{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.616272360069244,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002308136180034622,
      "grad_norm": 0.27939364314079285,
      "learning_rate": 4e-05,
      "loss": 0.3726,
      "step": 1
    },
    {
      "epoch": 0.004616272360069244,
      "grad_norm": 0.28028249740600586,
      "learning_rate": 8e-05,
      "loss": 0.3698,
      "step": 2
    },
    {
      "epoch": 0.006924408540103866,
      "grad_norm": 0.2671259641647339,
      "learning_rate": 0.00012,
      "loss": 0.3658,
      "step": 3
    },
    {
      "epoch": 0.009232544720138488,
      "grad_norm": 0.19880272448062897,
      "learning_rate": 0.00016,
      "loss": 0.3315,
      "step": 4
    },
    {
      "epoch": 0.01154068090017311,
      "grad_norm": 0.2226216048002243,
      "learning_rate": 0.0002,
      "loss": 0.2934,
      "step": 5
    },
    {
      "epoch": 0.013848817080207732,
      "grad_norm": 0.3117420971393585,
      "learning_rate": 0.0001999074074074074,
      "loss": 0.2647,
      "step": 6
    },
    {
      "epoch": 0.016156953260242354,
      "grad_norm": 0.3630733788013458,
      "learning_rate": 0.00019981481481481483,
      "loss": 0.2382,
      "step": 7
    },
    {
      "epoch": 0.018465089440276975,
      "grad_norm": 0.3318308889865875,
      "learning_rate": 0.00019972222222222223,
      "loss": 0.2215,
      "step": 8
    },
    {
      "epoch": 0.0207732256203116,
      "grad_norm": 0.10572563856840134,
      "learning_rate": 0.00019962962962962963,
      "loss": 0.2086,
      "step": 9
    },
    {
      "epoch": 0.02308136180034622,
      "grad_norm": 0.16791120171546936,
      "learning_rate": 0.00019953703703703705,
      "loss": 0.2088,
      "step": 10
    },
    {
      "epoch": 0.025389497980380843,
      "grad_norm": 0.0713539645075798,
      "learning_rate": 0.00019944444444444445,
      "loss": 0.2057,
      "step": 11
    },
    {
      "epoch": 0.027697634160415464,
      "grad_norm": 0.0983385220170021,
      "learning_rate": 0.00019935185185185184,
      "loss": 0.203,
      "step": 12
    },
    {
      "epoch": 0.030005770340450086,
      "grad_norm": 0.08346133679151535,
      "learning_rate": 0.00019925925925925927,
      "loss": 0.2009,
      "step": 13
    },
    {
      "epoch": 0.03231390652048471,
      "grad_norm": 0.07638801634311676,
      "learning_rate": 0.0001991666666666667,
      "loss": 0.1999,
      "step": 14
    },
    {
      "epoch": 0.03462204270051933,
      "grad_norm": 0.06702545285224915,
      "learning_rate": 0.0001990740740740741,
      "loss": 0.1929,
      "step": 15
    },
    {
      "epoch": 0.03693017888055395,
      "grad_norm": 0.08288706839084625,
      "learning_rate": 0.0001989814814814815,
      "loss": 0.1817,
      "step": 16
    },
    {
      "epoch": 0.03923831506058857,
      "grad_norm": 0.06332188099622726,
      "learning_rate": 0.0001988888888888889,
      "loss": 0.1909,
      "step": 17
    },
    {
      "epoch": 0.0415464512406232,
      "grad_norm": 0.05352393165230751,
      "learning_rate": 0.0001987962962962963,
      "loss": 0.1916,
      "step": 18
    },
    {
      "epoch": 0.04385458742065782,
      "grad_norm": 0.06091461330652237,
      "learning_rate": 0.0001987037037037037,
      "loss": 0.1894,
      "step": 19
    },
    {
      "epoch": 0.04616272360069244,
      "grad_norm": 0.05445568636059761,
      "learning_rate": 0.00019861111111111113,
      "loss": 0.1844,
      "step": 20
    },
    {
      "epoch": 0.048470859780727064,
      "grad_norm": 0.04017186164855957,
      "learning_rate": 0.00019851851851851853,
      "loss": 0.183,
      "step": 21
    },
    {
      "epoch": 0.050778995960761686,
      "grad_norm": 0.05530911684036255,
      "learning_rate": 0.00019842592592592592,
      "loss": 0.1832,
      "step": 22
    },
    {
      "epoch": 0.05308713214079631,
      "grad_norm": 0.05607778578996658,
      "learning_rate": 0.00019833333333333335,
      "loss": 0.1849,
      "step": 23
    },
    {
      "epoch": 0.05539526832083093,
      "grad_norm": 0.0359547920525074,
      "learning_rate": 0.00019824074074074074,
      "loss": 0.1844,
      "step": 24
    },
    {
      "epoch": 0.05770340450086555,
      "grad_norm": 0.05057906359434128,
      "learning_rate": 0.00019814814814814814,
      "loss": 0.1796,
      "step": 25
    },
    {
      "epoch": 0.06001154068090017,
      "grad_norm": 0.04395706206560135,
      "learning_rate": 0.00019805555555555557,
      "loss": 0.1766,
      "step": 26
    },
    {
      "epoch": 0.06231967686093479,
      "grad_norm": 0.04838942736387253,
      "learning_rate": 0.00019796296296296296,
      "loss": 0.1789,
      "step": 27
    },
    {
      "epoch": 0.06462781304096941,
      "grad_norm": 0.04146023094654083,
      "learning_rate": 0.00019787037037037036,
      "loss": 0.1845,
      "step": 28
    },
    {
      "epoch": 0.06693594922100404,
      "grad_norm": 0.06153571605682373,
      "learning_rate": 0.00019777777777777778,
      "loss": 0.1784,
      "step": 29
    },
    {
      "epoch": 0.06924408540103866,
      "grad_norm": 0.044718630611896515,
      "learning_rate": 0.0001976851851851852,
      "loss": 0.1831,
      "step": 30
    },
    {
      "epoch": 0.07155222158107329,
      "grad_norm": 0.04743555933237076,
      "learning_rate": 0.0001975925925925926,
      "loss": 0.181,
      "step": 31
    },
    {
      "epoch": 0.0738603577611079,
      "grad_norm": 0.058189600706100464,
      "learning_rate": 0.00019750000000000003,
      "loss": 0.1827,
      "step": 32
    },
    {
      "epoch": 0.07616849394114253,
      "grad_norm": 0.056668609380722046,
      "learning_rate": 0.00019740740740740743,
      "loss": 0.1761,
      "step": 33
    },
    {
      "epoch": 0.07847663012117714,
      "grad_norm": 0.047286469489336014,
      "learning_rate": 0.00019731481481481482,
      "loss": 0.1756,
      "step": 34
    },
    {
      "epoch": 0.08078476630121177,
      "grad_norm": 0.045554183423519135,
      "learning_rate": 0.00019722222222222225,
      "loss": 0.1704,
      "step": 35
    },
    {
      "epoch": 0.0830929024812464,
      "grad_norm": 0.07044310122728348,
      "learning_rate": 0.00019712962962962965,
      "loss": 0.1747,
      "step": 36
    },
    {
      "epoch": 0.08540103866128101,
      "grad_norm": 0.14839236438274384,
      "learning_rate": 0.00019703703703703704,
      "loss": 0.1835,
      "step": 37
    },
    {
      "epoch": 0.08770917484131564,
      "grad_norm": 0.07167937606573105,
      "learning_rate": 0.00019694444444444447,
      "loss": 0.1763,
      "step": 38
    },
    {
      "epoch": 0.09001731102135026,
      "grad_norm": 0.053169433027505875,
      "learning_rate": 0.00019685185185185186,
      "loss": 0.172,
      "step": 39
    },
    {
      "epoch": 0.09232544720138489,
      "grad_norm": 0.05013667047023773,
      "learning_rate": 0.00019675925925925926,
      "loss": 0.1733,
      "step": 40
    },
    {
      "epoch": 0.0946335833814195,
      "grad_norm": 0.04782355576753616,
      "learning_rate": 0.00019666666666666666,
      "loss": 0.1714,
      "step": 41
    },
    {
      "epoch": 0.09694171956145413,
      "grad_norm": 0.041702546179294586,
      "learning_rate": 0.00019657407407407408,
      "loss": 0.1672,
      "step": 42
    },
    {
      "epoch": 0.09924985574148874,
      "grad_norm": 0.043393392115831375,
      "learning_rate": 0.00019648148148148148,
      "loss": 0.1724,
      "step": 43
    },
    {
      "epoch": 0.10155799192152337,
      "grad_norm": 0.04598653316497803,
      "learning_rate": 0.00019638888888888888,
      "loss": 0.1639,
      "step": 44
    },
    {
      "epoch": 0.10386612810155799,
      "grad_norm": 0.041263360530138016,
      "learning_rate": 0.0001962962962962963,
      "loss": 0.1652,
      "step": 45
    },
    {
      "epoch": 0.10617426428159261,
      "grad_norm": 0.05192224681377411,
      "learning_rate": 0.0001962037037037037,
      "loss": 0.1763,
      "step": 46
    },
    {
      "epoch": 0.10848240046162723,
      "grad_norm": 0.07206670194864273,
      "learning_rate": 0.00019611111111111112,
      "loss": 0.1768,
      "step": 47
    },
    {
      "epoch": 0.11079053664166186,
      "grad_norm": 0.047010116279125214,
      "learning_rate": 0.00019601851851851855,
      "loss": 0.1625,
      "step": 48
    },
    {
      "epoch": 0.11309867282169649,
      "grad_norm": 0.0623844712972641,
      "learning_rate": 0.00019592592592592594,
      "loss": 0.1698,
      "step": 49
    },
    {
      "epoch": 0.1154068090017311,
      "grad_norm": 0.06913040578365326,
      "learning_rate": 0.00019583333333333334,
      "loss": 0.1694,
      "step": 50
    },
    {
      "epoch": 0.11771494518176573,
      "grad_norm": 0.08099398761987686,
      "learning_rate": 0.00019574074074074077,
      "loss": 0.1632,
      "step": 51
    },
    {
      "epoch": 0.12002308136180034,
      "grad_norm": 0.12050978094339371,
      "learning_rate": 0.00019564814814814816,
      "loss": 0.1677,
      "step": 52
    },
    {
      "epoch": 0.12233121754183497,
      "grad_norm": 0.08151208609342575,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.1654,
      "step": 53
    },
    {
      "epoch": 0.12463935372186959,
      "grad_norm": 0.08948655426502228,
      "learning_rate": 0.00019546296296296298,
      "loss": 0.1673,
      "step": 54
    },
    {
      "epoch": 0.12694748990190421,
      "grad_norm": 0.07907553017139435,
      "learning_rate": 0.00019537037037037038,
      "loss": 0.1697,
      "step": 55
    },
    {
      "epoch": 0.12925562608193883,
      "grad_norm": 0.06165081635117531,
      "learning_rate": 0.00019527777777777778,
      "loss": 0.1628,
      "step": 56
    },
    {
      "epoch": 0.13156376226197344,
      "grad_norm": 0.05658894032239914,
      "learning_rate": 0.0001951851851851852,
      "loss": 0.1719,
      "step": 57
    },
    {
      "epoch": 0.13387189844200809,
      "grad_norm": 0.06691501289606094,
      "learning_rate": 0.0001950925925925926,
      "loss": 0.1705,
      "step": 58
    },
    {
      "epoch": 0.1361800346220427,
      "grad_norm": 0.060781329870224,
      "learning_rate": 0.000195,
      "loss": 0.1618,
      "step": 59
    },
    {
      "epoch": 0.13848817080207732,
      "grad_norm": 0.054870542138814926,
      "learning_rate": 0.00019490740740740742,
      "loss": 0.1718,
      "step": 60
    },
    {
      "epoch": 0.14079630698211196,
      "grad_norm": 0.056277643889188766,
      "learning_rate": 0.00019481481481481482,
      "loss": 0.1611,
      "step": 61
    },
    {
      "epoch": 0.14310444316214657,
      "grad_norm": 0.042292430996894836,
      "learning_rate": 0.00019472222222222221,
      "loss": 0.1575,
      "step": 62
    },
    {
      "epoch": 0.1454125793421812,
      "grad_norm": 0.0642034038901329,
      "learning_rate": 0.00019462962962962964,
      "loss": 0.1642,
      "step": 63
    },
    {
      "epoch": 0.1477207155222158,
      "grad_norm": 0.05350774526596069,
      "learning_rate": 0.00019453703703703706,
      "loss": 0.1688,
      "step": 64
    },
    {
      "epoch": 0.15002885170225044,
      "grad_norm": 0.060144905000925064,
      "learning_rate": 0.00019444444444444446,
      "loss": 0.1649,
      "step": 65
    },
    {
      "epoch": 0.15233698788228506,
      "grad_norm": 0.05027244985103607,
      "learning_rate": 0.00019435185185185186,
      "loss": 0.1641,
      "step": 66
    },
    {
      "epoch": 0.15464512406231967,
      "grad_norm": 0.060575924813747406,
      "learning_rate": 0.00019425925925925928,
      "loss": 0.1663,
      "step": 67
    },
    {
      "epoch": 0.1569532602423543,
      "grad_norm": 0.05797772482037544,
      "learning_rate": 0.00019416666666666668,
      "loss": 0.162,
      "step": 68
    },
    {
      "epoch": 0.15926139642238893,
      "grad_norm": 0.07599963992834091,
      "learning_rate": 0.00019407407407407408,
      "loss": 0.1621,
      "step": 69
    },
    {
      "epoch": 0.16156953260242354,
      "grad_norm": 0.07238578796386719,
      "learning_rate": 0.0001939814814814815,
      "loss": 0.1607,
      "step": 70
    },
    {
      "epoch": 0.16387766878245816,
      "grad_norm": 0.09605788439512253,
      "learning_rate": 0.0001938888888888889,
      "loss": 0.1698,
      "step": 71
    },
    {
      "epoch": 0.1661858049624928,
      "grad_norm": 0.1585727035999298,
      "learning_rate": 0.0001937962962962963,
      "loss": 0.1596,
      "step": 72
    },
    {
      "epoch": 0.16849394114252741,
      "grad_norm": 0.1919248402118683,
      "learning_rate": 0.00019370370370370372,
      "loss": 0.1668,
      "step": 73
    },
    {
      "epoch": 0.17080207732256203,
      "grad_norm": 0.09726729989051819,
      "learning_rate": 0.00019361111111111112,
      "loss": 0.1601,
      "step": 74
    },
    {
      "epoch": 0.17311021350259664,
      "grad_norm": 0.06920541822910309,
      "learning_rate": 0.0001935185185185185,
      "loss": 0.1571,
      "step": 75
    },
    {
      "epoch": 0.17541834968263129,
      "grad_norm": 0.07686076313257217,
      "learning_rate": 0.00019342592592592594,
      "loss": 0.1595,
      "step": 76
    },
    {
      "epoch": 0.1777264858626659,
      "grad_norm": 0.05949549376964569,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.1577,
      "step": 77
    },
    {
      "epoch": 0.18003462204270052,
      "grad_norm": 0.060674380511045456,
      "learning_rate": 0.00019324074074074073,
      "loss": 0.1661,
      "step": 78
    },
    {
      "epoch": 0.18234275822273513,
      "grad_norm": 0.05908726900815964,
      "learning_rate": 0.00019314814814814816,
      "loss": 0.1615,
      "step": 79
    },
    {
      "epoch": 0.18465089440276977,
      "grad_norm": 0.059043094515800476,
      "learning_rate": 0.00019305555555555558,
      "loss": 0.1585,
      "step": 80
    },
    {
      "epoch": 0.1869590305828044,
      "grad_norm": 0.054791729897260666,
      "learning_rate": 0.00019296296296296298,
      "loss": 0.1554,
      "step": 81
    },
    {
      "epoch": 0.189267166762839,
      "grad_norm": 0.06371712684631348,
      "learning_rate": 0.0001928703703703704,
      "loss": 0.1574,
      "step": 82
    },
    {
      "epoch": 0.19157530294287364,
      "grad_norm": 0.05277816578745842,
      "learning_rate": 0.0001927777777777778,
      "loss": 0.1574,
      "step": 83
    },
    {
      "epoch": 0.19388343912290826,
      "grad_norm": 0.059945784509181976,
      "learning_rate": 0.0001926851851851852,
      "loss": 0.1599,
      "step": 84
    },
    {
      "epoch": 0.19619157530294287,
      "grad_norm": 0.05013200640678406,
      "learning_rate": 0.0001925925925925926,
      "loss": 0.16,
      "step": 85
    },
    {
      "epoch": 0.1984997114829775,
      "grad_norm": 0.05411451309919357,
      "learning_rate": 0.00019250000000000002,
      "loss": 0.1552,
      "step": 86
    },
    {
      "epoch": 0.20080784766301213,
      "grad_norm": 0.056629788130521774,
      "learning_rate": 0.00019240740740740741,
      "loss": 0.1621,
      "step": 87
    },
    {
      "epoch": 0.20311598384304674,
      "grad_norm": 0.05227493494749069,
      "learning_rate": 0.0001923148148148148,
      "loss": 0.1491,
      "step": 88
    },
    {
      "epoch": 0.20542412002308136,
      "grad_norm": 0.061280835419893265,
      "learning_rate": 0.00019222222222222224,
      "loss": 0.1558,
      "step": 89
    },
    {
      "epoch": 0.20773225620311597,
      "grad_norm": 0.04570850357413292,
      "learning_rate": 0.00019212962962962963,
      "loss": 0.1502,
      "step": 90
    },
    {
      "epoch": 0.21004039238315061,
      "grad_norm": 0.05651992931962013,
      "learning_rate": 0.00019203703703703703,
      "loss": 0.1611,
      "step": 91
    },
    {
      "epoch": 0.21234852856318523,
      "grad_norm": 0.05105481296777725,
      "learning_rate": 0.00019194444444444445,
      "loss": 0.1571,
      "step": 92
    },
    {
      "epoch": 0.21465666474321984,
      "grad_norm": 0.05344156175851822,
      "learning_rate": 0.00019185185185185185,
      "loss": 0.155,
      "step": 93
    },
    {
      "epoch": 0.21696480092325446,
      "grad_norm": 0.05218995362520218,
      "learning_rate": 0.00019175925925925925,
      "loss": 0.1546,
      "step": 94
    },
    {
      "epoch": 0.2192729371032891,
      "grad_norm": 0.0546402782201767,
      "learning_rate": 0.00019166666666666667,
      "loss": 0.1585,
      "step": 95
    },
    {
      "epoch": 0.22158107328332372,
      "grad_norm": 0.04980732128024101,
      "learning_rate": 0.0001915740740740741,
      "loss": 0.1533,
      "step": 96
    },
    {
      "epoch": 0.22388920946335833,
      "grad_norm": 0.05537516996264458,
      "learning_rate": 0.0001914814814814815,
      "loss": 0.1587,
      "step": 97
    },
    {
      "epoch": 0.22619734564339297,
      "grad_norm": 0.05610758066177368,
      "learning_rate": 0.00019138888888888892,
      "loss": 0.1512,
      "step": 98
    },
    {
      "epoch": 0.2285054818234276,
      "grad_norm": 0.06219853460788727,
      "learning_rate": 0.00019129629629629632,
      "loss": 0.1568,
      "step": 99
    },
    {
      "epoch": 0.2308136180034622,
      "grad_norm": 0.06942673027515411,
      "learning_rate": 0.0001912037037037037,
      "loss": 0.1515,
      "step": 100
    },
    {
      "epoch": 0.23312175418349682,
      "grad_norm": 0.0529804527759552,
      "learning_rate": 0.00019111111111111114,
      "loss": 0.1513,
      "step": 101
    },
    {
      "epoch": 0.23542989036353146,
      "grad_norm": 0.057867150753736496,
      "learning_rate": 0.00019101851851851853,
      "loss": 0.1506,
      "step": 102
    },
    {
      "epoch": 0.23773802654356607,
      "grad_norm": 0.056259866803884506,
      "learning_rate": 0.00019092592592592593,
      "loss": 0.1585,
      "step": 103
    },
    {
      "epoch": 0.2400461627236007,
      "grad_norm": 0.05715411156415939,
      "learning_rate": 0.00019083333333333336,
      "loss": 0.1571,
      "step": 104
    },
    {
      "epoch": 0.2423542989036353,
      "grad_norm": 0.06372460722923279,
      "learning_rate": 0.00019074074074074075,
      "loss": 0.1548,
      "step": 105
    },
    {
      "epoch": 0.24466243508366994,
      "grad_norm": 0.06821900606155396,
      "learning_rate": 0.00019064814814814815,
      "loss": 0.1544,
      "step": 106
    },
    {
      "epoch": 0.24697057126370456,
      "grad_norm": 0.06562413275241852,
      "learning_rate": 0.00019055555555555555,
      "loss": 0.1583,
      "step": 107
    },
    {
      "epoch": 0.24927870744373917,
      "grad_norm": 0.08347988873720169,
      "learning_rate": 0.00019046296296296297,
      "loss": 0.1514,
      "step": 108
    },
    {
      "epoch": 0.2515868436237738,
      "grad_norm": 0.05993643030524254,
      "learning_rate": 0.00019037037037037037,
      "loss": 0.1508,
      "step": 109
    },
    {
      "epoch": 0.25389497980380843,
      "grad_norm": 0.07057323306798935,
      "learning_rate": 0.00019027777777777776,
      "loss": 0.1525,
      "step": 110
    },
    {
      "epoch": 0.25620311598384304,
      "grad_norm": 0.09892449527978897,
      "learning_rate": 0.0001901851851851852,
      "loss": 0.1551,
      "step": 111
    },
    {
      "epoch": 0.25851125216387766,
      "grad_norm": 0.12110217660665512,
      "learning_rate": 0.0001900925925925926,
      "loss": 0.1528,
      "step": 112
    },
    {
      "epoch": 0.2608193883439123,
      "grad_norm": 0.07105842232704163,
      "learning_rate": 0.00019,
      "loss": 0.1584,
      "step": 113
    },
    {
      "epoch": 0.2631275245239469,
      "grad_norm": 0.06798320263624191,
      "learning_rate": 0.00018990740740740743,
      "loss": 0.1516,
      "step": 114
    },
    {
      "epoch": 0.26543566070398156,
      "grad_norm": 0.07161455601453781,
      "learning_rate": 0.00018981481481481483,
      "loss": 0.154,
      "step": 115
    },
    {
      "epoch": 0.26774379688401617,
      "grad_norm": 0.05709280073642731,
      "learning_rate": 0.00018972222222222223,
      "loss": 0.1554,
      "step": 116
    },
    {
      "epoch": 0.2700519330640508,
      "grad_norm": 0.060083746910095215,
      "learning_rate": 0.00018962962962962965,
      "loss": 0.1543,
      "step": 117
    },
    {
      "epoch": 0.2723600692440854,
      "grad_norm": 0.06796609610319138,
      "learning_rate": 0.00018953703703703705,
      "loss": 0.1493,
      "step": 118
    },
    {
      "epoch": 0.27466820542412,
      "grad_norm": 0.06265313923358917,
      "learning_rate": 0.00018944444444444445,
      "loss": 0.1555,
      "step": 119
    },
    {
      "epoch": 0.27697634160415463,
      "grad_norm": 0.060422059148550034,
      "learning_rate": 0.00018935185185185187,
      "loss": 0.1534,
      "step": 120
    },
    {
      "epoch": 0.27928447778418924,
      "grad_norm": 0.06933347135782242,
      "learning_rate": 0.00018925925925925927,
      "loss": 0.154,
      "step": 121
    },
    {
      "epoch": 0.2815926139642239,
      "grad_norm": 0.0509236641228199,
      "learning_rate": 0.00018916666666666667,
      "loss": 0.1554,
      "step": 122
    },
    {
      "epoch": 0.28390075014425853,
      "grad_norm": 0.07009289413690567,
      "learning_rate": 0.0001890740740740741,
      "loss": 0.1493,
      "step": 123
    },
    {
      "epoch": 0.28620888632429314,
      "grad_norm": 0.06411243230104446,
      "learning_rate": 0.0001889814814814815,
      "loss": 0.1547,
      "step": 124
    },
    {
      "epoch": 0.28851702250432776,
      "grad_norm": 0.047630466520786285,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.1528,
      "step": 125
    },
    {
      "epoch": 0.2908251586843624,
      "grad_norm": 0.061059992760419846,
      "learning_rate": 0.0001887962962962963,
      "loss": 0.1428,
      "step": 126
    },
    {
      "epoch": 0.293133294864397,
      "grad_norm": 0.058564092963933945,
      "learning_rate": 0.0001887037037037037,
      "loss": 0.149,
      "step": 127
    },
    {
      "epoch": 0.2954414310444316,
      "grad_norm": 0.06902942806482315,
      "learning_rate": 0.0001886111111111111,
      "loss": 0.1475,
      "step": 128
    },
    {
      "epoch": 0.2977495672244662,
      "grad_norm": 0.061034902930259705,
      "learning_rate": 0.00018851851851851853,
      "loss": 0.1505,
      "step": 129
    },
    {
      "epoch": 0.3000577034045009,
      "grad_norm": 0.0622573159635067,
      "learning_rate": 0.00018842592592592595,
      "loss": 0.1484,
      "step": 130
    },
    {
      "epoch": 0.3023658395845355,
      "grad_norm": 0.06985937803983688,
      "learning_rate": 0.00018833333333333335,
      "loss": 0.1521,
      "step": 131
    },
    {
      "epoch": 0.3046739757645701,
      "grad_norm": 0.06424791365861893,
      "learning_rate": 0.00018824074074074075,
      "loss": 0.1491,
      "step": 132
    },
    {
      "epoch": 0.30698211194460473,
      "grad_norm": 0.07435788959264755,
      "learning_rate": 0.00018814814814814817,
      "loss": 0.1472,
      "step": 133
    },
    {
      "epoch": 0.30929024812463934,
      "grad_norm": 0.09509983658790588,
      "learning_rate": 0.00018805555555555557,
      "loss": 0.1472,
      "step": 134
    },
    {
      "epoch": 0.31159838430467396,
      "grad_norm": 0.14597779512405396,
      "learning_rate": 0.00018796296296296296,
      "loss": 0.1518,
      "step": 135
    },
    {
      "epoch": 0.3139065204847086,
      "grad_norm": 0.12349999696016312,
      "learning_rate": 0.0001878703703703704,
      "loss": 0.1443,
      "step": 136
    },
    {
      "epoch": 0.31621465666474324,
      "grad_norm": 0.08212310075759888,
      "learning_rate": 0.00018777777777777779,
      "loss": 0.1512,
      "step": 137
    },
    {
      "epoch": 0.31852279284477786,
      "grad_norm": 0.11503876745700836,
      "learning_rate": 0.00018768518518518518,
      "loss": 0.1475,
      "step": 138
    },
    {
      "epoch": 0.3208309290248125,
      "grad_norm": 0.06587795913219452,
      "learning_rate": 0.0001875925925925926,
      "loss": 0.1437,
      "step": 139
    },
    {
      "epoch": 0.3231390652048471,
      "grad_norm": 0.07384002953767776,
      "learning_rate": 0.0001875,
      "loss": 0.1487,
      "step": 140
    },
    {
      "epoch": 0.3254472013848817,
      "grad_norm": 0.06668757647275925,
      "learning_rate": 0.0001874074074074074,
      "loss": 0.1489,
      "step": 141
    },
    {
      "epoch": 0.3277553375649163,
      "grad_norm": 0.06417941302061081,
      "learning_rate": 0.00018731481481481483,
      "loss": 0.146,
      "step": 142
    },
    {
      "epoch": 0.33006347374495093,
      "grad_norm": 0.06864254176616669,
      "learning_rate": 0.00018722222222222222,
      "loss": 0.1483,
      "step": 143
    },
    {
      "epoch": 0.3323716099249856,
      "grad_norm": 0.06642857939004898,
      "learning_rate": 0.00018712962962962962,
      "loss": 0.1486,
      "step": 144
    },
    {
      "epoch": 0.3346797461050202,
      "grad_norm": 0.06145348399877548,
      "learning_rate": 0.00018703703703703704,
      "loss": 0.1526,
      "step": 145
    },
    {
      "epoch": 0.33698788228505483,
      "grad_norm": 0.0777013823390007,
      "learning_rate": 0.00018694444444444447,
      "loss": 0.1524,
      "step": 146
    },
    {
      "epoch": 0.33929601846508944,
      "grad_norm": 0.053706854581832886,
      "learning_rate": 0.00018685185185185187,
      "loss": 0.1489,
      "step": 147
    },
    {
      "epoch": 0.34160415464512406,
      "grad_norm": 0.07778315246105194,
      "learning_rate": 0.0001867592592592593,
      "loss": 0.1526,
      "step": 148
    },
    {
      "epoch": 0.3439122908251587,
      "grad_norm": 0.07362362742424011,
      "learning_rate": 0.0001866666666666667,
      "loss": 0.1454,
      "step": 149
    },
    {
      "epoch": 0.3462204270051933,
      "grad_norm": 0.06637648493051529,
      "learning_rate": 0.00018657407407407408,
      "loss": 0.1467,
      "step": 150
    },
    {
      "epoch": 0.3485285631852279,
      "grad_norm": 0.07013659179210663,
      "learning_rate": 0.0001864814814814815,
      "loss": 0.1531,
      "step": 151
    },
    {
      "epoch": 0.35083669936526257,
      "grad_norm": 0.06309764087200165,
      "learning_rate": 0.0001863888888888889,
      "loss": 0.1496,
      "step": 152
    },
    {
      "epoch": 0.3531448355452972,
      "grad_norm": 0.06464076787233353,
      "learning_rate": 0.0001862962962962963,
      "loss": 0.1528,
      "step": 153
    },
    {
      "epoch": 0.3554529717253318,
      "grad_norm": 0.0690513625741005,
      "learning_rate": 0.0001862037037037037,
      "loss": 0.1475,
      "step": 154
    },
    {
      "epoch": 0.3577611079053664,
      "grad_norm": 0.056556567549705505,
      "learning_rate": 0.00018611111111111112,
      "loss": 0.1487,
      "step": 155
    },
    {
      "epoch": 0.36006924408540103,
      "grad_norm": 0.06378641724586487,
      "learning_rate": 0.00018601851851851852,
      "loss": 0.1487,
      "step": 156
    },
    {
      "epoch": 0.36237738026543564,
      "grad_norm": 0.06881760060787201,
      "learning_rate": 0.00018592592592592592,
      "loss": 0.1447,
      "step": 157
    },
    {
      "epoch": 0.36468551644547026,
      "grad_norm": 0.07467546314001083,
      "learning_rate": 0.00018583333333333334,
      "loss": 0.1543,
      "step": 158
    },
    {
      "epoch": 0.36699365262550493,
      "grad_norm": 0.05935186892747879,
      "learning_rate": 0.00018574074074074074,
      "loss": 0.1492,
      "step": 159
    },
    {
      "epoch": 0.36930178880553954,
      "grad_norm": 0.06921457499265671,
      "learning_rate": 0.00018564814814814814,
      "loss": 0.1515,
      "step": 160
    },
    {
      "epoch": 0.37160992498557416,
      "grad_norm": 0.05208181217312813,
      "learning_rate": 0.00018555555555555556,
      "loss": 0.1536,
      "step": 161
    },
    {
      "epoch": 0.3739180611656088,
      "grad_norm": 0.07313266396522522,
      "learning_rate": 0.00018546296296296298,
      "loss": 0.1401,
      "step": 162
    },
    {
      "epoch": 0.3762261973456434,
      "grad_norm": 0.06458649039268494,
      "learning_rate": 0.00018537037037037038,
      "loss": 0.1493,
      "step": 163
    },
    {
      "epoch": 0.378534333525678,
      "grad_norm": 0.0680573433637619,
      "learning_rate": 0.0001852777777777778,
      "loss": 0.1428,
      "step": 164
    },
    {
      "epoch": 0.3808424697057126,
      "grad_norm": 0.06942640990018845,
      "learning_rate": 0.0001851851851851852,
      "loss": 0.1544,
      "step": 165
    },
    {
      "epoch": 0.3831506058857473,
      "grad_norm": 0.09018060564994812,
      "learning_rate": 0.0001850925925925926,
      "loss": 0.1482,
      "step": 166
    },
    {
      "epoch": 0.3854587420657819,
      "grad_norm": 0.09488742798566818,
      "learning_rate": 0.00018500000000000002,
      "loss": 0.1487,
      "step": 167
    },
    {
      "epoch": 0.3877668782458165,
      "grad_norm": 0.05812246352434158,
      "learning_rate": 0.00018490740740740742,
      "loss": 0.1417,
      "step": 168
    },
    {
      "epoch": 0.39007501442585113,
      "grad_norm": 0.10994130373001099,
      "learning_rate": 0.00018481481481481482,
      "loss": 0.1498,
      "step": 169
    },
    {
      "epoch": 0.39238315060588574,
      "grad_norm": 0.09131181985139847,
      "learning_rate": 0.00018472222222222224,
      "loss": 0.1478,
      "step": 170
    },
    {
      "epoch": 0.39469128678592036,
      "grad_norm": 0.08217167854309082,
      "learning_rate": 0.00018462962962962964,
      "loss": 0.1475,
      "step": 171
    },
    {
      "epoch": 0.396999422965955,
      "grad_norm": 0.07954363524913788,
      "learning_rate": 0.00018453703703703704,
      "loss": 0.1496,
      "step": 172
    },
    {
      "epoch": 0.3993075591459896,
      "grad_norm": 0.06972556561231613,
      "learning_rate": 0.00018444444444444446,
      "loss": 0.1469,
      "step": 173
    },
    {
      "epoch": 0.40161569532602426,
      "grad_norm": 0.06256857514381409,
      "learning_rate": 0.00018435185185185186,
      "loss": 0.1423,
      "step": 174
    },
    {
      "epoch": 0.4039238315060589,
      "grad_norm": 0.06817682832479477,
      "learning_rate": 0.00018425925925925926,
      "loss": 0.1466,
      "step": 175
    },
    {
      "epoch": 0.4062319676860935,
      "grad_norm": 0.06916061788797379,
      "learning_rate": 0.00018416666666666665,
      "loss": 0.1501,
      "step": 176
    },
    {
      "epoch": 0.4085401038661281,
      "grad_norm": 0.06345528364181519,
      "learning_rate": 0.00018407407407407408,
      "loss": 0.1486,
      "step": 177
    },
    {
      "epoch": 0.4108482400461627,
      "grad_norm": 0.06836175918579102,
      "learning_rate": 0.0001839814814814815,
      "loss": 0.1429,
      "step": 178
    },
    {
      "epoch": 0.41315637622619733,
      "grad_norm": 0.06683284044265747,
      "learning_rate": 0.0001838888888888889,
      "loss": 0.1457,
      "step": 179
    },
    {
      "epoch": 0.41546451240623195,
      "grad_norm": 0.06601724028587341,
      "learning_rate": 0.00018379629629629632,
      "loss": 0.1486,
      "step": 180
    },
    {
      "epoch": 0.4177726485862666,
      "grad_norm": 0.06728939712047577,
      "learning_rate": 0.00018370370370370372,
      "loss": 0.148,
      "step": 181
    },
    {
      "epoch": 0.42008078476630123,
      "grad_norm": 0.06636571139097214,
      "learning_rate": 0.00018361111111111112,
      "loss": 0.1437,
      "step": 182
    },
    {
      "epoch": 0.42238892094633584,
      "grad_norm": 0.0734277218580246,
      "learning_rate": 0.00018351851851851854,
      "loss": 0.1432,
      "step": 183
    },
    {
      "epoch": 0.42469705712637046,
      "grad_norm": 0.05555666238069534,
      "learning_rate": 0.00018342592592592594,
      "loss": 0.1507,
      "step": 184
    },
    {
      "epoch": 0.4270051933064051,
      "grad_norm": 0.05956634506583214,
      "learning_rate": 0.00018333333333333334,
      "loss": 0.1429,
      "step": 185
    },
    {
      "epoch": 0.4293133294864397,
      "grad_norm": 0.0864204466342926,
      "learning_rate": 0.00018324074074074076,
      "loss": 0.149,
      "step": 186
    },
    {
      "epoch": 0.4316214656664743,
      "grad_norm": 0.07935954630374908,
      "learning_rate": 0.00018314814814814816,
      "loss": 0.1373,
      "step": 187
    },
    {
      "epoch": 0.4339296018465089,
      "grad_norm": 0.061793871223926544,
      "learning_rate": 0.00018305555555555555,
      "loss": 0.1449,
      "step": 188
    },
    {
      "epoch": 0.4362377380265436,
      "grad_norm": 0.06415320932865143,
      "learning_rate": 0.00018296296296296298,
      "loss": 0.1482,
      "step": 189
    },
    {
      "epoch": 0.4385458742065782,
      "grad_norm": 0.06927644461393356,
      "learning_rate": 0.00018287037037037038,
      "loss": 0.1432,
      "step": 190
    },
    {
      "epoch": 0.4408540103866128,
      "grad_norm": 0.0611124113202095,
      "learning_rate": 0.00018277777777777777,
      "loss": 0.1438,
      "step": 191
    },
    {
      "epoch": 0.44316214656664743,
      "grad_norm": 0.05857139825820923,
      "learning_rate": 0.0001826851851851852,
      "loss": 0.1439,
      "step": 192
    },
    {
      "epoch": 0.44547028274668204,
      "grad_norm": 0.053906142711639404,
      "learning_rate": 0.0001825925925925926,
      "loss": 0.1405,
      "step": 193
    },
    {
      "epoch": 0.44777841892671666,
      "grad_norm": 0.05838324874639511,
      "learning_rate": 0.0001825,
      "loss": 0.1419,
      "step": 194
    },
    {
      "epoch": 0.4500865551067513,
      "grad_norm": 0.05292156711220741,
      "learning_rate": 0.00018240740740740742,
      "loss": 0.1423,
      "step": 195
    },
    {
      "epoch": 0.45239469128678594,
      "grad_norm": 0.05878763273358345,
      "learning_rate": 0.00018231481481481484,
      "loss": 0.1421,
      "step": 196
    },
    {
      "epoch": 0.45470282746682056,
      "grad_norm": 0.0504269078373909,
      "learning_rate": 0.00018222222222222224,
      "loss": 0.1437,
      "step": 197
    },
    {
      "epoch": 0.4570109636468552,
      "grad_norm": 0.054045192897319794,
      "learning_rate": 0.00018212962962962963,
      "loss": 0.142,
      "step": 198
    },
    {
      "epoch": 0.4593190998268898,
      "grad_norm": 0.06022946164011955,
      "learning_rate": 0.00018203703703703706,
      "loss": 0.1468,
      "step": 199
    },
    {
      "epoch": 0.4616272360069244,
      "grad_norm": 0.07655218988656998,
      "learning_rate": 0.00018194444444444445,
      "loss": 0.1446,
      "step": 200
    },
    {
      "epoch": 0.463935372186959,
      "grad_norm": 0.0775132030248642,
      "learning_rate": 0.00018185185185185185,
      "loss": 0.1425,
      "step": 201
    },
    {
      "epoch": 0.46624350836699363,
      "grad_norm": 0.05718778818845749,
      "learning_rate": 0.00018175925925925928,
      "loss": 0.1467,
      "step": 202
    },
    {
      "epoch": 0.4685516445470283,
      "grad_norm": 0.0744498074054718,
      "learning_rate": 0.00018166666666666667,
      "loss": 0.1424,
      "step": 203
    },
    {
      "epoch": 0.4708597807270629,
      "grad_norm": 0.08696521818637848,
      "learning_rate": 0.00018157407407407407,
      "loss": 0.1447,
      "step": 204
    },
    {
      "epoch": 0.47316791690709753,
      "grad_norm": 0.09435555338859558,
      "learning_rate": 0.0001814814814814815,
      "loss": 0.1438,
      "step": 205
    },
    {
      "epoch": 0.47547605308713214,
      "grad_norm": 0.24541200697422028,
      "learning_rate": 0.0001813888888888889,
      "loss": 0.1507,
      "step": 206
    },
    {
      "epoch": 0.47778418926716676,
      "grad_norm": 0.1925467550754547,
      "learning_rate": 0.0001812962962962963,
      "loss": 0.145,
      "step": 207
    },
    {
      "epoch": 0.4800923254472014,
      "grad_norm": 0.1271919459104538,
      "learning_rate": 0.0001812037037037037,
      "loss": 0.1487,
      "step": 208
    },
    {
      "epoch": 0.482400461627236,
      "grad_norm": 0.07203220576047897,
      "learning_rate": 0.0001811111111111111,
      "loss": 0.1506,
      "step": 209
    },
    {
      "epoch": 0.4847085978072706,
      "grad_norm": 0.09411832690238953,
      "learning_rate": 0.0001810185185185185,
      "loss": 0.1459,
      "step": 210
    },
    {
      "epoch": 0.4870167339873053,
      "grad_norm": 0.06762874871492386,
      "learning_rate": 0.00018092592592592593,
      "loss": 0.1475,
      "step": 211
    },
    {
      "epoch": 0.4893248701673399,
      "grad_norm": 0.06898859888315201,
      "learning_rate": 0.00018083333333333336,
      "loss": 0.142,
      "step": 212
    },
    {
      "epoch": 0.4916330063473745,
      "grad_norm": 0.0666506364941597,
      "learning_rate": 0.00018074074074074075,
      "loss": 0.1408,
      "step": 213
    },
    {
      "epoch": 0.4939411425274091,
      "grad_norm": 0.06564829498529434,
      "learning_rate": 0.00018064814814814818,
      "loss": 0.1428,
      "step": 214
    },
    {
      "epoch": 0.49624927870744373,
      "grad_norm": 0.05717082694172859,
      "learning_rate": 0.00018055555555555557,
      "loss": 0.1441,
      "step": 215
    },
    {
      "epoch": 0.49855741488747835,
      "grad_norm": 0.07268943637609482,
      "learning_rate": 0.00018046296296296297,
      "loss": 0.1443,
      "step": 216
    },
    {
      "epoch": 0.500865551067513,
      "grad_norm": 0.05505065992474556,
      "learning_rate": 0.0001803703703703704,
      "loss": 0.1445,
      "step": 217
    },
    {
      "epoch": 0.5031736872475476,
      "grad_norm": 0.07598809152841568,
      "learning_rate": 0.0001802777777777778,
      "loss": 0.1462,
      "step": 218
    },
    {
      "epoch": 0.5054818234275822,
      "grad_norm": 0.052191078662872314,
      "learning_rate": 0.0001801851851851852,
      "loss": 0.1458,
      "step": 219
    },
    {
      "epoch": 0.5077899596076169,
      "grad_norm": 0.06750345230102539,
      "learning_rate": 0.0001800925925925926,
      "loss": 0.145,
      "step": 220
    },
    {
      "epoch": 0.5100980957876514,
      "grad_norm": 0.04916289821267128,
      "learning_rate": 0.00018,
      "loss": 0.143,
      "step": 221
    },
    {
      "epoch": 0.5124062319676861,
      "grad_norm": 0.06001339107751846,
      "learning_rate": 0.0001799074074074074,
      "loss": 0.1472,
      "step": 222
    },
    {
      "epoch": 0.5147143681477208,
      "grad_norm": 0.06306777894496918,
      "learning_rate": 0.0001798148148148148,
      "loss": 0.1478,
      "step": 223
    },
    {
      "epoch": 0.5170225043277553,
      "grad_norm": 0.05748048052191734,
      "learning_rate": 0.00017972222222222223,
      "loss": 0.1412,
      "step": 224
    },
    {
      "epoch": 0.51933064050779,
      "grad_norm": 0.06559883803129196,
      "learning_rate": 0.00017962962962962963,
      "loss": 0.1453,
      "step": 225
    },
    {
      "epoch": 0.5216387766878245,
      "grad_norm": 0.07112711668014526,
      "learning_rate": 0.00017953703703703702,
      "loss": 0.1406,
      "step": 226
    },
    {
      "epoch": 0.5239469128678592,
      "grad_norm": 0.0643484964966774,
      "learning_rate": 0.00017944444444444445,
      "loss": 0.1488,
      "step": 227
    },
    {
      "epoch": 0.5262550490478938,
      "grad_norm": 0.059827033430337906,
      "learning_rate": 0.00017935185185185187,
      "loss": 0.1475,
      "step": 228
    },
    {
      "epoch": 0.5285631852279284,
      "grad_norm": 0.06567304581403732,
      "learning_rate": 0.00017925925925925927,
      "loss": 0.1475,
      "step": 229
    },
    {
      "epoch": 0.5308713214079631,
      "grad_norm": 0.081722192466259,
      "learning_rate": 0.0001791666666666667,
      "loss": 0.1429,
      "step": 230
    },
    {
      "epoch": 0.5331794575879977,
      "grad_norm": 0.10949339717626572,
      "learning_rate": 0.0001790740740740741,
      "loss": 0.1343,
      "step": 231
    },
    {
      "epoch": 0.5354875937680323,
      "grad_norm": 0.09612349420785904,
      "learning_rate": 0.0001789814814814815,
      "loss": 0.141,
      "step": 232
    },
    {
      "epoch": 0.5377957299480669,
      "grad_norm": 0.06475210934877396,
      "learning_rate": 0.0001788888888888889,
      "loss": 0.1389,
      "step": 233
    },
    {
      "epoch": 0.5401038661281016,
      "grad_norm": 0.10206308960914612,
      "learning_rate": 0.0001787962962962963,
      "loss": 0.148,
      "step": 234
    },
    {
      "epoch": 0.5424120023081361,
      "grad_norm": 0.06186994910240173,
      "learning_rate": 0.0001787037037037037,
      "loss": 0.1461,
      "step": 235
    },
    {
      "epoch": 0.5447201384881708,
      "grad_norm": 0.06017153337597847,
      "learning_rate": 0.00017861111111111113,
      "loss": 0.1448,
      "step": 236
    },
    {
      "epoch": 0.5470282746682055,
      "grad_norm": 0.05920975282788277,
      "learning_rate": 0.00017851851851851853,
      "loss": 0.1443,
      "step": 237
    },
    {
      "epoch": 0.54933641084824,
      "grad_norm": 0.05842497572302818,
      "learning_rate": 0.00017842592592592593,
      "loss": 0.1414,
      "step": 238
    },
    {
      "epoch": 0.5516445470282747,
      "grad_norm": 0.05242815986275673,
      "learning_rate": 0.00017833333333333335,
      "loss": 0.1352,
      "step": 239
    },
    {
      "epoch": 0.5539526832083093,
      "grad_norm": 0.06514687091112137,
      "learning_rate": 0.00017824074074074075,
      "loss": 0.1427,
      "step": 240
    },
    {
      "epoch": 0.5562608193883439,
      "grad_norm": 0.06072159484028816,
      "learning_rate": 0.00017814814814814814,
      "loss": 0.1421,
      "step": 241
    },
    {
      "epoch": 0.5585689555683785,
      "grad_norm": 0.0503254234790802,
      "learning_rate": 0.00017805555555555554,
      "loss": 0.1357,
      "step": 242
    },
    {
      "epoch": 0.5608770917484132,
      "grad_norm": 0.06688246876001358,
      "learning_rate": 0.00017796296296296296,
      "loss": 0.1448,
      "step": 243
    },
    {
      "epoch": 0.5631852279284478,
      "grad_norm": 0.058162838220596313,
      "learning_rate": 0.0001778703703703704,
      "loss": 0.1447,
      "step": 244
    },
    {
      "epoch": 0.5654933641084824,
      "grad_norm": 0.059606194496154785,
      "learning_rate": 0.00017777777777777779,
      "loss": 0.1451,
      "step": 245
    },
    {
      "epoch": 0.5678015002885171,
      "grad_norm": 0.06822901964187622,
      "learning_rate": 0.0001776851851851852,
      "loss": 0.144,
      "step": 246
    },
    {
      "epoch": 0.5701096364685516,
      "grad_norm": 0.05813856050372124,
      "learning_rate": 0.0001775925925925926,
      "loss": 0.1457,
      "step": 247
    },
    {
      "epoch": 0.5724177726485863,
      "grad_norm": 0.0673782005906105,
      "learning_rate": 0.0001775,
      "loss": 0.1362,
      "step": 248
    },
    {
      "epoch": 0.5747259088286208,
      "grad_norm": 0.06300479918718338,
      "learning_rate": 0.00017740740740740743,
      "loss": 0.1418,
      "step": 249
    },
    {
      "epoch": 0.5770340450086555,
      "grad_norm": 0.05119998753070831,
      "learning_rate": 0.00017731481481481483,
      "loss": 0.1426,
      "step": 250
    },
    {
      "epoch": 0.5793421811886902,
      "grad_norm": 0.06682850420475006,
      "learning_rate": 0.00017722222222222222,
      "loss": 0.1388,
      "step": 251
    },
    {
      "epoch": 0.5816503173687247,
      "grad_norm": 0.055639732629060745,
      "learning_rate": 0.00017712962962962965,
      "loss": 0.1399,
      "step": 252
    },
    {
      "epoch": 0.5839584535487594,
      "grad_norm": 0.05591867119073868,
      "learning_rate": 0.00017703703703703704,
      "loss": 0.1416,
      "step": 253
    },
    {
      "epoch": 0.586266589728794,
      "grad_norm": 0.06745456904172897,
      "learning_rate": 0.00017694444444444444,
      "loss": 0.141,
      "step": 254
    },
    {
      "epoch": 0.5885747259088286,
      "grad_norm": 0.05160287395119667,
      "learning_rate": 0.00017685185185185187,
      "loss": 0.1452,
      "step": 255
    },
    {
      "epoch": 0.5908828620888632,
      "grad_norm": 0.06144549697637558,
      "learning_rate": 0.00017675925925925926,
      "loss": 0.1385,
      "step": 256
    },
    {
      "epoch": 0.5931909982688979,
      "grad_norm": 0.05618147552013397,
      "learning_rate": 0.00017666666666666666,
      "loss": 0.1427,
      "step": 257
    },
    {
      "epoch": 0.5954991344489324,
      "grad_norm": 0.05598113685846329,
      "learning_rate": 0.00017657407407407408,
      "loss": 0.1333,
      "step": 258
    },
    {
      "epoch": 0.5978072706289671,
      "grad_norm": 0.06003202497959137,
      "learning_rate": 0.00017648148148148148,
      "loss": 0.1436,
      "step": 259
    },
    {
      "epoch": 0.6001154068090018,
      "grad_norm": 0.05292029678821564,
      "learning_rate": 0.0001763888888888889,
      "loss": 0.1361,
      "step": 260
    },
    {
      "epoch": 0.6024235429890363,
      "grad_norm": 0.05425536260008812,
      "learning_rate": 0.0001762962962962963,
      "loss": 0.1393,
      "step": 261
    },
    {
      "epoch": 0.604731679169071,
      "grad_norm": 0.05251134932041168,
      "learning_rate": 0.00017620370370370373,
      "loss": 0.1397,
      "step": 262
    },
    {
      "epoch": 0.6070398153491056,
      "grad_norm": 0.05703645199537277,
      "learning_rate": 0.00017611111111111112,
      "loss": 0.1398,
      "step": 263
    },
    {
      "epoch": 0.6093479515291402,
      "grad_norm": 0.05381687358021736,
      "learning_rate": 0.00017601851851851852,
      "loss": 0.1406,
      "step": 264
    },
    {
      "epoch": 0.6116560877091748,
      "grad_norm": 0.06381919980049133,
      "learning_rate": 0.00017592592592592595,
      "loss": 0.1389,
      "step": 265
    },
    {
      "epoch": 0.6139642238892095,
      "grad_norm": 0.06035248190164566,
      "learning_rate": 0.00017583333333333334,
      "loss": 0.1341,
      "step": 266
    },
    {
      "epoch": 0.6162723600692441,
      "grad_norm": 0.052462413907051086,
      "learning_rate": 0.00017574074074074074,
      "loss": 0.1333,
      "step": 267
    },
    {
      "epoch": 0.6185804962492787,
      "grad_norm": 0.05380314961075783,
      "learning_rate": 0.00017564814814814816,
      "loss": 0.1448,
      "step": 268
    },
    {
      "epoch": 0.6208886324293134,
      "grad_norm": 0.06288813799619675,
      "learning_rate": 0.00017555555555555556,
      "loss": 0.141,
      "step": 269
    },
    {
      "epoch": 0.6231967686093479,
      "grad_norm": 0.061172183603048325,
      "learning_rate": 0.00017546296296296296,
      "loss": 0.1411,
      "step": 270
    },
    {
      "epoch": 0.6255049047893826,
      "grad_norm": 0.05253341794013977,
      "learning_rate": 0.00017537037037037038,
      "loss": 0.1316,
      "step": 271
    },
    {
      "epoch": 0.6278130409694171,
      "grad_norm": 0.05998392030596733,
      "learning_rate": 0.00017527777777777778,
      "loss": 0.1408,
      "step": 272
    },
    {
      "epoch": 0.6301211771494518,
      "grad_norm": 0.05555344000458717,
      "learning_rate": 0.00017518518518518518,
      "loss": 0.1344,
      "step": 273
    },
    {
      "epoch": 0.6324293133294865,
      "grad_norm": 0.0645773783326149,
      "learning_rate": 0.0001750925925925926,
      "loss": 0.1366,
      "step": 274
    },
    {
      "epoch": 0.634737449509521,
      "grad_norm": 0.06639505177736282,
      "learning_rate": 0.000175,
      "loss": 0.1352,
      "step": 275
    },
    {
      "epoch": 0.6370455856895557,
      "grad_norm": 0.07881718873977661,
      "learning_rate": 0.0001749074074074074,
      "loss": 0.1357,
      "step": 276
    },
    {
      "epoch": 0.6393537218695903,
      "grad_norm": 0.0673324391245842,
      "learning_rate": 0.00017481481481481482,
      "loss": 0.1415,
      "step": 277
    },
    {
      "epoch": 0.641661858049625,
      "grad_norm": 0.07648260146379471,
      "learning_rate": 0.00017472222222222224,
      "loss": 0.1392,
      "step": 278
    },
    {
      "epoch": 0.6439699942296595,
      "grad_norm": 0.07975548505783081,
      "learning_rate": 0.00017462962962962964,
      "loss": 0.1352,
      "step": 279
    },
    {
      "epoch": 0.6462781304096942,
      "grad_norm": 0.09713549166917801,
      "learning_rate": 0.00017453703703703707,
      "loss": 0.1347,
      "step": 280
    },
    {
      "epoch": 0.6485862665897288,
      "grad_norm": 0.14314262568950653,
      "learning_rate": 0.00017444444444444446,
      "loss": 0.1338,
      "step": 281
    },
    {
      "epoch": 0.6508944027697634,
      "grad_norm": 0.13428884744644165,
      "learning_rate": 0.00017435185185185186,
      "loss": 0.1416,
      "step": 282
    },
    {
      "epoch": 0.6532025389497981,
      "grad_norm": 0.08313377946615219,
      "learning_rate": 0.00017425925925925928,
      "loss": 0.1365,
      "step": 283
    },
    {
      "epoch": 0.6555106751298326,
      "grad_norm": 0.10860215872526169,
      "learning_rate": 0.00017416666666666668,
      "loss": 0.1431,
      "step": 284
    },
    {
      "epoch": 0.6578188113098673,
      "grad_norm": 0.06758227199316025,
      "learning_rate": 0.00017407407407407408,
      "loss": 0.1401,
      "step": 285
    },
    {
      "epoch": 0.6601269474899019,
      "grad_norm": 0.07106103748083115,
      "learning_rate": 0.0001739814814814815,
      "loss": 0.1395,
      "step": 286
    },
    {
      "epoch": 0.6624350836699365,
      "grad_norm": 0.07097764313220978,
      "learning_rate": 0.0001738888888888889,
      "loss": 0.1423,
      "step": 287
    },
    {
      "epoch": 0.6647432198499712,
      "grad_norm": 0.05738036707043648,
      "learning_rate": 0.0001737962962962963,
      "loss": 0.1301,
      "step": 288
    },
    {
      "epoch": 0.6670513560300058,
      "grad_norm": 0.05762964487075806,
      "learning_rate": 0.0001737037037037037,
      "loss": 0.1251,
      "step": 289
    },
    {
      "epoch": 0.6693594922100404,
      "grad_norm": 0.06736747920513153,
      "learning_rate": 0.00017361111111111112,
      "loss": 0.1455,
      "step": 290
    },
    {
      "epoch": 0.671667628390075,
      "grad_norm": 0.06712755560874939,
      "learning_rate": 0.00017351851851851851,
      "loss": 0.1358,
      "step": 291
    },
    {
      "epoch": 0.6739757645701097,
      "grad_norm": 0.056817635893821716,
      "learning_rate": 0.0001734259259259259,
      "loss": 0.1375,
      "step": 292
    },
    {
      "epoch": 0.6762839007501442,
      "grad_norm": 0.05745913088321686,
      "learning_rate": 0.00017333333333333334,
      "loss": 0.1344,
      "step": 293
    },
    {
      "epoch": 0.6785920369301789,
      "grad_norm": 0.055840883404016495,
      "learning_rate": 0.00017324074074074076,
      "loss": 0.1339,
      "step": 294
    },
    {
      "epoch": 0.6809001731102136,
      "grad_norm": 0.05386283993721008,
      "learning_rate": 0.00017314814814814816,
      "loss": 0.1412,
      "step": 295
    },
    {
      "epoch": 0.6832083092902481,
      "grad_norm": 0.05768018960952759,
      "learning_rate": 0.00017305555555555558,
      "loss": 0.1498,
      "step": 296
    },
    {
      "epoch": 0.6855164454702828,
      "grad_norm": 0.057531874626874924,
      "learning_rate": 0.00017296296296296298,
      "loss": 0.1352,
      "step": 297
    },
    {
      "epoch": 0.6878245816503173,
      "grad_norm": 0.059435050934553146,
      "learning_rate": 0.00017287037037037038,
      "loss": 0.1333,
      "step": 298
    },
    {
      "epoch": 0.690132717830352,
      "grad_norm": 0.06282006204128265,
      "learning_rate": 0.0001727777777777778,
      "loss": 0.1379,
      "step": 299
    },
    {
      "epoch": 0.6924408540103866,
      "grad_norm": 0.06693487614393234,
      "learning_rate": 0.0001726851851851852,
      "loss": 0.1345,
      "step": 300
    },
    {
      "epoch": 0.6947489901904212,
      "grad_norm": 0.05821982026100159,
      "learning_rate": 0.0001725925925925926,
      "loss": 0.1348,
      "step": 301
    },
    {
      "epoch": 0.6970571263704558,
      "grad_norm": 0.05834485590457916,
      "learning_rate": 0.00017250000000000002,
      "loss": 0.138,
      "step": 302
    },
    {
      "epoch": 0.6993652625504905,
      "grad_norm": 0.06818992644548416,
      "learning_rate": 0.00017240740740740742,
      "loss": 0.1444,
      "step": 303
    },
    {
      "epoch": 0.7016733987305251,
      "grad_norm": 0.053661372512578964,
      "learning_rate": 0.0001723148148148148,
      "loss": 0.1349,
      "step": 304
    },
    {
      "epoch": 0.7039815349105597,
      "grad_norm": 0.06325794756412506,
      "learning_rate": 0.00017222222222222224,
      "loss": 0.1412,
      "step": 305
    },
    {
      "epoch": 0.7062896710905944,
      "grad_norm": 0.060706935822963715,
      "learning_rate": 0.00017212962962962963,
      "loss": 0.1348,
      "step": 306
    },
    {
      "epoch": 0.7085978072706289,
      "grad_norm": 0.05550307407975197,
      "learning_rate": 0.00017203703703703703,
      "loss": 0.1385,
      "step": 307
    },
    {
      "epoch": 0.7109059434506636,
      "grad_norm": 0.054470013827085495,
      "learning_rate": 0.00017194444444444446,
      "loss": 0.132,
      "step": 308
    },
    {
      "epoch": 0.7132140796306982,
      "grad_norm": 0.06142938882112503,
      "learning_rate": 0.00017185185185185185,
      "loss": 0.142,
      "step": 309
    },
    {
      "epoch": 0.7155222158107328,
      "grad_norm": 0.051808521151542664,
      "learning_rate": 0.00017175925925925928,
      "loss": 0.1381,
      "step": 310
    },
    {
      "epoch": 0.7178303519907675,
      "grad_norm": 0.05493675172328949,
      "learning_rate": 0.00017166666666666667,
      "loss": 0.1382,
      "step": 311
    },
    {
      "epoch": 0.7201384881708021,
      "grad_norm": 0.061504606157541275,
      "learning_rate": 0.0001715740740740741,
      "loss": 0.1365,
      "step": 312
    },
    {
      "epoch": 0.7224466243508367,
      "grad_norm": 0.07524706423282623,
      "learning_rate": 0.0001714814814814815,
      "loss": 0.1347,
      "step": 313
    },
    {
      "epoch": 0.7247547605308713,
      "grad_norm": 0.06291554123163223,
      "learning_rate": 0.0001713888888888889,
      "loss": 0.1346,
      "step": 314
    },
    {
      "epoch": 0.727062896710906,
      "grad_norm": 0.05485260486602783,
      "learning_rate": 0.00017129629629629632,
      "loss": 0.1323,
      "step": 315
    },
    {
      "epoch": 0.7293710328909405,
      "grad_norm": 0.0769985169172287,
      "learning_rate": 0.00017120370370370371,
      "loss": 0.1292,
      "step": 316
    },
    {
      "epoch": 0.7316791690709752,
      "grad_norm": 0.07557163387537003,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.1456,
      "step": 317
    },
    {
      "epoch": 0.7339873052510099,
      "grad_norm": 0.051482971757650375,
      "learning_rate": 0.00017101851851851854,
      "loss": 0.1355,
      "step": 318
    },
    {
      "epoch": 0.7362954414310444,
      "grad_norm": 0.06656330078840256,
      "learning_rate": 0.00017092592592592593,
      "loss": 0.1344,
      "step": 319
    },
    {
      "epoch": 0.7386035776110791,
      "grad_norm": 0.06052062287926674,
      "learning_rate": 0.00017083333333333333,
      "loss": 0.136,
      "step": 320
    },
    {
      "epoch": 0.7409117137911136,
      "grad_norm": 0.05914086848497391,
      "learning_rate": 0.00017074074074074075,
      "loss": 0.1345,
      "step": 321
    },
    {
      "epoch": 0.7432198499711483,
      "grad_norm": 0.05831776559352875,
      "learning_rate": 0.00017064814814814815,
      "loss": 0.1382,
      "step": 322
    },
    {
      "epoch": 0.7455279861511829,
      "grad_norm": 0.06073298677802086,
      "learning_rate": 0.00017055555555555555,
      "loss": 0.1282,
      "step": 323
    },
    {
      "epoch": 0.7478361223312175,
      "grad_norm": 0.06719812005758286,
      "learning_rate": 0.00017046296296296297,
      "loss": 0.1346,
      "step": 324
    },
    {
      "epoch": 0.7501442585112522,
      "grad_norm": 0.06601303070783615,
      "learning_rate": 0.00017037037037037037,
      "loss": 0.1356,
      "step": 325
    },
    {
      "epoch": 0.7524523946912868,
      "grad_norm": 0.0537571907043457,
      "learning_rate": 0.0001702777777777778,
      "loss": 0.1326,
      "step": 326
    },
    {
      "epoch": 0.7547605308713214,
      "grad_norm": 0.06268756091594696,
      "learning_rate": 0.0001701851851851852,
      "loss": 0.1346,
      "step": 327
    },
    {
      "epoch": 0.757068667051356,
      "grad_norm": 0.06491613388061523,
      "learning_rate": 0.00017009259259259262,
      "loss": 0.1345,
      "step": 328
    },
    {
      "epoch": 0.7593768032313907,
      "grad_norm": 0.053328026086091995,
      "learning_rate": 0.00017,
      "loss": 0.1296,
      "step": 329
    },
    {
      "epoch": 0.7616849394114252,
      "grad_norm": 0.06269555538892746,
      "learning_rate": 0.00016990740740740744,
      "loss": 0.132,
      "step": 330
    },
    {
      "epoch": 0.7639930755914599,
      "grad_norm": 0.055639881640672684,
      "learning_rate": 0.00016981481481481483,
      "loss": 0.1319,
      "step": 331
    },
    {
      "epoch": 0.7663012117714946,
      "grad_norm": 0.051506321877241135,
      "learning_rate": 0.00016972222222222223,
      "loss": 0.1352,
      "step": 332
    },
    {
      "epoch": 0.7686093479515291,
      "grad_norm": 0.05923004820942879,
      "learning_rate": 0.00016962962962962963,
      "loss": 0.1379,
      "step": 333
    },
    {
      "epoch": 0.7709174841315638,
      "grad_norm": 0.05714567005634308,
      "learning_rate": 0.00016953703703703705,
      "loss": 0.1348,
      "step": 334
    },
    {
      "epoch": 0.7732256203115984,
      "grad_norm": 0.054765790700912476,
      "learning_rate": 0.00016944444444444445,
      "loss": 0.1405,
      "step": 335
    },
    {
      "epoch": 0.775533756491633,
      "grad_norm": 0.06484013050794601,
      "learning_rate": 0.00016935185185185185,
      "loss": 0.138,
      "step": 336
    },
    {
      "epoch": 0.7778418926716676,
      "grad_norm": 0.06624964624643326,
      "learning_rate": 0.00016925925925925927,
      "loss": 0.1366,
      "step": 337
    },
    {
      "epoch": 0.7801500288517023,
      "grad_norm": 0.059534184634685516,
      "learning_rate": 0.00016916666666666667,
      "loss": 0.1355,
      "step": 338
    },
    {
      "epoch": 0.7824581650317368,
      "grad_norm": 0.05257304385304451,
      "learning_rate": 0.00016907407407407406,
      "loss": 0.1324,
      "step": 339
    },
    {
      "epoch": 0.7847663012117715,
      "grad_norm": 0.06176763027906418,
      "learning_rate": 0.0001689814814814815,
      "loss": 0.1328,
      "step": 340
    },
    {
      "epoch": 0.7870744373918062,
      "grad_norm": 0.06330476701259613,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.1286,
      "step": 341
    },
    {
      "epoch": 0.7893825735718407,
      "grad_norm": 0.06781090050935745,
      "learning_rate": 0.0001687962962962963,
      "loss": 0.1351,
      "step": 342
    },
    {
      "epoch": 0.7916907097518754,
      "grad_norm": 0.07244280725717545,
      "learning_rate": 0.0001687037037037037,
      "loss": 0.1318,
      "step": 343
    },
    {
      "epoch": 0.79399884593191,
      "grad_norm": 0.06649068742990494,
      "learning_rate": 0.00016861111111111113,
      "loss": 0.1402,
      "step": 344
    },
    {
      "epoch": 0.7963069821119446,
      "grad_norm": 0.06082584336400032,
      "learning_rate": 0.00016851851851851853,
      "loss": 0.1305,
      "step": 345
    },
    {
      "epoch": 0.7986151182919792,
      "grad_norm": 0.06225201115012169,
      "learning_rate": 0.00016842592592592595,
      "loss": 0.1343,
      "step": 346
    },
    {
      "epoch": 0.8009232544720138,
      "grad_norm": 0.08486352860927582,
      "learning_rate": 0.00016833333333333335,
      "loss": 0.1354,
      "step": 347
    },
    {
      "epoch": 0.8032313906520485,
      "grad_norm": 0.10897298902273178,
      "learning_rate": 0.00016824074074074075,
      "loss": 0.133,
      "step": 348
    },
    {
      "epoch": 0.8055395268320831,
      "grad_norm": 0.12009774893522263,
      "learning_rate": 0.00016814814814814817,
      "loss": 0.135,
      "step": 349
    },
    {
      "epoch": 0.8078476630121177,
      "grad_norm": 0.11186303943395615,
      "learning_rate": 0.00016805555555555557,
      "loss": 0.1345,
      "step": 350
    },
    {
      "epoch": 0.8101557991921523,
      "grad_norm": 0.07706065475940704,
      "learning_rate": 0.00016796296296296297,
      "loss": 0.1273,
      "step": 351
    },
    {
      "epoch": 0.812463935372187,
      "grad_norm": 0.0626080259680748,
      "learning_rate": 0.0001678703703703704,
      "loss": 0.125,
      "step": 352
    },
    {
      "epoch": 0.8147720715522215,
      "grad_norm": 0.06778427958488464,
      "learning_rate": 0.0001677777777777778,
      "loss": 0.1243,
      "step": 353
    },
    {
      "epoch": 0.8170802077322562,
      "grad_norm": 0.06064091995358467,
      "learning_rate": 0.00016768518518518518,
      "loss": 0.1415,
      "step": 354
    },
    {
      "epoch": 0.8193883439122909,
      "grad_norm": 0.07406813651323318,
      "learning_rate": 0.00016759259259259258,
      "loss": 0.1362,
      "step": 355
    },
    {
      "epoch": 0.8216964800923254,
      "grad_norm": 0.06845991313457489,
      "learning_rate": 0.0001675,
      "loss": 0.1247,
      "step": 356
    },
    {
      "epoch": 0.8240046162723601,
      "grad_norm": 0.07087384909391403,
      "learning_rate": 0.0001674074074074074,
      "loss": 0.1388,
      "step": 357
    },
    {
      "epoch": 0.8263127524523947,
      "grad_norm": 0.060527075082063675,
      "learning_rate": 0.0001673148148148148,
      "loss": 0.1335,
      "step": 358
    },
    {
      "epoch": 0.8286208886324293,
      "grad_norm": 0.06400463730096817,
      "learning_rate": 0.00016722222222222222,
      "loss": 0.1311,
      "step": 359
    },
    {
      "epoch": 0.8309290248124639,
      "grad_norm": 0.05668658763170242,
      "learning_rate": 0.00016712962962962965,
      "loss": 0.1282,
      "step": 360
    },
    {
      "epoch": 0.8332371609924986,
      "grad_norm": 0.06403769552707672,
      "learning_rate": 0.00016703703703703705,
      "loss": 0.1284,
      "step": 361
    },
    {
      "epoch": 0.8355452971725332,
      "grad_norm": 0.0756312683224678,
      "learning_rate": 0.00016694444444444447,
      "loss": 0.1272,
      "step": 362
    },
    {
      "epoch": 0.8378534333525678,
      "grad_norm": 0.0708053857088089,
      "learning_rate": 0.00016685185185185187,
      "loss": 0.1343,
      "step": 363
    },
    {
      "epoch": 0.8401615695326025,
      "grad_norm": 0.06252378970384598,
      "learning_rate": 0.00016675925925925926,
      "loss": 0.135,
      "step": 364
    },
    {
      "epoch": 0.842469705712637,
      "grad_norm": 0.06451309472322464,
      "learning_rate": 0.0001666666666666667,
      "loss": 0.1405,
      "step": 365
    },
    {
      "epoch": 0.8447778418926717,
      "grad_norm": 0.06427218019962311,
      "learning_rate": 0.00016657407407407409,
      "loss": 0.1389,
      "step": 366
    },
    {
      "epoch": 0.8470859780727062,
      "grad_norm": 0.07281991839408875,
      "learning_rate": 0.00016648148148148148,
      "loss": 0.1303,
      "step": 367
    },
    {
      "epoch": 0.8493941142527409,
      "grad_norm": 0.06674263626337051,
      "learning_rate": 0.0001663888888888889,
      "loss": 0.1315,
      "step": 368
    },
    {
      "epoch": 0.8517022504327756,
      "grad_norm": 0.05987542122602463,
      "learning_rate": 0.0001662962962962963,
      "loss": 0.1401,
      "step": 369
    },
    {
      "epoch": 0.8540103866128101,
      "grad_norm": 0.058871105313301086,
      "learning_rate": 0.0001662037037037037,
      "loss": 0.1382,
      "step": 370
    },
    {
      "epoch": 0.8563185227928448,
      "grad_norm": 0.07391886413097382,
      "learning_rate": 0.00016611111111111113,
      "loss": 0.1329,
      "step": 371
    },
    {
      "epoch": 0.8586266589728794,
      "grad_norm": 0.06960415840148926,
      "learning_rate": 0.00016601851851851852,
      "loss": 0.1287,
      "step": 372
    },
    {
      "epoch": 0.860934795152914,
      "grad_norm": 0.05401191487908363,
      "learning_rate": 0.00016592592592592592,
      "loss": 0.1311,
      "step": 373
    },
    {
      "epoch": 0.8632429313329486,
      "grad_norm": 0.05892745032906532,
      "learning_rate": 0.00016583333333333334,
      "loss": 0.1354,
      "step": 374
    },
    {
      "epoch": 0.8655510675129833,
      "grad_norm": 0.05646412447094917,
      "learning_rate": 0.00016574074074074074,
      "loss": 0.1319,
      "step": 375
    },
    {
      "epoch": 0.8678592036930178,
      "grad_norm": 0.05770871415734291,
      "learning_rate": 0.00016564814814814817,
      "loss": 0.1346,
      "step": 376
    },
    {
      "epoch": 0.8701673398730525,
      "grad_norm": 0.056514665484428406,
      "learning_rate": 0.00016555555555555556,
      "loss": 0.1353,
      "step": 377
    },
    {
      "epoch": 0.8724754760530872,
      "grad_norm": 0.05817136541008949,
      "learning_rate": 0.00016546296296296299,
      "loss": 0.1341,
      "step": 378
    },
    {
      "epoch": 0.8747836122331217,
      "grad_norm": 0.05660689249634743,
      "learning_rate": 0.00016537037037037038,
      "loss": 0.127,
      "step": 379
    },
    {
      "epoch": 0.8770917484131564,
      "grad_norm": 0.056419163942337036,
      "learning_rate": 0.00016527777777777778,
      "loss": 0.1246,
      "step": 380
    },
    {
      "epoch": 0.879399884593191,
      "grad_norm": 0.05790020897984505,
      "learning_rate": 0.0001651851851851852,
      "loss": 0.1359,
      "step": 381
    },
    {
      "epoch": 0.8817080207732256,
      "grad_norm": 0.06118219345808029,
      "learning_rate": 0.0001650925925925926,
      "loss": 0.1337,
      "step": 382
    },
    {
      "epoch": 0.8840161569532602,
      "grad_norm": 0.06304623931646347,
      "learning_rate": 0.000165,
      "loss": 0.1297,
      "step": 383
    },
    {
      "epoch": 0.8863242931332949,
      "grad_norm": 0.06479991227388382,
      "learning_rate": 0.00016490740740740742,
      "loss": 0.1276,
      "step": 384
    },
    {
      "epoch": 0.8886324293133295,
      "grad_norm": 0.05389687418937683,
      "learning_rate": 0.00016481481481481482,
      "loss": 0.1287,
      "step": 385
    },
    {
      "epoch": 0.8909405654933641,
      "grad_norm": 0.06116378307342529,
      "learning_rate": 0.00016472222222222222,
      "loss": 0.135,
      "step": 386
    },
    {
      "epoch": 0.8932487016733988,
      "grad_norm": 0.07239045202732086,
      "learning_rate": 0.00016462962962962964,
      "loss": 0.1307,
      "step": 387
    },
    {
      "epoch": 0.8955568378534333,
      "grad_norm": 0.06322701275348663,
      "learning_rate": 0.00016453703703703704,
      "loss": 0.1296,
      "step": 388
    },
    {
      "epoch": 0.897864974033468,
      "grad_norm": 0.05430390313267708,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.1244,
      "step": 389
    },
    {
      "epoch": 0.9001731102135025,
      "grad_norm": 0.07260555773973465,
      "learning_rate": 0.00016435185185185186,
      "loss": 0.1358,
      "step": 390
    },
    {
      "epoch": 0.9024812463935372,
      "grad_norm": 0.058297574520111084,
      "learning_rate": 0.00016425925925925926,
      "loss": 0.1351,
      "step": 391
    },
    {
      "epoch": 0.9047893825735719,
      "grad_norm": 0.06689532101154327,
      "learning_rate": 0.00016416666666666668,
      "loss": 0.1239,
      "step": 392
    },
    {
      "epoch": 0.9070975187536064,
      "grad_norm": 0.07593647390604019,
      "learning_rate": 0.00016407407407407408,
      "loss": 0.1317,
      "step": 393
    },
    {
      "epoch": 0.9094056549336411,
      "grad_norm": 0.0626072883605957,
      "learning_rate": 0.0001639814814814815,
      "loss": 0.1312,
      "step": 394
    },
    {
      "epoch": 0.9117137911136757,
      "grad_norm": 0.05915417894721031,
      "learning_rate": 0.0001638888888888889,
      "loss": 0.1303,
      "step": 395
    },
    {
      "epoch": 0.9140219272937103,
      "grad_norm": 0.062168218195438385,
      "learning_rate": 0.00016379629629629632,
      "loss": 0.1351,
      "step": 396
    },
    {
      "epoch": 0.9163300634737449,
      "grad_norm": 0.05277865752577782,
      "learning_rate": 0.00016370370370370372,
      "loss": 0.1289,
      "step": 397
    },
    {
      "epoch": 0.9186381996537796,
      "grad_norm": 0.05689583346247673,
      "learning_rate": 0.00016361111111111112,
      "loss": 0.1272,
      "step": 398
    },
    {
      "epoch": 0.9209463358338142,
      "grad_norm": 0.061158064752817154,
      "learning_rate": 0.00016351851851851852,
      "loss": 0.1364,
      "step": 399
    },
    {
      "epoch": 0.9232544720138488,
      "grad_norm": 0.06694123148918152,
      "learning_rate": 0.00016342592592592594,
      "loss": 0.1298,
      "step": 400
    },
    {
      "epoch": 0.9255626081938835,
      "grad_norm": 0.06037142872810364,
      "learning_rate": 0.00016333333333333334,
      "loss": 0.1266,
      "step": 401
    },
    {
      "epoch": 0.927870744373918,
      "grad_norm": 0.05615115165710449,
      "learning_rate": 0.00016324074074074073,
      "loss": 0.1333,
      "step": 402
    },
    {
      "epoch": 0.9301788805539527,
      "grad_norm": 0.0579330176115036,
      "learning_rate": 0.00016314814814814816,
      "loss": 0.1242,
      "step": 403
    },
    {
      "epoch": 0.9324870167339873,
      "grad_norm": 0.06039915978908539,
      "learning_rate": 0.00016305555555555556,
      "loss": 0.131,
      "step": 404
    },
    {
      "epoch": 0.9347951529140219,
      "grad_norm": 0.053155019879341125,
      "learning_rate": 0.00016296296296296295,
      "loss": 0.1275,
      "step": 405
    },
    {
      "epoch": 0.9371032890940566,
      "grad_norm": 0.05309950187802315,
      "learning_rate": 0.00016287037037037038,
      "loss": 0.135,
      "step": 406
    },
    {
      "epoch": 0.9394114252740912,
      "grad_norm": 0.05852155759930611,
      "learning_rate": 0.00016277777777777777,
      "loss": 0.1293,
      "step": 407
    },
    {
      "epoch": 0.9417195614541258,
      "grad_norm": 0.06083635985851288,
      "learning_rate": 0.0001626851851851852,
      "loss": 0.136,
      "step": 408
    },
    {
      "epoch": 0.9440276976341604,
      "grad_norm": 0.05646814778447151,
      "learning_rate": 0.0001625925925925926,
      "loss": 0.1329,
      "step": 409
    },
    {
      "epoch": 0.9463358338141951,
      "grad_norm": 0.06743089854717255,
      "learning_rate": 0.00016250000000000002,
      "loss": 0.1273,
      "step": 410
    },
    {
      "epoch": 0.9486439699942296,
      "grad_norm": 0.06807095557451248,
      "learning_rate": 0.00016240740740740742,
      "loss": 0.1385,
      "step": 411
    },
    {
      "epoch": 0.9509521061742643,
      "grad_norm": 0.06895305216312408,
      "learning_rate": 0.00016231481481481484,
      "loss": 0.1287,
      "step": 412
    },
    {
      "epoch": 0.953260242354299,
      "grad_norm": 0.07045155763626099,
      "learning_rate": 0.00016222222222222224,
      "loss": 0.1302,
      "step": 413
    },
    {
      "epoch": 0.9555683785343335,
      "grad_norm": 0.054334819316864014,
      "learning_rate": 0.00016212962962962964,
      "loss": 0.1326,
      "step": 414
    },
    {
      "epoch": 0.9578765147143682,
      "grad_norm": 0.0692325085401535,
      "learning_rate": 0.00016203703703703706,
      "loss": 0.1342,
      "step": 415
    },
    {
      "epoch": 0.9601846508944027,
      "grad_norm": 0.06285296380519867,
      "learning_rate": 0.00016194444444444446,
      "loss": 0.1273,
      "step": 416
    },
    {
      "epoch": 0.9624927870744374,
      "grad_norm": 0.0595967173576355,
      "learning_rate": 0.00016185185185185185,
      "loss": 0.1296,
      "step": 417
    },
    {
      "epoch": 0.964800923254472,
      "grad_norm": 0.06929370015859604,
      "learning_rate": 0.00016175925925925928,
      "loss": 0.1317,
      "step": 418
    },
    {
      "epoch": 0.9671090594345066,
      "grad_norm": 0.07214278727769852,
      "learning_rate": 0.00016166666666666668,
      "loss": 0.1266,
      "step": 419
    },
    {
      "epoch": 0.9694171956145412,
      "grad_norm": 0.06406287848949432,
      "learning_rate": 0.00016157407407407407,
      "loss": 0.136,
      "step": 420
    },
    {
      "epoch": 0.9717253317945759,
      "grad_norm": 0.05603834241628647,
      "learning_rate": 0.0001614814814814815,
      "loss": 0.129,
      "step": 421
    },
    {
      "epoch": 0.9740334679746105,
      "grad_norm": 0.06080767512321472,
      "learning_rate": 0.0001613888888888889,
      "loss": 0.1254,
      "step": 422
    },
    {
      "epoch": 0.9763416041546451,
      "grad_norm": 0.0604126900434494,
      "learning_rate": 0.0001612962962962963,
      "loss": 0.1269,
      "step": 423
    },
    {
      "epoch": 0.9786497403346798,
      "grad_norm": 0.06764354556798935,
      "learning_rate": 0.0001612037037037037,
      "loss": 0.1338,
      "step": 424
    },
    {
      "epoch": 0.9809578765147143,
      "grad_norm": 0.057208798825740814,
      "learning_rate": 0.0001611111111111111,
      "loss": 0.1342,
      "step": 425
    },
    {
      "epoch": 0.983266012694749,
      "grad_norm": 0.05770406872034073,
      "learning_rate": 0.00016101851851851854,
      "loss": 0.1389,
      "step": 426
    },
    {
      "epoch": 0.9855741488747836,
      "grad_norm": 0.05363389477133751,
      "learning_rate": 0.00016092592592592593,
      "loss": 0.1269,
      "step": 427
    },
    {
      "epoch": 0.9878822850548182,
      "grad_norm": 0.06754197925329208,
      "learning_rate": 0.00016083333333333336,
      "loss": 0.1229,
      "step": 428
    },
    {
      "epoch": 0.9901904212348529,
      "grad_norm": 0.07205870002508163,
      "learning_rate": 0.00016074074074074075,
      "loss": 0.1277,
      "step": 429
    },
    {
      "epoch": 0.9924985574148875,
      "grad_norm": 0.06645669788122177,
      "learning_rate": 0.00016064814814814815,
      "loss": 0.137,
      "step": 430
    },
    {
      "epoch": 0.9948066935949221,
      "grad_norm": 0.0634750947356224,
      "learning_rate": 0.00016055555555555558,
      "loss": 0.1326,
      "step": 431
    },
    {
      "epoch": 0.9971148297749567,
      "grad_norm": 0.06000895798206329,
      "learning_rate": 0.00016046296296296297,
      "loss": 0.1341,
      "step": 432
    },
    {
      "epoch": 0.9994229659549914,
      "grad_norm": 0.06601957976818085,
      "learning_rate": 0.00016037037037037037,
      "loss": 0.1293,
      "step": 433
    },
    {
      "epoch": 1.001731102135026,
      "grad_norm": 0.11583652347326279,
      "learning_rate": 0.0001602777777777778,
      "loss": 0.1237,
      "step": 434
    },
    {
      "epoch": 1.0040392383150605,
      "grad_norm": 0.2113158404827118,
      "learning_rate": 0.0001601851851851852,
      "loss": 0.1402,
      "step": 435
    },
    {
      "epoch": 1.0063473744950953,
      "grad_norm": 0.15687702596187592,
      "learning_rate": 0.0001600925925925926,
      "loss": 0.1317,
      "step": 436
    },
    {
      "epoch": 1.0086555106751298,
      "grad_norm": 0.1036396473646164,
      "learning_rate": 0.00016,
      "loss": 0.1307,
      "step": 437
    },
    {
      "epoch": 1.0109636468551644,
      "grad_norm": 0.0797649472951889,
      "learning_rate": 0.0001599074074074074,
      "loss": 0.1288,
      "step": 438
    },
    {
      "epoch": 1.0132717830351992,
      "grad_norm": 0.09061572700738907,
      "learning_rate": 0.0001598148148148148,
      "loss": 0.1379,
      "step": 439
    },
    {
      "epoch": 1.0155799192152337,
      "grad_norm": 0.06906070560216904,
      "learning_rate": 0.00015972222222222223,
      "loss": 0.13,
      "step": 440
    },
    {
      "epoch": 1.0178880553952683,
      "grad_norm": 0.07185190171003342,
      "learning_rate": 0.00015962962962962963,
      "loss": 0.1299,
      "step": 441
    },
    {
      "epoch": 1.0201961915753028,
      "grad_norm": 0.07263153791427612,
      "learning_rate": 0.00015953703703703705,
      "loss": 0.1255,
      "step": 442
    },
    {
      "epoch": 1.0225043277553376,
      "grad_norm": 0.06459320336580276,
      "learning_rate": 0.00015944444444444445,
      "loss": 0.131,
      "step": 443
    },
    {
      "epoch": 1.0248124639353722,
      "grad_norm": 0.06365115940570831,
      "learning_rate": 0.00015935185185185187,
      "loss": 0.1326,
      "step": 444
    },
    {
      "epoch": 1.0271206001154067,
      "grad_norm": 0.06586971133947372,
      "learning_rate": 0.00015925925925925927,
      "loss": 0.129,
      "step": 445
    },
    {
      "epoch": 1.0294287362954415,
      "grad_norm": 0.06053543835878372,
      "learning_rate": 0.00015916666666666667,
      "loss": 0.1192,
      "step": 446
    },
    {
      "epoch": 1.031736872475476,
      "grad_norm": 0.06116175651550293,
      "learning_rate": 0.0001590740740740741,
      "loss": 0.1273,
      "step": 447
    },
    {
      "epoch": 1.0340450086555106,
      "grad_norm": 0.0620986744761467,
      "learning_rate": 0.0001589814814814815,
      "loss": 0.1236,
      "step": 448
    },
    {
      "epoch": 1.0363531448355452,
      "grad_norm": 0.05783109739422798,
      "learning_rate": 0.0001588888888888889,
      "loss": 0.1268,
      "step": 449
    },
    {
      "epoch": 1.03866128101558,
      "grad_norm": 0.059103306382894516,
      "learning_rate": 0.0001587962962962963,
      "loss": 0.1261,
      "step": 450
    },
    {
      "epoch": 1.0409694171956145,
      "grad_norm": 0.051785897463560104,
      "learning_rate": 0.0001587037037037037,
      "loss": 0.1234,
      "step": 451
    },
    {
      "epoch": 1.043277553375649,
      "grad_norm": 0.06014087796211243,
      "learning_rate": 0.0001586111111111111,
      "loss": 0.1276,
      "step": 452
    },
    {
      "epoch": 1.0455856895556839,
      "grad_norm": 0.057974543422460556,
      "learning_rate": 0.00015851851851851853,
      "loss": 0.1301,
      "step": 453
    },
    {
      "epoch": 1.0478938257357184,
      "grad_norm": 0.06165377050638199,
      "learning_rate": 0.00015842592592592593,
      "loss": 0.1297,
      "step": 454
    },
    {
      "epoch": 1.050201961915753,
      "grad_norm": 0.06181812658905983,
      "learning_rate": 0.00015833333333333332,
      "loss": 0.1261,
      "step": 455
    },
    {
      "epoch": 1.0525100980957875,
      "grad_norm": 0.06410099565982819,
      "learning_rate": 0.00015824074074074075,
      "loss": 0.1236,
      "step": 456
    },
    {
      "epoch": 1.0548182342758223,
      "grad_norm": 0.05882523953914642,
      "learning_rate": 0.00015814814814814815,
      "loss": 0.1275,
      "step": 457
    },
    {
      "epoch": 1.057126370455857,
      "grad_norm": 0.05501971393823624,
      "learning_rate": 0.00015805555555555557,
      "loss": 0.1261,
      "step": 458
    },
    {
      "epoch": 1.0594345066358914,
      "grad_norm": 0.05703722685575485,
      "learning_rate": 0.00015796296296296297,
      "loss": 0.1298,
      "step": 459
    },
    {
      "epoch": 1.0617426428159262,
      "grad_norm": 0.05989052355289459,
      "learning_rate": 0.0001578703703703704,
      "loss": 0.1324,
      "step": 460
    },
    {
      "epoch": 1.0640507789959608,
      "grad_norm": 0.06194508820772171,
      "learning_rate": 0.0001577777777777778,
      "loss": 0.1315,
      "step": 461
    },
    {
      "epoch": 1.0663589151759953,
      "grad_norm": 0.058788470923900604,
      "learning_rate": 0.0001576851851851852,
      "loss": 0.1294,
      "step": 462
    },
    {
      "epoch": 1.06866705135603,
      "grad_norm": 0.05487912893295288,
      "learning_rate": 0.0001575925925925926,
      "loss": 0.1279,
      "step": 463
    },
    {
      "epoch": 1.0709751875360647,
      "grad_norm": 0.062160443514585495,
      "learning_rate": 0.0001575,
      "loss": 0.1266,
      "step": 464
    },
    {
      "epoch": 1.0732833237160992,
      "grad_norm": 0.06403318792581558,
      "learning_rate": 0.00015740740740740743,
      "loss": 0.1257,
      "step": 465
    },
    {
      "epoch": 1.0755914598961338,
      "grad_norm": 0.05971311405301094,
      "learning_rate": 0.00015731481481481483,
      "loss": 0.1278,
      "step": 466
    },
    {
      "epoch": 1.0778995960761686,
      "grad_norm": 0.060210853815078735,
      "learning_rate": 0.00015722222222222223,
      "loss": 0.1191,
      "step": 467
    },
    {
      "epoch": 1.0802077322562031,
      "grad_norm": 0.07795955240726471,
      "learning_rate": 0.00015712962962962962,
      "loss": 0.1237,
      "step": 468
    },
    {
      "epoch": 1.0825158684362377,
      "grad_norm": 0.06535372138023376,
      "learning_rate": 0.00015703703703703705,
      "loss": 0.124,
      "step": 469
    },
    {
      "epoch": 1.0848240046162723,
      "grad_norm": 0.06483720242977142,
      "learning_rate": 0.00015694444444444444,
      "loss": 0.1332,
      "step": 470
    },
    {
      "epoch": 1.087132140796307,
      "grad_norm": 0.057768333703279495,
      "learning_rate": 0.00015685185185185184,
      "loss": 0.1268,
      "step": 471
    },
    {
      "epoch": 1.0894402769763416,
      "grad_norm": 0.06422332674264908,
      "learning_rate": 0.00015675925925925926,
      "loss": 0.1311,
      "step": 472
    },
    {
      "epoch": 1.0917484131563762,
      "grad_norm": 0.06580941379070282,
      "learning_rate": 0.00015666666666666666,
      "loss": 0.1306,
      "step": 473
    },
    {
      "epoch": 1.094056549336411,
      "grad_norm": 0.06723253428936005,
      "learning_rate": 0.00015657407407407409,
      "loss": 0.1199,
      "step": 474
    },
    {
      "epoch": 1.0963646855164455,
      "grad_norm": 0.05644245073199272,
      "learning_rate": 0.00015648148148148148,
      "loss": 0.1261,
      "step": 475
    },
    {
      "epoch": 1.09867282169648,
      "grad_norm": 0.05643429979681969,
      "learning_rate": 0.0001563888888888889,
      "loss": 0.1315,
      "step": 476
    },
    {
      "epoch": 1.1009809578765146,
      "grad_norm": 0.06453484296798706,
      "learning_rate": 0.0001562962962962963,
      "loss": 0.1285,
      "step": 477
    },
    {
      "epoch": 1.1032890940565494,
      "grad_norm": 0.06879416853189468,
      "learning_rate": 0.00015620370370370373,
      "loss": 0.1321,
      "step": 478
    },
    {
      "epoch": 1.105597230236584,
      "grad_norm": 0.06542493402957916,
      "learning_rate": 0.00015611111111111113,
      "loss": 0.1284,
      "step": 479
    },
    {
      "epoch": 1.1079053664166185,
      "grad_norm": 0.055879369378089905,
      "learning_rate": 0.00015601851851851852,
      "loss": 0.1371,
      "step": 480
    },
    {
      "epoch": 1.1102135025966533,
      "grad_norm": 0.05824080854654312,
      "learning_rate": 0.00015592592592592595,
      "loss": 0.129,
      "step": 481
    },
    {
      "epoch": 1.1125216387766879,
      "grad_norm": 0.06523554027080536,
      "learning_rate": 0.00015583333333333334,
      "loss": 0.1241,
      "step": 482
    },
    {
      "epoch": 1.1148297749567224,
      "grad_norm": 0.05699508264660835,
      "learning_rate": 0.00015574074074074074,
      "loss": 0.1237,
      "step": 483
    },
    {
      "epoch": 1.117137911136757,
      "grad_norm": 0.05805926024913788,
      "learning_rate": 0.00015564814814814817,
      "loss": 0.1302,
      "step": 484
    },
    {
      "epoch": 1.1194460473167918,
      "grad_norm": 0.05902712047100067,
      "learning_rate": 0.00015555555555555556,
      "loss": 0.1233,
      "step": 485
    },
    {
      "epoch": 1.1217541834968263,
      "grad_norm": 0.05701911076903343,
      "learning_rate": 0.00015546296296296296,
      "loss": 0.1292,
      "step": 486
    },
    {
      "epoch": 1.1240623196768609,
      "grad_norm": 0.057408515363931656,
      "learning_rate": 0.00015537037037037038,
      "loss": 0.1353,
      "step": 487
    },
    {
      "epoch": 1.1263704558568954,
      "grad_norm": 0.057401470839977264,
      "learning_rate": 0.00015527777777777778,
      "loss": 0.1232,
      "step": 488
    },
    {
      "epoch": 1.1286785920369302,
      "grad_norm": 0.062264181673526764,
      "learning_rate": 0.00015518518518518518,
      "loss": 0.124,
      "step": 489
    },
    {
      "epoch": 1.1309867282169648,
      "grad_norm": 0.06954425573348999,
      "learning_rate": 0.0001550925925925926,
      "loss": 0.1273,
      "step": 490
    },
    {
      "epoch": 1.1332948643969993,
      "grad_norm": 0.06275420635938644,
      "learning_rate": 0.000155,
      "loss": 0.1318,
      "step": 491
    },
    {
      "epoch": 1.1356030005770341,
      "grad_norm": 0.059264037758111954,
      "learning_rate": 0.00015490740740740742,
      "loss": 0.1289,
      "step": 492
    },
    {
      "epoch": 1.1379111367570687,
      "grad_norm": 0.06056118384003639,
      "learning_rate": 0.00015481481481481482,
      "loss": 0.1271,
      "step": 493
    },
    {
      "epoch": 1.1402192729371032,
      "grad_norm": 0.06082379072904587,
      "learning_rate": 0.00015472222222222225,
      "loss": 0.1288,
      "step": 494
    },
    {
      "epoch": 1.1425274091171378,
      "grad_norm": 0.06065623089671135,
      "learning_rate": 0.00015462962962962964,
      "loss": 0.1151,
      "step": 495
    },
    {
      "epoch": 1.1448355452971726,
      "grad_norm": 0.0642370954155922,
      "learning_rate": 0.00015453703703703704,
      "loss": 0.1252,
      "step": 496
    },
    {
      "epoch": 1.1471436814772071,
      "grad_norm": 0.05329868197441101,
      "learning_rate": 0.00015444444444444446,
      "loss": 0.1233,
      "step": 497
    },
    {
      "epoch": 1.1494518176572417,
      "grad_norm": 0.057288359850645065,
      "learning_rate": 0.00015435185185185186,
      "loss": 0.1253,
      "step": 498
    },
    {
      "epoch": 1.1517599538372765,
      "grad_norm": 0.06774638593196869,
      "learning_rate": 0.00015425925925925926,
      "loss": 0.1278,
      "step": 499
    },
    {
      "epoch": 1.154068090017311,
      "grad_norm": 0.06254816800355911,
      "learning_rate": 0.00015416666666666668,
      "loss": 0.1284,
      "step": 500
    },
    {
      "epoch": 1.1563762261973456,
      "grad_norm": 0.05919085815548897,
      "learning_rate": 0.00015407407407407408,
      "loss": 0.1286,
      "step": 501
    },
    {
      "epoch": 1.1586843623773802,
      "grad_norm": 0.07059169560670853,
      "learning_rate": 0.00015398148148148148,
      "loss": 0.13,
      "step": 502
    },
    {
      "epoch": 1.160992498557415,
      "grad_norm": 0.08087097108364105,
      "learning_rate": 0.0001538888888888889,
      "loss": 0.1321,
      "step": 503
    },
    {
      "epoch": 1.1633006347374495,
      "grad_norm": 0.09199072420597076,
      "learning_rate": 0.0001537962962962963,
      "loss": 0.1246,
      "step": 504
    },
    {
      "epoch": 1.165608770917484,
      "grad_norm": 0.11305177956819534,
      "learning_rate": 0.0001537037037037037,
      "loss": 0.131,
      "step": 505
    },
    {
      "epoch": 1.1679169070975188,
      "grad_norm": 0.09376886487007141,
      "learning_rate": 0.00015361111111111112,
      "loss": 0.1319,
      "step": 506
    },
    {
      "epoch": 1.1702250432775534,
      "grad_norm": 0.08503235876560211,
      "learning_rate": 0.00015351851851851852,
      "loss": 0.1242,
      "step": 507
    },
    {
      "epoch": 1.172533179457588,
      "grad_norm": 0.12891536951065063,
      "learning_rate": 0.00015342592592592594,
      "loss": 0.1309,
      "step": 508
    },
    {
      "epoch": 1.1748413156376225,
      "grad_norm": 0.08906060457229614,
      "learning_rate": 0.00015333333333333334,
      "loss": 0.1255,
      "step": 509
    },
    {
      "epoch": 1.1771494518176573,
      "grad_norm": 0.09136375784873962,
      "learning_rate": 0.00015324074074074076,
      "loss": 0.1279,
      "step": 510
    },
    {
      "epoch": 1.1794575879976918,
      "grad_norm": 0.12788566946983337,
      "learning_rate": 0.00015314814814814816,
      "loss": 0.1205,
      "step": 511
    },
    {
      "epoch": 1.1817657241777264,
      "grad_norm": 0.07637536525726318,
      "learning_rate": 0.00015305555555555556,
      "loss": 0.1303,
      "step": 512
    },
    {
      "epoch": 1.1840738603577612,
      "grad_norm": 0.06925041228532791,
      "learning_rate": 0.00015296296296296298,
      "loss": 0.1256,
      "step": 513
    },
    {
      "epoch": 1.1863819965377957,
      "grad_norm": 0.06945720314979553,
      "learning_rate": 0.00015287037037037038,
      "loss": 0.1268,
      "step": 514
    },
    {
      "epoch": 1.1886901327178303,
      "grad_norm": 0.06018393486738205,
      "learning_rate": 0.00015277777777777777,
      "loss": 0.1241,
      "step": 515
    },
    {
      "epoch": 1.1909982688978649,
      "grad_norm": 0.0644492357969284,
      "learning_rate": 0.0001526851851851852,
      "loss": 0.1305,
      "step": 516
    },
    {
      "epoch": 1.1933064050778996,
      "grad_norm": 0.0621226467192173,
      "learning_rate": 0.0001525925925925926,
      "loss": 0.1286,
      "step": 517
    },
    {
      "epoch": 1.1956145412579342,
      "grad_norm": 0.06589128822088242,
      "learning_rate": 0.0001525,
      "loss": 0.1226,
      "step": 518
    },
    {
      "epoch": 1.1979226774379688,
      "grad_norm": 0.07245851308107376,
      "learning_rate": 0.00015240740740740742,
      "loss": 0.1236,
      "step": 519
    },
    {
      "epoch": 1.2002308136180035,
      "grad_norm": 0.06897026300430298,
      "learning_rate": 0.00015231481481481481,
      "loss": 0.1221,
      "step": 520
    },
    {
      "epoch": 1.202538949798038,
      "grad_norm": 0.056689418852329254,
      "learning_rate": 0.0001522222222222222,
      "loss": 0.1322,
      "step": 521
    },
    {
      "epoch": 1.2048470859780727,
      "grad_norm": 0.06321047991514206,
      "learning_rate": 0.00015212962962962964,
      "loss": 0.1243,
      "step": 522
    },
    {
      "epoch": 1.2071552221581072,
      "grad_norm": 0.0733972117304802,
      "learning_rate": 0.00015203703703703703,
      "loss": 0.1206,
      "step": 523
    },
    {
      "epoch": 1.209463358338142,
      "grad_norm": 0.06292839348316193,
      "learning_rate": 0.00015194444444444446,
      "loss": 0.1231,
      "step": 524
    },
    {
      "epoch": 1.2117714945181766,
      "grad_norm": 0.06364362686872482,
      "learning_rate": 0.00015185185185185185,
      "loss": 0.1257,
      "step": 525
    },
    {
      "epoch": 1.2140796306982111,
      "grad_norm": 0.0657050833106041,
      "learning_rate": 0.00015175925925925928,
      "loss": 0.1236,
      "step": 526
    },
    {
      "epoch": 1.216387766878246,
      "grad_norm": 0.06318560242652893,
      "learning_rate": 0.00015166666666666668,
      "loss": 0.1287,
      "step": 527
    },
    {
      "epoch": 1.2186959030582805,
      "grad_norm": 0.056519605219364166,
      "learning_rate": 0.0001515740740740741,
      "loss": 0.1264,
      "step": 528
    },
    {
      "epoch": 1.221004039238315,
      "grad_norm": 0.06789349764585495,
      "learning_rate": 0.0001514814814814815,
      "loss": 0.1247,
      "step": 529
    },
    {
      "epoch": 1.2233121754183496,
      "grad_norm": 0.07219099998474121,
      "learning_rate": 0.0001513888888888889,
      "loss": 0.1278,
      "step": 530
    },
    {
      "epoch": 1.2256203115983844,
      "grad_norm": 0.057280562818050385,
      "learning_rate": 0.00015129629629629632,
      "loss": 0.1286,
      "step": 531
    },
    {
      "epoch": 1.227928447778419,
      "grad_norm": 0.07014620304107666,
      "learning_rate": 0.00015120370370370372,
      "loss": 0.1237,
      "step": 532
    },
    {
      "epoch": 1.2302365839584535,
      "grad_norm": 0.05971000716090202,
      "learning_rate": 0.0001511111111111111,
      "loss": 0.1218,
      "step": 533
    },
    {
      "epoch": 1.2325447201384883,
      "grad_norm": 0.07407764345407486,
      "learning_rate": 0.0001510185185185185,
      "loss": 0.1229,
      "step": 534
    },
    {
      "epoch": 1.2348528563185228,
      "grad_norm": 0.066289022564888,
      "learning_rate": 0.00015092592592592593,
      "loss": 0.1233,
      "step": 535
    },
    {
      "epoch": 1.2371609924985574,
      "grad_norm": 0.056438345462083817,
      "learning_rate": 0.00015083333333333333,
      "loss": 0.1284,
      "step": 536
    },
    {
      "epoch": 1.239469128678592,
      "grad_norm": 0.05904312804341316,
      "learning_rate": 0.00015074074074074073,
      "loss": 0.1311,
      "step": 537
    },
    {
      "epoch": 1.2417772648586267,
      "grad_norm": 0.056505776941776276,
      "learning_rate": 0.00015064814814814815,
      "loss": 0.1253,
      "step": 538
    },
    {
      "epoch": 1.2440854010386613,
      "grad_norm": 0.06539508700370789,
      "learning_rate": 0.00015055555555555555,
      "loss": 0.1215,
      "step": 539
    },
    {
      "epoch": 1.2463935372186958,
      "grad_norm": 0.05987078323960304,
      "learning_rate": 0.00015046296296296297,
      "loss": 0.1243,
      "step": 540
    },
    {
      "epoch": 1.2487016733987306,
      "grad_norm": 0.06465402990579605,
      "learning_rate": 0.00015037037037037037,
      "loss": 0.126,
      "step": 541
    },
    {
      "epoch": 1.2510098095787652,
      "grad_norm": 0.06128043308854103,
      "learning_rate": 0.0001502777777777778,
      "loss": 0.1299,
      "step": 542
    },
    {
      "epoch": 1.2533179457587997,
      "grad_norm": 0.06390910595655441,
      "learning_rate": 0.0001501851851851852,
      "loss": 0.1283,
      "step": 543
    },
    {
      "epoch": 1.2556260819388343,
      "grad_norm": 0.06390050798654556,
      "learning_rate": 0.00015009259259259262,
      "loss": 0.1259,
      "step": 544
    },
    {
      "epoch": 1.257934218118869,
      "grad_norm": 0.07520755380392075,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.1248,
      "step": 545
    },
    {
      "epoch": 1.2602423542989036,
      "grad_norm": 0.06576889753341675,
      "learning_rate": 0.0001499074074074074,
      "loss": 0.1244,
      "step": 546
    },
    {
      "epoch": 1.2625504904789382,
      "grad_norm": 0.06424735486507416,
      "learning_rate": 0.00014981481481481484,
      "loss": 0.1312,
      "step": 547
    },
    {
      "epoch": 1.264858626658973,
      "grad_norm": 0.06004881113767624,
      "learning_rate": 0.00014972222222222223,
      "loss": 0.1204,
      "step": 548
    },
    {
      "epoch": 1.2671667628390075,
      "grad_norm": 0.07299204170703888,
      "learning_rate": 0.00014962962962962963,
      "loss": 0.1255,
      "step": 549
    },
    {
      "epoch": 1.269474899019042,
      "grad_norm": 0.08276915550231934,
      "learning_rate": 0.00014953703703703705,
      "loss": 0.1258,
      "step": 550
    },
    {
      "epoch": 1.2717830351990767,
      "grad_norm": 0.06786096096038818,
      "learning_rate": 0.00014944444444444445,
      "loss": 0.127,
      "step": 551
    },
    {
      "epoch": 1.2740911713791114,
      "grad_norm": 0.06584415584802628,
      "learning_rate": 0.00014935185185185185,
      "loss": 0.1265,
      "step": 552
    },
    {
      "epoch": 1.276399307559146,
      "grad_norm": 0.07996676862239838,
      "learning_rate": 0.00014925925925925927,
      "loss": 0.1193,
      "step": 553
    },
    {
      "epoch": 1.2787074437391805,
      "grad_norm": 0.061803173273801804,
      "learning_rate": 0.00014916666666666667,
      "loss": 0.1221,
      "step": 554
    },
    {
      "epoch": 1.2810155799192153,
      "grad_norm": 0.07277137041091919,
      "learning_rate": 0.00014907407407407407,
      "loss": 0.1301,
      "step": 555
    },
    {
      "epoch": 1.28332371609925,
      "grad_norm": 0.05921773612499237,
      "learning_rate": 0.0001489814814814815,
      "loss": 0.1274,
      "step": 556
    },
    {
      "epoch": 1.2856318522792844,
      "grad_norm": 0.06268466264009476,
      "learning_rate": 0.0001488888888888889,
      "loss": 0.1255,
      "step": 557
    },
    {
      "epoch": 1.287939988459319,
      "grad_norm": 0.0675240159034729,
      "learning_rate": 0.0001487962962962963,
      "loss": 0.1255,
      "step": 558
    },
    {
      "epoch": 1.2902481246393538,
      "grad_norm": 0.06421612948179245,
      "learning_rate": 0.0001487037037037037,
      "loss": 0.1197,
      "step": 559
    },
    {
      "epoch": 1.2925562608193883,
      "grad_norm": 0.06285012513399124,
      "learning_rate": 0.00014861111111111113,
      "loss": 0.1248,
      "step": 560
    },
    {
      "epoch": 1.294864396999423,
      "grad_norm": 0.06570520251989365,
      "learning_rate": 0.00014851851851851853,
      "loss": 0.1231,
      "step": 561
    },
    {
      "epoch": 1.2971725331794577,
      "grad_norm": 0.06306800991296768,
      "learning_rate": 0.00014842592592592593,
      "loss": 0.1306,
      "step": 562
    },
    {
      "epoch": 1.2994806693594922,
      "grad_norm": 0.06336595863103867,
      "learning_rate": 0.00014833333333333335,
      "loss": 0.1205,
      "step": 563
    },
    {
      "epoch": 1.3017888055395268,
      "grad_norm": 0.059486471116542816,
      "learning_rate": 0.00014824074074074075,
      "loss": 0.1334,
      "step": 564
    },
    {
      "epoch": 1.3040969417195614,
      "grad_norm": 0.058824989944696426,
      "learning_rate": 0.00014814814814814815,
      "loss": 0.1174,
      "step": 565
    },
    {
      "epoch": 1.3064050778995961,
      "grad_norm": 0.06948786973953247,
      "learning_rate": 0.00014805555555555557,
      "loss": 0.1228,
      "step": 566
    },
    {
      "epoch": 1.3087132140796307,
      "grad_norm": 0.06240867078304291,
      "learning_rate": 0.00014796296296296297,
      "loss": 0.1316,
      "step": 567
    },
    {
      "epoch": 1.3110213502596653,
      "grad_norm": 0.06925071775913239,
      "learning_rate": 0.00014787037037037036,
      "loss": 0.1237,
      "step": 568
    },
    {
      "epoch": 1.3133294864397,
      "grad_norm": 0.06954062730073929,
      "learning_rate": 0.0001477777777777778,
      "loss": 0.1178,
      "step": 569
    },
    {
      "epoch": 1.3156376226197346,
      "grad_norm": 0.06070791557431221,
      "learning_rate": 0.00014768518518518519,
      "loss": 0.1254,
      "step": 570
    },
    {
      "epoch": 1.3179457587997692,
      "grad_norm": 0.054706308990716934,
      "learning_rate": 0.00014759259259259258,
      "loss": 0.1267,
      "step": 571
    },
    {
      "epoch": 1.3202538949798037,
      "grad_norm": 0.06145545840263367,
      "learning_rate": 0.0001475,
      "loss": 0.1297,
      "step": 572
    },
    {
      "epoch": 1.3225620311598385,
      "grad_norm": 0.05781954154372215,
      "learning_rate": 0.0001474074074074074,
      "loss": 0.1312,
      "step": 573
    },
    {
      "epoch": 1.324870167339873,
      "grad_norm": 0.06471090018749237,
      "learning_rate": 0.00014731481481481483,
      "loss": 0.1244,
      "step": 574
    },
    {
      "epoch": 1.3271783035199076,
      "grad_norm": 0.056357093155384064,
      "learning_rate": 0.00014722222222222223,
      "loss": 0.1243,
      "step": 575
    },
    {
      "epoch": 1.3294864396999424,
      "grad_norm": 0.06158432364463806,
      "learning_rate": 0.00014712962962962965,
      "loss": 0.1201,
      "step": 576
    },
    {
      "epoch": 1.331794575879977,
      "grad_norm": 0.05342824012041092,
      "learning_rate": 0.00014703703703703705,
      "loss": 0.1208,
      "step": 577
    },
    {
      "epoch": 1.3341027120600115,
      "grad_norm": 0.058397430926561356,
      "learning_rate": 0.00014694444444444447,
      "loss": 0.122,
      "step": 578
    },
    {
      "epoch": 1.336410848240046,
      "grad_norm": 0.05891604349017143,
      "learning_rate": 0.00014685185185185187,
      "loss": 0.1222,
      "step": 579
    },
    {
      "epoch": 1.3387189844200809,
      "grad_norm": 0.053171806037425995,
      "learning_rate": 0.00014675925925925927,
      "loss": 0.1271,
      "step": 580
    },
    {
      "epoch": 1.3410271206001154,
      "grad_norm": 0.055380042642354965,
      "learning_rate": 0.00014666666666666666,
      "loss": 0.1251,
      "step": 581
    },
    {
      "epoch": 1.34333525678015,
      "grad_norm": 0.06095922738313675,
      "learning_rate": 0.0001465740740740741,
      "loss": 0.1212,
      "step": 582
    },
    {
      "epoch": 1.3456433929601848,
      "grad_norm": 0.062225889414548874,
      "learning_rate": 0.00014648148148148148,
      "loss": 0.1238,
      "step": 583
    },
    {
      "epoch": 1.3479515291402193,
      "grad_norm": 0.0588654950261116,
      "learning_rate": 0.00014638888888888888,
      "loss": 0.1138,
      "step": 584
    },
    {
      "epoch": 1.3502596653202539,
      "grad_norm": 0.06028709560632706,
      "learning_rate": 0.0001462962962962963,
      "loss": 0.127,
      "step": 585
    },
    {
      "epoch": 1.3525678015002884,
      "grad_norm": 0.056191589683294296,
      "learning_rate": 0.0001462037037037037,
      "loss": 0.1267,
      "step": 586
    },
    {
      "epoch": 1.3548759376803232,
      "grad_norm": 0.055875811725854874,
      "learning_rate": 0.0001461111111111111,
      "loss": 0.1227,
      "step": 587
    },
    {
      "epoch": 1.3571840738603578,
      "grad_norm": 0.053982216864824295,
      "learning_rate": 0.00014601851851851852,
      "loss": 0.1198,
      "step": 588
    },
    {
      "epoch": 1.3594922100403923,
      "grad_norm": 0.05132046341896057,
      "learning_rate": 0.00014592592592592592,
      "loss": 0.1215,
      "step": 589
    },
    {
      "epoch": 1.3618003462204271,
      "grad_norm": 0.0577196329832077,
      "learning_rate": 0.00014583333333333335,
      "loss": 0.1176,
      "step": 590
    },
    {
      "epoch": 1.3641084824004617,
      "grad_norm": 0.058418143540620804,
      "learning_rate": 0.00014574074074074074,
      "loss": 0.1236,
      "step": 591
    },
    {
      "epoch": 1.3664166185804962,
      "grad_norm": 0.059597745537757874,
      "learning_rate": 0.00014564814814814817,
      "loss": 0.1206,
      "step": 592
    },
    {
      "epoch": 1.3687247547605308,
      "grad_norm": 0.058117616921663284,
      "learning_rate": 0.00014555555555555556,
      "loss": 0.1186,
      "step": 593
    },
    {
      "epoch": 1.3710328909405656,
      "grad_norm": 0.06531903892755508,
      "learning_rate": 0.000145462962962963,
      "loss": 0.1197,
      "step": 594
    },
    {
      "epoch": 1.3733410271206001,
      "grad_norm": 0.055493250489234924,
      "learning_rate": 0.00014537037037037039,
      "loss": 0.127,
      "step": 595
    },
    {
      "epoch": 1.3756491633006347,
      "grad_norm": 0.05752253904938698,
      "learning_rate": 0.00014527777777777778,
      "loss": 0.1256,
      "step": 596
    },
    {
      "epoch": 1.3779572994806695,
      "grad_norm": 0.060115355998277664,
      "learning_rate": 0.0001451851851851852,
      "loss": 0.1281,
      "step": 597
    },
    {
      "epoch": 1.380265435660704,
      "grad_norm": 0.058527398854494095,
      "learning_rate": 0.0001450925925925926,
      "loss": 0.1235,
      "step": 598
    },
    {
      "epoch": 1.3825735718407386,
      "grad_norm": 0.057766392827034,
      "learning_rate": 0.000145,
      "loss": 0.1252,
      "step": 599
    },
    {
      "epoch": 1.3848817080207732,
      "grad_norm": 0.054807595908641815,
      "learning_rate": 0.00014490740740740743,
      "loss": 0.1248,
      "step": 600
    },
    {
      "epoch": 1.3871898442008077,
      "grad_norm": 0.05749710649251938,
      "learning_rate": 0.00014481481481481482,
      "loss": 0.1233,
      "step": 601
    },
    {
      "epoch": 1.3894979803808425,
      "grad_norm": 0.05898619443178177,
      "learning_rate": 0.00014472222222222222,
      "loss": 0.1251,
      "step": 602
    },
    {
      "epoch": 1.391806116560877,
      "grad_norm": 0.05957065150141716,
      "learning_rate": 0.00014462962962962962,
      "loss": 0.1278,
      "step": 603
    },
    {
      "epoch": 1.3941142527409118,
      "grad_norm": 0.06282345950603485,
      "learning_rate": 0.00014453703703703704,
      "loss": 0.1141,
      "step": 604
    },
    {
      "epoch": 1.3964223889209464,
      "grad_norm": 0.05584244057536125,
      "learning_rate": 0.00014444444444444444,
      "loss": 0.1208,
      "step": 605
    },
    {
      "epoch": 1.398730525100981,
      "grad_norm": 0.06954366713762283,
      "learning_rate": 0.00014435185185185186,
      "loss": 0.1289,
      "step": 606
    },
    {
      "epoch": 1.4010386612810155,
      "grad_norm": 0.06893005222082138,
      "learning_rate": 0.00014425925925925926,
      "loss": 0.1293,
      "step": 607
    },
    {
      "epoch": 1.40334679746105,
      "grad_norm": 0.06348910182714462,
      "learning_rate": 0.00014416666666666668,
      "loss": 0.1202,
      "step": 608
    },
    {
      "epoch": 1.4056549336410848,
      "grad_norm": 0.07240862399339676,
      "learning_rate": 0.00014407407407407408,
      "loss": 0.1254,
      "step": 609
    },
    {
      "epoch": 1.4079630698211194,
      "grad_norm": 0.07549717277288437,
      "learning_rate": 0.0001439814814814815,
      "loss": 0.1198,
      "step": 610
    },
    {
      "epoch": 1.4102712060011542,
      "grad_norm": 0.07324759662151337,
      "learning_rate": 0.0001438888888888889,
      "loss": 0.1173,
      "step": 611
    },
    {
      "epoch": 1.4125793421811887,
      "grad_norm": 0.06188857927918434,
      "learning_rate": 0.0001437962962962963,
      "loss": 0.1228,
      "step": 612
    },
    {
      "epoch": 1.4148874783612233,
      "grad_norm": 0.06094217300415039,
      "learning_rate": 0.00014370370370370372,
      "loss": 0.1314,
      "step": 613
    },
    {
      "epoch": 1.4171956145412579,
      "grad_norm": 0.06285514682531357,
      "learning_rate": 0.00014361111111111112,
      "loss": 0.1175,
      "step": 614
    },
    {
      "epoch": 1.4195037507212924,
      "grad_norm": 0.06975087523460388,
      "learning_rate": 0.00014351851851851852,
      "loss": 0.1246,
      "step": 615
    },
    {
      "epoch": 1.4218118869013272,
      "grad_norm": 0.06220196187496185,
      "learning_rate": 0.00014342592592592594,
      "loss": 0.1163,
      "step": 616
    },
    {
      "epoch": 1.4241200230813618,
      "grad_norm": 0.07040389627218246,
      "learning_rate": 0.00014333333333333334,
      "loss": 0.1179,
      "step": 617
    },
    {
      "epoch": 1.4264281592613965,
      "grad_norm": 0.07850547134876251,
      "learning_rate": 0.00014324074074074074,
      "loss": 0.1225,
      "step": 618
    },
    {
      "epoch": 1.428736295441431,
      "grad_norm": 0.059273723512887955,
      "learning_rate": 0.00014314814814814816,
      "loss": 0.123,
      "step": 619
    },
    {
      "epoch": 1.4310444316214657,
      "grad_norm": 0.07133091986179352,
      "learning_rate": 0.00014305555555555556,
      "loss": 0.1247,
      "step": 620
    },
    {
      "epoch": 1.4333525678015002,
      "grad_norm": 0.07111719250679016,
      "learning_rate": 0.00014296296296296295,
      "loss": 0.1229,
      "step": 621
    },
    {
      "epoch": 1.4356607039815348,
      "grad_norm": 0.055620480328798294,
      "learning_rate": 0.00014287037037037038,
      "loss": 0.1238,
      "step": 622
    },
    {
      "epoch": 1.4379688401615696,
      "grad_norm": 0.07077746838331223,
      "learning_rate": 0.00014277777777777778,
      "loss": 0.1255,
      "step": 623
    },
    {
      "epoch": 1.4402769763416041,
      "grad_norm": 0.10745343565940857,
      "learning_rate": 0.0001426851851851852,
      "loss": 0.1156,
      "step": 624
    },
    {
      "epoch": 1.442585112521639,
      "grad_norm": 0.11927138268947601,
      "learning_rate": 0.0001425925925925926,
      "loss": 0.1264,
      "step": 625
    },
    {
      "epoch": 1.4448932487016735,
      "grad_norm": 0.07760899513959885,
      "learning_rate": 0.00014250000000000002,
      "loss": 0.1268,
      "step": 626
    },
    {
      "epoch": 1.447201384881708,
      "grad_norm": 0.07356236129999161,
      "learning_rate": 0.00014240740740740742,
      "loss": 0.1215,
      "step": 627
    },
    {
      "epoch": 1.4495095210617426,
      "grad_norm": 0.05896393582224846,
      "learning_rate": 0.00014231481481481482,
      "loss": 0.123,
      "step": 628
    },
    {
      "epoch": 1.4518176572417771,
      "grad_norm": 0.07980777323246002,
      "learning_rate": 0.00014222222222222224,
      "loss": 0.1223,
      "step": 629
    },
    {
      "epoch": 1.454125793421812,
      "grad_norm": 0.07899472117424011,
      "learning_rate": 0.00014212962962962964,
      "loss": 0.1231,
      "step": 630
    },
    {
      "epoch": 1.4564339296018465,
      "grad_norm": 0.06692565977573395,
      "learning_rate": 0.00014203703703703703,
      "loss": 0.1152,
      "step": 631
    },
    {
      "epoch": 1.4587420657818813,
      "grad_norm": 0.06786306947469711,
      "learning_rate": 0.00014194444444444446,
      "loss": 0.1271,
      "step": 632
    },
    {
      "epoch": 1.4610502019619158,
      "grad_norm": 0.06429289281368256,
      "learning_rate": 0.00014185185185185186,
      "loss": 0.1288,
      "step": 633
    },
    {
      "epoch": 1.4633583381419504,
      "grad_norm": 0.0748908519744873,
      "learning_rate": 0.00014175925925925925,
      "loss": 0.1225,
      "step": 634
    },
    {
      "epoch": 1.465666474321985,
      "grad_norm": 0.05763575807213783,
      "learning_rate": 0.00014166666666666668,
      "loss": 0.1211,
      "step": 635
    },
    {
      "epoch": 1.4679746105020195,
      "grad_norm": 0.08056017011404037,
      "learning_rate": 0.00014157407407407407,
      "loss": 0.1265,
      "step": 636
    },
    {
      "epoch": 1.4702827466820543,
      "grad_norm": 0.06612623482942581,
      "learning_rate": 0.00014148148148148147,
      "loss": 0.1272,
      "step": 637
    },
    {
      "epoch": 1.4725908828620888,
      "grad_norm": 0.06559476256370544,
      "learning_rate": 0.0001413888888888889,
      "loss": 0.123,
      "step": 638
    },
    {
      "epoch": 1.4748990190421236,
      "grad_norm": 0.05643089860677719,
      "learning_rate": 0.0001412962962962963,
      "loss": 0.1235,
      "step": 639
    },
    {
      "epoch": 1.4772071552221582,
      "grad_norm": 0.07214000076055527,
      "learning_rate": 0.00014120370370370372,
      "loss": 0.1191,
      "step": 640
    },
    {
      "epoch": 1.4795152914021927,
      "grad_norm": 0.05962374433875084,
      "learning_rate": 0.00014111111111111111,
      "loss": 0.1301,
      "step": 641
    },
    {
      "epoch": 1.4818234275822273,
      "grad_norm": 0.0669381320476532,
      "learning_rate": 0.00014101851851851854,
      "loss": 0.1299,
      "step": 642
    },
    {
      "epoch": 1.4841315637622619,
      "grad_norm": 0.06181809678673744,
      "learning_rate": 0.00014092592592592594,
      "loss": 0.1194,
      "step": 643
    },
    {
      "epoch": 1.4864396999422966,
      "grad_norm": 0.05997949466109276,
      "learning_rate": 0.00014083333333333336,
      "loss": 0.1206,
      "step": 644
    },
    {
      "epoch": 1.4887478361223312,
      "grad_norm": 0.0607033334672451,
      "learning_rate": 0.00014074074074074076,
      "loss": 0.1225,
      "step": 645
    },
    {
      "epoch": 1.491055972302366,
      "grad_norm": 0.057610418647527695,
      "learning_rate": 0.00014064814814814815,
      "loss": 0.1182,
      "step": 646
    },
    {
      "epoch": 1.4933641084824005,
      "grad_norm": 0.05941525474190712,
      "learning_rate": 0.00014055555555555555,
      "loss": 0.1222,
      "step": 647
    },
    {
      "epoch": 1.495672244662435,
      "grad_norm": 0.05772700533270836,
      "learning_rate": 0.00014046296296296298,
      "loss": 0.1234,
      "step": 648
    },
    {
      "epoch": 1.4979803808424696,
      "grad_norm": 0.06103130802512169,
      "learning_rate": 0.00014037037037037037,
      "loss": 0.1202,
      "step": 649
    },
    {
      "epoch": 1.5002885170225042,
      "grad_norm": 0.056772395968437195,
      "learning_rate": 0.00014027777777777777,
      "loss": 0.1191,
      "step": 650
    },
    {
      "epoch": 1.502596653202539,
      "grad_norm": 0.05476753041148186,
      "learning_rate": 0.0001401851851851852,
      "loss": 0.1243,
      "step": 651
    },
    {
      "epoch": 1.5049047893825735,
      "grad_norm": 0.05563043802976608,
      "learning_rate": 0.0001400925925925926,
      "loss": 0.128,
      "step": 652
    },
    {
      "epoch": 1.5072129255626083,
      "grad_norm": 0.058114081621170044,
      "learning_rate": 0.00014,
      "loss": 0.1186,
      "step": 653
    },
    {
      "epoch": 1.5095210617426429,
      "grad_norm": 0.05864199250936508,
      "learning_rate": 0.0001399074074074074,
      "loss": 0.1196,
      "step": 654
    },
    {
      "epoch": 1.5118291979226774,
      "grad_norm": 0.05820152536034584,
      "learning_rate": 0.0001398148148148148,
      "loss": 0.1232,
      "step": 655
    },
    {
      "epoch": 1.514137334102712,
      "grad_norm": 0.05895432084798813,
      "learning_rate": 0.00013972222222222223,
      "loss": 0.1195,
      "step": 656
    },
    {
      "epoch": 1.5164454702827466,
      "grad_norm": 0.05410686507821083,
      "learning_rate": 0.00013962962962962963,
      "loss": 0.1195,
      "step": 657
    },
    {
      "epoch": 1.5187536064627813,
      "grad_norm": 0.0652122050523758,
      "learning_rate": 0.00013953703703703705,
      "loss": 0.1189,
      "step": 658
    },
    {
      "epoch": 1.521061742642816,
      "grad_norm": 0.05396094173192978,
      "learning_rate": 0.00013944444444444445,
      "loss": 0.1204,
      "step": 659
    },
    {
      "epoch": 1.5233698788228507,
      "grad_norm": 0.05974944308400154,
      "learning_rate": 0.00013935185185185188,
      "loss": 0.1232,
      "step": 660
    },
    {
      "epoch": 1.5256780150028852,
      "grad_norm": 0.063397616147995,
      "learning_rate": 0.00013925925925925927,
      "loss": 0.1207,
      "step": 661
    },
    {
      "epoch": 1.5279861511829198,
      "grad_norm": 0.05819641798734665,
      "learning_rate": 0.00013916666666666667,
      "loss": 0.1275,
      "step": 662
    },
    {
      "epoch": 1.5302942873629544,
      "grad_norm": 0.05931056663393974,
      "learning_rate": 0.0001390740740740741,
      "loss": 0.1207,
      "step": 663
    },
    {
      "epoch": 1.532602423542989,
      "grad_norm": 0.06709395349025726,
      "learning_rate": 0.0001389814814814815,
      "loss": 0.1159,
      "step": 664
    },
    {
      "epoch": 1.5349105597230237,
      "grad_norm": 0.055006664246320724,
      "learning_rate": 0.0001388888888888889,
      "loss": 0.1183,
      "step": 665
    },
    {
      "epoch": 1.5372186959030583,
      "grad_norm": 0.06736339628696442,
      "learning_rate": 0.0001387962962962963,
      "loss": 0.118,
      "step": 666
    },
    {
      "epoch": 1.539526832083093,
      "grad_norm": 0.0584082156419754,
      "learning_rate": 0.0001387037037037037,
      "loss": 0.1145,
      "step": 667
    },
    {
      "epoch": 1.5418349682631276,
      "grad_norm": 0.0657401829957962,
      "learning_rate": 0.0001386111111111111,
      "loss": 0.1219,
      "step": 668
    },
    {
      "epoch": 1.5441431044431622,
      "grad_norm": 0.06403159350156784,
      "learning_rate": 0.0001385185185185185,
      "loss": 0.1152,
      "step": 669
    },
    {
      "epoch": 1.5464512406231967,
      "grad_norm": 0.06141997501254082,
      "learning_rate": 0.00013842592592592593,
      "loss": 0.1185,
      "step": 670
    },
    {
      "epoch": 1.5487593768032313,
      "grad_norm": 0.05843626707792282,
      "learning_rate": 0.00013833333333333333,
      "loss": 0.1289,
      "step": 671
    },
    {
      "epoch": 1.551067512983266,
      "grad_norm": 0.061746370047330856,
      "learning_rate": 0.00013824074074074075,
      "loss": 0.1187,
      "step": 672
    },
    {
      "epoch": 1.5533756491633006,
      "grad_norm": 0.0665271058678627,
      "learning_rate": 0.00013814814814814815,
      "loss": 0.125,
      "step": 673
    },
    {
      "epoch": 1.5556837853433354,
      "grad_norm": 0.05844796076416969,
      "learning_rate": 0.00013805555555555557,
      "loss": 0.1222,
      "step": 674
    },
    {
      "epoch": 1.55799192152337,
      "grad_norm": 0.06712702661752701,
      "learning_rate": 0.00013796296296296297,
      "loss": 0.1218,
      "step": 675
    },
    {
      "epoch": 1.5603000577034045,
      "grad_norm": 0.07478312402963638,
      "learning_rate": 0.0001378703703703704,
      "loss": 0.1218,
      "step": 676
    },
    {
      "epoch": 1.562608193883439,
      "grad_norm": 0.05990191549062729,
      "learning_rate": 0.0001377777777777778,
      "loss": 0.1195,
      "step": 677
    },
    {
      "epoch": 1.5649163300634736,
      "grad_norm": 0.05570739135146141,
      "learning_rate": 0.0001376851851851852,
      "loss": 0.1243,
      "step": 678
    },
    {
      "epoch": 1.5672244662435084,
      "grad_norm": 0.0645642802119255,
      "learning_rate": 0.0001375925925925926,
      "loss": 0.1219,
      "step": 679
    },
    {
      "epoch": 1.569532602423543,
      "grad_norm": 0.05905265733599663,
      "learning_rate": 0.0001375,
      "loss": 0.1197,
      "step": 680
    },
    {
      "epoch": 1.5718407386035778,
      "grad_norm": 0.059475161135196686,
      "learning_rate": 0.0001374074074074074,
      "loss": 0.1248,
      "step": 681
    },
    {
      "epoch": 1.5741488747836123,
      "grad_norm": 0.06028126925230026,
      "learning_rate": 0.00013731481481481483,
      "loss": 0.1199,
      "step": 682
    },
    {
      "epoch": 1.5764570109636469,
      "grad_norm": 0.06153789162635803,
      "learning_rate": 0.00013722222222222223,
      "loss": 0.1189,
      "step": 683
    },
    {
      "epoch": 1.5787651471436814,
      "grad_norm": 0.05840005353093147,
      "learning_rate": 0.00013712962962962962,
      "loss": 0.1205,
      "step": 684
    },
    {
      "epoch": 1.581073283323716,
      "grad_norm": 0.07097157090902328,
      "learning_rate": 0.00013703703703703705,
      "loss": 0.1205,
      "step": 685
    },
    {
      "epoch": 1.5833814195037508,
      "grad_norm": 0.058966848999261856,
      "learning_rate": 0.00013694444444444445,
      "loss": 0.1178,
      "step": 686
    },
    {
      "epoch": 1.5856895556837853,
      "grad_norm": 0.06321373581886292,
      "learning_rate": 0.00013685185185185184,
      "loss": 0.1241,
      "step": 687
    },
    {
      "epoch": 1.5879976918638201,
      "grad_norm": 0.06387839466333389,
      "learning_rate": 0.00013675925925925927,
      "loss": 0.1311,
      "step": 688
    },
    {
      "epoch": 1.5903058280438547,
      "grad_norm": 0.06395796686410904,
      "learning_rate": 0.00013666666666666666,
      "loss": 0.116,
      "step": 689
    },
    {
      "epoch": 1.5926139642238892,
      "grad_norm": 0.07225635647773743,
      "learning_rate": 0.0001365740740740741,
      "loss": 0.1164,
      "step": 690
    },
    {
      "epoch": 1.5949221004039238,
      "grad_norm": 0.06396475434303284,
      "learning_rate": 0.0001364814814814815,
      "loss": 0.113,
      "step": 691
    },
    {
      "epoch": 1.5972302365839584,
      "grad_norm": 0.06606000661849976,
      "learning_rate": 0.0001363888888888889,
      "loss": 0.1165,
      "step": 692
    },
    {
      "epoch": 1.599538372763993,
      "grad_norm": 0.058903004974126816,
      "learning_rate": 0.0001362962962962963,
      "loss": 0.1192,
      "step": 693
    },
    {
      "epoch": 1.6018465089440277,
      "grad_norm": 0.07133742421865463,
      "learning_rate": 0.0001362037037037037,
      "loss": 0.1104,
      "step": 694
    },
    {
      "epoch": 1.6041546451240625,
      "grad_norm": 0.07315586507320404,
      "learning_rate": 0.00013611111111111113,
      "loss": 0.1263,
      "step": 695
    },
    {
      "epoch": 1.606462781304097,
      "grad_norm": 0.06587207317352295,
      "learning_rate": 0.00013601851851851852,
      "loss": 0.1178,
      "step": 696
    },
    {
      "epoch": 1.6087709174841316,
      "grad_norm": 0.07077711820602417,
      "learning_rate": 0.00013592592592592592,
      "loss": 0.1187,
      "step": 697
    },
    {
      "epoch": 1.6110790536641661,
      "grad_norm": 0.06133824586868286,
      "learning_rate": 0.00013583333333333335,
      "loss": 0.1241,
      "step": 698
    },
    {
      "epoch": 1.6133871898442007,
      "grad_norm": 0.06736895442008972,
      "learning_rate": 0.00013574074074074074,
      "loss": 0.124,
      "step": 699
    },
    {
      "epoch": 1.6156953260242353,
      "grad_norm": 0.06552513688802719,
      "learning_rate": 0.00013564814814814814,
      "loss": 0.1233,
      "step": 700
    },
    {
      "epoch": 1.61800346220427,
      "grad_norm": 0.058084551244974136,
      "learning_rate": 0.00013555555555555556,
      "loss": 0.1192,
      "step": 701
    },
    {
      "epoch": 1.6203115983843048,
      "grad_norm": 0.08122546225786209,
      "learning_rate": 0.00013546296296296296,
      "loss": 0.1156,
      "step": 702
    },
    {
      "epoch": 1.6226197345643394,
      "grad_norm": 0.06033937260508537,
      "learning_rate": 0.00013537037037037036,
      "loss": 0.1217,
      "step": 703
    },
    {
      "epoch": 1.624927870744374,
      "grad_norm": 0.07198290526866913,
      "learning_rate": 0.00013527777777777778,
      "loss": 0.1177,
      "step": 704
    },
    {
      "epoch": 1.6272360069244085,
      "grad_norm": 0.08551137149333954,
      "learning_rate": 0.00013518518518518518,
      "loss": 0.1198,
      "step": 705
    },
    {
      "epoch": 1.629544143104443,
      "grad_norm": 0.06226940453052521,
      "learning_rate": 0.0001350925925925926,
      "loss": 0.1223,
      "step": 706
    },
    {
      "epoch": 1.6318522792844776,
      "grad_norm": 0.06362583488225937,
      "learning_rate": 0.00013500000000000003,
      "loss": 0.1217,
      "step": 707
    },
    {
      "epoch": 1.6341604154645124,
      "grad_norm": 0.06096767634153366,
      "learning_rate": 0.00013490740740740743,
      "loss": 0.124,
      "step": 708
    },
    {
      "epoch": 1.6364685516445472,
      "grad_norm": 0.060807421803474426,
      "learning_rate": 0.00013481481481481482,
      "loss": 0.119,
      "step": 709
    },
    {
      "epoch": 1.6387766878245817,
      "grad_norm": 0.06946613639593124,
      "learning_rate": 0.00013472222222222225,
      "loss": 0.1323,
      "step": 710
    },
    {
      "epoch": 1.6410848240046163,
      "grad_norm": 0.060641735792160034,
      "learning_rate": 0.00013462962962962964,
      "loss": 0.1164,
      "step": 711
    },
    {
      "epoch": 1.6433929601846509,
      "grad_norm": 0.060197897255420685,
      "learning_rate": 0.00013453703703703704,
      "loss": 0.1218,
      "step": 712
    },
    {
      "epoch": 1.6457010963646854,
      "grad_norm": 0.05965729057788849,
      "learning_rate": 0.00013444444444444447,
      "loss": 0.1227,
      "step": 713
    },
    {
      "epoch": 1.64800923254472,
      "grad_norm": 0.06232743710279465,
      "learning_rate": 0.00013435185185185186,
      "loss": 0.1186,
      "step": 714
    },
    {
      "epoch": 1.6503173687247548,
      "grad_norm": 0.06367014348506927,
      "learning_rate": 0.00013425925925925926,
      "loss": 0.122,
      "step": 715
    },
    {
      "epoch": 1.6526255049047895,
      "grad_norm": 0.06080668792128563,
      "learning_rate": 0.00013416666666666666,
      "loss": 0.1191,
      "step": 716
    },
    {
      "epoch": 1.654933641084824,
      "grad_norm": 0.06724482029676437,
      "learning_rate": 0.00013407407407407408,
      "loss": 0.1175,
      "step": 717
    },
    {
      "epoch": 1.6572417772648587,
      "grad_norm": 0.05660582333803177,
      "learning_rate": 0.00013398148148148148,
      "loss": 0.1264,
      "step": 718
    },
    {
      "epoch": 1.6595499134448932,
      "grad_norm": 0.06478651612997055,
      "learning_rate": 0.00013388888888888888,
      "loss": 0.1262,
      "step": 719
    },
    {
      "epoch": 1.6618580496249278,
      "grad_norm": 0.05838670954108238,
      "learning_rate": 0.0001337962962962963,
      "loss": 0.121,
      "step": 720
    },
    {
      "epoch": 1.6641661858049623,
      "grad_norm": 0.06722688674926758,
      "learning_rate": 0.0001337037037037037,
      "loss": 0.1201,
      "step": 721
    },
    {
      "epoch": 1.6664743219849971,
      "grad_norm": 0.06189953535795212,
      "learning_rate": 0.00013361111111111112,
      "loss": 0.1138,
      "step": 722
    },
    {
      "epoch": 1.668782458165032,
      "grad_norm": 0.07008249312639236,
      "learning_rate": 0.00013351851851851852,
      "loss": 0.1223,
      "step": 723
    },
    {
      "epoch": 1.6710905943450665,
      "grad_norm": 0.05773422122001648,
      "learning_rate": 0.00013342592592592594,
      "loss": 0.1192,
      "step": 724
    },
    {
      "epoch": 1.673398730525101,
      "grad_norm": 0.06355356425046921,
      "learning_rate": 0.00013333333333333334,
      "loss": 0.1199,
      "step": 725
    },
    {
      "epoch": 1.6757068667051356,
      "grad_norm": 0.05944075062870979,
      "learning_rate": 0.00013324074074074076,
      "loss": 0.1177,
      "step": 726
    },
    {
      "epoch": 1.6780150028851701,
      "grad_norm": 0.05986902862787247,
      "learning_rate": 0.00013314814814814816,
      "loss": 0.1224,
      "step": 727
    },
    {
      "epoch": 1.6803231390652047,
      "grad_norm": 0.062187157571315765,
      "learning_rate": 0.00013305555555555556,
      "loss": 0.1163,
      "step": 728
    },
    {
      "epoch": 1.6826312752452395,
      "grad_norm": 0.06979247182607651,
      "learning_rate": 0.00013296296296296298,
      "loss": 0.1209,
      "step": 729
    },
    {
      "epoch": 1.684939411425274,
      "grad_norm": 0.06368821114301682,
      "learning_rate": 0.00013287037037037038,
      "loss": 0.118,
      "step": 730
    },
    {
      "epoch": 1.6872475476053088,
      "grad_norm": 0.057999834418296814,
      "learning_rate": 0.00013277777777777778,
      "loss": 0.1209,
      "step": 731
    },
    {
      "epoch": 1.6895556837853434,
      "grad_norm": 0.06733234226703644,
      "learning_rate": 0.0001326851851851852,
      "loss": 0.1209,
      "step": 732
    },
    {
      "epoch": 1.691863819965378,
      "grad_norm": 0.0630391463637352,
      "learning_rate": 0.0001325925925925926,
      "loss": 0.1174,
      "step": 733
    },
    {
      "epoch": 1.6941719561454125,
      "grad_norm": 0.06213313341140747,
      "learning_rate": 0.0001325,
      "loss": 0.1194,
      "step": 734
    },
    {
      "epoch": 1.696480092325447,
      "grad_norm": 0.061045873910188675,
      "learning_rate": 0.00013240740740740742,
      "loss": 0.124,
      "step": 735
    },
    {
      "epoch": 1.6987882285054818,
      "grad_norm": 0.05852586030960083,
      "learning_rate": 0.00013231481481481482,
      "loss": 0.1207,
      "step": 736
    },
    {
      "epoch": 1.7010963646855164,
      "grad_norm": 0.062017105519771576,
      "learning_rate": 0.00013222222222222221,
      "loss": 0.1256,
      "step": 737
    },
    {
      "epoch": 1.7034045008655512,
      "grad_norm": 0.08607915043830872,
      "learning_rate": 0.00013212962962962964,
      "loss": 0.1215,
      "step": 738
    },
    {
      "epoch": 1.7057126370455857,
      "grad_norm": 0.09593923389911652,
      "learning_rate": 0.00013203703703703704,
      "loss": 0.1185,
      "step": 739
    },
    {
      "epoch": 1.7080207732256203,
      "grad_norm": 0.08136820793151855,
      "learning_rate": 0.00013194444444444446,
      "loss": 0.1258,
      "step": 740
    },
    {
      "epoch": 1.7103289094056549,
      "grad_norm": 0.055044420063495636,
      "learning_rate": 0.00013185185185185186,
      "loss": 0.1162,
      "step": 741
    },
    {
      "epoch": 1.7126370455856894,
      "grad_norm": 0.06979688256978989,
      "learning_rate": 0.00013175925925925928,
      "loss": 0.1142,
      "step": 742
    },
    {
      "epoch": 1.7149451817657242,
      "grad_norm": 0.0640416070818901,
      "learning_rate": 0.00013166666666666668,
      "loss": 0.1243,
      "step": 743
    },
    {
      "epoch": 1.7172533179457588,
      "grad_norm": 0.06633272022008896,
      "learning_rate": 0.00013157407407407407,
      "loss": 0.1259,
      "step": 744
    },
    {
      "epoch": 1.7195614541257935,
      "grad_norm": 0.06026218831539154,
      "learning_rate": 0.0001314814814814815,
      "loss": 0.1209,
      "step": 745
    },
    {
      "epoch": 1.721869590305828,
      "grad_norm": 0.06178969144821167,
      "learning_rate": 0.0001313888888888889,
      "loss": 0.1159,
      "step": 746
    },
    {
      "epoch": 1.7241777264858626,
      "grad_norm": 0.06139739230275154,
      "learning_rate": 0.0001312962962962963,
      "loss": 0.1158,
      "step": 747
    },
    {
      "epoch": 1.7264858626658972,
      "grad_norm": 0.06332992017269135,
      "learning_rate": 0.00013120370370370372,
      "loss": 0.1227,
      "step": 748
    },
    {
      "epoch": 1.7287939988459318,
      "grad_norm": 0.065108522772789,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.1252,
      "step": 749
    },
    {
      "epoch": 1.7311021350259665,
      "grad_norm": 0.06150335446000099,
      "learning_rate": 0.0001310185185185185,
      "loss": 0.1222,
      "step": 750
    },
    {
      "epoch": 1.733410271206001,
      "grad_norm": 0.05804193392395973,
      "learning_rate": 0.00013092592592592594,
      "loss": 0.1212,
      "step": 751
    },
    {
      "epoch": 1.7357184073860359,
      "grad_norm": 0.07085681706666946,
      "learning_rate": 0.00013083333333333333,
      "loss": 0.1253,
      "step": 752
    },
    {
      "epoch": 1.7380265435660704,
      "grad_norm": 0.059913937002420425,
      "learning_rate": 0.00013074074074074073,
      "loss": 0.1193,
      "step": 753
    },
    {
      "epoch": 1.740334679746105,
      "grad_norm": 0.06515415012836456,
      "learning_rate": 0.00013064814814814815,
      "loss": 0.1197,
      "step": 754
    },
    {
      "epoch": 1.7426428159261396,
      "grad_norm": 0.05975521728396416,
      "learning_rate": 0.00013055555555555555,
      "loss": 0.1189,
      "step": 755
    },
    {
      "epoch": 1.7449509521061741,
      "grad_norm": 0.06732716411352158,
      "learning_rate": 0.00013046296296296298,
      "loss": 0.1169,
      "step": 756
    },
    {
      "epoch": 1.747259088286209,
      "grad_norm": 0.0609743557870388,
      "learning_rate": 0.0001303703703703704,
      "loss": 0.1179,
      "step": 757
    },
    {
      "epoch": 1.7495672244662435,
      "grad_norm": 0.0883028507232666,
      "learning_rate": 0.0001302777777777778,
      "loss": 0.1278,
      "step": 758
    },
    {
      "epoch": 1.7518753606462782,
      "grad_norm": 0.14098505675792694,
      "learning_rate": 0.0001301851851851852,
      "loss": 0.1184,
      "step": 759
    },
    {
      "epoch": 1.7541834968263128,
      "grad_norm": 0.144779771566391,
      "learning_rate": 0.0001300925925925926,
      "loss": 0.121,
      "step": 760
    },
    {
      "epoch": 1.7564916330063474,
      "grad_norm": 0.08910983055830002,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.123,
      "step": 761
    },
    {
      "epoch": 1.758799769186382,
      "grad_norm": 0.0910792201757431,
      "learning_rate": 0.0001299074074074074,
      "loss": 0.1183,
      "step": 762
    },
    {
      "epoch": 1.7611079053664165,
      "grad_norm": 0.10936152935028076,
      "learning_rate": 0.0001298148148148148,
      "loss": 0.1149,
      "step": 763
    },
    {
      "epoch": 1.7634160415464513,
      "grad_norm": 0.06918296962976456,
      "learning_rate": 0.00012972222222222223,
      "loss": 0.1236,
      "step": 764
    },
    {
      "epoch": 1.7657241777264858,
      "grad_norm": 0.0938878059387207,
      "learning_rate": 0.00012962962962962963,
      "loss": 0.1247,
      "step": 765
    },
    {
      "epoch": 1.7680323139065206,
      "grad_norm": 0.07086901366710663,
      "learning_rate": 0.00012953703703703703,
      "loss": 0.1129,
      "step": 766
    },
    {
      "epoch": 1.7703404500865552,
      "grad_norm": 0.07435612380504608,
      "learning_rate": 0.00012944444444444445,
      "loss": 0.1138,
      "step": 767
    },
    {
      "epoch": 1.7726485862665897,
      "grad_norm": 0.06273084133863449,
      "learning_rate": 0.00012935185185185185,
      "loss": 0.1189,
      "step": 768
    },
    {
      "epoch": 1.7749567224466243,
      "grad_norm": 0.07695453613996506,
      "learning_rate": 0.00012925925925925925,
      "loss": 0.1185,
      "step": 769
    },
    {
      "epoch": 1.7772648586266588,
      "grad_norm": 0.06543347984552383,
      "learning_rate": 0.00012916666666666667,
      "loss": 0.1174,
      "step": 770
    },
    {
      "epoch": 1.7795729948066936,
      "grad_norm": 0.05636536702513695,
      "learning_rate": 0.00012907407407407407,
      "loss": 0.118,
      "step": 771
    },
    {
      "epoch": 1.7818811309867282,
      "grad_norm": 0.06446335464715958,
      "learning_rate": 0.0001289814814814815,
      "loss": 0.127,
      "step": 772
    },
    {
      "epoch": 1.784189267166763,
      "grad_norm": 0.06106099113821983,
      "learning_rate": 0.00012888888888888892,
      "loss": 0.1163,
      "step": 773
    },
    {
      "epoch": 1.7864974033467975,
      "grad_norm": 0.06457877904176712,
      "learning_rate": 0.00012879629629629631,
      "loss": 0.1152,
      "step": 774
    },
    {
      "epoch": 1.788805539526832,
      "grad_norm": 0.07122337073087692,
      "learning_rate": 0.0001287037037037037,
      "loss": 0.1182,
      "step": 775
    },
    {
      "epoch": 1.7911136757068666,
      "grad_norm": 0.06193533167243004,
      "learning_rate": 0.00012861111111111114,
      "loss": 0.1244,
      "step": 776
    },
    {
      "epoch": 1.7934218118869012,
      "grad_norm": 0.06358951330184937,
      "learning_rate": 0.00012851851851851853,
      "loss": 0.1213,
      "step": 777
    },
    {
      "epoch": 1.795729948066936,
      "grad_norm": 0.0628001019358635,
      "learning_rate": 0.00012842592592592593,
      "loss": 0.1185,
      "step": 778
    },
    {
      "epoch": 1.7980380842469705,
      "grad_norm": 0.057250965386629105,
      "learning_rate": 0.00012833333333333335,
      "loss": 0.121,
      "step": 779
    },
    {
      "epoch": 1.8003462204270053,
      "grad_norm": 0.05908717215061188,
      "learning_rate": 0.00012824074074074075,
      "loss": 0.1239,
      "step": 780
    },
    {
      "epoch": 1.8026543566070399,
      "grad_norm": 0.05842283368110657,
      "learning_rate": 0.00012814814814814815,
      "loss": 0.1166,
      "step": 781
    },
    {
      "epoch": 1.8049624927870744,
      "grad_norm": 0.0546472929418087,
      "learning_rate": 0.00012805555555555555,
      "loss": 0.1225,
      "step": 782
    },
    {
      "epoch": 1.807270628967109,
      "grad_norm": 0.06152014806866646,
      "learning_rate": 0.00012796296296296297,
      "loss": 0.1102,
      "step": 783
    },
    {
      "epoch": 1.8095787651471436,
      "grad_norm": 0.05704990774393082,
      "learning_rate": 0.00012787037037037037,
      "loss": 0.1198,
      "step": 784
    },
    {
      "epoch": 1.8118869013271783,
      "grad_norm": 0.06007629260420799,
      "learning_rate": 0.00012777777777777776,
      "loss": 0.1215,
      "step": 785
    },
    {
      "epoch": 1.814195037507213,
      "grad_norm": 0.05661753937602043,
      "learning_rate": 0.0001276851851851852,
      "loss": 0.1186,
      "step": 786
    },
    {
      "epoch": 1.8165031736872477,
      "grad_norm": 0.06338411569595337,
      "learning_rate": 0.00012759259259259258,
      "loss": 0.1157,
      "step": 787
    },
    {
      "epoch": 1.8188113098672822,
      "grad_norm": 0.0586630180478096,
      "learning_rate": 0.0001275,
      "loss": 0.1188,
      "step": 788
    },
    {
      "epoch": 1.8211194460473168,
      "grad_norm": 0.05543966591358185,
      "learning_rate": 0.0001274074074074074,
      "loss": 0.1204,
      "step": 789
    },
    {
      "epoch": 1.8234275822273514,
      "grad_norm": 0.062260620296001434,
      "learning_rate": 0.00012731481481481483,
      "loss": 0.1168,
      "step": 790
    },
    {
      "epoch": 1.825735718407386,
      "grad_norm": 0.06541585177183151,
      "learning_rate": 0.00012722222222222223,
      "loss": 0.1243,
      "step": 791
    },
    {
      "epoch": 1.8280438545874207,
      "grad_norm": 0.05417289212346077,
      "learning_rate": 0.00012712962962962965,
      "loss": 0.1193,
      "step": 792
    },
    {
      "epoch": 1.8303519907674553,
      "grad_norm": 0.060724470764398575,
      "learning_rate": 0.00012703703703703705,
      "loss": 0.1285,
      "step": 793
    },
    {
      "epoch": 1.83266012694749,
      "grad_norm": 0.061825208365917206,
      "learning_rate": 0.00012694444444444445,
      "loss": 0.122,
      "step": 794
    },
    {
      "epoch": 1.8349682631275246,
      "grad_norm": 0.05668256804347038,
      "learning_rate": 0.00012685185185185187,
      "loss": 0.1226,
      "step": 795
    },
    {
      "epoch": 1.8372763993075591,
      "grad_norm": 0.061974555253982544,
      "learning_rate": 0.00012675925925925927,
      "loss": 0.1114,
      "step": 796
    },
    {
      "epoch": 1.8395845354875937,
      "grad_norm": 0.05719340592622757,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.1167,
      "step": 797
    },
    {
      "epoch": 1.8418926716676283,
      "grad_norm": 0.0579003244638443,
      "learning_rate": 0.0001265740740740741,
      "loss": 0.1244,
      "step": 798
    },
    {
      "epoch": 1.844200807847663,
      "grad_norm": 0.05798196420073509,
      "learning_rate": 0.00012648148148148149,
      "loss": 0.1166,
      "step": 799
    },
    {
      "epoch": 1.8465089440276976,
      "grad_norm": 0.060376331210136414,
      "learning_rate": 0.00012638888888888888,
      "loss": 0.1205,
      "step": 800
    },
    {
      "epoch": 1.8488170802077324,
      "grad_norm": 0.05562356114387512,
      "learning_rate": 0.0001262962962962963,
      "loss": 0.1244,
      "step": 801
    },
    {
      "epoch": 1.851125216387767,
      "grad_norm": 0.05726703628897667,
      "learning_rate": 0.0001262037037037037,
      "loss": 0.1241,
      "step": 802
    },
    {
      "epoch": 1.8534333525678015,
      "grad_norm": 0.055833134800195694,
      "learning_rate": 0.0001261111111111111,
      "loss": 0.1183,
      "step": 803
    },
    {
      "epoch": 1.855741488747836,
      "grad_norm": 0.05612437054514885,
      "learning_rate": 0.00012601851851851853,
      "loss": 0.1188,
      "step": 804
    },
    {
      "epoch": 1.8580496249278706,
      "grad_norm": 0.05877692252397537,
      "learning_rate": 0.00012592592592592592,
      "loss": 0.1172,
      "step": 805
    },
    {
      "epoch": 1.8603577611079054,
      "grad_norm": 0.05604454129934311,
      "learning_rate": 0.00012583333333333335,
      "loss": 0.1171,
      "step": 806
    },
    {
      "epoch": 1.86266589728794,
      "grad_norm": 0.06181754171848297,
      "learning_rate": 0.00012574074074074074,
      "loss": 0.1134,
      "step": 807
    },
    {
      "epoch": 1.8649740334679747,
      "grad_norm": 0.06769300997257233,
      "learning_rate": 0.00012564814814814817,
      "loss": 0.1181,
      "step": 808
    },
    {
      "epoch": 1.8672821696480093,
      "grad_norm": 0.0570741631090641,
      "learning_rate": 0.00012555555555555557,
      "loss": 0.1266,
      "step": 809
    },
    {
      "epoch": 1.8695903058280439,
      "grad_norm": 0.06151634454727173,
      "learning_rate": 0.00012546296296296296,
      "loss": 0.1169,
      "step": 810
    },
    {
      "epoch": 1.8718984420080784,
      "grad_norm": 0.05545099079608917,
      "learning_rate": 0.0001253703703703704,
      "loss": 0.1154,
      "step": 811
    },
    {
      "epoch": 1.874206578188113,
      "grad_norm": 0.05808451771736145,
      "learning_rate": 0.00012527777777777778,
      "loss": 0.1139,
      "step": 812
    },
    {
      "epoch": 1.8765147143681478,
      "grad_norm": 0.06142181158065796,
      "learning_rate": 0.00012518518518518518,
      "loss": 0.1168,
      "step": 813
    },
    {
      "epoch": 1.8788228505481823,
      "grad_norm": 0.07300304621458054,
      "learning_rate": 0.0001250925925925926,
      "loss": 0.1146,
      "step": 814
    },
    {
      "epoch": 1.881130986728217,
      "grad_norm": 0.05860528349876404,
      "learning_rate": 0.000125,
      "loss": 0.1242,
      "step": 815
    },
    {
      "epoch": 1.8834391229082517,
      "grad_norm": 0.06621436029672623,
      "learning_rate": 0.0001249074074074074,
      "loss": 0.1192,
      "step": 816
    },
    {
      "epoch": 1.8857472590882862,
      "grad_norm": 0.053312331438064575,
      "learning_rate": 0.00012481481481481482,
      "loss": 0.1268,
      "step": 817
    },
    {
      "epoch": 1.8880553952683208,
      "grad_norm": 0.06686840951442719,
      "learning_rate": 0.00012472222222222222,
      "loss": 0.118,
      "step": 818
    },
    {
      "epoch": 1.8903635314483553,
      "grad_norm": 0.05725228786468506,
      "learning_rate": 0.00012462962962962962,
      "loss": 0.116,
      "step": 819
    },
    {
      "epoch": 1.8926716676283901,
      "grad_norm": 0.06405184417963028,
      "learning_rate": 0.00012453703703703704,
      "loss": 0.119,
      "step": 820
    },
    {
      "epoch": 1.8949798038084247,
      "grad_norm": 0.06503359973430634,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.1213,
      "step": 821
    },
    {
      "epoch": 1.8972879399884595,
      "grad_norm": 0.061056263744831085,
      "learning_rate": 0.00012435185185185186,
      "loss": 0.1159,
      "step": 822
    },
    {
      "epoch": 1.899596076168494,
      "grad_norm": 0.05711039900779724,
      "learning_rate": 0.0001242592592592593,
      "loss": 0.117,
      "step": 823
    },
    {
      "epoch": 1.9019042123485286,
      "grad_norm": 0.05892735347151756,
      "learning_rate": 0.00012416666666666669,
      "loss": 0.1157,
      "step": 824
    },
    {
      "epoch": 1.9042123485285631,
      "grad_norm": 0.05483463779091835,
      "learning_rate": 0.00012407407407407408,
      "loss": 0.1237,
      "step": 825
    },
    {
      "epoch": 1.9065204847085977,
      "grad_norm": 0.06040766462683678,
      "learning_rate": 0.0001239814814814815,
      "loss": 0.1195,
      "step": 826
    },
    {
      "epoch": 1.9088286208886325,
      "grad_norm": 0.05746941268444061,
      "learning_rate": 0.0001238888888888889,
      "loss": 0.117,
      "step": 827
    },
    {
      "epoch": 1.911136757068667,
      "grad_norm": 0.062030136585235596,
      "learning_rate": 0.0001237962962962963,
      "loss": 0.1167,
      "step": 828
    },
    {
      "epoch": 1.9134448932487018,
      "grad_norm": 0.063519686460495,
      "learning_rate": 0.0001237037037037037,
      "loss": 0.1252,
      "step": 829
    },
    {
      "epoch": 1.9157530294287364,
      "grad_norm": 0.06378965079784393,
      "learning_rate": 0.00012361111111111112,
      "loss": 0.1247,
      "step": 830
    },
    {
      "epoch": 1.918061165608771,
      "grad_norm": 0.058711323887109756,
      "learning_rate": 0.00012351851851851852,
      "loss": 0.1239,
      "step": 831
    },
    {
      "epoch": 1.9203693017888055,
      "grad_norm": 0.06036626920104027,
      "learning_rate": 0.00012342592592592592,
      "loss": 0.1198,
      "step": 832
    },
    {
      "epoch": 1.92267743796884,
      "grad_norm": 0.060680925846099854,
      "learning_rate": 0.00012333333333333334,
      "loss": 0.1175,
      "step": 833
    },
    {
      "epoch": 1.9249855741488748,
      "grad_norm": 0.06323889642953873,
      "learning_rate": 0.00012324074074074074,
      "loss": 0.1192,
      "step": 834
    },
    {
      "epoch": 1.9272937103289094,
      "grad_norm": 0.06004302576184273,
      "learning_rate": 0.00012314814814814813,
      "loss": 0.1161,
      "step": 835
    },
    {
      "epoch": 1.9296018465089442,
      "grad_norm": 0.062092848122119904,
      "learning_rate": 0.00012305555555555556,
      "loss": 0.1125,
      "step": 836
    },
    {
      "epoch": 1.9319099826889787,
      "grad_norm": 0.06276243925094604,
      "learning_rate": 0.00012296296296296296,
      "loss": 0.1226,
      "step": 837
    },
    {
      "epoch": 1.9342181188690133,
      "grad_norm": 0.06010131165385246,
      "learning_rate": 0.00012287037037037038,
      "loss": 0.1154,
      "step": 838
    },
    {
      "epoch": 1.9365262550490479,
      "grad_norm": 0.06013114005327225,
      "learning_rate": 0.0001227777777777778,
      "loss": 0.1216,
      "step": 839
    },
    {
      "epoch": 1.9388343912290824,
      "grad_norm": 0.06902510672807693,
      "learning_rate": 0.0001226851851851852,
      "loss": 0.1169,
      "step": 840
    },
    {
      "epoch": 1.9411425274091172,
      "grad_norm": 0.0635884702205658,
      "learning_rate": 0.0001225925925925926,
      "loss": 0.1199,
      "step": 841
    },
    {
      "epoch": 1.9434506635891518,
      "grad_norm": 0.063090480864048,
      "learning_rate": 0.00012250000000000002,
      "loss": 0.1174,
      "step": 842
    },
    {
      "epoch": 1.9457587997691865,
      "grad_norm": 0.06000242754817009,
      "learning_rate": 0.00012240740740740742,
      "loss": 0.1133,
      "step": 843
    },
    {
      "epoch": 1.948066935949221,
      "grad_norm": 0.057105567306280136,
      "learning_rate": 0.00012231481481481482,
      "loss": 0.1237,
      "step": 844
    },
    {
      "epoch": 1.9503750721292556,
      "grad_norm": 0.06667926162481308,
      "learning_rate": 0.00012222222222222224,
      "loss": 0.1162,
      "step": 845
    },
    {
      "epoch": 1.9526832083092902,
      "grad_norm": 0.057868074625730515,
      "learning_rate": 0.00012212962962962964,
      "loss": 0.1205,
      "step": 846
    },
    {
      "epoch": 1.9549913444893248,
      "grad_norm": 0.06870570033788681,
      "learning_rate": 0.00012203703703703704,
      "loss": 0.1171,
      "step": 847
    },
    {
      "epoch": 1.9572994806693593,
      "grad_norm": 0.06250982731580734,
      "learning_rate": 0.00012194444444444446,
      "loss": 0.1164,
      "step": 848
    },
    {
      "epoch": 1.959607616849394,
      "grad_norm": 0.05901934951543808,
      "learning_rate": 0.00012185185185185186,
      "loss": 0.118,
      "step": 849
    },
    {
      "epoch": 1.9619157530294289,
      "grad_norm": 0.07104910165071487,
      "learning_rate": 0.00012175925925925927,
      "loss": 0.1217,
      "step": 850
    },
    {
      "epoch": 1.9642238892094634,
      "grad_norm": 0.06285929679870605,
      "learning_rate": 0.00012166666666666667,
      "loss": 0.121,
      "step": 851
    },
    {
      "epoch": 1.966532025389498,
      "grad_norm": 0.06221470236778259,
      "learning_rate": 0.00012157407407407409,
      "loss": 0.1196,
      "step": 852
    },
    {
      "epoch": 1.9688401615695326,
      "grad_norm": 0.06815532594919205,
      "learning_rate": 0.00012148148148148149,
      "loss": 0.1183,
      "step": 853
    },
    {
      "epoch": 1.9711482977495671,
      "grad_norm": 0.05494093894958496,
      "learning_rate": 0.00012138888888888888,
      "loss": 0.1207,
      "step": 854
    },
    {
      "epoch": 1.9734564339296017,
      "grad_norm": 0.060239046812057495,
      "learning_rate": 0.00012129629629629631,
      "loss": 0.1189,
      "step": 855
    },
    {
      "epoch": 1.9757645701096365,
      "grad_norm": 0.056726496666669846,
      "learning_rate": 0.0001212037037037037,
      "loss": 0.1248,
      "step": 856
    },
    {
      "epoch": 1.9780727062896712,
      "grad_norm": 0.06294970214366913,
      "learning_rate": 0.0001211111111111111,
      "loss": 0.1211,
      "step": 857
    },
    {
      "epoch": 1.9803808424697058,
      "grad_norm": 0.06444509327411652,
      "learning_rate": 0.00012101851851851853,
      "loss": 0.1151,
      "step": 858
    },
    {
      "epoch": 1.9826889786497404,
      "grad_norm": 0.05366230756044388,
      "learning_rate": 0.00012092592592592594,
      "loss": 0.1201,
      "step": 859
    },
    {
      "epoch": 1.984997114829775,
      "grad_norm": 0.06061701104044914,
      "learning_rate": 0.00012083333333333333,
      "loss": 0.1167,
      "step": 860
    },
    {
      "epoch": 1.9873052510098095,
      "grad_norm": 0.057989396154880524,
      "learning_rate": 0.00012074074074074076,
      "loss": 0.1166,
      "step": 861
    },
    {
      "epoch": 1.989613387189844,
      "grad_norm": 0.06120140105485916,
      "learning_rate": 0.00012064814814814816,
      "loss": 0.1126,
      "step": 862
    },
    {
      "epoch": 1.9919215233698788,
      "grad_norm": 0.06389805674552917,
      "learning_rate": 0.00012055555555555555,
      "loss": 0.1187,
      "step": 863
    },
    {
      "epoch": 1.9942296595499136,
      "grad_norm": 0.06164228171110153,
      "learning_rate": 0.00012046296296296298,
      "loss": 0.1194,
      "step": 864
    },
    {
      "epoch": 1.9965377957299482,
      "grad_norm": 0.05837437883019447,
      "learning_rate": 0.00012037037037037037,
      "loss": 0.1226,
      "step": 865
    },
    {
      "epoch": 1.9988459319099827,
      "grad_norm": 0.06579951196908951,
      "learning_rate": 0.00012027777777777778,
      "loss": 0.117,
      "step": 866
    },
    {
      "epoch": 2.0011540680900173,
      "grad_norm": 0.06458060443401337,
      "learning_rate": 0.0001201851851851852,
      "loss": 0.1141,
      "step": 867
    },
    {
      "epoch": 2.003462204270052,
      "grad_norm": 0.07066481560468674,
      "learning_rate": 0.0001200925925925926,
      "loss": 0.1133,
      "step": 868
    },
    {
      "epoch": 2.0057703404500864,
      "grad_norm": 0.07935874164104462,
      "learning_rate": 0.00012,
      "loss": 0.1147,
      "step": 869
    },
    {
      "epoch": 2.008078476630121,
      "grad_norm": 0.06362339109182358,
      "learning_rate": 0.00011990740740740743,
      "loss": 0.1148,
      "step": 870
    },
    {
      "epoch": 2.010386612810156,
      "grad_norm": 0.07906344532966614,
      "learning_rate": 0.00011981481481481482,
      "loss": 0.1139,
      "step": 871
    },
    {
      "epoch": 2.0126947489901905,
      "grad_norm": 0.0744648352265358,
      "learning_rate": 0.00011972222222222222,
      "loss": 0.1184,
      "step": 872
    },
    {
      "epoch": 2.015002885170225,
      "grad_norm": 0.062084686011075974,
      "learning_rate": 0.00011962962962962962,
      "loss": 0.1183,
      "step": 873
    },
    {
      "epoch": 2.0173110213502596,
      "grad_norm": 0.08310504257678986,
      "learning_rate": 0.00011953703703703704,
      "loss": 0.1145,
      "step": 874
    },
    {
      "epoch": 2.019619157530294,
      "grad_norm": 0.07134054601192474,
      "learning_rate": 0.00011944444444444445,
      "loss": 0.1165,
      "step": 875
    },
    {
      "epoch": 2.0219272937103288,
      "grad_norm": 0.0694611668586731,
      "learning_rate": 0.00011935185185185185,
      "loss": 0.12,
      "step": 876
    },
    {
      "epoch": 2.0242354298903633,
      "grad_norm": 0.08106060326099396,
      "learning_rate": 0.00011925925925925927,
      "loss": 0.1073,
      "step": 877
    },
    {
      "epoch": 2.0265435660703983,
      "grad_norm": 0.06540350615978241,
      "learning_rate": 0.00011916666666666667,
      "loss": 0.1192,
      "step": 878
    },
    {
      "epoch": 2.028851702250433,
      "grad_norm": 0.0767219141125679,
      "learning_rate": 0.00011907407407407407,
      "loss": 0.1095,
      "step": 879
    },
    {
      "epoch": 2.0311598384304674,
      "grad_norm": 0.06539180874824524,
      "learning_rate": 0.0001189814814814815,
      "loss": 0.1203,
      "step": 880
    },
    {
      "epoch": 2.033467974610502,
      "grad_norm": 0.07096482813358307,
      "learning_rate": 0.00011888888888888889,
      "loss": 0.1158,
      "step": 881
    },
    {
      "epoch": 2.0357761107905366,
      "grad_norm": 0.07212911546230316,
      "learning_rate": 0.0001187962962962963,
      "loss": 0.114,
      "step": 882
    },
    {
      "epoch": 2.038084246970571,
      "grad_norm": 0.06715547293424606,
      "learning_rate": 0.00011870370370370371,
      "loss": 0.1127,
      "step": 883
    },
    {
      "epoch": 2.0403923831506057,
      "grad_norm": 0.07804431766271591,
      "learning_rate": 0.00011861111111111112,
      "loss": 0.1152,
      "step": 884
    },
    {
      "epoch": 2.0427005193306407,
      "grad_norm": 0.06067255511879921,
      "learning_rate": 0.00011851851851851852,
      "loss": 0.1144,
      "step": 885
    },
    {
      "epoch": 2.0450086555106752,
      "grad_norm": 0.07118728011846542,
      "learning_rate": 0.00011842592592592594,
      "loss": 0.1127,
      "step": 886
    },
    {
      "epoch": 2.04731679169071,
      "grad_norm": 0.07555549591779709,
      "learning_rate": 0.00011833333333333334,
      "loss": 0.1104,
      "step": 887
    },
    {
      "epoch": 2.0496249278707444,
      "grad_norm": 0.06712587177753448,
      "learning_rate": 0.00011824074074074074,
      "loss": 0.1157,
      "step": 888
    },
    {
      "epoch": 2.051933064050779,
      "grad_norm": 0.08028671145439148,
      "learning_rate": 0.00011814814814814816,
      "loss": 0.1144,
      "step": 889
    },
    {
      "epoch": 2.0542412002308135,
      "grad_norm": 0.07372070103883743,
      "learning_rate": 0.00011805555555555556,
      "loss": 0.1103,
      "step": 890
    },
    {
      "epoch": 2.056549336410848,
      "grad_norm": 0.0650438442826271,
      "learning_rate": 0.00011796296296296297,
      "loss": 0.1134,
      "step": 891
    },
    {
      "epoch": 2.058857472590883,
      "grad_norm": 0.06656362116336823,
      "learning_rate": 0.00011787037037037038,
      "loss": 0.1179,
      "step": 892
    },
    {
      "epoch": 2.0611656087709176,
      "grad_norm": 0.07259076833724976,
      "learning_rate": 0.00011777777777777779,
      "loss": 0.1074,
      "step": 893
    },
    {
      "epoch": 2.063473744950952,
      "grad_norm": 0.06991040706634521,
      "learning_rate": 0.00011768518518518519,
      "loss": 0.1174,
      "step": 894
    },
    {
      "epoch": 2.0657818811309867,
      "grad_norm": 0.061825189739465714,
      "learning_rate": 0.00011759259259259259,
      "loss": 0.1128,
      "step": 895
    },
    {
      "epoch": 2.0680900173110213,
      "grad_norm": 0.06519865989685059,
      "learning_rate": 0.00011750000000000001,
      "loss": 0.1111,
      "step": 896
    },
    {
      "epoch": 2.070398153491056,
      "grad_norm": 0.06219363212585449,
      "learning_rate": 0.00011740740740740741,
      "loss": 0.1149,
      "step": 897
    },
    {
      "epoch": 2.0727062896710904,
      "grad_norm": 0.06893327087163925,
      "learning_rate": 0.0001173148148148148,
      "loss": 0.111,
      "step": 898
    },
    {
      "epoch": 2.0750144258511254,
      "grad_norm": 0.06621815264225006,
      "learning_rate": 0.00011722222222222223,
      "loss": 0.1119,
      "step": 899
    },
    {
      "epoch": 2.07732256203116,
      "grad_norm": 0.06052153557538986,
      "learning_rate": 0.00011712962962962964,
      "loss": 0.1117,
      "step": 900
    },
    {
      "epoch": 2.0796306982111945,
      "grad_norm": 0.06153155490756035,
      "learning_rate": 0.00011703703703703704,
      "loss": 0.1114,
      "step": 901
    },
    {
      "epoch": 2.081938834391229,
      "grad_norm": 0.05995754897594452,
      "learning_rate": 0.00011694444444444446,
      "loss": 0.1157,
      "step": 902
    },
    {
      "epoch": 2.0842469705712636,
      "grad_norm": 0.07573620229959488,
      "learning_rate": 0.00011685185185185186,
      "loss": 0.1117,
      "step": 903
    },
    {
      "epoch": 2.086555106751298,
      "grad_norm": 0.07463104277849197,
      "learning_rate": 0.00011675925925925925,
      "loss": 0.116,
      "step": 904
    },
    {
      "epoch": 2.0888632429313327,
      "grad_norm": 0.06527411937713623,
      "learning_rate": 0.00011666666666666668,
      "loss": 0.1134,
      "step": 905
    },
    {
      "epoch": 2.0911713791113677,
      "grad_norm": 0.06653150916099548,
      "learning_rate": 0.00011657407407407408,
      "loss": 0.1103,
      "step": 906
    },
    {
      "epoch": 2.0934795152914023,
      "grad_norm": 0.0674898698925972,
      "learning_rate": 0.00011648148148148149,
      "loss": 0.1102,
      "step": 907
    },
    {
      "epoch": 2.095787651471437,
      "grad_norm": 0.06709618121385574,
      "learning_rate": 0.0001163888888888889,
      "loss": 0.1203,
      "step": 908
    },
    {
      "epoch": 2.0980957876514714,
      "grad_norm": 0.06608498096466064,
      "learning_rate": 0.00011629629629629631,
      "loss": 0.1134,
      "step": 909
    },
    {
      "epoch": 2.100403923831506,
      "grad_norm": 0.0731372982263565,
      "learning_rate": 0.0001162037037037037,
      "loss": 0.1146,
      "step": 910
    },
    {
      "epoch": 2.1027120600115405,
      "grad_norm": 0.06405247002840042,
      "learning_rate": 0.00011611111111111113,
      "loss": 0.118,
      "step": 911
    },
    {
      "epoch": 2.105020196191575,
      "grad_norm": 0.06920763105154037,
      "learning_rate": 0.00011601851851851853,
      "loss": 0.1172,
      "step": 912
    },
    {
      "epoch": 2.10732833237161,
      "grad_norm": 0.06977575272321701,
      "learning_rate": 0.00011592592592592592,
      "loss": 0.1127,
      "step": 913
    },
    {
      "epoch": 2.1096364685516447,
      "grad_norm": 0.06482131779193878,
      "learning_rate": 0.00011583333333333335,
      "loss": 0.1233,
      "step": 914
    },
    {
      "epoch": 2.111944604731679,
      "grad_norm": 0.06366951018571854,
      "learning_rate": 0.00011574074074074075,
      "loss": 0.1172,
      "step": 915
    },
    {
      "epoch": 2.114252740911714,
      "grad_norm": 0.06167622655630112,
      "learning_rate": 0.00011564814814814816,
      "loss": 0.1164,
      "step": 916
    },
    {
      "epoch": 2.1165608770917483,
      "grad_norm": 0.06380926072597504,
      "learning_rate": 0.00011555555555555555,
      "loss": 0.1162,
      "step": 917
    },
    {
      "epoch": 2.118869013271783,
      "grad_norm": 0.0627114549279213,
      "learning_rate": 0.00011546296296296298,
      "loss": 0.1047,
      "step": 918
    },
    {
      "epoch": 2.1211771494518175,
      "grad_norm": 0.06708511710166931,
      "learning_rate": 0.00011537037037037037,
      "loss": 0.1169,
      "step": 919
    },
    {
      "epoch": 2.1234852856318525,
      "grad_norm": 0.06480534374713898,
      "learning_rate": 0.00011527777777777777,
      "loss": 0.1123,
      "step": 920
    },
    {
      "epoch": 2.125793421811887,
      "grad_norm": 0.06327366083860397,
      "learning_rate": 0.0001151851851851852,
      "loss": 0.1075,
      "step": 921
    },
    {
      "epoch": 2.1281015579919216,
      "grad_norm": 0.07612491399049759,
      "learning_rate": 0.00011509259259259259,
      "loss": 0.1147,
      "step": 922
    },
    {
      "epoch": 2.130409694171956,
      "grad_norm": 0.06172380596399307,
      "learning_rate": 0.00011499999999999999,
      "loss": 0.1096,
      "step": 923
    },
    {
      "epoch": 2.1327178303519907,
      "grad_norm": 0.06699161976575851,
      "learning_rate": 0.00011490740740740741,
      "loss": 0.1135,
      "step": 924
    },
    {
      "epoch": 2.1350259665320253,
      "grad_norm": 0.06589633971452713,
      "learning_rate": 0.00011481481481481482,
      "loss": 0.1136,
      "step": 925
    },
    {
      "epoch": 2.13733410271206,
      "grad_norm": 0.06807922571897507,
      "learning_rate": 0.00011472222222222222,
      "loss": 0.114,
      "step": 926
    },
    {
      "epoch": 2.139642238892095,
      "grad_norm": 0.07797038555145264,
      "learning_rate": 0.00011462962962962965,
      "loss": 0.1135,
      "step": 927
    },
    {
      "epoch": 2.1419503750721294,
      "grad_norm": 0.0691361203789711,
      "learning_rate": 0.00011453703703703704,
      "loss": 0.1137,
      "step": 928
    },
    {
      "epoch": 2.144258511252164,
      "grad_norm": 0.08262480050325394,
      "learning_rate": 0.00011444444444444444,
      "loss": 0.1129,
      "step": 929
    },
    {
      "epoch": 2.1465666474321985,
      "grad_norm": 0.07052849233150482,
      "learning_rate": 0.00011435185185185186,
      "loss": 0.1193,
      "step": 930
    },
    {
      "epoch": 2.148874783612233,
      "grad_norm": 0.07403450459241867,
      "learning_rate": 0.00011425925925925926,
      "loss": 0.1088,
      "step": 931
    },
    {
      "epoch": 2.1511829197922676,
      "grad_norm": 0.07125364989042282,
      "learning_rate": 0.00011416666666666667,
      "loss": 0.1154,
      "step": 932
    },
    {
      "epoch": 2.153491055972302,
      "grad_norm": 0.07073383033275604,
      "learning_rate": 0.00011407407407407408,
      "loss": 0.1127,
      "step": 933
    },
    {
      "epoch": 2.155799192152337,
      "grad_norm": 0.06334490329027176,
      "learning_rate": 0.0001139814814814815,
      "loss": 0.1105,
      "step": 934
    },
    {
      "epoch": 2.1581073283323717,
      "grad_norm": 0.06710466742515564,
      "learning_rate": 0.00011388888888888889,
      "loss": 0.1169,
      "step": 935
    },
    {
      "epoch": 2.1604154645124063,
      "grad_norm": 0.06621242314577103,
      "learning_rate": 0.00011379629629629632,
      "loss": 0.1088,
      "step": 936
    },
    {
      "epoch": 2.162723600692441,
      "grad_norm": 0.06437657028436661,
      "learning_rate": 0.00011370370370370371,
      "loss": 0.1144,
      "step": 937
    },
    {
      "epoch": 2.1650317368724754,
      "grad_norm": 0.07260672748088837,
      "learning_rate": 0.00011361111111111111,
      "loss": 0.1151,
      "step": 938
    },
    {
      "epoch": 2.16733987305251,
      "grad_norm": 0.06671048700809479,
      "learning_rate": 0.0001135185185185185,
      "loss": 0.1053,
      "step": 939
    },
    {
      "epoch": 2.1696480092325445,
      "grad_norm": 0.07555361092090607,
      "learning_rate": 0.00011342592592592593,
      "loss": 0.119,
      "step": 940
    },
    {
      "epoch": 2.1719561454125795,
      "grad_norm": 0.06079293414950371,
      "learning_rate": 0.00011333333333333334,
      "loss": 0.1116,
      "step": 941
    },
    {
      "epoch": 2.174264281592614,
      "grad_norm": 0.07616889476776123,
      "learning_rate": 0.00011324074074074074,
      "loss": 0.1102,
      "step": 942
    },
    {
      "epoch": 2.1765724177726486,
      "grad_norm": 0.05999071151018143,
      "learning_rate": 0.00011314814814814816,
      "loss": 0.1104,
      "step": 943
    },
    {
      "epoch": 2.178880553952683,
      "grad_norm": 0.06329882144927979,
      "learning_rate": 0.00011305555555555556,
      "loss": 0.1186,
      "step": 944
    },
    {
      "epoch": 2.1811886901327178,
      "grad_norm": 0.06887472420930862,
      "learning_rate": 0.00011296296296296296,
      "loss": 0.1167,
      "step": 945
    },
    {
      "epoch": 2.1834968263127523,
      "grad_norm": 0.06195090711116791,
      "learning_rate": 0.00011287037037037038,
      "loss": 0.1132,
      "step": 946
    },
    {
      "epoch": 2.185804962492787,
      "grad_norm": 0.06101727485656738,
      "learning_rate": 0.00011277777777777778,
      "loss": 0.1098,
      "step": 947
    },
    {
      "epoch": 2.188113098672822,
      "grad_norm": 0.058897484093904495,
      "learning_rate": 0.00011268518518518519,
      "loss": 0.111,
      "step": 948
    },
    {
      "epoch": 2.1904212348528564,
      "grad_norm": 0.06654559820890427,
      "learning_rate": 0.0001125925925925926,
      "loss": 0.1147,
      "step": 949
    },
    {
      "epoch": 2.192729371032891,
      "grad_norm": 0.06810339540243149,
      "learning_rate": 0.00011250000000000001,
      "loss": 0.1115,
      "step": 950
    },
    {
      "epoch": 2.1950375072129256,
      "grad_norm": 0.06797947734594345,
      "learning_rate": 0.00011240740740740741,
      "loss": 0.1144,
      "step": 951
    },
    {
      "epoch": 2.19734564339296,
      "grad_norm": 0.062455661594867706,
      "learning_rate": 0.00011231481481481483,
      "loss": 0.1069,
      "step": 952
    },
    {
      "epoch": 2.1996537795729947,
      "grad_norm": 0.06143524497747421,
      "learning_rate": 0.00011222222222222223,
      "loss": 0.1176,
      "step": 953
    },
    {
      "epoch": 2.2019619157530292,
      "grad_norm": 0.06190888583660126,
      "learning_rate": 0.00011212962962962963,
      "loss": 0.1126,
      "step": 954
    },
    {
      "epoch": 2.2042700519330642,
      "grad_norm": 0.06230132281780243,
      "learning_rate": 0.00011203703703703705,
      "loss": 0.1228,
      "step": 955
    },
    {
      "epoch": 2.206578188113099,
      "grad_norm": 0.06454814225435257,
      "learning_rate": 0.00011194444444444445,
      "loss": 0.1182,
      "step": 956
    },
    {
      "epoch": 2.2088863242931334,
      "grad_norm": 0.06508689373731613,
      "learning_rate": 0.00011185185185185186,
      "loss": 0.1097,
      "step": 957
    },
    {
      "epoch": 2.211194460473168,
      "grad_norm": 0.06558562815189362,
      "learning_rate": 0.00011175925925925927,
      "loss": 0.113,
      "step": 958
    },
    {
      "epoch": 2.2135025966532025,
      "grad_norm": 0.06878868490457535,
      "learning_rate": 0.00011166666666666668,
      "loss": 0.1123,
      "step": 959
    },
    {
      "epoch": 2.215810732833237,
      "grad_norm": 0.06292112171649933,
      "learning_rate": 0.00011157407407407408,
      "loss": 0.1124,
      "step": 960
    },
    {
      "epoch": 2.2181188690132716,
      "grad_norm": 0.06617525219917297,
      "learning_rate": 0.0001114814814814815,
      "loss": 0.1217,
      "step": 961
    },
    {
      "epoch": 2.2204270051933066,
      "grad_norm": 0.07070888578891754,
      "learning_rate": 0.0001113888888888889,
      "loss": 0.1117,
      "step": 962
    },
    {
      "epoch": 2.222735141373341,
      "grad_norm": 0.058782633394002914,
      "learning_rate": 0.0001112962962962963,
      "loss": 0.1096,
      "step": 963
    },
    {
      "epoch": 2.2250432775533757,
      "grad_norm": 0.07320119440555573,
      "learning_rate": 0.00011120370370370369,
      "loss": 0.1038,
      "step": 964
    },
    {
      "epoch": 2.2273514137334103,
      "grad_norm": 0.06768988817930222,
      "learning_rate": 0.00011111111111111112,
      "loss": 0.1113,
      "step": 965
    },
    {
      "epoch": 2.229659549913445,
      "grad_norm": 0.06252680718898773,
      "learning_rate": 0.00011101851851851853,
      "loss": 0.1137,
      "step": 966
    },
    {
      "epoch": 2.2319676860934794,
      "grad_norm": 0.06756065785884857,
      "learning_rate": 0.00011092592592592592,
      "loss": 0.1129,
      "step": 967
    },
    {
      "epoch": 2.234275822273514,
      "grad_norm": 0.062187232077121735,
      "learning_rate": 0.00011083333333333335,
      "loss": 0.1033,
      "step": 968
    },
    {
      "epoch": 2.236583958453549,
      "grad_norm": 0.06824641674757004,
      "learning_rate": 0.00011074074074074075,
      "loss": 0.1089,
      "step": 969
    },
    {
      "epoch": 2.2388920946335835,
      "grad_norm": 0.06394147127866745,
      "learning_rate": 0.00011064814814814814,
      "loss": 0.1134,
      "step": 970
    },
    {
      "epoch": 2.241200230813618,
      "grad_norm": 0.06721079349517822,
      "learning_rate": 0.00011055555555555557,
      "loss": 0.117,
      "step": 971
    },
    {
      "epoch": 2.2435083669936526,
      "grad_norm": 0.06885454803705215,
      "learning_rate": 0.00011046296296296296,
      "loss": 0.1127,
      "step": 972
    },
    {
      "epoch": 2.245816503173687,
      "grad_norm": 0.07430989295244217,
      "learning_rate": 0.00011037037037037037,
      "loss": 0.1078,
      "step": 973
    },
    {
      "epoch": 2.2481246393537218,
      "grad_norm": 0.07048432528972626,
      "learning_rate": 0.00011027777777777779,
      "loss": 0.1068,
      "step": 974
    },
    {
      "epoch": 2.2504327755337563,
      "grad_norm": 0.06691984832286835,
      "learning_rate": 0.0001101851851851852,
      "loss": 0.1054,
      "step": 975
    },
    {
      "epoch": 2.252740911713791,
      "grad_norm": 0.08576422929763794,
      "learning_rate": 0.0001100925925925926,
      "loss": 0.1099,
      "step": 976
    },
    {
      "epoch": 2.255049047893826,
      "grad_norm": 0.06398241966962814,
      "learning_rate": 0.00011000000000000002,
      "loss": 0.1171,
      "step": 977
    },
    {
      "epoch": 2.2573571840738604,
      "grad_norm": 0.06927427649497986,
      "learning_rate": 0.00010990740740740741,
      "loss": 0.1091,
      "step": 978
    },
    {
      "epoch": 2.259665320253895,
      "grad_norm": 0.0643596425652504,
      "learning_rate": 0.00010981481481481481,
      "loss": 0.1092,
      "step": 979
    },
    {
      "epoch": 2.2619734564339296,
      "grad_norm": 0.06035730615258217,
      "learning_rate": 0.00010972222222222224,
      "loss": 0.1139,
      "step": 980
    },
    {
      "epoch": 2.264281592613964,
      "grad_norm": 0.06898952275514603,
      "learning_rate": 0.00010962962962962963,
      "loss": 0.1154,
      "step": 981
    },
    {
      "epoch": 2.2665897287939987,
      "grad_norm": 0.06056066229939461,
      "learning_rate": 0.00010953703703703704,
      "loss": 0.1117,
      "step": 982
    },
    {
      "epoch": 2.2688978649740337,
      "grad_norm": 0.0705919936299324,
      "learning_rate": 0.00010944444444444445,
      "loss": 0.115,
      "step": 983
    },
    {
      "epoch": 2.2712060011540682,
      "grad_norm": 0.06630073487758636,
      "learning_rate": 0.00010935185185185187,
      "loss": 0.1184,
      "step": 984
    },
    {
      "epoch": 2.273514137334103,
      "grad_norm": 0.06530337780714035,
      "learning_rate": 0.00010925925925925926,
      "loss": 0.1154,
      "step": 985
    },
    {
      "epoch": 2.2758222735141374,
      "grad_norm": 0.06615268439054489,
      "learning_rate": 0.00010916666666666666,
      "loss": 0.1082,
      "step": 986
    },
    {
      "epoch": 2.278130409694172,
      "grad_norm": 0.05847036466002464,
      "learning_rate": 0.00010907407407407408,
      "loss": 0.1194,
      "step": 987
    },
    {
      "epoch": 2.2804385458742065,
      "grad_norm": 0.07164110988378525,
      "learning_rate": 0.00010898148148148148,
      "loss": 0.1049,
      "step": 988
    },
    {
      "epoch": 2.282746682054241,
      "grad_norm": 0.0656791478395462,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.1163,
      "step": 989
    },
    {
      "epoch": 2.2850548182342756,
      "grad_norm": 0.06308694183826447,
      "learning_rate": 0.0001087962962962963,
      "loss": 0.1232,
      "step": 990
    },
    {
      "epoch": 2.2873629544143106,
      "grad_norm": 0.06776368618011475,
      "learning_rate": 0.00010870370370370371,
      "loss": 0.1123,
      "step": 991
    },
    {
      "epoch": 2.289671090594345,
      "grad_norm": 0.06645739823579788,
      "learning_rate": 0.00010861111111111111,
      "loss": 0.1166,
      "step": 992
    },
    {
      "epoch": 2.2919792267743797,
      "grad_norm": 0.07519471645355225,
      "learning_rate": 0.00010851851851851853,
      "loss": 0.1129,
      "step": 993
    },
    {
      "epoch": 2.2942873629544143,
      "grad_norm": 0.06778604537248611,
      "learning_rate": 0.00010842592592592593,
      "loss": 0.1111,
      "step": 994
    },
    {
      "epoch": 2.296595499134449,
      "grad_norm": 0.06861971318721771,
      "learning_rate": 0.00010833333333333333,
      "loss": 0.1211,
      "step": 995
    },
    {
      "epoch": 2.2989036353144834,
      "grad_norm": 0.07128448039293289,
      "learning_rate": 0.00010824074074074075,
      "loss": 0.1073,
      "step": 996
    },
    {
      "epoch": 2.3012117714945184,
      "grad_norm": 0.06735311448574066,
      "learning_rate": 0.00010814814814814815,
      "loss": 0.1086,
      "step": 997
    },
    {
      "epoch": 2.303519907674553,
      "grad_norm": 0.07986678183078766,
      "learning_rate": 0.00010805555555555556,
      "loss": 0.112,
      "step": 998
    },
    {
      "epoch": 2.3058280438545875,
      "grad_norm": 0.068412184715271,
      "learning_rate": 0.00010796296296296297,
      "loss": 0.1065,
      "step": 999
    },
    {
      "epoch": 2.308136180034622,
      "grad_norm": 0.0715126320719719,
      "learning_rate": 0.00010787037037037038,
      "loss": 0.1159,
      "step": 1000
    },
    {
      "epoch": 2.3104443162146566,
      "grad_norm": 0.0688762441277504,
      "learning_rate": 0.00010777777777777778,
      "loss": 0.1094,
      "step": 1001
    },
    {
      "epoch": 2.312752452394691,
      "grad_norm": 0.06288091093301773,
      "learning_rate": 0.0001076851851851852,
      "loss": 0.1143,
      "step": 1002
    },
    {
      "epoch": 2.3150605885747257,
      "grad_norm": 0.0668220967054367,
      "learning_rate": 0.0001075925925925926,
      "loss": 0.1084,
      "step": 1003
    },
    {
      "epoch": 2.3173687247547603,
      "grad_norm": 0.06132470816373825,
      "learning_rate": 0.0001075,
      "loss": 0.1183,
      "step": 1004
    },
    {
      "epoch": 2.3196768609347953,
      "grad_norm": 0.06608840078115463,
      "learning_rate": 0.00010740740740740742,
      "loss": 0.1134,
      "step": 1005
    },
    {
      "epoch": 2.32198499711483,
      "grad_norm": 0.06354363262653351,
      "learning_rate": 0.00010731481481481482,
      "loss": 0.1071,
      "step": 1006
    },
    {
      "epoch": 2.3242931332948644,
      "grad_norm": 0.06218728423118591,
      "learning_rate": 0.00010722222222222223,
      "loss": 0.1135,
      "step": 1007
    },
    {
      "epoch": 2.326601269474899,
      "grad_norm": 0.058630652725696564,
      "learning_rate": 0.00010712962962962963,
      "loss": 0.1111,
      "step": 1008
    },
    {
      "epoch": 2.3289094056549335,
      "grad_norm": 0.06434722989797592,
      "learning_rate": 0.00010703703703703705,
      "loss": 0.113,
      "step": 1009
    },
    {
      "epoch": 2.331217541834968,
      "grad_norm": 0.059792399406433105,
      "learning_rate": 0.00010694444444444445,
      "loss": 0.1147,
      "step": 1010
    },
    {
      "epoch": 2.333525678015003,
      "grad_norm": 0.06906107813119888,
      "learning_rate": 0.00010685185185185185,
      "loss": 0.1154,
      "step": 1011
    },
    {
      "epoch": 2.3358338141950377,
      "grad_norm": 0.06868558377027512,
      "learning_rate": 0.00010675925925925927,
      "loss": 0.111,
      "step": 1012
    },
    {
      "epoch": 2.338141950375072,
      "grad_norm": 0.07943182438611984,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.1152,
      "step": 1013
    },
    {
      "epoch": 2.340450086555107,
      "grad_norm": 0.07124125212430954,
      "learning_rate": 0.00010657407407407408,
      "loss": 0.114,
      "step": 1014
    },
    {
      "epoch": 2.3427582227351413,
      "grad_norm": 0.06394325941801071,
      "learning_rate": 0.00010648148148148149,
      "loss": 0.1118,
      "step": 1015
    },
    {
      "epoch": 2.345066358915176,
      "grad_norm": 0.06467465311288834,
      "learning_rate": 0.0001063888888888889,
      "loss": 0.1122,
      "step": 1016
    },
    {
      "epoch": 2.3473744950952105,
      "grad_norm": 0.060979247093200684,
      "learning_rate": 0.0001062962962962963,
      "loss": 0.1112,
      "step": 1017
    },
    {
      "epoch": 2.349682631275245,
      "grad_norm": 0.06247340515255928,
      "learning_rate": 0.00010620370370370372,
      "loss": 0.1092,
      "step": 1018
    },
    {
      "epoch": 2.35199076745528,
      "grad_norm": 0.06728631258010864,
      "learning_rate": 0.00010611111111111112,
      "loss": 0.1077,
      "step": 1019
    },
    {
      "epoch": 2.3542989036353146,
      "grad_norm": 0.061377499252557755,
      "learning_rate": 0.00010601851851851851,
      "loss": 0.1178,
      "step": 1020
    },
    {
      "epoch": 2.356607039815349,
      "grad_norm": 0.06732606887817383,
      "learning_rate": 0.00010592592592592594,
      "loss": 0.1075,
      "step": 1021
    },
    {
      "epoch": 2.3589151759953837,
      "grad_norm": 0.0579875186085701,
      "learning_rate": 0.00010583333333333334,
      "loss": 0.1147,
      "step": 1022
    },
    {
      "epoch": 2.3612233121754183,
      "grad_norm": 0.07123639434576035,
      "learning_rate": 0.00010574074074074075,
      "loss": 0.1122,
      "step": 1023
    },
    {
      "epoch": 2.363531448355453,
      "grad_norm": 0.06310222297906876,
      "learning_rate": 0.00010564814814814816,
      "loss": 0.1127,
      "step": 1024
    },
    {
      "epoch": 2.365839584535488,
      "grad_norm": 0.06181904301047325,
      "learning_rate": 0.00010555555555555557,
      "loss": 0.1142,
      "step": 1025
    },
    {
      "epoch": 2.3681477207155224,
      "grad_norm": 0.06272733211517334,
      "learning_rate": 0.00010546296296296296,
      "loss": 0.1131,
      "step": 1026
    },
    {
      "epoch": 2.370455856895557,
      "grad_norm": 0.05766041576862335,
      "learning_rate": 0.00010537037037037039,
      "loss": 0.1181,
      "step": 1027
    },
    {
      "epoch": 2.3727639930755915,
      "grad_norm": 0.060116246342659,
      "learning_rate": 0.00010527777777777779,
      "loss": 0.1062,
      "step": 1028
    },
    {
      "epoch": 2.375072129255626,
      "grad_norm": 0.06415055692195892,
      "learning_rate": 0.00010518518518518518,
      "loss": 0.1061,
      "step": 1029
    },
    {
      "epoch": 2.3773802654356606,
      "grad_norm": 0.06296518445014954,
      "learning_rate": 0.0001050925925925926,
      "loss": 0.1137,
      "step": 1030
    },
    {
      "epoch": 2.379688401615695,
      "grad_norm": 0.06534787267446518,
      "learning_rate": 0.000105,
      "loss": 0.1137,
      "step": 1031
    },
    {
      "epoch": 2.3819965377957297,
      "grad_norm": 0.06597837805747986,
      "learning_rate": 0.00010490740740740742,
      "loss": 0.1091,
      "step": 1032
    },
    {
      "epoch": 2.3843046739757647,
      "grad_norm": 0.06073552370071411,
      "learning_rate": 0.00010481481481481481,
      "loss": 0.1132,
      "step": 1033
    },
    {
      "epoch": 2.3866128101557993,
      "grad_norm": 0.06401500850915909,
      "learning_rate": 0.00010472222222222224,
      "loss": 0.1105,
      "step": 1034
    },
    {
      "epoch": 2.388920946335834,
      "grad_norm": 0.06349154561758041,
      "learning_rate": 0.00010462962962962963,
      "loss": 0.1191,
      "step": 1035
    },
    {
      "epoch": 2.3912290825158684,
      "grad_norm": 0.06778168678283691,
      "learning_rate": 0.00010453703703703703,
      "loss": 0.1134,
      "step": 1036
    },
    {
      "epoch": 2.393537218695903,
      "grad_norm": 0.060518573969602585,
      "learning_rate": 0.00010444444444444445,
      "loss": 0.1086,
      "step": 1037
    },
    {
      "epoch": 2.3958453548759375,
      "grad_norm": 0.06972122937440872,
      "learning_rate": 0.00010435185185185185,
      "loss": 0.116,
      "step": 1038
    },
    {
      "epoch": 2.3981534910559725,
      "grad_norm": 0.06940379738807678,
      "learning_rate": 0.00010425925925925926,
      "loss": 0.1107,
      "step": 1039
    },
    {
      "epoch": 2.400461627236007,
      "grad_norm": 0.08079859614372253,
      "learning_rate": 0.00010416666666666667,
      "loss": 0.1025,
      "step": 1040
    },
    {
      "epoch": 2.4027697634160416,
      "grad_norm": 0.07492893189191818,
      "learning_rate": 0.00010407407407407408,
      "loss": 0.1165,
      "step": 1041
    },
    {
      "epoch": 2.405077899596076,
      "grad_norm": 0.09707816690206528,
      "learning_rate": 0.00010398148148148148,
      "loss": 0.1138,
      "step": 1042
    },
    {
      "epoch": 2.4073860357761108,
      "grad_norm": 0.1533680260181427,
      "learning_rate": 0.0001038888888888889,
      "loss": 0.1105,
      "step": 1043
    },
    {
      "epoch": 2.4096941719561453,
      "grad_norm": 0.14268022775650024,
      "learning_rate": 0.0001037962962962963,
      "loss": 0.1111,
      "step": 1044
    },
    {
      "epoch": 2.41200230813618,
      "grad_norm": 0.08327072858810425,
      "learning_rate": 0.0001037037037037037,
      "loss": 0.1176,
      "step": 1045
    },
    {
      "epoch": 2.4143104443162144,
      "grad_norm": 0.113436259329319,
      "learning_rate": 0.00010361111111111112,
      "loss": 0.1098,
      "step": 1046
    },
    {
      "epoch": 2.4166185804962494,
      "grad_norm": 0.12621846795082092,
      "learning_rate": 0.00010351851851851852,
      "loss": 0.1157,
      "step": 1047
    },
    {
      "epoch": 2.418926716676284,
      "grad_norm": 0.09452440589666367,
      "learning_rate": 0.00010342592592592593,
      "loss": 0.1113,
      "step": 1048
    },
    {
      "epoch": 2.4212348528563186,
      "grad_norm": 0.13372133672237396,
      "learning_rate": 0.00010333333333333334,
      "loss": 0.1151,
      "step": 1049
    },
    {
      "epoch": 2.423542989036353,
      "grad_norm": 0.08389133214950562,
      "learning_rate": 0.00010324074074074075,
      "loss": 0.1089,
      "step": 1050
    },
    {
      "epoch": 2.4258511252163877,
      "grad_norm": 0.11222892254590988,
      "learning_rate": 0.00010314814814814815,
      "loss": 0.1108,
      "step": 1051
    },
    {
      "epoch": 2.4281592613964222,
      "grad_norm": 0.09788097441196442,
      "learning_rate": 0.00010305555555555555,
      "loss": 0.1176,
      "step": 1052
    },
    {
      "epoch": 2.4304673975764572,
      "grad_norm": 0.06743919849395752,
      "learning_rate": 0.00010296296296296297,
      "loss": 0.1241,
      "step": 1053
    },
    {
      "epoch": 2.432775533756492,
      "grad_norm": 0.09404382109642029,
      "learning_rate": 0.00010287037037037037,
      "loss": 0.1182,
      "step": 1054
    },
    {
      "epoch": 2.4350836699365264,
      "grad_norm": 0.0731574296951294,
      "learning_rate": 0.00010277777777777778,
      "loss": 0.1192,
      "step": 1055
    },
    {
      "epoch": 2.437391806116561,
      "grad_norm": 0.08034831285476685,
      "learning_rate": 0.00010268518518518519,
      "loss": 0.1102,
      "step": 1056
    },
    {
      "epoch": 2.4396999422965955,
      "grad_norm": 0.07376502454280853,
      "learning_rate": 0.0001025925925925926,
      "loss": 0.109,
      "step": 1057
    },
    {
      "epoch": 2.44200807847663,
      "grad_norm": 0.0669766217470169,
      "learning_rate": 0.0001025,
      "loss": 0.1195,
      "step": 1058
    },
    {
      "epoch": 2.4443162146566646,
      "grad_norm": 0.06764399260282516,
      "learning_rate": 0.00010240740740740742,
      "loss": 0.111,
      "step": 1059
    },
    {
      "epoch": 2.446624350836699,
      "grad_norm": 0.06752080470323563,
      "learning_rate": 0.00010231481481481482,
      "loss": 0.1118,
      "step": 1060
    },
    {
      "epoch": 2.448932487016734,
      "grad_norm": 0.06385441869497299,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.1107,
      "step": 1061
    },
    {
      "epoch": 2.4512406231967687,
      "grad_norm": 0.0694565549492836,
      "learning_rate": 0.00010212962962962964,
      "loss": 0.1101,
      "step": 1062
    },
    {
      "epoch": 2.4535487593768033,
      "grad_norm": 0.07093571126461029,
      "learning_rate": 0.00010203703703703704,
      "loss": 0.1134,
      "step": 1063
    },
    {
      "epoch": 2.455856895556838,
      "grad_norm": 0.07218756526708603,
      "learning_rate": 0.00010194444444444445,
      "loss": 0.1128,
      "step": 1064
    },
    {
      "epoch": 2.4581650317368724,
      "grad_norm": 0.06851860880851746,
      "learning_rate": 0.00010185185185185186,
      "loss": 0.1102,
      "step": 1065
    },
    {
      "epoch": 2.460473167916907,
      "grad_norm": 0.06919904053211212,
      "learning_rate": 0.00010175925925925927,
      "loss": 0.112,
      "step": 1066
    },
    {
      "epoch": 2.462781304096942,
      "grad_norm": 0.06780626624822617,
      "learning_rate": 0.00010166666666666667,
      "loss": 0.1126,
      "step": 1067
    },
    {
      "epoch": 2.4650894402769765,
      "grad_norm": 0.06722507625818253,
      "learning_rate": 0.00010157407407407409,
      "loss": 0.1106,
      "step": 1068
    },
    {
      "epoch": 2.467397576457011,
      "grad_norm": 0.0646774172782898,
      "learning_rate": 0.00010148148148148149,
      "loss": 0.1105,
      "step": 1069
    },
    {
      "epoch": 2.4697057126370456,
      "grad_norm": 0.06711084395647049,
      "learning_rate": 0.00010138888888888889,
      "loss": 0.1088,
      "step": 1070
    },
    {
      "epoch": 2.47201384881708,
      "grad_norm": 0.06366660445928574,
      "learning_rate": 0.00010129629629629631,
      "loss": 0.116,
      "step": 1071
    },
    {
      "epoch": 2.4743219849971148,
      "grad_norm": 0.06420040130615234,
      "learning_rate": 0.0001012037037037037,
      "loss": 0.1108,
      "step": 1072
    },
    {
      "epoch": 2.4766301211771493,
      "grad_norm": 0.05999176204204559,
      "learning_rate": 0.00010111111111111112,
      "loss": 0.1081,
      "step": 1073
    },
    {
      "epoch": 2.478938257357184,
      "grad_norm": 0.06310884654521942,
      "learning_rate": 0.00010101851851851851,
      "loss": 0.1143,
      "step": 1074
    },
    {
      "epoch": 2.481246393537219,
      "grad_norm": 0.0664728432893753,
      "learning_rate": 0.00010092592592592594,
      "loss": 0.112,
      "step": 1075
    },
    {
      "epoch": 2.4835545297172534,
      "grad_norm": 0.0687079057097435,
      "learning_rate": 0.00010083333333333334,
      "loss": 0.1121,
      "step": 1076
    },
    {
      "epoch": 2.485862665897288,
      "grad_norm": 0.0630941167473793,
      "learning_rate": 0.00010074074074074073,
      "loss": 0.1057,
      "step": 1077
    },
    {
      "epoch": 2.4881708020773226,
      "grad_norm": 0.0649518147110939,
      "learning_rate": 0.00010064814814814816,
      "loss": 0.1092,
      "step": 1078
    },
    {
      "epoch": 2.490478938257357,
      "grad_norm": 0.0614495687186718,
      "learning_rate": 0.00010055555555555555,
      "loss": 0.1148,
      "step": 1079
    },
    {
      "epoch": 2.4927870744373917,
      "grad_norm": 0.0600539930164814,
      "learning_rate": 0.00010046296296296296,
      "loss": 0.1177,
      "step": 1080
    },
    {
      "epoch": 2.4950952106174267,
      "grad_norm": 0.06395402550697327,
      "learning_rate": 0.00010037037037037038,
      "loss": 0.1087,
      "step": 1081
    },
    {
      "epoch": 2.4974033467974612,
      "grad_norm": 0.060378894209861755,
      "learning_rate": 0.00010027777777777779,
      "loss": 0.126,
      "step": 1082
    },
    {
      "epoch": 2.499711482977496,
      "grad_norm": 0.06973925232887268,
      "learning_rate": 0.00010018518518518518,
      "loss": 0.1058,
      "step": 1083
    },
    {
      "epoch": 2.5020196191575304,
      "grad_norm": 0.06307213753461838,
      "learning_rate": 0.00010009259259259261,
      "loss": 0.1166,
      "step": 1084
    },
    {
      "epoch": 2.504327755337565,
      "grad_norm": 0.06792129576206207,
      "learning_rate": 0.0001,
      "loss": 0.1063,
      "step": 1085
    },
    {
      "epoch": 2.5066358915175995,
      "grad_norm": 0.05908680334687233,
      "learning_rate": 9.990740740740742e-05,
      "loss": 0.1205,
      "step": 1086
    },
    {
      "epoch": 2.508944027697634,
      "grad_norm": 0.06841656565666199,
      "learning_rate": 9.981481481481481e-05,
      "loss": 0.1119,
      "step": 1087
    },
    {
      "epoch": 2.5112521638776686,
      "grad_norm": 0.0636344775557518,
      "learning_rate": 9.972222222222222e-05,
      "loss": 0.1183,
      "step": 1088
    },
    {
      "epoch": 2.513560300057703,
      "grad_norm": 0.06488758325576782,
      "learning_rate": 9.962962962962963e-05,
      "loss": 0.1114,
      "step": 1089
    },
    {
      "epoch": 2.515868436237738,
      "grad_norm": 0.0658671110868454,
      "learning_rate": 9.953703703703704e-05,
      "loss": 0.1183,
      "step": 1090
    },
    {
      "epoch": 2.5181765724177727,
      "grad_norm": 0.06591572612524033,
      "learning_rate": 9.944444444444446e-05,
      "loss": 0.1116,
      "step": 1091
    },
    {
      "epoch": 2.5204847085978073,
      "grad_norm": 0.06592145562171936,
      "learning_rate": 9.935185185185185e-05,
      "loss": 0.1148,
      "step": 1092
    },
    {
      "epoch": 2.522792844777842,
      "grad_norm": 0.06672943383455276,
      "learning_rate": 9.925925925925926e-05,
      "loss": 0.1089,
      "step": 1093
    },
    {
      "epoch": 2.5251009809578764,
      "grad_norm": 0.0707940086722374,
      "learning_rate": 9.916666666666667e-05,
      "loss": 0.1085,
      "step": 1094
    },
    {
      "epoch": 2.5274091171379114,
      "grad_norm": 0.08122927695512772,
      "learning_rate": 9.907407407407407e-05,
      "loss": 0.1154,
      "step": 1095
    },
    {
      "epoch": 2.529717253317946,
      "grad_norm": 0.06536000967025757,
      "learning_rate": 9.898148148148148e-05,
      "loss": 0.1188,
      "step": 1096
    },
    {
      "epoch": 2.5320253894979805,
      "grad_norm": 0.07824065536260605,
      "learning_rate": 9.888888888888889e-05,
      "loss": 0.1099,
      "step": 1097
    },
    {
      "epoch": 2.534333525678015,
      "grad_norm": 0.06915551424026489,
      "learning_rate": 9.87962962962963e-05,
      "loss": 0.1093,
      "step": 1098
    },
    {
      "epoch": 2.5366416618580496,
      "grad_norm": 0.06488841772079468,
      "learning_rate": 9.870370370370371e-05,
      "loss": 0.1141,
      "step": 1099
    },
    {
      "epoch": 2.538949798038084,
      "grad_norm": 0.06511961668729782,
      "learning_rate": 9.861111111111112e-05,
      "loss": 0.1127,
      "step": 1100
    },
    {
      "epoch": 2.5412579342181187,
      "grad_norm": 0.06521959602832794,
      "learning_rate": 9.851851851851852e-05,
      "loss": 0.1074,
      "step": 1101
    },
    {
      "epoch": 2.5435660703981533,
      "grad_norm": 0.07333683967590332,
      "learning_rate": 9.842592592592593e-05,
      "loss": 0.1149,
      "step": 1102
    },
    {
      "epoch": 2.545874206578188,
      "grad_norm": 0.06894991546869278,
      "learning_rate": 9.833333333333333e-05,
      "loss": 0.1091,
      "step": 1103
    },
    {
      "epoch": 2.548182342758223,
      "grad_norm": 0.06677638739347458,
      "learning_rate": 9.824074074074074e-05,
      "loss": 0.1135,
      "step": 1104
    },
    {
      "epoch": 2.5504904789382574,
      "grad_norm": 0.06655356287956238,
      "learning_rate": 9.814814814814815e-05,
      "loss": 0.1071,
      "step": 1105
    },
    {
      "epoch": 2.552798615118292,
      "grad_norm": 0.06044694036245346,
      "learning_rate": 9.805555555555556e-05,
      "loss": 0.1133,
      "step": 1106
    },
    {
      "epoch": 2.5551067512983265,
      "grad_norm": 0.07503234595060349,
      "learning_rate": 9.796296296296297e-05,
      "loss": 0.1086,
      "step": 1107
    },
    {
      "epoch": 2.557414887478361,
      "grad_norm": 0.06137268245220184,
      "learning_rate": 9.787037037037038e-05,
      "loss": 0.1084,
      "step": 1108
    },
    {
      "epoch": 2.559723023658396,
      "grad_norm": 0.06950575858354568,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.111,
      "step": 1109
    },
    {
      "epoch": 2.5620311598384307,
      "grad_norm": 0.06661064177751541,
      "learning_rate": 9.768518518518519e-05,
      "loss": 0.1099,
      "step": 1110
    },
    {
      "epoch": 2.564339296018465,
      "grad_norm": 0.07339566946029663,
      "learning_rate": 9.75925925925926e-05,
      "loss": 0.1093,
      "step": 1111
    },
    {
      "epoch": 2.5666474321985,
      "grad_norm": 0.07042503356933594,
      "learning_rate": 9.75e-05,
      "loss": 0.1114,
      "step": 1112
    },
    {
      "epoch": 2.5689555683785343,
      "grad_norm": 0.07155728340148926,
      "learning_rate": 9.740740740740741e-05,
      "loss": 0.1069,
      "step": 1113
    },
    {
      "epoch": 2.571263704558569,
      "grad_norm": 0.06994260847568512,
      "learning_rate": 9.731481481481482e-05,
      "loss": 0.1053,
      "step": 1114
    },
    {
      "epoch": 2.5735718407386035,
      "grad_norm": 0.06538574397563934,
      "learning_rate": 9.722222222222223e-05,
      "loss": 0.1058,
      "step": 1115
    },
    {
      "epoch": 2.575879976918638,
      "grad_norm": 0.07100135833024979,
      "learning_rate": 9.712962962962964e-05,
      "loss": 0.1101,
      "step": 1116
    },
    {
      "epoch": 2.5781881130986726,
      "grad_norm": 0.06175816059112549,
      "learning_rate": 9.703703703703704e-05,
      "loss": 0.1136,
      "step": 1117
    },
    {
      "epoch": 2.5804962492787076,
      "grad_norm": 0.07105552405118942,
      "learning_rate": 9.694444444444445e-05,
      "loss": 0.1076,
      "step": 1118
    },
    {
      "epoch": 2.582804385458742,
      "grad_norm": 0.06268072128295898,
      "learning_rate": 9.685185185185186e-05,
      "loss": 0.1134,
      "step": 1119
    },
    {
      "epoch": 2.5851125216387767,
      "grad_norm": 0.06977255642414093,
      "learning_rate": 9.675925925925926e-05,
      "loss": 0.1063,
      "step": 1120
    },
    {
      "epoch": 2.5874206578188113,
      "grad_norm": 0.064189113676548,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.1102,
      "step": 1121
    },
    {
      "epoch": 2.589728793998846,
      "grad_norm": 0.06389067322015762,
      "learning_rate": 9.657407407407408e-05,
      "loss": 0.113,
      "step": 1122
    },
    {
      "epoch": 2.592036930178881,
      "grad_norm": 0.07223134487867355,
      "learning_rate": 9.648148148148149e-05,
      "loss": 0.1125,
      "step": 1123
    },
    {
      "epoch": 2.5943450663589154,
      "grad_norm": 0.06277281790971756,
      "learning_rate": 9.63888888888889e-05,
      "loss": 0.1054,
      "step": 1124
    },
    {
      "epoch": 2.59665320253895,
      "grad_norm": 0.06246844306588173,
      "learning_rate": 9.62962962962963e-05,
      "loss": 0.1127,
      "step": 1125
    },
    {
      "epoch": 2.5989613387189845,
      "grad_norm": 0.06318633258342743,
      "learning_rate": 9.620370370370371e-05,
      "loss": 0.1115,
      "step": 1126
    },
    {
      "epoch": 2.601269474899019,
      "grad_norm": 0.06539614498615265,
      "learning_rate": 9.611111111111112e-05,
      "loss": 0.1053,
      "step": 1127
    },
    {
      "epoch": 2.6035776110790536,
      "grad_norm": 0.06377933919429779,
      "learning_rate": 9.601851851851851e-05,
      "loss": 0.1174,
      "step": 1128
    },
    {
      "epoch": 2.605885747259088,
      "grad_norm": 0.06457851827144623,
      "learning_rate": 9.592592592592593e-05,
      "loss": 0.1077,
      "step": 1129
    },
    {
      "epoch": 2.6081938834391227,
      "grad_norm": 0.06904317438602448,
      "learning_rate": 9.583333333333334e-05,
      "loss": 0.1105,
      "step": 1130
    },
    {
      "epoch": 2.6105020196191573,
      "grad_norm": 0.0636373981833458,
      "learning_rate": 9.574074074074075e-05,
      "loss": 0.1045,
      "step": 1131
    },
    {
      "epoch": 2.6128101557991923,
      "grad_norm": 0.06901761889457703,
      "learning_rate": 9.564814814814816e-05,
      "loss": 0.1128,
      "step": 1132
    },
    {
      "epoch": 2.615118291979227,
      "grad_norm": 0.06473273038864136,
      "learning_rate": 9.555555555555557e-05,
      "loss": 0.1103,
      "step": 1133
    },
    {
      "epoch": 2.6174264281592614,
      "grad_norm": 0.059912361204624176,
      "learning_rate": 9.546296296296297e-05,
      "loss": 0.1141,
      "step": 1134
    },
    {
      "epoch": 2.619734564339296,
      "grad_norm": 0.06659095734357834,
      "learning_rate": 9.537037037037038e-05,
      "loss": 0.112,
      "step": 1135
    },
    {
      "epoch": 2.6220427005193305,
      "grad_norm": 0.06047660857439041,
      "learning_rate": 9.527777777777777e-05,
      "loss": 0.11,
      "step": 1136
    },
    {
      "epoch": 2.6243508366993655,
      "grad_norm": 0.07176803052425385,
      "learning_rate": 9.518518518518518e-05,
      "loss": 0.1048,
      "step": 1137
    },
    {
      "epoch": 2.6266589728794,
      "grad_norm": 0.06270527094602585,
      "learning_rate": 9.50925925925926e-05,
      "loss": 0.1059,
      "step": 1138
    },
    {
      "epoch": 2.6289671090594346,
      "grad_norm": 0.06521376222372055,
      "learning_rate": 9.5e-05,
      "loss": 0.1151,
      "step": 1139
    },
    {
      "epoch": 2.631275245239469,
      "grad_norm": 0.0692153126001358,
      "learning_rate": 9.490740740740742e-05,
      "loss": 0.1039,
      "step": 1140
    },
    {
      "epoch": 2.6335833814195038,
      "grad_norm": 0.06468004733324051,
      "learning_rate": 9.481481481481483e-05,
      "loss": 0.1089,
      "step": 1141
    },
    {
      "epoch": 2.6358915175995383,
      "grad_norm": 0.06504610925912857,
      "learning_rate": 9.472222222222222e-05,
      "loss": 0.1094,
      "step": 1142
    },
    {
      "epoch": 2.638199653779573,
      "grad_norm": 0.06548403203487396,
      "learning_rate": 9.462962962962963e-05,
      "loss": 0.1064,
      "step": 1143
    },
    {
      "epoch": 2.6405077899596074,
      "grad_norm": 0.06968478858470917,
      "learning_rate": 9.453703703703705e-05,
      "loss": 0.1083,
      "step": 1144
    },
    {
      "epoch": 2.642815926139642,
      "grad_norm": 0.06335204839706421,
      "learning_rate": 9.444444444444444e-05,
      "loss": 0.1113,
      "step": 1145
    },
    {
      "epoch": 2.645124062319677,
      "grad_norm": 0.06333167105913162,
      "learning_rate": 9.435185185185185e-05,
      "loss": 0.1139,
      "step": 1146
    },
    {
      "epoch": 2.6474321984997116,
      "grad_norm": 0.06525623798370361,
      "learning_rate": 9.425925925925926e-05,
      "loss": 0.1134,
      "step": 1147
    },
    {
      "epoch": 2.649740334679746,
      "grad_norm": 0.06319650262594223,
      "learning_rate": 9.416666666666667e-05,
      "loss": 0.1185,
      "step": 1148
    },
    {
      "epoch": 2.6520484708597807,
      "grad_norm": 0.06441297382116318,
      "learning_rate": 9.407407407407408e-05,
      "loss": 0.1103,
      "step": 1149
    },
    {
      "epoch": 2.6543566070398152,
      "grad_norm": 0.06346457451581955,
      "learning_rate": 9.398148148148148e-05,
      "loss": 0.1164,
      "step": 1150
    },
    {
      "epoch": 2.6566647432198502,
      "grad_norm": 0.066008560359478,
      "learning_rate": 9.388888888888889e-05,
      "loss": 0.1143,
      "step": 1151
    },
    {
      "epoch": 2.658972879399885,
      "grad_norm": 0.06161175295710564,
      "learning_rate": 9.37962962962963e-05,
      "loss": 0.1118,
      "step": 1152
    },
    {
      "epoch": 2.6612810155799194,
      "grad_norm": 0.06405019760131836,
      "learning_rate": 9.37037037037037e-05,
      "loss": 0.1145,
      "step": 1153
    },
    {
      "epoch": 2.663589151759954,
      "grad_norm": 0.05822325870394707,
      "learning_rate": 9.361111111111111e-05,
      "loss": 0.1134,
      "step": 1154
    },
    {
      "epoch": 2.6658972879399885,
      "grad_norm": 0.06162920221686363,
      "learning_rate": 9.351851851851852e-05,
      "loss": 0.115,
      "step": 1155
    },
    {
      "epoch": 2.668205424120023,
      "grad_norm": 0.06294470280408859,
      "learning_rate": 9.342592592592593e-05,
      "loss": 0.1172,
      "step": 1156
    },
    {
      "epoch": 2.6705135603000576,
      "grad_norm": 0.06293560564517975,
      "learning_rate": 9.333333333333334e-05,
      "loss": 0.1051,
      "step": 1157
    },
    {
      "epoch": 2.672821696480092,
      "grad_norm": 0.06386113166809082,
      "learning_rate": 9.324074074074075e-05,
      "loss": 0.1137,
      "step": 1158
    },
    {
      "epoch": 2.6751298326601267,
      "grad_norm": 0.062412094324827194,
      "learning_rate": 9.314814814814815e-05,
      "loss": 0.1128,
      "step": 1159
    },
    {
      "epoch": 2.6774379688401617,
      "grad_norm": 0.06267069280147552,
      "learning_rate": 9.305555555555556e-05,
      "loss": 0.1068,
      "step": 1160
    },
    {
      "epoch": 2.6797461050201963,
      "grad_norm": 0.06829304248094559,
      "learning_rate": 9.296296296296296e-05,
      "loss": 0.1122,
      "step": 1161
    },
    {
      "epoch": 2.682054241200231,
      "grad_norm": 0.06650856137275696,
      "learning_rate": 9.287037037037037e-05,
      "loss": 0.1116,
      "step": 1162
    },
    {
      "epoch": 2.6843623773802654,
      "grad_norm": 0.06360474973917007,
      "learning_rate": 9.277777777777778e-05,
      "loss": 0.1077,
      "step": 1163
    },
    {
      "epoch": 2.6866705135603,
      "grad_norm": 0.0665355771780014,
      "learning_rate": 9.268518518518519e-05,
      "loss": 0.1003,
      "step": 1164
    },
    {
      "epoch": 2.6889786497403345,
      "grad_norm": 0.06642794609069824,
      "learning_rate": 9.25925925925926e-05,
      "loss": 0.1036,
      "step": 1165
    },
    {
      "epoch": 2.6912867859203695,
      "grad_norm": 0.06546643376350403,
      "learning_rate": 9.250000000000001e-05,
      "loss": 0.1125,
      "step": 1166
    },
    {
      "epoch": 2.693594922100404,
      "grad_norm": 0.06344105303287506,
      "learning_rate": 9.240740740740741e-05,
      "loss": 0.1041,
      "step": 1167
    },
    {
      "epoch": 2.6959030582804386,
      "grad_norm": 0.06820191442966461,
      "learning_rate": 9.231481481481482e-05,
      "loss": 0.1047,
      "step": 1168
    },
    {
      "epoch": 2.698211194460473,
      "grad_norm": 0.06149286776781082,
      "learning_rate": 9.222222222222223e-05,
      "loss": 0.1167,
      "step": 1169
    },
    {
      "epoch": 2.7005193306405078,
      "grad_norm": 0.0714329183101654,
      "learning_rate": 9.212962962962963e-05,
      "loss": 0.1134,
      "step": 1170
    },
    {
      "epoch": 2.7028274668205423,
      "grad_norm": 0.062424175441265106,
      "learning_rate": 9.203703703703704e-05,
      "loss": 0.1152,
      "step": 1171
    },
    {
      "epoch": 2.705135603000577,
      "grad_norm": 0.06830987334251404,
      "learning_rate": 9.194444444444445e-05,
      "loss": 0.107,
      "step": 1172
    },
    {
      "epoch": 2.7074437391806114,
      "grad_norm": 0.06229596212506294,
      "learning_rate": 9.185185185185186e-05,
      "loss": 0.1125,
      "step": 1173
    },
    {
      "epoch": 2.7097518753606464,
      "grad_norm": 0.07005751878023148,
      "learning_rate": 9.175925925925927e-05,
      "loss": 0.1136,
      "step": 1174
    },
    {
      "epoch": 2.712060011540681,
      "grad_norm": 0.06210910528898239,
      "learning_rate": 9.166666666666667e-05,
      "loss": 0.1131,
      "step": 1175
    },
    {
      "epoch": 2.7143681477207156,
      "grad_norm": 0.0742712989449501,
      "learning_rate": 9.157407407407408e-05,
      "loss": 0.1013,
      "step": 1176
    },
    {
      "epoch": 2.71667628390075,
      "grad_norm": 0.0671953409910202,
      "learning_rate": 9.148148148148149e-05,
      "loss": 0.1054,
      "step": 1177
    },
    {
      "epoch": 2.7189844200807847,
      "grad_norm": 0.06601515412330627,
      "learning_rate": 9.138888888888889e-05,
      "loss": 0.1109,
      "step": 1178
    },
    {
      "epoch": 2.7212925562608192,
      "grad_norm": 0.06075047329068184,
      "learning_rate": 9.12962962962963e-05,
      "loss": 0.1058,
      "step": 1179
    },
    {
      "epoch": 2.7236006924408542,
      "grad_norm": 0.06500991433858871,
      "learning_rate": 9.120370370370371e-05,
      "loss": 0.1069,
      "step": 1180
    },
    {
      "epoch": 2.725908828620889,
      "grad_norm": 0.06533980369567871,
      "learning_rate": 9.111111111111112e-05,
      "loss": 0.1164,
      "step": 1181
    },
    {
      "epoch": 2.7282169648009233,
      "grad_norm": 0.06360896676778793,
      "learning_rate": 9.101851851851853e-05,
      "loss": 0.1088,
      "step": 1182
    },
    {
      "epoch": 2.730525100980958,
      "grad_norm": 0.06060441955924034,
      "learning_rate": 9.092592592592593e-05,
      "loss": 0.1116,
      "step": 1183
    },
    {
      "epoch": 2.7328332371609925,
      "grad_norm": 0.06768076121807098,
      "learning_rate": 9.083333333333334e-05,
      "loss": 0.1071,
      "step": 1184
    },
    {
      "epoch": 2.735141373341027,
      "grad_norm": 0.06665431708097458,
      "learning_rate": 9.074074074074075e-05,
      "loss": 0.1156,
      "step": 1185
    },
    {
      "epoch": 2.7374495095210616,
      "grad_norm": 0.07042966783046722,
      "learning_rate": 9.064814814814814e-05,
      "loss": 0.1085,
      "step": 1186
    },
    {
      "epoch": 2.739757645701096,
      "grad_norm": 0.06716679781675339,
      "learning_rate": 9.055555555555556e-05,
      "loss": 0.1175,
      "step": 1187
    },
    {
      "epoch": 2.742065781881131,
      "grad_norm": 0.07337996363639832,
      "learning_rate": 9.046296296296297e-05,
      "loss": 0.1124,
      "step": 1188
    },
    {
      "epoch": 2.7443739180611657,
      "grad_norm": 0.06556232273578644,
      "learning_rate": 9.037037037037038e-05,
      "loss": 0.1061,
      "step": 1189
    },
    {
      "epoch": 2.7466820542412003,
      "grad_norm": 0.07632841914892197,
      "learning_rate": 9.027777777777779e-05,
      "loss": 0.1045,
      "step": 1190
    },
    {
      "epoch": 2.748990190421235,
      "grad_norm": 0.058858297765254974,
      "learning_rate": 9.01851851851852e-05,
      "loss": 0.1095,
      "step": 1191
    },
    {
      "epoch": 2.7512983266012694,
      "grad_norm": 0.0767543613910675,
      "learning_rate": 9.00925925925926e-05,
      "loss": 0.1097,
      "step": 1192
    },
    {
      "epoch": 2.753606462781304,
      "grad_norm": 0.06769121438264847,
      "learning_rate": 9e-05,
      "loss": 0.108,
      "step": 1193
    },
    {
      "epoch": 2.755914598961339,
      "grad_norm": 0.07070609927177429,
      "learning_rate": 8.99074074074074e-05,
      "loss": 0.1097,
      "step": 1194
    },
    {
      "epoch": 2.7582227351413735,
      "grad_norm": 0.06450021266937256,
      "learning_rate": 8.981481481481481e-05,
      "loss": 0.1035,
      "step": 1195
    },
    {
      "epoch": 2.760530871321408,
      "grad_norm": 0.06677190959453583,
      "learning_rate": 8.972222222222222e-05,
      "loss": 0.1161,
      "step": 1196
    },
    {
      "epoch": 2.7628390075014426,
      "grad_norm": 0.06439580023288727,
      "learning_rate": 8.962962962962963e-05,
      "loss": 0.1145,
      "step": 1197
    },
    {
      "epoch": 2.765147143681477,
      "grad_norm": 0.06436222791671753,
      "learning_rate": 8.953703703703705e-05,
      "loss": 0.1111,
      "step": 1198
    },
    {
      "epoch": 2.7674552798615117,
      "grad_norm": 0.06407810747623444,
      "learning_rate": 8.944444444444446e-05,
      "loss": 0.1095,
      "step": 1199
    },
    {
      "epoch": 2.7697634160415463,
      "grad_norm": 0.061969999223947525,
      "learning_rate": 8.935185185185185e-05,
      "loss": 0.1057,
      "step": 1200
    },
    {
      "epoch": 2.772071552221581,
      "grad_norm": 0.07271505147218704,
      "learning_rate": 8.925925925925926e-05,
      "loss": 0.105,
      "step": 1201
    },
    {
      "epoch": 2.7743796884016154,
      "grad_norm": 0.0687282532453537,
      "learning_rate": 8.916666666666667e-05,
      "loss": 0.111,
      "step": 1202
    },
    {
      "epoch": 2.7766878245816504,
      "grad_norm": 0.06695423275232315,
      "learning_rate": 8.907407407407407e-05,
      "loss": 0.1166,
      "step": 1203
    },
    {
      "epoch": 2.778995960761685,
      "grad_norm": 0.06060202792286873,
      "learning_rate": 8.898148148148148e-05,
      "loss": 0.1075,
      "step": 1204
    },
    {
      "epoch": 2.7813040969417195,
      "grad_norm": 0.07143843173980713,
      "learning_rate": 8.888888888888889e-05,
      "loss": 0.1094,
      "step": 1205
    },
    {
      "epoch": 2.783612233121754,
      "grad_norm": 0.06091539189219475,
      "learning_rate": 8.87962962962963e-05,
      "loss": 0.1069,
      "step": 1206
    },
    {
      "epoch": 2.7859203693017887,
      "grad_norm": 0.0703229308128357,
      "learning_rate": 8.870370370370371e-05,
      "loss": 0.1148,
      "step": 1207
    },
    {
      "epoch": 2.7882285054818237,
      "grad_norm": 0.06406431645154953,
      "learning_rate": 8.861111111111111e-05,
      "loss": 0.1046,
      "step": 1208
    },
    {
      "epoch": 2.790536641661858,
      "grad_norm": 0.07411704212427139,
      "learning_rate": 8.851851851851852e-05,
      "loss": 0.1097,
      "step": 1209
    },
    {
      "epoch": 2.792844777841893,
      "grad_norm": 0.06640252470970154,
      "learning_rate": 8.842592592592593e-05,
      "loss": 0.1051,
      "step": 1210
    },
    {
      "epoch": 2.7951529140219273,
      "grad_norm": 0.06520389020442963,
      "learning_rate": 8.833333333333333e-05,
      "loss": 0.1138,
      "step": 1211
    },
    {
      "epoch": 2.797461050201962,
      "grad_norm": 0.06444036960601807,
      "learning_rate": 8.824074074074074e-05,
      "loss": 0.1059,
      "step": 1212
    },
    {
      "epoch": 2.7997691863819965,
      "grad_norm": 0.0675966739654541,
      "learning_rate": 8.814814814814815e-05,
      "loss": 0.109,
      "step": 1213
    },
    {
      "epoch": 2.802077322562031,
      "grad_norm": 0.06585707515478134,
      "learning_rate": 8.805555555555556e-05,
      "loss": 0.1032,
      "step": 1214
    },
    {
      "epoch": 2.8043854587420656,
      "grad_norm": 0.06416294723749161,
      "learning_rate": 8.796296296296297e-05,
      "loss": 0.1039,
      "step": 1215
    },
    {
      "epoch": 2.8066935949221,
      "grad_norm": 0.0661073625087738,
      "learning_rate": 8.787037037037037e-05,
      "loss": 0.112,
      "step": 1216
    },
    {
      "epoch": 2.809001731102135,
      "grad_norm": 0.06835968792438507,
      "learning_rate": 8.777777777777778e-05,
      "loss": 0.1155,
      "step": 1217
    },
    {
      "epoch": 2.8113098672821697,
      "grad_norm": 0.06868423521518707,
      "learning_rate": 8.768518518518519e-05,
      "loss": 0.106,
      "step": 1218
    },
    {
      "epoch": 2.8136180034622043,
      "grad_norm": 0.063177689909935,
      "learning_rate": 8.759259259259259e-05,
      "loss": 0.1073,
      "step": 1219
    },
    {
      "epoch": 2.815926139642239,
      "grad_norm": 0.0732072964310646,
      "learning_rate": 8.75e-05,
      "loss": 0.1105,
      "step": 1220
    },
    {
      "epoch": 2.8182342758222734,
      "grad_norm": 0.06565076112747192,
      "learning_rate": 8.740740740740741e-05,
      "loss": 0.1129,
      "step": 1221
    },
    {
      "epoch": 2.8205424120023084,
      "grad_norm": 0.07367955893278122,
      "learning_rate": 8.731481481481482e-05,
      "loss": 0.105,
      "step": 1222
    },
    {
      "epoch": 2.822850548182343,
      "grad_norm": 0.06678752601146698,
      "learning_rate": 8.722222222222223e-05,
      "loss": 0.1098,
      "step": 1223
    },
    {
      "epoch": 2.8251586843623775,
      "grad_norm": 0.07375338673591614,
      "learning_rate": 8.712962962962964e-05,
      "loss": 0.1093,
      "step": 1224
    },
    {
      "epoch": 2.827466820542412,
      "grad_norm": 0.062444258481264114,
      "learning_rate": 8.703703703703704e-05,
      "loss": 0.1059,
      "step": 1225
    },
    {
      "epoch": 2.8297749567224466,
      "grad_norm": 0.06839457154273987,
      "learning_rate": 8.694444444444445e-05,
      "loss": 0.1113,
      "step": 1226
    },
    {
      "epoch": 2.832083092902481,
      "grad_norm": 0.06373050808906555,
      "learning_rate": 8.685185185185185e-05,
      "loss": 0.1096,
      "step": 1227
    },
    {
      "epoch": 2.8343912290825157,
      "grad_norm": 0.06499390304088593,
      "learning_rate": 8.675925925925926e-05,
      "loss": 0.1066,
      "step": 1228
    },
    {
      "epoch": 2.8366993652625503,
      "grad_norm": 0.062489721924066544,
      "learning_rate": 8.666666666666667e-05,
      "loss": 0.1187,
      "step": 1229
    },
    {
      "epoch": 2.839007501442585,
      "grad_norm": 0.0639164075255394,
      "learning_rate": 8.657407407407408e-05,
      "loss": 0.1114,
      "step": 1230
    },
    {
      "epoch": 2.84131563762262,
      "grad_norm": 0.06185689568519592,
      "learning_rate": 8.648148148148149e-05,
      "loss": 0.1179,
      "step": 1231
    },
    {
      "epoch": 2.8436237738026544,
      "grad_norm": 0.060767561197280884,
      "learning_rate": 8.63888888888889e-05,
      "loss": 0.107,
      "step": 1232
    },
    {
      "epoch": 2.845931909982689,
      "grad_norm": 0.06271563470363617,
      "learning_rate": 8.62962962962963e-05,
      "loss": 0.1114,
      "step": 1233
    },
    {
      "epoch": 2.8482400461627235,
      "grad_norm": 0.056873708963394165,
      "learning_rate": 8.620370370370371e-05,
      "loss": 0.1107,
      "step": 1234
    },
    {
      "epoch": 2.850548182342758,
      "grad_norm": 0.06316443532705307,
      "learning_rate": 8.611111111111112e-05,
      "loss": 0.1128,
      "step": 1235
    },
    {
      "epoch": 2.852856318522793,
      "grad_norm": 0.06665301322937012,
      "learning_rate": 8.601851851851852e-05,
      "loss": 0.1164,
      "step": 1236
    },
    {
      "epoch": 2.8551644547028276,
      "grad_norm": 0.06328419595956802,
      "learning_rate": 8.592592592592593e-05,
      "loss": 0.1124,
      "step": 1237
    },
    {
      "epoch": 2.857472590882862,
      "grad_norm": 0.06157767400145531,
      "learning_rate": 8.583333333333334e-05,
      "loss": 0.1046,
      "step": 1238
    },
    {
      "epoch": 2.8597807270628968,
      "grad_norm": 0.06175469607114792,
      "learning_rate": 8.574074074074075e-05,
      "loss": 0.1089,
      "step": 1239
    },
    {
      "epoch": 2.8620888632429313,
      "grad_norm": 0.06224953755736351,
      "learning_rate": 8.564814814814816e-05,
      "loss": 0.1117,
      "step": 1240
    },
    {
      "epoch": 2.864396999422966,
      "grad_norm": 0.0656314343214035,
      "learning_rate": 8.555555555555556e-05,
      "loss": 0.116,
      "step": 1241
    },
    {
      "epoch": 2.8667051356030004,
      "grad_norm": 0.06636226177215576,
      "learning_rate": 8.546296296296297e-05,
      "loss": 0.1122,
      "step": 1242
    },
    {
      "epoch": 2.869013271783035,
      "grad_norm": 0.06484440714120865,
      "learning_rate": 8.537037037037038e-05,
      "loss": 0.1116,
      "step": 1243
    },
    {
      "epoch": 2.8713214079630696,
      "grad_norm": 0.06339502334594727,
      "learning_rate": 8.527777777777777e-05,
      "loss": 0.1113,
      "step": 1244
    },
    {
      "epoch": 2.8736295441431046,
      "grad_norm": 0.06438542902469635,
      "learning_rate": 8.518518518518518e-05,
      "loss": 0.1096,
      "step": 1245
    },
    {
      "epoch": 2.875937680323139,
      "grad_norm": 0.06592599302530289,
      "learning_rate": 8.50925925925926e-05,
      "loss": 0.1149,
      "step": 1246
    },
    {
      "epoch": 2.8782458165031737,
      "grad_norm": 0.06493037939071655,
      "learning_rate": 8.5e-05,
      "loss": 0.1136,
      "step": 1247
    },
    {
      "epoch": 2.8805539526832082,
      "grad_norm": 0.06346163153648376,
      "learning_rate": 8.490740740740742e-05,
      "loss": 0.1073,
      "step": 1248
    },
    {
      "epoch": 2.882862088863243,
      "grad_norm": 0.06138644739985466,
      "learning_rate": 8.481481481481481e-05,
      "loss": 0.1082,
      "step": 1249
    },
    {
      "epoch": 2.885170225043278,
      "grad_norm": 0.06746942549943924,
      "learning_rate": 8.472222222222222e-05,
      "loss": 0.1169,
      "step": 1250
    },
    {
      "epoch": 2.8874783612233124,
      "grad_norm": 0.0599314346909523,
      "learning_rate": 8.462962962962964e-05,
      "loss": 0.1087,
      "step": 1251
    },
    {
      "epoch": 2.889786497403347,
      "grad_norm": 0.06530068069696426,
      "learning_rate": 8.453703703703703e-05,
      "loss": 0.1134,
      "step": 1252
    },
    {
      "epoch": 2.8920946335833815,
      "grad_norm": 0.06346049159765244,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.1044,
      "step": 1253
    },
    {
      "epoch": 2.894402769763416,
      "grad_norm": 0.06495968997478485,
      "learning_rate": 8.435185185185185e-05,
      "loss": 0.114,
      "step": 1254
    },
    {
      "epoch": 2.8967109059434506,
      "grad_norm": 0.06569454073905945,
      "learning_rate": 8.425925925925926e-05,
      "loss": 0.103,
      "step": 1255
    },
    {
      "epoch": 2.899019042123485,
      "grad_norm": 0.06635311245918274,
      "learning_rate": 8.416666666666668e-05,
      "loss": 0.1069,
      "step": 1256
    },
    {
      "epoch": 2.9013271783035197,
      "grad_norm": 0.0636238157749176,
      "learning_rate": 8.407407407407409e-05,
      "loss": 0.11,
      "step": 1257
    },
    {
      "epoch": 2.9036353144835543,
      "grad_norm": 0.06675776839256287,
      "learning_rate": 8.398148148148148e-05,
      "loss": 0.1104,
      "step": 1258
    },
    {
      "epoch": 2.9059434506635893,
      "grad_norm": 0.06708743423223495,
      "learning_rate": 8.38888888888889e-05,
      "loss": 0.1072,
      "step": 1259
    },
    {
      "epoch": 2.908251586843624,
      "grad_norm": 0.06758449226617813,
      "learning_rate": 8.379629629629629e-05,
      "loss": 0.1111,
      "step": 1260
    },
    {
      "epoch": 2.9105597230236584,
      "grad_norm": 0.0643402710556984,
      "learning_rate": 8.37037037037037e-05,
      "loss": 0.1092,
      "step": 1261
    },
    {
      "epoch": 2.912867859203693,
      "grad_norm": 0.06790553778409958,
      "learning_rate": 8.361111111111111e-05,
      "loss": 0.1173,
      "step": 1262
    },
    {
      "epoch": 2.9151759953837275,
      "grad_norm": 0.06296567618846893,
      "learning_rate": 8.351851851851852e-05,
      "loss": 0.1101,
      "step": 1263
    },
    {
      "epoch": 2.9174841315637625,
      "grad_norm": 0.07278038561344147,
      "learning_rate": 8.342592592592593e-05,
      "loss": 0.104,
      "step": 1264
    },
    {
      "epoch": 2.919792267743797,
      "grad_norm": 0.06712595373392105,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.1158,
      "step": 1265
    },
    {
      "epoch": 2.9221004039238316,
      "grad_norm": 0.0677039846777916,
      "learning_rate": 8.324074074074074e-05,
      "loss": 0.1041,
      "step": 1266
    },
    {
      "epoch": 2.924408540103866,
      "grad_norm": 0.0700831413269043,
      "learning_rate": 8.314814814814815e-05,
      "loss": 0.1021,
      "step": 1267
    },
    {
      "epoch": 2.9267166762839008,
      "grad_norm": 0.06155812740325928,
      "learning_rate": 8.305555555555556e-05,
      "loss": 0.1093,
      "step": 1268
    },
    {
      "epoch": 2.9290248124639353,
      "grad_norm": 0.0629374235868454,
      "learning_rate": 8.296296296296296e-05,
      "loss": 0.1117,
      "step": 1269
    },
    {
      "epoch": 2.93133294864397,
      "grad_norm": 0.06255346536636353,
      "learning_rate": 8.287037037037037e-05,
      "loss": 0.1112,
      "step": 1270
    },
    {
      "epoch": 2.9336410848240044,
      "grad_norm": 0.06587310880422592,
      "learning_rate": 8.277777777777778e-05,
      "loss": 0.1137,
      "step": 1271
    },
    {
      "epoch": 2.935949221004039,
      "grad_norm": 0.06792567670345306,
      "learning_rate": 8.268518518518519e-05,
      "loss": 0.107,
      "step": 1272
    },
    {
      "epoch": 2.938257357184074,
      "grad_norm": 0.07068953663110733,
      "learning_rate": 8.25925925925926e-05,
      "loss": 0.1055,
      "step": 1273
    },
    {
      "epoch": 2.9405654933641086,
      "grad_norm": 0.06486967206001282,
      "learning_rate": 8.25e-05,
      "loss": 0.111,
      "step": 1274
    },
    {
      "epoch": 2.942873629544143,
      "grad_norm": 0.06821799278259277,
      "learning_rate": 8.240740740740741e-05,
      "loss": 0.1162,
      "step": 1275
    },
    {
      "epoch": 2.9451817657241777,
      "grad_norm": 0.06846306473016739,
      "learning_rate": 8.231481481481482e-05,
      "loss": 0.1147,
      "step": 1276
    },
    {
      "epoch": 2.9474899019042122,
      "grad_norm": 0.06513357162475586,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.1004,
      "step": 1277
    },
    {
      "epoch": 2.9497980380842472,
      "grad_norm": 0.06985633075237274,
      "learning_rate": 8.212962962962963e-05,
      "loss": 0.1022,
      "step": 1278
    },
    {
      "epoch": 2.952106174264282,
      "grad_norm": 0.06936328858137131,
      "learning_rate": 8.203703703703704e-05,
      "loss": 0.1115,
      "step": 1279
    },
    {
      "epoch": 2.9544143104443163,
      "grad_norm": 0.06598280370235443,
      "learning_rate": 8.194444444444445e-05,
      "loss": 0.1106,
      "step": 1280
    },
    {
      "epoch": 2.956722446624351,
      "grad_norm": 0.0636172816157341,
      "learning_rate": 8.185185185185186e-05,
      "loss": 0.105,
      "step": 1281
    },
    {
      "epoch": 2.9590305828043855,
      "grad_norm": 0.061999183148145676,
      "learning_rate": 8.175925925925926e-05,
      "loss": 0.1183,
      "step": 1282
    },
    {
      "epoch": 2.96133871898442,
      "grad_norm": 0.06541929394006729,
      "learning_rate": 8.166666666666667e-05,
      "loss": 0.113,
      "step": 1283
    },
    {
      "epoch": 2.9636468551644546,
      "grad_norm": 0.06619175523519516,
      "learning_rate": 8.157407407407408e-05,
      "loss": 0.1067,
      "step": 1284
    },
    {
      "epoch": 2.965954991344489,
      "grad_norm": 0.0694909542798996,
      "learning_rate": 8.148148148148148e-05,
      "loss": 0.1085,
      "step": 1285
    },
    {
      "epoch": 2.9682631275245237,
      "grad_norm": 0.06206699088215828,
      "learning_rate": 8.138888888888889e-05,
      "loss": 0.1059,
      "step": 1286
    },
    {
      "epoch": 2.9705712637045587,
      "grad_norm": 0.0651930719614029,
      "learning_rate": 8.12962962962963e-05,
      "loss": 0.1006,
      "step": 1287
    },
    {
      "epoch": 2.9728793998845933,
      "grad_norm": 0.06138252466917038,
      "learning_rate": 8.120370370370371e-05,
      "loss": 0.1086,
      "step": 1288
    },
    {
      "epoch": 2.975187536064628,
      "grad_norm": 0.07034417241811752,
      "learning_rate": 8.111111111111112e-05,
      "loss": 0.1048,
      "step": 1289
    },
    {
      "epoch": 2.9774956722446624,
      "grad_norm": 0.06150440499186516,
      "learning_rate": 8.101851851851853e-05,
      "loss": 0.103,
      "step": 1290
    },
    {
      "epoch": 2.979803808424697,
      "grad_norm": 0.06783980876207352,
      "learning_rate": 8.092592592592593e-05,
      "loss": 0.1038,
      "step": 1291
    },
    {
      "epoch": 2.982111944604732,
      "grad_norm": 0.06272736936807632,
      "learning_rate": 8.083333333333334e-05,
      "loss": 0.1077,
      "step": 1292
    },
    {
      "epoch": 2.9844200807847665,
      "grad_norm": 0.0640345886349678,
      "learning_rate": 8.074074074074075e-05,
      "loss": 0.1168,
      "step": 1293
    },
    {
      "epoch": 2.986728216964801,
      "grad_norm": 0.06568410992622375,
      "learning_rate": 8.064814814814815e-05,
      "loss": 0.1073,
      "step": 1294
    },
    {
      "epoch": 2.9890363531448356,
      "grad_norm": 0.06602612882852554,
      "learning_rate": 8.055555555555556e-05,
      "loss": 0.1087,
      "step": 1295
    },
    {
      "epoch": 2.99134448932487,
      "grad_norm": 0.06466524302959442,
      "learning_rate": 8.046296296296297e-05,
      "loss": 0.1036,
      "step": 1296
    },
    {
      "epoch": 2.9936526255049047,
      "grad_norm": 0.06462538987398148,
      "learning_rate": 8.037037037037038e-05,
      "loss": 0.1004,
      "step": 1297
    },
    {
      "epoch": 2.9959607616849393,
      "grad_norm": 0.07028472423553467,
      "learning_rate": 8.027777777777779e-05,
      "loss": 0.1019,
      "step": 1298
    },
    {
      "epoch": 2.998268897864974,
      "grad_norm": 0.06252483278512955,
      "learning_rate": 8.018518518518519e-05,
      "loss": 0.1142,
      "step": 1299
    },
    {
      "epoch": 3.000577034045009,
      "grad_norm": 0.0764075294137001,
      "learning_rate": 8.00925925925926e-05,
      "loss": 0.1014,
      "step": 1300
    },
    {
      "epoch": 3.0028851702250434,
      "grad_norm": 0.07154536247253418,
      "learning_rate": 8e-05,
      "loss": 0.1041,
      "step": 1301
    },
    {
      "epoch": 3.005193306405078,
      "grad_norm": 0.06458856165409088,
      "learning_rate": 7.99074074074074e-05,
      "loss": 0.1062,
      "step": 1302
    },
    {
      "epoch": 3.0075014425851125,
      "grad_norm": 0.07254219800233841,
      "learning_rate": 7.981481481481481e-05,
      "loss": 0.1019,
      "step": 1303
    },
    {
      "epoch": 3.009809578765147,
      "grad_norm": 0.07390840351581573,
      "learning_rate": 7.972222222222223e-05,
      "loss": 0.1005,
      "step": 1304
    },
    {
      "epoch": 3.0121177149451817,
      "grad_norm": 0.06907027959823608,
      "learning_rate": 7.962962962962964e-05,
      "loss": 0.1022,
      "step": 1305
    },
    {
      "epoch": 3.014425851125216,
      "grad_norm": 0.07307944446802139,
      "learning_rate": 7.953703703703705e-05,
      "loss": 0.0998,
      "step": 1306
    },
    {
      "epoch": 3.016733987305251,
      "grad_norm": 0.06979411840438843,
      "learning_rate": 7.944444444444444e-05,
      "loss": 0.1037,
      "step": 1307
    },
    {
      "epoch": 3.0190421234852858,
      "grad_norm": 0.07093025743961334,
      "learning_rate": 7.935185185185185e-05,
      "loss": 0.1052,
      "step": 1308
    },
    {
      "epoch": 3.0213502596653203,
      "grad_norm": 0.07909522950649261,
      "learning_rate": 7.925925925925926e-05,
      "loss": 0.0966,
      "step": 1309
    },
    {
      "epoch": 3.023658395845355,
      "grad_norm": 0.07223410159349442,
      "learning_rate": 7.916666666666666e-05,
      "loss": 0.0999,
      "step": 1310
    },
    {
      "epoch": 3.0259665320253895,
      "grad_norm": 0.07958965003490448,
      "learning_rate": 7.907407407407407e-05,
      "loss": 0.1068,
      "step": 1311
    },
    {
      "epoch": 3.028274668205424,
      "grad_norm": 0.0808485671877861,
      "learning_rate": 7.898148148148148e-05,
      "loss": 0.0931,
      "step": 1312
    },
    {
      "epoch": 3.0305828043854586,
      "grad_norm": 0.07426121830940247,
      "learning_rate": 7.88888888888889e-05,
      "loss": 0.1058,
      "step": 1313
    },
    {
      "epoch": 3.0328909405654936,
      "grad_norm": 0.07947281002998352,
      "learning_rate": 7.87962962962963e-05,
      "loss": 0.1018,
      "step": 1314
    },
    {
      "epoch": 3.035199076745528,
      "grad_norm": 0.07325758039951324,
      "learning_rate": 7.870370370370372e-05,
      "loss": 0.1051,
      "step": 1315
    },
    {
      "epoch": 3.0375072129255627,
      "grad_norm": 0.07500126212835312,
      "learning_rate": 7.861111111111111e-05,
      "loss": 0.1109,
      "step": 1316
    },
    {
      "epoch": 3.0398153491055973,
      "grad_norm": 0.0704675242304802,
      "learning_rate": 7.851851851851852e-05,
      "loss": 0.1084,
      "step": 1317
    },
    {
      "epoch": 3.042123485285632,
      "grad_norm": 0.07710418105125427,
      "learning_rate": 7.842592592592592e-05,
      "loss": 0.1061,
      "step": 1318
    },
    {
      "epoch": 3.0444316214656664,
      "grad_norm": 0.07236240804195404,
      "learning_rate": 7.833333333333333e-05,
      "loss": 0.1114,
      "step": 1319
    },
    {
      "epoch": 3.046739757645701,
      "grad_norm": 0.06890179961919785,
      "learning_rate": 7.824074074074074e-05,
      "loss": 0.1034,
      "step": 1320
    },
    {
      "epoch": 3.049047893825736,
      "grad_norm": 0.07557398080825806,
      "learning_rate": 7.814814814814815e-05,
      "loss": 0.1034,
      "step": 1321
    },
    {
      "epoch": 3.0513560300057705,
      "grad_norm": 0.07659531384706497,
      "learning_rate": 7.805555555555556e-05,
      "loss": 0.1047,
      "step": 1322
    },
    {
      "epoch": 3.053664166185805,
      "grad_norm": 0.06749474257230759,
      "learning_rate": 7.796296296296297e-05,
      "loss": 0.1021,
      "step": 1323
    },
    {
      "epoch": 3.0559723023658396,
      "grad_norm": 0.07244992256164551,
      "learning_rate": 7.787037037037037e-05,
      "loss": 0.1082,
      "step": 1324
    },
    {
      "epoch": 3.058280438545874,
      "grad_norm": 0.07568838447332382,
      "learning_rate": 7.777777777777778e-05,
      "loss": 0.1059,
      "step": 1325
    },
    {
      "epoch": 3.0605885747259087,
      "grad_norm": 0.07257690280675888,
      "learning_rate": 7.768518518518519e-05,
      "loss": 0.1077,
      "step": 1326
    },
    {
      "epoch": 3.0628967109059433,
      "grad_norm": 0.06946852058172226,
      "learning_rate": 7.759259259259259e-05,
      "loss": 0.0993,
      "step": 1327
    },
    {
      "epoch": 3.065204847085978,
      "grad_norm": 0.0720793828368187,
      "learning_rate": 7.75e-05,
      "loss": 0.1023,
      "step": 1328
    },
    {
      "epoch": 3.067512983266013,
      "grad_norm": 0.06631141155958176,
      "learning_rate": 7.740740740740741e-05,
      "loss": 0.1109,
      "step": 1329
    },
    {
      "epoch": 3.0698211194460474,
      "grad_norm": 0.06919234991073608,
      "learning_rate": 7.731481481481482e-05,
      "loss": 0.107,
      "step": 1330
    },
    {
      "epoch": 3.072129255626082,
      "grad_norm": 0.06913310289382935,
      "learning_rate": 7.722222222222223e-05,
      "loss": 0.1064,
      "step": 1331
    },
    {
      "epoch": 3.0744373918061165,
      "grad_norm": 0.07767431437969208,
      "learning_rate": 7.712962962962963e-05,
      "loss": 0.1003,
      "step": 1332
    },
    {
      "epoch": 3.076745527986151,
      "grad_norm": 0.07922037690877914,
      "learning_rate": 7.703703703703704e-05,
      "loss": 0.0926,
      "step": 1333
    },
    {
      "epoch": 3.0790536641661856,
      "grad_norm": 0.07872515171766281,
      "learning_rate": 7.694444444444445e-05,
      "loss": 0.1067,
      "step": 1334
    },
    {
      "epoch": 3.0813618003462206,
      "grad_norm": 0.07913444191217422,
      "learning_rate": 7.685185185185185e-05,
      "loss": 0.0964,
      "step": 1335
    },
    {
      "epoch": 3.083669936526255,
      "grad_norm": 0.07317434251308441,
      "learning_rate": 7.675925925925926e-05,
      "loss": 0.1127,
      "step": 1336
    },
    {
      "epoch": 3.0859780727062898,
      "grad_norm": 0.07312801480293274,
      "learning_rate": 7.666666666666667e-05,
      "loss": 0.1029,
      "step": 1337
    },
    {
      "epoch": 3.0882862088863243,
      "grad_norm": 0.07101244479417801,
      "learning_rate": 7.657407407407408e-05,
      "loss": 0.1071,
      "step": 1338
    },
    {
      "epoch": 3.090594345066359,
      "grad_norm": 0.0750599354505539,
      "learning_rate": 7.648148148148149e-05,
      "loss": 0.101,
      "step": 1339
    },
    {
      "epoch": 3.0929024812463934,
      "grad_norm": 0.07678062468767166,
      "learning_rate": 7.638888888888889e-05,
      "loss": 0.1088,
      "step": 1340
    },
    {
      "epoch": 3.095210617426428,
      "grad_norm": 0.0794905498623848,
      "learning_rate": 7.62962962962963e-05,
      "loss": 0.1021,
      "step": 1341
    },
    {
      "epoch": 3.0975187536064626,
      "grad_norm": 0.06659753620624542,
      "learning_rate": 7.620370370370371e-05,
      "loss": 0.1038,
      "step": 1342
    },
    {
      "epoch": 3.0998268897864976,
      "grad_norm": 0.07662273198366165,
      "learning_rate": 7.61111111111111e-05,
      "loss": 0.1024,
      "step": 1343
    },
    {
      "epoch": 3.102135025966532,
      "grad_norm": 0.07258421927690506,
      "learning_rate": 7.601851851851852e-05,
      "loss": 0.1002,
      "step": 1344
    },
    {
      "epoch": 3.1044431621465667,
      "grad_norm": 0.07023981213569641,
      "learning_rate": 7.592592592592593e-05,
      "loss": 0.0975,
      "step": 1345
    },
    {
      "epoch": 3.1067512983266012,
      "grad_norm": 0.07126443833112717,
      "learning_rate": 7.583333333333334e-05,
      "loss": 0.1023,
      "step": 1346
    },
    {
      "epoch": 3.109059434506636,
      "grad_norm": 0.07090507447719574,
      "learning_rate": 7.574074074074075e-05,
      "loss": 0.1053,
      "step": 1347
    },
    {
      "epoch": 3.1113675706866704,
      "grad_norm": 0.06978990137577057,
      "learning_rate": 7.564814814814816e-05,
      "loss": 0.1071,
      "step": 1348
    },
    {
      "epoch": 3.1136757068667054,
      "grad_norm": 0.0739266574382782,
      "learning_rate": 7.555555555555556e-05,
      "loss": 0.1052,
      "step": 1349
    },
    {
      "epoch": 3.11598384304674,
      "grad_norm": 0.07017101347446442,
      "learning_rate": 7.546296296296297e-05,
      "loss": 0.1032,
      "step": 1350
    },
    {
      "epoch": 3.1182919792267745,
      "grad_norm": 0.07571917027235031,
      "learning_rate": 7.537037037037036e-05,
      "loss": 0.1006,
      "step": 1351
    },
    {
      "epoch": 3.120600115406809,
      "grad_norm": 0.07482227683067322,
      "learning_rate": 7.527777777777777e-05,
      "loss": 0.1041,
      "step": 1352
    },
    {
      "epoch": 3.1229082515868436,
      "grad_norm": 0.06881352514028549,
      "learning_rate": 7.518518518518519e-05,
      "loss": 0.1081,
      "step": 1353
    },
    {
      "epoch": 3.125216387766878,
      "grad_norm": 0.06834738701581955,
      "learning_rate": 7.50925925925926e-05,
      "loss": 0.1061,
      "step": 1354
    },
    {
      "epoch": 3.1275245239469127,
      "grad_norm": 0.06924191862344742,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.1094,
      "step": 1355
    },
    {
      "epoch": 3.1298326601269473,
      "grad_norm": 0.07394884526729584,
      "learning_rate": 7.490740740740742e-05,
      "loss": 0.101,
      "step": 1356
    },
    {
      "epoch": 3.1321407963069823,
      "grad_norm": 0.06924940645694733,
      "learning_rate": 7.481481481481481e-05,
      "loss": 0.0998,
      "step": 1357
    },
    {
      "epoch": 3.134448932487017,
      "grad_norm": 0.07207174599170685,
      "learning_rate": 7.472222222222223e-05,
      "loss": 0.1067,
      "step": 1358
    },
    {
      "epoch": 3.1367570686670514,
      "grad_norm": 0.07224036753177643,
      "learning_rate": 7.462962962962964e-05,
      "loss": 0.1035,
      "step": 1359
    },
    {
      "epoch": 3.139065204847086,
      "grad_norm": 0.06934022903442383,
      "learning_rate": 7.453703703703703e-05,
      "loss": 0.1066,
      "step": 1360
    },
    {
      "epoch": 3.1413733410271205,
      "grad_norm": 0.06953021138906479,
      "learning_rate": 7.444444444444444e-05,
      "loss": 0.11,
      "step": 1361
    },
    {
      "epoch": 3.143681477207155,
      "grad_norm": 0.07717426121234894,
      "learning_rate": 7.435185185185185e-05,
      "loss": 0.1002,
      "step": 1362
    },
    {
      "epoch": 3.14598961338719,
      "grad_norm": 0.07500791549682617,
      "learning_rate": 7.425925925925927e-05,
      "loss": 0.1025,
      "step": 1363
    },
    {
      "epoch": 3.1482977495672246,
      "grad_norm": 0.07164338231086731,
      "learning_rate": 7.416666666666668e-05,
      "loss": 0.0992,
      "step": 1364
    },
    {
      "epoch": 3.150605885747259,
      "grad_norm": 0.07608886063098907,
      "learning_rate": 7.407407407407407e-05,
      "loss": 0.0992,
      "step": 1365
    },
    {
      "epoch": 3.1529140219272938,
      "grad_norm": 0.0688973143696785,
      "learning_rate": 7.398148148148148e-05,
      "loss": 0.1019,
      "step": 1366
    },
    {
      "epoch": 3.1552221581073283,
      "grad_norm": 0.07253474742174149,
      "learning_rate": 7.38888888888889e-05,
      "loss": 0.1033,
      "step": 1367
    },
    {
      "epoch": 3.157530294287363,
      "grad_norm": 0.06953491270542145,
      "learning_rate": 7.379629629629629e-05,
      "loss": 0.0986,
      "step": 1368
    },
    {
      "epoch": 3.1598384304673974,
      "grad_norm": 0.07339481264352798,
      "learning_rate": 7.37037037037037e-05,
      "loss": 0.1075,
      "step": 1369
    },
    {
      "epoch": 3.162146566647432,
      "grad_norm": 0.07471857964992523,
      "learning_rate": 7.361111111111111e-05,
      "loss": 0.0985,
      "step": 1370
    },
    {
      "epoch": 3.164454702827467,
      "grad_norm": 0.07148949801921844,
      "learning_rate": 7.351851851851852e-05,
      "loss": 0.1091,
      "step": 1371
    },
    {
      "epoch": 3.1667628390075016,
      "grad_norm": 0.07141561061143875,
      "learning_rate": 7.342592592592593e-05,
      "loss": 0.101,
      "step": 1372
    },
    {
      "epoch": 3.169070975187536,
      "grad_norm": 0.07034918665885925,
      "learning_rate": 7.333333333333333e-05,
      "loss": 0.1098,
      "step": 1373
    },
    {
      "epoch": 3.1713791113675707,
      "grad_norm": 0.06759213656187057,
      "learning_rate": 7.324074074074074e-05,
      "loss": 0.1005,
      "step": 1374
    },
    {
      "epoch": 3.1736872475476052,
      "grad_norm": 0.0680166631937027,
      "learning_rate": 7.314814814814815e-05,
      "loss": 0.0997,
      "step": 1375
    },
    {
      "epoch": 3.17599538372764,
      "grad_norm": 0.06579965353012085,
      "learning_rate": 7.305555555555555e-05,
      "loss": 0.102,
      "step": 1376
    },
    {
      "epoch": 3.1783035199076743,
      "grad_norm": 0.06849706918001175,
      "learning_rate": 7.296296296296296e-05,
      "loss": 0.1066,
      "step": 1377
    },
    {
      "epoch": 3.1806116560877093,
      "grad_norm": 0.06967024505138397,
      "learning_rate": 7.287037037037037e-05,
      "loss": 0.1002,
      "step": 1378
    },
    {
      "epoch": 3.182919792267744,
      "grad_norm": 0.07741349935531616,
      "learning_rate": 7.277777777777778e-05,
      "loss": 0.1089,
      "step": 1379
    },
    {
      "epoch": 3.1852279284477785,
      "grad_norm": 0.07583953440189362,
      "learning_rate": 7.268518518518519e-05,
      "loss": 0.1071,
      "step": 1380
    },
    {
      "epoch": 3.187536064627813,
      "grad_norm": 0.0701722577214241,
      "learning_rate": 7.25925925925926e-05,
      "loss": 0.1056,
      "step": 1381
    },
    {
      "epoch": 3.1898442008078476,
      "grad_norm": 0.072073794901371,
      "learning_rate": 7.25e-05,
      "loss": 0.1041,
      "step": 1382
    },
    {
      "epoch": 3.192152336987882,
      "grad_norm": 0.07338391989469528,
      "learning_rate": 7.240740740740741e-05,
      "loss": 0.1087,
      "step": 1383
    },
    {
      "epoch": 3.1944604731679167,
      "grad_norm": 0.07224875688552856,
      "learning_rate": 7.231481481481481e-05,
      "loss": 0.1053,
      "step": 1384
    },
    {
      "epoch": 3.1967686093479517,
      "grad_norm": 0.06838536262512207,
      "learning_rate": 7.222222222222222e-05,
      "loss": 0.1027,
      "step": 1385
    },
    {
      "epoch": 3.1990767455279863,
      "grad_norm": 0.0706021785736084,
      "learning_rate": 7.212962962962963e-05,
      "loss": 0.1081,
      "step": 1386
    },
    {
      "epoch": 3.201384881708021,
      "grad_norm": 0.07160872966051102,
      "learning_rate": 7.203703703703704e-05,
      "loss": 0.1082,
      "step": 1387
    },
    {
      "epoch": 3.2036930178880554,
      "grad_norm": 0.0765438973903656,
      "learning_rate": 7.194444444444445e-05,
      "loss": 0.0991,
      "step": 1388
    },
    {
      "epoch": 3.20600115406809,
      "grad_norm": 0.0736602172255516,
      "learning_rate": 7.185185185185186e-05,
      "loss": 0.1066,
      "step": 1389
    },
    {
      "epoch": 3.2083092902481245,
      "grad_norm": 0.06911277025938034,
      "learning_rate": 7.175925925925926e-05,
      "loss": 0.1064,
      "step": 1390
    },
    {
      "epoch": 3.210617426428159,
      "grad_norm": 0.07101579010486603,
      "learning_rate": 7.166666666666667e-05,
      "loss": 0.1021,
      "step": 1391
    },
    {
      "epoch": 3.212925562608194,
      "grad_norm": 0.0704028531908989,
      "learning_rate": 7.157407407407408e-05,
      "loss": 0.1021,
      "step": 1392
    },
    {
      "epoch": 3.2152336987882286,
      "grad_norm": 0.07803825289011002,
      "learning_rate": 7.148148148148148e-05,
      "loss": 0.1065,
      "step": 1393
    },
    {
      "epoch": 3.217541834968263,
      "grad_norm": 0.07364324480295181,
      "learning_rate": 7.138888888888889e-05,
      "loss": 0.1072,
      "step": 1394
    },
    {
      "epoch": 3.2198499711482977,
      "grad_norm": 0.0733710452914238,
      "learning_rate": 7.12962962962963e-05,
      "loss": 0.1127,
      "step": 1395
    },
    {
      "epoch": 3.2221581073283323,
      "grad_norm": 0.07203733921051025,
      "learning_rate": 7.120370370370371e-05,
      "loss": 0.101,
      "step": 1396
    },
    {
      "epoch": 3.224466243508367,
      "grad_norm": 0.07261001318693161,
      "learning_rate": 7.111111111111112e-05,
      "loss": 0.105,
      "step": 1397
    },
    {
      "epoch": 3.2267743796884014,
      "grad_norm": 0.07318403571844101,
      "learning_rate": 7.101851851851852e-05,
      "loss": 0.1061,
      "step": 1398
    },
    {
      "epoch": 3.2290825158684364,
      "grad_norm": 0.06906497478485107,
      "learning_rate": 7.092592592592593e-05,
      "loss": 0.0963,
      "step": 1399
    },
    {
      "epoch": 3.231390652048471,
      "grad_norm": 0.07468712329864502,
      "learning_rate": 7.083333333333334e-05,
      "loss": 0.0991,
      "step": 1400
    },
    {
      "epoch": 3.2336987882285055,
      "grad_norm": 0.0731625035405159,
      "learning_rate": 7.074074074074074e-05,
      "loss": 0.106,
      "step": 1401
    },
    {
      "epoch": 3.23600692440854,
      "grad_norm": 0.07272157073020935,
      "learning_rate": 7.064814814814815e-05,
      "loss": 0.1073,
      "step": 1402
    },
    {
      "epoch": 3.2383150605885747,
      "grad_norm": 0.0785585343837738,
      "learning_rate": 7.055555555555556e-05,
      "loss": 0.0994,
      "step": 1403
    },
    {
      "epoch": 3.240623196768609,
      "grad_norm": 0.06936144828796387,
      "learning_rate": 7.046296296296297e-05,
      "loss": 0.1045,
      "step": 1404
    },
    {
      "epoch": 3.2429313329486438,
      "grad_norm": 0.07100614905357361,
      "learning_rate": 7.037037037037038e-05,
      "loss": 0.1045,
      "step": 1405
    },
    {
      "epoch": 3.2452394691286788,
      "grad_norm": 0.06655149906873703,
      "learning_rate": 7.027777777777778e-05,
      "loss": 0.1083,
      "step": 1406
    },
    {
      "epoch": 3.2475476053087133,
      "grad_norm": 0.07160137593746185,
      "learning_rate": 7.018518518518519e-05,
      "loss": 0.1069,
      "step": 1407
    },
    {
      "epoch": 3.249855741488748,
      "grad_norm": 0.06705298274755478,
      "learning_rate": 7.00925925925926e-05,
      "loss": 0.1001,
      "step": 1408
    },
    {
      "epoch": 3.2521638776687825,
      "grad_norm": 0.07310282438993454,
      "learning_rate": 7e-05,
      "loss": 0.1114,
      "step": 1409
    },
    {
      "epoch": 3.254472013848817,
      "grad_norm": 0.06919847428798676,
      "learning_rate": 6.99074074074074e-05,
      "loss": 0.1032,
      "step": 1410
    },
    {
      "epoch": 3.2567801500288516,
      "grad_norm": 0.08264300227165222,
      "learning_rate": 6.981481481481482e-05,
      "loss": 0.1026,
      "step": 1411
    },
    {
      "epoch": 3.259088286208886,
      "grad_norm": 0.06801755726337433,
      "learning_rate": 6.972222222222223e-05,
      "loss": 0.1102,
      "step": 1412
    },
    {
      "epoch": 3.261396422388921,
      "grad_norm": 0.07705879956483841,
      "learning_rate": 6.962962962962964e-05,
      "loss": 0.0935,
      "step": 1413
    },
    {
      "epoch": 3.2637045585689557,
      "grad_norm": 0.0774158388376236,
      "learning_rate": 6.953703703703705e-05,
      "loss": 0.0992,
      "step": 1414
    },
    {
      "epoch": 3.2660126947489903,
      "grad_norm": 0.07569294422864914,
      "learning_rate": 6.944444444444444e-05,
      "loss": 0.1109,
      "step": 1415
    },
    {
      "epoch": 3.268320830929025,
      "grad_norm": 0.07655805349349976,
      "learning_rate": 6.935185185185186e-05,
      "loss": 0.1042,
      "step": 1416
    },
    {
      "epoch": 3.2706289671090594,
      "grad_norm": 0.07557590305805206,
      "learning_rate": 6.925925925925925e-05,
      "loss": 0.1057,
      "step": 1417
    },
    {
      "epoch": 3.272937103289094,
      "grad_norm": 0.07483471184968948,
      "learning_rate": 6.916666666666666e-05,
      "loss": 0.0998,
      "step": 1418
    },
    {
      "epoch": 3.2752452394691285,
      "grad_norm": 0.07382563501596451,
      "learning_rate": 6.907407407407407e-05,
      "loss": 0.0976,
      "step": 1419
    },
    {
      "epoch": 3.2775533756491635,
      "grad_norm": 0.0768398866057396,
      "learning_rate": 6.898148148148148e-05,
      "loss": 0.107,
      "step": 1420
    },
    {
      "epoch": 3.279861511829198,
      "grad_norm": 0.07252249121665955,
      "learning_rate": 6.88888888888889e-05,
      "loss": 0.0948,
      "step": 1421
    },
    {
      "epoch": 3.2821696480092326,
      "grad_norm": 0.07219252735376358,
      "learning_rate": 6.87962962962963e-05,
      "loss": 0.1028,
      "step": 1422
    },
    {
      "epoch": 3.284477784189267,
      "grad_norm": 0.07036572694778442,
      "learning_rate": 6.87037037037037e-05,
      "loss": 0.1076,
      "step": 1423
    },
    {
      "epoch": 3.2867859203693017,
      "grad_norm": 0.07103049755096436,
      "learning_rate": 6.861111111111111e-05,
      "loss": 0.1016,
      "step": 1424
    },
    {
      "epoch": 3.2890940565493363,
      "grad_norm": 0.06950010359287262,
      "learning_rate": 6.851851851851852e-05,
      "loss": 0.1087,
      "step": 1425
    },
    {
      "epoch": 3.291402192729371,
      "grad_norm": 0.06824727356433868,
      "learning_rate": 6.842592592592592e-05,
      "loss": 0.1066,
      "step": 1426
    },
    {
      "epoch": 3.293710328909406,
      "grad_norm": 0.07123945653438568,
      "learning_rate": 6.833333333333333e-05,
      "loss": 0.0994,
      "step": 1427
    },
    {
      "epoch": 3.2960184650894404,
      "grad_norm": 0.07090997695922852,
      "learning_rate": 6.824074074074076e-05,
      "loss": 0.1028,
      "step": 1428
    },
    {
      "epoch": 3.298326601269475,
      "grad_norm": 0.07063266634941101,
      "learning_rate": 6.814814814814815e-05,
      "loss": 0.0998,
      "step": 1429
    },
    {
      "epoch": 3.3006347374495095,
      "grad_norm": 0.069894939661026,
      "learning_rate": 6.805555555555556e-05,
      "loss": 0.104,
      "step": 1430
    },
    {
      "epoch": 3.302942873629544,
      "grad_norm": 0.07208632677793503,
      "learning_rate": 6.796296296296296e-05,
      "loss": 0.0988,
      "step": 1431
    },
    {
      "epoch": 3.3052510098095786,
      "grad_norm": 0.07270758599042892,
      "learning_rate": 6.787037037037037e-05,
      "loss": 0.1087,
      "step": 1432
    },
    {
      "epoch": 3.307559145989613,
      "grad_norm": 0.0734015479683876,
      "learning_rate": 6.777777777777778e-05,
      "loss": 0.1057,
      "step": 1433
    },
    {
      "epoch": 3.309867282169648,
      "grad_norm": 0.06649303436279297,
      "learning_rate": 6.768518518518518e-05,
      "loss": 0.1025,
      "step": 1434
    },
    {
      "epoch": 3.3121754183496828,
      "grad_norm": 0.06872233003377914,
      "learning_rate": 6.759259259259259e-05,
      "loss": 0.105,
      "step": 1435
    },
    {
      "epoch": 3.3144835545297173,
      "grad_norm": 0.06888684630393982,
      "learning_rate": 6.750000000000001e-05,
      "loss": 0.111,
      "step": 1436
    },
    {
      "epoch": 3.316791690709752,
      "grad_norm": 0.06984144449234009,
      "learning_rate": 6.740740740740741e-05,
      "loss": 0.1079,
      "step": 1437
    },
    {
      "epoch": 3.3190998268897864,
      "grad_norm": 0.06998756527900696,
      "learning_rate": 6.731481481481482e-05,
      "loss": 0.1123,
      "step": 1438
    },
    {
      "epoch": 3.321407963069821,
      "grad_norm": 0.07279715687036514,
      "learning_rate": 6.722222222222223e-05,
      "loss": 0.1046,
      "step": 1439
    },
    {
      "epoch": 3.3237160992498556,
      "grad_norm": 0.0708356499671936,
      "learning_rate": 6.712962962962963e-05,
      "loss": 0.1064,
      "step": 1440
    },
    {
      "epoch": 3.32602423542989,
      "grad_norm": 0.07047084718942642,
      "learning_rate": 6.703703703703704e-05,
      "loss": 0.1088,
      "step": 1441
    },
    {
      "epoch": 3.328332371609925,
      "grad_norm": 0.0681910589337349,
      "learning_rate": 6.694444444444444e-05,
      "loss": 0.103,
      "step": 1442
    },
    {
      "epoch": 3.3306405077899597,
      "grad_norm": 0.07179426401853561,
      "learning_rate": 6.685185185185185e-05,
      "loss": 0.1083,
      "step": 1443
    },
    {
      "epoch": 3.3329486439699942,
      "grad_norm": 0.06806602329015732,
      "learning_rate": 6.675925925925926e-05,
      "loss": 0.1059,
      "step": 1444
    },
    {
      "epoch": 3.335256780150029,
      "grad_norm": 0.06857148557901382,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.0991,
      "step": 1445
    },
    {
      "epoch": 3.3375649163300634,
      "grad_norm": 0.07633858919143677,
      "learning_rate": 6.657407407407408e-05,
      "loss": 0.1049,
      "step": 1446
    },
    {
      "epoch": 3.339873052510098,
      "grad_norm": 0.07279085367918015,
      "learning_rate": 6.648148148148149e-05,
      "loss": 0.1071,
      "step": 1447
    },
    {
      "epoch": 3.342181188690133,
      "grad_norm": 0.07123171538114548,
      "learning_rate": 6.638888888888889e-05,
      "loss": 0.096,
      "step": 1448
    },
    {
      "epoch": 3.3444893248701675,
      "grad_norm": 0.06848891824483871,
      "learning_rate": 6.62962962962963e-05,
      "loss": 0.1041,
      "step": 1449
    },
    {
      "epoch": 3.346797461050202,
      "grad_norm": 0.06785860657691956,
      "learning_rate": 6.620370370370371e-05,
      "loss": 0.103,
      "step": 1450
    },
    {
      "epoch": 3.3491055972302366,
      "grad_norm": 0.06861650943756104,
      "learning_rate": 6.611111111111111e-05,
      "loss": 0.1011,
      "step": 1451
    },
    {
      "epoch": 3.351413733410271,
      "grad_norm": 0.07485319674015045,
      "learning_rate": 6.601851851851852e-05,
      "loss": 0.1069,
      "step": 1452
    },
    {
      "epoch": 3.3537218695903057,
      "grad_norm": 0.0701511949300766,
      "learning_rate": 6.592592592592593e-05,
      "loss": 0.0987,
      "step": 1453
    },
    {
      "epoch": 3.3560300057703403,
      "grad_norm": 0.07407116889953613,
      "learning_rate": 6.583333333333334e-05,
      "loss": 0.1095,
      "step": 1454
    },
    {
      "epoch": 3.358338141950375,
      "grad_norm": 0.0752105787396431,
      "learning_rate": 6.574074074074075e-05,
      "loss": 0.1019,
      "step": 1455
    },
    {
      "epoch": 3.36064627813041,
      "grad_norm": 0.0747295469045639,
      "learning_rate": 6.564814814814815e-05,
      "loss": 0.1027,
      "step": 1456
    },
    {
      "epoch": 3.3629544143104444,
      "grad_norm": 0.07597560435533524,
      "learning_rate": 6.555555555555556e-05,
      "loss": 0.1031,
      "step": 1457
    },
    {
      "epoch": 3.365262550490479,
      "grad_norm": 0.07227040827274323,
      "learning_rate": 6.546296296296297e-05,
      "loss": 0.0998,
      "step": 1458
    },
    {
      "epoch": 3.3675706866705135,
      "grad_norm": 0.06881266087293625,
      "learning_rate": 6.537037037037037e-05,
      "loss": 0.1107,
      "step": 1459
    },
    {
      "epoch": 3.369878822850548,
      "grad_norm": 0.07469842582941055,
      "learning_rate": 6.527777777777778e-05,
      "loss": 0.1017,
      "step": 1460
    },
    {
      "epoch": 3.3721869590305826,
      "grad_norm": 0.07287289947271347,
      "learning_rate": 6.51851851851852e-05,
      "loss": 0.1069,
      "step": 1461
    },
    {
      "epoch": 3.3744950952106176,
      "grad_norm": 0.07431170344352722,
      "learning_rate": 6.50925925925926e-05,
      "loss": 0.1001,
      "step": 1462
    },
    {
      "epoch": 3.376803231390652,
      "grad_norm": 0.06553962081670761,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.1044,
      "step": 1463
    },
    {
      "epoch": 3.3791113675706868,
      "grad_norm": 0.07578736543655396,
      "learning_rate": 6.49074074074074e-05,
      "loss": 0.1018,
      "step": 1464
    },
    {
      "epoch": 3.3814195037507213,
      "grad_norm": 0.07057691365480423,
      "learning_rate": 6.481481481481482e-05,
      "loss": 0.1064,
      "step": 1465
    },
    {
      "epoch": 3.383727639930756,
      "grad_norm": 0.06724168360233307,
      "learning_rate": 6.472222222222223e-05,
      "loss": 0.1103,
      "step": 1466
    },
    {
      "epoch": 3.3860357761107904,
      "grad_norm": 0.0718674287199974,
      "learning_rate": 6.462962962962962e-05,
      "loss": 0.1069,
      "step": 1467
    },
    {
      "epoch": 3.388343912290825,
      "grad_norm": 0.06954546272754669,
      "learning_rate": 6.453703703703703e-05,
      "loss": 0.1112,
      "step": 1468
    },
    {
      "epoch": 3.3906520484708595,
      "grad_norm": 0.07211090624332428,
      "learning_rate": 6.444444444444446e-05,
      "loss": 0.1,
      "step": 1469
    },
    {
      "epoch": 3.3929601846508946,
      "grad_norm": 0.07194823771715164,
      "learning_rate": 6.435185185185186e-05,
      "loss": 0.0981,
      "step": 1470
    },
    {
      "epoch": 3.395268320830929,
      "grad_norm": 0.07099419832229614,
      "learning_rate": 6.425925925925927e-05,
      "loss": 0.11,
      "step": 1471
    },
    {
      "epoch": 3.3975764570109637,
      "grad_norm": 0.07098103314638138,
      "learning_rate": 6.416666666666668e-05,
      "loss": 0.0995,
      "step": 1472
    },
    {
      "epoch": 3.3998845931909982,
      "grad_norm": 0.07512085884809494,
      "learning_rate": 6.407407407407407e-05,
      "loss": 0.1047,
      "step": 1473
    },
    {
      "epoch": 3.402192729371033,
      "grad_norm": 0.07270209491252899,
      "learning_rate": 6.398148148148148e-05,
      "loss": 0.1012,
      "step": 1474
    },
    {
      "epoch": 3.4045008655510673,
      "grad_norm": 0.07279909402132034,
      "learning_rate": 6.388888888888888e-05,
      "loss": 0.1009,
      "step": 1475
    },
    {
      "epoch": 3.4068090017311023,
      "grad_norm": 0.07017585635185242,
      "learning_rate": 6.379629629629629e-05,
      "loss": 0.106,
      "step": 1476
    },
    {
      "epoch": 3.409117137911137,
      "grad_norm": 0.07241550087928772,
      "learning_rate": 6.37037037037037e-05,
      "loss": 0.1008,
      "step": 1477
    },
    {
      "epoch": 3.4114252740911715,
      "grad_norm": 0.07290948927402496,
      "learning_rate": 6.361111111111111e-05,
      "loss": 0.1069,
      "step": 1478
    },
    {
      "epoch": 3.413733410271206,
      "grad_norm": 0.07255245745182037,
      "learning_rate": 6.351851851851852e-05,
      "loss": 0.1015,
      "step": 1479
    },
    {
      "epoch": 3.4160415464512406,
      "grad_norm": 0.07338615506887436,
      "learning_rate": 6.342592592592594e-05,
      "loss": 0.0962,
      "step": 1480
    },
    {
      "epoch": 3.418349682631275,
      "grad_norm": 0.06871872395277023,
      "learning_rate": 6.333333333333333e-05,
      "loss": 0.1084,
      "step": 1481
    },
    {
      "epoch": 3.4206578188113097,
      "grad_norm": 0.07104109227657318,
      "learning_rate": 6.324074074074074e-05,
      "loss": 0.0993,
      "step": 1482
    },
    {
      "epoch": 3.4229659549913443,
      "grad_norm": 0.0721510574221611,
      "learning_rate": 6.314814814814815e-05,
      "loss": 0.1048,
      "step": 1483
    },
    {
      "epoch": 3.4252740911713793,
      "grad_norm": 0.06988876312971115,
      "learning_rate": 6.305555555555555e-05,
      "loss": 0.1058,
      "step": 1484
    },
    {
      "epoch": 3.427582227351414,
      "grad_norm": 0.06912636011838913,
      "learning_rate": 6.296296296296296e-05,
      "loss": 0.1019,
      "step": 1485
    },
    {
      "epoch": 3.4298903635314484,
      "grad_norm": 0.07228203862905502,
      "learning_rate": 6.287037037037037e-05,
      "loss": 0.1007,
      "step": 1486
    },
    {
      "epoch": 3.432198499711483,
      "grad_norm": 0.0740547627210617,
      "learning_rate": 6.277777777777778e-05,
      "loss": 0.1035,
      "step": 1487
    },
    {
      "epoch": 3.4345066358915175,
      "grad_norm": 0.07571937888860703,
      "learning_rate": 6.26851851851852e-05,
      "loss": 0.1066,
      "step": 1488
    },
    {
      "epoch": 3.436814772071552,
      "grad_norm": 0.0718480721116066,
      "learning_rate": 6.259259259259259e-05,
      "loss": 0.1032,
      "step": 1489
    },
    {
      "epoch": 3.439122908251587,
      "grad_norm": 0.06957969814538956,
      "learning_rate": 6.25e-05,
      "loss": 0.1085,
      "step": 1490
    },
    {
      "epoch": 3.4414310444316216,
      "grad_norm": 0.07066766172647476,
      "learning_rate": 6.240740740740741e-05,
      "loss": 0.1065,
      "step": 1491
    },
    {
      "epoch": 3.443739180611656,
      "grad_norm": 0.07058632373809814,
      "learning_rate": 6.231481481481481e-05,
      "loss": 0.1052,
      "step": 1492
    },
    {
      "epoch": 3.4460473167916907,
      "grad_norm": 0.0716727003455162,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.1028,
      "step": 1493
    },
    {
      "epoch": 3.4483554529717253,
      "grad_norm": 0.07229859381914139,
      "learning_rate": 6.212962962962964e-05,
      "loss": 0.1044,
      "step": 1494
    },
    {
      "epoch": 3.45066358915176,
      "grad_norm": 0.06694754213094711,
      "learning_rate": 6.203703703703704e-05,
      "loss": 0.103,
      "step": 1495
    },
    {
      "epoch": 3.4529717253317944,
      "grad_norm": 0.07209664583206177,
      "learning_rate": 6.194444444444445e-05,
      "loss": 0.1012,
      "step": 1496
    },
    {
      "epoch": 3.455279861511829,
      "grad_norm": 0.0694059357047081,
      "learning_rate": 6.185185185185185e-05,
      "loss": 0.1097,
      "step": 1497
    },
    {
      "epoch": 3.457587997691864,
      "grad_norm": 0.07374472916126251,
      "learning_rate": 6.175925925925926e-05,
      "loss": 0.1035,
      "step": 1498
    },
    {
      "epoch": 3.4598961338718985,
      "grad_norm": 0.07131217420101166,
      "learning_rate": 6.166666666666667e-05,
      "loss": 0.1007,
      "step": 1499
    },
    {
      "epoch": 3.462204270051933,
      "grad_norm": 0.07447750121355057,
      "learning_rate": 6.157407407407407e-05,
      "loss": 0.0988,
      "step": 1500
    },
    {
      "epoch": 3.4645124062319677,
      "grad_norm": 0.07308261841535568,
      "learning_rate": 6.148148148148148e-05,
      "loss": 0.0987,
      "step": 1501
    },
    {
      "epoch": 3.466820542412002,
      "grad_norm": 0.07671162486076355,
      "learning_rate": 6.13888888888889e-05,
      "loss": 0.1063,
      "step": 1502
    },
    {
      "epoch": 3.4691286785920368,
      "grad_norm": 0.0752405971288681,
      "learning_rate": 6.12962962962963e-05,
      "loss": 0.103,
      "step": 1503
    },
    {
      "epoch": 3.4714368147720718,
      "grad_norm": 0.06844077259302139,
      "learning_rate": 6.120370370370371e-05,
      "loss": 0.0989,
      "step": 1504
    },
    {
      "epoch": 3.4737449509521063,
      "grad_norm": 0.07082361727952957,
      "learning_rate": 6.111111111111112e-05,
      "loss": 0.0983,
      "step": 1505
    },
    {
      "epoch": 3.476053087132141,
      "grad_norm": 0.07211093604564667,
      "learning_rate": 6.101851851851852e-05,
      "loss": 0.105,
      "step": 1506
    },
    {
      "epoch": 3.4783612233121755,
      "grad_norm": 0.07112585753202438,
      "learning_rate": 6.092592592592593e-05,
      "loss": 0.1107,
      "step": 1507
    },
    {
      "epoch": 3.48066935949221,
      "grad_norm": 0.07706272602081299,
      "learning_rate": 6.083333333333333e-05,
      "loss": 0.103,
      "step": 1508
    },
    {
      "epoch": 3.4829774956722446,
      "grad_norm": 0.07344915717840195,
      "learning_rate": 6.074074074074074e-05,
      "loss": 0.1027,
      "step": 1509
    },
    {
      "epoch": 3.485285631852279,
      "grad_norm": 0.07072657346725464,
      "learning_rate": 6.0648148148148154e-05,
      "loss": 0.0992,
      "step": 1510
    },
    {
      "epoch": 3.4875937680323137,
      "grad_norm": 0.07336150854825974,
      "learning_rate": 6.055555555555555e-05,
      "loss": 0.1036,
      "step": 1511
    },
    {
      "epoch": 3.4899019042123487,
      "grad_norm": 0.06873293966054916,
      "learning_rate": 6.046296296296297e-05,
      "loss": 0.1032,
      "step": 1512
    },
    {
      "epoch": 3.4922100403923833,
      "grad_norm": 0.07378534972667694,
      "learning_rate": 6.037037037037038e-05,
      "loss": 0.0985,
      "step": 1513
    },
    {
      "epoch": 3.494518176572418,
      "grad_norm": 0.06721527874469757,
      "learning_rate": 6.0277777777777776e-05,
      "loss": 0.11,
      "step": 1514
    },
    {
      "epoch": 3.4968263127524524,
      "grad_norm": 0.07103805989027023,
      "learning_rate": 6.018518518518519e-05,
      "loss": 0.1022,
      "step": 1515
    },
    {
      "epoch": 3.499134448932487,
      "grad_norm": 0.0758323222398758,
      "learning_rate": 6.00925925925926e-05,
      "loss": 0.0997,
      "step": 1516
    },
    {
      "epoch": 3.5014425851125215,
      "grad_norm": 0.0668865367770195,
      "learning_rate": 6e-05,
      "loss": 0.1041,
      "step": 1517
    },
    {
      "epoch": 3.5037507212925565,
      "grad_norm": 0.07440167665481567,
      "learning_rate": 5.990740740740741e-05,
      "loss": 0.1034,
      "step": 1518
    },
    {
      "epoch": 3.506058857472591,
      "grad_norm": 0.07085245847702026,
      "learning_rate": 5.981481481481481e-05,
      "loss": 0.1051,
      "step": 1519
    },
    {
      "epoch": 3.5083669936526256,
      "grad_norm": 0.07722411304712296,
      "learning_rate": 5.972222222222223e-05,
      "loss": 0.0964,
      "step": 1520
    },
    {
      "epoch": 3.51067512983266,
      "grad_norm": 0.07080914825201035,
      "learning_rate": 5.962962962962964e-05,
      "loss": 0.1046,
      "step": 1521
    },
    {
      "epoch": 3.5129832660126947,
      "grad_norm": 0.07207003980875015,
      "learning_rate": 5.9537037037037035e-05,
      "loss": 0.1021,
      "step": 1522
    },
    {
      "epoch": 3.5152914021927293,
      "grad_norm": 0.07423009723424911,
      "learning_rate": 5.9444444444444445e-05,
      "loss": 0.1006,
      "step": 1523
    },
    {
      "epoch": 3.517599538372764,
      "grad_norm": 0.07439469546079636,
      "learning_rate": 5.9351851851851856e-05,
      "loss": 0.1008,
      "step": 1524
    },
    {
      "epoch": 3.5199076745527984,
      "grad_norm": 0.06865911185741425,
      "learning_rate": 5.925925925925926e-05,
      "loss": 0.105,
      "step": 1525
    },
    {
      "epoch": 3.522215810732833,
      "grad_norm": 0.07315415889024734,
      "learning_rate": 5.916666666666667e-05,
      "loss": 0.1022,
      "step": 1526
    },
    {
      "epoch": 3.524523946912868,
      "grad_norm": 0.07312048971652985,
      "learning_rate": 5.907407407407408e-05,
      "loss": 0.0992,
      "step": 1527
    },
    {
      "epoch": 3.5268320830929025,
      "grad_norm": 0.08295368403196335,
      "learning_rate": 5.8981481481481485e-05,
      "loss": 0.1005,
      "step": 1528
    },
    {
      "epoch": 3.529140219272937,
      "grad_norm": 0.07426223158836365,
      "learning_rate": 5.8888888888888896e-05,
      "loss": 0.0979,
      "step": 1529
    },
    {
      "epoch": 3.5314483554529716,
      "grad_norm": 0.0769488736987114,
      "learning_rate": 5.879629629629629e-05,
      "loss": 0.1028,
      "step": 1530
    },
    {
      "epoch": 3.533756491633006,
      "grad_norm": 0.07679596543312073,
      "learning_rate": 5.8703703703703704e-05,
      "loss": 0.1097,
      "step": 1531
    },
    {
      "epoch": 3.536064627813041,
      "grad_norm": 0.07579246908426285,
      "learning_rate": 5.8611111111111114e-05,
      "loss": 0.1087,
      "step": 1532
    },
    {
      "epoch": 3.5383727639930758,
      "grad_norm": 0.0740475133061409,
      "learning_rate": 5.851851851851852e-05,
      "loss": 0.1026,
      "step": 1533
    },
    {
      "epoch": 3.5406809001731103,
      "grad_norm": 0.07354102283716202,
      "learning_rate": 5.842592592592593e-05,
      "loss": 0.1092,
      "step": 1534
    },
    {
      "epoch": 3.542989036353145,
      "grad_norm": 0.07092611491680145,
      "learning_rate": 5.833333333333334e-05,
      "loss": 0.0972,
      "step": 1535
    },
    {
      "epoch": 3.5452971725331794,
      "grad_norm": 0.07450656592845917,
      "learning_rate": 5.8240740740740743e-05,
      "loss": 0.102,
      "step": 1536
    },
    {
      "epoch": 3.547605308713214,
      "grad_norm": 0.07214991748332977,
      "learning_rate": 5.8148148148148154e-05,
      "loss": 0.0978,
      "step": 1537
    },
    {
      "epoch": 3.5499134448932486,
      "grad_norm": 0.0720483735203743,
      "learning_rate": 5.8055555555555565e-05,
      "loss": 0.1071,
      "step": 1538
    },
    {
      "epoch": 3.552221581073283,
      "grad_norm": 0.07301240414381027,
      "learning_rate": 5.796296296296296e-05,
      "loss": 0.1015,
      "step": 1539
    },
    {
      "epoch": 3.5545297172533177,
      "grad_norm": 0.06999611854553223,
      "learning_rate": 5.787037037037037e-05,
      "loss": 0.1039,
      "step": 1540
    },
    {
      "epoch": 3.5568378534333527,
      "grad_norm": 0.06919919699430466,
      "learning_rate": 5.7777777777777776e-05,
      "loss": 0.0994,
      "step": 1541
    },
    {
      "epoch": 3.5591459896133872,
      "grad_norm": 0.07497837394475937,
      "learning_rate": 5.768518518518519e-05,
      "loss": 0.1018,
      "step": 1542
    },
    {
      "epoch": 3.561454125793422,
      "grad_norm": 0.06988951563835144,
      "learning_rate": 5.75925925925926e-05,
      "loss": 0.0997,
      "step": 1543
    },
    {
      "epoch": 3.5637622619734564,
      "grad_norm": 0.06931570172309875,
      "learning_rate": 5.7499999999999995e-05,
      "loss": 0.1076,
      "step": 1544
    },
    {
      "epoch": 3.566070398153491,
      "grad_norm": 0.07441718876361847,
      "learning_rate": 5.740740740740741e-05,
      "loss": 0.1023,
      "step": 1545
    },
    {
      "epoch": 3.568378534333526,
      "grad_norm": 0.06647201627492905,
      "learning_rate": 5.731481481481482e-05,
      "loss": 0.1026,
      "step": 1546
    },
    {
      "epoch": 3.5706866705135605,
      "grad_norm": 0.07124849408864975,
      "learning_rate": 5.722222222222222e-05,
      "loss": 0.0969,
      "step": 1547
    },
    {
      "epoch": 3.572994806693595,
      "grad_norm": 0.07007735967636108,
      "learning_rate": 5.712962962962963e-05,
      "loss": 0.1056,
      "step": 1548
    },
    {
      "epoch": 3.5753029428736296,
      "grad_norm": 0.07627596706151962,
      "learning_rate": 5.703703703703704e-05,
      "loss": 0.1049,
      "step": 1549
    },
    {
      "epoch": 3.577611079053664,
      "grad_norm": 0.07640865445137024,
      "learning_rate": 5.6944444444444445e-05,
      "loss": 0.1017,
      "step": 1550
    },
    {
      "epoch": 3.5799192152336987,
      "grad_norm": 0.07730797678232193,
      "learning_rate": 5.6851851851851856e-05,
      "loss": 0.0958,
      "step": 1551
    },
    {
      "epoch": 3.5822273514137333,
      "grad_norm": 0.07217419147491455,
      "learning_rate": 5.675925925925925e-05,
      "loss": 0.1008,
      "step": 1552
    },
    {
      "epoch": 3.584535487593768,
      "grad_norm": 0.07629021257162094,
      "learning_rate": 5.666666666666667e-05,
      "loss": 0.0982,
      "step": 1553
    },
    {
      "epoch": 3.5868436237738024,
      "grad_norm": 0.06652412563562393,
      "learning_rate": 5.657407407407408e-05,
      "loss": 0.1043,
      "step": 1554
    },
    {
      "epoch": 3.5891517599538374,
      "grad_norm": 0.07747804373502731,
      "learning_rate": 5.648148148148148e-05,
      "loss": 0.1037,
      "step": 1555
    },
    {
      "epoch": 3.591459896133872,
      "grad_norm": 0.07528828829526901,
      "learning_rate": 5.638888888888889e-05,
      "loss": 0.1007,
      "step": 1556
    },
    {
      "epoch": 3.5937680323139065,
      "grad_norm": 0.07574134320020676,
      "learning_rate": 5.62962962962963e-05,
      "loss": 0.0998,
      "step": 1557
    },
    {
      "epoch": 3.596076168493941,
      "grad_norm": 0.07393226027488708,
      "learning_rate": 5.6203703703703704e-05,
      "loss": 0.1053,
      "step": 1558
    },
    {
      "epoch": 3.5983843046739756,
      "grad_norm": 0.07409621775150299,
      "learning_rate": 5.6111111111111114e-05,
      "loss": 0.1058,
      "step": 1559
    },
    {
      "epoch": 3.6006924408540106,
      "grad_norm": 0.06799048185348511,
      "learning_rate": 5.6018518518518525e-05,
      "loss": 0.1047,
      "step": 1560
    },
    {
      "epoch": 3.603000577034045,
      "grad_norm": 0.07236871868371964,
      "learning_rate": 5.592592592592593e-05,
      "loss": 0.105,
      "step": 1561
    },
    {
      "epoch": 3.6053087132140798,
      "grad_norm": 0.07578806579113007,
      "learning_rate": 5.583333333333334e-05,
      "loss": 0.1106,
      "step": 1562
    },
    {
      "epoch": 3.6076168493941143,
      "grad_norm": 0.07671520859003067,
      "learning_rate": 5.574074074074075e-05,
      "loss": 0.1017,
      "step": 1563
    },
    {
      "epoch": 3.609924985574149,
      "grad_norm": 0.07276728749275208,
      "learning_rate": 5.564814814814815e-05,
      "loss": 0.0997,
      "step": 1564
    },
    {
      "epoch": 3.6122331217541834,
      "grad_norm": 0.06788670271635056,
      "learning_rate": 5.555555555555556e-05,
      "loss": 0.1078,
      "step": 1565
    },
    {
      "epoch": 3.614541257934218,
      "grad_norm": 0.07857690006494522,
      "learning_rate": 5.546296296296296e-05,
      "loss": 0.1056,
      "step": 1566
    },
    {
      "epoch": 3.6168493941142525,
      "grad_norm": 0.07002672553062439,
      "learning_rate": 5.537037037037037e-05,
      "loss": 0.1006,
      "step": 1567
    },
    {
      "epoch": 3.619157530294287,
      "grad_norm": 0.07376798242330551,
      "learning_rate": 5.5277777777777783e-05,
      "loss": 0.0998,
      "step": 1568
    },
    {
      "epoch": 3.621465666474322,
      "grad_norm": 0.07370691001415253,
      "learning_rate": 5.518518518518519e-05,
      "loss": 0.1022,
      "step": 1569
    },
    {
      "epoch": 3.6237738026543567,
      "grad_norm": 0.07946460694074631,
      "learning_rate": 5.50925925925926e-05,
      "loss": 0.103,
      "step": 1570
    },
    {
      "epoch": 3.6260819388343912,
      "grad_norm": 0.07672355324029922,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.1078,
      "step": 1571
    },
    {
      "epoch": 3.628390075014426,
      "grad_norm": 0.07261677831411362,
      "learning_rate": 5.4907407407407406e-05,
      "loss": 0.1038,
      "step": 1572
    },
    {
      "epoch": 3.6306982111944603,
      "grad_norm": 0.07693244516849518,
      "learning_rate": 5.4814814814814817e-05,
      "loss": 0.1002,
      "step": 1573
    },
    {
      "epoch": 3.6330063473744953,
      "grad_norm": 0.07488128542900085,
      "learning_rate": 5.472222222222223e-05,
      "loss": 0.1008,
      "step": 1574
    },
    {
      "epoch": 3.63531448355453,
      "grad_norm": 0.07067075371742249,
      "learning_rate": 5.462962962962963e-05,
      "loss": 0.1092,
      "step": 1575
    },
    {
      "epoch": 3.6376226197345645,
      "grad_norm": 0.07352999597787857,
      "learning_rate": 5.453703703703704e-05,
      "loss": 0.1005,
      "step": 1576
    },
    {
      "epoch": 3.639930755914599,
      "grad_norm": 0.07692652195692062,
      "learning_rate": 5.4444444444444446e-05,
      "loss": 0.0969,
      "step": 1577
    },
    {
      "epoch": 3.6422388920946336,
      "grad_norm": 0.0784744843840599,
      "learning_rate": 5.4351851851851856e-05,
      "loss": 0.104,
      "step": 1578
    },
    {
      "epoch": 3.644547028274668,
      "grad_norm": 0.07523579895496368,
      "learning_rate": 5.425925925925927e-05,
      "loss": 0.1,
      "step": 1579
    },
    {
      "epoch": 3.6468551644547027,
      "grad_norm": 0.0789698138833046,
      "learning_rate": 5.4166666666666664e-05,
      "loss": 0.1067,
      "step": 1580
    },
    {
      "epoch": 3.6491633006347373,
      "grad_norm": 0.0747927576303482,
      "learning_rate": 5.4074074074074075e-05,
      "loss": 0.0965,
      "step": 1581
    },
    {
      "epoch": 3.651471436814772,
      "grad_norm": 0.07639984786510468,
      "learning_rate": 5.3981481481481486e-05,
      "loss": 0.1025,
      "step": 1582
    },
    {
      "epoch": 3.653779572994807,
      "grad_norm": 0.0713706761598587,
      "learning_rate": 5.388888888888889e-05,
      "loss": 0.0967,
      "step": 1583
    },
    {
      "epoch": 3.6560877091748414,
      "grad_norm": 0.07651326805353165,
      "learning_rate": 5.37962962962963e-05,
      "loss": 0.0974,
      "step": 1584
    },
    {
      "epoch": 3.658395845354876,
      "grad_norm": 0.07022444903850555,
      "learning_rate": 5.370370370370371e-05,
      "loss": 0.0989,
      "step": 1585
    },
    {
      "epoch": 3.6607039815349105,
      "grad_norm": 0.07200576364994049,
      "learning_rate": 5.3611111111111115e-05,
      "loss": 0.1063,
      "step": 1586
    },
    {
      "epoch": 3.663012117714945,
      "grad_norm": 0.07116062939167023,
      "learning_rate": 5.3518518518518525e-05,
      "loss": 0.0988,
      "step": 1587
    },
    {
      "epoch": 3.66532025389498,
      "grad_norm": 0.06890930235385895,
      "learning_rate": 5.342592592592592e-05,
      "loss": 0.1008,
      "step": 1588
    },
    {
      "epoch": 3.6676283900750146,
      "grad_norm": 0.07308963686227798,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.1021,
      "step": 1589
    },
    {
      "epoch": 3.669936526255049,
      "grad_norm": 0.0746755376458168,
      "learning_rate": 5.3240740740740744e-05,
      "loss": 0.0949,
      "step": 1590
    },
    {
      "epoch": 3.6722446624350837,
      "grad_norm": 0.07467921078205109,
      "learning_rate": 5.314814814814815e-05,
      "loss": 0.1046,
      "step": 1591
    },
    {
      "epoch": 3.6745527986151183,
      "grad_norm": 0.07346101850271225,
      "learning_rate": 5.305555555555556e-05,
      "loss": 0.1021,
      "step": 1592
    },
    {
      "epoch": 3.676860934795153,
      "grad_norm": 0.07120084017515182,
      "learning_rate": 5.296296296296297e-05,
      "loss": 0.1068,
      "step": 1593
    },
    {
      "epoch": 3.6791690709751874,
      "grad_norm": 0.07904088497161865,
      "learning_rate": 5.287037037037037e-05,
      "loss": 0.0976,
      "step": 1594
    },
    {
      "epoch": 3.681477207155222,
      "grad_norm": 0.07268280535936356,
      "learning_rate": 5.2777777777777784e-05,
      "loss": 0.1054,
      "step": 1595
    },
    {
      "epoch": 3.6837853433352565,
      "grad_norm": 0.07481575757265091,
      "learning_rate": 5.2685185185185194e-05,
      "loss": 0.1035,
      "step": 1596
    },
    {
      "epoch": 3.6860934795152915,
      "grad_norm": 0.07348473370075226,
      "learning_rate": 5.259259259259259e-05,
      "loss": 0.1051,
      "step": 1597
    },
    {
      "epoch": 3.688401615695326,
      "grad_norm": 0.07089775800704956,
      "learning_rate": 5.25e-05,
      "loss": 0.1037,
      "step": 1598
    },
    {
      "epoch": 3.6907097518753607,
      "grad_norm": 0.07000480592250824,
      "learning_rate": 5.2407407407407406e-05,
      "loss": 0.0967,
      "step": 1599
    },
    {
      "epoch": 3.693017888055395,
      "grad_norm": 0.0700959637761116,
      "learning_rate": 5.231481481481482e-05,
      "loss": 0.0967,
      "step": 1600
    },
    {
      "epoch": 3.6953260242354298,
      "grad_norm": 0.06892956793308258,
      "learning_rate": 5.222222222222223e-05,
      "loss": 0.1011,
      "step": 1601
    },
    {
      "epoch": 3.6976341604154648,
      "grad_norm": 0.07513794302940369,
      "learning_rate": 5.212962962962963e-05,
      "loss": 0.0935,
      "step": 1602
    },
    {
      "epoch": 3.6999422965954993,
      "grad_norm": 0.06690793484449387,
      "learning_rate": 5.203703703703704e-05,
      "loss": 0.0987,
      "step": 1603
    },
    {
      "epoch": 3.702250432775534,
      "grad_norm": 0.06937988102436066,
      "learning_rate": 5.194444444444445e-05,
      "loss": 0.1021,
      "step": 1604
    },
    {
      "epoch": 3.7045585689555685,
      "grad_norm": 0.06759417802095413,
      "learning_rate": 5.185185185185185e-05,
      "loss": 0.106,
      "step": 1605
    },
    {
      "epoch": 3.706866705135603,
      "grad_norm": 0.06969857215881348,
      "learning_rate": 5.175925925925926e-05,
      "loss": 0.1026,
      "step": 1606
    },
    {
      "epoch": 3.7091748413156376,
      "grad_norm": 0.0677732452750206,
      "learning_rate": 5.166666666666667e-05,
      "loss": 0.1094,
      "step": 1607
    },
    {
      "epoch": 3.711482977495672,
      "grad_norm": 0.07296530157327652,
      "learning_rate": 5.1574074074074075e-05,
      "loss": 0.1066,
      "step": 1608
    },
    {
      "epoch": 3.7137911136757067,
      "grad_norm": 0.07152239233255386,
      "learning_rate": 5.1481481481481486e-05,
      "loss": 0.1014,
      "step": 1609
    },
    {
      "epoch": 3.7160992498557412,
      "grad_norm": 0.0757564976811409,
      "learning_rate": 5.138888888888889e-05,
      "loss": 0.1054,
      "step": 1610
    },
    {
      "epoch": 3.7184073860357763,
      "grad_norm": 0.07009167969226837,
      "learning_rate": 5.12962962962963e-05,
      "loss": 0.102,
      "step": 1611
    },
    {
      "epoch": 3.720715522215811,
      "grad_norm": 0.07386514544487,
      "learning_rate": 5.120370370370371e-05,
      "loss": 0.102,
      "step": 1612
    },
    {
      "epoch": 3.7230236583958454,
      "grad_norm": 0.07167553901672363,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.0983,
      "step": 1613
    },
    {
      "epoch": 3.72533179457588,
      "grad_norm": 0.06751857697963715,
      "learning_rate": 5.101851851851852e-05,
      "loss": 0.109,
      "step": 1614
    },
    {
      "epoch": 3.7276399307559145,
      "grad_norm": 0.06852557510137558,
      "learning_rate": 5.092592592592593e-05,
      "loss": 0.1058,
      "step": 1615
    },
    {
      "epoch": 3.7299480669359495,
      "grad_norm": 0.07127327471971512,
      "learning_rate": 5.0833333333333333e-05,
      "loss": 0.1044,
      "step": 1616
    },
    {
      "epoch": 3.732256203115984,
      "grad_norm": 0.07348856329917908,
      "learning_rate": 5.0740740740740744e-05,
      "loss": 0.1053,
      "step": 1617
    },
    {
      "epoch": 3.7345643392960186,
      "grad_norm": 0.07111068814992905,
      "learning_rate": 5.0648148148148155e-05,
      "loss": 0.106,
      "step": 1618
    },
    {
      "epoch": 3.736872475476053,
      "grad_norm": 0.0730138048529625,
      "learning_rate": 5.055555555555556e-05,
      "loss": 0.1076,
      "step": 1619
    },
    {
      "epoch": 3.7391806116560877,
      "grad_norm": 0.07346248626708984,
      "learning_rate": 5.046296296296297e-05,
      "loss": 0.0974,
      "step": 1620
    },
    {
      "epoch": 3.7414887478361223,
      "grad_norm": 0.07277431339025497,
      "learning_rate": 5.0370370370370366e-05,
      "loss": 0.1087,
      "step": 1621
    },
    {
      "epoch": 3.743796884016157,
      "grad_norm": 0.07540902495384216,
      "learning_rate": 5.027777777777778e-05,
      "loss": 0.1049,
      "step": 1622
    },
    {
      "epoch": 3.7461050201961914,
      "grad_norm": 0.07597694545984268,
      "learning_rate": 5.018518518518519e-05,
      "loss": 0.104,
      "step": 1623
    },
    {
      "epoch": 3.748413156376226,
      "grad_norm": 0.07473307102918625,
      "learning_rate": 5.009259259259259e-05,
      "loss": 0.1022,
      "step": 1624
    },
    {
      "epoch": 3.750721292556261,
      "grad_norm": 0.07191503047943115,
      "learning_rate": 5e-05,
      "loss": 0.1086,
      "step": 1625
    },
    {
      "epoch": 3.7530294287362955,
      "grad_norm": 0.08716724067926407,
      "learning_rate": 4.9907407407407406e-05,
      "loss": 0.1015,
      "step": 1626
    },
    {
      "epoch": 3.75533756491633,
      "grad_norm": 0.07243862003087997,
      "learning_rate": 4.981481481481482e-05,
      "loss": 0.1006,
      "step": 1627
    },
    {
      "epoch": 3.7576457010963646,
      "grad_norm": 0.07888735830783844,
      "learning_rate": 4.972222222222223e-05,
      "loss": 0.1003,
      "step": 1628
    },
    {
      "epoch": 3.759953837276399,
      "grad_norm": 0.07057338207960129,
      "learning_rate": 4.962962962962963e-05,
      "loss": 0.1046,
      "step": 1629
    },
    {
      "epoch": 3.762261973456434,
      "grad_norm": 0.07319699972867966,
      "learning_rate": 4.9537037037037035e-05,
      "loss": 0.1087,
      "step": 1630
    },
    {
      "epoch": 3.7645701096364688,
      "grad_norm": 0.06961717456579208,
      "learning_rate": 4.9444444444444446e-05,
      "loss": 0.1029,
      "step": 1631
    },
    {
      "epoch": 3.7668782458165033,
      "grad_norm": 0.07344049960374832,
      "learning_rate": 4.935185185185186e-05,
      "loss": 0.0967,
      "step": 1632
    },
    {
      "epoch": 3.769186381996538,
      "grad_norm": 0.07431324571371078,
      "learning_rate": 4.925925925925926e-05,
      "loss": 0.1044,
      "step": 1633
    },
    {
      "epoch": 3.7714945181765724,
      "grad_norm": 0.07130073755979538,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 0.1042,
      "step": 1634
    },
    {
      "epoch": 3.773802654356607,
      "grad_norm": 0.07989674806594849,
      "learning_rate": 4.9074074074074075e-05,
      "loss": 0.1017,
      "step": 1635
    },
    {
      "epoch": 3.7761107905366416,
      "grad_norm": 0.07940138131380081,
      "learning_rate": 4.8981481481481486e-05,
      "loss": 0.0991,
      "step": 1636
    },
    {
      "epoch": 3.778418926716676,
      "grad_norm": 0.07546039670705795,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.0989,
      "step": 1637
    },
    {
      "epoch": 3.7807270628967107,
      "grad_norm": 0.07276611775159836,
      "learning_rate": 4.87962962962963e-05,
      "loss": 0.1028,
      "step": 1638
    },
    {
      "epoch": 3.7830351990767457,
      "grad_norm": 0.07599641382694244,
      "learning_rate": 4.8703703703703704e-05,
      "loss": 0.102,
      "step": 1639
    },
    {
      "epoch": 3.7853433352567802,
      "grad_norm": 0.0719391405582428,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.1021,
      "step": 1640
    },
    {
      "epoch": 3.787651471436815,
      "grad_norm": 0.0732007697224617,
      "learning_rate": 4.851851851851852e-05,
      "loss": 0.1005,
      "step": 1641
    },
    {
      "epoch": 3.7899596076168494,
      "grad_norm": 0.07403897494077682,
      "learning_rate": 4.842592592592593e-05,
      "loss": 0.1027,
      "step": 1642
    },
    {
      "epoch": 3.792267743796884,
      "grad_norm": 0.07413174211978912,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 0.1049,
      "step": 1643
    },
    {
      "epoch": 3.7945758799769185,
      "grad_norm": 0.07258404791355133,
      "learning_rate": 4.8240740740740744e-05,
      "loss": 0.099,
      "step": 1644
    },
    {
      "epoch": 3.7968840161569535,
      "grad_norm": 0.075963594019413,
      "learning_rate": 4.814814814814815e-05,
      "loss": 0.101,
      "step": 1645
    },
    {
      "epoch": 3.799192152336988,
      "grad_norm": 0.07136309146881104,
      "learning_rate": 4.805555555555556e-05,
      "loss": 0.0967,
      "step": 1646
    },
    {
      "epoch": 3.8015002885170226,
      "grad_norm": 0.07163260132074356,
      "learning_rate": 4.796296296296296e-05,
      "loss": 0.0963,
      "step": 1647
    },
    {
      "epoch": 3.803808424697057,
      "grad_norm": 0.07070951163768768,
      "learning_rate": 4.7870370370370373e-05,
      "loss": 0.1031,
      "step": 1648
    },
    {
      "epoch": 3.8061165608770917,
      "grad_norm": 0.07559061050415039,
      "learning_rate": 4.7777777777777784e-05,
      "loss": 0.0993,
      "step": 1649
    },
    {
      "epoch": 3.8084246970571263,
      "grad_norm": 0.06907986104488373,
      "learning_rate": 4.768518518518519e-05,
      "loss": 0.1016,
      "step": 1650
    },
    {
      "epoch": 3.810732833237161,
      "grad_norm": 0.07463349401950836,
      "learning_rate": 4.759259259259259e-05,
      "loss": 0.1007,
      "step": 1651
    },
    {
      "epoch": 3.8130409694171954,
      "grad_norm": 0.07462558895349503,
      "learning_rate": 4.75e-05,
      "loss": 0.1069,
      "step": 1652
    },
    {
      "epoch": 3.8153491055972304,
      "grad_norm": 0.07477578520774841,
      "learning_rate": 4.740740740740741e-05,
      "loss": 0.1038,
      "step": 1653
    },
    {
      "epoch": 3.817657241777265,
      "grad_norm": 0.07634717971086502,
      "learning_rate": 4.731481481481482e-05,
      "loss": 0.1112,
      "step": 1654
    },
    {
      "epoch": 3.8199653779572995,
      "grad_norm": 0.07466798275709152,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.1056,
      "step": 1655
    },
    {
      "epoch": 3.822273514137334,
      "grad_norm": 0.06952636688947678,
      "learning_rate": 4.712962962962963e-05,
      "loss": 0.0999,
      "step": 1656
    },
    {
      "epoch": 3.8245816503173686,
      "grad_norm": 0.07356712967157364,
      "learning_rate": 4.703703703703704e-05,
      "loss": 0.1027,
      "step": 1657
    },
    {
      "epoch": 3.826889786497403,
      "grad_norm": 0.07201752066612244,
      "learning_rate": 4.6944444444444446e-05,
      "loss": 0.0998,
      "step": 1658
    },
    {
      "epoch": 3.829197922677438,
      "grad_norm": 0.06879817694425583,
      "learning_rate": 4.685185185185185e-05,
      "loss": 0.1059,
      "step": 1659
    },
    {
      "epoch": 3.8315060588574728,
      "grad_norm": 0.0763455480337143,
      "learning_rate": 4.675925925925926e-05,
      "loss": 0.106,
      "step": 1660
    },
    {
      "epoch": 3.8338141950375073,
      "grad_norm": 0.07010699808597565,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.1028,
      "step": 1661
    },
    {
      "epoch": 3.836122331217542,
      "grad_norm": 0.08647926896810532,
      "learning_rate": 4.6574074074074076e-05,
      "loss": 0.0963,
      "step": 1662
    },
    {
      "epoch": 3.8384304673975764,
      "grad_norm": 0.07253182679414749,
      "learning_rate": 4.648148148148148e-05,
      "loss": 0.104,
      "step": 1663
    },
    {
      "epoch": 3.840738603577611,
      "grad_norm": 0.07866041362285614,
      "learning_rate": 4.638888888888889e-05,
      "loss": 0.1069,
      "step": 1664
    },
    {
      "epoch": 3.8430467397576455,
      "grad_norm": 0.07139114290475845,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.1029,
      "step": 1665
    },
    {
      "epoch": 3.84535487593768,
      "grad_norm": 0.07753301411867142,
      "learning_rate": 4.6203703703703705e-05,
      "loss": 0.1037,
      "step": 1666
    },
    {
      "epoch": 3.847663012117715,
      "grad_norm": 0.07359980046749115,
      "learning_rate": 4.6111111111111115e-05,
      "loss": 0.0991,
      "step": 1667
    },
    {
      "epoch": 3.8499711482977497,
      "grad_norm": 0.0773310735821724,
      "learning_rate": 4.601851851851852e-05,
      "loss": 0.1005,
      "step": 1668
    },
    {
      "epoch": 3.8522792844777842,
      "grad_norm": 0.07391587644815445,
      "learning_rate": 4.592592592592593e-05,
      "loss": 0.099,
      "step": 1669
    },
    {
      "epoch": 3.854587420657819,
      "grad_norm": 0.07471620291471481,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.0983,
      "step": 1670
    },
    {
      "epoch": 3.8568955568378533,
      "grad_norm": 0.0715847909450531,
      "learning_rate": 4.5740740740740745e-05,
      "loss": 0.0984,
      "step": 1671
    },
    {
      "epoch": 3.859203693017888,
      "grad_norm": 0.0717102661728859,
      "learning_rate": 4.564814814814815e-05,
      "loss": 0.1019,
      "step": 1672
    },
    {
      "epoch": 3.861511829197923,
      "grad_norm": 0.07856345176696777,
      "learning_rate": 4.555555555555556e-05,
      "loss": 0.1005,
      "step": 1673
    },
    {
      "epoch": 3.8638199653779575,
      "grad_norm": 0.07822638750076294,
      "learning_rate": 4.546296296296296e-05,
      "loss": 0.102,
      "step": 1674
    },
    {
      "epoch": 3.866128101557992,
      "grad_norm": 0.07480575144290924,
      "learning_rate": 4.5370370370370374e-05,
      "loss": 0.0998,
      "step": 1675
    },
    {
      "epoch": 3.8684362377380266,
      "grad_norm": 0.07065200805664062,
      "learning_rate": 4.527777777777778e-05,
      "loss": 0.1036,
      "step": 1676
    },
    {
      "epoch": 3.870744373918061,
      "grad_norm": 0.07694060355424881,
      "learning_rate": 4.518518518518519e-05,
      "loss": 0.1003,
      "step": 1677
    },
    {
      "epoch": 3.8730525100980957,
      "grad_norm": 0.07276439666748047,
      "learning_rate": 4.50925925925926e-05,
      "loss": 0.1037,
      "step": 1678
    },
    {
      "epoch": 3.8753606462781303,
      "grad_norm": 0.07501468062400818,
      "learning_rate": 4.5e-05,
      "loss": 0.1025,
      "step": 1679
    },
    {
      "epoch": 3.877668782458165,
      "grad_norm": 0.07596561312675476,
      "learning_rate": 4.490740740740741e-05,
      "loss": 0.1032,
      "step": 1680
    },
    {
      "epoch": 3.8799769186381994,
      "grad_norm": 0.07334397733211517,
      "learning_rate": 4.481481481481482e-05,
      "loss": 0.0987,
      "step": 1681
    },
    {
      "epoch": 3.8822850548182344,
      "grad_norm": 0.07101769000291824,
      "learning_rate": 4.472222222222223e-05,
      "loss": 0.1011,
      "step": 1682
    },
    {
      "epoch": 3.884593190998269,
      "grad_norm": 0.07970575988292694,
      "learning_rate": 4.462962962962963e-05,
      "loss": 0.1011,
      "step": 1683
    },
    {
      "epoch": 3.8869013271783035,
      "grad_norm": 0.0698528066277504,
      "learning_rate": 4.4537037037037036e-05,
      "loss": 0.1054,
      "step": 1684
    },
    {
      "epoch": 3.889209463358338,
      "grad_norm": 0.07277226448059082,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.1032,
      "step": 1685
    },
    {
      "epoch": 3.8915175995383726,
      "grad_norm": 0.07226983457803726,
      "learning_rate": 4.435185185185186e-05,
      "loss": 0.1114,
      "step": 1686
    },
    {
      "epoch": 3.8938257357184076,
      "grad_norm": 0.06960973888635635,
      "learning_rate": 4.425925925925926e-05,
      "loss": 0.1022,
      "step": 1687
    },
    {
      "epoch": 3.896133871898442,
      "grad_norm": 0.0731554850935936,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 0.0953,
      "step": 1688
    },
    {
      "epoch": 3.8984420080784767,
      "grad_norm": 0.0755450651049614,
      "learning_rate": 4.4074074074074076e-05,
      "loss": 0.0952,
      "step": 1689
    },
    {
      "epoch": 3.9007501442585113,
      "grad_norm": 0.07098191231489182,
      "learning_rate": 4.3981481481481486e-05,
      "loss": 0.1082,
      "step": 1690
    },
    {
      "epoch": 3.903058280438546,
      "grad_norm": 0.07729455828666687,
      "learning_rate": 4.388888888888889e-05,
      "loss": 0.0949,
      "step": 1691
    },
    {
      "epoch": 3.9053664166185804,
      "grad_norm": 0.07213867455720901,
      "learning_rate": 4.3796296296296294e-05,
      "loss": 0.1047,
      "step": 1692
    },
    {
      "epoch": 3.907674552798615,
      "grad_norm": 0.0729680210351944,
      "learning_rate": 4.3703703703703705e-05,
      "loss": 0.1019,
      "step": 1693
    },
    {
      "epoch": 3.9099826889786495,
      "grad_norm": 0.07491601258516312,
      "learning_rate": 4.3611111111111116e-05,
      "loss": 0.1044,
      "step": 1694
    },
    {
      "epoch": 3.912290825158684,
      "grad_norm": 0.07388050109148026,
      "learning_rate": 4.351851851851852e-05,
      "loss": 0.1093,
      "step": 1695
    },
    {
      "epoch": 3.914598961338719,
      "grad_norm": 0.07070640474557877,
      "learning_rate": 4.342592592592592e-05,
      "loss": 0.0946,
      "step": 1696
    },
    {
      "epoch": 3.9169070975187537,
      "grad_norm": 0.07265572249889374,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.0949,
      "step": 1697
    },
    {
      "epoch": 3.919215233698788,
      "grad_norm": 0.07613741606473923,
      "learning_rate": 4.3240740740740745e-05,
      "loss": 0.0965,
      "step": 1698
    },
    {
      "epoch": 3.9215233698788228,
      "grad_norm": 0.07195375859737396,
      "learning_rate": 4.314814814814815e-05,
      "loss": 0.0996,
      "step": 1699
    },
    {
      "epoch": 3.9238315060588573,
      "grad_norm": 0.07808765769004822,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.1019,
      "step": 1700
    },
    {
      "epoch": 3.9261396422388923,
      "grad_norm": 0.07393735647201538,
      "learning_rate": 4.296296296296296e-05,
      "loss": 0.1009,
      "step": 1701
    },
    {
      "epoch": 3.928447778418927,
      "grad_norm": 0.07203008979558945,
      "learning_rate": 4.2870370370370374e-05,
      "loss": 0.0988,
      "step": 1702
    },
    {
      "epoch": 3.9307559145989615,
      "grad_norm": 0.0719631165266037,
      "learning_rate": 4.277777777777778e-05,
      "loss": 0.1028,
      "step": 1703
    },
    {
      "epoch": 3.933064050778996,
      "grad_norm": 0.07469997555017471,
      "learning_rate": 4.268518518518519e-05,
      "loss": 0.104,
      "step": 1704
    },
    {
      "epoch": 3.9353721869590306,
      "grad_norm": 0.06861697882413864,
      "learning_rate": 4.259259259259259e-05,
      "loss": 0.1067,
      "step": 1705
    },
    {
      "epoch": 3.937680323139065,
      "grad_norm": 0.07139144837856293,
      "learning_rate": 4.25e-05,
      "loss": 0.1022,
      "step": 1706
    },
    {
      "epoch": 3.9399884593190997,
      "grad_norm": 0.07737748324871063,
      "learning_rate": 4.240740740740741e-05,
      "loss": 0.1045,
      "step": 1707
    },
    {
      "epoch": 3.9422965954991342,
      "grad_norm": 0.07521290332078934,
      "learning_rate": 4.231481481481482e-05,
      "loss": 0.0993,
      "step": 1708
    },
    {
      "epoch": 3.944604731679169,
      "grad_norm": 0.07259056717157364,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.1002,
      "step": 1709
    },
    {
      "epoch": 3.946912867859204,
      "grad_norm": 0.07190921157598495,
      "learning_rate": 4.212962962962963e-05,
      "loss": 0.1003,
      "step": 1710
    },
    {
      "epoch": 3.9492210040392384,
      "grad_norm": 0.07116570323705673,
      "learning_rate": 4.203703703703704e-05,
      "loss": 0.101,
      "step": 1711
    },
    {
      "epoch": 3.951529140219273,
      "grad_norm": 0.0680527612566948,
      "learning_rate": 4.194444444444445e-05,
      "loss": 0.0967,
      "step": 1712
    },
    {
      "epoch": 3.9538372763993075,
      "grad_norm": 0.07480891793966293,
      "learning_rate": 4.185185185185185e-05,
      "loss": 0.1045,
      "step": 1713
    },
    {
      "epoch": 3.956145412579342,
      "grad_norm": 0.07586976885795593,
      "learning_rate": 4.175925925925926e-05,
      "loss": 0.1027,
      "step": 1714
    },
    {
      "epoch": 3.958453548759377,
      "grad_norm": 0.07453934103250504,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.0984,
      "step": 1715
    },
    {
      "epoch": 3.9607616849394116,
      "grad_norm": 0.07293694466352463,
      "learning_rate": 4.1574074074074076e-05,
      "loss": 0.0911,
      "step": 1716
    },
    {
      "epoch": 3.963069821119446,
      "grad_norm": 0.07300060987472534,
      "learning_rate": 4.148148148148148e-05,
      "loss": 0.1068,
      "step": 1717
    },
    {
      "epoch": 3.9653779572994807,
      "grad_norm": 0.07118441164493561,
      "learning_rate": 4.138888888888889e-05,
      "loss": 0.1017,
      "step": 1718
    },
    {
      "epoch": 3.9676860934795153,
      "grad_norm": 0.07542804628610611,
      "learning_rate": 4.12962962962963e-05,
      "loss": 0.0975,
      "step": 1719
    },
    {
      "epoch": 3.96999422965955,
      "grad_norm": 0.07660391926765442,
      "learning_rate": 4.1203703703703705e-05,
      "loss": 0.1046,
      "step": 1720
    },
    {
      "epoch": 3.9723023658395844,
      "grad_norm": 0.07354540377855301,
      "learning_rate": 4.111111111111111e-05,
      "loss": 0.1049,
      "step": 1721
    },
    {
      "epoch": 3.974610502019619,
      "grad_norm": 0.07344894856214523,
      "learning_rate": 4.101851851851852e-05,
      "loss": 0.1021,
      "step": 1722
    },
    {
      "epoch": 3.9769186381996535,
      "grad_norm": 0.07788650691509247,
      "learning_rate": 4.092592592592593e-05,
      "loss": 0.1032,
      "step": 1723
    },
    {
      "epoch": 3.9792267743796885,
      "grad_norm": 0.07798030972480774,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 0.0997,
      "step": 1724
    },
    {
      "epoch": 3.981534910559723,
      "grad_norm": 0.07274124771356583,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.1021,
      "step": 1725
    },
    {
      "epoch": 3.9838430467397576,
      "grad_norm": 0.07510056346654892,
      "learning_rate": 4.064814814814815e-05,
      "loss": 0.0965,
      "step": 1726
    },
    {
      "epoch": 3.986151182919792,
      "grad_norm": 0.07315821200609207,
      "learning_rate": 4.055555555555556e-05,
      "loss": 0.1004,
      "step": 1727
    },
    {
      "epoch": 3.9884593190998268,
      "grad_norm": 0.07344626635313034,
      "learning_rate": 4.0462962962962963e-05,
      "loss": 0.0913,
      "step": 1728
    },
    {
      "epoch": 3.9907674552798618,
      "grad_norm": 0.06818577647209167,
      "learning_rate": 4.0370370370370374e-05,
      "loss": 0.1037,
      "step": 1729
    },
    {
      "epoch": 3.9930755914598963,
      "grad_norm": 0.07271457463502884,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.1002,
      "step": 1730
    },
    {
      "epoch": 3.995383727639931,
      "grad_norm": 0.07179445028305054,
      "learning_rate": 4.018518518518519e-05,
      "loss": 0.0921,
      "step": 1731
    },
    {
      "epoch": 3.9976918638199654,
      "grad_norm": 0.07152129709720612,
      "learning_rate": 4.009259259259259e-05,
      "loss": 0.0964,
      "step": 1732
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.08043841272592545,
      "learning_rate": 4e-05,
      "loss": 0.095,
      "step": 1733
    },
    {
      "epoch": 4.002308136180035,
      "grad_norm": 0.07450731098651886,
      "learning_rate": 3.990740740740741e-05,
      "loss": 0.094,
      "step": 1734
    },
    {
      "epoch": 4.004616272360069,
      "grad_norm": 0.07299776375293732,
      "learning_rate": 3.981481481481482e-05,
      "loss": 0.1014,
      "step": 1735
    },
    {
      "epoch": 4.006924408540104,
      "grad_norm": 0.07632085680961609,
      "learning_rate": 3.972222222222222e-05,
      "loss": 0.0916,
      "step": 1736
    },
    {
      "epoch": 4.009232544720138,
      "grad_norm": 0.07991166412830353,
      "learning_rate": 3.962962962962963e-05,
      "loss": 0.0914,
      "step": 1737
    },
    {
      "epoch": 4.011540680900173,
      "grad_norm": 0.08636566251516342,
      "learning_rate": 3.9537037037037036e-05,
      "loss": 0.0919,
      "step": 1738
    },
    {
      "epoch": 4.013848817080207,
      "grad_norm": 0.08197848498821259,
      "learning_rate": 3.944444444444445e-05,
      "loss": 0.095,
      "step": 1739
    },
    {
      "epoch": 4.016156953260242,
      "grad_norm": 0.07920564711093903,
      "learning_rate": 3.935185185185186e-05,
      "loss": 0.0946,
      "step": 1740
    },
    {
      "epoch": 4.018465089440277,
      "grad_norm": 0.07687513530254364,
      "learning_rate": 3.925925925925926e-05,
      "loss": 0.0974,
      "step": 1741
    },
    {
      "epoch": 4.020773225620312,
      "grad_norm": 0.07848761230707169,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.1076,
      "step": 1742
    },
    {
      "epoch": 4.0230813618003465,
      "grad_norm": 0.07819864153862,
      "learning_rate": 3.9074074074074076e-05,
      "loss": 0.102,
      "step": 1743
    },
    {
      "epoch": 4.025389497980381,
      "grad_norm": 0.08252421766519547,
      "learning_rate": 3.898148148148149e-05,
      "loss": 0.0977,
      "step": 1744
    },
    {
      "epoch": 4.027697634160416,
      "grad_norm": 0.08440664410591125,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.0905,
      "step": 1745
    },
    {
      "epoch": 4.03000577034045,
      "grad_norm": 0.08802264928817749,
      "learning_rate": 3.8796296296296295e-05,
      "loss": 0.1015,
      "step": 1746
    },
    {
      "epoch": 4.032313906520485,
      "grad_norm": 0.08528051525354385,
      "learning_rate": 3.8703703703703705e-05,
      "loss": 0.0943,
      "step": 1747
    },
    {
      "epoch": 4.034622042700519,
      "grad_norm": 0.0825500339269638,
      "learning_rate": 3.8611111111111116e-05,
      "loss": 0.092,
      "step": 1748
    },
    {
      "epoch": 4.036930178880554,
      "grad_norm": 0.08139495551586151,
      "learning_rate": 3.851851851851852e-05,
      "loss": 0.0976,
      "step": 1749
    },
    {
      "epoch": 4.039238315060588,
      "grad_norm": 0.08521188795566559,
      "learning_rate": 3.8425925925925924e-05,
      "loss": 0.0926,
      "step": 1750
    },
    {
      "epoch": 4.041546451240623,
      "grad_norm": 0.0821610763669014,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 0.0969,
      "step": 1751
    },
    {
      "epoch": 4.0438545874206575,
      "grad_norm": 0.08546742796897888,
      "learning_rate": 3.8240740740740745e-05,
      "loss": 0.093,
      "step": 1752
    },
    {
      "epoch": 4.046162723600692,
      "grad_norm": 0.08199765533208847,
      "learning_rate": 3.814814814814815e-05,
      "loss": 0.0952,
      "step": 1753
    },
    {
      "epoch": 4.048470859780727,
      "grad_norm": 0.08618094772100449,
      "learning_rate": 3.805555555555555e-05,
      "loss": 0.0947,
      "step": 1754
    },
    {
      "epoch": 4.050778995960762,
      "grad_norm": 0.08096267282962799,
      "learning_rate": 3.7962962962962964e-05,
      "loss": 0.0994,
      "step": 1755
    },
    {
      "epoch": 4.053087132140797,
      "grad_norm": 0.07903143763542175,
      "learning_rate": 3.7870370370370374e-05,
      "loss": 0.1038,
      "step": 1756
    },
    {
      "epoch": 4.055395268320831,
      "grad_norm": 0.08284498751163483,
      "learning_rate": 3.777777777777778e-05,
      "loss": 0.0894,
      "step": 1757
    },
    {
      "epoch": 4.057703404500866,
      "grad_norm": 0.0834072083234787,
      "learning_rate": 3.768518518518518e-05,
      "loss": 0.0967,
      "step": 1758
    },
    {
      "epoch": 4.0600115406809,
      "grad_norm": 0.08580133318901062,
      "learning_rate": 3.759259259259259e-05,
      "loss": 0.0924,
      "step": 1759
    },
    {
      "epoch": 4.062319676860935,
      "grad_norm": 0.08878447115421295,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0943,
      "step": 1760
    },
    {
      "epoch": 4.064627813040969,
      "grad_norm": 0.07999306917190552,
      "learning_rate": 3.740740740740741e-05,
      "loss": 0.0949,
      "step": 1761
    },
    {
      "epoch": 4.066935949221004,
      "grad_norm": 0.08124487102031708,
      "learning_rate": 3.731481481481482e-05,
      "loss": 0.097,
      "step": 1762
    },
    {
      "epoch": 4.0692440854010385,
      "grad_norm": 0.08802463859319687,
      "learning_rate": 3.722222222222222e-05,
      "loss": 0.0979,
      "step": 1763
    },
    {
      "epoch": 4.071552221581073,
      "grad_norm": 0.08120068162679672,
      "learning_rate": 3.712962962962963e-05,
      "loss": 0.1002,
      "step": 1764
    },
    {
      "epoch": 4.073860357761108,
      "grad_norm": 0.08201154321432114,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.0987,
      "step": 1765
    },
    {
      "epoch": 4.076168493941142,
      "grad_norm": 0.08792880922555923,
      "learning_rate": 3.694444444444445e-05,
      "loss": 0.0995,
      "step": 1766
    },
    {
      "epoch": 4.078476630121177,
      "grad_norm": 0.08233310282230377,
      "learning_rate": 3.685185185185185e-05,
      "loss": 0.0978,
      "step": 1767
    },
    {
      "epoch": 4.080784766301211,
      "grad_norm": 0.08558878302574158,
      "learning_rate": 3.675925925925926e-05,
      "loss": 0.0936,
      "step": 1768
    },
    {
      "epoch": 4.083092902481247,
      "grad_norm": 0.0798913836479187,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.0947,
      "step": 1769
    },
    {
      "epoch": 4.085401038661281,
      "grad_norm": 0.08219851553440094,
      "learning_rate": 3.6574074074074076e-05,
      "loss": 0.0958,
      "step": 1770
    },
    {
      "epoch": 4.087709174841316,
      "grad_norm": 0.07899602502584457,
      "learning_rate": 3.648148148148148e-05,
      "loss": 0.0969,
      "step": 1771
    },
    {
      "epoch": 4.0900173110213505,
      "grad_norm": 0.08629952371120453,
      "learning_rate": 3.638888888888889e-05,
      "loss": 0.0927,
      "step": 1772
    },
    {
      "epoch": 4.092325447201385,
      "grad_norm": 0.07797697931528091,
      "learning_rate": 3.62962962962963e-05,
      "loss": 0.0943,
      "step": 1773
    },
    {
      "epoch": 4.09463358338142,
      "grad_norm": 0.08745614439249039,
      "learning_rate": 3.6203703703703706e-05,
      "loss": 0.0953,
      "step": 1774
    },
    {
      "epoch": 4.096941719561454,
      "grad_norm": 0.0838518887758255,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.0997,
      "step": 1775
    },
    {
      "epoch": 4.099249855741489,
      "grad_norm": 0.08577281981706619,
      "learning_rate": 3.601851851851852e-05,
      "loss": 0.0985,
      "step": 1776
    },
    {
      "epoch": 4.101557991921523,
      "grad_norm": 0.08198212087154388,
      "learning_rate": 3.592592592592593e-05,
      "loss": 0.1035,
      "step": 1777
    },
    {
      "epoch": 4.103866128101558,
      "grad_norm": 0.08031629025936127,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 0.0951,
      "step": 1778
    },
    {
      "epoch": 4.106174264281592,
      "grad_norm": 0.08040511608123779,
      "learning_rate": 3.574074074074074e-05,
      "loss": 0.097,
      "step": 1779
    },
    {
      "epoch": 4.108482400461627,
      "grad_norm": 0.08482804894447327,
      "learning_rate": 3.564814814814815e-05,
      "loss": 0.1078,
      "step": 1780
    },
    {
      "epoch": 4.1107905366416615,
      "grad_norm": 0.08974684774875641,
      "learning_rate": 3.555555555555556e-05,
      "loss": 0.1025,
      "step": 1781
    },
    {
      "epoch": 4.113098672821696,
      "grad_norm": 0.08361063152551651,
      "learning_rate": 3.5462962962962964e-05,
      "loss": 0.0929,
      "step": 1782
    },
    {
      "epoch": 4.1154068090017315,
      "grad_norm": 0.08218709379434586,
      "learning_rate": 3.537037037037037e-05,
      "loss": 0.0981,
      "step": 1783
    },
    {
      "epoch": 4.117714945181766,
      "grad_norm": 0.08070368319749832,
      "learning_rate": 3.527777777777778e-05,
      "loss": 0.1006,
      "step": 1784
    },
    {
      "epoch": 4.120023081361801,
      "grad_norm": 0.0916745588183403,
      "learning_rate": 3.518518518518519e-05,
      "loss": 0.1001,
      "step": 1785
    },
    {
      "epoch": 4.122331217541835,
      "grad_norm": 0.07818419486284256,
      "learning_rate": 3.509259259259259e-05,
      "loss": 0.0967,
      "step": 1786
    },
    {
      "epoch": 4.12463935372187,
      "grad_norm": 0.08177449554204941,
      "learning_rate": 3.5e-05,
      "loss": 0.0972,
      "step": 1787
    },
    {
      "epoch": 4.126947489901904,
      "grad_norm": 0.08102845400571823,
      "learning_rate": 3.490740740740741e-05,
      "loss": 0.0947,
      "step": 1788
    },
    {
      "epoch": 4.129255626081939,
      "grad_norm": 0.08581893146038055,
      "learning_rate": 3.481481481481482e-05,
      "loss": 0.0913,
      "step": 1789
    },
    {
      "epoch": 4.131563762261973,
      "grad_norm": 0.08449820429086685,
      "learning_rate": 3.472222222222222e-05,
      "loss": 0.0982,
      "step": 1790
    },
    {
      "epoch": 4.133871898442008,
      "grad_norm": 0.08604080975055695,
      "learning_rate": 3.4629629629629626e-05,
      "loss": 0.093,
      "step": 1791
    },
    {
      "epoch": 4.1361800346220425,
      "grad_norm": 0.08496052026748657,
      "learning_rate": 3.453703703703704e-05,
      "loss": 0.0865,
      "step": 1792
    },
    {
      "epoch": 4.138488170802077,
      "grad_norm": 0.08148840814828873,
      "learning_rate": 3.444444444444445e-05,
      "loss": 0.0932,
      "step": 1793
    },
    {
      "epoch": 4.140796306982112,
      "grad_norm": 0.0862799659371376,
      "learning_rate": 3.435185185185185e-05,
      "loss": 0.0934,
      "step": 1794
    },
    {
      "epoch": 4.143104443162146,
      "grad_norm": 0.08250004053115845,
      "learning_rate": 3.425925925925926e-05,
      "loss": 0.0972,
      "step": 1795
    },
    {
      "epoch": 4.145412579342181,
      "grad_norm": 0.08421081304550171,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 0.0939,
      "step": 1796
    },
    {
      "epoch": 4.147720715522216,
      "grad_norm": 0.08294108510017395,
      "learning_rate": 3.4074074074074077e-05,
      "loss": 0.0919,
      "step": 1797
    },
    {
      "epoch": 4.150028851702251,
      "grad_norm": 0.08223319053649902,
      "learning_rate": 3.398148148148148e-05,
      "loss": 0.0929,
      "step": 1798
    },
    {
      "epoch": 4.152336987882285,
      "grad_norm": 0.08725429326295853,
      "learning_rate": 3.388888888888889e-05,
      "loss": 0.0942,
      "step": 1799
    },
    {
      "epoch": 4.15464512406232,
      "grad_norm": 0.083881676197052,
      "learning_rate": 3.3796296296296295e-05,
      "loss": 0.0939,
      "step": 1800
    },
    {
      "epoch": 4.1569532602423545,
      "grad_norm": 0.08893053978681564,
      "learning_rate": 3.3703703703703706e-05,
      "loss": 0.1024,
      "step": 1801
    },
    {
      "epoch": 4.159261396422389,
      "grad_norm": 0.07956549525260925,
      "learning_rate": 3.3611111111111116e-05,
      "loss": 0.0976,
      "step": 1802
    },
    {
      "epoch": 4.161569532602424,
      "grad_norm": 0.08945251256227493,
      "learning_rate": 3.351851851851852e-05,
      "loss": 0.1001,
      "step": 1803
    },
    {
      "epoch": 4.163877668782458,
      "grad_norm": 0.08129511028528214,
      "learning_rate": 3.3425925925925924e-05,
      "loss": 0.0921,
      "step": 1804
    },
    {
      "epoch": 4.166185804962493,
      "grad_norm": 0.08377901464700699,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0976,
      "step": 1805
    },
    {
      "epoch": 4.168493941142527,
      "grad_norm": 0.08749788999557495,
      "learning_rate": 3.3240740740740746e-05,
      "loss": 0.0858,
      "step": 1806
    },
    {
      "epoch": 4.170802077322562,
      "grad_norm": 0.08360150456428528,
      "learning_rate": 3.314814814814815e-05,
      "loss": 0.0986,
      "step": 1807
    },
    {
      "epoch": 4.173110213502596,
      "grad_norm": 0.08701533079147339,
      "learning_rate": 3.3055555555555553e-05,
      "loss": 0.0976,
      "step": 1808
    },
    {
      "epoch": 4.175418349682631,
      "grad_norm": 0.08529170602560043,
      "learning_rate": 3.2962962962962964e-05,
      "loss": 0.0925,
      "step": 1809
    },
    {
      "epoch": 4.1777264858626655,
      "grad_norm": 0.07833974063396454,
      "learning_rate": 3.2870370370370375e-05,
      "loss": 0.1057,
      "step": 1810
    },
    {
      "epoch": 4.180034622042701,
      "grad_norm": 0.08398272097110748,
      "learning_rate": 3.277777777777778e-05,
      "loss": 0.0941,
      "step": 1811
    },
    {
      "epoch": 4.1823427582227355,
      "grad_norm": 0.0835639089345932,
      "learning_rate": 3.268518518518518e-05,
      "loss": 0.0955,
      "step": 1812
    },
    {
      "epoch": 4.18465089440277,
      "grad_norm": 0.08440840244293213,
      "learning_rate": 3.25925925925926e-05,
      "loss": 0.1016,
      "step": 1813
    },
    {
      "epoch": 4.186959030582805,
      "grad_norm": 0.08645574748516083,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.0988,
      "step": 1814
    },
    {
      "epoch": 4.189267166762839,
      "grad_norm": 0.08334686607122421,
      "learning_rate": 3.240740740740741e-05,
      "loss": 0.0937,
      "step": 1815
    },
    {
      "epoch": 4.191575302942874,
      "grad_norm": 0.08069921284914017,
      "learning_rate": 3.231481481481481e-05,
      "loss": 0.0978,
      "step": 1816
    },
    {
      "epoch": 4.193883439122908,
      "grad_norm": 0.08343187719583511,
      "learning_rate": 3.222222222222223e-05,
      "loss": 0.0937,
      "step": 1817
    },
    {
      "epoch": 4.196191575302943,
      "grad_norm": 0.077997587621212,
      "learning_rate": 3.212962962962963e-05,
      "loss": 0.1026,
      "step": 1818
    },
    {
      "epoch": 4.198499711482977,
      "grad_norm": 0.08469629287719727,
      "learning_rate": 3.203703703703704e-05,
      "loss": 0.0926,
      "step": 1819
    },
    {
      "epoch": 4.200807847663012,
      "grad_norm": 0.08234648406505585,
      "learning_rate": 3.194444444444444e-05,
      "loss": 0.0992,
      "step": 1820
    },
    {
      "epoch": 4.2031159838430465,
      "grad_norm": 0.08333395421504974,
      "learning_rate": 3.185185185185185e-05,
      "loss": 0.0874,
      "step": 1821
    },
    {
      "epoch": 4.205424120023081,
      "grad_norm": 0.08219824731349945,
      "learning_rate": 3.175925925925926e-05,
      "loss": 0.097,
      "step": 1822
    },
    {
      "epoch": 4.207732256203116,
      "grad_norm": 0.08016586303710938,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.1047,
      "step": 1823
    },
    {
      "epoch": 4.21004039238315,
      "grad_norm": 0.08344191312789917,
      "learning_rate": 3.157407407407408e-05,
      "loss": 0.087,
      "step": 1824
    },
    {
      "epoch": 4.212348528563186,
      "grad_norm": 0.08203115314245224,
      "learning_rate": 3.148148148148148e-05,
      "loss": 0.0972,
      "step": 1825
    },
    {
      "epoch": 4.21465666474322,
      "grad_norm": 0.08124659955501556,
      "learning_rate": 3.138888888888889e-05,
      "loss": 0.0918,
      "step": 1826
    },
    {
      "epoch": 4.216964800923255,
      "grad_norm": 0.08218760043382645,
      "learning_rate": 3.1296296296296295e-05,
      "loss": 0.0997,
      "step": 1827
    },
    {
      "epoch": 4.219272937103289,
      "grad_norm": 0.07817607372999191,
      "learning_rate": 3.1203703703703706e-05,
      "loss": 0.1035,
      "step": 1828
    },
    {
      "epoch": 4.221581073283324,
      "grad_norm": 0.07928155362606049,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.0989,
      "step": 1829
    },
    {
      "epoch": 4.223889209463358,
      "grad_norm": 0.08378638327121735,
      "learning_rate": 3.101851851851852e-05,
      "loss": 0.0934,
      "step": 1830
    },
    {
      "epoch": 4.226197345643393,
      "grad_norm": 0.08161495625972748,
      "learning_rate": 3.0925925925925924e-05,
      "loss": 0.0982,
      "step": 1831
    },
    {
      "epoch": 4.228505481823428,
      "grad_norm": 0.08349762111902237,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 0.0943,
      "step": 1832
    },
    {
      "epoch": 4.230813618003462,
      "grad_norm": 0.08217111974954605,
      "learning_rate": 3.074074074074074e-05,
      "loss": 0.0958,
      "step": 1833
    },
    {
      "epoch": 4.233121754183497,
      "grad_norm": 0.08965833485126495,
      "learning_rate": 3.064814814814815e-05,
      "loss": 0.0979,
      "step": 1834
    },
    {
      "epoch": 4.235429890363531,
      "grad_norm": 0.08474937826395035,
      "learning_rate": 3.055555555555556e-05,
      "loss": 0.0983,
      "step": 1835
    },
    {
      "epoch": 4.237738026543566,
      "grad_norm": 0.0892176553606987,
      "learning_rate": 3.0462962962962964e-05,
      "loss": 0.0959,
      "step": 1836
    },
    {
      "epoch": 4.2400461627236,
      "grad_norm": 0.08364210277795792,
      "learning_rate": 3.037037037037037e-05,
      "loss": 0.0974,
      "step": 1837
    },
    {
      "epoch": 4.242354298903635,
      "grad_norm": 0.09258552640676498,
      "learning_rate": 3.0277777777777776e-05,
      "loss": 0.1005,
      "step": 1838
    },
    {
      "epoch": 4.24466243508367,
      "grad_norm": 0.08792300522327423,
      "learning_rate": 3.018518518518519e-05,
      "loss": 0.0943,
      "step": 1839
    },
    {
      "epoch": 4.246970571263705,
      "grad_norm": 0.07948849350214005,
      "learning_rate": 3.0092592592592593e-05,
      "loss": 0.0918,
      "step": 1840
    },
    {
      "epoch": 4.2492787074437395,
      "grad_norm": 0.08429206162691116,
      "learning_rate": 3e-05,
      "loss": 0.0953,
      "step": 1841
    },
    {
      "epoch": 4.251586843623774,
      "grad_norm": 0.08649352192878723,
      "learning_rate": 2.9907407407407405e-05,
      "loss": 0.0899,
      "step": 1842
    },
    {
      "epoch": 4.253894979803809,
      "grad_norm": 0.0803857296705246,
      "learning_rate": 2.981481481481482e-05,
      "loss": 0.0913,
      "step": 1843
    },
    {
      "epoch": 4.256203115983843,
      "grad_norm": 0.08873006701469421,
      "learning_rate": 2.9722222222222223e-05,
      "loss": 0.0894,
      "step": 1844
    },
    {
      "epoch": 4.258511252163878,
      "grad_norm": 0.0821148231625557,
      "learning_rate": 2.962962962962963e-05,
      "loss": 0.0979,
      "step": 1845
    },
    {
      "epoch": 4.260819388343912,
      "grad_norm": 0.08680618554353714,
      "learning_rate": 2.953703703703704e-05,
      "loss": 0.0909,
      "step": 1846
    },
    {
      "epoch": 4.263127524523947,
      "grad_norm": 0.08379969000816345,
      "learning_rate": 2.9444444444444448e-05,
      "loss": 0.0926,
      "step": 1847
    },
    {
      "epoch": 4.265435660703981,
      "grad_norm": 0.08866576850414276,
      "learning_rate": 2.9351851851851852e-05,
      "loss": 0.0939,
      "step": 1848
    },
    {
      "epoch": 4.267743796884016,
      "grad_norm": 0.07816161960363388,
      "learning_rate": 2.925925925925926e-05,
      "loss": 0.0929,
      "step": 1849
    },
    {
      "epoch": 4.2700519330640505,
      "grad_norm": 0.07924585044384003,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.0935,
      "step": 1850
    },
    {
      "epoch": 4.272360069244085,
      "grad_norm": 0.08286602050065994,
      "learning_rate": 2.9074074074074077e-05,
      "loss": 0.0977,
      "step": 1851
    },
    {
      "epoch": 4.27466820542412,
      "grad_norm": 0.08040314167737961,
      "learning_rate": 2.898148148148148e-05,
      "loss": 0.1027,
      "step": 1852
    },
    {
      "epoch": 4.276976341604154,
      "grad_norm": 0.08347471058368683,
      "learning_rate": 2.8888888888888888e-05,
      "loss": 0.0894,
      "step": 1853
    },
    {
      "epoch": 4.27928447778419,
      "grad_norm": 0.0840911865234375,
      "learning_rate": 2.87962962962963e-05,
      "loss": 0.0888,
      "step": 1854
    },
    {
      "epoch": 4.281592613964224,
      "grad_norm": 0.08661846816539764,
      "learning_rate": 2.8703703703703706e-05,
      "loss": 0.0962,
      "step": 1855
    },
    {
      "epoch": 4.283900750144259,
      "grad_norm": 0.08302923291921616,
      "learning_rate": 2.861111111111111e-05,
      "loss": 0.0914,
      "step": 1856
    },
    {
      "epoch": 4.286208886324293,
      "grad_norm": 0.08224494010210037,
      "learning_rate": 2.851851851851852e-05,
      "loss": 0.1001,
      "step": 1857
    },
    {
      "epoch": 4.288517022504328,
      "grad_norm": 0.08105452358722687,
      "learning_rate": 2.8425925925925928e-05,
      "loss": 0.0925,
      "step": 1858
    },
    {
      "epoch": 4.290825158684362,
      "grad_norm": 0.0788198858499527,
      "learning_rate": 2.8333333333333335e-05,
      "loss": 0.1007,
      "step": 1859
    },
    {
      "epoch": 4.293133294864397,
      "grad_norm": 0.07947266101837158,
      "learning_rate": 2.824074074074074e-05,
      "loss": 0.0994,
      "step": 1860
    },
    {
      "epoch": 4.2954414310444315,
      "grad_norm": 0.08335360884666443,
      "learning_rate": 2.814814814814815e-05,
      "loss": 0.0986,
      "step": 1861
    },
    {
      "epoch": 4.297749567224466,
      "grad_norm": 0.08139172941446304,
      "learning_rate": 2.8055555555555557e-05,
      "loss": 0.0986,
      "step": 1862
    },
    {
      "epoch": 4.300057703404501,
      "grad_norm": 0.0862392708659172,
      "learning_rate": 2.7962962962962965e-05,
      "loss": 0.0954,
      "step": 1863
    },
    {
      "epoch": 4.302365839584535,
      "grad_norm": 0.08523052930831909,
      "learning_rate": 2.7870370370370375e-05,
      "loss": 0.1042,
      "step": 1864
    },
    {
      "epoch": 4.30467397576457,
      "grad_norm": 0.08763065934181213,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0937,
      "step": 1865
    },
    {
      "epoch": 4.306982111944604,
      "grad_norm": 0.08660484105348587,
      "learning_rate": 2.7685185185185186e-05,
      "loss": 0.0952,
      "step": 1866
    },
    {
      "epoch": 4.30929024812464,
      "grad_norm": 0.08034911751747131,
      "learning_rate": 2.7592592592592594e-05,
      "loss": 0.1071,
      "step": 1867
    },
    {
      "epoch": 4.311598384304674,
      "grad_norm": 0.08272317796945572,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.0968,
      "step": 1868
    },
    {
      "epoch": 4.313906520484709,
      "grad_norm": 0.0832274928689003,
      "learning_rate": 2.7407407407407408e-05,
      "loss": 0.0941,
      "step": 1869
    },
    {
      "epoch": 4.3162146566647435,
      "grad_norm": 0.08511745184659958,
      "learning_rate": 2.7314814814814816e-05,
      "loss": 0.095,
      "step": 1870
    },
    {
      "epoch": 4.318522792844778,
      "grad_norm": 0.08415031433105469,
      "learning_rate": 2.7222222222222223e-05,
      "loss": 0.0972,
      "step": 1871
    },
    {
      "epoch": 4.320830929024813,
      "grad_norm": 0.0891333594918251,
      "learning_rate": 2.7129629629629634e-05,
      "loss": 0.0969,
      "step": 1872
    },
    {
      "epoch": 4.323139065204847,
      "grad_norm": 0.08300217986106873,
      "learning_rate": 2.7037037037037037e-05,
      "loss": 0.0952,
      "step": 1873
    },
    {
      "epoch": 4.325447201384882,
      "grad_norm": 0.0837203860282898,
      "learning_rate": 2.6944444444444445e-05,
      "loss": 0.0964,
      "step": 1874
    },
    {
      "epoch": 4.327755337564916,
      "grad_norm": 0.08347982913255692,
      "learning_rate": 2.6851851851851855e-05,
      "loss": 0.094,
      "step": 1875
    },
    {
      "epoch": 4.330063473744951,
      "grad_norm": 0.08197362720966339,
      "learning_rate": 2.6759259259259263e-05,
      "loss": 0.097,
      "step": 1876
    },
    {
      "epoch": 4.332371609924985,
      "grad_norm": 0.08208876848220825,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.0949,
      "step": 1877
    },
    {
      "epoch": 4.33467974610502,
      "grad_norm": 0.08793433755636215,
      "learning_rate": 2.6574074074074074e-05,
      "loss": 0.091,
      "step": 1878
    },
    {
      "epoch": 4.3369878822850545,
      "grad_norm": 0.086302749812603,
      "learning_rate": 2.6481481481481485e-05,
      "loss": 0.0898,
      "step": 1879
    },
    {
      "epoch": 4.339296018465089,
      "grad_norm": 0.08742526918649673,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 0.0927,
      "step": 1880
    },
    {
      "epoch": 4.341604154645124,
      "grad_norm": 0.08446653932332993,
      "learning_rate": 2.6296296296296296e-05,
      "loss": 0.0995,
      "step": 1881
    },
    {
      "epoch": 4.343912290825159,
      "grad_norm": 0.08229253441095352,
      "learning_rate": 2.6203703703703703e-05,
      "loss": 0.0981,
      "step": 1882
    },
    {
      "epoch": 4.346220427005194,
      "grad_norm": 0.08265943080186844,
      "learning_rate": 2.6111111111111114e-05,
      "loss": 0.0938,
      "step": 1883
    },
    {
      "epoch": 4.348528563185228,
      "grad_norm": 0.08226665109395981,
      "learning_rate": 2.601851851851852e-05,
      "loss": 0.0926,
      "step": 1884
    },
    {
      "epoch": 4.350836699365263,
      "grad_norm": 0.08619023859500885,
      "learning_rate": 2.5925925925925925e-05,
      "loss": 0.0849,
      "step": 1885
    },
    {
      "epoch": 4.353144835545297,
      "grad_norm": 0.08823271840810776,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 0.0913,
      "step": 1886
    },
    {
      "epoch": 4.355452971725332,
      "grad_norm": 0.07949109375476837,
      "learning_rate": 2.5740740740740743e-05,
      "loss": 0.0899,
      "step": 1887
    },
    {
      "epoch": 4.357761107905366,
      "grad_norm": 0.08263242989778519,
      "learning_rate": 2.564814814814815e-05,
      "loss": 0.0932,
      "step": 1888
    },
    {
      "epoch": 4.360069244085401,
      "grad_norm": 0.09101185202598572,
      "learning_rate": 2.5555555555555554e-05,
      "loss": 0.0909,
      "step": 1889
    },
    {
      "epoch": 4.3623773802654355,
      "grad_norm": 0.08915597200393677,
      "learning_rate": 2.5462962962962965e-05,
      "loss": 0.0973,
      "step": 1890
    },
    {
      "epoch": 4.36468551644547,
      "grad_norm": 0.08705798536539078,
      "learning_rate": 2.5370370370370372e-05,
      "loss": 0.093,
      "step": 1891
    },
    {
      "epoch": 4.366993652625505,
      "grad_norm": 0.08258432149887085,
      "learning_rate": 2.527777777777778e-05,
      "loss": 0.0977,
      "step": 1892
    },
    {
      "epoch": 4.369301788805539,
      "grad_norm": 0.08850180357694626,
      "learning_rate": 2.5185185185185183e-05,
      "loss": 0.0984,
      "step": 1893
    },
    {
      "epoch": 4.371609924985574,
      "grad_norm": 0.08396847546100616,
      "learning_rate": 2.5092592592592594e-05,
      "loss": 0.0998,
      "step": 1894
    },
    {
      "epoch": 4.373918061165609,
      "grad_norm": 0.08599070459604263,
      "learning_rate": 2.5e-05,
      "loss": 0.0894,
      "step": 1895
    },
    {
      "epoch": 4.376226197345644,
      "grad_norm": 0.0811719074845314,
      "learning_rate": 2.490740740740741e-05,
      "loss": 0.0985,
      "step": 1896
    },
    {
      "epoch": 4.378534333525678,
      "grad_norm": 0.08406871557235718,
      "learning_rate": 2.4814814814814816e-05,
      "loss": 0.0952,
      "step": 1897
    },
    {
      "epoch": 4.380842469705713,
      "grad_norm": 0.08697894215583801,
      "learning_rate": 2.4722222222222223e-05,
      "loss": 0.0976,
      "step": 1898
    },
    {
      "epoch": 4.3831506058857475,
      "grad_norm": 0.0841851755976677,
      "learning_rate": 2.462962962962963e-05,
      "loss": 0.0912,
      "step": 1899
    },
    {
      "epoch": 4.385458742065782,
      "grad_norm": 0.08065333217382431,
      "learning_rate": 2.4537037037037038e-05,
      "loss": 0.097,
      "step": 1900
    },
    {
      "epoch": 4.387766878245817,
      "grad_norm": 0.0868447944521904,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.0891,
      "step": 1901
    },
    {
      "epoch": 4.390075014425851,
      "grad_norm": 0.08101449906826019,
      "learning_rate": 2.4351851851851852e-05,
      "loss": 0.091,
      "step": 1902
    },
    {
      "epoch": 4.392383150605886,
      "grad_norm": 0.08216770738363266,
      "learning_rate": 2.425925925925926e-05,
      "loss": 0.1032,
      "step": 1903
    },
    {
      "epoch": 4.39469128678592,
      "grad_norm": 0.08083458989858627,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.0923,
      "step": 1904
    },
    {
      "epoch": 4.396999422965955,
      "grad_norm": 0.08226275444030762,
      "learning_rate": 2.4074074074074074e-05,
      "loss": 0.0908,
      "step": 1905
    },
    {
      "epoch": 4.399307559145989,
      "grad_norm": 0.08051864057779312,
      "learning_rate": 2.398148148148148e-05,
      "loss": 0.0963,
      "step": 1906
    },
    {
      "epoch": 4.401615695326024,
      "grad_norm": 0.07815883308649063,
      "learning_rate": 2.3888888888888892e-05,
      "loss": 0.0994,
      "step": 1907
    },
    {
      "epoch": 4.4039238315060585,
      "grad_norm": 0.08551505953073502,
      "learning_rate": 2.3796296296296296e-05,
      "loss": 0.0975,
      "step": 1908
    },
    {
      "epoch": 4.406231967686093,
      "grad_norm": 0.0826415941119194,
      "learning_rate": 2.3703703703703707e-05,
      "loss": 0.104,
      "step": 1909
    },
    {
      "epoch": 4.4085401038661285,
      "grad_norm": 0.08316341787576675,
      "learning_rate": 2.361111111111111e-05,
      "loss": 0.0961,
      "step": 1910
    },
    {
      "epoch": 4.410848240046163,
      "grad_norm": 0.0814400389790535,
      "learning_rate": 2.351851851851852e-05,
      "loss": 0.0957,
      "step": 1911
    },
    {
      "epoch": 4.413156376226198,
      "grad_norm": 0.08011600375175476,
      "learning_rate": 2.3425925925925925e-05,
      "loss": 0.0938,
      "step": 1912
    },
    {
      "epoch": 4.415464512406232,
      "grad_norm": 0.07995935529470444,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.0904,
      "step": 1913
    },
    {
      "epoch": 4.417772648586267,
      "grad_norm": 0.08290641009807587,
      "learning_rate": 2.324074074074074e-05,
      "loss": 0.098,
      "step": 1914
    },
    {
      "epoch": 4.420080784766301,
      "grad_norm": 0.08179384469985962,
      "learning_rate": 2.314814814814815e-05,
      "loss": 0.089,
      "step": 1915
    },
    {
      "epoch": 4.422388920946336,
      "grad_norm": 0.08062732219696045,
      "learning_rate": 2.3055555555555558e-05,
      "loss": 0.0983,
      "step": 1916
    },
    {
      "epoch": 4.42469705712637,
      "grad_norm": 0.08097699284553528,
      "learning_rate": 2.2962962962962965e-05,
      "loss": 0.0996,
      "step": 1917
    },
    {
      "epoch": 4.427005193306405,
      "grad_norm": 0.0813087522983551,
      "learning_rate": 2.2870370370370372e-05,
      "loss": 0.0954,
      "step": 1918
    },
    {
      "epoch": 4.4293133294864395,
      "grad_norm": 0.08182039856910706,
      "learning_rate": 2.277777777777778e-05,
      "loss": 0.0844,
      "step": 1919
    },
    {
      "epoch": 4.431621465666474,
      "grad_norm": 0.0812203511595726,
      "learning_rate": 2.2685185185185187e-05,
      "loss": 0.0958,
      "step": 1920
    },
    {
      "epoch": 4.433929601846509,
      "grad_norm": 0.07774294167757034,
      "learning_rate": 2.2592592592592594e-05,
      "loss": 0.0989,
      "step": 1921
    },
    {
      "epoch": 4.436237738026543,
      "grad_norm": 0.08072575181722641,
      "learning_rate": 2.25e-05,
      "loss": 0.0897,
      "step": 1922
    },
    {
      "epoch": 4.438545874206579,
      "grad_norm": 0.07977304607629776,
      "learning_rate": 2.240740740740741e-05,
      "loss": 0.0979,
      "step": 1923
    },
    {
      "epoch": 4.440854010386613,
      "grad_norm": 0.08006980270147324,
      "learning_rate": 2.2314814814814816e-05,
      "loss": 0.0939,
      "step": 1924
    },
    {
      "epoch": 4.443162146566648,
      "grad_norm": 0.08098588138818741,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.0949,
      "step": 1925
    },
    {
      "epoch": 4.445470282746682,
      "grad_norm": 0.0841732993721962,
      "learning_rate": 2.212962962962963e-05,
      "loss": 0.0987,
      "step": 1926
    },
    {
      "epoch": 4.447778418926717,
      "grad_norm": 0.08010715246200562,
      "learning_rate": 2.2037037037037038e-05,
      "loss": 0.093,
      "step": 1927
    },
    {
      "epoch": 4.450086555106751,
      "grad_norm": 0.08065465092658997,
      "learning_rate": 2.1944444444444445e-05,
      "loss": 0.0937,
      "step": 1928
    },
    {
      "epoch": 4.452394691286786,
      "grad_norm": 0.08388105779886246,
      "learning_rate": 2.1851851851851852e-05,
      "loss": 0.0887,
      "step": 1929
    },
    {
      "epoch": 4.454702827466821,
      "grad_norm": 0.08158472925424576,
      "learning_rate": 2.175925925925926e-05,
      "loss": 0.0996,
      "step": 1930
    },
    {
      "epoch": 4.457010963646855,
      "grad_norm": 0.0839938223361969,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 0.0977,
      "step": 1931
    },
    {
      "epoch": 4.45931909982689,
      "grad_norm": 0.0807269811630249,
      "learning_rate": 2.1574074074074074e-05,
      "loss": 0.0963,
      "step": 1932
    },
    {
      "epoch": 4.461627236006924,
      "grad_norm": 0.0856325775384903,
      "learning_rate": 2.148148148148148e-05,
      "loss": 0.091,
      "step": 1933
    },
    {
      "epoch": 4.463935372186959,
      "grad_norm": 0.08232910186052322,
      "learning_rate": 2.138888888888889e-05,
      "loss": 0.0987,
      "step": 1934
    },
    {
      "epoch": 4.466243508366993,
      "grad_norm": 0.08208692073822021,
      "learning_rate": 2.1296296296296296e-05,
      "loss": 0.0917,
      "step": 1935
    },
    {
      "epoch": 4.468551644547028,
      "grad_norm": 0.08460013568401337,
      "learning_rate": 2.1203703703703703e-05,
      "loss": 0.0932,
      "step": 1936
    },
    {
      "epoch": 4.4708597807270625,
      "grad_norm": 0.08107100427150726,
      "learning_rate": 2.111111111111111e-05,
      "loss": 0.0924,
      "step": 1937
    },
    {
      "epoch": 4.473167916907098,
      "grad_norm": 0.08559548854827881,
      "learning_rate": 2.101851851851852e-05,
      "loss": 0.0985,
      "step": 1938
    },
    {
      "epoch": 4.4754760530871325,
      "grad_norm": 0.08470273017883301,
      "learning_rate": 2.0925925925925925e-05,
      "loss": 0.0985,
      "step": 1939
    },
    {
      "epoch": 4.477784189267167,
      "grad_norm": 0.0793725922703743,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0998,
      "step": 1940
    },
    {
      "epoch": 4.480092325447202,
      "grad_norm": 0.08811813592910767,
      "learning_rate": 2.074074074074074e-05,
      "loss": 0.0935,
      "step": 1941
    },
    {
      "epoch": 4.482400461627236,
      "grad_norm": 0.0838591456413269,
      "learning_rate": 2.064814814814815e-05,
      "loss": 0.0972,
      "step": 1942
    },
    {
      "epoch": 4.484708597807271,
      "grad_norm": 0.08574528992176056,
      "learning_rate": 2.0555555555555555e-05,
      "loss": 0.0856,
      "step": 1943
    },
    {
      "epoch": 4.487016733987305,
      "grad_norm": 0.08581560105085373,
      "learning_rate": 2.0462962962962965e-05,
      "loss": 0.0925,
      "step": 1944
    },
    {
      "epoch": 4.48932487016734,
      "grad_norm": 0.08401628583669662,
      "learning_rate": 2.037037037037037e-05,
      "loss": 0.0913,
      "step": 1945
    },
    {
      "epoch": 4.491633006347374,
      "grad_norm": 0.08299828320741653,
      "learning_rate": 2.027777777777778e-05,
      "loss": 0.0957,
      "step": 1946
    },
    {
      "epoch": 4.493941142527409,
      "grad_norm": 0.07773829996585846,
      "learning_rate": 2.0185185185185187e-05,
      "loss": 0.1006,
      "step": 1947
    },
    {
      "epoch": 4.4962492787074435,
      "grad_norm": 0.0813092365860939,
      "learning_rate": 2.0092592592592594e-05,
      "loss": 0.0957,
      "step": 1948
    },
    {
      "epoch": 4.498557414887478,
      "grad_norm": 0.07776177674531937,
      "learning_rate": 2e-05,
      "loss": 0.0977,
      "step": 1949
    },
    {
      "epoch": 4.500865551067513,
      "grad_norm": 0.08157168328762054,
      "learning_rate": 1.990740740740741e-05,
      "loss": 0.0905,
      "step": 1950
    },
    {
      "epoch": 4.503173687247548,
      "grad_norm": 0.08216714859008789,
      "learning_rate": 1.9814814814814816e-05,
      "loss": 0.0969,
      "step": 1951
    },
    {
      "epoch": 4.505481823427582,
      "grad_norm": 0.08214664459228516,
      "learning_rate": 1.9722222222222224e-05,
      "loss": 0.0922,
      "step": 1952
    },
    {
      "epoch": 4.507789959607617,
      "grad_norm": 0.08025806397199631,
      "learning_rate": 1.962962962962963e-05,
      "loss": 0.0945,
      "step": 1953
    },
    {
      "epoch": 4.510098095787652,
      "grad_norm": 0.08112560212612152,
      "learning_rate": 1.9537037037037038e-05,
      "loss": 0.0972,
      "step": 1954
    },
    {
      "epoch": 4.512406231967686,
      "grad_norm": 0.08225028961896896,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.0948,
      "step": 1955
    },
    {
      "epoch": 4.514714368147721,
      "grad_norm": 0.08371096849441528,
      "learning_rate": 1.9351851851851853e-05,
      "loss": 0.0981,
      "step": 1956
    },
    {
      "epoch": 4.517022504327755,
      "grad_norm": 0.08655606955289841,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.0932,
      "step": 1957
    },
    {
      "epoch": 4.51933064050779,
      "grad_norm": 0.08565112948417664,
      "learning_rate": 1.9166666666666667e-05,
      "loss": 0.0922,
      "step": 1958
    },
    {
      "epoch": 4.5216387766878245,
      "grad_norm": 0.08371853083372116,
      "learning_rate": 1.9074074074074075e-05,
      "loss": 0.0914,
      "step": 1959
    },
    {
      "epoch": 4.523946912867859,
      "grad_norm": 0.08019944280385971,
      "learning_rate": 1.8981481481481482e-05,
      "loss": 0.0959,
      "step": 1960
    },
    {
      "epoch": 4.526255049047894,
      "grad_norm": 0.08441659808158875,
      "learning_rate": 1.888888888888889e-05,
      "loss": 0.0927,
      "step": 1961
    },
    {
      "epoch": 4.528563185227928,
      "grad_norm": 0.08053024858236313,
      "learning_rate": 1.8796296296296296e-05,
      "loss": 0.0933,
      "step": 1962
    },
    {
      "epoch": 4.530871321407963,
      "grad_norm": 0.08257871121168137,
      "learning_rate": 1.8703703703703704e-05,
      "loss": 0.0989,
      "step": 1963
    },
    {
      "epoch": 4.533179457587997,
      "grad_norm": 0.07784412056207657,
      "learning_rate": 1.861111111111111e-05,
      "loss": 0.1028,
      "step": 1964
    },
    {
      "epoch": 4.535487593768032,
      "grad_norm": 0.08487147092819214,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0877,
      "step": 1965
    },
    {
      "epoch": 4.537795729948067,
      "grad_norm": 0.08670619875192642,
      "learning_rate": 1.8425925925925926e-05,
      "loss": 0.0913,
      "step": 1966
    },
    {
      "epoch": 4.540103866128102,
      "grad_norm": 0.08406758308410645,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.0933,
      "step": 1967
    },
    {
      "epoch": 4.5424120023081365,
      "grad_norm": 0.0869966372847557,
      "learning_rate": 1.824074074074074e-05,
      "loss": 0.0921,
      "step": 1968
    },
    {
      "epoch": 4.544720138488171,
      "grad_norm": 0.09148359298706055,
      "learning_rate": 1.814814814814815e-05,
      "loss": 0.0933,
      "step": 1969
    },
    {
      "epoch": 4.547028274668206,
      "grad_norm": 0.08378588408231735,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 0.1013,
      "step": 1970
    },
    {
      "epoch": 4.54933641084824,
      "grad_norm": 0.08789436519145966,
      "learning_rate": 1.7962962962962965e-05,
      "loss": 0.0908,
      "step": 1971
    },
    {
      "epoch": 4.551644547028275,
      "grad_norm": 0.08298159390687943,
      "learning_rate": 1.787037037037037e-05,
      "loss": 0.0957,
      "step": 1972
    },
    {
      "epoch": 4.553952683208309,
      "grad_norm": 0.08107160776853561,
      "learning_rate": 1.777777777777778e-05,
      "loss": 0.0948,
      "step": 1973
    },
    {
      "epoch": 4.556260819388344,
      "grad_norm": 0.08678272366523743,
      "learning_rate": 1.7685185185185184e-05,
      "loss": 0.0918,
      "step": 1974
    },
    {
      "epoch": 4.558568955568378,
      "grad_norm": 0.08468597382307053,
      "learning_rate": 1.7592592592592595e-05,
      "loss": 0.0863,
      "step": 1975
    },
    {
      "epoch": 4.560877091748413,
      "grad_norm": 0.08039715886116028,
      "learning_rate": 1.75e-05,
      "loss": 0.0822,
      "step": 1976
    },
    {
      "epoch": 4.5631852279284475,
      "grad_norm": 0.0823940709233284,
      "learning_rate": 1.740740740740741e-05,
      "loss": 0.0939,
      "step": 1977
    },
    {
      "epoch": 4.565493364108482,
      "grad_norm": 0.08963687717914581,
      "learning_rate": 1.7314814814814813e-05,
      "loss": 0.1038,
      "step": 1978
    },
    {
      "epoch": 4.5678015002885175,
      "grad_norm": 0.08346489816904068,
      "learning_rate": 1.7222222222222224e-05,
      "loss": 0.0972,
      "step": 1979
    },
    {
      "epoch": 4.570109636468551,
      "grad_norm": 0.08698716759681702,
      "learning_rate": 1.712962962962963e-05,
      "loss": 0.093,
      "step": 1980
    },
    {
      "epoch": 4.572417772648587,
      "grad_norm": 0.09004484117031097,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 0.0961,
      "step": 1981
    },
    {
      "epoch": 4.574725908828621,
      "grad_norm": 0.08015062659978867,
      "learning_rate": 1.6944444444444446e-05,
      "loss": 0.0926,
      "step": 1982
    },
    {
      "epoch": 4.577034045008656,
      "grad_norm": 0.08224586397409439,
      "learning_rate": 1.6851851851851853e-05,
      "loss": 0.0907,
      "step": 1983
    },
    {
      "epoch": 4.57934218118869,
      "grad_norm": 0.08724253624677658,
      "learning_rate": 1.675925925925926e-05,
      "loss": 0.0867,
      "step": 1984
    },
    {
      "epoch": 4.581650317368725,
      "grad_norm": 0.07941649854183197,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0911,
      "step": 1985
    },
    {
      "epoch": 4.583958453548759,
      "grad_norm": 0.0814509466290474,
      "learning_rate": 1.6574074074074075e-05,
      "loss": 0.0942,
      "step": 1986
    },
    {
      "epoch": 4.586266589728794,
      "grad_norm": 0.08643992245197296,
      "learning_rate": 1.6481481481481482e-05,
      "loss": 0.1009,
      "step": 1987
    },
    {
      "epoch": 4.5885747259088285,
      "grad_norm": 0.08291400969028473,
      "learning_rate": 1.638888888888889e-05,
      "loss": 0.0937,
      "step": 1988
    },
    {
      "epoch": 4.590882862088863,
      "grad_norm": 0.07834281772375107,
      "learning_rate": 1.62962962962963e-05,
      "loss": 0.0957,
      "step": 1989
    },
    {
      "epoch": 4.593190998268898,
      "grad_norm": 0.08046448975801468,
      "learning_rate": 1.6203703703703704e-05,
      "loss": 0.0988,
      "step": 1990
    },
    {
      "epoch": 4.595499134448932,
      "grad_norm": 0.08041954040527344,
      "learning_rate": 1.6111111111111115e-05,
      "loss": 0.0958,
      "step": 1991
    },
    {
      "epoch": 4.597807270628967,
      "grad_norm": 0.07888827472925186,
      "learning_rate": 1.601851851851852e-05,
      "loss": 0.0985,
      "step": 1992
    },
    {
      "epoch": 4.600115406809001,
      "grad_norm": 0.08588582277297974,
      "learning_rate": 1.5925925925925926e-05,
      "loss": 0.0879,
      "step": 1993
    },
    {
      "epoch": 4.602423542989037,
      "grad_norm": 0.08303145319223404,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.0954,
      "step": 1994
    },
    {
      "epoch": 4.604731679169071,
      "grad_norm": 0.08368394523859024,
      "learning_rate": 1.574074074074074e-05,
      "loss": 0.0956,
      "step": 1995
    },
    {
      "epoch": 4.607039815349106,
      "grad_norm": 0.07943252474069595,
      "learning_rate": 1.5648148148148148e-05,
      "loss": 0.0919,
      "step": 1996
    },
    {
      "epoch": 4.6093479515291405,
      "grad_norm": 0.0844736322760582,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.0951,
      "step": 1997
    },
    {
      "epoch": 4.611656087709175,
      "grad_norm": 0.08402314782142639,
      "learning_rate": 1.5462962962962962e-05,
      "loss": 0.0943,
      "step": 1998
    },
    {
      "epoch": 4.61396422388921,
      "grad_norm": 0.08060862869024277,
      "learning_rate": 1.537037037037037e-05,
      "loss": 0.0957,
      "step": 1999
    },
    {
      "epoch": 4.616272360069244,
      "grad_norm": 0.08488716185092926,
      "learning_rate": 1.527777777777778e-05,
      "loss": 0.0942,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 2165,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.498798314937909e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}