad_norm": 1.5385454732666586,
      "learning_rate": 7.01714834305732e-07,
      "loss": 0.7982,
      "step": 8129
    },
    {
      "epoch": 0.7331920458132299,
      "grad_norm": 1.8148015635230812,
      "learning_rate": 7.012705000995544e-07,
      "loss": 0.706,
      "step": 8130
    },
    {
      "epoch": 0.7332822293367002,
      "grad_norm": 1.6114706080469992,
      "learning_rate": 7.008262767081392e-07,
      "loss": 0.7384,
      "step": 8131
    },
    {
      "epoch": 0.7333724128601704,
      "grad_norm": 22.10920315193815,
      "learning_rate": 7.003821641693892e-07,
      "loss": 0.7989,
      "step": 8132
    },
    {
      "epoch": 0.7334625963836408,
      "grad_norm": 1.8031321507250975,
      "learning_rate": 6.999381625211993e-07,
      "loss": 0.7319,
      "step": 8133
    },
    {
      "epoch": 0.733552779907111,
      "grad_norm": 1.7871083077182606,
      "learning_rate": 6.994942718014536e-07,
      "loss": 0.6677,
      "step": 8134
    },
    {
      "epoch": 0.7336429634305812,
      "grad_norm": 2.723177669734599,
      "learning_rate": 6.990504920480282e-07,
      "loss": 0.6543,
      "step": 8135
    },
    {
      "epoch": 0.7337331469540515,
      "grad_norm": 1.7817483171476742,
      "learning_rate": 6.986068232987879e-07,
      "loss": 0.7435,
      "step": 8136
    },
    {
      "epoch": 0.7338233304775218,
      "grad_norm": 1.6112738952758883,
      "learning_rate": 6.981632655915888e-07,
      "loss": 0.6973,
      "step": 8137
    },
    {
      "epoch": 0.733913514000992,
      "grad_norm": 5.411407480254874,
      "learning_rate": 6.977198189642783e-07,
      "loss": 0.8087,
      "step": 8138
    },
    {
      "epoch": 0.7340036975244623,
      "grad_norm": 1.9616657743665318,
      "learning_rate": 6.972764834546935e-07,
      "loss": 0.7705,
      "step": 8139
    },
    {
      "epoch": 0.7340938810479325,
      "grad_norm": 2.6309235358963585,
      "learning_rate": 6.96833259100663e-07,
      "loss": 0.7633,
      "step": 8140
    },
    {
      "epoch": 0.7341840645714028,
      "grad_norm": 1.9864353087247244,
      "learning_rate": 6.96390145940003e-07,
      "loss": 0.7199,
      "step": 8141
    },
    {
      "epoch": 0.7342742480948731,
      "grad_norm": 32.52130086213688,
      "learning_rate": 6.959471440105253e-07,
      "loss": 0.6989,
      "step": 8142
    },
    {
      "epoch": 0.7343644316183433,
      "grad_norm": 1.7578167724557452,
      "learning_rate": 6.955042533500261e-07,
      "loss": 0.7789,
      "step": 8143
    },
    {
      "epoch": 0.7344546151418135,
      "grad_norm": 2.490139493443663,
      "learning_rate": 6.950614739962986e-07,
      "loss": 0.7036,
      "step": 8144
    },
    {
      "epoch": 0.7345447986652839,
      "grad_norm": 1.5640749051481468,
      "learning_rate": 6.946188059871198e-07,
      "loss": 0.6358,
      "step": 8145
    },
    {
      "epoch": 0.7346349821887541,
      "grad_norm": 1.4949113042540918,
      "learning_rate": 6.941762493602638e-07,
      "loss": 0.8593,
      "step": 8146
    },
    {
      "epoch": 0.7347251657122243,
      "grad_norm": 1.633511772269806,
      "learning_rate": 6.937338041534899e-07,
      "loss": 0.6837,
      "step": 8147
    },
    {
      "epoch": 0.7348153492356947,
      "grad_norm": 1.923292986029553,
      "learning_rate": 6.932914704045505e-07,
      "loss": 0.7437,
      "step": 8148
    },
    {
      "epoch": 0.7349055327591649,
      "grad_norm": 1.6665650654659843,
      "learning_rate": 6.928492481511878e-07,
      "loss": 0.773,
      "step": 8149
    },
    {
      "epoch": 0.7349957162826352,
      "grad_norm": 1.9280264187779526,
      "learning_rate": 6.924071374311349e-07,
      "loss": 0.7005,
      "step": 8150
    },
    {
      "epoch": 0.7350858998061054,
      "grad_norm": 0.5969104227070275,
      "learning_rate": 6.919651382821157e-07,
      "loss": 0.5584,
      "step": 8151
    },
    {
      "epoch": 0.7351760833295757,
      "grad_norm": 1.602139252809907,
      "learning_rate": 6.915232507418425e-07,
      "loss": 0.7167,
      "step": 8152
    },
    {
      "epoch": 0.735266266853046,
      "grad_norm": 1.960822964549126,
      "learning_rate": 6.910814748480204e-07,
      "loss": 0.636,
      "step": 8153
    },
    {
      "epoch": 0.7353564503765162,
      "grad_norm": 2.4866011140823168,
      "learning_rate": 6.906398106383445e-07,
      "loss": 0.7658,
      "step": 8154
    },
    {
      "epoch": 0.7354466338999864,
      "grad_norm": 1.474907002507664,
      "learning_rate": 6.901982581504994e-07,
      "loss": 0.7425,
      "step": 8155
    },
    {
      "epoch": 0.7355368174234568,
      "grad_norm": 5.697223776950352,
      "learning_rate": 6.897568174221611e-07,
      "loss": 0.6195,
      "step": 8156
    },
    {
      "epoch": 0.735627000946927,
      "grad_norm": 2.000777927740006,
      "learning_rate": 6.893154884909966e-07,
      "loss": 0.8127,
      "step": 8157
    },
    {
      "epoch": 0.7357171844703972,
      "grad_norm": 0.605751217817656,
      "learning_rate": 6.888742713946602e-07,
      "loss": 0.5837,
      "step": 8158
    },
    {
      "epoch": 0.7358073679938675,
      "grad_norm": 1.9491632182251282,
      "learning_rate": 6.884331661708018e-07,
      "loss": 0.7466,
      "step": 8159
    },
    {
      "epoch": 0.7358975515173378,
      "grad_norm": 1.5161410367578159,
      "learning_rate": 6.879921728570561e-07,
      "loss": 0.7448,
      "step": 8160
    },
    {
      "epoch": 0.735987735040808,
      "grad_norm": 1.7205226426833813,
      "learning_rate": 6.875512914910539e-07,
      "loss": 0.8202,
      "step": 8161
    },
    {
      "epoch": 0.7360779185642783,
      "grad_norm": 1.648481467867426,
      "learning_rate": 6.871105221104119e-07,
      "loss": 0.694,
      "step": 8162
    },
    {
      "epoch": 0.7361681020877485,
      "grad_norm": 4.583247212843541,
      "learning_rate": 6.866698647527391e-07,
      "loss": 0.7179,
      "step": 8163
    },
    {
      "epoch": 0.7362582856112189,
      "grad_norm": 1.5706382906951408,
      "learning_rate": 6.862293194556353e-07,
      "loss": 0.7695,
      "step": 8164
    },
    {
      "epoch": 0.7363484691346891,
      "grad_norm": 1.5817665161167682,
      "learning_rate": 6.857888862566896e-07,
      "loss": 0.6861,
      "step": 8165
    },
    {
      "epoch": 0.7364386526581593,
      "grad_norm": 2.0800466042579453,
      "learning_rate": 6.853485651934836e-07,
      "loss": 0.7096,
      "step": 8166
    },
    {
      "epoch": 0.7365288361816296,
      "grad_norm": 1.628630909873223,
      "learning_rate": 6.849083563035855e-07,
      "loss": 0.6036,
      "step": 8167
    },
    {
      "epoch": 0.7366190197050999,
      "grad_norm": 0.713245752794608,
      "learning_rate": 6.844682596245592e-07,
      "loss": 0.6608,
      "step": 8168
    },
    {
      "epoch": 0.7367092032285701,
      "grad_norm": 1.8601250938680567,
      "learning_rate": 6.840282751939539e-07,
      "loss": 0.8066,
      "step": 8169
    },
    {
      "epoch": 0.7367993867520404,
      "grad_norm": 1.5207278681029928,
      "learning_rate": 6.835884030493126e-07,
      "loss": 0.7902,
      "step": 8170
    },
    {
      "epoch": 0.7368895702755107,
      "grad_norm": 2.118516455637963,
      "learning_rate": 6.831486432281672e-07,
      "loss": 0.7669,
      "step": 8171
    },
    {
      "epoch": 0.736979753798981,
      "grad_norm": 1.9552719523167306,
      "learning_rate": 6.827089957680407e-07,
      "loss": 0.76,
      "step": 8172
    },
    {
      "epoch": 0.7370699373224512,
      "grad_norm": 1.5867730556343842,
      "learning_rate": 6.822694607064461e-07,
      "loss": 0.8613,
      "step": 8173
    },
    {
      "epoch": 0.7371601208459214,
      "grad_norm": 2.0269184585226907,
      "learning_rate": 6.818300380808877e-07,
      "loss": 0.7662,
      "step": 8174
    },
    {
      "epoch": 0.7372503043693918,
      "grad_norm": 1.5350038098465053,
      "learning_rate": 6.813907279288574e-07,
      "loss": 0.7366,
      "step": 8175
    },
    {
      "epoch": 0.737340487892862,
      "grad_norm": 1.708162679134532,
      "learning_rate": 6.809515302878422e-07,
      "loss": 0.7723,
      "step": 8176
    },
    {
      "epoch": 0.7374306714163322,
      "grad_norm": 3.668470906912148,
      "learning_rate": 6.80512445195315e-07,
      "loss": 0.7355,
      "step": 8177
    },
    {
      "epoch": 0.7375208549398025,
      "grad_norm": 1.599993947136874,
      "learning_rate": 6.800734726887416e-07,
      "loss": 0.7647,
      "step": 8178
    },
    {
      "epoch": 0.7376110384632728,
      "grad_norm": 2.673807170554533,
      "learning_rate": 6.796346128055775e-07,
      "loss": 0.7455,
      "step": 8179
    },
    {
      "epoch": 0.737701221986743,
      "grad_norm": 1.6304121854847093,
      "learning_rate": 6.791958655832684e-07,
      "loss": 0.7167,
      "step": 8180
    },
    {
      "epoch": 0.7377914055102133,
      "grad_norm": 2.0080178003700024,
      "learning_rate": 6.787572310592518e-07,
      "loss": 0.7108,
      "step": 8181
    },
    {
      "epoch": 0.7378815890336835,
      "grad_norm": 2.0817387645606527,
      "learning_rate": 6.783187092709521e-07,
      "loss": 0.835,
      "step": 8182
    },
    {
      "epoch": 0.7379717725571538,
      "grad_norm": 2.1522102660726317,
      "learning_rate": 6.778803002557891e-07,
      "loss": 0.7532,
      "step": 8183
    },
    {
      "epoch": 0.7380619560806241,
      "grad_norm": 1.4316511999470878,
      "learning_rate": 6.774420040511686e-07,
      "loss": 0.7814,
      "step": 8184
    },
    {
      "epoch": 0.7381521396040943,
      "grad_norm": 1.7027491583535737,
      "learning_rate": 6.770038206944886e-07,
      "loss": 0.7533,
      "step": 8185
    },
    {
      "epoch": 0.7382423231275645,
      "grad_norm": 1.5839800852236532,
      "learning_rate": 6.765657502231375e-07,
      "loss": 0.7595,
      "step": 8186
    },
    {
      "epoch": 0.7383325066510349,
      "grad_norm": 1.5813420807937977,
      "learning_rate": 6.761277926744939e-07,
      "loss": 0.812,
      "step": 8187
    },
    {
      "epoch": 0.7384226901745051,
      "grad_norm": 1.4708644986115331,
      "learning_rate": 6.756899480859268e-07,
      "loss": 0.7373,
      "step": 8188
    },
    {
      "epoch": 0.7385128736979754,
      "grad_norm": 1.8755853375083649,
      "learning_rate": 6.752522164947956e-07,
      "loss": 0.7931,
      "step": 8189
    },
    {
      "epoch": 0.7386030572214456,
      "grad_norm": 1.4868103127440815,
      "learning_rate": 6.748145979384498e-07,
      "loss": 0.7987,
      "step": 8190
    },
    {
      "epoch": 0.7386932407449159,
      "grad_norm": 2.178202680518834,
      "learning_rate": 6.743770924542303e-07,
      "loss": 0.69,
      "step": 8191
    },
    {
      "epoch": 0.7387834242683862,
      "grad_norm": 0.6343628764755779,
      "learning_rate": 6.739397000794658e-07,
      "loss": 0.5704,
      "step": 8192
    },
    {
      "epoch": 0.7388736077918564,
      "grad_norm": 1.9463946538247325,
      "learning_rate": 6.735024208514782e-07,
      "loss": 0.7236,
      "step": 8193
    },
    {
      "epoch": 0.7389637913153266,
      "grad_norm": 1.5291358210114436,
      "learning_rate": 6.73065254807578e-07,
      "loss": 0.6608,
      "step": 8194
    },
    {
      "epoch": 0.739053974838797,
      "grad_norm": 1.579097485935137,
      "learning_rate": 6.726282019850669e-07,
      "loss": 0.8008,
      "step": 8195
    },
    {
      "epoch": 0.7391441583622672,
      "grad_norm": 1.688505791344396,
      "learning_rate": 6.721912624212376e-07,
      "loss": 0.8133,
      "step": 8196
    },
    {
      "epoch": 0.7392343418857374,
      "grad_norm": 1.9607320731113331,
      "learning_rate": 6.717544361533696e-07,
      "loss": 0.6448,
      "step": 8197
    },
    {
      "epoch": 0.7393245254092078,
      "grad_norm": 1.39929224244067,
      "learning_rate": 6.713177232187386e-07,
      "loss": 0.7295,
      "step": 8198
    },
    {
      "epoch": 0.739414708932678,
      "grad_norm": 1.3837078665613727,
      "learning_rate": 6.708811236546041e-07,
      "loss": 0.7365,
      "step": 8199
    },
    {
      "epoch": 0.7395048924561483,
      "grad_norm": 2.320145417950567,
      "learning_rate": 6.704446374982224e-07,
      "loss": 0.7233,
      "step": 8200
    },
    {
      "epoch": 0.7395950759796185,
      "grad_norm": 1.9815809749601696,
      "learning_rate": 6.700082647868346e-07,
      "loss": 0.7836,
      "step": 8201
    },
    {
      "epoch": 0.7396852595030888,
      "grad_norm": 1.4614154217547546,
      "learning_rate": 6.695720055576751e-07,
      "loss": 0.8015,
      "step": 8202
    },
    {
      "epoch": 0.7397754430265591,
      "grad_norm": 2.481396407028651,
      "learning_rate": 6.691358598479679e-07,
      "loss": 0.7598,
      "step": 8203
    },
    {
      "epoch": 0.7398656265500293,
      "grad_norm": 1.916872725259969,
      "learning_rate": 6.686998276949276e-07,
      "loss": 0.7384,
      "step": 8204
    },
    {
      "epoch": 0.7399558100734995,
      "grad_norm": 1.8563485398772819,
      "learning_rate": 6.682639091357587e-07,
      "loss": 0.5858,
      "step": 8205
    },
    {
      "epoch": 0.7400459935969699,
      "grad_norm": 1.680003291762624,
      "learning_rate": 6.678281042076568e-07,
      "loss": 0.7356,
      "step": 8206
    },
    {
      "epoch": 0.7401361771204401,
      "grad_norm": 1.6838375189450165,
      "learning_rate": 6.673924129478059e-07,
      "loss": 0.7488,
      "step": 8207
    },
    {
      "epoch": 0.7402263606439103,
      "grad_norm": 2.1540665559607355,
      "learning_rate": 6.669568353933824e-07,
      "loss": 0.6671,
      "step": 8208
    },
    {
      "epoch": 0.7403165441673806,
      "grad_norm": 1.7435739927439895,
      "learning_rate": 6.665213715815519e-07,
      "loss": 0.6809,
      "step": 8209
    },
    {
      "epoch": 0.7404067276908509,
      "grad_norm": 1.7559221063425565,
      "learning_rate": 6.660860215494706e-07,
      "loss": 0.7619,
      "step": 8210
    },
    {
      "epoch": 0.7404969112143212,
      "grad_norm": 1.5941842927305623,
      "learning_rate": 6.656507853342852e-07,
      "loss": 0.701,
      "step": 8211
    },
    {
      "epoch": 0.7405870947377914,
      "grad_norm": 1.4839228694387177,
      "learning_rate": 6.652156629731323e-07,
      "loss": 0.7562,
      "step": 8212
    },
    {
      "epoch": 0.7406772782612616,
      "grad_norm": 2.346765536461574,
      "learning_rate": 6.647806545031396e-07,
      "loss": 0.8102,
      "step": 8213
    },
    {
      "epoch": 0.740767461784732,
      "grad_norm": 1.5356664998677017,
      "learning_rate": 6.643457599614224e-07,
      "loss": 0.6725,
      "step": 8214
    },
    {
      "epoch": 0.7408576453082022,
      "grad_norm": 1.6333969749326451,
      "learning_rate": 6.63910979385091e-07,
      "loss": 0.8017,
      "step": 8215
    },
    {
      "epoch": 0.7409478288316724,
      "grad_norm": 2.6579067673244414,
      "learning_rate": 6.634763128112409e-07,
      "loss": 0.8359,
      "step": 8216
    },
    {
      "epoch": 0.7410380123551427,
      "grad_norm": 2.377534618105597,
      "learning_rate": 6.630417602769622e-07,
      "loss": 0.7364,
      "step": 8217
    },
    {
      "epoch": 0.741128195878613,
      "grad_norm": 1.6361335673335595,
      "learning_rate": 6.62607321819332e-07,
      "loss": 0.7532,
      "step": 8218
    },
    {
      "epoch": 0.7412183794020832,
      "grad_norm": 1.3517455048780025,
      "learning_rate": 6.621729974754196e-07,
      "loss": 0.7468,
      "step": 8219
    },
    {
      "epoch": 0.7413085629255535,
      "grad_norm": 1.7071789431021467,
      "learning_rate": 6.617387872822835e-07,
      "loss": 0.6972,
      "step": 8220
    },
    {
      "epoch": 0.7413987464490238,
      "grad_norm": 1.847016302341458,
      "learning_rate": 6.613046912769731e-07,
      "loss": 0.719,
      "step": 8221
    },
    {
      "epoch": 0.741488929972494,
      "grad_norm": 1.732687326627828,
      "learning_rate": 6.608707094965289e-07,
      "loss": 0.8135,
      "step": 8222
    },
    {
      "epoch": 0.7415791134959643,
      "grad_norm": 1.9871006066511503,
      "learning_rate": 6.604368419779787e-07,
      "loss": 0.6764,
      "step": 8223
    },
    {
      "epoch": 0.7416692970194345,
      "grad_norm": 1.635182244412722,
      "learning_rate": 6.600030887583434e-07,
      "loss": 0.7595,
      "step": 8224
    },
    {
      "epoch": 0.7417594805429049,
      "grad_norm": 1.777378291538495,
      "learning_rate": 6.595694498746336e-07,
      "loss": 0.8562,
      "step": 8225
    },
    {
      "epoch": 0.7418496640663751,
      "grad_norm": 2.3766553029256388,
      "learning_rate": 6.591359253638491e-07,
      "loss": 0.7226,
      "step": 8226
    },
    {
      "epoch": 0.7419398475898453,
      "grad_norm": 3.2521823011869753,
      "learning_rate": 6.587025152629808e-07,
      "loss": 0.6376,
      "step": 8227
    },
    {
      "epoch": 0.7420300311133156,
      "grad_norm": 2.315004590485143,
      "learning_rate": 6.582692196090107e-07,
      "loss": 0.7908,
      "step": 8228
    },
    {
      "epoch": 0.7421202146367859,
      "grad_norm": 1.6013732798377067,
      "learning_rate": 6.578360384389074e-07,
      "loss": 0.7656,
      "step": 8229
    },
    {
      "epoch": 0.7422103981602561,
      "grad_norm": 1.7080198868793126,
      "learning_rate": 6.574029717896355e-07,
      "loss": 0.7265,
      "step": 8230
    },
    {
      "epoch": 0.7423005816837264,
      "grad_norm": 1.6119818453748527,
      "learning_rate": 6.569700196981436e-07,
      "loss": 0.7418,
      "step": 8231
    },
    {
      "epoch": 0.7423907652071966,
      "grad_norm": 1.7908351728247345,
      "learning_rate": 6.565371822013763e-07,
      "loss": 0.6535,
      "step": 8232
    },
    {
      "epoch": 0.742480948730667,
      "grad_norm": 12.156929055985547,
      "learning_rate": 6.561044593362636e-07,
      "loss": 0.8018,
      "step": 8233
    },
    {
      "epoch": 0.7425711322541372,
      "grad_norm": 1.8779939272521353,
      "learning_rate": 6.556718511397288e-07,
      "loss": 0.7296,
      "step": 8234
    },
    {
      "epoch": 0.7426613157776074,
      "grad_norm": 1.5798682225478349,
      "learning_rate": 6.552393576486843e-07,
      "loss": 0.8616,
      "step": 8235
    },
    {
      "epoch": 0.7427514993010776,
      "grad_norm": 2.1564510085911093,
      "learning_rate": 6.548069789000325e-07,
      "loss": 0.7614,
      "step": 8236
    },
    {
      "epoch": 0.742841682824548,
      "grad_norm": 1.883509273525464,
      "learning_rate": 6.543747149306673e-07,
      "loss": 0.7317,
      "step": 8237
    },
    {
      "epoch": 0.7429318663480182,
      "grad_norm": 1.6967439716817125,
      "learning_rate": 6.5394256577747e-07,
      "loss": 0.7984,
      "step": 8238
    },
    {
      "epoch": 0.7430220498714885,
      "grad_norm": 2.250817044578271,
      "learning_rate": 6.535105314773161e-07,
      "loss": 0.777,
      "step": 8239
    },
    {
      "epoch": 0.7431122333949587,
      "grad_norm": 4.068454779771301,
      "learning_rate": 6.530786120670677e-07,
      "loss": 0.6782,
      "step": 8240
    },
    {
      "epoch": 0.743202416918429,
      "grad_norm": 1.818143346227753,
      "learning_rate": 6.526468075835787e-07,
      "loss": 0.8058,
      "step": 8241
    },
    {
      "epoch": 0.7432926004418993,
      "grad_norm": 2.4704503831017193,
      "learning_rate": 6.522151180636937e-07,
      "loss": 0.7933,
      "step": 8242
    },
    {
      "epoch": 0.7433827839653695,
      "grad_norm": 3.071108649311591,
      "learning_rate": 6.517835435442461e-07,
      "loss": 0.7111,
      "step": 8243
    },
    {
      "epoch": 0.7434729674888398,
      "grad_norm": 2.119661806769973,
      "learning_rate": 6.513520840620606e-07,
      "loss": 0.7681,
      "step": 8244
    },
    {
      "epoch": 0.7435631510123101,
      "grad_norm": 1.7156398029070938,
      "learning_rate": 6.509207396539525e-07,
      "loss": 0.712,
      "step": 8245
    },
    {
      "epoch": 0.7436533345357803,
      "grad_norm": 0.6221837489782868,
      "learning_rate": 6.50489510356724e-07,
      "loss": 0.5526,
      "step": 8246
    },
    {
      "epoch": 0.7437435180592505,
      "grad_norm": 2.459084530730606,
      "learning_rate": 6.500583962071732e-07,
      "loss": 0.6795,
      "step": 8247
    },
    {
      "epoch": 0.7438337015827209,
      "grad_norm": 2.227316852408509,
      "learning_rate": 6.496273972420827e-07,
      "loss": 0.7714,
      "step": 8248
    },
    {
      "epoch": 0.7439238851061911,
      "grad_norm": 1.822407119016834,
      "learning_rate": 6.491965134982287e-07,
      "loss": 0.7993,
      "step": 8249
    },
    {
      "epoch": 0.7440140686296614,
      "grad_norm": 1.9855572039995761,
      "learning_rate": 6.487657450123765e-07,
      "loss": 0.714,
      "step": 8250
    },
    {
      "epoch": 0.7441042521531316,
      "grad_norm": 1.5944643102834186,
      "learning_rate": 6.483350918212814e-07,
      "loss": 0.7604,
      "step": 8251
    },
    {
      "epoch": 0.7441944356766019,
      "grad_norm": 0.7090487280189506,
      "learning_rate": 6.479045539616898e-07,
      "loss": 0.5848,
      "step": 8252
    },
    {
      "epoch": 0.7442846192000722,
      "grad_norm": 1.9646196180642352,
      "learning_rate": 6.474741314703358e-07,
      "loss": 0.7527,
      "step": 8253
    },
    {
      "epoch": 0.7443748027235424,
      "grad_norm": 2.3720317160559,
      "learning_rate": 6.47043824383948e-07,
      "loss": 0.8075,
      "step": 8254
    },
    {
      "epoch": 0.7444649862470126,
      "grad_norm": 1.8416253472205324,
      "learning_rate": 6.466136327392399e-07,
      "loss": 0.7839,
      "step": 8255
    },
    {
      "epoch": 0.744555169770483,
      "grad_norm": 1.7568031634528092,
      "learning_rate": 6.461835565729206e-07,
      "loss": 0.783,
      "step": 8256
    },
    {
      "epoch": 0.7446453532939532,
      "grad_norm": 2.2260064752558244,
      "learning_rate": 6.457535959216844e-07,
      "loss": 0.7766,
      "step": 8257
    },
    {
      "epoch": 0.7447355368174234,
      "grad_norm": 1.8363454162653454,
      "learning_rate": 6.453237508222186e-07,
      "loss": 0.7083,
      "step": 8258
    },
    {
      "epoch": 0.7448257203408937,
      "grad_norm": 1.5645942386225131,
      "learning_rate": 6.448940213112e-07,
      "loss": 0.7708,
      "step": 8259
    },
    {
      "epoch": 0.744915903864364,
      "grad_norm": 1.4534262068118426,
      "learning_rate": 6.444644074252954e-07,
      "loss": 0.832,
      "step": 8260
    },
    {
      "epoch": 0.7450060873878342,
      "grad_norm": 1.9339592482248833,
      "learning_rate": 6.440349092011628e-07,
      "loss": 0.8197,
      "step": 8261
    },
    {
      "epoch": 0.7450962709113045,
      "grad_norm": 1.4766983696323626,
      "learning_rate": 6.436055266754475e-07,
      "loss": 0.7392,
      "step": 8262
    },
    {
      "epoch": 0.7451864544347747,
      "grad_norm": 1.8503687104531419,
      "learning_rate": 6.431762598847879e-07,
      "loss": 0.7207,
      "step": 8263
    },
    {
      "epoch": 0.7452766379582451,
      "grad_norm": 1.8111283604100152,
      "learning_rate": 6.427471088658111e-07,
      "loss": 0.73,
      "step": 8264
    },
    {
      "epoch": 0.7453668214817153,
      "grad_norm": 2.1594953577972564,
      "learning_rate": 6.42318073655135e-07,
      "loss": 0.7483,
      "step": 8265
    },
    {
      "epoch": 0.7454570050051855,
      "grad_norm": 3.4597557237424654,
      "learning_rate": 6.41889154289367e-07,
      "loss": 0.7161,
      "step": 8266
    },
    {
      "epoch": 0.7455471885286559,
      "grad_norm": 1.780526733793348,
      "learning_rate": 6.414603508051055e-07,
      "loss": 0.8072,
      "step": 8267
    },
    {
      "epoch": 0.7456373720521261,
      "grad_norm": 0.635181068038768,
      "learning_rate": 6.410316632389365e-07,
      "loss": 0.6046,
      "step": 8268
    },
    {
      "epoch": 0.7457275555755963,
      "grad_norm": 2.4916668291208817,
      "learning_rate": 6.406030916274406e-07,
      "loss": 0.755,
      "step": 8269
    },
    {
      "epoch": 0.7458177390990666,
      "grad_norm": 2.101936051874132,
      "learning_rate": 6.401746360071831e-07,
      "loss": 0.7088,
      "step": 8270
    },
    {
      "epoch": 0.7459079226225369,
      "grad_norm": 1.7283917571259921,
      "learning_rate": 6.397462964147251e-07,
      "loss": 0.7114,
      "step": 8271
    },
    {
      "epoch": 0.7459981061460071,
      "grad_norm": 2.353280792550355,
      "learning_rate": 6.393180728866128e-07,
      "loss": 0.7387,
      "step": 8272
    },
    {
      "epoch": 0.7460882896694774,
      "grad_norm": 1.8774281991164201,
      "learning_rate": 6.388899654593853e-07,
      "loss": 0.7808,
      "step": 8273
    },
    {
      "epoch": 0.7461784731929476,
      "grad_norm": 2.109600026878788,
      "learning_rate": 6.384619741695709e-07,
      "loss": 0.7744,
      "step": 8274
    },
    {
      "epoch": 0.746268656716418,
      "grad_norm": 1.4604990092022443,
      "learning_rate": 6.380340990536883e-07,
      "loss": 0.7485,
      "step": 8275
    },
    {
      "epoch": 0.7463588402398882,
      "grad_norm": 2.9339503491861034,
      "learning_rate": 6.37606340148247e-07,
      "loss": 0.7542,
      "step": 8276
    },
    {
      "epoch": 0.7464490237633584,
      "grad_norm": 4.036587752041523,
      "learning_rate": 6.371786974897433e-07,
      "loss": 0.6554,
      "step": 8277
    },
    {
      "epoch": 0.7465392072868287,
      "grad_norm": 1.413003345704218,
      "learning_rate": 6.367511711146691e-07,
      "loss": 0.7776,
      "step": 8278
    },
    {
      "epoch": 0.746629390810299,
      "grad_norm": 2.389354193514777,
      "learning_rate": 6.363237610595014e-07,
      "loss": 0.6961,
      "step": 8279
    },
    {
      "epoch": 0.7467195743337692,
      "grad_norm": 2.049090402287221,
      "learning_rate": 6.358964673607094e-07,
      "loss": 0.7334,
      "step": 8280
    },
    {
      "epoch": 0.7468097578572395,
      "grad_norm": 1.6133558796392242,
      "learning_rate": 6.354692900547525e-07,
      "loss": 0.7804,
      "step": 8281
    },
    {
      "epoch": 0.7468999413807097,
      "grad_norm": 1.7010280754198912,
      "learning_rate": 6.350422291780797e-07,
      "loss": 0.7244,
      "step": 8282
    },
    {
      "epoch": 0.74699012490418,
      "grad_norm": 2.115803959770186,
      "learning_rate": 6.346152847671302e-07,
      "loss": 0.8476,
      "step": 8283
    },
    {
      "epoch": 0.7470803084276503,
      "grad_norm": 2.996636889786775,
      "learning_rate": 6.34188456858334e-07,
      "loss": 0.7451,
      "step": 8284
    },
    {
      "epoch": 0.7471704919511205,
      "grad_norm": 1.7681238501697738,
      "learning_rate": 6.337617454881081e-07,
      "loss": 0.7005,
      "step": 8285
    },
    {
      "epoch": 0.7472606754745907,
      "grad_norm": 2.860956973848677,
      "learning_rate": 6.333351506928651e-07,
      "loss": 0.6637,
      "step": 8286
    },
    {
      "epoch": 0.7473508589980611,
      "grad_norm": 1.6660511151700523,
      "learning_rate": 6.329086725090018e-07,
      "loss": 0.7185,
      "step": 8287
    },
    {
      "epoch": 0.7474410425215313,
      "grad_norm": 1.8854076813822678,
      "learning_rate": 6.324823109729087e-07,
      "loss": 0.7118,
      "step": 8288
    },
    {
      "epoch": 0.7475312260450016,
      "grad_norm": 3.7873449463785165,
      "learning_rate": 6.320560661209653e-07,
      "loss": 0.6806,
      "step": 8289
    },
    {
      "epoch": 0.7476214095684719,
      "grad_norm": 2.481379304298425,
      "learning_rate": 6.316299379895411e-07,
      "loss": 0.6332,
      "step": 8290
    },
    {
      "epoch": 0.7477115930919421,
      "grad_norm": 2.1807155577425767,
      "learning_rate": 6.312039266149965e-07,
      "loss": 0.7857,
      "step": 8291
    },
    {
      "epoch": 0.7478017766154124,
      "grad_norm": 2.595875524072892,
      "learning_rate": 6.307780320336789e-07,
      "loss": 0.7832,
      "step": 8292
    },
    {
      "epoch": 0.7478919601388826,
      "grad_norm": 1.8894167104969595,
      "learning_rate": 6.303522542819306e-07,
      "loss": 0.8082,
      "step": 8293
    },
    {
      "epoch": 0.7479821436623529,
      "grad_norm": 2.127853609277064,
      "learning_rate": 6.299265933960796e-07,
      "loss": 0.7223,
      "step": 8294
    },
    {
      "epoch": 0.7480723271858232,
      "grad_norm": 2.3311135996168972,
      "learning_rate": 6.295010494124462e-07,
      "loss": 0.6992,
      "step": 8295
    },
    {
      "epoch": 0.7481625107092934,
      "grad_norm": 1.450307760791889,
      "learning_rate": 6.290756223673399e-07,
      "loss": 0.7244,
      "step": 8296
    },
    {
      "epoch": 0.7482526942327636,
      "grad_norm": 1.3491882020355885,
      "learning_rate": 6.28650312297061e-07,
      "loss": 0.7803,
      "step": 8297
    },
    {
      "epoch": 0.748342877756234,
      "grad_norm": 1.9534566368834088,
      "learning_rate": 6.282251192378987e-07,
      "loss": 0.6688,
      "step": 8298
    },
    {
      "epoch": 0.7484330612797042,
      "grad_norm": 3.90024452910068,
      "learning_rate": 6.278000432261334e-07,
      "loss": 0.6775,
      "step": 8299
    },
    {
      "epoch": 0.7485232448031744,
      "grad_norm": 2.512506957542693,
      "learning_rate": 6.273750842980345e-07,
      "loss": 0.7684,
      "step": 8300
    },
    {
      "epoch": 0.7486134283266447,
      "grad_norm": 2.122797216986253,
      "learning_rate": 6.269502424898625e-07,
      "loss": 0.7752,
      "step": 8301
    },
    {
      "epoch": 0.748703611850115,
      "grad_norm": 1.9709609801534569,
      "learning_rate": 6.265255178378663e-07,
      "loss": 0.7907,
      "step": 8302
    },
    {
      "epoch": 0.7487937953735853,
      "grad_norm": 1.6170977139081313,
      "learning_rate": 6.261009103782861e-07,
      "loss": 0.6413,
      "step": 8303
    },
    {
      "epoch": 0.7488839788970555,
      "grad_norm": 1.7940271781470245,
      "learning_rate": 6.256764201473519e-07,
      "loss": 0.7922,
      "step": 8304
    },
    {
      "epoch": 0.7489741624205257,
      "grad_norm": 0.6226104595046502,
      "learning_rate": 6.252520471812835e-07,
      "loss": 0.5728,
      "step": 8305
    },
    {
      "epoch": 0.7490643459439961,
      "grad_norm": 1.8396408714319477,
      "learning_rate": 6.248277915162912e-07,
      "loss": 0.7637,
      "step": 8306
    },
    {
      "epoch": 0.7491545294674663,
      "grad_norm": 1.5255866907694418,
      "learning_rate": 6.244036531885731e-07,
      "loss": 0.6052,
      "step": 8307
    },
    {
      "epoch": 0.7492447129909365,
      "grad_norm": 1.6938606760949237,
      "learning_rate": 6.239796322343216e-07,
      "loss": 0.6827,
      "step": 8308
    },
    {
      "epoch": 0.7493348965144068,
      "grad_norm": 1.7216263025481489,
      "learning_rate": 6.235557286897137e-07,
      "loss": 0.6851,
      "step": 8309
    },
    {
      "epoch": 0.7494250800378771,
      "grad_norm": 1.811530741962194,
      "learning_rate": 6.231319425909223e-07,
      "loss": 0.7432,
      "step": 8310
    },
    {
      "epoch": 0.7495152635613473,
      "grad_norm": 1.6236629120317274,
      "learning_rate": 6.227082739741045e-07,
      "loss": 0.7812,
      "step": 8311
    },
    {
      "epoch": 0.7496054470848176,
      "grad_norm": 1.530543183663073,
      "learning_rate": 6.222847228754113e-07,
      "loss": 0.772,
      "step": 8312
    },
    {
      "epoch": 0.7496956306082878,
      "grad_norm": 1.8976404848225206,
      "learning_rate": 6.218612893309823e-07,
      "loss": 0.8451,
      "step": 8313
    },
    {
      "epoch": 0.7497858141317582,
      "grad_norm": 1.537000795283512,
      "learning_rate": 6.214379733769468e-07,
      "loss": 0.6896,
      "step": 8314
    },
    {
      "epoch": 0.7498759976552284,
      "grad_norm": 1.5899292269136505,
      "learning_rate": 6.21014775049425e-07,
      "loss": 0.7792,
      "step": 8315
    },
    {
      "epoch": 0.7499661811786986,
      "grad_norm": 0.5726503244382285,
      "learning_rate": 6.205916943845267e-07,
      "loss": 0.564,
      "step": 8316
    },
    {
      "epoch": 0.750056364702169,
      "grad_norm": 1.758203895227975,
      "learning_rate": 6.201687314183504e-07,
      "loss": 0.7928,
      "step": 8317
    },
    {
      "epoch": 0.7501465482256392,
      "grad_norm": 0.6511719280899658,
      "learning_rate": 6.197458861869862e-07,
      "loss": 0.6116,
      "step": 8318
    },
    {
      "epoch": 0.7502367317491094,
      "grad_norm": 1.8642151609390545,
      "learning_rate": 6.193231587265138e-07,
      "loss": 0.7923,
      "step": 8319
    },
    {
      "epoch": 0.7503269152725797,
      "grad_norm": 2.566315206599537,
      "learning_rate": 6.189005490730024e-07,
      "loss": 0.7302,
      "step": 8320
    },
    {
      "epoch": 0.75041709879605,
      "grad_norm": 3.1794009067806464,
      "learning_rate": 6.184780572625115e-07,
      "loss": 0.7695,
      "step": 8321
    },
    {
      "epoch": 0.7505072823195202,
      "grad_norm": 1.7683521241110387,
      "learning_rate": 6.180556833310902e-07,
      "loss": 0.6485,
      "step": 8322
    },
    {
      "epoch": 0.7505974658429905,
      "grad_norm": 1.5472825409374553,
      "learning_rate": 6.176334273147788e-07,
      "loss": 0.7851,
      "step": 8323
    },
    {
      "epoch": 0.7506876493664607,
      "grad_norm": 2.0376667492741327,
      "learning_rate": 6.172112892496042e-07,
      "loss": 0.6713,
      "step": 8324
    },
    {
      "epoch": 0.750777832889931,
      "grad_norm": 1.627425071526747,
      "learning_rate": 6.167892691715883e-07,
      "loss": 0.7906,
      "step": 8325
    },
    {
      "epoch": 0.7508680164134013,
      "grad_norm": 1.6000064641106682,
      "learning_rate": 6.163673671167378e-07,
      "loss": 0.7325,
      "step": 8326
    },
    {
      "epoch": 0.7509581999368715,
      "grad_norm": 1.5734741873380047,
      "learning_rate": 6.15945583121054e-07,
      "loss": 0.7896,
      "step": 8327
    },
    {
      "epoch": 0.7510483834603418,
      "grad_norm": 2.167758910875172,
      "learning_rate": 6.15523917220524e-07,
      "loss": 0.6974,
      "step": 8328
    },
    {
      "epoch": 0.7511385669838121,
      "grad_norm": 3.1875951416177735,
      "learning_rate": 6.151023694511273e-07,
      "loss": 0.8122,
      "step": 8329
    },
    {
      "epoch": 0.7512287505072823,
      "grad_norm": 1.8190047515001782,
      "learning_rate": 6.146809398488328e-07,
      "loss": 0.6913,
      "step": 8330
    },
    {
      "epoch": 0.7513189340307526,
      "grad_norm": 1.6372853757257932,
      "learning_rate": 6.142596284495989e-07,
      "loss": 0.7793,
      "step": 8331
    },
    {
      "epoch": 0.7514091175542228,
      "grad_norm": 1.7077836881477526,
      "learning_rate": 6.138384352893751e-07,
      "loss": 0.7063,
      "step": 8332
    },
    {
      "epoch": 0.7514993010776931,
      "grad_norm": 8.797097022065007,
      "learning_rate": 6.134173604040987e-07,
      "loss": 0.7413,
      "step": 8333
    },
    {
      "epoch": 0.7515894846011634,
      "grad_norm": 1.958863944005836,
      "learning_rate": 6.129964038296984e-07,
      "loss": 0.7555,
      "step": 8334
    },
    {
      "epoch": 0.7516796681246336,
      "grad_norm": 2.0657878447963265,
      "learning_rate": 6.12575565602093e-07,
      "loss": 0.8165,
      "step": 8335
    },
    {
      "epoch": 0.7517698516481038,
      "grad_norm": 1.253025636990669,
      "learning_rate": 6.121548457571905e-07,
      "loss": 0.8183,
      "step": 8336
    },
    {
      "epoch": 0.7518600351715742,
      "grad_norm": 4.825689695409036,
      "learning_rate": 6.11734244330889e-07,
      "loss": 0.6973,
      "step": 8337
    },
    {
      "epoch": 0.7519502186950444,
      "grad_norm": 1.7793289667411576,
      "learning_rate": 6.113137613590773e-07,
      "loss": 0.7613,
      "step": 8338
    },
    {
      "epoch": 0.7520404022185146,
      "grad_norm": 1.597454403389215,
      "learning_rate": 6.108933968776313e-07,
      "loss": 0.8063,
      "step": 8339
    },
    {
      "epoch": 0.752130585741985,
      "grad_norm": 3.473085872477111,
      "learning_rate": 6.104731509224212e-07,
      "loss": 0.7456,
      "step": 8340
    },
    {
      "epoch": 0.7522207692654552,
      "grad_norm": 2.6965715646497475,
      "learning_rate": 6.100530235293027e-07,
      "loss": 0.679,
      "step": 8341
    },
    {
      "epoch": 0.7523109527889255,
      "grad_norm": 1.8342916700421807,
      "learning_rate": 6.096330147341253e-07,
      "loss": 0.7701,
      "step": 8342
    },
    {
      "epoch": 0.7524011363123957,
      "grad_norm": 2.2777207129652117,
      "learning_rate": 6.09213124572725e-07,
      "loss": 0.8067,
      "step": 8343
    },
    {
      "epoch": 0.752491319835866,
      "grad_norm": 1.3897843552453661,
      "learning_rate": 6.087933530809297e-07,
      "loss": 0.8231,
      "step": 8344
    },
    {
      "epoch": 0.7525815033593363,
      "grad_norm": 1.532839514506945,
      "learning_rate": 6.083737002945566e-07,
      "loss": 0.7372,
      "step": 8345
    },
    {
      "epoch": 0.7526716868828065,
      "grad_norm": 1.636599515882726,
      "learning_rate": 6.079541662494126e-07,
      "loss": 0.7043,
      "step": 8346
    },
    {
      "epoch": 0.7527618704062767,
      "grad_norm": 2.874567911600935,
      "learning_rate": 6.075347509812954e-07,
      "loss": 0.7558,
      "step": 8347
    },
    {
      "epoch": 0.7528520539297471,
      "grad_norm": 1.7763767168776146,
      "learning_rate": 6.0711545452599e-07,
      "loss": 0.7036,
      "step": 8348
    },
    {
      "epoch": 0.7529422374532173,
      "grad_norm": 1.9700673857651587,
      "learning_rate": 6.066962769192756e-07,
      "loss": 0.6716,
      "step": 8349
    },
    {
      "epoch": 0.7530324209766875,
      "grad_norm": 1.9313101799239503,
      "learning_rate": 6.062772181969167e-07,
      "loss": 0.6858,
      "step": 8350
    },
    {
      "epoch": 0.7531226045001578,
      "grad_norm": 1.4454759659987535,
      "learning_rate": 6.058582783946706e-07,
      "loss": 0.8137,
      "step": 8351
    },
    {
      "epoch": 0.7532127880236281,
      "grad_norm": 1.8370053696032165,
      "learning_rate": 6.054394575482833e-07,
      "loss": 0.7413,
      "step": 8352
    },
    {
      "epoch": 0.7533029715470984,
      "grad_norm": 1.578518733394652,
      "learning_rate": 6.05020755693491e-07,
      "loss": 0.6647,
      "step": 8353
    },
    {
      "epoch": 0.7533931550705686,
      "grad_norm": 2.223938476569147,
      "learning_rate": 6.046021728660198e-07,
      "loss": 0.6631,
      "step": 8354
    },
    {
      "epoch": 0.7534833385940388,
      "grad_norm": 2.187087755867989,
      "learning_rate": 6.041837091015858e-07,
      "loss": 0.785,
      "step": 8355
    },
    {
      "epoch": 0.7535735221175092,
      "grad_norm": 1.5051399224210598,
      "learning_rate": 6.037653644358931e-07,
      "loss": 0.8001,
      "step": 8356
    },
    {
      "epoch": 0.7536637056409794,
      "grad_norm": 1.6509915493680138,
      "learning_rate": 6.033471389046393e-07,
      "loss": 0.7617,
      "step": 8357
    },
    {
      "epoch": 0.7537538891644496,
      "grad_norm": 1.524898200467674,
      "learning_rate": 6.029290325435084e-07,
      "loss": 0.7041,
      "step": 8358
    },
    {
      "epoch": 0.7538440726879199,
      "grad_norm": 1.9431762609042373,
      "learning_rate": 6.025110453881756e-07,
      "loss": 0.808,
      "step": 8359
    },
    {
      "epoch": 0.7539342562113902,
      "grad_norm": 2.0020556614366343,
      "learning_rate": 6.020931774743061e-07,
      "loss": 0.6865,
      "step": 8360
    },
    {
      "epoch": 0.7540244397348604,
      "grad_norm": 1.5789887736563435,
      "learning_rate": 6.016754288375546e-07,
      "loss": 0.6706,
      "step": 8361
    },
    {
      "epoch": 0.7541146232583307,
      "grad_norm": 1.7388477559585542,
      "learning_rate": 6.012577995135665e-07,
      "loss": 0.7397,
      "step": 8362
    },
    {
      "epoch": 0.754204806781801,
      "grad_norm": 2.646531581105738,
      "learning_rate": 6.008402895379743e-07,
      "loss": 0.6922,
      "step": 8363
    },
    {
      "epoch": 0.7542949903052713,
      "grad_norm": 2.2536692899701785,
      "learning_rate": 6.004228989464047e-07,
      "loss": 0.8294,
      "step": 8364
    },
    {
      "epoch": 0.7543851738287415,
      "grad_norm": 2.122756671460596,
      "learning_rate": 6.000056277744692e-07,
      "loss": 0.7703,
      "step": 8365
    },
    {
      "epoch": 0.7544753573522117,
      "grad_norm": 2.0582163861162988,
      "learning_rate": 5.995884760577745e-07,
      "loss": 0.7213,
      "step": 8366
    },
    {
      "epoch": 0.7545655408756821,
      "grad_norm": 0.6580971425926941,
      "learning_rate": 5.99171443831912e-07,
      "loss": 0.5645,
      "step": 8367
    },
    {
      "epoch": 0.7546557243991523,
      "grad_norm": 1.4523683444316406,
      "learning_rate": 5.98754531132466e-07,
      "loss": 0.7456,
      "step": 8368
    },
    {
      "epoch": 0.7547459079226225,
      "grad_norm": 2.8304406901986874,
      "learning_rate": 5.983377379950099e-07,
      "loss": 0.7526,
      "step": 8369
    },
    {
      "epoch": 0.7548360914460928,
      "grad_norm": 1.632346784987412,
      "learning_rate": 5.979210644551067e-07,
      "loss": 0.687,
      "step": 8370
    },
    {
      "epoch": 0.7549262749695631,
      "grad_norm": 2.175086521751655,
      "learning_rate": 5.975045105483091e-07,
      "loss": 0.7422,
      "step": 8371
    },
    {
      "epoch": 0.7550164584930333,
      "grad_norm": 0.6019151322878091,
      "learning_rate": 5.970880763101607e-07,
      "loss": 0.588,
      "step": 8372
    },
    {
      "epoch": 0.7551066420165036,
      "grad_norm": 2.0189609570896714,
      "learning_rate": 5.966717617761925e-07,
      "loss": 0.7668,
      "step": 8373
    },
    {
      "epoch": 0.7551968255399738,
      "grad_norm": 1.817307837086083,
      "learning_rate": 5.962555669819276e-07,
      "loss": 0.728,
      "step": 8374
    },
    {
      "epoch": 0.7552870090634441,
      "grad_norm": 1.5241528518798746,
      "learning_rate": 5.958394919628777e-07,
      "loss": 0.771,
      "step": 8375
    },
    {
      "epoch": 0.7553771925869144,
      "grad_norm": 1.3467878114311365,
      "learning_rate": 5.954235367545451e-07,
      "loss": 0.6918,
      "step": 8376
    },
    {
      "epoch": 0.7554673761103846,
      "grad_norm": 1.430313223567127,
      "learning_rate": 5.950077013924213e-07,
      "loss": 0.7429,
      "step": 8377
    },
    {
      "epoch": 0.7555575596338548,
      "grad_norm": 1.722793123858325,
      "learning_rate": 5.945919859119865e-07,
      "loss": 0.8295,
      "step": 8378
    },
    {
      "epoch": 0.7556477431573252,
      "grad_norm": 1.9414184078703192,
      "learning_rate": 5.94176390348714e-07,
      "loss": 0.6385,
      "step": 8379
    },
    {
      "epoch": 0.7557379266807954,
      "grad_norm": 1.2882172096116824,
      "learning_rate": 5.937609147380622e-07,
      "loss": 0.8539,
      "step": 8380
    },
    {
      "epoch": 0.7558281102042657,
      "grad_norm": 1.5560352866724299,
      "learning_rate": 5.933455591154844e-07,
      "loss": 0.6925,
      "step": 8381
    },
    {
      "epoch": 0.7559182937277359,
      "grad_norm": 2.831514129781252,
      "learning_rate": 5.929303235164191e-07,
      "loss": 0.7167,
      "step": 8382
    },
    {
      "epoch": 0.7560084772512062,
      "grad_norm": 5.54605053429291,
      "learning_rate": 5.92515207976297e-07,
      "loss": 0.7196,
      "step": 8383
    },
    {
      "epoch": 0.7560986607746765,
      "grad_norm": 1.757100685956032,
      "learning_rate": 5.921002125305383e-07,
      "loss": 0.7383,
      "step": 8384
    },
    {
      "epoch": 0.7561888442981467,
      "grad_norm": 1.9935297016182454,
      "learning_rate": 5.916853372145525e-07,
      "loss": 0.6707,
      "step": 8385
    },
    {
      "epoch": 0.756279027821617,
      "grad_norm": 2.1036067507246536,
      "learning_rate": 5.912705820637389e-07,
      "loss": 0.6846,
      "step": 8386
    },
    {
      "epoch": 0.7563692113450873,
      "grad_norm": 1.7648816189496253,
      "learning_rate": 5.908559471134871e-07,
      "loss": 0.8119,
      "step": 8387
    },
    {
      "epoch": 0.7564593948685575,
      "grad_norm": 2.5136342197730777,
      "learning_rate": 5.904414323991764e-07,
      "loss": 0.6872,
      "step": 8388
    },
    {
      "epoch": 0.7565495783920277,
      "grad_norm": 2.0853008770372243,
      "learning_rate": 5.900270379561743e-07,
      "loss": 0.7538,
      "step": 8389
    },
    {
      "epoch": 0.7566397619154981,
      "grad_norm": 2.2341992502517405,
      "learning_rate": 5.896127638198399e-07,
      "loss": 0.7611,
      "step": 8390
    },
    {
      "epoch": 0.7567299454389683,
      "grad_norm": 2.0525088054732943,
      "learning_rate": 5.89198610025521e-07,
      "loss": 0.75,
      "step": 8391
    },
    {
      "epoch": 0.7568201289624386,
      "grad_norm": 2.140693941885651,
      "learning_rate": 5.887845766085559e-07,
      "loss": 0.815,
      "step": 8392
    },
    {
      "epoch": 0.7569103124859088,
      "grad_norm": 1.8525885463253053,
      "learning_rate": 5.883706636042722e-07,
      "loss": 0.7058,
      "step": 8393
    },
    {
      "epoch": 0.7570004960093791,
      "grad_norm": 3.659195715912653,
      "learning_rate": 5.879568710479879e-07,
      "loss": 0.6897,
      "step": 8394
    },
    {
      "epoch": 0.7570906795328494,
      "grad_norm": 3.5881235943093643,
      "learning_rate": 5.875431989750078e-07,
      "loss": 0.8357,
      "step": 8395
    },
    {
      "epoch": 0.7571808630563196,
      "grad_norm": 1.3869472691469218,
      "learning_rate": 5.871296474206313e-07,
      "loss": 0.6823,
      "step": 8396
    },
    {
      "epoch": 0.7572710465797898,
      "grad_norm": 2.298636455210492,
      "learning_rate": 5.867162164201427e-07,
      "loss": 0.7526,
      "step": 8397
    },
    {
      "epoch": 0.7573612301032602,
      "grad_norm": 1.8925932420061897,
      "learning_rate": 5.863029060088205e-07,
      "loss": 0.7738,
      "step": 8398
    },
    {
      "epoch": 0.7574514136267304,
      "grad_norm": 1.7336430422706128,
      "learning_rate": 5.858897162219289e-07,
      "loss": 0.6454,
      "step": 8399
    },
    {
      "epoch": 0.7575415971502006,
      "grad_norm": 1.9415517096263375,
      "learning_rate": 5.854766470947238e-07,
      "loss": 0.7963,
      "step": 8400
    },
    {
      "epoch": 0.7576317806736709,
      "grad_norm": 2.0219135215898434,
      "learning_rate": 5.850636986624511e-07,
      "loss": 0.7358,
      "step": 8401
    },
    {
      "epoch": 0.7577219641971412,
      "grad_norm": 1.858425122148145,
      "learning_rate": 5.846508709603453e-07,
      "loss": 0.7546,
      "step": 8402
    },
    {
      "epoch": 0.7578121477206115,
      "grad_norm": 1.5266112305155994,
      "learning_rate": 5.842381640236318e-07,
      "loss": 0.7223,
      "step": 8403
    },
    {
      "epoch": 0.7579023312440817,
      "grad_norm": 1.5809623969357722,
      "learning_rate": 5.838255778875242e-07,
      "loss": 0.7477,
      "step": 8404
    },
    {
      "epoch": 0.7579925147675519,
      "grad_norm": 1.669399960701534,
      "learning_rate": 5.83413112587227e-07,
      "loss": 0.7684,
      "step": 8405
    },
    {
      "epoch": 0.7580826982910223,
      "grad_norm": 1.6935854082265487,
      "learning_rate": 5.830007681579338e-07,
      "loss": 0.7351,
      "step": 8406
    },
    {
      "epoch": 0.7581728818144925,
      "grad_norm": 2.26493288189135,
      "learning_rate": 5.825885446348284e-07,
      "loss": 0.7611,
      "step": 8407
    },
    {
      "epoch": 0.7582630653379627,
      "grad_norm": 1.585626130004047,
      "learning_rate": 5.821764420530842e-07,
      "loss": 0.6876,
      "step": 8408
    },
    {
      "epoch": 0.7583532488614331,
      "grad_norm": 1.7754441390664575,
      "learning_rate": 5.817644604478633e-07,
      "loss": 0.7662,
      "step": 8409
    },
    {
      "epoch": 0.7584434323849033,
      "grad_norm": 2.6897640007045385,
      "learning_rate": 5.81352599854319e-07,
      "loss": 0.6939,
      "step": 8410
    },
    {
      "epoch": 0.7585336159083735,
      "grad_norm": 1.872671143333722,
      "learning_rate": 5.809408603075938e-07,
      "loss": 0.781,
      "step": 8411
    },
    {
      "epoch": 0.7586237994318438,
      "grad_norm": 2.2098569352461195,
      "learning_rate": 5.805292418428176e-07,
      "loss": 0.7276,
      "step": 8412
    },
    {
      "epoch": 0.7587139829553141,
      "grad_norm": 2.986594608010953,
      "learning_rate": 5.801177444951148e-07,
      "loss": 0.737,
      "step": 8413
    },
    {
      "epoch": 0.7588041664787843,
      "grad_norm": 1.2082549376652243,
      "learning_rate": 5.797063682995944e-07,
      "loss": 0.7776,
      "step": 8414
    },
    {
      "epoch": 0.7588943500022546,
      "grad_norm": 2.20329305670664,
      "learning_rate": 5.792951132913584e-07,
      "loss": 0.8158,
      "step": 8415
    },
    {
      "epoch": 0.7589845335257248,
      "grad_norm": 1.9177226046546254,
      "learning_rate": 5.788839795054968e-07,
      "loss": 0.8063,
      "step": 8416
    },
    {
      "epoch": 0.7590747170491952,
      "grad_norm": 1.602928732593652,
      "learning_rate": 5.784729669770898e-07,
      "loss": 0.6636,
      "step": 8417
    },
    {
      "epoch": 0.7591649005726654,
      "grad_norm": 1.6011520976408404,
      "learning_rate": 5.780620757412084e-07,
      "loss": 0.7478,
      "step": 8418
    },
    {
      "epoch": 0.7592550840961356,
      "grad_norm": 1.9428452008044645,
      "learning_rate": 5.776513058329098e-07,
      "loss": 0.7672,
      "step": 8419
    },
    {
      "epoch": 0.7593452676196059,
      "grad_norm": 2.414560309874399,
      "learning_rate": 5.772406572872459e-07,
      "loss": 0.7668,
      "step": 8420
    },
    {
      "epoch": 0.7594354511430762,
      "grad_norm": 2.2628129015625724,
      "learning_rate": 5.768301301392535e-07,
      "loss": 0.7738,
      "step": 8421
    },
    {
      "epoch": 0.7595256346665464,
      "grad_norm": 1.5909460496944359,
      "learning_rate": 5.764197244239615e-07,
      "loss": 0.7568,
      "step": 8422
    },
    {
      "epoch": 0.7596158181900167,
      "grad_norm": 1.8387195043673645,
      "learning_rate": 5.760094401763884e-07,
      "loss": 0.7866,
      "step": 8423
    },
    {
      "epoch": 0.7597060017134869,
      "grad_norm": 1.7832588111898275,
      "learning_rate": 5.755992774315414e-07,
      "loss": 0.736,
      "step": 8424
    },
    {
      "epoch": 0.7597961852369572,
      "grad_norm": 1.837524637070099,
      "learning_rate": 5.751892362244183e-07,
      "loss": 0.7366,
      "step": 8425
    },
    {
      "epoch": 0.7598863687604275,
      "grad_norm": 1.861970324373035,
      "learning_rate": 5.747793165900065e-07,
      "loss": 0.8038,
      "step": 8426
    },
    {
      "epoch": 0.7599765522838977,
      "grad_norm": 1.5714266036999756,
      "learning_rate": 5.743695185632806e-07,
      "loss": 0.8038,
      "step": 8427
    },
    {
      "epoch": 0.7600667358073679,
      "grad_norm": 1.9887960850370976,
      "learning_rate": 5.739598421792098e-07,
      "loss": 0.7889,
      "step": 8428
    },
    {
      "epoch": 0.7601569193308383,
      "grad_norm": 2.5314209491789663,
      "learning_rate": 5.735502874727474e-07,
      "loss": 0.7267,
      "step": 8429
    },
    {
      "epoch": 0.7602471028543085,
      "grad_norm": 1.6418482988188126,
      "learning_rate": 5.731408544788398e-07,
      "loss": 0.8757,
      "step": 8430
    },
    {
      "epoch": 0.7603372863777788,
      "grad_norm": 3.410524818105288,
      "learning_rate": 5.727315432324225e-07,
      "loss": 0.6644,
      "step": 8431
    },
    {
      "epoch": 0.760427469901249,
      "grad_norm": 1.9364489504002174,
      "learning_rate": 5.723223537684196e-07,
      "loss": 0.7356,
      "step": 8432
    },
    {
      "epoch": 0.7605176534247193,
      "grad_norm": 2.3058893722968863,
      "learning_rate": 5.719132861217462e-07,
      "loss": 0.8225,
      "step": 8433
    },
    {
      "epoch": 0.7606078369481896,
      "grad_norm": 1.6310260253551265,
      "learning_rate": 5.715043403273044e-07,
      "loss": 0.7131,
      "step": 8434
    },
    {
      "epoch": 0.7606980204716598,
      "grad_norm": 2.3634457869827674,
      "learning_rate": 5.710955164199902e-07,
      "loss": 0.7105,
      "step": 8435
    },
    {
      "epoch": 0.7607882039951301,
      "grad_norm": 1.6942083744982765,
      "learning_rate": 5.706868144346841e-07,
      "loss": 0.7666,
      "step": 8436
    },
    {
      "epoch": 0.7608783875186004,
      "grad_norm": 1.654834358247031,
      "learning_rate": 5.702782344062613e-07,
      "loss": 0.747,
      "step": 8437
    },
    {
      "epoch": 0.7609685710420706,
      "grad_norm": 2.180911250439527,
      "learning_rate": 5.698697763695826e-07,
      "loss": 0.7266,
      "step": 8438
    },
    {
      "epoch": 0.7610587545655408,
      "grad_norm": 2.1550767443239187,
      "learning_rate": 5.694614403595002e-07,
      "loss": 0.7726,
      "step": 8439
    },
    {
      "epoch": 0.7611489380890112,
      "grad_norm": 1.599152104899688,
      "learning_rate": 5.690532264108554e-07,
      "loss": 0.661,
      "step": 8440
    },
    {
      "epoch": 0.7612391216124814,
      "grad_norm": 1.9049556277456463,
      "learning_rate": 5.686451345584795e-07,
      "loss": 0.7882,
      "step": 8441
    },
    {
      "epoch": 0.7613293051359517,
      "grad_norm": 1.865769745852192,
      "learning_rate": 5.682371648371933e-07,
      "loss": 0.8434,
      "step": 8442
    },
    {
      "epoch": 0.7614194886594219,
      "grad_norm": 1.5914735430580629,
      "learning_rate": 5.678293172818074e-07,
      "loss": 0.7099,
      "step": 8443
    },
    {
      "epoch": 0.7615096721828922,
      "grad_norm": 3.950920364739794,
      "learning_rate": 5.674215919271204e-07,
      "loss": 0.7615,
      "step": 8444
    },
    {
      "epoch": 0.7615998557063625,
      "grad_norm": 1.9823827880614024,
      "learning_rate": 5.670139888079224e-07,
      "loss": 0.6776,
      "step": 8445
    },
    {
      "epoch": 0.7616900392298327,
      "grad_norm": 2.3504948521119307,
      "learning_rate": 5.666065079589924e-07,
      "loss": 0.8065,
      "step": 8446
    },
    {
      "epoch": 0.7617802227533029,
      "grad_norm": 1.7372362746287877,
      "learning_rate": 5.661991494150986e-07,
      "loss": 0.6969,
      "step": 8447
    },
    {
      "epoch": 0.7618704062767733,
      "grad_norm": 1.4973729493977153,
      "learning_rate": 5.657919132109999e-07,
      "loss": 0.7932,
      "step": 8448
    },
    {
      "epoch": 0.7619605898002435,
      "grad_norm": 1.9103951334043079,
      "learning_rate": 5.653847993814421e-07,
      "loss": 0.7787,
      "step": 8449
    },
    {
      "epoch": 0.7620507733237137,
      "grad_norm": 1.6961236235080048,
      "learning_rate": 5.649778079611647e-07,
      "loss": 0.7603,
      "step": 8450
    },
    {
      "epoch": 0.762140956847184,
      "grad_norm": 1.9286043779898867,
      "learning_rate": 5.645709389848923e-07,
      "loss": 0.7294,
      "step": 8451
    },
    {
      "epoch": 0.7622311403706543,
      "grad_norm": 1.7361250974303701,
      "learning_rate": 5.641641924873435e-07,
      "loss": 0.7056,
      "step": 8452
    },
    {
      "epoch": 0.7623213238941245,
      "grad_norm": 4.976415706846999,
      "learning_rate": 5.637575685032217e-07,
      "loss": 0.7495,
      "step": 8453
    },
    {
      "epoch": 0.7624115074175948,
      "grad_norm": 1.7050277859837248,
      "learning_rate": 5.633510670672246e-07,
      "loss": 0.828,
      "step": 8454
    },
    {
      "epoch": 0.762501690941065,
      "grad_norm": 0.6225787469930434,
      "learning_rate": 5.629446882140354e-07,
      "loss": 0.605,
      "step": 8455
    },
    {
      "epoch": 0.7625918744645354,
      "grad_norm": 1.4894309421651215,
      "learning_rate": 5.625384319783295e-07,
      "loss": 0.7592,
      "step": 8456
    },
    {
      "epoch": 0.7626820579880056,
      "grad_norm": 2.1932570857367635,
      "learning_rate": 5.621322983947705e-07,
      "loss": 0.7016,
      "step": 8457
    },
    {
      "epoch": 0.7627722415114758,
      "grad_norm": 1.8286759940013944,
      "learning_rate": 5.617262874980122e-07,
      "loss": 0.7844,
      "step": 8458
    },
    {
      "epoch": 0.7628624250349462,
      "grad_norm": 1.665627974459227,
      "learning_rate": 5.613203993226981e-07,
      "loss": 0.793,
      "step": 8459
    },
    {
      "epoch": 0.7629526085584164,
      "grad_norm": 4.809343367900767,
      "learning_rate": 5.609146339034599e-07,
      "loss": 0.7965,
      "step": 8460
    },
    {
      "epoch": 0.7630427920818866,
      "grad_norm": 2.090678224898197,
      "learning_rate": 5.605089912749199e-07,
      "loss": 0.6914,
      "step": 8461
    },
    {
      "epoch": 0.7631329756053569,
      "grad_norm": 2.4769167481262744,
      "learning_rate": 5.601034714716901e-07,
      "loss": 0.6907,
      "step": 8462
    },
    {
      "epoch": 0.7632231591288272,
      "grad_norm": 1.3958410101532375,
      "learning_rate": 5.59698074528372e-07,
      "loss": 0.7204,
      "step": 8463
    },
    {
      "epoch": 0.7633133426522974,
      "grad_norm": 1.5312037558287315,
      "learning_rate": 5.592928004795555e-07,
      "loss": 0.7206,
      "step": 8464
    },
    {
      "epoch": 0.7634035261757677,
      "grad_norm": 4.96999933622728,
      "learning_rate": 5.58887649359822e-07,
      "loss": 0.756,
      "step": 8465
    },
    {
      "epoch": 0.7634937096992379,
      "grad_norm": 1.659011841311513,
      "learning_rate": 5.584826212037393e-07,
      "loss": 0.6012,
      "step": 8466
    },
    {
      "epoch": 0.7635838932227083,
      "grad_norm": 1.8445977024127342,
      "learning_rate": 5.580777160458689e-07,
      "loss": 0.6704,
      "step": 8467
    },
    {
      "epoch": 0.7636740767461785,
      "grad_norm": 1.7283020922131183,
      "learning_rate": 5.576729339207574e-07,
      "loss": 0.7721,
      "step": 8468
    },
    {
      "epoch": 0.7637642602696487,
      "grad_norm": 2.8520552536051436,
      "learning_rate": 5.572682748629449e-07,
      "loss": 0.8093,
      "step": 8469
    },
    {
      "epoch": 0.763854443793119,
      "grad_norm": 1.571777116219112,
      "learning_rate": 5.568637389069582e-07,
      "loss": 0.7311,
      "step": 8470
    },
    {
      "epoch": 0.7639446273165893,
      "grad_norm": 1.6290119969281578,
      "learning_rate": 5.564593260873145e-07,
      "loss": 0.7631,
      "step": 8471
    },
    {
      "epoch": 0.7640348108400595,
      "grad_norm": 1.6900881240937582,
      "learning_rate": 5.560550364385206e-07,
      "loss": 0.7887,
      "step": 8472
    },
    {
      "epoch": 0.7641249943635298,
      "grad_norm": 2.0048592664076414,
      "learning_rate": 5.556508699950728e-07,
      "loss": 0.8161,
      "step": 8473
    },
    {
      "epoch": 0.764215177887,
      "grad_norm": 1.7452219721881375,
      "learning_rate": 5.552468267914577e-07,
      "loss": 0.6992,
      "step": 8474
    },
    {
      "epoch": 0.7643053614104703,
      "grad_norm": 2.253604121593949,
      "learning_rate": 5.548429068621481e-07,
      "loss": 0.6645,
      "step": 8475
    },
    {
      "epoch": 0.7643955449339406,
      "grad_norm": 0.6256901506842782,
      "learning_rate": 5.544391102416115e-07,
      "loss": 0.5749,
      "step": 8476
    },
    {
      "epoch": 0.7644857284574108,
      "grad_norm": 1.8990817134727014,
      "learning_rate": 5.540354369643003e-07,
      "loss": 0.8047,
      "step": 8477
    },
    {
      "epoch": 0.764575911980881,
      "grad_norm": 1.9904894604493726,
      "learning_rate": 5.536318870646586e-07,
      "loss": 0.6037,
      "step": 8478
    },
    {
      "epoch": 0.7646660955043514,
      "grad_norm": 2.420395823965913,
      "learning_rate": 5.532284605771194e-07,
      "loss": 0.6558,
      "step": 8479
    },
    {
      "epoch": 0.7647562790278216,
      "grad_norm": 1.5892032017114601,
      "learning_rate": 5.528251575361052e-07,
      "loss": 0.7123,
      "step": 8480
    },
    {
      "epoch": 0.7648464625512919,
      "grad_norm": 1.4486235761405866,
      "learning_rate": 5.524219779760284e-07,
      "loss": 0.7432,
      "step": 8481
    },
    {
      "epoch": 0.7649366460747622,
      "grad_norm": 1.8120968468955203,
      "learning_rate": 5.520189219312907e-07,
      "loss": 0.774,
      "step": 8482
    },
    {
      "epoch": 0.7650268295982324,
      "grad_norm": 1.83683027921053,
      "learning_rate": 5.516159894362817e-07,
      "loss": 0.6303,
      "step": 8483
    },
    {
      "epoch": 0.7651170131217027,
      "grad_norm": 4.8677940909125805,
      "learning_rate": 5.512131805253839e-07,
      "loss": 0.683,
      "step": 8484
    },
    {
      "epoch": 0.7652071966451729,
      "grad_norm": 2.486639753951138,
      "learning_rate": 5.508104952329653e-07,
      "loss": 0.7581,
      "step": 8485
    },
    {
      "epoch": 0.7652973801686432,
      "grad_norm": 2.1465275823463537,
      "learning_rate": 5.504079335933862e-07,
      "loss": 0.7758,
      "step": 8486
    },
    {
      "epoch": 0.7653875636921135,
      "grad_norm": 1.9411908342982598,
      "learning_rate": 5.500054956409952e-07,
      "loss": 0.7741,
      "step": 8487
    },
    {
      "epoch": 0.7654777472155837,
      "grad_norm": 2.71707777138476,
      "learning_rate": 5.496031814101303e-07,
      "loss": 0.7089,
      "step": 8488
    },
    {
      "epoch": 0.7655679307390539,
      "grad_norm": 3.969061651533868,
      "learning_rate": 5.492009909351203e-07,
      "loss": 0.756,
      "step": 8489
    },
    {
      "epoch": 0.7656581142625243,
      "grad_norm": 1.7562005684300284,
      "learning_rate": 5.4879892425028e-07,
      "loss": 0.6685,
      "step": 8490
    },
    {
      "epoch": 0.7657482977859945,
      "grad_norm": 2.2383632678172187,
      "learning_rate": 5.483969813899184e-07,
      "loss": 0.882,
      "step": 8491
    },
    {
      "epoch": 0.7658384813094647,
      "grad_norm": 1.4240759279389412,
      "learning_rate": 5.479951623883299e-07,
      "loss": 0.8282,
      "step": 8492
    },
    {
      "epoch": 0.765928664832935,
      "grad_norm": 2.130701494530655,
      "learning_rate": 5.475934672798004e-07,
      "loss": 0.6828,
      "step": 8493
    },
    {
      "epoch": 0.7660188483564053,
      "grad_norm": 1.5858423364209657,
      "learning_rate": 5.471918960986047e-07,
      "loss": 0.7118,
      "step": 8494
    },
    {
      "epoch": 0.7661090318798756,
      "grad_norm": 1.6612530843621434,
      "learning_rate": 5.467904488790071e-07,
      "loss": 0.8428,
      "step": 8495
    },
    {
      "epoch": 0.7661992154033458,
      "grad_norm": 2.070693992158517,
      "learning_rate": 5.463891256552615e-07,
      "loss": 0.7559,
      "step": 8496
    },
    {
      "epoch": 0.766289398926816,
      "grad_norm": 1.664557441570534,
      "learning_rate": 5.459879264616107e-07,
      "loss": 0.6507,
      "step": 8497
    },
    {
      "epoch": 0.7663795824502864,
      "grad_norm": 1.908865338371306,
      "learning_rate": 5.455868513322874e-07,
      "loss": 0.6307,
      "step": 8498
    },
    {
      "epoch": 0.7664697659737566,
      "grad_norm": 1.78534375297425,
      "learning_rate": 5.451859003015143e-07,
      "loss": 0.6414,
      "step": 8499
    },
    {
      "epoch": 0.7665599494972268,
      "grad_norm": 1.796414125466219,
      "learning_rate": 5.447850734035009e-07,
      "loss": 0.781,
      "step": 8500
    },
    {
      "epoch": 0.7666501330206971,
      "grad_norm": 1.7245973476452778,
      "learning_rate": 5.443843706724494e-07,
      "loss": 0.726,
      "step": 8501
    },
    {
      "epoch": 0.7667403165441674,
      "grad_norm": 1.6424812351821299,
      "learning_rate": 5.439837921425494e-07,
      "loss": 0.6743,
      "step": 8502
    },
    {
      "epoch": 0.7668305000676376,
      "grad_norm": 5.401236971711661,
      "learning_rate": 5.435833378479807e-07,
      "loss": 0.7332,
      "step": 8503
    },
    {
      "epoch": 0.7669206835911079,
      "grad_norm": 1.7052272456135398,
      "learning_rate": 5.431830078229128e-07,
      "loss": 0.607,
      "step": 8504
    },
    {
      "epoch": 0.7670108671145782,
      "grad_norm": 1.7175777946300073,
      "learning_rate": 5.427828021015022e-07,
      "loss": 0.7628,
      "step": 8505
    },
    {
      "epoch": 0.7671010506380485,
      "grad_norm": 2.0431253581077375,
      "learning_rate": 5.42382720717899e-07,
      "loss": 0.7764,
      "step": 8506
    },
    {
      "epoch": 0.7671912341615187,
      "grad_norm": 3.3508023909517566,
      "learning_rate": 5.419827637062384e-07,
      "loss": 0.777,
      "step": 8507
    },
    {
      "epoch": 0.7672814176849889,
      "grad_norm": 1.8063043394791722,
      "learning_rate": 5.415829311006487e-07,
      "loss": 0.8895,
      "step": 8508
    },
    {
      "epoch": 0.7673716012084593,
      "grad_norm": 1.7759552950473483,
      "learning_rate": 5.411832229352447e-07,
      "loss": 0.8288,
      "step": 8509
    },
    {
      "epoch": 0.7674617847319295,
      "grad_norm": 1.9806000254083567,
      "learning_rate": 5.407836392441319e-07,
      "loss": 0.8433,
      "step": 8510
    },
    {
      "epoch": 0.7675519682553997,
      "grad_norm": 1.5047625593278275,
      "learning_rate": 5.403841800614049e-07,
      "loss": 0.7604,
      "step": 8511
    },
    {
      "epoch": 0.76764215177887,
      "grad_norm": 1.3665050438160777,
      "learning_rate": 5.39984845421148e-07,
      "loss": 0.7028,
      "step": 8512
    },
    {
      "epoch": 0.7677323353023403,
      "grad_norm": 2.651964388025685,
      "learning_rate": 5.395856353574344e-07,
      "loss": 0.73,
      "step": 8513
    },
    {
      "epoch": 0.7678225188258105,
      "grad_norm": 0.591405934544105,
      "learning_rate": 5.391865499043275e-07,
      "loss": 0.5665,
      "step": 8514
    },
    {
      "epoch": 0.7679127023492808,
      "grad_norm": 1.5989060118954737,
      "learning_rate": 5.387875890958788e-07,
      "loss": 0.7531,
      "step": 8515
    },
    {
      "epoch": 0.768002885872751,
      "grad_norm": 1.4901682987787999,
      "learning_rate": 5.383887529661298e-07,
      "loss": 0.7709,
      "step": 8516
    },
    {
      "epoch": 0.7680930693962214,
      "grad_norm": 2.0652000786603417,
      "learning_rate": 5.379900415491116e-07,
      "loss": 0.7043,
      "step": 8517
    },
    {
      "epoch": 0.7681832529196916,
      "grad_norm": 1.9768896020717235,
      "learning_rate": 5.375914548788447e-07,
      "loss": 0.695,
      "step": 8518
    },
    {
      "epoch": 0.7682734364431618,
      "grad_norm": 1.602967478726254,
      "learning_rate": 5.371929929893384e-07,
      "loss": 0.6252,
      "step": 8519
    },
    {
      "epoch": 0.768363619966632,
      "grad_norm": 1.8688908552587091,
      "learning_rate": 5.367946559145917e-07,
      "loss": 0.7085,
      "step": 8520
    },
    {
      "epoch": 0.7684538034901024,
      "grad_norm": 1.5000707291775945,
      "learning_rate": 5.363964436885935e-07,
      "loss": 0.7652,
      "step": 8521
    },
    {
      "epoch": 0.7685439870135726,
      "grad_norm": 1.3917068388084315,
      "learning_rate": 5.359983563453199e-07,
      "loss": 0.8292,
      "step": 8522
    },
    {
      "epoch": 0.7686341705370429,
      "grad_norm": 1.8090001220281338,
      "learning_rate": 5.356003939187402e-07,
      "loss": 0.8613,
      "step": 8523
    },
    {
      "epoch": 0.7687243540605131,
      "grad_norm": 1.6907111098382663,
      "learning_rate": 5.352025564428082e-07,
      "loss": 0.7237,
      "step": 8524
    },
    {
      "epoch": 0.7688145375839834,
      "grad_norm": 1.6933669075803846,
      "learning_rate": 5.348048439514723e-07,
      "loss": 0.788,
      "step": 8525
    },
    {
      "epoch": 0.7689047211074537,
      "grad_norm": 1.860722414233792,
      "learning_rate": 5.344072564786653e-07,
      "loss": 0.6274,
      "step": 8526
    },
    {
      "epoch": 0.7689949046309239,
      "grad_norm": 1.4575728568029778,
      "learning_rate": 5.340097940583123e-07,
      "loss": 0.7578,
      "step": 8527
    },
    {
      "epoch": 0.7690850881543942,
      "grad_norm": 2.850492260900837,
      "learning_rate": 5.336124567243275e-07,
      "loss": 0.7063,
      "step": 8528
    },
    {
      "epoch": 0.7691752716778645,
      "grad_norm": 1.8224873791167375,
      "learning_rate": 5.33215244510613e-07,
      "loss": 0.7518,
      "step": 8529
    },
    {
      "epoch": 0.7692654552013347,
      "grad_norm": 1.768083194630837,
      "learning_rate": 5.328181574510624e-07,
      "loss": 0.7433,
      "step": 8530
    },
    {
      "epoch": 0.769355638724805,
      "grad_norm": 1.7319450195210935,
      "learning_rate": 5.324211955795559e-07,
      "loss": 0.7949,
      "step": 8531
    },
    {
      "epoch": 0.7694458222482753,
      "grad_norm": 2.69752981782825,
      "learning_rate": 5.320243589299651e-07,
      "loss": 0.6755,
      "step": 8532
    },
    {
      "epoch": 0.7695360057717455,
      "grad_norm": 1.7231486129388112,
      "learning_rate": 5.316276475361505e-07,
      "loss": 0.7027,
      "step": 8533
    },
    {
      "epoch": 0.7696261892952158,
      "grad_norm": 8.357350441167606,
      "learning_rate": 5.312310614319613e-07,
      "loss": 0.7886,
      "step": 8534
    },
    {
      "epoch": 0.769716372818686,
      "grad_norm": 3.148421748952536,
      "learning_rate": 5.308346006512367e-07,
      "loss": 0.7203,
      "step": 8535
    },
    {
      "epoch": 0.7698065563421563,
      "grad_norm": 1.6468324485652635,
      "learning_rate": 5.30438265227805e-07,
      "loss": 0.678,
      "step": 8536
    },
    {
      "epoch": 0.7698967398656266,
      "grad_norm": 0.6169067661798605,
      "learning_rate": 5.300420551954837e-07,
      "loss": 0.5604,
      "step": 8537
    },
    {
      "epoch": 0.7699869233890968,
      "grad_norm": 1.4348283072181347,
      "learning_rate": 5.296459705880798e-07,
      "loss": 0.6746,
      "step": 8538
    },
    {
      "epoch": 0.770077106912567,
      "grad_norm": 1.4488356256072277,
      "learning_rate": 5.292500114393881e-07,
      "loss": 0.7995,
      "step": 8539
    },
    {
      "epoch": 0.7701672904360374,
      "grad_norm": 2.1588268638758734,
      "learning_rate": 5.288541777831963e-07,
      "loss": 0.7712,
      "step": 8540
    },
    {
      "epoch": 0.7702574739595076,
      "grad_norm": 1.5050520062253234,
      "learning_rate": 5.284584696532772e-07,
      "loss": 0.799,
      "step": 8541
    },
    {
      "epoch": 0.7703476574829778,
      "grad_norm": 1.6770184269383481,
      "learning_rate": 5.280628870833954e-07,
      "loss": 0.7509,
      "step": 8542
    },
    {
      "epoch": 0.7704378410064481,
      "grad_norm": 1.6465961603882335,
      "learning_rate": 5.276674301073045e-07,
      "loss": 0.7755,
      "step": 8543
    },
    {
      "epoch": 0.7705280245299184,
      "grad_norm": 1.4643531893425907,
      "learning_rate": 5.272720987587467e-07,
      "loss": 0.7623,
      "step": 8544
    },
    {
      "epoch": 0.7706182080533887,
      "grad_norm": 2.5539068960128906,
      "learning_rate": 5.268768930714545e-07,
      "loss": 0.7018,
      "step": 8545
    },
    {
      "epoch": 0.7707083915768589,
      "grad_norm": 1.5091390358164647,
      "learning_rate": 5.264818130791473e-07,
      "loss": 0.7764,
      "step": 8546
    },
    {
      "epoch": 0.7707985751003291,
      "grad_norm": 1.7641727923234494,
      "learning_rate": 5.260868588155378e-07,
      "loss": 0.7546,
      "step": 8547
    },
    {
      "epoch": 0.7708887586237995,
      "grad_norm": 1.3783976884107418,
      "learning_rate": 5.256920303143242e-07,
      "loss": 0.7753,
      "step": 8548
    },
    {
      "epoch": 0.7709789421472697,
      "grad_norm": 1.3802500976385583,
      "learning_rate": 5.252973276091956e-07,
      "loss": 0.8591,
      "step": 8549
    },
    {
      "epoch": 0.7710691256707399,
      "grad_norm": 1.5187424522180952,
      "learning_rate": 5.249027507338307e-07,
      "loss": 0.8177,
      "step": 8550
    },
    {
      "epoch": 0.7711593091942102,
      "grad_norm": 1.41440034487103,
      "learning_rate": 5.245082997218966e-07,
      "loss": 0.719,
      "step": 8551
    },
    {
      "epoch": 0.7712494927176805,
      "grad_norm": 1.7651239545051975,
      "learning_rate": 5.241139746070499e-07,
      "loss": 0.7725,
      "step": 8552
    },
    {
      "epoch": 0.7713396762411507,
      "grad_norm": 1.480217582714901,
      "learning_rate": 5.237197754229376e-07,
      "loss": 0.6284,
      "step": 8553
    },
    {
      "epoch": 0.771429859764621,
      "grad_norm": 3.174106057542244,
      "learning_rate": 5.233257022031931e-07,
      "loss": 0.6536,
      "step": 8554
    },
    {
      "epoch": 0.7715200432880913,
      "grad_norm": 2.3493323412505935,
      "learning_rate": 5.229317549814432e-07,
      "loss": 0.7592,
      "step": 8555
    },
    {
      "epoch": 0.7716102268115616,
      "grad_norm": 2.3461141234182405,
      "learning_rate": 5.225379337912998e-07,
      "loss": 0.8271,
      "step": 8556
    },
    {
      "epoch": 0.7717004103350318,
      "grad_norm": 1.6607931912335898,
      "learning_rate": 5.221442386663663e-07,
      "loss": 0.7602,
      "step": 8557
    },
    {
      "epoch": 0.771790593858502,
      "grad_norm": 1.4710973284774587,
      "learning_rate": 5.217506696402354e-07,
      "loss": 0.7877,
      "step": 8558
    },
    {
      "epoch": 0.7718807773819724,
      "grad_norm": 3.3906955975333197,
      "learning_rate": 5.213572267464883e-07,
      "loss": 0.7675,
      "step": 8559
    },
    {
      "epoch": 0.7719709609054426,
      "grad_norm": 3.588741361527259,
      "learning_rate": 5.209639100186965e-07,
      "loss": 0.7241,
      "step": 8560
    },
    {
      "epoch": 0.7720611444289128,
      "grad_norm": 0.6946997567904656,
      "learning_rate": 5.205707194904179e-07,
      "loss": 0.6412,
      "step": 8561
    },
    {
      "epoch": 0.7721513279523831,
      "grad_norm": 1.9676174963411865,
      "learning_rate": 5.201776551952042e-07,
      "loss": 0.6729,
      "step": 8562
    },
    {
      "epoch": 0.7722415114758534,
      "grad_norm": 2.3930957467818503,
      "learning_rate": 5.197847171665914e-07,
      "loss": 0.7125,
      "step": 8563
    },
    {
      "epoch": 0.7723316949993236,
      "grad_norm": 1.9792608004486492,
      "learning_rate": 5.193919054381095e-07,
      "loss": 0.7361,
      "step": 8564
    },
    {
      "epoch": 0.7724218785227939,
      "grad_norm": 1.6998041657240237,
      "learning_rate": 5.189992200432738e-07,
      "loss": 0.7123,
      "step": 8565
    },
    {
      "epoch": 0.7725120620462641,
      "grad_norm": 2.5323424630907083,
      "learning_rate": 5.186066610155906e-07,
      "loss": 0.763,
      "step": 8566
    },
    {
      "epoch": 0.7726022455697344,
      "grad_norm": 1.766831989770129,
      "learning_rate": 5.182142283885555e-07,
      "loss": 0.8017,
      "step": 8567
    },
    {
      "epoch": 0.7726924290932047,
      "grad_norm": 0.6630467219266584,
      "learning_rate": 5.178219221956528e-07,
      "loss": 0.5313,
      "step": 8568
    },
    {
      "epoch": 0.7727826126166749,
      "grad_norm": 1.6882377001058577,
      "learning_rate": 5.174297424703565e-07,
      "loss": 0.7676,
      "step": 8569
    },
    {
      "epoch": 0.7728727961401451,
      "grad_norm": 1.6395852563532498,
      "learning_rate": 5.170376892461299e-07,
      "loss": 0.6777,
      "step": 8570
    },
    {
      "epoch": 0.7729629796636155,
      "grad_norm": 0.7245286757781512,
      "learning_rate": 5.16645762556424e-07,
      "loss": 0.6309,
      "step": 8571
    },
    {
      "epoch": 0.7730531631870857,
      "grad_norm": 1.511899794351193,
      "learning_rate": 5.162539624346809e-07,
      "loss": 0.7939,
      "step": 8572
    },
    {
      "epoch": 0.773143346710556,
      "grad_norm": 2.5908255578831656,
      "learning_rate": 5.158622889143309e-07,
      "loss": 0.8087,
      "step": 8573
    },
    {
      "epoch": 0.7732335302340262,
      "grad_norm": 1.8598775745422482,
      "learning_rate": 5.154707420287939e-07,
      "loss": 0.7067,
      "step": 8574
    },
    {
      "epoch": 0.7733237137574965,
      "grad_norm": 2.574474825589482,
      "learning_rate": 5.150793218114793e-07,
      "loss": 0.6759,
      "step": 8575
    },
    {
      "epoch": 0.7734138972809668,
      "grad_norm": 1.9355816727328914,
      "learning_rate": 5.146880282957837e-07,
      "loss": 0.7376,
      "step": 8576
    },
    {
      "epoch": 0.773504080804437,
      "grad_norm": 1.624941091203303,
      "learning_rate": 5.142968615150964e-07,
      "loss": 0.7364,
      "step": 8577
    },
    {
      "epoch": 0.7735942643279073,
      "grad_norm": 3.765608031681816,
      "learning_rate": 5.139058215027921e-07,
      "loss": 0.6778,
      "step": 8578
    },
    {
      "epoch": 0.7736844478513776,
      "grad_norm": 1.3820018655209572,
      "learning_rate": 5.135149082922383e-07,
      "loss": 0.7338,
      "step": 8579
    },
    {
      "epoch": 0.7737746313748478,
      "grad_norm": 5.983021874498908,
      "learning_rate": 5.131241219167879e-07,
      "loss": 0.6776,
      "step": 8580
    },
    {
      "epoch": 0.773864814898318,
      "grad_norm": 0.8074418861334399,
      "learning_rate": 5.127334624097869e-07,
      "loss": 0.6229,
      "step": 8581
    },
    {
      "epoch": 0.7739549984217884,
      "grad_norm": 2.4218302076566016,
      "learning_rate": 5.123429298045672e-07,
      "loss": 0.8022,
      "step": 8582
    },
    {
      "epoch": 0.7740451819452586,
      "grad_norm": 1.5218867307363602,
      "learning_rate": 5.119525241344515e-07,
      "loss": 0.6365,
      "step": 8583
    },
    {
      "epoch": 0.7741353654687289,
      "grad_norm": 1.584716101049618,
      "learning_rate": 5.115622454327515e-07,
      "loss": 0.7527,
      "step": 8584
    },
    {
      "epoch": 0.7742255489921991,
      "grad_norm": 2.7347424069571282,
      "learning_rate": 5.11172093732768e-07,
      "loss": 0.7326,
      "step": 8585
    },
    {
      "epoch": 0.7743157325156694,
      "grad_norm": 1.7522404497833086,
      "learning_rate": 5.107820690677911e-07,
      "loss": 0.7749,
      "step": 8586
    },
    {
      "epoch": 0.7744059160391397,
      "grad_norm": 1.4978052613377726,
      "learning_rate": 5.103921714710991e-07,
      "loss": 0.6935,
      "step": 8587
    },
    {
      "epoch": 0.7744960995626099,
      "grad_norm": 1.9411100165404012,
      "learning_rate": 5.100024009759605e-07,
      "loss": 0.7608,
      "step": 8588
    },
    {
      "epoch": 0.7745862830860801,
      "grad_norm": 1.9400106591000665,
      "learning_rate": 5.09612757615633e-07,
      "loss": 0.7841,
      "step": 8589
    },
    {
      "epoch": 0.7746764666095505,
      "grad_norm": 1.6598639735825098,
      "learning_rate": 5.092232414233628e-07,
      "loss": 0.7001,
      "step": 8590
    },
    {
      "epoch": 0.7747666501330207,
      "grad_norm": 1.7819889610453814,
      "learning_rate": 5.088338524323858e-07,
      "loss": 0.7196,
      "step": 8591
    },
    {
      "epoch": 0.7748568336564909,
      "grad_norm": 2.5295503349416024,
      "learning_rate": 5.084445906759271e-07,
      "loss": 0.8067,
      "step": 8592
    },
    {
      "epoch": 0.7749470171799612,
      "grad_norm": 1.6715882670437816,
      "learning_rate": 5.080554561871995e-07,
      "loss": 0.809,
      "step": 8593
    },
    {
      "epoch": 0.7750372007034315,
      "grad_norm": 1.4466587704268683,
      "learning_rate": 5.076664489994078e-07,
      "loss": 0.7778,
      "step": 8594
    },
    {
      "epoch": 0.7751273842269017,
      "grad_norm": 1.850116019863832,
      "learning_rate": 5.07277569145742e-07,
      "loss": 0.7159,
      "step": 8595
    },
    {
      "epoch": 0.775217567750372,
      "grad_norm": 3.431363402335748,
      "learning_rate": 5.068888166593861e-07,
      "loss": 0.6649,
      "step": 8596
    },
    {
      "epoch": 0.7753077512738422,
      "grad_norm": 1.7826000250831002,
      "learning_rate": 5.065001915735087e-07,
      "loss": 0.7384,
      "step": 8597
    },
    {
      "epoch": 0.7753979347973126,
      "grad_norm": 5.098998004991493,
      "learning_rate": 5.061116939212702e-07,
      "loss": 0.7114,
      "step": 8598
    },
    {
      "epoch": 0.7754881183207828,
      "grad_norm": 1.7736318204515933,
      "learning_rate": 5.05723323735819e-07,
      "loss": 0.7436,
      "step": 8599
    },
    {
      "epoch": 0.775578301844253,
      "grad_norm": 0.6251621751187475,
      "learning_rate": 5.053350810502932e-07,
      "loss": 0.5837,
      "step": 8600
    },
    {
      "epoch": 0.7756684853677234,
      "grad_norm": 2.599471342920478,
      "learning_rate": 5.049469658978202e-07,
      "loss": 0.7412,
      "step": 8601
    },
    {
      "epoch": 0.7757586688911936,
      "grad_norm": 0.6989244875456864,
      "learning_rate": 5.045589783115147e-07,
      "loss": 0.5317,
      "step": 8602
    },
    {
      "epoch": 0.7758488524146638,
      "grad_norm": 10.290030489654894,
      "learning_rate": 5.041711183244842e-07,
      "loss": 0.7889,
      "step": 8603
    },
    {
      "epoch": 0.7759390359381341,
      "grad_norm": 4.053137216424583,
      "learning_rate": 5.037833859698211e-07,
      "loss": 0.7631,
      "step": 8604
    },
    {
      "epoch": 0.7760292194616044,
      "grad_norm": 1.7330734593032076,
      "learning_rate": 5.033957812806096e-07,
      "loss": 0.8131,
      "step": 8605
    },
    {
      "epoch": 0.7761194029850746,
      "grad_norm": 2.3871991911720563,
      "learning_rate": 5.030083042899223e-07,
      "loss": 0.7956,
      "step": 8606
    },
    {
      "epoch": 0.7762095865085449,
      "grad_norm": 1.727330654030806,
      "learning_rate": 5.026209550308207e-07,
      "loss": 0.7424,
      "step": 8607
    },
    {
      "epoch": 0.7762997700320151,
      "grad_norm": 4.516270574865451,
      "learning_rate": 5.022337335363558e-07,
      "loss": 0.8075,
      "step": 8608
    },
    {
      "epoch": 0.7763899535554855,
      "grad_norm": 1.5780577692497824,
      "learning_rate": 5.018466398395677e-07,
      "loss": 0.7719,
      "step": 8609
    },
    {
      "epoch": 0.7764801370789557,
      "grad_norm": 4.780702048424025,
      "learning_rate": 5.01459673973484e-07,
      "loss": 0.6859,
      "step": 8610
    },
    {
      "epoch": 0.7765703206024259,
      "grad_norm": 2.2360222353230594,
      "learning_rate": 5.01072835971125e-07,
      "loss": 0.8152,
      "step": 8611
    },
    {
      "epoch": 0.7766605041258962,
      "grad_norm": 1.766170248727003,
      "learning_rate": 5.006861258654959e-07,
      "loss": 0.6522,
      "step": 8612
    },
    {
      "epoch": 0.7767506876493665,
      "grad_norm": 2.1653206263194384,
      "learning_rate": 5.002995436895938e-07,
      "loss": 0.7767,
      "step": 8613
    },
    {
      "epoch": 0.7768408711728367,
      "grad_norm": 1.9540514770372603,
      "learning_rate": 4.999130894764039e-07,
      "loss": 0.7055,
      "step": 8614
    },
    {
      "epoch": 0.776931054696307,
      "grad_norm": 1.8119290866786089,
      "learning_rate": 4.995267632589006e-07,
      "loss": 0.6916,
      "step": 8615
    },
    {
      "epoch": 0.7770212382197772,
      "grad_norm": 0.7177774234583203,
      "learning_rate": 4.99140565070048e-07,
      "loss": 0.581,
      "step": 8616
    },
    {
      "epoch": 0.7771114217432475,
      "grad_norm": 1.4173968714584115,
      "learning_rate": 4.987544949427969e-07,
      "loss": 0.7691,
      "step": 8617
    },
    {
      "epoch": 0.7772016052667178,
      "grad_norm": 0.6608049264248516,
      "learning_rate": 4.98368552910091e-07,
      "loss": 0.5881,
      "step": 8618
    },
    {
      "epoch": 0.777291788790188,
      "grad_norm": 4.009528255729855,
      "learning_rate": 4.979827390048596e-07,
      "loss": 0.8631,
      "step": 8619
    },
    {
      "epoch": 0.7773819723136582,
      "grad_norm": 1.5309564737303236,
      "learning_rate": 4.975970532600231e-07,
      "loss": 0.7139,
      "step": 8620
    },
    {
      "epoch": 0.7774721558371286,
      "grad_norm": 2.9006857159964503,
      "learning_rate": 4.972114957084901e-07,
      "loss": 0.7701,
      "step": 8621
    },
    {
      "epoch": 0.7775623393605988,
      "grad_norm": 2.495543226683589,
      "learning_rate": 4.968260663831585e-07,
      "loss": 0.7269,
      "step": 8622
    },
    {
      "epoch": 0.777652522884069,
      "grad_norm": 1.3279929151489671,
      "learning_rate": 4.964407653169154e-07,
      "loss": 0.8581,
      "step": 8623
    },
    {
      "epoch": 0.7777427064075394,
      "grad_norm": 1.4047838409279774,
      "learning_rate": 4.960555925426366e-07,
      "loss": 0.7149,
      "step": 8624
    },
    {
      "epoch": 0.7778328899310096,
      "grad_norm": 2.06312089301957,
      "learning_rate": 4.956705480931876e-07,
      "loss": 0.8047,
      "step": 8625
    },
    {
      "epoch": 0.7779230734544799,
      "grad_norm": 1.3619435040076078,
      "learning_rate": 4.952856320014225e-07,
      "loss": 0.7275,
      "step": 8626
    },
    {
      "epoch": 0.7780132569779501,
      "grad_norm": 1.3467592211744535,
      "learning_rate": 4.949008443001838e-07,
      "loss": 0.6854,
      "step": 8627
    },
    {
      "epoch": 0.7781034405014204,
      "grad_norm": 1.72517677728177,
      "learning_rate": 4.945161850223041e-07,
      "loss": 0.7599,
      "step": 8628
    },
    {
      "epoch": 0.7781936240248907,
      "grad_norm": 2.1349208597349794,
      "learning_rate": 4.941316542006044e-07,
      "loss": 0.6763,
      "step": 8629
    },
    {
      "epoch": 0.7782838075483609,
      "grad_norm": 1.3845918059379214,
      "learning_rate": 4.937472518678956e-07,
      "loss": 0.7708,
      "step": 8630
    },
    {
      "epoch": 0.7783739910718311,
      "grad_norm": 1.994831081495667,
      "learning_rate": 4.93362978056977e-07,
      "loss": 0.746,
      "step": 8631
    },
    {
      "epoch": 0.7784641745953015,
      "grad_norm": 1.7329799093104363,
      "learning_rate": 4.929788328006355e-07,
      "loss": 0.6989,
      "step": 8632
    },
    {
      "epoch": 0.7785543581187717,
      "grad_norm": 2.0241081864238137,
      "learning_rate": 4.925948161316506e-07,
      "loss": 0.8192,
      "step": 8633
    },
    {
      "epoch": 0.778644541642242,
      "grad_norm": 1.6553687144177933,
      "learning_rate": 4.922109280827868e-07,
      "loss": 0.736,
      "step": 8634
    },
    {
      "epoch": 0.7787347251657122,
      "grad_norm": 2.0061855032467713,
      "learning_rate": 4.918271686868016e-07,
      "loss": 0.7418,
      "step": 8635
    },
    {
      "epoch": 0.7788249086891825,
      "grad_norm": 1.9415453855213651,
      "learning_rate": 4.914435379764379e-07,
      "loss": 0.79,
      "step": 8636
    },
    {
      "epoch": 0.7789150922126528,
      "grad_norm": 1.4463026624371127,
      "learning_rate": 4.910600359844294e-07,
      "loss": 0.7389,
      "step": 8637
    },
    {
      "epoch": 0.779005275736123,
      "grad_norm": 2.0438704242282966,
      "learning_rate": 4.90676662743499e-07,
      "loss": 0.6554,
      "step": 8638
    },
    {
      "epoch": 0.7790954592595932,
      "grad_norm": 1.7675057580792035,
      "learning_rate": 4.902934182863581e-07,
      "loss": 0.7415,
      "step": 8639
    },
    {
      "epoch": 0.7791856427830636,
      "grad_norm": 2.0440190316077556,
      "learning_rate": 4.899103026457069e-07,
      "loss": 0.6234,
      "step": 8640
    },
    {
      "epoch": 0.7792758263065338,
      "grad_norm": 0.592184185532124,
      "learning_rate": 4.895273158542361e-07,
      "loss": 0.5306,
      "step": 8641
    },
    {
      "epoch": 0.779366009830004,
      "grad_norm": 1.6062110985227247,
      "learning_rate": 4.891444579446227e-07,
      "loss": 0.7614,
      "step": 8642
    },
    {
      "epoch": 0.7794561933534743,
      "grad_norm": 2.1391039268210443,
      "learning_rate": 4.887617289495349e-07,
      "loss": 0.7037,
      "step": 8643
    },
    {
      "epoch": 0.7795463768769446,
      "grad_norm": 0.5934040668022907,
      "learning_rate": 4.883791289016292e-07,
      "loss": 0.5118,
      "step": 8644
    },
    {
      "epoch": 0.7796365604004148,
      "grad_norm": 2.4071274185655827,
      "learning_rate": 4.879966578335514e-07,
      "loss": 0.7549,
      "step": 8645
    },
    {
      "epoch": 0.7797267439238851,
      "grad_norm": 1.9352849788917201,
      "learning_rate": 4.876143157779358e-07,
      "loss": 0.6435,
      "step": 8646
    },
    {
      "epoch": 0.7798169274473554,
      "grad_norm": 1.9524767601007498,
      "learning_rate": 4.872321027674058e-07,
      "loss": 0.6371,
      "step": 8647
    },
    {
      "epoch": 0.7799071109708257,
      "grad_norm": 1.810674208162321,
      "learning_rate": 4.868500188345748e-07,
      "loss": 0.8186,
      "step": 8648
    },
    {
      "epoch": 0.7799972944942959,
      "grad_norm": 1.6645542188381195,
      "learning_rate": 4.864680640120425e-07,
      "loss": 0.8082,
      "step": 8649
    },
    {
      "epoch": 0.7800874780177661,
      "grad_norm": 2.2073466657909706,
      "learning_rate": 4.860862383324016e-07,
      "loss": 0.7558,
      "step": 8650
    },
    {
      "epoch": 0.7801776615412365,
      "grad_norm": 1.479242949690271,
      "learning_rate": 4.857045418282295e-07,
      "loss": 0.7398,
      "step": 8651
    },
    {
      "epoch": 0.7802678450647067,
      "grad_norm": 2.1800821145367544,
      "learning_rate": 4.853229745320966e-07,
      "loss": 0.6959,
      "step": 8652
    },
    {
      "epoch": 0.7803580285881769,
      "grad_norm": 1.712843438676244,
      "learning_rate": 4.849415364765587e-07,
      "loss": 0.7306,
      "step": 8653
    },
    {
      "epoch": 0.7804482121116472,
      "grad_norm": 2.1750648181947105,
      "learning_rate": 4.845602276941631e-07,
      "loss": 0.8074,
      "step": 8654
    },
    {
      "epoch": 0.7805383956351175,
      "grad_norm": 1.5946131594084407,
      "learning_rate": 4.841790482174449e-07,
      "loss": 0.7903,
      "step": 8655
    },
    {
      "epoch": 0.7806285791585877,
      "grad_norm": 1.8204430136327803,
      "learning_rate": 4.837979980789282e-07,
      "loss": 0.7383,
      "step": 8656
    },
    {
      "epoch": 0.780718762682058,
      "grad_norm": 1.6126053620448118,
      "learning_rate": 4.834170773111273e-07,
      "loss": 0.688,
      "step": 8657
    },
    {
      "epoch": 0.7808089462055282,
      "grad_norm": 1.6347419983171692,
      "learning_rate": 4.830362859465431e-07,
      "loss": 0.7653,
      "step": 8658
    },
    {
      "epoch": 0.7808991297289986,
      "grad_norm": 1.9614090675437696,
      "learning_rate": 4.826556240176675e-07,
      "loss": 0.8129,
      "step": 8659
    },
    {
      "epoch": 0.7809893132524688,
      "grad_norm": 1.7983076272399368,
      "learning_rate": 4.822750915569807e-07,
      "loss": 0.7645,
      "step": 8660
    },
    {
      "epoch": 0.781079496775939,
      "grad_norm": 3.437707235851929,
      "learning_rate": 4.818946885969514e-07,
      "loss": 0.821,
      "step": 8661
    },
    {
      "epoch": 0.7811696802994093,
      "grad_norm": 3.0805936102603906,
      "learning_rate": 4.815144151700383e-07,
      "loss": 0.7907,
      "step": 8662
    },
    {
      "epoch": 0.7812598638228796,
      "grad_norm": 2.218947925604702,
      "learning_rate": 4.811342713086885e-07,
      "loss": 0.7489,
      "step": 8663
    },
    {
      "epoch": 0.7813500473463498,
      "grad_norm": 2.352574634401831,
      "learning_rate": 4.807542570453367e-07,
      "loss": 0.7869,
      "step": 8664
    },
    {
      "epoch": 0.7814402308698201,
      "grad_norm": 1.7684012672445661,
      "learning_rate": 4.803743724124098e-07,
      "loss": 0.7135,
      "step": 8665
    },
    {
      "epoch": 0.7815304143932903,
      "grad_norm": 1.9793183787239101,
      "learning_rate": 4.799946174423192e-07,
      "loss": 0.7548,
      "step": 8666
    },
    {
      "epoch": 0.7816205979167606,
      "grad_norm": 0.5954394898423907,
      "learning_rate": 4.796149921674706e-07,
      "loss": 0.5326,
      "step": 8667
    },
    {
      "epoch": 0.7817107814402309,
      "grad_norm": 1.5845743719759182,
      "learning_rate": 4.792354966202534e-07,
      "loss": 0.7046,
      "step": 8668
    },
    {
      "epoch": 0.7818009649637011,
      "grad_norm": 2.6825385316497132,
      "learning_rate": 4.788561308330489e-07,
      "loss": 0.5763,
      "step": 8669
    },
    {
      "epoch": 0.7818911484871713,
      "grad_norm": 2.2435253020173933,
      "learning_rate": 4.784768948382272e-07,
      "loss": 0.6961,
      "step": 8670
    },
    {
      "epoch": 0.7819813320106417,
      "grad_norm": 1.5067292111983148,
      "learning_rate": 4.780977886681461e-07,
      "loss": 0.7539,
      "step": 8671
    },
    {
      "epoch": 0.7820715155341119,
      "grad_norm": 1.8595111059814708,
      "learning_rate": 4.777188123551541e-07,
      "loss": 0.6964,
      "step": 8672
    },
    {
      "epoch": 0.7821616990575821,
      "grad_norm": 2.222375835301608,
      "learning_rate": 4.773399659315856e-07,
      "loss": 0.7777,
      "step": 8673
    },
    {
      "epoch": 0.7822518825810525,
      "grad_norm": 1.97360955799867,
      "learning_rate": 4.769612494297681e-07,
      "loss": 0.7285,
      "step": 8674
    },
    {
      "epoch": 0.7823420661045227,
      "grad_norm": 0.6427443037266595,
      "learning_rate": 4.765826628820142e-07,
      "loss": 0.5549,
      "step": 8675
    },
    {
      "epoch": 0.782432249627993,
      "grad_norm": 1.6707269880487237,
      "learning_rate": 4.7620420632062775e-07,
      "loss": 0.6595,
      "step": 8676
    },
    {
      "epoch": 0.7825224331514632,
      "grad_norm": 2.1358201056745654,
      "learning_rate": 4.758258797779002e-07,
      "loss": 0.704,
      "step": 8677
    },
    {
      "epoch": 0.7826126166749335,
      "grad_norm": 0.6167947026239253,
      "learning_rate": 4.7544768328611317e-07,
      "loss": 0.5651,
      "step": 8678
    },
    {
      "epoch": 0.7827028001984038,
      "grad_norm": 2.013858819788224,
      "learning_rate": 4.750696168775359e-07,
      "loss": 0.7293,
      "step": 8679
    },
    {
      "epoch": 0.782792983721874,
      "grad_norm": 0.6019759800735239,
      "learning_rate": 4.746916805844279e-07,
      "loss": 0.5525,
      "step": 8680
    },
    {
      "epoch": 0.7828831672453442,
      "grad_norm": 1.470356488760643,
      "learning_rate": 4.743138744390356e-07,
      "loss": 0.7749,
      "step": 8681
    },
    {
      "epoch": 0.7829733507688146,
      "grad_norm": 1.6662593184809382,
      "learning_rate": 4.739361984735959e-07,
      "loss": 0.8235,
      "step": 8682
    },
    {
      "epoch": 0.7830635342922848,
      "grad_norm": 3.0502257091617095,
      "learning_rate": 4.7355865272033455e-07,
      "loss": 0.7023,
      "step": 8683
    },
    {
      "epoch": 0.783153717815755,
      "grad_norm": 1.6083398100701145,
      "learning_rate": 4.7318123721146563e-07,
      "loss": 0.7425,
      "step": 8684
    },
    {
      "epoch": 0.7832439013392253,
      "grad_norm": 1.4944542250299402,
      "learning_rate": 4.728039519791924e-07,
      "loss": 0.7365,
      "step": 8685
    },
    {
      "epoch": 0.7833340848626956,
      "grad_norm": 0.6641565592853083,
      "learning_rate": 4.72426797055707e-07,
      "loss": 0.5989,
      "step": 8686
    },
    {
      "epoch": 0.7834242683861659,
      "grad_norm": 1.9379615849131269,
      "learning_rate": 4.720497724731904e-07,
      "loss": 0.7091,
      "step": 8687
    },
    {
      "epoch": 0.7835144519096361,
      "grad_norm": 1.7768988754702562,
      "learning_rate": 4.7167287826381153e-07,
      "loss": 0.7827,
      "step": 8688
    },
    {
      "epoch": 0.7836046354331063,
      "grad_norm": 1.653217057835971,
      "learning_rate": 4.712961144597307e-07,
      "loss": 0.7505,
      "step": 8689
    },
    {
      "epoch": 0.7836948189565767,
      "grad_norm": 2.206398770338771,
      "learning_rate": 4.7091948109309343e-07,
      "loss": 0.7893,
      "step": 8690
    },
    {
      "epoch": 0.7837850024800469,
      "grad_norm": 1.8629370797384905,
      "learning_rate": 4.705429781960384e-07,
      "loss": 0.7549,
      "step": 8691
    },
    {
      "epoch": 0.7838751860035171,
      "grad_norm": 1.7103713679694328,
      "learning_rate": 4.7016660580068923e-07,
      "loss": 0.7446,
      "step": 8692
    },
    {
      "epoch": 0.7839653695269874,
      "grad_norm": 1.7416576041513632,
      "learning_rate": 4.6979036393916093e-07,
      "loss": 0.7762,
      "step": 8693
    },
    {
      "epoch": 0.7840555530504577,
      "grad_norm": 1.6958197612672368,
      "learning_rate": 4.6941425264355603e-07,
      "loss": 0.8224,
      "step": 8694
    },
    {
      "epoch": 0.7841457365739279,
      "grad_norm": 1.5705971531006546,
      "learning_rate": 4.6903827194596666e-07,
      "loss": 0.7618,
      "step": 8695
    },
    {
      "epoch": 0.7842359200973982,
      "grad_norm": 2.0630364153963012,
      "learning_rate": 4.686624218784743e-07,
      "loss": 0.8034,
      "step": 8696
    },
    {
      "epoch": 0.7843261036208685,
      "grad_norm": 2.3013449012462663,
      "learning_rate": 4.6828670247314696e-07,
      "loss": 0.744,
      "step": 8697
    },
    {
      "epoch": 0.7844162871443388,
      "grad_norm": 2.251411630948507,
      "learning_rate": 4.679111137620442e-07,
      "loss": 0.761,
      "step": 8698
    },
    {
      "epoch": 0.784506470667809,
      "grad_norm": 1.6418186749924473,
      "learning_rate": 4.67535655777213e-07,
      "loss": 0.7101,
      "step": 8699
    },
    {
      "epoch": 0.7845966541912792,
      "grad_norm": 1.45953407809236,
      "learning_rate": 4.6716032855068956e-07,
      "loss": 0.7543,
      "step": 8700
    },
    {
      "epoch": 0.7846868377147496,
      "grad_norm": 2.267701138257335,
      "learning_rate": 4.6678513211449867e-07,
      "loss": 0.7889,
      "step": 8701
    },
    {
      "epoch": 0.7847770212382198,
      "grad_norm": 1.7508033543373973,
      "learning_rate": 4.6641006650065516e-07,
      "loss": 0.7184,
      "step": 8702
    },
    {
      "epoch": 0.78486720476169,
      "grad_norm": 3.667131712350613,
      "learning_rate": 4.6603513174115973e-07,
      "loss": 0.7331,
      "step": 8703
    },
    {
      "epoch": 0.7849573882851603,
      "grad_norm": 2.7590319825296223,
      "learning_rate": 4.6566032786800625e-07,
      "loss": 0.7628,
      "step": 8704
    },
    {
      "epoch": 0.7850475718086306,
      "grad_norm": 1.607151532528184,
      "learning_rate": 4.6528565491317274e-07,
      "loss": 0.7294,
      "step": 8705
    },
    {
      "epoch": 0.7851377553321008,
      "grad_norm": 2.0628178669444033,
      "learning_rate": 4.649111129086305e-07,
      "loss": 0.8147,
      "step": 8706
    },
    {
      "epoch": 0.7852279388555711,
      "grad_norm": 1.6962523777607028,
      "learning_rate": 4.6453670188633596e-07,
      "loss": 0.7754,
      "step": 8707
    },
    {
      "epoch": 0.7853181223790413,
      "grad_norm": 2.0568919376961383,
      "learning_rate": 4.641624218782365e-07,
      "loss": 0.7594,
      "step": 8708
    },
    {
      "epoch": 0.7854083059025116,
      "grad_norm": 1.4479670710217922,
      "learning_rate": 4.6378827291626765e-07,
      "loss": 0.7167,
      "step": 8709
    },
    {
      "epoch": 0.7854984894259819,
      "grad_norm": 1.3543231531349416,
      "learning_rate": 4.634142550323541e-07,
      "loss": 0.8013,
      "step": 8710
    },
    {
      "epoch": 0.7855886729494521,
      "grad_norm": 3.334000536583924,
      "learning_rate": 4.6304036825840943e-07,
      "loss": 0.7974,
      "step": 8711
    },
    {
      "epoch": 0.7856788564729223,
      "grad_norm": 1.4068523812223386,
      "learning_rate": 4.626666126263341e-07,
      "loss": 0.6527,
      "step": 8712
    },
    {
      "epoch": 0.7857690399963927,
      "grad_norm": 1.6887386156174204,
      "learning_rate": 4.622929881680213e-07,
      "loss": 0.772,
      "step": 8713
    },
    {
      "epoch": 0.7858592235198629,
      "grad_norm": 4.8401146813669795,
      "learning_rate": 4.6191949491534887e-07,
      "loss": 0.6652,
      "step": 8714
    },
    {
      "epoch": 0.7859494070433332,
      "grad_norm": 1.8306839468423592,
      "learning_rate": 4.6154613290018617e-07,
      "loss": 0.7582,
      "step": 8715
    },
    {
      "epoch": 0.7860395905668034,
      "grad_norm": 1.6520487204057017,
      "learning_rate": 4.6117290215439043e-07,
      "loss": 0.6977,
      "step": 8716
    },
    {
      "epoch": 0.7861297740902737,
      "grad_norm": 2.4218325703503565,
      "learning_rate": 4.6079980270980744e-07,
      "loss": 0.6118,
      "step": 8717
    },
    {
      "epoch": 0.786219957613744,
      "grad_norm": 3.374476922084468,
      "learning_rate": 4.6042683459827245e-07,
      "loss": 0.7238,
      "step": 8718
    },
    {
      "epoch": 0.7863101411372142,
      "grad_norm": 0.7037134674983901,
      "learning_rate": 4.600539978516098e-07,
      "loss": 0.5635,
      "step": 8719
    },
    {
      "epoch": 0.7864003246606845,
      "grad_norm": 1.7941858486503204,
      "learning_rate": 4.5968129250163004e-07,
      "loss": 0.7093,
      "step": 8720
    },
    {
      "epoch": 0.7864905081841548,
      "grad_norm": 1.6791904867421625,
      "learning_rate": 4.5930871858013653e-07,
      "loss": 0.6815,
      "step": 8721
    },
    {
      "epoch": 0.786580691707625,
      "grad_norm": 0.65441019016437,
      "learning_rate": 4.589362761189182e-07,
      "loss": 0.6212,
      "step": 8722
    },
    {
      "epoch": 0.7866708752310952,
      "grad_norm": 1.6325432222261447,
      "learning_rate": 4.585639651497539e-07,
      "loss": 0.8151,
      "step": 8723
    },
    {
      "epoch": 0.7867610587545656,
      "grad_norm": 1.5040256997763597,
      "learning_rate": 4.581917857044115e-07,
      "loss": 0.702,
      "step": 8724
    },
    {
      "epoch": 0.7868512422780358,
      "grad_norm": 2.0651760658040215,
      "learning_rate": 4.5781973781464734e-07,
      "loss": 0.7831,
      "step": 8725
    },
    {
      "epoch": 0.7869414258015061,
      "grad_norm": 2.2499690583538707,
      "learning_rate": 4.574478215122073e-07,
      "loss": 0.7871,
      "step": 8726
    },
    {
      "epoch": 0.7870316093249763,
      "grad_norm": 2.553606463715014,
      "learning_rate": 4.5707603682882357e-07,
      "loss": 0.7284,
      "step": 8727
    },
    {
      "epoch": 0.7871217928484466,
      "grad_norm": 3.885329967414412,
      "learning_rate": 4.56704383796221e-07,
      "loss": 0.6149,
      "step": 8728
    },
    {
      "epoch": 0.7872119763719169,
      "grad_norm": 1.9967231728588146,
      "learning_rate": 4.5633286244610956e-07,
      "loss": 0.7868,
      "step": 8729
    },
    {
      "epoch": 0.7873021598953871,
      "grad_norm": 1.4236882982289052,
      "learning_rate": 4.5596147281018993e-07,
      "loss": 0.813,
      "step": 8730
    },
    {
      "epoch": 0.7873923434188573,
      "grad_norm": 1.8112112264649196,
      "learning_rate": 4.5559021492015137e-07,
      "loss": 0.8532,
      "step": 8731
    },
    {
      "epoch": 0.7874825269423277,
      "grad_norm": 1.4191571764685604,
      "learning_rate": 4.552190888076712e-07,
      "loss": 0.6567,
      "step": 8732
    },
    {
      "epoch": 0.7875727104657979,
      "grad_norm": 1.5848677461483567,
      "learning_rate": 4.548480945044164e-07,
      "loss": 0.7003,
      "step": 8733
    },
    {
      "epoch": 0.7876628939892681,
      "grad_norm": 1.569265747399692,
      "learning_rate": 4.54477232042042e-07,
      "loss": 0.7448,
      "step": 8734
    },
    {
      "epoch": 0.7877530775127384,
      "grad_norm": 1.5297133851710536,
      "learning_rate": 4.541065014521921e-07,
      "loss": 0.7457,
      "step": 8735
    },
    {
      "epoch": 0.7878432610362087,
      "grad_norm": 3.6656992821024805,
      "learning_rate": 4.5373590276649996e-07,
      "loss": 0.6978,
      "step": 8736
    },
    {
      "epoch": 0.787933444559679,
      "grad_norm": 1.390868647801113,
      "learning_rate": 4.533654360165862e-07,
      "loss": 0.7171,
      "step": 8737
    },
    {
      "epoch": 0.7880236280831492,
      "grad_norm": 1.805025605736016,
      "learning_rate": 4.5299510123406115e-07,
      "loss": 0.8025,
      "step": 8738
    },
    {
      "epoch": 0.7881138116066194,
      "grad_norm": 1.450169007547624,
      "learning_rate": 4.5262489845052456e-07,
      "loss": 0.7179,
      "step": 8739
    },
    {
      "epoch": 0.7882039951300898,
      "grad_norm": 1.6259025121557638,
      "learning_rate": 4.5225482769756353e-07,
      "loss": 0.7136,
      "step": 8740
    },
    {
      "epoch": 0.78829417865356,
      "grad_norm": 2.3519914305589227,
      "learning_rate": 4.5188488900675545e-07,
      "loss": 0.6775,
      "step": 8741
    },
    {
      "epoch": 0.7883843621770302,
      "grad_norm": 1.9646056620629158,
      "learning_rate": 4.5151508240966363e-07,
      "loss": 0.7798,
      "step": 8742
    },
    {
      "epoch": 0.7884745457005006,
      "grad_norm": 1.6325495019858742,
      "learning_rate": 4.511454079378445e-07,
      "loss": 0.7912,
      "step": 8743
    },
    {
      "epoch": 0.7885647292239708,
      "grad_norm": 1.9879546431197057,
      "learning_rate": 4.507758656228382e-07,
      "loss": 0.6555,
      "step": 8744
    },
    {
      "epoch": 0.788654912747441,
      "grad_norm": 2.7329110722679144,
      "learning_rate": 4.5040645549617864e-07,
      "loss": 0.8109,
      "step": 8745
    },
    {
      "epoch": 0.7887450962709113,
      "grad_norm": 1.5338301427370629,
      "learning_rate": 4.5003717758938384e-07,
      "loss": 0.6638,
      "step": 8746
    },
    {
      "epoch": 0.7888352797943816,
      "grad_norm": 1.6273657377445199,
      "learning_rate": 4.4966803193396365e-07,
      "loss": 0.6876,
      "step": 8747
    },
    {
      "epoch": 0.7889254633178518,
      "grad_norm": 2.0706343490615207,
      "learning_rate": 4.492990185614154e-07,
      "loss": 0.7546,
      "step": 8748
    },
    {
      "epoch": 0.7890156468413221,
      "grad_norm": 1.512680733341421,
      "learning_rate": 4.489301375032255e-07,
      "loss": 0.7387,
      "step": 8749
    },
    {
      "epoch": 0.7891058303647923,
      "grad_norm": 2.4286331882157155,
      "learning_rate": 4.4856138879086857e-07,
      "loss": 0.7535,
      "step": 8750
    },
    {
      "epoch": 0.7891960138882627,
      "grad_norm": 3.0359133999702856,
      "learning_rate": 4.481927724558092e-07,
      "loss": 0.6964,
      "step": 8751
    },
    {
      "epoch": 0.7892861974117329,
      "grad_norm": 1.5672713861045338,
      "learning_rate": 4.478242885294985e-07,
      "loss": 0.7165,
      "step": 8752
    },
    {
      "epoch": 0.7893763809352031,
      "grad_norm": 2.5092940187985784,
      "learning_rate": 4.474559370433779e-07,
      "loss": 0.7096,
      "step": 8753
    },
    {
      "epoch": 0.7894665644586734,
      "grad_norm": 3.0898743060080474,
      "learning_rate": 4.470877180288777e-07,
      "loss": 0.6935,
      "step": 8754
    },
    {
      "epoch": 0.7895567479821437,
      "grad_norm": 3.7923101920079305,
      "learning_rate": 4.4671963151741574e-07,
      "loss": 0.7164,
      "step": 8755
    },
    {
      "epoch": 0.7896469315056139,
      "grad_norm": 1.420902464556149,
      "learning_rate": 4.4635167754039973e-07,
      "loss": 0.7231,
      "step": 8756
    },
    {
      "epoch": 0.7897371150290842,
      "grad_norm": 1.3793659254500688,
      "learning_rate": 4.459838561292253e-07,
      "loss": 0.7699,
      "step": 8757
    },
    {
      "epoch": 0.7898272985525544,
      "grad_norm": 2.8192193192509594,
      "learning_rate": 4.456161673152774e-07,
      "loss": 0.6668,
      "step": 8758
    },
    {
      "epoch": 0.7899174820760247,
      "grad_norm": 1.6125394210136756,
      "learning_rate": 4.4524861112992806e-07,
      "loss": 0.7358,
      "step": 8759
    },
    {
      "epoch": 0.790007665599495,
      "grad_norm": 1.789038445590031,
      "learning_rate": 4.448811876045411e-07,
      "loss": 0.7431,
      "step": 8760
    },
    {
      "epoch": 0.7900978491229652,
      "grad_norm": 2.7182552720615805,
      "learning_rate": 4.445138967704647e-07,
      "loss": 0.8156,
      "step": 8761
    },
    {
      "epoch": 0.7901880326464354,
      "grad_norm": 6.026146503438844,
      "learning_rate": 4.4414673865904075e-07,
      "loss": 0.7194,
      "step": 8762
    },
    {
      "epoch": 0.7902782161699058,
      "grad_norm": 1.4684697147528656,
      "learning_rate": 4.437797133015955e-07,
      "loss": 0.7609,
      "step": 8763
    },
    {
      "epoch": 0.790368399693376,
      "grad_norm": 1.7909199097328146,
      "learning_rate": 4.4341282072944586e-07,
      "loss": 0.6673,
      "step": 8764
    },
    {
      "epoch": 0.7904585832168463,
      "grad_norm": 2.346739528113693,
      "learning_rate": 4.430460609738973e-07,
      "loss": 0.7675,
      "step": 8765
    },
    {
      "epoch": 0.7905487667403166,
      "grad_norm": 0.7128509928516924,
      "learning_rate": 4.4267943406624386e-07,
      "loss": 0.6098,
      "step": 8766
    },
    {
      "epoch": 0.7906389502637868,
      "grad_norm": 1.855316862615539,
      "learning_rate": 4.4231294003776853e-07,
      "loss": 0.8565,
      "step": 8767
    },
    {
      "epoch": 0.7907291337872571,
      "grad_norm": 3.1390032020691057,
      "learning_rate": 4.419465789197416e-07,
      "loss": 0.7873,
      "step": 8768
    },
    {
      "epoch": 0.7908193173107273,
      "grad_norm": 1.646104943120167,
      "learning_rate": 4.415803507434237e-07,
      "loss": 0.7377,
      "step": 8769
    },
    {
      "epoch": 0.7909095008341976,
      "grad_norm": 1.7233681797117577,
      "learning_rate": 4.4121425554006307e-07,
      "loss": 0.7651,
      "step": 8770
    },
    {
      "epoch": 0.7909996843576679,
      "grad_norm": 1.9149154319715487,
      "learning_rate": 4.4084829334089744e-07,
      "loss": 0.7306,
      "step": 8771
    },
    {
      "epoch": 0.7910898678811381,
      "grad_norm": 1.5313839756490315,
      "learning_rate": 4.404824641771525e-07,
      "loss": 0.6858,
      "step": 8772
    },
    {
      "epoch": 0.7911800514046083,
      "grad_norm": 1.5495771446559188,
      "learning_rate": 4.4011676808004327e-07,
      "loss": 0.8125,
      "step": 8773
    },
    {
      "epoch": 0.7912702349280787,
      "grad_norm": 6.832809831368819,
      "learning_rate": 4.3975120508077145e-07,
      "loss": 0.7547,
      "step": 8774
    },
    {
      "epoch": 0.7913604184515489,
      "grad_norm": 1.6595449242092695,
      "learning_rate": 4.39385775210531e-07,
      "loss": 0.7929,
      "step": 8775
    },
    {
      "epoch": 0.7914506019750192,
      "grad_norm": 1.8408459568008189,
      "learning_rate": 4.390204785005003e-07,
      "loss": 0.7625,
      "step": 8776
    },
    {
      "epoch": 0.7915407854984894,
      "grad_norm": 1.4297815510816596,
      "learning_rate": 4.386553149818504e-07,
      "loss": 0.7457,
      "step": 8777
    },
    {
      "epoch": 0.7916309690219597,
      "grad_norm": 2.4428207816219896,
      "learning_rate": 4.3829028468573793e-07,
      "loss": 0.6125,
      "step": 8778
    },
    {
      "epoch": 0.79172115254543,
      "grad_norm": 2.140610186671772,
      "learning_rate": 4.3792538764330935e-07,
      "loss": 0.6843,
      "step": 8779
    },
    {
      "epoch": 0.7918113360689002,
      "grad_norm": 1.802626121965576,
      "learning_rate": 4.3756062388569994e-07,
      "loss": 0.7228,
      "step": 8780
    },
    {
      "epoch": 0.7919015195923704,
      "grad_norm": 0.7314155684612795,
      "learning_rate": 4.3719599344403346e-07,
      "loss": 0.5802,
      "step": 8781
    },
    {
      "epoch": 0.7919917031158408,
      "grad_norm": 1.4109256156682795,
      "learning_rate": 4.3683149634942243e-07,
      "loss": 0.7262,
      "step": 8782
    },
    {
      "epoch": 0.792081886639311,
      "grad_norm": 6.376945049286617,
      "learning_rate": 4.364671326329663e-07,
      "loss": 0.6254,
      "step": 8783
    },
    {
      "epoch": 0.7921720701627812,
      "grad_norm": 1.948857520503587,
      "learning_rate": 4.3610290232575673e-07,
      "loss": 0.7642,
      "step": 8784
    },
    {
      "epoch": 0.7922622536862515,
      "grad_norm": 2.313529738994174,
      "learning_rate": 4.357388054588702e-07,
      "loss": 0.7498,
      "step": 8785
    },
    {
      "epoch": 0.7923524372097218,
      "grad_norm": 1.646465912304737,
      "learning_rate": 4.3537484206337405e-07,
      "loss": 0.6891,
      "step": 8786
    },
    {
      "epoch": 0.792442620733192,
      "grad_norm": 1.7617854002125166,
      "learning_rate": 4.3501101217032366e-07,
      "loss": 0.776,
      "step": 8787
    },
    {
      "epoch": 0.7925328042566623,
      "grad_norm": 1.8964357222534372,
      "learning_rate": 4.346473158107629e-07,
      "loss": 0.6758,
      "step": 8788
    },
    {
      "epoch": 0.7926229877801325,
      "grad_norm": 2.0499405363809347,
      "learning_rate": 4.342837530157244e-07,
      "loss": 0.7494,
      "step": 8789
    },
    {
      "epoch": 0.7927131713036029,
      "grad_norm": 1.721467107240872,
      "learning_rate": 4.3392032381622987e-07,
      "loss": 0.6465,
      "step": 8790
    },
    {
      "epoch": 0.7928033548270731,
      "grad_norm": 1.4297442815792683,
      "learning_rate": 4.3355702824328765e-07,
      "loss": 0.7058,
      "step": 8791
    },
    {
      "epoch": 0.7928935383505433,
      "grad_norm": 1.4774469219555892,
      "learning_rate": 4.3319386632789823e-07,
      "loss": 0.7877,
      "step": 8792
    },
    {
      "epoch": 0.7929837218740137,
      "grad_norm": 1.8220514333104492,
      "learning_rate": 4.328308381010466e-07,
      "loss": 0.74,
      "step": 8793
    },
    {
      "epoch": 0.7930739053974839,
      "grad_norm": 1.5606955217553478,
      "learning_rate": 4.3246794359370933e-07,
      "loss": 0.7511,
      "step": 8794
    },
    {
      "epoch": 0.7931640889209541,
      "grad_norm": 2.6015844029143205,
      "learning_rate": 4.3210518283685025e-07,
      "loss": 0.7231,
      "step": 8795
    },
    {
      "epoch": 0.7932542724444244,
      "grad_norm": 6.2020665570144216,
      "learning_rate": 4.317425558614225e-07,
      "loss": 0.7406,
      "step": 8796
    },
    {
      "epoch": 0.7933444559678947,
      "grad_norm": 1.3523699162341207,
      "learning_rate": 4.3138006269836744e-07,
      "loss": 0.7076,
      "step": 8797
    },
    {
      "epoch": 0.793434639491365,
      "grad_norm": 1.9458135243073946,
      "learning_rate": 4.3101770337861376e-07,
      "loss": 0.7326,
      "step": 8798
    },
    {
      "epoch": 0.7935248230148352,
      "grad_norm": 0.6723476122745871,
      "learning_rate": 4.30655477933082e-07,
      "loss": 0.5942,
      "step": 8799
    },
    {
      "epoch": 0.7936150065383054,
      "grad_norm": 1.571481829211406,
      "learning_rate": 4.30293386392677e-07,
      "loss": 0.7695,
      "step": 8800
    },
    {
      "epoch": 0.7937051900617758,
      "grad_norm": 1.52425014633621,
      "learning_rate": 4.299314287882967e-07,
      "loss": 0.7629,
      "step": 8801
    },
    {
      "epoch": 0.793795373585246,
      "grad_norm": 2.0318252922555895,
      "learning_rate": 4.2956960515082353e-07,
      "loss": 0.7776,
      "step": 8802
    },
    {
      "epoch": 0.7938855571087162,
      "grad_norm": 1.5292246132801872,
      "learning_rate": 4.29207915511131e-07,
      "loss": 0.7032,
      "step": 8803
    },
    {
      "epoch": 0.7939757406321865,
      "grad_norm": 1.8953173078332917,
      "learning_rate": 4.2884635990008024e-07,
      "loss": 0.804,
      "step": 8804
    },
    {
      "epoch": 0.7940659241556568,
      "grad_norm": 2.201612453828009,
      "learning_rate": 4.284849383485214e-07,
      "loss": 0.7113,
      "step": 8805
    },
    {
      "epoch": 0.794156107679127,
      "grad_norm": 2.003572967952237,
      "learning_rate": 4.2812365088729296e-07,
      "loss": 0.7658,
      "step": 8806
    },
    {
      "epoch": 0.7942462912025973,
      "grad_norm": 1.5508236050526982,
      "learning_rate": 4.2776249754722227e-07,
      "loss": 0.7633,
      "step": 8807
    },
    {
      "epoch": 0.7943364747260675,
      "grad_norm": 2.1510445607853246,
      "learning_rate": 4.27401478359124e-07,
      "loss": 0.8139,
      "step": 8808
    },
    {
      "epoch": 0.7944266582495378,
      "grad_norm": 3.11102680440236,
      "learning_rate": 4.2704059335380283e-07,
      "loss": 0.8463,
      "step": 8809
    },
    {
      "epoch": 0.7945168417730081,
      "grad_norm": 1.470724604679009,
      "learning_rate": 4.266798425620515e-07,
      "loss": 0.8469,
      "step": 8810
    },
    {
      "epoch": 0.7946070252964783,
      "grad_norm": 1.5296689649010327,
      "learning_rate": 4.263192260146511e-07,
      "loss": 0.7331,
      "step": 8811
    },
    {
      "epoch": 0.7946972088199485,
      "grad_norm": 3.340599786504476,
      "learning_rate": 4.2595874374237216e-07,
      "loss": 0.7943,
      "step": 8812
    },
    {
      "epoch": 0.7947873923434189,
      "grad_norm": 1.3986158204002022,
      "learning_rate": 4.255983957759712e-07,
      "loss": 0.7419,
      "step": 8813
    },
    {
      "epoch": 0.7948775758668891,
      "grad_norm": 1.3619692372336376,
      "learning_rate": 4.2523818214619745e-07,
      "loss": 0.6836,
      "step": 8814
    },
    {
      "epoch": 0.7949677593903594,
      "grad_norm": 1.6091140239183876,
      "learning_rate": 4.24878102883784e-07,
      "loss": 0.7927,
      "step": 8815
    },
    {
      "epoch": 0.7950579429138297,
      "grad_norm": 1.5094521259955902,
      "learning_rate": 4.24518158019457e-07,
      "loss": 0.7513,
      "step": 8816
    },
    {
      "epoch": 0.7951481264372999,
      "grad_norm": 1.6231627715559176,
      "learning_rate": 4.241583475839274e-07,
      "loss": 0.6863,
      "step": 8817
    },
    {
      "epoch": 0.7952383099607702,
      "grad_norm": 1.9094543415573113,
      "learning_rate": 4.237986716078965e-07,
      "loss": 0.791,
      "step": 8818
    },
    {
      "epoch": 0.7953284934842404,
      "grad_norm": 3.044374192797108,
      "learning_rate": 4.2343913012205433e-07,
      "loss": 0.7309,
      "step": 8819
    },
    {
      "epoch": 0.7954186770077107,
      "grad_norm": 1.7435103385056465,
      "learning_rate": 4.230797231570784e-07,
      "loss": 0.721,
      "step": 8820
    },
    {
      "epoch": 0.795508860531181,
      "grad_norm": 2.326648013232347,
      "learning_rate": 4.227204507436357e-07,
      "loss": 0.7504,
      "step": 8821
    },
    {
      "epoch": 0.7955990440546512,
      "grad_norm": 1.41331757371741,
      "learning_rate": 4.223613129123811e-07,
      "loss": 0.6635,
      "step": 8822
    },
    {
      "epoch": 0.7956892275781214,
      "grad_norm": 1.6788418082485952,
      "learning_rate": 4.220023096939589e-07,
      "loss": 0.6526,
      "step": 8823
    },
    {
      "epoch": 0.7957794111015918,
      "grad_norm": 1.8070822675984888,
      "learning_rate": 4.21643441119e-07,
      "loss": 0.7392,
      "step": 8824
    },
    {
      "epoch": 0.795869594625062,
      "grad_norm": 0.6269965231067438,
      "learning_rate": 4.212847072181256e-07,
      "loss": 0.6301,
      "step": 8825
    },
    {
      "epoch": 0.7959597781485322,
      "grad_norm": 1.545250395943431,
      "learning_rate": 4.2092610802194505e-07,
      "loss": 0.7128,
      "step": 8826
    },
    {
      "epoch": 0.7960499616720025,
      "grad_norm": 1.5020365400428517,
      "learning_rate": 4.2056764356105587e-07,
      "loss": 0.6588,
      "step": 8827
    },
    {
      "epoch": 0.7961401451954728,
      "grad_norm": 1.71620371078892,
      "learning_rate": 4.202093138660443e-07,
      "loss": 0.7898,
      "step": 8828
    },
    {
      "epoch": 0.7962303287189431,
      "grad_norm": 6.202699737961418,
      "learning_rate": 4.198511189674854e-07,
      "loss": 0.779,
      "step": 8829
    },
    {
      "epoch": 0.7963205122424133,
      "grad_norm": 1.5860459614764932,
      "learning_rate": 4.1949305889594066e-07,
      "loss": 0.7595,
      "step": 8830
    },
    {
      "epoch": 0.7964106957658835,
      "grad_norm": 1.5794253138327585,
      "learning_rate": 4.191351336819642e-07,
      "loss": 0.7366,
      "step": 8831
    },
    {
      "epoch": 0.7965008792893539,
      "grad_norm": 1.4846010537936347,
      "learning_rate": 4.187773433560939e-07,
      "loss": 0.6853,
      "step": 8832
    },
    {
      "epoch": 0.7965910628128241,
      "grad_norm": 2.5594585325887786,
      "learning_rate": 4.184196879488604e-07,
      "loss": 0.7445,
      "step": 8833
    },
    {
      "epoch": 0.7966812463362943,
      "grad_norm": 2.553715698937961,
      "learning_rate": 4.1806216749077936e-07,
      "loss": 0.8077,
      "step": 8834
    },
    {
      "epoch": 0.7967714298597646,
      "grad_norm": 1.7371747213278963,
      "learning_rate": 4.177047820123569e-07,
      "loss": 0.7879,
      "step": 8835
    },
    {
      "epoch": 0.7968616133832349,
      "grad_norm": 1.6586656767764516,
      "learning_rate": 4.1734753154408733e-07,
      "loss": 0.6174,
      "step": 8836
    },
    {
      "epoch": 0.7969517969067051,
      "grad_norm": 4.5169848866052575,
      "learning_rate": 4.169904161164528e-07,
      "loss": 0.7271,
      "step": 8837
    },
    {
      "epoch": 0.7970419804301754,
      "grad_norm": 2.126151614326521,
      "learning_rate": 4.1663343575992526e-07,
      "loss": 0.8584,
      "step": 8838
    },
    {
      "epoch": 0.7971321639536457,
      "grad_norm": 2.154701008246045,
      "learning_rate": 4.1627659050496275e-07,
      "loss": 0.7617,
      "step": 8839
    },
    {
      "epoch": 0.797222347477116,
      "grad_norm": 1.5509615776472785,
      "learning_rate": 4.1591988038201453e-07,
      "loss": 0.6555,
      "step": 8840
    },
    {
      "epoch": 0.7973125310005862,
      "grad_norm": 3.052562550146777,
      "learning_rate": 4.155633054215164e-07,
      "loss": 0.7902,
      "step": 8841
    },
    {
      "epoch": 0.7974027145240564,
      "grad_norm": 2.157114491416056,
      "learning_rate": 4.152068656538934e-07,
      "loss": 0.6527,
      "step": 8842
    },
    {
      "epoch": 0.7974928980475268,
      "grad_norm": 1.4670982000537913,
      "learning_rate": 4.148505611095594e-07,
      "loss": 0.8034,
      "step": 8843
    },
    {
      "epoch": 0.797583081570997,
      "grad_norm": 1.940381030781506,
      "learning_rate": 4.1449439181891563e-07,
      "loss": 0.6251,
      "step": 8844
    },
    {
      "epoch": 0.7976732650944672,
      "grad_norm": 1.618502022738131,
      "learning_rate": 4.14138357812353e-07,
      "loss": 0.7626,
      "step": 8845
    },
    {
      "epoch": 0.7977634486179375,
      "grad_norm": 1.3760317486105487,
      "learning_rate": 4.137824591202506e-07,
      "loss": 0.7968,
      "step": 8846
    },
    {
      "epoch": 0.7978536321414078,
      "grad_norm": 4.245337902410084,
      "learning_rate": 4.134266957729737e-07,
      "loss": 0.7311,
      "step": 8847
    },
    {
      "epoch": 0.797943815664878,
      "grad_norm": 1.7015971058530814,
      "learning_rate": 4.1307106780088065e-07,
      "loss": 0.8345,
      "step": 8848
    },
    {
      "epoch": 0.7980339991883483,
      "grad_norm": 1.4704381291131166,
      "learning_rate": 4.1271557523431387e-07,
      "loss": 0.7558,
      "step": 8849
    },
    {
      "epoch": 0.7981241827118185,
      "grad_norm": 1.5884616102047024,
      "learning_rate": 4.1236021810360634e-07,
      "loss": 0.7314,
      "step": 8850
    },
    {
      "epoch": 0.7982143662352889,
      "grad_norm": 2.4278619415084433,
      "learning_rate": 4.120049964390793e-07,
      "loss": 0.6903,
      "step": 8851
    },
    {
      "epoch": 0.7983045497587591,
      "grad_norm": 1.7592939449631309,
      "learning_rate": 4.116499102710418e-07,
      "loss": 0.6937,
      "step": 8852
    },
    {
      "epoch": 0.7983947332822293,
      "grad_norm": 1.9978111926080038,
      "learning_rate": 4.112949596297928e-07,
      "loss": 0.7174,
      "step": 8853
    },
    {
      "epoch": 0.7984849168056996,
      "grad_norm": 1.7290178334478616,
      "learning_rate": 4.1094014454561664e-07,
      "loss": 0.6866,
      "step": 8854
    },
    {
      "epoch": 0.7985751003291699,
      "grad_norm": 1.9191368880993538,
      "learning_rate": 4.1058546504879057e-07,
      "loss": 0.6485,
      "step": 8855
    },
    {
      "epoch": 0.7986652838526401,
      "grad_norm": 1.983406070636315,
      "learning_rate": 4.1023092116957583e-07,
      "loss": 0.8076,
      "step": 8856
    },
    {
      "epoch": 0.7987554673761104,
      "grad_norm": 1.509229083540741,
      "learning_rate": 4.098765129382249e-07,
      "loss": 0.7851,
      "step": 8857
    },
    {
      "epoch": 0.7988456508995806,
      "grad_norm": 1.8579132360766195,
      "learning_rate": 4.0952224038497764e-07,
      "loss": 0.816,
      "step": 8858
    },
    {
      "epoch": 0.7989358344230509,
      "grad_norm": 2.735131731225793,
      "learning_rate": 4.091681035400627e-07,
      "loss": 0.7417,
      "step": 8859
    },
    {
      "epoch": 0.7990260179465212,
      "grad_norm": 1.750194402523547,
      "learning_rate": 4.088141024336971e-07,
      "loss": 0.7787,
      "step": 8860
    },
    {
      "epoch": 0.7991162014699914,
      "grad_norm": 1.4078011329183506,
      "learning_rate": 4.0846023709608636e-07,
      "loss": 0.8218,
      "step": 8861
    },
    {
      "epoch": 0.7992063849934617,
      "grad_norm": 2.374541288549047,
      "learning_rate": 4.081065075574226e-07,
      "loss": 0.8208,
      "step": 8862
    },
    {
      "epoch": 0.799296568516932,
      "grad_norm": 2.1012672869296294,
      "learning_rate": 4.077529138478906e-07,
      "loss": 0.6755,
      "step": 8863
    },
    {
      "epoch": 0.7993867520404022,
      "grad_norm": 0.6806728849564423,
      "learning_rate": 4.073994559976588e-07,
      "loss": 0.5417,
      "step": 8864
    },
    {
      "epoch": 0.7994769355638724,
      "grad_norm": 2.2745706677111857,
      "learning_rate": 4.0704613403688716e-07,
      "loss": 0.708,
      "step": 8865
    },
    {
      "epoch": 0.7995671190873428,
      "grad_norm": 2.2803544677452394,
      "learning_rate": 4.0669294799572264e-07,
      "loss": 0.8015,
      "step": 8866
    },
    {
      "epoch": 0.799657302610813,
      "grad_norm": 5.023558619627523,
      "learning_rate": 4.0633989790430113e-07,
      "loss": 0.8472,
      "step": 8867
    },
    {
      "epoch": 0.7997474861342833,
      "grad_norm": 2.035041791219654,
      "learning_rate": 4.059869837927477e-07,
      "loss": 0.7564,
      "step": 8868
    },
    {
      "epoch": 0.7998376696577535,
      "grad_norm": 2.5705711907065476,
      "learning_rate": 4.056342056911728e-07,
      "loss": 0.7299,
      "step": 8869
    },
    {
      "epoch": 0.7999278531812238,
      "grad_norm": 1.767597413352978,
      "learning_rate": 4.052815636296798e-07,
      "loss": 0.7624,
      "step": 8870
    },
    {
      "epoch": 0.8000180367046941,
      "grad_norm": 1.4076123314407574,
      "learning_rate": 4.0492905763835593e-07,
      "loss": 0.7776,
      "step": 8871
    },
    {
      "epoch": 0.8001082202281643,
      "grad_norm": 1.5564545217426604,
      "learning_rate": 4.0457668774728115e-07,
      "loss": 0.772,
      "step": 8872
    },
    {
      "epoch": 0.8001984037516345,
      "grad_norm": 1.4495496346435635,
      "learning_rate": 4.0422445398651985e-07,
      "loss": 0.7606,
      "step": 8873
    },
    {
      "epoch": 0.8002885872751049,
      "grad_norm": 1.640557278643229,
      "learning_rate": 4.0387235638612706e-07,
      "loss": 0.6695,
      "step": 8874
    },
    {
      "epoch": 0.8003787707985751,
      "grad_norm": 2.000593931701115,
      "learning_rate": 4.0352039497614586e-07,
      "loss": 0.7503,
      "step": 8875
    },
    {
      "epoch": 0.8004689543220453,
      "grad_norm": 2.563964611279389,
      "learning_rate": 4.031685697866074e-07,
      "loss": 0.7337,
      "step": 8876
    },
    {
      "epoch": 0.8005591378455156,
      "grad_norm": 1.6230445246673717,
      "learning_rate": 4.0281688084753165e-07,
      "loss": 0.7837,
      "step": 8877
    },
    {
      "epoch": 0.8006493213689859,
      "grad_norm": 1.682972521244237,
      "learning_rate": 4.0246532818892675e-07,
      "loss": 0.799,
      "step": 8878
    },
    {
      "epoch": 0.8007395048924562,
      "grad_norm": 2.2613311066429103,
      "learning_rate": 4.0211391184078814e-07,
      "loss": 0.7845,
      "step": 8879
    },
    {
      "epoch": 0.8008296884159264,
      "grad_norm": 1.7441134221090406,
      "learning_rate": 4.0176263183310135e-07,
      "loss": 0.7819,
      "step": 8880
    },
    {
      "epoch": 0.8009198719393966,
      "grad_norm": 1.798081299930645,
      "learning_rate": 4.0141148819583925e-07,
      "loss": 0.8055,
      "step": 8881
    },
    {
      "epoch": 0.801010055462867,
      "grad_norm": 0.5984480356429442,
      "learning_rate": 4.010604809589637e-07,
      "loss": 0.5702,
      "step": 8882
    },
    {
      "epoch": 0.8011002389863372,
      "grad_norm": 1.6631776131468716,
      "learning_rate": 4.0070961015242475e-07,
      "loss": 0.7486,
      "step": 8883
    },
    {
      "epoch": 0.8011904225098074,
      "grad_norm": 1.5527723775567672,
      "learning_rate": 4.0035887580615933e-07,
      "loss": 0.8174,
      "step": 8884
    },
    {
      "epoch": 0.8012806060332777,
      "grad_norm": 2.021771662103939,
      "learning_rate": 4.0000827795009594e-07,
      "loss": 0.7979,
      "step": 8885
    },
    {
      "epoch": 0.801370789556748,
      "grad_norm": 3.473187743837478,
      "learning_rate": 3.996578166141475e-07,
      "loss": 0.7111,
      "step": 8886
    },
    {
      "epoch": 0.8014609730802182,
      "grad_norm": 10.55436040793593,
      "learning_rate": 3.9930749182821955e-07,
      "loss": 0.751,
      "step": 8887
    },
    {
      "epoch": 0.8015511566036885,
      "grad_norm": 0.6039270276520986,
      "learning_rate": 3.9895730362220116e-07,
      "loss": 0.5948,
      "step": 8888
    },
    {
      "epoch": 0.8016413401271588,
      "grad_norm": 2.2497100643272137,
      "learning_rate": 3.986072520259749e-07,
      "loss": 0.7849,
      "step": 8889
    },
    {
      "epoch": 0.801731523650629,
      "grad_norm": 1.6158192324141953,
      "learning_rate": 3.9825733706940736e-07,
      "loss": 0.7385,
      "step": 8890
    },
    {
      "epoch": 0.8018217071740993,
      "grad_norm": 1.4623749540945081,
      "learning_rate": 3.979075587823557e-07,
      "loss": 0.7658,
      "step": 8891
    },
    {
      "epoch": 0.8019118906975695,
      "grad_norm": 2.4073903738000686,
      "learning_rate": 3.9755791719466504e-07,
      "loss": 0.6279,
      "step": 8892
    },
    {
      "epoch": 0.8020020742210399,
      "grad_norm": 1.913463327475568,
      "learning_rate": 3.9720841233616875e-07,
      "loss": 0.7107,
      "step": 8893
    },
    {
      "epoch": 0.8020922577445101,
      "grad_norm": 1.622035770863784,
      "learning_rate": 3.968590442366888e-07,
      "loss": 0.7181,
      "step": 8894
    },
    {
      "epoch": 0.8021824412679803,
      "grad_norm": 1.5350992517348845,
      "learning_rate": 3.9650981292603423e-07,
      "loss": 0.7245,
      "step": 8895
    },
    {
      "epoch": 0.8022726247914506,
      "grad_norm": 1.919203289056355,
      "learning_rate": 3.961607184340041e-07,
      "loss": 0.7283,
      "step": 8896
    },
    {
      "epoch": 0.8023628083149209,
      "grad_norm": 1.3457471067995241,
      "learning_rate": 3.9581176079038505e-07,
      "loss": 0.6432,
      "step": 8897
    },
    {
      "epoch": 0.8024529918383911,
      "grad_norm": 1.4481085871925845,
      "learning_rate": 3.954629400249516e-07,
      "loss": 0.6941,
      "step": 8898
    },
    {
      "epoch": 0.8025431753618614,
      "grad_norm": 1.9065897279279196,
      "learning_rate": 3.9511425616746787e-07,
      "loss": 0.8131,
      "step": 8899
    },
    {
      "epoch": 0.8026333588853316,
      "grad_norm": 2.4165453935207863,
      "learning_rate": 3.947657092476853e-07,
      "loss": 0.8099,
      "step": 8900
    },
    {
      "epoch": 0.802723542408802,
      "grad_norm": 1.8668096950208144,
      "learning_rate": 3.944172992953425e-07,
      "loss": 0.759,
      "step": 8901
    },
    {
      "epoch": 0.8028137259322722,
      "grad_norm": 1.601405140659299,
      "learning_rate": 3.9406902634017e-07,
      "loss": 0.7018,
      "step": 8902
    },
    {
      "epoch": 0.8029039094557424,
      "grad_norm": 1.482193076026525,
      "learning_rate": 3.9372089041188227e-07,
      "loss": 0.7816,
      "step": 8903
    },
    {
      "epoch": 0.8029940929792126,
      "grad_norm": 4.95356608932825,
      "learning_rate": 3.9337289154018593e-07,
      "loss": 0.6648,
      "step": 8904
    },
    {
      "epoch": 0.803084276502683,
      "grad_norm": 1.5079854757282394,
      "learning_rate": 3.930250297547728e-07,
      "loss": 0.747,
      "step": 8905
    },
    {
      "epoch": 0.8031744600261532,
      "grad_norm": 2.5062563812859553,
      "learning_rate": 3.9267730508532513e-07,
      "loss": 0.6297,
      "step": 8906
    },
    {
      "epoch": 0.8032646435496235,
      "grad_norm": 1.6771029437638356,
      "learning_rate": 3.923297175615121e-07,
      "loss": 0.7357,
      "step": 8907
    },
    {
      "epoch": 0.8033548270730937,
      "grad_norm": 1.597604242425022,
      "learning_rate": 3.9198226721299243e-07,
      "loss": 0.7897,
      "step": 8908
    },
    {
      "epoch": 0.803445010596564,
      "grad_norm": 2.0812452024112424,
      "learning_rate": 3.916349540694128e-07,
      "loss": 0.7263,
      "step": 8909
    },
    {
      "epoch": 0.8035351941200343,
      "grad_norm": 2.3494834453927442,
      "learning_rate": 3.912877781604063e-07,
      "loss": 0.7558,
      "step": 8910
    },
    {
      "epoch": 0.8036253776435045,
      "grad_norm": 1.7072436726472848,
      "learning_rate": 3.909407395155977e-07,
      "loss": 0.7769,
      "step": 8911
    },
    {
      "epoch": 0.8037155611669748,
      "grad_norm": 8.03460029752059,
      "learning_rate": 3.9059383816459725e-07,
      "loss": 0.694,
      "step": 8912
    },
    {
      "epoch": 0.8038057446904451,
      "grad_norm": 5.920106810044392,
      "learning_rate": 3.902470741370045e-07,
      "loss": 0.811,
      "step": 8913
    },
    {
      "epoch": 0.8038959282139153,
      "grad_norm": 1.4600352392777585,
      "learning_rate": 3.8990044746240746e-07,
      "loss": 0.7336,
      "step": 8914
    },
    {
      "epoch": 0.8039861117373855,
      "grad_norm": 2.1813821528497113,
      "learning_rate": 3.8955395817038237e-07,
      "loss": 0.8449,
      "step": 8915
    },
    {
      "epoch": 0.8040762952608559,
      "grad_norm": 1.6497728624928927,
      "learning_rate": 3.892076062904934e-07,
      "loss": 0.7161,
      "step": 8916
    },
    {
      "epoch": 0.8041664787843261,
      "grad_norm": 2.0288073598258882,
      "learning_rate": 3.8886139185229384e-07,
      "loss": 0.7055,
      "step": 8917
    },
    {
      "epoch": 0.8042566623077964,
      "grad_norm": 1.9452950963233793,
      "learning_rate": 3.8851531488532284e-07,
      "loss": 0.6608,
      "step": 8918
    },
    {
      "epoch": 0.8043468458312666,
      "grad_norm": 1.8943814247900874,
      "learning_rate": 3.88169375419112e-07,
      "loss": 0.77,
      "step": 8919
    },
    {
      "epoch": 0.8044370293547369,
      "grad_norm": 2.0823857504838914,
      "learning_rate": 3.8782357348317717e-07,
      "loss": 0.7197,
      "step": 8920
    },
    {
      "epoch": 0.8045272128782072,
      "grad_norm": 2.3950922497083567,
      "learning_rate": 3.8747790910702437e-07,
      "loss": 0.7955,
      "step": 8921
    },
    {
      "epoch": 0.8046173964016774,
      "grad_norm": 1.6103501559888111,
      "learning_rate": 3.8713238232014776e-07,
      "loss": 0.7487,
      "step": 8922
    },
    {
      "epoch": 0.8047075799251476,
      "grad_norm": 1.8007396767672685,
      "learning_rate": 3.867869931520296e-07,
      "loss": 0.8069,
      "step": 8923
    },
    {
      "epoch": 0.804797763448618,
      "grad_norm": 2.0385196591688466,
      "learning_rate": 3.864417416321406e-07,
      "loss": 0.7613,
      "step": 8924
    },
    {
      "epoch": 0.8048879469720882,
      "grad_norm": 9.820804653414472,
      "learning_rate": 3.8609662778993847e-07,
      "loss": 0.7273,
      "step": 8925
    },
    {
      "epoch": 0.8049781304955584,
      "grad_norm": 0.576914370299934,
      "learning_rate": 3.85751651654872e-07,
      "loss": 0.5579,
      "step": 8926
    },
    {
      "epoch": 0.8050683140190287,
      "grad_norm": 1.724509697492834,
      "learning_rate": 3.8540681325637505e-07,
      "loss": 0.8591,
      "step": 8927
    },
    {
      "epoch": 0.805158497542499,
      "grad_norm": 1.560952984768391,
      "learning_rate": 3.8506211262387155e-07,
      "loss": 0.7611,
      "step": 8928
    },
    {
      "epoch": 0.8052486810659693,
      "grad_norm": 14.166656374460109,
      "learning_rate": 3.847175497867732e-07,
      "loss": 0.7627,
      "step": 8929
    },
    {
      "epoch": 0.8053388645894395,
      "grad_norm": 1.7804042079688118,
      "learning_rate": 3.843731247744801e-07,
      "loss": 0.6948,
      "step": 8930
    },
    {
      "epoch": 0.8054290481129097,
      "grad_norm": 1.6358350311804246,
      "learning_rate": 3.8402883761638047e-07,
      "loss": 0.7911,
      "step": 8931
    },
    {
      "epoch": 0.8055192316363801,
      "grad_norm": 0.6699928045954001,
      "learning_rate": 3.8368468834185076e-07,
      "loss": 0.6138,
      "step": 8932
    },
    {
      "epoch": 0.8056094151598503,
      "grad_norm": 3.5444603317393644,
      "learning_rate": 3.8334067698025583e-07,
      "loss": 0.7943,
      "step": 8933
    },
    {
      "epoch": 0.8056995986833205,
      "grad_norm": 1.553905808116453,
      "learning_rate": 3.8299680356094897e-07,
      "loss": 0.6819,
      "step": 8934
    },
    {
      "epoch": 0.8057897822067909,
      "grad_norm": 2.0362501371323907,
      "learning_rate": 3.8265306811327024e-07,
      "loss": 0.6447,
      "step": 8935
    },
    {
      "epoch": 0.8058799657302611,
      "grad_norm": 3.907218873984875,
      "learning_rate": 3.8230947066654994e-07,
      "loss": 0.7184,
      "step": 8936
    },
    {
      "epoch": 0.8059701492537313,
      "grad_norm": 3.1405408857478214,
      "learning_rate": 3.819660112501053e-07,
      "loss": 0.8117,
      "step": 8937
    },
    {
      "epoch": 0.8060603327772016,
      "grad_norm": 0.6187502716526244,
      "learning_rate": 3.816226898932422e-07,
      "loss": 0.5976,
      "step": 8938
    },
    {
      "epoch": 0.8061505163006719,
      "grad_norm": 2.1046471629422725,
      "learning_rate": 3.812795066252557e-07,
      "loss": 0.7398,
      "step": 8939
    },
    {
      "epoch": 0.8062406998241421,
      "grad_norm": 2.0255973217517087,
      "learning_rate": 3.8093646147542577e-07,
      "loss": 0.7624,
      "step": 8940
    },
    {
      "epoch": 0.8063308833476124,
      "grad_norm": 1.7914161359112444,
      "learning_rate": 3.805935544730259e-07,
      "loss": 0.7204,
      "step": 8941
    },
    {
      "epoch": 0.8064210668710826,
      "grad_norm": 1.650505034865749,
      "learning_rate": 3.802507856473118e-07,
      "loss": 0.684,
      "step": 8942
    },
    {
      "epoch": 0.806511250394553,
      "grad_norm": 2.368922236881593,
      "learning_rate": 3.7990815502753317e-07,
      "loss": 0.7537,
      "step": 8943
    },
    {
      "epoch": 0.8066014339180232,
      "grad_norm": 1.7024215505397562,
      "learning_rate": 3.795656626429231e-07,
      "loss": 0.7539,
      "step": 8944
    },
    {
      "epoch": 0.8066916174414934,
      "grad_norm": 2.3999497805745404,
      "learning_rate": 3.792233085227059e-07,
      "loss": 0.6745,
      "step": 8945
    },
    {
      "epoch": 0.8067818009649637,
      "grad_norm": 2.0059669413785373,
      "learning_rate": 3.788810926960928e-07,
      "loss": 0.7612,
      "step": 8946
    },
    {
      "epoch": 0.806871984488434,
      "grad_norm": 1.646620847718365,
      "learning_rate": 3.785390151922836e-07,
      "loss": 0.7357,
      "step": 8947
    },
    {
      "epoch": 0.8069621680119042,
      "grad_norm": 2.2159681476824002,
      "learning_rate": 3.781970760404665e-07,
      "loss": 0.735,
      "step": 8948
    },
    {
      "epoch": 0.8070523515353745,
      "grad_norm": 1.502019317716066,
      "learning_rate": 3.778552752698176e-07,
      "loss": 0.825,
      "step": 8949
    },
    {
      "epoch": 0.8071425350588447,
      "grad_norm": 1.4848296422889304,
      "learning_rate": 3.775136129095007e-07,
      "loss": 0.6584,
      "step": 8950
    },
    {
      "epoch": 0.807232718582315,
      "grad_norm": 3.3066251949304046,
      "learning_rate": 3.771720889886685e-07,
      "loss": 0.7198,
      "step": 8951
    },
    {
      "epoch": 0.8073229021057853,
      "grad_norm": 0.695908794384892,
      "learning_rate": 3.7683070353646194e-07,
      "loss": 0.5963,
      "step": 8952
    },
    {
      "epoch": 0.8074130856292555,
      "grad_norm": 1.7116088943364458,
      "learning_rate": 3.7648945658200983e-07,
      "loss": 0.712,
      "step": 8953
    },
    {
      "epoch": 0.8075032691527257,
      "grad_norm": 2.780903569623316,
      "learning_rate": 3.761483481544292e-07,
      "loss": 0.6376,
      "step": 8954
    },
    {
      "epoch": 0.8075934526761961,
      "grad_norm": 2.0484588552099448,
      "learning_rate": 3.7580737828282525e-07,
      "loss": 0.72,
      "step": 8955
    },
    {
      "epoch": 0.8076836361996663,
      "grad_norm": 1.7649055973342684,
      "learning_rate": 3.754665469962921e-07,
      "loss": 0.728,
      "step": 8956
    },
    {
      "epoch": 0.8077738197231366,
      "grad_norm": 1.7440271629003037,
      "learning_rate": 3.7512585432390973e-07,
      "loss": 0.7079,
      "step": 8957
    },
    {
      "epoch": 0.8078640032466069,
      "grad_norm": 1.8907388700132173,
      "learning_rate": 3.7478530029474987e-07,
      "loss": 0.6881,
      "step": 8958
    },
    {
      "epoch": 0.8079541867700771,
      "grad_norm": 0.6228735512278912,
      "learning_rate": 3.7444488493786854e-07,
      "loss": 0.628,
      "step": 8959
    },
    {
      "epoch": 0.8080443702935474,
      "grad_norm": 2.072602240199876,
      "learning_rate": 3.7410460828231405e-07,
      "loss": 0.7982,
      "step": 8960
    },
    {
      "epoch": 0.8081345538170176,
      "grad_norm": 1.4529514003956296,
      "learning_rate": 3.737644703571188e-07,
      "loss": 0.6839,
      "step": 8961
    },
    {
      "epoch": 0.8082247373404879,
      "grad_norm": 1.7459924314319566,
      "learning_rate": 3.734244711913059e-07,
      "loss": 0.8246,
      "step": 8962
    },
    {
      "epoch": 0.8083149208639582,
      "grad_norm": 1.5793995007032435,
      "learning_rate": 3.7308461081388584e-07,
      "loss": 0.7248,
      "step": 8963
    },
    {
      "epoch": 0.8084051043874284,
      "grad_norm": 1.6671575141362371,
      "learning_rate": 3.727448892538576e-07,
      "loss": 0.6822,
      "step": 8964
    },
    {
      "epoch": 0.8084952879108986,
      "grad_norm": 1.5156827306075835,
      "learning_rate": 3.724053065402086e-07,
      "loss": 0.8343,
      "step": 8965
    },
    {
      "epoch": 0.808585471434369,
      "grad_norm": 1.5648432136949906,
      "learning_rate": 3.7206586270191285e-07,
      "loss": 0.7129,
      "step": 8966
    },
    {
      "epoch": 0.8086756549578392,
      "grad_norm": 1.5623588498256928,
      "learning_rate": 3.7172655776793385e-07,
      "loss": 0.7967,
      "step": 8967
    },
    {
      "epoch": 0.8087658384813095,
      "grad_norm": 1.6749547952274109,
      "learning_rate": 3.7138739176722323e-07,
      "loss": 0.643,
      "step": 8968
    },
    {
      "epoch": 0.8088560220047797,
      "grad_norm": 1.8922417403746576,
      "learning_rate": 3.710483647287206e-07,
      "loss": 0.7133,
      "step": 8969
    },
    {
      "epoch": 0.80894620552825,
      "grad_norm": 1.6267721343374693,
      "learning_rate": 3.707094766813532e-07,
      "loss": 0.6463,
      "step": 8970
    },
    {
      "epoch": 0.8090363890517203,
      "grad_norm": 1.6754463056693736,
      "learning_rate": 3.7037072765403754e-07,
      "loss": 0.7136,
      "step": 8971
    },
    {
      "epoch": 0.8091265725751905,
      "grad_norm": 2.022931363346707,
      "learning_rate": 3.700321176756762e-07,
      "loss": 0.7081,
      "step": 8972
    },
    {
      "epoch": 0.8092167560986607,
      "grad_norm": 1.710414789208866,
      "learning_rate": 3.69693646775163e-07,
      "loss": 0.7943,
      "step": 8973
    },
    {
      "epoch": 0.8093069396221311,
      "grad_norm": 2.0044577510774557,
      "learning_rate": 3.693553149813764e-07,
      "loss": 0.7358,
      "step": 8974
    },
    {
      "epoch": 0.8093971231456013,
      "grad_norm": 0.5813860805737492,
      "learning_rate": 3.690171223231866e-07,
      "loss": 0.5515,
      "step": 8975
    },
    {
      "epoch": 0.8094873066690715,
      "grad_norm": 1.3744143192344687,
      "learning_rate": 3.6867906882944854e-07,
      "loss": 0.8379,
      "step": 8976
    },
    {
      "epoch": 0.8095774901925418,
      "grad_norm": 2.223372895300106,
      "learning_rate": 3.6834115452900737e-07,
      "loss": 0.7447,
      "step": 8977
    },
    {
      "epoch": 0.8096676737160121,
      "grad_norm": 1.5264606703468795,
      "learning_rate": 3.680033794506958e-07,
      "loss": 0.6422,
      "step": 8978
    },
    {
      "epoch": 0.8097578572394823,
      "grad_norm": 2.026351891885043,
      "learning_rate": 3.676657436233346e-07,
      "loss": 0.5768,
      "step": 8979
    },
    {
      "epoch": 0.8098480407629526,
      "grad_norm": 1.5840443553927663,
      "learning_rate": 3.6732824707573305e-07,
      "loss": 0.7421,
      "step": 8980
    },
    {
      "epoch": 0.8099382242864229,
      "grad_norm": 1.768808583613755,
      "learning_rate": 3.6699088983668716e-07,
      "loss": 0.7844,
      "step": 8981
    },
    {
      "epoch": 0.8100284078098932,
      "grad_norm": 1.6802753218343496,
      "learning_rate": 3.6665367193498376e-07,
      "loss": 0.7008,
      "step": 8982
    },
    {
      "epoch": 0.8101185913333634,
      "grad_norm": 2.44653562723239,
      "learning_rate": 3.663165933993948e-07,
      "loss": 0.7472,
      "step": 8983
    },
    {
      "epoch": 0.8102087748568336,
      "grad_norm": 1.851577629961093,
      "learning_rate": 3.659796542586822e-07,
      "loss": 0.7677,
      "step": 8984
    },
    {
      "epoch": 0.810298958380304,
      "grad_norm": 1.5317361021163174,
      "learning_rate": 3.6564285454159526e-07,
      "loss": 0.6995,
      "step": 8985
    },
    {
      "epoch": 0.8103891419037742,
      "grad_norm": 1.5105303842114453,
      "learning_rate": 3.653061942768718e-07,
      "loss": 0.7912,
      "step": 8986
    },
    {
      "epoch": 0.8104793254272444,
      "grad_norm": 1.7815911485007607,
      "learning_rate": 3.649696734932375e-07,
      "loss": 0.7026,
      "step": 8987
    },
    {
      "epoch": 0.8105695089507147,
      "grad_norm": 3.68166134775335,
      "learning_rate": 3.646332922194064e-07,
      "loss": 0.6778,
      "step": 8988
    },
    {
      "epoch": 0.810659692474185,
      "grad_norm": 1.7539423405729635,
      "learning_rate": 3.6429705048407943e-07,
      "loss": 0.7311,
      "step": 8989
    },
    {
      "epoch": 0.8107498759976552,
      "grad_norm": 1.9367382644094893,
      "learning_rate": 3.6396094831594804e-07,
      "loss": 0.7433,
      "step": 8990
    },
    {
      "epoch": 0.8108400595211255,
      "grad_norm": 1.6923705561172608,
      "learning_rate": 3.6362498574368926e-07,
      "loss": 0.6963,
      "step": 8991
    },
    {
      "epoch": 0.8109302430445957,
      "grad_norm": 0.7048758853393342,
      "learning_rate": 3.6328916279596935e-07,
      "loss": 0.6246,
      "step": 8992
    },
    {
      "epoch": 0.811020426568066,
      "grad_norm": 2.060904492626544,
      "learning_rate": 3.6295347950144305e-07,
      "loss": 0.7952,
      "step": 8993
    },
    {
      "epoch": 0.8111106100915363,
      "grad_norm": 2.1802734899892497,
      "learning_rate": 3.626179358887522e-07,
      "loss": 0.5975,
      "step": 8994
    },
    {
      "epoch": 0.8112007936150065,
      "grad_norm": 1.7705184506971492,
      "learning_rate": 3.6228253198652816e-07,
      "loss": 0.8036,
      "step": 8995
    },
    {
      "epoch": 0.8112909771384768,
      "grad_norm": 2.3810303073573933,
      "learning_rate": 3.6194726782338767e-07,
      "loss": 0.6944,
      "step": 8996
    },
    {
      "epoch": 0.8113811606619471,
      "grad_norm": 6.675191515057476,
      "learning_rate": 3.6161214342793953e-07,
      "loss": 0.7233,
      "step": 8997
    },
    {
      "epoch": 0.8114713441854173,
      "grad_norm": 1.884904261416752,
      "learning_rate": 3.612771588287764e-07,
      "loss": 0.7741,
      "step": 8998
    },
    {
      "epoch": 0.8115615277088876,
      "grad_norm": 2.096435305207895,
      "learning_rate": 3.609423140544827e-07,
      "loss": 0.8106,
      "step": 8999
    },
    {
      "epoch": 0.8116517112323578,
      "grad_norm": 2.196565101004393,
      "learning_rate": 3.6060760913362787e-07,
      "loss": 0.7617,
      "step": 9000
    },
    {
      "epoch": 0.8117418947558281,
      "grad_norm": 1.5441572416662612,
      "learning_rate": 3.6027304409477146e-07,
      "loss": 0.7501,
      "step": 9001
    },
    {
      "epoch": 0.8118320782792984,
      "grad_norm": 1.8787373488679822,
      "learning_rate": 3.599386189664604e-07,
      "loss": 0.7159,
      "step": 9002
    },
    {
      "epoch": 0.8119222618027686,
      "grad_norm": 1.9866242164625152,
      "learning_rate": 3.5960433377722945e-07,
      "loss": 0.7391,
      "step": 9003
    },
    {
      "epoch": 0.8120124453262388,
      "grad_norm": 1.7391086623653393,
      "learning_rate": 3.5927018855560174e-07,
      "loss": 0.8046,
      "step": 9004
    },
    {
      "epoch": 0.8121026288497092,
      "grad_norm": 1.571283448317736,
      "learning_rate": 3.5893618333008904e-07,
      "loss": 0.7339,
      "step": 9005
    },
    {
      "epoch": 0.8121928123731794,
      "grad_norm": 1.687092201973984,
      "learning_rate": 3.586023181291893e-07,
      "loss": 0.8372,
      "step": 9006
    },
    {
      "epoch": 0.8122829958966497,
      "grad_norm": 0.7575160754421141,
      "learning_rate": 3.5826859298139044e-07,
      "loss": 0.6114,
      "step": 9007
    },
    {
      "epoch": 0.81237317942012,
      "grad_norm": 1.526948297141175,
      "learning_rate": 3.5793500791516773e-07,
      "loss": 0.7273,
      "step": 9008
    },
    {
      "epoch": 0.8124633629435902,
      "grad_norm": 1.5398202957232827,
      "learning_rate": 3.5760156295898415e-07,
      "loss": 0.716,
      "step": 9009
    },
    {
      "epoch": 0.8125535464670605,
      "grad_norm": 2.341379416372138,
      "learning_rate": 3.5726825814129203e-07,
      "loss": 0.7428,
      "step": 9010
    },
    {
      "epoch": 0.8126437299905307,
      "grad_norm": 0.699564097274256,
      "learning_rate": 3.5693509349052886e-07,
      "loss": 0.6132,
      "step": 9011
    },
    {
      "epoch": 0.812733913514001,
      "grad_norm": 1.6238879286614254,
      "learning_rate": 3.5660206903512433e-07,
      "loss": 0.8151,
      "step": 9012
    },
    {
      "epoch": 0.8128240970374713,
      "grad_norm": 2.196043279794585,
      "learning_rate": 3.56269184803492e-07,
      "loss": 0.7969,
      "step": 9013
    },
    {
      "epoch": 0.8129142805609415,
      "grad_norm": 1.5189801077127223,
      "learning_rate": 3.5593644082403727e-07,
      "loss": 0.7661,
      "step": 9014
    },
    {
      "epoch": 0.8130044640844117,
      "grad_norm": 1.9531112670415636,
      "learning_rate": 3.5560383712514994e-07,
      "loss": 0.8424,
      "step": 9015
    },
    {
      "epoch": 0.8130946476078821,
      "grad_norm": 1.6173505470531806,
      "learning_rate": 3.5527137373521066e-07,
      "loss": 0.7474,
      "step": 9016
    },
    {
      "epoch": 0.8131848311313523,
      "grad_norm": 1.938098814934103,
      "learning_rate": 3.5493905068258645e-07,
      "loss": 0.7036,
      "step": 9017
    },
    {
      "epoch": 0.8132750146548225,
      "grad_norm": 1.464149982195943,
      "learning_rate": 3.546068679956333e-07,
      "loss": 0.7603,
      "step": 9018
    },
    {
      "epoch": 0.8133651981782928,
      "grad_norm": 1.6774919221459,
      "learning_rate": 3.5427482570269487e-07,
      "loss": 0.8079,
      "step": 9019
    },
    {
      "epoch": 0.8134553817017631,
      "grad_norm": 2.956021619044059,
      "learning_rate": 3.539429238321026e-07,
      "loss": 0.5955,
      "step": 9020
    },
    {
      "epoch": 0.8135455652252334,
      "grad_norm": 2.2517352300671774,
      "learning_rate": 3.536111624121769e-07,
      "loss": 0.5979,
      "step": 9021
    },
    {
      "epoch": 0.8136357487487036,
      "grad_norm": 2.1054239073668546,
      "learning_rate": 3.532795414712244e-07,
      "loss": 0.7076,
      "step": 9022
    },
    {
      "epoch": 0.8137259322721738,
      "grad_norm": 1.4273415584787499,
      "learning_rate": 3.5294806103754124e-07,
      "loss": 0.7642,
      "step": 9023
    },
    {
      "epoch": 0.8138161157956442,
      "grad_norm": 1.6657549351151488,
      "learning_rate": 3.526167211394115e-07,
      "loss": 0.7597,
      "step": 9024
    },
    {
      "epoch": 0.8139062993191144,
      "grad_norm": 4.095787414934989,
      "learning_rate": 3.522855218051066e-07,
      "loss": 0.6854,
      "step": 9025
    },
    {
      "epoch": 0.8139964828425846,
      "grad_norm": 2.00451139418384,
      "learning_rate": 3.5195446306288633e-07,
      "loss": 0.7551,
      "step": 9026
    },
    {
      "epoch": 0.8140866663660549,
      "grad_norm": 1.415816341715057,
      "learning_rate": 3.51623544940999e-07,
      "loss": 0.6894,
      "step": 9027
    },
    {
      "epoch": 0.8141768498895252,
      "grad_norm": 1.460000293418123,
      "learning_rate": 3.5129276746767886e-07,
      "loss": 0.7317,
      "step": 9028
    },
    {
      "epoch": 0.8142670334129954,
      "grad_norm": 1.9172547792600703,
      "learning_rate": 3.5096213067115165e-07,
      "loss": 0.7216,
      "step": 9029
    },
    {
      "epoch": 0.8143572169364657,
      "grad_norm": 1.7061774521380255,
      "learning_rate": 3.506316345796272e-07,
      "loss": 0.7661,
      "step": 9030
    },
    {
      "epoch": 0.814447400459936,
      "grad_norm": 0.596168936291154,
      "learning_rate": 3.5030127922130714e-07,
      "loss": 0.5357,
      "step": 9031
    },
    {
      "epoch": 0.8145375839834063,
      "grad_norm": 1.5019224086840017,
      "learning_rate": 3.4997106462437784e-07,
      "loss": 0.6373,
      "step": 9032
    },
    {
      "epoch": 0.8146277675068765,
      "grad_norm": 0.6558043919793448,
      "learning_rate": 3.496409908170157e-07,
      "loss": 0.562,
      "step": 9033
    },
    {
      "epoch": 0.8147179510303467,
      "grad_norm": 2.925791680555506,
      "learning_rate": 3.493110578273839e-07,
      "loss": 0.7112,
      "step": 9034
    },
    {
      "epoch": 0.8148081345538171,
      "grad_norm": 1.8813791477825998,
      "learning_rate": 3.489812656836346e-07,
      "loss": 0.8039,
      "step": 9035
    },
    {
      "epoch": 0.8148983180772873,
      "grad_norm": 1.516059970379144,
      "learning_rate": 3.486516144139078e-07,
      "loss": 0.6593,
      "step": 9036
    },
    {
      "epoch": 0.8149885016007575,
      "grad_norm": 1.5736450967525832,
      "learning_rate": 3.4832210404632957e-07,
      "loss": 0.7143,
      "step": 9037
    },
    {
      "epoch": 0.8150786851242278,
      "grad_norm": 1.4341287472174808,
      "learning_rate": 3.479927346090179e-07,
      "loss": 0.7554,
      "step": 9038
    },
    {
      "epoch": 0.8151688686476981,
      "grad_norm": 1.8439580509023379,
      "learning_rate": 3.4766350613007455e-07,
      "loss": 0.7337,
      "step": 9039
    },
    {
      "epoch": 0.8152590521711683,
      "grad_norm": 1.7027525888349662,
      "learning_rate": 3.4733441863759173e-07,
      "loss": 0.7369,
      "step": 9040
    },
    {
      "epoch": 0.8153492356946386,
      "grad_norm": 2.4131584573405815,
      "learning_rate": 3.4700547215964916e-07,
      "loss": 0.6397,
      "step": 9041
    },
    {
      "epoch": 0.8154394192181088,
      "grad_norm": 1.4857212587208652,
      "learning_rate": 3.46676666724314e-07,
      "loss": 0.7237,
      "step": 9042
    },
    {
      "epoch": 0.8155296027415792,
      "grad_norm": 2.862233218625403,
      "learning_rate": 3.463480023596421e-07,
      "loss": 0.74,
      "step": 9043
    },
    {
      "epoch": 0.8156197862650494,
      "grad_norm": 2.207185506915646,
      "learning_rate": 3.460194790936772e-07,
      "loss": 0.8182,
      "step": 9044
    },
    {
      "epoch": 0.8157099697885196,
      "grad_norm": 1.7461629081531003,
      "learning_rate": 3.456910969544495e-07,
      "loss": 0.7352,
      "step": 9045
    },
    {
      "epoch": 0.8158001533119899,
      "grad_norm": 1.9660025908646146,
      "learning_rate": 3.4536285596997994e-07,
      "loss": 0.7755,
      "step": 9046
    },
    {
      "epoch": 0.8158903368354602,
      "grad_norm": 1.9467227221065015,
      "learning_rate": 3.450347561682747e-07,
      "loss": 0.7038,
      "step": 9047
    },
    {
      "epoch": 0.8159805203589304,
      "grad_norm": 5.353457372961534,
      "learning_rate": 3.4470679757732945e-07,
      "loss": 0.6873,
      "step": 9048
    },
    {
      "epoch": 0.8160707038824007,
      "grad_norm": 2.1674721638829126,
      "learning_rate": 3.4437898022512735e-07,
      "loss": 0.8272,
      "step": 9049
    },
    {
      "epoch": 0.8161608874058709,
      "grad_norm": 2.0505327927991015,
      "learning_rate": 3.4405130413963977e-07,
      "loss": 0.7476,
      "step": 9050
    },
    {
      "epoch": 0.8162510709293412,
      "grad_norm": 2.3642690060337563,
      "learning_rate": 3.437237693488262e-07,
      "loss": 0.8141,
      "step": 9051
    },
    {
      "epoch": 0.8163412544528115,
      "grad_norm": 1.3951212171987424,
      "learning_rate": 3.433963758806322e-07,
      "loss": 0.7596,
      "step": 9052
    },
    {
      "epoch": 0.8164314379762817,
      "grad_norm": 1.654655480727923,
      "learning_rate": 3.430691237629948e-07,
      "loss": 0.6081,
      "step": 9053
    },
    {
      "epoch": 0.816521621499752,
      "grad_norm": 2.919222520240734,
      "learning_rate": 3.427420130238354e-07,
      "loss": 0.6971,
      "step": 9054
    },
    {
      "epoch": 0.8166118050232223,
      "grad_norm": 1.5404000448840582,
      "learning_rate": 3.424150436910658e-07,
      "loss": 0.8312,
      "step": 9055
    },
    {
      "epoch": 0.8167019885466925,
      "grad_norm": 23.757982057106222,
      "learning_rate": 3.420882157925842e-07,
      "loss": 0.8024,
      "step": 9056
    },
    {
      "epoch": 0.8167921720701627,
      "grad_norm": 2.557289594868263,
      "learning_rate": 3.417615293562777e-07,
      "loss": 0.7208,
      "step": 9057
    },
    {
      "epoch": 0.8168823555936331,
      "grad_norm": 1.5946941197231952,
      "learning_rate": 3.4143498441002105e-07,
      "loss": 0.6537,
      "step": 9058
    },
    {
      "epoch": 0.8169725391171033,
      "grad_norm": 1.5509834061699692,
      "learning_rate": 3.411085809816767e-07,
      "loss": 0.7521,
      "step": 9059
    },
    {
      "epoch": 0.8170627226405736,
      "grad_norm": 1.8995152934343131,
      "learning_rate": 3.407823190990953e-07,
      "loss": 0.8651,
      "step": 9060
    },
    {
      "epoch": 0.8171529061640438,
      "grad_norm": 1.9305294532139525,
      "learning_rate": 3.4045619879011577e-07,
      "loss": 0.8303,
      "step": 9061
    },
    {
      "epoch": 0.8172430896875141,
      "grad_norm": 1.9987817868871522,
      "learning_rate": 3.4013022008256334e-07,
      "loss": 0.6733,
      "step": 9062
    },
    {
      "epoch": 0.8173332732109844,
      "grad_norm": 1.8595972008529604,
      "learning_rate": 3.398043830042532e-07,
      "loss": 0.683,
      "step": 9063
    },
    {
      "epoch": 0.8174234567344546,
      "grad_norm": 1.7666665440085507,
      "learning_rate": 3.394786875829871e-07,
      "loss": 0.7432,
      "step": 9064
    },
    {
      "epoch": 0.8175136402579248,
      "grad_norm": 1.5500956444302765,
      "learning_rate": 3.3915313384655564e-07,
      "loss": 0.6271,
      "step": 9065
    },
    {
      "epoch": 0.8176038237813952,
      "grad_norm": 1.3626100040790277,
      "learning_rate": 3.388277218227369e-07,
      "loss": 0.7593,
      "step": 9066
    },
    {
      "epoch": 0.8176940073048654,
      "grad_norm": 2.1017732762708237,
      "learning_rate": 3.3850245153929557e-07,
      "loss": 0.7647,
      "step": 9067
    },
    {
      "epoch": 0.8177841908283356,
      "grad_norm": 2.269135484970347,
      "learning_rate": 3.381773230239875e-07,
      "loss": 0.7046,
      "step": 9068
    },
    {
      "epoch": 0.8178743743518059,
      "grad_norm": 1.4935165797495575,
      "learning_rate": 3.3785233630455247e-07,
      "loss": 0.7742,
      "step": 9069
    },
    {
      "epoch": 0.8179645578752762,
      "grad_norm": 1.7686384024309398,
      "learning_rate": 3.375274914087221e-07,
      "loss": 0.7552,
      "step": 9070
    },
    {
      "epoch": 0.8180547413987465,
      "grad_norm": 1.391066791779237,
      "learning_rate": 3.3720278836421234e-07,
      "loss": 0.7167,
      "step": 9071
    },
    {
      "epoch": 0.8181449249222167,
      "grad_norm": 1.7626718450870327,
      "learning_rate": 3.368782271987294e-07,
      "loss": 0.7698,
      "step": 9072
    },
    {
      "epoch": 0.8182351084456869,
      "grad_norm": 1.6233502596782785,
      "learning_rate": 3.3655380793996636e-07,
      "loss": 0.7514,
      "step": 9073
    },
    {
      "epoch": 0.8183252919691573,
      "grad_norm": 1.5089275605173345,
      "learning_rate": 3.362295306156047e-07,
      "loss": 0.6623,
      "step": 9074
    },
    {
      "epoch": 0.8184154754926275,
      "grad_norm": 1.6083046029379195,
      "learning_rate": 3.3590539525331327e-07,
      "loss": 0.7562,
      "step": 9075
    },
    {
      "epoch": 0.8185056590160977,
      "grad_norm": 1.6315800884579617,
      "learning_rate": 3.3558140188074967e-07,
      "loss": 0.6935,
      "step": 9076
    },
    {
      "epoch": 0.8185958425395681,
      "grad_norm": 4.340499869534803,
      "learning_rate": 3.3525755052555817e-07,
      "loss": 0.7744,
      "step": 9077
    },
    {
      "epoch": 0.8186860260630383,
      "grad_norm": 0.7487076512685199,
      "learning_rate": 3.3493384121537147e-07,
      "loss": 0.6493,
      "step": 9078
    },
    {
      "epoch": 0.8187762095865085,
      "grad_norm": 1.8901584144082282,
      "learning_rate": 3.3461027397781075e-07,
      "loss": 0.7436,
      "step": 9079
    },
    {
      "epoch": 0.8188663931099788,
      "grad_norm": 4.076658723396637,
      "learning_rate": 3.3428684884048397e-07,
      "loss": 0.7274,
      "step": 9080
    },
    {
      "epoch": 0.8189565766334491,
      "grad_norm": 1.520209938032089,
      "learning_rate": 3.3396356583098826e-07,
      "loss": 0.7459,
      "step": 9081
    },
    {
      "epoch": 0.8190467601569194,
      "grad_norm": 1.7606503107745286,
      "learning_rate": 3.3364042497690736e-07,
      "loss": 0.7802,
      "step": 9082
    },
    {
      "epoch": 0.8191369436803896,
      "grad_norm": 1.5050599268138067,
      "learning_rate": 3.3331742630581405e-07,
      "loss": 0.8001,
      "step": 9083
    },
    {
      "epoch": 0.8192271272038598,
      "grad_norm": 1.418384753698564,
      "learning_rate": 3.3299456984526717e-07,
      "loss": 0.6907,
      "step": 9084
    },
    {
      "epoch": 0.8193173107273302,
      "grad_norm": 1.4994439842952403,
      "learning_rate": 3.3267185562281605e-07,
      "loss": 0.7832,
      "step": 9085
    },
    {
      "epoch": 0.8194074942508004,
      "grad_norm": 1.7293099340200386,
      "learning_rate": 3.3234928366599514e-07,
      "loss": 0.7381,
      "step": 9086
    },
    {
      "epoch": 0.8194976777742706,
      "grad_norm": 1.7411700643469337,
      "learning_rate": 3.3202685400232946e-07,
      "loss": 0.7817,
      "step": 9087
    },
    {
      "epoch": 0.8195878612977409,
      "grad_norm": 2.3663940106966725,
      "learning_rate": 3.317045666593292e-07,
      "loss": 0.7739,
      "step": 9088
    },
    {
      "epoch": 0.8196780448212112,
      "grad_norm": 2.0965771163633353,
      "learning_rate": 3.3138242166449426e-07,
      "loss": 0.7515,
      "step": 9089
    },
    {
      "epoch": 0.8197682283446814,
      "grad_norm": 1.5306632805769105,
      "learning_rate": 3.310604190453117e-07,
      "loss": 0.808,
      "step": 9090
    },
    {
      "epoch": 0.8198584118681517,
      "grad_norm": 1.4677692953877433,
      "learning_rate": 3.307385588292566e-07,
      "loss": 0.694,
      "step": 9091
    },
    {
      "epoch": 0.8199485953916219,
      "grad_norm": 1.6699635639898587,
      "learning_rate": 3.304168410437924e-07,
      "loss": 0.739,
      "step": 9092
    },
    {
      "epoch": 0.8200387789150922,
      "grad_norm": 0.6317343062627954,
      "learning_rate": 3.300952657163687e-07,
      "loss": 0.6038,
      "step": 9093
    },
    {
      "epoch": 0.8201289624385625,
      "grad_norm": 1.6275945637737548,
      "learning_rate": 3.297738328744248e-07,
      "loss": 0.726,
      "step": 9094
    },
    {
      "epoch": 0.8202191459620327,
      "grad_norm": 1.813732714580507,
      "learning_rate": 3.2945254254538714e-07,
      "loss": 0.7367,
      "step": 9095
    },
    {
      "epoch": 0.820309329485503,
      "grad_norm": 1.417278447539762,
      "learning_rate": 3.2913139475666963e-07,
      "loss": 0.71,
      "step": 9096
    },
    {
      "epoch": 0.8203995130089733,
      "grad_norm": 2.1779722622663074,
      "learning_rate": 3.288103895356749e-07,
      "loss": 0.6935,
      "step": 9097
    },
    {
      "epoch": 0.8204896965324435,
      "grad_norm": 1.9197287632169577,
      "learning_rate": 3.284895269097927e-07,
      "loss": 0.676,
      "step": 9098
    },
    {
      "epoch": 0.8205798800559138,
      "grad_norm": 0.6697927189211097,
      "learning_rate": 3.281688069063999e-07,
      "loss": 0.5815,
      "step": 9099
    },
    {
      "epoch": 0.8206700635793841,
      "grad_norm": 1.7671761240032275,
      "learning_rate": 3.2784822955286396e-07,
      "loss": 0.5691,
      "step": 9100
    },
    {
      "epoch": 0.8207602471028543,
      "grad_norm": 1.4860937507700787,
      "learning_rate": 3.275277948765365e-07,
      "loss": 0.8154,
      "step": 9101
    },
    {
      "epoch": 0.8208504306263246,
      "grad_norm": 2.9520807589724094,
      "learning_rate": 3.2720750290475964e-07,
      "loss": 0.784,
      "step": 9102
    },
    {
      "epoch": 0.8209406141497948,
      "grad_norm": 2.009407330337447,
      "learning_rate": 3.268873536648622e-07,
      "loss": 0.7467,
      "step": 9103
    },
    {
      "epoch": 0.8210307976732651,
      "grad_norm": 1.509424484421106,
      "learning_rate": 3.265673471841612e-07,
      "loss": 0.7174,
      "step": 9104
    },
    {
      "epoch": 0.8211209811967354,
      "grad_norm": 0.6269200396497961,
      "learning_rate": 3.262474834899616e-07,
      "loss": 0.5404,
      "step": 9105
    },
    {
      "epoch": 0.8212111647202056,
      "grad_norm": 0.6751323958643191,
      "learning_rate": 3.2592776260955534e-07,
      "loss": 0.6166,
      "step": 9106
    },
    {
      "epoch": 0.8213013482436758,
      "grad_norm": 1.806311137078808,
      "learning_rate": 3.256081845702239e-07,
      "loss": 0.7864,
      "step": 9107
    },
    {
      "epoch": 0.8213915317671462,
      "grad_norm": 4.445677194233971,
      "learning_rate": 3.2528874939923335e-07,
      "loss": 0.7924,
      "step": 9108
    },
    {
      "epoch": 0.8214817152906164,
      "grad_norm": 1.9646306007607839,
      "learning_rate": 3.2496945712384217e-07,
      "loss": 0.702,
      "step": 9109
    },
    {
      "epoch": 0.8215718988140867,
      "grad_norm": 1.8690520358243154,
      "learning_rate": 3.246503077712923e-07,
      "loss": 0.7256,
      "step": 9110
    },
    {
      "epoch": 0.8216620823375569,
      "grad_norm": 1.640916852877455,
      "learning_rate": 3.2433130136881625e-07,
      "loss": 0.8359,
      "step": 9111
    },
    {
      "epoch": 0.8217522658610272,
      "grad_norm": 2.50192024870452,
      "learning_rate": 3.2401243794363287e-07,
      "loss": 0.7424,
      "step": 9112
    },
    {
      "epoch": 0.8218424493844975,
      "grad_norm": 1.7308810670699084,
      "learning_rate": 3.236937175229495e-07,
      "loss": 0.8035,
      "step": 9113
    },
    {
      "epoch": 0.8219326329079677,
      "grad_norm": 1.8242764514888934,
      "learning_rate": 3.233751401339615e-07,
      "loss": 0.7188,
      "step": 9114
    },
    {
      "epoch": 0.8220228164314379,
      "grad_norm": 2.0331311245206565,
      "learning_rate": 3.2305670580385157e-07,
      "loss": 0.8242,
      "step": 9115
    },
    {
      "epoch": 0.8221129999549083,
      "grad_norm": 1.44970152183506,
      "learning_rate": 3.227384145597898e-07,
      "loss": 0.7447,
      "step": 9116
    },
    {
      "epoch": 0.8222031834783785,
      "grad_norm": 2.0669331158362567,
      "learning_rate": 3.224202664289346e-07,
      "loss": 0.7336,
      "step": 9117
    },
    {
      "epoch": 0.8222933670018487,
      "grad_norm": 3.0094836063930366,
      "learning_rate": 3.2210226143843257e-07,
      "loss": 0.6888,
      "step": 9118
    },
    {
      "epoch": 0.822383550525319,
      "grad_norm": 0.6532242366471108,
      "learning_rate": 3.217843996154173e-07,
      "loss": 0.6134,
      "step": 9119
    },
    {
      "epoch": 0.8224737340487893,
      "grad_norm": 0.6052895065296732,
      "learning_rate": 3.2146668098701055e-07,
      "loss": 0.529,
      "step": 9120
    },
    {
      "epoch": 0.8225639175722596,
      "grad_norm": 2.317532090476195,
      "learning_rate": 3.2114910558032215e-07,
      "loss": 0.8727,
      "step": 9121
    },
    {
      "epoch": 0.8226541010957298,
      "grad_norm": 3.6471693951423196,
      "learning_rate": 3.2083167342244945e-07,
      "loss": 0.7781,
      "step": 9122
    },
    {
      "epoch": 0.8227442846192,
      "grad_norm": 1.8660273439176316,
      "learning_rate": 3.205143845404763e-07,
      "loss": 0.8424,
      "step": 9123
    },
    {
      "epoch": 0.8228344681426704,
      "grad_norm": 1.573516689152635,
      "learning_rate": 3.201972389614773e-07,
      "loss": 0.7193,
      "step": 9124
    },
    {
      "epoch": 0.8229246516661406,
      "grad_norm": 1.9430609852579617,
      "learning_rate": 3.198802367125115e-07,
      "loss": 0.7419,
      "step": 9125
    },
    {
      "epoch": 0.8230148351896108,
      "grad_norm": 1.6915126000638974,
      "learning_rate": 3.195633778206288e-07,
      "loss": 0.754,
      "step": 9126
    },
    {
      "epoch": 0.8231050187130812,
      "grad_norm": 2.3006599018627907,
      "learning_rate": 3.19246662312864e-07,
      "loss": 0.8215,
      "step": 9127
    },
    {
      "epoch": 0.8231952022365514,
      "grad_norm": 0.6421700199096412,
      "learning_rate": 3.189300902162417e-07,
      "loss": 0.5791,
      "step": 9128
    },
    {
      "epoch": 0.8232853857600216,
      "grad_norm": 2.484736674151663,
      "learning_rate": 3.1861366155777327e-07,
      "loss": 0.7603,
      "step": 9129
    },
    {
      "epoch": 0.8233755692834919,
      "grad_norm": 1.4676921362577664,
      "learning_rate": 3.182973763644583e-07,
      "loss": 0.7542,
      "step": 9130
    },
    {
      "epoch": 0.8234657528069622,
      "grad_norm": 2.6704537699714876,
      "learning_rate": 3.1798123466328463e-07,
      "loss": 0.6714,
      "step": 9131
    },
    {
      "epoch": 0.8235559363304324,
      "grad_norm": 2.0459160341949105,
      "learning_rate": 3.17665236481226e-07,
      "loss": 0.7491,
      "step": 9132
    },
    {
      "epoch": 0.8236461198539027,
      "grad_norm": 2.6125869626990017,
      "learning_rate": 3.1734938184524576e-07,
      "loss": 0.7084,
      "step": 9133
    },
    {
      "epoch": 0.8237363033773729,
      "grad_norm": 1.8572843964056986,
      "learning_rate": 3.1703367078229427e-07,
      "loss": 0.7572,
      "step": 9134
    },
    {
      "epoch": 0.8238264869008433,
      "grad_norm": 1.7625283286340878,
      "learning_rate": 3.167181033193096e-07,
      "loss": 0.7075,
      "step": 9135
    },
    {
      "epoch": 0.8239166704243135,
      "grad_norm": 1.615517827801343,
      "learning_rate": 3.16402679483218e-07,
      "loss": 0.7591,
      "step": 9136
    },
    {
      "epoch": 0.8240068539477837,
      "grad_norm": 2.277226911266312,
      "learning_rate": 3.1608739930093366e-07,
      "loss": 0.6789,
      "step": 9137
    },
    {
      "epoch": 0.824097037471254,
      "grad_norm": 1.8507196624672708,
      "learning_rate": 3.157722627993562e-07,
      "loss": 0.6581,
      "step": 9138
    },
    {
      "epoch": 0.8241872209947243,
      "grad_norm": 1.773587123928038,
      "learning_rate": 3.1545727000537727e-07,
      "loss": 0.7051,
      "step": 9139
    },
    {
      "epoch": 0.8242774045181945,
      "grad_norm": 2.0967011790424497,
      "learning_rate": 3.151424209458713e-07,
      "loss": 0.7425,
      "step": 9140
    },
    {
      "epoch": 0.8243675880416648,
      "grad_norm": 1.7427104626916379,
      "learning_rate": 3.148277156477053e-07,
      "loss": 0.7593,
      "step": 9141
    },
    {
      "epoch": 0.824457771565135,
      "grad_norm": 2.215480598511111,
      "learning_rate": 3.145131541377299e-07,
      "loss": 0.7475,
      "step": 9142
    },
    {
      "epoch": 0.8245479550886053,
      "grad_norm": 1.8117153837198203,
      "learning_rate": 3.1419873644278606e-07,
      "loss": 0.7124,
      "step": 9143
    },
    {
      "epoch": 0.8246381386120756,
      "grad_norm": 7.987625565235901,
      "learning_rate": 3.1388446258970147e-07,
      "loss": 0.768,
      "step": 9144
    },
    {
      "epoch": 0.8247283221355458,
      "grad_norm": 2.537282184591778,
      "learning_rate": 3.1357033260529145e-07,
      "loss": 0.7285,
      "step": 9145
    },
    {
      "epoch": 0.824818505659016,
      "grad_norm": 1.5990210488742924,
      "learning_rate": 3.1325634651636025e-07,
      "loss": 0.8233,
      "step": 9146
    },
    {
      "epoch": 0.8249086891824864,
      "grad_norm": 1.5067844344983488,
      "learning_rate": 3.1294250434969694e-07,
      "loss": 0.7883,
      "step": 9147
    },
    {
      "epoch": 0.8249988727059566,
      "grad_norm": 2.898784112898185,
      "learning_rate": 3.1262880613208274e-07,
      "loss": 0.7873,
      "step": 9148
    },
    {
      "epoch": 0.8250890562294269,
      "grad_norm": 2.396459105217729,
      "learning_rate": 3.123152518902823e-07,
      "loss": 0.7835,
      "step": 9149
    },
    {
      "epoch": 0.8251792397528972,
      "grad_norm": 1.4525586737175338,
      "learning_rate": 3.1200184165105017e-07,
      "loss": 0.6976,
      "step": 9150
    },
    {
      "epoch": 0.8252694232763674,
      "grad_norm": 1.765264491182887,
      "learning_rate": 3.116885754411287e-07,
      "loss": 0.7543,
      "step": 9151
    },
    {
      "epoch": 0.8253596067998377,
      "grad_norm": 5.9664292542235255,
      "learning_rate": 3.1137545328724703e-07,
      "loss": 0.6791,
      "step": 9152
    },
    {
      "epoch": 0.8254497903233079,
      "grad_norm": 2.204862410043516,
      "learning_rate": 3.1106247521612285e-07,
      "loss": 0.7489,
      "step": 9153
    },
    {
      "epoch": 0.8255399738467782,
      "grad_norm": 2.3825531364992023,
      "learning_rate": 3.107496412544612e-07,
      "loss": 0.762,
      "step": 9154
    },
    {
      "epoch": 0.8256301573702485,
      "grad_norm": 2.4262976841512915,
      "learning_rate": 3.1043695142895397e-07,
      "loss": 0.7707,
      "step": 9155
    },
    {
      "epoch": 0.8257203408937187,
      "grad_norm": 1.6280733535980634,
      "learning_rate": 3.101244057662828e-07,
      "loss": 0.6669,
      "step": 9156
    },
    {
      "epoch": 0.8258105244171889,
      "grad_norm": 1.7166212597492838,
      "learning_rate": 3.098120042931152e-07,
      "loss": 0.7339,
      "step": 9157
    },
    {
      "epoch": 0.8259007079406593,
      "grad_norm": 1.6184003032469076,
      "learning_rate": 3.0949974703610647e-07,
      "loss": 0.7883,
      "step": 9158
    },
    {
      "epoch": 0.8259908914641295,
      "grad_norm": 1.6975052535996862,
      "learning_rate": 3.0918763402190107e-07,
      "loss": 0.7356,
      "step": 9159
    },
    {
      "epoch": 0.8260810749875998,
      "grad_norm": 1.8056865750565565,
      "learning_rate": 3.088756652771296e-07,
      "loss": 0.7015,
      "step": 9160
    },
    {
      "epoch": 0.82617125851107,
      "grad_norm": 1.8063741621485783,
      "learning_rate": 3.0856384082841147e-07,
      "loss": 0.8425,
      "step": 9161
    },
    {
      "epoch": 0.8262614420345403,
      "grad_norm": 1.6681120406698289,
      "learning_rate": 3.0825216070235207e-07,
      "loss": 0.6907,
      "step": 9162
    },
    {
      "epoch": 0.8263516255580106,
      "grad_norm": 1.899282008584382,
      "learning_rate": 3.0794062492554764e-07,
      "loss": 0.8588,
      "step": 9163
    },
    {
      "epoch": 0.8264418090814808,
      "grad_norm": 2.0313163159622345,
      "learning_rate": 3.076292335245783e-07,
      "loss": 0.7612,
      "step": 9164
    },
    {
      "epoch": 0.826531992604951,
      "grad_norm": 2.3711683585229557,
      "learning_rate": 3.073179865260145e-07,
      "loss": 0.8181,
      "step": 9165
    },
    {
      "epoch": 0.8266221761284214,
      "grad_norm": 1.5741878534886773,
      "learning_rate": 3.070068839564135e-07,
      "loss": 0.7528,
      "step": 9166
    },
    {
      "epoch": 0.8267123596518916,
      "grad_norm": 1.9222950670126724,
      "learning_rate": 3.0669592584232006e-07,
      "loss": 0.7872,
      "step": 9167
    },
    {
      "epoch": 0.8268025431753618,
      "grad_norm": 1.9267121515987822,
      "learning_rate": 3.063851122102672e-07,
      "loss": 0.6594,
      "step": 9168
    },
    {
      "epoch": 0.8268927266988321,
      "grad_norm": 1.9696091487842478,
      "learning_rate": 3.06074443086775e-07,
      "loss": 0.6699,
      "step": 9169
    },
    {
      "epoch": 0.8269829102223024,
      "grad_norm": 0.6155530800417682,
      "learning_rate": 3.057639184983514e-07,
      "loss": 0.531,
      "step": 9170
    },
    {
      "epoch": 0.8270730937457726,
      "grad_norm": 1.3824553378653222,
      "learning_rate": 3.054535384714927e-07,
      "loss": 0.752,
      "step": 9171
    },
    {
      "epoch": 0.8271632772692429,
      "grad_norm": 1.732069734482204,
      "learning_rate": 3.0514330303268135e-07,
      "loss": 0.9105,
      "step": 9172
    },
    {
      "epoch": 0.8272534607927132,
      "grad_norm": 1.8389841327159022,
      "learning_rate": 3.0483321220838876e-07,
      "loss": 0.6989,
      "step": 9173
    },
    {
      "epoch": 0.8273436443161835,
      "grad_norm": 2.1105027186582506,
      "learning_rate": 3.045232660250734e-07,
      "loss": 0.6029,
      "step": 9174
    },
    {
      "epoch": 0.8274338278396537,
      "grad_norm": 1.609109282547883,
      "learning_rate": 3.0421346450918185e-07,
      "loss": 0.7768,
      "step": 9175
    },
    {
      "epoch": 0.8275240113631239,
      "grad_norm": 1.8297184577492505,
      "learning_rate": 3.039038076871485e-07,
      "loss": 0.6936,
      "step": 9176
    },
    {
      "epoch": 0.8276141948865943,
      "grad_norm": 1.6705011445046973,
      "learning_rate": 3.035942955853934e-07,
      "loss": 0.7638,
      "step": 9177
    },
    {
      "epoch": 0.8277043784100645,
      "grad_norm": 1.7129839498744481,
      "learning_rate": 3.0328492823032804e-07,
      "loss": 0.7215,
      "step": 9178
    },
    {
      "epoch": 0.8277945619335347,
      "grad_norm": 1.7294588263474935,
      "learning_rate": 3.029757056483471e-07,
      "loss": 0.684,
      "step": 9179
    },
    {
      "epoch": 0.827884745457005,
      "grad_norm": 1.7514687913776255,
      "learning_rate": 3.026666278658372e-07,
      "loss": 0.6338,
      "step": 9180
    },
    {
      "epoch": 0.8279749289804753,
      "grad_norm": 3.786888647311435,
      "learning_rate": 3.023576949091691e-07,
      "loss": 0.7445,
      "step": 9181
    },
    {
      "epoch": 0.8280651125039455,
      "grad_norm": 2.081133163884304,
      "learning_rate": 3.020489068047032e-07,
      "loss": 0.6705,
      "step": 9182
    },
    {
      "epoch": 0.8281552960274158,
      "grad_norm": 1.7102321058992502,
      "learning_rate": 3.017402635787869e-07,
      "loss": 0.7085,
      "step": 9183
    },
    {
      "epoch": 0.828245479550886,
      "grad_norm": 1.6536037970378894,
      "learning_rate": 3.0143176525775537e-07,
      "loss": 0.7225,
      "step": 9184
    },
    {
      "epoch": 0.8283356630743564,
      "grad_norm": 1.658716201035425,
      "learning_rate": 3.0112341186793155e-07,
      "loss": 0.6924,
      "step": 9185
    },
    {
      "epoch": 0.8284258465978266,
      "grad_norm": 1.5037463610978463,
      "learning_rate": 3.008152034356264e-07,
      "loss": 0.6819,
      "step": 9186
    },
    {
      "epoch": 0.8285160301212968,
      "grad_norm": 0.6314499396766116,
      "learning_rate": 3.005071399871366e-07,
      "loss": 0.5644,
      "step": 9187
    },
    {
      "epoch": 0.828606213644767,
      "grad_norm": 1.934140338129567,
      "learning_rate": 3.0019922154874853e-07,
      "loss": 0.8073,
      "step": 9188
    },
    {
      "epoch": 0.8286963971682374,
      "grad_norm": 1.8218328978186948,
      "learning_rate": 2.998914481467356e-07,
      "loss": 0.7077,
      "step": 9189
    },
    {
      "epoch": 0.8287865806917076,
      "grad_norm": 1.822021271737434,
      "learning_rate": 2.9958381980735837e-07,
      "loss": 0.6956,
      "step": 9190
    },
    {
      "epoch": 0.8288767642151779,
      "grad_norm": 2.3751287927339804,
      "learning_rate": 2.992763365568658e-07,
      "loss": 0.6725,
      "step": 9191
    },
    {
      "epoch": 0.8289669477386481,
      "grad_norm": 1.9298633078361724,
      "learning_rate": 2.98968998421494e-07,
      "loss": 0.7461,
      "step": 9192
    },
    {
      "epoch": 0.8290571312621184,
      "grad_norm": 1.6449049572605738,
      "learning_rate": 2.98661805427467e-07,
      "loss": 0.736,
      "step": 9193
    },
    {
      "epoch": 0.8291473147855887,
      "grad_norm": 1.7212419651262274,
      "learning_rate": 2.9835475760099483e-07,
      "loss": 0.7595,
      "step": 9194
    },
    {
      "epoch": 0.8292374983090589,
      "grad_norm": 1.5821860155297913,
      "learning_rate": 2.9804785496827856e-07,
      "loss": 0.7618,
      "step": 9195
    },
    {
      "epoch": 0.8293276818325293,
      "grad_norm": 2.0995371535543907,
      "learning_rate": 2.977410975555028e-07,
      "loss": 0.7449,
      "step": 9196
    },
    {
      "epoch": 0.8294178653559995,
      "grad_norm": 1.8106325987240637,
      "learning_rate": 2.9743448538884376e-07,
      "loss": 0.8047,
      "step": 9197
    },
    {
      "epoch": 0.8295080488794697,
      "grad_norm": 1.8683724573799116,
      "learning_rate": 2.9712801849446154e-07,
      "loss": 0.7346,
      "step": 9198
    },
    {
      "epoch": 0.82959823240294,
      "grad_norm": 1.8556463859253407,
      "learning_rate": 2.9682169689850665e-07,
      "loss": 0.693,
      "step": 9199
    },
    {
      "epoch": 0.8296884159264103,
      "grad_norm": 0.6645600866917726,
      "learning_rate": 2.9651552062711573e-07,
      "loss": 0.5276,
      "step": 9200
    },
    {
      "epoch": 0.8297785994498805,
      "grad_norm": 1.9548817782742793,
      "learning_rate": 2.9620948970641333e-07,
      "loss": 0.8197,
      "step": 9201
    },
    {
      "epoch": 0.8298687829733508,
      "grad_norm": 4.681648162084567,
      "learning_rate": 2.959036041625125e-07,
      "loss": 0.761,
      "step": 9202
    },
    {
      "epoch": 0.829958966496821,
      "grad_norm": 1.6874025811226612,
      "learning_rate": 2.95597864021512e-07,
      "loss": 0.6553,
      "step": 9203
    },
    {
      "epoch": 0.8300491500202913,
      "grad_norm": 2.471201198869774,
      "learning_rate": 2.9529226930949966e-07,
      "loss": 0.7905,
      "step": 9204
    },
    {
      "epoch": 0.8301393335437616,
      "grad_norm": 1.9208319089584973,
      "learning_rate": 2.949868200525505e-07,
      "loss": 0.7599,
      "step": 9205
    },
    {
      "epoch": 0.8302295170672318,
      "grad_norm": 5.106752428342291,
      "learning_rate": 2.9468151627672734e-07,
      "loss": 0.7371,
      "step": 9206
    },
    {
      "epoch": 0.830319700590702,
      "grad_norm": 1.7513744542556184,
      "learning_rate": 2.9437635800808026e-07,
      "loss": 0.7,
      "step": 9207
    },
    {
      "epoch": 0.8304098841141724,
      "grad_norm": 2.06077815922809,
      "learning_rate": 2.940713452726473e-07,
      "loss": 0.7425,
      "step": 9208
    },
    {
      "epoch": 0.8305000676376426,
      "grad_norm": 1.8237849175063068,
      "learning_rate": 2.937664780964526e-07,
      "loss": 0.7401,
      "step": 9209
    },
    {
      "epoch": 0.8305902511611128,
      "grad_norm": 1.8907280885949282,
      "learning_rate": 2.9346175650551133e-07,
      "loss": 0.7549,
      "step": 9210
    },
    {
      "epoch": 0.8306804346845831,
      "grad_norm": 1.552327727804834,
      "learning_rate": 2.931571805258215e-07,
      "loss": 0.7662,
      "step": 9211
    },
    {
      "epoch": 0.8307706182080534,
      "grad_norm": 1.981646907773126,
      "learning_rate": 2.9285275018337353e-07,
      "loss": 0.7948,
      "step": 9212
    },
    {
      "epoch": 0.8308608017315237,
      "grad_norm": 1.7204289645513589,
      "learning_rate": 2.9254846550414146e-07,
      "loss": 0.8434,
      "step": 9213
    },
    {
      "epoch": 0.8309509852549939,
      "grad_norm": 2.5539317281764364,
      "learning_rate": 2.922443265140893e-07,
      "loss": 0.7981,
      "step": 9214
    },
    {
      "epoch": 0.8310411687784641,
      "grad_norm": 1.703773515053434,
      "learning_rate": 2.919403332391674e-07,
      "loss": 0.7456,
      "step": 9215
    },
    {
      "epoch": 0.8311313523019345,
      "grad_norm": 2.4414930648627093,
      "learning_rate": 2.9163648570531464e-07,
      "loss": 0.7662,
      "step": 9216
    },
    {
      "epoch": 0.8312215358254047,
      "grad_norm": 2.5230136196450763,
      "learning_rate": 2.9133278393845717e-07,
      "loss": 0.6807,
      "step": 9217
    },
    {
      "epoch": 0.8313117193488749,
      "grad_norm": 2.140719891971808,
      "learning_rate": 2.9102922796450703e-07,
      "loss": 0.8185,
      "step": 9218
    },
    {
      "epoch": 0.8314019028723453,
      "grad_norm": 1.7295897858543057,
      "learning_rate": 2.907258178093672e-07,
      "loss": 0.7031,
      "step": 9219
    },
    {
      "epoch": 0.8314920863958155,
      "grad_norm": 1.4420931040247225,
      "learning_rate": 2.904225534989251e-07,
      "loss": 0.6694,
      "step": 9220
    },
    {
      "epoch": 0.8315822699192857,
      "grad_norm": 2.6403431939530604,
      "learning_rate": 2.901194350590572e-07,
      "loss": 0.6463,
      "step": 9221
    },
    {
      "epoch": 0.831672453442756,
      "grad_norm": 2.0129768659586036,
      "learning_rate": 2.898164625156274e-07,
      "loss": 0.6269,
      "step": 9222
    },
    {
      "epoch": 0.8317626369662263,
      "grad_norm": 2.2665448656267912,
      "learning_rate": 2.8951363589448676e-07,
      "loss": 0.7009,
      "step": 9223
    },
    {
      "epoch": 0.8318528204896966,
      "grad_norm": 1.5298897287111877,
      "learning_rate": 2.8921095522147434e-07,
      "loss": 0.7406,
      "step": 9224
    },
    {
      "epoch": 0.8319430040131668,
      "grad_norm": 1.6578206945678788,
      "learning_rate": 2.8890842052241683e-07,
      "loss": 0.7722,
      "step": 9225
    },
    {
      "epoch": 0.832033187536637,
      "grad_norm": 0.5919065971762193,
      "learning_rate": 2.886060318231267e-07,
      "loss": 0.5269,
      "step": 9226
    },
    {
      "epoch": 0.8321233710601074,
      "grad_norm": 5.257006556509195,
      "learning_rate": 2.8830378914940755e-07,
      "loss": 0.7713,
      "step": 9227
    },
    {
      "epoch": 0.8322135545835776,
      "grad_norm": 7.561867679611637,
      "learning_rate": 2.8800169252704675e-07,
      "loss": 0.7303,
      "step": 9228
    },
    {
      "epoch": 0.8323037381070478,
      "grad_norm": 1.6614617454310412,
      "learning_rate": 2.8769974198182143e-07,
      "loss": 0.7689,
      "step": 9229
    },
    {
      "epoch": 0.8323939216305181,
      "grad_norm": 4.307080166416283,
      "learning_rate": 2.873979375394955e-07,
      "loss": 0.7448,
      "step": 9230
    },
    {
      "epoch": 0.8324841051539884,
      "grad_norm": 1.8545400122190372,
      "learning_rate": 2.870962792258209e-07,
      "loss": 0.7781,
      "step": 9231
    },
    {
      "epoch": 0.8325742886774586,
      "grad_norm": 3.22804887355029,
      "learning_rate": 2.8679476706653716e-07,
      "loss": 0.7534,
      "step": 9232
    },
    {
      "epoch": 0.8326644722009289,
      "grad_norm": 1.9227089661924286,
      "learning_rate": 2.864934010873692e-07,
      "loss": 0.7683,
      "step": 9233
    },
    {
      "epoch": 0.8327546557243991,
      "grad_norm": 1.463942675333289,
      "learning_rate": 2.8619218131403357e-07,
      "loss": 0.719,
      "step": 9234
    },
    {
      "epoch": 0.8328448392478695,
      "grad_norm": 1.7477349200083103,
      "learning_rate": 2.858911077722299e-07,
      "loss": 0.7675,
      "step": 9235
    },
    {
      "epoch": 0.8329350227713397,
      "grad_norm": 2.7008454517946747,
      "learning_rate": 2.855901804876493e-07,
      "loss": 0.6003,
      "step": 9236
    },
    {
      "epoch": 0.8330252062948099,
      "grad_norm": 1.5499192155274593,
      "learning_rate": 2.852893994859673e-07,
      "loss": 0.7855,
      "step": 9237
    },
    {
      "epoch": 0.8331153898182801,
      "grad_norm": 2.162003511974718,
      "learning_rate": 2.849887647928484e-07,
      "loss": 0.6875,
      "step": 9238
    },
    {
      "epoch": 0.8332055733417505,
      "grad_norm": 2.3787672381194915,
      "learning_rate": 2.8468827643394465e-07,
      "loss": 0.648,
      "step": 9239
    },
    {
      "epoch": 0.8332957568652207,
      "grad_norm": 1.6357816139138437,
      "learning_rate": 2.843879344348954e-07,
      "loss": 0.7531,
      "step": 9240
    },
    {
      "epoch": 0.833385940388691,
      "grad_norm": 1.5003356558044354,
      "learning_rate": 2.840877388213272e-07,
      "loss": 0.7154,
      "step": 9241
    },
    {
      "epoch": 0.8334761239121612,
      "grad_norm": 1.9869390427790805,
      "learning_rate": 2.8378768961885515e-07,
      "loss": 0.7481,
      "step": 9242
    },
    {
      "epoch": 0.8335663074356315,
      "grad_norm": 3.411496942122094,
      "learning_rate": 2.8348778685307983e-07,
      "loss": 0.7757,
      "step": 9243
    },
    {
      "epoch": 0.8336564909591018,
      "grad_norm": 1.769206777301514,
      "learning_rate": 2.831880305495915e-07,
      "loss": 0.8201,
      "step": 9244
    },
    {
      "epoch": 0.833746674482572,
      "grad_norm": 0.6753146153619444,
      "learning_rate": 2.828884207339668e-07,
      "loss": 0.6119,
      "step": 9245
    },
    {
      "epoch": 0.8338368580060423,
      "grad_norm": 1.9785943123928784,
      "learning_rate": 2.8258895743177014e-07,
      "loss": 0.8252,
      "step": 9246
    },
    {
      "epoch": 0.8339270415295126,
      "grad_norm": 1.7044685034288278,
      "learning_rate": 2.8228964066855356e-07,
      "loss": 0.8191,
      "step": 9247
    },
    {
      "epoch": 0.8340172250529828,
      "grad_norm": 1.9661059108158863,
      "learning_rate": 2.819904704698555e-07,
      "loss": 0.7236,
      "step": 9248
    },
    {
      "epoch": 0.834107408576453,
      "grad_norm": 0.6393461255576148,
      "learning_rate": 2.8169144686120437e-07,
      "loss": 0.5279,
      "step": 9249
    },
    {
      "epoch": 0.8341975920999234,
      "grad_norm": 6.020366751819695,
      "learning_rate": 2.8139256986811254e-07,
      "loss": 0.6576,
      "step": 9250
    },
    {
      "epoch": 0.8342877756233936,
      "grad_norm": 1.7496935031063647,
      "learning_rate": 2.8109383951608424e-07,
      "loss": 0.7943,
      "step": 9251
    },
    {
      "epoch": 0.8343779591468639,
      "grad_norm": 1.7436568560043828,
      "learning_rate": 2.8079525583060683e-07,
      "loss": 0.8077,
      "step": 9252
    },
    {
      "epoch": 0.8344681426703341,
      "grad_norm": 1.38309036308044,
      "learning_rate": 2.804968188371577e-07,
      "loss": 0.7534,
      "step": 9253
    },
    {
      "epoch": 0.8345583261938044,
      "grad_norm": 1.6589100188030783,
      "learning_rate": 2.801985285612014e-07,
      "loss": 0.7469,
      "step": 9254
    },
    {
      "epoch": 0.8346485097172747,
      "grad_norm": 2.016136873291519,
      "learning_rate": 2.7990038502818934e-07,
      "loss": 0.716,
      "step": 9255
    },
    {
      "epoch": 0.8347386932407449,
      "grad_norm": 1.4972276499187265,
      "learning_rate": 2.796023882635612e-07,
      "loss": 0.7223,
      "step": 9256
    },
    {
      "epoch": 0.8348288767642151,
      "grad_norm": 1.541257116544185,
      "learning_rate": 2.7930453829274323e-07,
      "loss": 0.7941,
      "step": 9257
    },
    {
      "epoch": 0.8349190602876855,
      "grad_norm": 4.297594044950653,
      "learning_rate": 2.7900683514115054e-07,
      "loss": 0.7894,
      "step": 9258
    },
    {
      "epoch": 0.8350092438111557,
      "grad_norm": 1.925957785788453,
      "learning_rate": 2.787092788341836e-07,
      "loss": 0.6832,
      "step": 9259
    },
    {
      "epoch": 0.8350994273346259,
      "grad_norm": 1.5920945339745671,
      "learning_rate": 2.7841186939723195e-07,
      "loss": 0.7445,
      "step": 9260
    },
    {
      "epoch": 0.8351896108580962,
      "grad_norm": 1.7894156161650898,
      "learning_rate": 2.7811460685567255e-07,
      "loss": 0.7188,
      "step": 9261
    },
    {
      "epoch": 0.8352797943815665,
      "grad_norm": 4.856020048335763,
      "learning_rate": 2.778174912348692e-07,
      "loss": 0.7414,
      "step": 9262
    },
    {
      "epoch": 0.8353699779050368,
      "grad_norm": 2.298074940833154,
      "learning_rate": 2.7752052256017354e-07,
      "loss": 0.7093,
      "step": 9263
    },
    {
      "epoch": 0.835460161428507,
      "grad_norm": 1.8023234815578149,
      "learning_rate": 2.7722370085692493e-07,
      "loss": 0.7159,
      "step": 9264
    },
    {
      "epoch": 0.8355503449519772,
      "grad_norm": 2.0807657314202226,
      "learning_rate": 2.769270261504486e-07,
      "loss": 0.719,
      "step": 9265
    },
    {
      "epoch": 0.8356405284754476,
      "grad_norm": 1.597059366889161,
      "learning_rate": 2.7663049846606015e-07,
      "loss": 0.7048,
      "step": 9266
    },
    {
      "epoch": 0.8357307119989178,
      "grad_norm": 1.529461029713524,
      "learning_rate": 2.763341178290592e-07,
      "loss": 0.7611,
      "step": 9267
    },
    {
      "epoch": 0.835820895522388,
      "grad_norm": 2.7385623803896455,
      "learning_rate": 2.7603788426473663e-07,
      "loss": 0.7713,
      "step": 9268
    },
    {
      "epoch": 0.8359110790458584,
      "grad_norm": 2.128728176683796,
      "learning_rate": 2.7574179779836695e-07,
      "loss": 0.6791,
      "step": 9269
    },
    {
      "epoch": 0.8360012625693286,
      "grad_norm": 1.6917395846932863,
      "learning_rate": 2.754458584552146e-07,
      "loss": 0.7053,
      "step": 9270
    },
    {
      "epoch": 0.8360914460927988,
      "grad_norm": 1.9918004277001802,
      "learning_rate": 2.751500662605308e-07,
      "loss": 0.731,
      "step": 9271
    },
    {
      "epoch": 0.8361816296162691,
      "grad_norm": 1.6842957911507148,
      "learning_rate": 2.7485442123955383e-07,
      "loss": 0.7039,
      "step": 9272
    },
    {
      "epoch": 0.8362718131397394,
      "grad_norm": 1.7618941327612716,
      "learning_rate": 2.7455892341751075e-07,
      "loss": 0.6272,
      "step": 9273
    },
    {
      "epoch": 0.8363619966632097,
      "grad_norm": 1.7653426898327067,
      "learning_rate": 2.7426357281961365e-07,
      "loss": 0.6177,
      "step": 9274
    },
    {
      "epoch": 0.8364521801866799,
      "grad_norm": 1.891646054305426,
      "learning_rate": 2.7396836947106416e-07,
      "loss": 0.7467,
      "step": 9275
    },
    {
      "epoch": 0.8365423637101501,
      "grad_norm": 1.7503199284890862,
      "learning_rate": 2.736733133970506e-07,
      "loss": 0.8399,
      "step": 9276
    },
    {
      "epoch": 0.8366325472336205,
      "grad_norm": 1.4494988923872907,
      "learning_rate": 2.7337840462274896e-07,
      "loss": 0.7829,
      "step": 9277
    },
    {
      "epoch": 0.8367227307570907,
      "grad_norm": 1.7701561773762977,
      "learning_rate": 2.730836431733221e-07,
      "loss": 0.7151,
      "step": 9278
    },
    {
      "epoch": 0.8368129142805609,
      "grad_norm": 2.8234283903109043,
      "learning_rate": 2.727890290739212e-07,
      "loss": 0.7956,
      "step": 9279
    },
    {
      "epoch": 0.8369030978040312,
      "grad_norm": 1.5043624384204664,
      "learning_rate": 2.7249456234968395e-07,
      "loss": 0.7798,
      "step": 9280
    },
    {
      "epoch": 0.8369932813275015,
      "grad_norm": 1.5911926246002253,
      "learning_rate": 2.722002430257364e-07,
      "loss": 0.8133,
      "step": 9281
    },
    {
      "epoch": 0.8370834648509717,
      "grad_norm": 1.9851218192437332,
      "learning_rate": 2.7190607112719035e-07,
      "loss": 0.7028,
      "step": 9282
    },
    {
      "epoch": 0.837173648374442,
      "grad_norm": 2.7266795835523254,
      "learning_rate": 2.716120466791476e-07,
      "loss": 0.666,
      "step": 9283
    },
    {
      "epoch": 0.8372638318979122,
      "grad_norm": 2.102182630065748,
      "learning_rate": 2.7131816970669483e-07,
      "loss": 0.7201,
      "step": 9284
    },
    {
      "epoch": 0.8373540154213825,
      "grad_norm": 1.875152200243588,
      "learning_rate": 2.7102444023490777e-07,
      "loss": 0.7509,
      "step": 9285
    },
    {
      "epoch": 0.8374441989448528,
      "grad_norm": 1.5712358028283342,
      "learning_rate": 2.70730858288849e-07,
      "loss": 0.8002,
      "step": 9286
    },
    {
      "epoch": 0.837534382468323,
      "grad_norm": 1.7010280754198912,
      "learning_rate": 2.704374238935685e-07,
      "loss": 0.6813,
      "step": 9287
    },
    {
      "epoch": 0.8376245659917932,
      "grad_norm": 0.673134200441465,
      "learning_rate": 2.70144137074104e-07,
      "loss": 0.575,
      "step": 9288
    },
    {
      "epoch": 0.8377147495152636,
      "grad_norm": 1.9876980928752233,
      "learning_rate": 2.6985099785547926e-07,
      "loss": 0.7639,
      "step": 9289
    },
    {
      "epoch": 0.8378049330387338,
      "grad_norm": 4.934612733744285,
      "learning_rate": 2.695580062627083e-07,
      "loss": 0.7723,
      "step": 9290
    },
    {
      "epoch": 0.8378951165622041,
      "grad_norm": 3.4182625431770033,
      "learning_rate": 2.692651623207891e-07,
      "loss": 0.7395,
      "step": 9291
    },
    {
      "epoch": 0.8379853000856744,
      "grad_norm": 1.4147588374725875,
      "learning_rate": 2.689724660547097e-07,
      "loss": 0.7315,
      "step": 9292
    },
    {
      "epoch": 0.8380754836091446,
      "grad_norm": 0.6212014638718933,
      "learning_rate": 2.686799174894441e-07,
      "loss": 0.5717,
      "step": 9293
    },
    {
      "epoch": 0.8381656671326149,
      "grad_norm": 0.777884175671443,
      "learning_rate": 2.683875166499545e-07,
      "loss": 0.5724,
      "step": 9294
    },
    {
      "epoch": 0.8382558506560851,
      "grad_norm": 1.790301231419395,
      "learning_rate": 2.680952635611899e-07,
      "loss": 0.7514,
      "step": 9295
    },
    {
      "epoch": 0.8383460341795554,
      "grad_norm": 1.7962399894844225,
      "learning_rate": 2.678031582480875e-07,
      "loss": 0.6713,
      "step": 9296
    },
    {
      "epoch": 0.8384362177030257,
      "grad_norm": 1.6070659331425725,
      "learning_rate": 2.6751120073557e-07,
      "loss": 0.7855,
      "step": 9297
    },
    {
      "epoch": 0.8385264012264959,
      "grad_norm": 1.6821495287424182,
      "learning_rate": 2.672193910485505e-07,
      "loss": 0.7118,
      "step": 9298
    },
    {
      "epoch": 0.8386165847499661,
      "grad_norm": 1.7972898584865773,
      "learning_rate": 2.669277292119265e-07,
      "loss": 0.7433,
      "step": 9299
    },
    {
      "epoch": 0.8387067682734365,
      "grad_norm": 2.080183687352206,
      "learning_rate": 2.666362152505848e-07,
      "loss": 0.7475,
      "step": 9300
    },
    {
      "epoch": 0.8387969517969067,
      "grad_norm": 4.8248024776527645,
      "learning_rate": 2.663448491893989e-07,
      "loss": 0.6784,
      "step": 9301
    },
    {
      "epoch": 0.838887135320377,
      "grad_norm": 1.7832227791479172,
      "learning_rate": 2.6605363105322974e-07,
      "loss": 0.7208,
      "step": 9302
    },
    {
      "epoch": 0.8389773188438472,
      "grad_norm": 2.261175376963977,
      "learning_rate": 2.657625608669263e-07,
      "loss": 0.7438,
      "step": 9303
    },
    {
      "epoch": 0.8390675023673175,
      "grad_norm": 1.8801100080851203,
      "learning_rate": 2.654716386553224e-07,
      "loss": 0.6587,
      "step": 9304
    },
    {
      "epoch": 0.8391576858907878,
      "grad_norm": 1.8339921100034902,
      "learning_rate": 2.651808644432436e-07,
      "loss": 0.7356,
      "step": 9305
    },
    {
      "epoch": 0.839247869414258,
      "grad_norm": 1.7313132539617162,
      "learning_rate": 2.6489023825549807e-07,
      "loss": 0.8783,
      "step": 9306
    },
    {
      "epoch": 0.8393380529377282,
      "grad_norm": 1.97216319352526,
      "learning_rate": 2.6459976011688547e-07,
      "loss": 0.6715,
      "step": 9307
    },
    {
      "epoch": 0.8394282364611986,
      "grad_norm": 1.9126619158251796,
      "learning_rate": 2.6430943005219e-07,
      "loss": 0.7382,
      "step": 9308
    },
    {
      "epoch": 0.8395184199846688,
      "grad_norm": 1.401248217061769,
      "learning_rate": 2.6401924808618447e-07,
      "loss": 0.7706,
      "step": 9309
    },
    {
      "epoch": 0.839608603508139,
      "grad_norm": 2.6324715322782914,
      "learning_rate": 2.637292142436287e-07,
      "loss": 0.7428,
      "step": 9310
    },
    {
      "epoch": 0.8396987870316093,
      "grad_norm": 1.980481631144932,
      "learning_rate": 2.6343932854927e-07,
      "loss": 0.7137,
      "step": 9311
    },
    {
      "epoch": 0.8397889705550796,
      "grad_norm": 0.6829212259284325,
      "learning_rate": 2.6314959102784316e-07,
      "loss": 0.5697,
      "step": 9312
    },
    {
      "epoch": 0.8398791540785498,
      "grad_norm": 3.64536288177953,
      "learning_rate": 2.6286000170407074e-07,
      "loss": 0.744,
      "step": 9313
    },
    {
      "epoch": 0.8399693376020201,
      "grad_norm": 2.1333110073033974,
      "learning_rate": 2.625705606026607e-07,
      "loss": 0.7191,
      "step": 9314
    },
    {
      "epoch": 0.8400595211254904,
      "grad_norm": 2.5374479354252566,
      "learning_rate": 2.622812677483106e-07,
      "loss": 0.818,
      "step": 9315
    },
    {
      "epoch": 0.8401497046489607,
      "grad_norm": 1.9130295437985705,
      "learning_rate": 2.6199212316570453e-07,
      "loss": 0.8007,
      "step": 9316
    },
    {
      "epoch": 0.8402398881724309,
      "grad_norm": 1.643899683065852,
      "learning_rate": 2.617031268795138e-07,
      "loss": 0.7191,
      "step": 9317
    },
    {
      "epoch": 0.8403300716959011,
      "grad_norm": 3.30464406807859,
      "learning_rate": 2.614142789143976e-07,
      "loss": 0.7708,
      "step": 9318
    },
    {
      "epoch": 0.8404202552193715,
      "grad_norm": 1.482379173900191,
      "learning_rate": 2.6112557929500047e-07,
      "loss": 0.7223,
      "step": 9319
    },
    {
      "epoch": 0.8405104387428417,
      "grad_norm": 1.935226583299986,
      "learning_rate": 2.6083702804595817e-07,
      "loss": 0.7676,
      "step": 9320
    },
    {
      "epoch": 0.8406006222663119,
      "grad_norm": 1.4922639517647172,
      "learning_rate": 2.6054862519188915e-07,
      "loss": 0.7258,
      "step": 9321
    },
    {
      "epoch": 0.8406908057897822,
      "grad_norm": 1.6628619237478983,
      "learning_rate": 2.6026037075740357e-07,
      "loss": 0.6832,
      "step": 9322
    },
    {
      "epoch": 0.8407809893132525,
      "grad_norm": 1.863537209002425,
      "learning_rate": 2.5997226476709524e-07,
      "loss": 0.7058,
      "step": 9323
    },
    {
      "epoch": 0.8408711728367227,
      "grad_norm": 1.9057116764583442,
      "learning_rate": 2.5968430724554856e-07,
      "loss": 0.7366,
      "step": 9324
    },
    {
      "epoch": 0.840961356360193,
      "grad_norm": 1.9196328831410387,
      "learning_rate": 2.5939649821733225e-07,
      "loss": 0.6773,
      "step": 9325
    },
    {
      "epoch": 0.8410515398836632,
      "grad_norm": 1.9055224421632748,
      "learning_rate": 2.5910883770700433e-07,
      "loss": 0.7115,
      "step": 9326
    },
    {
      "epoch": 0.8411417234071336,
      "grad_norm": 1.3258109184366837,
      "learning_rate": 2.5882132573910965e-07,
      "loss": 0.583,
      "step": 9327
    },
    {
      "epoch": 0.8412319069306038,
      "grad_norm": 1.2947322194494761,
      "learning_rate": 2.585339623381801e-07,
      "loss": 0.7787,
      "step": 9328
    },
    {
      "epoch": 0.841322090454074,
      "grad_norm": 1.5025109890807287,
      "learning_rate": 2.582467475287358e-07,
      "loss": 0.7199,
      "step": 9329
    },
    {
      "epoch": 0.8414122739775443,
      "grad_norm": 1.7643151615061532,
      "learning_rate": 2.5795968133528224e-07,
      "loss": 0.7523,
      "step": 9330
    },
    {
      "epoch": 0.8415024575010146,
      "grad_norm": 1.6343045513075465,
      "learning_rate": 2.576727637823144e-07,
      "loss": 0.7439,
      "step": 9331
    },
    {
      "epoch": 0.8415926410244848,
      "grad_norm": 1.8844369552321534,
      "learning_rate": 2.5738599489431335e-07,
      "loss": 0.7856,
      "step": 9332
    },
    {
      "epoch": 0.8416828245479551,
      "grad_norm": 1.426006477020235,
      "learning_rate": 2.5709937469574794e-07,
      "loss": 0.8043,
      "step": 9333
    },
    {
      "epoch": 0.8417730080714253,
      "grad_norm": 1.4045461292860044,
      "learning_rate": 2.568129032110742e-07,
      "loss": 0.7828,
      "step": 9334
    },
    {
      "epoch": 0.8418631915948956,
      "grad_norm": 1.5630132976461764,
      "learning_rate": 2.5652658046473565e-07,
      "loss": 0.8218,
      "step": 9335
    },
    {
      "epoch": 0.8419533751183659,
      "grad_norm": 1.7567322527074627,
      "learning_rate": 2.5624040648116184e-07,
      "loss": 0.7424,
      "step": 9336
    },
    {
      "epoch": 0.8420435586418361,
      "grad_norm": 1.7255742623711927,
      "learning_rate": 2.5595438128477245e-07,
      "loss": 0.6871,
      "step": 9337
    },
    {
      "epoch": 0.8421337421653065,
      "grad_norm": 3.18020753099315,
      "learning_rate": 2.5566850489997096e-07,
      "loss": 0.7515,
      "step": 9338
    },
    {
      "epoch": 0.8422239256887767,
      "grad_norm": 2.241834446561792,
      "learning_rate": 2.5538277735115166e-07,
      "loss": 0.7705,
      "step": 9339
    },
    {
      "epoch": 0.8423141092122469,
      "grad_norm": 2.6873967350482233,
      "learning_rate": 2.5509719866269306e-07,
      "loss": 0.7945,
      "step": 9340
    },
    {
      "epoch": 0.8424042927357172,
      "grad_norm": 1.70875968109938,
      "learning_rate": 2.548117688589628e-07,
      "loss": 0.8048,
      "step": 9341
    },
    {
      "epoch": 0.8424944762591875,
      "grad_norm": 2.138935842728024,
      "learning_rate": 2.545264879643152e-07,
      "loss": 0.7268,
      "step": 9342
    },
    {
      "epoch": 0.8425846597826577,
      "grad_norm": 1.7807109083913413,
      "learning_rate": 2.542413560030923e-07,
      "loss": 0.7043,
      "step": 9343
    },
    {
      "epoch": 0.842674843306128,
      "grad_norm": 4.291880531290806,
      "learning_rate": 2.53956372999623e-07,
      "loss": 0.6352,
      "step": 9344
    },
    {
      "epoch": 0.8427650268295982,
      "grad_norm": 0.6494759895799677,
      "learning_rate": 2.5367153897822293e-07,
      "loss": 0.6094,
      "step": 9345
    },
    {
      "epoch": 0.8428552103530685,
      "grad_norm": 1.5905184427179637,
      "learning_rate": 2.5338685396319715e-07,
      "loss": 0.7353,
      "step": 9346
    },
    {
      "epoch": 0.8429453938765388,
      "grad_norm": 2.1814178926840158,
      "learning_rate": 2.531023179788352e-07,
      "loss": 0.6777,
      "step": 9347
    },
    {
      "epoch": 0.843035577400009,
      "grad_norm": 1.5783790157330775,
      "learning_rate": 2.528179310494158e-07,
      "loss": 0.718,
      "step": 9348
    },
    {
      "epoch": 0.8431257609234792,
      "grad_norm": 1.4129623011009815,
      "learning_rate": 2.5253369319920436e-07,
      "loss": 0.7628,
      "step": 9349
    },
    {
      "epoch": 0.8432159444469496,
      "grad_norm": 2.717087511413229,
      "learning_rate": 2.522496044524538e-07,
      "loss": 0.7444,
      "step": 9350
    },
    {
      "epoch": 0.8433061279704198,
      "grad_norm": 4.379911880651869,
      "learning_rate": 2.5196566483340386e-07,
      "loss": 0.8318,
      "step": 9351
    },
    {
      "epoch": 0.84339631149389,
      "grad_norm": 1.6630255822045104,
      "learning_rate": 2.516818743662825e-07,
      "loss": 0.761,
      "step": 9352
    },
    {
      "epoch": 0.8434864950173603,
      "grad_norm": 1.8903317184593815,
      "learning_rate": 2.5139823307530285e-07,
      "loss": 0.7409,
      "step": 9353
    },
    {
      "epoch": 0.8435766785408306,
      "grad_norm": 9.285166931788135,
      "learning_rate": 2.5111474098466836e-07,
      "loss": 0.778,
      "step": 9354
    },
    {
      "epoch": 0.8436668620643009,
      "grad_norm": 1.6293689976132788,
      "learning_rate": 2.50831398118567e-07,
      "loss": 0.7904,
      "step": 9355
    },
    {
      "epoch": 0.8437570455877711,
      "grad_norm": 1.6423794039925537,
      "learning_rate": 2.5054820450117576e-07,
      "loss": 0.6875,
      "step": 9356
    },
    {
      "epoch": 0.8438472291112413,
      "grad_norm": 1.684121458543797,
      "learning_rate": 2.502651601566579e-07,
      "loss": 0.7027,
      "step": 9357
    },
    {
      "epoch": 0.8439374126347117,
      "grad_norm": 1.7573166889729972,
      "learning_rate": 2.499822651091645e-07,
      "loss": 0.6952,
      "step": 9358
    },
    {
      "epoch": 0.8440275961581819,
      "grad_norm": 1.8819978463204168,
      "learning_rate": 2.496995193828344e-07,
      "loss": 0.7618,
      "step": 9359
    },
    {
      "epoch": 0.8441177796816521,
      "grad_norm": 1.7494845312349914,
      "learning_rate": 2.494169230017913e-07,
      "loss": 0.7402,
      "step": 9360
    },
    {
      "epoch": 0.8442079632051224,
      "grad_norm": 1.5103606202699835,
      "learning_rate": 2.491344759901499e-07,
      "loss": 0.7,
      "step": 9361
    },
    {
      "epoch": 0.8442981467285927,
      "grad_norm": 5.780230458030996,
      "learning_rate": 2.488521783720088e-07,
      "loss": 0.7362,
      "step": 9362
    },
    {
      "epoch": 0.844388330252063,
      "grad_norm": 2.4286358388004783,
      "learning_rate": 2.4857003017145526e-07,
      "loss": 0.7803,
      "step": 9363
    },
    {
      "epoch": 0.8444785137755332,
      "grad_norm": 1.7182098927206522,
      "learning_rate": 2.482880314125644e-07,
      "loss": 0.7788,
      "step": 9364
    },
    {
      "epoch": 0.8445686972990035,
      "grad_norm": 1.5982256110323616,
      "learning_rate": 2.4800618211939726e-07,
      "loss": 0.7331,
      "step": 9365
    },
    {
      "epoch": 0.8446588808224738,
      "grad_norm": 2.253824692015794,
      "learning_rate": 2.477244823160034e-07,
      "loss": 0.7491,
      "step": 9366
    },
    {
      "epoch": 0.844749064345944,
      "grad_norm": 1.6865227306653845,
      "learning_rate": 2.474429320264184e-07,
      "loss": 0.7663,
      "step": 9367
    },
    {
      "epoch": 0.8448392478694142,
      "grad_norm": 1.6725622172964267,
      "learning_rate": 2.47161531274666e-07,
      "loss": 0.6999,
      "step": 9368
    },
    {
      "epoch": 0.8449294313928846,
      "grad_norm": 1.6425875596581752,
      "learning_rate": 2.4688028008475714e-07,
      "loss": 0.6717,
      "step": 9369
    },
    {
      "epoch": 0.8450196149163548,
      "grad_norm": 2.018567798057024,
      "learning_rate": 2.465991784806891e-07,
      "loss": 0.7369,
      "step": 9370
    },
    {
      "epoch": 0.845109798439825,
      "grad_norm": 1.9134675015893716,
      "learning_rate": 2.463182264864472e-07,
      "loss": 0.8187,
      "step": 9371
    },
    {
      "epoch": 0.8451999819632953,
      "grad_norm": 1.9523843810641746,
      "learning_rate": 2.460374241260039e-07,
      "loss": 0.6747,
      "step": 9372
    },
    {
      "epoch": 0.8452901654867656,
      "grad_norm": 2.1627320975406135,
      "learning_rate": 2.4575677142331884e-07,
      "loss": 0.811,
      "step": 9373
    },
    {
      "epoch": 0.8453803490102358,
      "grad_norm": 1.9115521843622367,
      "learning_rate": 2.454762684023395e-07,
      "loss": 0.6841,
      "step": 9374
    },
    {
      "epoch": 0.8454705325337061,
      "grad_norm": 3.7412044372741105,
      "learning_rate": 2.4519591508699823e-07,
      "loss": 0.6903,
      "step": 9375
    },
    {
      "epoch": 0.8455607160571763,
      "grad_norm": 1.839886642571609,
      "learning_rate": 2.4491571150121815e-07,
      "loss": 0.789,
      "step": 9376
    },
    {
      "epoch": 0.8456508995806467,
      "grad_norm": 3.1888395000654604,
      "learning_rate": 2.446356576689062e-07,
      "loss": 0.7379,
      "step": 9377
    },
    {
      "epoch": 0.8457410831041169,
      "grad_norm": 1.9126360502034148,
      "learning_rate": 2.4435575361395976e-07,
      "loss": 0.7542,
      "step": 9378
    },
    {
      "epoch": 0.8458312666275871,
      "grad_norm": 1.741141171857562,
      "learning_rate": 2.440759993602606e-07,
      "loss": 0.7194,
      "step": 9379
    },
    {
      "epoch": 0.8459214501510574,
      "grad_norm": 1.6927305327695943,
      "learning_rate": 2.437963949316793e-07,
      "loss": 0.6902,
      "step": 9380
    },
    {
      "epoch": 0.8460116336745277,
      "grad_norm": 2.022668876623272,
      "learning_rate": 2.435169403520729e-07,
      "loss": 0.7315,
      "step": 9381
    },
    {
      "epoch": 0.8461018171979979,
      "grad_norm": 1.8043115397631815,
      "learning_rate": 2.4323763564528653e-07,
      "loss": 0.7864,
      "step": 9382
    },
    {
      "epoch": 0.8461920007214682,
      "grad_norm": 1.6393854454018986,
      "learning_rate": 2.429584808351517e-07,
      "loss": 0.7592,
      "step": 9383
    },
    {
      "epoch": 0.8462821842449384,
      "grad_norm": 2.1482643751269195,
      "learning_rate": 2.42679475945488e-07,
      "loss": 0.8616,
      "step": 9384
    },
    {
      "epoch": 0.8463723677684087,
      "grad_norm": 1.3767278392276534,
      "learning_rate": 2.424006210001008e-07,
      "loss": 0.74,
      "step": 9385
    },
    {
      "epoch": 0.846462551291879,
      "grad_norm": 2.1353170480209367,
      "learning_rate": 2.421219160227839e-07,
      "loss": 0.769,
      "step": 9386
    },
    {
      "epoch": 0.8465527348153492,
      "grad_norm": 1.5077942367791692,
      "learning_rate": 2.4184336103731785e-07,
      "loss": 0.7357,
      "step": 9387
    },
    {
      "epoch": 0.8466429183388196,
      "grad_norm": 1.516306615717498,
      "learning_rate": 2.4156495606747065e-07,
      "loss": 0.7611,
      "step": 9388
    },
    {
      "epoch": 0.8467331018622898,
      "grad_norm": 1.9195165662316436,
      "learning_rate": 2.412867011369972e-07,
      "loss": 0.8571,
      "step": 9389
    },
    {
      "epoch": 0.84682328538576,
      "grad_norm": 3.043068097002294,
      "learning_rate": 2.4100859626963997e-07,
      "loss": 0.7524,
      "step": 9390
    },
    {
      "epoch": 0.8469134689092302,
      "grad_norm": 1.384208190374749,
      "learning_rate": 2.407306414891288e-07,
      "loss": 0.7386,
      "step": 9391
    },
    {
      "epoch": 0.8470036524327006,
      "grad_norm": 1.9430663841635987,
      "learning_rate": 2.4045283681917893e-07,
      "loss": 0.8436,
      "step": 9392
    },
    {
      "epoch": 0.8470938359561708,
      "grad_norm": 1.7600718580011894,
      "learning_rate": 2.4017518228349586e-07,
      "loss": 0.8555,
      "step": 9393
    },
    {
      "epoch": 0.8471840194796411,
      "grad_norm": 0.6109632189086028,
      "learning_rate": 2.3989767790576887e-07,
      "loss": 0.5921,
      "step": 9394
    },
    {
      "epoch": 0.8472742030031113,
      "grad_norm": 1.9329453739513203,
      "learning_rate": 2.396203237096781e-07,
      "loss": 0.7107,
      "step": 9395
    },
    {
      "epoch": 0.8473643865265816,
      "grad_norm": 1.8386194646727236,
      "learning_rate": 2.393431197188873e-07,
      "loss": 0.7346,
      "step": 9396
    },
    {
      "epoch": 0.8474545700500519,
      "grad_norm": 2.626936697642642,
      "learning_rate": 2.3906606595705004e-07,
      "loss": 0.8105,
      "step": 9397
    },
    {
      "epoch": 0.8475447535735221,
      "grad_norm": 2.2599478263233994,
      "learning_rate": 2.387891624478056e-07,
      "loss": 0.7935,
      "step": 9398
    },
    {
      "epoch": 0.8476349370969923,
      "grad_norm": 1.5846739748357093,
      "learning_rate": 2.3851240921478075e-07,
      "loss": 0.7614,
      "step": 9399
    },
    {
      "epoch": 0.8477251206204627,
      "grad_norm": 1.6295903002242873,
      "learning_rate": 2.3823580628159057e-07,
      "loss": 0.7248,
      "step": 9400
    },
    {
      "epoch": 0.8478153041439329,
      "grad_norm": 2.089432663459882,
      "learning_rate": 2.3795935367183517e-07,
      "loss": 0.7257,
      "step": 9401
    },
    {
      "epoch": 0.8479054876674031,
      "grad_norm": 1.539761457174851,
      "learning_rate": 2.376830514091035e-07,
      "loss": 0.7621,
      "step": 9402
    },
    {
      "epoch": 0.8479956711908734,
      "grad_norm": 1.6138787075399306,
      "learning_rate": 2.3740689951697135e-07,
      "loss": 0.8238,
      "step": 9403
    },
    {
      "epoch": 0.8480858547143437,
      "grad_norm": 2.0399090993487814,
      "learning_rate": 2.371308980190012e-07,
      "loss": 0.7412,
      "step": 9404
    },
    {
      "epoch": 0.848176038237814,
      "grad_norm": 0.634334265054081,
      "learning_rate": 2.3685504693874337e-07,
      "loss": 0.598,
      "step": 9405
    },
    {
      "epoch": 0.8482662217612842,
      "grad_norm": 3.9825161057007072,
      "learning_rate": 2.3657934629973497e-07,
      "loss": 0.7779,
      "step": 9406
    },
    {
      "epoch": 0.8483564052847544,
      "grad_norm": 1.9266880832898556,
      "learning_rate": 2.3630379612549944e-07,
      "loss": 0.7472,
      "step": 9407
    },
    {
      "epoch": 0.8484465888082248,
      "grad_norm": 0.7355881268667882,
      "learning_rate": 2.3602839643954997e-07,
      "loss": 0.6217,
      "step": 9408
    },
    {
      "epoch": 0.848536772331695,
      "grad_norm": 1.3924968422900135,
      "learning_rate": 2.3575314726538308e-07,
      "loss": 0.7451,
      "step": 9409
    },
    {
      "epoch": 0.8486269558551652,
      "grad_norm": 1.7149822900374838,
      "learning_rate": 2.3547804862648645e-07,
      "loss": 0.7814,
      "step": 9410
    },
    {
      "epoch": 0.8487171393786356,
      "grad_norm": 1.5580492335266616,
      "learning_rate": 2.3520310054633174e-07,
      "loss": 0.7927,
      "step": 9411
    },
    {
      "epoch": 0.8488073229021058,
      "grad_norm": 1.7802017875478127,
      "learning_rate": 2.3492830304837973e-07,
      "loss": 0.7063,
      "step": 9412
    },
    {
      "epoch": 0.848897506425576,
      "grad_norm": 1.8297280350258134,
      "learning_rate": 2.3465365615607723e-07,
      "loss": 0.6889,
      "step": 9413
    },
    {
      "epoch": 0.8489876899490463,
      "grad_norm": 3.3992318463714333,
      "learning_rate": 2.3437915989285884e-07,
      "loss": 0.7364,
      "step": 9414
    },
    {
      "epoch": 0.8490778734725166,
      "grad_norm": 2.517032111289301,
      "learning_rate": 2.3410481428214647e-07,
      "loss": 0.8081,
      "step": 9415
    },
    {
      "epoch": 0.8491680569959869,
      "grad_norm": 1.8413518406400344,
      "learning_rate": 2.338306193473476e-07,
      "loss": 0.771,
      "step": 9416
    },
    {
      "epoch": 0.8492582405194571,
      "grad_norm": 2.2026481382885366,
      "learning_rate": 2.3355657511185957e-07,
      "loss": 0.7422,
      "step": 9417
    },
    {
      "epoch": 0.8493484240429273,
      "grad_norm": 0.6590915423198875,
      "learning_rate": 2.3328268159906428e-07,
      "loss": 0.5627,
      "step": 9418
    },
    {
      "epoch": 0.8494386075663977,
      "grad_norm": 2.207401318877147,
      "learning_rate": 2.330089388323322e-07,
      "loss": 0.793,
      "step": 9419
    },
    {
      "epoch": 0.8495287910898679,
      "grad_norm": 1.8090994274214198,
      "learning_rate": 2.327353468350204e-07,
      "loss": 0.7791,
      "step": 9420
    },
    {
      "epoch": 0.8496189746133381,
      "grad_norm": 1.6078225212415087,
      "learning_rate": 2.3246190563047352e-07,
      "loss": 0.7376,
      "step": 9421
    },
    {
      "epoch": 0.8497091581368084,
      "grad_norm": 1.8986521783960848,
      "learning_rate": 2.3218861524202293e-07,
      "loss": 0.726,
      "step": 9422
    },
    {
      "epoch": 0.8497993416602787,
      "grad_norm": 6.229653465956851,
      "learning_rate": 2.3191547569298775e-07,
      "loss": 0.7527,
      "step": 9423
    },
    {
      "epoch": 0.8498895251837489,
      "grad_norm": 1.8221857476589913,
      "learning_rate": 2.3164248700667245e-07,
      "loss": 0.7351,
      "step": 9424
    },
    {
      "epoch": 0.8499797087072192,
      "grad_norm": 2.6674855186977413,
      "learning_rate": 2.313696492063717e-07,
      "loss": 0.7773,
      "step": 9425
    },
    {
      "epoch": 0.8500698922306894,
      "grad_norm": 1.953569529490869,
      "learning_rate": 2.3109696231536401e-07,
      "loss": 0.7309,
      "step": 9426
    },
    {
      "epoch": 0.8501600757541597,
      "grad_norm": 1.6853868888170271,
      "learning_rate": 2.3082442635691722e-07,
      "loss": 0.7041,
      "step": 9427
    },
    {
      "epoch": 0.85025025927763,
      "grad_norm": 1.8668234242740294,
      "learning_rate": 2.305520413542854e-07,
      "loss": 0.7538,
      "step": 9428
    },
    {
      "epoch": 0.8503404428011002,
      "grad_norm": 1.2842897317245523,
      "learning_rate": 2.3027980733071018e-07,
      "loss": 0.6647,
      "step": 9429
    },
    {
      "epoch": 0.8504306263245704,
      "grad_norm": 0.5984034635660331,
      "learning_rate": 2.3000772430942027e-07,
      "loss": 0.5448,
      "step": 9430
    },
    {
      "epoch": 0.8505208098480408,
      "grad_norm": 1.632178662674417,
      "learning_rate": 2.2973579231363028e-07,
      "loss": 0.7589,
      "step": 9431
    },
    {
      "epoch": 0.850610993371511,
      "grad_norm": 1.6543449385865774,
      "learning_rate": 2.2946401136654446e-07,
      "loss": 0.7683,
      "step": 9432
    },
    {
      "epoch": 0.8507011768949813,
      "grad_norm": 1.5110072943298674,
      "learning_rate": 2.2919238149135077e-07,
      "loss": 0.7717,
      "step": 9433
    },
    {
      "epoch": 0.8507913604184516,
      "grad_norm": 1.4625740880214104,
      "learning_rate": 2.289209027112282e-07,
      "loss": 0.7462,
      "step": 9434
    },
    {
      "epoch": 0.8508815439419218,
      "grad_norm": 2.1816559247169427,
      "learning_rate": 2.2864957504933934e-07,
      "loss": 0.8046,
      "step": 9435
    },
    {
      "epoch": 0.8509717274653921,
      "grad_norm": 2.765295752630778,
      "learning_rate": 2.2837839852883589e-07,
      "loss": 0.7251,
      "step": 9436
    },
    {
      "epoch": 0.8510619109888623,
      "grad_norm": 1.8785823296951212,
      "learning_rate": 2.2810737317285623e-07,
      "loss": 0.7426,
      "step": 9437
    },
    {
      "epoch": 0.8511520945123326,
      "grad_norm": 1.4158247615172286,
      "learning_rate": 2.278364990045254e-07,
      "loss": 0.684,
      "step": 9438
    },
    {
      "epoch": 0.8512422780358029,
      "grad_norm": 2.940119772307362,
      "learning_rate": 2.2756577604695625e-07,
      "loss": 0.7357,
      "step": 9439
    },
    {
      "epoch": 0.8513324615592731,
      "grad_norm": 2.064269838163654,
      "learning_rate": 2.2729520432324855e-07,
      "loss": 0.7638,
      "step": 9440
    },
    {
      "epoch": 0.8514226450827433,
      "grad_norm": 1.5329329137139,
      "learning_rate": 2.2702478385648826e-07,
      "loss": 0.6625,
      "step": 9441
    },
    {
      "epoch": 0.8515128286062137,
      "grad_norm": 1.7051107746507135,
      "learning_rate": 2.2675451466974938e-07,
      "loss": 0.7403,
      "step": 9442
    },
    {
      "epoch": 0.8516030121296839,
      "grad_norm": 1.5982479874329565,
      "learning_rate": 2.26484396786093e-07,
      "loss": 0.6793,
      "step": 9443
    },
    {
      "epoch": 0.8516931956531542,
      "grad_norm": 0.6925774907229166,
      "learning_rate": 2.2621443022856667e-07,
      "loss": 0.6032,
      "step": 9444
    },
    {
      "epoch": 0.8517833791766244,
      "grad_norm": 2.0761647551765265,
      "learning_rate": 2.2594461502020646e-07,
      "loss": 0.7836,
      "step": 9445
    },
    {
      "epoch": 0.8518735627000947,
      "grad_norm": 2.213467824863633,
      "learning_rate": 2.2567495118403278e-07,
      "loss": 0.6216,
      "step": 9446
    },
    {
      "epoch": 0.851963746223565,
      "grad_norm": 2.709681767011193,
      "learning_rate": 2.254054387430566e-07,
      "loss": 0.7569,
      "step": 9447
    },
    {
      "epoch": 0.8520539297470352,
      "grad_norm": 2.030321349091964,
      "learning_rate": 2.2513607772027243e-07,
      "loss": 0.7294,
      "step": 9448
    },
    {
      "epoch": 0.8521441132705054,
      "grad_norm": 2.0394875971217568,
      "learning_rate": 2.2486686813866562e-07,
      "loss": 0.7079,
      "step": 9449
    },
    {
      "epoch": 0.8522342967939758,
      "grad_norm": 2.1106929476449863,
      "learning_rate": 2.2459781002120514e-07,
      "loss": 0.799,
      "step": 9450
    },
    {
      "epoch": 0.852324480317446,
      "grad_norm": 1.518860421267808,
      "learning_rate": 2.243289033908491e-07,
      "loss": 0.8103,
      "step": 9451
    },
    {
      "epoch": 0.8524146638409162,
      "grad_norm": 2.322547632392116,
      "learning_rate": 2.2406014827054176e-07,
      "loss": 0.7597,
      "step": 9452
    },
    {
      "epoch": 0.8525048473643865,
      "grad_norm": 4.586899820508282,
      "learning_rate": 2.2379154468321525e-07,
      "loss": 0.8079,
      "step": 9453
    },
    {
      "epoch": 0.8525950308878568,
      "grad_norm": 1.4304846537199196,
      "learning_rate": 2.2352309265178793e-07,
      "loss": 0.7794,
      "step": 9454
    },
    {
      "epoch": 0.852685214411327,
      "grad_norm": 2.466493565267772,
      "learning_rate": 2.2325479219916565e-07,
      "loss": 0.8106,
      "step": 9455
    },
    {
      "epoch": 0.8527753979347973,
      "grad_norm": 1.622196640432757,
      "learning_rate": 2.229866433482419e-07,
      "loss": 0.7758,
      "step": 9456
    },
    {
      "epoch": 0.8528655814582676,
      "grad_norm": 1.8522940691363614,
      "learning_rate": 2.2271864612189552e-07,
      "loss": 0.7958,
      "step": 9457
    },
    {
      "epoch": 0.8529557649817379,
      "grad_norm": 1.7680368744944612,
      "learning_rate": 2.2245080054299415e-07,
      "loss": 0.7761,
      "step": 9458
    },
    {
      "epoch": 0.8530459485052081,
      "grad_norm": 1.53198567059541,
      "learning_rate": 2.2218310663439198e-07,
      "loss": 0.6977,
      "step": 9459
    },
    {
      "epoch": 0.8531361320286783,
      "grad_norm": 2.860537100023412,
      "learning_rate": 2.2191556441892968e-07,
      "loss": 0.7665,
      "step": 9460
    },
    {
      "epoch": 0.8532263155521487,
      "grad_norm": 1.5179202603592463,
      "learning_rate": 2.216481739194358e-07,
      "loss": 0.7473,
      "step": 9461
    },
    {
      "epoch": 0.8533164990756189,
      "grad_norm": 3.137099227537364,
      "learning_rate": 2.2138093515872592e-07,
      "loss": 0.6914,
      "step": 9462
    },
    {
      "epoch": 0.8534066825990891,
      "grad_norm": 1.3629227743514396,
      "learning_rate": 2.2111384815960087e-07,
      "loss": 0.8273,
      "step": 9463
    },
    {
      "epoch": 0.8534968661225594,
      "grad_norm": 1.5409272027121292,
      "learning_rate": 2.208469129448518e-07,
      "loss": 0.667,
      "step": 9464
    },
    {
      "epoch": 0.8535870496460297,
      "grad_norm": 0.6303877351840946,
      "learning_rate": 2.2058012953725357e-07,
      "loss": 0.5688,
      "step": 9465
    },
    {
      "epoch": 0.8536772331695,
      "grad_norm": 1.7400241986049956,
      "learning_rate": 2.203134979595711e-07,
      "loss": 0.7139,
      "step": 9466
    },
    {
      "epoch": 0.8537674166929702,
      "grad_norm": 1.564594619581589,
      "learning_rate": 2.2004701823455374e-07,
      "loss": 0.6995,
      "step": 9467
    },
    {
      "epoch": 0.8538576002164404,
      "grad_norm": 1.7550435955795067,
      "learning_rate": 2.1978069038493906e-07,
      "loss": 0.7725,
      "step": 9468
    },
    {
      "epoch": 0.8539477837399108,
      "grad_norm": 1.9004108863056275,
      "learning_rate": 2.1951451443345225e-07,
      "loss": 0.7469,
      "step": 9469
    },
    {
      "epoch": 0.854037967263381,
      "grad_norm": 1.6914319043053005,
      "learning_rate": 2.1924849040280425e-07,
      "loss": 0.7308,
      "step": 9470
    },
    {
      "epoch": 0.8541281507868512,
      "grad_norm": 3.019489243623563,
      "learning_rate": 2.1898261831569465e-07,
      "loss": 0.8269,
      "step": 9471
    },
    {
      "epoch": 0.8542183343103215,
      "grad_norm": 3.0843453851300358,
      "learning_rate": 2.1871689819480798e-07,
      "loss": 0.7501,
      "step": 9472
    },
    {
      "epoch": 0.8543085178337918,
      "grad_norm": 1.3781889868173758,
      "learning_rate": 2.1845133006281745e-07,
      "loss": 0.7299,
      "step": 9473
    },
    {
      "epoch": 0.854398701357262,
      "grad_norm": 1.546704292511109,
      "learning_rate": 2.1818591394238294e-07,
      "loss": 0.6021,
      "step": 9474
    },
    {
      "epoch": 0.8544888848807323,
      "grad_norm": 1.6897191891063017,
      "learning_rate": 2.1792064985615076e-07,
      "loss": 0.7256,
      "step": 9475
    },
    {
      "epoch": 0.8545790684042025,
      "grad_norm": 1.5773066296680398,
      "learning_rate": 2.1765553782675528e-07,
      "loss": 0.7146,
      "step": 9476
    },
    {
      "epoch": 0.8546692519276728,
      "grad_norm": 1.4765675862542607,
      "learning_rate": 2.1739057787681703e-07,
      "loss": 0.8183,
      "step": 9477
    },
    {
      "epoch": 0.8547594354511431,
      "grad_norm": 2.1032448313613403,
      "learning_rate": 2.1712577002894372e-07,
      "loss": 0.7709,
      "step": 9478
    },
    {
      "epoch": 0.8548496189746133,
      "grad_norm": 1.5801676758992729,
      "learning_rate": 2.1686111430573105e-07,
      "loss": 0.7166,
      "step": 9479
    },
    {
      "epoch": 0.8549398024980835,
      "grad_norm": 1.4922940680511907,
      "learning_rate": 2.165966107297592e-07,
      "loss": 0.7979,
      "step": 9480
    },
    {
      "epoch": 0.8550299860215539,
      "grad_norm": 4.53787084487974,
      "learning_rate": 2.16332259323599e-07,
      "loss": 0.7719,
      "step": 9481
    },
    {
      "epoch": 0.8551201695450241,
      "grad_norm": 1.8486174649802036,
      "learning_rate": 2.1606806010980504e-07,
      "loss": 0.6939,
      "step": 9482
    },
    {
      "epoch": 0.8552103530684944,
      "grad_norm": 2.1335970934098207,
      "learning_rate": 2.1580401311092067e-07,
      "loss": 0.7381,
      "step": 9483
    },
    {
      "epoch": 0.8553005365919647,
      "grad_norm": 1.496497834219036,
      "learning_rate": 2.1554011834947604e-07,
      "loss": 0.7336,
      "step": 9484
    },
    {
      "epoch": 0.8553907201154349,
      "grad_norm": 1.8884883187851569,
      "learning_rate": 2.1527637584798764e-07,
      "loss": 0.7733,
      "step": 9485
    },
    {
      "epoch": 0.8554809036389052,
      "grad_norm": 1.7885640878518023,
      "learning_rate": 2.150127856289603e-07,
      "loss": 0.8098,
      "step": 9486
    },
    {
      "epoch": 0.8555710871623754,
      "grad_norm": 0.6207392178527381,
      "learning_rate": 2.1474934771488363e-07,
      "loss": 0.5572,
      "step": 9487
    },
    {
      "epoch": 0.8556612706858457,
      "grad_norm": 1.5424276827733914,
      "learning_rate": 2.1448606212823715e-07,
      "loss": 0.7912,
      "step": 9488
    },
    {
      "epoch": 0.855751454209316,
      "grad_norm": 1.672651378013361,
      "learning_rate": 2.1422292889148452e-07,
      "loss": 0.7287,
      "step": 9489
    },
    {
      "epoch": 0.8558416377327862,
      "grad_norm": 1.594767002917467,
      "learning_rate": 2.139599480270784e-07,
      "loss": 0.7625,
      "step": 9490
    },
    {
      "epoch": 0.8559318212562564,
      "grad_norm": 3.5225623285154755,
      "learning_rate": 2.1369711955745773e-07,
      "loss": 0.7344,
      "step": 9491
    },
    {
      "epoch": 0.8560220047797268,
      "grad_norm": 2.4228510704793433,
      "learning_rate": 2.1343444350504813e-07,
      "loss": 0.7002,
      "step": 9492
    },
    {
      "epoch": 0.856112188303197,
      "grad_norm": 1.5468876385413302,
      "learning_rate": 2.1317191989226302e-07,
      "loss": 0.7466,
      "step": 9493
    },
    {
      "epoch": 0.8562023718266673,
      "grad_norm": 2.3289478051318198,
      "learning_rate": 2.129095487415027e-07,
      "loss": 0.73,
      "step": 9494
    },
    {
      "epoch": 0.8562925553501375,
      "grad_norm": 2.0543029193169065,
      "learning_rate": 2.1264733007515257e-07,
      "loss": 0.7568,
      "step": 9495
    },
    {
      "epoch": 0.8563827388736078,
      "grad_norm": 1.944975858308078,
      "learning_rate": 2.1238526391558852e-07,
      "loss": 0.7263,
      "step": 9496
    },
    {
      "epoch": 0.8564729223970781,
      "grad_norm": 2.079810813956545,
      "learning_rate": 2.1212335028517003e-07,
      "loss": 0.882,
      "step": 9497
    },
    {
      "epoch": 0.8565631059205483,
      "grad_norm": 1.9501171296961375,
      "learning_rate": 2.1186158920624563e-07,
      "loss": 0.7512,
      "step": 9498
    },
    {
      "epoch": 0.8566532894440185,
      "grad_norm": 1.6482067941065273,
      "learning_rate": 2.1159998070115015e-07,
      "loss": 0.7452,
      "step": 9499
    },
    {
      "epoch": 0.8567434729674889,
      "grad_norm": 1.778375615323674,
      "learning_rate": 2.113385247922055e-07,
      "loss": 0.727,
      "step": 9500
    },
    {
      "epoch": 0.8568336564909591,
      "grad_norm": 2.147693852122851,
      "learning_rate": 2.1107722150172068e-07,
      "loss": 0.7518,
      "step": 9501
    },
    {
      "epoch": 0.8569238400144293,
      "grad_norm": 4.973837880645031,
      "learning_rate": 2.108160708519906e-07,
      "loss": 0.7813,
      "step": 9502
    },
    {
      "epoch": 0.8570140235378996,
      "grad_norm": 2.088633762780262,
      "learning_rate": 2.1055507286529984e-07,
      "loss": 0.7263,
      "step": 9503
    },
    {
      "epoch": 0.8571042070613699,
      "grad_norm": 1.6757613316488162,
      "learning_rate": 2.1029422756391612e-07,
      "loss": 0.7653,
      "step": 9504
    },
    {
      "epoch": 0.8571943905848401,
      "grad_norm": 1.5391759685379167,
      "learning_rate": 2.1003353497009812e-07,
      "loss": 0.6586,
      "step": 9505
    },
    {
      "epoch": 0.8572845741083104,
      "grad_norm": 1.6006813416550414,
      "learning_rate": 2.0977299510608825e-07,
      "loss": 0.6686,
      "step": 9506
    },
    {
      "epoch": 0.8573747576317807,
      "grad_norm": 1.9097676574234301,
      "learning_rate": 2.0951260799411784e-07,
      "loss": 0.7761,
      "step": 9507
    },
    {
      "epoch": 0.857464941155251,
      "grad_norm": 1.9755452670909466,
      "learning_rate": 2.0925237365640424e-07,
      "loss": 0.737,
      "step": 9508
    },
    {
      "epoch": 0.8575551246787212,
      "grad_norm": 0.6519577574751045,
      "learning_rate": 2.0899229211515211e-07,
      "loss": 0.5761,
      "step": 9509
    },
    {
      "epoch": 0.8576453082021914,
      "grad_norm": 15.564333888087946,
      "learning_rate": 2.0873236339255306e-07,
      "loss": 0.6725,
      "step": 9510
    },
    {
      "epoch": 0.8577354917256618,
      "grad_norm": 1.490516086014249,
      "learning_rate": 2.0847258751078644e-07,
      "loss": 0.6909,
      "step": 9511
    },
    {
      "epoch": 0.857825675249132,
      "grad_norm": 4.01094916965762,
      "learning_rate": 2.082129644920163e-07,
      "loss": 0.7208,
      "step": 9512
    },
    {
      "epoch": 0.8579158587726022,
      "grad_norm": 1.6207536823278537,
      "learning_rate": 2.0795349435839605e-07,
      "loss": 0.7546,
      "step": 9513
    },
    {
      "epoch": 0.8580060422960725,
      "grad_norm": 2.3506815835014616,
      "learning_rate": 2.0769417713206484e-07,
      "loss": 0.7167,
      "step": 9514
    },
    {
      "epoch": 0.8580962258195428,
      "grad_norm": 1.8406527091306177,
      "learning_rate": 2.074350128351492e-07,
      "loss": 0.806,
      "step": 9515
    },
    {
      "epoch": 0.858186409343013,
      "grad_norm": 1.479248510254559,
      "learning_rate": 2.0717600148976256e-07,
      "loss": 0.6651,
      "step": 9516
    },
    {
      "epoch": 0.8582765928664833,
      "grad_norm": 3.2389542601589567,
      "learning_rate": 2.0691714311800436e-07,
      "loss": 0.6828,
      "step": 9517
    },
    {
      "epoch": 0.8583667763899535,
      "grad_norm": 2.9497158818900697,
      "learning_rate": 2.066584377419631e-07,
      "loss": 0.8022,
      "step": 9518
    },
    {
      "epoch": 0.8584569599134239,
      "grad_norm": 1.6220745015420397,
      "learning_rate": 2.0639988538371167e-07,
      "loss": 0.7651,
      "step": 9519
    },
    {
      "epoch": 0.8585471434368941,
      "grad_norm": 2.6408064740364803,
      "learning_rate": 2.0614148606531258e-07,
      "loss": 0.7117,
      "step": 9520
    },
    {
      "epoch": 0.8586373269603643,
      "grad_norm": 2.466512221149301,
      "learning_rate": 2.0588323980881285e-07,
      "loss": 0.778,
      "step": 9521
    },
    {
      "epoch": 0.8587275104838346,
      "grad_norm": 1.69444308862823,
      "learning_rate": 2.0562514663624752e-07,
      "loss": 0.7315,
      "step": 9522
    },
    {
      "epoch": 0.8588176940073049,
      "grad_norm": 1.4594202169401544,
      "learning_rate": 2.0536720656963902e-07,
      "loss": 0.8104,
      "step": 9523
    },
    {
      "epoch": 0.8589078775307751,
      "grad_norm": 2.050361168842937,
      "learning_rate": 2.051094196309957e-07,
      "loss": 0.7692,
      "step": 9524
    },
    {
      "epoch": 0.8589980610542454,
      "grad_norm": 1.5379771941965767,
      "learning_rate": 2.0485178584231378e-07,
      "loss": 0.7041,
      "step": 9525
    },
    {
      "epoch": 0.8590882445777156,
      "grad_norm": 0.6710979270583021,
      "learning_rate": 2.0459430522557587e-07,
      "loss": 0.5782,
      "step": 9526
    },
    {
      "epoch": 0.8591784281011859,
      "grad_norm": 3.164995974135889,
      "learning_rate": 2.0433697780275195e-07,
      "loss": 0.7478,
      "step": 9527
    },
    {
      "epoch": 0.8592686116246562,
      "grad_norm": 1.624872716173565,
      "learning_rate": 2.040798035957978e-07,
      "loss": 0.7626,
      "step": 9528
    },
    {
      "epoch": 0.8593587951481264,
      "grad_norm": 0.7312720466615948,
      "learning_rate": 2.038227826266574e-07,
      "loss": 0.5936,
      "step": 9529
    },
    {
      "epoch": 0.8594489786715968,
      "grad_norm": 4.0396775269933025,
      "learning_rate": 2.0356591491726126e-07,
      "loss": 0.752,
      "step": 9530
    },
    {
      "epoch": 0.859539162195067,
      "grad_norm": 1.433735687131205,
      "learning_rate": 2.033092004895267e-07,
      "loss": 0.7846,
      "step": 9531
    },
    {
      "epoch": 0.8596293457185372,
      "grad_norm": 1.689927368634814,
      "learning_rate": 2.03052639365358e-07,
      "loss": 0.7649,
      "step": 9532
    },
    {
      "epoch": 0.8597195292420075,
      "grad_norm": 1.7168354815589484,
      "learning_rate": 2.0279623156664694e-07,
      "loss": 0.7648,
      "step": 9533
    },
    {
      "epoch": 0.8598097127654778,
      "grad_norm": 1.6062650538826353,
      "learning_rate": 2.0253997711527005e-07,
      "loss": 0.8061,
      "step": 9534
    },
    {
      "epoch": 0.859899896288948,
      "grad_norm": 1.9741729646482713,
      "learning_rate": 2.0228387603309428e-07,
      "loss": 0.7917,
      "step": 9535
    },
    {
      "epoch": 0.8599900798124183,
      "grad_norm": 1.521451937485175,
      "learning_rate": 2.0202792834197035e-07,
      "loss": 0.7654,
      "step": 9536
    },
    {
      "epoch": 0.8600802633358885,
      "grad_norm": 1.8987412700238924,
      "learning_rate": 2.017721340637375e-07,
      "loss": 0.6742,
      "step": 9537
    },
    {
      "epoch": 0.8601704468593588,
      "grad_norm": 1.6969252970566777,
      "learning_rate": 2.0151649322022134e-07,
      "loss": 0.7332,
      "step": 9538
    },
    {
      "epoch": 0.8602606303828291,
      "grad_norm": 1.9114402402875565,
      "learning_rate": 2.012610058332349e-07,
      "loss": 0.5579,
      "step": 9539
    },
    {
      "epoch": 0.8603508139062993,
      "grad_norm": 0.6774845157416305,
      "learning_rate": 2.010056719245774e-07,
      "loss": 0.5741,
      "step": 9540
    },
    {
      "epoch": 0.8604409974297695,
      "grad_norm": 1.57436027553286,
      "learning_rate": 2.0075049151603563e-07,
      "loss": 0.7671,
      "step": 9541
    },
    {
      "epoch": 0.8605311809532399,
      "grad_norm": 1.5576528515215677,
      "learning_rate": 2.0049546462938326e-07,
      "loss": 0.7518,
      "step": 9542
    },
    {
      "epoch": 0.8606213644767101,
      "grad_norm": 1.4836734921045487,
      "learning_rate": 2.0024059128637961e-07,
      "loss": 0.7652,
      "step": 9543
    },
    {
      "epoch": 0.8607115480001803,
      "grad_norm": 1.472188860237337,
      "learning_rate": 1.99985871508773e-07,
      "loss": 0.6123,
      "step": 9544
    },
    {
      "epoch": 0.8608017315236506,
      "grad_norm": 1.795163741227692,
      "learning_rate": 1.9973130531829674e-07,
      "loss": 0.7714,
      "step": 9545
    },
    {
      "epoch": 0.8608919150471209,
      "grad_norm": 1.6970331978863564,
      "learning_rate": 1.994768927366721e-07,
      "loss": 0.6977,
      "step": 9546
    },
    {
      "epoch": 0.8609820985705912,
      "grad_norm": 7.329449174072531,
      "learning_rate": 1.992226337856069e-07,
      "loss": 0.7147,
      "step": 9547
    },
    {
      "epoch": 0.8610722820940614,
      "grad_norm": 1.5726354318104048,
      "learning_rate": 1.9896852848679592e-07,
      "loss": 0.7,
      "step": 9548
    },
    {
      "epoch": 0.8611624656175316,
      "grad_norm": 2.2087526343199864,
      "learning_rate": 1.9871457686192094e-07,
      "loss": 0.7554,
      "step": 9549
    },
    {
      "epoch": 0.861252649141002,
      "grad_norm": 1.8128015990735886,
      "learning_rate": 1.984607789326509e-07,
      "loss": 0.7418,
      "step": 9550
    },
    {
      "epoch": 0.8613428326644722,
      "grad_norm": 3.5602002668307025,
      "learning_rate": 1.982071347206402e-07,
      "loss": 0.6941,
      "step": 9551
    },
    {
      "epoch": 0.8614330161879424,
      "grad_norm": 1.4572267784667612,
      "learning_rate": 1.9795364424753202e-07,
      "loss": 0.7194,
      "step": 9552
    },
    {
      "epoch": 0.8615231997114128,
      "grad_norm": 1.9157282218068592,
      "learning_rate": 1.9770030753495505e-07,
      "loss": 0.7106,
      "step": 9553
    },
    {
      "epoch": 0.861613383234883,
      "grad_norm": 1.918100138901195,
      "learning_rate": 1.9744712460452573e-07,
      "loss": 0.7034,
      "step": 9554
    },
    {
      "epoch": 0.8617035667583532,
      "grad_norm": 1.8361988287457145,
      "learning_rate": 1.9719409547784703e-07,
      "loss": 0.7287,
      "step": 9555
    },
    {
      "epoch": 0.8617937502818235,
      "grad_norm": 2.285464952272335,
      "learning_rate": 1.9694122017650837e-07,
      "loss": 0.7682,
      "step": 9556
    },
    {
      "epoch": 0.8618839338052938,
      "grad_norm": 1.5531488980885861,
      "learning_rate": 1.9668849872208738e-07,
      "loss": 0.7606,
      "step": 9557
    },
    {
      "epoch": 0.8619741173287641,
      "grad_norm": 1.6241184190786933,
      "learning_rate": 1.9643593113614632e-07,
      "loss": 0.591,
      "step": 9558
    },
    {
      "epoch": 0.8620643008522343,
      "grad_norm": 1.4560611553291543,
      "learning_rate": 1.961835174402371e-07,
      "loss": 0.7839,
      "step": 9559
    },
    {
      "epoch": 0.8621544843757045,
      "grad_norm": 1.9716371226522416,
      "learning_rate": 1.9593125765589535e-07,
      "loss": 0.7149,
      "step": 9560
    },
    {
      "epoch": 0.8622446678991749,
      "grad_norm": 1.5763198045706772,
      "learning_rate": 1.9567915180464721e-07,
      "loss": 0.6932,
      "step": 9561
    },
    {
      "epoch": 0.8623348514226451,
      "grad_norm": 1.970969568696003,
      "learning_rate": 1.9542719990800217e-07,
      "loss": 0.7307,
      "step": 9562
    },
    {
      "epoch": 0.8624250349461153,
      "grad_norm": 2.3400120746268414,
      "learning_rate": 1.9517540198745896e-07,
      "loss": 0.8137,
      "step": 9563
    },
    {
      "epoch": 0.8625152184695856,
      "grad_norm": 8.874250353462962,
      "learning_rate": 1.94923758064502e-07,
      "loss": 0.7444,
      "step": 9564
    },
    {
      "epoch": 0.8626054019930559,
      "grad_norm": 1.680035364369157,
      "learning_rate": 1.9467226816060322e-07,
      "loss": 0.7113,
      "step": 9565
    },
    {
      "epoch": 0.8626955855165261,
      "grad_norm": 2.0699918678603093,
      "learning_rate": 1.9442093229722122e-07,
      "loss": 0.6841,
      "step": 9566
    },
    {
      "epoch": 0.8627857690399964,
      "grad_norm": 2.308141467203276,
      "learning_rate": 1.9416975049580085e-07,
      "loss": 0.8247,
      "step": 9567
    },
    {
      "epoch": 0.8628759525634666,
      "grad_norm": 7.375884180893242,
      "learning_rate": 1.9391872277777456e-07,
      "loss": 0.7311,
      "step": 9568
    },
    {
      "epoch": 0.862966136086937,
      "grad_norm": 3.31536626066907,
      "learning_rate": 1.9366784916456158e-07,
      "loss": 0.7318,
      "step": 9569
    },
    {
      "epoch": 0.8630563196104072,
      "grad_norm": 1.5631207568184642,
      "learning_rate": 1.9341712967756774e-07,
      "loss": 0.6978,
      "step": 9570
    },
    {
      "epoch": 0.8631465031338774,
      "grad_norm": 2.053964814127087,
      "learning_rate": 1.9316656433818566e-07,
      "loss": 0.6543,
      "step": 9571
    },
    {
      "epoch": 0.8632366866573477,
      "grad_norm": 1.6832919458771647,
      "learning_rate": 1.929161531677954e-07,
      "loss": 0.6446,
      "step": 9572
    },
    {
      "epoch": 0.863326870180818,
      "grad_norm": 1.6490615458989017,
      "learning_rate": 1.9266589618776251e-07,
      "loss": 0.6989,
      "step": 9573
    },
    {
      "epoch": 0.8634170537042882,
      "grad_norm": 1.9483487743008385,
      "learning_rate": 1.924157934194417e-07,
      "loss": 0.8004,
      "step": 9574
    },
    {
      "epoch": 0.8635072372277585,
      "grad_norm": 2.2201180337337285,
      "learning_rate": 1.9216584488417142e-07,
      "loss": 0.6357,
      "step": 9575
    },
    {
      "epoch": 0.8635974207512288,
      "grad_norm": 2.6053242564705377,
      "learning_rate": 1.919160506032802e-07,
      "loss": 0.7995,
      "step": 9576
    },
    {
      "epoch": 0.863687604274699,
      "grad_norm": 1.5642207017093226,
      "learning_rate": 1.916664105980812e-07,
      "loss": 0.805,
      "step": 9577
    },
    {
      "epoch": 0.8637777877981693,
      "grad_norm": 1.8201597747257754,
      "learning_rate": 1.914169248898747e-07,
      "loss": 0.7816,
      "step": 9578
    },
    {
      "epoch": 0.8638679713216395,
      "grad_norm": 1.269560358740806,
      "learning_rate": 1.9116759349994882e-07,
      "loss": 0.6568,
      "step": 9579
    },
    {
      "epoch": 0.8639581548451098,
      "grad_norm": 1.6039209631808486,
      "learning_rate": 1.9091841644957763e-07,
      "loss": 0.7837,
      "step": 9580
    },
    {
      "epoch": 0.8640483383685801,
      "grad_norm": 1.4520751277886945,
      "learning_rate": 1.9066939376002278e-07,
      "loss": 0.8276,
      "step": 9581
    },
    {
      "epoch": 0.8641385218920503,
      "grad_norm": 2.083647538968717,
      "learning_rate": 1.9042052545253085e-07,
      "loss": 0.6615,
      "step": 9582
    },
    {
      "epoch": 0.8642287054155205,
      "grad_norm": 1.6985873046617967,
      "learning_rate": 1.901718115483384e-07,
      "loss": 0.7224,
      "step": 9583
    },
    {
      "epoch": 0.8643188889389909,
      "grad_norm": 1.8287111304553059,
      "learning_rate": 1.8992325206866598e-07,
      "loss": 0.7274,
      "step": 9584
    },
    {
      "epoch": 0.8644090724624611,
      "grad_norm": 1.3599967849917838,
      "learning_rate": 1.8967484703472225e-07,
      "loss": 0.7191,
      "step": 9585
    },
    {
      "epoch": 0.8644992559859314,
      "grad_norm": 1.3796908818135445,
      "learning_rate": 1.8942659646770288e-07,
      "loss": 0.7904,
      "step": 9586
    },
    {
      "epoch": 0.8645894395094016,
      "grad_norm": 1.877915722045706,
      "learning_rate": 1.8917850038878936e-07,
      "loss": 0.8086,
      "step": 9587
    },
    {
      "epoch": 0.8646796230328719,
      "grad_norm": 1.5409984515143356,
      "learning_rate": 1.8893055881915121e-07,
      "loss": 0.6515,
      "step": 9588
    },
    {
      "epoch": 0.8647698065563422,
      "grad_norm": 2.2127559449169123,
      "learning_rate": 1.886827717799442e-07,
      "loss": 0.6921,
      "step": 9589
    },
    {
      "epoch": 0.8648599900798124,
      "grad_norm": 2.0001144376439752,
      "learning_rate": 1.884351392923096e-07,
      "loss": 0.7596,
      "step": 9590
    },
    {
      "epoch": 0.8649501736032826,
      "grad_norm": 3.0910497113231767,
      "learning_rate": 1.8818766137737896e-07,
      "loss": 0.7805,
      "step": 9591
    },
    {
      "epoch": 0.865040357126753,
      "grad_norm": 18.22003724145695,
      "learning_rate": 1.8794033805626653e-07,
      "loss": 0.7243,
      "step": 9592
    },
    {
      "epoch": 0.8651305406502232,
      "grad_norm": 1.5745552403828484,
      "learning_rate": 1.876931693500763e-07,
      "loss": 0.7376,
      "step": 9593
    },
    {
      "epoch": 0.8652207241736934,
      "grad_norm": 1.7747612268737873,
      "learning_rate": 1.8744615527989783e-07,
      "loss": 0.592,
      "step": 9594
    },
    {
      "epoch": 0.8653109076971637,
      "grad_norm": 2.0017416046793595,
      "learning_rate": 1.871992958668076e-07,
      "loss": 0.8334,
      "step": 9595
    },
    {
      "epoch": 0.865401091220634,
      "grad_norm": 2.5188383820506144,
      "learning_rate": 1.8695259113186944e-07,
      "loss": 0.789,
      "step": 9596
    },
    {
      "epoch": 0.8654912747441043,
      "grad_norm": 2.0461225218496,
      "learning_rate": 1.8670604109613252e-07,
      "loss": 0.7074,
      "step": 9597
    },
    {
      "epoch": 0.8655814582675745,
      "grad_norm": 1.5564019799228959,
      "learning_rate": 1.8645964578063533e-07,
      "loss": 0.7892,
      "step": 9598
    },
    {
      "epoch": 0.8656716417910447,
      "grad_norm": 1.5296517419580886,
      "learning_rate": 1.862134052064006e-07,
      "loss": 0.7994,
      "step": 9599
    },
    {
      "epoch": 0.8657618253145151,
      "grad_norm": 18.91748036162379,
      "learning_rate": 1.8596731939443932e-07,
      "loss": 0.7861,
      "step": 9600
    },
    {
      "epoch": 0.8658520088379853,
      "grad_norm": 1.6493068048126664,
      "learning_rate": 1.857213883657487e-07,
      "loss": 0.7741,
      "step": 9601
    },
    {
      "epoch": 0.8659421923614555,
      "grad_norm": 0.6426544839773066,
      "learning_rate": 1.8547561214131303e-07,
      "loss": 0.5507,
      "step": 9602
    },
    {
      "epoch": 0.8660323758849259,
      "grad_norm": 121.27577045725168,
      "learning_rate": 1.8522999074210355e-07,
      "loss": 0.8205,
      "step": 9603
    },
    {
      "epoch": 0.8661225594083961,
      "grad_norm": 2.4428632371232095,
      "learning_rate": 1.849845241890775e-07,
      "loss": 0.782,
      "step": 9604
    },
    {
      "epoch": 0.8662127429318663,
      "grad_norm": 1.6560970451695787,
      "learning_rate": 1.8473921250317992e-07,
      "loss": 0.744,
      "step": 9605
    },
    {
      "epoch": 0.8663029264553366,
      "grad_norm": 2.2276567948939987,
      "learning_rate": 1.8449405570534225e-07,
      "loss": 0.675,
      "step": 9606
    },
    {
      "epoch": 0.8663931099788069,
      "grad_norm": 1.5434795837782116,
      "learning_rate": 1.8424905381648204e-07,
      "loss": 0.7891,
      "step": 9607
    },
    {
      "epoch": 0.8664832935022772,
      "grad_norm": 2.1399099450459342,
      "learning_rate": 1.8400420685750452e-07,
      "loss": 0.7002,
      "step": 9608
    },
    {
      "epoch": 0.8665734770257474,
      "grad_norm": 2.3775313088715166,
      "learning_rate": 1.8375951484930142e-07,
      "loss": 0.6811,
      "step": 9609
    },
    {
      "epoch": 0.8666636605492176,
      "grad_norm": 4.952784577714055,
      "learning_rate": 1.8351497781275094e-07,
      "loss": 0.7471,
      "step": 9610
    },
    {
      "epoch": 0.866753844072688,
      "grad_norm": 2.0286221451818878,
      "learning_rate": 1.8327059576871907e-07,
      "loss": 0.7277,
      "step": 9611
    },
    {
      "epoch": 0.8668440275961582,
      "grad_norm": 1.4389819923106528,
      "learning_rate": 1.8302636873805665e-07,
      "loss": 0.7938,
      "step": 9612
    },
    {
      "epoch": 0.8669342111196284,
      "grad_norm": 1.7905377298495473,
      "learning_rate": 1.8278229674160373e-07,
      "loss": 0.8214,
      "step": 9613
    },
    {
      "epoch": 0.8670243946430987,
      "grad_norm": 1.332524541088512,
      "learning_rate": 1.825383798001845e-07,
      "loss": 0.6767,
      "step": 9614
    },
    {
      "epoch": 0.867114578166569,
      "grad_norm": 1.6568840540606526,
      "learning_rate": 1.8229461793461297e-07,
      "loss": 0.689,
      "step": 9615
    },
    {
      "epoch": 0.8672047616900392,
      "grad_norm": 1.625694639950157,
      "learning_rate": 1.8205101116568698e-07,
      "loss": 0.7586,
      "step": 9616
    },
    {
      "epoch": 0.8672949452135095,
      "grad_norm": 1.9310857363473708,
      "learning_rate": 1.818075595141928e-07,
      "loss": 0.7943,
      "step": 9617
    },
    {
      "epoch": 0.8673851287369797,
      "grad_norm": 2.398830915858477,
      "learning_rate": 1.8156426300090288e-07,
      "loss": 0.741,
      "step": 9618
    },
    {
      "epoch": 0.86747531226045,
      "grad_norm": 1.4665032917119725,
      "learning_rate": 1.8132112164657686e-07,
      "loss": 0.7309,
      "step": 9619
    },
    {
      "epoch": 0.8675654957839203,
      "grad_norm": 1.6716369744678294,
      "learning_rate": 1.8107813547196106e-07,
      "loss": 0.6381,
      "step": 9620
    },
    {
      "epoch": 0.8676556793073905,
      "grad_norm": 1.9359033219812671,
      "learning_rate": 1.8083530449778817e-07,
      "loss": 0.7828,
      "step": 9621
    },
    {
      "epoch": 0.8677458628308607,
      "grad_norm": 1.843809870378247,
      "learning_rate": 1.8059262874477787e-07,
      "loss": 0.6682,
      "step": 9622
    },
    {
      "epoch": 0.8678360463543311,
      "grad_norm": 3.598133366164062,
      "learning_rate": 1.8035010823363627e-07,
      "loss": 0.7281,
      "step": 9623
    },
    {
      "epoch": 0.8679262298778013,
      "grad_norm": 6.672613098626718,
      "learning_rate": 1.8010774298505705e-07,
      "loss": 0.782,
      "step": 9624
    },
    {
      "epoch": 0.8680164134012716,
      "grad_norm": 2.6549014988100383,
      "learning_rate": 1.7986553301972007e-07,
      "loss": 0.835,
      "step": 9625
    },
    {
      "epoch": 0.8681065969247419,
      "grad_norm": 1.8346115193774595,
      "learning_rate": 1.7962347835829171e-07,
      "loss": 0.7354,
      "step": 9626
    },
    {
      "epoch": 0.8681967804482121,
      "grad_norm": 1.5058158343757992,
      "learning_rate": 1.793815790214257e-07,
      "loss": 0.7486,
      "step": 9627
    },
    {
      "epoch": 0.8682869639716824,
      "grad_norm": 1.8445313947678363,
      "learning_rate": 1.791398350297626e-07,
      "loss": 0.7045,
      "step": 9628
    },
    {
      "epoch": 0.8683771474951526,
      "grad_norm": 1.9296046559701485,
      "learning_rate": 1.7889824640392813e-07,
      "loss": 0.7496,
      "step": 9629
    },
    {
      "epoch": 0.868467331018623,
      "grad_norm": 1.8461265764453787,
      "learning_rate": 1.7865681316453741e-07,
      "loss": 0.7241,
      "step": 9630
    },
    {
      "epoch": 0.8685575145420932,
      "grad_norm": 1.7209060324151793,
      "learning_rate": 1.7841553533218968e-07,
      "loss": 0.7436,
      "step": 9631
    },
    {
      "epoch": 0.8686476980655634,
      "grad_norm": 24.576709471316537,
      "learning_rate": 1.7817441292747292e-07,
      "loss": 0.8472,
      "step": 9632
    },
    {
      "epoch": 0.8687378815890336,
      "grad_norm": 1.6920055006221884,
      "learning_rate": 1.779334459709607e-07,
      "loss": 0.7359,
      "step": 9633
    },
    {
      "epoch": 0.868828065112504,
      "grad_norm": 2.0043141803131914,
      "learning_rate": 1.7769263448321347e-07,
      "loss": 0.7528,
      "step": 9634
    },
    {
      "epoch": 0.8689182486359742,
      "grad_norm": 3.294935189103213,
      "learning_rate": 1.7745197848477879e-07,
      "loss": 0.736,
      "step": 9635
    },
    {
      "epoch": 0.8690084321594445,
      "grad_norm": 2.0380077922848026,
      "learning_rate": 1.7721147799619063e-07,
      "loss": 0.8448,
      "step": 9636
    },
    {
      "epoch": 0.8690986156829147,
      "grad_norm": 0.5848510333256652,
      "learning_rate": 1.769711330379704e-07,
      "loss": 0.5475,
      "step": 9637
    },
    {
      "epoch": 0.869188799206385,
      "grad_norm": 1.9005310696714481,
      "learning_rate": 1.767309436306248e-07,
      "loss": 0.7923,
      "step": 9638
    },
    {
      "epoch": 0.8692789827298553,
      "grad_norm": 1.9816514195186359,
      "learning_rate": 1.764909097946483e-07,
      "loss": 0.8082,
      "step": 9639
    },
    {
      "epoch": 0.8693691662533255,
      "grad_norm": 1.9500200539560326,
      "learning_rate": 1.7625103155052236e-07,
      "loss": 0.8105,
      "step": 9640
    },
    {
      "epoch": 0.8694593497767957,
      "grad_norm": 1.7428377588721822,
      "learning_rate": 1.760113089187143e-07,
      "loss": 0.7582,
      "step": 9641
    },
    {
      "epoch": 0.8695495333002661,
      "grad_norm": 2.3408107383382637,
      "learning_rate": 1.7577174191967868e-07,
      "loss": 0.7699,
      "step": 9642
    },
    {
      "epoch": 0.8696397168237363,
      "grad_norm": 1.6780147296045334,
      "learning_rate": 1.755323305738574e-07,
      "loss": 0.7556,
      "step": 9643
    },
    {
      "epoch": 0.8697299003472065,
      "grad_norm": 1.4758195572218562,
      "learning_rate": 1.7529307490167677e-07,
      "loss": 0.7759,
      "step": 9644
    },
    {
      "epoch": 0.8698200838706768,
      "grad_norm": 3.2416853550337112,
      "learning_rate": 1.7505397492355288e-07,
      "loss": 0.7281,
      "step": 9645
    },
    {
      "epoch": 0.8699102673941471,
      "grad_norm": 1.8376163030654333,
      "learning_rate": 1.7481503065988589e-07,
      "loss": 0.776,
      "step": 9646
    },
    {
      "epoch": 0.8700004509176174,
      "grad_norm": 1.374688850223531,
      "learning_rate": 1.7457624213106526e-07,
      "loss": 0.7107,
      "step": 9647
    },
    {
      "epoch": 0.8700906344410876,
      "grad_norm": 1.750400565488853,
      "learning_rate": 1.7433760935746465e-07,
      "loss": 0.7469,
      "step": 9648
    },
    {
      "epoch": 0.8701808179645579,
      "grad_norm": 1.9030024897789115,
      "learning_rate": 1.740991323594456e-07,
      "loss": 0.6901,
      "step": 9649
    },
    {
      "epoch": 0.8702710014880282,
      "grad_norm": 1.585458241476097,
      "learning_rate": 1.7386081115735651e-07,
      "loss": 0.7232,
      "step": 9650
    },
    {
      "epoch": 0.8703611850114984,
      "grad_norm": 2.285613498424103,
      "learning_rate": 1.736226457715324e-07,
      "loss": 0.7063,
      "step": 9651
    },
    {
      "epoch": 0.8704513685349686,
      "grad_norm": 1.3814545769032274,
      "learning_rate": 1.7338463622229505e-07,
      "loss": 0.6974,
      "step": 9652
    },
    {
      "epoch": 0.870541552058439,
      "grad_norm": 14.874818945031262,
      "learning_rate": 1.7314678252995152e-07,
      "loss": 0.689,
      "step": 9653
    },
    {
      "epoch": 0.8706317355819092,
      "grad_norm": 1.6093788517285523,
      "learning_rate": 1.7290908471479847e-07,
      "loss": 0.6759,
      "step": 9654
    },
    {
      "epoch": 0.8707219191053794,
      "grad_norm": 1.9872108681017617,
      "learning_rate": 1.7267154279711637e-07,
      "loss": 0.7412,
      "step": 9655
    },
    {
      "epoch": 0.8708121026288497,
      "grad_norm": 1.406736374562119,
      "learning_rate": 1.724341567971741e-07,
      "loss": 0.791,
      "step": 9656
    },
    {
      "epoch": 0.87090228615232,
      "grad_norm": 1.6558265864569648,
      "learning_rate": 1.7219692673522657e-07,
      "loss": 0.8435,
      "step": 9657
    },
    {
      "epoch": 0.8709924696757902,
      "grad_norm": 1.6822053003334916,
      "learning_rate": 1.7195985263151558e-07,
      "loss": 0.7636,
      "step": 9658
    },
    {
      "epoch": 0.8710826531992605,
      "grad_norm": 2.083305384130556,
      "learning_rate": 1.7172293450626985e-07,
      "loss": 0.7919,
      "step": 9659
    },
    {
      "epoch": 0.8711728367227307,
      "grad_norm": 2.0553552971053124,
      "learning_rate": 1.7148617237970475e-07,
      "loss": 0.7449,
      "step": 9660
    },
    {
      "epoch": 0.8712630202462011,
      "grad_norm": 1.748513135049652,
      "learning_rate": 1.7124956627202102e-07,
      "loss": 0.7087,
      "step": 9661
    },
    {
      "epoch": 0.8713532037696713,
      "grad_norm": 3.790193809646818,
      "learning_rate": 1.7101311620340852e-07,
      "loss": 0.7604,
      "step": 9662
    },
    {
      "epoch": 0.8714433872931415,
      "grad_norm": 1.7654347275017255,
      "learning_rate": 1.7077682219404155e-07,
      "loss": 0.692,
      "step": 9663
    },
    {
      "epoch": 0.8715335708166118,
      "grad_norm": 7.017574863449005,
      "learning_rate": 1.705406842640824e-07,
      "loss": 0.8705,
      "step": 9664
    },
    {
      "epoch": 0.8716237543400821,
      "grad_norm": 1.4697324835065533,
      "learning_rate": 1.7030470243367946e-07,
      "loss": 0.7343,
      "step": 9665
    },
    {
      "epoch": 0.8717139378635523,
      "grad_norm": 1.3513886885704387,
      "learning_rate": 1.7006887672296834e-07,
      "loss": 0.6685,
      "step": 9666
    },
    {
      "epoch": 0.8718041213870226,
      "grad_norm": 1.8014329901775352,
      "learning_rate": 1.6983320715207094e-07,
      "loss": 0.6879,
      "step": 9667
    },
    {
      "epoch": 0.8718943049104928,
      "grad_norm": 1.4387890385055095,
      "learning_rate": 1.6959769374109523e-07,
      "loss": 0.7631,
      "step": 9668
    },
    {
      "epoch": 0.8719844884339631,
      "grad_norm": 1.5468017098375197,
      "learning_rate": 1.6936233651013754e-07,
      "loss": 0.6812,
      "step": 9669
    },
    {
      "epoch": 0.8720746719574334,
      "grad_norm": 1.5188123870720218,
      "learning_rate": 1.691271354792787e-07,
      "loss": 0.7744,
      "step": 9670
    },
    {
      "epoch": 0.8721648554809036,
      "grad_norm": 1.462819972452116,
      "learning_rate": 1.6889209066858866e-07,
      "loss": 0.6238,
      "step": 9671
    },
    {
      "epoch": 0.872255039004374,
      "grad_norm": 1.6413453473463075,
      "learning_rate": 1.6865720209812185e-07,
      "loss": 0.7892,
      "step": 9672
    },
    {
      "epoch": 0.8723452225278442,
      "grad_norm": 2.0028006494879502,
      "learning_rate": 1.684224697879204e-07,
      "loss": 0.7107,
      "step": 9673
    },
    {
      "epoch": 0.8724354060513144,
      "grad_norm": 3.102701972332143,
      "learning_rate": 1.6818789375801302e-07,
      "loss": 0.819,
      "step": 9674
    },
    {
      "epoch": 0.8725255895747847,
      "grad_norm": 1.4437590627674834,
      "learning_rate": 1.679534740284152e-07,
      "loss": 0.6993,
      "step": 9675
    },
    {
      "epoch": 0.872615773098255,
      "grad_norm": 1.660313282439961,
      "learning_rate": 1.6771921061912853e-07,
      "loss": 0.6855,
      "step": 9676
    },
    {
      "epoch": 0.8727059566217252,
      "grad_norm": 1.8919997458387208,
      "learning_rate": 1.6748510355014234e-07,
      "loss": 0.6816,
      "step": 9677
    },
    {
      "epoch": 0.8727961401451955,
      "grad_norm": 1.6745089166904443,
      "learning_rate": 1.6725115284143132e-07,
      "loss": 0.7552,
      "step": 9678
    },
    {
      "epoch": 0.8728863236686657,
      "grad_norm": 2.6901600115808977,
      "learning_rate": 1.670173585129575e-07,
      "loss": 0.72,
      "step": 9679
    },
    {
      "epoch": 0.872976507192136,
      "grad_norm": 1.624874256844332,
      "learning_rate": 1.667837205846696e-07,
      "loss": 0.802,
      "step": 9680
    },
    {
      "epoch": 0.8730666907156063,
      "grad_norm": 2.1371016638729463,
      "learning_rate": 1.6655023907650278e-07,
      "loss": 0.7498,
      "step": 9681
    },
    {
      "epoch": 0.8731568742390765,
      "grad_norm": 2.2272748911746243,
      "learning_rate": 1.6631691400837954e-07,
      "loss": 0.7816,
      "step": 9682
    },
    {
      "epoch": 0.8732470577625467,
      "grad_norm": 1.5216085557123478,
      "learning_rate": 1.6608374540020752e-07,
      "loss": 0.7949,
      "step": 9683
    },
    {
      "epoch": 0.8733372412860171,
      "grad_norm": 0.6888906978340736,
      "learning_rate": 1.658507332718828e-07,
      "loss": 0.5272,
      "step": 9684
    },
    {
      "epoch": 0.8734274248094873,
      "grad_norm": 1.4758763409376898,
      "learning_rate": 1.656178776432864e-07,
      "loss": 0.6788,
      "step": 9685
    },
    {
      "epoch": 0.8735176083329576,
      "grad_norm": 1.4873511744827956,
      "learning_rate": 1.6538517853428814e-07,
      "loss": 0.723,
      "step": 9686
    },
    {
      "epoch": 0.8736077918564278,
      "grad_norm": 2.483014387945553,
      "learning_rate": 1.6515263596474194e-07,
      "loss": 0.724,
      "step": 9687
    },
    {
      "epoch": 0.8736979753798981,
      "grad_norm": 2.996941119629134,
      "learning_rate": 1.6492024995449017e-07,
      "loss": 0.7305,
      "step": 9688
    },
    {
      "epoch": 0.8737881589033684,
      "grad_norm": 1.671636546590342,
      "learning_rate": 1.6468802052336116e-07,
      "loss": 0.7075,
      "step": 9689
    },
    {
      "epoch": 0.8738783424268386,
      "grad_norm": 1.6219685962990686,
      "learning_rate": 1.6445594769116998e-07,
      "loss": 0.7627,
      "step": 9690
    },
    {
      "epoch": 0.8739685259503088,
      "grad_norm": 1.633362891779618,
      "learning_rate": 1.6422403147771836e-07,
      "loss": 0.8148,
      "step": 9691
    },
    {
      "epoch": 0.8740587094737792,
      "grad_norm": 1.8995039342491573,
      "learning_rate": 1.6399227190279485e-07,
      "loss": 0.707,
      "step": 9692
    },
    {
      "epoch": 0.8741488929972494,
      "grad_norm": 1.8762463559821194,
      "learning_rate": 1.637606689861748e-07,
      "loss": 0.6744,
      "step": 9693
    },
    {
      "epoch": 0.8742390765207196,
      "grad_norm": 36.32381862916793,
      "learning_rate": 1.6352922274761883e-07,
      "loss": 0.7013,
      "step": 9694
    },
    {
      "epoch": 0.87432926004419,
      "grad_norm": 1.8112651301476663,
      "learning_rate": 1.6329793320687602e-07,
      "loss": 0.6072,
      "step": 9695
    },
    {
      "epoch": 0.8744194435676602,
      "grad_norm": 1.8617830471088603,
      "learning_rate": 1.630668003836808e-07,
      "loss": 0.781,
      "step": 9696
    },
    {
      "epoch": 0.8745096270911304,
      "grad_norm": 5.424334493152457,
      "learning_rate": 1.62835824297755e-07,
      "loss": 0.7471,
      "step": 9697
    },
    {
      "epoch": 0.8745998106146007,
      "grad_norm": 1.5054179730881718,
      "learning_rate": 1.626050049688066e-07,
      "loss": 0.7544,
      "step": 9698
    },
    {
      "epoch": 0.874689994138071,
      "grad_norm": 11.110374523755059,
      "learning_rate": 1.623743424165309e-07,
      "loss": 0.7194,
      "step": 9699
    },
    {
      "epoch": 0.8747801776615413,
      "grad_norm": 1.7610605029360067,
      "learning_rate": 1.6214383666060826e-07,
      "loss": 0.6953,
      "step": 9700
    },
    {
      "epoch": 0.8748703611850115,
      "grad_norm": 2.576172615380864,
      "learning_rate": 1.619134877207078e-07,
      "loss": 0.7638,
      "step": 9701
    },
    {
      "epoch": 0.8749605447084817,
      "grad_norm": 1.6181154399190116,
      "learning_rate": 1.616832956164831e-07,
      "loss": 0.7447,
      "step": 9702
    },
    {
      "epoch": 0.8750507282319521,
      "grad_norm": 1.5755547469931679,
      "learning_rate": 1.6145326036757667e-07,
      "loss": 0.7135,
      "step": 9703
    },
    {
      "epoch": 0.8751409117554223,
      "grad_norm": 9.796919778481199,
      "learning_rate": 1.612233819936155e-07,
      "loss": 0.7081,
      "step": 9704
    },
    {
      "epoch": 0.8752310952788925,
      "grad_norm": 1.7368741973622523,
      "learning_rate": 1.6099366051421414e-07,
      "loss": 0.7408,
      "step": 9705
    },
    {
      "epoch": 0.8753212788023628,
      "grad_norm": 1.5222628246161212,
      "learning_rate": 1.6076409594897378e-07,
      "loss": 0.803,
      "step": 9706
    },
    {
      "epoch": 0.8754114623258331,
      "grad_norm": 0.662764481167924,
      "learning_rate": 1.605346883174823e-07,
      "loss": 0.5273,
      "step": 9707
    },
    {
      "epoch": 0.8755016458493033,
      "grad_norm": 1.8881484275761018,
      "learning_rate": 1.6030543763931427e-07,
      "loss": 0.7169,
      "step": 9708
    },
    {
      "epoch": 0.8755918293727736,
      "grad_norm": 1.6722401728182343,
      "learning_rate": 1.600763439340298e-07,
      "loss": 0.7618,
      "step": 9709
    },
    {
      "epoch": 0.8756820128962438,
      "grad_norm": 1.6538780790226366,
      "learning_rate": 1.5984740722117707e-07,
      "loss": 0.7962,
      "step": 9710
    },
    {
      "epoch": 0.8757721964197142,
      "grad_norm": 1.7133292981606154,
      "learning_rate": 1.5961862752028998e-07,
      "loss": 0.7566,
      "step": 9711
    },
    {
      "epoch": 0.8758623799431844,
      "grad_norm": 1.386938201303147,
      "learning_rate": 1.5939000485088937e-07,
      "loss": 0.6526,
      "step": 9712
    },
    {
      "epoch": 0.8759525634666546,
      "grad_norm": 1.9298917222162386,
      "learning_rate": 1.5916153923248254e-07,
      "loss": 0.8387,
      "step": 9713
    },
    {
      "epoch": 0.8760427469901249,
      "grad_norm": 1.530222022209803,
      "learning_rate": 1.5893323068456342e-07,
      "loss": 0.7842,
      "step": 9714
    },
    {
      "epoch": 0.8761329305135952,
      "grad_norm": 1.4892002582372563,
      "learning_rate": 1.5870507922661248e-07,
      "loss": 0.7901,
      "step": 9715
    },
    {
      "epoch": 0.8762231140370654,
      "grad_norm": 0.6234892706540567,
      "learning_rate": 1.5847708487809763e-07,
      "loss": 0.5683,
      "step": 9716
    },
    {
      "epoch": 0.8763132975605357,
      "grad_norm": 4.647726565579515,
      "learning_rate": 1.5824924765847113e-07,
      "loss": 0.8231,
      "step": 9717
    },
    {
      "epoch": 0.8764034810840059,
      "grad_norm": 1.4324830453345332,
      "learning_rate": 1.5802156758717478e-07,
      "loss": 0.7221,
      "step": 9718
    },
    {
      "epoch": 0.8764936646074762,
      "grad_norm": 1.7653248624660827,
      "learning_rate": 1.5779404468363433e-07,
      "loss": 0.7452,
      "step": 9719
    },
    {
      "epoch": 0.8765838481309465,
      "grad_norm": 1.827634305389303,
      "learning_rate": 1.5756667896726405e-07,
      "loss": 0.797,
      "step": 9720
    },
    {
      "epoch": 0.8766740316544167,
      "grad_norm": 8.762533312419961,
      "learning_rate": 1.5733947045746377e-07,
      "loss": 0.7622,
      "step": 9721
    },
    {
      "epoch": 0.876764215177887,
      "grad_norm": 1.452594291003297,
      "learning_rate": 1.5711241917362018e-07,
      "loss": 0.809,
      "step": 9722
    },
    {
      "epoch": 0.8768543987013573,
      "grad_norm": 2.512908226744956,
      "learning_rate": 1.5688552513510688e-07,
      "loss": 0.6875,
      "step": 9723
    },
    {
      "epoch": 0.8769445822248275,
      "grad_norm": 2.874082170016599,
      "learning_rate": 1.5665878836128266e-07,
      "loss": 0.7663,
      "step": 9724
    },
    {
      "epoch": 0.8770347657482978,
      "grad_norm": 1.4906593206841083,
      "learning_rate": 1.5643220887149554e-07,
      "loss": 0.7908,
      "step": 9725
    },
    {
      "epoch": 0.8771249492717681,
      "grad_norm": 0.6630898027489976,
      "learning_rate": 1.562057866850772e-07,
      "loss": 0.5884,
      "step": 9726
    },
    {
      "epoch": 0.8772151327952383,
      "grad_norm": 1.795268792134131,
      "learning_rate": 1.5597952182134777e-07,
      "loss": 0.6175,
      "step": 9727
    },
    {
      "epoch": 0.8773053163187086,
      "grad_norm": 1.5146665393724075,
      "learning_rate": 1.557534142996133e-07,
      "loss": 0.7995,
      "step": 9728
    },
    {
      "epoch": 0.8773954998421788,
      "grad_norm": 1.9785769002149214,
      "learning_rate": 1.5552746413916662e-07,
      "loss": 0.7775,
      "step": 9729
    },
    {
      "epoch": 0.8774856833656491,
      "grad_norm": 1.7064712967782099,
      "learning_rate": 1.5530167135928697e-07,
      "loss": 0.7375,
      "step": 9730
    },
    {
      "epoch": 0.8775758668891194,
      "grad_norm": 1.874984486833611,
      "learning_rate": 1.5507603597924068e-07,
      "loss": 0.6923,
      "step": 9731
    },
    {
      "epoch": 0.8776660504125896,
      "grad_norm": 3.1019190047688214,
      "learning_rate": 1.548505580182793e-07,
      "loss": 0.7063,
      "step": 9732
    },
    {
      "epoch": 0.8777562339360598,
      "grad_norm": 1.5553904722051666,
      "learning_rate": 1.5462523749564271e-07,
      "loss": 0.7453,
      "step": 9733
    },
    {
      "epoch": 0.8778464174595302,
      "grad_norm": 1.532245857112423,
      "learning_rate": 1.5440007443055602e-07,
      "loss": 0.7397,
      "step": 9734
    },
    {
      "epoch": 0.8779366009830004,
      "grad_norm": 4.2810417180317595,
      "learning_rate": 1.541750688422314e-07,
      "loss": 0.6796,
      "step": 9735
    },
    {
      "epoch": 0.8780267845064706,
      "grad_norm": 1.975243471324827,
      "learning_rate": 1.5395022074986797e-07,
      "loss": 0.747,
      "step": 9736
    },
    {
      "epoch": 0.8781169680299409,
      "grad_norm": 1.6615936877882418,
      "learning_rate": 1.5372553017265033e-07,
      "loss": 0.6873,
      "step": 9737
    },
    {
      "epoch": 0.8782071515534112,
      "grad_norm": 2.2253046063000705,
      "learning_rate": 1.5350099712975116e-07,
      "loss": 0.8035,
      "step": 9738
    },
    {
      "epoch": 0.8782973350768815,
      "grad_norm": 1.7992364376522167,
      "learning_rate": 1.5327662164032785e-07,
      "loss": 0.737,
      "step": 9739
    },
    {
      "epoch": 0.8783875186003517,
      "grad_norm": 1.6033246278559525,
      "learning_rate": 1.5305240372352656e-07,
      "loss": 0.5937,
      "step": 9740
    },
    {
      "epoch": 0.8784777021238219,
      "grad_norm": 1.9009613089800474,
      "learning_rate": 1.5282834339847738e-07,
      "loss": 0.7582,
      "step": 9741
    },
    {
      "epoch": 0.8785678856472923,
      "grad_norm": 1.6957550174459504,
      "learning_rate": 1.526044406842999e-07,
      "loss": 0.7231,
      "step": 9742
    },
    {
      "epoch": 0.8786580691707625,
      "grad_norm": 1.770480880069563,
      "learning_rate": 1.523806956000977e-07,
      "loss": 0.6735,
      "step": 9743
    },
    {
      "epoch": 0.8787482526942327,
      "grad_norm": 1.8580417501236717,
      "learning_rate": 1.5215710816496197e-07,
      "loss": 0.7967,
      "step": 9744
    },
    {
      "epoch": 0.8788384362177031,
      "grad_norm": 1.752031713655928,
      "learning_rate": 1.5193367839797077e-07,
      "loss": 0.6919,
      "step": 9745
    },
    {
      "epoch": 0.8789286197411733,
      "grad_norm": 1.4559439112318848,
      "learning_rate": 1.5171040631818842e-07,
      "loss": 0.7431,
      "step": 9746
    },
    {
      "epoch": 0.8790188032646435,
      "grad_norm": 1.7108261559892184,
      "learning_rate": 1.5148729194466547e-07,
      "loss": 0.6751,
      "step": 9747
    },
    {
      "epoch": 0.8791089867881138,
      "grad_norm": 1.601741650493658,
      "learning_rate": 1.5126433529643956e-07,
      "loss": 0.8128,
      "step": 9748
    },
    {
      "epoch": 0.8791991703115841,
      "grad_norm": 1.4143557586911946,
      "learning_rate": 1.5104153639253436e-07,
      "loss": 0.7416,
      "step": 9749
    },
    {
      "epoch": 0.8792893538350544,
      "grad_norm": 0.6129256997608706,
      "learning_rate": 1.5081889525196002e-07,
      "loss": 0.6297,
      "step": 9750
    },
    {
      "epoch": 0.8793795373585246,
      "grad_norm": 2.559025246147193,
      "learning_rate": 1.5059641189371398e-07,
      "loss": 0.798,
      "step": 9751
    },
    {
      "epoch": 0.8794697208819948,
      "grad_norm": 1.5716538654802552,
      "learning_rate": 1.503740863367795e-07,
      "loss": 0.7599,
      "step": 9752
    },
    {
      "epoch": 0.8795599044054652,
      "grad_norm": 1.830016828261582,
      "learning_rate": 1.50151918600127e-07,
      "loss": 0.7739,
      "step": 9753
    },
    {
      "epoch": 0.8796500879289354,
      "grad_norm": 2.1958441582641544,
      "learning_rate": 1.4992990870271217e-07,
      "loss": 0.7909,
      "step": 9754
    },
    {
      "epoch": 0.8797402714524056,
      "grad_norm": 1.992536686232875,
      "learning_rate": 1.497080566634794e-07,
      "loss": 0.7496,
      "step": 9755
    },
    {
      "epoch": 0.8798304549758759,
      "grad_norm": 1.5232942171451695,
      "learning_rate": 1.4948636250135693e-07,
      "loss": 0.7552,
      "step": 9756
    },
    {
      "epoch": 0.8799206384993462,
      "grad_norm": 1.7841202214963534,
      "learning_rate": 1.4926482623526249e-07,
      "loss": 0.8089,
      "step": 9757
    },
    {
      "epoch": 0.8800108220228164,
      "grad_norm": 1.8257693249921456,
      "learning_rate": 1.4904344788409694e-07,
      "loss": 0.7461,
      "step": 9758
    },
    {
      "epoch": 0.8801010055462867,
      "grad_norm": 1.572896018324304,
      "learning_rate": 1.4882222746675143e-07,
      "loss": 0.7095,
      "step": 9759
    },
    {
      "epoch": 0.8801911890697569,
      "grad_norm": 1.6419847212578929,
      "learning_rate": 1.4860116500210018e-07,
      "loss": 0.7874,
      "step": 9760
    },
    {
      "epoch": 0.8802813725932273,
      "grad_norm": 6.253419779740553,
      "learning_rate": 1.4838026050900632e-07,
      "loss": 0.815,
      "step": 9761
    },
    {
      "epoch": 0.8803715561166975,
      "grad_norm": 6.160835843042087,
      "learning_rate": 1.481595140063181e-07,
      "loss": 0.792,
      "step": 9762
    },
    {
      "epoch": 0.8804617396401677,
      "grad_norm": 1.8405275147404017,
      "learning_rate": 1.4793892551287136e-07,
      "loss": 0.8342,
      "step": 9763
    },
    {
      "epoch": 0.880551923163638,
      "grad_norm": 1.7766249324460661,
      "learning_rate": 1.4771849504748768e-07,
      "loss": 0.6792,
      "step": 9764
    },
    {
      "epoch": 0.8806421066871083,
      "grad_norm": 1.5015771045431072,
      "learning_rate": 1.4749822262897517e-07,
      "loss": 0.8217,
      "step": 9765
    },
    {
      "epoch": 0.8807322902105785,
      "grad_norm": 1.889809361586898,
      "learning_rate": 1.4727810827612895e-07,
      "loss": 0.6904,
      "step": 9766
    },
    {
      "epoch": 0.8808224737340488,
      "grad_norm": 2.2177858003769795,
      "learning_rate": 1.470581520077303e-07,
      "loss": 0.7534,
      "step": 9767
    },
    {
      "epoch": 0.8809126572575191,
      "grad_norm": 2.0210639848202,
      "learning_rate": 1.4683835384254705e-07,
      "loss": 0.7433,
      "step": 9768
    },
    {
      "epoch": 0.8810028407809893,
      "grad_norm": 1.6698541995607064,
      "learning_rate": 1.4661871379933376e-07,
      "loss": 0.8276,
      "step": 9769
    },
    {
      "epoch": 0.8810930243044596,
      "grad_norm": 1.8135838392139223,
      "learning_rate": 1.4639923189683169e-07,
      "loss": 0.794,
      "step": 9770
    },
    {
      "epoch": 0.8811832078279298,
      "grad_norm": 1.739061981776898,
      "learning_rate": 1.461799081537669e-07,
      "loss": 0.797,
      "step": 9771
    },
    {
      "epoch": 0.8812733913514001,
      "grad_norm": 1.6202602219861733,
      "learning_rate": 1.4596074258885514e-07,
      "loss": 0.8117,
      "step": 9772
    },
    {
      "epoch": 0.8813635748748704,
      "grad_norm": 1.7644067122676579,
      "learning_rate": 1.4574173522079502e-07,
      "loss": 0.7195,
      "step": 9773
    },
    {
      "epoch": 0.8814537583983406,
      "grad_norm": 2.1493615954736245,
      "learning_rate": 1.4552288606827513e-07,
      "loss": 0.7229,
      "step": 9774
    },
    {
      "epoch": 0.8815439419218108,
      "grad_norm": 1.651619682657408,
      "learning_rate": 1.4530419514996761e-07,
      "loss": 0.8232,
      "step": 9775
    },
    {
      "epoch": 0.8816341254452812,
      "grad_norm": 0.691620120144549,
      "learning_rate": 1.4508566248453291e-07,
      "loss": 0.5889,
      "step": 9776
    },
    {
      "epoch": 0.8817243089687514,
      "grad_norm": 1.922744492261669,
      "learning_rate": 1.448672880906172e-07,
      "loss": 0.7422,
      "step": 9777
    },
    {
      "epoch": 0.8818144924922217,
      "grad_norm": 1.5518276422152006,
      "learning_rate": 1.4464907198685382e-07,
      "loss": 0.6982,
      "step": 9778
    },
    {
      "epoch": 0.8819046760156919,
      "grad_norm": 1.736261115481428,
      "learning_rate": 1.444310141918621e-07,
      "loss": 0.7478,
      "step": 9779
    },
    {
      "epoch": 0.8819948595391622,
      "grad_norm": 2.636172813128436,
      "learning_rate": 1.4421311472424735e-07,
      "loss": 0.7774,
      "step": 9780
    },
    {
      "epoch": 0.8820850430626325,
      "grad_norm": 1.8886605768349052,
      "learning_rate": 1.4399537360260273e-07,
      "loss": 0.781,
      "step": 9781
    },
    {
      "epoch": 0.8821752265861027,
      "grad_norm": 2.608211052153971,
      "learning_rate": 1.4377779084550645e-07,
      "loss": 0.77,
      "step": 9782
    },
    {
      "epoch": 0.8822654101095729,
      "grad_norm": 1.722752298112487,
      "learning_rate": 1.4356036647152413e-07,
      "loss": 0.7377,
      "step": 9783
    },
    {
      "epoch": 0.8823555936330433,
      "grad_norm": 1.4473833249889874,
      "learning_rate": 1.4334310049920785e-07,
      "loss": 0.7249,
      "step": 9784
    },
    {
      "epoch": 0.8824457771565135,
      "grad_norm": 0.5985411282846871,
      "learning_rate": 1.431259929470956e-07,
      "loss": 0.5294,
      "step": 9785
    },
    {
      "epoch": 0.8825359606799837,
      "grad_norm": 1.8155390156068119,
      "learning_rate": 1.4290904383371237e-07,
      "loss": 0.8244,
      "step": 9786
    },
    {
      "epoch": 0.882626144203454,
      "grad_norm": 2.72197462002528,
      "learning_rate": 1.4269225317756961e-07,
      "loss": 0.7298,
      "step": 9787
    },
    {
      "epoch": 0.8827163277269243,
      "grad_norm": 2.1109981367210504,
      "learning_rate": 1.424756209971645e-07,
      "loss": 0.7132,
      "step": 9788
    },
    {
      "epoch": 0.8828065112503946,
      "grad_norm": 1.8710838747659875,
      "learning_rate": 1.4225914731098199e-07,
      "loss": 0.7441,
      "step": 9789
    },
    {
      "epoch": 0.8828966947738648,
      "grad_norm": 1.7346564399633781,
      "learning_rate": 1.4204283213749248e-07,
      "loss": 0.6801,
      "step": 9790
    },
    {
      "epoch": 0.8829868782973351,
      "grad_norm": 1.6829003412950747,
      "learning_rate": 1.4182667549515315e-07,
      "loss": 0.803,
      "step": 9791
    },
    {
      "epoch": 0.8830770618208054,
      "grad_norm": 1.8180401416329468,
      "learning_rate": 1.4161067740240752e-07,
      "loss": 0.7263,
      "step": 9792
    },
    {
      "epoch": 0.8831672453442756,
      "grad_norm": 1.858136317439663,
      "learning_rate": 1.4139483787768614e-07,
      "loss": 0.7389,
      "step": 9793
    },
    {
      "epoch": 0.8832574288677458,
      "grad_norm": 2.4387995850074518,
      "learning_rate": 1.4117915693940584e-07,
      "loss": 0.6566,
      "step": 9794
    },
    {
      "epoch": 0.8833476123912162,
      "grad_norm": 1.4107859043201307,
      "learning_rate": 1.409636346059684e-07,
      "loss": 0.7417,
      "step": 9795
    },
    {
      "epoch": 0.8834377959146864,
      "grad_norm": 2.6960928409279745,
      "learning_rate": 1.4074827089576501e-07,
      "loss": 0.7803,
      "step": 9796
    },
    {
      "epoch": 0.8835279794381566,
      "grad_norm": 1.83585464615239,
      "learning_rate": 1.4053306582717085e-07,
      "loss": 0.7594,
      "step": 9797
    },
    {
      "epoch": 0.8836181629616269,
      "grad_norm": 1.4997422473705186,
      "learning_rate": 1.4031801941854827e-07,
      "loss": 0.7745,
      "step": 9798
    },
    {
      "epoch": 0.8837083464850972,
      "grad_norm": 1.478917983019152,
      "learning_rate": 1.401031316882466e-07,
      "loss": 0.717,
      "step": 9799
    },
    {
      "epoch": 0.8837985300085675,
      "grad_norm": 1.537817126807648,
      "learning_rate": 1.39888402654601e-07,
      "loss": 0.762,
      "step": 9800
    },
    {
      "epoch": 0.8838887135320377,
      "grad_norm": 0.6106584677607396,
      "learning_rate": 1.3967383233593344e-07,
      "loss": 0.5876,
      "step": 9801
    },
    {
      "epoch": 0.8839788970555079,
      "grad_norm": 3.1733437130328195,
      "learning_rate": 1.3945942075055218e-07,
      "loss": 0.7076,
      "step": 9802
    },
    {
      "epoch": 0.8840690805789783,
      "grad_norm": 1.9322047321131726,
      "learning_rate": 1.3924516791675212e-07,
      "loss": 0.7667,
      "step": 9803
    },
    {
      "epoch": 0.8841592641024485,
      "grad_norm": 1.4818502537830336,
      "learning_rate": 1.3903107385281487e-07,
      "loss": 0.6547,
      "step": 9804
    },
    {
      "epoch": 0.8842494476259187,
      "grad_norm": 1.9360153909181792,
      "learning_rate": 1.3881713857700717e-07,
      "loss": 0.7585,
      "step": 9805
    },
    {
      "epoch": 0.884339631149389,
      "grad_norm": 4.380274835263911,
      "learning_rate": 1.3860336210758372e-07,
      "loss": 0.7236,
      "step": 9806
    },
    {
      "epoch": 0.8844298146728593,
      "grad_norm": 1.7619858093166574,
      "learning_rate": 1.3838974446278506e-07,
      "loss": 0.7083,
      "step": 9807
    },
    {
      "epoch": 0.8845199981963295,
      "grad_norm": 1.543945544132454,
      "learning_rate": 1.3817628566083817e-07,
      "loss": 0.7224,
      "step": 9808
    },
    {
      "epoch": 0.8846101817197998,
      "grad_norm": 1.8106252906385616,
      "learning_rate": 1.3796298571995712e-07,
      "loss": 0.7616,
      "step": 9809
    },
    {
      "epoch": 0.88470036524327,
      "grad_norm": 4.771665653215772,
      "learning_rate": 1.377498446583405e-07,
      "loss": 0.647,
      "step": 9810
    },
    {
      "epoch": 0.8847905487667403,
      "grad_norm": 2.154158863616722,
      "learning_rate": 1.3753686249417596e-07,
      "loss": 0.7212,
      "step": 9811
    },
    {
      "epoch": 0.8848807322902106,
      "grad_norm": 48.10991545246936,
      "learning_rate": 1.373240392456354e-07,
      "loss": 0.7017,
      "step": 9812
    },
    {
      "epoch": 0.8849709158136808,
      "grad_norm": 1.6888980725454192,
      "learning_rate": 1.37111374930879e-07,
      "loss": 0.6827,
      "step": 9813
    },
    {
      "epoch": 0.885061099337151,
      "grad_norm": 1.5222381565825482,
      "learning_rate": 1.3689886956805176e-07,
      "loss": 0.6727,
      "step": 9814
    },
    {
      "epoch": 0.8851512828606214,
      "grad_norm": 1.467738351002792,
      "learning_rate": 1.3668652317528585e-07,
      "loss": 0.7374,
      "step": 9815
    },
    {
      "epoch": 0.8852414663840916,
      "grad_norm": 1.5774552841015317,
      "learning_rate": 1.3647433577070012e-07,
      "loss": 0.7653,
      "step": 9816
    },
    {
      "epoch": 0.8853316499075619,
      "grad_norm": 0.7203149758252937,
      "learning_rate": 1.3626230737239942e-07,
      "loss": 0.5495,
      "step": 9817
    },
    {
      "epoch": 0.8854218334310322,
      "grad_norm": 0.7035542767130476,
      "learning_rate": 1.3605043799847527e-07,
      "loss": 0.557,
      "step": 9818
    },
    {
      "epoch": 0.8855120169545024,
      "grad_norm": 1.4712089084276707,
      "learning_rate": 1.3583872766700567e-07,
      "loss": 0.6825,
      "step": 9819
    },
    {
      "epoch": 0.8856022004779727,
      "grad_norm": 1.5835429186724652,
      "learning_rate": 1.3562717639605437e-07,
      "loss": 0.7282,
      "step": 9820
    },
    {
      "epoch": 0.8856923840014429,
      "grad_norm": 2.987954594329686,
      "learning_rate": 1.3541578420367229e-07,
      "loss": 0.7162,
      "step": 9821
    },
    {
      "epoch": 0.8857825675249132,
      "grad_norm": 1.7401541574744213,
      "learning_rate": 1.3520455110789697e-07,
      "loss": 0.7624,
      "step": 9822
    },
    {
      "epoch": 0.8858727510483835,
      "grad_norm": 1.4851112547534482,
      "learning_rate": 1.3499347712675158e-07,
      "loss": 0.7289,
      "step": 9823
    },
    {
      "epoch": 0.8859629345718537,
      "grad_norm": 1.5313821073882035,
      "learning_rate": 1.3478256227824635e-07,
      "loss": 0.7082,
      "step": 9824
    },
    {
      "epoch": 0.8860531180953239,
      "grad_norm": 1.845136993615249,
      "learning_rate": 1.3457180658037759e-07,
      "loss": 0.6553,
      "step": 9825
    },
    {
      "epoch": 0.8861433016187943,
      "grad_norm": 2.2747750558836812,
      "learning_rate": 1.3436121005112843e-07,
      "loss": 0.7779,
      "step": 9826
    },
    {
      "epoch": 0.8862334851422645,
      "grad_norm": 1.663428172247141,
      "learning_rate": 1.3415077270846719e-07,
      "loss": 0.7142,
      "step": 9827
    },
    {
      "epoch": 0.8863236686657348,
      "grad_norm": 2.283289689902251,
      "learning_rate": 1.3394049457035105e-07,
      "loss": 0.8599,
      "step": 9828
    },
    {
      "epoch": 0.886413852189205,
      "grad_norm": 1.5920854739817278,
      "learning_rate": 1.3373037565472034e-07,
      "loss": 0.7572,
      "step": 9829
    },
    {
      "epoch": 0.8865040357126753,
      "grad_norm": 1.471779559821419,
      "learning_rate": 1.3352041597950537e-07,
      "loss": 0.7264,
      "step": 9830
    },
    {
      "epoch": 0.8865942192361456,
      "grad_norm": 2.3937341754280705,
      "learning_rate": 1.333106155626196e-07,
      "loss": 0.7379,
      "step": 9831
    },
    {
      "epoch": 0.8866844027596158,
      "grad_norm": 1.6694091723681728,
      "learning_rate": 1.331009744219651e-07,
      "loss": 0.6841,
      "step": 9832
    },
    {
      "epoch": 0.886774586283086,
      "grad_norm": 0.672835639134027,
      "learning_rate": 1.3289149257542943e-07,
      "loss": 0.6125,
      "step": 9833
    },
    {
      "epoch": 0.8868647698065564,
      "grad_norm": 1.6340265460571943,
      "learning_rate": 1.3268217004088666e-07,
      "loss": 0.6474,
      "step": 9834
    },
    {
      "epoch": 0.8869549533300266,
      "grad_norm": 1.6358968997497116,
      "learning_rate": 1.3247300683619788e-07,
      "loss": 0.636,
      "step": 9835
    },
    {
      "epoch": 0.8870451368534968,
      "grad_norm": 2.712699647227484,
      "learning_rate": 1.3226400297920903e-07,
      "loss": 0.7215,
      "step": 9836
    },
    {
      "epoch": 0.8871353203769671,
      "grad_norm": 1.7267698275680208,
      "learning_rate": 1.3205515848775428e-07,
      "loss": 0.7287,
      "step": 9837
    },
    {
      "epoch": 0.8872255039004374,
      "grad_norm": 1.680802370062659,
      "learning_rate": 1.3184647337965316e-07,
      "loss": 0.6843,
      "step": 9838
    },
    {
      "epoch": 0.8873156874239077,
      "grad_norm": 1.4381644330510055,
      "learning_rate": 1.3163794767271163e-07,
      "loss": 0.6942,
      "step": 9839
    },
    {
      "epoch": 0.8874058709473779,
      "grad_norm": 1.503884293036752,
      "learning_rate": 1.314295813847226e-07,
      "loss": 0.7154,
      "step": 9840
    },
    {
      "epoch": 0.8874960544708482,
      "grad_norm": 1.8145945055588844,
      "learning_rate": 1.3122137453346515e-07,
      "loss": 0.8437,
      "step": 9841
    },
    {
      "epoch": 0.8875862379943185,
      "grad_norm": 2.826412209162835,
      "learning_rate": 1.3101332713670376e-07,
      "loss": 0.8053,
      "step": 9842
    },
    {
      "epoch": 0.8876764215177887,
      "grad_norm": 0.6626496937613404,
      "learning_rate": 1.3080543921219133e-07,
      "loss": 0.6065,
      "step": 9843
    },
    {
      "epoch": 0.8877666050412589,
      "grad_norm": 1.69507607775973,
      "learning_rate": 1.3059771077766478e-07,
      "loss": 0.7379,
      "step": 9844
    },
    {
      "epoch": 0.8878567885647293,
      "grad_norm": 2.6673285636404738,
      "learning_rate": 1.3039014185085018e-07,
      "loss": 0.7846,
      "step": 9845
    },
    {
      "epoch": 0.8879469720881995,
      "grad_norm": 2.1825897692978637,
      "learning_rate": 1.301827324494571e-07,
      "loss": 0.7545,
      "step": 9846
    },
    {
      "epoch": 0.8880371556116697,
      "grad_norm": 2.754255037337585,
      "learning_rate": 1.2997548259118342e-07,
      "loss": 0.7831,
      "step": 9847
    },
    {
      "epoch": 0.88812733913514,
      "grad_norm": 1.8050525510751603,
      "learning_rate": 1.2976839229371272e-07,
      "loss": 0.7775,
      "step": 9848
    },
    {
      "epoch": 0.8882175226586103,
      "grad_norm": 0.6247232062634187,
      "learning_rate": 1.2956146157471515e-07,
      "loss": 0.5957,
      "step": 9849
    },
    {
      "epoch": 0.8883077061820805,
      "grad_norm": 1.577280555128022,
      "learning_rate": 1.2935469045184745e-07,
      "loss": 0.7633,
      "step": 9850
    },
    {
      "epoch": 0.8883978897055508,
      "grad_norm": 2.5261337945137474,
      "learning_rate": 1.291480789427517e-07,
      "loss": 0.7209,
      "step": 9851
    },
    {
      "epoch": 0.888488073229021,
      "grad_norm": 1.6150486757483877,
      "learning_rate": 1.2894162706505807e-07,
      "loss": 0.784,
      "step": 9852
    },
    {
      "epoch": 0.8885782567524914,
      "grad_norm": 1.8199715356382493,
      "learning_rate": 1.2873533483638155e-07,
      "loss": 0.7397,
      "step": 9853
    },
    {
      "epoch": 0.8886684402759616,
      "grad_norm": 1.7046509652283028,
      "learning_rate": 1.285292022743243e-07,
      "loss": 0.6918,
      "step": 9854
    },
    {
      "epoch": 0.8887586237994318,
      "grad_norm": 1.5600860258982916,
      "learning_rate": 1.2832322939647467e-07,
      "loss": 0.6935,
      "step": 9855
    },
    {
      "epoch": 0.8888488073229021,
      "grad_norm": 1.90782856852445,
      "learning_rate": 1.281174162204075e-07,
      "loss": 0.7284,
      "step": 9856
    },
    {
      "epoch": 0.8889389908463724,
      "grad_norm": 1.9801798051053028,
      "learning_rate": 1.2791176276368366e-07,
      "loss": 0.6228,
      "step": 9857
    },
    {
      "epoch": 0.8890291743698426,
      "grad_norm": 1.720775381902812,
      "learning_rate": 1.2770626904385128e-07,
      "loss": 0.7555,
      "step": 9858
    },
    {
      "epoch": 0.8891193578933129,
      "grad_norm": 1.6964026958241774,
      "learning_rate": 1.2750093507844306e-07,
      "loss": 0.6061,
      "step": 9859
    },
    {
      "epoch": 0.8892095414167831,
      "grad_norm": 1.4190966951357697,
      "learning_rate": 1.272957608849805e-07,
      "loss": 0.7811,
      "step": 9860
    },
    {
      "epoch": 0.8892997249402534,
      "grad_norm": 1.4722086178053062,
      "learning_rate": 1.270907464809694e-07,
      "loss": 0.7187,
      "step": 9861
    },
    {
      "epoch": 0.8893899084637237,
      "grad_norm": 1.3992353752507434,
      "learning_rate": 1.2688589188390285e-07,
      "loss": 0.7452,
      "step": 9862
    },
    {
      "epoch": 0.8894800919871939,
      "grad_norm": 1.8540147726299612,
      "learning_rate": 1.2668119711126023e-07,
      "loss": 0.7563,
      "step": 9863
    },
    {
      "epoch": 0.8895702755106643,
      "grad_norm": 4.7638754713180305,
      "learning_rate": 1.2647666218050735e-07,
      "loss": 0.7452,
      "step": 9864
    },
    {
      "epoch": 0.8896604590341345,
      "grad_norm": 2.079273451494557,
      "learning_rate": 1.2627228710909643e-07,
      "loss": 0.7059,
      "step": 9865
    },
    {
      "epoch": 0.8897506425576047,
      "grad_norm": 1.5217622593786502,
      "learning_rate": 1.260680719144649e-07,
      "loss": 0.7113,
      "step": 9866
    },
    {
      "epoch": 0.889840826081075,
      "grad_norm": 3.571615903573436,
      "learning_rate": 1.2586401661403877e-07,
      "loss": 0.675,
      "step": 9867
    },
    {
      "epoch": 0.8899310096045453,
      "grad_norm": 1.8392330379059119,
      "learning_rate": 1.2566012122522817e-07,
      "loss": 0.7218,
      "step": 9868
    },
    {
      "epoch": 0.8900211931280155,
      "grad_norm": 1.872703226365218,
      "learning_rate": 1.254563857654316e-07,
      "loss": 0.7164,
      "step": 9869
    },
    {
      "epoch": 0.8901113766514858,
      "grad_norm": 1.946761361566733,
      "learning_rate": 1.2525281025203205e-07,
      "loss": 0.6339,
      "step": 9870
    },
    {
      "epoch": 0.890201560174956,
      "grad_norm": 4.017218010340308,
      "learning_rate": 1.2504939470240006e-07,
      "loss": 0.7306,
      "step": 9871
    },
    {
      "epoch": 0.8902917436984263,
      "grad_norm": 1.684928630370117,
      "learning_rate": 1.2484613913389196e-07,
      "loss": 0.81,
      "step": 9872
    },
    {
      "epoch": 0.8903819272218966,
      "grad_norm": 1.7403530161753518,
      "learning_rate": 1.2464304356385057e-07,
      "loss": 0.6587,
      "step": 9873
    },
    {
      "epoch": 0.8904721107453668,
      "grad_norm": 2.4116931139118214,
      "learning_rate": 1.2444010800960558e-07,
      "loss": 0.8036,
      "step": 9874
    },
    {
      "epoch": 0.890562294268837,
      "grad_norm": 1.565870769040371,
      "learning_rate": 1.2423733248847267e-07,
      "loss": 0.7296,
      "step": 9875
    },
    {
      "epoch": 0.8906524777923074,
      "grad_norm": 2.592749161256407,
      "learning_rate": 1.2403471701775293e-07,
      "loss": 0.6867,
      "step": 9876
    },
    {
      "epoch": 0.8907426613157776,
      "grad_norm": 1.7156256281394713,
      "learning_rate": 1.2383226161473515e-07,
      "loss": 0.7879,
      "step": 9877
    },
    {
      "epoch": 0.8908328448392479,
      "grad_norm": 7.60203796969764,
      "learning_rate": 1.2362996629669376e-07,
      "loss": 0.8206,
      "step": 9878
    },
    {
      "epoch": 0.8909230283627181,
      "grad_norm": 1.4700075001427966,
      "learning_rate": 1.2342783108089007e-07,
      "loss": 0.7544,
      "step": 9879
    },
    {
      "epoch": 0.8910132118861884,
      "grad_norm": 1.8300344814116256,
      "learning_rate": 1.2322585598457135e-07,
      "loss": 0.8013,
      "step": 9880
    },
    {
      "epoch": 0.8911033954096587,
      "grad_norm": 1.3368098662115617,
      "learning_rate": 1.2302404102497034e-07,
      "loss": 0.7188,
      "step": 9881
    },
    {
      "epoch": 0.8911935789331289,
      "grad_norm": 5.576377787353985,
      "learning_rate": 1.228223862193083e-07,
      "loss": 0.6749,
      "step": 9882
    },
    {
      "epoch": 0.8912837624565991,
      "grad_norm": 2.340826525498899,
      "learning_rate": 1.2262089158479038e-07,
      "loss": 0.7545,
      "step": 9883
    },
    {
      "epoch": 0.8913739459800695,
      "grad_norm": 1.8184840633857207,
      "learning_rate": 1.2241955713861042e-07,
      "loss": 0.7905,
      "step": 9884
    },
    {
      "epoch": 0.8914641295035397,
      "grad_norm": 2.2048387287702593,
      "learning_rate": 1.222183828979464e-07,
      "loss": 0.6879,
      "step": 9885
    },
    {
      "epoch": 0.8915543130270099,
      "grad_norm": 1.4477015365211938,
      "learning_rate": 1.2201736887996372e-07,
      "loss": 0.741,
      "step": 9886
    },
    {
      "epoch": 0.8916444965504803,
      "grad_norm": 1.819033128127406,
      "learning_rate": 1.2181651510181444e-07,
      "loss": 0.8218,
      "step": 9887
    },
    {
      "epoch": 0.8917346800739505,
      "grad_norm": 1.6691827934325334,
      "learning_rate": 1.2161582158063622e-07,
      "loss": 0.7745,
      "step": 9888
    },
    {
      "epoch": 0.8918248635974207,
      "grad_norm": 2.064982682864383,
      "learning_rate": 1.214152883335533e-07,
      "loss": 0.6486,
      "step": 9889
    },
    {
      "epoch": 0.891915047120891,
      "grad_norm": 1.8783714500876536,
      "learning_rate": 1.2121491537767648e-07,
      "loss": 0.6922,
      "step": 9890
    },
    {
      "epoch": 0.8920052306443613,
      "grad_norm": 1.8737356373425418,
      "learning_rate": 1.2101470273010294e-07,
      "loss": 0.7755,
      "step": 9891
    },
    {
      "epoch": 0.8920954141678316,
      "grad_norm": 1.781550264983963,
      "learning_rate": 1.2081465040791528e-07,
      "loss": 0.7222,
      "step": 9892
    },
    {
      "epoch": 0.8921855976913018,
      "grad_norm": 1.6157014072435458,
      "learning_rate": 1.2061475842818335e-07,
      "loss": 0.7098,
      "step": 9893
    },
    {
      "epoch": 0.892275781214772,
      "grad_norm": 2.1117253525348825,
      "learning_rate": 1.2041502680796313e-07,
      "loss": 0.7216,
      "step": 9894
    },
    {
      "epoch": 0.8923659647382424,
      "grad_norm": 1.5456963875184853,
      "learning_rate": 1.2021545556429648e-07,
      "loss": 0.7484,
      "step": 9895
    },
    {
      "epoch": 0.8924561482617126,
      "grad_norm": 1.690977539987142,
      "learning_rate": 1.2001604471421245e-07,
      "loss": 0.7437,
      "step": 9896
    },
    {
      "epoch": 0.8925463317851828,
      "grad_norm": 1.4310365576029729,
      "learning_rate": 1.1981679427472567e-07,
      "loss": 0.7947,
      "step": 9897
    },
    {
      "epoch": 0.8926365153086531,
      "grad_norm": 1.822018981792843,
      "learning_rate": 1.196177042628368e-07,
      "loss": 0.7972,
      "step": 9898
    },
    {
      "epoch": 0.8927266988321234,
      "grad_norm": 2.1254329240400662,
      "learning_rate": 1.194187746955344e-07,
      "loss": 0.7454,
      "step": 9899
    },
    {
      "epoch": 0.8928168823555936,
      "grad_norm": 3.503640869705694,
      "learning_rate": 1.1922000558979094e-07,
      "loss": 0.7201,
      "step": 9900
    },
    {
      "epoch": 0.8929070658790639,
      "grad_norm": 1.7198040591124042,
      "learning_rate": 1.1902139696256752e-07,
      "loss": 0.7763,
      "step": 9901
    },
    {
      "epoch": 0.8929972494025341,
      "grad_norm": 1.9341646218717066,
      "learning_rate": 1.188229488308099e-07,
      "loss": 0.6788,
      "step": 9902
    },
    {
      "epoch": 0.8930874329260045,
      "grad_norm": 0.5679382127111091,
      "learning_rate": 1.1862466121145098e-07,
      "loss": 0.5573,
      "step": 9903
    },
    {
      "epoch": 0.8931776164494747,
      "grad_norm": 2.0262896493071776,
      "learning_rate": 1.184265341214099e-07,
      "loss": 0.6917,
      "step": 9904
    },
    {
      "epoch": 0.8932677999729449,
      "grad_norm": 1.616067323325708,
      "learning_rate": 1.182285675775918e-07,
      "loss": 0.7673,
      "step": 9905
    },
    {
      "epoch": 0.8933579834964152,
      "grad_norm": 1.9262397019236646,
      "learning_rate": 1.1803076159688851e-07,
      "loss": 0.7307,
      "step": 9906
    },
    {
      "epoch": 0.8934481670198855,
      "grad_norm": 1.6032861134137768,
      "learning_rate": 1.1783311619617741e-07,
      "loss": 0.7785,
      "step": 9907
    },
    {
      "epoch": 0.8935383505433557,
      "grad_norm": 1.6700069651201472,
      "learning_rate": 1.1763563139232302e-07,
      "loss": 0.7153,
      "step": 9908
    },
    {
      "epoch": 0.893628534066826,
      "grad_norm": 2.6884124448913034,
      "learning_rate": 1.1743830720217562e-07,
      "loss": 0.6856,
      "step": 9909
    },
    {
      "epoch": 0.8937187175902963,
      "grad_norm": 1.9776535942913946,
      "learning_rate": 1.1724114364257243e-07,
      "loss": 0.7256,
      "step": 9910
    },
    {
      "epoch": 0.8938089011137665,
      "grad_norm": 2.5834990007336933,
      "learning_rate": 1.1704414073033619e-07,
      "loss": 0.7529,
      "step": 9911
    },
    {
      "epoch": 0.8938990846372368,
      "grad_norm": 1.8933179608020942,
      "learning_rate": 1.1684729848227636e-07,
      "loss": 0.7856,
      "step": 9912
    },
    {
      "epoch": 0.893989268160707,
      "grad_norm": 1.9084496236403257,
      "learning_rate": 1.1665061691518884e-07,
      "loss": 0.8034,
      "step": 9913
    },
    {
      "epoch": 0.8940794516841774,
      "grad_norm": 1.684708440894968,
      "learning_rate": 1.1645409604585532e-07,
      "loss": 0.743,
      "step": 9914
    },
    {
      "epoch": 0.8941696352076476,
      "grad_norm": 1.620282514804178,
      "learning_rate": 1.162577358910437e-07,
      "loss": 0.754,
      "step": 9915
    },
    {
      "epoch": 0.8942598187311178,
      "grad_norm": 2.594002033535786,
      "learning_rate": 1.160615364675095e-07,
      "loss": 0.7221,
      "step": 9916
    },
    {
      "epoch": 0.894350002254588,
      "grad_norm": 0.611250875546505,
      "learning_rate": 1.1586549779199262e-07,
      "loss": 0.5596,
      "step": 9917
    },
    {
      "epoch": 0.8944401857780584,
      "grad_norm": 1.9325299043303266,
      "learning_rate": 1.1566961988122037e-07,
      "loss": 0.8039,
      "step": 9918
    },
    {
      "epoch": 0.8945303693015286,
      "grad_norm": 2.076803260319565,
      "learning_rate": 1.1547390275190627e-07,
      "loss": 0.7133,
      "step": 9919
    },
    {
      "epoch": 0.8946205528249989,
      "grad_norm": 1.7855284321662241,
      "learning_rate": 1.1527834642075007e-07,
      "loss": 0.6857,
      "step": 9920
    },
    {
      "epoch": 0.8947107363484691,
      "grad_norm": 5.7395538310857965,
      "learning_rate": 1.1508295090443797e-07,
      "loss": 0.7299,
      "step": 9921
    },
    {
      "epoch": 0.8948009198719394,
      "grad_norm": 1.432185174855645,
      "learning_rate": 1.148877162196411e-07,
      "loss": 0.6966,
      "step": 9922
    },
    {
      "epoch": 0.8948911033954097,
      "grad_norm": 2.3939004039076783,
      "learning_rate": 1.1469264238301924e-07,
      "loss": 0.6785,
      "step": 9923
    },
    {
      "epoch": 0.8949812869188799,
      "grad_norm": 1.7632296357314134,
      "learning_rate": 1.1449772941121638e-07,
      "loss": 0.7689,
      "step": 9924
    },
    {
      "epoch": 0.8950714704423501,
      "grad_norm": 1.8639319861000194,
      "learning_rate": 1.1430297732086369e-07,
      "loss": 0.717,
      "step": 9925
    },
    {
      "epoch": 0.8951616539658205,
      "grad_norm": 1.802025884105804,
      "learning_rate": 1.1410838612857876e-07,
      "loss": 0.7359,
      "step": 9926
    },
    {
      "epoch": 0.8952518374892907,
      "grad_norm": 1.8508241364263303,
      "learning_rate": 1.1391395585096497e-07,
      "loss": 0.8417,
      "step": 9927
    },
    {
      "epoch": 0.895342021012761,
      "grad_norm": 1.8986451463207492,
      "learning_rate": 1.1371968650461216e-07,
      "loss": 0.6751,
      "step": 9928
    },
    {
      "epoch": 0.8954322045362312,
      "grad_norm": 0.7220418018989085,
      "learning_rate": 1.1352557810609687e-07,
      "loss": 0.5721,
      "step": 9929
    },
    {
      "epoch": 0.8955223880597015,
      "grad_norm": 17.610727578100743,
      "learning_rate": 1.1333163067198048e-07,
      "loss": 0.7304,
      "step": 9930
    },
    {
      "epoch": 0.8956125715831718,
      "grad_norm": 1.4886595077621543,
      "learning_rate": 1.1313784421881311e-07,
      "loss": 0.8081,
      "step": 9931
    },
    {
      "epoch": 0.895702755106642,
      "grad_norm": 2.3725692959183142,
      "learning_rate": 1.1294421876312865e-07,
      "loss": 0.7772,
      "step": 9932
    },
    {
      "epoch": 0.8957929386301122,
      "grad_norm": 1.8549463440160383,
      "learning_rate": 1.1275075432144831e-07,
      "loss": 0.7162,
      "step": 9933
    },
    {
      "epoch": 0.8958831221535826,
      "grad_norm": 2.4487545662604187,
      "learning_rate": 1.1255745091028002e-07,
      "loss": 0.7597,
      "step": 9934
    },
    {
      "epoch": 0.8959733056770528,
      "grad_norm": 1.9496339650056527,
      "learning_rate": 1.1236430854611723e-07,
      "loss": 0.7531,
      "step": 9935
    },
    {
      "epoch": 0.896063489200523,
      "grad_norm": 27.04595991555682,
      "learning_rate": 1.1217132724544032e-07,
      "loss": 0.7073,
      "step": 9936
    },
    {
      "epoch": 0.8961536727239934,
      "grad_norm": 3.7835335889968498,
      "learning_rate": 1.1197850702471434e-07,
      "loss": 0.7918,
      "step": 9937
    },
    {
      "epoch": 0.8962438562474636,
      "grad_norm": 1.8028678803278533,
      "learning_rate": 1.1178584790039348e-07,
      "loss": 0.8098,
      "step": 9938
    },
    {
      "epoch": 0.8963340397709338,
      "grad_norm": 1.8014619082897765,
      "learning_rate": 1.1159334988891478e-07,
      "loss": 0.6877,
      "step": 9939
    },
    {
      "epoch": 0.8964242232944041,
      "grad_norm": 1.9431669973875967,
      "learning_rate": 1.1140101300670446e-07,
      "loss": 0.7213,
      "step": 9940
    },
    {
      "epoch": 0.8965144068178744,
      "grad_norm": 1.5681113383069398,
      "learning_rate": 1.1120883727017338e-07,
      "loss": 0.6048,
      "step": 9941
    },
    {
      "epoch": 0.8966045903413447,
      "grad_norm": 1.6141507297268083,
      "learning_rate": 1.1101682269571889e-07,
      "loss": 0.7968,
      "step": 9942
    },
    {
      "epoch": 0.8966947738648149,
      "grad_norm": 2.126449875718704,
      "learning_rate": 1.1082496929972473e-07,
      "loss": 0.7226,
      "step": 9943
    },
    {
      "epoch": 0.8967849573882851,
      "grad_norm": 2.03151126061712,
      "learning_rate": 1.1063327709856096e-07,
      "loss": 0.7176,
      "step": 9944
    },
    {
      "epoch": 0.8968751409117555,
      "grad_norm": 1.8867163490295928,
      "learning_rate": 1.1044174610858403e-07,
      "loss": 0.6283,
      "step": 9945
    },
    {
      "epoch": 0.8969653244352257,
      "grad_norm": 1.8198867103403746,
      "learning_rate": 1.1025037634613643e-07,
      "loss": 0.699,
      "step": 9946
    },
    {
      "epoch": 0.8970555079586959,
      "grad_norm": 3.4500276647370773,
      "learning_rate": 1.1005916782754643e-07,
      "loss": 0.7676,
      "step": 9947
    },
    {
      "epoch": 0.8971456914821662,
      "grad_norm": 7.585731071389967,
      "learning_rate": 1.0986812056912919e-07,
      "loss": 0.7291,
      "step": 9948
    },
    {
      "epoch": 0.8972358750056365,
      "grad_norm": 1.8741163078701397,
      "learning_rate": 1.0967723458718613e-07,
      "loss": 0.6245,
      "step": 9949
    },
    {
      "epoch": 0.8973260585291067,
      "grad_norm": 3.0133231277067942,
      "learning_rate": 1.0948650989800445e-07,
      "loss": 0.7108,
      "step": 9950
    },
    {
      "epoch": 0.897416242052577,
      "grad_norm": 3.011042300005411,
      "learning_rate": 1.0929594651785823e-07,
      "loss": 0.7267,
      "step": 9951
    },
    {
      "epoch": 0.8975064255760472,
      "grad_norm": 1.5638617113250255,
      "learning_rate": 1.0910554446300646e-07,
      "loss": 0.7934,
      "step": 9952
    },
    {
      "epoch": 0.8975966090995176,
      "grad_norm": 1.7386769048051967,
      "learning_rate": 1.089153037496966e-07,
      "loss": 0.642,
      "step": 9953
    },
    {
      "epoch": 0.8976867926229878,
      "grad_norm": 2.935306785445538,
      "learning_rate": 1.0872522439415966e-07,
      "loss": 0.7655,
      "step": 9954
    },
    {
      "epoch": 0.897776976146458,
      "grad_norm": 1.9409930827643016,
      "learning_rate": 1.0853530641261554e-07,
      "loss": 0.6957,
      "step": 9955
    },
    {
      "epoch": 0.8978671596699282,
      "grad_norm": 1.5062523711253137,
      "learning_rate": 1.083455498212682e-07,
      "loss": 0.7565,
      "step": 9956
    },
    {
      "epoch": 0.8979573431933986,
      "grad_norm": 1.7100825154880195,
      "learning_rate": 1.0815595463630911e-07,
      "loss": 0.7787,
      "step": 9957
    },
    {
      "epoch": 0.8980475267168688,
      "grad_norm": 2.0788886022553394,
      "learning_rate": 1.0796652087391556e-07,
      "loss": 0.6139,
      "step": 9958
    },
    {
      "epoch": 0.8981377102403391,
      "grad_norm": 1.5259280460837434,
      "learning_rate": 1.0777724855025083e-07,
      "loss": 0.84,
      "step": 9959
    },
    {
      "epoch": 0.8982278937638094,
      "grad_norm": 1.9060359271807379,
      "learning_rate": 1.075881376814649e-07,
      "loss": 0.7449,
      "step": 9960
    },
    {
      "epoch": 0.8983180772872796,
      "grad_norm": 1.802989144058255,
      "learning_rate": 1.0739918828369377e-07,
      "loss": 0.7975,
      "step": 9961
    },
    {
      "epoch": 0.8984082608107499,
      "grad_norm": 1.6083602669129216,
      "learning_rate": 1.0721040037305983e-07,
      "loss": 0.7501,
      "step": 9962
    },
    {
      "epoch": 0.8984984443342201,
      "grad_norm": 1.8286092393980697,
      "learning_rate": 1.0702177396567114e-07,
      "loss": 0.6924,
      "step": 9963
    },
    {
      "epoch": 0.8985886278576904,
      "grad_norm": 0.663706897442434,
      "learning_rate": 1.0683330907762233e-07,
      "loss": 0.5632,
      "step": 9964
    },
    {
      "epoch": 0.8986788113811607,
      "grad_norm": 1.4315657654637381,
      "learning_rate": 1.0664500572499435e-07,
      "loss": 0.6738,
      "step": 9965
    },
    {
      "epoch": 0.8987689949046309,
      "grad_norm": 1.813518435354795,
      "learning_rate": 1.0645686392385455e-07,
      "loss": 0.6843,
      "step": 9966
    },
    {
      "epoch": 0.8988591784281011,
      "grad_norm": 1.4687139628433028,
      "learning_rate": 1.0626888369025588e-07,
      "loss": 0.6668,
      "step": 9967
    },
    {
      "epoch": 0.8989493619515715,
      "grad_norm": 1.6484168896584401,
      "learning_rate": 1.0608106504023817e-07,
      "loss": 0.7675,
      "step": 9968
    },
    {
      "epoch": 0.8990395454750417,
      "grad_norm": 1.4149199357313578,
      "learning_rate": 1.0589340798982637e-07,
      "loss": 0.7334,
      "step": 9969
    },
    {
      "epoch": 0.899129728998512,
      "grad_norm": 2.5004429424803627,
      "learning_rate": 1.057059125550337e-07,
      "loss": 0.6841,
      "step": 9970
    },
    {
      "epoch": 0.8992199125219822,
      "grad_norm": 1.4252784155860303,
      "learning_rate": 1.0551857875185732e-07,
      "loss": 0.7513,
      "step": 9971
    },
    {
      "epoch": 0.8993100960454525,
      "grad_norm": 1.7472617661220102,
      "learning_rate": 1.0533140659628181e-07,
      "loss": 0.6255,
      "step": 9972
    },
    {
      "epoch": 0.8994002795689228,
      "grad_norm": 1.8957676579205014,
      "learning_rate": 1.0514439610427772e-07,
      "loss": 0.8005,
      "step": 9973
    },
    {
      "epoch": 0.899490463092393,
      "grad_norm": 4.016823675060114,
      "learning_rate": 1.0495754729180206e-07,
      "loss": 0.7687,
      "step": 9974
    },
    {
      "epoch": 0.8995806466158632,
      "grad_norm": 1.536868783039241,
      "learning_rate": 1.0477086017479741e-07,
      "loss": 0.8419,
      "step": 9975
    },
    {
      "epoch": 0.8996708301393336,
      "grad_norm": 2.086547419499394,
      "learning_rate": 1.0458433476919327e-07,
      "loss": 0.6991,
      "step": 9976
    },
    {
      "epoch": 0.8997610136628038,
      "grad_norm": 1.58090674855082,
      "learning_rate": 1.0439797109090509e-07,
      "loss": 0.7696,
      "step": 9977
    },
    {
      "epoch": 0.899851197186274,
      "grad_norm": 2.287154929393946,
      "learning_rate": 1.0421176915583396e-07,
      "loss": 0.7165,
      "step": 9978
    },
    {
      "epoch": 0.8999413807097443,
      "grad_norm": 1.7102736487245918,
      "learning_rate": 1.0402572897986828e-07,
      "loss": 0.774,
      "step": 9979
    },
    {
      "epoch": 0.9000315642332146,
      "grad_norm": 0.5891380888911254,
      "learning_rate": 1.0383985057888134e-07,
      "loss": 0.5682,
      "step": 9980
    },
    {
      "epoch": 0.9001217477566849,
      "grad_norm": 2.828961833417277,
      "learning_rate": 1.036541339687338e-07,
      "loss": 0.647,
      "step": 9981
    },
    {
      "epoch": 0.9002119312801551,
      "grad_norm": 0.651134375011307,
      "learning_rate": 1.0346857916527186e-07,
      "loss": 0.5384,
      "step": 9982
    },
    {
      "epoch": 0.9003021148036254,
      "grad_norm": 1.9460178919417586,
      "learning_rate": 1.0328318618432819e-07,
      "loss": 0.7407,
      "step": 9983
    },
    {
      "epoch": 0.9003922983270957,
      "grad_norm": 1.5820173098103032,
      "learning_rate": 1.0309795504172148e-07,
      "loss": 0.804,
      "step": 9984
    },
    {
      "epoch": 0.9004824818505659,
      "grad_norm": 2.014003722618322,
      "learning_rate": 1.0291288575325685e-07,
      "loss": 0.7192,
      "step": 9985
    },
    {
      "epoch": 0.9005726653740361,
      "grad_norm": 1.6812852933788285,
      "learning_rate": 1.0272797833472502e-07,
      "loss": 0.7064,
      "step": 9986
    },
    {
      "epoch": 0.9006628488975065,
      "grad_norm": 0.6703743254775353,
      "learning_rate": 1.0254323280190335e-07,
      "loss": 0.548,
      "step": 9987
    },
    {
      "epoch": 0.9007530324209767,
      "grad_norm": 2.1997268117000486,
      "learning_rate": 1.023586491705557e-07,
      "loss": 0.7206,
      "step": 9988
    },
    {
      "epoch": 0.9008432159444469,
      "grad_norm": 2.722860539350877,
      "learning_rate": 1.0217422745643145e-07,
      "loss": 0.6826,
      "step": 9989
    },
    {
      "epoch": 0.9009333994679172,
      "grad_norm": 1.9743937177456121,
      "learning_rate": 1.0198996767526691e-07,
      "loss": 0.6701,
      "step": 9990
    },
    {
      "epoch": 0.9010235829913875,
      "grad_norm": 2.0428173818729793,
      "learning_rate": 1.018058698427835e-07,
      "loss": 0.7209,
      "step": 9991
    },
    {
      "epoch": 0.9011137665148578,
      "grad_norm": 1.66965579762992,
      "learning_rate": 1.0162193397469021e-07,
      "loss": 0.7883,
      "step": 9992
    },
    {
      "epoch": 0.901203950038328,
      "grad_norm": 1.724014888269877,
      "learning_rate": 1.0143816008668049e-07,
      "loss": 0.6773,
      "step": 9993
    },
    {
      "epoch": 0.9012941335617982,
      "grad_norm": 0.6407301514089334,
      "learning_rate": 1.0125454819443624e-07,
      "loss": 0.5866,
      "step": 9994
    },
    {
      "epoch": 0.9013843170852686,
      "grad_norm": 2.367221139599571,
      "learning_rate": 1.0107109831362315e-07,
      "loss": 0.6518,
      "step": 9995
    },
    {
      "epoch": 0.9014745006087388,
      "grad_norm": 1.5064915222020583,
      "learning_rate": 1.0088781045989447e-07,
      "loss": 0.7473,
      "step": 9996
    },
    {
      "epoch": 0.901564684132209,
      "grad_norm": 5.905207789267354,
      "learning_rate": 1.0070468464888926e-07,
      "loss": 0.7937,
      "step": 9997
    },
    {
      "epoch": 0.9016548676556793,
      "grad_norm": 2.765027083942011,
      "learning_rate": 1.0052172089623324e-07,
      "loss": 0.7611,
      "step": 9998
    },
    {
      "epoch": 0.9017450511791496,
      "grad_norm": 1.843657022491779,
      "learning_rate": 1.0033891921753746e-07,
      "loss": 0.7776,
      "step": 9999
    },
    {
      "epoch": 0.9018352347026198,
      "grad_norm": 3.6849747109030635,
      "learning_rate": 1.0015627962839968e-07,
      "loss": 0.6554,
      "step": 10000
    },
    {
      "epoch": 0.9019254182260901,
      "grad_norm": 3.747951456657351,
      "learning_rate": 9.99738021444041e-08,
      "loss": 0.7146,
      "step": 10001
    },
    {
      "epoch": 0.9020156017495603,
      "grad_norm": 1.7298476781654912,
      "learning_rate": 9.979148678112003e-08,
      "loss": 0.6728,
      "step": 10002
    },
    {
      "epoch": 0.9021057852730306,
      "grad_norm": 1.8457460105798114,
      "learning_rate": 9.960933355410417e-08,
      "loss": 0.8424,
      "step": 10003
    },
    {
      "epoch": 0.9021959687965009,
      "grad_norm": 1.5615499278309242,
      "learning_rate": 9.942734247889828e-08,
      "loss": 0.6561,
      "step": 10004
    },
    {
      "epoch": 0.9022861523199711,
      "grad_norm": 2.9696212293264783,
      "learning_rate": 9.92455135710315e-08,
      "loss": 0.748,
      "step": 10005
    },
    {
      "epoch": 0.9023763358434415,
      "grad_norm": 2.908293467105427,
      "learning_rate": 9.906384684601787e-08,
      "loss": 0.6741,
      "step": 10006
    },
    {
      "epoch": 0.9024665193669117,
      "grad_norm": 1.6428339302010562,
      "learning_rate": 9.8882342319359e-08,
      "loss": 0.7827,
      "step": 10007
    },
    {
      "epoch": 0.9025567028903819,
      "grad_norm": 1.3607239880911421,
      "learning_rate": 9.870100000654048e-08,
      "loss": 0.6894,
      "step": 10008
    },
    {
      "epoch": 0.9026468864138522,
      "grad_norm": 1.388098103255375,
      "learning_rate": 9.851981992303704e-08,
      "loss": 0.7136,
      "step": 10009
    },
    {
      "epoch": 0.9027370699373225,
      "grad_norm": 1.5689119384589558,
      "learning_rate": 9.833880208430678e-08,
      "loss": 0.7439,
      "step": 10010
    },
    {
      "epoch": 0.9028272534607927,
      "grad_norm": 1.7784403677169691,
      "learning_rate": 9.815794650579601e-08,
      "loss": 0.7354,
      "step": 10011
    },
    {
      "epoch": 0.902917436984263,
      "grad_norm": 1.725229637783135,
      "learning_rate": 9.797725320293548e-08,
      "loss": 0.7152,
      "step": 10012
    },
    {
      "epoch": 0.9030076205077332,
      "grad_norm": 3.6838698541555397,
      "learning_rate": 9.779672219114354e-08,
      "loss": 0.8146,
      "step": 10013
    },
    {
      "epoch": 0.9030978040312035,
      "grad_norm": 2.0806320097722977,
      "learning_rate": 9.761635348582386e-08,
      "loss": 0.7212,
      "step": 10014
    },
    {
      "epoch": 0.9031879875546738,
      "grad_norm": 1.6960240290442046,
      "learning_rate": 9.743614710236658e-08,
      "loss": 0.7722,
      "step": 10015
    },
    {
      "epoch": 0.903278171078144,
      "grad_norm": 1.7033031265481824,
      "learning_rate": 9.725610305614806e-08,
      "loss": 0.6999,
      "step": 10016
    },
    {
      "epoch": 0.9033683546016142,
      "grad_norm": 1.5940760110754768,
      "learning_rate": 9.707622136253002e-08,
      "loss": 0.7329,
      "step": 10017
    },
    {
      "epoch": 0.9034585381250846,
      "grad_norm": 1.6933525463701342,
      "learning_rate": 9.689650203686173e-08,
      "loss": 0.7639,
      "step": 10018
    },
    {
      "epoch": 0.9035487216485548,
      "grad_norm": 2.014309831086462,
      "learning_rate": 9.671694509447715e-08,
      "loss": 0.7176,
      "step": 10019
    },
    {
      "epoch": 0.903638905172025,
      "grad_norm": 2.347301995942629,
      "learning_rate": 9.653755055069757e-08,
      "loss": 0.7571,
      "step": 10020
    },
    {
      "epoch": 0.9037290886954953,
      "grad_norm": 1.269824138014422,
      "learning_rate": 9.635831842082987e-08,
      "loss": 0.7153,
      "step": 10021
    },
    {
      "epoch": 0.9038192722189656,
      "grad_norm": 1.8839217594568531,
      "learning_rate": 9.617924872016691e-08,
      "loss": 0.7513,
      "step": 10022
    },
    {
      "epoch": 0.9039094557424359,
      "grad_norm": 1.793340573786663,
      "learning_rate": 9.600034146398806e-08,
      "loss": 0.7809,
      "step": 10023
    },
    {
      "epoch": 0.9039996392659061,
      "grad_norm": 0.6539114656775874,
      "learning_rate": 9.582159666755863e-08,
      "loss": 0.5817,
      "step": 10024
    },
    {
      "epoch": 0.9040898227893763,
      "grad_norm": 1.4187618171098928,
      "learning_rate": 9.564301434612976e-08,
      "loss": 0.712,
      "step": 10025
    },
    {
      "epoch": 0.9041800063128467,
      "grad_norm": 1.7171486764255661,
      "learning_rate": 9.546459451494015e-08,
      "loss": 0.7831,
      "step": 10026
    },
    {
      "epoch": 0.9042701898363169,
      "grad_norm": 0.6311818293785038,
      "learning_rate": 9.528633718921231e-08,
      "loss": 0.6054,
      "step": 10027
    },
    {
      "epoch": 0.9043603733597871,
      "grad_norm": 1.48419782685446,
      "learning_rate": 9.510824238415672e-08,
      "loss": 0.7895,
      "step": 10028
    },
    {
      "epoch": 0.9044505568832575,
      "grad_norm": 3.131194117612776,
      "learning_rate": 9.493031011496944e-08,
      "loss": 0.7706,
      "step": 10029
    },
    {
      "epoch": 0.9045407404067277,
      "grad_norm": 1.700786139534198,
      "learning_rate": 9.475254039683234e-08,
      "loss": 0.5982,
      "step": 10030
    },
    {
      "epoch": 0.904630923930198,
      "grad_norm": 1.5128351076765765,
      "learning_rate": 9.45749332449144e-08,
      "loss": 0.7059,
      "step": 10031
    },
    {
      "epoch": 0.9047211074536682,
      "grad_norm": 0.6480231282468932,
      "learning_rate": 9.439748867436903e-08,
      "loss": 0.5896,
      "step": 10032
    },
    {
      "epoch": 0.9048112909771385,
      "grad_norm": 2.0503414009335326,
      "learning_rate": 9.42202067003377e-08,
      "loss": 0.7047,
      "step": 10033
    },
    {
      "epoch": 0.9049014745006088,
      "grad_norm": 1.7442484071584052,
      "learning_rate": 9.404308733794652e-08,
      "loss": 0.8742,
      "step": 10034
    },
    {
      "epoch": 0.904991658024079,
      "grad_norm": 1.4454301940836258,
      "learning_rate": 9.38661306023083e-08,
      "loss": 0.7602,
      "step": 10035
    },
    {
      "epoch": 0.9050818415475492,
      "grad_norm": 3.7778010757201606,
      "learning_rate": 9.368933650852229e-08,
      "loss": 0.731,
      "step": 10036
    },
    {
      "epoch": 0.9051720250710196,
      "grad_norm": 0.7060057014999672,
      "learning_rate": 9.351270507167352e-08,
      "loss": 0.5941,
      "step": 10037
    },
    {
      "epoch": 0.9052622085944898,
      "grad_norm": 2.442264302340218,
      "learning_rate": 9.333623630683285e-08,
      "loss": 0.6818,
      "step": 10038
    },
    {
      "epoch": 0.90535239211796,
      "grad_norm": 1.729833344166451,
      "learning_rate": 9.315993022905799e-08,
      "loss": 0.5916,
      "step": 10039
    },
    {
      "epoch": 0.9054425756414303,
      "grad_norm": 2.239117903732719,
      "learning_rate": 9.298378685339158e-08,
      "loss": 0.7193,
      "step": 10040
    },
    {
      "epoch": 0.9055327591649006,
      "grad_norm": 2.352879760445539,
      "learning_rate": 9.280780619486406e-08,
      "loss": 0.8034,
      "step": 10041
    },
    {
      "epoch": 0.9056229426883708,
      "grad_norm": 1.8520070117834182,
      "learning_rate": 9.26319882684905e-08,
      "loss": 0.8219,
      "step": 10042
    },
    {
      "epoch": 0.9057131262118411,
      "grad_norm": 1.7230276109738059,
      "learning_rate": 9.245633308927293e-08,
      "loss": 0.7077,
      "step": 10043
    },
    {
      "epoch": 0.9058033097353113,
      "grad_norm": 1.677708369132338,
      "learning_rate": 9.228084067219888e-08,
      "loss": 0.7216,
      "step": 10044
    },
    {
      "epoch": 0.9058934932587817,
      "grad_norm": 1.5823975692309926,
      "learning_rate": 9.210551103224284e-08,
      "loss": 0.6817,
      "step": 10045
    },
    {
      "epoch": 0.9059836767822519,
      "grad_norm": 1.529223599876882,
      "learning_rate": 9.193034418436463e-08,
      "loss": 0.7918,
      "step": 10046
    },
    {
      "epoch": 0.9060738603057221,
      "grad_norm": 0.6578894300623235,
      "learning_rate": 9.175534014351005e-08,
      "loss": 0.6142,
      "step": 10047
    },
    {
      "epoch": 0.9061640438291924,
      "grad_norm": 2.3137519128297948,
      "learning_rate": 9.158049892461228e-08,
      "loss": 0.6832,
      "step": 10048
    },
    {
      "epoch": 0.9062542273526627,
      "grad_norm": 1.751353353626744,
      "learning_rate": 9.140582054258871e-08,
      "loss": 0.6774,
      "step": 10049
    },
    {
      "epoch": 0.9063444108761329,
      "grad_norm": 1.8273782753750976,
      "learning_rate": 9.123130501234499e-08,
      "loss": 0.749,
      "step": 10050
    },
    {
      "epoch": 0.9064345943996032,
      "grad_norm": 1.7451661652201276,
      "learning_rate": 9.105695234877098e-08,
      "loss": 0.7898,
      "step": 10051
    },
    {
      "epoch": 0.9065247779230734,
      "grad_norm": 5.391452360930278,
      "learning_rate": 9.088276256674344e-08,
      "loss": 0.7636,
      "step": 10052
    },
    {
      "epoch": 0.9066149614465437,
      "grad_norm": 1.4700652381797192,
      "learning_rate": 9.070873568112536e-08,
      "loss": 0.7247,
      "step": 10053
    },
    {
      "epoch": 0.906705144970014,
      "grad_norm": 3.059100697331818,
      "learning_rate": 9.053487170676577e-08,
      "loss": 0.8001,
      "step": 10054
    },
    {
      "epoch": 0.9067953284934842,
      "grad_norm": 4.097113239685195,
      "learning_rate": 9.036117065849968e-08,
      "loss": 0.7996,
      "step": 10055
    },
    {
      "epoch": 0.9068855120169546,
      "grad_norm": 2.1400289332136273,
      "learning_rate": 9.018763255114837e-08,
      "loss": 0.7199,
      "step": 10056
    },
    {
      "epoch": 0.9069756955404248,
      "grad_norm": 1.6123473486983548,
      "learning_rate": 9.00142573995184e-08,
      "loss": 0.712,
      "step": 10057
    },
    {
      "epoch": 0.907065879063895,
      "grad_norm": 11.121367472263696,
      "learning_rate": 8.984104521840375e-08,
      "loss": 0.8176,
      "step": 10058
    },
    {
      "epoch": 0.9071560625873653,
      "grad_norm": 3.2772146933860484,
      "learning_rate": 8.966799602258346e-08,
      "loss": 0.7602,
      "step": 10059
    },
    {
      "epoch": 0.9072462461108356,
      "grad_norm": 1.5705735478320588,
      "learning_rate": 8.949510982682329e-08,
      "loss": 0.7862,
      "step": 10060
    },
    {
      "epoch": 0.9073364296343058,
      "grad_norm": 2.4703290693872284,
      "learning_rate": 8.932238664587499e-08,
      "loss": 0.8082,
      "step": 10061
    },
    {
      "epoch": 0.9074266131577761,
      "grad_norm": 2.4139036848548576,
      "learning_rate": 8.914982649447567e-08,
      "loss": 0.8326,
      "step": 10062
    },
    {
      "epoch": 0.9075167966812463,
      "grad_norm": 2.968615558993311,
      "learning_rate": 8.897742938734975e-08,
      "loss": 0.8075,
      "step": 10063
    },
    {
      "epoch": 0.9076069802047166,
      "grad_norm": 1.8464077689313165,
      "learning_rate": 8.880519533920661e-08,
      "loss": 0.7722,
      "step": 10064
    },
    {
      "epoch": 0.9076971637281869,
      "grad_norm": 1.324739831897294,
      "learning_rate": 8.863312436474268e-08,
      "loss": 0.7436,
      "step": 10065
    },
    {
      "epoch": 0.9077873472516571,
      "grad_norm": 2.110327950433327,
      "learning_rate": 8.846121647863936e-08,
      "loss": 0.7492,
      "step": 10066
    },
    {
      "epoch": 0.9078775307751273,
      "grad_norm": 1.4801388174113883,
      "learning_rate": 8.828947169556555e-08,
      "loss": 0.7435,
      "step": 10067
    },
    {
      "epoch": 0.9079677142985977,
      "grad_norm": 1.4477663396652023,
      "learning_rate": 8.81178900301749e-08,
      "loss": 0.7065,
      "step": 10068
    },
    {
      "epoch": 0.9080578978220679,
      "grad_norm": 1.3873268191196717,
      "learning_rate": 8.794647149710787e-08,
      "loss": 0.654,
      "step": 10069
    },
    {
      "epoch": 0.9081480813455381,
      "grad_norm": 1.7109597958022165,
      "learning_rate": 8.777521611099081e-08,
      "loss": 0.7379,
      "step": 10070
    },
    {
      "epoch": 0.9082382648690084,
      "grad_norm": 1.6024858743696686,
      "learning_rate": 8.760412388643624e-08,
      "loss": 0.6801,
      "step": 10071
    },
    {
      "epoch": 0.9083284483924787,
      "grad_norm": 1.624154310980288,
      "learning_rate": 8.74331948380429e-08,
      "loss": 0.7555,
      "step": 10072
    },
    {
      "epoch": 0.908418631915949,
      "grad_norm": 1.825178265683335,
      "learning_rate": 8.726242898039516e-08,
      "loss": 0.7416,
      "step": 10073
    },
    {
      "epoch": 0.9085088154394192,
      "grad_norm": 2.5436959113291713,
      "learning_rate": 8.709182632806334e-08,
      "loss": 0.7497,
      "step": 10074
    },
    {
      "epoch": 0.9085989989628894,
      "grad_norm": 1.8626923660170218,
      "learning_rate": 8.692138689560469e-08,
      "loss": 0.7498,
      "step": 10075
    },
    {
      "epoch": 0.9086891824863598,
      "grad_norm": 2.044582101891365,
      "learning_rate": 8.675111069756203e-08,
      "loss": 0.7261,
      "step": 10076
    },
    {
      "epoch": 0.90877936600983,
      "grad_norm": 1.6143451720113522,
      "learning_rate": 8.658099774846395e-08,
      "loss": 0.7114,
      "step": 10077
    },
    {
      "epoch": 0.9088695495333002,
      "grad_norm": 1.3916659370389994,
      "learning_rate": 8.641104806282595e-08,
      "loss": 0.682,
      "step": 10078
    },
    {
      "epoch": 0.9089597330567706,
      "grad_norm": 1.9734709313249759,
      "learning_rate": 8.624126165514845e-08,
      "loss": 0.749,
      "step": 10079
    },
    {
      "epoch": 0.9090499165802408,
      "grad_norm": 8.523295624248865,
      "learning_rate": 8.607163853991917e-08,
      "loss": 0.7073,
      "step": 10080
    },
    {
      "epoch": 0.909140100103711,
      "grad_norm": 1.6535993995054545,
      "learning_rate": 8.590217873161054e-08,
      "loss": 0.5869,
      "step": 10081
    },
    {
      "epoch": 0.9092302836271813,
      "grad_norm": 1.598226580682881,
      "learning_rate": 8.573288224468255e-08,
      "loss": 0.7547,
      "step": 10082
    },
    {
      "epoch": 0.9093204671506516,
      "grad_norm": 2.157128307199626,
      "learning_rate": 8.556374909358011e-08,
      "loss": 0.6706,
      "step": 10083
    },
    {
      "epoch": 0.9094106506741219,
      "grad_norm": 1.7471039514517575,
      "learning_rate": 8.539477929273476e-08,
      "loss": 0.7134,
      "step": 10084
    },
    {
      "epoch": 0.9095008341975921,
      "grad_norm": 1.4007456394653968,
      "learning_rate": 8.522597285656386e-08,
      "loss": 0.7712,
      "step": 10085
    },
    {
      "epoch": 0.9095910177210623,
      "grad_norm": 0.6590590529082633,
      "learning_rate": 8.505732979947078e-08,
      "loss": 0.5899,
      "step": 10086
    },
    {
      "epoch": 0.9096812012445327,
      "grad_norm": 1.6553709468421782,
      "learning_rate": 8.488885013584557e-08,
      "loss": 0.7561,
      "step": 10087
    },
    {
      "epoch": 0.9097713847680029,
      "grad_norm": 1.7692182318775873,
      "learning_rate": 8.472053388006295e-08,
      "loss": 0.6819,
      "step": 10088
    },
    {
      "epoch": 0.9098615682914731,
      "grad_norm": 1.8732096071326423,
      "learning_rate": 8.455238104648565e-08,
      "loss": 0.8271,
      "step": 10089
    },
    {
      "epoch": 0.9099517518149434,
      "grad_norm": 2.1893105779519217,
      "learning_rate": 8.438439164946043e-08,
      "loss": 0.6118,
      "step": 10090
    },
    {
      "epoch": 0.9100419353384137,
      "grad_norm": 1.7090504312010661,
      "learning_rate": 8.42165657033218e-08,
      "loss": 0.7531,
      "step": 10091
    },
    {
      "epoch": 0.9101321188618839,
      "grad_norm": 1.9390659157033587,
      "learning_rate": 8.4048903222389e-08,
      "loss": 0.778,
      "step": 10092
    },
    {
      "epoch": 0.9102223023853542,
      "grad_norm": 1.4883305098423905,
      "learning_rate": 8.388140422096856e-08,
      "loss": 0.7928,
      "step": 10093
    },
    {
      "epoch": 0.9103124859088244,
      "grad_norm": 11.822240940675123,
      "learning_rate": 8.371406871335173e-08,
      "loss": 0.7602,
      "step": 10094
    },
    {
      "epoch": 0.9104026694322948,
      "grad_norm": 1.6725341353190293,
      "learning_rate": 8.354689671381732e-08,
      "loss": 0.7411,
      "step": 10095
    },
    {
      "epoch": 0.910492852955765,
      "grad_norm": 1.8561667324103943,
      "learning_rate": 8.337988823662834e-08,
      "loss": 0.7359,
      "step": 10096
    },
    {
      "epoch": 0.9105830364792352,
      "grad_norm": 2.083421997726714,
      "learning_rate": 8.321304329603607e-08,
      "loss": 0.698,
      "step": 10097
    },
    {
      "epoch": 0.9106732200027055,
      "grad_norm": 0.6226802452757548,
      "learning_rate": 8.304636190627557e-08,
      "loss": 0.5865,
      "step": 10098
    },
    {
      "epoch": 0.9107634035261758,
      "grad_norm": 2.3960933599481056,
      "learning_rate": 8.287984408156945e-08,
      "loss": 0.76,
      "step": 10099
    },
    {
      "epoch": 0.910853587049646,
      "grad_norm": 1.6436578238863175,
      "learning_rate": 8.271348983612591e-08,
      "loss": 0.6505,
      "step": 10100
    },
    {
      "epoch": 0.9109437705731163,
      "grad_norm": 1.5175174813936902,
      "learning_rate": 8.254729918413938e-08,
      "loss": 0.7041,
      "step": 10101
    },
    {
      "epoch": 0.9110339540965866,
      "grad_norm": 1.4198801488399302,
      "learning_rate": 8.238127213979006e-08,
      "loss": 0.7126,
      "step": 10102
    },
    {
      "epoch": 0.9111241376200568,
      "grad_norm": 1.947056429080023,
      "learning_rate": 8.221540871724398e-08,
      "loss": 0.809,
      "step": 10103
    },
    {
      "epoch": 0.9112143211435271,
      "grad_norm": 1.6641824168665122,
      "learning_rate": 8.2049708930654e-08,
      "loss": 0.6301,
      "step": 10104
    },
    {
      "epoch": 0.9113045046669973,
      "grad_norm": 1.47540059872812,
      "learning_rate": 8.188417279415793e-08,
      "loss": 0.7061,
      "step": 10105
    },
    {
      "epoch": 0.9113946881904677,
      "grad_norm": 2.1829392982742215,
      "learning_rate": 8.171880032188117e-08,
      "loss": 0.8337,
      "step": 10106
    },
    {
      "epoch": 0.9114848717139379,
      "grad_norm": 1.4915300288922788,
      "learning_rate": 8.155359152793351e-08,
      "loss": 0.7201,
      "step": 10107
    },
    {
      "epoch": 0.9115750552374081,
      "grad_norm": 1.7745576925306974,
      "learning_rate": 8.138854642641147e-08,
      "loss": 0.7335,
      "step": 10108
    },
    {
      "epoch": 0.9116652387608783,
      "grad_norm": 1.6131263640086493,
      "learning_rate": 8.122366503139777e-08,
      "loss": 0.6686,
      "step": 10109
    },
    {
      "epoch": 0.9117554222843487,
      "grad_norm": 1.920657384706786,
      "learning_rate": 8.105894735696117e-08,
      "loss": 0.681,
      "step": 10110
    },
    {
      "epoch": 0.9118456058078189,
      "grad_norm": 2.230331710640762,
      "learning_rate": 8.089439341715576e-08,
      "loss": 0.7217,
      "step": 10111
    },
    {
      "epoch": 0.9119357893312892,
      "grad_norm": 1.7148974156074552,
      "learning_rate": 8.073000322602319e-08,
      "loss": 0.7713,
      "step": 10112
    },
    {
      "epoch": 0.9120259728547594,
      "grad_norm": 1.6520941074528324,
      "learning_rate": 8.056577679758891e-08,
      "loss": 0.7572,
      "step": 10113
    },
    {
      "epoch": 0.9121161563782297,
      "grad_norm": 2.032197701846365,
      "learning_rate": 8.040171414586638e-08,
      "loss": 0.6951,
      "step": 10114
    },
    {
      "epoch": 0.9122063399017,
      "grad_norm": 1.4524770132796714,
      "learning_rate": 8.023781528485419e-08,
      "loss": 0.7048,
      "step": 10115
    },
    {
      "epoch": 0.9122965234251702,
      "grad_norm": 2.074311047815539,
      "learning_rate": 8.00740802285369e-08,
      "loss": 0.7648,
      "step": 10116
    },
    {
      "epoch": 0.9123867069486404,
      "grad_norm": 1.5385027801964615,
      "learning_rate": 7.99105089908858e-08,
      "loss": 0.6065,
      "step": 10117
    },
    {
      "epoch": 0.9124768904721108,
      "grad_norm": 2.049715008348233,
      "learning_rate": 7.974710158585685e-08,
      "loss": 0.7156,
      "step": 10118
    },
    {
      "epoch": 0.912567073995581,
      "grad_norm": 1.822641021233228,
      "learning_rate": 7.958385802739375e-08,
      "loss": 0.766,
      "step": 10119
    },
    {
      "epoch": 0.9126572575190512,
      "grad_norm": 4.70257463751885,
      "learning_rate": 7.942077832942452e-08,
      "loss": 0.7983,
      "step": 10120
    },
    {
      "epoch": 0.9127474410425215,
      "grad_norm": 0.8115873346065262,
      "learning_rate": 7.925786250586508e-08,
      "loss": 0.5817,
      "step": 10121
    },
    {
      "epoch": 0.9128376245659918,
      "grad_norm": 0.6574136544233258,
      "learning_rate": 7.909511057061524e-08,
      "loss": 0.5805,
      "step": 10122
    },
    {
      "epoch": 0.9129278080894621,
      "grad_norm": 2.374675527794205,
      "learning_rate": 7.893252253756234e-08,
      "loss": 0.7258,
      "step": 10123
    },
    {
      "epoch": 0.9130179916129323,
      "grad_norm": 1.9448874135516434,
      "learning_rate": 7.877009842057925e-08,
      "loss": 0.6702,
      "step": 10124
    },
    {
      "epoch": 0.9131081751364026,
      "grad_norm": 1.6209515244932775,
      "learning_rate": 7.860783823352512e-08,
      "loss": 0.7659,
      "step": 10125
    },
    {
      "epoch": 0.9131983586598729,
      "grad_norm": 1.5933340689259723,
      "learning_rate": 7.844574199024445e-08,
      "loss": 0.75,
      "step": 10126
    },
    {
      "epoch": 0.9132885421833431,
      "grad_norm": 1.6022764521936126,
      "learning_rate": 7.82838097045686e-08,
      "loss": 0.7854,
      "step": 10127
    },
    {
      "epoch": 0.9133787257068133,
      "grad_norm": 1.7242159540510649,
      "learning_rate": 7.812204139031454e-08,
      "loss": 0.7336,
      "step": 10128
    },
    {
      "epoch": 0.9134689092302837,
      "grad_norm": 1.9355886937925308,
      "learning_rate": 7.796043706128474e-08,
      "loss": 0.7532,
      "step": 10129
    },
    {
      "epoch": 0.9135590927537539,
      "grad_norm": 1.94321313048976,
      "learning_rate": 7.779899673126844e-08,
      "loss": 0.7679,
      "step": 10130
    },
    {
      "epoch": 0.9136492762772241,
      "grad_norm": 1.5455555538583063,
      "learning_rate": 7.76377204140406e-08,
      "loss": 0.7817,
      "step": 10131
    },
    {
      "epoch": 0.9137394598006944,
      "grad_norm": 1.860039087520607,
      "learning_rate": 7.74766081233622e-08,
      "loss": 0.729,
      "step": 10132
    },
    {
      "epoch": 0.9138296433241647,
      "grad_norm": 1.607697510754755,
      "learning_rate": 7.73156598729805e-08,
      "loss": 0.8075,
      "step": 10133
    },
    {
      "epoch": 0.913919826847635,
      "grad_norm": 1.7741607669978898,
      "learning_rate": 7.715487567662849e-08,
      "loss": 0.7934,
      "step": 10134
    },
    {
      "epoch": 0.9140100103711052,
      "grad_norm": 2.9069754402690005,
      "learning_rate": 7.69942555480243e-08,
      "loss": 0.7087,
      "step": 10135
    },
    {
      "epoch": 0.9141001938945754,
      "grad_norm": 1.741320954771476,
      "learning_rate": 7.68337995008741e-08,
      "loss": 0.8454,
      "step": 10136
    },
    {
      "epoch": 0.9141903774180458,
      "grad_norm": 1.4730091431043275,
      "learning_rate": 7.667350754886803e-08,
      "loss": 0.5965,
      "step": 10137
    },
    {
      "epoch": 0.914280560941516,
      "grad_norm": 3.4128763682978094,
      "learning_rate": 7.651337970568361e-08,
      "loss": 0.7344,
      "step": 10138
    },
    {
      "epoch": 0.9143707444649862,
      "grad_norm": 1.6330733387510128,
      "learning_rate": 7.635341598498368e-08,
      "loss": 0.7052,
      "step": 10139
    },
    {
      "epoch": 0.9144609279884565,
      "grad_norm": 2.391449929317947,
      "learning_rate": 7.61936164004171e-08,
      "loss": 0.7376,
      "step": 10140
    },
    {
      "epoch": 0.9145511115119268,
      "grad_norm": 1.5511886868590101,
      "learning_rate": 7.603398096561875e-08,
      "loss": 0.7878,
      "step": 10141
    },
    {
      "epoch": 0.914641295035397,
      "grad_norm": 1.7728665142683304,
      "learning_rate": 7.587450969420994e-08,
      "loss": 0.7479,
      "step": 10142
    },
    {
      "epoch": 0.9147314785588673,
      "grad_norm": 0.6780831266914432,
      "learning_rate": 7.571520259979757e-08,
      "loss": 0.5607,
      "step": 10143
    },
    {
      "epoch": 0.9148216620823375,
      "grad_norm": 1.9668185735753778,
      "learning_rate": 7.555605969597455e-08,
      "loss": 0.5784,
      "step": 10144
    },
    {
      "epoch": 0.9149118456058078,
      "grad_norm": 1.7150915570745953,
      "learning_rate": 7.539708099631959e-08,
      "loss": 0.8246,
      "step": 10145
    },
    {
      "epoch": 0.9150020291292781,
      "grad_norm": 1.494242107460504,
      "learning_rate": 7.52382665143978e-08,
      "loss": 0.7469,
      "step": 10146
    },
    {
      "epoch": 0.9150922126527483,
      "grad_norm": 1.7897678633894891,
      "learning_rate": 7.507961626376014e-08,
      "loss": 0.7967,
      "step": 10147
    },
    {
      "epoch": 0.9151823961762187,
      "grad_norm": 1.714831237052155,
      "learning_rate": 7.492113025794378e-08,
      "loss": 0.7794,
      "step": 10148
    },
    {
      "epoch": 0.9152725796996889,
      "grad_norm": 1.6389472739098394,
      "learning_rate": 7.476280851047101e-08,
      "loss": 0.6894,
      "step": 10149
    },
    {
      "epoch": 0.9153627632231591,
      "grad_norm": 1.1950329721762147,
      "learning_rate": 7.460465103485125e-08,
      "loss": 0.7622,
      "step": 10150
    },
    {
      "epoch": 0.9154529467466294,
      "grad_norm": 1.7998701260655239,
      "learning_rate": 7.444665784457948e-08,
      "loss": 0.7323,
      "step": 10151
    },
    {
      "epoch": 0.9155431302700997,
      "grad_norm": 2.048968111243397,
      "learning_rate": 7.42888289531356e-08,
      "loss": 0.698,
      "step": 10152
    },
    {
      "epoch": 0.9156333137935699,
      "grad_norm": 1.583525453589982,
      "learning_rate": 7.41311643739877e-08,
      "loss": 0.6621,
      "step": 10153
    },
    {
      "epoch": 0.9157234973170402,
      "grad_norm": 1.5532338616579457,
      "learning_rate": 7.39736641205877e-08,
      "loss": 0.7812,
      "step": 10154
    },
    {
      "epoch": 0.9158136808405104,
      "grad_norm": 1.7062411018111612,
      "learning_rate": 7.381632820637462e-08,
      "loss": 0.7649,
      "step": 10155
    },
    {
      "epoch": 0.9159038643639807,
      "grad_norm": 1.8126211454748908,
      "learning_rate": 7.365915664477352e-08,
      "loss": 0.7325,
      "step": 10156
    },
    {
      "epoch": 0.915994047887451,
      "grad_norm": 1.845447211714376,
      "learning_rate": 7.350214944919474e-08,
      "loss": 0.7455,
      "step": 10157
    },
    {
      "epoch": 0.9160842314109212,
      "grad_norm": 1.8209946299792676,
      "learning_rate": 7.334530663303539e-08,
      "loss": 0.7739,
      "step": 10158
    },
    {
      "epoch": 0.9161744149343914,
      "grad_norm": 1.9670905127943434,
      "learning_rate": 7.318862820967742e-08,
      "loss": 0.75,
      "step": 10159
    },
    {
      "epoch": 0.9162645984578618,
      "grad_norm": 1.6085909813314718,
      "learning_rate": 7.303211419249056e-08,
      "loss": 0.7703,
      "step": 10160
    },
    {
      "epoch": 0.916354781981332,
      "grad_norm": 1.615431417234369,
      "learning_rate": 7.287576459482858e-08,
      "loss": 0.7577,
      "step": 10161
    },
    {
      "epoch": 0.9164449655048023,
      "grad_norm": 1.926823270278433,
      "learning_rate": 7.271957943003259e-08,
      "loss": 0.7793,
      "step": 10162
    },
    {
      "epoch": 0.9165351490282725,
      "grad_norm": 1.92230877206739,
      "learning_rate": 7.256355871142883e-08,
      "loss": 0.7401,
      "step": 10163
    },
    {
      "epoch": 0.9166253325517428,
      "grad_norm": 2.736212499927058,
      "learning_rate": 7.240770245233019e-08,
      "loss": 0.7935,
      "step": 10164
    },
    {
      "epoch": 0.9167155160752131,
      "grad_norm": 1.5745429753446794,
      "learning_rate": 7.225201066603492e-08,
      "loss": 0.6541,
      "step": 10165
    },
    {
      "epoch": 0.9168056995986833,
      "grad_norm": 1.7494678369374608,
      "learning_rate": 7.209648336582774e-08,
      "loss": 0.8492,
      "step": 10166
    },
    {
      "epoch": 0.9168958831221535,
      "grad_norm": 2.121321467475305,
      "learning_rate": 7.19411205649787e-08,
      "loss": 0.726,
      "step": 10167
    },
    {
      "epoch": 0.9169860666456239,
      "grad_norm": 1.861621045083243,
      "learning_rate": 7.178592227674474e-08,
      "loss": 0.7644,
      "step": 10168
    },
    {
      "epoch": 0.9170762501690941,
      "grad_norm": 1.9528220590255834,
      "learning_rate": 7.163088851436771e-08,
      "loss": 0.6553,
      "step": 10169
    },
    {
      "epoch": 0.9171664336925643,
      "grad_norm": 1.5117748161387343,
      "learning_rate": 7.147601929107639e-08,
      "loss": 0.7061,
      "step": 10170
    },
    {
      "epoch": 0.9172566172160346,
      "grad_norm": 1.8043370422662492,
      "learning_rate": 7.132131462008461e-08,
      "loss": 0.7913,
      "step": 10171
    },
    {
      "epoch": 0.9173468007395049,
      "grad_norm": 1.704832288710357,
      "learning_rate": 7.116677451459297e-08,
      "loss": 0.7468,
      "step": 10172
    },
    {
      "epoch": 0.9174369842629752,
      "grad_norm": 1.3182735612985732,
      "learning_rate": 7.101239898778799e-08,
      "loss": 0.7645,
      "step": 10173
    },
    {
      "epoch": 0.9175271677864454,
      "grad_norm": 2.261585712637907,
      "learning_rate": 7.085818805284094e-08,
      "loss": 0.7778,
      "step": 10174
    },
    {
      "epoch": 0.9176173513099157,
      "grad_norm": 1.7345045574015094,
      "learning_rate": 7.070414172291083e-08,
      "loss": 0.8227,
      "step": 10175
    },
    {
      "epoch": 0.917707534833386,
      "grad_norm": 1.3759941928402764,
      "learning_rate": 7.055026001114095e-08,
      "loss": 0.6866,
      "step": 10176
    },
    {
      "epoch": 0.9177977183568562,
      "grad_norm": 0.6451019072794656,
      "learning_rate": 7.039654293066211e-08,
      "loss": 0.5658,
      "step": 10177
    },
    {
      "epoch": 0.9178879018803264,
      "grad_norm": 1.72037540220314,
      "learning_rate": 7.024299049459003e-08,
      "loss": 0.6629,
      "step": 10178
    },
    {
      "epoch": 0.9179780854037968,
      "grad_norm": 2.1460147027954535,
      "learning_rate": 7.008960271602627e-08,
      "loss": 0.6579,
      "step": 10179
    },
    {
      "epoch": 0.918068268927267,
      "grad_norm": 1.8221648782272213,
      "learning_rate": 6.993637960805921e-08,
      "loss": 0.7064,
      "step": 10180
    },
    {
      "epoch": 0.9181584524507372,
      "grad_norm": 1.8056433322143568,
      "learning_rate": 6.97833211837624e-08,
      "loss": 0.7664,
      "step": 10181
    },
    {
      "epoch": 0.9182486359742075,
      "grad_norm": 2.227856176565968,
      "learning_rate": 6.963042745619562e-08,
      "loss": 0.6879,
      "step": 10182
    },
    {
      "epoch": 0.9183388194976778,
      "grad_norm": 1.7226979776116476,
      "learning_rate": 6.947769843840511e-08,
      "loss": 0.7096,
      "step": 10183
    },
    {
      "epoch": 0.918429003021148,
      "grad_norm": 1.7909425410644215,
      "learning_rate": 6.9325134143422e-08,
      "loss": 0.7731,
      "step": 10184
    },
    {
      "epoch": 0.9185191865446183,
      "grad_norm": 1.8104622665767216,
      "learning_rate": 6.917273458426387e-08,
      "loss": 0.7228,
      "step": 10185
    },
    {
      "epoch": 0.9186093700680885,
      "grad_norm": 0.6868450122496921,
      "learning_rate": 6.902049977393476e-08,
      "loss": 0.5541,
      "step": 10186
    },
    {
      "epoch": 0.9186995535915589,
      "grad_norm": 2.838333672885423,
      "learning_rate": 6.886842972542362e-08,
      "loss": 0.7465,
      "step": 10187
    },
    {
      "epoch": 0.9187897371150291,
      "grad_norm": 1.5145886210786987,
      "learning_rate": 6.871652445170672e-08,
      "loss": 0.6299,
      "step": 10188
    },
    {
      "epoch": 0.9188799206384993,
      "grad_norm": 1.6008433146101866,
      "learning_rate": 6.856478396574416e-08,
      "loss": 0.8267,
      "step": 10189
    },
    {
      "epoch": 0.9189701041619696,
      "grad_norm": 3.0932066469175106,
      "learning_rate": 6.841320828048491e-08,
      "loss": 0.8012,
      "step": 10190
    },
    {
      "epoch": 0.9190602876854399,
      "grad_norm": 1.7535647461236372,
      "learning_rate": 6.826179740886062e-08,
      "loss": 0.738,
      "step": 10191
    },
    {
      "epoch": 0.9191504712089101,
      "grad_norm": 1.6276627512242892,
      "learning_rate": 6.811055136379184e-08,
      "loss": 0.8058,
      "step": 10192
    },
    {
      "epoch": 0.9192406547323804,
      "grad_norm": 3.3941655812557165,
      "learning_rate": 6.79594701581827e-08,
      "loss": 0.7182,
      "step": 10193
    },
    {
      "epoch": 0.9193308382558506,
      "grad_norm": 1.7065387076350593,
      "learning_rate": 6.780855380492511e-08,
      "loss": 0.7478,
      "step": 10194
    },
    {
      "epoch": 0.919421021779321,
      "grad_norm": 2.6771850467660405,
      "learning_rate": 6.765780231689544e-08,
      "loss": 0.7546,
      "step": 10195
    },
    {
      "epoch": 0.9195112053027912,
      "grad_norm": 2.1390989112402194,
      "learning_rate": 6.750721570695695e-08,
      "loss": 0.7967,
      "step": 10196
    },
    {
      "epoch": 0.9196013888262614,
      "grad_norm": 3.150010844620702,
      "learning_rate": 6.735679398795868e-08,
      "loss": 0.7249,
      "step": 10197
    },
    {
      "epoch": 0.9196915723497318,
      "grad_norm": 3.209736768916171,
      "learning_rate": 6.720653717273506e-08,
      "loss": 0.6805,
      "step": 10198
    },
    {
      "epoch": 0.919781755873202,
      "grad_norm": 1.5663483340241935,
      "learning_rate": 6.705644527410714e-08,
      "loss": 0.7422,
      "step": 10199
    },
    {
      "epoch": 0.9198719393966722,
      "grad_norm": 1.966154961638492,
      "learning_rate": 6.690651830488136e-08,
      "loss": 0.7099,
      "step": 10200
    },
    {
      "epoch": 0.9199621229201425,
      "grad_norm": 2.354851738379152,
      "learning_rate": 6.675675627785037e-08,
      "loss": 0.7401,
      "step": 10201
    },
    {
      "epoch": 0.9200523064436128,
      "grad_norm": 4.750634201275768,
      "learning_rate": 6.660715920579263e-08,
      "loss": 0.7603,
      "step": 10202
    },
    {
      "epoch": 0.920142489967083,
      "grad_norm": 0.5940027953621565,
      "learning_rate": 6.645772710147279e-08,
      "loss": 0.5773,
      "step": 10203
    },
    {
      "epoch": 0.9202326734905533,
      "grad_norm": 1.2817653456964042,
      "learning_rate": 6.630845997764112e-08,
      "loss": 0.7156,
      "step": 10204
    },
    {
      "epoch": 0.9203228570140235,
      "grad_norm": 1.8001776819482216,
      "learning_rate": 6.615935784703409e-08,
      "loss": 0.6833,
      "step": 10205
    },
    {
      "epoch": 0.9204130405374938,
      "grad_norm": 1.664495707255701,
      "learning_rate": 6.601042072237328e-08,
      "loss": 0.6673,
      "step": 10206
    },
    {
      "epoch": 0.9205032240609641,
      "grad_norm": 1.767338149498865,
      "learning_rate": 6.586164861636767e-08,
      "loss": 0.7692,
      "step": 10207
    },
    {
      "epoch": 0.9205934075844343,
      "grad_norm": 2.131454650360685,
      "learning_rate": 6.571304154171065e-08,
      "loss": 0.6563,
      "step": 10208
    },
    {
      "epoch": 0.9206835911079045,
      "grad_norm": 2.020406212517751,
      "learning_rate": 6.556459951108273e-08,
      "loss": 0.7623,
      "step": 10209
    },
    {
      "epoch": 0.9207737746313749,
      "grad_norm": 2.054367098466691,
      "learning_rate": 6.541632253714957e-08,
      "loss": 0.7227,
      "step": 10210
    },
    {
      "epoch": 0.9208639581548451,
      "grad_norm": 1.8904634517031378,
      "learning_rate": 6.526821063256261e-08,
      "loss": 0.6692,
      "step": 10211
    },
    {
      "epoch": 0.9209541416783154,
      "grad_norm": 2.672480107546534,
      "learning_rate": 6.512026380996016e-08,
      "loss": 0.6523,
      "step": 10212
    },
    {
      "epoch": 0.9210443252017856,
      "grad_norm": 1.6917028717651634,
      "learning_rate": 6.49724820819657e-08,
      "loss": 0.7153,
      "step": 10213
    },
    {
      "epoch": 0.9211345087252559,
      "grad_norm": 1.7490367963441535,
      "learning_rate": 6.48248654611887e-08,
      "loss": 0.7742,
      "step": 10214
    },
    {
      "epoch": 0.9212246922487262,
      "grad_norm": 2.020499906529059,
      "learning_rate": 6.467741396022419e-08,
      "loss": 0.7892,
      "step": 10215
    },
    {
      "epoch": 0.9213148757721964,
      "grad_norm": 1.8611954184394817,
      "learning_rate": 6.453012759165455e-08,
      "loss": 0.7511,
      "step": 10216
    },
    {
      "epoch": 0.9214050592956666,
      "grad_norm": 2.564224686220878,
      "learning_rate": 6.438300636804639e-08,
      "loss": 0.7065,
      "step": 10217
    },
    {
      "epoch": 0.921495242819137,
      "grad_norm": 1.6772174505150654,
      "learning_rate": 6.423605030195278e-08,
      "loss": 0.7767,
      "step": 10218
    },
    {
      "epoch": 0.9215854263426072,
      "grad_norm": 2.266480698623455,
      "learning_rate": 6.408925940591304e-08,
      "loss": 0.8174,
      "step": 10219
    },
    {
      "epoch": 0.9216756098660774,
      "grad_norm": 1.6556753205212162,
      "learning_rate": 6.394263369245222e-08,
      "loss": 0.6729,
      "step": 10220
    },
    {
      "epoch": 0.9217657933895478,
      "grad_norm": 1.7346565774076375,
      "learning_rate": 6.379617317408126e-08,
      "loss": 0.7463,
      "step": 10221
    },
    {
      "epoch": 0.921855976913018,
      "grad_norm": 1.9201782924601571,
      "learning_rate": 6.364987786329723e-08,
      "loss": 0.731,
      "step": 10222
    },
    {
      "epoch": 0.9219461604364882,
      "grad_norm": 1.6591627312074997,
      "learning_rate": 6.350374777258193e-08,
      "loss": 0.7364,
      "step": 10223
    },
    {
      "epoch": 0.9220363439599585,
      "grad_norm": 1.6919067910811225,
      "learning_rate": 6.335778291440519e-08,
      "loss": 0.7206,
      "step": 10224
    },
    {
      "epoch": 0.9221265274834288,
      "grad_norm": 1.9600491278193963,
      "learning_rate": 6.321198330122057e-08,
      "loss": 0.6492,
      "step": 10225
    },
    {
      "epoch": 0.9222167110068991,
      "grad_norm": 1.7003643514968156,
      "learning_rate": 6.306634894546902e-08,
      "loss": 0.6758,
      "step": 10226
    },
    {
      "epoch": 0.9223068945303693,
      "grad_norm": 1.5207539716245735,
      "learning_rate": 6.292087985957661e-08,
      "loss": 0.6808,
      "step": 10227
    },
    {
      "epoch": 0.9223970780538395,
      "grad_norm": 2.2418081780145553,
      "learning_rate": 6.277557605595585e-08,
      "loss": 0.7821,
      "step": 10228
    },
    {
      "epoch": 0.9224872615773099,
      "grad_norm": 1.3234320810470521,
      "learning_rate": 6.263043754700481e-08,
      "loss": 0.6669,
      "step": 10229
    },
    {
      "epoch": 0.9225774451007801,
      "grad_norm": 1.7381701487493733,
      "learning_rate": 6.248546434510671e-08,
      "loss": 0.7403,
      "step": 10230
    },
    {
      "epoch": 0.9226676286242503,
      "grad_norm": 1.8067570829230213,
      "learning_rate": 6.234065646263298e-08,
      "loss": 0.6738,
      "step": 10231
    },
    {
      "epoch": 0.9227578121477206,
      "grad_norm": 1.8878232514925046,
      "learning_rate": 6.219601391193796e-08,
      "loss": 0.647,
      "step": 10232
    },
    {
      "epoch": 0.9228479956711909,
      "grad_norm": 2.2614929402739006,
      "learning_rate": 6.205153670536423e-08,
      "loss": 0.6678,
      "step": 10233
    },
    {
      "epoch": 0.9229381791946611,
      "grad_norm": 1.5892839125597005,
      "learning_rate": 6.190722485523902e-08,
      "loss": 0.7758,
      "step": 10234
    },
    {
      "epoch": 0.9230283627181314,
      "grad_norm": 1.4959777308963829,
      "learning_rate": 6.176307837387607e-08,
      "loss": 0.7166,
      "step": 10235
    },
    {
      "epoch": 0.9231185462416016,
      "grad_norm": 1.6951785408218987,
      "learning_rate": 6.16190972735744e-08,
      "loss": 0.6894,
      "step": 10236
    },
    {
      "epoch": 0.923208729765072,
      "grad_norm": 1.6419411602420166,
      "learning_rate": 6.147528156661974e-08,
      "loss": 0.7679,
      "step": 10237
    },
    {
      "epoch": 0.9232989132885422,
      "grad_norm": 1.9168401998937499,
      "learning_rate": 6.133163126528273e-08,
      "loss": 0.6613,
      "step": 10238
    },
    {
      "epoch": 0.9233890968120124,
      "grad_norm": 1.8954786311187557,
      "learning_rate": 6.11881463818209e-08,
      "loss": 0.7461,
      "step": 10239
    },
    {
      "epoch": 0.9234792803354827,
      "grad_norm": 1.980130198658082,
      "learning_rate": 6.104482692847668e-08,
      "loss": 0.7311,
      "step": 10240
    },
    {
      "epoch": 0.923569463858953,
      "grad_norm": 1.5446890596493412,
      "learning_rate": 6.090167291747917e-08,
      "loss": 0.7072,
      "step": 10241
    },
    {
      "epoch": 0.9236596473824232,
      "grad_norm": 12.802583457586854,
      "learning_rate": 6.075868436104303e-08,
      "loss": 0.6382,
      "step": 10242
    },
    {
      "epoch": 0.9237498309058935,
      "grad_norm": 1.8267719445670485,
      "learning_rate": 6.061586127136875e-08,
      "loss": 0.7287,
      "step": 10243
    },
    {
      "epoch": 0.9238400144293638,
      "grad_norm": 1.5313403823024903,
      "learning_rate": 6.047320366064324e-08,
      "loss": 0.7058,
      "step": 10244
    },
    {
      "epoch": 0.923930197952834,
      "grad_norm": 1.671801342633983,
      "learning_rate": 6.033071154103786e-08,
      "loss": 0.7499,
      "step": 10245
    },
    {
      "epoch": 0.9240203814763043,
      "grad_norm": 1.6145223134313218,
      "learning_rate": 6.018838492471178e-08,
      "loss": 0.7604,
      "step": 10246
    },
    {
      "epoch": 0.9241105649997745,
      "grad_norm": 0.620269468337918,
      "learning_rate": 6.00462238238082e-08,
      "loss": 0.5799,
      "step": 10247
    },
    {
      "epoch": 0.9242007485232449,
      "grad_norm": 2.407934218689972,
      "learning_rate": 5.990422825045827e-08,
      "loss": 0.7054,
      "step": 10248
    },
    {
      "epoch": 0.9242909320467151,
      "grad_norm": 1.6405412925074074,
      "learning_rate": 5.976239821677675e-08,
      "loss": 0.7194,
      "step": 10249
    },
    {
      "epoch": 0.9243811155701853,
      "grad_norm": 6.011365933441599,
      "learning_rate": 5.962073373486598e-08,
      "loss": 0.6826,
      "step": 10250
    },
    {
      "epoch": 0.9244712990936556,
      "grad_norm": 1.4167248863524031,
      "learning_rate": 5.947923481681316e-08,
      "loss": 0.7027,
      "step": 10251
    },
    {
      "epoch": 0.9245614826171259,
      "grad_norm": 1.5555101834598088,
      "learning_rate": 5.933790147469198e-08,
      "loss": 0.7114,
      "step": 10252
    },
    {
      "epoch": 0.9246516661405961,
      "grad_norm": 2.9667210120320417,
      "learning_rate": 5.9196733720561665e-08,
      "loss": 0.6923,
      "step": 10253
    },
    {
      "epoch": 0.9247418496640664,
      "grad_norm": 3.7566831164492105,
      "learning_rate": 5.905573156646793e-08,
      "loss": 0.8244,
      "step": 10254
    },
    {
      "epoch": 0.9248320331875366,
      "grad_norm": 2.0725295377865085,
      "learning_rate": 5.8914895024441134e-08,
      "loss": 0.8157,
      "step": 10255
    },
    {
      "epoch": 0.9249222167110069,
      "grad_norm": 4.894237025399466,
      "learning_rate": 5.877422410649857e-08,
      "loss": 0.7716,
      "step": 10256
    },
    {
      "epoch": 0.9250124002344772,
      "grad_norm": 1.8016861567941345,
      "learning_rate": 5.863371882464285e-08,
      "loss": 0.7575,
      "step": 10257
    },
    {
      "epoch": 0.9251025837579474,
      "grad_norm": 20.519386614876705,
      "learning_rate": 5.849337919086283e-08,
      "loss": 0.7196,
      "step": 10258
    },
    {
      "epoch": 0.9251927672814176,
      "grad_norm": 2.691531152138524,
      "learning_rate": 5.835320521713316e-08,
      "loss": 0.7174,
      "step": 10259
    },
    {
      "epoch": 0.925282950804888,
      "grad_norm": 1.5884076504901827,
      "learning_rate": 5.8213196915414264e-08,
      "loss": 0.7248,
      "step": 10260
    },
    {
      "epoch": 0.9253731343283582,
      "grad_norm": 2.8987077859467147,
      "learning_rate": 5.807335429765237e-08,
      "loss": 0.795,
      "step": 10261
    },
    {
      "epoch": 0.9254633178518284,
      "grad_norm": 1.8905258152694115,
      "learning_rate": 5.7933677375779034e-08,
      "loss": 0.7891,
      "step": 10262
    },
    {
      "epoch": 0.9255535013752987,
      "grad_norm": 1.55723026456528,
      "learning_rate": 5.77941661617134e-08,
      "loss": 0.7918,
      "step": 10263
    },
    {
      "epoch": 0.925643684898769,
      "grad_norm": 1.7537645992179836,
      "learning_rate": 5.765482066735816e-08,
      "loss": 0.6745,
      "step": 10264
    },
    {
      "epoch": 0.9257338684222393,
      "grad_norm": 2.1344094358824717,
      "learning_rate": 5.7515640904604256e-08,
      "loss": 0.744,
      "step": 10265
    },
    {
      "epoch": 0.9258240519457095,
      "grad_norm": 1.6543569722794824,
      "learning_rate": 5.7376626885326187e-08,
      "loss": 0.7265,
      "step": 10266
    },
    {
      "epoch": 0.9259142354691798,
      "grad_norm": 1.6043476729226065,
      "learning_rate": 5.723777862138601e-08,
      "loss": 0.8096,
      "step": 10267
    },
    {
      "epoch": 0.9260044189926501,
      "grad_norm": 2.0512895807867753,
      "learning_rate": 5.7099096124630705e-08,
      "loss": 0.8202,
      "step": 10268
    },
    {
      "epoch": 0.9260946025161203,
      "grad_norm": 1.2720153449980316,
      "learning_rate": 5.696057940689347e-08,
      "loss": 0.7492,
      "step": 10269
    },
    {
      "epoch": 0.9261847860395905,
      "grad_norm": 2.039637108296184,
      "learning_rate": 5.6822228479993736e-08,
      "loss": 0.7625,
      "step": 10270
    },
    {
      "epoch": 0.9262749695630609,
      "grad_norm": 2.867943271740973,
      "learning_rate": 5.668404335573584e-08,
      "loss": 0.7818,
      "step": 10271
    },
    {
      "epoch": 0.9263651530865311,
      "grad_norm": 1.612171521108567,
      "learning_rate": 5.654602404591058e-08,
      "loss": 0.6857,
      "step": 10272
    },
    {
      "epoch": 0.9264553366100013,
      "grad_norm": 3.091004203159186,
      "learning_rate": 5.640817056229474e-08,
      "loss": 0.7986,
      "step": 10273
    },
    {
      "epoch": 0.9265455201334716,
      "grad_norm": 1.4968462214560505,
      "learning_rate": 5.6270482916650706e-08,
      "loss": 0.7841,
      "step": 10274
    },
    {
      "epoch": 0.9266357036569419,
      "grad_norm": 1.6384501256025201,
      "learning_rate": 5.613296112072663e-08,
      "loss": 0.8174,
      "step": 10275
    },
    {
      "epoch": 0.9267258871804122,
      "grad_norm": 6.044554587993753,
      "learning_rate": 5.59956051862569e-08,
      "loss": 0.7479,
      "step": 10276
    },
    {
      "epoch": 0.9268160707038824,
      "grad_norm": 1.6864831474541122,
      "learning_rate": 5.585841512496081e-08,
      "loss": 0.8059,
      "step": 10277
    },
    {
      "epoch": 0.9269062542273526,
      "grad_norm": 2.8807693312199203,
      "learning_rate": 5.5721390948545e-08,
      "loss": 0.8498,
      "step": 10278
    },
    {
      "epoch": 0.926996437750823,
      "grad_norm": 1.8198210089562852,
      "learning_rate": 5.558453266870056e-08,
      "loss": 0.7858,
      "step": 10279
    },
    {
      "epoch": 0.9270866212742932,
      "grad_norm": 1.8804611148845636,
      "learning_rate": 5.544784029710525e-08,
      "loss": 0.6835,
      "step": 10280
    },
    {
      "epoch": 0.9271768047977634,
      "grad_norm": 3.056153552946018,
      "learning_rate": 5.531131384542242e-08,
      "loss": 0.8152,
      "step": 10281
    },
    {
      "epoch": 0.9272669883212337,
      "grad_norm": 1.442457523877538,
      "learning_rate": 5.51749533253012e-08,
      "loss": 0.6921,
      "step": 10282
    },
    {
      "epoch": 0.927357171844704,
      "grad_norm": 1.892254718118945,
      "learning_rate": 5.503875874837649e-08,
      "loss": 0.6841,
      "step": 10283
    },
    {
      "epoch": 0.9274473553681742,
      "grad_norm": 1.6368964560822319,
      "learning_rate": 5.4902730126269225e-08,
      "loss": 0.7077,
      "step": 10284
    },
    {
      "epoch": 0.9275375388916445,
      "grad_norm": 1.9767986654087997,
      "learning_rate": 5.476686747058656e-08,
      "loss": 0.7782,
      "step": 10285
    },
    {
      "epoch": 0.9276277224151147,
      "grad_norm": 2.1450254777582325,
      "learning_rate": 5.4631170792920124e-08,
      "loss": 0.7199,
      "step": 10286
    },
    {
      "epoch": 0.927717905938585,
      "grad_norm": 1.792535534217862,
      "learning_rate": 5.449564010484953e-08,
      "loss": 0.7187,
      "step": 10287
    },
    {
      "epoch": 0.9278080894620553,
      "grad_norm": 2.1646480938571186,
      "learning_rate": 5.436027541793775e-08,
      "loss": 0.767,
      "step": 10288
    },
    {
      "epoch": 0.9278982729855255,
      "grad_norm": 1.671393423153366,
      "learning_rate": 5.4225076743735554e-08,
      "loss": 0.7193,
      "step": 10289
    },
    {
      "epoch": 0.9279884565089958,
      "grad_norm": 1.5318293351363437,
      "learning_rate": 5.409004409377882e-08,
      "loss": 0.6649,
      "step": 10290
    },
    {
      "epoch": 0.9280786400324661,
      "grad_norm": 1.5609700914556524,
      "learning_rate": 5.3955177479589e-08,
      "loss": 0.6672,
      "step": 10291
    },
    {
      "epoch": 0.9281688235559363,
      "grad_norm": 11.256284865558857,
      "learning_rate": 5.3820476912674e-08,
      "loss": 0.748,
      "step": 10292
    },
    {
      "epoch": 0.9282590070794066,
      "grad_norm": 1.494000676401335,
      "learning_rate": 5.3685942404527063e-08,
      "loss": 0.7697,
      "step": 10293
    },
    {
      "epoch": 0.9283491906028769,
      "grad_norm": 1.8823219388361678,
      "learning_rate": 5.355157396662702e-08,
      "loss": 0.7226,
      "step": 10294
    },
    {
      "epoch": 0.9284393741263471,
      "grad_norm": 1.5368275947597423,
      "learning_rate": 5.34173716104398e-08,
      "loss": 0.7257,
      "step": 10295
    },
    {
      "epoch": 0.9285295576498174,
      "grad_norm": 1.8164323702339003,
      "learning_rate": 5.328333534741536e-08,
      "loss": 0.7517,
      "step": 10296
    },
    {
      "epoch": 0.9286197411732876,
      "grad_norm": 3.236537634192432,
      "learning_rate": 5.314946518899099e-08,
      "loss": 0.7115,
      "step": 10297
    },
    {
      "epoch": 0.928709924696758,
      "grad_norm": 1.7972771236099105,
      "learning_rate": 5.301576114658912e-08,
      "loss": 0.8106,
      "step": 10298
    },
    {
      "epoch": 0.9288001082202282,
      "grad_norm": 2.284871297170808,
      "learning_rate": 5.288222323161795e-08,
      "loss": 0.77,
      "step": 10299
    },
    {
      "epoch": 0.9288902917436984,
      "grad_norm": 2.124882414313367,
      "learning_rate": 5.274885145547214e-08,
      "loss": 0.7939,
      "step": 10300
    },
    {
      "epoch": 0.9289804752671686,
      "grad_norm": 2.1237361740012517,
      "learning_rate": 5.261564582953082e-08,
      "loss": 0.7339,
      "step": 10301
    },
    {
      "epoch": 0.929070658790639,
      "grad_norm": 2.141246844554921,
      "learning_rate": 5.248260636516066e-08,
      "loss": 0.7202,
      "step": 10302
    },
    {
      "epoch": 0.9291608423141092,
      "grad_norm": 1.9415950568204998,
      "learning_rate": 5.2349733073712824e-08,
      "loss": 0.6897,
      "step": 10303
    },
    {
      "epoch": 0.9292510258375795,
      "grad_norm": 1.5568977631781136,
      "learning_rate": 5.221702596652533e-08,
      "loss": 0.8305,
      "step": 10304
    },
    {
      "epoch": 0.9293412093610497,
      "grad_norm": 2.523570810109452,
      "learning_rate": 5.208448505492091e-08,
      "loss": 0.6886,
      "step": 10305
    },
    {
      "epoch": 0.92943139288452,
      "grad_norm": 1.6920204368962308,
      "learning_rate": 5.1952110350208965e-08,
      "loss": 0.7246,
      "step": 10306
    },
    {
      "epoch": 0.9295215764079903,
      "grad_norm": 1.8001505311965,
      "learning_rate": 5.181990186368446e-08,
      "loss": 0.6975,
      "step": 10307
    },
    {
      "epoch": 0.9296117599314605,
      "grad_norm": 1.7763141710151304,
      "learning_rate": 5.1687859606627915e-08,
      "loss": 0.8376,
      "step": 10308
    },
    {
      "epoch": 0.9297019434549307,
      "grad_norm": 2.1570611063967884,
      "learning_rate": 5.1555983590306327e-08,
      "loss": 0.653,
      "step": 10309
    },
    {
      "epoch": 0.9297921269784011,
      "grad_norm": 0.6560112882332002,
      "learning_rate": 5.1424273825971806e-08,
      "loss": 0.5785,
      "step": 10310
    },
    {
      "epoch": 0.9298823105018713,
      "grad_norm": 1.7284477608499786,
      "learning_rate": 5.1292730324862475e-08,
      "loss": 0.7021,
      "step": 10311
    },
    {
      "epoch": 0.9299724940253415,
      "grad_norm": 1.5548757314550716,
      "learning_rate": 5.116135309820224e-08,
      "loss": 0.7573,
      "step": 10312
    },
    {
      "epoch": 0.9300626775488118,
      "grad_norm": 1.9448470818978423,
      "learning_rate": 5.103014215720147e-08,
      "loss": 0.7744,
      "step": 10313
    },
    {
      "epoch": 0.9301528610722821,
      "grad_norm": 1.7589068587790848,
      "learning_rate": 5.0899097513055214e-08,
      "loss": 0.7986,
      "step": 10314
    },
    {
      "epoch": 0.9302430445957524,
      "grad_norm": 1.6799912288981764,
      "learning_rate": 5.076821917694563e-08,
      "loss": 0.73,
      "step": 10315
    },
    {
      "epoch": 0.9303332281192226,
      "grad_norm": 1.8659311321343466,
      "learning_rate": 5.063750716003889e-08,
      "loss": 0.7882,
      "step": 10316
    },
    {
      "epoch": 0.9304234116426929,
      "grad_norm": 3.8007950503186,
      "learning_rate": 5.050696147348921e-08,
      "loss": 0.725,
      "step": 10317
    },
    {
      "epoch": 0.9305135951661632,
      "grad_norm": 1.9819140455621636,
      "learning_rate": 5.037658212843454e-08,
      "loss": 0.7133,
      "step": 10318
    },
    {
      "epoch": 0.9306037786896334,
      "grad_norm": 1.457697248247251,
      "learning_rate": 5.0246369136000444e-08,
      "loss": 0.6321,
      "step": 10319
    },
    {
      "epoch": 0.9306939622131036,
      "grad_norm": 1.6127563805028902,
      "learning_rate": 5.011632250729691e-08,
      "loss": 0.8191,
      "step": 10320
    },
    {
      "epoch": 0.930784145736574,
      "grad_norm": 2.164322159265606,
      "learning_rate": 4.998644225342019e-08,
      "loss": 0.8322,
      "step": 10321
    },
    {
      "epoch": 0.9308743292600442,
      "grad_norm": 1.9301407625325406,
      "learning_rate": 4.9856728385452296e-08,
      "loss": 0.7444,
      "step": 10322
    },
    {
      "epoch": 0.9309645127835144,
      "grad_norm": 1.8356358057054085,
      "learning_rate": 4.9727180914461485e-08,
      "loss": 0.8316,
      "step": 10323
    },
    {
      "epoch": 0.9310546963069847,
      "grad_norm": 1.9371306774939663,
      "learning_rate": 4.959779985150137e-08,
      "loss": 0.7511,
      "step": 10324
    },
    {
      "epoch": 0.931144879830455,
      "grad_norm": 1.9850403278587838,
      "learning_rate": 4.9468585207611105e-08,
      "loss": 0.6936,
      "step": 10325
    },
    {
      "epoch": 0.9312350633539253,
      "grad_norm": 1.8668593114264578,
      "learning_rate": 4.9339536993816764e-08,
      "loss": 0.7205,
      "step": 10326
    },
    {
      "epoch": 0.9313252468773955,
      "grad_norm": 1.7280021637620449,
      "learning_rate": 4.921065522112844e-08,
      "loss": 0.6937,
      "step": 10327
    },
    {
      "epoch": 0.9314154304008657,
      "grad_norm": 1.5885567669593734,
      "learning_rate": 4.908193990054377e-08,
      "loss": 0.7865,
      "step": 10328
    },
    {
      "epoch": 0.9315056139243361,
      "grad_norm": 1.6347573848685322,
      "learning_rate": 4.89533910430453e-08,
      "loss": 0.7579,
      "step": 10329
    },
    {
      "epoch": 0.9315957974478063,
      "grad_norm": 1.4752191154532377,
      "learning_rate": 4.8825008659601376e-08,
      "loss": 0.7318,
      "step": 10330
    },
    {
      "epoch": 0.9316859809712765,
      "grad_norm": 1.8904082749418518,
      "learning_rate": 4.869679276116634e-08,
      "loss": 0.7517,
      "step": 10331
    },
    {
      "epoch": 0.9317761644947468,
      "grad_norm": 1.492482101358665,
      "learning_rate": 4.856874335868055e-08,
      "loss": 0.7558,
      "step": 10332
    },
    {
      "epoch": 0.9318663480182171,
      "grad_norm": 2.6986985319458903,
      "learning_rate": 4.844086046306928e-08,
      "loss": 0.8481,
      "step": 10333
    },
    {
      "epoch": 0.9319565315416873,
      "grad_norm": 1.6472966734665067,
      "learning_rate": 4.8313144085244896e-08,
      "loss": 0.74,
      "step": 10334
    },
    {
      "epoch": 0.9320467150651576,
      "grad_norm": 0.6498171136400057,
      "learning_rate": 4.818559423610424e-08,
      "loss": 0.5746,
      "step": 10335
    },
    {
      "epoch": 0.9321368985886278,
      "grad_norm": 1.6110906252526889,
      "learning_rate": 4.8058210926531284e-08,
      "loss": 0.6853,
      "step": 10336
    },
    {
      "epoch": 0.9322270821120981,
      "grad_norm": 1.5233570566702126,
      "learning_rate": 4.7930994167394435e-08,
      "loss": 0.7766,
      "step": 10337
    },
    {
      "epoch": 0.9323172656355684,
      "grad_norm": 1.504583746164301,
      "learning_rate": 4.7803943969548786e-08,
      "loss": 0.7859,
      "step": 10338
    },
    {
      "epoch": 0.9324074491590386,
      "grad_norm": 1.9236056591313202,
      "learning_rate": 4.7677060343834784e-08,
      "loss": 0.6879,
      "step": 10339
    },
    {
      "epoch": 0.932497632682509,
      "grad_norm": 1.7384241624012278,
      "learning_rate": 4.75503433010791e-08,
      "loss": 0.7361,
      "step": 10340
    },
    {
      "epoch": 0.9325878162059792,
      "grad_norm": 2.335851457216808,
      "learning_rate": 4.742379285209419e-08,
      "loss": 0.7263,
      "step": 10341
    },
    {
      "epoch": 0.9326779997294494,
      "grad_norm": 2.9161711317271903,
      "learning_rate": 4.72974090076772e-08,
      "loss": 0.8408,
      "step": 10342
    },
    {
      "epoch": 0.9327681832529197,
      "grad_norm": 1.4658208817818266,
      "learning_rate": 4.717119177861262e-08,
      "loss": 0.7003,
      "step": 10343
    },
    {
      "epoch": 0.93285836677639,
      "grad_norm": 1.779460677511776,
      "learning_rate": 4.70451411756696e-08,
      "loss": 0.7576,
      "step": 10344
    },
    {
      "epoch": 0.9329485502998602,
      "grad_norm": 1.6395312341784403,
      "learning_rate": 4.691925720960355e-08,
      "loss": 0.7799,
      "step": 10345
    },
    {
      "epoch": 0.9330387338233305,
      "grad_norm": 2.133945262088931,
      "learning_rate": 4.6793539891155645e-08,
      "loss": 0.7691,
      "step": 10346
    },
    {
      "epoch": 0.9331289173468007,
      "grad_norm": 3.285434053447986,
      "learning_rate": 4.6667989231052864e-08,
      "loss": 0.7234,
      "step": 10347
    },
    {
      "epoch": 0.933219100870271,
      "grad_norm": 1.7229867907840821,
      "learning_rate": 4.654260524000797e-08,
      "loss": 0.755,
      "step": 10348
    },
    {
      "epoch": 0.9333092843937413,
      "grad_norm": 1.56817344618059,
      "learning_rate": 4.6417387928719076e-08,
      "loss": 0.7307,
      "step": 10349
    },
    {
      "epoch": 0.9333994679172115,
      "grad_norm": 1.7758946138809042,
      "learning_rate": 4.629233730787052e-08,
      "loss": 0.6154,
      "step": 10350
    },
    {
      "epoch": 0.9334896514406817,
      "grad_norm": 5.983697041089314,
      "learning_rate": 4.616745338813266e-08,
      "loss": 0.7049,
      "step": 10351
    },
    {
      "epoch": 0.9335798349641521,
      "grad_norm": 2.073866877191085,
      "learning_rate": 4.6042736180160744e-08,
      "loss": 0.7592,
      "step": 10352
    },
    {
      "epoch": 0.9336700184876223,
      "grad_norm": 1.923842996452673,
      "learning_rate": 4.591818569459671e-08,
      "loss": 0.8596,
      "step": 10353
    },
    {
      "epoch": 0.9337602020110926,
      "grad_norm": 1.635778407364854,
      "learning_rate": 4.5793801942067614e-08,
      "loss": 0.7645,
      "step": 10354
    },
    {
      "epoch": 0.9338503855345628,
      "grad_norm": 2.4120831816328323,
      "learning_rate": 4.566958493318673e-08,
      "loss": 0.6128,
      "step": 10355
    },
    {
      "epoch": 0.9339405690580331,
      "grad_norm": 0.605535318960522,
      "learning_rate": 4.554553467855316e-08,
      "loss": 0.5991,
      "step": 10356
    },
    {
      "epoch": 0.9340307525815034,
      "grad_norm": 6.319829028750439,
      "learning_rate": 4.5421651188751074e-08,
      "loss": 0.7609,
      "step": 10357
    },
    {
      "epoch": 0.9341209361049736,
      "grad_norm": 2.84741695534144,
      "learning_rate": 4.529793447435137e-08,
      "loss": 0.7551,
      "step": 10358
    },
    {
      "epoch": 0.9342111196284438,
      "grad_norm": 2.4635679688504633,
      "learning_rate": 4.5174384545909824e-08,
      "loss": 0.7453,
      "step": 10359
    },
    {
      "epoch": 0.9343013031519142,
      "grad_norm": 2.384862399500395,
      "learning_rate": 4.505100141396867e-08,
      "loss": 0.7855,
      "step": 10360
    },
    {
      "epoch": 0.9343914866753844,
      "grad_norm": 2.1198813554565166,
      "learning_rate": 4.492778508905548e-08,
      "loss": 0.7803,
      "step": 10361
    },
    {
      "epoch": 0.9344816701988546,
      "grad_norm": 1.6827642605667097,
      "learning_rate": 4.480473558168385e-08,
      "loss": 0.6821,
      "step": 10362
    },
    {
      "epoch": 0.934571853722325,
      "grad_norm": 5.149439897911911,
      "learning_rate": 4.4681852902352936e-08,
      "loss": 0.6461,
      "step": 10363
    },
    {
      "epoch": 0.9346620372457952,
      "grad_norm": 1.8402471727749312,
      "learning_rate": 4.455913706154812e-08,
      "loss": 0.6639,
      "step": 10364
    },
    {
      "epoch": 0.9347522207692655,
      "grad_norm": 1.8403376019748534,
      "learning_rate": 4.443658806973949e-08,
      "loss": 0.7672,
      "step": 10365
    },
    {
      "epoch": 0.9348424042927357,
      "grad_norm": 0.6668038947211347,
      "learning_rate": 4.431420593738444e-08,
      "loss": 0.6012,
      "step": 10366
    },
    {
      "epoch": 0.934932587816206,
      "grad_norm": 1.8609023032395196,
      "learning_rate": 4.419199067492485e-08,
      "loss": 0.6551,
      "step": 10367
    },
    {
      "epoch": 0.9350227713396763,
      "grad_norm": 1.5949103675219736,
      "learning_rate": 4.4069942292788596e-08,
      "loss": 0.7489,
      "step": 10368
    },
    {
      "epoch": 0.9351129548631465,
      "grad_norm": 1.784144609479167,
      "learning_rate": 4.39480608013898e-08,
      "loss": 0.7479,
      "step": 10369
    },
    {
      "epoch": 0.9352031383866167,
      "grad_norm": 1.8320431576730223,
      "learning_rate": 4.3826346211128126e-08,
      "loss": 0.8332,
      "step": 10370
    },
    {
      "epoch": 0.9352933219100871,
      "grad_norm": 1.683586740147276,
      "learning_rate": 4.370479853238884e-08,
      "loss": 0.7473,
      "step": 10371
    },
    {
      "epoch": 0.9353835054335573,
      "grad_norm": 1.6185405433674784,
      "learning_rate": 4.3583417775542756e-08,
      "loss": 0.706,
      "step": 10372
    },
    {
      "epoch": 0.9354736889570275,
      "grad_norm": 1.9927486452158596,
      "learning_rate": 4.3462203950947575e-08,
      "loss": 0.6935,
      "step": 10373
    },
    {
      "epoch": 0.9355638724804978,
      "grad_norm": 1.8676488458025018,
      "learning_rate": 4.3341157068944814e-08,
      "loss": 0.7291,
      "step": 10374
    },
    {
      "epoch": 0.9356540560039681,
      "grad_norm": 1.7322714481179327,
      "learning_rate": 4.322027713986376e-08,
      "loss": 0.7574,
      "step": 10375
    },
    {
      "epoch": 0.9357442395274383,
      "grad_norm": 1.912077454406696,
      "learning_rate": 4.309956417401816e-08,
      "loss": 0.8135,
      "step": 10376
    },
    {
      "epoch": 0.9358344230509086,
      "grad_norm": 1.831062825502749,
      "learning_rate": 4.297901818170801e-08,
      "loss": 0.7128,
      "step": 10377
    },
    {
      "epoch": 0.9359246065743788,
      "grad_norm": 1.659780015387024,
      "learning_rate": 4.285863917321886e-08,
      "loss": 0.673,
      "step": 10378
    },
    {
      "epoch": 0.9360147900978492,
      "grad_norm": 1.943958531322012,
      "learning_rate": 4.2738427158822253e-08,
      "loss": 0.6816,
      "step": 10379
    },
    {
      "epoch": 0.9361049736213194,
      "grad_norm": 1.6434414621157776,
      "learning_rate": 4.261838214877511e-08,
      "loss": 0.7539,
      "step": 10380
    },
    {
      "epoch": 0.9361951571447896,
      "grad_norm": 2.1056341838777923,
      "learning_rate": 4.249850415332079e-08,
      "loss": 0.7489,
      "step": 10381
    },
    {
      "epoch": 0.9362853406682599,
      "grad_norm": 1.7858656928043597,
      "learning_rate": 4.237879318268756e-08,
      "loss": 0.8012,
      "step": 10382
    },
    {
      "epoch": 0.9363755241917302,
      "grad_norm": 1.482108624599734,
      "learning_rate": 4.225924924708968e-08,
      "loss": 0.6722,
      "step": 10383
    },
    {
      "epoch": 0.9364657077152004,
      "grad_norm": 2.9195967171065855,
      "learning_rate": 4.2139872356727665e-08,
      "loss": 0.7558,
      "step": 10384
    },
    {
      "epoch": 0.9365558912386707,
      "grad_norm": 1.7444305349849674,
      "learning_rate": 4.202066252178738e-08,
      "loss": 0.7077,
      "step": 10385
    },
    {
      "epoch": 0.936646074762141,
      "grad_norm": 1.745580542326402,
      "learning_rate": 4.1901619752440445e-08,
      "loss": 0.8365,
      "step": 10386
    },
    {
      "epoch": 0.9367362582856112,
      "grad_norm": 1.5501137937797926,
      "learning_rate": 4.178274405884363e-08,
      "loss": 0.7416,
      "step": 10387
    },
    {
      "epoch": 0.9368264418090815,
      "grad_norm": 2.2904487032006213,
      "learning_rate": 4.166403545114105e-08,
      "loss": 0.7511,
      "step": 10388
    },
    {
      "epoch": 0.9369166253325517,
      "grad_norm": 1.7328483804852537,
      "learning_rate": 4.154549393946083e-08,
      "loss": 0.7976,
      "step": 10389
    },
    {
      "epoch": 0.937006808856022,
      "grad_norm": 2.3828804381644226,
      "learning_rate": 4.14271195339182e-08,
      "loss": 0.7746,
      "step": 10390
    },
    {
      "epoch": 0.9370969923794923,
      "grad_norm": 2.032664114882403,
      "learning_rate": 4.1308912244613084e-08,
      "loss": 0.7101,
      "step": 10391
    },
    {
      "epoch": 0.9371871759029625,
      "grad_norm": 1.7418335696228664,
      "learning_rate": 4.1190872081631636e-08,
      "loss": 0.6996,
      "step": 10392
    },
    {
      "epoch": 0.9372773594264328,
      "grad_norm": 1.7432764866582544,
      "learning_rate": 4.107299905504558e-08,
      "loss": 0.8099,
      "step": 10393
    },
    {
      "epoch": 0.9373675429499031,
      "grad_norm": 1.5867137042988537,
      "learning_rate": 4.095529317491286e-08,
      "loss": 0.7387,
      "step": 10394
    },
    {
      "epoch": 0.9374577264733733,
      "grad_norm": 1.8340470990879634,
      "learning_rate": 4.0837754451276575e-08,
      "loss": 0.7506,
      "step": 10395
    },
    {
      "epoch": 0.9375479099968436,
      "grad_norm": 2.508420496172864,
      "learning_rate": 4.072038289416557e-08,
      "loss": 0.7307,
      "step": 10396
    },
    {
      "epoch": 0.9376380935203138,
      "grad_norm": 1.526899190141929,
      "learning_rate": 4.0603178513595185e-08,
      "loss": 0.7434,
      "step": 10397
    },
    {
      "epoch": 0.9377282770437841,
      "grad_norm": 2.9663342736156944,
      "learning_rate": 4.0486141319565624e-08,
      "loss": 0.7512,
      "step": 10398
    },
    {
      "epoch": 0.9378184605672544,
      "grad_norm": 1.8408071016601553,
      "learning_rate": 4.0369271322062916e-08,
      "loss": 0.686,
      "step": 10399
    },
    {
      "epoch": 0.9379086440907246,
      "grad_norm": 1.709892058060896,
      "learning_rate": 4.0252568531059295e-08,
      "loss": 0.7647,
      "step": 10400
    },
    {
      "epoch": 0.9379988276141948,
      "grad_norm": 10.446913808282128,
      "learning_rate": 4.013603295651235e-08,
      "loss": 0.7437,
      "step": 10401
    },
    {
      "epoch": 0.9380890111376652,
      "grad_norm": 1.59670492263124,
      "learning_rate": 4.001966460836592e-08,
      "loss": 0.7732,
      "step": 10402
    },
    {
      "epoch": 0.9381791946611354,
      "grad_norm": 1.7046646018723768,
      "learning_rate": 3.990346349654894e-08,
      "loss": 0.8196,
      "step": 10403
    },
    {
      "epoch": 0.9382693781846057,
      "grad_norm": 1.6221066906620238,
      "learning_rate": 3.9787429630975924e-08,
      "loss": 0.7178,
      "step": 10404
    },
    {
      "epoch": 0.9383595617080759,
      "grad_norm": 2.767369985998312,
      "learning_rate": 3.967156302154828e-08,
      "loss": 0.7322,
      "step": 10405
    },
    {
      "epoch": 0.9384497452315462,
      "grad_norm": 3.8504385710768365,
      "learning_rate": 3.955586367815189e-08,
      "loss": 0.7004,
      "step": 10406
    },
    {
      "epoch": 0.9385399287550165,
      "grad_norm": 1.7259290404822996,
      "learning_rate": 3.944033161065907e-08,
      "loss": 0.8125,
      "step": 10407
    },
    {
      "epoch": 0.9386301122784867,
      "grad_norm": 1.5786796577108104,
      "learning_rate": 3.93249668289275e-08,
      "loss": 0.6277,
      "step": 10408
    },
    {
      "epoch": 0.9387202958019569,
      "grad_norm": 1.5362252354934383,
      "learning_rate": 3.920976934280063e-08,
      "loss": 0.6706,
      "step": 10409
    },
    {
      "epoch": 0.9388104793254273,
      "grad_norm": 1.7435164237051552,
      "learning_rate": 3.909473916210815e-08,
      "loss": 0.8267,
      "step": 10410
    },
    {
      "epoch": 0.9389006628488975,
      "grad_norm": 3.3303110885796103,
      "learning_rate": 3.897987629666488e-08,
      "loss": 0.7004,
      "step": 10411
    },
    {
      "epoch": 0.9389908463723677,
      "grad_norm": 1.5055403909627236,
      "learning_rate": 3.886518075627143e-08,
      "loss": 0.8152,
      "step": 10412
    },
    {
      "epoch": 0.9390810298958381,
      "grad_norm": 2.1890320998374264,
      "learning_rate": 3.875065255071419e-08,
      "loss": 0.7445,
      "step": 10413
    },
    {
      "epoch": 0.9391712134193083,
      "grad_norm": 2.18147275820153,
      "learning_rate": 3.863629168976579e-08,
      "loss": 0.7165,
      "step": 10414
    },
    {
      "epoch": 0.9392613969427785,
      "grad_norm": 1.7718893006839265,
      "learning_rate": 3.852209818318375e-08,
      "loss": 0.7989,
      "step": 10415
    },
    {
      "epoch": 0.9393515804662488,
      "grad_norm": 12.323202835505063,
      "learning_rate": 3.840807204071161e-08,
      "loss": 0.6505,
      "step": 10416
    },
    {
      "epoch": 0.9394417639897191,
      "grad_norm": 1.6350645751333361,
      "learning_rate": 3.829421327207894e-08,
      "loss": 0.7259,
      "step": 10417
    },
    {
      "epoch": 0.9395319475131894,
      "grad_norm": 1.57013652062166,
      "learning_rate": 3.8180521887000825e-08,
      "loss": 0.6983,
      "step": 10418
    },
    {
      "epoch": 0.9396221310366596,
      "grad_norm": 0.5915721554781802,
      "learning_rate": 3.806699789517775e-08,
      "loss": 0.61,
      "step": 10419
    },
    {
      "epoch": 0.9397123145601298,
      "grad_norm": 1.5615393164843048,
      "learning_rate": 3.7953641306296635e-08,
      "loss": 0.7259,
      "step": 10420
    },
    {
      "epoch": 0.9398024980836002,
      "grad_norm": 1.7261535022014212,
      "learning_rate": 3.784045213002951e-08,
      "loss": 0.5902,
      "step": 10421
    },
    {
      "epoch": 0.9398926816070704,
      "grad_norm": 2.0025756463515556,
      "learning_rate": 3.7727430376033986e-08,
      "loss": 0.7571,
      "step": 10422
    },
    {
      "epoch": 0.9399828651305406,
      "grad_norm": 1.682093117475776,
      "learning_rate": 3.7614576053954126e-08,
      "loss": 0.7475,
      "step": 10423
    },
    {
      "epoch": 0.9400730486540109,
      "grad_norm": 1.6846318065089436,
      "learning_rate": 3.75018891734189e-08,
      "loss": 0.7397,
      "step": 10424
    },
    {
      "epoch": 0.9401632321774812,
      "grad_norm": 1.7974707901110634,
      "learning_rate": 3.738936974404372e-08,
      "loss": 0.7905,
      "step": 10425
    },
    {
      "epoch": 0.9402534157009514,
      "grad_norm": 1.693714003514733,
      "learning_rate": 3.7277017775429354e-08,
      "loss": 0.8071,
      "step": 10426
    },
    {
      "epoch": 0.9403435992244217,
      "grad_norm": 1.5505987487563555,
      "learning_rate": 3.7164833277162136e-08,
      "loss": 0.7681,
      "step": 10427
    },
    {
      "epoch": 0.9404337827478919,
      "grad_norm": 1.3566327113198904,
      "learning_rate": 3.705281625881418e-08,
      "loss": 0.7246,
      "step": 10428
    },
    {
      "epoch": 0.9405239662713623,
      "grad_norm": 1.6081891181341386,
      "learning_rate": 3.694096672994362e-08,
      "loss": 0.6529,
      "step": 10429
    },
    {
      "epoch": 0.9406141497948325,
      "grad_norm": 1.5016181482965159,
      "learning_rate": 3.682928470009394e-08,
      "loss": 0.7985,
      "step": 10430
    },
    {
      "epoch": 0.9407043333183027,
      "grad_norm": 2.4337906612375377,
      "learning_rate": 3.6717770178794406e-08,
      "loss": 0.705,
      "step": 10431
    },
    {
      "epoch": 0.940794516841773,
      "grad_norm": 1.4060176657439167,
      "learning_rate": 3.6606423175560287e-08,
      "loss": 0.8492,
      "step": 10432
    },
    {
      "epoch": 0.9408847003652433,
      "grad_norm": 0.6257891679470035,
      "learning_rate": 3.649524369989221e-08,
      "loss": 0.5463,
      "step": 10433
    },
    {
      "epoch": 0.9409748838887135,
      "grad_norm": 2.8590040956503215,
      "learning_rate": 3.638423176127636e-08,
      "loss": 0.7769,
      "step": 10434
    },
    {
      "epoch": 0.9410650674121838,
      "grad_norm": 1.4823858485411543,
      "learning_rate": 3.6273387369185396e-08,
      "loss": 0.7801,
      "step": 10435
    },
    {
      "epoch": 0.9411552509356541,
      "grad_norm": 1.41830734539915,
      "learning_rate": 3.616271053307685e-08,
      "loss": 0.6871,
      "step": 10436
    },
    {
      "epoch": 0.9412454344591243,
      "grad_norm": 1.8111186189567383,
      "learning_rate": 3.6052201262394275e-08,
      "loss": 0.7644,
      "step": 10437
    },
    {
      "epoch": 0.9413356179825946,
      "grad_norm": 1.6004801744915274,
      "learning_rate": 3.5941859566566816e-08,
      "loss": 0.7994,
      "step": 10438
    },
    {
      "epoch": 0.9414258015060648,
      "grad_norm": 0.6852883180108273,
      "learning_rate": 3.583168545500981e-08,
      "loss": 0.5685,
      "step": 10439
    },
    {
      "epoch": 0.9415159850295352,
      "grad_norm": 1.5791245873446995,
      "learning_rate": 3.5721678937123746e-08,
      "loss": 0.6898,
      "step": 10440
    },
    {
      "epoch": 0.9416061685530054,
      "grad_norm": 1.8147809376153317,
      "learning_rate": 3.561184002229467e-08,
      "loss": 0.8261,
      "step": 10441
    },
    {
      "epoch": 0.9416963520764756,
      "grad_norm": 1.461806502937177,
      "learning_rate": 3.550216871989531e-08,
      "loss": 0.6912,
      "step": 10442
    },
    {
      "epoch": 0.9417865355999459,
      "grad_norm": 0.6391463777296976,
      "learning_rate": 3.539266503928262e-08,
      "loss": 0.6469,
      "step": 10443
    },
    {
      "epoch": 0.9418767191234162,
      "grad_norm": 1.6647261608024921,
      "learning_rate": 3.528332898980091e-08,
      "loss": 0.7566,
      "step": 10444
    },
    {
      "epoch": 0.9419669026468864,
      "grad_norm": 1.8013435859378992,
      "learning_rate": 3.517416058077849e-08,
      "loss": 0.7133,
      "step": 10445
    },
    {
      "epoch": 0.9420570861703567,
      "grad_norm": 2.076942968137428,
      "learning_rate": 3.506515982153102e-08,
      "loss": 0.6504,
      "step": 10446
    },
    {
      "epoch": 0.9421472696938269,
      "grad_norm": 2.0096265857821014,
      "learning_rate": 3.495632672135862e-08,
      "loss": 0.7492,
      "step": 10447
    },
    {
      "epoch": 0.9422374532172972,
      "grad_norm": 1.4837032203402316,
      "learning_rate": 3.4847661289547417e-08,
      "loss": 0.7754,
      "step": 10448
    },
    {
      "epoch": 0.9423276367407675,
      "grad_norm": 2.2388453014202225,
      "learning_rate": 3.473916353536932e-08,
      "loss": 0.6986,
      "step": 10449
    },
    {
      "epoch": 0.9424178202642377,
      "grad_norm": 1.5973461941507532,
      "learning_rate": 3.463083346808249e-08,
      "loss": 0.7122,
      "step": 10450
    },
    {
      "epoch": 0.9425080037877079,
      "grad_norm": 1.8656692397401065,
      "learning_rate": 3.452267109692975e-08,
      "loss": 0.6451,
      "step": 10451
    },
    {
      "epoch": 0.9425981873111783,
      "grad_norm": 1.7660918167240025,
      "learning_rate": 3.441467643114016e-08,
      "loss": 0.7455,
      "step": 10452
    },
    {
      "epoch": 0.9426883708346485,
      "grad_norm": 0.5907831101487887,
      "learning_rate": 3.430684947992857e-08,
      "loss": 0.5576,
      "step": 10453
    },
    {
      "epoch": 0.9427785543581187,
      "grad_norm": 4.6740197199420574,
      "learning_rate": 3.419919025249518e-08,
      "loss": 0.7598,
      "step": 10454
    },
    {
      "epoch": 0.942868737881589,
      "grad_norm": 1.6137888112265466,
      "learning_rate": 3.40916987580262e-08,
      "loss": 0.7636,
      "step": 10455
    },
    {
      "epoch": 0.9429589214050593,
      "grad_norm": 2.3212905004334,
      "learning_rate": 3.398437500569362e-08,
      "loss": 0.7232,
      "step": 10456
    },
    {
      "epoch": 0.9430491049285296,
      "grad_norm": 1.5563248489948218,
      "learning_rate": 3.3877219004654347e-08,
      "loss": 0.6242,
      "step": 10457
    },
    {
      "epoch": 0.9431392884519998,
      "grad_norm": 1.865791788744699,
      "learning_rate": 3.3770230764051946e-08,
      "loss": 0.7764,
      "step": 10458
    },
    {
      "epoch": 0.9432294719754701,
      "grad_norm": 1.8477356714191933,
      "learning_rate": 3.366341029301534e-08,
      "loss": 0.7667,
      "step": 10459
    },
    {
      "epoch": 0.9433196554989404,
      "grad_norm": 1.9273354099125706,
      "learning_rate": 3.355675760065857e-08,
      "loss": 0.7548,
      "step": 10460
    },
    {
      "epoch": 0.9434098390224106,
      "grad_norm": 0.5762930015357463,
      "learning_rate": 3.345027269608236e-08,
      "loss": 0.5251,
      "step": 10461
    },
    {
      "epoch": 0.9435000225458808,
      "grad_norm": 0.7446820549679796,
      "learning_rate": 3.334395558837211e-08,
      "loss": 0.5926,
      "step": 10462
    },
    {
      "epoch": 0.9435902060693512,
      "grad_norm": 1.6438499362285917,
      "learning_rate": 3.3237806286599667e-08,
      "loss": 0.6529,
      "step": 10463
    },
    {
      "epoch": 0.9436803895928214,
      "grad_norm": 1.7726311552003944,
      "learning_rate": 3.313182479982224e-08,
      "loss": 0.7335,
      "step": 10464
    },
    {
      "epoch": 0.9437705731162916,
      "grad_norm": 1.6469660695789947,
      "learning_rate": 3.302601113708259e-08,
      "loss": 0.8127,
      "step": 10465
    },
    {
      "epoch": 0.9438607566397619,
      "grad_norm": 1.2854894420645722,
      "learning_rate": 3.292036530740972e-08,
      "loss": 0.7944,
      "step": 10466
    },
    {
      "epoch": 0.9439509401632322,
      "grad_norm": 1.7167404911993935,
      "learning_rate": 3.2814887319817294e-08,
      "loss": 0.7345,
      "step": 10467
    },
    {
      "epoch": 0.9440411236867025,
      "grad_norm": 1.737495776384009,
      "learning_rate": 3.270957718330591e-08,
      "loss": 0.7872,
      "step": 10468
    },
    {
      "epoch": 0.9441313072101727,
      "grad_norm": 1.3483500588224127,
      "learning_rate": 3.260443490686082e-08,
      "loss": 0.7915,
      "step": 10469
    },
    {
      "epoch": 0.9442214907336429,
      "grad_norm": 1.9241353210874232,
      "learning_rate": 3.249946049945351e-08,
      "loss": 0.7637,
      "step": 10470
    },
    {
      "epoch": 0.9443116742571133,
      "grad_norm": 1.8883103001876895,
      "learning_rate": 3.239465397004082e-08,
      "loss": 0.7595,
      "step": 10471
    },
    {
      "epoch": 0.9444018577805835,
      "grad_norm": 2.3252498615503963,
      "learning_rate": 3.229001532756559e-08,
      "loss": 0.7083,
      "step": 10472
    },
    {
      "epoch": 0.9444920413040537,
      "grad_norm": 4.238211044925456,
      "learning_rate": 3.218554458095602e-08,
      "loss": 0.7422,
      "step": 10473
    },
    {
      "epoch": 0.944582224827524,
      "grad_norm": 6.713508015960888,
      "learning_rate": 3.20812417391263e-08,
      "loss": 0.7478,
      "step": 10474
    },
    {
      "epoch": 0.9446724083509943,
      "grad_norm": 1.2552812114573975,
      "learning_rate": 3.1977106810975764e-08,
      "loss": 0.7375,
      "step": 10475
    },
    {
      "epoch": 0.9447625918744645,
      "grad_norm": 0.7820750648198288,
      "learning_rate": 3.187313980539042e-08,
      "loss": 0.6218,
      "step": 10476
    },
    {
      "epoch": 0.9448527753979348,
      "grad_norm": 1.4874871004971801,
      "learning_rate": 3.176934073124071e-08,
      "loss": 0.8269,
      "step": 10477
    },
    {
      "epoch": 0.944942958921405,
      "grad_norm": 1.7121517801919042,
      "learning_rate": 3.166570959738357e-08,
      "loss": 0.7491,
      "step": 10478
    },
    {
      "epoch": 0.9450331424448754,
      "grad_norm": 1.6830512144033423,
      "learning_rate": 3.1562246412661476e-08,
      "loss": 0.6827,
      "step": 10479
    },
    {
      "epoch": 0.9451233259683456,
      "grad_norm": 1.90855774566381,
      "learning_rate": 3.145895118590225e-08,
      "loss": 0.7812,
      "step": 10480
    },
    {
      "epoch": 0.9452135094918158,
      "grad_norm": 1.7192473645608175,
      "learning_rate": 3.135582392591996e-08,
      "loss": 0.6559,
      "step": 10481
    },
    {
      "epoch": 0.9453036930152862,
      "grad_norm": 0.6027120283215532,
      "learning_rate": 3.125286464151333e-08,
      "loss": 0.536,
      "step": 10482
    },
    {
      "epoch": 0.9453938765387564,
      "grad_norm": 1.8087604355263884,
      "learning_rate": 3.115007334146824e-08,
      "loss": 0.7197,
      "step": 10483
    },
    {
      "epoch": 0.9454840600622266,
      "grad_norm": 0.6309510391770325,
      "learning_rate": 3.104745003455478e-08,
      "loss": 0.5571,
      "step": 10484
    },
    {
      "epoch": 0.9455742435856969,
      "grad_norm": 1.5176813392446569,
      "learning_rate": 3.094499472952972e-08,
      "loss": 0.7659,
      "step": 10485
    },
    {
      "epoch": 0.9456644271091672,
      "grad_norm": 1.9007280134529712,
      "learning_rate": 3.084270743513495e-08,
      "loss": 0.717,
      "step": 10486
    },
    {
      "epoch": 0.9457546106326374,
      "grad_norm": 1.9698326524827512,
      "learning_rate": 3.074058816009817e-08,
      "loss": 0.6375,
      "step": 10487
    },
    {
      "epoch": 0.9458447941561077,
      "grad_norm": 1.8377417606464332,
      "learning_rate": 3.063863691313284e-08,
      "loss": 0.6977,
      "step": 10488
    },
    {
      "epoch": 0.9459349776795779,
      "grad_norm": 1.748057990864622,
      "learning_rate": 3.0536853702937794e-08,
      "loss": 0.6827,
      "step": 10489
    },
    {
      "epoch": 0.9460251612030482,
      "grad_norm": 2.338013405949882,
      "learning_rate": 3.043523853819807e-08,
      "loss": 0.7451,
      "step": 10490
    },
    {
      "epoch": 0.9461153447265185,
      "grad_norm": 1.6670910136109407,
      "learning_rate": 3.0333791427583855e-08,
      "loss": 0.771,
      "step": 10491
    },
    {
      "epoch": 0.9462055282499887,
      "grad_norm": 1.6647599598829679,
      "learning_rate": 3.023251237975111e-08,
      "loss": 0.866,
      "step": 10492
    },
    {
      "epoch": 0.946295711773459,
      "grad_norm": 0.6561555340305126,
      "learning_rate": 3.0131401403341584e-08,
      "loss": 0.5941,
      "step": 10493
    },
    {
      "epoch": 0.9463858952969293,
      "grad_norm": 2.681837815093837,
      "learning_rate": 3.00304585069826e-08,
      "loss": 0.7255,
      "step": 10494
    },
    {
      "epoch": 0.9464760788203995,
      "grad_norm": 1.6653159708340168,
      "learning_rate": 2.992968369928728e-08,
      "loss": 0.7353,
      "step": 10495
    },
    {
      "epoch": 0.9465662623438698,
      "grad_norm": 1.744516022577722,
      "learning_rate": 2.982907698885429e-08,
      "loss": 0.7635,
      "step": 10496
    },
    {
      "epoch": 0.94665644586734,
      "grad_norm": 2.157718655854151,
      "learning_rate": 2.9728638384267645e-08,
      "loss": 0.7546,
      "step": 10497
    },
    {
      "epoch": 0.9467466293908103,
      "grad_norm": 2.0688040366131197,
      "learning_rate": 2.962836789409784e-08,
      "loss": 0.7498,
      "step": 10498
    },
    {
      "epoch": 0.9468368129142806,
      "grad_norm": 1.472375412934201,
      "learning_rate": 2.95282655268998e-08,
      "loss": 0.8396,
      "step": 10499
    },
    {
      "epoch": 0.9469269964377508,
      "grad_norm": 1.9726088112140907,
      "learning_rate": 2.942833129121558e-08,
      "loss": 0.747,
      "step": 10500
    },
    {
      "epoch": 0.947017179961221,
      "grad_norm": 1.937608100582644,
      "learning_rate": 2.9328565195571475e-08,
      "loss": 0.6732,
      "step": 10501
    },
    {
      "epoch": 0.9471073634846914,
      "grad_norm": 1.4972153883860433,
      "learning_rate": 2.9228967248480675e-08,
      "loss": 0.8139,
      "step": 10502
    },
    {
      "epoch": 0.9471975470081616,
      "grad_norm": 1.9009437500980895,
      "learning_rate": 2.912953745844082e-08,
      "loss": 0.7835,
      "step": 10503
    },
    {
      "epoch": 0.9472877305316318,
      "grad_norm": 1.9079497841718762,
      "learning_rate": 2.9030275833936247e-08,
      "loss": 0.7368,
      "step": 10504
    },
    {
      "epoch": 0.9473779140551022,
      "grad_norm": 2.014153705008608,
      "learning_rate": 2.893118238343617e-08,
      "loss": 0.6473,
      "step": 10505
    },
    {
      "epoch": 0.9474680975785724,
      "grad_norm": 2.19247035338172,
      "learning_rate": 2.8832257115396052e-08,
      "loss": 0.6572,
      "step": 10506
    },
    {
      "epoch": 0.9475582811020427,
      "grad_norm": 2.096350236656028,
      "learning_rate": 2.873350003825692e-08,
      "loss": 0.7263,
      "step": 10507
    },
    {
      "epoch": 0.9476484646255129,
      "grad_norm": 0.5640374471326202,
      "learning_rate": 2.8634911160444696e-08,
      "loss": 0.4722,
      "step": 10508
    },
    {
      "epoch": 0.9477386481489832,
      "grad_norm": 1.6810853331520286,
      "learning_rate": 2.853649049037199e-08,
      "loss": 0.8082,
      "step": 10509
    },
    {
      "epoch": 0.9478288316724535,
      "grad_norm": 1.9509959098581429,
      "learning_rate": 2.8438238036436525e-08,
      "loss": 0.798,
      "step": 10510
    },
    {
      "epoch": 0.9479190151959237,
      "grad_norm": 4.02244090357508,
      "learning_rate": 2.834015380702137e-08,
      "loss": 0.7411,
      "step": 10511
    },
    {
      "epoch": 0.9480091987193939,
      "grad_norm": 0.5810850247709946,
      "learning_rate": 2.824223781049606e-08,
      "loss": 0.5129,
      "step": 10512
    },
    {
      "epoch": 0.9480993822428643,
      "grad_norm": 1.9664096545802618,
      "learning_rate": 2.8144490055215465e-08,
      "loss": 0.8091,
      "step": 10513
    },
    {
      "epoch": 0.9481895657663345,
      "grad_norm": 1.5789267137079992,
      "learning_rate": 2.8046910549519355e-08,
      "loss": 0.7795,
      "step": 10514
    },
    {
      "epoch": 0.9482797492898047,
      "grad_norm": 1.6489220941946858,
      "learning_rate": 2.794949930173418e-08,
      "loss": 0.7498,
      "step": 10515
    },
    {
      "epoch": 0.948369932813275,
      "grad_norm": 1.9333975027659942,
      "learning_rate": 2.7852256320171296e-08,
      "loss": 0.7777,
      "step": 10516
    },
    {
      "epoch": 0.9484601163367453,
      "grad_norm": 1.7623007208529973,
      "learning_rate": 2.775518161312851e-08,
      "loss": 0.6101,
      "step": 10517
    },
    {
      "epoch": 0.9485502998602156,
      "grad_norm": 1.5160294612497869,
      "learning_rate": 2.76582751888883e-08,
      "loss": 0.7725,
      "step": 10518
    },
    {
      "epoch": 0.9486404833836858,
      "grad_norm": 1.9778263439092363,
      "learning_rate": 2.756153705571962e-08,
      "loss": 0.7621,
      "step": 10519
    },
    {
      "epoch": 0.948730666907156,
      "grad_norm": 1.4449172020656393,
      "learning_rate": 2.74649672218763e-08,
      "loss": 0.7268,
      "step": 10520
    },
    {
      "epoch": 0.9488208504306264,
      "grad_norm": 1.5953060108915909,
      "learning_rate": 2.7368565695598424e-08,
      "loss": 0.733,
      "step": 10521
    },
    {
      "epoch": 0.9489110339540966,
      "grad_norm": 1.7679237322906243,
      "learning_rate": 2.727233248511185e-08,
      "loss": 0.6716,
      "step": 10522
    },
    {
      "epoch": 0.9490012174775668,
      "grad_norm": 2.111525957718064,
      "learning_rate": 2.71762675986269e-08,
      "loss": 0.7013,
      "step": 10523
    },
    {
      "epoch": 0.9490914010010371,
      "grad_norm": 1.4699285121324879,
      "learning_rate": 2.7080371044341242e-08,
      "loss": 0.7874,
      "step": 10524
    },
    {
      "epoch": 0.9491815845245074,
      "grad_norm": 2.094799974128821,
      "learning_rate": 2.6984642830436556e-08,
      "loss": 0.7762,
      "step": 10525
    },
    {
      "epoch": 0.9492717680479776,
      "grad_norm": 1.4774720957746459,
      "learning_rate": 2.688908296508141e-08,
      "loss": 0.73,
      "step": 10526
    },
    {
      "epoch": 0.9493619515714479,
      "grad_norm": 2.573405066299106,
      "learning_rate": 2.679369145642929e-08,
      "loss": 0.7972,
      "step": 10527
    },
    {
      "epoch": 0.9494521350949181,
      "grad_norm": 2.0484931897494087,
      "learning_rate": 2.669846831261946e-08,
      "loss": 0.7647,
      "step": 10528
    },
    {
      "epoch": 0.9495423186183884,
      "grad_norm": 2.032559134360935,
      "learning_rate": 2.6603413541776976e-08,
      "loss": 0.7332,
      "step": 10529
    },
    {
      "epoch": 0.9496325021418587,
      "grad_norm": 1.5531771430428978,
      "learning_rate": 2.6508527152012683e-08,
      "loss": 0.6973,
      "step": 10530
    },
    {
      "epoch": 0.9497226856653289,
      "grad_norm": 2.090385238515631,
      "learning_rate": 2.641380915142233e-08,
      "loss": 0.6495,
      "step": 10531
    },
    {
      "epoch": 0.9498128691887993,
      "grad_norm": 2.05711980818143,
      "learning_rate": 2.6319259548088334e-08,
      "loss": 0.7683,
      "step": 10532
    },
    {
      "epoch": 0.9499030527122695,
      "grad_norm": 1.8578114707693678,
      "learning_rate": 2.6224878350077585e-08,
      "loss": 0.7446,
      "step": 10533
    },
    {
      "epoch": 0.9499932362357397,
      "grad_norm": 1.700347174929081,
      "learning_rate": 2.6130665565443633e-08,
      "loss": 0.6619,
      "step": 10534
    },
    {
      "epoch": 0.95008341975921,
      "grad_norm": 1.8610352228675802,
      "learning_rate": 2.603662120222494e-08,
      "loss": 0.7473,
      "step": 10535
    },
    {
      "epoch": 0.9501736032826803,
      "grad_norm": 2.1912681821687876,
      "learning_rate": 2.59427452684462e-08,
      "loss": 0.5933,
      "step": 10536
    },
    {
      "epoch": 0.9502637868061505,
      "grad_norm": 1.4266078277439451,
      "learning_rate": 2.5849037772117443e-08,
      "loss": 0.7273,
      "step": 10537
    },
    {
      "epoch": 0.9503539703296208,
      "grad_norm": 2.31945322483048,
      "learning_rate": 2.575549872123384e-08,
      "loss": 0.7674,
      "step": 10538
    },
    {
      "epoch": 0.950444153853091,
      "grad_norm": 1.3741631562230736,
      "learning_rate": 2.5662128123776994e-08,
      "loss": 0.7509,
      "step": 10539
    },
    {
      "epoch": 0.9505343373765613,
      "grad_norm": 1.4835920176161965,
      "learning_rate": 2.5568925987713875e-08,
      "loss": 0.7426,
      "step": 10540
    },
    {
      "epoch": 0.9506245209000316,
      "grad_norm": 2.6462204016758704,
      "learning_rate": 2.5475892320996785e-08,
      "loss": 0.7181,
      "step": 10541
    },
    {
      "epoch": 0.9507147044235018,
      "grad_norm": 0.6745697363256169,
      "learning_rate": 2.5383027131564038e-08,
      "loss": 0.5615,
      "step": 10542
    },
    {
      "epoch": 0.950804887946972,
      "grad_norm": 1.7559320861222418,
      "learning_rate": 2.52903304273393e-08,
      "loss": 0.7189,
      "step": 10543
    },
    {
      "epoch": 0.9508950714704424,
      "grad_norm": 1.499544233381407,
      "learning_rate": 2.519780221623202e-08,
      "loss": 0.734,
      "step": 10544
    },
    {
      "epoch": 0.9509852549939126,
      "grad_norm": 1.5818648639355721,
      "learning_rate": 2.510544250613722e-08,
      "loss": 0.7055,
      "step": 10545
    },
    {
      "epoch": 0.9510754385173829,
      "grad_norm": 2.6235919082214387,
      "learning_rate": 2.501325130493548e-08,
      "loss": 0.8444,
      "step": 10546
    },
    {
      "epoch": 0.9511656220408531,
      "grad_norm": 0.7146215041246727,
      "learning_rate": 2.4921228620493395e-08,
      "loss": 0.5784,
      "step": 10547
    },
    {
      "epoch": 0.9512558055643234,
      "grad_norm": 0.7399201946785228,
      "learning_rate": 2.4829374460662244e-08,
      "loss": 0.6149,
      "step": 10548
    },
    {
      "epoch": 0.9513459890877937,
      "grad_norm": 1.5452553332547967,
      "learning_rate": 2.473768883327976e-08,
      "loss": 0.7634,
      "step": 10549
    },
    {
      "epoch": 0.9514361726112639,
      "grad_norm": 2.077544740821715,
      "learning_rate": 2.464617174616923e-08,
      "loss": 0.7093,
      "step": 10550
    },
    {
      "epoch": 0.9515263561347341,
      "grad_norm": 1.631717588560704,
      "learning_rate": 2.455482320713953e-08,
      "loss": 0.7436,
      "step": 10551
    },
    {
      "epoch": 0.9516165396582045,
      "grad_norm": 2.9959794760027783,
      "learning_rate": 2.4463643223984643e-08,
      "loss": 0.8071,
      "step": 10552
    },
    {
      "epoch": 0.9517067231816747,
      "grad_norm": 3.8256728434457283,
      "learning_rate": 2.4372631804484567e-08,
      "loss": 0.817,
      "step": 10553
    },
    {
      "epoch": 0.9517969067051449,
      "grad_norm": 3.190263784139195,
      "learning_rate": 2.4281788956405313e-08,
      "loss": 0.8232,
      "step": 10554
    },
    {
      "epoch": 0.9518870902286153,
      "grad_norm": 9.745174289669485,
      "learning_rate": 2.4191114687497572e-08,
      "loss": 0.6774,
      "step": 10555
    },
    {
      "epoch": 0.9519772737520855,
      "grad_norm": 2.152553984245304,
      "learning_rate": 2.4100609005498706e-08,
      "loss": 0.7068,
      "step": 10556
    },
    {
      "epoch": 0.9520674572755558,
      "grad_norm": 1.5565845666074454,
      "learning_rate": 2.4010271918130764e-08,
      "loss": 0.6772,
      "step": 10557
    },
    {
      "epoch": 0.952157640799026,
      "grad_norm": 2.0232119637065833,
      "learning_rate": 2.39201034331018e-08,
      "loss": 0.7064,
      "step": 10558
    },
    {
      "epoch": 0.9522478243224963,
      "grad_norm": 1.9502979417748674,
      "learning_rate": 2.3830103558105663e-08,
      "loss": 0.7363,
      "step": 10559
    },
    {
      "epoch": 0.9523380078459666,
      "grad_norm": 2.037623222697111,
      "learning_rate": 2.374027230082154e-08,
      "loss": 0.755,
      "step": 10560
    },
    {
      "epoch": 0.9524281913694368,
      "grad_norm": 0.6790832815488199,
      "learning_rate": 2.365060966891441e-08,
      "loss": 0.5953,
      "step": 10561
    },
    {
      "epoch": 0.952518374892907,
      "grad_norm": 1.6507939133742873,
      "learning_rate": 2.3561115670034827e-08,
      "loss": 0.741,
      "step": 10562
    },
    {
      "epoch": 0.9526085584163774,
      "grad_norm": 1.4376525797930273,
      "learning_rate": 2.3471790311818675e-08,
      "loss": 0.7646,
      "step": 10563
    },
    {
      "epoch": 0.9526987419398476,
      "grad_norm": 2.2150574165635293,
      "learning_rate": 2.338263360188808e-08,
      "loss": 0.7354,
      "step": 10564
    },
    {
      "epoch": 0.9527889254633178,
      "grad_norm": 1.740652460124788,
      "learning_rate": 2.329364554784985e-08,
      "loss": 0.6826,
      "step": 10565
    },
    {
      "epoch": 0.9528791089867881,
      "grad_norm": 2.7825849094099846,
      "learning_rate": 2.3204826157297465e-08,
      "loss": 0.8036,
      "step": 10566
    },
    {
      "epoch": 0.9529692925102584,
      "grad_norm": 1.7117509693506816,
      "learning_rate": 2.3116175437809082e-08,
      "loss": 0.6817,
      "step": 10567
    },
    {
      "epoch": 0.9530594760337286,
      "grad_norm": 0.6421505743395298,
      "learning_rate": 2.30276933969491e-08,
      "loss": 0.5957,
      "step": 10568
    },
    {
      "epoch": 0.9531496595571989,
      "grad_norm": 1.4013681232303061,
      "learning_rate": 2.2939380042267255e-08,
      "loss": 0.801,
      "step": 10569
    },
    {
      "epoch": 0.9532398430806691,
      "grad_norm": 1.8140077567896769,
      "learning_rate": 2.2851235381298627e-08,
      "loss": 0.651,
      "step": 10570
    },
    {
      "epoch": 0.9533300266041395,
      "grad_norm": 1.8066240629519568,
      "learning_rate": 2.2763259421564986e-08,
      "loss": 0.675,
      "step": 10571
    },
    {
      "epoch": 0.9534202101276097,
      "grad_norm": 1.9744809615491716,
      "learning_rate": 2.2675452170571873e-08,
      "loss": 0.7132,
      "step": 10572
    },
    {
      "epoch": 0.9535103936510799,
      "grad_norm": 2.0304319641843804,
      "learning_rate": 2.2587813635812414e-08,
      "loss": 0.7418,
      "step": 10573
    },
    {
      "epoch": 0.9536005771745502,
      "grad_norm": 2.097215659473023,
      "learning_rate": 2.2500343824763958e-08,
      "loss": 0.8328,
      "step": 10574
    },
    {
      "epoch": 0.9536907606980205,
      "grad_norm": 2.0083098630470255,
      "learning_rate": 2.2413042744890088e-08,
      "loss": 0.6948,
      "step": 10575
    },
    {
      "epoch": 0.9537809442214907,
      "grad_norm": 1.7483813429363053,
      "learning_rate": 2.2325910403639514e-08,
      "loss": 0.7316,
      "step": 10576
    },
    {
      "epoch": 0.953871127744961,
      "grad_norm": 1.6512306017028633,
      "learning_rate": 2.223894680844718e-08,
      "loss": 0.7038,
      "step": 10577
    },
    {
      "epoch": 0.9539613112684313,
      "grad_norm": 2.025922508625592,
      "learning_rate": 2.2152151966733146e-08,
      "loss": 0.5833,
      "step": 10578
    },
    {
      "epoch": 0.9540514947919015,
      "grad_norm": 2.3377740587129168,
      "learning_rate": 2.2065525885903267e-08,
      "loss": 0.7817,
      "step": 10579
    },
    {
      "epoch": 0.9541416783153718,
      "grad_norm": 1.5527902653070138,
      "learning_rate": 2.1979068573348747e-08,
      "loss": 0.7907,
      "step": 10580
    },
    {
      "epoch": 0.954231861838842,
      "grad_norm": 2.066500396602952,
      "learning_rate": 2.1892780036447013e-08,
      "loss": 0.7119,
      "step": 10581
    },
    {
      "epoch": 0.9543220453623124,
      "grad_norm": 1.7791106778819137,
      "learning_rate": 2.1806660282560175e-08,
      "loss": 0.8867,
      "step": 10582
    },
    {
      "epoch": 0.9544122288857826,
      "grad_norm": 1.4764075626848279,
      "learning_rate": 2.1720709319037024e-08,
      "loss": 0.7679,
      "step": 10583
    },
    {
      "epoch": 0.9545024124092528,
      "grad_norm": 2.431553850578963,
      "learning_rate": 2.1634927153211023e-08,
      "loss": 0.7113,
      "step": 10584
    },
    {
      "epoch": 0.954592595932723,
      "grad_norm": 2.308624938650893,
      "learning_rate": 2.1549313792401437e-08,
      "loss": 0.8086,
      "step": 10585
    },
    {
      "epoch": 0.9546827794561934,
      "grad_norm": 1.9586537957925236,
      "learning_rate": 2.1463869243913746e-08,
      "loss": 0.8374,
      "step": 10586
    },
    {
      "epoch": 0.9547729629796636,
      "grad_norm": 2.3137566528554405,
      "learning_rate": 2.1378593515037902e-08,
      "loss": 0.7234,
      "step": 10587
    },
    {
      "epoch": 0.9548631465031339,
      "grad_norm": 1.50381675553016,
      "learning_rate": 2.129348661305075e-08,
      "loss": 0.8026,
      "step": 10588
    },
    {
      "epoch": 0.9549533300266041,
      "grad_norm": 2.431740044089583,
      "learning_rate": 2.1208548545213813e-08,
      "loss": 0.7803,
      "step": 10589
    },
    {
      "epoch": 0.9550435135500744,
      "grad_norm": 1.9128373565330163,
      "learning_rate": 2.1123779318774404e-08,
      "loss": 0.702,
      "step": 10590
    },
    {
      "epoch": 0.9551336970735447,
      "grad_norm": 6.906026845235209,
      "learning_rate": 2.1039178940965408e-08,
      "loss": 0.688,
      "step": 10591
    },
    {
      "epoch": 0.9552238805970149,
      "grad_norm": 1.4249345329371235,
      "learning_rate": 2.0954747419005712e-08,
      "loss": 0.8095,
      "step": 10592
    },
    {
      "epoch": 0.9553140641204851,
      "grad_norm": 2.4503762948802845,
      "learning_rate": 2.087048476009934e-08,
      "loss": 0.7249,
      "step": 10593
    },
    {
      "epoch": 0.9554042476439555,
      "grad_norm": 1.5707065220786365,
      "learning_rate": 2.0786390971435862e-08,
      "loss": 0.72,
      "step": 10594
    },
    {
      "epoch": 0.9554944311674257,
      "grad_norm": 2.669668435358319,
      "learning_rate": 2.070246606019088e-08,
      "loss": 0.7073,
      "step": 10595
    },
    {
      "epoch": 0.955584614690896,
      "grad_norm": 2.438838004631651,
      "learning_rate": 2.0618710033525112e-08,
      "loss": 0.665,
      "step": 10596
    },
    {
      "epoch": 0.9556747982143662,
      "grad_norm": 1.6710123706635946,
      "learning_rate": 2.053512289858528e-08,
      "loss": 0.7123,
      "step": 10597
    },
    {
      "epoch": 0.9557649817378365,
      "grad_norm": 2.042731481016028,
      "learning_rate": 2.0451704662503456e-08,
      "loss": 0.7865,
      "step": 10598
    },
    {
      "epoch": 0.9558551652613068,
      "grad_norm": 1.893237366236614,
      "learning_rate": 2.0368455332397282e-08,
      "loss": 0.8276,
      "step": 10599
    },
    {
      "epoch": 0.955945348784777,
      "grad_norm": 1.6779712514221945,
      "learning_rate": 2.0285374915369967e-08,
      "loss": 0.7588,
      "step": 10600
    },
    {
      "epoch": 0.9560355323082473,
      "grad_norm": 2.2615873993729365,
      "learning_rate": 2.020246341851073e-08,
      "loss": 0.7888,
      "step": 10601
    },
    {
      "epoch": 0.9561257158317176,
      "grad_norm": 2.8924569588164935,
      "learning_rate": 2.0119720848893463e-08,
      "loss": 0.7863,
      "step": 10602
    },
    {
      "epoch": 0.9562158993551878,
      "grad_norm": 1.7946692819283425,
      "learning_rate": 2.0037147213578964e-08,
      "loss": 0.6935,
      "step": 10603
    },
    {
      "epoch": 0.956306082878658,
      "grad_norm": 1.6126222904171619,
      "learning_rate": 1.9954742519612265e-08,
      "loss": 0.7718,
      "step": 10604
    },
    {
      "epoch": 0.9563962664021284,
      "grad_norm": 1.5888844438118694,
      "learning_rate": 1.9872506774024633e-08,
      "loss": 0.7209,
      "step": 10605
    },
    {
      "epoch": 0.9564864499255986,
      "grad_norm": 2.010220045314621,
      "learning_rate": 1.979043998383334e-08,
      "loss": 0.7523,
      "step": 10606
    },
    {
      "epoch": 0.9565766334490688,
      "grad_norm": 3.1462993318784545,
      "learning_rate": 1.970854215604034e-08,
      "loss": 0.6856,
      "step": 10607
    },
    {
      "epoch": 0.9566668169725391,
      "grad_norm": 1.5979566221783135,
      "learning_rate": 1.9626813297633826e-08,
      "loss": 0.7158,
      "step": 10608
    },
    {
      "epoch": 0.9567570004960094,
      "grad_norm": 2.2543661352479836,
      "learning_rate": 1.954525341558688e-08,
      "loss": 0.646,
      "step": 10609
    },
    {
      "epoch": 0.9568471840194797,
      "grad_norm": 2.048227576724493,
      "learning_rate": 1.9463862516859498e-08,
      "loss": 0.7142,
      "step": 10610
    },
    {
      "epoch": 0.9569373675429499,
      "grad_norm": 1.7010917776108092,
      "learning_rate": 1.938264060839545e-08,
      "loss": 0.8337,
      "step": 10611
    },
    {
      "epoch": 0.9570275510664201,
      "grad_norm": 1.7336412544497921,
      "learning_rate": 1.9301587697126086e-08,
      "loss": 0.7192,
      "step": 10612
    },
    {
      "epoch": 0.9571177345898905,
      "grad_norm": 1.9826144120248896,
      "learning_rate": 1.9220703789966318e-08,
      "loss": 0.6987,
      "step": 10613
    },
    {
      "epoch": 0.9572079181133607,
      "grad_norm": 1.6090121554620278,
      "learning_rate": 1.913998889381818e-08,
      "loss": 0.7053,
      "step": 10614
    },
    {
      "epoch": 0.9572981016368309,
      "grad_norm": 2.3261747448292542,
      "learning_rate": 1.9059443015568387e-08,
      "loss": 0.7614,
      "step": 10615
    },
    {
      "epoch": 0.9573882851603012,
      "grad_norm": 1.9831911295178082,
      "learning_rate": 1.8979066162089884e-08,
      "loss": 0.7597,
      "step": 10616
    },
    {
      "epoch": 0.9574784686837715,
      "grad_norm": 4.675847128752175,
      "learning_rate": 1.889885834024052e-08,
      "loss": 0.7492,
      "step": 10617
    },
    {
      "epoch": 0.9575686522072417,
      "grad_norm": 2.2544331851956114,
      "learning_rate": 1.8818819556864374e-08,
      "loss": 0.6975,
      "step": 10618
    },
    {
      "epoch": 0.957658835730712,
      "grad_norm": 2.2158499018449116,
      "learning_rate": 1.873894981879065e-08,
      "loss": 0.7315,
      "step": 10619
    },
    {
      "epoch": 0.9577490192541822,
      "grad_norm": 0.6347621741567858,
      "learning_rate": 1.8659249132834342e-08,
      "loss": 0.5911,
      "step": 10620
    },
    {
      "epoch": 0.9578392027776526,
      "grad_norm": 1.521912657724184,
      "learning_rate": 1.857971750579579e-08,
      "loss": 0.7237,
      "step": 10621
    },
    {
      "epoch": 0.9579293863011228,
      "grad_norm": 1.4164812677277911,
      "learning_rate": 1.8500354944461116e-08,
      "loss": 0.6287,
      "step": 10622
    },
    {
      "epoch": 0.958019569824593,
      "grad_norm": 1.4988144321924703,
      "learning_rate": 1.8421161455602242e-08,
      "loss": 0.755,
      "step": 10623
    },
    {
      "epoch": 0.9581097533480634,
      "grad_norm": 1.5944925710051077,
      "learning_rate": 1.834213704597598e-08,
      "loss": 0.7695,
      "step": 10624
    },
    {
      "epoch": 0.9581999368715336,
      "grad_norm": 1.5014048990828524,
      "learning_rate": 1.8263281722325385e-08,
      "loss": 0.7553,
      "step": 10625
    },
    {
      "epoch": 0.9582901203950038,
      "grad_norm": 1.950328136614288,
      "learning_rate": 1.818459549137885e-08,
      "loss": 0.7563,
      "step": 10626
    },
    {
      "epoch": 0.9583803039184741,
      "grad_norm": 1.6985086294543972,
      "learning_rate": 1.8106078359850117e-08,
      "loss": 0.7332,
      "step": 10627
    },
    {
      "epoch": 0.9584704874419444,
      "grad_norm": 3.9597959660457764,
      "learning_rate": 1.802773033443894e-08,
      "loss": 0.7651,
      "step": 10628
    },
    {
      "epoch": 0.9585606709654146,
      "grad_norm": 1.646049469224135,
      "learning_rate": 1.7949551421830413e-08,
      "loss": 0.7132,
      "step": 10629
    },
    {
      "epoch": 0.9586508544888849,
      "grad_norm": 1.526466137002843,
      "learning_rate": 1.7871541628694752e-08,
      "loss": 0.7056,
      "step": 10630
    },
    {
      "epoch": 0.9587410380123551,
      "grad_norm": 2.8257579698096706,
      "learning_rate": 1.779370096168864e-08,
      "loss": 0.6581,
      "step": 10631
    },
    {
      "epoch": 0.9588312215358255,
      "grad_norm": 1.9124450700638052,
      "learning_rate": 1.771602942745387e-08,
      "loss": 0.7109,
      "step": 10632
    },
    {
      "epoch": 0.9589214050592957,
      "grad_norm": 4.0607981711862235,
      "learning_rate": 1.763852703261759e-08,
      "loss": 0.7276,
      "step": 10633
    },
    {
      "epoch": 0.9590115885827659,
      "grad_norm": 1.7406516382997281,
      "learning_rate": 1.756119378379295e-08,
      "loss": 0.8086,
      "step": 10634
    },
    {
      "epoch": 0.9591017721062362,
      "grad_norm": 1.5786917395877442,
      "learning_rate": 1.7484029687578005e-08,
      "loss": 0.8344,
      "step": 10635
    },
    {
      "epoch": 0.9591919556297065,
      "grad_norm": 1.495829187810709,
      "learning_rate": 1.740703475055727e-08,
      "loss": 0.7214,
      "step": 10636
    },
    {
      "epoch": 0.9592821391531767,
      "grad_norm": 2.740965829184635,
      "learning_rate": 1.7330208979300153e-08,
      "loss": 0.7821,
      "step": 10637
    },
    {
      "epoch": 0.959372322676647,
      "grad_norm": 2.501741660932082,
      "learning_rate": 1.725355238036208e-08,
      "loss": 0.7883,
      "step": 10638
    },
    {
      "epoch": 0.9594625062001172,
      "grad_norm": 1.6341986362274645,
      "learning_rate": 1.7177064960283594e-08,
      "loss": 0.7862,
      "step": 10639
    },
    {
      "epoch": 0.9595526897235875,
      "grad_norm": 11.387089938203799,
      "learning_rate": 1.7100746725591253e-08,
      "loss": 0.7415,
      "step": 10640
    },
    {
      "epoch": 0.9596428732470578,
      "grad_norm": 2.1626133663031486,
      "learning_rate": 1.7024597682796517e-08,
      "loss": 0.7405,
      "step": 10641
    },
    {
      "epoch": 0.959733056770528,
      "grad_norm": 1.618629365645727,
      "learning_rate": 1.6948617838397293e-08,
      "loss": 0.8166,
      "step": 10642
    },
    {
      "epoch": 0.9598232402939982,
      "grad_norm": 1.446565404605978,
      "learning_rate": 1.6872807198876404e-08,
      "loss": 0.7747,
      "step": 10643
    },
    {
      "epoch": 0.9599134238174686,
      "grad_norm": 1.8317786329282504,
      "learning_rate": 1.679716577070245e-08,
      "loss": 0.8462,
      "step": 10644
    },
    {
      "epoch": 0.9600036073409388,
      "grad_norm": 1.7152434906340486,
      "learning_rate": 1.6721693560329596e-08,
      "loss": 0.8339,
      "step": 10645
    },
    {
      "epoch": 0.960093790864409,
      "grad_norm": 1.8512463521052138,
      "learning_rate": 1.6646390574197366e-08,
      "loss": 0.6552,
      "step": 10646
    },
    {
      "epoch": 0.9601839743878793,
      "grad_norm": 0.6906491460937533,
      "learning_rate": 1.6571256818731504e-08,
      "loss": 0.5966,
      "step": 10647
    },
    {
      "epoch": 0.9602741579113496,
      "grad_norm": 1.8240021734308378,
      "learning_rate": 1.6496292300342218e-08,
      "loss": 0.6761,
      "step": 10648
    },
    {
      "epoch": 0.9603643414348199,
      "grad_norm": 2.244657744080715,
      "learning_rate": 1.642149702542639e-08,
      "loss": 0.6867,
      "step": 10649
    },
    {
      "epoch": 0.9604545249582901,
      "grad_norm": 1.614630257529653,
      "learning_rate": 1.634687100036558e-08,
      "loss": 0.7088,
      "step": 10650
    },
    {
      "epoch": 0.9605447084817604,
      "grad_norm": 2.5580740131410904,
      "learning_rate": 1.627241423152781e-08,
      "loss": 0.6778,
      "step": 10651
    },
    {
      "epoch": 0.9606348920052307,
      "grad_norm": 2.3130631663219368,
      "learning_rate": 1.619812672526555e-08,
      "loss": 0.7943,
      "step": 10652
    },
    {
      "epoch": 0.9607250755287009,
      "grad_norm": 1.8690589878984933,
      "learning_rate": 1.6124008487917727e-08,
      "loss": 0.7296,
      "step": 10653
    },
    {
      "epoch": 0.9608152590521711,
      "grad_norm": 0.6470098520778774,
      "learning_rate": 1.6050059525808623e-08,
      "loss": 0.5946,
      "step": 10654
    },
    {
      "epoch": 0.9609054425756415,
      "grad_norm": 0.6367551641056378,
      "learning_rate": 1.597627984524763e-08,
      "loss": 0.5904,
      "step": 10655
    },
    {
      "epoch": 0.9609956260991117,
      "grad_norm": 1.6083547821327702,
      "learning_rate": 1.590266945253038e-08,
      "loss": 0.7872,
      "step": 10656
    },
    {
      "epoch": 0.9610858096225819,
      "grad_norm": 1.3904396105001082,
      "learning_rate": 1.582922835393763e-08,
      "loss": 0.7598,
      "step": 10657
    },
    {
      "epoch": 0.9611759931460522,
      "grad_norm": 1.529464692984304,
      "learning_rate": 1.5755956555735473e-08,
      "loss": 0.7197,
      "step": 10658
    },
    {
      "epoch": 0.9612661766695225,
      "grad_norm": 0.717739057690671,
      "learning_rate": 1.5682854064176244e-08,
      "loss": 0.6649,
      "step": 10659
    },
    {
      "epoch": 0.9613563601929928,
      "grad_norm": 1.9214733913702517,
      "learning_rate": 1.5609920885497395e-08,
      "loss": 0.6793,
      "step": 10660
    },
    {
      "epoch": 0.961446543716463,
      "grad_norm": 1.6805075129778286,
      "learning_rate": 1.5537157025921732e-08,
      "loss": 0.7651,
      "step": 10661
    },
    {
      "epoch": 0.9615367272399332,
      "grad_norm": 1.8461781692736583,
      "learning_rate": 1.5464562491658285e-08,
      "loss": 0.7253,
      "step": 10662
    },
    {
      "epoch": 0.9616269107634036,
      "grad_norm": 2.034154839622603,
      "learning_rate": 1.5392137288900764e-08,
      "loss": 0.7784,
      "step": 10663
    },
    {
      "epoch": 0.9617170942868738,
      "grad_norm": 0.598452144075107,
      "learning_rate": 1.531988142382934e-08,
      "loss": 0.5797,
      "step": 10664
    },
    {
      "epoch": 0.961807277810344,
      "grad_norm": 4.464694016449779,
      "learning_rate": 1.5247794902608634e-08,
      "loss": 0.6745,
      "step": 10665
    },
    {
      "epoch": 0.9618974613338143,
      "grad_norm": 1.8527781041208742,
      "learning_rate": 1.5175877731390398e-08,
      "loss": 0.7043,
      "step": 10666
    },
    {
      "epoch": 0.9619876448572846,
      "grad_norm": 1.7744990461255024,
      "learning_rate": 1.510412991631016e-08,
      "loss": 0.7745,
      "step": 10667
    },
    {
      "epoch": 0.9620778283807548,
      "grad_norm": 1.553950916413565,
      "learning_rate": 1.503255146349014e-08,
      "loss": 0.7725,
      "step": 10668
    },
    {
      "epoch": 0.9621680119042251,
      "grad_norm": 1.8108153571079084,
      "learning_rate": 1.4961142379037893e-08,
      "loss": 0.7125,
      "step": 10669
    },
    {
      "epoch": 0.9622581954276953,
      "grad_norm": 2.428221428918979,
      "learning_rate": 1.4889902669046327e-08,
      "loss": 0.7836,
      "step": 10670
    },
    {
      "epoch": 0.9623483789511657,
      "grad_norm": 1.9210321702740705,
      "learning_rate": 1.4818832339594135e-08,
      "loss": 0.746,
      "step": 10671
    },
    {
      "epoch": 0.9624385624746359,
      "grad_norm": 1.4607099060008923,
      "learning_rate": 1.474793139674535e-08,
      "loss": 0.6977,
      "step": 10672
    },
    {
      "epoch": 0.9625287459981061,
      "grad_norm": 1.6284532534595528,
      "learning_rate": 1.4677199846549581e-08,
      "loss": 0.744,
      "step": 10673
    },
    {
      "epoch": 0.9626189295215765,
      "grad_norm": 1.9763577924426485,
      "learning_rate": 1.4606637695042224e-08,
      "loss": 0.7461,
      "step": 10674
    },
    {
      "epoch": 0.9627091130450467,
      "grad_norm": 4.008762299126556,
      "learning_rate": 1.4536244948243793e-08,
      "loss": 0.8401,
      "step": 10675
    },
    {
      "epoch": 0.9627992965685169,
      "grad_norm": 1.7764481184749126,
      "learning_rate": 1.4466021612160595e-08,
      "loss": 0.7112,
      "step": 10676
    },
    {
      "epoch": 0.9628894800919872,
      "grad_norm": 1.7688511768963748,
      "learning_rate": 1.4395967692784505e-08,
      "loss": 0.7103,
      "step": 10677
    },
    {
      "epoch": 0.9629796636154575,
      "grad_norm": 1.7159168821981294,
      "learning_rate": 1.4326083196092963e-08,
      "loss": 0.6724,
      "step": 10678
    },
    {
      "epoch": 0.9630698471389277,
      "grad_norm": 1.4316536146878303,
      "learning_rate": 1.42563681280492e-08,
      "loss": 0.7533,
      "step": 10679
    },
    {
      "epoch": 0.963160030662398,
      "grad_norm": 1.9399310211817737,
      "learning_rate": 1.4186822494600902e-08,
      "loss": 0.72,
      "step": 10680
    },
    {
      "epoch": 0.9632502141858682,
      "grad_norm": 2.560176633463256,
      "learning_rate": 1.4117446301682877e-08,
      "loss": 0.7326,
      "step": 10681
    },
    {
      "epoch": 0.9633403977093385,
      "grad_norm": 1.645088300798938,
      "learning_rate": 1.4048239555214392e-08,
      "loss": 0.7006,
      "step": 10682
    },
    {
      "epoch": 0.9634305812328088,
      "grad_norm": 1.5294399072789415,
      "learning_rate": 1.3979202261100497e-08,
      "loss": 0.7776,
      "step": 10683
    },
    {
      "epoch": 0.963520764756279,
      "grad_norm": 2.870730173354934,
      "learning_rate": 1.3910334425231817e-08,
      "loss": 0.8088,
      "step": 10684
    },
    {
      "epoch": 0.9636109482797492,
      "grad_norm": 1.4973428556496713,
      "learning_rate": 1.384163605348454e-08,
      "loss": 0.739,
      "step": 10685
    },
    {
      "epoch": 0.9637011318032196,
      "grad_norm": 2.041664889068868,
      "learning_rate": 1.3773107151720642e-08,
      "loss": 0.8022,
      "step": 10686
    },
    {
      "epoch": 0.9637913153266898,
      "grad_norm": 1.9095914352792256,
      "learning_rate": 1.3704747725787003e-08,
      "loss": 0.8065,
      "step": 10687
    },
    {
      "epoch": 0.9638814988501601,
      "grad_norm": 0.7331864092409571,
      "learning_rate": 1.3636557781516512e-08,
      "loss": 0.5873,
      "step": 10688
    },
    {
      "epoch": 0.9639716823736303,
      "grad_norm": 2.1066932868281625,
      "learning_rate": 1.3568537324727847e-08,
      "loss": 0.747,
      "step": 10689
    },
    {
      "epoch": 0.9640618658971006,
      "grad_norm": 1.5228754033640253,
      "learning_rate": 1.3500686361224589e-08,
      "loss": 0.7426,
      "step": 10690
    },
    {
      "epoch": 0.9641520494205709,
      "grad_norm": 1.9533623513007323,
      "learning_rate": 1.3433004896796108e-08,
      "loss": 0.7416,
      "step": 10691
    },
    {
      "epoch": 0.9642422329440411,
      "grad_norm": 3.6299562444230347,
      "learning_rate": 1.336549293721756e-08,
      "loss": 0.6326,
      "step": 10692
    },
    {
      "epoch": 0.9643324164675113,
      "grad_norm": 0.6788187723462701,
      "learning_rate": 1.3298150488249227e-08,
      "loss": 0.5731,
      "step": 10693
    },
    {
      "epoch": 0.9644225999909817,
      "grad_norm": 2.179236143133989,
      "learning_rate": 1.3230977555637401e-08,
      "loss": 0.817,
      "step": 10694
    },
    {
      "epoch": 0.9645127835144519,
      "grad_norm": 1.4571262359339012,
      "learning_rate": 1.3163974145113499e-08,
      "loss": 0.7001,
      "step": 10695
    },
    {
      "epoch": 0.9646029670379221,
      "grad_norm": 12.855658660478138,
      "learning_rate": 1.3097140262394723e-08,
      "loss": 0.643,
      "step": 10696
    },
    {
      "epoch": 0.9646931505613925,
      "grad_norm": 2.5832543412303783,
      "learning_rate": 1.303047591318318e-08,
      "loss": 0.7784,
      "step": 10697
    },
    {
      "epoch": 0.9647833340848627,
      "grad_norm": 1.6439125183757528,
      "learning_rate": 1.2963981103167875e-08,
      "loss": 0.6747,
      "step": 10698
    },
    {
      "epoch": 0.964873517608333,
      "grad_norm": 1.5277248989934,
      "learning_rate": 1.2897655838021825e-08,
      "loss": 0.7268,
      "step": 10699
    },
    {
      "epoch": 0.9649637011318032,
      "grad_norm": 1.9845379739954845,
      "learning_rate": 1.2831500123404726e-08,
      "loss": 0.7518,
      "step": 10700
    },
    {
      "epoch": 0.9650538846552735,
      "grad_norm": 2.3822939449199887,
      "learning_rate": 1.2765513964961172e-08,
      "loss": 0.8944,
      "step": 10701
    },
    {
      "epoch": 0.9651440681787438,
      "grad_norm": 1.7544302720406173,
      "learning_rate": 1.2699697368321549e-08,
      "loss": 0.7409,
      "step": 10702
    },
    {
      "epoch": 0.965234251702214,
      "grad_norm": 1.615644668075702,
      "learning_rate": 1.2634050339101366e-08,
      "loss": 0.7273,
      "step": 10703
    },
    {
      "epoch": 0.9653244352256842,
      "grad_norm": 2.761973842726258,
      "learning_rate": 1.2568572882902361e-08,
      "loss": 0.721,
      "step": 10704
    },
    {
      "epoch": 0.9654146187491546,
      "grad_norm": 1.8485689066517836,
      "learning_rate": 1.2503265005311402e-08,
      "loss": 0.7168,
      "step": 10705
    },
    {
      "epoch": 0.9655048022726248,
      "grad_norm": 1.67022416797447,
      "learning_rate": 1.2438126711900698e-08,
      "loss": 0.7084,
      "step": 10706
    },
    {
      "epoch": 0.965594985796095,
      "grad_norm": 1.5215717334666936,
      "learning_rate": 1.2373158008228247e-08,
      "loss": 0.5909,
      "step": 10707
    },
    {
      "epoch": 0.9656851693195653,
      "grad_norm": 1.3852357913618831,
      "learning_rate": 1.2308358899837833e-08,
      "loss": 0.7212,
      "step": 10708
    },
    {
      "epoch": 0.9657753528430356,
      "grad_norm": 1.7907066286664466,
      "learning_rate": 1.224372939225815e-08,
      "loss": 0.7195,
      "step": 10709
    },
    {
      "epoch": 0.9658655363665059,
      "grad_norm": 1.7178305507445044,
      "learning_rate": 1.2179269491003674e-08,
      "loss": 0.823,
      "step": 10710
    },
    {
      "epoch": 0.9659557198899761,
      "grad_norm": 1.9530364359803234,
      "learning_rate": 1.2114979201574894e-08,
      "loss": 0.6143,
      "step": 10711
    },
    {
      "epoch": 0.9660459034134463,
      "grad_norm": 1.6054079364480165,
      "learning_rate": 1.2050858529456975e-08,
      "loss": 0.7611,
      "step": 10712
    },
    {
      "epoch": 0.9661360869369167,
      "grad_norm": 1.6747598447440573,
      "learning_rate": 1.1986907480121545e-08,
      "loss": 0.7456,
      "step": 10713
    },
    {
      "epoch": 0.9662262704603869,
      "grad_norm": 1.781153559165384,
      "learning_rate": 1.192312605902468e-08,
      "loss": 0.7664,
      "step": 10714
    },
    {
      "epoch": 0.9663164539838571,
      "grad_norm": 0.7918972800476787,
      "learning_rate": 1.1859514271608917e-08,
      "loss": 0.6028,
      "step": 10715
    },
    {
      "epoch": 0.9664066375073274,
      "grad_norm": 2.5515846751383022,
      "learning_rate": 1.1796072123301914e-08,
      "loss": 0.7772,
      "step": 10716
    },
    {
      "epoch": 0.9664968210307977,
      "grad_norm": 0.6404533388843415,
      "learning_rate": 1.1732799619516897e-08,
      "loss": 0.5821,
      "step": 10717
    },
    {
      "epoch": 0.9665870045542679,
      "grad_norm": 2.9414513167820786,
      "learning_rate": 1.1669696765652659e-08,
      "loss": 0.7299,
      "step": 10718
    },
    {
      "epoch": 0.9666771880777382,
      "grad_norm": 2.0485038973524707,
      "learning_rate": 1.1606763567093336e-08,
      "loss": 0.7723,
      "step": 10719
    },
    {
      "epoch": 0.9667673716012085,
      "grad_norm": 0.6738306902408145,
      "learning_rate": 1.1544000029208857e-08,
      "loss": 0.5729,
      "step": 10720
    },
    {
      "epoch": 0.9668575551246787,
      "grad_norm": 1.7577705208746066,
      "learning_rate": 1.148140615735449e-08,
      "loss": 0.6021,
      "step": 10721
    },
    {
      "epoch": 0.966947738648149,
      "grad_norm": 1.4784462835663368,
      "learning_rate": 1.1418981956871076e-08,
      "loss": 0.7209,
      "step": 10722
    },
    {
      "epoch": 0.9670379221716192,
      "grad_norm": 2.135013101792169,
      "learning_rate": 1.1356727433085245e-08,
      "loss": 0.7328,
      "step": 10723
    },
    {
      "epoch": 0.9671281056950896,
      "grad_norm": 1.482641553119506,
      "learning_rate": 1.1294642591308524e-08,
      "loss": 0.7666,
      "step": 10724
    },
    {
      "epoch": 0.9672182892185598,
      "grad_norm": 2.3651001039251836,
      "learning_rate": 1.1232727436838452e-08,
      "loss": 0.7333,
      "step": 10725
    },
    {
      "epoch": 0.96730847274203,
      "grad_norm": 2.2081374915272005,
      "learning_rate": 1.1170981974958138e-08,
      "loss": 0.7884,
      "step": 10726
    },
    {
      "epoch": 0.9673986562655003,
      "grad_norm": 1.5557747877507224,
      "learning_rate": 1.1109406210936035e-08,
      "loss": 0.8096,
      "step": 10727
    },
    {
      "epoch": 0.9674888397889706,
      "grad_norm": 2.383465086578616,
      "learning_rate": 1.1048000150025939e-08,
      "loss": 0.8601,
      "step": 10728
    },
    {
      "epoch": 0.9675790233124408,
      "grad_norm": 3.3507228413082957,
      "learning_rate": 1.0986763797467213e-08,
      "loss": 0.6412,
      "step": 10729
    },
    {
      "epoch": 0.9676692068359111,
      "grad_norm": 0.5876172719746302,
      "learning_rate": 1.0925697158485459e-08,
      "loss": 0.5446,
      "step": 10730
    },
    {
      "epoch": 0.9677593903593813,
      "grad_norm": 1.679767127700147,
      "learning_rate": 1.0864800238290727e-08,
      "loss": 0.7506,
      "step": 10731
    },
    {
      "epoch": 0.9678495738828516,
      "grad_norm": 1.7244335184264588,
      "learning_rate": 1.0804073042079309e-08,
      "loss": 0.7021,
      "step": 10732
    },
    {
      "epoch": 0.9679397574063219,
      "grad_norm": 0.6419482172852338,
      "learning_rate": 1.0743515575032392e-08,
      "loss": 0.5888,
      "step": 10733
    },
    {
      "epoch": 0.9680299409297921,
      "grad_norm": 2.155078182528151,
      "learning_rate": 1.0683127842317619e-08,
      "loss": 0.7554,
      "step": 10734
    },
    {
      "epoch": 0.9681201244532623,
      "grad_norm": 3.7654410036333408,
      "learning_rate": 1.0622909849087314e-08,
      "loss": 0.7727,
      "step": 10735
    },
    {
      "epoch": 0.9682103079767327,
      "grad_norm": 1.5890161108036875,
      "learning_rate": 1.0562861600479588e-08,
      "loss": 0.8039,
      "step": 10736
    },
    {
      "epoch": 0.9683004915002029,
      "grad_norm": 1.9585801503363982,
      "learning_rate": 1.0502983101618345e-08,
      "loss": 0.836,
      "step": 10737
    },
    {
      "epoch": 0.9683906750236732,
      "grad_norm": 1.921360039919608,
      "learning_rate": 1.0443274357612386e-08,
      "loss": 0.624,
      "step": 10738
    },
    {
      "epoch": 0.9684808585471434,
      "grad_norm": 1.7377243006988359,
      "learning_rate": 1.0383735373556524e-08,
      "loss": 0.7576,
      "step": 10739
    },
    {
      "epoch": 0.9685710420706137,
      "grad_norm": 1.4371717326972153,
      "learning_rate": 1.0324366154531139e-08,
      "loss": 0.7735,
      "step": 10740
    },
    {
      "epoch": 0.968661225594084,
      "grad_norm": 1.9124420780561056,
      "learning_rate": 1.0265166705601735e-08,
      "loss": 0.7096,
      "step": 10741
    },
    {
      "epoch": 0.9687514091175542,
      "grad_norm": 0.704815929407543,
      "learning_rate": 1.0206137031819606e-08,
      "loss": 0.611,
      "step": 10742
    },
    {
      "epoch": 0.9688415926410244,
      "grad_norm": 4.683187510074372,
      "learning_rate": 1.0147277138221388e-08,
      "loss": 0.7568,
      "step": 10743
    },
    {
      "epoch": 0.9689317761644948,
      "grad_norm": 4.331253628639804,
      "learning_rate": 1.0088587029829287e-08,
      "loss": 0.7783,
      "step": 10744
    },
    {
      "epoch": 0.969021959687965,
      "grad_norm": 1.5501619346690523,
      "learning_rate": 1.003006671165152e-08,
      "loss": 0.8472,
      "step": 10745
    },
    {
      "epoch": 0.9691121432114352,
      "grad_norm": 1.8738210150333503,
      "learning_rate": 9.971716188680978e-09,
      "loss": 0.7377,
      "step": 10746
    },
    {
      "epoch": 0.9692023267349056,
      "grad_norm": 2.0451516620760266,
      "learning_rate": 9.91353546589635e-09,
      "loss": 0.7097,
      "step": 10747
    },
    {
      "epoch": 0.9692925102583758,
      "grad_norm": 1.8587483223533627,
      "learning_rate": 9.855524548262106e-09,
      "loss": 0.6804,
      "step": 10748
    },
    {
      "epoch": 0.969382693781846,
      "grad_norm": 1.796376830994295,
      "learning_rate": 9.797683440728288e-09,
      "loss": 0.6909,
      "step": 10749
    },
    {
      "epoch": 0.9694728773053163,
      "grad_norm": 1.6614969741329957,
      "learning_rate": 9.740012148229836e-09,
      "loss": 0.6564,
      "step": 10750
    },
    {
      "epoch": 0.9695630608287866,
      "grad_norm": 1.519165858095714,
      "learning_rate": 9.682510675687705e-09,
      "loss": 0.6945,
      "step": 10751
    },
    {
      "epoch": 0.9696532443522569,
      "grad_norm": 1.9564041076994525,
      "learning_rate": 9.625179028008191e-09,
      "loss": 0.6648,
      "step": 10752
    },
    {
      "epoch": 0.9697434278757271,
      "grad_norm": 1.7239090219995925,
      "learning_rate": 9.568017210083379e-09,
      "loss": 0.5888,
      "step": 10753
    },
    {
      "epoch": 0.9698336113991973,
      "grad_norm": 2.2980120692388524,
      "learning_rate": 9.511025226790259e-09,
      "loss": 0.722,
      "step": 10754
    },
    {
      "epoch": 0.9699237949226677,
      "grad_norm": 1.643291595127542,
      "learning_rate": 9.454203082992052e-09,
      "loss": 0.7175,
      "step": 10755
    },
    {
      "epoch": 0.9700139784461379,
      "grad_norm": 2.3026787748734483,
      "learning_rate": 9.3975507835371e-09,
      "loss": 0.6551,
      "step": 10756
    },
    {
      "epoch": 0.9701041619696081,
      "grad_norm": 0.6311331470845365,
      "learning_rate": 9.341068333259094e-09,
      "loss": 0.5651,
      "step": 10757
    },
    {
      "epoch": 0.9701943454930784,
      "grad_norm": 1.9300755407937187,
      "learning_rate": 9.28475573697729e-09,
      "loss": 0.7197,
      "step": 10758
    },
    {
      "epoch": 0.9702845290165487,
      "grad_norm": 1.6047035491348796,
      "learning_rate": 9.228612999497177e-09,
      "loss": 0.7451,
      "step": 10759
    },
    {
      "epoch": 0.970374712540019,
      "grad_norm": 2.8831709145314126,
      "learning_rate": 9.172640125608478e-09,
      "loss": 0.6822,
      "step": 10760
    },
    {
      "epoch": 0.9704648960634892,
      "grad_norm": 1.5711863349634967,
      "learning_rate": 9.116837120087817e-09,
      "loss": 0.757,
      "step": 10761
    },
    {
      "epoch": 0.9705550795869594,
      "grad_norm": 1.9869835596465166,
      "learning_rate": 9.061203987695832e-09,
      "loss": 0.8113,
      "step": 10762
    },
    {
      "epoch": 0.9706452631104298,
      "grad_norm": 2.0541034056831253,
      "learning_rate": 9.005740733180055e-09,
      "loss": 0.69,
      "step": 10763
    },
    {
      "epoch": 0.9707354466339,
      "grad_norm": 1.8491966487980405,
      "learning_rate": 8.950447361272483e-09,
      "loss": 0.8259,
      "step": 10764
    },
    {
      "epoch": 0.9708256301573702,
      "grad_norm": 4.153950219874959,
      "learning_rate": 8.895323876691784e-09,
      "loss": 0.7487,
      "step": 10765
    },
    {
      "epoch": 0.9709158136808405,
      "grad_norm": 1.9915024121705889,
      "learning_rate": 8.840370284140419e-09,
      "loss": 0.7972,
      "step": 10766
    },
    {
      "epoch": 0.9710059972043108,
      "grad_norm": 2.0604160068409336,
      "learning_rate": 8.78558658830797e-09,
      "loss": 0.6615,
      "step": 10767
    },
    {
      "epoch": 0.971096180727781,
      "grad_norm": 0.7035399590011113,
      "learning_rate": 8.730972793868696e-09,
      "loss": 0.5771,
      "step": 10768
    },
    {
      "epoch": 0.9711863642512513,
      "grad_norm": 2.612171159700685,
      "learning_rate": 8.67652890548265e-09,
      "loss": 0.7062,
      "step": 10769
    },
    {
      "epoch": 0.9712765477747216,
      "grad_norm": 1.5133861247418106,
      "learning_rate": 8.622254927795004e-09,
      "loss": 0.7244,
      "step": 10770
    },
    {
      "epoch": 0.9713667312981918,
      "grad_norm": 2.1349578240403035,
      "learning_rate": 8.568150865436941e-09,
      "loss": 0.6935,
      "step": 10771
    },
    {
      "epoch": 0.9714569148216621,
      "grad_norm": 1.5044145790158443,
      "learning_rate": 8.514216723024991e-09,
      "loss": 0.754,
      "step": 10772
    },
    {
      "epoch": 0.9715470983451323,
      "grad_norm": 2.392743636654017,
      "learning_rate": 8.460452505161031e-09,
      "loss": 0.6631,
      "step": 10773
    },
    {
      "epoch": 0.9716372818686027,
      "grad_norm": 2.6536261670423023,
      "learning_rate": 8.4068582164325e-09,
      "loss": 0.7657,
      "step": 10774
    },
    {
      "epoch": 0.9717274653920729,
      "grad_norm": 1.6307471687868937,
      "learning_rate": 8.353433861412406e-09,
      "loss": 0.7236,
      "step": 10775
    },
    {
      "epoch": 0.9718176489155431,
      "grad_norm": 1.530870118021958,
      "learning_rate": 8.300179444658883e-09,
      "loss": 0.7016,
      "step": 10776
    },
    {
      "epoch": 0.9719078324390134,
      "grad_norm": 4.226432304703947,
      "learning_rate": 8.247094970716296e-09,
      "loss": 0.8376,
      "step": 10777
    },
    {
      "epoch": 0.9719980159624837,
      "grad_norm": 1.5191077890179983,
      "learning_rate": 8.19418044411413e-09,
      "loss": 0.7777,
      "step": 10778
    },
    {
      "epoch": 0.9720881994859539,
      "grad_norm": 1.380970951736089,
      "learning_rate": 8.141435869367219e-09,
      "loss": 0.7438,
      "step": 10779
    },
    {
      "epoch": 0.9721783830094242,
      "grad_norm": 1.4741653279064275,
      "learning_rate": 8.088861250975742e-09,
      "loss": 0.7782,
      "step": 10780
    },
    {
      "epoch": 0.9722685665328944,
      "grad_norm": 23.179376788446692,
      "learning_rate": 8.036456593426111e-09,
      "loss": 0.708,
      "step": 10781
    },
    {
      "epoch": 0.9723587500563647,
      "grad_norm": 1.8353901148537886,
      "learning_rate": 7.984221901189415e-09,
      "loss": 0.7773,
      "step": 10782
    },
    {
      "epoch": 0.972448933579835,
      "grad_norm": 9.556805679202599,
      "learning_rate": 7.932157178722976e-09,
      "loss": 0.6645,
      "step": 10783
    },
    {
      "epoch": 0.9725391171033052,
      "grad_norm": 2.3147344361937763,
      "learning_rate": 7.880262430468799e-09,
      "loss": 0.7102,
      "step": 10784
    },
    {
      "epoch": 0.9726293006267754,
      "grad_norm": 1.788115271468833,
      "learning_rate": 7.828537660855339e-09,
      "loss": 0.6754,
      "step": 10785
    },
    {
      "epoch": 0.9727194841502458,
      "grad_norm": 1.9429803066423812,
      "learning_rate": 7.776982874295512e-09,
      "loss": 0.7573,
      "step": 10786
    },
    {
      "epoch": 0.972809667673716,
      "grad_norm": 2.20410231969277,
      "learning_rate": 7.725598075188688e-09,
      "loss": 0.6501,
      "step": 10787
    },
    {
      "epoch": 0.9728998511971862,
      "grad_norm": 1.4370804879431438,
      "learning_rate": 7.674383267918916e-09,
      "loss": 0.6924,
      "step": 10788
    },
    {
      "epoch": 0.9729900347206565,
      "grad_norm": 1.7302217673236235,
      "learning_rate": 7.623338456856476e-09,
      "loss": 0.7301,
      "step": 10789
    },
    {
      "epoch": 0.9730802182441268,
      "grad_norm": 2.1345691642859363,
      "learning_rate": 7.572463646356554e-09,
      "loss": 0.6343,
      "step": 10790
    },
    {
      "epoch": 0.9731704017675971,
      "grad_norm": 1.8366180457218022,
      "learning_rate": 7.521758840760339e-09,
      "loss": 0.7127,
      "step": 10791
    },
    {
      "epoch": 0.9732605852910673,
      "grad_norm": 2.271026171507846,
      "learning_rate": 7.471224044393931e-09,
      "loss": 0.7011,
      "step": 10792
    },
    {
      "epoch": 0.9733507688145376,
      "grad_norm": 1.6238343386224934,
      "learning_rate": 7.420859261569434e-09,
      "loss": 0.7983,
      "step": 10793
    },
    {
      "epoch": 0.9734409523380079,
      "grad_norm": 2.0063567468878687,
      "learning_rate": 7.370664496584078e-09,
      "loss": 0.8009,
      "step": 10794
    },
    {
      "epoch": 0.9735311358614781,
      "grad_norm": 1.95546874756152,
      "learning_rate": 7.3206397537211026e-09,
      "loss": 0.6645,
      "step": 10795
    },
    {
      "epoch": 0.9736213193849483,
      "grad_norm": 10.058003244195833,
      "learning_rate": 7.270785037248428e-09,
      "loss": 0.7126,
      "step": 10796
    },
    {
      "epoch": 0.9737115029084187,
      "grad_norm": 3.3321450181614405,
      "learning_rate": 7.221100351420428e-09,
      "loss": 0.7041,
      "step": 10797
    },
    {
      "epoch": 0.9738016864318889,
      "grad_norm": 1.6111982813451895,
      "learning_rate": 7.171585700475935e-09,
      "loss": 0.668,
      "step": 10798
    },
    {
      "epoch": 0.9738918699553591,
      "grad_norm": 1.7437305545064468,
      "learning_rate": 7.122241088640235e-09,
      "loss": 0.8021,
      "step": 10799
    },
    {
      "epoch": 0.9739820534788294,
      "grad_norm": 1.7736444562570324,
      "learning_rate": 7.073066520123516e-09,
      "loss": 0.8626,
      "step": 10800
    },
    {
      "epoch": 0.9740722370022997,
      "grad_norm": 2.7758553295090453,
      "learning_rate": 7.0240619991217555e-09,
      "loss": 0.6355,
      "step": 10801
    },
    {
      "epoch": 0.97416242052577,
      "grad_norm": 1.5957507122886248,
      "learning_rate": 6.975227529816052e-09,
      "loss": 0.8114,
      "step": 10802
    },
    {
      "epoch": 0.9742526040492402,
      "grad_norm": 1.8109734124349437,
      "learning_rate": 6.926563116373296e-09,
      "loss": 0.7858,
      "step": 10803
    },
    {
      "epoch": 0.9743427875727104,
      "grad_norm": 1.3717494168510609,
      "learning_rate": 6.878068762945943e-09,
      "loss": 0.6758,
      "step": 10804
    },
    {
      "epoch": 0.9744329710961808,
      "grad_norm": 1.4784917590131257,
      "learning_rate": 6.829744473671794e-09,
      "loss": 0.7783,
      "step": 10805
    },
    {
      "epoch": 0.974523154619651,
      "grad_norm": 0.5872639455767469,
      "learning_rate": 6.781590252674219e-09,
      "loss": 0.4789,
      "step": 10806
    },
    {
      "epoch": 0.9746133381431212,
      "grad_norm": 5.263185926663888,
      "learning_rate": 6.733606104061484e-09,
      "loss": 0.7488,
      "step": 10807
    },
    {
      "epoch": 0.9747035216665915,
      "grad_norm": 1.4017653268541188,
      "learning_rate": 6.6857920319283165e-09,
      "loss": 0.6851,
      "step": 10808
    },
    {
      "epoch": 0.9747937051900618,
      "grad_norm": 2.09713676171848,
      "learning_rate": 6.638148040354563e-09,
      "loss": 0.7154,
      "step": 10809
    },
    {
      "epoch": 0.974883888713532,
      "grad_norm": 1.75487425329194,
      "learning_rate": 6.590674133405194e-09,
      "loss": 0.7014,
      "step": 10810
    },
    {
      "epoch": 0.9749740722370023,
      "grad_norm": 1.9228979350204367,
      "learning_rate": 6.5433703151311914e-09,
      "loss": 0.7505,
      "step": 10811
    },
    {
      "epoch": 0.9750642557604725,
      "grad_norm": 1.7165583422041186,
      "learning_rate": 6.49623658956866e-09,
      "loss": 0.7411,
      "step": 10812
    },
    {
      "epoch": 0.9751544392839429,
      "grad_norm": 2.46219208239659,
      "learning_rate": 6.44927296073905e-09,
      "loss": 0.6164,
      "step": 10813
    },
    {
      "epoch": 0.9752446228074131,
      "grad_norm": 1.5607184362340722,
      "learning_rate": 6.402479432649821e-09,
      "loss": 0.8231,
      "step": 10814
    },
    {
      "epoch": 0.9753348063308833,
      "grad_norm": 1.7195223373466346,
      "learning_rate": 6.355856009293781e-09,
      "loss": 0.7323,
      "step": 10815
    },
    {
      "epoch": 0.9754249898543537,
      "grad_norm": 1.932512015420096,
      "learning_rate": 6.3094026946488575e-09,
      "loss": 0.7168,
      "step": 10816
    },
    {
      "epoch": 0.9755151733778239,
      "grad_norm": 1.5082856063598078,
      "learning_rate": 6.2631194926787704e-09,
      "loss": 0.7419,
      "step": 10817
    },
    {
      "epoch": 0.9756053569012941,
      "grad_norm": 1.619209021215413,
      "learning_rate": 6.217006407332581e-09,
      "loss": 0.6363,
      "step": 10818
    },
    {
      "epoch": 0.9756955404247644,
      "grad_norm": 1.4322897986197571,
      "learning_rate": 6.1710634425453654e-09,
      "loss": 0.6858,
      "step": 10819
    },
    {
      "epoch": 0.9757857239482347,
      "grad_norm": 1.7728754572966938,
      "learning_rate": 6.1252906022366544e-09,
      "loss": 0.7622,
      "step": 10820
    },
    {
      "epoch": 0.9758759074717049,
      "grad_norm": 11.09447483327498,
      "learning_rate": 6.079687890312213e-09,
      "loss": 0.7173,
      "step": 10821
    },
    {
      "epoch": 0.9759660909951752,
      "grad_norm": 1.958919870183641,
      "learning_rate": 6.034255310663372e-09,
      "loss": 0.7202,
      "step": 10822
    },
    {
      "epoch": 0.9760562745186454,
      "grad_norm": 1.8062385400322751,
      "learning_rate": 5.988992867166143e-09,
      "loss": 0.8371,
      "step": 10823
    },
    {
      "epoch": 0.9761464580421157,
      "grad_norm": 1.9324874641633667,
      "learning_rate": 5.943900563682991e-09,
      "loss": 0.7777,
      "step": 10824
    },
    {
      "epoch": 0.976236641565586,
      "grad_norm": 0.7148394975379406,
      "learning_rate": 5.898978404061506e-09,
      "loss": 0.6563,
      "step": 10825
    },
    {
      "epoch": 0.9763268250890562,
      "grad_norm": 2.2830329094084245,
      "learning_rate": 5.85422639213462e-09,
      "loss": 0.7316,
      "step": 10826
    },
    {
      "epoch": 0.9764170086125264,
      "grad_norm": 1.4901769384512182,
      "learning_rate": 5.809644531720614e-09,
      "loss": 0.7511,
      "step": 10827
    },
    {
      "epoch": 0.9765071921359968,
      "grad_norm": 1.7502234180068135,
      "learning_rate": 5.765232826623556e-09,
      "loss": 0.7682,
      "step": 10828
    },
    {
      "epoch": 0.976597375659467,
      "grad_norm": 1.9641462722429481,
      "learning_rate": 5.720991280633081e-09,
      "loss": 0.7192,
      "step": 10829
    },
    {
      "epoch": 0.9766875591829373,
      "grad_norm": 1.4464543136781844,
      "learning_rate": 5.676919897523724e-09,
      "loss": 0.7233,
      "step": 10830
    },
    {
      "epoch": 0.9767777427064075,
      "grad_norm": 1.4986485114904133,
      "learning_rate": 5.633018681056256e-09,
      "loss": 0.7434,
      "step": 10831
    },
    {
      "epoch": 0.9768679262298778,
      "grad_norm": 3.029304745416114,
      "learning_rate": 5.589287634976569e-09,
      "loss": 0.7406,
      "step": 10832
    },
    {
      "epoch": 0.9769581097533481,
      "grad_norm": 1.6372246518027331,
      "learning_rate": 5.5457267630159014e-09,
      "loss": 0.7106,
      "step": 10833
    },
    {
      "epoch": 0.9770482932768183,
      "grad_norm": 2.7176701823888347,
      "learning_rate": 5.5023360688910555e-09,
      "loss": 0.7772,
      "step": 10834
    },
    {
      "epoch": 0.9771384768002885,
      "grad_norm": 1.6235441875923615,
      "learning_rate": 5.459115556304183e-09,
      "loss": 0.7883,
      "step": 10835
    },
    {
      "epoch": 0.9772286603237589,
      "grad_norm": 1.9686595351006793,
      "learning_rate": 5.416065228943889e-09,
      "loss": 0.6708,
      "step": 10836
    },
    {
      "epoch": 0.9773188438472291,
      "grad_norm": 1.7057386868820281,
      "learning_rate": 5.373185090482568e-09,
      "loss": 0.6933,
      "step": 10837
    },
    {
      "epoch": 0.9774090273706993,
      "grad_norm": 1.6592449963465314,
      "learning_rate": 5.330475144579516e-09,
      "loss": 0.7191,
      "step": 10838
    },
    {
      "epoch": 0.9774992108941697,
      "grad_norm": 1.4490107055384425,
      "learning_rate": 5.2879353948787065e-09,
      "loss": 0.8264,
      "step": 10839
    },
    {
      "epoch": 0.9775893944176399,
      "grad_norm": 2.035310062153377,
      "learning_rate": 5.245565845010125e-09,
      "loss": 0.7988,
      "step": 10840
    },
    {
      "epoch": 0.9776795779411102,
      "grad_norm": 1.4763523335724602,
      "learning_rate": 5.2033664985886575e-09,
      "loss": 0.6713,
      "step": 10841
    },
    {
      "epoch": 0.9777697614645804,
      "grad_norm": 1.6927489134007974,
      "learning_rate": 5.161337359215201e-09,
      "loss": 0.8827,
      "step": 10842
    },
    {
      "epoch": 0.9778599449880507,
      "grad_norm": 1.9069270823219988,
      "learning_rate": 5.119478430475999e-09,
      "loss": 0.7516,
      "step": 10843
    },
    {
      "epoch": 0.977950128511521,
      "grad_norm": 1.9759608623397102,
      "learning_rate": 5.077789715942416e-09,
      "loss": 0.6881,
      "step": 10844
    },
    {
      "epoch": 0.9780403120349912,
      "grad_norm": 2.0825024155766467,
      "learning_rate": 5.036271219171606e-09,
      "loss": 0.7641,
      "step": 10845
    },
    {
      "epoch": 0.9781304955584614,
      "grad_norm": 4.777327979647417,
      "learning_rate": 4.994922943706514e-09,
      "loss": 0.7766,
      "step": 10846
    },
    {
      "epoch": 0.9782206790819318,
      "grad_norm": 1.6546479160087648,
      "learning_rate": 4.953744893074763e-09,
      "loss": 0.7197,
      "step": 10847
    },
    {
      "epoch": 0.978310862605402,
      "grad_norm": 2.379892630198769,
      "learning_rate": 4.912737070789985e-09,
      "loss": 0.6671,
      "step": 10848
    },
    {
      "epoch": 0.9784010461288722,
      "grad_norm": 1.7104222466783965,
      "learning_rate": 4.871899480351604e-09,
      "loss": 0.8161,
      "step": 10849
    },
    {
      "epoch": 0.9784912296523425,
      "grad_norm": 1.645380014046907,
      "learning_rate": 4.831232125243501e-09,
      "loss": 0.7555,
      "step": 10850
    },
    {
      "epoch": 0.9785814131758128,
      "grad_norm": 2.6032788912379683,
      "learning_rate": 4.7907350089360086e-09,
      "loss": 0.8421,
      "step": 10851
    },
    {
      "epoch": 0.978671596699283,
      "grad_norm": 1.676623721751425,
      "learning_rate": 4.750408134884365e-09,
      "loss": 0.6383,
      "step": 10852
    },
    {
      "epoch": 0.9787617802227533,
      "grad_norm": 1.6367325884614115,
      "learning_rate": 4.710251506529816e-09,
      "loss": 0.6494,
      "step": 10853
    },
    {
      "epoch": 0.9788519637462235,
      "grad_norm": 1.5424827099886773,
      "learning_rate": 4.6702651272982894e-09,
      "loss": 0.7559,
      "step": 10854
    },
    {
      "epoch": 0.9789421472696939,
      "grad_norm": 1.6682745409783541,
      "learning_rate": 4.630449000602166e-09,
      "loss": 0.7148,
      "step": 10855
    },
    {
      "epoch": 0.9790323307931641,
      "grad_norm": 2.7187278132793975,
      "learning_rate": 4.590803129838283e-09,
      "loss": 0.8061,
      "step": 10856
    },
    {
      "epoch": 0.9791225143166343,
      "grad_norm": 1.562310550648651,
      "learning_rate": 4.551327518389714e-09,
      "loss": 0.7475,
      "step": 10857
    },
    {
      "epoch": 0.9792126978401046,
      "grad_norm": 2.0062508651398785,
      "learning_rate": 4.512022169624652e-09,
      "loss": 0.7306,
      "step": 10858
    },
    {
      "epoch": 0.9793028813635749,
      "grad_norm": 1.7148320712515186,
      "learning_rate": 4.472887086896637e-09,
      "loss": 0.7126,
      "step": 10859
    },
    {
      "epoch": 0.9793930648870451,
      "grad_norm": 1.7837455400866948,
      "learning_rate": 4.433922273545443e-09,
      "loss": 0.6905,
      "step": 10860
    },
    {
      "epoch": 0.9794832484105154,
      "grad_norm": 0.6064161702155302,
      "learning_rate": 4.395127732895299e-09,
      "loss": 0.5813,
      "step": 10861
    },
    {
      "epoch": 0.9795734319339856,
      "grad_norm": 2.0932073316900937,
      "learning_rate": 4.356503468256445e-09,
      "loss": 0.8219,
      "step": 10862
    },
    {
      "epoch": 0.979663615457456,
      "grad_norm": 2.079210499802806,
      "learning_rate": 4.318049482924913e-09,
      "loss": 0.7516,
      "step": 10863
    },
    {
      "epoch": 0.9797537989809262,
      "grad_norm": 1.8250502461871532,
      "learning_rate": 4.279765780181188e-09,
      "loss": 0.7079,
      "step": 10864
    },
    {
      "epoch": 0.9798439825043964,
      "grad_norm": 2.279250431323097,
      "learning_rate": 4.241652363291992e-09,
      "loss": 0.7569,
      "step": 10865
    },
    {
      "epoch": 0.9799341660278668,
      "grad_norm": 9.945852164634347,
      "learning_rate": 4.203709235509834e-09,
      "loss": 0.7128,
      "step": 10866
    },
    {
      "epoch": 0.980024349551337,
      "grad_norm": 2.3447646932746564,
      "learning_rate": 4.165936400071679e-09,
      "loss": 0.7587,
      "step": 10867
    },
    {
      "epoch": 0.9801145330748072,
      "grad_norm": 1.6579230872975275,
      "learning_rate": 4.12833386020095e-09,
      "loss": 0.6896,
      "step": 10868
    },
    {
      "epoch": 0.9802047165982775,
      "grad_norm": 2.5042510130054687,
      "learning_rate": 4.090901619105746e-09,
      "loss": 0.7807,
      "step": 10869
    },
    {
      "epoch": 0.9802949001217478,
      "grad_norm": 1.5765347163584902,
      "learning_rate": 4.053639679980181e-09,
      "loss": 0.6323,
      "step": 10870
    },
    {
      "epoch": 0.980385083645218,
      "grad_norm": 1.8081460155501283,
      "learning_rate": 4.01654804600371e-09,
      "loss": 0.7486,
      "step": 10871
    },
    {
      "epoch": 0.9804752671686883,
      "grad_norm": 2.1295955556947805,
      "learning_rate": 3.9796267203409114e-09,
      "loss": 0.7497,
      "step": 10872
    },
    {
      "epoch": 0.9805654506921585,
      "grad_norm": 1.4725089163906946,
      "learning_rate": 3.942875706142379e-09,
      "loss": 0.7177,
      "step": 10873
    },
    {
      "epoch": 0.9806556342156288,
      "grad_norm": 2.130882535785851,
      "learning_rate": 3.906295006543825e-09,
      "loss": 0.801,
      "step": 10874
    },
    {
      "epoch": 0.9807458177390991,
      "grad_norm": 1.739003783517487,
      "learning_rate": 3.8698846246665305e-09,
      "loss": 0.7197,
      "step": 10875
    },
    {
      "epoch": 0.9808360012625693,
      "grad_norm": 1.8417205632926523,
      "learning_rate": 3.833644563617344e-09,
      "loss": 0.7434,
      "step": 10876
    },
    {
      "epoch": 0.9809261847860395,
      "grad_norm": 3.2176163538099405,
      "learning_rate": 3.797574826488237e-09,
      "loss": 0.7541,
      "step": 10877
    },
    {
      "epoch": 0.9810163683095099,
      "grad_norm": 3.6058192056958482,
      "learning_rate": 3.761675416356969e-09,
      "loss": 0.6681,
      "step": 10878
    },
    {
      "epoch": 0.9811065518329801,
      "grad_norm": 1.5328290154892386,
      "learning_rate": 3.725946336286867e-09,
      "loss": 0.6482,
      "step": 10879
    },
    {
      "epoch": 0.9811967353564504,
      "grad_norm": 2.0690698889188623,
      "learning_rate": 3.6903875893261604e-09,
      "loss": 0.7158,
      "step": 10880
    },
    {
      "epoch": 0.9812869188799206,
      "grad_norm": 1.6577482104981414,
      "learning_rate": 3.6549991785093105e-09,
      "loss": 0.6726,
      "step": 10881
    },
    {
      "epoch": 0.9813771024033909,
      "grad_norm": 2.2940394746398884,
      "learning_rate": 3.6197811068554575e-09,
      "loss": 0.7306,
      "step": 10882
    },
    {
      "epoch": 0.9814672859268612,
      "grad_norm": 1.6061535045568327,
      "learning_rate": 3.584733377369975e-09,
      "loss": 0.8262,
      "step": 10883
    },
    {
      "epoch": 0.9815574694503314,
      "grad_norm": 5.295078209145432,
      "learning_rate": 3.549855993043138e-09,
      "loss": 0.7227,
      "step": 10884
    },
    {
      "epoch": 0.9816476529738016,
      "grad_norm": 1.7860938350039073,
      "learning_rate": 3.5151489568507887e-09,
      "loss": 0.7235,
      "step": 10885
    },
    {
      "epoch": 0.981737836497272,
      "grad_norm": 2.060975291474668,
      "learning_rate": 3.4806122717545572e-09,
      "loss": 0.7572,
      "step": 10886
    },
    {
      "epoch": 0.9818280200207422,
      "grad_norm": 1.7461632494994814,
      "learning_rate": 3.446245940701198e-09,
      "loss": 0.7034,
      "step": 10887
    },
    {
      "epoch": 0.9819182035442124,
      "grad_norm": 1.4830490614993446,
      "learning_rate": 3.41204996662281e-09,
      "loss": 0.6799,
      "step": 10888
    },
    {
      "epoch": 0.9820083870676828,
      "grad_norm": 1.352125061794074,
      "learning_rate": 3.3780243524375028e-09,
      "loss": 0.7367,
      "step": 10889
    },
    {
      "epoch": 0.982098570591153,
      "grad_norm": 1.7363411009296146,
      "learning_rate": 3.3441691010485107e-09,
      "loss": 0.7674,
      "step": 10890
    },
    {
      "epoch": 0.9821887541146233,
      "grad_norm": 1.7068797028964613,
      "learning_rate": 3.3104842153444113e-09,
      "loss": 0.8164,
      "step": 10891
    },
    {
      "epoch": 0.9822789376380935,
      "grad_norm": 1.7088444418157152,
      "learning_rate": 3.27696969819935e-09,
      "loss": 0.736,
      "step": 10892
    },
    {
      "epoch": 0.9823691211615638,
      "grad_norm": 1.3714084535780762,
      "learning_rate": 3.2436255524732615e-09,
      "loss": 0.6951,
      "step": 10893
    },
    {
      "epoch": 0.9824593046850341,
      "grad_norm": 1.87713812671529,
      "learning_rate": 3.210451781010759e-09,
      "loss": 0.7212,
      "step": 10894
    },
    {
      "epoch": 0.9825494882085043,
      "grad_norm": 4.085270843678658,
      "learning_rate": 3.1774483866426895e-09,
      "loss": 0.6031,
      "step": 10895
    },
    {
      "epoch": 0.9826396717319745,
      "grad_norm": 1.6827452041360411,
      "learning_rate": 3.144615372185244e-09,
      "loss": 0.7063,
      "step": 10896
    },
    {
      "epoch": 0.9827298552554449,
      "grad_norm": 1.5584950810227223,
      "learning_rate": 3.1119527404399604e-09,
      "loss": 0.83,
      "step": 10897
    },
    {
      "epoch": 0.9828200387789151,
      "grad_norm": 2.798681943474024,
      "learning_rate": 3.0794604941932754e-09,
      "loss": 0.7427,
      "step": 10898
    },
    {
      "epoch": 0.9829102223023853,
      "grad_norm": 1.6002770452253634,
      "learning_rate": 3.0471386362180827e-09,
      "loss": 0.6404,
      "step": 10899
    },
    {
      "epoch": 0.9830004058258556,
      "grad_norm": 3.529951143656116,
      "learning_rate": 3.0149871692719542e-09,
      "loss": 0.7912,
      "step": 10900
    },
    {
      "epoch": 0.9830905893493259,
      "grad_norm": 1.4290831075250665,
      "learning_rate": 2.9830060960984728e-09,
      "loss": 0.7255,
      "step": 10901
    },
    {
      "epoch": 0.9831807728727961,
      "grad_norm": 0.7025350426862991,
      "learning_rate": 2.9511954194263442e-09,
      "loss": 0.5842,
      "step": 10902
    },
    {
      "epoch": 0.9832709563962664,
      "grad_norm": 1.9453218145798585,
      "learning_rate": 2.9195551419698426e-09,
      "loss": 0.7411,
      "step": 10903
    },
    {
      "epoch": 0.9833611399197366,
      "grad_norm": 9.30486583779032,
      "learning_rate": 2.888085266428808e-09,
      "loss": 0.7013,
      "step": 10904
    },
    {
      "epoch": 0.983451323443207,
      "grad_norm": 2.1352836629493854,
      "learning_rate": 2.8567857954882037e-09,
      "loss": 0.697,
      "step": 10905
    },
    {
      "epoch": 0.9835415069666772,
      "grad_norm": 1.724727363049713,
      "learning_rate": 2.82565673181856e-09,
      "loss": 0.7939,
      "step": 10906
    },
    {
      "epoch": 0.9836316904901474,
      "grad_norm": 1.7111493680580014,
      "learning_rate": 2.7946980780764184e-09,
      "loss": 0.7604,
      "step": 10907
    },
    {
      "epoch": 0.9837218740136177,
      "grad_norm": 1.9099656460453645,
      "learning_rate": 2.763909836903e-09,
      "loss": 0.7975,
      "step": 10908
    },
    {
      "epoch": 0.983812057537088,
      "grad_norm": 1.63821270116652,
      "learning_rate": 2.7332920109255364e-09,
      "loss": 0.6547,
      "step": 10909
    },
    {
      "epoch": 0.9839022410605582,
      "grad_norm": 1.5839570556994047,
      "learning_rate": 2.702844602756382e-09,
      "loss": 0.7188,
      "step": 10910
    },
    {
      "epoch": 0.9839924245840285,
      "grad_norm": 2.319545220820906,
      "learning_rate": 2.6725676149936814e-09,
      "loss": 0.7201,
      "step": 10911
    },
    {
      "epoch": 0.9840826081074988,
      "grad_norm": 1.8305952548615387,
      "learning_rate": 2.642461050220479e-09,
      "loss": 0.8057,
      "step": 10912
    },
    {
      "epoch": 0.984172791630969,
      "grad_norm": 1.4865178752790595,
      "learning_rate": 2.612524911005609e-09,
      "loss": 0.7224,
      "step": 10913
    },
    {
      "epoch": 0.9842629751544393,
      "grad_norm": 1.5606357897659338,
      "learning_rate": 2.582759199903917e-09,
      "loss": 0.6157,
      "step": 10914
    },
    {
      "epoch": 0.9843531586779095,
      "grad_norm": 1.5910744742100886,
      "learning_rate": 2.553163919454704e-09,
      "loss": 0.7008,
      "step": 10915
    },
    {
      "epoch": 0.9844433422013799,
      "grad_norm": 1.7751352527290891,
      "learning_rate": 2.523739072183506e-09,
      "loss": 0.7218,
      "step": 10916
    },
    {
      "epoch": 0.9845335257248501,
      "grad_norm": 1.5937780489977778,
      "learning_rate": 2.4944846606007597e-09,
      "loss": 0.6983,
      "step": 10917
    },
    {
      "epoch": 0.9846237092483203,
      "grad_norm": 0.6148667220946401,
      "learning_rate": 2.46540068720269e-09,
      "loss": 0.6153,
      "step": 10918
    },
    {
      "epoch": 0.9847138927717906,
      "grad_norm": 3.172243895478004,
      "learning_rate": 2.4364871544708674e-09,
      "loss": 0.7792,
      "step": 10919
    },
    {
      "epoch": 0.9848040762952609,
      "grad_norm": 2.225852451544658,
      "learning_rate": 2.4077440648726523e-09,
      "loss": 0.6886,
      "step": 10920
    },
    {
      "epoch": 0.9848942598187311,
      "grad_norm": 1.7896540301047275,
      "learning_rate": 2.379171420860082e-09,
      "loss": 0.8055,
      "step": 10921
    },
    {
      "epoch": 0.9849844433422014,
      "grad_norm": 2.2736821518181336,
      "learning_rate": 2.3507692248714296e-09,
      "loss": 0.6619,
      "step": 10922
    },
    {
      "epoch": 0.9850746268656716,
      "grad_norm": 1.7295402291861177,
      "learning_rate": 2.322537479330089e-09,
      "loss": 0.7519,
      "step": 10923
    },
    {
      "epoch": 0.9851648103891419,
      "grad_norm": 1.6034401656318058,
      "learning_rate": 2.2944761866450223e-09,
      "loss": 0.8156,
      "step": 10924
    },
    {
      "epoch": 0.9852549939126122,
      "grad_norm": 2.711261740870576,
      "learning_rate": 2.266585349210315e-09,
      "loss": 0.7231,
      "step": 10925
    },
    {
      "epoch": 0.9853451774360824,
      "grad_norm": 2.3548285530029753,
      "learning_rate": 2.2388649694060623e-09,
      "loss": 0.6987,
      "step": 10926
    },
    {
      "epoch": 0.9854353609595526,
      "grad_norm": 1.759859105886264,
      "learning_rate": 2.211315049597262e-09,
      "loss": 0.8,
      "step": 10927
    },
    {
      "epoch": 0.985525544483023,
      "grad_norm": 1.599356793221361,
      "learning_rate": 2.1839355921349224e-09,
      "loss": 0.7741,
      "step": 10928
    },
    {
      "epoch": 0.9856157280064932,
      "grad_norm": 2.0545282918653602,
      "learning_rate": 2.156726599354952e-09,
      "loss": 0.7307,
      "step": 10929
    },
    {
      "epoch": 0.9857059115299635,
      "grad_norm": 1.8234716678611795,
      "learning_rate": 2.129688073578828e-09,
      "loss": 0.7634,
      "step": 10930
    },
    {
      "epoch": 0.9857960950534337,
      "grad_norm": 2.3925940688244274,
      "learning_rate": 2.1028200171142597e-09,
      "loss": 0.8723,
      "step": 10931
    },
    {
      "epoch": 0.985886278576904,
      "grad_norm": 2.2961762072439336,
      "learning_rate": 2.076122432253191e-09,
      "loss": 0.7215,
      "step": 10932
    },
    {
      "epoch": 0.9859764621003743,
      "grad_norm": 2.3707241920418887,
      "learning_rate": 2.0495953212738005e-09,
      "loss": 0.7746,
      "step": 10933
    },
    {
      "epoch": 0.9860666456238445,
      "grad_norm": 1.399453758214474,
      "learning_rate": 2.0232386864396102e-09,
      "loss": 0.8193,
      "step": 10934
    },
    {
      "epoch": 0.9861568291473148,
      "grad_norm": 2.1748357118844046,
      "learning_rate": 1.9970525299992656e-09,
      "loss": 0.7622,
      "step": 10935
    },
    {
      "epoch": 0.9862470126707851,
      "grad_norm": 1.9646864234626482,
      "learning_rate": 1.9710368541874245e-09,
      "loss": 0.7008,
      "step": 10936
    },
    {
      "epoch": 0.9863371961942553,
      "grad_norm": 0.5725794377665201,
      "learning_rate": 1.945191661223644e-09,
      "loss": 0.5866,
      "step": 10937
    },
    {
      "epoch": 0.9864273797177255,
      "grad_norm": 1.4975333595699638,
      "learning_rate": 1.9195169533132714e-09,
      "loss": 0.788,
      "step": 10938
    },
    {
      "epoch": 0.9865175632411959,
      "grad_norm": 0.5882927162704472,
      "learning_rate": 1.894012732646999e-09,
      "loss": 0.5937,
      "step": 10939
    },
    {
      "epoch": 0.9866077467646661,
      "grad_norm": 2.180133842244296,
      "learning_rate": 1.8686790014010854e-09,
      "loss": 0.6916,
      "step": 10940
    },
    {
      "epoch": 0.9866979302881363,
      "grad_norm": 0.6039181450268233,
      "learning_rate": 1.8435157617369134e-09,
      "loss": 0.5589,
      "step": 10941
    },
    {
      "epoch": 0.9867881138116066,
      "grad_norm": 1.6782300436719133,
      "learning_rate": 1.818523015801876e-09,
      "loss": 0.8787,
      "step": 10942
    },
    {
      "epoch": 0.9868782973350769,
      "grad_norm": 1.901443236975038,
      "learning_rate": 1.7937007657282677e-09,
      "loss": 0.7003,
      "step": 10943
    },
    {
      "epoch": 0.9869684808585472,
      "grad_norm": 1.9950574360851892,
      "learning_rate": 1.7690490136341718e-09,
      "loss": 0.685,
      "step": 10944
    },
    {
      "epoch": 0.9870586643820174,
      "grad_norm": 1.8250007341984356,
      "learning_rate": 1.744567761622795e-09,
      "loss": 0.7371,
      "step": 10945
    },
    {
      "epoch": 0.9871488479054876,
      "grad_norm": 1.6770867373350993,
      "learning_rate": 1.7202570117831327e-09,
      "loss": 0.7271,
      "step": 10946
    },
    {
      "epoch": 0.987239031428958,
      "grad_norm": 2.489400135998505,
      "learning_rate": 1.696116766189526e-09,
      "loss": 0.7189,
      "step": 10947
    },
    {
      "epoch": 0.9873292149524282,
      "grad_norm": 8.794480534881975,
      "learning_rate": 1.6721470269021042e-09,
      "loss": 0.7815,
      "step": 10948
    },
    {
      "epoch": 0.9874193984758984,
      "grad_norm": 2.0191886212767387,
      "learning_rate": 1.6483477959654546e-09,
      "loss": 0.7094,
      "step": 10949
    },
    {
      "epoch": 0.9875095819993687,
      "grad_norm": 1.850999126757679,
      "learning_rate": 1.6247190754106187e-09,
      "loss": 0.7208,
      "step": 10950
    },
    {
      "epoch": 0.987599765522839,
      "grad_norm": 1.4416916314023405,
      "learning_rate": 1.6012608672537619e-09,
      "loss": 0.7052,
      "step": 10951
    },
    {
      "epoch": 0.9876899490463092,
      "grad_norm": 3.390206306728996,
      "learning_rate": 1.5779731734963942e-09,
      "loss": 0.7501,
      "step": 10952
    },
    {
      "epoch": 0.9877801325697795,
      "grad_norm": 16.449375054915176,
      "learning_rate": 1.5548559961253705e-09,
      "loss": 0.8042,
      "step": 10953
    },
    {
      "epoch": 0.9878703160932497,
      "grad_norm": 1.6388717729497286,
      "learning_rate": 1.5319093371135573e-09,
      "loss": 0.6655,
      "step": 10954
    },
    {
      "epoch": 0.9879604996167201,
      "grad_norm": 1.6509729926651633,
      "learning_rate": 1.5091331984184997e-09,
      "loss": 0.8263,
      "step": 10955
    },
    {
      "epoch": 0.9880506831401903,
      "grad_norm": 1.5346685901704915,
      "learning_rate": 1.486527581983754e-09,
      "loss": 0.735,
      "step": 10956
    },
    {
      "epoch": 0.9881408666636605,
      "grad_norm": 2.990982012032381,
      "learning_rate": 1.4640924897382223e-09,
      "loss": 0.8026,
      "step": 10957
    },
    {
      "epoch": 0.9882310501871309,
      "grad_norm": 1.5502667474406508,
      "learning_rate": 1.4418279235961506e-09,
      "loss": 0.78,
      "step": 10958
    },
    {
      "epoch": 0.9883212337106011,
      "grad_norm": 1.7147745105433767,
      "learning_rate": 1.4197338854573526e-09,
      "loss": 0.6441,
      "step": 10959
    },
    {
      "epoch": 0.9884114172340713,
      "grad_norm": 1.9126335571129551,
      "learning_rate": 1.3978103772067651e-09,
      "loss": 0.6974,
      "step": 10960
    },
    {
      "epoch": 0.9885016007575416,
      "grad_norm": 1.864346245835838,
      "learning_rate": 1.3760574007153358e-09,
      "loss": 0.6603,
      "step": 10961
    },
    {
      "epoch": 0.9885917842810119,
      "grad_norm": 2.390197516224327,
      "learning_rate": 1.3544749578389137e-09,
      "loss": 0.8192,
      "step": 10962
    },
    {
      "epoch": 0.9886819678044821,
      "grad_norm": 1.6184299874375352,
      "learning_rate": 1.3330630504189143e-09,
      "loss": 0.7921,
      "step": 10963
    },
    {
      "epoch": 0.9887721513279524,
      "grad_norm": 1.953329212956862,
      "learning_rate": 1.3118216802827652e-09,
      "loss": 0.7068,
      "step": 10964
    },
    {
      "epoch": 0.9888623348514226,
      "grad_norm": 2.05256061199614,
      "learning_rate": 1.2907508492425722e-09,
      "loss": 0.7096,
      "step": 10965
    },
    {
      "epoch": 0.988952518374893,
      "grad_norm": 1.7610216474356062,
      "learning_rate": 1.2698505590962305e-09,
      "loss": 0.7026,
      "step": 10966
    },
    {
      "epoch": 0.9890427018983632,
      "grad_norm": 1.6598972971295007,
      "learning_rate": 1.2491208116272022e-09,
      "loss": 0.7372,
      "step": 10967
    },
    {
      "epoch": 0.9891328854218334,
      "grad_norm": 1.7804664092908455,
      "learning_rate": 1.2285616086040728e-09,
      "loss": 0.7719,
      "step": 10968
    },
    {
      "epoch": 0.9892230689453037,
      "grad_norm": 1.3741372609659843,
      "learning_rate": 1.2081729517812167e-09,
      "loss": 0.8028,
      "step": 10969
    },
    {
      "epoch": 0.989313252468774,
      "grad_norm": 1.5744842229726828,
      "learning_rate": 1.1879548428983533e-09,
      "loss": 0.7382,
      "step": 10970
    },
    {
      "epoch": 0.9894034359922442,
      "grad_norm": 1.6779851759300108,
      "learning_rate": 1.167907283680547e-09,
      "loss": 0.6901,
      "step": 10971
    },
    {
      "epoch": 0.9894936195157145,
      "grad_norm": 1.5785808848989833,
      "learning_rate": 1.1480302758382077e-09,
      "loss": 0.6806,
      "step": 10972
    },
    {
      "epoch": 0.9895838030391847,
      "grad_norm": 1.6010758567771433,
      "learning_rate": 1.1283238210675338e-09,
      "loss": 0.7297,
      "step": 10973
    },
    {
      "epoch": 0.989673986562655,
      "grad_norm": 2.4648308851870593,
      "learning_rate": 1.1087879210498474e-09,
      "loss": 0.6691,
      "step": 10974
    },
    {
      "epoch": 0.9897641700861253,
      "grad_norm": 1.6413242847979286,
      "learning_rate": 1.0894225774522592e-09,
      "loss": 0.7855,
      "step": 10975
    },
    {
      "epoch": 0.9898543536095955,
      "grad_norm": 0.6960678712292174,
      "learning_rate": 1.070227791927003e-09,
      "loss": 0.5953,
      "step": 10976
    },
    {
      "epoch": 0.9899445371330657,
      "grad_norm": 3.2196331108954657,
      "learning_rate": 1.0512035661118802e-09,
      "loss": 0.8404,
      "step": 10977
    },
    {
      "epoch": 0.9900347206565361,
      "grad_norm": 2.067619328220716,
      "learning_rate": 1.0323499016300364e-09,
      "loss": 0.761,
      "step": 10978
    },
    {
      "epoch": 0.9901249041800063,
      "grad_norm": 1.7667894152817591,
      "learning_rate": 1.013666800090407e-09,
      "loss": 0.6918,
      "step": 10979
    },
    {
      "epoch": 0.9902150877034765,
      "grad_norm": 8.16153426014326,
      "learning_rate": 9.951542630870502e-10,
      "loss": 0.724,
      "step": 10980
    },
    {
      "epoch": 0.9903052712269468,
      "grad_norm": 2.056028791315081,
      "learning_rate": 9.768122921995915e-10,
      "loss": 0.78,
      "step": 10981
    },
    {
      "epoch": 0.9903954547504171,
      "grad_norm": 2.143559247752837,
      "learning_rate": 9.58640888992779e-10,
      "loss": 0.6807,
      "step": 10982
    },
    {
      "epoch": 0.9904856382738874,
      "grad_norm": 1.7149895191150406,
      "learning_rate": 9.40640055017594e-10,
      "loss": 0.6549,
      "step": 10983
    },
    {
      "epoch": 0.9905758217973576,
      "grad_norm": 1.622834817228605,
      "learning_rate": 9.228097918094757e-10,
      "loss": 0.7401,
      "step": 10984
    },
    {
      "epoch": 0.9906660053208279,
      "grad_norm": 1.9520071874509657,
      "learning_rate": 9.051501008900952e-10,
      "loss": 0.7115,
      "step": 10985
    },
    {
      "epoch": 0.9907561888442982,
      "grad_norm": 2.115787057008684,
      "learning_rate": 8.876609837662475e-10,
      "loss": 0.7544,
      "step": 10986
    },
    {
      "epoch": 0.9908463723677684,
      "grad_norm": 1.4885380240578883,
      "learning_rate": 8.70342441930294e-10,
      "loss": 0.8266,
      "step": 10987
    },
    {
      "epoch": 0.9909365558912386,
      "grad_norm": 1.7601122922187329,
      "learning_rate": 8.531944768594979e-10,
      "loss": 0.71,
      "step": 10988
    },
    {
      "epoch": 0.991026739414709,
      "grad_norm": 3.1405049011648942,
      "learning_rate": 8.362170900175769e-10,
      "loss": 0.8199,
      "step": 10989
    },
    {
      "epoch": 0.9911169229381792,
      "grad_norm": 7.27508074004474,
      "learning_rate": 8.194102828527061e-10,
      "loss": 0.697,
      "step": 10990
    },
    {
      "epoch": 0.9912071064616494,
      "grad_norm": 1.7379634266594928,
      "learning_rate": 8.027740567992936e-10,
      "loss": 0.7826,
      "step": 10991
    },
    {
      "epoch": 0.9912972899851197,
      "grad_norm": 1.5112236848595717,
      "learning_rate": 7.863084132766484e-10,
      "loss": 0.7373,
      "step": 10992
    },
    {
      "epoch": 0.99138747350859,
      "grad_norm": 1.6801104345809221,
      "learning_rate": 7.700133536896469e-10,
      "loss": 0.7556,
      "step": 10993
    },
    {
      "epoch": 0.9914776570320603,
      "grad_norm": 1.5012617367707306,
      "learning_rate": 7.538888794287324e-10,
      "loss": 0.7104,
      "step": 10994
    },
    {
      "epoch": 0.9915678405555305,
      "grad_norm": 1.5444484134702354,
      "learning_rate": 7.379349918696931e-10,
      "loss": 0.7621,
      "step": 10995
    },
    {
      "epoch": 0.9916580240790007,
      "grad_norm": 1.8276799630244926,
      "learning_rate": 7.221516923738846e-10,
      "loss": 0.715,
      "step": 10996
    },
    {
      "epoch": 0.9917482076024711,
      "grad_norm": 2.075874544171306,
      "learning_rate": 7.065389822880075e-10,
      "loss": 0.8462,
      "step": 10997
    },
    {
      "epoch": 0.9918383911259413,
      "grad_norm": 3.246765875061049,
      "learning_rate": 6.910968629443292e-10,
      "loss": 0.8174,
      "step": 10998
    },
    {
      "epoch": 0.9919285746494115,
      "grad_norm": 1.8482880429542856,
      "learning_rate": 6.758253356602406e-10,
      "loss": 0.7359,
      "step": 10999
    },
    {
      "epoch": 0.9920187581728818,
      "grad_norm": 1.5529229967244007,
      "learning_rate": 6.607244017389213e-10,
      "loss": 0.8321,
      "step": 11000
    },
    {
      "epoch": 0.9921089416963521,
      "grad_norm": 2.5034358256659064,
      "learning_rate": 6.457940624686742e-10,
      "loss": 0.7616,
      "step": 11001
    },
    {
      "epoch": 0.9921991252198223,
      "grad_norm": 1.9188651109393027,
      "learning_rate": 6.310343191238132e-10,
      "loss": 0.8215,
      "step": 11002
    },
    {
      "epoch": 0.9922893087432926,
      "grad_norm": 2.1598319556954566,
      "learning_rate": 6.164451729635534e-10,
      "loss": 0.6369,
      "step": 11003
    },
    {
      "epoch": 0.9923794922667628,
      "grad_norm": 2.431816125381106,
      "learning_rate": 6.020266252324546e-10,
      "loss": 0.6157,
      "step": 11004
    },
    {
      "epoch": 0.9924696757902332,
      "grad_norm": 1.719567745918533,
      "learning_rate": 5.877786771610882e-10,
      "loss": 0.7122,
      "step": 11005
    },
    {
      "epoch": 0.9925598593137034,
      "grad_norm": 0.7498214827114977,
      "learning_rate": 5.737013299651483e-10,
      "loss": 0.6468,
      "step": 11006
    },
    {
      "epoch": 0.9926500428371736,
      "grad_norm": 1.718025956669432,
      "learning_rate": 5.597945848458963e-10,
      "loss": 0.7439,
      "step": 11007
    },
    {
      "epoch": 0.992740226360644,
      "grad_norm": 1.6026969055511502,
      "learning_rate": 5.460584429894944e-10,
      "loss": 0.7118,
      "step": 11008
    },
    {
      "epoch": 0.9928304098841142,
      "grad_norm": 3.4075961471505387,
      "learning_rate": 5.32492905568338e-10,
      "loss": 0.7428,
      "step": 11009
    },
    {
      "epoch": 0.9929205934075844,
      "grad_norm": 1.768433961165373,
      "learning_rate": 5.190979737399459e-10,
      "loss": 0.7758,
      "step": 11010
    },
    {
      "epoch": 0.9930107769310547,
      "grad_norm": 1.4809893092993727,
      "learning_rate": 5.058736486469594e-10,
      "loss": 0.6298,
      "step": 11011
    },
    {
      "epoch": 0.993100960454525,
      "grad_norm": 1.810364747994854,
      "learning_rate": 4.928199314180314e-10,
      "loss": 0.771,
      "step": 11012
    },
    {
      "epoch": 0.9931911439779952,
      "grad_norm": 2.46605303279905,
      "learning_rate": 4.799368231669376e-10,
      "loss": 0.6588,
      "step": 11013
    },
    {
      "epoch": 0.9932813275014655,
      "grad_norm": 2.061229198714046,
      "learning_rate": 4.672243249927988e-10,
      "loss": 0.7297,
      "step": 11014
    },
    {
      "epoch": 0.9933715110249357,
      "grad_norm": 1.5755100302417804,
      "learning_rate": 4.546824379803027e-10,
      "loss": 0.7684,
      "step": 11015
    },
    {
      "epoch": 0.993461694548406,
      "grad_norm": 2.229662854834342,
      "learning_rate": 4.4231116319970454e-10,
      "loss": 0.7653,
      "step": 11016
    },
    {
      "epoch": 0.9935518780718763,
      "grad_norm": 1.894435290236495,
      "learning_rate": 4.3011050170660423e-10,
      "loss": 0.7452,
      "step": 11017
    },
    {
      "epoch": 0.9936420615953465,
      "grad_norm": 2.047780775630415,
      "learning_rate": 4.18080454542169e-10,
      "loss": 0.7346,
      "step": 11018
    },
    {
      "epoch": 0.9937322451188167,
      "grad_norm": 0.6637858768328998,
      "learning_rate": 4.0622102273246694e-10,
      "loss": 0.5764,
      "step": 11019
    },
    {
      "epoch": 0.9938224286422871,
      "grad_norm": 2.3251051807197696,
      "learning_rate": 3.945322072897994e-10,
      "loss": 0.7986,
      "step": 11020
    },
    {
      "epoch": 0.9939126121657573,
      "grad_norm": 3.665058317878408,
      "learning_rate": 3.830140092111467e-10,
      "loss": 0.7969,
      "step": 11021
    },
    {
      "epoch": 0.9940027956892276,
      "grad_norm": 2.0885763442917424,
      "learning_rate": 3.7166642947972225e-10,
      "loss": 0.7409,
      "step": 11022
    },
    {
      "epoch": 0.9940929792126978,
      "grad_norm": 1.9752998390027356,
      "learning_rate": 3.604894690634186e-10,
      "loss": 0.7172,
      "step": 11023
    },
    {
      "epoch": 0.9941831627361681,
      "grad_norm": 6.669215223822077,
      "learning_rate": 3.494831289161393e-10,
      "loss": 0.8261,
      "step": 11024
    },
    {
      "epoch": 0.9942733462596384,
      "grad_norm": 1.6077599430669491,
      "learning_rate": 3.3864740997668897e-10,
      "loss": 0.7749,
      "step": 11025
    },
    {
      "epoch": 0.9943635297831086,
      "grad_norm": 1.5590965588111132,
      "learning_rate": 3.279823131701054e-10,
      "loss": 0.7638,
      "step": 11026
    },
    {
      "epoch": 0.9944537133065788,
      "grad_norm": 2.242018955820958,
      "learning_rate": 3.1748783940610536e-10,
      "loss": 0.6963,
      "step": 11027
    },
    {
      "epoch": 0.9945438968300492,
      "grad_norm": 1.5928932579902233,
      "learning_rate": 3.071639895801947e-10,
      "loss": 0.703,
      "step": 11028
    },
    {
      "epoch": 0.9946340803535194,
      "grad_norm": 1.8864102634629585,
      "learning_rate": 2.9701076457322447e-10,
      "loss": 0.806,
      "step": 11029
    },
    {
      "epoch": 0.9947242638769896,
      "grad_norm": 2.057433871090668,
      "learning_rate": 2.870281652513906e-10,
      "loss": 0.781,
      "step": 11030
    },
    {
      "epoch": 0.99481444740046,
      "grad_norm": 1.76422955217873,
      "learning_rate": 2.772161924669003e-10,
      "loss": 0.7278,
      "step": 11031
    },
    {
      "epoch": 0.9949046309239302,
      "grad_norm": 1.8716929199107668,
      "learning_rate": 2.6757484705641765e-10,
      "loss": 0.7257,
      "step": 11032
    },
    {
      "epoch": 0.9949948144474005,
      "grad_norm": 1.872316729307563,
      "learning_rate": 2.58104129843062e-10,
      "loss": 0.7434,
      "step": 11033
    },
    {
      "epoch": 0.9950849979708707,
      "grad_norm": 1.96596445084852,
      "learning_rate": 2.4880404163463154e-10,
      "loss": 0.6951,
      "step": 11034
    },
    {
      "epoch": 0.995175181494341,
      "grad_norm": 1.9286616142822997,
      "learning_rate": 2.3967458322471377e-10,
      "loss": 0.7297,
      "step": 11035
    },
    {
      "epoch": 0.9952653650178113,
      "grad_norm": 1.590921847180835,
      "learning_rate": 2.307157553922412e-10,
      "loss": 0.7539,
      "step": 11036
    },
    {
      "epoch": 0.9953555485412815,
      "grad_norm": 1.4965913507815454,
      "learning_rate": 2.2192755890193538e-10,
      "loss": 0.7818,
      "step": 11037
    },
    {
      "epoch": 0.9954457320647517,
      "grad_norm": 0.6218019682206751,
      "learning_rate": 2.133099945034189e-10,
      "loss": 0.546,
      "step": 11038
    },
    {
      "epoch": 0.9955359155882221,
      "grad_norm": 2.8135717257367863,
      "learning_rate": 2.048630629318815e-10,
      "loss": 0.7628,
      "step": 11039
    },
    {
      "epoch": 0.9956260991116923,
      "grad_norm": 1.4040218503099466,
      "learning_rate": 1.965867649080799e-10,
      "loss": 0.7359,
      "step": 11040
    },
    {
      "epoch": 0.9957162826351625,
      "grad_norm": 0.6129166558051956,
      "learning_rate": 1.8848110113856008e-10,
      "loss": 0.5637,
      "step": 11041
    },
    {
      "epoch": 0.9958064661586328,
      "grad_norm": 1.6625131190649993,
      "learning_rate": 1.8054607231454687e-10,
      "loss": 0.78,
      "step": 11042
    },
    {
      "epoch": 0.9958966496821031,
      "grad_norm": 1.5638271798802144,
      "learning_rate": 1.7278167911327635e-10,
      "loss": 0.6959,
      "step": 11043
    },
    {
      "epoch": 0.9959868332055734,
      "grad_norm": 1.6053699175261067,
      "learning_rate": 1.6518792219710753e-10,
      "loss": 0.7748,
      "step": 11044
    },
    {
      "epoch": 0.9960770167290436,
      "grad_norm": 1.6930098827454405,
      "learning_rate": 1.5776480221418865e-10,
      "loss": 0.7465,
      "step": 11045
    },
    {
      "epoch": 0.9961672002525138,
      "grad_norm": 3.9131547861347147,
      "learning_rate": 1.505123197977909e-10,
      "loss": 0.7448,
      "step": 11046
    },
    {
      "epoch": 0.9962573837759842,
      "grad_norm": 1.724372061173526,
      "learning_rate": 1.4343047556675258e-10,
      "loss": 0.678,
      "step": 11047
    },
    {
      "epoch": 0.9963475672994544,
      "grad_norm": 2.2647573980296283,
      "learning_rate": 1.3651927012503506e-10,
      "loss": 0.7437,
      "step": 11048
    },
    {
      "epoch": 0.9964377508229246,
      "grad_norm": 1.8777323840872533,
      "learning_rate": 1.297787040630549e-10,
      "loss": 0.7462,
      "step": 11049
    },
    {
      "epoch": 0.9965279343463949,
      "grad_norm": 2.683422810581046,
      "learning_rate": 1.2320877795524153e-10,
      "loss": 0.7676,
      "step": 11050
    },
    {
      "epoch": 0.9966181178698652,
      "grad_norm": 2.016188668384235,
      "learning_rate": 1.1680949236247962e-10,
      "loss": 0.7245,
      "step": 11051
    },
    {
      "epoch": 0.9967083013933354,
      "grad_norm": 1.57868600070773,
      "learning_rate": 1.1058084783099886e-10,
      "loss": 0.7477,
      "step": 11052
    },
    {
      "epoch": 0.9967984849168057,
      "grad_norm": 2.211507339641303,
      "learning_rate": 1.0452284489170793e-10,
      "loss": 0.7533,
      "step": 11053
    },
    {
      "epoch": 0.996888668440276,
      "grad_norm": 2.184767515091988,
      "learning_rate": 9.86354840621928e-11,
      "loss": 0.7262,
      "step": 11054
    },
    {
      "epoch": 0.9969788519637462,
      "grad_norm": 3.4887947823723167,
      "learning_rate": 9.291876584427427e-11,
      "loss": 0.6192,
      "step": 11055
    },
    {
      "epoch": 0.9970690354872165,
      "grad_norm": 2.131693340290717,
      "learning_rate": 8.737269072578435e-11,
      "loss": 0.7135,
      "step": 11056
    },
    {
      "epoch": 0.9971592190106867,
      "grad_norm": 1.4994367495658116,
      "learning_rate": 8.199725918012212e-11,
      "loss": 0.7389,
      "step": 11057
    },
    {
      "epoch": 0.9972494025341571,
      "grad_norm": 1.6607929758979938,
      "learning_rate": 7.679247166603175e-11,
      "loss": 0.7444,
      "step": 11058
    },
    {
      "epoch": 0.9973395860576273,
      "grad_norm": 1.844617477915555,
      "learning_rate": 7.17583286273804e-11,
      "loss": 0.7445,
      "step": 11059
    },
    {
      "epoch": 0.9974297695810975,
      "grad_norm": 1.809757854956306,
      "learning_rate": 6.689483049360233e-11,
      "loss": 0.7657,
      "step": 11060
    },
    {
      "epoch": 0.9975199531045678,
      "grad_norm": 2.3715028614121025,
      "learning_rate": 6.220197768014302e-11,
      "loss": 0.715,
      "step": 11061
    },
    {
      "epoch": 0.9976101366280381,
      "grad_norm": 1.999793876516644,
      "learning_rate": 5.7679770587126806e-11,
      "loss": 0.7602,
      "step": 11062
    },
    {
      "epoch": 0.9977003201515083,
      "grad_norm": 0.6164481164520272,
      "learning_rate": 5.33282096002452e-11,
      "loss": 0.6658,
      "step": 11063
    },
    {
      "epoch": 0.9977905036749786,
      "grad_norm": 1.9915265950489418,
      "learning_rate": 4.914729509120086e-11,
      "loss": 0.7134,
      "step": 11064
    },
    {
      "epoch": 0.9978806871984488,
      "grad_norm": 1.4740987740008686,
      "learning_rate": 4.513702741637537e-11,
      "loss": 0.6669,
      "step": 11065
    },
    {
      "epoch": 0.9979708707219191,
      "grad_norm": 3.2393231711638113,
      "learning_rate": 4.129740691816153e-11,
      "loss": 0.8042,
      "step": 11066
    },
    {
      "epoch": 0.9980610542453894,
      "grad_norm": 1.3799733039789281,
      "learning_rate": 3.762843392429715e-11,
      "loss": 0.7601,
      "step": 11067
    },
    {
      "epoch": 0.9981512377688596,
      "grad_norm": 1.8508771440776053,
      "learning_rate": 3.413010874742106e-11,
      "loss": 0.7199,
      "step": 11068
    },
    {
      "epoch": 0.9982414212923298,
      "grad_norm": 2.778526344101661,
      "learning_rate": 3.080243168618324e-11,
      "loss": 0.7348,
      "step": 11069
    },
    {
      "epoch": 0.9983316048158002,
      "grad_norm": 1.915173079933426,
      "learning_rate": 2.7645403024800783e-11,
      "loss": 0.6885,
      "step": 11070
    },
    {
      "epoch": 0.9984217883392704,
      "grad_norm": 1.357126239445417,
      "learning_rate": 2.4659023032391756e-11,
      "loss": 0.8092,
      "step": 11071
    },
    {
      "epoch": 0.9985119718627407,
      "grad_norm": 1.6403973376035073,
      "learning_rate": 2.1843291963863364e-11,
      "loss": 0.7861,
      "step": 11072
    },
    {
      "epoch": 0.9986021553862109,
      "grad_norm": 1.6125888770398173,
      "learning_rate": 1.9198210059245822e-11,
      "loss": 0.7571,
      "step": 11073
    },
    {
      "epoch": 0.9986923389096812,
      "grad_norm": 1.9919347864568018,
      "learning_rate": 1.672377754458054e-11,
      "loss": 0.7436,
      "step": 11074
    },
    {
      "epoch": 0.9987825224331515,
      "grad_norm": 20.00877188006394,
      "learning_rate": 1.4419994630809895e-11,
      "loss": 0.7323,
      "step": 11075
    },
    {
      "epoch": 0.9988727059566217,
      "grad_norm": 2.1621274594151094,
      "learning_rate": 1.2286861514443358e-11,
      "loss": 0.6932,
      "step": 11076
    },
    {
      "epoch": 0.998962889480092,
      "grad_norm": 1.3248851078532171,
      "learning_rate": 1.0324378377779553e-11,
      "loss": 0.7789,
      "step": 11077
    },
    {
      "epoch": 0.9990530730035623,
      "grad_norm": 2.974754284881345,
      "learning_rate": 8.532545388018064e-12,
      "loss": 0.7394,
      "step": 11078
    },
    {
      "epoch": 0.9991432565270325,
      "grad_norm": 1.8211917963722504,
      "learning_rate": 6.911362697925582e-12,
      "loss": 0.5767,
      "step": 11079
    },
    {
      "epoch": 0.9992334400505027,
      "grad_norm": 2.0852892150447815,
      "learning_rate": 5.46083044605794e-12,
      "loss": 0.717,
      "step": 11080
    },
    {
      "epoch": 0.9993236235739731,
      "grad_norm": 1.6458990429987688,
      "learning_rate": 4.1809487563160276e-12,
      "loss": 0.7922,
      "step": 11081
    },
    {
      "epoch": 0.9994138070974433,
      "grad_norm": 2.6489534578767273,
      "learning_rate": 3.0717177375017e-12,
      "loss": 0.7442,
      "step": 11082
    },
    {
      "epoch": 0.9995039906209136,
      "grad_norm": 1.8609619420878256,
      "learning_rate": 2.1331374846500495e-12,
      "loss": 0.7363,
      "step": 11083
    },
    {
      "epoch": 0.9995941741443838,
      "grad_norm": 2.0582749990415974,
      "learning_rate": 1.3652080774750885e-12,
      "loss": 0.838,
      "step": 11084
    },
    {
      "epoch": 0.9996843576678541,
      "grad_norm": 1.3818844628322762,
      "learning_rate": 7.679295817020204e-13,
      "loss": 0.6564,
      "step": 11085
    },
    {
      "epoch": 0.9997745411913244,
      "grad_norm": 1.7941582085669492,
      "learning_rate": 3.413020484011042e-13,
      "loss": 0.7195,
      "step": 11086
    },
    {
      "epoch": 0.9998647247147946,
      "grad_norm": 1.6475317758323569,
      "learning_rate": 8.53255139876552e-14,
      "loss": 0.7831,
      "step": 11087
    },
    {
      "epoch": 0.9999549082382648,
      "grad_norm": 3.1586735551382463,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 11088
    },
    {
      "epoch": 0.9999549082382648,
      "step": 11088,
      "total_flos": 1.347768590492762e+18,
      "train_loss": 0.5537439968522966,
      "train_runtime": 116542.4248,
      "train_samples_per_second": 5.709,
      "train_steps_per_second": 0.095
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 11088,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.347768590492762e+18,
  "train_batch_size": 5,
  "trial_name": null,
  "trial_params": null
}