{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.28835063437139563,
  "eval_steps": 500,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014417531718569781,
      "grad_norm": 106.5,
      "kd_loss": 1.1900196075439453,
      "learning_rate": 4.444444444444445e-07,
      "loss": 1.389,
      "step": 5,
      "student_loss": 0.1422676295042038,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0028835063437139563,
      "grad_norm": 75.5,
      "kd_loss": 1.0197750329971313,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.3585,
      "step": 10,
      "student_loss": 0.20726262032985687,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.004325259515570935,
      "grad_norm": 67.0,
      "kd_loss": 0.8719732165336609,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 1.2979,
      "step": 15,
      "student_loss": 0.07172369956970215,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0057670126874279125,
      "grad_norm": 88.0,
      "kd_loss": 0.7114419341087341,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 1.1904,
      "step": 20,
      "student_loss": 0.13332775235176086,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.00720876585928489,
      "grad_norm": 40.5,
      "kd_loss": 0.6378957033157349,
      "learning_rate": 2.666666666666667e-06,
      "loss": 1.0315,
      "step": 25,
      "student_loss": 0.1825275719165802,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.00865051903114187,
      "grad_norm": 36.0,
      "kd_loss": 0.48297253251075745,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.8442,
      "step": 30,
      "student_loss": 0.16044040024280548,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.010092272202998846,
      "grad_norm": 30.375,
      "kd_loss": 0.4690523147583008,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.7189,
      "step": 35,
      "student_loss": 0.02417413890361786,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.011534025374855825,
      "grad_norm": 10.75,
      "kd_loss": 0.3731468915939331,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.5653,
      "step": 40,
      "student_loss": 0.07280340045690536,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.012975778546712802,
      "grad_norm": 9.25,
      "kd_loss": 0.7187333703041077,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.463,
      "step": 45,
      "student_loss": 0.10900051146745682,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.01441753171856978,
      "grad_norm": 8.875,
      "kd_loss": 0.39285412430763245,
      "learning_rate": 5.444444444444445e-06,
      "loss": 0.4475,
      "step": 50,
      "student_loss": 0.019839322194457054,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.015859284890426758,
      "grad_norm": 11.5,
      "kd_loss": 0.19283825159072876,
      "learning_rate": 6e-06,
      "loss": 0.3823,
      "step": 55,
      "student_loss": 0.0010118820937350392,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.01730103806228374,
      "grad_norm": 12.9375,
      "kd_loss": 0.16523021459579468,
      "learning_rate": 6.555555555555556e-06,
      "loss": 0.3287,
      "step": 60,
      "student_loss": 0.0016304850578308105,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.018742791234140715,
      "grad_norm": 7.34375,
      "kd_loss": 0.14898601174354553,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.3321,
      "step": 65,
      "student_loss": 0.027957376092672348,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.020184544405997693,
      "grad_norm": 6.96875,
      "kd_loss": 0.34543901681900024,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.3072,
      "step": 70,
      "student_loss": 0.013891056180000305,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.02162629757785467,
      "grad_norm": 10.875,
      "kd_loss": 0.11955085396766663,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.2997,
      "step": 75,
      "student_loss": 0.001077663735486567,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.02306805074971165,
      "grad_norm": 8.125,
      "kd_loss": 0.10641466826200485,
      "learning_rate": 8.777777777777778e-06,
      "loss": 0.274,
      "step": 80,
      "student_loss": 0.0010437737219035625,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.024509803921568627,
      "grad_norm": 10.0625,
      "kd_loss": 0.06137283146381378,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.2777,
      "step": 85,
      "student_loss": 0.0004318205174058676,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.025951557093425604,
      "grad_norm": 7.5,
      "kd_loss": 0.3231976330280304,
      "learning_rate": 9.88888888888889e-06,
      "loss": 0.2554,
      "step": 90,
      "student_loss": 0.010668772272765636,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.027393310265282585,
      "grad_norm": 7.8125,
      "kd_loss": 0.2638072371482849,
      "learning_rate": 9.99995337991255e-06,
      "loss": 0.2571,
      "step": 95,
      "student_loss": 0.013187691569328308,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.02883506343713956,
      "grad_norm": 7.15625,
      "kd_loss": 0.08886470645666122,
      "learning_rate": 9.999763987297266e-06,
      "loss": 0.2321,
      "step": 100,
      "student_loss": 0.0006429512286558747,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.03027681660899654,
      "grad_norm": 6.5625,
      "kd_loss": 0.24982710182666779,
      "learning_rate": 9.999428913912876e-06,
      "loss": 0.2416,
      "step": 105,
      "student_loss": 0.0025494606234133244,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.031718569780853516,
      "grad_norm": 5.875,
      "kd_loss": 0.26601096987724304,
      "learning_rate": 9.998948169522587e-06,
      "loss": 0.2443,
      "step": 110,
      "student_loss": 0.036904845386743546,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.03316032295271049,
      "grad_norm": 6.34375,
      "kd_loss": 0.3104342222213745,
      "learning_rate": 9.998321768134101e-06,
      "loss": 0.237,
      "step": 115,
      "student_loss": 0.0034773952793329954,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.03460207612456748,
      "grad_norm": 5.4375,
      "kd_loss": 0.13216911256313324,
      "learning_rate": 9.997549727999212e-06,
      "loss": 0.2046,
      "step": 120,
      "student_loss": 0.0008047327864915133,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.036043829296424454,
      "grad_norm": 10.9375,
      "kd_loss": 0.39640313386917114,
      "learning_rate": 9.996632071613257e-06,
      "loss": 0.2315,
      "step": 125,
      "student_loss": 0.055598437786102295,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.03748558246828143,
      "grad_norm": 7.375,
      "kd_loss": 0.15535855293273926,
      "learning_rate": 9.995568825714479e-06,
      "loss": 0.2439,
      "step": 130,
      "student_loss": 0.0005180384614504874,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.03892733564013841,
      "grad_norm": 8.875,
      "kd_loss": 0.2404610514640808,
      "learning_rate": 9.994360021283235e-06,
      "loss": 0.2227,
      "step": 135,
      "student_loss": 0.0007398177986033261,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.040369088811995385,
      "grad_norm": 6.65625,
      "kd_loss": 0.11759798973798752,
      "learning_rate": 9.993005693541098e-06,
      "loss": 0.2056,
      "step": 140,
      "student_loss": 0.0005030568572692573,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.04181084198385236,
      "grad_norm": 10.1875,
      "kd_loss": 0.495441734790802,
      "learning_rate": 9.991505881949837e-06,
      "loss": 0.2072,
      "step": 145,
      "student_loss": 0.04584295675158501,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.04325259515570934,
      "grad_norm": 7.5625,
      "kd_loss": 0.06494709104299545,
      "learning_rate": 9.989860630210251e-06,
      "loss": 0.208,
      "step": 150,
      "student_loss": 0.0006991412374190986,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.04469434832756632,
      "grad_norm": 6.34375,
      "kd_loss": 0.11540775746107101,
      "learning_rate": 9.98806998626091e-06,
      "loss": 0.2123,
      "step": 155,
      "student_loss": 0.0013254352379590273,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0461361014994233,
      "grad_norm": 7.78125,
      "kd_loss": 0.20968985557556152,
      "learning_rate": 9.98613400227676e-06,
      "loss": 0.2308,
      "step": 160,
      "student_loss": 0.07876866310834885,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.04757785467128028,
      "grad_norm": 5.125,
      "kd_loss": 0.19647686183452606,
      "learning_rate": 9.984052734667586e-06,
      "loss": 0.2057,
      "step": 165,
      "student_loss": 0.00031577967456541955,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.049019607843137254,
      "grad_norm": 6.0625,
      "kd_loss": 0.0746464729309082,
      "learning_rate": 9.981826244076387e-06,
      "loss": 0.2068,
      "step": 170,
      "student_loss": 0.0005107919569127262,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.05046136101499423,
      "grad_norm": 9.5,
      "kd_loss": 0.13447396457195282,
      "learning_rate": 9.979454595377594e-06,
      "loss": 0.1983,
      "step": 175,
      "student_loss": 0.012695364654064178,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.05190311418685121,
      "grad_norm": 7.09375,
      "kd_loss": 0.2970976233482361,
      "learning_rate": 9.97693785767519e-06,
      "loss": 0.1904,
      "step": 180,
      "student_loss": 0.0004823607159778476,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.05334486735870819,
      "grad_norm": 6.5,
      "kd_loss": 0.16425344347953796,
      "learning_rate": 9.9742761043007e-06,
      "loss": 0.2085,
      "step": 185,
      "student_loss": 0.04301092401146889,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.05478662053056517,
      "grad_norm": 7.96875,
      "kd_loss": 0.2780059278011322,
      "learning_rate": 9.971469412811032e-06,
      "loss": 0.1921,
      "step": 190,
      "student_loss": 0.011953173205256462,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.056228373702422146,
      "grad_norm": 5.625,
      "kd_loss": 0.3551367223262787,
      "learning_rate": 9.968517864986247e-06,
      "loss": 0.1926,
      "step": 195,
      "student_loss": 0.0191096980124712,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.05767012687427912,
      "grad_norm": 8.0625,
      "kd_loss": 0.13143359124660492,
      "learning_rate": 9.965421546827153e-06,
      "loss": 0.2061,
      "step": 200,
      "student_loss": 0.0003951227408833802,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0591118800461361,
      "grad_norm": 5.4375,
      "kd_loss": 0.17285136878490448,
      "learning_rate": 9.962180548552812e-06,
      "loss": 0.1878,
      "step": 205,
      "student_loss": 0.04404015466570854,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06055363321799308,
      "grad_norm": 5.8125,
      "kd_loss": 0.21570773422718048,
      "learning_rate": 9.958794964597906e-06,
      "loss": 0.1884,
      "step": 210,
      "student_loss": 0.0004762043245136738,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.061995386389850055,
      "grad_norm": 6.3125,
      "kd_loss": 0.14821867644786835,
      "learning_rate": 9.955264893609986e-06,
      "loss": 0.2088,
      "step": 215,
      "student_loss": 0.00042462380952201784,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06343713956170703,
      "grad_norm": 5.53125,
      "kd_loss": 0.21854154765605927,
      "learning_rate": 9.951590438446597e-06,
      "loss": 0.1834,
      "step": 220,
      "student_loss": 0.034548722207546234,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06487889273356401,
      "grad_norm": 5.375,
      "kd_loss": 0.09108870476484299,
      "learning_rate": 9.947771706172285e-06,
      "loss": 0.1893,
      "step": 225,
      "student_loss": 0.03484133258461952,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06632064590542099,
      "grad_norm": 9.75,
      "kd_loss": 0.23377805948257446,
      "learning_rate": 9.943808808055472e-06,
      "loss": 0.182,
      "step": 230,
      "student_loss": 0.0028649731539189816,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06776239907727798,
      "grad_norm": 7.4375,
      "kd_loss": 0.4445033073425293,
      "learning_rate": 9.93970185956522e-06,
      "loss": 0.1916,
      "step": 235,
      "student_loss": 0.016616689041256905,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.06920415224913495,
      "grad_norm": 8.9375,
      "kd_loss": 0.1569254994392395,
      "learning_rate": 9.93545098036786e-06,
      "loss": 0.1856,
      "step": 240,
      "student_loss": 0.005397707223892212,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07064590542099193,
      "grad_norm": 10.0,
      "kd_loss": 0.05294733867049217,
      "learning_rate": 9.931056294323507e-06,
      "loss": 0.1755,
      "step": 245,
      "student_loss": 0.0006755735375918448,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07208765859284891,
      "grad_norm": 11.9375,
      "kd_loss": 0.06715790927410126,
      "learning_rate": 9.926517929482454e-06,
      "loss": 0.1881,
      "step": 250,
      "student_loss": 0.0005509246839210391,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07352941176470588,
      "grad_norm": 9.75,
      "kd_loss": 0.06327904760837555,
      "learning_rate": 9.921836018081438e-06,
      "loss": 0.1683,
      "step": 255,
      "student_loss": 0.00027642049826681614,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07497116493656286,
      "grad_norm": 7.59375,
      "kd_loss": 0.12199637293815613,
      "learning_rate": 9.917010696539793e-06,
      "loss": 0.2046,
      "step": 260,
      "student_loss": 0.042497601360082626,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07641291810841984,
      "grad_norm": 7.65625,
      "kd_loss": 0.052612222731113434,
      "learning_rate": 9.912042105455462e-06,
      "loss": 0.1824,
      "step": 265,
      "student_loss": 0.00036533031379804015,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07785467128027682,
      "grad_norm": 4.84375,
      "kd_loss": 0.050872161984443665,
      "learning_rate": 9.906930389600917e-06,
      "loss": 0.1905,
      "step": 270,
      "student_loss": 0.0008181569282896817,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.07929642445213379,
      "grad_norm": 7.34375,
      "kd_loss": 0.13403832912445068,
      "learning_rate": 9.901675697918925e-06,
      "loss": 0.1834,
      "step": 275,
      "student_loss": 0.050149545073509216,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08073817762399077,
      "grad_norm": 6.78125,
      "kd_loss": 0.10233883559703827,
      "learning_rate": 9.896278183518216e-06,
      "loss": 0.1848,
      "step": 280,
      "student_loss": 0.0005086572491563857,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08217993079584775,
      "grad_norm": 5.625,
      "kd_loss": 0.09045373648405075,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.2056,
      "step": 285,
      "student_loss": 0.0005186001653783023,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08362168396770472,
      "grad_norm": 10.4375,
      "kd_loss": 0.07550664991140366,
      "learning_rate": 9.885055319798514e-06,
      "loss": 0.1994,
      "step": 290,
      "student_loss": 0.0020688113290816545,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0850634371395617,
      "grad_norm": 5.375,
      "kd_loss": 0.07470615953207016,
      "learning_rate": 9.879230297486034e-06,
      "loss": 0.2036,
      "step": 295,
      "student_loss": 0.0005224258056841791,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08650519031141868,
      "grad_norm": 4.96875,
      "kd_loss": 0.254043310880661,
      "learning_rate": 9.873263106458353e-06,
      "loss": 0.1766,
      "step": 300,
      "student_loss": 0.022904984652996063,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08794694348327567,
      "grad_norm": 8.625,
      "kd_loss": 0.12328250706195831,
      "learning_rate": 9.86715392058467e-06,
      "loss": 0.1687,
      "step": 305,
      "student_loss": 0.0030305401887744665,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.08938869665513265,
      "grad_norm": 8.5625,
      "kd_loss": 0.08438524603843689,
      "learning_rate": 9.860902917871566e-06,
      "loss": 0.179,
      "step": 310,
      "student_loss": 0.0004664761363528669,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09083044982698962,
      "grad_norm": 5.84375,
      "kd_loss": 0.21459946036338806,
      "learning_rate": 9.854510280457819e-06,
      "loss": 0.1756,
      "step": 315,
      "student_loss": 0.00737906526774168,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.0922722029988466,
      "grad_norm": 7.28125,
      "kd_loss": 0.07853322476148605,
      "learning_rate": 9.847976194609083e-06,
      "loss": 0.1793,
      "step": 320,
      "student_loss": 0.0004535326734185219,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09371395617070358,
      "grad_norm": 7.0625,
      "kd_loss": 0.36134564876556396,
      "learning_rate": 9.841300850712479e-06,
      "loss": 0.1574,
      "step": 325,
      "student_loss": 0.0006651441799476743,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09515570934256055,
      "grad_norm": 5.59375,
      "kd_loss": 0.2662656009197235,
      "learning_rate": 9.834484443271035e-06,
      "loss": 0.1744,
      "step": 330,
      "student_loss": 0.0028211227618157864,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09659746251441753,
      "grad_norm": 8.0625,
      "kd_loss": 0.09443436563014984,
      "learning_rate": 9.827527170898019e-06,
      "loss": 0.1489,
      "step": 335,
      "student_loss": 0.0005554864183068275,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09803921568627451,
      "grad_norm": 5.71875,
      "kd_loss": 0.10154180973768234,
      "learning_rate": 9.820429236311158e-06,
      "loss": 0.1851,
      "step": 340,
      "student_loss": 0.009886357001960278,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.09948096885813149,
      "grad_norm": 3.40625,
      "kd_loss": 0.13887977600097656,
      "learning_rate": 9.81319084632673e-06,
      "loss": 0.1744,
      "step": 345,
      "student_loss": 0.006124563980847597,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10092272202998846,
      "grad_norm": 4.71875,
      "kd_loss": 0.2847428023815155,
      "learning_rate": 9.805812211853534e-06,
      "loss": 0.1624,
      "step": 350,
      "student_loss": 0.00044082352542318404,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10236447520184544,
      "grad_norm": 7.0,
      "kd_loss": 0.0902910828590393,
      "learning_rate": 9.798293547886748e-06,
      "loss": 0.1686,
      "step": 355,
      "student_loss": 0.00043379762792028487,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10380622837370242,
      "grad_norm": 7.46875,
      "kd_loss": 0.05925482511520386,
      "learning_rate": 9.790635073501662e-06,
      "loss": 0.1726,
      "step": 360,
      "student_loss": 0.0004483895900193602,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1052479815455594,
      "grad_norm": 9.75,
      "kd_loss": 0.1501902937889099,
      "learning_rate": 9.782837011847294e-06,
      "loss": 0.1935,
      "step": 365,
      "student_loss": 0.04511932656168938,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10668973471741638,
      "grad_norm": 5.5,
      "kd_loss": 0.2575005292892456,
      "learning_rate": 9.774899590139897e-06,
      "loss": 0.1727,
      "step": 370,
      "student_loss": 0.033754754811525345,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10813148788927336,
      "grad_norm": 7.0,
      "kd_loss": 0.1478898823261261,
      "learning_rate": 9.766823039656328e-06,
      "loss": 0.1646,
      "step": 375,
      "student_loss": 0.031900398433208466,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.10957324106113034,
      "grad_norm": 4.0625,
      "kd_loss": 0.05213665962219238,
      "learning_rate": 9.758607595727309e-06,
      "loss": 0.1731,
      "step": 380,
      "student_loss": 0.00042764144018292427,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11101499423298732,
      "grad_norm": 7.4375,
      "kd_loss": 0.1557849943637848,
      "learning_rate": 9.75025349773058e-06,
      "loss": 0.1576,
      "step": 385,
      "student_loss": 0.00038361549377441406,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11245674740484429,
      "grad_norm": 7.34375,
      "kd_loss": 0.04977763444185257,
      "learning_rate": 9.741760989083914e-06,
      "loss": 0.1599,
      "step": 390,
      "student_loss": 0.00024680295609869063,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11389850057670127,
      "grad_norm": 5.5,
      "kd_loss": 0.09320851415395737,
      "learning_rate": 9.73313031723803e-06,
      "loss": 0.1557,
      "step": 395,
      "student_loss": 0.00043702221591956913,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11534025374855825,
      "grad_norm": 4.0625,
      "kd_loss": 0.0984533280134201,
      "learning_rate": 9.724361733669383e-06,
      "loss": 0.1313,
      "step": 400,
      "student_loss": 0.002326528076082468,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11678200692041522,
      "grad_norm": 5.6875,
      "kd_loss": 0.08723900467157364,
      "learning_rate": 9.715455493872827e-06,
      "loss": 0.1661,
      "step": 405,
      "student_loss": 0.011903912760317326,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1182237600922722,
      "grad_norm": 6.84375,
      "kd_loss": 0.20244774222373962,
      "learning_rate": 9.70641185735419e-06,
      "loss": 0.1729,
      "step": 410,
      "student_loss": 0.01531198713928461,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.11966551326412918,
      "grad_norm": 5.90625,
      "kd_loss": 0.08771727979183197,
      "learning_rate": 9.697231087622691e-06,
      "loss": 0.1872,
      "step": 415,
      "student_loss": 0.0038373665884137154,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12110726643598616,
      "grad_norm": 6.03125,
      "kd_loss": 0.23439179360866547,
      "learning_rate": 9.687913452183277e-06,
      "loss": 0.1579,
      "step": 420,
      "student_loss": 0.06521515548229218,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12254901960784313,
      "grad_norm": 10.25,
      "kd_loss": 0.02720467746257782,
      "learning_rate": 9.678459222528825e-06,
      "loss": 0.1542,
      "step": 425,
      "student_loss": 0.0021591742988675833,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12399077277970011,
      "grad_norm": 4.625,
      "kd_loss": 0.26806876063346863,
      "learning_rate": 9.668868674132224e-06,
      "loss": 0.1984,
      "step": 430,
      "student_loss": 0.06153964251279831,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1254325259515571,
      "grad_norm": 5.875,
      "kd_loss": 0.0505138635635376,
      "learning_rate": 9.65914208643836e-06,
      "loss": 0.1399,
      "step": 435,
      "student_loss": 0.0006324865389615297,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12687427912341406,
      "grad_norm": 5.6875,
      "kd_loss": 0.25690194964408875,
      "learning_rate": 9.649279742855964e-06,
      "loss": 0.1766,
      "step": 440,
      "student_loss": 0.014252256602048874,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12831603229527105,
      "grad_norm": 13.9375,
      "kd_loss": 0.16726160049438477,
      "learning_rate": 9.639281930749363e-06,
      "loss": 0.1683,
      "step": 445,
      "student_loss": 0.030716991052031517,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.12975778546712802,
      "grad_norm": 5.875,
      "kd_loss": 0.14118966460227966,
      "learning_rate": 9.629148941430096e-06,
      "loss": 0.1852,
      "step": 450,
      "student_loss": 0.0006644712993875146,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.131199538638985,
      "grad_norm": 5.71875,
      "kd_loss": 0.0657167136669159,
      "learning_rate": 9.618881070148436e-06,
      "loss": 0.1697,
      "step": 455,
      "student_loss": 0.0011592389782890677,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.13264129181084197,
      "grad_norm": 8.625,
      "kd_loss": 0.2250318080186844,
      "learning_rate": 9.608478616084784e-06,
      "loss": 0.1424,
      "step": 460,
      "student_loss": 0.009063606150448322,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.13408304498269896,
      "grad_norm": 4.96875,
      "kd_loss": 0.23261451721191406,
      "learning_rate": 9.597941882340947e-06,
      "loss": 0.19,
      "step": 465,
      "student_loss": 0.00037093457649461925,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.13552479815455595,
      "grad_norm": 5.65625,
      "kd_loss": 0.06862400472164154,
      "learning_rate": 9.587271175931313e-06,
      "loss": 0.1546,
      "step": 470,
      "student_loss": 0.0003174058219883591,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.13696655132641292,
      "grad_norm": 6.9375,
      "kd_loss": 0.1504378765821457,
      "learning_rate": 9.5764668077739e-06,
      "loss": 0.137,
      "step": 475,
      "student_loss": 0.00043892176472581923,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1384083044982699,
      "grad_norm": 9.5,
      "kd_loss": 0.19803652167320251,
      "learning_rate": 9.5655290926813e-06,
      "loss": 0.1684,
      "step": 480,
      "student_loss": 0.03657232224941254,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.13985005767012687,
      "grad_norm": 10.6875,
      "kd_loss": 0.042275670915842056,
      "learning_rate": 9.554458349351507e-06,
      "loss": 0.1598,
      "step": 485,
      "student_loss": 0.0004415225994307548,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14129181084198386,
      "grad_norm": 5.25,
      "kd_loss": 0.08490484207868576,
      "learning_rate": 9.54325490035863e-06,
      "loss": 0.1747,
      "step": 490,
      "student_loss": 0.0003650383441708982,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14273356401384082,
      "grad_norm": 9.125,
      "kd_loss": 0.22001218795776367,
      "learning_rate": 9.531919072143488e-06,
      "loss": 0.1684,
      "step": 495,
      "student_loss": 0.0004516220942605287,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14417531718569782,
      "grad_norm": 7.5625,
      "kd_loss": 0.1156071275472641,
      "learning_rate": 9.520451195004107e-06,
      "loss": 0.1612,
      "step": 500,
      "student_loss": 0.0002619729784782976,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14561707035755478,
      "grad_norm": 7.96875,
      "kd_loss": 0.18259596824645996,
      "learning_rate": 9.508851603086094e-06,
      "loss": 0.1792,
      "step": 505,
      "student_loss": 0.00036563287721946836,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 5.15625,
      "kd_loss": 0.06558777391910553,
      "learning_rate": 9.497120634372894e-06,
      "loss": 0.1732,
      "step": 510,
      "student_loss": 0.0003081738541368395,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14850057670126873,
      "grad_norm": 11.3125,
      "kd_loss": 0.072016641497612,
      "learning_rate": 9.48525863067595e-06,
      "loss": 0.1545,
      "step": 515,
      "student_loss": 0.016884997487068176,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.14994232987312572,
      "grad_norm": 12.3125,
      "kd_loss": 0.24426941573619843,
      "learning_rate": 9.473265937624748e-06,
      "loss": 0.1545,
      "step": 520,
      "student_loss": 0.040650855749845505,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1513840830449827,
      "grad_norm": 10.0625,
      "kd_loss": 0.1018366813659668,
      "learning_rate": 9.461142904656722e-06,
      "loss": 0.1785,
      "step": 525,
      "student_loss": 0.00040851361700333655,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.15282583621683968,
      "grad_norm": 7.0625,
      "kd_loss": 0.10763265192508698,
      "learning_rate": 9.448889885007105e-06,
      "loss": 0.1533,
      "step": 530,
      "student_loss": 0.0003649446298368275,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.15426758938869667,
      "grad_norm": 4.28125,
      "kd_loss": 0.20555651187896729,
      "learning_rate": 9.436507235698613e-06,
      "loss": 0.1602,
      "step": 535,
      "student_loss": 0.0006306799477897584,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.15570934256055363,
      "grad_norm": 7.40625,
      "kd_loss": 0.31379157304763794,
      "learning_rate": 9.423995317531049e-06,
      "loss": 0.1694,
      "step": 540,
      "student_loss": 0.031822286546230316,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.15715109573241062,
      "grad_norm": 4.53125,
      "kd_loss": 0.074574314057827,
      "learning_rate": 9.4113544950708e-06,
      "loss": 0.1485,
      "step": 545,
      "student_loss": 0.001338273286819458,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.15859284890426759,
      "grad_norm": 5.09375,
      "kd_loss": 0.11674503982067108,
      "learning_rate": 9.398585136640195e-06,
      "loss": 0.1392,
      "step": 550,
      "student_loss": 0.00020110022160224617,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16003460207612458,
      "grad_norm": 5.5,
      "kd_loss": 0.38831356167793274,
      "learning_rate": 9.38568761430679e-06,
      "loss": 0.1626,
      "step": 555,
      "student_loss": 0.06176994368433952,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16147635524798154,
      "grad_norm": 5.21875,
      "kd_loss": 0.17765283584594727,
      "learning_rate": 9.372662303872514e-06,
      "loss": 0.1692,
      "step": 560,
      "student_loss": 0.0007789277588017285,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16291810841983853,
      "grad_norm": 6.78125,
      "kd_loss": 0.18269239366054535,
      "learning_rate": 9.359509584862735e-06,
      "loss": 0.156,
      "step": 565,
      "student_loss": 0.029893578961491585,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1643598615916955,
      "grad_norm": 4.96875,
      "kd_loss": 0.13837981224060059,
      "learning_rate": 9.346229840515186e-06,
      "loss": 0.1754,
      "step": 570,
      "student_loss": 0.010819545015692711,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16580161476355249,
      "grad_norm": 6.6875,
      "kd_loss": 0.05965150147676468,
      "learning_rate": 9.332823457768805e-06,
      "loss": 0.1521,
      "step": 575,
      "student_loss": 0.0004287472111172974,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16724336793540945,
      "grad_norm": 5.5,
      "kd_loss": 0.11529169976711273,
      "learning_rate": 9.31929082725246e-06,
      "loss": 0.1443,
      "step": 580,
      "student_loss": 0.0018629044061526656,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.16868512110726644,
      "grad_norm": 6.4375,
      "kd_loss": 0.09178049862384796,
      "learning_rate": 9.305632343273567e-06,
      "loss": 0.1411,
      "step": 585,
      "student_loss": 0.00033143372274935246,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1701268742791234,
      "grad_norm": 7.3125,
      "kd_loss": 0.09623868018388748,
      "learning_rate": 9.291848403806604e-06,
      "loss": 0.135,
      "step": 590,
      "student_loss": 0.0005691334954462945,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1715686274509804,
      "grad_norm": 8.8125,
      "kd_loss": 0.1206836849451065,
      "learning_rate": 9.277939410481507e-06,
      "loss": 0.1615,
      "step": 595,
      "student_loss": 0.00032701450982131064,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.17301038062283736,
      "grad_norm": 10.0,
      "kd_loss": 0.15421448647975922,
      "learning_rate": 9.263905768571977e-06,
      "loss": 0.1599,
      "step": 600,
      "student_loss": 0.043646011501550674,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.17445213379469435,
      "grad_norm": 4.3125,
      "kd_loss": 0.17744074761867523,
      "learning_rate": 9.249747886983664e-06,
      "loss": 0.1579,
      "step": 605,
      "student_loss": 0.11804817616939545,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.17589388696655134,
      "grad_norm": 7.75,
      "kd_loss": 0.21163012087345123,
      "learning_rate": 9.235466178242255e-06,
      "loss": 0.1584,
      "step": 610,
      "student_loss": 0.024440325796604156,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1773356401384083,
      "grad_norm": 5.96875,
      "kd_loss": 0.05635638162493706,
      "learning_rate": 9.221061058481456e-06,
      "loss": 0.1395,
      "step": 615,
      "student_loss": 0.00018186365196015686,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1787773933102653,
      "grad_norm": 4.90625,
      "kd_loss": 0.13311146199703217,
      "learning_rate": 9.206532947430865e-06,
      "loss": 0.1565,
      "step": 620,
      "student_loss": 0.011785196140408516,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.18021914648212226,
      "grad_norm": 6.65625,
      "kd_loss": 0.0831049308180809,
      "learning_rate": 9.191882268403743e-06,
      "loss": 0.1376,
      "step": 625,
      "student_loss": 0.014004797674715519,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.18166089965397925,
      "grad_norm": 7.46875,
      "kd_loss": 0.16148407757282257,
      "learning_rate": 9.177109448284673e-06,
      "loss": 0.1595,
      "step": 630,
      "student_loss": 0.0004211153427604586,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1831026528258362,
      "grad_norm": 6.125,
      "kd_loss": 0.1255534440279007,
      "learning_rate": 9.162214917517139e-06,
      "loss": 0.1768,
      "step": 635,
      "student_loss": 0.0005924052675254643,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1845444059976932,
      "grad_norm": 4.625,
      "kd_loss": 0.10638347268104553,
      "learning_rate": 9.14719911009096e-06,
      "loss": 0.1418,
      "step": 640,
      "student_loss": 0.0003365401062183082,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.18598615916955016,
      "grad_norm": 9.8125,
      "kd_loss": 0.1964510828256607,
      "learning_rate": 9.132062463529666e-06,
      "loss": 0.1397,
      "step": 645,
      "student_loss": 0.0003920219314750284,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.18742791234140715,
      "grad_norm": 4.53125,
      "kd_loss": 0.05526847392320633,
      "learning_rate": 9.11680541887774e-06,
      "loss": 0.1444,
      "step": 650,
      "student_loss": 0.0005510112387128174,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.18886966551326412,
      "grad_norm": 6.6875,
      "kd_loss": 0.15618346631526947,
      "learning_rate": 9.101428420687759e-06,
      "loss": 0.1537,
      "step": 655,
      "student_loss": 0.0003180911298841238,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.1903114186851211,
      "grad_norm": 5.875,
      "kd_loss": 0.10197541862726212,
      "learning_rate": 9.085931917007464e-06,
      "loss": 0.1473,
      "step": 660,
      "student_loss": 0.00045391038293018937,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.19175317185697807,
      "grad_norm": 6.375,
      "kd_loss": 0.25619009137153625,
      "learning_rate": 9.070316359366681e-06,
      "loss": 0.1586,
      "step": 665,
      "student_loss": 0.0004292258818168193,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.19319492502883506,
      "grad_norm": 5.46875,
      "kd_loss": 0.16420753300189972,
      "learning_rate": 9.054582202764175e-06,
      "loss": 0.144,
      "step": 670,
      "student_loss": 0.001433389843441546,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.19463667820069205,
      "grad_norm": 5.84375,
      "kd_loss": 0.07009908556938171,
      "learning_rate": 9.038729905654393e-06,
      "loss": 0.1567,
      "step": 675,
      "student_loss": 0.0005579895805567503,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 6.5,
      "kd_loss": 0.28905627131462097,
      "learning_rate": 9.022759929934109e-06,
      "loss": 0.1337,
      "step": 680,
      "student_loss": 0.11294573545455933,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.197520184544406,
      "grad_norm": 7.0625,
      "kd_loss": 0.15458427369594574,
      "learning_rate": 9.006672740928952e-06,
      "loss": 0.1567,
      "step": 685,
      "student_loss": 0.03980840742588043,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.19896193771626297,
      "grad_norm": 5.9375,
      "kd_loss": 0.37066850066185,
      "learning_rate": 8.990468807379861e-06,
      "loss": 0.1569,
      "step": 690,
      "student_loss": 0.011855090036988258,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20040369088811996,
      "grad_norm": 7.09375,
      "kd_loss": 0.09913132339715958,
      "learning_rate": 8.974148601429426e-06,
      "loss": 0.1661,
      "step": 695,
      "student_loss": 0.004078580066561699,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20184544405997693,
      "grad_norm": 4.71875,
      "kd_loss": 0.07698167860507965,
      "learning_rate": 8.957712598608123e-06,
      "loss": 0.152,
      "step": 700,
      "student_loss": 0.00035562299308367074,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20328719723183392,
      "grad_norm": 9.3125,
      "kd_loss": 0.055959805846214294,
      "learning_rate": 8.941161277820462e-06,
      "loss": 0.1509,
      "step": 705,
      "student_loss": 0.00035906920675188303,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20472895040369088,
      "grad_norm": 7.6875,
      "kd_loss": 0.28692546486854553,
      "learning_rate": 8.924495121331038e-06,
      "loss": 0.1526,
      "step": 710,
      "student_loss": 0.0032384570222347975,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20617070357554787,
      "grad_norm": 5.40625,
      "kd_loss": 0.07432126998901367,
      "learning_rate": 8.907714614750473e-06,
      "loss": 0.1613,
      "step": 715,
      "student_loss": 0.003424170194193721,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20761245674740483,
      "grad_norm": 5.625,
      "kd_loss": 0.08819170296192169,
      "learning_rate": 8.890820247021263e-06,
      "loss": 0.1564,
      "step": 720,
      "student_loss": 0.0003675469197332859,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.20905420991926182,
      "grad_norm": 4.375,
      "kd_loss": 0.04775778204202652,
      "learning_rate": 8.873812510403545e-06,
      "loss": 0.1344,
      "step": 725,
      "student_loss": 0.000303939770674333,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2104959630911188,
      "grad_norm": 6.4375,
      "kd_loss": 0.10333582013845444,
      "learning_rate": 8.85669190046074e-06,
      "loss": 0.1585,
      "step": 730,
      "student_loss": 0.0029086838476359844,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.21193771626297578,
      "grad_norm": 6.4375,
      "kd_loss": 0.10097087919712067,
      "learning_rate": 8.839458916045117e-06,
      "loss": 0.166,
      "step": 735,
      "student_loss": 0.001680055633187294,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.21337946943483277,
      "grad_norm": 4.5,
      "kd_loss": 0.21172094345092773,
      "learning_rate": 8.822114059283267e-06,
      "loss": 0.1458,
      "step": 740,
      "student_loss": 0.003110807156190276,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.21482122260668973,
      "grad_norm": 4.5,
      "kd_loss": 0.12226401269435883,
      "learning_rate": 8.804657835561456e-06,
      "loss": 0.1497,
      "step": 745,
      "student_loss": 0.044773317873477936,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.21626297577854672,
      "grad_norm": 4.90625,
      "kd_loss": 0.4294062852859497,
      "learning_rate": 8.787090753510917e-06,
      "loss": 0.1377,
      "step": 750,
      "student_loss": 0.01304060872644186,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2177047289504037,
      "grad_norm": 5.0625,
      "kd_loss": 0.045374076813459396,
      "learning_rate": 8.76941332499301e-06,
      "loss": 0.1639,
      "step": 755,
      "student_loss": 0.03447943180799484,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.21914648212226068,
      "grad_norm": 4.9375,
      "kd_loss": 0.17385347187519073,
      "learning_rate": 8.751626065084328e-06,
      "loss": 0.1486,
      "step": 760,
      "student_loss": 0.05372428148984909,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.22058823529411764,
      "grad_norm": 6.40625,
      "kd_loss": 0.19620856642723083,
      "learning_rate": 8.733729492061675e-06,
      "loss": 0.1476,
      "step": 765,
      "student_loss": 0.002973183523863554,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.22202998846597463,
      "grad_norm": 6.625,
      "kd_loss": 0.2309700846672058,
      "learning_rate": 8.715724127386971e-06,
      "loss": 0.1584,
      "step": 770,
      "student_loss": 0.0005056107183918357,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2234717416378316,
      "grad_norm": 4.5,
      "kd_loss": 0.17416110634803772,
      "learning_rate": 8.697610495692055e-06,
      "loss": 0.1332,
      "step": 775,
      "student_loss": 0.0009080126183107495,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.22491349480968859,
      "grad_norm": 5.96875,
      "kd_loss": 0.14442984759807587,
      "learning_rate": 8.679389124763396e-06,
      "loss": 0.1744,
      "step": 780,
      "student_loss": 0.00044986195280216634,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.22635524798154555,
      "grad_norm": 3.90625,
      "kd_loss": 0.03409692645072937,
      "learning_rate": 8.661060545526724e-06,
      "loss": 0.1469,
      "step": 785,
      "student_loss": 0.00031174320611171424,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.22779700115340254,
      "grad_norm": 5.875,
      "kd_loss": 0.0466926172375679,
      "learning_rate": 8.64262529203155e-06,
      "loss": 0.126,
      "step": 790,
      "student_loss": 0.008866921067237854,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2292387543252595,
      "grad_norm": 4.65625,
      "kd_loss": 0.0940956249833107,
      "learning_rate": 8.624083901435613e-06,
      "loss": 0.1439,
      "step": 795,
      "student_loss": 0.00031902300543151796,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2306805074971165,
      "grad_norm": 6.5,
      "kd_loss": 0.34910041093826294,
      "learning_rate": 8.60543691398922e-06,
      "loss": 0.1535,
      "step": 800,
      "student_loss": 0.0551641546189785,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.23212226066897348,
      "grad_norm": 6.6875,
      "kd_loss": 0.03965107351541519,
      "learning_rate": 8.586684873019513e-06,
      "loss": 0.1491,
      "step": 805,
      "student_loss": 0.00032568341703154147,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.23356401384083045,
      "grad_norm": 6.1875,
      "kd_loss": 0.11863593012094498,
      "learning_rate": 8.567828324914633e-06,
      "loss": 0.1286,
      "step": 810,
      "student_loss": 0.00037313165375962853,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.23500576701268744,
      "grad_norm": 4.4375,
      "kd_loss": 0.053709935396909714,
      "learning_rate": 8.548867819107803e-06,
      "loss": 0.1463,
      "step": 815,
      "student_loss": 0.0006219578790478408,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2364475201845444,
      "grad_norm": 5.59375,
      "kd_loss": 0.10856286436319351,
      "learning_rate": 8.52980390806131e-06,
      "loss": 0.1392,
      "step": 820,
      "student_loss": 0.00545511394739151,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2378892733564014,
      "grad_norm": 6.78125,
      "kd_loss": 0.13921646773815155,
      "learning_rate": 8.51063714725042e-06,
      "loss": 0.1519,
      "step": 825,
      "student_loss": 0.0005082834395579994,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.23933102652825836,
      "grad_norm": 4.15625,
      "kd_loss": 0.1973619908094406,
      "learning_rate": 8.49136809514719e-06,
      "loss": 0.1542,
      "step": 830,
      "student_loss": 0.007410020101815462,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.24077277970011535,
      "grad_norm": 6.5625,
      "kd_loss": 0.09400839358568192,
      "learning_rate": 8.471997313204183e-06,
      "loss": 0.1306,
      "step": 835,
      "student_loss": 0.01405125018209219,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2422145328719723,
      "grad_norm": 4.65625,
      "kd_loss": 0.10384070873260498,
      "learning_rate": 8.452525365838123e-06,
      "loss": 0.145,
      "step": 840,
      "student_loss": 0.00034653532202355564,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2436562860438293,
      "grad_norm": 3.90625,
      "kd_loss": 0.0835334062576294,
      "learning_rate": 8.432952820413447e-06,
      "loss": 0.1523,
      "step": 845,
      "student_loss": 0.0004449452389962971,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.24509803921568626,
      "grad_norm": 8.0,
      "kd_loss": 0.185490220785141,
      "learning_rate": 8.41328024722577e-06,
      "loss": 0.1493,
      "step": 850,
      "student_loss": 0.010087787173688412,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.24653979238754326,
      "grad_norm": 4.09375,
      "kd_loss": 0.06114596873521805,
      "learning_rate": 8.393508219485263e-06,
      "loss": 0.1445,
      "step": 855,
      "student_loss": 0.0001966914423974231,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.24798154555940022,
      "grad_norm": 6.8125,
      "kd_loss": 0.09195505827665329,
      "learning_rate": 8.373637313299966e-06,
      "loss": 0.1609,
      "step": 860,
      "student_loss": 0.0012263396056368947,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2494232987312572,
      "grad_norm": 4.3125,
      "kd_loss": 0.08556319028139114,
      "learning_rate": 8.353668107658984e-06,
      "loss": 0.1457,
      "step": 865,
      "student_loss": 0.00025935019948519766,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2508650519031142,
      "grad_norm": 4.65625,
      "kd_loss": 0.16815349459648132,
      "learning_rate": 8.333601184415632e-06,
      "loss": 0.1477,
      "step": 870,
      "student_loss": 0.0003948257945012301,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.25230680507497116,
      "grad_norm": 7.3125,
      "kd_loss": 0.04278595373034477,
      "learning_rate": 8.31343712827047e-06,
      "loss": 0.1363,
      "step": 875,
      "student_loss": 0.0002142190933227539,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2537485582468281,
      "grad_norm": 5.625,
      "kd_loss": 0.11660173535346985,
      "learning_rate": 8.293176526754274e-06,
      "loss": 0.1448,
      "step": 880,
      "student_loss": 0.0004427894309628755,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.25519031141868515,
      "grad_norm": 5.40625,
      "kd_loss": 0.3428073525428772,
      "learning_rate": 8.27281997021091e-06,
      "loss": 0.1216,
      "step": 885,
      "student_loss": 0.015522399917244911,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2566320645905421,
      "grad_norm": 6.125,
      "kd_loss": 0.0416768416762352,
      "learning_rate": 8.25236805178014e-06,
      "loss": 0.1405,
      "step": 890,
      "student_loss": 0.000336356955813244,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.25807381776239907,
      "grad_norm": 5.53125,
      "kd_loss": 0.18904848396778107,
      "learning_rate": 8.231821367380335e-06,
      "loss": 0.1546,
      "step": 895,
      "student_loss": 0.0075630671344697475,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.25951557093425603,
      "grad_norm": 7.15625,
      "kd_loss": 0.12547728419303894,
      "learning_rate": 8.211180515691108e-06,
      "loss": 0.146,
      "step": 900,
      "student_loss": 0.03521721065044403,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.26095732410611305,
      "grad_norm": 6.5625,
      "kd_loss": 0.1317462921142578,
      "learning_rate": 8.190446098135883e-06,
      "loss": 0.1288,
      "step": 905,
      "student_loss": 0.0030942773446440697,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.26239907727797,
      "grad_norm": 6.40625,
      "kd_loss": 0.024655191227793694,
      "learning_rate": 8.16961871886435e-06,
      "loss": 0.1313,
      "step": 910,
      "student_loss": 0.00026106953737325966,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.263840830449827,
      "grad_norm": 4.0625,
      "kd_loss": 0.16816310584545135,
      "learning_rate": 8.148698984734887e-06,
      "loss": 0.1178,
      "step": 915,
      "student_loss": 0.0004576483916025609,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.26528258362168394,
      "grad_norm": 7.5,
      "kd_loss": 0.11437579244375229,
      "learning_rate": 8.127687505296852e-06,
      "loss": 0.1448,
      "step": 920,
      "student_loss": 0.014312313869595528,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.26672433679354096,
      "grad_norm": 5.78125,
      "kd_loss": 0.033377278596162796,
      "learning_rate": 8.106584892772844e-06,
      "loss": 0.1286,
      "step": 925,
      "student_loss": 0.0002718448522500694,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2681660899653979,
      "grad_norm": 4.5625,
      "kd_loss": 0.06759815663099289,
      "learning_rate": 8.08539176204085e-06,
      "loss": 0.1349,
      "step": 930,
      "student_loss": 0.004725519567728043,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2696078431372549,
      "grad_norm": 4.59375,
      "kd_loss": 0.06565514206886292,
      "learning_rate": 8.064108730616336e-06,
      "loss": 0.1316,
      "step": 935,
      "student_loss": 0.0022741896100342274,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2710495963091119,
      "grad_norm": 4.8125,
      "kd_loss": 0.22540642321109772,
      "learning_rate": 8.042736418634252e-06,
      "loss": 0.1477,
      "step": 940,
      "student_loss": 0.009700382128357887,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.27249134948096887,
      "grad_norm": 7.25,
      "kd_loss": 0.06281514465808868,
      "learning_rate": 8.02127544883096e-06,
      "loss": 0.1468,
      "step": 945,
      "student_loss": 0.007055414374917746,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.27393310265282583,
      "grad_norm": 15.125,
      "kd_loss": 0.1174190565943718,
      "learning_rate": 7.999726446526095e-06,
      "loss": 0.1342,
      "step": 950,
      "student_loss": 0.00020863223471678793,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2753748558246828,
      "grad_norm": 7.8125,
      "kd_loss": 0.10327854752540588,
      "learning_rate": 7.978090039604342e-06,
      "loss": 0.1317,
      "step": 955,
      "student_loss": 0.0008837580680847168,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2768166089965398,
      "grad_norm": 11.0625,
      "kd_loss": 0.10824739933013916,
      "learning_rate": 7.956366858497138e-06,
      "loss": 0.1331,
      "step": 960,
      "student_loss": 0.00035057624336332083,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2782583621683968,
      "grad_norm": 5.6875,
      "kd_loss": 0.1710205078125,
      "learning_rate": 7.934557536164304e-06,
      "loss": 0.1318,
      "step": 965,
      "student_loss": 0.01099088229238987,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.27970011534025374,
      "grad_norm": 6.40625,
      "kd_loss": 0.08352657407522202,
      "learning_rate": 7.91266270807561e-06,
      "loss": 0.1408,
      "step": 970,
      "student_loss": 0.00031413231045007706,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2811418685121107,
      "grad_norm": 4.0,
      "kd_loss": 0.0860796719789505,
      "learning_rate": 7.890683012192249e-06,
      "loss": 0.1406,
      "step": 975,
      "student_loss": 0.017684049904346466,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2825836216839677,
      "grad_norm": 4.1875,
      "kd_loss": 0.0853007584810257,
      "learning_rate": 7.868619088948246e-06,
      "loss": 0.1386,
      "step": 980,
      "student_loss": 0.00036458708927966654,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2840253748558247,
      "grad_norm": 5.28125,
      "kd_loss": 0.0649181604385376,
      "learning_rate": 7.846471581231814e-06,
      "loss": 0.158,
      "step": 985,
      "student_loss": 0.0002832816680893302,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.28546712802768165,
      "grad_norm": 5.875,
      "kd_loss": 0.07041647285223007,
      "learning_rate": 7.824241134366604e-06,
      "loss": 0.1144,
      "step": 990,
      "student_loss": 0.0019226326839998364,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.2869088811995386,
      "grad_norm": 5.1875,
      "kd_loss": 0.15222607553005219,
      "learning_rate": 7.801928396092908e-06,
      "loss": 0.1402,
      "step": 995,
      "student_loss": 0.0024301602970808744,
      "teacher_loss": 0.0
    },
    {
      "epoch": 0.28835063437139563,
      "grad_norm": 3.875,
      "kd_loss": 0.14844445884227753,
      "learning_rate": 7.779534016548791e-06,
      "loss": 0.1352,
      "step": 1000,
      "student_loss": 0.005971855483949184,
      "teacher_loss": 0.0
    }
  ],
  "logging_steps": 5,
  "max_steps": 3000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}