{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 279,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010752688172043012,
      "grad_norm": 19.659445278083677,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.8801538944244385,
      "step": 1
    },
    {
      "epoch": 0.021505376344086023,
      "grad_norm": 21.183622990197563,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.9106118679046631,
      "step": 2
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 9.680562864297194,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.89518141746521,
      "step": 3
    },
    {
      "epoch": 0.043010752688172046,
      "grad_norm": 8.463557758001873,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.8813375234603882,
      "step": 4
    },
    {
      "epoch": 0.053763440860215055,
      "grad_norm": 8.000585049846563,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.8478757739067078,
      "step": 5
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 6.392760393586216,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7342901825904846,
      "step": 6
    },
    {
      "epoch": 0.07526881720430108,
      "grad_norm": 4.291231505071156,
      "learning_rate": 5e-06,
      "loss": 0.6894474029541016,
      "step": 7
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 3.3545639671049603,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.6144750118255615,
      "step": 8
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 1.3577798099653526,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 0.5370485186576843,
      "step": 9
    },
    {
      "epoch": 0.10752688172043011,
      "grad_norm": 1.2774532761940443,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.5807953476905823,
      "step": 10
    },
    {
      "epoch": 0.11827956989247312,
      "grad_norm": 1.0447472651159857,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.5280188918113708,
      "step": 11
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 1.3301070676426008,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.5441749691963196,
      "step": 12
    },
    {
      "epoch": 0.13978494623655913,
      "grad_norm": 1.0565188353675214,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.5112073421478271,
      "step": 13
    },
    {
      "epoch": 0.15053763440860216,
      "grad_norm": 1.0753192556611026,
      "learning_rate": 1e-05,
      "loss": 0.5184136629104614,
      "step": 14
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 0.8209457367101312,
      "learning_rate": 9.999648647603774e-06,
      "loss": 0.5023149251937866,
      "step": 15
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 0.7282985426170734,
      "learning_rate": 9.998594639794502e-06,
      "loss": 0.48974722623825073,
      "step": 16
    },
    {
      "epoch": 0.1827956989247312,
      "grad_norm": 0.7177574512039909,
      "learning_rate": 9.996838124703448e-06,
      "loss": 0.47443556785583496,
      "step": 17
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 0.7867041042254088,
      "learning_rate": 9.994379349192927e-06,
      "loss": 0.48248669505119324,
      "step": 18
    },
    {
      "epoch": 0.20430107526881722,
      "grad_norm": 0.7192865433337082,
      "learning_rate": 9.991218658821609e-06,
      "loss": 0.46316730976104736,
      "step": 19
    },
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 0.8279557032400371,
      "learning_rate": 9.987356497795944e-06,
      "loss": 0.5312884449958801,
      "step": 20
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 0.7907337063911983,
      "learning_rate": 9.982793408907747e-06,
      "loss": 0.48629915714263916,
      "step": 21
    },
    {
      "epoch": 0.23655913978494625,
      "grad_norm": 0.634234358721607,
      "learning_rate": 9.977530033457906e-06,
      "loss": 0.47711604833602905,
      "step": 22
    },
    {
      "epoch": 0.24731182795698925,
      "grad_norm": 0.7748372497671804,
      "learning_rate": 9.971567111166246e-06,
      "loss": 0.4741020202636719,
      "step": 23
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.6968307669517513,
      "learning_rate": 9.964905480067585e-06,
      "loss": 0.48086652159690857,
      "step": 24
    },
    {
      "epoch": 0.26881720430107525,
      "grad_norm": 0.6205197336090342,
      "learning_rate": 9.957546076393944e-06,
      "loss": 0.45759013295173645,
      "step": 25
    },
    {
      "epoch": 0.27956989247311825,
      "grad_norm": 0.5942106287222008,
      "learning_rate": 9.949489934442966e-06,
      "loss": 0.454744428396225,
      "step": 26
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 0.6374339549026193,
      "learning_rate": 9.940738186432565e-06,
      "loss": 0.4751560091972351,
      "step": 27
    },
    {
      "epoch": 0.3010752688172043,
      "grad_norm": 0.5187971760627564,
      "learning_rate": 9.931292062341793e-06,
      "loss": 0.39658498764038086,
      "step": 28
    },
    {
      "epoch": 0.3118279569892473,
      "grad_norm": 0.5360240389517962,
      "learning_rate": 9.921152889737985e-06,
      "loss": 0.46163371205329895,
      "step": 29
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.5585203893563241,
      "learning_rate": 9.910322093590177e-06,
      "loss": 0.46062612533569336,
      "step": 30
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.5610561072744311,
      "learning_rate": 9.898801196068839e-06,
      "loss": 0.4484734833240509,
      "step": 31
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 0.5533641378752504,
      "learning_rate": 9.886591816331953e-06,
      "loss": 0.41486942768096924,
      "step": 32
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 0.5365723882895206,
      "learning_rate": 9.87369567029745e-06,
      "loss": 0.43502286076545715,
      "step": 33
    },
    {
      "epoch": 0.3655913978494624,
      "grad_norm": 0.5536007189636761,
      "learning_rate": 9.860114570402055e-06,
      "loss": 0.4295029640197754,
      "step": 34
    },
    {
      "epoch": 0.3763440860215054,
      "grad_norm": 0.537839879636324,
      "learning_rate": 9.845850425346563e-06,
      "loss": 0.45029592514038086,
      "step": 35
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 0.5299693550104672,
      "learning_rate": 9.830905239827592e-06,
      "loss": 0.38297146558761597,
      "step": 36
    },
    {
      "epoch": 0.3978494623655914,
      "grad_norm": 0.5451092272235792,
      "learning_rate": 9.815281114255841e-06,
      "loss": 0.4300140142440796,
      "step": 37
    },
    {
      "epoch": 0.40860215053763443,
      "grad_norm": 0.6192689983212899,
      "learning_rate": 9.798980244460892e-06,
      "loss": 0.47214269638061523,
      "step": 38
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 0.5870329353069306,
      "learning_rate": 9.782004921382612e-06,
      "loss": 0.43783068656921387,
      "step": 39
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 0.5903637860392952,
      "learning_rate": 9.764357530749178e-06,
      "loss": 0.4084509611129761,
      "step": 40
    },
    {
      "epoch": 0.44086021505376344,
      "grad_norm": 0.61705449873411,
      "learning_rate": 9.74604055274178e-06,
      "loss": 0.42812418937683105,
      "step": 41
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 1.156325694795339,
      "learning_rate": 9.727056561646067e-06,
      "loss": 0.4196915626525879,
      "step": 42
    },
    {
      "epoch": 0.46236559139784944,
      "grad_norm": 0.580641483078281,
      "learning_rate": 9.707408225490343e-06,
      "loss": 0.4251587986946106,
      "step": 43
    },
    {
      "epoch": 0.4731182795698925,
      "grad_norm": 0.5510691280763641,
      "learning_rate": 9.687098305670606e-06,
      "loss": 0.4205142855644226,
      "step": 44
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 0.5225683335691166,
      "learning_rate": 9.66612965656245e-06,
      "loss": 0.37856176495552063,
      "step": 45
    },
    {
      "epoch": 0.4946236559139785,
      "grad_norm": 0.5740926054795668,
      "learning_rate": 9.644505225119922e-06,
      "loss": 0.4447917342185974,
      "step": 46
    },
    {
      "epoch": 0.5053763440860215,
      "grad_norm": 0.5873250280785997,
      "learning_rate": 9.622228050461345e-06,
      "loss": 0.3922080397605896,
      "step": 47
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.6219913849412524,
      "learning_rate": 9.599301263442194e-06,
      "loss": 0.43804931640625,
      "step": 48
    },
    {
      "epoch": 0.5268817204301075,
      "grad_norm": 0.5613998719760911,
      "learning_rate": 9.575728086215093e-06,
      "loss": 0.39639532566070557,
      "step": 49
    },
    {
      "epoch": 0.5376344086021505,
      "grad_norm": 0.5892845271021349,
      "learning_rate": 9.551511831776966e-06,
      "loss": 0.4323637783527374,
      "step": 50
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 0.6452327069861976,
      "learning_rate": 9.526655903503423e-06,
      "loss": 0.4111916422843933,
      "step": 51
    },
    {
      "epoch": 0.5591397849462365,
      "grad_norm": 0.6139840452181067,
      "learning_rate": 9.501163794670445e-06,
      "loss": 0.41024255752563477,
      "step": 52
    },
    {
      "epoch": 0.5698924731182796,
      "grad_norm": 0.5311229042653062,
      "learning_rate": 9.475039087963443e-06,
      "loss": 0.39047467708587646,
      "step": 53
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 0.5686779350086011,
      "learning_rate": 9.448285454973739e-06,
      "loss": 0.4083505868911743,
      "step": 54
    },
    {
      "epoch": 0.5913978494623656,
      "grad_norm": 0.5795028580373828,
      "learning_rate": 9.420906655682553e-06,
      "loss": 0.41430288553237915,
      "step": 55
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 0.5426200748349074,
      "learning_rate": 9.392906537932582e-06,
      "loss": 0.39692986011505127,
      "step": 56
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 0.5096666206067523,
      "learning_rate": 9.364289036887214e-06,
      "loss": 0.39456355571746826,
      "step": 57
    },
    {
      "epoch": 0.6236559139784946,
      "grad_norm": 0.6214693543974321,
      "learning_rate": 9.335058174477472e-06,
      "loss": 0.43065640330314636,
      "step": 58
    },
    {
      "epoch": 0.6344086021505376,
      "grad_norm": 0.6080775052566749,
      "learning_rate": 9.305218058836778e-06,
      "loss": 0.4233320653438568,
      "step": 59
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.5891380183146641,
      "learning_rate": 9.274772883723587e-06,
      "loss": 0.44017481803894043,
      "step": 60
    },
    {
      "epoch": 0.6559139784946236,
      "grad_norm": 0.5317932432917034,
      "learning_rate": 9.24372692793199e-06,
      "loss": 0.39426931738853455,
      "step": 61
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.6464375491620328,
      "learning_rate": 9.21208455469037e-06,
      "loss": 0.4302670657634735,
      "step": 62
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 0.6191245668003084,
      "learning_rate": 9.179850211048193e-06,
      "loss": 0.3996105492115021,
      "step": 63
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 0.5209850545783036,
      "learning_rate": 9.14702842725101e-06,
      "loss": 0.3926236629486084,
      "step": 64
    },
    {
      "epoch": 0.6989247311827957,
      "grad_norm": 0.5469602060829455,
      "learning_rate": 9.113623816103775e-06,
      "loss": 0.4129854142665863,
      "step": 65
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 0.6436176685575749,
      "learning_rate": 9.079641072322555e-06,
      "loss": 0.4262722432613373,
      "step": 66
    },
    {
      "epoch": 0.7204301075268817,
      "grad_norm": 0.6241121753870925,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.4259442687034607,
      "step": 67
    },
    {
      "epoch": 0.7311827956989247,
      "grad_norm": 0.520706057326722,
      "learning_rate": 9.009960371307798e-06,
      "loss": 0.40881264209747314,
      "step": 68
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 0.5406305590251429,
      "learning_rate": 8.974272207066767e-06,
      "loss": 0.4136165976524353,
      "step": 69
    },
    {
      "epoch": 0.7526881720430108,
      "grad_norm": 0.5589669382622552,
      "learning_rate": 8.938025494800454e-06,
      "loss": 0.3692484498023987,
      "step": 70
    },
    {
      "epoch": 0.7634408602150538,
      "grad_norm": 0.5484586018961858,
      "learning_rate": 8.901225328656543e-06,
      "loss": 0.392721563577652,
      "step": 71
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.5720189126246105,
      "learning_rate": 8.863876880565656e-06,
      "loss": 0.4131432771682739,
      "step": 72
    },
    {
      "epoch": 0.7849462365591398,
      "grad_norm": 0.5475650690180592,
      "learning_rate": 8.825985399514488e-06,
      "loss": 0.39399850368499756,
      "step": 73
    },
    {
      "epoch": 0.7956989247311828,
      "grad_norm": 0.571446157132029,
      "learning_rate": 8.787556210808101e-06,
      "loss": 0.3765556216239929,
      "step": 74
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 0.5096610649520538,
      "learning_rate": 8.748594715321512e-06,
      "loss": 0.38086819648742676,
      "step": 75
    },
    {
      "epoch": 0.8172043010752689,
      "grad_norm": 0.5696584948279285,
      "learning_rate": 8.70910638874064e-06,
      "loss": 0.40993741154670715,
      "step": 76
    },
    {
      "epoch": 0.8279569892473119,
      "grad_norm": 0.5448260281274895,
      "learning_rate": 8.669096780792754e-06,
      "loss": 0.3912901282310486,
      "step": 77
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 0.584810442023009,
      "learning_rate": 8.628571514466502e-06,
      "loss": 0.39722931385040283,
      "step": 78
    },
    {
      "epoch": 0.8494623655913979,
      "grad_norm": 0.5590387314291252,
      "learning_rate": 8.587536285221656e-06,
      "loss": 0.38008394837379456,
      "step": 79
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 0.5199976478034525,
      "learning_rate": 8.545996860188668e-06,
      "loss": 0.3779188394546509,
      "step": 80
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 0.6237910452530793,
      "learning_rate": 8.503959077358143e-06,
      "loss": 0.40831005573272705,
      "step": 81
    },
    {
      "epoch": 0.8817204301075269,
      "grad_norm": 0.5103230716146216,
      "learning_rate": 8.46142884476038e-06,
      "loss": 0.3830574154853821,
      "step": 82
    },
    {
      "epoch": 0.8924731182795699,
      "grad_norm": 0.6022470754794768,
      "learning_rate": 8.418412139635026e-06,
      "loss": 0.3760561943054199,
      "step": 83
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 0.5676501864622416,
      "learning_rate": 8.374915007591053e-06,
      "loss": 0.3981160521507263,
      "step": 84
    },
    {
      "epoch": 0.9139784946236559,
      "grad_norm": 0.62733814659435,
      "learning_rate": 8.330943561757092e-06,
      "loss": 0.38736072182655334,
      "step": 85
    },
    {
      "epoch": 0.9247311827956989,
      "grad_norm": 0.581917413328386,
      "learning_rate": 8.286503981922284e-06,
      "loss": 0.40054696798324585,
      "step": 86
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 0.5069721523225358,
      "learning_rate": 8.241602513667775e-06,
      "loss": 0.36416178941726685,
      "step": 87
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 0.6484709424945745,
      "learning_rate": 8.19624546748895e-06,
      "loss": 0.3703634738922119,
      "step": 88
    },
    {
      "epoch": 0.956989247311828,
      "grad_norm": 0.5102251649420729,
      "learning_rate": 8.150439217908557e-06,
      "loss": 0.39143842458724976,
      "step": 89
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 0.5485517368687335,
      "learning_rate": 8.104190202580811e-06,
      "loss": 0.3922935128211975,
      "step": 90
    },
    {
      "epoch": 0.978494623655914,
      "grad_norm": 0.5680193851368315,
      "learning_rate": 8.057504921386661e-06,
      "loss": 0.3872210383415222,
      "step": 91
    },
    {
      "epoch": 0.989247311827957,
      "grad_norm": 0.5405961708214808,
      "learning_rate": 8.010389935520269e-06,
      "loss": 0.4088854193687439,
      "step": 92
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5234958056904553,
      "learning_rate": 7.962851866566912e-06,
      "loss": 0.3855600953102112,
      "step": 93
    },
    {
      "epoch": 1.010752688172043,
      "grad_norm": 0.5751268630788401,
      "learning_rate": 7.914897395572362e-06,
      "loss": 0.32572415471076965,
      "step": 94
    },
    {
      "epoch": 1.021505376344086,
      "grad_norm": 0.5181295042077261,
      "learning_rate": 7.866533262103937e-06,
      "loss": 0.3046826720237732,
      "step": 95
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 0.5508917328926465,
      "learning_rate": 7.817766263303312e-06,
      "loss": 0.29789605736732483,
      "step": 96
    },
    {
      "epoch": 1.043010752688172,
      "grad_norm": 0.5870437887252276,
      "learning_rate": 7.768603252931243e-06,
      "loss": 0.3265905976295471,
      "step": 97
    },
    {
      "epoch": 1.053763440860215,
      "grad_norm": 0.5133318653975599,
      "learning_rate": 7.719051140404327e-06,
      "loss": 0.31667008996009827,
      "step": 98
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 0.5679147172610466,
      "learning_rate": 7.669116889823955e-06,
      "loss": 0.2714020311832428,
      "step": 99
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 0.6225396396665326,
      "learning_rate": 7.6188075189975644e-06,
      "loss": 0.3139406740665436,
      "step": 100
    },
    {
      "epoch": 1.086021505376344,
      "grad_norm": 0.714553736071244,
      "learning_rate": 7.568130098452352e-06,
      "loss": 0.2961091995239258,
      "step": 101
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 0.523889168005371,
      "learning_rate": 7.517091750441576e-06,
      "loss": 0.27245283126831055,
      "step": 102
    },
    {
      "epoch": 1.10752688172043,
      "grad_norm": 0.639295771779283,
      "learning_rate": 7.465699647943586e-06,
      "loss": 0.29422637820243835,
      "step": 103
    },
    {
      "epoch": 1.118279569892473,
      "grad_norm": 0.6150692376811179,
      "learning_rate": 7.413961013653725e-06,
      "loss": 0.29594993591308594,
      "step": 104
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 0.6901790920161586,
      "learning_rate": 7.361883118969248e-06,
      "loss": 0.3163909912109375,
      "step": 105
    },
    {
      "epoch": 1.139784946236559,
      "grad_norm": 0.5947317629259291,
      "learning_rate": 7.309473282967387e-06,
      "loss": 0.2941242456436157,
      "step": 106
    },
    {
      "epoch": 1.1505376344086022,
      "grad_norm": 0.5873657820382887,
      "learning_rate": 7.256738871376733e-06,
      "loss": 0.29748812317848206,
      "step": 107
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 0.5914901548215277,
      "learning_rate": 7.203687295542032e-06,
      "loss": 0.3174619972705841,
      "step": 108
    },
    {
      "epoch": 1.1720430107526882,
      "grad_norm": 0.5757880846821776,
      "learning_rate": 7.1503260113826035e-06,
      "loss": 0.31319767236709595,
      "step": 109
    },
    {
      "epoch": 1.1827956989247312,
      "grad_norm": 0.7019483863871168,
      "learning_rate": 7.09666251834447e-06,
      "loss": 0.3302695155143738,
      "step": 110
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 0.5613831627968954,
      "learning_rate": 7.042704358346375e-06,
      "loss": 0.30153942108154297,
      "step": 111
    },
    {
      "epoch": 1.2043010752688172,
      "grad_norm": 0.5906296533769152,
      "learning_rate": 6.988459114719849e-06,
      "loss": 0.3321286141872406,
      "step": 112
    },
    {
      "epoch": 1.2150537634408602,
      "grad_norm": 0.6185930436417313,
      "learning_rate": 6.933934411143419e-06,
      "loss": 0.3183019161224365,
      "step": 113
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 0.5223193968800454,
      "learning_rate": 6.879137910571191e-06,
      "loss": 0.28522804379463196,
      "step": 114
    },
    {
      "epoch": 1.2365591397849462,
      "grad_norm": 0.5306097120719284,
      "learning_rate": 6.824077314155877e-06,
      "loss": 0.27248305082321167,
      "step": 115
    },
    {
      "epoch": 1.2473118279569892,
      "grad_norm": 0.5410094711890291,
      "learning_rate": 6.768760360166471e-06,
      "loss": 0.29798364639282227,
      "step": 116
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 0.594316663926518,
      "learning_rate": 6.713194822900707e-06,
      "loss": 0.27839338779449463,
      "step": 117
    },
    {
      "epoch": 1.2688172043010753,
      "grad_norm": 0.6110653082771621,
      "learning_rate": 6.657388511592453e-06,
      "loss": 0.30802613496780396,
      "step": 118
    },
    {
      "epoch": 1.2795698924731183,
      "grad_norm": 0.5260300979877027,
      "learning_rate": 6.601349269314188e-06,
      "loss": 0.29552769660949707,
      "step": 119
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 0.541255561766811,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.28333544731140137,
      "step": 120
    },
    {
      "epoch": 1.3010752688172043,
      "grad_norm": 0.5631658544435217,
      "learning_rate": 6.488603526712391e-06,
      "loss": 0.3072783946990967,
      "step": 121
    },
    {
      "epoch": 1.3118279569892473,
      "grad_norm": 0.5999697642319023,
      "learning_rate": 6.431912871783587e-06,
      "loss": 0.2956047058105469,
      "step": 122
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 0.5550402984265157,
      "learning_rate": 6.3750209744473105e-06,
      "loss": 0.3064526319503784,
      "step": 123
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.6418624402271341,
      "learning_rate": 6.3179358303453386e-06,
      "loss": 0.30682218074798584,
      "step": 124
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 0.5305043074037712,
      "learning_rate": 6.260665462278544e-06,
      "loss": 0.3036462068557739,
      "step": 125
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 0.5732839993822031,
      "learning_rate": 6.203217919079343e-06,
      "loss": 0.32605141401290894,
      "step": 126
    },
    {
      "epoch": 1.3655913978494625,
      "grad_norm": 0.5272514261375651,
      "learning_rate": 6.145601274480521e-06,
      "loss": 0.2800780236721039,
      "step": 127
    },
    {
      "epoch": 1.3763440860215055,
      "grad_norm": 0.5383888711112677,
      "learning_rate": 6.08782362598054e-06,
      "loss": 0.29685452580451965,
      "step": 128
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 0.5260069329849035,
      "learning_rate": 6.029893093705492e-06,
      "loss": 0.30100512504577637,
      "step": 129
    },
    {
      "epoch": 1.3978494623655915,
      "grad_norm": 0.5194176965790988,
      "learning_rate": 5.971817819267914e-06,
      "loss": 0.28969407081604004,
      "step": 130
    },
    {
      "epoch": 1.4086021505376345,
      "grad_norm": 0.5795379145956892,
      "learning_rate": 5.9136059646225375e-06,
      "loss": 0.3088012635707855,
      "step": 131
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 0.5605938731509191,
      "learning_rate": 5.855265710919211e-06,
      "loss": 0.2954477369785309,
      "step": 132
    },
    {
      "epoch": 1.4301075268817205,
      "grad_norm": 0.5513949309182947,
      "learning_rate": 5.796805257353109e-06,
      "loss": 0.31166109442710876,
      "step": 133
    },
    {
      "epoch": 1.4408602150537635,
      "grad_norm": 0.6271394444282732,
      "learning_rate": 5.738232820012407e-06,
      "loss": 0.33265459537506104,
      "step": 134
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 0.5513926189063966,
      "learning_rate": 5.679556630723592e-06,
      "loss": 0.271618515253067,
      "step": 135
    },
    {
      "epoch": 1.4623655913978495,
      "grad_norm": 0.5799088021416959,
      "learning_rate": 5.620784935894548e-06,
      "loss": 0.2902123034000397,
      "step": 136
    },
    {
      "epoch": 1.4731182795698925,
      "grad_norm": 0.5662089313253346,
      "learning_rate": 5.561925995355595e-06,
      "loss": 0.2844197154045105,
      "step": 137
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 0.9937555515678825,
      "learning_rate": 5.5029880811986546e-06,
      "loss": 0.2922922372817993,
      "step": 138
    },
    {
      "epoch": 1.4946236559139785,
      "grad_norm": 0.591368004900944,
      "learning_rate": 5.443979476614674e-06,
      "loss": 0.28497251868247986,
      "step": 139
    },
    {
      "epoch": 1.5053763440860215,
      "grad_norm": 0.6076112891374724,
      "learning_rate": 5.384908474729501e-06,
      "loss": 0.31142377853393555,
      "step": 140
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 0.5765965412966652,
      "learning_rate": 5.325783377438357e-06,
      "loss": 0.3015650510787964,
      "step": 141
    },
    {
      "epoch": 1.5268817204301075,
      "grad_norm": 0.5232918774810716,
      "learning_rate": 5.266612494239088e-06,
      "loss": 0.275716096162796,
      "step": 142
    },
    {
      "epoch": 1.5376344086021505,
      "grad_norm": 0.5680197113944581,
      "learning_rate": 5.207404141064334e-06,
      "loss": 0.2873622179031372,
      "step": 143
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 0.49399262225350316,
      "learning_rate": 5.148166639112799e-06,
      "loss": 0.26688140630722046,
      "step": 144
    },
    {
      "epoch": 1.5591397849462365,
      "grad_norm": 0.5949244178581083,
      "learning_rate": 5.088908313679788e-06,
      "loss": 0.29102981090545654,
      "step": 145
    },
    {
      "epoch": 1.5698924731182795,
      "grad_norm": 0.6289302836737687,
      "learning_rate": 5.029637492987153e-06,
      "loss": 0.2931191027164459,
      "step": 146
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 0.6088833365757095,
      "learning_rate": 4.970362507012848e-06,
      "loss": 0.2844638228416443,
      "step": 147
    },
    {
      "epoch": 1.5913978494623655,
      "grad_norm": 0.6971118150263104,
      "learning_rate": 4.911091686320213e-06,
      "loss": 0.3226025104522705,
      "step": 148
    },
    {
      "epoch": 1.6021505376344085,
      "grad_norm": 0.5612811561354285,
      "learning_rate": 4.8518333608872015e-06,
      "loss": 0.28764593601226807,
      "step": 149
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 0.6065928167636621,
      "learning_rate": 4.792595858935668e-06,
      "loss": 0.3277990221977234,
      "step": 150
    },
    {
      "epoch": 1.6236559139784945,
      "grad_norm": 0.5135483316527438,
      "learning_rate": 4.733387505760913e-06,
      "loss": 0.2962847948074341,
      "step": 151
    },
    {
      "epoch": 1.6344086021505375,
      "grad_norm": 0.6160906809336857,
      "learning_rate": 4.674216622561645e-06,
      "loss": 0.294375479221344,
      "step": 152
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 0.6058840168616271,
      "learning_rate": 4.6150915252705005e-06,
      "loss": 0.3054540157318115,
      "step": 153
    },
    {
      "epoch": 1.6559139784946235,
      "grad_norm": 0.5298336251451402,
      "learning_rate": 4.556020523385326e-06,
      "loss": 0.30477648973464966,
      "step": 154
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.5241054596736249,
      "learning_rate": 4.497011918801347e-06,
      "loss": 0.2855217456817627,
      "step": 155
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 0.48994778885097257,
      "learning_rate": 4.438074004644407e-06,
      "loss": 0.30031388998031616,
      "step": 156
    },
    {
      "epoch": 1.6881720430107527,
      "grad_norm": 0.6034882888193746,
      "learning_rate": 4.379215064105454e-06,
      "loss": 0.3072971999645233,
      "step": 157
    },
    {
      "epoch": 1.6989247311827957,
      "grad_norm": 0.5057395078651564,
      "learning_rate": 4.32044336927641e-06,
      "loss": 0.27074217796325684,
      "step": 158
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 0.548027010828449,
      "learning_rate": 4.261767179987595e-06,
      "loss": 0.28892627358436584,
      "step": 159
    },
    {
      "epoch": 1.7204301075268817,
      "grad_norm": 0.5949799690870136,
      "learning_rate": 4.203194742646893e-06,
      "loss": 0.31755733489990234,
      "step": 160
    },
    {
      "epoch": 1.7311827956989247,
      "grad_norm": 0.5988557766268561,
      "learning_rate": 4.1447342890807905e-06,
      "loss": 0.28983038663864136,
      "step": 161
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 0.5153375977084437,
      "learning_rate": 4.086394035377463e-06,
      "loss": 0.2821354269981384,
      "step": 162
    },
    {
      "epoch": 1.7526881720430108,
      "grad_norm": 0.5758453410329892,
      "learning_rate": 4.028182180732088e-06,
      "loss": 0.26659733057022095,
      "step": 163
    },
    {
      "epoch": 1.7634408602150538,
      "grad_norm": 0.5515188533069812,
      "learning_rate": 3.970106906294509e-06,
      "loss": 0.29241591691970825,
      "step": 164
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 0.5465960078319545,
      "learning_rate": 3.912176374019462e-06,
      "loss": 0.28333449363708496,
      "step": 165
    },
    {
      "epoch": 1.7849462365591398,
      "grad_norm": 0.5696267902501757,
      "learning_rate": 3.85439872551948e-06,
      "loss": 0.28454482555389404,
      "step": 166
    },
    {
      "epoch": 1.7956989247311828,
      "grad_norm": 0.5537225691065368,
      "learning_rate": 3.796782080920659e-06,
      "loss": 0.29573291540145874,
      "step": 167
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 0.5317821119388748,
      "learning_rate": 3.7393345377214584e-06,
      "loss": 0.3006817400455475,
      "step": 168
    },
    {
      "epoch": 1.817204301075269,
      "grad_norm": 0.5193143293546626,
      "learning_rate": 3.682064169654663e-06,
      "loss": 0.28759250044822693,
      "step": 169
    },
    {
      "epoch": 1.827956989247312,
      "grad_norm": 0.5024936169709733,
      "learning_rate": 3.6249790255526916e-06,
      "loss": 0.279812753200531,
      "step": 170
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 0.522744432939167,
      "learning_rate": 3.568087128216414e-06,
      "loss": 0.2699696123600006,
      "step": 171
    },
    {
      "epoch": 1.849462365591398,
      "grad_norm": 0.6740080747223203,
      "learning_rate": 3.511396473287611e-06,
      "loss": 0.3212928771972656,
      "step": 172
    },
    {
      "epoch": 1.860215053763441,
      "grad_norm": 0.5393424030521861,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.28264331817626953,
      "step": 173
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 0.5389635395816308,
      "learning_rate": 3.398650730685813e-06,
      "loss": 0.2982739508152008,
      "step": 174
    },
    {
      "epoch": 1.881720430107527,
      "grad_norm": 0.5527956538218779,
      "learning_rate": 3.3426114884075488e-06,
      "loss": 0.2892792820930481,
      "step": 175
    },
    {
      "epoch": 1.89247311827957,
      "grad_norm": 0.5783362380337232,
      "learning_rate": 3.2868051770992935e-06,
      "loss": 0.2867039442062378,
      "step": 176
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 0.4505107682619091,
      "learning_rate": 3.2312396398335312e-06,
      "loss": 0.2998119592666626,
      "step": 177
    },
    {
      "epoch": 1.913978494623656,
      "grad_norm": 0.544159970587014,
      "learning_rate": 3.175922685844125e-06,
      "loss": 0.2946498394012451,
      "step": 178
    },
    {
      "epoch": 1.924731182795699,
      "grad_norm": 0.5601933382330992,
      "learning_rate": 3.1208620894288105e-06,
      "loss": 0.30848872661590576,
      "step": 179
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 0.5692086999688832,
      "learning_rate": 3.0660655888565827e-06,
      "loss": 0.30292433500289917,
      "step": 180
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 0.5204759182695003,
      "learning_rate": 3.0115408852801535e-06,
      "loss": 0.2992181181907654,
      "step": 181
    },
    {
      "epoch": 1.956989247311828,
      "grad_norm": 0.547974934732699,
      "learning_rate": 2.9572956416536267e-06,
      "loss": 0.3052282929420471,
      "step": 182
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 0.5036888640423901,
      "learning_rate": 2.9033374816555338e-06,
      "loss": 0.27619490027427673,
      "step": 183
    },
    {
      "epoch": 1.978494623655914,
      "grad_norm": 0.5857097319940581,
      "learning_rate": 2.8496739886173994e-06,
      "loss": 0.2969588041305542,
      "step": 184
    },
    {
      "epoch": 1.989247311827957,
      "grad_norm": 0.5125923920870736,
      "learning_rate": 2.7963127044579697e-06,
      "loss": 0.28236857056617737,
      "step": 185
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.5460296926482203,
      "learning_rate": 2.743261128623269e-06,
      "loss": 0.2802872657775879,
      "step": 186
    },
    {
      "epoch": 2.010752688172043,
      "grad_norm": 0.5627765344162555,
      "learning_rate": 2.6905267170326143e-06,
      "loss": 0.2465861439704895,
      "step": 187
    },
    {
      "epoch": 2.021505376344086,
      "grad_norm": 0.5241819547135816,
      "learning_rate": 2.6381168810307536e-06,
      "loss": 0.2593002915382385,
      "step": 188
    },
    {
      "epoch": 2.032258064516129,
      "grad_norm": 0.5146155536713976,
      "learning_rate": 2.5860389863462765e-06,
      "loss": 0.22058740258216858,
      "step": 189
    },
    {
      "epoch": 2.043010752688172,
      "grad_norm": 0.527929699946712,
      "learning_rate": 2.534300352056416e-06,
      "loss": 0.24380052089691162,
      "step": 190
    },
    {
      "epoch": 2.053763440860215,
      "grad_norm": 0.5680981651809573,
      "learning_rate": 2.4829082495584244e-06,
      "loss": 0.2181762158870697,
      "step": 191
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 0.5210266779766828,
      "learning_rate": 2.4318699015476495e-06,
      "loss": 0.21483556926250458,
      "step": 192
    },
    {
      "epoch": 2.075268817204301,
      "grad_norm": 0.5228849523676178,
      "learning_rate": 2.3811924810024385e-06,
      "loss": 0.2233523428440094,
      "step": 193
    },
    {
      "epoch": 2.086021505376344,
      "grad_norm": 1.2592287942902403,
      "learning_rate": 2.330883110176049e-06,
      "loss": 0.21375712752342224,
      "step": 194
    },
    {
      "epoch": 2.096774193548387,
      "grad_norm": 0.5541159570889329,
      "learning_rate": 2.2809488595956746e-06,
      "loss": 0.22336843609809875,
      "step": 195
    },
    {
      "epoch": 2.10752688172043,
      "grad_norm": 0.6275490404272853,
      "learning_rate": 2.2313967470687593e-06,
      "loss": 0.23877255618572235,
      "step": 196
    },
    {
      "epoch": 2.118279569892473,
      "grad_norm": 0.5287134027247969,
      "learning_rate": 2.18223373669669e-06,
      "loss": 0.23572032153606415,
      "step": 197
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 0.6005092115812523,
      "learning_rate": 2.1334667378960642e-06,
      "loss": 0.25271525979042053,
      "step": 198
    },
    {
      "epoch": 2.139784946236559,
      "grad_norm": 0.5729023596023511,
      "learning_rate": 2.0851026044276405e-06,
      "loss": 0.2403365820646286,
      "step": 199
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 0.502960648659189,
      "learning_rate": 2.0371481334330913e-06,
      "loss": 0.22194445133209229,
      "step": 200
    },
    {
      "epoch": 2.161290322580645,
      "grad_norm": 1.0792975178234447,
      "learning_rate": 1.9896100644797316e-06,
      "loss": 0.20418012142181396,
      "step": 201
    },
    {
      "epoch": 2.172043010752688,
      "grad_norm": 0.563900355756192,
      "learning_rate": 1.9424950786133414e-06,
      "loss": 0.2197030484676361,
      "step": 202
    },
    {
      "epoch": 2.182795698924731,
      "grad_norm": 0.6049813205320395,
      "learning_rate": 1.8958097974191909e-06,
      "loss": 0.23065771162509918,
      "step": 203
    },
    {
      "epoch": 2.193548387096774,
      "grad_norm": 0.5693258359321052,
      "learning_rate": 1.8495607820914451e-06,
      "loss": 0.22469733655452728,
      "step": 204
    },
    {
      "epoch": 2.204301075268817,
      "grad_norm": 0.49687778546847566,
      "learning_rate": 1.8037545325110506e-06,
      "loss": 0.22447475790977478,
      "step": 205
    },
    {
      "epoch": 2.21505376344086,
      "grad_norm": 0.6870685608162285,
      "learning_rate": 1.7583974863322272e-06,
      "loss": 0.2232748568058014,
      "step": 206
    },
    {
      "epoch": 2.225806451612903,
      "grad_norm": 0.5587791759635568,
      "learning_rate": 1.7134960180777171e-06,
      "loss": 0.21617698669433594,
      "step": 207
    },
    {
      "epoch": 2.236559139784946,
      "grad_norm": 0.5406726897347297,
      "learning_rate": 1.6690564382429104e-06,
      "loss": 0.22749440371990204,
      "step": 208
    },
    {
      "epoch": 2.247311827956989,
      "grad_norm": 0.6059659457413509,
      "learning_rate": 1.6250849924089485e-06,
      "loss": 0.24092203378677368,
      "step": 209
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 0.5440152722934113,
      "learning_rate": 1.581587860364977e-06,
      "loss": 0.20948372781276703,
      "step": 210
    },
    {
      "epoch": 2.268817204301075,
      "grad_norm": 0.48547134095880384,
      "learning_rate": 1.5385711552396227e-06,
      "loss": 0.21325388550758362,
      "step": 211
    },
    {
      "epoch": 2.279569892473118,
      "grad_norm": 0.6324122951694761,
      "learning_rate": 1.4960409226418576e-06,
      "loss": 0.22965453565120697,
      "step": 212
    },
    {
      "epoch": 2.2903225806451615,
      "grad_norm": 0.5469586332098659,
      "learning_rate": 1.4540031398113335e-06,
      "loss": 0.22315660119056702,
      "step": 213
    },
    {
      "epoch": 2.3010752688172045,
      "grad_norm": 0.7520121411441703,
      "learning_rate": 1.4124637147783431e-06,
      "loss": 0.23476293683052063,
      "step": 214
    },
    {
      "epoch": 2.3118279569892475,
      "grad_norm": 0.535533468892341,
      "learning_rate": 1.371428485533498e-06,
      "loss": 0.21404291689395905,
      "step": 215
    },
    {
      "epoch": 2.3225806451612905,
      "grad_norm": 0.5561238179269572,
      "learning_rate": 1.3309032192072463e-06,
      "loss": 0.23378810286521912,
      "step": 216
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.5663331713554731,
      "learning_rate": 1.29089361125936e-06,
      "loss": 0.2105513960123062,
      "step": 217
    },
    {
      "epoch": 2.3440860215053765,
      "grad_norm": 0.5851084790781899,
      "learning_rate": 1.251405284678488e-06,
      "loss": 0.22501936554908752,
      "step": 218
    },
    {
      "epoch": 2.3548387096774195,
      "grad_norm": 0.5546173616985522,
      "learning_rate": 1.2124437891918995e-06,
      "loss": 0.21492747962474823,
      "step": 219
    },
    {
      "epoch": 2.3655913978494625,
      "grad_norm": 0.5542629930888266,
      "learning_rate": 1.1740146004855141e-06,
      "loss": 0.22757509350776672,
      "step": 220
    },
    {
      "epoch": 2.3763440860215055,
      "grad_norm": 0.5368948762143404,
      "learning_rate": 1.1361231194343436e-06,
      "loss": 0.20776845514774323,
      "step": 221
    },
    {
      "epoch": 2.3870967741935485,
      "grad_norm": 0.5362067849303692,
      "learning_rate": 1.0987746713434578e-06,
      "loss": 0.21643775701522827,
      "step": 222
    },
    {
      "epoch": 2.3978494623655915,
      "grad_norm": 0.7478981160056104,
      "learning_rate": 1.0619745051995473e-06,
      "loss": 0.2321006953716278,
      "step": 223
    },
    {
      "epoch": 2.4086021505376345,
      "grad_norm": 0.5589077284378241,
      "learning_rate": 1.0257277929332332e-06,
      "loss": 0.2335091233253479,
      "step": 224
    },
    {
      "epoch": 2.4193548387096775,
      "grad_norm": 0.5449990060855128,
      "learning_rate": 9.900396286922025e-07,
      "loss": 0.22067061066627502,
      "step": 225
    },
    {
      "epoch": 2.4301075268817205,
      "grad_norm": 0.545790815861183,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.24003136157989502,
      "step": 226
    },
    {
      "epoch": 2.4408602150537635,
      "grad_norm": 0.5197037503280709,
      "learning_rate": 9.203589276774438e-07,
      "loss": 0.2232522964477539,
      "step": 227
    },
    {
      "epoch": 2.4516129032258065,
      "grad_norm": 0.5783224759821376,
      "learning_rate": 8.86376183896226e-07,
      "loss": 0.2168661653995514,
      "step": 228
    },
    {
      "epoch": 2.4623655913978495,
      "grad_norm": 0.5772298663365216,
      "learning_rate": 8.529715727489912e-07,
      "loss": 0.22855976223945618,
      "step": 229
    },
    {
      "epoch": 2.4731182795698925,
      "grad_norm": 0.5901747678986362,
      "learning_rate": 8.201497889518073e-07,
      "loss": 0.23543155193328857,
      "step": 230
    },
    {
      "epoch": 2.4838709677419355,
      "grad_norm": 0.4859350321815666,
      "learning_rate": 7.879154453096305e-07,
      "loss": 0.20444487035274506,
      "step": 231
    },
    {
      "epoch": 2.4946236559139785,
      "grad_norm": 0.5733816209081586,
      "learning_rate": 7.562730720680111e-07,
      "loss": 0.22317257523536682,
      "step": 232
    },
    {
      "epoch": 2.5053763440860215,
      "grad_norm": 0.6158850373560834,
      "learning_rate": 7.25227116276413e-07,
      "loss": 0.21304547786712646,
      "step": 233
    },
    {
      "epoch": 2.5161290322580645,
      "grad_norm": 0.5303929149495401,
      "learning_rate": 6.947819411632223e-07,
      "loss": 0.22398605942726135,
      "step": 234
    },
    {
      "epoch": 2.5268817204301075,
      "grad_norm": 0.5807287890882401,
      "learning_rate": 6.649418255225298e-07,
      "loss": 0.22487413883209229,
      "step": 235
    },
    {
      "epoch": 2.5376344086021505,
      "grad_norm": 0.5151341689499322,
      "learning_rate": 6.357109631127889e-07,
      "loss": 0.20791815221309662,
      "step": 236
    },
    {
      "epoch": 2.5483870967741935,
      "grad_norm": 0.5723989318815612,
      "learning_rate": 6.07093462067419e-07,
      "loss": 0.22853252291679382,
      "step": 237
    },
    {
      "epoch": 2.5591397849462365,
      "grad_norm": 0.5151529413040788,
      "learning_rate": 5.79093344317449e-07,
      "loss": 0.23513737320899963,
      "step": 238
    },
    {
      "epoch": 2.5698924731182795,
      "grad_norm": 0.6125760680997326,
      "learning_rate": 5.517145450262639e-07,
      "loss": 0.23416566848754883,
      "step": 239
    },
    {
      "epoch": 2.5806451612903225,
      "grad_norm": 0.5169337829356161,
      "learning_rate": 5.249609120365579e-07,
      "loss": 0.2236918807029724,
      "step": 240
    },
    {
      "epoch": 2.5913978494623655,
      "grad_norm": 0.5150874552544314,
      "learning_rate": 4.988362053295564e-07,
      "loss": 0.22128576040267944,
      "step": 241
    },
    {
      "epoch": 2.6021505376344085,
      "grad_norm": 0.6254184464054292,
      "learning_rate": 4.733440964965791e-07,
      "loss": 0.23455382883548737,
      "step": 242
    },
    {
      "epoch": 2.6129032258064515,
      "grad_norm": 0.5612944471466359,
      "learning_rate": 4.484881682230341e-07,
      "loss": 0.22176575660705566,
      "step": 243
    },
    {
      "epoch": 2.6236559139784945,
      "grad_norm": 0.5178079276744024,
      "learning_rate": 4.242719137849077e-07,
      "loss": 0.21548646688461304,
      "step": 244
    },
    {
      "epoch": 2.6344086021505375,
      "grad_norm": 0.5659047109155118,
      "learning_rate": 4.00698736557808e-07,
      "loss": 0.23057113587856293,
      "step": 245
    },
    {
      "epoch": 2.6451612903225805,
      "grad_norm": 0.5098023427109231,
      "learning_rate": 3.777719495386567e-07,
      "loss": 0.2156354784965515,
      "step": 246
    },
    {
      "epoch": 2.6559139784946235,
      "grad_norm": 0.49515330466811,
      "learning_rate": 3.5549477488007853e-07,
      "loss": 0.19704020023345947,
      "step": 247
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.5297791040512543,
      "learning_rate": 3.3387034343755063e-07,
      "loss": 0.21541355550289154,
      "step": 248
    },
    {
      "epoch": 2.6774193548387095,
      "grad_norm": 0.5134148705186912,
      "learning_rate": 3.1290169432939556e-07,
      "loss": 0.20781481266021729,
      "step": 249
    },
    {
      "epoch": 2.688172043010753,
      "grad_norm": 0.6591675869543023,
      "learning_rate": 2.925917745096568e-07,
      "loss": 0.23452350497245789,
      "step": 250
    },
    {
      "epoch": 2.698924731182796,
      "grad_norm": 0.5940779586998381,
      "learning_rate": 2.7294343835393366e-07,
      "loss": 0.23721283674240112,
      "step": 251
    },
    {
      "epoch": 2.709677419354839,
      "grad_norm": 0.6904575634426214,
      "learning_rate": 2.539594472582213e-07,
      "loss": 0.23339498043060303,
      "step": 252
    },
    {
      "epoch": 2.720430107526882,
      "grad_norm": 0.549700857643845,
      "learning_rate": 2.3564246925082358e-07,
      "loss": 0.2422715723514557,
      "step": 253
    },
    {
      "epoch": 2.731182795698925,
      "grad_norm": 0.551951066945312,
      "learning_rate": 2.179950786173879e-07,
      "loss": 0.22343403100967407,
      "step": 254
    },
    {
      "epoch": 2.741935483870968,
      "grad_norm": 0.5112230065122543,
      "learning_rate": 2.01019755539108e-07,
      "loss": 0.2022184431552887,
      "step": 255
    },
    {
      "epoch": 2.752688172043011,
      "grad_norm": 0.6011326795775324,
      "learning_rate": 1.8471888574415953e-07,
      "loss": 0.21843938529491425,
      "step": 256
    },
    {
      "epoch": 2.763440860215054,
      "grad_norm": 0.5598013580182175,
      "learning_rate": 1.690947601724091e-07,
      "loss": 0.2313247174024582,
      "step": 257
    },
    {
      "epoch": 2.774193548387097,
      "grad_norm": 0.579260801737671,
      "learning_rate": 1.5414957465343883e-07,
      "loss": 0.23432737588882446,
      "step": 258
    },
    {
      "epoch": 2.78494623655914,
      "grad_norm": 0.6080486926763811,
      "learning_rate": 1.3988542959794627e-07,
      "loss": 0.2501685619354248,
      "step": 259
    },
    {
      "epoch": 2.795698924731183,
      "grad_norm": 0.6231912247704571,
      "learning_rate": 1.2630432970255014e-07,
      "loss": 0.22736617922782898,
      "step": 260
    },
    {
      "epoch": 2.806451612903226,
      "grad_norm": 0.6526633675286432,
      "learning_rate": 1.1340818366804728e-07,
      "loss": 0.2217504233121872,
      "step": 261
    },
    {
      "epoch": 2.817204301075269,
      "grad_norm": 0.5774442194442434,
      "learning_rate": 1.0119880393116177e-07,
      "loss": 0.21880705654621124,
      "step": 262
    },
    {
      "epoch": 2.827956989247312,
      "grad_norm": 0.5510924764674725,
      "learning_rate": 8.967790640982466e-08,
      "loss": 0.23220205307006836,
      "step": 263
    },
    {
      "epoch": 2.838709677419355,
      "grad_norm": 0.5940155308333281,
      "learning_rate": 7.884711026201586e-08,
      "loss": 0.22368629276752472,
      "step": 264
    },
    {
      "epoch": 2.849462365591398,
      "grad_norm": 0.48018958702032566,
      "learning_rate": 6.870793765820783e-08,
      "loss": 0.2191024124622345,
      "step": 265
    },
    {
      "epoch": 2.860215053763441,
      "grad_norm": 0.5348968055346366,
      "learning_rate": 5.92618135674361e-08,
      "loss": 0.2141711413860321,
      "step": 266
    },
    {
      "epoch": 2.870967741935484,
      "grad_norm": 0.5058952998875674,
      "learning_rate": 5.0510065557034526e-08,
      "loss": 0.21273520588874817,
      "step": 267
    },
    {
      "epoch": 2.881720430107527,
      "grad_norm": 0.5748923929140294,
      "learning_rate": 4.245392360605727e-08,
      "loss": 0.2339775264263153,
      "step": 268
    },
    {
      "epoch": 2.89247311827957,
      "grad_norm": 0.543587597117276,
      "learning_rate": 3.5094519932415417e-08,
      "loss": 0.22123411297798157,
      "step": 269
    },
    {
      "epoch": 2.903225806451613,
      "grad_norm": 0.5451715623802244,
      "learning_rate": 2.843288883375539e-08,
      "loss": 0.22225311398506165,
      "step": 270
    },
    {
      "epoch": 2.913978494623656,
      "grad_norm": 0.5639808086322913,
      "learning_rate": 2.2469966542096323e-08,
      "loss": 0.220473051071167,
      "step": 271
    },
    {
      "epoch": 2.924731182795699,
      "grad_norm": 0.5603998497829358,
      "learning_rate": 1.7206591092253642e-08,
      "loss": 0.22855965793132782,
      "step": 272
    },
    {
      "epoch": 2.935483870967742,
      "grad_norm": 0.5700123425140283,
      "learning_rate": 1.264350220405719e-08,
      "loss": 0.226323202252388,
      "step": 273
    },
    {
      "epoch": 2.946236559139785,
      "grad_norm": 0.5193108167577539,
      "learning_rate": 8.781341178393244e-09,
      "loss": 0.21938025951385498,
      "step": 274
    },
    {
      "epoch": 2.956989247311828,
      "grad_norm": 0.6419513279403505,
      "learning_rate": 5.620650807073857e-09,
      "loss": 0.24610912799835205,
      "step": 275
    },
    {
      "epoch": 2.967741935483871,
      "grad_norm": 0.6308682159730912,
      "learning_rate": 3.1618752965534295e-09,
      "loss": 0.22690746188163757,
      "step": 276
    },
    {
      "epoch": 2.978494623655914,
      "grad_norm": 0.5263395659056124,
      "learning_rate": 1.4053602054991954e-09,
      "loss": 0.20472858846187592,
      "step": 277
    },
    {
      "epoch": 2.989247311827957,
      "grad_norm": 0.5329091838420249,
      "learning_rate": 3.513523962256349e-10,
      "loss": 0.2296607792377472,
      "step": 278
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6481688008102051,
      "learning_rate": 0.0,
      "loss": 0.22974838316440582,
      "step": 279
    }
  ],
  "logging_steps": 1,
  "max_steps": 279,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 142925655572480.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}