{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8193643098562698,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005462428732375132,
      "grad_norm": 9.9274875537958e-05,
      "learning_rate": 2e-05,
      "loss": 11.5,
      "step": 1
    },
    {
      "epoch": 0.0010924857464750264,
      "grad_norm": 8.736815652810037e-05,
      "learning_rate": 4e-05,
      "loss": 11.5,
      "step": 2
    },
    {
      "epoch": 0.0016387286197125396,
      "grad_norm": 9.104706259677187e-05,
      "learning_rate": 6e-05,
      "loss": 11.5,
      "step": 3
    },
    {
      "epoch": 0.0021849714929500527,
      "grad_norm": 8.75555124366656e-05,
      "learning_rate": 8e-05,
      "loss": 11.5,
      "step": 4
    },
    {
      "epoch": 0.002731214366187566,
      "grad_norm": 0.00015634573355782777,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 5
    },
    {
      "epoch": 0.0032774572394250793,
      "grad_norm": 6.726804713252932e-05,
      "learning_rate": 0.00012,
      "loss": 11.5,
      "step": 6
    },
    {
      "epoch": 0.0038237001126625926,
      "grad_norm": 8.250509563367814e-05,
      "learning_rate": 0.00014,
      "loss": 11.5,
      "step": 7
    },
    {
      "epoch": 0.004369942985900105,
      "grad_norm": 8.2841856055893e-05,
      "learning_rate": 0.00016,
      "loss": 11.5,
      "step": 8
    },
    {
      "epoch": 0.004916185859137619,
      "grad_norm": 7.09330925019458e-05,
      "learning_rate": 0.00018,
      "loss": 11.5,
      "step": 9
    },
    {
      "epoch": 0.005462428732375132,
      "grad_norm": 6.527874211315066e-05,
      "learning_rate": 0.0002,
      "loss": 11.5,
      "step": 10
    },
    {
      "epoch": 0.006008671605612645,
      "grad_norm": 0.00012205211533000693,
      "learning_rate": 0.00019999977772170748,
      "loss": 11.5,
      "step": 11
    },
    {
      "epoch": 0.0065549144788501585,
      "grad_norm": 8.729364344617352e-05,
      "learning_rate": 0.00019999911088781805,
      "loss": 11.5,
      "step": 12
    },
    {
      "epoch": 0.007101157352087672,
      "grad_norm": 8.821779192658141e-05,
      "learning_rate": 0.0001999979995012962,
      "loss": 11.5,
      "step": 13
    },
    {
      "epoch": 0.007647400225325185,
      "grad_norm": 0.00010153299808735028,
      "learning_rate": 0.00019999644356708261,
      "loss": 11.5,
      "step": 14
    },
    {
      "epoch": 0.008193643098562698,
      "grad_norm": 0.00010531259613344446,
      "learning_rate": 0.00019999444309209432,
      "loss": 11.5,
      "step": 15
    },
    {
      "epoch": 0.00873988597180021,
      "grad_norm": 0.00010518762428546324,
      "learning_rate": 0.0001999919980852246,
      "loss": 11.5,
      "step": 16
    },
    {
      "epoch": 0.009286128845037725,
      "grad_norm": 9.567197412252426e-05,
      "learning_rate": 0.00019998910855734288,
      "loss": 11.5,
      "step": 17
    },
    {
      "epoch": 0.009832371718275237,
      "grad_norm": 0.0001835573639255017,
      "learning_rate": 0.0001999857745212947,
      "loss": 11.5,
      "step": 18
    },
    {
      "epoch": 0.010378614591512752,
      "grad_norm": 0.0002136574184987694,
      "learning_rate": 0.00019998199599190178,
      "loss": 11.5,
      "step": 19
    },
    {
      "epoch": 0.010924857464750264,
      "grad_norm": 0.00010190153261646628,
      "learning_rate": 0.0001999777729859618,
      "loss": 11.5,
      "step": 20
    },
    {
      "epoch": 0.011471100337987778,
      "grad_norm": 0.00012312202306929976,
      "learning_rate": 0.00019997310552224846,
      "loss": 11.5,
      "step": 21
    },
    {
      "epoch": 0.01201734321122529,
      "grad_norm": 7.525176624767482e-05,
      "learning_rate": 0.00019996799362151122,
      "loss": 11.5,
      "step": 22
    },
    {
      "epoch": 0.012563586084462805,
      "grad_norm": 0.00012282673560548574,
      "learning_rate": 0.00019996243730647538,
      "loss": 11.5,
      "step": 23
    },
    {
      "epoch": 0.013109828957700317,
      "grad_norm": 9.848399349721149e-05,
      "learning_rate": 0.00019995643660184191,
      "loss": 11.5,
      "step": 24
    },
    {
      "epoch": 0.013656071830937831,
      "grad_norm": 9.144116484094411e-05,
      "learning_rate": 0.00019994999153428737,
      "loss": 11.5,
      "step": 25
    },
    {
      "epoch": 0.014202314704175344,
      "grad_norm": 0.0002811682061292231,
      "learning_rate": 0.00019994310213246368,
      "loss": 11.5,
      "step": 26
    },
    {
      "epoch": 0.014748557577412858,
      "grad_norm": 0.0004516992485150695,
      "learning_rate": 0.00019993576842699816,
      "loss": 11.5,
      "step": 27
    },
    {
      "epoch": 0.01529480045065037,
      "grad_norm": 0.00014355852908920497,
      "learning_rate": 0.0001999279904504933,
      "loss": 11.5,
      "step": 28
    },
    {
      "epoch": 0.015841043323887884,
      "grad_norm": 0.00010362818284193054,
      "learning_rate": 0.00019991976823752653,
      "loss": 11.5,
      "step": 29
    },
    {
      "epoch": 0.016387286197125397,
      "grad_norm": 0.00012890249490737915,
      "learning_rate": 0.00019991110182465032,
      "loss": 11.5,
      "step": 30
    },
    {
      "epoch": 0.01693352907036291,
      "grad_norm": 0.0001249217166332528,
      "learning_rate": 0.00019990199125039174,
      "loss": 11.5,
      "step": 31
    },
    {
      "epoch": 0.01747977194360042,
      "grad_norm": 0.00019562743545975536,
      "learning_rate": 0.00019989243655525247,
      "loss": 11.5,
      "step": 32
    },
    {
      "epoch": 0.018026014816837937,
      "grad_norm": 0.00015727915160823613,
      "learning_rate": 0.00019988243778170853,
      "loss": 11.5,
      "step": 33
    },
    {
      "epoch": 0.01857225769007545,
      "grad_norm": 0.00015102663019206375,
      "learning_rate": 0.0001998719949742101,
      "loss": 11.5,
      "step": 34
    },
    {
      "epoch": 0.019118500563312962,
      "grad_norm": 0.00024164341448340565,
      "learning_rate": 0.0001998611081791814,
      "loss": 11.5,
      "step": 35
    },
    {
      "epoch": 0.019664743436550475,
      "grad_norm": 0.00024841472622938454,
      "learning_rate": 0.00019984977744502038,
      "loss": 11.5,
      "step": 36
    },
    {
      "epoch": 0.02021098630978799,
      "grad_norm": 0.00020485970890149474,
      "learning_rate": 0.00019983800282209857,
      "loss": 11.5,
      "step": 37
    },
    {
      "epoch": 0.020757229183025503,
      "grad_norm": 0.00030442627030424774,
      "learning_rate": 0.00019982578436276082,
      "loss": 11.5,
      "step": 38
    },
    {
      "epoch": 0.021303472056263015,
      "grad_norm": 0.00017282631597481668,
      "learning_rate": 0.00019981312212132512,
      "loss": 11.5,
      "step": 39
    },
    {
      "epoch": 0.021849714929500528,
      "grad_norm": 0.00023939376114867628,
      "learning_rate": 0.00019980001615408228,
      "loss": 11.5,
      "step": 40
    },
    {
      "epoch": 0.022395957802738044,
      "grad_norm": 0.00030217712628655136,
      "learning_rate": 0.00019978646651929572,
      "loss": 11.5,
      "step": 41
    },
    {
      "epoch": 0.022942200675975556,
      "grad_norm": 0.00027799236704595387,
      "learning_rate": 0.00019977247327720128,
      "loss": 11.5,
      "step": 42
    },
    {
      "epoch": 0.02348844354921307,
      "grad_norm": 0.00022140814689919353,
      "learning_rate": 0.0001997580364900068,
      "loss": 11.5,
      "step": 43
    },
    {
      "epoch": 0.02403468642245058,
      "grad_norm": 0.00044123470433987677,
      "learning_rate": 0.000199743156221892,
      "loss": 11.5,
      "step": 44
    },
    {
      "epoch": 0.024580929295688097,
      "grad_norm": 0.00030213367426767945,
      "learning_rate": 0.00019972783253900808,
      "loss": 11.5,
      "step": 45
    },
    {
      "epoch": 0.02512717216892561,
      "grad_norm": 0.0003121906192973256,
      "learning_rate": 0.00019971206550947748,
      "loss": 11.5,
      "step": 46
    },
    {
      "epoch": 0.02567341504216312,
      "grad_norm": 0.0003478510770946741,
      "learning_rate": 0.00019969585520339354,
      "loss": 11.5,
      "step": 47
    },
    {
      "epoch": 0.026219657915400634,
      "grad_norm": 0.00044502029777504504,
      "learning_rate": 0.0001996792016928203,
      "loss": 11.5,
      "step": 48
    },
    {
      "epoch": 0.026765900788638147,
      "grad_norm": 0.00028886072686873376,
      "learning_rate": 0.00019966210505179197,
      "loss": 11.5,
      "step": 49
    },
    {
      "epoch": 0.027312143661875662,
      "grad_norm": 0.00041241306462325156,
      "learning_rate": 0.00019964456535631286,
      "loss": 11.5,
      "step": 50
    },
    {
      "epoch": 0.027858386535113175,
      "grad_norm": 0.0002404985425528139,
      "learning_rate": 0.0001996265826843568,
      "loss": 11.5,
      "step": 51
    },
    {
      "epoch": 0.028404629408350687,
      "grad_norm": 0.0004042983055114746,
      "learning_rate": 0.00019960815711586696,
      "loss": 11.5,
      "step": 52
    },
    {
      "epoch": 0.0289508722815882,
      "grad_norm": 0.00032950201421044767,
      "learning_rate": 0.00019958928873275539,
      "loss": 11.5,
      "step": 53
    },
    {
      "epoch": 0.029497115154825716,
      "grad_norm": 0.00036795888445340097,
      "learning_rate": 0.00019956997761890277,
      "loss": 11.5,
      "step": 54
    },
    {
      "epoch": 0.030043358028063228,
      "grad_norm": 0.0002794052707031369,
      "learning_rate": 0.00019955022386015792,
      "loss": 11.5,
      "step": 55
    },
    {
      "epoch": 0.03058960090130074,
      "grad_norm": 0.0004315042751841247,
      "learning_rate": 0.00019953002754433743,
      "loss": 11.5,
      "step": 56
    },
    {
      "epoch": 0.031135843774538253,
      "grad_norm": 0.00041837862227112055,
      "learning_rate": 0.00019950938876122542,
      "loss": 11.5,
      "step": 57
    },
    {
      "epoch": 0.03168208664777577,
      "grad_norm": 0.00045636837603524327,
      "learning_rate": 0.00019948830760257291,
      "loss": 11.5,
      "step": 58
    },
    {
      "epoch": 0.03222832952101328,
      "grad_norm": 0.0004122463578823954,
      "learning_rate": 0.0001994667841620976,
      "loss": 11.5,
      "step": 59
    },
    {
      "epoch": 0.032774572394250794,
      "grad_norm": 0.00032944034319370985,
      "learning_rate": 0.00019944481853548335,
      "loss": 11.5,
      "step": 60
    },
    {
      "epoch": 0.033320815267488306,
      "grad_norm": 0.0004234000516589731,
      "learning_rate": 0.00019942241082037982,
      "loss": 11.5,
      "step": 61
    },
    {
      "epoch": 0.03386705814072582,
      "grad_norm": 0.0004000971093773842,
      "learning_rate": 0.00019939956111640197,
      "loss": 11.5,
      "step": 62
    },
    {
      "epoch": 0.03441330101396333,
      "grad_norm": 0.0003795691882260144,
      "learning_rate": 0.00019937626952512964,
      "loss": 11.5,
      "step": 63
    },
    {
      "epoch": 0.03495954388720084,
      "grad_norm": 0.00040721616824157536,
      "learning_rate": 0.0001993525361501072,
      "loss": 11.5,
      "step": 64
    },
    {
      "epoch": 0.03550578676043836,
      "grad_norm": 0.000611544237472117,
      "learning_rate": 0.00019932836109684286,
      "loss": 11.5,
      "step": 65
    },
    {
      "epoch": 0.036052029633675875,
      "grad_norm": 0.00045183743350207806,
      "learning_rate": 0.00019930374447280845,
      "loss": 11.5,
      "step": 66
    },
    {
      "epoch": 0.03659827250691339,
      "grad_norm": 0.0006656524492427707,
      "learning_rate": 0.00019927868638743875,
      "loss": 11.5,
      "step": 67
    },
    {
      "epoch": 0.0371445153801509,
      "grad_norm": 0.0006330456235446036,
      "learning_rate": 0.0001992531869521312,
      "loss": 11.5,
      "step": 68
    },
    {
      "epoch": 0.03769075825338841,
      "grad_norm": 0.0004605160793289542,
      "learning_rate": 0.00019922724628024515,
      "loss": 11.5,
      "step": 69
    },
    {
      "epoch": 0.038237001126625925,
      "grad_norm": 0.0004875901504419744,
      "learning_rate": 0.0001992008644871016,
      "loss": 11.5,
      "step": 70
    },
    {
      "epoch": 0.03878324399986344,
      "grad_norm": 0.0012211997527629137,
      "learning_rate": 0.00019917404168998256,
      "loss": 11.5,
      "step": 71
    },
    {
      "epoch": 0.03932948687310095,
      "grad_norm": 0.000653817318379879,
      "learning_rate": 0.0001991467780081305,
      "loss": 11.5,
      "step": 72
    },
    {
      "epoch": 0.03987572974633847,
      "grad_norm": 0.0003804317966569215,
      "learning_rate": 0.00019911907356274795,
      "loss": 11.5,
      "step": 73
    },
    {
      "epoch": 0.04042197261957598,
      "grad_norm": 0.0004734350077342242,
      "learning_rate": 0.00019909092847699683,
      "loss": 11.5,
      "step": 74
    },
    {
      "epoch": 0.040968215492813494,
      "grad_norm": 0.0006534051499329507,
      "learning_rate": 0.00019906234287599798,
      "loss": 11.5,
      "step": 75
    },
    {
      "epoch": 0.041514458366051006,
      "grad_norm": 0.0009523346088826656,
      "learning_rate": 0.00019903331688683057,
      "loss": 11.5,
      "step": 76
    },
    {
      "epoch": 0.04206070123928852,
      "grad_norm": 0.0007973241154104471,
      "learning_rate": 0.00019900385063853154,
      "loss": 11.5,
      "step": 77
    },
    {
      "epoch": 0.04260694411252603,
      "grad_norm": 0.0008113976800814271,
      "learning_rate": 0.00019897394426209505,
      "loss": 11.5,
      "step": 78
    },
    {
      "epoch": 0.04315318698576354,
      "grad_norm": 0.0006288820877671242,
      "learning_rate": 0.00019894359789047187,
      "loss": 11.5,
      "step": 79
    },
    {
      "epoch": 0.043699429859001056,
      "grad_norm": 0.0004901019274257123,
      "learning_rate": 0.00019891281165856873,
      "loss": 11.5,
      "step": 80
    },
    {
      "epoch": 0.04424567273223857,
      "grad_norm": 0.0007199411047622561,
      "learning_rate": 0.00019888158570324795,
      "loss": 11.5,
      "step": 81
    },
    {
      "epoch": 0.04479191560547609,
      "grad_norm": 0.0004847659438382834,
      "learning_rate": 0.0001988499201633265,
      "loss": 11.5,
      "step": 82
    },
    {
      "epoch": 0.0453381584787136,
      "grad_norm": 0.0005560573772527277,
      "learning_rate": 0.00019881781517957562,
      "loss": 11.5,
      "step": 83
    },
    {
      "epoch": 0.04588440135195111,
      "grad_norm": 0.0005896556540392339,
      "learning_rate": 0.0001987852708947202,
      "loss": 11.5,
      "step": 84
    },
    {
      "epoch": 0.046430644225188625,
      "grad_norm": 0.000917259429115802,
      "learning_rate": 0.00019875228745343794,
      "loss": 11.5,
      "step": 85
    },
    {
      "epoch": 0.04697688709842614,
      "grad_norm": 0.0006931269890628755,
      "learning_rate": 0.0001987188650023589,
      "loss": 11.5,
      "step": 86
    },
    {
      "epoch": 0.04752312997166365,
      "grad_norm": 0.0009154531289823353,
      "learning_rate": 0.0001986850036900648,
      "loss": 11.5,
      "step": 87
    },
    {
      "epoch": 0.04806937284490116,
      "grad_norm": 0.0008669509552419186,
      "learning_rate": 0.00019865070366708836,
      "loss": 11.5,
      "step": 88
    },
    {
      "epoch": 0.048615615718138674,
      "grad_norm": 0.0009474908001720905,
      "learning_rate": 0.00019861596508591255,
      "loss": 11.5,
      "step": 89
    },
    {
      "epoch": 0.049161858591376194,
      "grad_norm": 0.0008921613334678113,
      "learning_rate": 0.00019858078810097002,
      "loss": 11.5,
      "step": 90
    },
    {
      "epoch": 0.049708101464613706,
      "grad_norm": 0.0008722237544134259,
      "learning_rate": 0.00019854517286864245,
      "loss": 11.5,
      "step": 91
    },
    {
      "epoch": 0.05025434433785122,
      "grad_norm": 0.0008749884436838329,
      "learning_rate": 0.0001985091195472596,
      "loss": 11.5,
      "step": 92
    },
    {
      "epoch": 0.05080058721108873,
      "grad_norm": 0.0009783711284399033,
      "learning_rate": 0.0001984726282970989,
      "loss": 11.5,
      "step": 93
    },
    {
      "epoch": 0.05134683008432624,
      "grad_norm": 0.0008170564542524517,
      "learning_rate": 0.0001984356992803847,
      "loss": 11.5,
      "step": 94
    },
    {
      "epoch": 0.051893072957563756,
      "grad_norm": 0.0011803907109424472,
      "learning_rate": 0.00019839833266128724,
      "loss": 11.5,
      "step": 95
    },
    {
      "epoch": 0.05243931583080127,
      "grad_norm": 0.0006016236147843301,
      "learning_rate": 0.00019836052860592237,
      "loss": 11.5,
      "step": 96
    },
    {
      "epoch": 0.05298555870403878,
      "grad_norm": 0.0010825739009305835,
      "learning_rate": 0.0001983222872823505,
      "loss": 11.5,
      "step": 97
    },
    {
      "epoch": 0.05353180157727629,
      "grad_norm": 0.0013415748253464699,
      "learning_rate": 0.00019828360886057594,
      "loss": 11.5,
      "step": 98
    },
    {
      "epoch": 0.05407804445051381,
      "grad_norm": 0.0011260845931246877,
      "learning_rate": 0.00019824449351254616,
      "loss": 11.5,
      "step": 99
    },
    {
      "epoch": 0.054624287323751325,
      "grad_norm": 0.0016935975290834904,
      "learning_rate": 0.00019820494141215104,
      "loss": 11.5,
      "step": 100
    },
    {
      "epoch": 0.05517053019698884,
      "grad_norm": 0.0010154311312362552,
      "learning_rate": 0.000198164952735222,
      "loss": 11.5,
      "step": 101
    },
    {
      "epoch": 0.05571677307022635,
      "grad_norm": 0.0011301799677312374,
      "learning_rate": 0.00019812452765953135,
      "loss": 11.5,
      "step": 102
    },
    {
      "epoch": 0.05626301594346386,
      "grad_norm": 0.0010727025801315904,
      "learning_rate": 0.00019808366636479147,
      "loss": 11.5,
      "step": 103
    },
    {
      "epoch": 0.056809258816701375,
      "grad_norm": 0.0008837860077619553,
      "learning_rate": 0.00019804236903265388,
      "loss": 11.5,
      "step": 104
    },
    {
      "epoch": 0.05735550168993889,
      "grad_norm": 0.0008506904123350978,
      "learning_rate": 0.00019800063584670863,
      "loss": 11.5,
      "step": 105
    },
    {
      "epoch": 0.0579017445631764,
      "grad_norm": 0.00114605447743088,
      "learning_rate": 0.00019795846699248332,
      "loss": 11.5,
      "step": 106
    },
    {
      "epoch": 0.05844798743641392,
      "grad_norm": 0.0010288127232342958,
      "learning_rate": 0.00019791586265744237,
      "loss": 11.5,
      "step": 107
    },
    {
      "epoch": 0.05899423030965143,
      "grad_norm": 0.002101024379953742,
      "learning_rate": 0.00019787282303098617,
      "loss": 11.5,
      "step": 108
    },
    {
      "epoch": 0.059540473182888944,
      "grad_norm": 0.0011782986111938953,
      "learning_rate": 0.0001978293483044502,
      "loss": 11.5,
      "step": 109
    },
    {
      "epoch": 0.060086716056126456,
      "grad_norm": 0.0011132174404338002,
      "learning_rate": 0.00019778543867110426,
      "loss": 11.5,
      "step": 110
    },
    {
      "epoch": 0.06063295892936397,
      "grad_norm": 0.0011150569189339876,
      "learning_rate": 0.00019774109432615147,
      "loss": 11.5,
      "step": 111
    },
    {
      "epoch": 0.06117920180260148,
      "grad_norm": 0.001215057447552681,
      "learning_rate": 0.00019769631546672756,
      "loss": 11.5,
      "step": 112
    },
    {
      "epoch": 0.06172544467583899,
      "grad_norm": 0.0011679328745231032,
      "learning_rate": 0.00019765110229189988,
      "loss": 11.5,
      "step": 113
    },
    {
      "epoch": 0.062271687549076506,
      "grad_norm": 0.0012629639822989702,
      "learning_rate": 0.00019760545500266657,
      "loss": 11.5,
      "step": 114
    },
    {
      "epoch": 0.06281793042231403,
      "grad_norm": 0.00198923796415329,
      "learning_rate": 0.00019755937380195568,
      "loss": 11.5,
      "step": 115
    },
    {
      "epoch": 0.06336417329555154,
      "grad_norm": 0.001028231461532414,
      "learning_rate": 0.00019751285889462423,
      "loss": 11.5,
      "step": 116
    },
    {
      "epoch": 0.06391041616878905,
      "grad_norm": 0.0012992372503504157,
      "learning_rate": 0.0001974659104874573,
      "loss": 11.5,
      "step": 117
    },
    {
      "epoch": 0.06445665904202656,
      "grad_norm": 0.0017271611141040921,
      "learning_rate": 0.0001974185287891671,
      "loss": 11.5,
      "step": 118
    },
    {
      "epoch": 0.06500290191526407,
      "grad_norm": 0.0015606487868353724,
      "learning_rate": 0.0001973707140103921,
      "loss": 11.5,
      "step": 119
    },
    {
      "epoch": 0.06554914478850159,
      "grad_norm": 0.004301104228943586,
      "learning_rate": 0.00019732246636369605,
      "loss": 11.5,
      "step": 120
    },
    {
      "epoch": 0.0660953876617391,
      "grad_norm": 0.001471640425734222,
      "learning_rate": 0.00019727378606356703,
      "loss": 11.5,
      "step": 121
    },
    {
      "epoch": 0.06664163053497661,
      "grad_norm": 0.002853769576177001,
      "learning_rate": 0.00019722467332641656,
      "loss": 11.5,
      "step": 122
    },
    {
      "epoch": 0.06718787340821412,
      "grad_norm": 0.0014008270809426904,
      "learning_rate": 0.00019717512837057855,
      "loss": 11.5,
      "step": 123
    },
    {
      "epoch": 0.06773411628145164,
      "grad_norm": 0.0018577142618596554,
      "learning_rate": 0.0001971251514163083,
      "loss": 11.5,
      "step": 124
    },
    {
      "epoch": 0.06828035915468915,
      "grad_norm": 0.002127281157299876,
      "learning_rate": 0.0001970747426857817,
      "loss": 11.5,
      "step": 125
    },
    {
      "epoch": 0.06882660202792666,
      "grad_norm": 0.0019906344823539257,
      "learning_rate": 0.00019702390240309404,
      "loss": 11.5,
      "step": 126
    },
    {
      "epoch": 0.06937284490116417,
      "grad_norm": 0.002931496361270547,
      "learning_rate": 0.0001969726307942592,
      "loss": 11.5,
      "step": 127
    },
    {
      "epoch": 0.06991908777440169,
      "grad_norm": 0.0023249045480042696,
      "learning_rate": 0.00019692092808720846,
      "loss": 11.5,
      "step": 128
    },
    {
      "epoch": 0.07046533064763921,
      "grad_norm": 0.001595000154338777,
      "learning_rate": 0.0001968687945117896,
      "loss": 11.5,
      "step": 129
    },
    {
      "epoch": 0.07101157352087673,
      "grad_norm": 0.0013937182957306504,
      "learning_rate": 0.00019681623029976588,
      "loss": 11.5,
      "step": 130
    },
    {
      "epoch": 0.07155781639411424,
      "grad_norm": 0.0013084670063108206,
      "learning_rate": 0.00019676323568481498,
      "loss": 11.5,
      "step": 131
    },
    {
      "epoch": 0.07210405926735175,
      "grad_norm": 0.001346346689388156,
      "learning_rate": 0.00019670981090252792,
      "loss": 11.5,
      "step": 132
    },
    {
      "epoch": 0.07265030214058926,
      "grad_norm": 0.0018107227515429258,
      "learning_rate": 0.00019665595619040808,
      "loss": 11.5,
      "step": 133
    },
    {
      "epoch": 0.07319654501382677,
      "grad_norm": 0.001764649641700089,
      "learning_rate": 0.0001966016717878702,
      "loss": 11.5,
      "step": 134
    },
    {
      "epoch": 0.07374278788706429,
      "grad_norm": 0.0024073971435427666,
      "learning_rate": 0.00019654695793623907,
      "loss": 11.5,
      "step": 135
    },
    {
      "epoch": 0.0742890307603018,
      "grad_norm": 0.0017252506222575903,
      "learning_rate": 0.0001964918148787488,
      "loss": 11.5,
      "step": 136
    },
    {
      "epoch": 0.07483527363353931,
      "grad_norm": 0.0016268540639430285,
      "learning_rate": 0.00019643624286054144,
      "loss": 11.5,
      "step": 137
    },
    {
      "epoch": 0.07538151650677682,
      "grad_norm": 0.002305061323568225,
      "learning_rate": 0.00019638024212866606,
      "loss": 11.5,
      "step": 138
    },
    {
      "epoch": 0.07592775938001434,
      "grad_norm": 0.0018998737214133143,
      "learning_rate": 0.0001963238129320776,
      "loss": 11.5,
      "step": 139
    },
    {
      "epoch": 0.07647400225325185,
      "grad_norm": 0.00191564264241606,
      "learning_rate": 0.00019626695552163578,
      "loss": 11.5,
      "step": 140
    },
    {
      "epoch": 0.07702024512648936,
      "grad_norm": 0.0026092506013810635,
      "learning_rate": 0.00019620967015010395,
      "loss": 11.5,
      "step": 141
    },
    {
      "epoch": 0.07756648799972687,
      "grad_norm": 0.0031949372496455908,
      "learning_rate": 0.00019615195707214803,
      "loss": 11.5,
      "step": 142
    },
    {
      "epoch": 0.07811273087296439,
      "grad_norm": 0.001672394573688507,
      "learning_rate": 0.0001960938165443353,
      "loss": 11.5,
      "step": 143
    },
    {
      "epoch": 0.0786589737462019,
      "grad_norm": 0.004255365114659071,
      "learning_rate": 0.00019603524882513327,
      "loss": 11.5,
      "step": 144
    },
    {
      "epoch": 0.07920521661943941,
      "grad_norm": 0.0025640197563916445,
      "learning_rate": 0.0001959762541749086,
      "loss": 11.5,
      "step": 145
    },
    {
      "epoch": 0.07975145949267694,
      "grad_norm": 0.0024671240244060755,
      "learning_rate": 0.00019591683285592593,
      "loss": 11.5,
      "step": 146
    },
    {
      "epoch": 0.08029770236591445,
      "grad_norm": 0.0031216409988701344,
      "learning_rate": 0.00019585698513234663,
      "loss": 11.5,
      "step": 147
    },
    {
      "epoch": 0.08084394523915196,
      "grad_norm": 0.001815325696952641,
      "learning_rate": 0.0001957967112702277,
      "loss": 11.5,
      "step": 148
    },
    {
      "epoch": 0.08139018811238947,
      "grad_norm": 0.003914890345185995,
      "learning_rate": 0.00019573601153752052,
      "loss": 11.5,
      "step": 149
    },
    {
      "epoch": 0.08193643098562699,
      "grad_norm": 0.002451361855491996,
      "learning_rate": 0.00019567488620406983,
      "loss": 11.5,
      "step": 150
    },
    {
      "epoch": 0.0824826738588645,
      "grad_norm": 0.004037910606712103,
      "learning_rate": 0.00019561333554161224,
      "loss": 11.5,
      "step": 151
    },
    {
      "epoch": 0.08302891673210201,
      "grad_norm": 0.0032814564183354378,
      "learning_rate": 0.0001955513598237753,
      "loss": 11.5,
      "step": 152
    },
    {
      "epoch": 0.08357515960533952,
      "grad_norm": 0.0032473402097821236,
      "learning_rate": 0.00019548895932607621,
      "loss": 11.5,
      "step": 153
    },
    {
      "epoch": 0.08412140247857704,
      "grad_norm": 0.0035192391369491816,
      "learning_rate": 0.00019542613432592038,
      "loss": 11.5,
      "step": 154
    },
    {
      "epoch": 0.08466764535181455,
      "grad_norm": 0.0035076031927019358,
      "learning_rate": 0.00019536288510260056,
      "loss": 11.5,
      "step": 155
    },
    {
      "epoch": 0.08521388822505206,
      "grad_norm": 0.003550986759364605,
      "learning_rate": 0.00019529921193729534,
      "loss": 11.5,
      "step": 156
    },
    {
      "epoch": 0.08576013109828957,
      "grad_norm": 0.003494930686429143,
      "learning_rate": 0.00019523511511306793,
      "loss": 11.5,
      "step": 157
    },
    {
      "epoch": 0.08630637397152709,
      "grad_norm": 0.004425590392202139,
      "learning_rate": 0.000195170594914865,
      "loss": 11.5,
      "step": 158
    },
    {
      "epoch": 0.0868526168447646,
      "grad_norm": 0.0033555696718394756,
      "learning_rate": 0.00019510565162951537,
      "loss": 11.5,
      "step": 159
    },
    {
      "epoch": 0.08739885971800211,
      "grad_norm": 0.004704903811216354,
      "learning_rate": 0.00019504028554572864,
      "loss": 11.5,
      "step": 160
    },
    {
      "epoch": 0.08794510259123962,
      "grad_norm": 0.004017127677798271,
      "learning_rate": 0.00019497449695409408,
      "loss": 11.5,
      "step": 161
    },
    {
      "epoch": 0.08849134546447714,
      "grad_norm": 0.0034707789309322834,
      "learning_rate": 0.00019490828614707916,
      "loss": 11.5,
      "step": 162
    },
    {
      "epoch": 0.08903758833771466,
      "grad_norm": 0.0036698062904179096,
      "learning_rate": 0.00019484165341902845,
      "loss": 11.5,
      "step": 163
    },
    {
      "epoch": 0.08958383121095218,
      "grad_norm": 0.0036391145549714565,
      "learning_rate": 0.00019477459906616206,
      "loss": 11.5,
      "step": 164
    },
    {
      "epoch": 0.09013007408418969,
      "grad_norm": 0.0055595445446670055,
      "learning_rate": 0.00019470712338657458,
      "loss": 11.5,
      "step": 165
    },
    {
      "epoch": 0.0906763169574272,
      "grad_norm": 0.0034312040079385042,
      "learning_rate": 0.0001946392266802336,
      "loss": 11.5,
      "step": 166
    },
    {
      "epoch": 0.09122255983066471,
      "grad_norm": 0.004627106245607138,
      "learning_rate": 0.0001945709092489783,
      "loss": 11.5,
      "step": 167
    },
    {
      "epoch": 0.09176880270390222,
      "grad_norm": 0.005648384336382151,
      "learning_rate": 0.00019450217139651844,
      "loss": 11.5,
      "step": 168
    },
    {
      "epoch": 0.09231504557713974,
      "grad_norm": 0.00625839177519083,
      "learning_rate": 0.0001944330134284326,
      "loss": 11.5,
      "step": 169
    },
    {
      "epoch": 0.09286128845037725,
      "grad_norm": 0.004797192756086588,
      "learning_rate": 0.00019436343565216711,
      "loss": 11.5,
      "step": 170
    },
    {
      "epoch": 0.09340753132361476,
      "grad_norm": 0.005659387446939945,
      "learning_rate": 0.00019429343837703455,
      "loss": 11.5,
      "step": 171
    },
    {
      "epoch": 0.09395377419685227,
      "grad_norm": 0.005367459263652563,
      "learning_rate": 0.0001942230219142124,
      "loss": 11.5,
      "step": 172
    },
    {
      "epoch": 0.09450001707008979,
      "grad_norm": 0.005297891329973936,
      "learning_rate": 0.0001941521865767417,
      "loss": 11.5,
      "step": 173
    },
    {
      "epoch": 0.0950462599433273,
      "grad_norm": 0.005610900465399027,
      "learning_rate": 0.0001940809326795256,
      "loss": 11.5,
      "step": 174
    },
    {
      "epoch": 0.09559250281656481,
      "grad_norm": 0.00406970689073205,
      "learning_rate": 0.000194009260539328,
      "loss": 11.5,
      "step": 175
    },
    {
      "epoch": 0.09613874568980232,
      "grad_norm": 0.00735284062102437,
      "learning_rate": 0.0001939371704747721,
      "loss": 11.5,
      "step": 176
    },
    {
      "epoch": 0.09668498856303984,
      "grad_norm": 0.008486613631248474,
      "learning_rate": 0.00019386466280633906,
      "loss": 11.5,
      "step": 177
    },
    {
      "epoch": 0.09723123143627735,
      "grad_norm": 0.00998898409307003,
      "learning_rate": 0.00019379173785636646,
      "loss": 11.5,
      "step": 178
    },
    {
      "epoch": 0.09777747430951486,
      "grad_norm": 0.0067891585640609264,
      "learning_rate": 0.000193718395949047,
      "loss": 11.5,
      "step": 179
    },
    {
      "epoch": 0.09832371718275239,
      "grad_norm": 0.0062019010074436665,
      "learning_rate": 0.00019364463741042694,
      "loss": 11.5,
      "step": 180
    },
    {
      "epoch": 0.0988699600559899,
      "grad_norm": 0.005536094307899475,
      "learning_rate": 0.00019357046256840473,
      "loss": 11.5,
      "step": 181
    },
    {
      "epoch": 0.09941620292922741,
      "grad_norm": 0.004772233311086893,
      "learning_rate": 0.00019349587175272948,
      "loss": 11.5,
      "step": 182
    },
    {
      "epoch": 0.09996244580246492,
      "grad_norm": 0.006683160550892353,
      "learning_rate": 0.0001934208652949996,
      "loss": 11.5,
      "step": 183
    },
    {
      "epoch": 0.10050868867570244,
      "grad_norm": 0.005879896227270365,
      "learning_rate": 0.00019334544352866127,
      "loss": 11.5,
      "step": 184
    },
    {
      "epoch": 0.10105493154893995,
      "grad_norm": 0.006935118697583675,
      "learning_rate": 0.00019326960678900688,
      "loss": 11.5,
      "step": 185
    },
    {
      "epoch": 0.10160117442217746,
      "grad_norm": 0.007274080999195576,
      "learning_rate": 0.00019319335541317361,
      "loss": 11.5,
      "step": 186
    },
    {
      "epoch": 0.10214741729541497,
      "grad_norm": 0.003108037170022726,
      "learning_rate": 0.00019311668974014208,
      "loss": 11.5,
      "step": 187
    },
    {
      "epoch": 0.10269366016865249,
      "grad_norm": 0.007204362656921148,
      "learning_rate": 0.00019303961011073447,
      "loss": 11.5,
      "step": 188
    },
    {
      "epoch": 0.10323990304189,
      "grad_norm": 0.0062853507697582245,
      "learning_rate": 0.00019296211686761346,
      "loss": 11.5,
      "step": 189
    },
    {
      "epoch": 0.10378614591512751,
      "grad_norm": 0.0054272557608783245,
      "learning_rate": 0.00019288421035528028,
      "loss": 11.5,
      "step": 190
    },
    {
      "epoch": 0.10433238878836502,
      "grad_norm": 0.004476429894566536,
      "learning_rate": 0.00019280589092007352,
      "loss": 11.5,
      "step": 191
    },
    {
      "epoch": 0.10487863166160254,
      "grad_norm": 0.00726770143955946,
      "learning_rate": 0.00019272715891016735,
      "loss": 11.5,
      "step": 192
    },
    {
      "epoch": 0.10542487453484005,
      "grad_norm": 0.004088373389095068,
      "learning_rate": 0.00019264801467557007,
      "loss": 11.5,
      "step": 193
    },
    {
      "epoch": 0.10597111740807756,
      "grad_norm": 0.00533740408718586,
      "learning_rate": 0.00019256845856812266,
      "loss": 11.5,
      "step": 194
    },
    {
      "epoch": 0.10651736028131507,
      "grad_norm": 0.004800862167030573,
      "learning_rate": 0.000192488490941497,
      "loss": 11.5,
      "step": 195
    },
    {
      "epoch": 0.10706360315455259,
      "grad_norm": 0.004992916714400053,
      "learning_rate": 0.00019240811215119448,
      "loss": 11.5,
      "step": 196
    },
    {
      "epoch": 0.10760984602779011,
      "grad_norm": 0.0068426732905209064,
      "learning_rate": 0.00019232732255454422,
      "loss": 11.5,
      "step": 197
    },
    {
      "epoch": 0.10815608890102762,
      "grad_norm": 0.0042264219373464584,
      "learning_rate": 0.00019224612251070175,
      "loss": 11.5,
      "step": 198
    },
    {
      "epoch": 0.10870233177426514,
      "grad_norm": 0.004886781331151724,
      "learning_rate": 0.0001921645123806472,
      "loss": 11.5,
      "step": 199
    },
    {
      "epoch": 0.10924857464750265,
      "grad_norm": 0.007674403954297304,
      "learning_rate": 0.0001920824925271838,
      "loss": 11.5,
      "step": 200
    },
    {
      "epoch": 0.10979481752074016,
      "grad_norm": 0.006116423290222883,
      "learning_rate": 0.0001920000633149362,
      "loss": 11.5,
      "step": 201
    },
    {
      "epoch": 0.11034106039397767,
      "grad_norm": 0.005647597834467888,
      "learning_rate": 0.00019191722511034884,
      "loss": 11.5,
      "step": 202
    },
    {
      "epoch": 0.11088730326721519,
      "grad_norm": 0.006665877532213926,
      "learning_rate": 0.00019183397828168448,
      "loss": 11.5,
      "step": 203
    },
    {
      "epoch": 0.1114335461404527,
      "grad_norm": 0.004790004342794418,
      "learning_rate": 0.00019175032319902234,
      "loss": 11.5,
      "step": 204
    },
    {
      "epoch": 0.11197978901369021,
      "grad_norm": 0.0058875554241240025,
      "learning_rate": 0.00019166626023425662,
      "loss": 11.5,
      "step": 205
    },
    {
      "epoch": 0.11252603188692772,
      "grad_norm": 0.00861747283488512,
      "learning_rate": 0.00019158178976109476,
      "loss": 11.5,
      "step": 206
    },
    {
      "epoch": 0.11307227476016524,
      "grad_norm": 0.006387247238308191,
      "learning_rate": 0.0001914969121550558,
      "loss": 11.5,
      "step": 207
    },
    {
      "epoch": 0.11361851763340275,
      "grad_norm": 0.006305269431322813,
      "learning_rate": 0.00019141162779346874,
      "loss": 11.5,
      "step": 208
    },
    {
      "epoch": 0.11416476050664026,
      "grad_norm": 0.00655133742839098,
      "learning_rate": 0.00019132593705547082,
      "loss": 11.5,
      "step": 209
    },
    {
      "epoch": 0.11471100337987777,
      "grad_norm": 0.005687771365046501,
      "learning_rate": 0.00019123984032200586,
      "loss": 11.5,
      "step": 210
    },
    {
      "epoch": 0.11525724625311529,
      "grad_norm": 0.006879627238959074,
      "learning_rate": 0.00019115333797582254,
      "loss": 11.5,
      "step": 211
    },
    {
      "epoch": 0.1158034891263528,
      "grad_norm": 0.006566868629306555,
      "learning_rate": 0.00019106643040147278,
      "loss": 11.5,
      "step": 212
    },
    {
      "epoch": 0.11634973199959031,
      "grad_norm": 0.004991826601326466,
      "learning_rate": 0.00019097911798530987,
      "loss": 11.5,
      "step": 213
    },
    {
      "epoch": 0.11689597487282784,
      "grad_norm": 0.005583076737821102,
      "learning_rate": 0.00019089140111548696,
      "loss": 11.5,
      "step": 214
    },
    {
      "epoch": 0.11744221774606535,
      "grad_norm": 0.004320789128541946,
      "learning_rate": 0.00019080328018195513,
      "loss": 11.5,
      "step": 215
    },
    {
      "epoch": 0.11798846061930286,
      "grad_norm": 0.005942629184573889,
      "learning_rate": 0.0001907147555764618,
      "loss": 11.5,
      "step": 216
    },
    {
      "epoch": 0.11853470349254037,
      "grad_norm": 0.004765739664435387,
      "learning_rate": 0.00019062582769254895,
      "loss": 11.5,
      "step": 217
    },
    {
      "epoch": 0.11908094636577789,
      "grad_norm": 0.0062084924429655075,
      "learning_rate": 0.00019053649692555135,
      "loss": 11.5,
      "step": 218
    },
    {
      "epoch": 0.1196271892390154,
      "grad_norm": 0.0039649344980716705,
      "learning_rate": 0.00019044676367259476,
      "loss": 11.5,
      "step": 219
    },
    {
      "epoch": 0.12017343211225291,
      "grad_norm": 0.003176108468323946,
      "learning_rate": 0.00019035662833259432,
      "loss": 11.5,
      "step": 220
    },
    {
      "epoch": 0.12071967498549042,
      "grad_norm": 0.0056105321273207664,
      "learning_rate": 0.00019026609130625257,
      "loss": 11.5,
      "step": 221
    },
    {
      "epoch": 0.12126591785872794,
      "grad_norm": 0.005996837746351957,
      "learning_rate": 0.00019017515299605788,
      "loss": 11.5,
      "step": 222
    },
    {
      "epoch": 0.12181216073196545,
      "grad_norm": 0.00406813295558095,
      "learning_rate": 0.00019008381380628247,
      "loss": 11.5,
      "step": 223
    },
    {
      "epoch": 0.12235840360520296,
      "grad_norm": 0.005131961777806282,
      "learning_rate": 0.00018999207414298067,
      "loss": 11.5,
      "step": 224
    },
    {
      "epoch": 0.12290464647844047,
      "grad_norm": 0.004345993977040052,
      "learning_rate": 0.00018989993441398726,
      "loss": 11.5,
      "step": 225
    },
    {
      "epoch": 0.12345088935167799,
      "grad_norm": 0.008761870674788952,
      "learning_rate": 0.00018980739502891546,
      "loss": 11.5,
      "step": 226
    },
    {
      "epoch": 0.1239971322249155,
      "grad_norm": 0.009130410850048065,
      "learning_rate": 0.0001897144563991552,
      "loss": 11.5,
      "step": 227
    },
    {
      "epoch": 0.12454337509815301,
      "grad_norm": 0.007637005299329758,
      "learning_rate": 0.00018962111893787128,
      "loss": 11.5,
      "step": 228
    },
    {
      "epoch": 0.12508961797139054,
      "grad_norm": 0.0027051966171711683,
      "learning_rate": 0.00018952738306000151,
      "loss": 11.5,
      "step": 229
    },
    {
      "epoch": 0.12563586084462805,
      "grad_norm": 0.002841376233845949,
      "learning_rate": 0.00018943324918225494,
      "loss": 11.5,
      "step": 230
    },
    {
      "epoch": 0.12618210371786556,
      "grad_norm": 0.0025662824045866728,
      "learning_rate": 0.0001893387177231099,
      "loss": 11.5,
      "step": 231
    },
    {
      "epoch": 0.12672834659110307,
      "grad_norm": 0.0028462784830480814,
      "learning_rate": 0.0001892437891028122,
      "loss": 11.5,
      "step": 232
    },
    {
      "epoch": 0.1272745894643406,
      "grad_norm": 0.002275710226967931,
      "learning_rate": 0.0001891484637433733,
      "loss": 11.5,
      "step": 233
    },
    {
      "epoch": 0.1278208323375781,
      "grad_norm": 0.0018196929013356566,
      "learning_rate": 0.00018905274206856837,
      "loss": 11.5,
      "step": 234
    },
    {
      "epoch": 0.1283670752108156,
      "grad_norm": 0.0028202449902892113,
      "learning_rate": 0.00018895662450393438,
      "loss": 11.5,
      "step": 235
    },
    {
      "epoch": 0.12891331808405312,
      "grad_norm": 0.002004245761781931,
      "learning_rate": 0.00018886011147676833,
      "loss": 11.5,
      "step": 236
    },
    {
      "epoch": 0.12945956095729064,
      "grad_norm": 0.0031620205845683813,
      "learning_rate": 0.00018876320341612522,
      "loss": 11.5,
      "step": 237
    },
    {
      "epoch": 0.13000580383052815,
      "grad_norm": 0.0033896085806190968,
      "learning_rate": 0.00018866590075281624,
      "loss": 11.5,
      "step": 238
    },
    {
      "epoch": 0.13055204670376566,
      "grad_norm": 0.004640740342438221,
      "learning_rate": 0.00018856820391940674,
      "loss": 11.5,
      "step": 239
    },
    {
      "epoch": 0.13109828957700317,
      "grad_norm": 0.0019844146445393562,
      "learning_rate": 0.00018847011335021449,
      "loss": 11.5,
      "step": 240
    },
    {
      "epoch": 0.1316445324502407,
      "grad_norm": 0.0028746069874614477,
      "learning_rate": 0.00018837162948130752,
      "loss": 11.5,
      "step": 241
    },
    {
      "epoch": 0.1321907753234782,
      "grad_norm": 0.0025174636393785477,
      "learning_rate": 0.00018827275275050233,
      "loss": 11.5,
      "step": 242
    },
    {
      "epoch": 0.1327370181967157,
      "grad_norm": 0.0033687143586575985,
      "learning_rate": 0.00018817348359736203,
      "loss": 11.5,
      "step": 243
    },
    {
      "epoch": 0.13328326106995322,
      "grad_norm": 0.002695529954507947,
      "learning_rate": 0.00018807382246319412,
      "loss": 11.5,
      "step": 244
    },
    {
      "epoch": 0.13382950394319074,
      "grad_norm": 0.005114860832691193,
      "learning_rate": 0.00018797376979104872,
      "loss": 11.5,
      "step": 245
    },
    {
      "epoch": 0.13437574681642825,
      "grad_norm": 0.005076493136584759,
      "learning_rate": 0.00018787332602571662,
      "loss": 11.5,
      "step": 246
    },
    {
      "epoch": 0.13492198968966576,
      "grad_norm": 0.004942054394632578,
      "learning_rate": 0.00018777249161372713,
      "loss": 11.5,
      "step": 247
    },
    {
      "epoch": 0.13546823256290327,
      "grad_norm": 0.0032974916975945234,
      "learning_rate": 0.00018767126700334634,
      "loss": 11.5,
      "step": 248
    },
    {
      "epoch": 0.13601447543614079,
      "grad_norm": 0.007932610809803009,
      "learning_rate": 0.0001875696526445749,
      "loss": 11.5,
      "step": 249
    },
    {
      "epoch": 0.1365607183093783,
      "grad_norm": 0.005659975577145815,
      "learning_rate": 0.0001874676489891461,
      "loss": 11.5,
      "step": 250
    },
    {
      "epoch": 0.1371069611826158,
      "grad_norm": 0.004244368523359299,
      "learning_rate": 0.00018736525649052394,
      "loss": 11.5,
      "step": 251
    },
    {
      "epoch": 0.13765320405585332,
      "grad_norm": 0.003855164395645261,
      "learning_rate": 0.00018726247560390099,
      "loss": 11.5,
      "step": 252
    },
    {
      "epoch": 0.13819944692909084,
      "grad_norm": 0.00465209037065506,
      "learning_rate": 0.00018715930678619644,
      "loss": 11.5,
      "step": 253
    },
    {
      "epoch": 0.13874568980232835,
      "grad_norm": 0.004538490902632475,
      "learning_rate": 0.00018705575049605413,
      "loss": 11.5,
      "step": 254
    },
    {
      "epoch": 0.13929193267556586,
      "grad_norm": 0.0031845690682530403,
      "learning_rate": 0.00018695180719384029,
      "loss": 11.5,
      "step": 255
    },
    {
      "epoch": 0.13983817554880337,
      "grad_norm": 0.004388829693198204,
      "learning_rate": 0.00018684747734164177,
      "loss": 11.5,
      "step": 256
    },
    {
      "epoch": 0.1403844184220409,
      "grad_norm": 0.004310618620365858,
      "learning_rate": 0.00018674276140326376,
      "loss": 11.5,
      "step": 257
    },
    {
      "epoch": 0.14093066129527843,
      "grad_norm": 0.004943131003528833,
      "learning_rate": 0.00018663765984422786,
      "loss": 11.5,
      "step": 258
    },
    {
      "epoch": 0.14147690416851594,
      "grad_norm": 0.0034593376331031322,
      "learning_rate": 0.00018653217313177004,
      "loss": 11.5,
      "step": 259
    },
    {
      "epoch": 0.14202314704175345,
      "grad_norm": 0.003939722664654255,
      "learning_rate": 0.00018642630173483832,
      "loss": 11.5,
      "step": 260
    },
    {
      "epoch": 0.14256938991499096,
      "grad_norm": 0.0035885698162019253,
      "learning_rate": 0.00018632004612409103,
      "loss": 11.5,
      "step": 261
    },
    {
      "epoch": 0.14311563278822848,
      "grad_norm": 0.004717244766652584,
      "learning_rate": 0.00018621340677189453,
      "loss": 11.5,
      "step": 262
    },
    {
      "epoch": 0.143661875661466,
      "grad_norm": 0.004850431811064482,
      "learning_rate": 0.00018610638415232097,
      "loss": 11.5,
      "step": 263
    },
    {
      "epoch": 0.1442081185347035,
      "grad_norm": 0.0043102228082716465,
      "learning_rate": 0.00018599897874114652,
      "loss": 11.5,
      "step": 264
    },
    {
      "epoch": 0.144754361407941,
      "grad_norm": 0.004483947530388832,
      "learning_rate": 0.00018589119101584898,
      "loss": 11.5,
      "step": 265
    },
    {
      "epoch": 0.14530060428117852,
      "grad_norm": 0.00466382410377264,
      "learning_rate": 0.00018578302145560584,
      "loss": 11.5,
      "step": 266
    },
    {
      "epoch": 0.14584684715441604,
      "grad_norm": 0.005623066332191229,
      "learning_rate": 0.00018567447054129195,
      "loss": 11.5,
      "step": 267
    },
    {
      "epoch": 0.14639309002765355,
      "grad_norm": 0.0039781732484698296,
      "learning_rate": 0.00018556553875547754,
      "loss": 11.5,
      "step": 268
    },
    {
      "epoch": 0.14693933290089106,
      "grad_norm": 0.008801842108368874,
      "learning_rate": 0.00018545622658242607,
      "loss": 11.5,
      "step": 269
    },
    {
      "epoch": 0.14748557577412857,
      "grad_norm": 0.0027506386395543814,
      "learning_rate": 0.00018534653450809197,
      "loss": 11.5,
      "step": 270
    },
    {
      "epoch": 0.1480318186473661,
      "grad_norm": 0.0014627018244937062,
      "learning_rate": 0.00018523646302011867,
      "loss": 11.5,
      "step": 271
    },
    {
      "epoch": 0.1485780615206036,
      "grad_norm": 0.003000725293532014,
      "learning_rate": 0.00018512601260783606,
      "loss": 11.5,
      "step": 272
    },
    {
      "epoch": 0.1491243043938411,
      "grad_norm": 0.0037209002766758204,
      "learning_rate": 0.00018501518376225887,
      "loss": 11.5,
      "step": 273
    },
    {
      "epoch": 0.14967054726707862,
      "grad_norm": 0.00450047105550766,
      "learning_rate": 0.00018490397697608395,
      "loss": 11.5,
      "step": 274
    },
    {
      "epoch": 0.15021679014031614,
      "grad_norm": 0.004258057568222284,
      "learning_rate": 0.0001847923927436884,
      "loss": 11.5,
      "step": 275
    },
    {
      "epoch": 0.15076303301355365,
      "grad_norm": 0.005219324957579374,
      "learning_rate": 0.00018468043156112728,
      "loss": 11.5,
      "step": 276
    },
    {
      "epoch": 0.15130927588679116,
      "grad_norm": 0.007191560231149197,
      "learning_rate": 0.0001845680939261314,
      "loss": 11.5,
      "step": 277
    },
    {
      "epoch": 0.15185551876002867,
      "grad_norm": 0.013161702081561089,
      "learning_rate": 0.00018445538033810515,
      "loss": 11.5,
      "step": 278
    },
    {
      "epoch": 0.1524017616332662,
      "grad_norm": 0.0031373328529298306,
      "learning_rate": 0.00018434229129812418,
      "loss": 11.5,
      "step": 279
    },
    {
      "epoch": 0.1529480045065037,
      "grad_norm": 0.004784329328685999,
      "learning_rate": 0.0001842288273089332,
      "loss": 11.5,
      "step": 280
    },
    {
      "epoch": 0.1534942473797412,
      "grad_norm": 0.0026535531505942345,
      "learning_rate": 0.00018411498887494396,
      "loss": 11.5,
      "step": 281
    },
    {
      "epoch": 0.15404049025297872,
      "grad_norm": 0.0031883949413895607,
      "learning_rate": 0.00018400077650223263,
      "loss": 11.5,
      "step": 282
    },
    {
      "epoch": 0.15458673312621624,
      "grad_norm": 0.0028871088288724422,
      "learning_rate": 0.0001838861906985379,
      "loss": 11.5,
      "step": 283
    },
    {
      "epoch": 0.15513297599945375,
      "grad_norm": 0.0023424867540597916,
      "learning_rate": 0.00018377123197325842,
      "loss": 11.5,
      "step": 284
    },
    {
      "epoch": 0.15567921887269126,
      "grad_norm": 0.0029161826241761446,
      "learning_rate": 0.00018365590083745085,
      "loss": 11.5,
      "step": 285
    },
    {
      "epoch": 0.15622546174592877,
      "grad_norm": 0.002998099895194173,
      "learning_rate": 0.00018354019780382735,
      "loss": 11.5,
      "step": 286
    },
    {
      "epoch": 0.15677170461916629,
      "grad_norm": 0.001905181328766048,
      "learning_rate": 0.0001834241233867533,
      "loss": 11.5,
      "step": 287
    },
    {
      "epoch": 0.1573179474924038,
      "grad_norm": 0.002156859962269664,
      "learning_rate": 0.00018330767810224524,
      "loss": 11.5,
      "step": 288
    },
    {
      "epoch": 0.1578641903656413,
      "grad_norm": 0.002918890444561839,
      "learning_rate": 0.0001831908624679683,
      "loss": 11.5,
      "step": 289
    },
    {
      "epoch": 0.15841043323887882,
      "grad_norm": 0.003527702996507287,
      "learning_rate": 0.0001830736770032341,
      "loss": 11.5,
      "step": 290
    },
    {
      "epoch": 0.15895667611211636,
      "grad_norm": 0.002203888026997447,
      "learning_rate": 0.0001829561222289984,
      "loss": 11.5,
      "step": 291
    },
    {
      "epoch": 0.15950291898535388,
      "grad_norm": 0.002634590258821845,
      "learning_rate": 0.00018283819866785853,
      "loss": 11.5,
      "step": 292
    },
    {
      "epoch": 0.1600491618585914,
      "grad_norm": 0.004594285041093826,
      "learning_rate": 0.0001827199068440516,
      "loss": 11.5,
      "step": 293
    },
    {
      "epoch": 0.1605954047318289,
      "grad_norm": 0.0033522641751915216,
      "learning_rate": 0.00018260124728345162,
      "loss": 11.5,
      "step": 294
    },
    {
      "epoch": 0.1611416476050664,
      "grad_norm": 0.0031829439103603363,
      "learning_rate": 0.00018248222051356754,
      "loss": 11.5,
      "step": 295
    },
    {
      "epoch": 0.16168789047830393,
      "grad_norm": 0.004577325191348791,
      "learning_rate": 0.00018236282706354063,
      "loss": 11.5,
      "step": 296
    },
    {
      "epoch": 0.16223413335154144,
      "grad_norm": 0.0022684200666844845,
      "learning_rate": 0.00018224306746414238,
      "loss": 11.5,
      "step": 297
    },
    {
      "epoch": 0.16278037622477895,
      "grad_norm": 0.005714023485779762,
      "learning_rate": 0.00018212294224777197,
      "loss": 11.5,
      "step": 298
    },
    {
      "epoch": 0.16332661909801646,
      "grad_norm": 0.004955889191478491,
      "learning_rate": 0.00018200245194845399,
      "loss": 11.5,
      "step": 299
    },
    {
      "epoch": 0.16387286197125397,
      "grad_norm": 0.005777426529675722,
      "learning_rate": 0.00018188159710183594,
      "loss": 11.5,
      "step": 300
    },
    {
      "epoch": 0.1644191048444915,
      "grad_norm": 0.0035528906155377626,
      "learning_rate": 0.000181760378245186,
      "loss": 11.5,
      "step": 301
    },
    {
      "epoch": 0.164965347717729,
      "grad_norm": 0.002069627633318305,
      "learning_rate": 0.00018163879591739067,
      "loss": 11.5,
      "step": 302
    },
    {
      "epoch": 0.1655115905909665,
      "grad_norm": 0.003029781859368086,
      "learning_rate": 0.0001815168506589521,
      "loss": 11.5,
      "step": 303
    },
    {
      "epoch": 0.16605783346420402,
      "grad_norm": 0.0020544673316180706,
      "learning_rate": 0.000181394543011986,
      "loss": 11.5,
      "step": 304
    },
    {
      "epoch": 0.16660407633744154,
      "grad_norm": 0.004713078960776329,
      "learning_rate": 0.00018127187352021907,
      "loss": 11.5,
      "step": 305
    },
    {
      "epoch": 0.16715031921067905,
      "grad_norm": 0.0019469287944957614,
      "learning_rate": 0.0001811488427289866,
      "loss": 11.5,
      "step": 306
    },
    {
      "epoch": 0.16769656208391656,
      "grad_norm": 0.0028484466020017862,
      "learning_rate": 0.00018102545118523007,
      "loss": 11.5,
      "step": 307
    },
    {
      "epoch": 0.16824280495715407,
      "grad_norm": 0.0017475357744842768,
      "learning_rate": 0.00018090169943749476,
      "loss": 11.5,
      "step": 308
    },
    {
      "epoch": 0.1687890478303916,
      "grad_norm": 0.003255147021263838,
      "learning_rate": 0.00018077758803592718,
      "loss": 11.5,
      "step": 309
    },
    {
      "epoch": 0.1693352907036291,
      "grad_norm": 0.005029660649597645,
      "learning_rate": 0.00018065311753227273,
      "loss": 11.5,
      "step": 310
    },
    {
      "epoch": 0.1698815335768666,
      "grad_norm": 0.0024491765070706606,
      "learning_rate": 0.0001805282884798732,
      "loss": 11.5,
      "step": 311
    },
    {
      "epoch": 0.17042777645010412,
      "grad_norm": 0.002166147343814373,
      "learning_rate": 0.00018040310143366446,
      "loss": 11.5,
      "step": 312
    },
    {
      "epoch": 0.17097401932334164,
      "grad_norm": 0.003262473735958338,
      "learning_rate": 0.00018027755695017368,
      "loss": 11.5,
      "step": 313
    },
    {
      "epoch": 0.17152026219657915,
      "grad_norm": 0.005922917742282152,
      "learning_rate": 0.00018015165558751717,
      "loss": 11.5,
      "step": 314
    },
    {
      "epoch": 0.17206650506981666,
      "grad_norm": 0.0025184715632349253,
      "learning_rate": 0.00018002539790539773,
      "loss": 11.5,
      "step": 315
    },
    {
      "epoch": 0.17261274794305417,
      "grad_norm": 0.00378241715952754,
      "learning_rate": 0.00017989878446510215,
      "loss": 11.5,
      "step": 316
    },
    {
      "epoch": 0.17315899081629169,
      "grad_norm": 0.002998926443979144,
      "learning_rate": 0.00017977181582949888,
      "loss": 11.5,
      "step": 317
    },
    {
      "epoch": 0.1737052336895292,
      "grad_norm": 0.0031140188220888376,
      "learning_rate": 0.0001796444925630353,
      "loss": 11.5,
      "step": 318
    },
    {
      "epoch": 0.1742514765627667,
      "grad_norm": 0.016652435064315796,
      "learning_rate": 0.00017951681523173542,
      "loss": 11.5,
      "step": 319
    },
    {
      "epoch": 0.17479771943600422,
      "grad_norm": 0.001634707092307508,
      "learning_rate": 0.0001793887844031972,
      "loss": 11.5,
      "step": 320
    },
    {
      "epoch": 0.17534396230924174,
      "grad_norm": 0.0048229871317744255,
      "learning_rate": 0.00017926040064659014,
      "loss": 11.5,
      "step": 321
    },
    {
      "epoch": 0.17589020518247925,
      "grad_norm": 0.0018697242485359311,
      "learning_rate": 0.0001791316645326526,
      "loss": 11.5,
      "step": 322
    },
    {
      "epoch": 0.17643644805571676,
      "grad_norm": 0.002689193934202194,
      "learning_rate": 0.00017900257663368963,
      "loss": 11.5,
      "step": 323
    },
    {
      "epoch": 0.17698269092895427,
      "grad_norm": 0.003174367593601346,
      "learning_rate": 0.0001788731375235698,
      "loss": 11.5,
      "step": 324
    },
    {
      "epoch": 0.1775289338021918,
      "grad_norm": 0.0027308883145451546,
      "learning_rate": 0.00017874334777772327,
      "loss": 11.5,
      "step": 325
    },
    {
      "epoch": 0.17807517667542933,
      "grad_norm": 0.006789851933717728,
      "learning_rate": 0.00017861320797313892,
      "loss": 11.5,
      "step": 326
    },
    {
      "epoch": 0.17862141954866684,
      "grad_norm": 0.00444045988842845,
      "learning_rate": 0.0001784827186883618,
      "loss": 11.5,
      "step": 327
    },
    {
      "epoch": 0.17916766242190435,
      "grad_norm": 0.0024206070229411125,
      "learning_rate": 0.00017835188050349064,
      "loss": 11.5,
      "step": 328
    },
    {
      "epoch": 0.17971390529514186,
      "grad_norm": 0.00218951771967113,
      "learning_rate": 0.00017822069400017516,
      "loss": 11.5,
      "step": 329
    },
    {
      "epoch": 0.18026014816837937,
      "grad_norm": 0.003052780404686928,
      "learning_rate": 0.00017808915976161362,
      "loss": 11.5,
      "step": 330
    },
    {
      "epoch": 0.1808063910416169,
      "grad_norm": 0.0030408899765461683,
      "learning_rate": 0.00017795727837255015,
      "loss": 11.5,
      "step": 331
    },
    {
      "epoch": 0.1813526339148544,
      "grad_norm": 0.0030923194717615843,
      "learning_rate": 0.00017782505041927216,
      "loss": 11.5,
      "step": 332
    },
    {
      "epoch": 0.1818988767880919,
      "grad_norm": 0.002059780992567539,
      "learning_rate": 0.00017769247648960774,
      "loss": 11.5,
      "step": 333
    },
    {
      "epoch": 0.18244511966132942,
      "grad_norm": 0.005412459839135408,
      "learning_rate": 0.00017755955717292296,
      "loss": 11.5,
      "step": 334
    },
    {
      "epoch": 0.18299136253456694,
      "grad_norm": 0.0026739363092929125,
      "learning_rate": 0.00017742629306011944,
      "loss": 11.5,
      "step": 335
    },
    {
      "epoch": 0.18353760540780445,
      "grad_norm": 0.002516842680051923,
      "learning_rate": 0.00017729268474363154,
      "loss": 11.5,
      "step": 336
    },
    {
      "epoch": 0.18408384828104196,
      "grad_norm": 0.0025833912659436464,
      "learning_rate": 0.0001771587328174239,
      "loss": 11.5,
      "step": 337
    },
    {
      "epoch": 0.18463009115427947,
      "grad_norm": 0.00230031437240541,
      "learning_rate": 0.0001770244378769885,
      "loss": 11.5,
      "step": 338
    },
    {
      "epoch": 0.185176334027517,
      "grad_norm": 0.0030858968384563923,
      "learning_rate": 0.0001768898005193425,
      "loss": 11.5,
      "step": 339
    },
    {
      "epoch": 0.1857225769007545,
      "grad_norm": 0.002494861138984561,
      "learning_rate": 0.000176754821343025,
      "loss": 11.5,
      "step": 340
    },
    {
      "epoch": 0.186268819773992,
      "grad_norm": 0.0028091270942240953,
      "learning_rate": 0.0001766195009480949,
      "loss": 11.5,
      "step": 341
    },
    {
      "epoch": 0.18681506264722952,
      "grad_norm": 0.003020921954885125,
      "learning_rate": 0.0001764838399361279,
      "loss": 11.5,
      "step": 342
    },
    {
      "epoch": 0.18736130552046704,
      "grad_norm": 0.006396893877536058,
      "learning_rate": 0.00017634783891021393,
      "loss": 11.5,
      "step": 343
    },
    {
      "epoch": 0.18790754839370455,
      "grad_norm": 0.004477501381188631,
      "learning_rate": 0.00017621149847495458,
      "loss": 11.5,
      "step": 344
    },
    {
      "epoch": 0.18845379126694206,
      "grad_norm": 0.004152910318225622,
      "learning_rate": 0.00017607481923646016,
      "loss": 11.5,
      "step": 345
    },
    {
      "epoch": 0.18900003414017957,
      "grad_norm": 0.0053338236175477505,
      "learning_rate": 0.0001759378018023473,
      "loss": 11.5,
      "step": 346
    },
    {
      "epoch": 0.18954627701341709,
      "grad_norm": 0.00430544838309288,
      "learning_rate": 0.00017580044678173592,
      "loss": 11.5,
      "step": 347
    },
    {
      "epoch": 0.1900925198866546,
      "grad_norm": 0.005962560418993235,
      "learning_rate": 0.00017566275478524693,
      "loss": 11.5,
      "step": 348
    },
    {
      "epoch": 0.1906387627598921,
      "grad_norm": 0.003991235513240099,
      "learning_rate": 0.0001755247264249991,
      "loss": 11.5,
      "step": 349
    },
    {
      "epoch": 0.19118500563312962,
      "grad_norm": 0.004732074216008186,
      "learning_rate": 0.0001753863623146066,
      "loss": 11.5,
      "step": 350
    },
    {
      "epoch": 0.19173124850636714,
      "grad_norm": 0.0029028833378106356,
      "learning_rate": 0.00017524766306917618,
      "loss": 11.5,
      "step": 351
    },
    {
      "epoch": 0.19227749137960465,
      "grad_norm": 0.0033172841649502516,
      "learning_rate": 0.0001751086293053045,
      "loss": 11.5,
      "step": 352
    },
    {
      "epoch": 0.19282373425284216,
      "grad_norm": 0.0026081353425979614,
      "learning_rate": 0.0001749692616410753,
      "loss": 11.5,
      "step": 353
    },
    {
      "epoch": 0.19336997712607967,
      "grad_norm": 0.0030906107276678085,
      "learning_rate": 0.00017482956069605668,
      "loss": 11.5,
      "step": 354
    },
    {
      "epoch": 0.19391621999931719,
      "grad_norm": 0.0024727594573050737,
      "learning_rate": 0.00017468952709129846,
      "loss": 11.5,
      "step": 355
    },
    {
      "epoch": 0.1944624628725547,
      "grad_norm": 0.002029180061072111,
      "learning_rate": 0.00017454916144932922,
      "loss": 11.5,
      "step": 356
    },
    {
      "epoch": 0.1950087057457922,
      "grad_norm": 0.0027439759578555822,
      "learning_rate": 0.0001744084643941536,
      "loss": 11.5,
      "step": 357
    },
    {
      "epoch": 0.19555494861902972,
      "grad_norm": 0.002265994204208255,
      "learning_rate": 0.00017426743655124974,
      "loss": 11.5,
      "step": 358
    },
    {
      "epoch": 0.19610119149226726,
      "grad_norm": 0.0018808526219800115,
      "learning_rate": 0.0001741260785475661,
      "loss": 11.5,
      "step": 359
    },
    {
      "epoch": 0.19664743436550478,
      "grad_norm": 0.003247732063755393,
      "learning_rate": 0.00017398439101151905,
      "loss": 11.5,
      "step": 360
    },
    {
      "epoch": 0.1971936772387423,
      "grad_norm": 0.0024454586673527956,
      "learning_rate": 0.00017384237457298987,
      "loss": 11.5,
      "step": 361
    },
    {
      "epoch": 0.1977399201119798,
      "grad_norm": 0.0023723850026726723,
      "learning_rate": 0.00017370002986332193,
      "loss": 11.5,
      "step": 362
    },
    {
      "epoch": 0.1982861629852173,
      "grad_norm": 0.0023238896392285824,
      "learning_rate": 0.00017355735751531807,
      "loss": 11.5,
      "step": 363
    },
    {
      "epoch": 0.19883240585845482,
      "grad_norm": 0.0037743044085800648,
      "learning_rate": 0.00017341435816323756,
      "loss": 11.5,
      "step": 364
    },
    {
      "epoch": 0.19937864873169234,
      "grad_norm": 0.003563903272151947,
      "learning_rate": 0.00017327103244279348,
      "loss": 11.5,
      "step": 365
    },
    {
      "epoch": 0.19992489160492985,
      "grad_norm": 0.0030282444786280394,
      "learning_rate": 0.00017312738099114973,
      "loss": 11.5,
      "step": 366
    },
    {
      "epoch": 0.20047113447816736,
      "grad_norm": 0.002511603059247136,
      "learning_rate": 0.00017298340444691835,
      "loss": 11.5,
      "step": 367
    },
    {
      "epoch": 0.20101737735140487,
      "grad_norm": 0.0025034563150256872,
      "learning_rate": 0.00017283910345015647,
      "loss": 11.5,
      "step": 368
    },
    {
      "epoch": 0.2015636202246424,
      "grad_norm": 0.0036226874217391014,
      "learning_rate": 0.0001726944786423637,
      "loss": 11.5,
      "step": 369
    },
    {
      "epoch": 0.2021098630978799,
      "grad_norm": 0.002845035633072257,
      "learning_rate": 0.00017254953066647913,
      "loss": 11.5,
      "step": 370
    },
    {
      "epoch": 0.2026561059711174,
      "grad_norm": 0.0018421168206259608,
      "learning_rate": 0.00017240426016687863,
      "loss": 11.5,
      "step": 371
    },
    {
      "epoch": 0.20320234884435492,
      "grad_norm": 0.002946121385321021,
      "learning_rate": 0.00017225866778937165,
      "loss": 11.5,
      "step": 372
    },
    {
      "epoch": 0.20374859171759244,
      "grad_norm": 0.0040790969505906105,
      "learning_rate": 0.00017211275418119876,
      "loss": 11.5,
      "step": 373
    },
    {
      "epoch": 0.20429483459082995,
      "grad_norm": 0.001974373823031783,
      "learning_rate": 0.0001719665199910285,
      "loss": 11.5,
      "step": 374
    },
    {
      "epoch": 0.20484107746406746,
      "grad_norm": 0.005413064733147621,
      "learning_rate": 0.00017181996586895454,
      "loss": 11.5,
      "step": 375
    },
    {
      "epoch": 0.20484107746406746,
      "eval_loss": 11.5,
      "eval_runtime": 23.1724,
      "eval_samples_per_second": 133.089,
      "eval_steps_per_second": 66.545,
      "step": 375
    },
    {
      "epoch": 0.20538732033730497,
      "grad_norm": 0.003951927646994591,
      "learning_rate": 0.00017167309246649297,
      "loss": 11.5,
      "step": 376
    },
    {
      "epoch": 0.2059335632105425,
      "grad_norm": 0.011415340937674046,
      "learning_rate": 0.0001715259004365791,
      "loss": 11.5,
      "step": 377
    },
    {
      "epoch": 0.20647980608378,
      "grad_norm": 0.007708726916462183,
      "learning_rate": 0.00017137839043356484,
      "loss": 11.5,
      "step": 378
    },
    {
      "epoch": 0.2070260489570175,
      "grad_norm": 0.003585737431421876,
      "learning_rate": 0.00017123056311321562,
      "loss": 11.5,
      "step": 379
    },
    {
      "epoch": 0.20757229183025502,
      "grad_norm": 0.002808233490213752,
      "learning_rate": 0.0001710824191327075,
      "loss": 11.5,
      "step": 380
    },
    {
      "epoch": 0.20811853470349254,
      "grad_norm": 0.001834494643844664,
      "learning_rate": 0.00017093395915062428,
      "loss": 11.5,
      "step": 381
    },
    {
      "epoch": 0.20866477757673005,
      "grad_norm": 0.004499961156398058,
      "learning_rate": 0.00017078518382695465,
      "loss": 11.5,
      "step": 382
    },
    {
      "epoch": 0.20921102044996756,
      "grad_norm": 0.003169924020767212,
      "learning_rate": 0.00017063609382308908,
      "loss": 11.5,
      "step": 383
    },
    {
      "epoch": 0.20975726332320507,
      "grad_norm": 0.002276839455589652,
      "learning_rate": 0.00017048668980181698,
      "loss": 11.5,
      "step": 384
    },
    {
      "epoch": 0.21030350619644259,
      "grad_norm": 0.003066045232117176,
      "learning_rate": 0.00017033697242732377,
      "loss": 11.5,
      "step": 385
    },
    {
      "epoch": 0.2108497490696801,
      "grad_norm": 0.002703852951526642,
      "learning_rate": 0.0001701869423651879,
      "loss": 11.5,
      "step": 386
    },
    {
      "epoch": 0.2113959919429176,
      "grad_norm": 0.004381160717457533,
      "learning_rate": 0.00017003660028237793,
      "loss": 11.5,
      "step": 387
    },
    {
      "epoch": 0.21194223481615512,
      "grad_norm": 0.001723824068903923,
      "learning_rate": 0.00016988594684724947,
      "loss": 11.5,
      "step": 388
    },
    {
      "epoch": 0.21248847768939264,
      "grad_norm": 0.0019858500454574823,
      "learning_rate": 0.00016973498272954222,
      "loss": 11.5,
      "step": 389
    },
    {
      "epoch": 0.21303472056263015,
      "grad_norm": 0.002614605240523815,
      "learning_rate": 0.00016958370860037717,
      "loss": 11.5,
      "step": 390
    },
    {
      "epoch": 0.21358096343586766,
      "grad_norm": 0.0018274768954142928,
      "learning_rate": 0.00016943212513225345,
      "loss": 11.5,
      "step": 391
    },
    {
      "epoch": 0.21412720630910517,
      "grad_norm": 0.0032833365257829428,
      "learning_rate": 0.00016928023299904533,
      "loss": 11.5,
      "step": 392
    },
    {
      "epoch": 0.2146734491823427,
      "grad_norm": 0.004258325789123774,
      "learning_rate": 0.0001691280328759992,
      "loss": 11.5,
      "step": 393
    },
    {
      "epoch": 0.21521969205558023,
      "grad_norm": 0.0031729962211102247,
      "learning_rate": 0.00016897552543973084,
      "loss": 11.5,
      "step": 394
    },
    {
      "epoch": 0.21576593492881774,
      "grad_norm": 0.0021603042259812355,
      "learning_rate": 0.00016882271136822206,
      "loss": 11.5,
      "step": 395
    },
    {
      "epoch": 0.21631217780205525,
      "grad_norm": 0.003850439563393593,
      "learning_rate": 0.0001686695913408179,
      "loss": 11.5,
      "step": 396
    },
    {
      "epoch": 0.21685842067529276,
      "grad_norm": 0.0033242283388972282,
      "learning_rate": 0.0001685161660382235,
      "loss": 11.5,
      "step": 397
    },
    {
      "epoch": 0.21740466354853027,
      "grad_norm": 0.005488258320838213,
      "learning_rate": 0.00016836243614250113,
      "loss": 11.5,
      "step": 398
    },
    {
      "epoch": 0.2179509064217678,
      "grad_norm": 0.005948767997324467,
      "learning_rate": 0.00016820840233706719,
      "loss": 11.5,
      "step": 399
    },
    {
      "epoch": 0.2184971492950053,
      "grad_norm": 0.007582229562103748,
      "learning_rate": 0.0001680540653066891,
      "loss": 11.5,
      "step": 400
    },
    {
      "epoch": 0.2190433921682428,
      "grad_norm": 0.0026957413647323847,
      "learning_rate": 0.00016789942573748232,
      "loss": 11.5,
      "step": 401
    },
    {
      "epoch": 0.21958963504148032,
      "grad_norm": 0.00211122608743608,
      "learning_rate": 0.0001677444843169072,
      "loss": 11.5,
      "step": 402
    },
    {
      "epoch": 0.22013587791471784,
      "grad_norm": 0.001941608265042305,
      "learning_rate": 0.00016758924173376603,
      "loss": 11.5,
      "step": 403
    },
    {
      "epoch": 0.22068212078795535,
      "grad_norm": 0.002123320708051324,
      "learning_rate": 0.0001674336986781999,
      "loss": 11.5,
      "step": 404
    },
    {
      "epoch": 0.22122836366119286,
      "grad_norm": 0.0017500972608104348,
      "learning_rate": 0.00016727785584168581,
      "loss": 11.5,
      "step": 405
    },
    {
      "epoch": 0.22177460653443037,
      "grad_norm": 0.0038807126693427563,
      "learning_rate": 0.0001671217139170333,
      "loss": 11.5,
      "step": 406
    },
    {
      "epoch": 0.2223208494076679,
      "grad_norm": 0.0017392388544976711,
      "learning_rate": 0.00016696527359838154,
      "loss": 11.5,
      "step": 407
    },
    {
      "epoch": 0.2228670922809054,
      "grad_norm": 0.004699967801570892,
      "learning_rate": 0.00016680853558119632,
      "loss": 11.5,
      "step": 408
    },
    {
      "epoch": 0.2234133351541429,
      "grad_norm": 0.0019580137450248003,
      "learning_rate": 0.0001666515005622668,
      "loss": 11.5,
      "step": 409
    },
    {
      "epoch": 0.22395957802738042,
      "grad_norm": 0.003774157725274563,
      "learning_rate": 0.0001664941692397025,
      "loss": 11.5,
      "step": 410
    },
    {
      "epoch": 0.22450582090061794,
      "grad_norm": 0.0014043869450688362,
      "learning_rate": 0.00016633654231293013,
      "loss": 11.5,
      "step": 411
    },
    {
      "epoch": 0.22505206377385545,
      "grad_norm": 0.0032073333859443665,
      "learning_rate": 0.00016617862048269065,
      "loss": 11.5,
      "step": 412
    },
    {
      "epoch": 0.22559830664709296,
      "grad_norm": 0.002138904295861721,
      "learning_rate": 0.00016602040445103588,
      "loss": 11.5,
      "step": 413
    },
    {
      "epoch": 0.22614454952033047,
      "grad_norm": 0.0015345015563070774,
      "learning_rate": 0.00016586189492132566,
      "loss": 11.5,
      "step": 414
    },
    {
      "epoch": 0.22669079239356799,
      "grad_norm": 0.00214973627589643,
      "learning_rate": 0.00016570309259822453,
      "loss": 11.5,
      "step": 415
    },
    {
      "epoch": 0.2272370352668055,
      "grad_norm": 0.0017527873860672116,
      "learning_rate": 0.0001655439981876987,
      "loss": 11.5,
      "step": 416
    },
    {
      "epoch": 0.227783278140043,
      "grad_norm": 0.0026633741799741983,
      "learning_rate": 0.00016538461239701277,
      "loss": 11.5,
      "step": 417
    },
    {
      "epoch": 0.22832952101328052,
      "grad_norm": 0.003344568656757474,
      "learning_rate": 0.00016522493593472683,
      "loss": 11.5,
      "step": 418
    },
    {
      "epoch": 0.22887576388651804,
      "grad_norm": 0.0020820987410843372,
      "learning_rate": 0.0001650649695106931,
      "loss": 11.5,
      "step": 419
    },
    {
      "epoch": 0.22942200675975555,
      "grad_norm": 0.0032771332189440727,
      "learning_rate": 0.00016490471383605288,
      "loss": 11.5,
      "step": 420
    },
    {
      "epoch": 0.22996824963299306,
      "grad_norm": 0.0025349047500640154,
      "learning_rate": 0.00016474416962323325,
      "loss": 11.5,
      "step": 421
    },
    {
      "epoch": 0.23051449250623057,
      "grad_norm": 0.0018015307141467929,
      "learning_rate": 0.00016458333758594414,
      "loss": 11.5,
      "step": 422
    },
    {
      "epoch": 0.23106073537946809,
      "grad_norm": 0.002641750266775489,
      "learning_rate": 0.00016442221843917496,
      "loss": 11.5,
      "step": 423
    },
    {
      "epoch": 0.2316069782527056,
      "grad_norm": 0.00211442937143147,
      "learning_rate": 0.00016426081289919143,
      "loss": 11.5,
      "step": 424
    },
    {
      "epoch": 0.2321532211259431,
      "grad_norm": 0.0010476912138983607,
      "learning_rate": 0.0001640991216835326,
      "loss": 11.5,
      "step": 425
    },
    {
      "epoch": 0.23269946399918062,
      "grad_norm": 0.003821734571829438,
      "learning_rate": 0.00016393714551100734,
      "loss": 11.5,
      "step": 426
    },
    {
      "epoch": 0.23324570687241816,
      "grad_norm": 0.006275371182709932,
      "learning_rate": 0.0001637748851016914,
      "loss": 11.5,
      "step": 427
    },
    {
      "epoch": 0.23379194974565567,
      "grad_norm": 0.014230850152671337,
      "learning_rate": 0.00016361234117692413,
      "loss": 11.5,
      "step": 428
    },
    {
      "epoch": 0.2343381926188932,
      "grad_norm": 0.0029954607598483562,
      "learning_rate": 0.00016344951445930526,
      "loss": 11.5,
      "step": 429
    },
    {
      "epoch": 0.2348844354921307,
      "grad_norm": 0.0029866104014217854,
      "learning_rate": 0.0001632864056726917,
      "loss": 11.5,
      "step": 430
    },
    {
      "epoch": 0.2354306783653682,
      "grad_norm": 0.001989017240703106,
      "learning_rate": 0.00016312301554219426,
      "loss": 11.5,
      "step": 431
    },
    {
      "epoch": 0.23597692123860572,
      "grad_norm": 0.002801706548780203,
      "learning_rate": 0.00016295934479417453,
      "loss": 11.5,
      "step": 432
    },
    {
      "epoch": 0.23652316411184324,
      "grad_norm": 0.004477903712540865,
      "learning_rate": 0.00016279539415624164,
      "loss": 11.5,
      "step": 433
    },
    {
      "epoch": 0.23706940698508075,
      "grad_norm": 0.0018590339459478855,
      "learning_rate": 0.0001626311643572489,
      "loss": 11.5,
      "step": 434
    },
    {
      "epoch": 0.23761564985831826,
      "grad_norm": 0.0038578978274017572,
      "learning_rate": 0.00016246665612729074,
      "loss": 11.5,
      "step": 435
    },
    {
      "epoch": 0.23816189273155577,
      "grad_norm": 0.002760401228442788,
      "learning_rate": 0.00016230187019769928,
      "loss": 11.5,
      "step": 436
    },
    {
      "epoch": 0.2387081356047933,
      "grad_norm": 0.002469010651111603,
      "learning_rate": 0.00016213680730104124,
      "loss": 11.5,
      "step": 437
    },
    {
      "epoch": 0.2392543784780308,
      "grad_norm": 0.002470056526362896,
      "learning_rate": 0.0001619714681711146,
      "loss": 11.5,
      "step": 438
    },
    {
      "epoch": 0.2398006213512683,
      "grad_norm": 0.00203881086781621,
      "learning_rate": 0.00016180585354294536,
      "loss": 11.5,
      "step": 439
    },
    {
      "epoch": 0.24034686422450582,
      "grad_norm": 0.0033059976994991302,
      "learning_rate": 0.00016163996415278424,
      "loss": 11.5,
      "step": 440
    },
    {
      "epoch": 0.24089310709774334,
      "grad_norm": 0.0031290745828300714,
      "learning_rate": 0.00016147380073810346,
      "loss": 11.5,
      "step": 441
    },
    {
      "epoch": 0.24143934997098085,
      "grad_norm": 0.003260674886405468,
      "learning_rate": 0.0001613073640375934,
      "loss": 11.5,
      "step": 442
    },
    {
      "epoch": 0.24198559284421836,
      "grad_norm": 0.0029559109825640917,
      "learning_rate": 0.00016114065479115946,
      "loss": 11.5,
      "step": 443
    },
    {
      "epoch": 0.24253183571745587,
      "grad_norm": 0.00397954136133194,
      "learning_rate": 0.00016097367373991842,
      "loss": 11.5,
      "step": 444
    },
    {
      "epoch": 0.24307807859069339,
      "grad_norm": 0.003918122500181198,
      "learning_rate": 0.00016080642162619565,
      "loss": 11.5,
      "step": 445
    },
    {
      "epoch": 0.2436243214639309,
      "grad_norm": 0.002881030086427927,
      "learning_rate": 0.0001606388991935214,
      "loss": 11.5,
      "step": 446
    },
    {
      "epoch": 0.2441705643371684,
      "grad_norm": 0.00418865168467164,
      "learning_rate": 0.0001604711071866277,
      "loss": 11.5,
      "step": 447
    },
    {
      "epoch": 0.24471680721040592,
      "grad_norm": 0.00576843973249197,
      "learning_rate": 0.00016030304635144494,
      "loss": 11.5,
      "step": 448
    },
    {
      "epoch": 0.24526305008364344,
      "grad_norm": 0.004736778326332569,
      "learning_rate": 0.00016013471743509862,
      "loss": 11.5,
      "step": 449
    },
    {
      "epoch": 0.24580929295688095,
      "grad_norm": 0.007776481099426746,
      "learning_rate": 0.00015996612118590603,
      "loss": 11.5,
      "step": 450
    },
    {
      "epoch": 0.24635553583011846,
      "grad_norm": 0.002060589147731662,
      "learning_rate": 0.00015979725835337294,
      "loss": 11.5,
      "step": 451
    },
    {
      "epoch": 0.24690177870335597,
      "grad_norm": 0.0021606662776321173,
      "learning_rate": 0.00015962812968819016,
      "loss": 11.5,
      "step": 452
    },
    {
      "epoch": 0.24744802157659349,
      "grad_norm": 0.0032455027103424072,
      "learning_rate": 0.0001594587359422303,
      "loss": 11.5,
      "step": 453
    },
    {
      "epoch": 0.247994264449831,
      "grad_norm": 0.002954303054139018,
      "learning_rate": 0.0001592890778685444,
      "loss": 11.5,
      "step": 454
    },
    {
      "epoch": 0.2485405073230685,
      "grad_norm": 0.002560376189649105,
      "learning_rate": 0.00015911915622135862,
      "loss": 11.5,
      "step": 455
    },
    {
      "epoch": 0.24908675019630602,
      "grad_norm": 0.002631485229358077,
      "learning_rate": 0.00015894897175607086,
      "loss": 11.5,
      "step": 456
    },
    {
      "epoch": 0.24963299306954354,
      "grad_norm": 0.0025433823466300964,
      "learning_rate": 0.00015877852522924732,
      "loss": 11.5,
      "step": 457
    },
    {
      "epoch": 0.2501792359427811,
      "grad_norm": 0.002800427610054612,
      "learning_rate": 0.00015860781739861928,
      "loss": 11.5,
      "step": 458
    },
    {
      "epoch": 0.2507254788160186,
      "grad_norm": 0.008643715642392635,
      "learning_rate": 0.00015843684902307962,
      "loss": 11.5,
      "step": 459
    },
    {
      "epoch": 0.2512717216892561,
      "grad_norm": 0.0031836614944040775,
      "learning_rate": 0.00015826562086267956,
      "loss": 11.5,
      "step": 460
    },
    {
      "epoch": 0.2518179645624936,
      "grad_norm": 0.004226151388138533,
      "learning_rate": 0.00015809413367862512,
      "loss": 11.5,
      "step": 461
    },
    {
      "epoch": 0.2523642074357311,
      "grad_norm": 0.002880207961425185,
      "learning_rate": 0.00015792238823327388,
      "loss": 11.5,
      "step": 462
    },
    {
      "epoch": 0.25291045030896864,
      "grad_norm": 0.0020592620130628347,
      "learning_rate": 0.00015775038529013152,
      "loss": 11.5,
      "step": 463
    },
    {
      "epoch": 0.25345669318220615,
      "grad_norm": 0.0028274275828152895,
      "learning_rate": 0.0001575781256138485,
      "loss": 11.5,
      "step": 464
    },
    {
      "epoch": 0.25400293605544366,
      "grad_norm": 0.0028644087724387646,
      "learning_rate": 0.00015740560997021648,
      "loss": 11.5,
      "step": 465
    },
    {
      "epoch": 0.2545491789286812,
      "grad_norm": 0.001992451027035713,
      "learning_rate": 0.00015723283912616513,
      "loss": 11.5,
      "step": 466
    },
    {
      "epoch": 0.2550954218019187,
      "grad_norm": 0.00540782418102026,
      "learning_rate": 0.00015705981384975866,
      "loss": 11.5,
      "step": 467
    },
    {
      "epoch": 0.2556416646751562,
      "grad_norm": 0.002315417630597949,
      "learning_rate": 0.0001568865349101923,
      "loss": 11.5,
      "step": 468
    },
    {
      "epoch": 0.2561879075483937,
      "grad_norm": 0.0015409248881042004,
      "learning_rate": 0.00015671300307778898,
      "loss": 11.5,
      "step": 469
    },
    {
      "epoch": 0.2567341504216312,
      "grad_norm": 0.0036984607577323914,
      "learning_rate": 0.00015653921912399589,
      "loss": 11.5,
      "step": 470
    },
    {
      "epoch": 0.25728039329486874,
      "grad_norm": 0.0012320042587816715,
      "learning_rate": 0.00015636518382138107,
      "loss": 11.5,
      "step": 471
    },
    {
      "epoch": 0.25782663616810625,
      "grad_norm": 0.003385546151548624,
      "learning_rate": 0.0001561908979436299,
      "loss": 11.5,
      "step": 472
    },
    {
      "epoch": 0.25837287904134376,
      "grad_norm": 0.002360150683671236,
      "learning_rate": 0.00015601636226554168,
      "loss": 11.5,
      "step": 473
    },
    {
      "epoch": 0.2589191219145813,
      "grad_norm": 0.0020710271783173084,
      "learning_rate": 0.00015584157756302634,
      "loss": 11.5,
      "step": 474
    },
    {
      "epoch": 0.2594653647878188,
      "grad_norm": 0.0033624109346419573,
      "learning_rate": 0.0001556665446131007,
      "loss": 11.5,
      "step": 475
    },
    {
      "epoch": 0.2600116076610563,
      "grad_norm": 0.006726677995175123,
      "learning_rate": 0.00015549126419388536,
      "loss": 11.5,
      "step": 476
    },
    {
      "epoch": 0.2605578505342938,
      "grad_norm": 0.0050238389521837234,
      "learning_rate": 0.0001553157370846009,
      "loss": 11.5,
      "step": 477
    },
    {
      "epoch": 0.2611040934075313,
      "grad_norm": 0.015043279156088829,
      "learning_rate": 0.00015513996406556465,
      "loss": 11.5,
      "step": 478
    },
    {
      "epoch": 0.26165033628076884,
      "grad_norm": 0.003200068837031722,
      "learning_rate": 0.00015496394591818716,
      "loss": 11.5,
      "step": 479
    },
    {
      "epoch": 0.26219657915400635,
      "grad_norm": 0.0013106224359944463,
      "learning_rate": 0.0001547876834249687,
      "loss": 11.5,
      "step": 480
    },
    {
      "epoch": 0.26274282202724386,
      "grad_norm": 0.0020858210045844316,
      "learning_rate": 0.00015461117736949577,
      "loss": 11.5,
      "step": 481
    },
    {
      "epoch": 0.2632890649004814,
      "grad_norm": 0.0023034035693854094,
      "learning_rate": 0.00015443442853643762,
      "loss": 11.5,
      "step": 482
    },
    {
      "epoch": 0.2638353077737189,
      "grad_norm": 0.0026517340447753668,
      "learning_rate": 0.00015425743771154294,
      "loss": 11.5,
      "step": 483
    },
    {
      "epoch": 0.2643815506469564,
      "grad_norm": 0.0014967182651162148,
      "learning_rate": 0.00015408020568163602,
      "loss": 11.5,
      "step": 484
    },
    {
      "epoch": 0.2649277935201939,
      "grad_norm": 0.0022742683067917824,
      "learning_rate": 0.00015390273323461352,
      "loss": 11.5,
      "step": 485
    },
    {
      "epoch": 0.2654740363934314,
      "grad_norm": 0.005552656948566437,
      "learning_rate": 0.0001537250211594409,
      "loss": 11.5,
      "step": 486
    },
    {
      "epoch": 0.26602027926666894,
      "grad_norm": 0.003581566736102104,
      "learning_rate": 0.0001535470702461489,
      "loss": 11.5,
      "step": 487
    },
    {
      "epoch": 0.26656652213990645,
      "grad_norm": 0.0035480069927871227,
      "learning_rate": 0.00015336888128583,
      "loss": 11.5,
      "step": 488
    },
    {
      "epoch": 0.26711276501314396,
      "grad_norm": 0.002717837691307068,
      "learning_rate": 0.000153190455070635,
      "loss": 11.5,
      "step": 489
    },
    {
      "epoch": 0.2676590078863815,
      "grad_norm": 0.003168510738760233,
      "learning_rate": 0.00015301179239376938,
      "loss": 11.5,
      "step": 490
    },
    {
      "epoch": 0.268205250759619,
      "grad_norm": 0.002508896868675947,
      "learning_rate": 0.00015283289404948976,
      "loss": 11.5,
      "step": 491
    },
    {
      "epoch": 0.2687514936328565,
      "grad_norm": 0.004662887193262577,
      "learning_rate": 0.0001526537608331006,
      "loss": 11.5,
      "step": 492
    },
    {
      "epoch": 0.269297736506094,
      "grad_norm": 0.0009710693848319352,
      "learning_rate": 0.00015247439354095041,
      "loss": 11.5,
      "step": 493
    },
    {
      "epoch": 0.2698439793793315,
      "grad_norm": 0.002795287175104022,
      "learning_rate": 0.00015229479297042823,
      "loss": 11.5,
      "step": 494
    },
    {
      "epoch": 0.27039022225256903,
      "grad_norm": 0.002313773613423109,
      "learning_rate": 0.00015211495991996027,
      "loss": 11.5,
      "step": 495
    },
    {
      "epoch": 0.27093646512580655,
      "grad_norm": 0.0031737987883388996,
      "learning_rate": 0.0001519348951890062,
      "loss": 11.5,
      "step": 496
    },
    {
      "epoch": 0.27148270799904406,
      "grad_norm": 0.0037419102154672146,
      "learning_rate": 0.0001517545995780556,
      "loss": 11.5,
      "step": 497
    },
    {
      "epoch": 0.27202895087228157,
      "grad_norm": 0.0037045928183943033,
      "learning_rate": 0.00015157407388862452,
      "loss": 11.5,
      "step": 498
    },
    {
      "epoch": 0.2725751937455191,
      "grad_norm": 0.0032189227640628815,
      "learning_rate": 0.00015139331892325179,
      "loss": 11.5,
      "step": 499
    },
    {
      "epoch": 0.2731214366187566,
      "grad_norm": 0.006522863171994686,
      "learning_rate": 0.0001512123354854955,
      "loss": 11.5,
      "step": 500
    },
    {
      "epoch": 0.2736676794919941,
      "grad_norm": 0.0023784793447703123,
      "learning_rate": 0.0001510311243799295,
      "loss": 11.5,
      "step": 501
    },
    {
      "epoch": 0.2742139223652316,
      "grad_norm": 0.002028838964179158,
      "learning_rate": 0.00015084968641213958,
      "loss": 11.5,
      "step": 502
    },
    {
      "epoch": 0.27476016523846913,
      "grad_norm": 0.00381656875833869,
      "learning_rate": 0.00015066802238872023,
      "loss": 11.5,
      "step": 503
    },
    {
      "epoch": 0.27530640811170665,
      "grad_norm": 0.002986014587804675,
      "learning_rate": 0.0001504861331172709,
      "loss": 11.5,
      "step": 504
    },
    {
      "epoch": 0.27585265098494416,
      "grad_norm": 0.0024798542726784945,
      "learning_rate": 0.0001503040194063922,
      "loss": 11.5,
      "step": 505
    },
    {
      "epoch": 0.27639889385818167,
      "grad_norm": 0.0014252073597162962,
      "learning_rate": 0.00015012168206568268,
      "loss": 11.5,
      "step": 506
    },
    {
      "epoch": 0.2769451367314192,
      "grad_norm": 0.0026844933163374662,
      "learning_rate": 0.00014993912190573505,
      "loss": 11.5,
      "step": 507
    },
    {
      "epoch": 0.2774913796046567,
      "grad_norm": 0.0019227571319788694,
      "learning_rate": 0.00014975633973813242,
      "loss": 11.5,
      "step": 508
    },
    {
      "epoch": 0.2780376224778942,
      "grad_norm": 0.0032376281451433897,
      "learning_rate": 0.00014957333637544503,
      "loss": 11.5,
      "step": 509
    },
    {
      "epoch": 0.2785838653511317,
      "grad_norm": 0.0022490976843982935,
      "learning_rate": 0.00014939011263122634,
      "loss": 11.5,
      "step": 510
    },
    {
      "epoch": 0.27913010822436923,
      "grad_norm": 0.002172824228182435,
      "learning_rate": 0.0001492066693200096,
      "loss": 11.5,
      "step": 511
    },
    {
      "epoch": 0.27967635109760675,
      "grad_norm": 0.0022969343699514866,
      "learning_rate": 0.00014902300725730413,
      "loss": 11.5,
      "step": 512
    },
    {
      "epoch": 0.2802225939708443,
      "grad_norm": 0.001222333637997508,
      "learning_rate": 0.00014883912725959167,
      "loss": 11.5,
      "step": 513
    },
    {
      "epoch": 0.2807688368440818,
      "grad_norm": 0.0023129659239202738,
      "learning_rate": 0.00014865503014432292,
      "loss": 11.5,
      "step": 514
    },
    {
      "epoch": 0.28131507971731934,
      "grad_norm": 0.004228978417813778,
      "learning_rate": 0.00014847071672991367,
      "loss": 11.5,
      "step": 515
    },
    {
      "epoch": 0.28186132259055685,
      "grad_norm": 0.002110213041305542,
      "learning_rate": 0.0001482861878357414,
      "loss": 11.5,
      "step": 516
    },
    {
      "epoch": 0.28240756546379436,
      "grad_norm": 0.002400135388597846,
      "learning_rate": 0.00014810144428214144,
      "loss": 11.5,
      "step": 517
    },
    {
      "epoch": 0.2829538083370319,
      "grad_norm": 0.01990536041557789,
      "learning_rate": 0.0001479164868904034,
      "loss": 11.5,
      "step": 518
    },
    {
      "epoch": 0.2835000512102694,
      "grad_norm": 0.002049579517915845,
      "learning_rate": 0.00014773131648276758,
      "loss": 11.5,
      "step": 519
    },
    {
      "epoch": 0.2840462940835069,
      "grad_norm": 0.0029685578774660826,
      "learning_rate": 0.00014754593388242117,
      "loss": 11.5,
      "step": 520
    },
    {
      "epoch": 0.2845925369567444,
      "grad_norm": 0.001267296727746725,
      "learning_rate": 0.0001473603399134948,
      "loss": 11.5,
      "step": 521
    },
    {
      "epoch": 0.2851387798299819,
      "grad_norm": 0.0016093715094029903,
      "learning_rate": 0.0001471745354010586,
      "loss": 11.5,
      "step": 522
    },
    {
      "epoch": 0.28568502270321944,
      "grad_norm": 0.0042183720506727695,
      "learning_rate": 0.00014698852117111884,
      "loss": 11.5,
      "step": 523
    },
    {
      "epoch": 0.28623126557645695,
      "grad_norm": 0.0011894983472302556,
      "learning_rate": 0.000146802298050614,
      "loss": 11.5,
      "step": 524
    },
    {
      "epoch": 0.28677750844969446,
      "grad_norm": 0.0014218876603990793,
      "learning_rate": 0.0001466158668674112,
      "loss": 11.5,
      "step": 525
    },
    {
      "epoch": 0.287323751322932,
      "grad_norm": 0.002285819500684738,
      "learning_rate": 0.00014642922845030257,
      "loss": 11.5,
      "step": 526
    },
    {
      "epoch": 0.2878699941961695,
      "grad_norm": 0.007943695411086082,
      "learning_rate": 0.0001462423836290015,
      "loss": 11.5,
      "step": 527
    },
    {
      "epoch": 0.288416237069407,
      "grad_norm": 0.0014261419419199228,
      "learning_rate": 0.00014605533323413887,
      "loss": 11.5,
      "step": 528
    },
    {
      "epoch": 0.2889624799426445,
      "grad_norm": 0.0015476415865123272,
      "learning_rate": 0.00014586807809725962,
      "loss": 11.5,
      "step": 529
    },
    {
      "epoch": 0.289508722815882,
      "grad_norm": 0.0018527792999520898,
      "learning_rate": 0.00014568061905081875,
      "loss": 11.5,
      "step": 530
    },
    {
      "epoch": 0.29005496568911954,
      "grad_norm": 0.0033635981380939484,
      "learning_rate": 0.00014549295692817778,
      "loss": 11.5,
      "step": 531
    },
    {
      "epoch": 0.29060120856235705,
      "grad_norm": 0.00130665663164109,
      "learning_rate": 0.00014530509256360102,
      "loss": 11.5,
      "step": 532
    },
    {
      "epoch": 0.29114745143559456,
      "grad_norm": 0.006648363079875708,
      "learning_rate": 0.00014511702679225193,
      "loss": 11.5,
      "step": 533
    },
    {
      "epoch": 0.2916936943088321,
      "grad_norm": 0.002225110772997141,
      "learning_rate": 0.0001449287604501893,
      "loss": 11.5,
      "step": 534
    },
    {
      "epoch": 0.2922399371820696,
      "grad_norm": 0.003732465673238039,
      "learning_rate": 0.00014474029437436348,
      "loss": 11.5,
      "step": 535
    },
    {
      "epoch": 0.2927861800553071,
      "grad_norm": 0.0017800740897655487,
      "learning_rate": 0.00014455162940261285,
      "loss": 11.5,
      "step": 536
    },
    {
      "epoch": 0.2933324229285446,
      "grad_norm": 0.0030014649964869022,
      "learning_rate": 0.0001443627663736599,
      "loss": 11.5,
      "step": 537
    },
    {
      "epoch": 0.2938786658017821,
      "grad_norm": 0.002414504997432232,
      "learning_rate": 0.00014417370612710778,
      "loss": 11.5,
      "step": 538
    },
    {
      "epoch": 0.29442490867501964,
      "grad_norm": 0.0026760674081742764,
      "learning_rate": 0.00014398444950343623,
      "loss": 11.5,
      "step": 539
    },
    {
      "epoch": 0.29497115154825715,
      "grad_norm": 0.0026608051266521215,
      "learning_rate": 0.00014379499734399798,
      "loss": 11.5,
      "step": 540
    },
    {
      "epoch": 0.29551739442149466,
      "grad_norm": 0.003985922783613205,
      "learning_rate": 0.0001436053504910151,
      "loss": 11.5,
      "step": 541
    },
    {
      "epoch": 0.2960636372947322,
      "grad_norm": 0.004381160251796246,
      "learning_rate": 0.0001434155097875752,
      "loss": 11.5,
      "step": 542
    },
    {
      "epoch": 0.2966098801679697,
      "grad_norm": 0.004655285272747278,
      "learning_rate": 0.00014322547607762762,
      "loss": 11.5,
      "step": 543
    },
    {
      "epoch": 0.2971561230412072,
      "grad_norm": 0.0037039557937532663,
      "learning_rate": 0.0001430352502059797,
      "loss": 11.5,
      "step": 544
    },
    {
      "epoch": 0.2977023659144447,
      "grad_norm": 0.002017107792198658,
      "learning_rate": 0.0001428448330182931,
      "loss": 11.5,
      "step": 545
    },
    {
      "epoch": 0.2982486087876822,
      "grad_norm": 0.005651818588376045,
      "learning_rate": 0.00014265422536107993,
      "loss": 11.5,
      "step": 546
    },
    {
      "epoch": 0.29879485166091974,
      "grad_norm": 0.00297833071090281,
      "learning_rate": 0.00014246342808169914,
      "loss": 11.5,
      "step": 547
    },
    {
      "epoch": 0.29934109453415725,
      "grad_norm": 0.0013742482988163829,
      "learning_rate": 0.00014227244202835257,
      "loss": 11.5,
      "step": 548
    },
    {
      "epoch": 0.29988733740739476,
      "grad_norm": 0.005741320084780455,
      "learning_rate": 0.0001420812680500813,
      "loss": 11.5,
      "step": 549
    },
    {
      "epoch": 0.3004335802806323,
      "grad_norm": 0.002907536691054702,
      "learning_rate": 0.00014188990699676184,
      "loss": 11.5,
      "step": 550
    },
    {
      "epoch": 0.3009798231538698,
      "grad_norm": 0.0013665088918060064,
      "learning_rate": 0.00014169835971910238,
      "loss": 11.5,
      "step": 551
    },
    {
      "epoch": 0.3015260660271073,
      "grad_norm": 0.0019254100043326616,
      "learning_rate": 0.0001415066270686389,
      "loss": 11.5,
      "step": 552
    },
    {
      "epoch": 0.3020723089003448,
      "grad_norm": 0.002163645578548312,
      "learning_rate": 0.00014131470989773158,
      "loss": 11.5,
      "step": 553
    },
    {
      "epoch": 0.3026185517735823,
      "grad_norm": 0.001554999384097755,
      "learning_rate": 0.0001411226090595608,
      "loss": 11.5,
      "step": 554
    },
    {
      "epoch": 0.30316479464681984,
      "grad_norm": 0.00214815023355186,
      "learning_rate": 0.00014093032540812348,
      "loss": 11.5,
      "step": 555
    },
    {
      "epoch": 0.30371103752005735,
      "grad_norm": 0.002626813715323806,
      "learning_rate": 0.0001407378597982293,
      "loss": 11.5,
      "step": 556
    },
    {
      "epoch": 0.30425728039329486,
      "grad_norm": 0.0015932084061205387,
      "learning_rate": 0.00014054521308549673,
      "loss": 11.5,
      "step": 557
    },
    {
      "epoch": 0.3048035232665324,
      "grad_norm": 0.002143952064216137,
      "learning_rate": 0.0001403523861263495,
      "loss": 11.5,
      "step": 558
    },
    {
      "epoch": 0.3053497661397699,
      "grad_norm": 0.003965519834309816,
      "learning_rate": 0.00014015937977801256,
      "loss": 11.5,
      "step": 559
    },
    {
      "epoch": 0.3058960090130074,
      "grad_norm": 0.0026599925477057695,
      "learning_rate": 0.00013996619489850822,
      "loss": 11.5,
      "step": 560
    },
    {
      "epoch": 0.3064422518862449,
      "grad_norm": 0.003134147496894002,
      "learning_rate": 0.00013977283234665273,
      "loss": 11.5,
      "step": 561
    },
    {
      "epoch": 0.3069884947594824,
      "grad_norm": 0.0036417001392692327,
      "learning_rate": 0.00013957929298205195,
      "loss": 11.5,
      "step": 562
    },
    {
      "epoch": 0.30753473763271993,
      "grad_norm": 0.0020678199362009764,
      "learning_rate": 0.00013938557766509792,
      "loss": 11.5,
      "step": 563
    },
    {
      "epoch": 0.30808098050595745,
      "grad_norm": 0.0024986094795167446,
      "learning_rate": 0.0001391916872569648,
      "loss": 11.5,
      "step": 564
    },
    {
      "epoch": 0.30862722337919496,
      "grad_norm": 0.004739161115139723,
      "learning_rate": 0.00013899762261960518,
      "loss": 11.5,
      "step": 565
    },
    {
      "epoch": 0.30917346625243247,
      "grad_norm": 0.002017226070165634,
      "learning_rate": 0.0001388033846157462,
      "loss": 11.5,
      "step": 566
    },
    {
      "epoch": 0.30971970912567,
      "grad_norm": 0.0017808328848332167,
      "learning_rate": 0.0001386089741088857,
      "loss": 11.5,
      "step": 567
    },
    {
      "epoch": 0.3102659519989075,
      "grad_norm": 0.0038176323287189007,
      "learning_rate": 0.00013841439196328836,
      "loss": 11.5,
      "step": 568
    },
    {
      "epoch": 0.310812194872145,
      "grad_norm": 0.0009957607835531235,
      "learning_rate": 0.00013821963904398193,
      "loss": 11.5,
      "step": 569
    },
    {
      "epoch": 0.3113584377453825,
      "grad_norm": 0.0035520957317203283,
      "learning_rate": 0.00013802471621675338,
      "loss": 11.5,
      "step": 570
    },
    {
      "epoch": 0.31190468061862003,
      "grad_norm": 0.0024566673673689365,
      "learning_rate": 0.00013782962434814492,
      "loss": 11.5,
      "step": 571
    },
    {
      "epoch": 0.31245092349185755,
      "grad_norm": 0.004485953599214554,
      "learning_rate": 0.00013763436430545034,
      "loss": 11.5,
      "step": 572
    },
    {
      "epoch": 0.31299716636509506,
      "grad_norm": 0.002457631519064307,
      "learning_rate": 0.00013743893695671096,
      "loss": 11.5,
      "step": 573
    },
    {
      "epoch": 0.31354340923833257,
      "grad_norm": 0.003239212790504098,
      "learning_rate": 0.00013724334317071198,
      "loss": 11.5,
      "step": 574
    },
    {
      "epoch": 0.3140896521115701,
      "grad_norm": 0.002238941378891468,
      "learning_rate": 0.00013704758381697844,
      "loss": 11.5,
      "step": 575
    },
    {
      "epoch": 0.3146358949848076,
      "grad_norm": 0.0036554813850671053,
      "learning_rate": 0.00013685165976577146,
      "loss": 11.5,
      "step": 576
    },
    {
      "epoch": 0.3151821378580451,
      "grad_norm": 0.003536693751811981,
      "learning_rate": 0.0001366555718880843,
      "loss": 11.5,
      "step": 577
    },
    {
      "epoch": 0.3157283807312826,
      "grad_norm": 0.0024959612637758255,
      "learning_rate": 0.00013645932105563844,
      "loss": 11.5,
      "step": 578
    },
    {
      "epoch": 0.31627462360452013,
      "grad_norm": 0.004410578403621912,
      "learning_rate": 0.00013626290814088005,
      "loss": 11.5,
      "step": 579
    },
    {
      "epoch": 0.31682086647775765,
      "grad_norm": 0.003481909865513444,
      "learning_rate": 0.00013606633401697557,
      "loss": 11.5,
      "step": 580
    },
    {
      "epoch": 0.3173671093509952,
      "grad_norm": 0.0011030982714146376,
      "learning_rate": 0.00013586959955780824,
      "loss": 11.5,
      "step": 581
    },
    {
      "epoch": 0.3179133522242327,
      "grad_norm": 0.0021160771138966084,
      "learning_rate": 0.00013567270563797398,
      "loss": 11.5,
      "step": 582
    },
    {
      "epoch": 0.31845959509747024,
      "grad_norm": 0.0029451404698193073,
      "learning_rate": 0.00013547565313277776,
      "loss": 11.5,
      "step": 583
    },
    {
      "epoch": 0.31900583797070775,
      "grad_norm": 0.0013868617825210094,
      "learning_rate": 0.00013527844291822948,
      "loss": 11.5,
      "step": 584
    },
    {
      "epoch": 0.31955208084394526,
      "grad_norm": 0.003411195008084178,
      "learning_rate": 0.0001350810758710401,
      "loss": 11.5,
      "step": 585
    },
    {
      "epoch": 0.3200983237171828,
      "grad_norm": 0.0016142602544277906,
      "learning_rate": 0.00013488355286861783,
      "loss": 11.5,
      "step": 586
    },
    {
      "epoch": 0.3206445665904203,
      "grad_norm": 0.0031888813246041536,
      "learning_rate": 0.0001346858747890642,
      "loss": 11.5,
      "step": 587
    },
    {
      "epoch": 0.3211908094636578,
      "grad_norm": 0.002269251737743616,
      "learning_rate": 0.00013448804251117003,
      "loss": 11.5,
      "step": 588
    },
    {
      "epoch": 0.3217370523368953,
      "grad_norm": 0.003011634573340416,
      "learning_rate": 0.0001342900569144119,
      "loss": 11.5,
      "step": 589
    },
    {
      "epoch": 0.3222832952101328,
      "grad_norm": 0.0031441410537809134,
      "learning_rate": 0.0001340919188789477,
      "loss": 11.5,
      "step": 590
    },
    {
      "epoch": 0.32282953808337034,
      "grad_norm": 0.001507241278886795,
      "learning_rate": 0.00013389362928561317,
      "loss": 11.5,
      "step": 591
    },
    {
      "epoch": 0.32337578095660785,
      "grad_norm": 0.0034554002340883017,
      "learning_rate": 0.00013369518901591772,
      "loss": 11.5,
      "step": 592
    },
    {
      "epoch": 0.32392202382984536,
      "grad_norm": 0.0020458416547626257,
      "learning_rate": 0.00013349659895204067,
      "loss": 11.5,
      "step": 593
    },
    {
      "epoch": 0.3244682667030829,
      "grad_norm": 0.005565323401242495,
      "learning_rate": 0.0001332978599768272,
      "loss": 11.5,
      "step": 594
    },
    {
      "epoch": 0.3250145095763204,
      "grad_norm": 0.004285743925720453,
      "learning_rate": 0.00013309897297378455,
      "loss": 11.5,
      "step": 595
    },
    {
      "epoch": 0.3255607524495579,
      "grad_norm": 0.002867820905521512,
      "learning_rate": 0.00013289993882707797,
      "loss": 11.5,
      "step": 596
    },
    {
      "epoch": 0.3261069953227954,
      "grad_norm": 0.00545547204092145,
      "learning_rate": 0.00013270075842152678,
      "loss": 11.5,
      "step": 597
    },
    {
      "epoch": 0.3266532381960329,
      "grad_norm": 0.0020265954080969095,
      "learning_rate": 0.00013250143264260074,
      "loss": 11.5,
      "step": 598
    },
    {
      "epoch": 0.32719948106927044,
      "grad_norm": 0.003085668431594968,
      "learning_rate": 0.0001323019623764156,
      "loss": 11.5,
      "step": 599
    },
    {
      "epoch": 0.32774572394250795,
      "grad_norm": 0.009509916417300701,
      "learning_rate": 0.00013210234850972964,
      "loss": 11.5,
      "step": 600
    },
    {
      "epoch": 0.32829196681574546,
      "grad_norm": 0.00851928535848856,
      "learning_rate": 0.0001319025919299394,
      "loss": 11.5,
      "step": 601
    },
    {
      "epoch": 0.328838209688983,
      "grad_norm": 0.002103782957419753,
      "learning_rate": 0.00013170269352507597,
      "loss": 11.5,
      "step": 602
    },
    {
      "epoch": 0.3293844525622205,
      "grad_norm": 0.002564853522926569,
      "learning_rate": 0.0001315026541838008,
      "loss": 11.5,
      "step": 603
    },
    {
      "epoch": 0.329930695435458,
      "grad_norm": 0.004507384262979031,
      "learning_rate": 0.00013130247479540202,
      "loss": 11.5,
      "step": 604
    },
    {
      "epoch": 0.3304769383086955,
      "grad_norm": 0.0014069717144593596,
      "learning_rate": 0.00013110215624979025,
      "loss": 11.5,
      "step": 605
    },
    {
      "epoch": 0.331023181181933,
      "grad_norm": 0.00138050876557827,
      "learning_rate": 0.00013090169943749476,
      "loss": 11.5,
      "step": 606
    },
    {
      "epoch": 0.33156942405517054,
      "grad_norm": 0.00118165067397058,
      "learning_rate": 0.00013070110524965954,
      "loss": 11.5,
      "step": 607
    },
    {
      "epoch": 0.33211566692840805,
      "grad_norm": 0.0023713954724371433,
      "learning_rate": 0.00013050037457803924,
      "loss": 11.5,
      "step": 608
    },
    {
      "epoch": 0.33266190980164556,
      "grad_norm": 0.002364475978538394,
      "learning_rate": 0.0001302995083149953,
      "loss": 11.5,
      "step": 609
    },
    {
      "epoch": 0.3332081526748831,
      "grad_norm": 0.0015375670045614243,
      "learning_rate": 0.0001300985073534919,
      "loss": 11.5,
      "step": 610
    },
    {
      "epoch": 0.3337543955481206,
      "grad_norm": 0.006613594479858875,
      "learning_rate": 0.00012989737258709203,
      "loss": 11.5,
      "step": 611
    },
    {
      "epoch": 0.3343006384213581,
      "grad_norm": 0.006146575789898634,
      "learning_rate": 0.00012969610490995358,
      "loss": 11.5,
      "step": 612
    },
    {
      "epoch": 0.3348468812945956,
      "grad_norm": 0.0019026859663426876,
      "learning_rate": 0.00012949470521682528,
      "loss": 11.5,
      "step": 613
    },
    {
      "epoch": 0.3353931241678331,
      "grad_norm": 0.006842575501650572,
      "learning_rate": 0.0001292931744030427,
      "loss": 11.5,
      "step": 614
    },
    {
      "epoch": 0.33593936704107064,
      "grad_norm": 0.0016394400736317039,
      "learning_rate": 0.0001290915133645243,
      "loss": 11.5,
      "step": 615
    },
    {
      "epoch": 0.33648560991430815,
      "grad_norm": 0.0014736796729266644,
      "learning_rate": 0.00012888972299776754,
      "loss": 11.5,
      "step": 616
    },
    {
      "epoch": 0.33703185278754566,
      "grad_norm": 0.0035685778129845858,
      "learning_rate": 0.00012868780419984482,
      "loss": 11.5,
      "step": 617
    },
    {
      "epoch": 0.3375780956607832,
      "grad_norm": 0.0040477244183421135,
      "learning_rate": 0.00012848575786839943,
      "loss": 11.5,
      "step": 618
    },
    {
      "epoch": 0.3381243385340207,
      "grad_norm": 0.003374082036316395,
      "learning_rate": 0.0001282835849016416,
      "loss": 11.5,
      "step": 619
    },
    {
      "epoch": 0.3386705814072582,
      "grad_norm": 0.00518795195966959,
      "learning_rate": 0.00012808128619834461,
      "loss": 11.5,
      "step": 620
    },
    {
      "epoch": 0.3392168242804957,
      "grad_norm": 0.0020359072368592024,
      "learning_rate": 0.0001278788626578407,
      "loss": 11.5,
      "step": 621
    },
    {
      "epoch": 0.3397630671537332,
      "grad_norm": 0.0021134570706635714,
      "learning_rate": 0.00012767631518001698,
      "loss": 11.5,
      "step": 622
    },
    {
      "epoch": 0.34030931002697073,
      "grad_norm": 0.0029832003638148308,
      "learning_rate": 0.00012747364466531163,
      "loss": 11.5,
      "step": 623
    },
    {
      "epoch": 0.34085555290020825,
      "grad_norm": 0.0018310737796127796,
      "learning_rate": 0.00012727085201470973,
      "loss": 11.5,
      "step": 624
    },
    {
      "epoch": 0.34140179577344576,
      "grad_norm": 0.0025357643608003855,
      "learning_rate": 0.00012706793812973941,
      "loss": 11.5,
      "step": 625
    },
    {
      "epoch": 0.34194803864668327,
      "grad_norm": 0.010150302201509476,
      "learning_rate": 0.0001268649039124677,
      "loss": 11.5,
      "step": 626
    },
    {
      "epoch": 0.3424942815199208,
      "grad_norm": 0.002274255035445094,
      "learning_rate": 0.00012666175026549662,
      "loss": 11.5,
      "step": 627
    },
    {
      "epoch": 0.3430405243931583,
      "grad_norm": 0.002047842601314187,
      "learning_rate": 0.000126458478091959,
      "loss": 11.5,
      "step": 628
    },
    {
      "epoch": 0.3435867672663958,
      "grad_norm": 0.002137575764209032,
      "learning_rate": 0.00012625508829551473,
      "loss": 11.5,
      "step": 629
    },
    {
      "epoch": 0.3441330101396333,
      "grad_norm": 0.0017129582120105624,
      "learning_rate": 0.00012605158178034654,
      "loss": 11.5,
      "step": 630
    },
    {
      "epoch": 0.34467925301287083,
      "grad_norm": 0.0020056625362485647,
      "learning_rate": 0.00012584795945115603,
      "loss": 11.5,
      "step": 631
    },
    {
      "epoch": 0.34522549588610835,
      "grad_norm": 0.0026283313054591417,
      "learning_rate": 0.0001256442222131597,
      "loss": 11.5,
      "step": 632
    },
    {
      "epoch": 0.34577173875934586,
      "grad_norm": 0.0035744968336075544,
      "learning_rate": 0.0001254403709720848,
      "loss": 11.5,
      "step": 633
    },
    {
      "epoch": 0.34631798163258337,
      "grad_norm": 0.003465071087703109,
      "learning_rate": 0.0001252364066341655,
      "loss": 11.5,
      "step": 634
    },
    {
      "epoch": 0.3468642245058209,
      "grad_norm": 0.0027587225195020437,
      "learning_rate": 0.00012503233010613865,
      "loss": 11.5,
      "step": 635
    },
    {
      "epoch": 0.3474104673790584,
      "grad_norm": 0.0029067157302051783,
      "learning_rate": 0.00012482814229523997,
      "loss": 11.5,
      "step": 636
    },
    {
      "epoch": 0.3479567102522959,
      "grad_norm": 0.006217729765921831,
      "learning_rate": 0.00012462384410919975,
      "loss": 11.5,
      "step": 637
    },
    {
      "epoch": 0.3485029531255334,
      "grad_norm": 0.00708062294870615,
      "learning_rate": 0.00012441943645623903,
      "loss": 11.5,
      "step": 638
    },
    {
      "epoch": 0.34904919599877093,
      "grad_norm": 0.0032513788901269436,
      "learning_rate": 0.00012421492024506555,
      "loss": 11.5,
      "step": 639
    },
    {
      "epoch": 0.34959543887200845,
      "grad_norm": 0.0035173557698726654,
      "learning_rate": 0.00012401029638486953,
      "loss": 11.5,
      "step": 640
    },
    {
      "epoch": 0.35014168174524596,
      "grad_norm": 0.002301457803696394,
      "learning_rate": 0.0001238055657853198,
      "loss": 11.5,
      "step": 641
    },
    {
      "epoch": 0.35068792461848347,
      "grad_norm": 0.0031268279999494553,
      "learning_rate": 0.00012360072935655982,
      "loss": 11.5,
      "step": 642
    },
    {
      "epoch": 0.351234167491721,
      "grad_norm": 0.004037101753056049,
      "learning_rate": 0.00012339578800920332,
      "loss": 11.5,
      "step": 643
    },
    {
      "epoch": 0.3517804103649585,
      "grad_norm": 0.002754442160949111,
      "learning_rate": 0.00012319074265433063,
      "loss": 11.5,
      "step": 644
    },
    {
      "epoch": 0.352326653238196,
      "grad_norm": 0.005021520424634218,
      "learning_rate": 0.00012298559420348437,
      "loss": 11.5,
      "step": 645
    },
    {
      "epoch": 0.3528728961114335,
      "grad_norm": 0.0016287361504510045,
      "learning_rate": 0.00012278034356866545,
      "loss": 11.5,
      "step": 646
    },
    {
      "epoch": 0.35341913898467103,
      "grad_norm": 0.004090526141226292,
      "learning_rate": 0.00012257499166232907,
      "loss": 11.5,
      "step": 647
    },
    {
      "epoch": 0.35396538185790855,
      "grad_norm": 0.0028432963881641626,
      "learning_rate": 0.0001223695393973807,
      "loss": 11.5,
      "step": 648
    },
    {
      "epoch": 0.3545116247311461,
      "grad_norm": 0.00406615249812603,
      "learning_rate": 0.0001221639876871719,
      "loss": 11.5,
      "step": 649
    },
    {
      "epoch": 0.3550578676043836,
      "grad_norm": 0.006397001910954714,
      "learning_rate": 0.0001219583374454963,
      "loss": 11.5,
      "step": 650
    },
    {
      "epoch": 0.35560411047762114,
      "grad_norm": 0.0029726503416895866,
      "learning_rate": 0.00012175258958658564,
      "loss": 11.5,
      "step": 651
    },
    {
      "epoch": 0.35615035335085865,
      "grad_norm": 0.002462679985910654,
      "learning_rate": 0.00012154674502510555,
      "loss": 11.5,
      "step": 652
    },
    {
      "epoch": 0.35669659622409616,
      "grad_norm": 0.001187213696539402,
      "learning_rate": 0.00012134080467615159,
      "loss": 11.5,
      "step": 653
    },
    {
      "epoch": 0.3572428390973337,
      "grad_norm": 0.0044397213496267796,
      "learning_rate": 0.00012113476945524513,
      "loss": 11.5,
      "step": 654
    },
    {
      "epoch": 0.3577890819705712,
      "grad_norm": 0.002027162816375494,
      "learning_rate": 0.00012092864027832933,
      "loss": 11.5,
      "step": 655
    },
    {
      "epoch": 0.3583353248438087,
      "grad_norm": 0.0020707782823592424,
      "learning_rate": 0.000120722418061765,
      "loss": 11.5,
      "step": 656
    },
    {
      "epoch": 0.3588815677170462,
      "grad_norm": 0.00276152859441936,
      "learning_rate": 0.0001205161037223266,
      "loss": 11.5,
      "step": 657
    },
    {
      "epoch": 0.3594278105902837,
      "grad_norm": 0.0026741467881947756,
      "learning_rate": 0.00012030969817719808,
      "loss": 11.5,
      "step": 658
    },
    {
      "epoch": 0.35997405346352124,
      "grad_norm": 0.0012922851601615548,
      "learning_rate": 0.00012010320234396894,
      "loss": 11.5,
      "step": 659
    },
    {
      "epoch": 0.36052029633675875,
      "grad_norm": 0.0010917502222582698,
      "learning_rate": 0.00011989661714062999,
      "loss": 11.5,
      "step": 660
    },
    {
      "epoch": 0.36106653920999626,
      "grad_norm": 0.002443787408992648,
      "learning_rate": 0.0001196899434855693,
      "loss": 11.5,
      "step": 661
    },
    {
      "epoch": 0.3616127820832338,
      "grad_norm": 0.0027405829168856144,
      "learning_rate": 0.00011948318229756827,
      "loss": 11.5,
      "step": 662
    },
    {
      "epoch": 0.3621590249564713,
      "grad_norm": 0.0017419673968106508,
      "learning_rate": 0.00011927633449579735,
      "loss": 11.5,
      "step": 663
    },
    {
      "epoch": 0.3627052678297088,
      "grad_norm": 0.004113049246370792,
      "learning_rate": 0.0001190694009998121,
      "loss": 11.5,
      "step": 664
    },
    {
      "epoch": 0.3632515107029463,
      "grad_norm": 0.00292083784006536,
      "learning_rate": 0.00011886238272954897,
      "loss": 11.5,
      "step": 665
    },
    {
      "epoch": 0.3637977535761838,
      "grad_norm": 0.0036332150921225548,
      "learning_rate": 0.00011865528060532127,
      "loss": 11.5,
      "step": 666
    },
    {
      "epoch": 0.36434399644942134,
      "grad_norm": 0.0018869400955736637,
      "learning_rate": 0.0001184480955478152,
      "loss": 11.5,
      "step": 667
    },
    {
      "epoch": 0.36489023932265885,
      "grad_norm": 0.0028332872316241264,
      "learning_rate": 0.00011824082847808558,
      "loss": 11.5,
      "step": 668
    },
    {
      "epoch": 0.36543648219589636,
      "grad_norm": 0.0015665607061237097,
      "learning_rate": 0.00011803348031755179,
      "loss": 11.5,
      "step": 669
    },
    {
      "epoch": 0.3659827250691339,
      "grad_norm": 0.0010158073855563998,
      "learning_rate": 0.0001178260519879937,
      "loss": 11.5,
      "step": 670
    },
    {
      "epoch": 0.3665289679423714,
      "grad_norm": 0.0029538969974964857,
      "learning_rate": 0.00011761854441154767,
      "loss": 11.5,
      "step": 671
    },
    {
      "epoch": 0.3670752108156089,
      "grad_norm": 0.005724848248064518,
      "learning_rate": 0.00011741095851070228,
      "loss": 11.5,
      "step": 672
    },
    {
      "epoch": 0.3676214536888464,
      "grad_norm": 0.004290254786610603,
      "learning_rate": 0.00011720329520829429,
      "loss": 11.5,
      "step": 673
    },
    {
      "epoch": 0.3681676965620839,
      "grad_norm": 0.00148143304977566,
      "learning_rate": 0.0001169955554275046,
      "loss": 11.5,
      "step": 674
    },
    {
      "epoch": 0.36871393943532144,
      "grad_norm": 0.004428256768733263,
      "learning_rate": 0.0001167877400918541,
      "loss": 11.5,
      "step": 675
    },
    {
      "epoch": 0.36926018230855895,
      "grad_norm": 0.0018742958782240748,
      "learning_rate": 0.00011657985012519952,
      "loss": 11.5,
      "step": 676
    },
    {
      "epoch": 0.36980642518179646,
      "grad_norm": 0.0029305454809218645,
      "learning_rate": 0.00011637188645172944,
      "loss": 11.5,
      "step": 677
    },
    {
      "epoch": 0.370352668055034,
      "grad_norm": 0.010208102874457836,
      "learning_rate": 0.00011616384999596006,
      "loss": 11.5,
      "step": 678
    },
    {
      "epoch": 0.3708989109282715,
      "grad_norm": 0.0019220743561163545,
      "learning_rate": 0.00011595574168273111,
      "loss": 11.5,
      "step": 679
    },
    {
      "epoch": 0.371445153801509,
      "grad_norm": 0.002812942024320364,
      "learning_rate": 0.0001157475624372018,
      "loss": 11.5,
      "step": 680
    },
    {
      "epoch": 0.3719913966747465,
      "grad_norm": 0.0021047613117843866,
      "learning_rate": 0.0001155393131848467,
      "loss": 11.5,
      "step": 681
    },
    {
      "epoch": 0.372537639547984,
      "grad_norm": 0.003757823957130313,
      "learning_rate": 0.00011533099485145155,
      "loss": 11.5,
      "step": 682
    },
    {
      "epoch": 0.37308388242122154,
      "grad_norm": 0.0015496105188503861,
      "learning_rate": 0.00011512260836310924,
      "loss": 11.5,
      "step": 683
    },
    {
      "epoch": 0.37363012529445905,
      "grad_norm": 0.0028103413060307503,
      "learning_rate": 0.00011491415464621562,
      "loss": 11.5,
      "step": 684
    },
    {
      "epoch": 0.37417636816769656,
      "grad_norm": 0.0022213319316506386,
      "learning_rate": 0.00011470563462746541,
      "loss": 11.5,
      "step": 685
    },
    {
      "epoch": 0.3747226110409341,
      "grad_norm": 0.0024270599242299795,
      "learning_rate": 0.00011449704923384812,
      "loss": 11.5,
      "step": 686
    },
    {
      "epoch": 0.3752688539141716,
      "grad_norm": 0.0023061565589159727,
      "learning_rate": 0.00011428839939264382,
      "loss": 11.5,
      "step": 687
    },
    {
      "epoch": 0.3758150967874091,
      "grad_norm": 0.004723786376416683,
      "learning_rate": 0.0001140796860314191,
      "loss": 11.5,
      "step": 688
    },
    {
      "epoch": 0.3763613396606466,
      "grad_norm": 0.003816710552200675,
      "learning_rate": 0.00011387091007802297,
      "loss": 11.5,
      "step": 689
    },
    {
      "epoch": 0.3769075825338841,
      "grad_norm": 0.0021447124890983105,
      "learning_rate": 0.0001136620724605827,
      "loss": 11.5,
      "step": 690
    },
    {
      "epoch": 0.37745382540712163,
      "grad_norm": 0.0023786923848092556,
      "learning_rate": 0.00011345317410749964,
      "loss": 11.5,
      "step": 691
    },
    {
      "epoch": 0.37800006828035915,
      "grad_norm": 0.0026864660903811455,
      "learning_rate": 0.00011324421594744516,
      "loss": 11.5,
      "step": 692
    },
    {
      "epoch": 0.37854631115359666,
      "grad_norm": 0.0022893056739121675,
      "learning_rate": 0.00011303519890935656,
      "loss": 11.5,
      "step": 693
    },
    {
      "epoch": 0.37909255402683417,
      "grad_norm": 0.0033118838910013437,
      "learning_rate": 0.00011282612392243286,
      "loss": 11.5,
      "step": 694
    },
    {
      "epoch": 0.3796387969000717,
      "grad_norm": 0.003728573676198721,
      "learning_rate": 0.00011261699191613066,
      "loss": 11.5,
      "step": 695
    },
    {
      "epoch": 0.3801850397733092,
      "grad_norm": 0.0051893084309995174,
      "learning_rate": 0.00011240780382016005,
      "loss": 11.5,
      "step": 696
    },
    {
      "epoch": 0.3807312826465467,
      "grad_norm": 0.0030115193221718073,
      "learning_rate": 0.00011219856056448051,
      "loss": 11.5,
      "step": 697
    },
    {
      "epoch": 0.3812775255197842,
      "grad_norm": 0.0027597760781645775,
      "learning_rate": 0.00011198926307929664,
      "loss": 11.5,
      "step": 698
    },
    {
      "epoch": 0.38182376839302173,
      "grad_norm": 0.0021454724483191967,
      "learning_rate": 0.00011177991229505431,
      "loss": 11.5,
      "step": 699
    },
    {
      "epoch": 0.38237001126625925,
      "grad_norm": 0.004616680555045605,
      "learning_rate": 0.00011157050914243614,
      "loss": 11.5,
      "step": 700
    },
    {
      "epoch": 0.38291625413949676,
      "grad_norm": 0.0030226109083741903,
      "learning_rate": 0.00011136105455235766,
      "loss": 11.5,
      "step": 701
    },
    {
      "epoch": 0.38346249701273427,
      "grad_norm": 0.001542450743727386,
      "learning_rate": 0.00011115154945596305,
      "loss": 11.5,
      "step": 702
    },
    {
      "epoch": 0.3840087398859718,
      "grad_norm": 0.003065952565521002,
      "learning_rate": 0.00011094199478462095,
      "loss": 11.5,
      "step": 703
    },
    {
      "epoch": 0.3845549827592093,
      "grad_norm": 0.002049930626526475,
      "learning_rate": 0.00011073239146992054,
      "loss": 11.5,
      "step": 704
    },
    {
      "epoch": 0.3851012256324468,
      "grad_norm": 0.0017114533111453056,
      "learning_rate": 0.00011052274044366711,
      "loss": 11.5,
      "step": 705
    },
    {
      "epoch": 0.3856474685056843,
      "grad_norm": 0.001665680087171495,
      "learning_rate": 0.00011031304263787812,
      "loss": 11.5,
      "step": 706
    },
    {
      "epoch": 0.38619371137892183,
      "grad_norm": 0.0026139637921005487,
      "learning_rate": 0.00011010329898477891,
      "loss": 11.5,
      "step": 707
    },
    {
      "epoch": 0.38673995425215935,
      "grad_norm": 0.0017987636383622885,
      "learning_rate": 0.0001098935104167988,
      "loss": 11.5,
      "step": 708
    },
    {
      "epoch": 0.38728619712539686,
      "grad_norm": 0.003949366044253111,
      "learning_rate": 0.00010968367786656663,
      "loss": 11.5,
      "step": 709
    },
    {
      "epoch": 0.38783243999863437,
      "grad_norm": 0.0018144213827326894,
      "learning_rate": 0.00010947380226690684,
      "loss": 11.5,
      "step": 710
    },
    {
      "epoch": 0.3883786828718719,
      "grad_norm": 0.003763078711926937,
      "learning_rate": 0.00010926388455083522,
      "loss": 11.5,
      "step": 711
    },
    {
      "epoch": 0.3889249257451094,
      "grad_norm": 0.001604754594154656,
      "learning_rate": 0.00010905392565155477,
      "loss": 11.5,
      "step": 712
    },
    {
      "epoch": 0.3894711686183469,
      "grad_norm": 0.002445596968755126,
      "learning_rate": 0.00010884392650245165,
      "loss": 11.5,
      "step": 713
    },
    {
      "epoch": 0.3900174114915844,
      "grad_norm": 0.001502310042269528,
      "learning_rate": 0.00010863388803709089,
      "loss": 11.5,
      "step": 714
    },
    {
      "epoch": 0.39056365436482193,
      "grad_norm": 0.0023849355056881905,
      "learning_rate": 0.00010842381118921232,
      "loss": 11.5,
      "step": 715
    },
    {
      "epoch": 0.39110989723805945,
      "grad_norm": 0.002108894055709243,
      "learning_rate": 0.00010821369689272638,
      "loss": 11.5,
      "step": 716
    },
    {
      "epoch": 0.391656140111297,
      "grad_norm": 0.0023284833878278732,
      "learning_rate": 0.00010800354608171003,
      "loss": 11.5,
      "step": 717
    },
    {
      "epoch": 0.3922023829845345,
      "grad_norm": 0.0021306509152054787,
      "learning_rate": 0.00010779335969040252,
      "loss": 11.5,
      "step": 718
    },
    {
      "epoch": 0.39274862585777204,
      "grad_norm": 0.0015543142799288034,
      "learning_rate": 0.00010758313865320134,
      "loss": 11.5,
      "step": 719
    },
    {
      "epoch": 0.39329486873100955,
      "grad_norm": 0.0019943786319345236,
      "learning_rate": 0.00010737288390465792,
      "loss": 11.5,
      "step": 720
    },
    {
      "epoch": 0.39384111160424706,
      "grad_norm": 0.0019235857762396336,
      "learning_rate": 0.00010716259637947357,
      "loss": 11.5,
      "step": 721
    },
    {
      "epoch": 0.3943873544774846,
      "grad_norm": 0.0029933659825474024,
      "learning_rate": 0.00010695227701249537,
      "loss": 11.5,
      "step": 722
    },
    {
      "epoch": 0.3949335973507221,
      "grad_norm": 0.0025009207893162966,
      "learning_rate": 0.00010674192673871191,
      "loss": 11.5,
      "step": 723
    },
    {
      "epoch": 0.3954798402239596,
      "grad_norm": 0.0019206060096621513,
      "learning_rate": 0.00010653154649324917,
      "loss": 11.5,
      "step": 724
    },
    {
      "epoch": 0.3960260830971971,
      "grad_norm": 0.0011178608983755112,
      "learning_rate": 0.00010632113721136636,
      "loss": 11.5,
      "step": 725
    },
    {
      "epoch": 0.3965723259704346,
      "grad_norm": 0.01080204639583826,
      "learning_rate": 0.00010611069982845183,
      "loss": 11.5,
      "step": 726
    },
    {
      "epoch": 0.39711856884367214,
      "grad_norm": 0.0017544220900163054,
      "learning_rate": 0.00010590023528001884,
      "loss": 11.5,
      "step": 727
    },
    {
      "epoch": 0.39766481171690965,
      "grad_norm": 0.0017221475718542933,
      "learning_rate": 0.00010568974450170139,
      "loss": 11.5,
      "step": 728
    },
    {
      "epoch": 0.39821105459014716,
      "grad_norm": 0.0040596588514745235,
      "learning_rate": 0.00010547922842925008,
      "loss": 11.5,
      "step": 729
    },
    {
      "epoch": 0.3987572974633847,
      "grad_norm": 0.001911718281917274,
      "learning_rate": 0.00010526868799852796,
      "loss": 11.5,
      "step": 730
    },
    {
      "epoch": 0.3993035403366222,
      "grad_norm": 0.002136989962309599,
      "learning_rate": 0.0001050581241455064,
      "loss": 11.5,
      "step": 731
    },
    {
      "epoch": 0.3998497832098597,
      "grad_norm": 0.004930080845952034,
      "learning_rate": 0.00010484753780626089,
      "loss": 11.5,
      "step": 732
    },
    {
      "epoch": 0.4003960260830972,
      "grad_norm": 0.009830904193222523,
      "learning_rate": 0.00010463692991696685,
      "loss": 11.5,
      "step": 733
    },
    {
      "epoch": 0.4009422689563347,
      "grad_norm": 0.001292129047214985,
      "learning_rate": 0.00010442630141389549,
      "loss": 11.5,
      "step": 734
    },
    {
      "epoch": 0.40148851182957224,
      "grad_norm": 0.001807724591344595,
      "learning_rate": 0.00010421565323340971,
      "loss": 11.5,
      "step": 735
    },
    {
      "epoch": 0.40203475470280975,
      "grad_norm": 0.0027543313335627317,
      "learning_rate": 0.00010400498631195992,
      "loss": 11.5,
      "step": 736
    },
    {
      "epoch": 0.40258099757604726,
      "grad_norm": 0.007954576052725315,
      "learning_rate": 0.00010379430158607975,
      "loss": 11.5,
      "step": 737
    },
    {
      "epoch": 0.4031272404492848,
      "grad_norm": 0.0016194587806239724,
      "learning_rate": 0.000103583599992382,
      "loss": 11.5,
      "step": 738
    },
    {
      "epoch": 0.4036734833225223,
      "grad_norm": 0.003970892168581486,
      "learning_rate": 0.0001033728824675545,
      "loss": 11.5,
      "step": 739
    },
    {
      "epoch": 0.4042197261957598,
      "grad_norm": 0.0018762516556307673,
      "learning_rate": 0.0001031621499483559,
      "loss": 11.5,
      "step": 740
    },
    {
      "epoch": 0.4047659690689973,
      "grad_norm": 0.008109194226562977,
      "learning_rate": 0.00010295140337161146,
      "loss": 11.5,
      "step": 741
    },
    {
      "epoch": 0.4053122119422348,
      "grad_norm": 0.0014864240074530244,
      "learning_rate": 0.00010274064367420897,
      "loss": 11.5,
      "step": 742
    },
    {
      "epoch": 0.40585845481547234,
      "grad_norm": 0.004298364277929068,
      "learning_rate": 0.00010252987179309459,
      "loss": 11.5,
      "step": 743
    },
    {
      "epoch": 0.40640469768870985,
      "grad_norm": 0.0025850574020296335,
      "learning_rate": 0.00010231908866526851,
      "loss": 11.5,
      "step": 744
    },
    {
      "epoch": 0.40695094056194736,
      "grad_norm": 0.0029821840580552816,
      "learning_rate": 0.00010210829522778111,
      "loss": 11.5,
      "step": 745
    },
    {
      "epoch": 0.4074971834351849,
      "grad_norm": 0.0059503004886209965,
      "learning_rate": 0.00010189749241772844,
      "loss": 11.5,
      "step": 746
    },
    {
      "epoch": 0.4080434263084224,
      "grad_norm": 0.005914604291319847,
      "learning_rate": 0.00010168668117224825,
      "loss": 11.5,
      "step": 747
    },
    {
      "epoch": 0.4085896691816599,
      "grad_norm": 0.0038348566740751266,
      "learning_rate": 0.00010147586242851585,
      "loss": 11.5,
      "step": 748
    },
    {
      "epoch": 0.4091359120548974,
      "grad_norm": 0.004920531529933214,
      "learning_rate": 0.00010126503712373982,
      "loss": 11.5,
      "step": 749
    },
    {
      "epoch": 0.4096821549281349,
      "grad_norm": 0.0033745449036359787,
      "learning_rate": 0.00010105420619515798,
      "loss": 11.5,
      "step": 750
    },
    {
      "epoch": 0.4096821549281349,
      "eval_loss": 11.5,
      "eval_runtime": 21.2365,
      "eval_samples_per_second": 145.222,
      "eval_steps_per_second": 72.611,
      "step": 750
    },
    {
      "epoch": 0.41022839780137244,
      "grad_norm": 0.003717338666319847,
      "learning_rate": 0.00010084337058003303,
      "loss": 11.5,
      "step": 751
    },
    {
      "epoch": 0.41077464067460995,
      "grad_norm": 0.0023081412073224783,
      "learning_rate": 0.00010063253121564868,
      "loss": 11.5,
      "step": 752
    },
    {
      "epoch": 0.41132088354784746,
      "grad_norm": 0.003727917093783617,
      "learning_rate": 0.00010042168903930514,
      "loss": 11.5,
      "step": 753
    },
    {
      "epoch": 0.411867126421085,
      "grad_norm": 0.0019546893890947104,
      "learning_rate": 0.00010021084498831522,
      "loss": 11.5,
      "step": 754
    },
    {
      "epoch": 0.4124133692943225,
      "grad_norm": 0.0014350549317896366,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 755
    },
    {
      "epoch": 0.41295961216756,
      "grad_norm": 0.0014874525368213654,
      "learning_rate": 9.97891550116848e-05,
      "loss": 11.5,
      "step": 756
    },
    {
      "epoch": 0.4135058550407975,
      "grad_norm": 0.001366943004541099,
      "learning_rate": 9.957831096069488e-05,
      "loss": 11.5,
      "step": 757
    },
    {
      "epoch": 0.414052097914035,
      "grad_norm": 0.002336501143872738,
      "learning_rate": 9.936746878435136e-05,
      "loss": 11.5,
      "step": 758
    },
    {
      "epoch": 0.41459834078727253,
      "grad_norm": 0.0020807066466659307,
      "learning_rate": 9.915662941996699e-05,
      "loss": 11.5,
      "step": 759
    },
    {
      "epoch": 0.41514458366051005,
      "grad_norm": 0.0016819059383124113,
      "learning_rate": 9.894579380484204e-05,
      "loss": 11.5,
      "step": 760
    },
    {
      "epoch": 0.41569082653374756,
      "grad_norm": 0.0013762516900897026,
      "learning_rate": 9.873496287626019e-05,
      "loss": 11.5,
      "step": 761
    },
    {
      "epoch": 0.41623706940698507,
      "grad_norm": 0.0007033822475932539,
      "learning_rate": 9.852413757148417e-05,
      "loss": 11.5,
      "step": 762
    },
    {
      "epoch": 0.4167833122802226,
      "grad_norm": 0.0022579808719456196,
      "learning_rate": 9.831331882775178e-05,
      "loss": 11.5,
      "step": 763
    },
    {
      "epoch": 0.4173295551534601,
      "grad_norm": 0.0015905239852145314,
      "learning_rate": 9.81025075822716e-05,
      "loss": 11.5,
      "step": 764
    },
    {
      "epoch": 0.4178757980266976,
      "grad_norm": 0.0072204419411718845,
      "learning_rate": 9.789170477221891e-05,
      "loss": 11.5,
      "step": 765
    },
    {
      "epoch": 0.4184220408999351,
      "grad_norm": 0.0017316995654255152,
      "learning_rate": 9.76809113347315e-05,
      "loss": 11.5,
      "step": 766
    },
    {
      "epoch": 0.41896828377317263,
      "grad_norm": 0.0013655085349455476,
      "learning_rate": 9.747012820690543e-05,
      "loss": 11.5,
      "step": 767
    },
    {
      "epoch": 0.41951452664641015,
      "grad_norm": 0.0029249212238937616,
      "learning_rate": 9.725935632579104e-05,
      "loss": 11.5,
      "step": 768
    },
    {
      "epoch": 0.42006076951964766,
      "grad_norm": 0.0023902684915810823,
      "learning_rate": 9.704859662838855e-05,
      "loss": 11.5,
      "step": 769
    },
    {
      "epoch": 0.42060701239288517,
      "grad_norm": 0.0013596831122413278,
      "learning_rate": 9.683785005164411e-05,
      "loss": 11.5,
      "step": 770
    },
    {
      "epoch": 0.4211532552661227,
      "grad_norm": 0.003812977345660329,
      "learning_rate": 9.662711753244551e-05,
      "loss": 11.5,
      "step": 771
    },
    {
      "epoch": 0.4216994981393602,
      "grad_norm": 0.0010871372651308775,
      "learning_rate": 9.641640000761802e-05,
      "loss": 11.5,
      "step": 772
    },
    {
      "epoch": 0.4222457410125977,
      "grad_norm": 0.0023194472305476665,
      "learning_rate": 9.620569841392029e-05,
      "loss": 11.5,
      "step": 773
    },
    {
      "epoch": 0.4227919838858352,
      "grad_norm": 0.0012818895047530532,
      "learning_rate": 9.59950136880401e-05,
      "loss": 11.5,
      "step": 774
    },
    {
      "epoch": 0.42333822675907273,
      "grad_norm": 0.001139130094088614,
      "learning_rate": 9.57843467665903e-05,
      "loss": 11.5,
      "step": 775
    },
    {
      "epoch": 0.42388446963231025,
      "grad_norm": 0.0016325972974300385,
      "learning_rate": 9.557369858610453e-05,
      "loss": 11.5,
      "step": 776
    },
    {
      "epoch": 0.42443071250554776,
      "grad_norm": 0.009854095987975597,
      "learning_rate": 9.53630700830332e-05,
      "loss": 11.5,
      "step": 777
    },
    {
      "epoch": 0.42497695537878527,
      "grad_norm": 0.0010259324917569757,
      "learning_rate": 9.51524621937391e-05,
      "loss": 11.5,
      "step": 778
    },
    {
      "epoch": 0.4255231982520228,
      "grad_norm": 0.002525655785575509,
      "learning_rate": 9.494187585449358e-05,
      "loss": 11.5,
      "step": 779
    },
    {
      "epoch": 0.4260694411252603,
      "grad_norm": 0.003286036429926753,
      "learning_rate": 9.473131200147205e-05,
      "loss": 11.5,
      "step": 780
    },
    {
      "epoch": 0.4266156839984978,
      "grad_norm": 0.0026976955123245716,
      "learning_rate": 9.452077157074994e-05,
      "loss": 11.5,
      "step": 781
    },
    {
      "epoch": 0.4271619268717353,
      "grad_norm": 0.0028888965025544167,
      "learning_rate": 9.431025549829862e-05,
      "loss": 11.5,
      "step": 782
    },
    {
      "epoch": 0.42770816974497283,
      "grad_norm": 0.001401234301738441,
      "learning_rate": 9.409976471998118e-05,
      "loss": 11.5,
      "step": 783
    },
    {
      "epoch": 0.42825441261821034,
      "grad_norm": 0.0016947650583460927,
      "learning_rate": 9.388930017154819e-05,
      "loss": 11.5,
      "step": 784
    },
    {
      "epoch": 0.42880065549144786,
      "grad_norm": 0.003407653421163559,
      "learning_rate": 9.367886278863366e-05,
      "loss": 11.5,
      "step": 785
    },
    {
      "epoch": 0.4293468983646854,
      "grad_norm": 0.002984801074489951,
      "learning_rate": 9.346845350675088e-05,
      "loss": 11.5,
      "step": 786
    },
    {
      "epoch": 0.42989314123792294,
      "grad_norm": 0.0025760724674910307,
      "learning_rate": 9.325807326128814e-05,
      "loss": 11.5,
      "step": 787
    },
    {
      "epoch": 0.43043938411116045,
      "grad_norm": 0.0018182594794780016,
      "learning_rate": 9.304772298750463e-05,
      "loss": 11.5,
      "step": 788
    },
    {
      "epoch": 0.43098562698439796,
      "grad_norm": 0.004073329735547304,
      "learning_rate": 9.283740362052642e-05,
      "loss": 11.5,
      "step": 789
    },
    {
      "epoch": 0.4315318698576355,
      "grad_norm": 0.0013142618117854,
      "learning_rate": 9.26271160953421e-05,
      "loss": 11.5,
      "step": 790
    },
    {
      "epoch": 0.432078112730873,
      "grad_norm": 0.00762759055942297,
      "learning_rate": 9.241686134679867e-05,
      "loss": 11.5,
      "step": 791
    },
    {
      "epoch": 0.4326243556041105,
      "grad_norm": 0.0018035795073956251,
      "learning_rate": 9.220664030959749e-05,
      "loss": 11.5,
      "step": 792
    },
    {
      "epoch": 0.433170598477348,
      "grad_norm": 0.004614766221493483,
      "learning_rate": 9.199645391828999e-05,
      "loss": 11.5,
      "step": 793
    },
    {
      "epoch": 0.4337168413505855,
      "grad_norm": 0.001580020529218018,
      "learning_rate": 9.178630310727365e-05,
      "loss": 11.5,
      "step": 794
    },
    {
      "epoch": 0.43426308422382304,
      "grad_norm": 0.007021275348961353,
      "learning_rate": 9.157618881078772e-05,
      "loss": 11.5,
      "step": 795
    },
    {
      "epoch": 0.43480932709706055,
      "grad_norm": 0.003164935391396284,
      "learning_rate": 9.136611196290915e-05,
      "loss": 11.5,
      "step": 796
    },
    {
      "epoch": 0.43535556997029806,
      "grad_norm": 0.005368545185774565,
      "learning_rate": 9.115607349754834e-05,
      "loss": 11.5,
      "step": 797
    },
    {
      "epoch": 0.4359018128435356,
      "grad_norm": 0.003806391032412648,
      "learning_rate": 9.094607434844523e-05,
      "loss": 11.5,
      "step": 798
    },
    {
      "epoch": 0.4364480557167731,
      "grad_norm": 0.004586106166243553,
      "learning_rate": 9.07361154491648e-05,
      "loss": 11.5,
      "step": 799
    },
    {
      "epoch": 0.4369942985900106,
      "grad_norm": 0.008655227720737457,
      "learning_rate": 9.052619773309317e-05,
      "loss": 11.5,
      "step": 800
    },
    {
      "epoch": 0.4375405414632481,
      "grad_norm": 0.0050348243676126,
      "learning_rate": 9.031632213343339e-05,
      "loss": 11.5,
      "step": 801
    },
    {
      "epoch": 0.4380867843364856,
      "grad_norm": 0.002646834123879671,
      "learning_rate": 9.01064895832012e-05,
      "loss": 11.5,
      "step": 802
    },
    {
      "epoch": 0.43863302720972314,
      "grad_norm": 0.0021842329297214746,
      "learning_rate": 8.98967010152211e-05,
      "loss": 11.5,
      "step": 803
    },
    {
      "epoch": 0.43917927008296065,
      "grad_norm": 0.002736287657171488,
      "learning_rate": 8.968695736212193e-05,
      "loss": 11.5,
      "step": 804
    },
    {
      "epoch": 0.43972551295619816,
      "grad_norm": 0.0029469120781868696,
      "learning_rate": 8.947725955633294e-05,
      "loss": 11.5,
      "step": 805
    },
    {
      "epoch": 0.4402717558294357,
      "grad_norm": 0.0013820442836731672,
      "learning_rate": 8.926760853007946e-05,
      "loss": 11.5,
      "step": 806
    },
    {
      "epoch": 0.4408179987026732,
      "grad_norm": 0.0015859166160225868,
      "learning_rate": 8.905800521537905e-05,
      "loss": 11.5,
      "step": 807
    },
    {
      "epoch": 0.4413642415759107,
      "grad_norm": 0.001276521012187004,
      "learning_rate": 8.884845054403699e-05,
      "loss": 11.5,
      "step": 808
    },
    {
      "epoch": 0.4419104844491482,
      "grad_norm": 0.0020217278506606817,
      "learning_rate": 8.863894544764236e-05,
      "loss": 11.5,
      "step": 809
    },
    {
      "epoch": 0.4424567273223857,
      "grad_norm": 0.002158909337595105,
      "learning_rate": 8.84294908575639e-05,
      "loss": 11.5,
      "step": 810
    },
    {
      "epoch": 0.44300297019562324,
      "grad_norm": 0.0013419922906905413,
      "learning_rate": 8.822008770494572e-05,
      "loss": 11.5,
      "step": 811
    },
    {
      "epoch": 0.44354921306886075,
      "grad_norm": 0.00219023902900517,
      "learning_rate": 8.801073692070337e-05,
      "loss": 11.5,
      "step": 812
    },
    {
      "epoch": 0.44409545594209826,
      "grad_norm": 0.002086648717522621,
      "learning_rate": 8.780143943551954e-05,
      "loss": 11.5,
      "step": 813
    },
    {
      "epoch": 0.4446416988153358,
      "grad_norm": 0.0017264597117900848,
      "learning_rate": 8.759219617983999e-05,
      "loss": 11.5,
      "step": 814
    },
    {
      "epoch": 0.4451879416885733,
      "grad_norm": 0.002369537716731429,
      "learning_rate": 8.738300808386935e-05,
      "loss": 11.5,
      "step": 815
    },
    {
      "epoch": 0.4457341845618108,
      "grad_norm": 0.0037914803251624107,
      "learning_rate": 8.717387607756713e-05,
      "loss": 11.5,
      "step": 816
    },
    {
      "epoch": 0.4462804274350483,
      "grad_norm": 0.004660093225538731,
      "learning_rate": 8.696480109064342e-05,
      "loss": 11.5,
      "step": 817
    },
    {
      "epoch": 0.4468266703082858,
      "grad_norm": 0.001320662209764123,
      "learning_rate": 8.675578405255485e-05,
      "loss": 11.5,
      "step": 818
    },
    {
      "epoch": 0.44737291318152334,
      "grad_norm": 0.0016007827362045646,
      "learning_rate": 8.654682589250038e-05,
      "loss": 11.5,
      "step": 819
    },
    {
      "epoch": 0.44791915605476085,
      "grad_norm": 0.0043623121455311775,
      "learning_rate": 8.633792753941733e-05,
      "loss": 11.5,
      "step": 820
    },
    {
      "epoch": 0.44846539892799836,
      "grad_norm": 0.0022066624369472265,
      "learning_rate": 8.612908992197705e-05,
      "loss": 11.5,
      "step": 821
    },
    {
      "epoch": 0.4490116418012359,
      "grad_norm": 0.0029318437445908785,
      "learning_rate": 8.592031396858093e-05,
      "loss": 11.5,
      "step": 822
    },
    {
      "epoch": 0.4495578846744734,
      "grad_norm": 0.0019769661594182253,
      "learning_rate": 8.571160060735624e-05,
      "loss": 11.5,
      "step": 823
    },
    {
      "epoch": 0.4501041275477109,
      "grad_norm": 0.002241168636828661,
      "learning_rate": 8.550295076615188e-05,
      "loss": 11.5,
      "step": 824
    },
    {
      "epoch": 0.4506503704209484,
      "grad_norm": 0.0015900206053629518,
      "learning_rate": 8.529436537253458e-05,
      "loss": 11.5,
      "step": 825
    },
    {
      "epoch": 0.4511966132941859,
      "grad_norm": 0.004321118351072073,
      "learning_rate": 8.508584535378439e-05,
      "loss": 11.5,
      "step": 826
    },
    {
      "epoch": 0.45174285616742343,
      "grad_norm": 0.004054056480526924,
      "learning_rate": 8.487739163689079e-05,
      "loss": 11.5,
      "step": 827
    },
    {
      "epoch": 0.45228909904066095,
      "grad_norm": 0.01161109283566475,
      "learning_rate": 8.466900514854847e-05,
      "loss": 11.5,
      "step": 828
    },
    {
      "epoch": 0.45283534191389846,
      "grad_norm": 0.0018767928704619408,
      "learning_rate": 8.446068681515334e-05,
      "loss": 11.5,
      "step": 829
    },
    {
      "epoch": 0.45338158478713597,
      "grad_norm": 0.002357539953663945,
      "learning_rate": 8.425243756279824e-05,
      "loss": 11.5,
      "step": 830
    },
    {
      "epoch": 0.4539278276603735,
      "grad_norm": 0.003999626263976097,
      "learning_rate": 8.404425831726894e-05,
      "loss": 11.5,
      "step": 831
    },
    {
      "epoch": 0.454474070533611,
      "grad_norm": 0.00198917044326663,
      "learning_rate": 8.383615000404e-05,
      "loss": 11.5,
      "step": 832
    },
    {
      "epoch": 0.4550203134068485,
      "grad_norm": 0.00371006247587502,
      "learning_rate": 8.362811354827059e-05,
      "loss": 11.5,
      "step": 833
    },
    {
      "epoch": 0.455566556280086,
      "grad_norm": 0.0016412147087976336,
      "learning_rate": 8.342014987480047e-05,
      "loss": 11.5,
      "step": 834
    },
    {
      "epoch": 0.45611279915332353,
      "grad_norm": 0.003618762828409672,
      "learning_rate": 8.321225990814591e-05,
      "loss": 11.5,
      "step": 835
    },
    {
      "epoch": 0.45665904202656105,
      "grad_norm": 0.0014960282715037465,
      "learning_rate": 8.300444457249543e-05,
      "loss": 11.5,
      "step": 836
    },
    {
      "epoch": 0.45720528489979856,
      "grad_norm": 0.0037192620802670717,
      "learning_rate": 8.279670479170573e-05,
      "loss": 11.5,
      "step": 837
    },
    {
      "epoch": 0.45775152777303607,
      "grad_norm": 0.003394707338884473,
      "learning_rate": 8.258904148929775e-05,
      "loss": 11.5,
      "step": 838
    },
    {
      "epoch": 0.4582977706462736,
      "grad_norm": 0.002499793656170368,
      "learning_rate": 8.238145558845235e-05,
      "loss": 11.5,
      "step": 839
    },
    {
      "epoch": 0.4588440135195111,
      "grad_norm": 0.004915180616080761,
      "learning_rate": 8.217394801200631e-05,
      "loss": 11.5,
      "step": 840
    },
    {
      "epoch": 0.4593902563927486,
      "grad_norm": 0.001847822917625308,
      "learning_rate": 8.196651968244826e-05,
      "loss": 11.5,
      "step": 841
    },
    {
      "epoch": 0.4599364992659861,
      "grad_norm": 0.002306227572262287,
      "learning_rate": 8.175917152191447e-05,
      "loss": 11.5,
      "step": 842
    },
    {
      "epoch": 0.46048274213922363,
      "grad_norm": 0.0034235112834721804,
      "learning_rate": 8.15519044521848e-05,
      "loss": 11.5,
      "step": 843
    },
    {
      "epoch": 0.46102898501246115,
      "grad_norm": 0.0020738046150654554,
      "learning_rate": 8.134471939467874e-05,
      "loss": 11.5,
      "step": 844
    },
    {
      "epoch": 0.46157522788569866,
      "grad_norm": 0.002567283809185028,
      "learning_rate": 8.113761727045105e-05,
      "loss": 11.5,
      "step": 845
    },
    {
      "epoch": 0.46212147075893617,
      "grad_norm": 0.0037481554318219423,
      "learning_rate": 8.093059900018792e-05,
      "loss": 11.5,
      "step": 846
    },
    {
      "epoch": 0.4626677136321737,
      "grad_norm": 0.0034958485048264265,
      "learning_rate": 8.072366550420266e-05,
      "loss": 11.5,
      "step": 847
    },
    {
      "epoch": 0.4632139565054112,
      "grad_norm": 0.0022758147679269314,
      "learning_rate": 8.051681770243175e-05,
      "loss": 11.5,
      "step": 848
    },
    {
      "epoch": 0.4637601993786487,
      "grad_norm": 0.0028618639335036278,
      "learning_rate": 8.031005651443073e-05,
      "loss": 11.5,
      "step": 849
    },
    {
      "epoch": 0.4643064422518862,
      "grad_norm": 0.0067398035898804665,
      "learning_rate": 8.010338285937006e-05,
      "loss": 11.5,
      "step": 850
    },
    {
      "epoch": 0.46485268512512373,
      "grad_norm": 0.0016845482168719172,
      "learning_rate": 7.989679765603108e-05,
      "loss": 11.5,
      "step": 851
    },
    {
      "epoch": 0.46539892799836124,
      "grad_norm": 0.00151927606202662,
      "learning_rate": 7.969030182280192e-05,
      "loss": 11.5,
      "step": 852
    },
    {
      "epoch": 0.46594517087159876,
      "grad_norm": 0.006853247992694378,
      "learning_rate": 7.948389627767343e-05,
      "loss": 11.5,
      "step": 853
    },
    {
      "epoch": 0.4664914137448363,
      "grad_norm": 0.0017898344667628407,
      "learning_rate": 7.927758193823501e-05,
      "loss": 11.5,
      "step": 854
    },
    {
      "epoch": 0.46703765661807384,
      "grad_norm": 0.001284423633478582,
      "learning_rate": 7.907135972167069e-05,
      "loss": 11.5,
      "step": 855
    },
    {
      "epoch": 0.46758389949131135,
      "grad_norm": 0.002788104582577944,
      "learning_rate": 7.88652305447549e-05,
      "loss": 11.5,
      "step": 856
    },
    {
      "epoch": 0.46813014236454886,
      "grad_norm": 0.0020991170313209295,
      "learning_rate": 7.865919532384844e-05,
      "loss": 11.5,
      "step": 857
    },
    {
      "epoch": 0.4686763852377864,
      "grad_norm": 0.001541205681860447,
      "learning_rate": 7.845325497489449e-05,
      "loss": 11.5,
      "step": 858
    },
    {
      "epoch": 0.4692226281110239,
      "grad_norm": 0.0010202117264270782,
      "learning_rate": 7.82474104134144e-05,
      "loss": 11.5,
      "step": 859
    },
    {
      "epoch": 0.4697688709842614,
      "grad_norm": 0.0009887984488159418,
      "learning_rate": 7.804166255450373e-05,
      "loss": 11.5,
      "step": 860
    },
    {
      "epoch": 0.4703151138574989,
      "grad_norm": 0.001964594703167677,
      "learning_rate": 7.783601231282812e-05,
      "loss": 11.5,
      "step": 861
    },
    {
      "epoch": 0.4708613567307364,
      "grad_norm": 0.0022888982202857733,
      "learning_rate": 7.763046060261932e-05,
      "loss": 11.5,
      "step": 862
    },
    {
      "epoch": 0.47140759960397394,
      "grad_norm": 0.003015857422724366,
      "learning_rate": 7.742500833767094e-05,
      "loss": 11.5,
      "step": 863
    },
    {
      "epoch": 0.47195384247721145,
      "grad_norm": 0.00468654278665781,
      "learning_rate": 7.721965643133458e-05,
      "loss": 11.5,
      "step": 864
    },
    {
      "epoch": 0.47250008535044896,
      "grad_norm": 0.0031884205527603626,
      "learning_rate": 7.701440579651566e-05,
      "loss": 11.5,
      "step": 865
    },
    {
      "epoch": 0.4730463282236865,
      "grad_norm": 0.0022365872282534838,
      "learning_rate": 7.680925734566937e-05,
      "loss": 11.5,
      "step": 866
    },
    {
      "epoch": 0.473592571096924,
      "grad_norm": 0.0034705346915870905,
      "learning_rate": 7.660421199079669e-05,
      "loss": 11.5,
      "step": 867
    },
    {
      "epoch": 0.4741388139701615,
      "grad_norm": 0.001237342949025333,
      "learning_rate": 7.639927064344022e-05,
      "loss": 11.5,
      "step": 868
    },
    {
      "epoch": 0.474685056843399,
      "grad_norm": 0.0015939086442813277,
      "learning_rate": 7.619443421468021e-05,
      "loss": 11.5,
      "step": 869
    },
    {
      "epoch": 0.4752312997166365,
      "grad_norm": 0.0009356584050692618,
      "learning_rate": 7.598970361513051e-05,
      "loss": 11.5,
      "step": 870
    },
    {
      "epoch": 0.47577754258987404,
      "grad_norm": 0.0035895020700991154,
      "learning_rate": 7.578507975493448e-05,
      "loss": 11.5,
      "step": 871
    },
    {
      "epoch": 0.47632378546311155,
      "grad_norm": 0.0029749858658760786,
      "learning_rate": 7.558056354376098e-05,
      "loss": 11.5,
      "step": 872
    },
    {
      "epoch": 0.47687002833634906,
      "grad_norm": 0.0038305483758449554,
      "learning_rate": 7.537615589080027e-05,
      "loss": 11.5,
      "step": 873
    },
    {
      "epoch": 0.4774162712095866,
      "grad_norm": 0.0013644895516335964,
      "learning_rate": 7.517185770476006e-05,
      "loss": 11.5,
      "step": 874
    },
    {
      "epoch": 0.4779625140828241,
      "grad_norm": 0.002218692097812891,
      "learning_rate": 7.496766989386136e-05,
      "loss": 11.5,
      "step": 875
    },
    {
      "epoch": 0.4785087569560616,
      "grad_norm": 0.002175125991925597,
      "learning_rate": 7.476359336583454e-05,
      "loss": 11.5,
      "step": 876
    },
    {
      "epoch": 0.4790549998292991,
      "grad_norm": 0.0069276695139706135,
      "learning_rate": 7.455962902791522e-05,
      "loss": 11.5,
      "step": 877
    },
    {
      "epoch": 0.4796012427025366,
      "grad_norm": 0.003694717073813081,
      "learning_rate": 7.435577778684033e-05,
      "loss": 11.5,
      "step": 878
    },
    {
      "epoch": 0.48014748557577414,
      "grad_norm": 0.003659107256680727,
      "learning_rate": 7.415204054884399e-05,
      "loss": 11.5,
      "step": 879
    },
    {
      "epoch": 0.48069372844901165,
      "grad_norm": 0.009692396968603134,
      "learning_rate": 7.394841821965345e-05,
      "loss": 11.5,
      "step": 880
    },
    {
      "epoch": 0.48123997132224916,
      "grad_norm": 0.004759619478136301,
      "learning_rate": 7.374491170448525e-05,
      "loss": 11.5,
      "step": 881
    },
    {
      "epoch": 0.4817862141954867,
      "grad_norm": 0.0019755198154598475,
      "learning_rate": 7.3541521908041e-05,
      "loss": 11.5,
      "step": 882
    },
    {
      "epoch": 0.4823324570687242,
      "grad_norm": 0.0028185860719531775,
      "learning_rate": 7.33382497345034e-05,
      "loss": 11.5,
      "step": 883
    },
    {
      "epoch": 0.4828786999419617,
      "grad_norm": 0.0017185299657285213,
      "learning_rate": 7.313509608753231e-05,
      "loss": 11.5,
      "step": 884
    },
    {
      "epoch": 0.4834249428151992,
      "grad_norm": 0.0023241250310093164,
      "learning_rate": 7.293206187026061e-05,
      "loss": 11.5,
      "step": 885
    },
    {
      "epoch": 0.4839711856884367,
      "grad_norm": 0.005199312698096037,
      "learning_rate": 7.27291479852903e-05,
      "loss": 11.5,
      "step": 886
    },
    {
      "epoch": 0.48451742856167423,
      "grad_norm": 0.0031441054306924343,
      "learning_rate": 7.252635533468843e-05,
      "loss": 11.5,
      "step": 887
    },
    {
      "epoch": 0.48506367143491175,
      "grad_norm": 0.002316902158781886,
      "learning_rate": 7.232368481998309e-05,
      "loss": 11.5,
      "step": 888
    },
    {
      "epoch": 0.48560991430814926,
      "grad_norm": 0.004939797800034285,
      "learning_rate": 7.212113734215932e-05,
      "loss": 11.5,
      "step": 889
    },
    {
      "epoch": 0.48615615718138677,
      "grad_norm": 0.004743826575577259,
      "learning_rate": 7.191871380165538e-05,
      "loss": 11.5,
      "step": 890
    },
    {
      "epoch": 0.4867024000546243,
      "grad_norm": 0.0031508817337453365,
      "learning_rate": 7.17164150983584e-05,
      "loss": 11.5,
      "step": 891
    },
    {
      "epoch": 0.4872486429278618,
      "grad_norm": 0.0028478377498686314,
      "learning_rate": 7.151424213160061e-05,
      "loss": 11.5,
      "step": 892
    },
    {
      "epoch": 0.4877948858010993,
      "grad_norm": 0.00233198469504714,
      "learning_rate": 7.131219580015521e-05,
      "loss": 11.5,
      "step": 893
    },
    {
      "epoch": 0.4883411286743368,
      "grad_norm": 0.003270956687629223,
      "learning_rate": 7.11102770022325e-05,
      "loss": 11.5,
      "step": 894
    },
    {
      "epoch": 0.48888737154757433,
      "grad_norm": 0.0023021420929580927,
      "learning_rate": 7.090848663547574e-05,
      "loss": 11.5,
      "step": 895
    },
    {
      "epoch": 0.48943361442081185,
      "grad_norm": 0.002850915305316448,
      "learning_rate": 7.070682559695736e-05,
      "loss": 11.5,
      "step": 896
    },
    {
      "epoch": 0.48997985729404936,
      "grad_norm": 0.002281076041981578,
      "learning_rate": 7.050529478317476e-05,
      "loss": 11.5,
      "step": 897
    },
    {
      "epoch": 0.49052610016728687,
      "grad_norm": 0.002127870684489608,
      "learning_rate": 7.03038950900464e-05,
      "loss": 11.5,
      "step": 898
    },
    {
      "epoch": 0.4910723430405244,
      "grad_norm": 0.0067464690655469894,
      "learning_rate": 7.010262741290798e-05,
      "loss": 11.5,
      "step": 899
    },
    {
      "epoch": 0.4916185859137619,
      "grad_norm": 0.0044977557845413685,
      "learning_rate": 6.990149264650814e-05,
      "loss": 11.5,
      "step": 900
    },
    {
      "epoch": 0.4921648287869994,
      "grad_norm": 0.003738689236342907,
      "learning_rate": 6.970049168500474e-05,
      "loss": 11.5,
      "step": 901
    },
    {
      "epoch": 0.4927110716602369,
      "grad_norm": 0.002164067467674613,
      "learning_rate": 6.94996254219608e-05,
      "loss": 11.5,
      "step": 902
    },
    {
      "epoch": 0.49325731453347443,
      "grad_norm": 0.0021851700730621815,
      "learning_rate": 6.929889475034048e-05,
      "loss": 11.5,
      "step": 903
    },
    {
      "epoch": 0.49380355740671195,
      "grad_norm": 0.0018468326888978481,
      "learning_rate": 6.909830056250527e-05,
      "loss": 11.5,
      "step": 904
    },
    {
      "epoch": 0.49434980027994946,
      "grad_norm": 0.001695907791145146,
      "learning_rate": 6.889784375020978e-05,
      "loss": 11.5,
      "step": 905
    },
    {
      "epoch": 0.49489604315318697,
      "grad_norm": 0.00273995753377676,
      "learning_rate": 6.869752520459803e-05,
      "loss": 11.5,
      "step": 906
    },
    {
      "epoch": 0.4954422860264245,
      "grad_norm": 0.0014393012970685959,
      "learning_rate": 6.849734581619918e-05,
      "loss": 11.5,
      "step": 907
    },
    {
      "epoch": 0.495988528899662,
      "grad_norm": 0.001313905231654644,
      "learning_rate": 6.829730647492404e-05,
      "loss": 11.5,
      "step": 908
    },
    {
      "epoch": 0.4965347717728995,
      "grad_norm": 0.0011272575939074159,
      "learning_rate": 6.80974080700606e-05,
      "loss": 11.5,
      "step": 909
    },
    {
      "epoch": 0.497081014646137,
      "grad_norm": 0.0011534381192177534,
      "learning_rate": 6.789765149027039e-05,
      "loss": 11.5,
      "step": 910
    },
    {
      "epoch": 0.49762725751937453,
      "grad_norm": 0.0017280761385336518,
      "learning_rate": 6.769803762358443e-05,
      "loss": 11.5,
      "step": 911
    },
    {
      "epoch": 0.49817350039261205,
      "grad_norm": 0.0009949959348887205,
      "learning_rate": 6.749856735739928e-05,
      "loss": 11.5,
      "step": 912
    },
    {
      "epoch": 0.49871974326584956,
      "grad_norm": 0.0018224769737571478,
      "learning_rate": 6.729924157847323e-05,
      "loss": 11.5,
      "step": 913
    },
    {
      "epoch": 0.49926598613908707,
      "grad_norm": 0.004745212849229574,
      "learning_rate": 6.710006117292209e-05,
      "loss": 11.5,
      "step": 914
    },
    {
      "epoch": 0.4998122290123246,
      "grad_norm": 0.002042470732703805,
      "learning_rate": 6.690102702621548e-05,
      "loss": 11.5,
      "step": 915
    },
    {
      "epoch": 0.5003584718855622,
      "grad_norm": 0.0013022255152463913,
      "learning_rate": 6.670214002317278e-05,
      "loss": 11.5,
      "step": 916
    },
    {
      "epoch": 0.5009047147587996,
      "grad_norm": 0.0018200019840151072,
      "learning_rate": 6.650340104795932e-05,
      "loss": 11.5,
      "step": 917
    },
    {
      "epoch": 0.5014509576320372,
      "grad_norm": 0.002653895178809762,
      "learning_rate": 6.630481098408228e-05,
      "loss": 11.5,
      "step": 918
    },
    {
      "epoch": 0.5019972005052746,
      "grad_norm": 0.002124528866261244,
      "learning_rate": 6.610637071438686e-05,
      "loss": 11.5,
      "step": 919
    },
    {
      "epoch": 0.5025434433785122,
      "grad_norm": 0.001327026286162436,
      "learning_rate": 6.590808112105232e-05,
      "loss": 11.5,
      "step": 920
    },
    {
      "epoch": 0.5030896862517497,
      "grad_norm": 0.0020096751395612955,
      "learning_rate": 6.570994308558812e-05,
      "loss": 11.5,
      "step": 921
    },
    {
      "epoch": 0.5036359291249872,
      "grad_norm": 0.002027923474088311,
      "learning_rate": 6.551195748882997e-05,
      "loss": 11.5,
      "step": 922
    },
    {
      "epoch": 0.5041821719982247,
      "grad_norm": 0.002560134045779705,
      "learning_rate": 6.531412521093586e-05,
      "loss": 11.5,
      "step": 923
    },
    {
      "epoch": 0.5047284148714622,
      "grad_norm": 0.00554626202210784,
      "learning_rate": 6.51164471313822e-05,
      "loss": 11.5,
      "step": 924
    },
    {
      "epoch": 0.5052746577446997,
      "grad_norm": 0.00247046141885221,
      "learning_rate": 6.491892412895995e-05,
      "loss": 11.5,
      "step": 925
    },
    {
      "epoch": 0.5058209006179373,
      "grad_norm": 0.005176033359020948,
      "learning_rate": 6.472155708177052e-05,
      "loss": 11.5,
      "step": 926
    },
    {
      "epoch": 0.5063671434911747,
      "grad_norm": 0.0045324754901230335,
      "learning_rate": 6.452434686722224e-05,
      "loss": 11.5,
      "step": 927
    },
    {
      "epoch": 0.5069133863644123,
      "grad_norm": 0.006474991794675589,
      "learning_rate": 6.432729436202604e-05,
      "loss": 11.5,
      "step": 928
    },
    {
      "epoch": 0.5074596292376498,
      "grad_norm": 0.002311764983460307,
      "learning_rate": 6.41304004421918e-05,
      "loss": 11.5,
      "step": 929
    },
    {
      "epoch": 0.5080058721108873,
      "grad_norm": 0.004412028938531876,
      "learning_rate": 6.393366598302446e-05,
      "loss": 11.5,
      "step": 930
    },
    {
      "epoch": 0.5085521149841248,
      "grad_norm": 0.0330335758626461,
      "learning_rate": 6.373709185911998e-05,
      "loss": 11.5,
      "step": 931
    },
    {
      "epoch": 0.5090983578573623,
      "grad_norm": 0.002150138607248664,
      "learning_rate": 6.354067894436155e-05,
      "loss": 11.5,
      "step": 932
    },
    {
      "epoch": 0.5096446007305998,
      "grad_norm": 0.0033174867276102304,
      "learning_rate": 6.334442811191576e-05,
      "loss": 11.5,
      "step": 933
    },
    {
      "epoch": 0.5101908436038374,
      "grad_norm": 0.0026628931518644094,
      "learning_rate": 6.314834023422858e-05,
      "loss": 11.5,
      "step": 934
    },
    {
      "epoch": 0.5107370864770748,
      "grad_norm": 0.0008951456402428448,
      "learning_rate": 6.295241618302156e-05,
      "loss": 11.5,
      "step": 935
    },
    {
      "epoch": 0.5112833293503124,
      "grad_norm": 0.0018861317075788975,
      "learning_rate": 6.275665682928803e-05,
      "loss": 11.5,
      "step": 936
    },
    {
      "epoch": 0.5118295722235499,
      "grad_norm": 0.0014663169858977199,
      "learning_rate": 6.256106304328905e-05,
      "loss": 11.5,
      "step": 937
    },
    {
      "epoch": 0.5123758150967874,
      "grad_norm": 0.0024490831419825554,
      "learning_rate": 6.23656356945497e-05,
      "loss": 11.5,
      "step": 938
    },
    {
      "epoch": 0.5129220579700249,
      "grad_norm": 0.004993111360818148,
      "learning_rate": 6.21703756518551e-05,
      "loss": 11.5,
      "step": 939
    },
    {
      "epoch": 0.5134683008432624,
      "grad_norm": 0.0028289591427892447,
      "learning_rate": 6.197528378324665e-05,
      "loss": 11.5,
      "step": 940
    },
    {
      "epoch": 0.5140145437164999,
      "grad_norm": 0.004099354147911072,
      "learning_rate": 6.17803609560181e-05,
      "loss": 11.5,
      "step": 941
    },
    {
      "epoch": 0.5145607865897375,
      "grad_norm": 0.002568903611972928,
      "learning_rate": 6.158560803671168e-05,
      "loss": 11.5,
      "step": 942
    },
    {
      "epoch": 0.5151070294629749,
      "grad_norm": 0.009959476999938488,
      "learning_rate": 6.139102589111435e-05,
      "loss": 11.5,
      "step": 943
    },
    {
      "epoch": 0.5156532723362125,
      "grad_norm": 0.0035951989702880383,
      "learning_rate": 6.119661538425381e-05,
      "loss": 11.5,
      "step": 944
    },
    {
      "epoch": 0.51619951520945,
      "grad_norm": 0.004883930087089539,
      "learning_rate": 6.100237738039484e-05,
      "loss": 11.5,
      "step": 945
    },
    {
      "epoch": 0.5167457580826875,
      "grad_norm": 0.003818538738414645,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 11.5,
      "step": 946
    },
    {
      "epoch": 0.517292000955925,
      "grad_norm": 0.006517360452562571,
      "learning_rate": 6.061442233490211e-05,
      "loss": 11.5,
      "step": 947
    },
    {
      "epoch": 0.5178382438291625,
      "grad_norm": 0.004829299170523882,
      "learning_rate": 6.042070701794806e-05,
      "loss": 11.5,
      "step": 948
    },
    {
      "epoch": 0.5183844867024,
      "grad_norm": 0.0044110058806836605,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 11.5,
      "step": 949
    },
    {
      "epoch": 0.5189307295756376,
      "grad_norm": 0.0046128565445542336,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 11.5,
      "step": 950
    },
    {
      "epoch": 0.519476972448875,
      "grad_norm": 0.0031397435814142227,
      "learning_rate": 5.98406202219875e-05,
      "loss": 11.5,
      "step": 951
    },
    {
      "epoch": 0.5200232153221126,
      "grad_norm": 0.002906328532844782,
      "learning_rate": 5.964761387365052e-05,
      "loss": 11.5,
      "step": 952
    },
    {
      "epoch": 0.52056945819535,
      "grad_norm": 0.002064706524834037,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 11.5,
      "step": 953
    },
    {
      "epoch": 0.5211157010685876,
      "grad_norm": 0.009561535902321339,
      "learning_rate": 5.926214020177074e-05,
      "loss": 11.5,
      "step": 954
    },
    {
      "epoch": 0.5216619439418251,
      "grad_norm": 0.0026621257420629263,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 11.5,
      "step": 955
    },
    {
      "epoch": 0.5222081868150626,
      "grad_norm": 0.000764953438192606,
      "learning_rate": 5.887739094043923e-05,
      "loss": 11.5,
      "step": 956
    },
    {
      "epoch": 0.5227544296883002,
      "grad_norm": 0.002469574334099889,
      "learning_rate": 5.868529010226845e-05,
      "loss": 11.5,
      "step": 957
    },
    {
      "epoch": 0.5233006725615377,
      "grad_norm": 0.002084027510136366,
      "learning_rate": 5.849337293136112e-05,
      "loss": 11.5,
      "step": 958
    },
    {
      "epoch": 0.5238469154347752,
      "grad_norm": 0.005116146523505449,
      "learning_rate": 5.830164028089766e-05,
      "loss": 11.5,
      "step": 959
    },
    {
      "epoch": 0.5243931583080127,
      "grad_norm": 0.0023875755723565817,
      "learning_rate": 5.811009300323818e-05,
      "loss": 11.5,
      "step": 960
    },
    {
      "epoch": 0.5249394011812503,
      "grad_norm": 0.0029359874315559864,
      "learning_rate": 5.791873194991872e-05,
      "loss": 11.5,
      "step": 961
    },
    {
      "epoch": 0.5254856440544877,
      "grad_norm": 0.0021241111680865288,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 11.5,
      "step": 962
    },
    {
      "epoch": 0.5260318869277253,
      "grad_norm": 0.002936464734375477,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 11.5,
      "step": 963
    },
    {
      "epoch": 0.5265781298009627,
      "grad_norm": 0.0025741513818502426,
      "learning_rate": 5.734577463892008e-05,
      "loss": 11.5,
      "step": 964
    },
    {
      "epoch": 0.5271243726742003,
      "grad_norm": 0.001486259396187961,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 11.5,
      "step": 965
    },
    {
      "epoch": 0.5276706155474378,
      "grad_norm": 0.0027294056490063667,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 11.5,
      "step": 966
    },
    {
      "epoch": 0.5282168584206753,
      "grad_norm": 0.003999914508312941,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 11.5,
      "step": 967
    },
    {
      "epoch": 0.5287631012939128,
      "grad_norm": 0.0012990074465051293,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 11.5,
      "step": 968
    },
    {
      "epoch": 0.5293093441671504,
      "grad_norm": 0.0018185500521212816,
      "learning_rate": 5.639464950898491e-05,
      "loss": 11.5,
      "step": 969
    },
    {
      "epoch": 0.5298555870403878,
      "grad_norm": 0.0013664327561855316,
      "learning_rate": 5.620500265600206e-05,
      "loss": 11.5,
      "step": 970
    },
    {
      "epoch": 0.5304018299136254,
      "grad_norm": 0.0010281425202265382,
      "learning_rate": 5.601555049656382e-05,
      "loss": 11.5,
      "step": 971
    },
    {
      "epoch": 0.5309480727868628,
      "grad_norm": 0.0011611772933974862,
      "learning_rate": 5.58262938728922e-05,
      "loss": 11.5,
      "step": 972
    },
    {
      "epoch": 0.5314943156601004,
      "grad_norm": 0.006015506573021412,
      "learning_rate": 5.563723362634008e-05,
      "loss": 11.5,
      "step": 973
    },
    {
      "epoch": 0.5320405585333379,
      "grad_norm": 0.0014010408194735646,
      "learning_rate": 5.544837059738719e-05,
      "loss": 11.5,
      "step": 974
    },
    {
      "epoch": 0.5325868014065754,
      "grad_norm": 0.004722512327134609,
      "learning_rate": 5.525970562563656e-05,
      "loss": 11.5,
      "step": 975
    },
    {
      "epoch": 0.5331330442798129,
      "grad_norm": 0.004967433866113424,
      "learning_rate": 5.507123954981073e-05,
      "loss": 11.5,
      "step": 976
    },
    {
      "epoch": 0.5336792871530505,
      "grad_norm": 0.005346799734979868,
      "learning_rate": 5.488297320774807e-05,
      "loss": 11.5,
      "step": 977
    },
    {
      "epoch": 0.5342255300262879,
      "grad_norm": 0.0014861986273899674,
      "learning_rate": 5.4694907436399e-05,
      "loss": 11.5,
      "step": 978
    },
    {
      "epoch": 0.5347717728995255,
      "grad_norm": 0.004273305181413889,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 11.5,
      "step": 979
    },
    {
      "epoch": 0.535318015772763,
      "grad_norm": 0.004509516526013613,
      "learning_rate": 5.431938094918132e-05,
      "loss": 11.5,
      "step": 980
    },
    {
      "epoch": 0.5358642586460005,
      "grad_norm": 0.0021610937546938658,
      "learning_rate": 5.41319219027404e-05,
      "loss": 11.5,
      "step": 981
    },
    {
      "epoch": 0.536410501519238,
      "grad_norm": 0.0038977896329015493,
      "learning_rate": 5.394466676586114e-05,
      "loss": 11.5,
      "step": 982
    },
    {
      "epoch": 0.5369567443924755,
      "grad_norm": 0.00370238465256989,
      "learning_rate": 5.375761637099854e-05,
      "loss": 11.5,
      "step": 983
    },
    {
      "epoch": 0.537502987265713,
      "grad_norm": 0.0015931488014757633,
      "learning_rate": 5.357077154969742e-05,
      "loss": 11.5,
      "step": 984
    },
    {
      "epoch": 0.5380492301389506,
      "grad_norm": 0.00264379964210093,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 11.5,
      "step": 985
    },
    {
      "epoch": 0.538595473012188,
      "grad_norm": 0.0021742002572864294,
      "learning_rate": 5.3197701949386e-05,
      "loss": 11.5,
      "step": 986
    },
    {
      "epoch": 0.5391417158854256,
      "grad_norm": 0.0049259718507528305,
      "learning_rate": 5.301147882888116e-05,
      "loss": 11.5,
      "step": 987
    },
    {
      "epoch": 0.539687958758663,
      "grad_norm": 0.0023208707571029663,
      "learning_rate": 5.28254645989414e-05,
      "loss": 11.5,
      "step": 988
    },
    {
      "epoch": 0.5402342016319006,
      "grad_norm": 0.006813263054937124,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 11.5,
      "step": 989
    },
    {
      "epoch": 0.5407804445051381,
      "grad_norm": 0.004512079991400242,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 11.5,
      "step": 990
    },
    {
      "epoch": 0.5413266873783756,
      "grad_norm": 0.0020569567568600178,
      "learning_rate": 5.226868351723244e-05,
      "loss": 11.5,
      "step": 991
    },
    {
      "epoch": 0.5418729302516131,
      "grad_norm": 0.0020024082623422146,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 11.5,
      "step": 992
    },
    {
      "epoch": 0.5424191731248507,
      "grad_norm": 0.0009753123740665615,
      "learning_rate": 5.189855571785859e-05,
      "loss": 11.5,
      "step": 993
    },
    {
      "epoch": 0.5429654159980881,
      "grad_norm": 0.00147640286013484,
      "learning_rate": 5.171381216425863e-05,
      "loss": 11.5,
      "step": 994
    },
    {
      "epoch": 0.5435116588713257,
      "grad_norm": 0.003999672364443541,
      "learning_rate": 5.152928327008635e-05,
      "loss": 11.5,
      "step": 995
    },
    {
      "epoch": 0.5440579017445631,
      "grad_norm": 0.002149751642718911,
      "learning_rate": 5.134496985567714e-05,
      "loss": 11.5,
      "step": 996
    },
    {
      "epoch": 0.5446041446178007,
      "grad_norm": 0.0029203195590525866,
      "learning_rate": 5.116087274040837e-05,
      "loss": 11.5,
      "step": 997
    },
    {
      "epoch": 0.5451503874910382,
      "grad_norm": 0.003084364114329219,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 11.5,
      "step": 998
    },
    {
      "epoch": 0.5456966303642757,
      "grad_norm": 0.00491071492433548,
      "learning_rate": 5.07933306799904e-05,
      "loss": 11.5,
      "step": 999
    },
    {
      "epoch": 0.5462428732375132,
      "grad_norm": 0.0036079217679798603,
      "learning_rate": 5.060988736877366e-05,
      "loss": 11.5,
      "step": 1000
    },
    {
      "epoch": 0.5467891161107508,
      "grad_norm": 0.002479356713593006,
      "learning_rate": 5.042666362455498e-05,
      "loss": 11.5,
      "step": 1001
    },
    {
      "epoch": 0.5473353589839882,
      "grad_norm": 0.0016236052615568042,
      "learning_rate": 5.024366026186755e-05,
      "loss": 11.5,
      "step": 1002
    },
    {
      "epoch": 0.5478816018572258,
      "grad_norm": 0.0022359138820320368,
      "learning_rate": 5.006087809426496e-05,
      "loss": 11.5,
      "step": 1003
    },
    {
      "epoch": 0.5484278447304632,
      "grad_norm": 0.00341726909391582,
      "learning_rate": 4.987831793431731e-05,
      "loss": 11.5,
      "step": 1004
    },
    {
      "epoch": 0.5489740876037008,
      "grad_norm": 0.0018256044713780284,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 11.5,
      "step": 1005
    },
    {
      "epoch": 0.5495203304769383,
      "grad_norm": 0.0009432027582079172,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 11.5,
      "step": 1006
    },
    {
      "epoch": 0.5500665733501758,
      "grad_norm": 0.002036880236119032,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 11.5,
      "step": 1007
    },
    {
      "epoch": 0.5506128162234133,
      "grad_norm": 0.0018898161360993981,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 11.5,
      "step": 1008
    },
    {
      "epoch": 0.5511590590966509,
      "grad_norm": 0.0018555504502728581,
      "learning_rate": 4.896887562007054e-05,
      "loss": 11.5,
      "step": 1009
    },
    {
      "epoch": 0.5517053019698883,
      "grad_norm": 0.002257602522149682,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 11.5,
      "step": 1010
    },
    {
      "epoch": 0.5522515448431259,
      "grad_norm": 0.002359110862016678,
      "learning_rate": 4.860668107674823e-05,
      "loss": 11.5,
      "step": 1011
    },
    {
      "epoch": 0.5527977877163633,
      "grad_norm": 0.002735028974711895,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 11.5,
      "step": 1012
    },
    {
      "epoch": 0.5533440305896009,
      "grad_norm": 0.0018301908858120441,
      "learning_rate": 4.824540042194443e-05,
      "loss": 11.5,
      "step": 1013
    },
    {
      "epoch": 0.5538902734628384,
      "grad_norm": 0.0033904314041137695,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 11.5,
      "step": 1014
    },
    {
      "epoch": 0.5544365163360759,
      "grad_norm": 0.00229973834939301,
      "learning_rate": 4.788504008003978e-05,
      "loss": 11.5,
      "step": 1015
    },
    {
      "epoch": 0.5549827592093134,
      "grad_norm": 0.0017884359695017338,
      "learning_rate": 4.770520702957182e-05,
      "loss": 11.5,
      "step": 1016
    },
    {
      "epoch": 0.555529002082551,
      "grad_norm": 0.0014823866076767445,
      "learning_rate": 4.752560645904962e-05,
      "loss": 11.5,
      "step": 1017
    },
    {
      "epoch": 0.5560752449557884,
      "grad_norm": 0.006923645734786987,
      "learning_rate": 4.734623916689941e-05,
      "loss": 11.5,
      "step": 1018
    },
    {
      "epoch": 0.556621487829026,
      "grad_norm": 0.001784989726729691,
      "learning_rate": 4.716710595051022e-05,
      "loss": 11.5,
      "step": 1019
    },
    {
      "epoch": 0.5571677307022634,
      "grad_norm": 0.002050504321232438,
      "learning_rate": 4.698820760623064e-05,
      "loss": 11.5,
      "step": 1020
    },
    {
      "epoch": 0.557713973575501,
      "grad_norm": 0.0017426966223865747,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 11.5,
      "step": 1021
    },
    {
      "epoch": 0.5582602164487385,
      "grad_norm": 0.0015206191455945373,
      "learning_rate": 4.663111871417e-05,
      "loss": 11.5,
      "step": 1022
    },
    {
      "epoch": 0.558806459321976,
      "grad_norm": 0.0022413195110857487,
      "learning_rate": 4.645292975385111e-05,
      "loss": 11.5,
      "step": 1023
    },
    {
      "epoch": 0.5593527021952135,
      "grad_norm": 0.001363492221571505,
      "learning_rate": 4.627497884055912e-05,
      "loss": 11.5,
      "step": 1024
    },
    {
      "epoch": 0.5598989450684511,
      "grad_norm": 0.004290115553885698,
      "learning_rate": 4.609726676538652e-05,
      "loss": 11.5,
      "step": 1025
    },
    {
      "epoch": 0.5604451879416886,
      "grad_norm": 0.0031121852807700634,
      "learning_rate": 4.591979431836402e-05,
      "loss": 11.5,
      "step": 1026
    },
    {
      "epoch": 0.5609914308149261,
      "grad_norm": 0.006989837158471346,
      "learning_rate": 4.574256228845706e-05,
      "loss": 11.5,
      "step": 1027
    },
    {
      "epoch": 0.5615376736881637,
      "grad_norm": 0.001483198837377131,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 11.5,
      "step": 1028
    },
    {
      "epoch": 0.5620839165614011,
      "grad_norm": 0.001577763818204403,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 11.5,
      "step": 1029
    },
    {
      "epoch": 0.5626301594346387,
      "grad_norm": 0.0024241406936198473,
      "learning_rate": 4.521231657503132e-05,
      "loss": 11.5,
      "step": 1030
    },
    {
      "epoch": 0.5631764023078761,
      "grad_norm": 0.000545271672308445,
      "learning_rate": 4.503605408181286e-05,
      "loss": 11.5,
      "step": 1031
    },
    {
      "epoch": 0.5637226451811137,
      "grad_norm": 0.0013636453077197075,
      "learning_rate": 4.486003593443537e-05,
      "loss": 11.5,
      "step": 1032
    },
    {
      "epoch": 0.5642688880543512,
      "grad_norm": 0.0015294887125492096,
      "learning_rate": 4.468426291539914e-05,
      "loss": 11.5,
      "step": 1033
    },
    {
      "epoch": 0.5648151309275887,
      "grad_norm": 0.0024410763289779425,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 11.5,
      "step": 1034
    },
    {
      "epoch": 0.5653613738008262,
      "grad_norm": 0.0010288477642461658,
      "learning_rate": 4.433345538689929e-05,
      "loss": 11.5,
      "step": 1035
    },
    {
      "epoch": 0.5659076166740638,
      "grad_norm": 0.0026750138495117426,
      "learning_rate": 4.415842243697369e-05,
      "loss": 11.5,
      "step": 1036
    },
    {
      "epoch": 0.5664538595473012,
      "grad_norm": 0.002512294100597501,
      "learning_rate": 4.39836377344583e-05,
      "loss": 11.5,
      "step": 1037
    },
    {
      "epoch": 0.5670001024205388,
      "grad_norm": 0.0034719575196504593,
      "learning_rate": 4.380910205637012e-05,
      "loss": 11.5,
      "step": 1038
    },
    {
      "epoch": 0.5675463452937762,
      "grad_norm": 0.0036240341141819954,
      "learning_rate": 4.363481617861893e-05,
      "loss": 11.5,
      "step": 1039
    },
    {
      "epoch": 0.5680925881670138,
      "grad_norm": 0.0053438954055309296,
      "learning_rate": 4.346078087600412e-05,
      "loss": 11.5,
      "step": 1040
    },
    {
      "epoch": 0.5686388310402513,
      "grad_norm": 0.002039511688053608,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 11.5,
      "step": 1041
    },
    {
      "epoch": 0.5691850739134888,
      "grad_norm": 0.002616970567032695,
      "learning_rate": 4.311346508980772e-05,
      "loss": 11.5,
      "step": 1042
    },
    {
      "epoch": 0.5697313167867263,
      "grad_norm": 0.0023614356759935617,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 11.5,
      "step": 1043
    },
    {
      "epoch": 0.5702775596599639,
      "grad_norm": 0.0031840763986110687,
      "learning_rate": 4.27671608738349e-05,
      "loss": 11.5,
      "step": 1044
    },
    {
      "epoch": 0.5708238025332013,
      "grad_norm": 0.004273599945008755,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 11.5,
      "step": 1045
    },
    {
      "epoch": 0.5713700454064389,
      "grad_norm": 0.002781930845230818,
      "learning_rate": 4.242187438615153e-05,
      "loss": 11.5,
      "step": 1046
    },
    {
      "epoch": 0.5719162882796763,
      "grad_norm": 0.003205101238563657,
      "learning_rate": 4.224961470986849e-05,
      "loss": 11.5,
      "step": 1047
    },
    {
      "epoch": 0.5724625311529139,
      "grad_norm": 0.0037823196034878492,
      "learning_rate": 4.207761176672614e-05,
      "loss": 11.5,
      "step": 1048
    },
    {
      "epoch": 0.5730087740261514,
      "grad_norm": 0.007255354430526495,
      "learning_rate": 4.190586632137491e-05,
      "loss": 11.5,
      "step": 1049
    },
    {
      "epoch": 0.5735550168993889,
      "grad_norm": 0.005401613190770149,
      "learning_rate": 4.173437913732048e-05,
      "loss": 11.5,
      "step": 1050
    },
    {
      "epoch": 0.5741012597726264,
      "grad_norm": 0.0018057918641716242,
      "learning_rate": 4.156315097692037e-05,
      "loss": 11.5,
      "step": 1051
    },
    {
      "epoch": 0.574647502645864,
      "grad_norm": 0.0018210355192422867,
      "learning_rate": 4.139218260138074e-05,
      "loss": 11.5,
      "step": 1052
    },
    {
      "epoch": 0.5751937455191014,
      "grad_norm": 0.003044363111257553,
      "learning_rate": 4.12214747707527e-05,
      "loss": 11.5,
      "step": 1053
    },
    {
      "epoch": 0.575739988392339,
      "grad_norm": 0.0023737873416393995,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 11.5,
      "step": 1054
    },
    {
      "epoch": 0.5762862312655764,
      "grad_norm": 0.0015267801936715841,
      "learning_rate": 4.088084377864135e-05,
      "loss": 11.5,
      "step": 1055
    },
    {
      "epoch": 0.576832474138814,
      "grad_norm": 0.0017467316938564181,
      "learning_rate": 4.07109221314556e-05,
      "loss": 11.5,
      "step": 1056
    },
    {
      "epoch": 0.5773787170120515,
      "grad_norm": 0.0023274230770766735,
      "learning_rate": 4.054126405776971e-05,
      "loss": 11.5,
      "step": 1057
    },
    {
      "epoch": 0.577924959885289,
      "grad_norm": 0.0017256352584809065,
      "learning_rate": 4.037187031180985e-05,
      "loss": 11.5,
      "step": 1058
    },
    {
      "epoch": 0.5784712027585265,
      "grad_norm": 0.004267829470336437,
      "learning_rate": 4.020274164662707e-05,
      "loss": 11.5,
      "step": 1059
    },
    {
      "epoch": 0.579017445631764,
      "grad_norm": 0.0015294020995497704,
      "learning_rate": 4.003387881409397e-05,
      "loss": 11.5,
      "step": 1060
    },
    {
      "epoch": 0.5795636885050015,
      "grad_norm": 0.0022894744761288166,
      "learning_rate": 3.986528256490141e-05,
      "loss": 11.5,
      "step": 1061
    },
    {
      "epoch": 0.5801099313782391,
      "grad_norm": 0.0018242521909996867,
      "learning_rate": 3.969695364855511e-05,
      "loss": 11.5,
      "step": 1062
    },
    {
      "epoch": 0.5806561742514765,
      "grad_norm": 0.002613277640193701,
      "learning_rate": 3.952889281337235e-05,
      "loss": 11.5,
      "step": 1063
    },
    {
      "epoch": 0.5812024171247141,
      "grad_norm": 0.0041056349873542786,
      "learning_rate": 3.93611008064786e-05,
      "loss": 11.5,
      "step": 1064
    },
    {
      "epoch": 0.5817486599979516,
      "grad_norm": 0.0035323777701705694,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 11.5,
      "step": 1065
    },
    {
      "epoch": 0.5822949028711891,
      "grad_norm": 0.00199114135466516,
      "learning_rate": 3.90263262600816e-05,
      "loss": 11.5,
      "step": 1066
    },
    {
      "epoch": 0.5828411457444266,
      "grad_norm": 0.016857963055372238,
      "learning_rate": 3.88593452088406e-05,
      "loss": 11.5,
      "step": 1067
    },
    {
      "epoch": 0.5833873886176641,
      "grad_norm": 0.0008053850033320487,
      "learning_rate": 3.869263596240661e-05,
      "loss": 11.5,
      "step": 1068
    },
    {
      "epoch": 0.5839336314909016,
      "grad_norm": 0.0014982972061261535,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 11.5,
      "step": 1069
    },
    {
      "epoch": 0.5844798743641392,
      "grad_norm": 0.0020326015073806047,
      "learning_rate": 3.836003584721577e-05,
      "loss": 11.5,
      "step": 1070
    },
    {
      "epoch": 0.5850261172373766,
      "grad_norm": 0.0008912864723242819,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 11.5,
      "step": 1071
    },
    {
      "epoch": 0.5855723601106142,
      "grad_norm": 0.0018782146507874131,
      "learning_rate": 3.802853182888543e-05,
      "loss": 11.5,
      "step": 1072
    },
    {
      "epoch": 0.5861186029838517,
      "grad_norm": 0.002121982164680958,
      "learning_rate": 3.786319269895877e-05,
      "loss": 11.5,
      "step": 1073
    },
    {
      "epoch": 0.5866648458570892,
      "grad_norm": 0.003669917117804289,
      "learning_rate": 3.769812980230074e-05,
      "loss": 11.5,
      "step": 1074
    },
    {
      "epoch": 0.5872110887303267,
      "grad_norm": 0.0025308290496468544,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 11.5,
      "step": 1075
    },
    {
      "epoch": 0.5877573316035642,
      "grad_norm": 0.007471106480807066,
      "learning_rate": 3.736883564275112e-05,
      "loss": 11.5,
      "step": 1076
    },
    {
      "epoch": 0.5883035744768017,
      "grad_norm": 0.007432878948748112,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 11.5,
      "step": 1077
    },
    {
      "epoch": 0.5888498173500393,
      "grad_norm": 0.003973603714257479,
      "learning_rate": 3.704065520582549e-05,
      "loss": 11.5,
      "step": 1078
    },
    {
      "epoch": 0.5893960602232767,
      "grad_norm": 0.002147914608940482,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 11.5,
      "step": 1079
    },
    {
      "epoch": 0.5899423030965143,
      "grad_norm": 0.0017495207721367478,
      "learning_rate": 3.671359432730834e-05,
      "loss": 11.5,
      "step": 1080
    },
    {
      "epoch": 0.5904885459697518,
      "grad_norm": 0.0033344049006700516,
      "learning_rate": 3.655048554069478e-05,
      "loss": 11.5,
      "step": 1081
    },
    {
      "epoch": 0.5910347888429893,
      "grad_norm": 0.003967491444200277,
      "learning_rate": 3.638765882307589e-05,
      "loss": 11.5,
      "step": 1082
    },
    {
      "epoch": 0.5915810317162268,
      "grad_norm": 0.004562418907880783,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 11.5,
      "step": 1083
    },
    {
      "epoch": 0.5921272745894643,
      "grad_norm": 0.0011576596880331635,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 11.5,
      "step": 1084
    },
    {
      "epoch": 0.5926735174627018,
      "grad_norm": 0.004044550936669111,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 11.5,
      "step": 1085
    },
    {
      "epoch": 0.5932197603359394,
      "grad_norm": 0.0031230563763529062,
      "learning_rate": 3.573918710080857e-05,
      "loss": 11.5,
      "step": 1086
    },
    {
      "epoch": 0.5937660032091768,
      "grad_norm": 0.0018693240126594901,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 11.5,
      "step": 1087
    },
    {
      "epoch": 0.5943122460824144,
      "grad_norm": 0.00299852155148983,
      "learning_rate": 3.541666241405588e-05,
      "loss": 11.5,
      "step": 1088
    },
    {
      "epoch": 0.5948584889556519,
      "grad_norm": 0.0014317899476736784,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 11.5,
      "step": 1089
    },
    {
      "epoch": 0.5954047318288894,
      "grad_norm": 0.0012998664751648903,
      "learning_rate": 3.509528616394716e-05,
      "loss": 11.5,
      "step": 1090
    },
    {
      "epoch": 0.5959509747021269,
      "grad_norm": 0.001895720954053104,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 11.5,
      "step": 1091
    },
    {
      "epoch": 0.5964972175753644,
      "grad_norm": 0.004137916956096888,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 11.5,
      "step": 1092
    },
    {
      "epoch": 0.597043460448602,
      "grad_norm": 0.0020725708454847336,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 11.5,
      "step": 1093
    },
    {
      "epoch": 0.5975897033218395,
      "grad_norm": 0.004418713506311178,
      "learning_rate": 3.445600181230134e-05,
      "loss": 11.5,
      "step": 1094
    },
    {
      "epoch": 0.598135946195077,
      "grad_norm": 0.006596324499696493,
      "learning_rate": 3.429690740177549e-05,
      "loss": 11.5,
      "step": 1095
    },
    {
      "epoch": 0.5986821890683145,
      "grad_norm": 0.0028011086396872997,
      "learning_rate": 3.413810507867436e-05,
      "loss": 11.5,
      "step": 1096
    },
    {
      "epoch": 0.5992284319415521,
      "grad_norm": 0.012573237530887127,
      "learning_rate": 3.397959554896415e-05,
      "loss": 11.5,
      "step": 1097
    },
    {
      "epoch": 0.5997746748147895,
      "grad_norm": 0.002526806900277734,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 11.5,
      "step": 1098
    },
    {
      "epoch": 0.6003209176880271,
      "grad_norm": 0.003810478840023279,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 11.5,
      "step": 1099
    },
    {
      "epoch": 0.6008671605612645,
      "grad_norm": 0.003942031878978014,
      "learning_rate": 3.350583076029754e-05,
      "loss": 11.5,
      "step": 1100
    },
    {
      "epoch": 0.6014134034345021,
      "grad_norm": 0.0023176928516477346,
      "learning_rate": 3.334849943773323e-05,
      "loss": 11.5,
      "step": 1101
    },
    {
      "epoch": 0.6019596463077396,
      "grad_norm": 0.0016617953078821301,
      "learning_rate": 3.319146441880371e-05,
      "loss": 11.5,
      "step": 1102
    },
    {
      "epoch": 0.6025058891809771,
      "grad_norm": 0.003315573325380683,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 11.5,
      "step": 1103
    },
    {
      "epoch": 0.6030521320542146,
      "grad_norm": 0.001432799850590527,
      "learning_rate": 3.28782860829667e-05,
      "loss": 11.5,
      "step": 1104
    },
    {
      "epoch": 0.6035983749274522,
      "grad_norm": 0.0018534609116613865,
      "learning_rate": 3.272214415831418e-05,
      "loss": 11.5,
      "step": 1105
    },
    {
      "epoch": 0.6041446178006896,
      "grad_norm": 0.0014482801780104637,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 11.5,
      "step": 1106
    },
    {
      "epoch": 0.6046908606739272,
      "grad_norm": 0.0017829852877184749,
      "learning_rate": 3.241075826623401e-05,
      "loss": 11.5,
      "step": 1107
    },
    {
      "epoch": 0.6052371035471646,
      "grad_norm": 0.0007371989777311683,
      "learning_rate": 3.225551568309284e-05,
      "loss": 11.5,
      "step": 1108
    },
    {
      "epoch": 0.6057833464204022,
      "grad_norm": 0.012006135657429695,
      "learning_rate": 3.210057426251773e-05,
      "loss": 11.5,
      "step": 1109
    },
    {
      "epoch": 0.6063295892936397,
      "grad_norm": 0.0022680237889289856,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 11.5,
      "step": 1110
    },
    {
      "epoch": 0.6068758321668772,
      "grad_norm": 0.0010752554517239332,
      "learning_rate": 3.179159766293282e-05,
      "loss": 11.5,
      "step": 1111
    },
    {
      "epoch": 0.6074220750401147,
      "grad_norm": 0.0020027230493724346,
      "learning_rate": 3.163756385749889e-05,
      "loss": 11.5,
      "step": 1112
    },
    {
      "epoch": 0.6079683179133523,
      "grad_norm": 0.002606367925181985,
      "learning_rate": 3.148383396177653e-05,
      "loss": 11.5,
      "step": 1113
    },
    {
      "epoch": 0.6085145607865897,
      "grad_norm": 0.0007325381156988442,
      "learning_rate": 3.133040865918213e-05,
      "loss": 11.5,
      "step": 1114
    },
    {
      "epoch": 0.6090608036598273,
      "grad_norm": 0.0012890608049929142,
      "learning_rate": 3.117728863177796e-05,
      "loss": 11.5,
      "step": 1115
    },
    {
      "epoch": 0.6096070465330647,
      "grad_norm": 0.0016919784247875214,
      "learning_rate": 3.102447456026919e-05,
      "loss": 11.5,
      "step": 1116
    },
    {
      "epoch": 0.6101532894063023,
      "grad_norm": 0.0019321856088936329,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 11.5,
      "step": 1117
    },
    {
      "epoch": 0.6106995322795398,
      "grad_norm": 0.001461253734305501,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 11.5,
      "step": 1118
    },
    {
      "epoch": 0.6112457751527773,
      "grad_norm": 0.0024798670783638954,
      "learning_rate": 3.056787486774656e-05,
      "loss": 11.5,
      "step": 1119
    },
    {
      "epoch": 0.6117920180260148,
      "grad_norm": 0.0030659267213195562,
      "learning_rate": 3.041629139962283e-05,
      "loss": 11.5,
      "step": 1120
    },
    {
      "epoch": 0.6123382608992524,
      "grad_norm": 0.0033599124290049076,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 11.5,
      "step": 1121
    },
    {
      "epoch": 0.6128845037724898,
      "grad_norm": 0.004811499733477831,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 11.5,
      "step": 1122
    },
    {
      "epoch": 0.6134307466457274,
      "grad_norm": 0.002827427117154002,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 11.5,
      "step": 1123
    },
    {
      "epoch": 0.6139769895189648,
      "grad_norm": 0.0018454660894349217,
      "learning_rate": 2.98130576348121e-05,
      "loss": 11.5,
      "step": 1124
    },
    {
      "epoch": 0.6145232323922024,
      "grad_norm": 0.004254624247550964,
      "learning_rate": 2.966302757267625e-05,
      "loss": 11.5,
      "step": 1125
    },
    {
      "epoch": 0.6145232323922024,
      "eval_loss": 11.5,
      "eval_runtime": 22.4577,
      "eval_samples_per_second": 137.325,
      "eval_steps_per_second": 68.662,
      "step": 1125
    },
    {
      "epoch": 0.6150694752654399,
      "grad_norm": 0.01993197202682495,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 11.5,
      "step": 1126
    },
    {
      "epoch": 0.6156157181386774,
      "grad_norm": 0.004962220322340727,
      "learning_rate": 2.936390617691097e-05,
      "loss": 11.5,
      "step": 1127
    },
    {
      "epoch": 0.6161619610119149,
      "grad_norm": 0.0021024374291300774,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 11.5,
      "step": 1128
    },
    {
      "epoch": 0.6167082038851525,
      "grad_norm": 0.003987603355199099,
      "learning_rate": 2.906604084937572e-05,
      "loss": 11.5,
      "step": 1129
    },
    {
      "epoch": 0.6172544467583899,
      "grad_norm": 0.0037599438801407814,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 11.5,
      "step": 1130
    },
    {
      "epoch": 0.6178006896316275,
      "grad_norm": 0.0014475536299869418,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 11.5,
      "step": 1131
    },
    {
      "epoch": 0.6183469325048649,
      "grad_norm": 0.0021651582792401314,
      "learning_rate": 2.862160956643517e-05,
      "loss": 11.5,
      "step": 1132
    },
    {
      "epoch": 0.6188931753781025,
      "grad_norm": 0.011577312834560871,
      "learning_rate": 2.847409956342092e-05,
      "loss": 11.5,
      "step": 1133
    },
    {
      "epoch": 0.61943941825134,
      "grad_norm": 0.0025699161924421787,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 11.5,
      "step": 1134
    },
    {
      "epoch": 0.6199856611245775,
      "grad_norm": 0.0021698148921132088,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 11.5,
      "step": 1135
    },
    {
      "epoch": 0.620531903997815,
      "grad_norm": 0.003978729248046875,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 11.5,
      "step": 1136
    },
    {
      "epoch": 0.6210781468710526,
      "grad_norm": 0.0019385089399293065,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 11.5,
      "step": 1137
    },
    {
      "epoch": 0.62162438974429,
      "grad_norm": 0.0026736219879239798,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 11.5,
      "step": 1138
    },
    {
      "epoch": 0.6221706326175276,
      "grad_norm": 0.0061402530409395695,
      "learning_rate": 2.759573983312138e-05,
      "loss": 11.5,
      "step": 1139
    },
    {
      "epoch": 0.622716875490765,
      "grad_norm": 0.003079845104366541,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 11.5,
      "step": 1140
    },
    {
      "epoch": 0.6232631183640026,
      "grad_norm": 0.0016338963760063052,
      "learning_rate": 2.730552135763632e-05,
      "loss": 11.5,
      "step": 1141
    },
    {
      "epoch": 0.6238093612372401,
      "grad_norm": 0.0019486056407913566,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 11.5,
      "step": 1142
    },
    {
      "epoch": 0.6243556041104776,
      "grad_norm": 0.003156149759888649,
      "learning_rate": 2.701659555308169e-05,
      "loss": 11.5,
      "step": 1143
    },
    {
      "epoch": 0.6249018469837151,
      "grad_norm": 0.006141822785139084,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 11.5,
      "step": 1144
    },
    {
      "epoch": 0.6254480898569527,
      "grad_norm": 0.004009457770735025,
      "learning_rate": 2.672896755720654e-05,
      "loss": 11.5,
      "step": 1145
    },
    {
      "epoch": 0.6259943327301901,
      "grad_norm": 0.0022867745719850063,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 11.5,
      "step": 1146
    },
    {
      "epoch": 0.6265405756034277,
      "grad_norm": 0.002311904449015856,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 11.5,
      "step": 1147
    },
    {
      "epoch": 0.6270868184766651,
      "grad_norm": 0.005323686636984348,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 11.5,
      "step": 1148
    },
    {
      "epoch": 0.6276330613499027,
      "grad_norm": 0.0063576726242899895,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 11.5,
      "step": 1149
    },
    {
      "epoch": 0.6281793042231402,
      "grad_norm": 0.009111561812460423,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 11.5,
      "step": 1150
    },
    {
      "epoch": 0.6287255470963777,
      "grad_norm": 0.0015762292314320803,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 11.5,
      "step": 1151
    },
    {
      "epoch": 0.6292717899696152,
      "grad_norm": 0.0038388462271541357,
      "learning_rate": 2.57325634487503e-05,
      "loss": 11.5,
      "step": 1152
    },
    {
      "epoch": 0.6298180328428528,
      "grad_norm": 0.0025379660073667765,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 11.5,
      "step": 1153
    },
    {
      "epoch": 0.6303642757160902,
      "grad_norm": 0.0019404747290536761,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 11.5,
      "step": 1154
    },
    {
      "epoch": 0.6309105185893278,
      "grad_norm": 0.0014938137028366327,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 11.5,
      "step": 1155
    },
    {
      "epoch": 0.6314567614625652,
      "grad_norm": 0.0011000343365594745,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 11.5,
      "step": 1156
    },
    {
      "epoch": 0.6320030043358028,
      "grad_norm": 0.002351970411837101,
      "learning_rate": 2.503073835892471e-05,
      "loss": 11.5,
      "step": 1157
    },
    {
      "epoch": 0.6325492472090403,
      "grad_norm": 0.0016858282033354044,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 11.5,
      "step": 1158
    },
    {
      "epoch": 0.6330954900822778,
      "grad_norm": 0.0027756807394325733,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 11.5,
      "step": 1159
    },
    {
      "epoch": 0.6336417329555153,
      "grad_norm": 0.0020845020189881325,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 11.5,
      "step": 1160
    },
    {
      "epoch": 0.6341879758287529,
      "grad_norm": 0.0010510800639167428,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 11.5,
      "step": 1161
    },
    {
      "epoch": 0.6347342187019904,
      "grad_norm": 0.0016489658737555146,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 11.5,
      "step": 1162
    },
    {
      "epoch": 0.6352804615752279,
      "grad_norm": 0.003023509168997407,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 11.5,
      "step": 1163
    },
    {
      "epoch": 0.6358267044484655,
      "grad_norm": 0.0023103358689695597,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 11.5,
      "step": 1164
    },
    {
      "epoch": 0.6363729473217029,
      "grad_norm": 0.0019096818286925554,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 11.5,
      "step": 1165
    },
    {
      "epoch": 0.6369191901949405,
      "grad_norm": 0.00205929740332067,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 11.5,
      "step": 1166
    },
    {
      "epoch": 0.6374654330681779,
      "grad_norm": 0.0013338574208319187,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 11.5,
      "step": 1167
    },
    {
      "epoch": 0.6380116759414155,
      "grad_norm": 0.0013540113577619195,
      "learning_rate": 2.351616006387214e-05,
      "loss": 11.5,
      "step": 1168
    },
    {
      "epoch": 0.638557918814653,
      "grad_norm": 0.0031189974397420883,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 11.5,
      "step": 1169
    },
    {
      "epoch": 0.6391041616878905,
      "grad_norm": 0.0036993299145251513,
      "learning_rate": 2.324517865697501e-05,
      "loss": 11.5,
      "step": 1170
    },
    {
      "epoch": 0.639650404561128,
      "grad_norm": 0.001113213482312858,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 11.5,
      "step": 1171
    },
    {
      "epoch": 0.6401966474343656,
      "grad_norm": 0.005887668114155531,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 11.5,
      "step": 1172
    },
    {
      "epoch": 0.640742890307603,
      "grad_norm": 0.0030743663664907217,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 11.5,
      "step": 1173
    },
    {
      "epoch": 0.6412891331808406,
      "grad_norm": 0.0017425680998712778,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 11.5,
      "step": 1174
    },
    {
      "epoch": 0.641835376054078,
      "grad_norm": 0.003690354758873582,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 11.5,
      "step": 1175
    },
    {
      "epoch": 0.6423816189273156,
      "grad_norm": 0.0026895971968770027,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 11.5,
      "step": 1176
    },
    {
      "epoch": 0.6429278618005531,
      "grad_norm": 0.004942162428051233,
      "learning_rate": 2.230752351039228e-05,
      "loss": 11.5,
      "step": 1177
    },
    {
      "epoch": 0.6434741046737906,
      "grad_norm": 0.002101318910717964,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 11.5,
      "step": 1178
    },
    {
      "epoch": 0.6440203475470281,
      "grad_norm": 0.002412347588688135,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 11.5,
      "step": 1179
    },
    {
      "epoch": 0.6445665904202657,
      "grad_norm": 0.0019196484936401248,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 11.5,
      "step": 1180
    },
    {
      "epoch": 0.6451128332935031,
      "grad_norm": 0.002362537896260619,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 11.5,
      "step": 1181
    },
    {
      "epoch": 0.6456590761667407,
      "grad_norm": 0.0012380859116092324,
      "learning_rate": 2.164811949650942e-05,
      "loss": 11.5,
      "step": 1182
    },
    {
      "epoch": 0.6462053190399781,
      "grad_norm": 0.004010463133454323,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 11.5,
      "step": 1183
    },
    {
      "epoch": 0.6467515619132157,
      "grad_norm": 0.002289370633661747,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 11.5,
      "step": 1184
    },
    {
      "epoch": 0.6472978047864532,
      "grad_norm": 0.0024632876738905907,
      "learning_rate": 2.125665222227675e-05,
      "loss": 11.5,
      "step": 1185
    },
    {
      "epoch": 0.6478440476596907,
      "grad_norm": 0.004943888168781996,
      "learning_rate": 2.112686247643024e-05,
      "loss": 11.5,
      "step": 1186
    },
    {
      "epoch": 0.6483902905329282,
      "grad_norm": 0.001005526282824576,
      "learning_rate": 2.09974233663104e-05,
      "loss": 11.5,
      "step": 1187
    },
    {
      "epoch": 0.6489365334061657,
      "grad_norm": 0.0031702059786766768,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 11.5,
      "step": 1188
    },
    {
      "epoch": 0.6494827762794032,
      "grad_norm": 0.002402954502031207,
      "learning_rate": 2.073959935340988e-05,
      "loss": 11.5,
      "step": 1189
    },
    {
      "epoch": 0.6500290191526408,
      "grad_norm": 0.0025990318972617388,
      "learning_rate": 2.06112155968028e-05,
      "loss": 11.5,
      "step": 1190
    },
    {
      "epoch": 0.6505752620258782,
      "grad_norm": 0.0025058884639292955,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 11.5,
      "step": 1191
    },
    {
      "epoch": 0.6511215048991158,
      "grad_norm": 0.0019212894840165973,
      "learning_rate": 2.035550743696468e-05,
      "loss": 11.5,
      "step": 1192
    },
    {
      "epoch": 0.6516677477723533,
      "grad_norm": 0.0030080482829362154,
      "learning_rate": 2.022818417050113e-05,
      "loss": 11.5,
      "step": 1193
    },
    {
      "epoch": 0.6522139906455908,
      "grad_norm": 0.0028067021630704403,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 11.5,
      "step": 1194
    },
    {
      "epoch": 0.6527602335188283,
      "grad_norm": 0.0024176568258553743,
      "learning_rate": 1.99746020946023e-05,
      "loss": 11.5,
      "step": 1195
    },
    {
      "epoch": 0.6533064763920658,
      "grad_norm": 0.009284674189984798,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 11.5,
      "step": 1196
    },
    {
      "epoch": 0.6538527192653033,
      "grad_norm": 0.007358079310506582,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 11.5,
      "step": 1197
    },
    {
      "epoch": 0.6543989621385409,
      "grad_norm": 0.004378767684102058,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 11.5,
      "step": 1198
    },
    {
      "epoch": 0.6549452050117783,
      "grad_norm": 0.004543717484921217,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 11.5,
      "step": 1199
    },
    {
      "epoch": 0.6554914478850159,
      "grad_norm": 0.006057028193026781,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 11.5,
      "step": 1200
    },
    {
      "epoch": 0.6560376907582534,
      "grad_norm": 0.0023776718880981207,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 11.5,
      "step": 1201
    },
    {
      "epoch": 0.6565839336314909,
      "grad_norm": 0.0070073967799544334,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 11.5,
      "step": 1202
    },
    {
      "epoch": 0.6571301765047284,
      "grad_norm": 0.0012188195250928402,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 11.5,
      "step": 1203
    },
    {
      "epoch": 0.657676419377966,
      "grad_norm": 0.00563446618616581,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 11.5,
      "step": 1204
    },
    {
      "epoch": 0.6582226622512034,
      "grad_norm": 0.001230606809258461,
      "learning_rate": 1.872812647978095e-05,
      "loss": 11.5,
      "step": 1205
    },
    {
      "epoch": 0.658768905124441,
      "grad_norm": 0.0020121014676988125,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 11.5,
      "step": 1206
    },
    {
      "epoch": 0.6593151479976784,
      "grad_norm": 0.001542790443636477,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 11.5,
      "step": 1207
    },
    {
      "epoch": 0.659861390870916,
      "grad_norm": 0.006217401474714279,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 11.5,
      "step": 1208
    },
    {
      "epoch": 0.6604076337441535,
      "grad_norm": 0.0017873792676255107,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 11.5,
      "step": 1209
    },
    {
      "epoch": 0.660953876617391,
      "grad_norm": 0.0028278431855142117,
      "learning_rate": 1.811840289816409e-05,
      "loss": 11.5,
      "step": 1210
    },
    {
      "epoch": 0.6615001194906285,
      "grad_norm": 0.0017044382402673364,
      "learning_rate": 1.799754805154603e-05,
      "loss": 11.5,
      "step": 1211
    },
    {
      "epoch": 0.662046362363866,
      "grad_norm": 0.002223801100626588,
      "learning_rate": 1.787705775222802e-05,
      "loss": 11.5,
      "step": 1212
    },
    {
      "epoch": 0.6625926052371035,
      "grad_norm": 0.003799920901656151,
      "learning_rate": 1.775693253585763e-05,
      "loss": 11.5,
      "step": 1213
    },
    {
      "epoch": 0.6631388481103411,
      "grad_norm": 0.0018733303295448422,
      "learning_rate": 1.763717293645939e-05,
      "loss": 11.5,
      "step": 1214
    },
    {
      "epoch": 0.6636850909835785,
      "grad_norm": 0.0033246867824345827,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 11.5,
      "step": 1215
    },
    {
      "epoch": 0.6642313338568161,
      "grad_norm": 0.0015196403255686164,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 11.5,
      "step": 1216
    },
    {
      "epoch": 0.6647775767300536,
      "grad_norm": 0.0034531813580542803,
      "learning_rate": 1.728009315594843e-05,
      "loss": 11.5,
      "step": 1217
    },
    {
      "epoch": 0.6653238196032911,
      "grad_norm": 0.0014630970545113087,
      "learning_rate": 1.716180133214149e-05,
      "loss": 11.5,
      "step": 1218
    },
    {
      "epoch": 0.6658700624765286,
      "grad_norm": 0.001782647566869855,
      "learning_rate": 1.704387777100165e-05,
      "loss": 11.5,
      "step": 1219
    },
    {
      "epoch": 0.6664163053497661,
      "grad_norm": 0.0033411027397960424,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 11.5,
      "step": 1220
    },
    {
      "epoch": 0.6669625482230036,
      "grad_norm": 0.0017641630256548524,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 11.5,
      "step": 1221
    },
    {
      "epoch": 0.6675087910962412,
      "grad_norm": 0.0016836110735312104,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 11.5,
      "step": 1222
    },
    {
      "epoch": 0.6680550339694786,
      "grad_norm": 0.001815141411498189,
      "learning_rate": 1.65758766132467e-05,
      "loss": 11.5,
      "step": 1223
    },
    {
      "epoch": 0.6686012768427162,
      "grad_norm": 0.0071573322638869286,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 11.5,
      "step": 1224
    },
    {
      "epoch": 0.6691475197159537,
      "grad_norm": 0.003100442932918668,
      "learning_rate": 1.634409916254914e-05,
      "loss": 11.5,
      "step": 1225
    },
    {
      "epoch": 0.6696937625891912,
      "grad_norm": 0.002849602373316884,
      "learning_rate": 1.622876802674158e-05,
      "loss": 11.5,
      "step": 1226
    },
    {
      "epoch": 0.6702400054624287,
      "grad_norm": 0.008768129162490368,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 11.5,
      "step": 1227
    },
    {
      "epoch": 0.6707862483356662,
      "grad_norm": 0.0010701478458940983,
      "learning_rate": 1.599922349776738e-05,
      "loss": 11.5,
      "step": 1228
    },
    {
      "epoch": 0.6713324912089038,
      "grad_norm": 0.0024875360541045666,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 11.5,
      "step": 1229
    },
    {
      "epoch": 0.6718787340821413,
      "grad_norm": 0.0019850365351885557,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 11.5,
      "step": 1230
    },
    {
      "epoch": 0.6724249769553788,
      "grad_norm": 0.0021345640998333693,
      "learning_rate": 1.565770870187585e-05,
      "loss": 11.5,
      "step": 1231
    },
    {
      "epoch": 0.6729712198286163,
      "grad_norm": 0.002326137153431773,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 11.5,
      "step": 1232
    },
    {
      "epoch": 0.6735174627018539,
      "grad_norm": 0.0021549456287175417,
      "learning_rate": 1.543190607386861e-05,
      "loss": 11.5,
      "step": 1233
    },
    {
      "epoch": 0.6740637055750913,
      "grad_norm": 0.003295585745945573,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 11.5,
      "step": 1234
    },
    {
      "epoch": 0.6746099484483289,
      "grad_norm": 0.0020066702272742987,
      "learning_rate": 1.520760725631164e-05,
      "loss": 11.5,
      "step": 1235
    },
    {
      "epoch": 0.6751561913215663,
      "grad_norm": 0.0010160175152122974,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 11.5,
      "step": 1236
    },
    {
      "epoch": 0.6757024341948039,
      "grad_norm": 0.0013291870709508657,
      "learning_rate": 1.498481623774115e-05,
      "loss": 11.5,
      "step": 1237
    },
    {
      "epoch": 0.6762486770680414,
      "grad_norm": 0.003264233935624361,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 11.5,
      "step": 1238
    },
    {
      "epoch": 0.6767949199412789,
      "grad_norm": 0.0037207063287496567,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 11.5,
      "step": 1239
    },
    {
      "epoch": 0.6773411628145164,
      "grad_norm": 0.001333065563812852,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 11.5,
      "step": 1240
    },
    {
      "epoch": 0.677887405687754,
      "grad_norm": 0.001744848326779902,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 11.5,
      "step": 1241
    },
    {
      "epoch": 0.6784336485609914,
      "grad_norm": 0.0029564390424638987,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 11.5,
      "step": 1242
    },
    {
      "epoch": 0.678979891434229,
      "grad_norm": 0.0034123205114156008,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 11.5,
      "step": 1243
    },
    {
      "epoch": 0.6795261343074664,
      "grad_norm": 0.003555154660716653,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 11.5,
      "step": 1244
    },
    {
      "epoch": 0.680072377180704,
      "grad_norm": 0.004074727185070515,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 11.5,
      "step": 1245
    },
    {
      "epoch": 0.6806186200539415,
      "grad_norm": 0.004739739466458559,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 11.5,
      "step": 1246
    },
    {
      "epoch": 0.681164862927179,
      "grad_norm": 0.003729431889951229,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 11.5,
      "step": 1247
    },
    {
      "epoch": 0.6817111058004165,
      "grad_norm": 0.004022814333438873,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 11.5,
      "step": 1248
    },
    {
      "epoch": 0.6822573486736541,
      "grad_norm": 0.005161876790225506,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 11.5,
      "step": 1249
    },
    {
      "epoch": 0.6828035915468915,
      "grad_norm": 0.005452809389680624,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 11.5,
      "step": 1250
    },
    {
      "epoch": 0.6833498344201291,
      "grad_norm": 0.0016456928569823503,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 11.5,
      "step": 1251
    },
    {
      "epoch": 0.6838960772933665,
      "grad_norm": 0.0014586971374228597,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 11.5,
      "step": 1252
    },
    {
      "epoch": 0.6844423201666041,
      "grad_norm": 0.0033780052326619625,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 11.5,
      "step": 1253
    },
    {
      "epoch": 0.6849885630398416,
      "grad_norm": 0.005720363464206457,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 11.5,
      "step": 1254
    },
    {
      "epoch": 0.6855348059130791,
      "grad_norm": 0.0010830963728949428,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 11.5,
      "step": 1255
    },
    {
      "epoch": 0.6860810487863166,
      "grad_norm": 0.0019973397720605135,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 11.5,
      "step": 1256
    },
    {
      "epoch": 0.6866272916595542,
      "grad_norm": 0.0020665244664996862,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 11.5,
      "step": 1257
    },
    {
      "epoch": 0.6871735345327916,
      "grad_norm": 0.002097556833177805,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 11.5,
      "step": 1258
    },
    {
      "epoch": 0.6877197774060292,
      "grad_norm": 0.004246579948812723,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 11.5,
      "step": 1259
    },
    {
      "epoch": 0.6882660202792666,
      "grad_norm": 0.001359708490781486,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 11.5,
      "step": 1260
    },
    {
      "epoch": 0.6888122631525042,
      "grad_norm": 0.0016796435229480267,
      "learning_rate": 1.243034735542512e-05,
      "loss": 11.5,
      "step": 1261
    },
    {
      "epoch": 0.6893585060257417,
      "grad_norm": 0.0020303584169596434,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 11.5,
      "step": 1262
    },
    {
      "epoch": 0.6899047488989792,
      "grad_norm": 0.0012736147036775947,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 11.5,
      "step": 1263
    },
    {
      "epoch": 0.6904509917722167,
      "grad_norm": 0.001467080437578261,
      "learning_rate": 1.212667397428342e-05,
      "loss": 11.5,
      "step": 1264
    },
    {
      "epoch": 0.6909972346454543,
      "grad_norm": 0.0026581105776131153,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 11.5,
      "step": 1265
    },
    {
      "epoch": 0.6915434775186917,
      "grad_norm": 0.001862226752564311,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 11.5,
      "step": 1266
    },
    {
      "epoch": 0.6920897203919293,
      "grad_norm": 0.0014995734672993422,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 11.5,
      "step": 1267
    },
    {
      "epoch": 0.6926359632651667,
      "grad_norm": 0.002457674825564027,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 11.5,
      "step": 1268
    },
    {
      "epoch": 0.6931822061384043,
      "grad_norm": 0.0030555473640561104,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 11.5,
      "step": 1269
    },
    {
      "epoch": 0.6937284490116418,
      "grad_norm": 0.0017758248141035438,
      "learning_rate": 1.152988664978556e-05,
      "loss": 11.5,
      "step": 1270
    },
    {
      "epoch": 0.6942746918848793,
      "grad_norm": 0.0028779946733266115,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 11.5,
      "step": 1271
    },
    {
      "epoch": 0.6948209347581168,
      "grad_norm": 0.001942069036886096,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 11.5,
      "step": 1272
    },
    {
      "epoch": 0.6953671776313544,
      "grad_norm": 0.001584741403348744,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 11.5,
      "step": 1273
    },
    {
      "epoch": 0.6959134205045918,
      "grad_norm": 0.0018744285916909575,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 11.5,
      "step": 1274
    },
    {
      "epoch": 0.6964596633778294,
      "grad_norm": 0.0031281188130378723,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 11.5,
      "step": 1275
    },
    {
      "epoch": 0.6970059062510668,
      "grad_norm": 0.005564101040363312,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 11.5,
      "step": 1276
    },
    {
      "epoch": 0.6975521491243044,
      "grad_norm": 0.00928118173032999,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 11.5,
      "step": 1277
    },
    {
      "epoch": 0.6980983919975419,
      "grad_norm": 0.0020212356466799974,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 11.5,
      "step": 1278
    },
    {
      "epoch": 0.6986446348707794,
      "grad_norm": 0.0035546557046473026,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 11.5,
      "step": 1279
    },
    {
      "epoch": 0.6991908777440169,
      "grad_norm": 0.0012950057862326503,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 11.5,
      "step": 1280
    },
    {
      "epoch": 0.6997371206172545,
      "grad_norm": 0.0023903818801045418,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 11.5,
      "step": 1281
    },
    {
      "epoch": 0.7002833634904919,
      "grad_norm": 0.0023539774119853973,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 11.5,
      "step": 1282
    },
    {
      "epoch": 0.7008296063637295,
      "grad_norm": 0.0024071242660284042,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 11.5,
      "step": 1283
    },
    {
      "epoch": 0.7013758492369669,
      "grad_norm": 0.0020267264917492867,
      "learning_rate": 1.019260497108453e-05,
      "loss": 11.5,
      "step": 1284
    },
    {
      "epoch": 0.7019220921102045,
      "grad_norm": 0.0017884878907352686,
      "learning_rate": 1.010006558601274e-05,
      "loss": 11.5,
      "step": 1285
    },
    {
      "epoch": 0.702468334983442,
      "grad_norm": 0.0013069588458165526,
      "learning_rate": 1.000792585701934e-05,
      "loss": 11.5,
      "step": 1286
    },
    {
      "epoch": 0.7030145778566795,
      "grad_norm": 0.002535565523430705,
      "learning_rate": 9.91618619371757e-06,
      "loss": 11.5,
      "step": 1287
    },
    {
      "epoch": 0.703560820729917,
      "grad_norm": 0.0012303863186389208,
      "learning_rate": 9.82484700394215e-06,
      "loss": 11.5,
      "step": 1288
    },
    {
      "epoch": 0.7041070636031546,
      "grad_norm": 0.0023976543452590704,
      "learning_rate": 9.73390869374743e-06,
      "loss": 11.5,
      "step": 1289
    },
    {
      "epoch": 0.704653306476392,
      "grad_norm": 0.0012473637470975518,
      "learning_rate": 9.643371667405698e-06,
      "loss": 11.5,
      "step": 1290
    },
    {
      "epoch": 0.7051995493496296,
      "grad_norm": 0.002648539375513792,
      "learning_rate": 9.553236327405246e-06,
      "loss": 11.5,
      "step": 1291
    },
    {
      "epoch": 0.705745792222867,
      "grad_norm": 0.0024053948000073433,
      "learning_rate": 9.463503074448677e-06,
      "loss": 11.5,
      "step": 1292
    },
    {
      "epoch": 0.7062920350961046,
      "grad_norm": 0.002721801633015275,
      "learning_rate": 9.374172307451068e-06,
      "loss": 11.5,
      "step": 1293
    },
    {
      "epoch": 0.7068382779693421,
      "grad_norm": 0.0042703221552073956,
      "learning_rate": 9.285244423538197e-06,
      "loss": 11.5,
      "step": 1294
    },
    {
      "epoch": 0.7073845208425796,
      "grad_norm": 0.0027522039599716663,
      "learning_rate": 9.196719818044886e-06,
      "loss": 11.5,
      "step": 1295
    },
    {
      "epoch": 0.7079307637158171,
      "grad_norm": 0.001288115861825645,
      "learning_rate": 9.108598884513053e-06,
      "loss": 11.5,
      "step": 1296
    },
    {
      "epoch": 0.7084770065890547,
      "grad_norm": 0.0031574699096381664,
      "learning_rate": 9.020882014690136e-06,
      "loss": 11.5,
      "step": 1297
    },
    {
      "epoch": 0.7090232494622922,
      "grad_norm": 0.0028882541228085756,
      "learning_rate": 8.933569598527247e-06,
      "loss": 11.5,
      "step": 1298
    },
    {
      "epoch": 0.7095694923355297,
      "grad_norm": 0.005438880994915962,
      "learning_rate": 8.846662024177477e-06,
      "loss": 11.5,
      "step": 1299
    },
    {
      "epoch": 0.7101157352087673,
      "grad_norm": 0.005320105701684952,
      "learning_rate": 8.760159677994172e-06,
      "loss": 11.5,
      "step": 1300
    },
    {
      "epoch": 0.7106619780820047,
      "grad_norm": 0.004087544046342373,
      "learning_rate": 8.674062944529216e-06,
      "loss": 11.5,
      "step": 1301
    },
    {
      "epoch": 0.7112082209552423,
      "grad_norm": 0.001674605067819357,
      "learning_rate": 8.588372206531292e-06,
      "loss": 11.5,
      "step": 1302
    },
    {
      "epoch": 0.7117544638284797,
      "grad_norm": 0.0029745162464678288,
      "learning_rate": 8.503087844944213e-06,
      "loss": 11.5,
      "step": 1303
    },
    {
      "epoch": 0.7123007067017173,
      "grad_norm": 0.001884937402792275,
      "learning_rate": 8.418210238905256e-06,
      "loss": 11.5,
      "step": 1304
    },
    {
      "epoch": 0.7128469495749548,
      "grad_norm": 0.00207031168974936,
      "learning_rate": 8.333739765743398e-06,
      "loss": 11.5,
      "step": 1305
    },
    {
      "epoch": 0.7133931924481923,
      "grad_norm": 0.0032965540885925293,
      "learning_rate": 8.249676800977658e-06,
      "loss": 11.5,
      "step": 1306
    },
    {
      "epoch": 0.7139394353214298,
      "grad_norm": 0.00273863784968853,
      "learning_rate": 8.16602171831553e-06,
      "loss": 11.5,
      "step": 1307
    },
    {
      "epoch": 0.7144856781946674,
      "grad_norm": 0.0008363326196558774,
      "learning_rate": 8.082774889651168e-06,
      "loss": 11.5,
      "step": 1308
    },
    {
      "epoch": 0.7150319210679048,
      "grad_norm": 0.003077114699408412,
      "learning_rate": 7.999936685063835e-06,
      "loss": 11.5,
      "step": 1309
    },
    {
      "epoch": 0.7155781639411424,
      "grad_norm": 0.0018011052161455154,
      "learning_rate": 7.91750747281621e-06,
      "loss": 11.5,
      "step": 1310
    },
    {
      "epoch": 0.7161244068143798,
      "grad_norm": 0.003721567802131176,
      "learning_rate": 7.835487619352811e-06,
      "loss": 11.5,
      "step": 1311
    },
    {
      "epoch": 0.7166706496876174,
      "grad_norm": 0.003996845334768295,
      "learning_rate": 7.753877489298244e-06,
      "loss": 11.5,
      "step": 1312
    },
    {
      "epoch": 0.7172168925608549,
      "grad_norm": 0.0011683859629556537,
      "learning_rate": 7.67267744545579e-06,
      "loss": 11.5,
      "step": 1313
    },
    {
      "epoch": 0.7177631354340924,
      "grad_norm": 0.0014867286663502455,
      "learning_rate": 7.591887848805545e-06,
      "loss": 11.5,
      "step": 1314
    },
    {
      "epoch": 0.7183093783073299,
      "grad_norm": 0.001102230278775096,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 11.5,
      "step": 1315
    },
    {
      "epoch": 0.7188556211805675,
      "grad_norm": 0.004628360737115145,
      "learning_rate": 7.431541431877342e-06,
      "loss": 11.5,
      "step": 1316
    },
    {
      "epoch": 0.7194018640538049,
      "grad_norm": 0.0019795906264334917,
      "learning_rate": 7.351985324429933e-06,
      "loss": 11.5,
      "step": 1317
    },
    {
      "epoch": 0.7199481069270425,
      "grad_norm": 0.002365355845540762,
      "learning_rate": 7.272841089832694e-06,
      "loss": 11.5,
      "step": 1318
    },
    {
      "epoch": 0.7204943498002799,
      "grad_norm": 0.0020862857345491648,
      "learning_rate": 7.194109079926514e-06,
      "loss": 11.5,
      "step": 1319
    },
    {
      "epoch": 0.7210405926735175,
      "grad_norm": 0.0019436441361904144,
      "learning_rate": 7.115789644719728e-06,
      "loss": 11.5,
      "step": 1320
    },
    {
      "epoch": 0.721586835546755,
      "grad_norm": 0.0034929795656353235,
      "learning_rate": 7.037883132386547e-06,
      "loss": 11.5,
      "step": 1321
    },
    {
      "epoch": 0.7221330784199925,
      "grad_norm": 0.001760690938681364,
      "learning_rate": 6.960389889265517e-06,
      "loss": 11.5,
      "step": 1322
    },
    {
      "epoch": 0.72267932129323,
      "grad_norm": 0.0011968996841460466,
      "learning_rate": 6.883310259857944e-06,
      "loss": 11.5,
      "step": 1323
    },
    {
      "epoch": 0.7232255641664675,
      "grad_norm": 0.0017747837118804455,
      "learning_rate": 6.806644586826383e-06,
      "loss": 11.5,
      "step": 1324
    },
    {
      "epoch": 0.723771807039705,
      "grad_norm": 0.001513978699222207,
      "learning_rate": 6.730393210993147e-06,
      "loss": 11.5,
      "step": 1325
    },
    {
      "epoch": 0.7243180499129426,
      "grad_norm": 0.0025924539659172297,
      "learning_rate": 6.654556471338746e-06,
      "loss": 11.5,
      "step": 1326
    },
    {
      "epoch": 0.72486429278618,
      "grad_norm": 0.004879363812506199,
      "learning_rate": 6.579134705000412e-06,
      "loss": 11.5,
      "step": 1327
    },
    {
      "epoch": 0.7254105356594176,
      "grad_norm": 0.002875222824513912,
      "learning_rate": 6.504128247270546e-06,
      "loss": 11.5,
      "step": 1328
    },
    {
      "epoch": 0.7259567785326551,
      "grad_norm": 0.00199029128998518,
      "learning_rate": 6.429537431595312e-06,
      "loss": 11.5,
      "step": 1329
    },
    {
      "epoch": 0.7265030214058926,
      "grad_norm": 0.0019077645847573876,
      "learning_rate": 6.355362589573077e-06,
      "loss": 11.5,
      "step": 1330
    },
    {
      "epoch": 0.7270492642791301,
      "grad_norm": 0.003867127699777484,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 11.5,
      "step": 1331
    },
    {
      "epoch": 0.7275955071523676,
      "grad_norm": 0.0024787657894194126,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 11.5,
      "step": 1332
    },
    {
      "epoch": 0.7281417500256051,
      "grad_norm": 0.0018038952257484198,
      "learning_rate": 6.135337193660962e-06,
      "loss": 11.5,
      "step": 1333
    },
    {
      "epoch": 0.7286879928988427,
      "grad_norm": 0.0014144019223749638,
      "learning_rate": 6.062829525227909e-06,
      "loss": 11.5,
      "step": 1334
    },
    {
      "epoch": 0.7292342357720801,
      "grad_norm": 0.002350360155105591,
      "learning_rate": 5.990739460672024e-06,
      "loss": 11.5,
      "step": 1335
    },
    {
      "epoch": 0.7297804786453177,
      "grad_norm": 0.0027264589443802834,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 11.5,
      "step": 1336
    },
    {
      "epoch": 0.7303267215185552,
      "grad_norm": 0.0017241844907402992,
      "learning_rate": 5.84781342325833e-06,
      "loss": 11.5,
      "step": 1337
    },
    {
      "epoch": 0.7308729643917927,
      "grad_norm": 0.011842361651360989,
      "learning_rate": 5.77697808578761e-06,
      "loss": 11.5,
      "step": 1338
    },
    {
      "epoch": 0.7314192072650302,
      "grad_norm": 0.0012703301617875695,
      "learning_rate": 5.706561622965467e-06,
      "loss": 11.5,
      "step": 1339
    },
    {
      "epoch": 0.7319654501382677,
      "grad_norm": 0.004821242764592171,
      "learning_rate": 5.636564347832907e-06,
      "loss": 11.5,
      "step": 1340
    },
    {
      "epoch": 0.7325116930115052,
      "grad_norm": 0.0015547401271760464,
      "learning_rate": 5.566986571567401e-06,
      "loss": 11.5,
      "step": 1341
    },
    {
      "epoch": 0.7330579358847428,
      "grad_norm": 0.00522242859005928,
      "learning_rate": 5.497828603481569e-06,
      "loss": 11.5,
      "step": 1342
    },
    {
      "epoch": 0.7336041787579802,
      "grad_norm": 0.0024683407973498106,
      "learning_rate": 5.429090751021704e-06,
      "loss": 11.5,
      "step": 1343
    },
    {
      "epoch": 0.7341504216312178,
      "grad_norm": 0.003259873017668724,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 11.5,
      "step": 1344
    },
    {
      "epoch": 0.7346966645044553,
      "grad_norm": 0.0033022002317011356,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 11.5,
      "step": 1345
    },
    {
      "epoch": 0.7352429073776928,
      "grad_norm": 0.003327632322907448,
      "learning_rate": 5.225400933837954e-06,
      "loss": 11.5,
      "step": 1346
    },
    {
      "epoch": 0.7357891502509303,
      "grad_norm": 0.004901455715298653,
      "learning_rate": 5.158346580971573e-06,
      "loss": 11.5,
      "step": 1347
    },
    {
      "epoch": 0.7363353931241678,
      "grad_norm": 0.0030267355032265186,
      "learning_rate": 5.091713852920854e-06,
      "loss": 11.5,
      "step": 1348
    },
    {
      "epoch": 0.7368816359974053,
      "grad_norm": 0.003071550512686372,
      "learning_rate": 5.025503045905933e-06,
      "loss": 11.5,
      "step": 1349
    },
    {
      "epoch": 0.7374278788706429,
      "grad_norm": 0.009210243821144104,
      "learning_rate": 4.959714454271369e-06,
      "loss": 11.5,
      "step": 1350
    },
    {
      "epoch": 0.7379741217438803,
      "grad_norm": 0.002266655210405588,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 11.5,
      "step": 1351
    },
    {
      "epoch": 0.7385203646171179,
      "grad_norm": 0.0026669465005397797,
      "learning_rate": 4.829405085134997e-06,
      "loss": 11.5,
      "step": 1352
    },
    {
      "epoch": 0.7390666074903554,
      "grad_norm": 0.0018494194373488426,
      "learning_rate": 4.764884886932086e-06,
      "loss": 11.5,
      "step": 1353
    },
    {
      "epoch": 0.7396128503635929,
      "grad_norm": 0.0018054916290566325,
      "learning_rate": 4.700788062704687e-06,
      "loss": 11.5,
      "step": 1354
    },
    {
      "epoch": 0.7401590932368304,
      "grad_norm": 0.002083915052935481,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 11.5,
      "step": 1355
    },
    {
      "epoch": 0.740705336110068,
      "grad_norm": 0.0020263842307031155,
      "learning_rate": 4.573865674079625e-06,
      "loss": 11.5,
      "step": 1356
    },
    {
      "epoch": 0.7412515789833054,
      "grad_norm": 0.0007900174241513014,
      "learning_rate": 4.511040673923828e-06,
      "loss": 11.5,
      "step": 1357
    },
    {
      "epoch": 0.741797821856543,
      "grad_norm": 0.0016930592246353626,
      "learning_rate": 4.448640176224694e-06,
      "loss": 11.5,
      "step": 1358
    },
    {
      "epoch": 0.7423440647297804,
      "grad_norm": 0.0021725897677242756,
      "learning_rate": 4.386664458387779e-06,
      "loss": 11.5,
      "step": 1359
    },
    {
      "epoch": 0.742890307603018,
      "grad_norm": 0.0012811250053346157,
      "learning_rate": 4.325113795930203e-06,
      "loss": 11.5,
      "step": 1360
    },
    {
      "epoch": 0.7434365504762555,
      "grad_norm": 0.000873478886205703,
      "learning_rate": 4.263988462479484e-06,
      "loss": 11.5,
      "step": 1361
    },
    {
      "epoch": 0.743982793349493,
      "grad_norm": 0.0012780260294675827,
      "learning_rate": 4.203288729772326e-06,
      "loss": 11.5,
      "step": 1362
    },
    {
      "epoch": 0.7445290362227305,
      "grad_norm": 0.0012301707174628973,
      "learning_rate": 4.143014867653383e-06,
      "loss": 11.5,
      "step": 1363
    },
    {
      "epoch": 0.745075279095968,
      "grad_norm": 0.0012873925734311342,
      "learning_rate": 4.083167144074073e-06,
      "loss": 11.5,
      "step": 1364
    },
    {
      "epoch": 0.7456215219692055,
      "grad_norm": 0.001136487117037177,
      "learning_rate": 4.023745825091407e-06,
      "loss": 11.5,
      "step": 1365
    },
    {
      "epoch": 0.7461677648424431,
      "grad_norm": 0.0018665916286408901,
      "learning_rate": 3.964751174866765e-06,
      "loss": 11.5,
      "step": 1366
    },
    {
      "epoch": 0.7467140077156806,
      "grad_norm": 0.0019229714525863528,
      "learning_rate": 3.906183455664725e-06,
      "loss": 11.5,
      "step": 1367
    },
    {
      "epoch": 0.7472602505889181,
      "grad_norm": 0.0014188254717737436,
      "learning_rate": 3.84804292785198e-06,
      "loss": 11.5,
      "step": 1368
    },
    {
      "epoch": 0.7478064934621557,
      "grad_norm": 0.004028795752674341,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 11.5,
      "step": 1369
    },
    {
      "epoch": 0.7483527363353931,
      "grad_norm": 0.0023594757076352835,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 11.5,
      "step": 1370
    },
    {
      "epoch": 0.7488989792086307,
      "grad_norm": 0.009972302243113518,
      "learning_rate": 3.676187067922421e-06,
      "loss": 11.5,
      "step": 1371
    },
    {
      "epoch": 0.7494452220818681,
      "grad_norm": 0.002551661804318428,
      "learning_rate": 3.619757871333973e-06,
      "loss": 11.5,
      "step": 1372
    },
    {
      "epoch": 0.7499914649551057,
      "grad_norm": 0.0025232841726392508,
      "learning_rate": 3.563757139458579e-06,
      "loss": 11.5,
      "step": 1373
    },
    {
      "epoch": 0.7505377078283432,
      "grad_norm": 0.001582469791173935,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 11.5,
      "step": 1374
    },
    {
      "epoch": 0.7510839507015807,
      "grad_norm": 0.0017557555111125112,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 11.5,
      "step": 1375
    },
    {
      "epoch": 0.7516301935748182,
      "grad_norm": 0.003967929165810347,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 11.5,
      "step": 1376
    },
    {
      "epoch": 0.7521764364480558,
      "grad_norm": 0.004053074400871992,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 11.5,
      "step": 1377
    },
    {
      "epoch": 0.7527226793212932,
      "grad_norm": 0.0012636987958103418,
      "learning_rate": 3.290189097472096e-06,
      "loss": 11.5,
      "step": 1378
    },
    {
      "epoch": 0.7532689221945308,
      "grad_norm": 0.002674862276762724,
      "learning_rate": 3.236764315185037e-06,
      "loss": 11.5,
      "step": 1379
    },
    {
      "epoch": 0.7538151650677682,
      "grad_norm": 0.0026120080146938562,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 11.5,
      "step": 1380
    },
    {
      "epoch": 0.7543614079410058,
      "grad_norm": 0.003058419330045581,
      "learning_rate": 3.131205488210409e-06,
      "loss": 11.5,
      "step": 1381
    },
    {
      "epoch": 0.7549076508142433,
      "grad_norm": 0.002422794234007597,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 11.5,
      "step": 1382
    },
    {
      "epoch": 0.7554538936874808,
      "grad_norm": 0.0058337366208434105,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 11.5,
      "step": 1383
    },
    {
      "epoch": 0.7560001365607183,
      "grad_norm": 0.0036612991243600845,
      "learning_rate": 2.976097596905969e-06,
      "loss": 11.5,
      "step": 1384
    },
    {
      "epoch": 0.7565463794339559,
      "grad_norm": 0.0014994586817920208,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 11.5,
      "step": 1385
    },
    {
      "epoch": 0.7570926223071933,
      "grad_norm": 0.0012648140545934439,
      "learning_rate": 2.874848583691714e-06,
      "loss": 11.5,
      "step": 1386
    },
    {
      "epoch": 0.7576388651804309,
      "grad_norm": 0.002087618922814727,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 11.5,
      "step": 1387
    },
    {
      "epoch": 0.7581851080536683,
      "grad_norm": 0.002688927110284567,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 11.5,
      "step": 1388
    },
    {
      "epoch": 0.7587313509269059,
      "grad_norm": 0.0030269331764429808,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 11.5,
      "step": 1389
    },
    {
      "epoch": 0.7592775938001434,
      "grad_norm": 0.0015822380082681775,
      "learning_rate": 2.677533636303964e-06,
      "loss": 11.5,
      "step": 1390
    },
    {
      "epoch": 0.7598238366733809,
      "grad_norm": 0.0023680240847170353,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 11.5,
      "step": 1391
    },
    {
      "epoch": 0.7603700795466184,
      "grad_norm": 0.002223307266831398,
      "learning_rate": 2.581471210832931e-06,
      "loss": 11.5,
      "step": 1392
    },
    {
      "epoch": 0.760916322419856,
      "grad_norm": 0.004201092757284641,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 11.5,
      "step": 1393
    },
    {
      "epoch": 0.7614625652930934,
      "grad_norm": 0.003599813673645258,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 11.5,
      "step": 1394
    },
    {
      "epoch": 0.762008808166331,
      "grad_norm": 0.002943573286756873,
      "learning_rate": 2.440626198044327e-06,
      "loss": 11.5,
      "step": 1395
    },
    {
      "epoch": 0.7625550510395684,
      "grad_norm": 0.0038293672259896994,
      "learning_rate": 2.394544997333437e-06,
      "loss": 11.5,
      "step": 1396
    },
    {
      "epoch": 0.763101293912806,
      "grad_norm": 0.0035623155999928713,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 11.5,
      "step": 1397
    },
    {
      "epoch": 0.7636475367860435,
      "grad_norm": 0.003958249930292368,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 11.5,
      "step": 1398
    },
    {
      "epoch": 0.764193779659281,
      "grad_norm": 0.003409895347431302,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 11.5,
      "step": 1399
    },
    {
      "epoch": 0.7647400225325185,
      "grad_norm": 0.005488795693963766,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 11.5,
      "step": 1400
    },
    {
      "epoch": 0.7652862654057561,
      "grad_norm": 0.002642770064994693,
      "learning_rate": 2.170651695549786e-06,
      "loss": 11.5,
      "step": 1401
    },
    {
      "epoch": 0.7658325082789935,
      "grad_norm": 0.0027073374949395657,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 11.5,
      "step": 1402
    },
    {
      "epoch": 0.7663787511522311,
      "grad_norm": 0.0018250963184982538,
      "learning_rate": 2.084137342557646e-06,
      "loss": 11.5,
      "step": 1403
    },
    {
      "epoch": 0.7669249940254685,
      "grad_norm": 0.0019291441421955824,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 11.5,
      "step": 1404
    },
    {
      "epoch": 0.7674712368987061,
      "grad_norm": 0.0037627327255904675,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 11.5,
      "step": 1405
    },
    {
      "epoch": 0.7680174797719436,
      "grad_norm": 0.0021866937167942524,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 11.5,
      "step": 1406
    },
    {
      "epoch": 0.7685637226451811,
      "grad_norm": 0.001909253653138876,
      "learning_rate": 1.916333635208556e-06,
      "loss": 11.5,
      "step": 1407
    },
    {
      "epoch": 0.7691099655184186,
      "grad_norm": 0.0016098282067105174,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 11.5,
      "step": 1408
    },
    {
      "epoch": 0.7696562083916562,
      "grad_norm": 0.002866283291950822,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 11.5,
      "step": 1409
    },
    {
      "epoch": 0.7702024512648936,
      "grad_norm": 0.0018965203780680895,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 11.5,
      "step": 1410
    },
    {
      "epoch": 0.7707486941381312,
      "grad_norm": 0.004066252615302801,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 11.5,
      "step": 1411
    },
    {
      "epoch": 0.7712949370113686,
      "grad_norm": 0.0016708888579159975,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 11.5,
      "step": 1412
    },
    {
      "epoch": 0.7718411798846062,
      "grad_norm": 0.0012090769596397877,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 11.5,
      "step": 1413
    },
    {
      "epoch": 0.7723874227578437,
      "grad_norm": 0.002874663332477212,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 11.5,
      "step": 1414
    },
    {
      "epoch": 0.7729336656310812,
      "grad_norm": 0.0012420472921803594,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 11.5,
      "step": 1415
    },
    {
      "epoch": 0.7734799085043187,
      "grad_norm": 0.0016741782892495394,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 11.5,
      "step": 1416
    },
    {
      "epoch": 0.7740261513775563,
      "grad_norm": 0.0024936567060649395,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 11.5,
      "step": 1417
    },
    {
      "epoch": 0.7745723942507937,
      "grad_norm": 0.0026969050522893667,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 11.5,
      "step": 1418
    },
    {
      "epoch": 0.7751186371240313,
      "grad_norm": 0.0017574471421539783,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 11.5,
      "step": 1419
    },
    {
      "epoch": 0.7756648799972687,
      "grad_norm": 0.0018026907928287983,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 11.5,
      "step": 1420
    },
    {
      "epoch": 0.7762111228705063,
      "grad_norm": 0.0016529866261407733,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 11.5,
      "step": 1421
    },
    {
      "epoch": 0.7767573657437438,
      "grad_norm": 0.001537367352284491,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 11.5,
      "step": 1422
    },
    {
      "epoch": 0.7773036086169813,
      "grad_norm": 0.0054277097806334496,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 11.5,
      "step": 1423
    },
    {
      "epoch": 0.7778498514902188,
      "grad_norm": 0.007898773066699505,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 11.5,
      "step": 1424
    },
    {
      "epoch": 0.7783960943634564,
      "grad_norm": 0.0021550608798861504,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 11.5,
      "step": 1425
    },
    {
      "epoch": 0.7789423372366938,
      "grad_norm": 0.006450321059674025,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 11.5,
      "step": 1426
    },
    {
      "epoch": 0.7794885801099314,
      "grad_norm": 0.00852119643241167,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 11.5,
      "step": 1427
    },
    {
      "epoch": 0.7800348229831688,
      "grad_norm": 0.00133515486959368,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 11.5,
      "step": 1428
    },
    {
      "epoch": 0.7805810658564064,
      "grad_norm": 0.0024400546681135893,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 11.5,
      "step": 1429
    },
    {
      "epoch": 0.7811273087296439,
      "grad_norm": 0.004377436824142933,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 11.5,
      "step": 1430
    },
    {
      "epoch": 0.7816735516028814,
      "grad_norm": 0.0019139524083584547,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 11.5,
      "step": 1431
    },
    {
      "epoch": 0.7822197944761189,
      "grad_norm": 0.00161735771689564,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 11.5,
      "step": 1432
    },
    {
      "epoch": 0.7827660373493565,
      "grad_norm": 0.0012120873434469104,
      "learning_rate": 9.96149361468457e-07,
      "loss": 11.5,
      "step": 1433
    },
    {
      "epoch": 0.783312280222594,
      "grad_norm": 0.0018187157111242414,
      "learning_rate": 9.66683113169431e-07,
      "loss": 11.5,
      "step": 1434
    },
    {
      "epoch": 0.7838585230958315,
      "grad_norm": 0.0035522240214049816,
      "learning_rate": 9.376571240020227e-07,
      "loss": 11.5,
      "step": 1435
    },
    {
      "epoch": 0.784404765969069,
      "grad_norm": 0.0015070607187226415,
      "learning_rate": 9.090715230031688e-07,
      "loss": 11.5,
      "step": 1436
    },
    {
      "epoch": 0.7849510088423065,
      "grad_norm": 0.002900484949350357,
      "learning_rate": 8.809264372520609e-07,
      "loss": 11.5,
      "step": 1437
    },
    {
      "epoch": 0.7854972517155441,
      "grad_norm": 0.0027923565357923508,
      "learning_rate": 8.532219918695128e-07,
      "loss": 11.5,
      "step": 1438
    },
    {
      "epoch": 0.7860434945887815,
      "grad_norm": 0.0015748300356790423,
      "learning_rate": 8.259583100174606e-07,
      "loss": 11.5,
      "step": 1439
    },
    {
      "epoch": 0.7865897374620191,
      "grad_norm": 0.0015470330836251378,
      "learning_rate": 7.991355128984079e-07,
      "loss": 11.5,
      "step": 1440
    },
    {
      "epoch": 0.7871359803352566,
      "grad_norm": 0.004884340334683657,
      "learning_rate": 7.727537197548707e-07,
      "loss": 11.5,
      "step": 1441
    },
    {
      "epoch": 0.7876822232084941,
      "grad_norm": 0.008151402696967125,
      "learning_rate": 7.468130478688218e-07,
      "loss": 11.5,
      "step": 1442
    },
    {
      "epoch": 0.7882284660817316,
      "grad_norm": 0.002272314392030239,
      "learning_rate": 7.213136125612586e-07,
      "loss": 11.5,
      "step": 1443
    },
    {
      "epoch": 0.7887747089549692,
      "grad_norm": 0.0017025255365297198,
      "learning_rate": 6.962555271915805e-07,
      "loss": 11.5,
      "step": 1444
    },
    {
      "epoch": 0.7893209518282066,
      "grad_norm": 0.0034414255060255527,
      "learning_rate": 6.716389031571568e-07,
      "loss": 11.5,
      "step": 1445
    },
    {
      "epoch": 0.7898671947014442,
      "grad_norm": 0.007159316446632147,
      "learning_rate": 6.474638498928265e-07,
      "loss": 11.5,
      "step": 1446
    },
    {
      "epoch": 0.7904134375746816,
      "grad_norm": 0.006766141392290592,
      "learning_rate": 6.237304748703543e-07,
      "loss": 11.5,
      "step": 1447
    },
    {
      "epoch": 0.7909596804479192,
      "grad_norm": 0.010525074787437916,
      "learning_rate": 6.004388835980423e-07,
      "loss": 11.5,
      "step": 1448
    },
    {
      "epoch": 0.7915059233211567,
      "grad_norm": 0.005451246630400419,
      "learning_rate": 5.77589179620186e-07,
      "loss": 11.5,
      "step": 1449
    },
    {
      "epoch": 0.7920521661943942,
      "grad_norm": 0.013137973845005035,
      "learning_rate": 5.55181464516652e-07,
      "loss": 11.5,
      "step": 1450
    },
    {
      "epoch": 0.7925984090676317,
      "grad_norm": 0.0019559848587960005,
      "learning_rate": 5.332158379024122e-07,
      "loss": 11.5,
      "step": 1451
    },
    {
      "epoch": 0.7931446519408692,
      "grad_norm": 0.008733430877327919,
      "learning_rate": 5.116923974270993e-07,
      "loss": 11.5,
      "step": 1452
    },
    {
      "epoch": 0.7936908948141067,
      "grad_norm": 0.002527567557990551,
      "learning_rate": 4.906112387745965e-07,
      "loss": 11.5,
      "step": 1453
    },
    {
      "epoch": 0.7942371376873443,
      "grad_norm": 0.0024396120570600033,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 11.5,
      "step": 1454
    },
    {
      "epoch": 0.7947833805605817,
      "grad_norm": 0.0027662746142596006,
      "learning_rate": 4.497761398421063e-07,
      "loss": 11.5,
      "step": 1455
    },
    {
      "epoch": 0.7953296234338193,
      "grad_norm": 0.0022224104031920433,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 11.5,
      "step": 1456
    },
    {
      "epoch": 0.7958758663070568,
      "grad_norm": 0.0017662117024883628,
      "learning_rate": 4.107112672446123e-07,
      "loss": 11.5,
      "step": 1457
    },
    {
      "epoch": 0.7964221091802943,
      "grad_norm": 0.001358191017061472,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 11.5,
      "step": 1458
    },
    {
      "epoch": 0.7969683520535318,
      "grad_norm": 0.0024535360280424356,
      "learning_rate": 3.734173156432208e-07,
      "loss": 11.5,
      "step": 1459
    },
    {
      "epoch": 0.7975145949267693,
      "grad_norm": 0.002833048813045025,
      "learning_rate": 3.554346436871581e-07,
      "loss": 11.5,
      "step": 1460
    },
    {
      "epoch": 0.7980608378000068,
      "grad_norm": 0.0010908719850704074,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 11.5,
      "step": 1461
    },
    {
      "epoch": 0.7986070806732444,
      "grad_norm": 0.0018547059735283256,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 11.5,
      "step": 1462
    },
    {
      "epoch": 0.7991533235464818,
      "grad_norm": 0.001917566522024572,
      "learning_rate": 3.041447966064648e-07,
      "loss": 11.5,
      "step": 1463
    },
    {
      "epoch": 0.7996995664197194,
      "grad_norm": 0.0030424429569393396,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 11.5,
      "step": 1464
    },
    {
      "epoch": 0.8002458092929569,
      "grad_norm": 0.005918130744248629,
      "learning_rate": 2.721674609919345e-07,
      "loss": 11.5,
      "step": 1465
    },
    {
      "epoch": 0.8007920521661944,
      "grad_norm": 0.0014608923811465502,
      "learning_rate": 2.568437781080069e-07,
      "loss": 11.5,
      "step": 1466
    },
    {
      "epoch": 0.8013382950394319,
      "grad_norm": 0.0040291957557201385,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 11.5,
      "step": 1467
    },
    {
      "epoch": 0.8018845379126694,
      "grad_norm": 0.0024590615648776293,
      "learning_rate": 2.275267227987321e-07,
      "loss": 11.5,
      "step": 1468
    },
    {
      "epoch": 0.8024307807859069,
      "grad_norm": 0.003163374960422516,
      "learning_rate": 2.135334807042866e-07,
      "loss": 11.5,
      "step": 1469
    },
    {
      "epoch": 0.8029770236591445,
      "grad_norm": 0.00189702317584306,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 11.5,
      "step": 1470
    },
    {
      "epoch": 0.8035232665323819,
      "grad_norm": 0.0016610570019111037,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 11.5,
      "step": 1471
    },
    {
      "epoch": 0.8040695094056195,
      "grad_norm": 0.0025671569164842367,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 11.5,
      "step": 1472
    },
    {
      "epoch": 0.804615752278857,
      "grad_norm": 0.0021813735365867615,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 11.5,
      "step": 1473
    },
    {
      "epoch": 0.8051619951520945,
      "grad_norm": 0.0020996739622205496,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 11.5,
      "step": 1474
    },
    {
      "epoch": 0.805708238025332,
      "grad_norm": 0.00536866532638669,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 11.5,
      "step": 1475
    },
    {
      "epoch": 0.8062544808985695,
      "grad_norm": 0.011212555691599846,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 11.5,
      "step": 1476
    },
    {
      "epoch": 0.806800723771807,
      "grad_norm": 0.002728024497628212,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 11.5,
      "step": 1477
    },
    {
      "epoch": 0.8073469666450446,
      "grad_norm": 0.002138944808393717,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 11.5,
      "step": 1478
    },
    {
      "epoch": 0.807893209518282,
      "grad_norm": 0.0023431165609508753,
      "learning_rate": 9.800874960826933e-08,
      "loss": 11.5,
      "step": 1479
    },
    {
      "epoch": 0.8084394523915196,
      "grad_norm": 0.0023863911628723145,
      "learning_rate": 8.889817534969425e-08,
      "loss": 11.5,
      "step": 1480
    },
    {
      "epoch": 0.808985695264757,
      "grad_norm": 0.001407451112754643,
      "learning_rate": 8.023176247348163e-08,
      "loss": 11.5,
      "step": 1481
    },
    {
      "epoch": 0.8095319381379946,
      "grad_norm": 0.007113595027476549,
      "learning_rate": 7.200954950673522e-08,
      "loss": 11.5,
      "step": 1482
    },
    {
      "epoch": 0.8100781810112321,
      "grad_norm": 0.00435604527592659,
      "learning_rate": 6.423157300184946e-08,
      "loss": 11.5,
      "step": 1483
    },
    {
      "epoch": 0.8106244238844696,
      "grad_norm": 0.013751745223999023,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 11.5,
      "step": 1484
    },
    {
      "epoch": 0.8111706667577071,
      "grad_norm": 0.0026584560982882977,
      "learning_rate": 5.000846571264761e-08,
      "loss": 11.5,
      "step": 1485
    },
    {
      "epoch": 0.8117169096309447,
      "grad_norm": 0.0026074391789734364,
      "learning_rate": 4.35633981580974e-08,
      "loss": 11.5,
      "step": 1486
    },
    {
      "epoch": 0.8122631525041821,
      "grad_norm": 0.002102084457874298,
      "learning_rate": 3.756269352462871e-08,
      "loss": 11.5,
      "step": 1487
    },
    {
      "epoch": 0.8128093953774197,
      "grad_norm": 0.0010915385792031884,
      "learning_rate": 3.20063784888025e-08,
      "loss": 11.5,
      "step": 1488
    },
    {
      "epoch": 0.8133556382506572,
      "grad_norm": 0.005016356706619263,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 11.5,
      "step": 1489
    },
    {
      "epoch": 0.8139018811238947,
      "grad_norm": 0.01016772910952568,
      "learning_rate": 2.222701403818972e-08,
      "loss": 11.5,
      "step": 1490
    },
    {
      "epoch": 0.8144481239971322,
      "grad_norm": 0.004224858712404966,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 11.5,
      "step": 1491
    },
    {
      "epoch": 0.8149943668703697,
      "grad_norm": 0.0016458936734125018,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 11.5,
      "step": 1492
    },
    {
      "epoch": 0.8155406097436072,
      "grad_norm": 0.00637467997148633,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 11.5,
      "step": 1493
    },
    {
      "epoch": 0.8160868526168448,
      "grad_norm": 0.003266803454607725,
      "learning_rate": 8.001914775401798e-09,
      "loss": 11.5,
      "step": 1494
    },
    {
      "epoch": 0.8166330954900822,
      "grad_norm": 0.003603791818022728,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 11.5,
      "step": 1495
    },
    {
      "epoch": 0.8171793383633198,
      "grad_norm": 0.004938117228448391,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 11.5,
      "step": 1496
    },
    {
      "epoch": 0.8177255812365573,
      "grad_norm": 0.010525913909077644,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 11.5,
      "step": 1497
    },
    {
      "epoch": 0.8182718241097948,
      "grad_norm": 0.0025527190882712603,
      "learning_rate": 8.891121819565306e-10,
      "loss": 11.5,
      "step": 1498
    },
    {
      "epoch": 0.8188180669830323,
      "grad_norm": 0.004775170236825943,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 11.5,
      "step": 1499
    },
    {
      "epoch": 0.8193643098562698,
      "grad_norm": 0.0029178583063185215,
      "learning_rate": 0.0,
      "loss": 11.5,
      "step": 1500
    },
    {
      "epoch": 0.8193643098562698,
      "eval_loss": 11.5,
      "eval_runtime": 23.077,
      "eval_samples_per_second": 133.64,
      "eval_steps_per_second": 66.82,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 367686683492352.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}