{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 846,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0035460992907801418,
      "grad_norm": 14.04724696500921,
      "learning_rate": 0.0,
      "loss": 1.8587,
      "step": 1
    },
    {
      "epoch": 0.0070921985815602835,
      "grad_norm": 14.602415786818106,
      "learning_rate": 2.3529411764705883e-07,
      "loss": 1.9007,
      "step": 2
    },
    {
      "epoch": 0.010638297872340425,
      "grad_norm": 14.76042346081964,
      "learning_rate": 4.7058823529411767e-07,
      "loss": 1.9022,
      "step": 3
    },
    {
      "epoch": 0.014184397163120567,
      "grad_norm": 14.481514822839404,
      "learning_rate": 7.058823529411766e-07,
      "loss": 1.9044,
      "step": 4
    },
    {
      "epoch": 0.01773049645390071,
      "grad_norm": 14.264365275251663,
      "learning_rate": 9.411764705882353e-07,
      "loss": 1.9279,
      "step": 5
    },
    {
      "epoch": 0.02127659574468085,
      "grad_norm": 13.7952134429284,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 1.9164,
      "step": 6
    },
    {
      "epoch": 0.024822695035460994,
      "grad_norm": 14.28102655974937,
      "learning_rate": 1.4117647058823531e-06,
      "loss": 1.9369,
      "step": 7
    },
    {
      "epoch": 0.028368794326241134,
      "grad_norm": 12.30872463822712,
      "learning_rate": 1.6470588235294118e-06,
      "loss": 1.8994,
      "step": 8
    },
    {
      "epoch": 0.031914893617021274,
      "grad_norm": 11.611863485631176,
      "learning_rate": 1.8823529411764707e-06,
      "loss": 1.845,
      "step": 9
    },
    {
      "epoch": 0.03546099290780142,
      "grad_norm": 8.00669314869585,
      "learning_rate": 2.1176470588235296e-06,
      "loss": 1.7459,
      "step": 10
    },
    {
      "epoch": 0.03900709219858156,
      "grad_norm": 7.053422475492129,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 1.6914,
      "step": 11
    },
    {
      "epoch": 0.0425531914893617,
      "grad_norm": 6.82884339451663,
      "learning_rate": 2.5882352941176473e-06,
      "loss": 1.7283,
      "step": 12
    },
    {
      "epoch": 0.04609929078014184,
      "grad_norm": 3.283572533292923,
      "learning_rate": 2.8235294117647062e-06,
      "loss": 1.6065,
      "step": 13
    },
    {
      "epoch": 0.04964539007092199,
      "grad_norm": 2.685692798754663,
      "learning_rate": 3.058823529411765e-06,
      "loss": 1.5898,
      "step": 14
    },
    {
      "epoch": 0.05319148936170213,
      "grad_norm": 2.5132795524411335,
      "learning_rate": 3.2941176470588236e-06,
      "loss": 1.5992,
      "step": 15
    },
    {
      "epoch": 0.05673758865248227,
      "grad_norm": 2.1153617069636295,
      "learning_rate": 3.529411764705883e-06,
      "loss": 1.5119,
      "step": 16
    },
    {
      "epoch": 0.06028368794326241,
      "grad_norm": 2.1115729682872924,
      "learning_rate": 3.7647058823529414e-06,
      "loss": 1.5371,
      "step": 17
    },
    {
      "epoch": 0.06382978723404255,
      "grad_norm": 2.830840809822407,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.525,
      "step": 18
    },
    {
      "epoch": 0.0673758865248227,
      "grad_norm": 3.1825497271706578,
      "learning_rate": 4.235294117647059e-06,
      "loss": 1.5146,
      "step": 19
    },
    {
      "epoch": 0.07092198581560284,
      "grad_norm": 2.934535682603558,
      "learning_rate": 4.4705882352941184e-06,
      "loss": 1.4848,
      "step": 20
    },
    {
      "epoch": 0.07446808510638298,
      "grad_norm": 2.4904059244565486,
      "learning_rate": 4.705882352941177e-06,
      "loss": 1.4281,
      "step": 21
    },
    {
      "epoch": 0.07801418439716312,
      "grad_norm": 1.965660747063884,
      "learning_rate": 4.941176470588236e-06,
      "loss": 1.4339,
      "step": 22
    },
    {
      "epoch": 0.08156028368794327,
      "grad_norm": 1.601702138082789,
      "learning_rate": 5.176470588235295e-06,
      "loss": 1.4012,
      "step": 23
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": 1.1765313349167572,
      "learning_rate": 5.411764705882353e-06,
      "loss": 1.3595,
      "step": 24
    },
    {
      "epoch": 0.08865248226950355,
      "grad_norm": 1.181473952846224,
      "learning_rate": 5.6470588235294125e-06,
      "loss": 1.3754,
      "step": 25
    },
    {
      "epoch": 0.09219858156028368,
      "grad_norm": 1.2232305593346886,
      "learning_rate": 5.882352941176471e-06,
      "loss": 1.3597,
      "step": 26
    },
    {
      "epoch": 0.09574468085106383,
      "grad_norm": 1.200373129757842,
      "learning_rate": 6.11764705882353e-06,
      "loss": 1.3325,
      "step": 27
    },
    {
      "epoch": 0.09929078014184398,
      "grad_norm": 0.9974822666519331,
      "learning_rate": 6.352941176470589e-06,
      "loss": 1.303,
      "step": 28
    },
    {
      "epoch": 0.10283687943262411,
      "grad_norm": 0.8261258243562135,
      "learning_rate": 6.588235294117647e-06,
      "loss": 1.3272,
      "step": 29
    },
    {
      "epoch": 0.10638297872340426,
      "grad_norm": 0.6976037915101265,
      "learning_rate": 6.8235294117647065e-06,
      "loss": 1.2773,
      "step": 30
    },
    {
      "epoch": 0.1099290780141844,
      "grad_norm": 0.7026199238348708,
      "learning_rate": 7.058823529411766e-06,
      "loss": 1.2758,
      "step": 31
    },
    {
      "epoch": 0.11347517730496454,
      "grad_norm": 0.680951408768359,
      "learning_rate": 7.294117647058823e-06,
      "loss": 1.264,
      "step": 32
    },
    {
      "epoch": 0.11702127659574468,
      "grad_norm": 0.669311532690681,
      "learning_rate": 7.529411764705883e-06,
      "loss": 1.2209,
      "step": 33
    },
    {
      "epoch": 0.12056737588652482,
      "grad_norm": 0.6005160404767125,
      "learning_rate": 7.764705882352941e-06,
      "loss": 1.1828,
      "step": 34
    },
    {
      "epoch": 0.12411347517730496,
      "grad_norm": 0.7326603813019236,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2431,
      "step": 35
    },
    {
      "epoch": 0.1276595744680851,
      "grad_norm": 0.5835009779668501,
      "learning_rate": 8.23529411764706e-06,
      "loss": 1.241,
      "step": 36
    },
    {
      "epoch": 0.13120567375886524,
      "grad_norm": 0.5378943901741212,
      "learning_rate": 8.470588235294118e-06,
      "loss": 1.1968,
      "step": 37
    },
    {
      "epoch": 0.1347517730496454,
      "grad_norm": 0.572579019039888,
      "learning_rate": 8.705882352941177e-06,
      "loss": 1.2327,
      "step": 38
    },
    {
      "epoch": 0.13829787234042554,
      "grad_norm": 0.5071056767777127,
      "learning_rate": 8.941176470588237e-06,
      "loss": 1.1909,
      "step": 39
    },
    {
      "epoch": 0.14184397163120568,
      "grad_norm": 0.5076864263245315,
      "learning_rate": 9.176470588235294e-06,
      "loss": 1.1914,
      "step": 40
    },
    {
      "epoch": 0.1453900709219858,
      "grad_norm": 0.4768242650196604,
      "learning_rate": 9.411764705882354e-06,
      "loss": 1.2212,
      "step": 41
    },
    {
      "epoch": 0.14893617021276595,
      "grad_norm": 0.5175086914458062,
      "learning_rate": 9.647058823529412e-06,
      "loss": 1.1533,
      "step": 42
    },
    {
      "epoch": 0.1524822695035461,
      "grad_norm": 0.469141750165397,
      "learning_rate": 9.882352941176472e-06,
      "loss": 1.1962,
      "step": 43
    },
    {
      "epoch": 0.15602836879432624,
      "grad_norm": 0.4713622627510269,
      "learning_rate": 1.011764705882353e-05,
      "loss": 1.202,
      "step": 44
    },
    {
      "epoch": 0.1595744680851064,
      "grad_norm": 0.43318402798290195,
      "learning_rate": 1.035294117647059e-05,
      "loss": 1.1557,
      "step": 45
    },
    {
      "epoch": 0.16312056737588654,
      "grad_norm": 0.4320649172294336,
      "learning_rate": 1.0588235294117648e-05,
      "loss": 1.1794,
      "step": 46
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.4115353237514731,
      "learning_rate": 1.0823529411764706e-05,
      "loss": 1.1802,
      "step": 47
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 0.4298696060251368,
      "learning_rate": 1.1058823529411766e-05,
      "loss": 1.1723,
      "step": 48
    },
    {
      "epoch": 0.17375886524822695,
      "grad_norm": 0.42417079988997497,
      "learning_rate": 1.1294117647058825e-05,
      "loss": 1.1788,
      "step": 49
    },
    {
      "epoch": 0.1773049645390071,
      "grad_norm": 0.4067328480323003,
      "learning_rate": 1.1529411764705882e-05,
      "loss": 1.1431,
      "step": 50
    },
    {
      "epoch": 0.18085106382978725,
      "grad_norm": 0.3874360377877721,
      "learning_rate": 1.1764705882352942e-05,
      "loss": 1.1142,
      "step": 51
    },
    {
      "epoch": 0.18439716312056736,
      "grad_norm": 0.4675215039516683,
      "learning_rate": 1.2e-05,
      "loss": 1.1023,
      "step": 52
    },
    {
      "epoch": 0.1879432624113475,
      "grad_norm": 0.4205563606474083,
      "learning_rate": 1.223529411764706e-05,
      "loss": 1.1854,
      "step": 53
    },
    {
      "epoch": 0.19148936170212766,
      "grad_norm": 0.41018245492071165,
      "learning_rate": 1.2470588235294119e-05,
      "loss": 1.1658,
      "step": 54
    },
    {
      "epoch": 0.1950354609929078,
      "grad_norm": 0.3620344825134141,
      "learning_rate": 1.2705882352941177e-05,
      "loss": 1.1242,
      "step": 55
    },
    {
      "epoch": 0.19858156028368795,
      "grad_norm": 0.40939886980642687,
      "learning_rate": 1.2941176470588238e-05,
      "loss": 1.1287,
      "step": 56
    },
    {
      "epoch": 0.20212765957446807,
      "grad_norm": 0.39840938018772826,
      "learning_rate": 1.3176470588235294e-05,
      "loss": 1.1755,
      "step": 57
    },
    {
      "epoch": 0.20567375886524822,
      "grad_norm": 0.3972864117185851,
      "learning_rate": 1.3411764705882353e-05,
      "loss": 1.1524,
      "step": 58
    },
    {
      "epoch": 0.20921985815602837,
      "grad_norm": 0.35963503957203713,
      "learning_rate": 1.3647058823529413e-05,
      "loss": 1.1172,
      "step": 59
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 0.3934205934197654,
      "learning_rate": 1.3882352941176471e-05,
      "loss": 1.099,
      "step": 60
    },
    {
      "epoch": 0.21631205673758866,
      "grad_norm": 0.36473534875394586,
      "learning_rate": 1.4117647058823532e-05,
      "loss": 1.094,
      "step": 61
    },
    {
      "epoch": 0.2198581560283688,
      "grad_norm": 0.38310616016878557,
      "learning_rate": 1.435294117647059e-05,
      "loss": 1.1315,
      "step": 62
    },
    {
      "epoch": 0.22340425531914893,
      "grad_norm": 0.38310118046511205,
      "learning_rate": 1.4588235294117647e-05,
      "loss": 1.0979,
      "step": 63
    },
    {
      "epoch": 0.22695035460992907,
      "grad_norm": 0.40071709534499794,
      "learning_rate": 1.4823529411764707e-05,
      "loss": 1.1292,
      "step": 64
    },
    {
      "epoch": 0.23049645390070922,
      "grad_norm": 0.40343417157026384,
      "learning_rate": 1.5058823529411765e-05,
      "loss": 1.1497,
      "step": 65
    },
    {
      "epoch": 0.23404255319148937,
      "grad_norm": 0.38981040775595505,
      "learning_rate": 1.5294117647058822e-05,
      "loss": 1.0976,
      "step": 66
    },
    {
      "epoch": 0.2375886524822695,
      "grad_norm": 0.4844512170653719,
      "learning_rate": 1.5529411764705882e-05,
      "loss": 1.129,
      "step": 67
    },
    {
      "epoch": 0.24113475177304963,
      "grad_norm": 0.3730642827040671,
      "learning_rate": 1.5764705882352943e-05,
      "loss": 1.0611,
      "step": 68
    },
    {
      "epoch": 0.24468085106382978,
      "grad_norm": 0.38672469429859196,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.0984,
      "step": 69
    },
    {
      "epoch": 0.24822695035460993,
      "grad_norm": 0.4027051220542017,
      "learning_rate": 1.623529411764706e-05,
      "loss": 1.0971,
      "step": 70
    },
    {
      "epoch": 0.25177304964539005,
      "grad_norm": 0.38792933242967387,
      "learning_rate": 1.647058823529412e-05,
      "loss": 1.1201,
      "step": 71
    },
    {
      "epoch": 0.2553191489361702,
      "grad_norm": 0.40861480699220193,
      "learning_rate": 1.670588235294118e-05,
      "loss": 1.1313,
      "step": 72
    },
    {
      "epoch": 0.25886524822695034,
      "grad_norm": 0.3649020493754554,
      "learning_rate": 1.6941176470588237e-05,
      "loss": 1.0918,
      "step": 73
    },
    {
      "epoch": 0.2624113475177305,
      "grad_norm": 0.38869825009731374,
      "learning_rate": 1.7176470588235293e-05,
      "loss": 1.0971,
      "step": 74
    },
    {
      "epoch": 0.26595744680851063,
      "grad_norm": 0.3869921912718602,
      "learning_rate": 1.7411764705882353e-05,
      "loss": 1.0756,
      "step": 75
    },
    {
      "epoch": 0.2695035460992908,
      "grad_norm": 0.37677316889298695,
      "learning_rate": 1.7647058823529414e-05,
      "loss": 1.0792,
      "step": 76
    },
    {
      "epoch": 0.2730496453900709,
      "grad_norm": 0.36310119840141764,
      "learning_rate": 1.7882352941176474e-05,
      "loss": 1.095,
      "step": 77
    },
    {
      "epoch": 0.2765957446808511,
      "grad_norm": 0.3817251391796483,
      "learning_rate": 1.811764705882353e-05,
      "loss": 1.1147,
      "step": 78
    },
    {
      "epoch": 0.2801418439716312,
      "grad_norm": 0.38683476125531163,
      "learning_rate": 1.8352941176470587e-05,
      "loss": 1.1098,
      "step": 79
    },
    {
      "epoch": 0.28368794326241137,
      "grad_norm": 0.36958806331472616,
      "learning_rate": 1.8588235294117647e-05,
      "loss": 1.0726,
      "step": 80
    },
    {
      "epoch": 0.2872340425531915,
      "grad_norm": 0.39165579816640134,
      "learning_rate": 1.8823529411764708e-05,
      "loss": 1.0831,
      "step": 81
    },
    {
      "epoch": 0.2907801418439716,
      "grad_norm": 0.4058344969874132,
      "learning_rate": 1.9058823529411764e-05,
      "loss": 1.0735,
      "step": 82
    },
    {
      "epoch": 0.29432624113475175,
      "grad_norm": 0.4030952561691915,
      "learning_rate": 1.9294117647058825e-05,
      "loss": 1.0924,
      "step": 83
    },
    {
      "epoch": 0.2978723404255319,
      "grad_norm": 0.42274227026304756,
      "learning_rate": 1.9529411764705885e-05,
      "loss": 1.1099,
      "step": 84
    },
    {
      "epoch": 0.30141843971631205,
      "grad_norm": 0.40979457721603246,
      "learning_rate": 1.9764705882352945e-05,
      "loss": 1.0618,
      "step": 85
    },
    {
      "epoch": 0.3049645390070922,
      "grad_norm": 0.385822581453486,
      "learning_rate": 2e-05,
      "loss": 1.0432,
      "step": 86
    },
    {
      "epoch": 0.30851063829787234,
      "grad_norm": 0.4283573534577983,
      "learning_rate": 1.99999147881843e-05,
      "loss": 1.0875,
      "step": 87
    },
    {
      "epoch": 0.3120567375886525,
      "grad_norm": 0.41563178697289727,
      "learning_rate": 1.9999659154189404e-05,
      "loss": 1.0805,
      "step": 88
    },
    {
      "epoch": 0.31560283687943264,
      "grad_norm": 0.3899019912130811,
      "learning_rate": 1.9999233102371918e-05,
      "loss": 1.0768,
      "step": 89
    },
    {
      "epoch": 0.3191489361702128,
      "grad_norm": 0.3992722195667444,
      "learning_rate": 1.9998636639992776e-05,
      "loss": 1.1252,
      "step": 90
    },
    {
      "epoch": 0.32269503546099293,
      "grad_norm": 0.41052367607056267,
      "learning_rate": 1.9997869777217106e-05,
      "loss": 1.069,
      "step": 91
    },
    {
      "epoch": 0.3262411347517731,
      "grad_norm": 0.41728106430532,
      "learning_rate": 1.9996932527114064e-05,
      "loss": 1.0592,
      "step": 92
    },
    {
      "epoch": 0.32978723404255317,
      "grad_norm": 0.3912128813820819,
      "learning_rate": 1.99958249056566e-05,
      "loss": 1.0442,
      "step": 93
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.40709314071102765,
      "learning_rate": 1.9994546931721204e-05,
      "loss": 1.1249,
      "step": 94
    },
    {
      "epoch": 0.33687943262411346,
      "grad_norm": 0.4223398268189614,
      "learning_rate": 1.9993098627087576e-05,
      "loss": 1.0949,
      "step": 95
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 0.44711495627147035,
      "learning_rate": 1.999148001643824e-05,
      "loss": 1.0807,
      "step": 96
    },
    {
      "epoch": 0.34397163120567376,
      "grad_norm": 0.39883160757012304,
      "learning_rate": 1.998969112735816e-05,
      "loss": 1.1175,
      "step": 97
    },
    {
      "epoch": 0.3475177304964539,
      "grad_norm": 0.4124247088388006,
      "learning_rate": 1.9987731990334225e-05,
      "loss": 1.0469,
      "step": 98
    },
    {
      "epoch": 0.35106382978723405,
      "grad_norm": 0.8676420162826889,
      "learning_rate": 1.9985602638754758e-05,
      "loss": 1.0547,
      "step": 99
    },
    {
      "epoch": 0.3546099290780142,
      "grad_norm": 0.4539645796907816,
      "learning_rate": 1.9983303108908946e-05,
      "loss": 1.071,
      "step": 100
    },
    {
      "epoch": 0.35815602836879434,
      "grad_norm": 0.4103701258745225,
      "learning_rate": 1.998083343998621e-05,
      "loss": 1.0597,
      "step": 101
    },
    {
      "epoch": 0.3617021276595745,
      "grad_norm": 0.38720212977229745,
      "learning_rate": 1.9978193674075547e-05,
      "loss": 1.052,
      "step": 102
    },
    {
      "epoch": 0.36524822695035464,
      "grad_norm": 0.41572032365047884,
      "learning_rate": 1.9975383856164798e-05,
      "loss": 1.0686,
      "step": 103
    },
    {
      "epoch": 0.36879432624113473,
      "grad_norm": 0.4082614940841916,
      "learning_rate": 1.9972404034139913e-05,
      "loss": 1.0855,
      "step": 104
    },
    {
      "epoch": 0.3723404255319149,
      "grad_norm": 0.37549114805028977,
      "learning_rate": 1.996925425878409e-05,
      "loss": 1.047,
      "step": 105
    },
    {
      "epoch": 0.375886524822695,
      "grad_norm": 0.4610243331143955,
      "learning_rate": 1.9965934583776948e-05,
      "loss": 1.0636,
      "step": 106
    },
    {
      "epoch": 0.37943262411347517,
      "grad_norm": 0.44503123216260176,
      "learning_rate": 1.99624450656936e-05,
      "loss": 1.0526,
      "step": 107
    },
    {
      "epoch": 0.3829787234042553,
      "grad_norm": 0.4149811610983327,
      "learning_rate": 1.995878576400367e-05,
      "loss": 1.0076,
      "step": 108
    },
    {
      "epoch": 0.38652482269503546,
      "grad_norm": 0.3918182369922544,
      "learning_rate": 1.9954956741070312e-05,
      "loss": 1.0797,
      "step": 109
    },
    {
      "epoch": 0.3900709219858156,
      "grad_norm": 0.435027379258559,
      "learning_rate": 1.9950958062149126e-05,
      "loss": 1.0864,
      "step": 110
    },
    {
      "epoch": 0.39361702127659576,
      "grad_norm": 0.3879332371338635,
      "learning_rate": 1.9946789795387048e-05,
      "loss": 1.0949,
      "step": 111
    },
    {
      "epoch": 0.3971631205673759,
      "grad_norm": 0.4109671535287323,
      "learning_rate": 1.9942452011821195e-05,
      "loss": 1.0456,
      "step": 112
    },
    {
      "epoch": 0.40070921985815605,
      "grad_norm": 0.39837338883362244,
      "learning_rate": 1.993794478537765e-05,
      "loss": 1.0678,
      "step": 113
    },
    {
      "epoch": 0.40425531914893614,
      "grad_norm": 0.48242943738691013,
      "learning_rate": 1.9933268192870202e-05,
      "loss": 1.0972,
      "step": 114
    },
    {
      "epoch": 0.4078014184397163,
      "grad_norm": 0.4147047600582376,
      "learning_rate": 1.9928422313999038e-05,
      "loss": 1.0471,
      "step": 115
    },
    {
      "epoch": 0.41134751773049644,
      "grad_norm": 0.4185632224192961,
      "learning_rate": 1.9923407231349386e-05,
      "loss": 1.0918,
      "step": 116
    },
    {
      "epoch": 0.4148936170212766,
      "grad_norm": 0.41380413853065684,
      "learning_rate": 1.991822303039011e-05,
      "loss": 1.0642,
      "step": 117
    },
    {
      "epoch": 0.41843971631205673,
      "grad_norm": 0.40984905075653666,
      "learning_rate": 1.991286979947224e-05,
      "loss": 1.0911,
      "step": 118
    },
    {
      "epoch": 0.4219858156028369,
      "grad_norm": 0.4346507481006875,
      "learning_rate": 1.990734762982748e-05,
      "loss": 1.0524,
      "step": 119
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 0.3802797846709125,
      "learning_rate": 1.9901656615566655e-05,
      "loss": 1.0371,
      "step": 120
    },
    {
      "epoch": 0.42907801418439717,
      "grad_norm": 0.43994501408678205,
      "learning_rate": 1.9895796853678096e-05,
      "loss": 1.0402,
      "step": 121
    },
    {
      "epoch": 0.4326241134751773,
      "grad_norm": 0.47314319261278487,
      "learning_rate": 1.988976844402599e-05,
      "loss": 1.0624,
      "step": 122
    },
    {
      "epoch": 0.43617021276595747,
      "grad_norm": 0.43887512296501896,
      "learning_rate": 1.988357148934869e-05,
      "loss": 1.0642,
      "step": 123
    },
    {
      "epoch": 0.4397163120567376,
      "grad_norm": 0.4128076475964722,
      "learning_rate": 1.9877206095256935e-05,
      "loss": 1.0363,
      "step": 124
    },
    {
      "epoch": 0.4432624113475177,
      "grad_norm": 0.4348332361717801,
      "learning_rate": 1.9870672370232097e-05,
      "loss": 1.0219,
      "step": 125
    },
    {
      "epoch": 0.44680851063829785,
      "grad_norm": 0.4035353186595829,
      "learning_rate": 1.986397042562428e-05,
      "loss": 1.0277,
      "step": 126
    },
    {
      "epoch": 0.450354609929078,
      "grad_norm": 0.4054498285955409,
      "learning_rate": 1.985710037565047e-05,
      "loss": 1.0883,
      "step": 127
    },
    {
      "epoch": 0.45390070921985815,
      "grad_norm": 0.39687366636112803,
      "learning_rate": 1.9850062337392536e-05,
      "loss": 1.0607,
      "step": 128
    },
    {
      "epoch": 0.4574468085106383,
      "grad_norm": 0.4405171428453929,
      "learning_rate": 1.9842856430795298e-05,
      "loss": 1.0564,
      "step": 129
    },
    {
      "epoch": 0.46099290780141844,
      "grad_norm": 0.41170770013631397,
      "learning_rate": 1.9835482778664426e-05,
      "loss": 1.0311,
      "step": 130
    },
    {
      "epoch": 0.4645390070921986,
      "grad_norm": 0.39274848149852354,
      "learning_rate": 1.9827941506664378e-05,
      "loss": 1.0626,
      "step": 131
    },
    {
      "epoch": 0.46808510638297873,
      "grad_norm": 0.43188276490314575,
      "learning_rate": 1.9820232743316244e-05,
      "loss": 1.0597,
      "step": 132
    },
    {
      "epoch": 0.4716312056737589,
      "grad_norm": 0.4032899104523295,
      "learning_rate": 1.981235661999558e-05,
      "loss": 1.06,
      "step": 133
    },
    {
      "epoch": 0.475177304964539,
      "grad_norm": 0.45040621241859885,
      "learning_rate": 1.9804313270930134e-05,
      "loss": 1.078,
      "step": 134
    },
    {
      "epoch": 0.4787234042553192,
      "grad_norm": 0.38105861324144824,
      "learning_rate": 1.979610283319758e-05,
      "loss": 1.0653,
      "step": 135
    },
    {
      "epoch": 0.48226950354609927,
      "grad_norm": 0.4048312669954708,
      "learning_rate": 1.978772544672318e-05,
      "loss": 1.0469,
      "step": 136
    },
    {
      "epoch": 0.4858156028368794,
      "grad_norm": 0.39934964581590104,
      "learning_rate": 1.9779181254277397e-05,
      "loss": 1.0482,
      "step": 137
    },
    {
      "epoch": 0.48936170212765956,
      "grad_norm": 0.4297490738832722,
      "learning_rate": 1.9770470401473462e-05,
      "loss": 1.0751,
      "step": 138
    },
    {
      "epoch": 0.4929078014184397,
      "grad_norm": 0.3732722270246798,
      "learning_rate": 1.9761593036764894e-05,
      "loss": 1.02,
      "step": 139
    },
    {
      "epoch": 0.49645390070921985,
      "grad_norm": 0.40981082693027104,
      "learning_rate": 1.975254931144296e-05,
      "loss": 1.0506,
      "step": 140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3811499359219313,
      "learning_rate": 1.9743339379634117e-05,
      "loss": 1.0495,
      "step": 141
    },
    {
      "epoch": 0.5035460992907801,
      "grad_norm": 0.38491189378621854,
      "learning_rate": 1.9733963398297365e-05,
      "loss": 1.0351,
      "step": 142
    },
    {
      "epoch": 0.5070921985815603,
      "grad_norm": 0.4247342696761821,
      "learning_rate": 1.972442152722158e-05,
      "loss": 1.04,
      "step": 143
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 0.41437878465719885,
      "learning_rate": 1.97147139290228e-05,
      "loss": 1.0916,
      "step": 144
    },
    {
      "epoch": 0.5141843971631206,
      "grad_norm": 0.43538254663755344,
      "learning_rate": 1.9704840769141435e-05,
      "loss": 1.0462,
      "step": 145
    },
    {
      "epoch": 0.5177304964539007,
      "grad_norm": 0.4133957050527475,
      "learning_rate": 1.969480221583946e-05,
      "loss": 1.0684,
      "step": 146
    },
    {
      "epoch": 0.5212765957446809,
      "grad_norm": 0.4321179837333326,
      "learning_rate": 1.968459844019755e-05,
      "loss": 1.0537,
      "step": 147
    },
    {
      "epoch": 0.524822695035461,
      "grad_norm": 0.4199126244655049,
      "learning_rate": 1.9674229616112147e-05,
      "loss": 1.0694,
      "step": 148
    },
    {
      "epoch": 0.5283687943262412,
      "grad_norm": 0.41968308383381936,
      "learning_rate": 1.9663695920292528e-05,
      "loss": 1.0374,
      "step": 149
    },
    {
      "epoch": 0.5319148936170213,
      "grad_norm": 0.3937539527163739,
      "learning_rate": 1.9652997532257752e-05,
      "loss": 1.0097,
      "step": 150
    },
    {
      "epoch": 0.5354609929078015,
      "grad_norm": 0.4151219034363994,
      "learning_rate": 1.964213463433364e-05,
      "loss": 1.0286,
      "step": 151
    },
    {
      "epoch": 0.5390070921985816,
      "grad_norm": 0.4040400586893122,
      "learning_rate": 1.9631107411649632e-05,
      "loss": 1.0483,
      "step": 152
    },
    {
      "epoch": 0.5425531914893617,
      "grad_norm": 0.3716515261787462,
      "learning_rate": 1.9619916052135676e-05,
      "loss": 1.054,
      "step": 153
    },
    {
      "epoch": 0.5460992907801419,
      "grad_norm": 0.4650548851701862,
      "learning_rate": 1.9608560746518977e-05,
      "loss": 1.0405,
      "step": 154
    },
    {
      "epoch": 0.549645390070922,
      "grad_norm": 0.44014703147533524,
      "learning_rate": 1.959704168832078e-05,
      "loss": 1.065,
      "step": 155
    },
    {
      "epoch": 0.5531914893617021,
      "grad_norm": 0.4012529413831958,
      "learning_rate": 1.9585359073853052e-05,
      "loss": 1.0896,
      "step": 156
    },
    {
      "epoch": 0.5567375886524822,
      "grad_norm": 0.4252692558854423,
      "learning_rate": 1.957351310221516e-05,
      "loss": 1.0305,
      "step": 157
    },
    {
      "epoch": 0.5602836879432624,
      "grad_norm": 0.41637802948134767,
      "learning_rate": 1.9561503975290446e-05,
      "loss": 1.0678,
      "step": 158
    },
    {
      "epoch": 0.5638297872340425,
      "grad_norm": 0.41613274402508954,
      "learning_rate": 1.954933189774282e-05,
      "loss": 1.0199,
      "step": 159
    },
    {
      "epoch": 0.5673758865248227,
      "grad_norm": 0.39382583802044363,
      "learning_rate": 1.9536997077013236e-05,
      "loss": 1.0495,
      "step": 160
    },
    {
      "epoch": 0.5709219858156028,
      "grad_norm": 0.388499486904763,
      "learning_rate": 1.95244997233162e-05,
      "loss": 1.0014,
      "step": 161
    },
    {
      "epoch": 0.574468085106383,
      "grad_norm": 0.38512135545026266,
      "learning_rate": 1.951184004963615e-05,
      "loss": 1.0718,
      "step": 162
    },
    {
      "epoch": 0.5780141843971631,
      "grad_norm": 0.3783281093275966,
      "learning_rate": 1.9499018271723836e-05,
      "loss": 1.0206,
      "step": 163
    },
    {
      "epoch": 0.5815602836879432,
      "grad_norm": 0.43369685190384033,
      "learning_rate": 1.9486034608092657e-05,
      "loss": 1.0498,
      "step": 164
    },
    {
      "epoch": 0.5851063829787234,
      "grad_norm": 0.3987508299716515,
      "learning_rate": 1.9472889280014924e-05,
      "loss": 1.0351,
      "step": 165
    },
    {
      "epoch": 0.5886524822695035,
      "grad_norm": 0.4100807327115829,
      "learning_rate": 1.945958251151809e-05,
      "loss": 1.0499,
      "step": 166
    },
    {
      "epoch": 0.5921985815602837,
      "grad_norm": 0.4256425875406832,
      "learning_rate": 1.944611452938094e-05,
      "loss": 1.0487,
      "step": 167
    },
    {
      "epoch": 0.5957446808510638,
      "grad_norm": 0.44803839464658235,
      "learning_rate": 1.943248556312971e-05,
      "loss": 0.9881,
      "step": 168
    },
    {
      "epoch": 0.599290780141844,
      "grad_norm": 0.40259897757633056,
      "learning_rate": 1.9418695845034197e-05,
      "loss": 1.0168,
      "step": 169
    },
    {
      "epoch": 0.6028368794326241,
      "grad_norm": 0.42470303559371697,
      "learning_rate": 1.9404745610103785e-05,
      "loss": 1.0281,
      "step": 170
    },
    {
      "epoch": 0.6063829787234043,
      "grad_norm": 0.41403135143146674,
      "learning_rate": 1.9390635096083443e-05,
      "loss": 1.0543,
      "step": 171
    },
    {
      "epoch": 0.6099290780141844,
      "grad_norm": 0.4204328398131905,
      "learning_rate": 1.937636454344967e-05,
      "loss": 1.0365,
      "step": 172
    },
    {
      "epoch": 0.6134751773049646,
      "grad_norm": 0.46115860711121315,
      "learning_rate": 1.936193419540641e-05,
      "loss": 1.0458,
      "step": 173
    },
    {
      "epoch": 0.6170212765957447,
      "grad_norm": 0.4378596032908916,
      "learning_rate": 1.9347344297880892e-05,
      "loss": 1.0556,
      "step": 174
    },
    {
      "epoch": 0.6205673758865248,
      "grad_norm": 0.5113254010648783,
      "learning_rate": 1.9332595099519454e-05,
      "loss": 1.0638,
      "step": 175
    },
    {
      "epoch": 0.624113475177305,
      "grad_norm": 0.4512162294917552,
      "learning_rate": 1.9317686851683285e-05,
      "loss": 1.0662,
      "step": 176
    },
    {
      "epoch": 0.6276595744680851,
      "grad_norm": 0.4304702959646727,
      "learning_rate": 1.9302619808444158e-05,
      "loss": 1.0468,
      "step": 177
    },
    {
      "epoch": 0.6312056737588653,
      "grad_norm": 0.4719059649772538,
      "learning_rate": 1.9287394226580102e-05,
      "loss": 1.0701,
      "step": 178
    },
    {
      "epoch": 0.6347517730496454,
      "grad_norm": 0.3734601257160578,
      "learning_rate": 1.9272010365571002e-05,
      "loss": 1.0236,
      "step": 179
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 0.4651134237014968,
      "learning_rate": 1.9256468487594215e-05,
      "loss": 1.0541,
      "step": 180
    },
    {
      "epoch": 0.6418439716312057,
      "grad_norm": 0.4042541523513242,
      "learning_rate": 1.9240768857520062e-05,
      "loss": 1.0466,
      "step": 181
    },
    {
      "epoch": 0.6453900709219859,
      "grad_norm": 0.4243886495965767,
      "learning_rate": 1.922491174290734e-05,
      "loss": 1.0424,
      "step": 182
    },
    {
      "epoch": 0.648936170212766,
      "grad_norm": 0.40111191394993834,
      "learning_rate": 1.9208897413998758e-05,
      "loss": 1.0218,
      "step": 183
    },
    {
      "epoch": 0.6524822695035462,
      "grad_norm": 0.3940895377839162,
      "learning_rate": 1.9192726143716322e-05,
      "loss": 1.0594,
      "step": 184
    },
    {
      "epoch": 0.6560283687943262,
      "grad_norm": 0.42271957323204373,
      "learning_rate": 1.9176398207656693e-05,
      "loss": 1.0432,
      "step": 185
    },
    {
      "epoch": 0.6595744680851063,
      "grad_norm": 0.37986382559991433,
      "learning_rate": 1.9159913884086486e-05,
      "loss": 1.0228,
      "step": 186
    },
    {
      "epoch": 0.6631205673758865,
      "grad_norm": 0.4074315975311201,
      "learning_rate": 1.9143273453937533e-05,
      "loss": 1.0569,
      "step": 187
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.40463938046019843,
      "learning_rate": 1.9126477200802084e-05,
      "loss": 1.0295,
      "step": 188
    },
    {
      "epoch": 0.6702127659574468,
      "grad_norm": 0.4497742000064388,
      "learning_rate": 1.9109525410927988e-05,
      "loss": 1.0472,
      "step": 189
    },
    {
      "epoch": 0.6737588652482269,
      "grad_norm": 0.4079176238827453,
      "learning_rate": 1.9092418373213795e-05,
      "loss": 1.0033,
      "step": 190
    },
    {
      "epoch": 0.6773049645390071,
      "grad_norm": 0.38168468355602525,
      "learning_rate": 1.9075156379203867e-05,
      "loss": 1.018,
      "step": 191
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 0.4270641992780327,
      "learning_rate": 1.9057739723083366e-05,
      "loss": 1.0285,
      "step": 192
    },
    {
      "epoch": 0.6843971631205674,
      "grad_norm": 0.4468028496736141,
      "learning_rate": 1.904016870167327e-05,
      "loss": 1.0169,
      "step": 193
    },
    {
      "epoch": 0.6879432624113475,
      "grad_norm": 0.388728811354985,
      "learning_rate": 1.902244361442531e-05,
      "loss": 1.0109,
      "step": 194
    },
    {
      "epoch": 0.6914893617021277,
      "grad_norm": 0.4366235722900726,
      "learning_rate": 1.9004564763416856e-05,
      "loss": 1.0279,
      "step": 195
    },
    {
      "epoch": 0.6950354609929078,
      "grad_norm": 0.4005450474464167,
      "learning_rate": 1.8986532453345785e-05,
      "loss": 1.0614,
      "step": 196
    },
    {
      "epoch": 0.6985815602836879,
      "grad_norm": 0.4283032818637797,
      "learning_rate": 1.8968346991525267e-05,
      "loss": 0.9702,
      "step": 197
    },
    {
      "epoch": 0.7021276595744681,
      "grad_norm": 0.4149454329152974,
      "learning_rate": 1.895000868787855e-05,
      "loss": 1.0506,
      "step": 198
    },
    {
      "epoch": 0.7056737588652482,
      "grad_norm": 0.3908431105870882,
      "learning_rate": 1.8931517854933664e-05,
      "loss": 1.0352,
      "step": 199
    },
    {
      "epoch": 0.7092198581560284,
      "grad_norm": 0.4122069231448336,
      "learning_rate": 1.89128748078181e-05,
      "loss": 1.0082,
      "step": 200
    },
    {
      "epoch": 0.7127659574468085,
      "grad_norm": 0.39206516921546697,
      "learning_rate": 1.8894079864253434e-05,
      "loss": 1.0089,
      "step": 201
    },
    {
      "epoch": 0.7163120567375887,
      "grad_norm": 0.4017261077067384,
      "learning_rate": 1.887513334454992e-05,
      "loss": 1.0385,
      "step": 202
    },
    {
      "epoch": 0.7198581560283688,
      "grad_norm": 0.4144866828019059,
      "learning_rate": 1.8856035571601027e-05,
      "loss": 1.012,
      "step": 203
    },
    {
      "epoch": 0.723404255319149,
      "grad_norm": 0.4476529046794145,
      "learning_rate": 1.8836786870877942e-05,
      "loss": 1.0259,
      "step": 204
    },
    {
      "epoch": 0.7269503546099291,
      "grad_norm": 0.4139864974336738,
      "learning_rate": 1.8817387570424003e-05,
      "loss": 1.0546,
      "step": 205
    },
    {
      "epoch": 0.7304964539007093,
      "grad_norm": 0.41744847245347383,
      "learning_rate": 1.879783800084914e-05,
      "loss": 1.0575,
      "step": 206
    },
    {
      "epoch": 0.7340425531914894,
      "grad_norm": 0.41092263512752675,
      "learning_rate": 1.8778138495324213e-05,
      "loss": 0.9908,
      "step": 207
    },
    {
      "epoch": 0.7375886524822695,
      "grad_norm": 0.4699649880193204,
      "learning_rate": 1.8758289389575355e-05,
      "loss": 1.0347,
      "step": 208
    },
    {
      "epoch": 0.7411347517730497,
      "grad_norm": 0.43956419673006875,
      "learning_rate": 1.873829102187823e-05,
      "loss": 1.0577,
      "step": 209
    },
    {
      "epoch": 0.7446808510638298,
      "grad_norm": 0.42241045409865247,
      "learning_rate": 1.8718143733052278e-05,
      "loss": 1.0152,
      "step": 210
    },
    {
      "epoch": 0.74822695035461,
      "grad_norm": 0.43120791636280065,
      "learning_rate": 1.8697847866454922e-05,
      "loss": 0.9924,
      "step": 211
    },
    {
      "epoch": 0.75177304964539,
      "grad_norm": 0.4347253404784015,
      "learning_rate": 1.867740376797568e-05,
      "loss": 1.008,
      "step": 212
    },
    {
      "epoch": 0.7553191489361702,
      "grad_norm": 0.4312933323093618,
      "learning_rate": 1.865681178603031e-05,
      "loss": 0.9941,
      "step": 213
    },
    {
      "epoch": 0.7588652482269503,
      "grad_norm": 0.4114365381511061,
      "learning_rate": 1.8636072271554842e-05,
      "loss": 1.0626,
      "step": 214
    },
    {
      "epoch": 0.7624113475177305,
      "grad_norm": 0.4629217798127864,
      "learning_rate": 1.8615185577999616e-05,
      "loss": 1.0296,
      "step": 215
    },
    {
      "epoch": 0.7659574468085106,
      "grad_norm": 0.44094993883878913,
      "learning_rate": 1.8594152061323245e-05,
      "loss": 1.0399,
      "step": 216
    },
    {
      "epoch": 0.7695035460992907,
      "grad_norm": 0.415176959865907,
      "learning_rate": 1.857297207998656e-05,
      "loss": 1.035,
      "step": 217
    },
    {
      "epoch": 0.7730496453900709,
      "grad_norm": 0.44266355517212597,
      "learning_rate": 1.85516459949465e-05,
      "loss": 1.1049,
      "step": 218
    },
    {
      "epoch": 0.776595744680851,
      "grad_norm": 0.44043411082498807,
      "learning_rate": 1.853017416964994e-05,
      "loss": 1.015,
      "step": 219
    },
    {
      "epoch": 0.7801418439716312,
      "grad_norm": 0.4033941099550602,
      "learning_rate": 1.850855697002753e-05,
      "loss": 1.0421,
      "step": 220
    },
    {
      "epoch": 0.7836879432624113,
      "grad_norm": 0.43450208494361897,
      "learning_rate": 1.8486794764487436e-05,
      "loss": 1.0474,
      "step": 221
    },
    {
      "epoch": 0.7872340425531915,
      "grad_norm": 0.6137072500977133,
      "learning_rate": 1.8464887923909062e-05,
      "loss": 1.0659,
      "step": 222
    },
    {
      "epoch": 0.7907801418439716,
      "grad_norm": 0.42816908651382335,
      "learning_rate": 1.844283682163675e-05,
      "loss": 1.0588,
      "step": 223
    },
    {
      "epoch": 0.7943262411347518,
      "grad_norm": 0.3865343363054826,
      "learning_rate": 1.8420641833473386e-05,
      "loss": 1.0228,
      "step": 224
    },
    {
      "epoch": 0.7978723404255319,
      "grad_norm": 0.42101602612274497,
      "learning_rate": 1.839830333767402e-05,
      "loss": 0.9757,
      "step": 225
    },
    {
      "epoch": 0.8014184397163121,
      "grad_norm": 0.40910650357636097,
      "learning_rate": 1.8375821714939408e-05,
      "loss": 1.0464,
      "step": 226
    },
    {
      "epoch": 0.8049645390070922,
      "grad_norm": 0.42468493209973357,
      "learning_rate": 1.835319734840953e-05,
      "loss": 1.0311,
      "step": 227
    },
    {
      "epoch": 0.8085106382978723,
      "grad_norm": 0.38532684525673616,
      "learning_rate": 1.833043062365706e-05,
      "loss": 1.0266,
      "step": 228
    },
    {
      "epoch": 0.8120567375886525,
      "grad_norm": 0.4203249202064101,
      "learning_rate": 1.8307521928680782e-05,
      "loss": 1.033,
      "step": 229
    },
    {
      "epoch": 0.8156028368794326,
      "grad_norm": 0.4110077616808033,
      "learning_rate": 1.8284471653898995e-05,
      "loss": 1.0488,
      "step": 230
    },
    {
      "epoch": 0.8191489361702128,
      "grad_norm": 0.4088394328094259,
      "learning_rate": 1.8261280192142857e-05,
      "loss": 0.9693,
      "step": 231
    },
    {
      "epoch": 0.8226950354609929,
      "grad_norm": 0.3999605154188856,
      "learning_rate": 1.823794793864968e-05,
      "loss": 1.0256,
      "step": 232
    },
    {
      "epoch": 0.8262411347517731,
      "grad_norm": 0.4089862024662302,
      "learning_rate": 1.82144752910562e-05,
      "loss": 1.0152,
      "step": 233
    },
    {
      "epoch": 0.8297872340425532,
      "grad_norm": 0.3889593277582271,
      "learning_rate": 1.81908626493918e-05,
      "loss": 1.003,
      "step": 234
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.41794204162797643,
      "learning_rate": 1.8167110416071696e-05,
      "loss": 1.0425,
      "step": 235
    },
    {
      "epoch": 0.8368794326241135,
      "grad_norm": 0.3952447291799417,
      "learning_rate": 1.8143218995890072e-05,
      "loss": 1.0034,
      "step": 236
    },
    {
      "epoch": 0.8404255319148937,
      "grad_norm": 0.45317681752619565,
      "learning_rate": 1.8119188796013186e-05,
      "loss": 1.0461,
      "step": 237
    },
    {
      "epoch": 0.8439716312056738,
      "grad_norm": 0.3711961454153343,
      "learning_rate": 1.8095020225972436e-05,
      "loss": 1.0439,
      "step": 238
    },
    {
      "epoch": 0.8475177304964538,
      "grad_norm": 0.4755874761259998,
      "learning_rate": 1.807071369765736e-05,
      "loss": 1.0436,
      "step": 239
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 0.4196318651413355,
      "learning_rate": 1.804626962530865e-05,
      "loss": 1.0486,
      "step": 240
    },
    {
      "epoch": 0.8546099290780141,
      "grad_norm": 0.4125254595290099,
      "learning_rate": 1.8021688425511058e-05,
      "loss": 1.0293,
      "step": 241
    },
    {
      "epoch": 0.8581560283687943,
      "grad_norm": 0.39729235078930125,
      "learning_rate": 1.7996970517186315e-05,
      "loss": 0.9931,
      "step": 242
    },
    {
      "epoch": 0.8617021276595744,
      "grad_norm": 0.40981989014199277,
      "learning_rate": 1.7972116321585997e-05,
      "loss": 1.0016,
      "step": 243
    },
    {
      "epoch": 0.8652482269503546,
      "grad_norm": 0.3941668554247862,
      "learning_rate": 1.794712626228433e-05,
      "loss": 1.0114,
      "step": 244
    },
    {
      "epoch": 0.8687943262411347,
      "grad_norm": 0.4273992838829838,
      "learning_rate": 1.7922000765170976e-05,
      "loss": 1.0231,
      "step": 245
    },
    {
      "epoch": 0.8723404255319149,
      "grad_norm": 0.408707494956399,
      "learning_rate": 1.789674025844378e-05,
      "loss": 1.021,
      "step": 246
    },
    {
      "epoch": 0.875886524822695,
      "grad_norm": 0.39186386718572935,
      "learning_rate": 1.7871345172601475e-05,
      "loss": 1.0313,
      "step": 247
    },
    {
      "epoch": 0.8794326241134752,
      "grad_norm": 0.4041443699597106,
      "learning_rate": 1.7845815940436336e-05,
      "loss": 1.0161,
      "step": 248
    },
    {
      "epoch": 0.8829787234042553,
      "grad_norm": 0.3791911252140322,
      "learning_rate": 1.7820152997026807e-05,
      "loss": 1.0213,
      "step": 249
    },
    {
      "epoch": 0.8865248226950354,
      "grad_norm": 0.3812683533757764,
      "learning_rate": 1.7794356779730084e-05,
      "loss": 1.0071,
      "step": 250
    },
    {
      "epoch": 0.8900709219858156,
      "grad_norm": 0.391304314088133,
      "learning_rate": 1.7768427728174677e-05,
      "loss": 1.0088,
      "step": 251
    },
    {
      "epoch": 0.8936170212765957,
      "grad_norm": 0.34306587428261137,
      "learning_rate": 1.7742366284252895e-05,
      "loss": 0.9865,
      "step": 252
    },
    {
      "epoch": 0.8971631205673759,
      "grad_norm": 0.3991855772440667,
      "learning_rate": 1.7716172892113326e-05,
      "loss": 1.0133,
      "step": 253
    },
    {
      "epoch": 0.900709219858156,
      "grad_norm": 0.3837234039192193,
      "learning_rate": 1.768984799815328e-05,
      "loss": 0.993,
      "step": 254
    },
    {
      "epoch": 0.9042553191489362,
      "grad_norm": 0.42478259818829744,
      "learning_rate": 1.7663392051011156e-05,
      "loss": 1.0383,
      "step": 255
    },
    {
      "epoch": 0.9078014184397163,
      "grad_norm": 0.37024984262507593,
      "learning_rate": 1.7636805501558804e-05,
      "loss": 1.0335,
      "step": 256
    },
    {
      "epoch": 0.9113475177304965,
      "grad_norm": 0.4073494702670319,
      "learning_rate": 1.7610088802893867e-05,
      "loss": 1.0341,
      "step": 257
    },
    {
      "epoch": 0.9148936170212766,
      "grad_norm": 0.41045424811733167,
      "learning_rate": 1.7583242410332016e-05,
      "loss": 1.0435,
      "step": 258
    },
    {
      "epoch": 0.9184397163120568,
      "grad_norm": 0.35484485597404164,
      "learning_rate": 1.7556266781399227e-05,
      "loss": 1.0778,
      "step": 259
    },
    {
      "epoch": 0.9219858156028369,
      "grad_norm": 0.39414786905368276,
      "learning_rate": 1.752916237582396e-05,
      "loss": 0.9929,
      "step": 260
    },
    {
      "epoch": 0.925531914893617,
      "grad_norm": 0.3779847959746389,
      "learning_rate": 1.7501929655529344e-05,
      "loss": 1.0518,
      "step": 261
    },
    {
      "epoch": 0.9290780141843972,
      "grad_norm": 0.4023732096344426,
      "learning_rate": 1.7474569084625282e-05,
      "loss": 1.024,
      "step": 262
    },
    {
      "epoch": 0.9326241134751773,
      "grad_norm": 0.38745144745402504,
      "learning_rate": 1.7447081129400562e-05,
      "loss": 1.0152,
      "step": 263
    },
    {
      "epoch": 0.9361702127659575,
      "grad_norm": 0.35292649898123035,
      "learning_rate": 1.74194662583149e-05,
      "loss": 1.0142,
      "step": 264
    },
    {
      "epoch": 0.9397163120567376,
      "grad_norm": 0.41603250139588377,
      "learning_rate": 1.7391724941990952e-05,
      "loss": 1.0182,
      "step": 265
    },
    {
      "epoch": 0.9432624113475178,
      "grad_norm": 0.36463091568100964,
      "learning_rate": 1.7363857653206307e-05,
      "loss": 1.0292,
      "step": 266
    },
    {
      "epoch": 0.9468085106382979,
      "grad_norm": 0.40201841807122,
      "learning_rate": 1.7335864866885424e-05,
      "loss": 1.027,
      "step": 267
    },
    {
      "epoch": 0.950354609929078,
      "grad_norm": 0.4073367489856179,
      "learning_rate": 1.7307747060091525e-05,
      "loss": 0.989,
      "step": 268
    },
    {
      "epoch": 0.9539007092198581,
      "grad_norm": 0.4328837367699121,
      "learning_rate": 1.7279504712018495e-05,
      "loss": 1.0748,
      "step": 269
    },
    {
      "epoch": 0.9574468085106383,
      "grad_norm": 0.3992266115353178,
      "learning_rate": 1.7251138303982675e-05,
      "loss": 1.0325,
      "step": 270
    },
    {
      "epoch": 0.9609929078014184,
      "grad_norm": 0.378699543152666,
      "learning_rate": 1.72226483194147e-05,
      "loss": 1.0129,
      "step": 271
    },
    {
      "epoch": 0.9645390070921985,
      "grad_norm": 0.36186590779535055,
      "learning_rate": 1.7194035243851225e-05,
      "loss": 0.9969,
      "step": 272
    },
    {
      "epoch": 0.9680851063829787,
      "grad_norm": 0.3752934769207618,
      "learning_rate": 1.7165299564926683e-05,
      "loss": 0.9828,
      "step": 273
    },
    {
      "epoch": 0.9716312056737588,
      "grad_norm": 0.40860538489185705,
      "learning_rate": 1.7136441772364943e-05,
      "loss": 1.0382,
      "step": 274
    },
    {
      "epoch": 0.975177304964539,
      "grad_norm": 0.3568597745180973,
      "learning_rate": 1.710746235797099e-05,
      "loss": 1.0251,
      "step": 275
    },
    {
      "epoch": 0.9787234042553191,
      "grad_norm": 0.4276859934230444,
      "learning_rate": 1.7078361815622518e-05,
      "loss": 1.0205,
      "step": 276
    },
    {
      "epoch": 0.9822695035460993,
      "grad_norm": 0.37449732400229285,
      "learning_rate": 1.704914064126155e-05,
      "loss": 1.0266,
      "step": 277
    },
    {
      "epoch": 0.9858156028368794,
      "grad_norm": 0.389616566339835,
      "learning_rate": 1.7019799332885944e-05,
      "loss": 1.0229,
      "step": 278
    },
    {
      "epoch": 0.9893617021276596,
      "grad_norm": 0.3879192737138395,
      "learning_rate": 1.6990338390540935e-05,
      "loss": 1.0243,
      "step": 279
    },
    {
      "epoch": 0.9929078014184397,
      "grad_norm": 0.41573175808695956,
      "learning_rate": 1.6960758316310597e-05,
      "loss": 1.0229,
      "step": 280
    },
    {
      "epoch": 0.9964539007092199,
      "grad_norm": 0.38073253927449124,
      "learning_rate": 1.69310596143093e-05,
      "loss": 0.9925,
      "step": 281
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4255526467602786,
      "learning_rate": 1.6901242790673108e-05,
      "loss": 1.026,
      "step": 282
    },
    {
      "epoch": 1.00354609929078,
      "grad_norm": 0.42162634204753163,
      "learning_rate": 1.6871308353551153e-05,
      "loss": 1.0029,
      "step": 283
    },
    {
      "epoch": 1.0070921985815602,
      "grad_norm": 0.4337284231713999,
      "learning_rate": 1.684125681309699e-05,
      "loss": 0.9403,
      "step": 284
    },
    {
      "epoch": 1.0106382978723405,
      "grad_norm": 0.3708412522789234,
      "learning_rate": 1.6811088681459876e-05,
      "loss": 0.9679,
      "step": 285
    },
    {
      "epoch": 1.0141843971631206,
      "grad_norm": 0.4547631877983983,
      "learning_rate": 1.678080447277608e-05,
      "loss": 0.998,
      "step": 286
    },
    {
      "epoch": 1.0177304964539007,
      "grad_norm": 0.376112068736744,
      "learning_rate": 1.6750404703160063e-05,
      "loss": 0.993,
      "step": 287
    },
    {
      "epoch": 1.0212765957446808,
      "grad_norm": 0.4164541233002394,
      "learning_rate": 1.6719889890695756e-05,
      "loss": 0.9793,
      "step": 288
    },
    {
      "epoch": 1.024822695035461,
      "grad_norm": 0.4464551981366061,
      "learning_rate": 1.6689260555427666e-05,
      "loss": 0.9713,
      "step": 289
    },
    {
      "epoch": 1.0283687943262412,
      "grad_norm": 0.41796256047468283,
      "learning_rate": 1.665851721935205e-05,
      "loss": 0.9949,
      "step": 290
    },
    {
      "epoch": 1.0319148936170213,
      "grad_norm": 0.3889331388007683,
      "learning_rate": 1.6627660406408e-05,
      "loss": 1.0553,
      "step": 291
    },
    {
      "epoch": 1.0354609929078014,
      "grad_norm": 0.42334438714966177,
      "learning_rate": 1.6596690642468537e-05,
      "loss": 0.9754,
      "step": 292
    },
    {
      "epoch": 1.0390070921985815,
      "grad_norm": 0.42310995735693097,
      "learning_rate": 1.6565608455331622e-05,
      "loss": 0.9374,
      "step": 293
    },
    {
      "epoch": 1.0425531914893618,
      "grad_norm": 0.3692897902651878,
      "learning_rate": 1.6534414374711167e-05,
      "loss": 0.9771,
      "step": 294
    },
    {
      "epoch": 1.0460992907801419,
      "grad_norm": 0.3925797062980314,
      "learning_rate": 1.650310893222803e-05,
      "loss": 0.9761,
      "step": 295
    },
    {
      "epoch": 1.049645390070922,
      "grad_norm": 0.39833455353073705,
      "learning_rate": 1.6471692661400925e-05,
      "loss": 1.0217,
      "step": 296
    },
    {
      "epoch": 1.053191489361702,
      "grad_norm": 0.3819281134424317,
      "learning_rate": 1.6440166097637355e-05,
      "loss": 0.9801,
      "step": 297
    },
    {
      "epoch": 1.0567375886524824,
      "grad_norm": 0.3740823723188146,
      "learning_rate": 1.640852977822446e-05,
      "loss": 0.9807,
      "step": 298
    },
    {
      "epoch": 1.0602836879432624,
      "grad_norm": 0.3935831640578161,
      "learning_rate": 1.637678424231989e-05,
      "loss": 0.9534,
      "step": 299
    },
    {
      "epoch": 1.0638297872340425,
      "grad_norm": 0.4130459791663398,
      "learning_rate": 1.6344930030942593e-05,
      "loss": 0.9945,
      "step": 300
    },
    {
      "epoch": 1.0673758865248226,
      "grad_norm": 0.4172895714087477,
      "learning_rate": 1.6312967686963606e-05,
      "loss": 0.968,
      "step": 301
    },
    {
      "epoch": 1.070921985815603,
      "grad_norm": 0.3854672364656851,
      "learning_rate": 1.6280897755096805e-05,
      "loss": 0.961,
      "step": 302
    },
    {
      "epoch": 1.074468085106383,
      "grad_norm": 0.4336599329155487,
      "learning_rate": 1.6248720781889612e-05,
      "loss": 0.9568,
      "step": 303
    },
    {
      "epoch": 1.0780141843971631,
      "grad_norm": 0.42322982607889414,
      "learning_rate": 1.6216437315713693e-05,
      "loss": 0.9821,
      "step": 304
    },
    {
      "epoch": 1.0815602836879432,
      "grad_norm": 0.3974435049130651,
      "learning_rate": 1.61840479067556e-05,
      "loss": 0.9453,
      "step": 305
    },
    {
      "epoch": 1.0851063829787233,
      "grad_norm": 0.4019963999487,
      "learning_rate": 1.6151553107007395e-05,
      "loss": 0.9374,
      "step": 306
    },
    {
      "epoch": 1.0886524822695036,
      "grad_norm": 0.3932799513813298,
      "learning_rate": 1.6118953470257267e-05,
      "loss": 0.9708,
      "step": 307
    },
    {
      "epoch": 1.0921985815602837,
      "grad_norm": 0.3986498595695824,
      "learning_rate": 1.608624955208006e-05,
      "loss": 0.9646,
      "step": 308
    },
    {
      "epoch": 1.0957446808510638,
      "grad_norm": 0.368172249858901,
      "learning_rate": 1.6053441909827823e-05,
      "loss": 0.995,
      "step": 309
    },
    {
      "epoch": 1.099290780141844,
      "grad_norm": 0.4263975876112546,
      "learning_rate": 1.6020531102620303e-05,
      "loss": 1.0011,
      "step": 310
    },
    {
      "epoch": 1.1028368794326242,
      "grad_norm": 0.3482330087299883,
      "learning_rate": 1.5987517691335443e-05,
      "loss": 0.9654,
      "step": 311
    },
    {
      "epoch": 1.1063829787234043,
      "grad_norm": 0.4368586512892545,
      "learning_rate": 1.595440223859977e-05,
      "loss": 0.9683,
      "step": 312
    },
    {
      "epoch": 1.1099290780141844,
      "grad_norm": 0.35850477996330843,
      "learning_rate": 1.5921185308778863e-05,
      "loss": 0.97,
      "step": 313
    },
    {
      "epoch": 1.1134751773049645,
      "grad_norm": 0.4255767896591614,
      "learning_rate": 1.5887867467967703e-05,
      "loss": 0.9879,
      "step": 314
    },
    {
      "epoch": 1.1170212765957448,
      "grad_norm": 0.37321781456573017,
      "learning_rate": 1.585444928398103e-05,
      "loss": 0.9675,
      "step": 315
    },
    {
      "epoch": 1.1205673758865249,
      "grad_norm": 0.37089000074077766,
      "learning_rate": 1.5820931326343673e-05,
      "loss": 0.9362,
      "step": 316
    },
    {
      "epoch": 1.124113475177305,
      "grad_norm": 0.5469094383974885,
      "learning_rate": 1.5787314166280833e-05,
      "loss": 0.9871,
      "step": 317
    },
    {
      "epoch": 1.127659574468085,
      "grad_norm": 0.3929479871521509,
      "learning_rate": 1.5753598376708365e-05,
      "loss": 0.952,
      "step": 318
    },
    {
      "epoch": 1.1312056737588652,
      "grad_norm": 0.3861132355230585,
      "learning_rate": 1.5719784532223e-05,
      "loss": 0.9804,
      "step": 319
    },
    {
      "epoch": 1.1347517730496455,
      "grad_norm": 0.3699956189442316,
      "learning_rate": 1.5685873209092547e-05,
      "loss": 0.9658,
      "step": 320
    },
    {
      "epoch": 1.1382978723404256,
      "grad_norm": 0.4266810239187451,
      "learning_rate": 1.5651864985246095e-05,
      "loss": 0.9693,
      "step": 321
    },
    {
      "epoch": 1.1418439716312057,
      "grad_norm": 0.38488281487698434,
      "learning_rate": 1.5617760440264143e-05,
      "loss": 0.9705,
      "step": 322
    },
    {
      "epoch": 1.1453900709219857,
      "grad_norm": 0.43051269351008986,
      "learning_rate": 1.558356015536873e-05,
      "loss": 0.9813,
      "step": 323
    },
    {
      "epoch": 1.148936170212766,
      "grad_norm": 0.3894392288886495,
      "learning_rate": 1.5549264713413536e-05,
      "loss": 0.9521,
      "step": 324
    },
    {
      "epoch": 1.1524822695035462,
      "grad_norm": 0.41603671070271653,
      "learning_rate": 1.551487469887393e-05,
      "loss": 0.966,
      "step": 325
    },
    {
      "epoch": 1.1560283687943262,
      "grad_norm": 0.4076462857668706,
      "learning_rate": 1.5480390697837033e-05,
      "loss": 1.0079,
      "step": 326
    },
    {
      "epoch": 1.1595744680851063,
      "grad_norm": 0.38313030955774846,
      "learning_rate": 1.5445813297991713e-05,
      "loss": 0.9585,
      "step": 327
    },
    {
      "epoch": 1.1631205673758864,
      "grad_norm": 0.4134651737788734,
      "learning_rate": 1.541114308861857e-05,
      "loss": 0.9736,
      "step": 328
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 0.4212975648234023,
      "learning_rate": 1.5376380660579907e-05,
      "loss": 1.0212,
      "step": 329
    },
    {
      "epoch": 1.1702127659574468,
      "grad_norm": 0.39702588616324036,
      "learning_rate": 1.5341526606309646e-05,
      "loss": 0.9744,
      "step": 330
    },
    {
      "epoch": 1.173758865248227,
      "grad_norm": 0.3720449347403283,
      "learning_rate": 1.5306581519803233e-05,
      "loss": 0.9566,
      "step": 331
    },
    {
      "epoch": 1.177304964539007,
      "grad_norm": 0.40095384584626675,
      "learning_rate": 1.5271545996607525e-05,
      "loss": 1.0078,
      "step": 332
    },
    {
      "epoch": 1.1808510638297873,
      "grad_norm": 0.4089399913580165,
      "learning_rate": 1.5236420633810633e-05,
      "loss": 0.9953,
      "step": 333
    },
    {
      "epoch": 1.1843971631205674,
      "grad_norm": 0.4073214342319023,
      "learning_rate": 1.520120603003174e-05,
      "loss": 1.0064,
      "step": 334
    },
    {
      "epoch": 1.1879432624113475,
      "grad_norm": 0.39061623369847304,
      "learning_rate": 1.516590278541092e-05,
      "loss": 0.9624,
      "step": 335
    },
    {
      "epoch": 1.1914893617021276,
      "grad_norm": 0.43052959263616075,
      "learning_rate": 1.513051150159888e-05,
      "loss": 0.9548,
      "step": 336
    },
    {
      "epoch": 1.1950354609929077,
      "grad_norm": 0.4235116096113713,
      "learning_rate": 1.5095032781746732e-05,
      "loss": 0.9411,
      "step": 337
    },
    {
      "epoch": 1.198581560283688,
      "grad_norm": 0.38060647472893777,
      "learning_rate": 1.5059467230495703e-05,
      "loss": 0.9588,
      "step": 338
    },
    {
      "epoch": 1.202127659574468,
      "grad_norm": 0.48596035224216816,
      "learning_rate": 1.5023815453966837e-05,
      "loss": 0.9988,
      "step": 339
    },
    {
      "epoch": 1.2056737588652482,
      "grad_norm": 0.4167021821865609,
      "learning_rate": 1.4988078059750652e-05,
      "loss": 1.0026,
      "step": 340
    },
    {
      "epoch": 1.2092198581560283,
      "grad_norm": 0.44545889732960464,
      "learning_rate": 1.4952255656896801e-05,
      "loss": 0.9888,
      "step": 341
    },
    {
      "epoch": 1.2127659574468086,
      "grad_norm": 0.3918874394085114,
      "learning_rate": 1.4916348855903679e-05,
      "loss": 0.9889,
      "step": 342
    },
    {
      "epoch": 1.2163120567375887,
      "grad_norm": 0.36397855903707554,
      "learning_rate": 1.488035826870803e-05,
      "loss": 0.9798,
      "step": 343
    },
    {
      "epoch": 1.2198581560283688,
      "grad_norm": 0.4098853286858338,
      "learning_rate": 1.484428450867451e-05,
      "loss": 0.9582,
      "step": 344
    },
    {
      "epoch": 1.2234042553191489,
      "grad_norm": 0.38453575428102105,
      "learning_rate": 1.480812819058524e-05,
      "loss": 1.0062,
      "step": 345
    },
    {
      "epoch": 1.226950354609929,
      "grad_norm": 0.433496258525714,
      "learning_rate": 1.4771889930629317e-05,
      "loss": 0.9442,
      "step": 346
    },
    {
      "epoch": 1.2304964539007093,
      "grad_norm": 0.38804486850428566,
      "learning_rate": 1.473557034639233e-05,
      "loss": 0.9674,
      "step": 347
    },
    {
      "epoch": 1.2340425531914894,
      "grad_norm": 0.35527739750700876,
      "learning_rate": 1.4699170056845822e-05,
      "loss": 0.9545,
      "step": 348
    },
    {
      "epoch": 1.2375886524822695,
      "grad_norm": 0.38950854347971275,
      "learning_rate": 1.4662689682336746e-05,
      "loss": 0.9605,
      "step": 349
    },
    {
      "epoch": 1.2411347517730495,
      "grad_norm": 0.3832098944579604,
      "learning_rate": 1.4626129844576894e-05,
      "loss": 0.9749,
      "step": 350
    },
    {
      "epoch": 1.2446808510638299,
      "grad_norm": 0.4336296794079613,
      "learning_rate": 1.4589491166632292e-05,
      "loss": 0.9415,
      "step": 351
    },
    {
      "epoch": 1.24822695035461,
      "grad_norm": 0.40843255555461383,
      "learning_rate": 1.4552774272912602e-05,
      "loss": 0.9766,
      "step": 352
    },
    {
      "epoch": 1.25177304964539,
      "grad_norm": 0.44986141537811064,
      "learning_rate": 1.4515979789160455e-05,
      "loss": 0.9856,
      "step": 353
    },
    {
      "epoch": 1.2553191489361701,
      "grad_norm": 0.458707236262281,
      "learning_rate": 1.4479108342440804e-05,
      "loss": 0.9937,
      "step": 354
    },
    {
      "epoch": 1.2588652482269502,
      "grad_norm": 0.40472496228744964,
      "learning_rate": 1.4442160561130238e-05,
      "loss": 0.9655,
      "step": 355
    },
    {
      "epoch": 1.2624113475177305,
      "grad_norm": 0.39162624954597314,
      "learning_rate": 1.4405137074906259e-05,
      "loss": 0.9413,
      "step": 356
    },
    {
      "epoch": 1.2659574468085106,
      "grad_norm": 0.4534609974783642,
      "learning_rate": 1.4368038514736566e-05,
      "loss": 0.9883,
      "step": 357
    },
    {
      "epoch": 1.2695035460992907,
      "grad_norm": 0.3718189815457106,
      "learning_rate": 1.4330865512868293e-05,
      "loss": 0.9619,
      "step": 358
    },
    {
      "epoch": 1.273049645390071,
      "grad_norm": 0.4649383944211333,
      "learning_rate": 1.429361870281724e-05,
      "loss": 1.0076,
      "step": 359
    },
    {
      "epoch": 1.2765957446808511,
      "grad_norm": 0.3949660291531871,
      "learning_rate": 1.4256298719357063e-05,
      "loss": 0.9752,
      "step": 360
    },
    {
      "epoch": 1.2801418439716312,
      "grad_norm": 0.4077416176647928,
      "learning_rate": 1.421890619850848e-05,
      "loss": 0.9496,
      "step": 361
    },
    {
      "epoch": 1.2836879432624113,
      "grad_norm": 0.3895313217970244,
      "learning_rate": 1.4181441777528402e-05,
      "loss": 0.9754,
      "step": 362
    },
    {
      "epoch": 1.2872340425531914,
      "grad_norm": 0.4495084176311799,
      "learning_rate": 1.4143906094899103e-05,
      "loss": 0.9408,
      "step": 363
    },
    {
      "epoch": 1.2907801418439715,
      "grad_norm": 0.40503459294390115,
      "learning_rate": 1.410629979031731e-05,
      "loss": 0.9305,
      "step": 364
    },
    {
      "epoch": 1.2943262411347518,
      "grad_norm": 0.4102179939736063,
      "learning_rate": 1.4068623504683334e-05,
      "loss": 0.9684,
      "step": 365
    },
    {
      "epoch": 1.297872340425532,
      "grad_norm": 0.3987705956597129,
      "learning_rate": 1.4030877880090104e-05,
      "loss": 1.0156,
      "step": 366
    },
    {
      "epoch": 1.301418439716312,
      "grad_norm": 0.42916989583146437,
      "learning_rate": 1.3993063559812268e-05,
      "loss": 0.9572,
      "step": 367
    },
    {
      "epoch": 1.3049645390070923,
      "grad_norm": 0.3909945273536003,
      "learning_rate": 1.3955181188295203e-05,
      "loss": 0.9646,
      "step": 368
    },
    {
      "epoch": 1.3085106382978724,
      "grad_norm": 0.44340897137841345,
      "learning_rate": 1.391723141114404e-05,
      "loss": 1.0004,
      "step": 369
    },
    {
      "epoch": 1.3120567375886525,
      "grad_norm": 0.38336773400029367,
      "learning_rate": 1.3879214875112665e-05,
      "loss": 0.9815,
      "step": 370
    },
    {
      "epoch": 1.3156028368794326,
      "grad_norm": 0.4587120808854636,
      "learning_rate": 1.3841132228092684e-05,
      "loss": 0.9451,
      "step": 371
    },
    {
      "epoch": 1.3191489361702127,
      "grad_norm": 0.42258328196940764,
      "learning_rate": 1.3802984119102403e-05,
      "loss": 0.9619,
      "step": 372
    },
    {
      "epoch": 1.322695035460993,
      "grad_norm": 0.397151037289473,
      "learning_rate": 1.3764771198275745e-05,
      "loss": 0.9866,
      "step": 373
    },
    {
      "epoch": 1.326241134751773,
      "grad_norm": 0.37879764171065744,
      "learning_rate": 1.3726494116851189e-05,
      "loss": 0.9678,
      "step": 374
    },
    {
      "epoch": 1.3297872340425532,
      "grad_norm": 0.4055657975764546,
      "learning_rate": 1.368815352716065e-05,
      "loss": 0.9504,
      "step": 375
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.37133045086049504,
      "learning_rate": 1.3649750082618387e-05,
      "loss": 0.9689,
      "step": 376
    },
    {
      "epoch": 1.3368794326241136,
      "grad_norm": 0.41029972517648766,
      "learning_rate": 1.361128443770984e-05,
      "loss": 0.9708,
      "step": 377
    },
    {
      "epoch": 1.3404255319148937,
      "grad_norm": 0.39076722012763027,
      "learning_rate": 1.3572757247980502e-05,
      "loss": 0.9743,
      "step": 378
    },
    {
      "epoch": 1.3439716312056738,
      "grad_norm": 0.3448538829992877,
      "learning_rate": 1.353416917002473e-05,
      "loss": 0.9395,
      "step": 379
    },
    {
      "epoch": 1.3475177304964538,
      "grad_norm": 0.37047746355086686,
      "learning_rate": 1.3495520861474565e-05,
      "loss": 0.9485,
      "step": 380
    },
    {
      "epoch": 1.351063829787234,
      "grad_norm": 0.42680045310215664,
      "learning_rate": 1.3456812980988513e-05,
      "loss": 0.9995,
      "step": 381
    },
    {
      "epoch": 1.3546099290780143,
      "grad_norm": 0.3414703460078149,
      "learning_rate": 1.3418046188240328e-05,
      "loss": 0.9613,
      "step": 382
    },
    {
      "epoch": 1.3581560283687943,
      "grad_norm": 0.3829959439836121,
      "learning_rate": 1.3379221143907775e-05,
      "loss": 0.9859,
      "step": 383
    },
    {
      "epoch": 1.3617021276595744,
      "grad_norm": 0.3701947232252572,
      "learning_rate": 1.3340338509661355e-05,
      "loss": 0.9833,
      "step": 384
    },
    {
      "epoch": 1.3652482269503547,
      "grad_norm": 0.36949452405651734,
      "learning_rate": 1.3301398948153042e-05,
      "loss": 0.95,
      "step": 385
    },
    {
      "epoch": 1.3687943262411348,
      "grad_norm": 0.3611222088473356,
      "learning_rate": 1.326240312300498e-05,
      "loss": 0.9872,
      "step": 386
    },
    {
      "epoch": 1.372340425531915,
      "grad_norm": 0.3859473860907529,
      "learning_rate": 1.3223351698798186e-05,
      "loss": 0.9391,
      "step": 387
    },
    {
      "epoch": 1.375886524822695,
      "grad_norm": 0.3582103645991868,
      "learning_rate": 1.318424534106121e-05,
      "loss": 0.9636,
      "step": 388
    },
    {
      "epoch": 1.3794326241134751,
      "grad_norm": 0.3366736526064197,
      "learning_rate": 1.3145084716258804e-05,
      "loss": 0.9617,
      "step": 389
    },
    {
      "epoch": 1.3829787234042552,
      "grad_norm": 0.36060765339239736,
      "learning_rate": 1.3105870491780559e-05,
      "loss": 0.9599,
      "step": 390
    },
    {
      "epoch": 1.3865248226950355,
      "grad_norm": 0.398088313286727,
      "learning_rate": 1.3066603335929522e-05,
      "loss": 0.981,
      "step": 391
    },
    {
      "epoch": 1.3900709219858156,
      "grad_norm": 0.36529659957165594,
      "learning_rate": 1.302728391791083e-05,
      "loss": 0.964,
      "step": 392
    },
    {
      "epoch": 1.3936170212765957,
      "grad_norm": 0.4007319301068806,
      "learning_rate": 1.2987912907820274e-05,
      "loss": 0.9867,
      "step": 393
    },
    {
      "epoch": 1.397163120567376,
      "grad_norm": 0.4247614902726926,
      "learning_rate": 1.2948490976632914e-05,
      "loss": 0.9446,
      "step": 394
    },
    {
      "epoch": 1.400709219858156,
      "grad_norm": 0.3547395206863501,
      "learning_rate": 1.2909018796191616e-05,
      "loss": 0.9643,
      "step": 395
    },
    {
      "epoch": 1.4042553191489362,
      "grad_norm": 0.6673933827613349,
      "learning_rate": 1.2869497039195609e-05,
      "loss": 0.9537,
      "step": 396
    },
    {
      "epoch": 1.4078014184397163,
      "grad_norm": 0.35359698348808816,
      "learning_rate": 1.2829926379189025e-05,
      "loss": 0.9844,
      "step": 397
    },
    {
      "epoch": 1.4113475177304964,
      "grad_norm": 0.36149036424597636,
      "learning_rate": 1.2790307490549432e-05,
      "loss": 0.9827,
      "step": 398
    },
    {
      "epoch": 1.4148936170212765,
      "grad_norm": 0.3831897208194427,
      "learning_rate": 1.2750641048476309e-05,
      "loss": 0.9742,
      "step": 399
    },
    {
      "epoch": 1.4184397163120568,
      "grad_norm": 0.3871487792244679,
      "learning_rate": 1.2710927728979568e-05,
      "loss": 0.953,
      "step": 400
    },
    {
      "epoch": 1.4219858156028369,
      "grad_norm": 0.37024401544958657,
      "learning_rate": 1.2671168208868024e-05,
      "loss": 0.9524,
      "step": 401
    },
    {
      "epoch": 1.425531914893617,
      "grad_norm": 0.4267642235307254,
      "learning_rate": 1.2631363165737854e-05,
      "loss": 0.9504,
      "step": 402
    },
    {
      "epoch": 1.4290780141843973,
      "grad_norm": 0.3381259448637118,
      "learning_rate": 1.2591513277961058e-05,
      "loss": 0.9685,
      "step": 403
    },
    {
      "epoch": 1.4326241134751774,
      "grad_norm": 0.3623948447486684,
      "learning_rate": 1.2551619224673894e-05,
      "loss": 0.9721,
      "step": 404
    },
    {
      "epoch": 1.4361702127659575,
      "grad_norm": 0.3376920646108776,
      "learning_rate": 1.2511681685765308e-05,
      "loss": 0.9697,
      "step": 405
    },
    {
      "epoch": 1.4397163120567376,
      "grad_norm": 0.3711071679343133,
      "learning_rate": 1.2471701341865342e-05,
      "loss": 0.9814,
      "step": 406
    },
    {
      "epoch": 1.4432624113475176,
      "grad_norm": 0.3650675243183325,
      "learning_rate": 1.2431678874333532e-05,
      "loss": 0.9713,
      "step": 407
    },
    {
      "epoch": 1.4468085106382977,
      "grad_norm": 0.3862656526697159,
      "learning_rate": 1.2391614965247307e-05,
      "loss": 0.9974,
      "step": 408
    },
    {
      "epoch": 1.450354609929078,
      "grad_norm": 0.3532038311894071,
      "learning_rate": 1.2351510297390348e-05,
      "loss": 0.9807,
      "step": 409
    },
    {
      "epoch": 1.4539007092198581,
      "grad_norm": 0.3467889591941661,
      "learning_rate": 1.2311365554240972e-05,
      "loss": 0.9281,
      "step": 410
    },
    {
      "epoch": 1.4574468085106382,
      "grad_norm": 0.36275768850044293,
      "learning_rate": 1.2271181419960473e-05,
      "loss": 0.9743,
      "step": 411
    },
    {
      "epoch": 1.4609929078014185,
      "grad_norm": 0.35517643206705524,
      "learning_rate": 1.2230958579381458e-05,
      "loss": 0.9801,
      "step": 412
    },
    {
      "epoch": 1.4645390070921986,
      "grad_norm": 0.3877376505719052,
      "learning_rate": 1.2190697717996179e-05,
      "loss": 0.992,
      "step": 413
    },
    {
      "epoch": 1.4680851063829787,
      "grad_norm": 0.3562682093151937,
      "learning_rate": 1.2150399521944858e-05,
      "loss": 0.9653,
      "step": 414
    },
    {
      "epoch": 1.4716312056737588,
      "grad_norm": 0.3604528130360619,
      "learning_rate": 1.211006467800399e-05,
      "loss": 0.9431,
      "step": 415
    },
    {
      "epoch": 1.475177304964539,
      "grad_norm": 0.3612796477592489,
      "learning_rate": 1.2069693873574628e-05,
      "loss": 0.9477,
      "step": 416
    },
    {
      "epoch": 1.4787234042553192,
      "grad_norm": 0.34195694361156626,
      "learning_rate": 1.2029287796670684e-05,
      "loss": 0.9421,
      "step": 417
    },
    {
      "epoch": 1.4822695035460993,
      "grad_norm": 0.38109197935236333,
      "learning_rate": 1.1988847135907188e-05,
      "loss": 0.9541,
      "step": 418
    },
    {
      "epoch": 1.4858156028368794,
      "grad_norm": 0.3498732999579304,
      "learning_rate": 1.1948372580488574e-05,
      "loss": 0.9705,
      "step": 419
    },
    {
      "epoch": 1.4893617021276595,
      "grad_norm": 0.39752070002562495,
      "learning_rate": 1.190786482019691e-05,
      "loss": 0.935,
      "step": 420
    },
    {
      "epoch": 1.4929078014184398,
      "grad_norm": 0.3524932281108635,
      "learning_rate": 1.1867324545380159e-05,
      "loss": 0.9549,
      "step": 421
    },
    {
      "epoch": 1.49645390070922,
      "grad_norm": 0.406904792088018,
      "learning_rate": 1.1826752446940401e-05,
      "loss": 0.9501,
      "step": 422
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.3836320671730948,
      "learning_rate": 1.1786149216322079e-05,
      "loss": 0.9843,
      "step": 423
    },
    {
      "epoch": 1.50354609929078,
      "grad_norm": 0.3453064612911423,
      "learning_rate": 1.1745515545500186e-05,
      "loss": 0.988,
      "step": 424
    },
    {
      "epoch": 1.5070921985815602,
      "grad_norm": 0.4169790496181414,
      "learning_rate": 1.17048521269685e-05,
      "loss": 0.9426,
      "step": 425
    },
    {
      "epoch": 1.5106382978723403,
      "grad_norm": 0.35431862853342666,
      "learning_rate": 1.1664159653727767e-05,
      "loss": 1.0007,
      "step": 426
    },
    {
      "epoch": 1.5141843971631206,
      "grad_norm": 0.4403221223658087,
      "learning_rate": 1.1623438819273887e-05,
      "loss": 0.9394,
      "step": 427
    },
    {
      "epoch": 1.5177304964539007,
      "grad_norm": 0.35909026402669597,
      "learning_rate": 1.1582690317586116e-05,
      "loss": 0.9688,
      "step": 428
    },
    {
      "epoch": 1.521276595744681,
      "grad_norm": 0.3616088953094568,
      "learning_rate": 1.1541914843115212e-05,
      "loss": 0.9548,
      "step": 429
    },
    {
      "epoch": 1.524822695035461,
      "grad_norm": 0.41651072781255166,
      "learning_rate": 1.1501113090771618e-05,
      "loss": 0.9434,
      "step": 430
    },
    {
      "epoch": 1.5283687943262412,
      "grad_norm": 0.3533959037818844,
      "learning_rate": 1.146028575591362e-05,
      "loss": 0.9627,
      "step": 431
    },
    {
      "epoch": 1.5319148936170213,
      "grad_norm": 0.3987483236015162,
      "learning_rate": 1.1419433534335483e-05,
      "loss": 0.9786,
      "step": 432
    },
    {
      "epoch": 1.5354609929078014,
      "grad_norm": 0.39355195566700935,
      "learning_rate": 1.1378557122255596e-05,
      "loss": 0.9607,
      "step": 433
    },
    {
      "epoch": 1.5390070921985815,
      "grad_norm": 0.3719202018138273,
      "learning_rate": 1.133765721630462e-05,
      "loss": 0.9938,
      "step": 434
    },
    {
      "epoch": 1.5425531914893615,
      "grad_norm": 0.3553320294386332,
      "learning_rate": 1.129673451351361e-05,
      "loss": 0.9611,
      "step": 435
    },
    {
      "epoch": 1.5460992907801419,
      "grad_norm": 0.4083603203915794,
      "learning_rate": 1.1255789711302123e-05,
      "loss": 0.9735,
      "step": 436
    },
    {
      "epoch": 1.549645390070922,
      "grad_norm": 0.3653527658924869,
      "learning_rate": 1.1214823507466345e-05,
      "loss": 0.9837,
      "step": 437
    },
    {
      "epoch": 1.5531914893617023,
      "grad_norm": 0.3737537963108637,
      "learning_rate": 1.1173836600167202e-05,
      "loss": 0.9586,
      "step": 438
    },
    {
      "epoch": 1.5567375886524824,
      "grad_norm": 0.393363237627448,
      "learning_rate": 1.1132829687918456e-05,
      "loss": 0.9328,
      "step": 439
    },
    {
      "epoch": 1.5602836879432624,
      "grad_norm": 0.3816275461313281,
      "learning_rate": 1.1091803469574789e-05,
      "loss": 0.9539,
      "step": 440
    },
    {
      "epoch": 1.5638297872340425,
      "grad_norm": 0.37772612682001405,
      "learning_rate": 1.1050758644319917e-05,
      "loss": 0.9284,
      "step": 441
    },
    {
      "epoch": 1.5673758865248226,
      "grad_norm": 0.37139778430983705,
      "learning_rate": 1.1009695911654651e-05,
      "loss": 0.9592,
      "step": 442
    },
    {
      "epoch": 1.5709219858156027,
      "grad_norm": 0.3789450188184618,
      "learning_rate": 1.0968615971385e-05,
      "loss": 0.9496,
      "step": 443
    },
    {
      "epoch": 1.574468085106383,
      "grad_norm": 0.3490092134445974,
      "learning_rate": 1.092751952361022e-05,
      "loss": 0.9441,
      "step": 444
    },
    {
      "epoch": 1.5780141843971631,
      "grad_norm": 0.37172081990209344,
      "learning_rate": 1.08864072687109e-05,
      "loss": 0.9737,
      "step": 445
    },
    {
      "epoch": 1.5815602836879432,
      "grad_norm": 0.3689748363727634,
      "learning_rate": 1.0845279907337016e-05,
      "loss": 0.9625,
      "step": 446
    },
    {
      "epoch": 1.5851063829787235,
      "grad_norm": 0.36769682217189764,
      "learning_rate": 1.0804138140395995e-05,
      "loss": 0.9448,
      "step": 447
    },
    {
      "epoch": 1.5886524822695036,
      "grad_norm": 0.36948009642812196,
      "learning_rate": 1.0762982669040772e-05,
      "loss": 0.9422,
      "step": 448
    },
    {
      "epoch": 1.5921985815602837,
      "grad_norm": 0.3874510000854089,
      "learning_rate": 1.0721814194657832e-05,
      "loss": 0.9559,
      "step": 449
    },
    {
      "epoch": 1.5957446808510638,
      "grad_norm": 0.35123043713553576,
      "learning_rate": 1.0680633418855266e-05,
      "loss": 0.9909,
      "step": 450
    },
    {
      "epoch": 1.599290780141844,
      "grad_norm": 0.3424584678806653,
      "learning_rate": 1.0639441043450809e-05,
      "loss": 0.987,
      "step": 451
    },
    {
      "epoch": 1.602836879432624,
      "grad_norm": 0.3540879822796454,
      "learning_rate": 1.0598237770459887e-05,
      "loss": 0.9642,
      "step": 452
    },
    {
      "epoch": 1.6063829787234043,
      "grad_norm": 0.341535695117283,
      "learning_rate": 1.0557024302083635e-05,
      "loss": 0.9458,
      "step": 453
    },
    {
      "epoch": 1.6099290780141844,
      "grad_norm": 0.3401330561400603,
      "learning_rate": 1.0515801340696946e-05,
      "loss": 0.9537,
      "step": 454
    },
    {
      "epoch": 1.6134751773049647,
      "grad_norm": 0.3793952706621045,
      "learning_rate": 1.0474569588836503e-05,
      "loss": 0.9788,
      "step": 455
    },
    {
      "epoch": 1.6170212765957448,
      "grad_norm": 0.3601958693890712,
      "learning_rate": 1.043332974918879e-05,
      "loss": 0.9419,
      "step": 456
    },
    {
      "epoch": 1.6205673758865249,
      "grad_norm": 0.36751037561946204,
      "learning_rate": 1.0392082524578135e-05,
      "loss": 0.954,
      "step": 457
    },
    {
      "epoch": 1.624113475177305,
      "grad_norm": 0.3996259740255699,
      "learning_rate": 1.0350828617954712e-05,
      "loss": 0.97,
      "step": 458
    },
    {
      "epoch": 1.627659574468085,
      "grad_norm": 0.3742204360804281,
      "learning_rate": 1.0309568732382582e-05,
      "loss": 0.9609,
      "step": 459
    },
    {
      "epoch": 1.6312056737588652,
      "grad_norm": 0.3363933659013848,
      "learning_rate": 1.0268303571027696e-05,
      "loss": 0.9292,
      "step": 460
    },
    {
      "epoch": 1.6347517730496453,
      "grad_norm": 0.3852661006055739,
      "learning_rate": 1.0227033837145924e-05,
      "loss": 0.9579,
      "step": 461
    },
    {
      "epoch": 1.6382978723404256,
      "grad_norm": 0.3594510036459689,
      "learning_rate": 1.0185760234071054e-05,
      "loss": 0.9486,
      "step": 462
    },
    {
      "epoch": 1.6418439716312057,
      "grad_norm": 0.38093406145891096,
      "learning_rate": 1.0144483465202818e-05,
      "loss": 0.9515,
      "step": 463
    },
    {
      "epoch": 1.645390070921986,
      "grad_norm": 0.37391762202041084,
      "learning_rate": 1.01032042339949e-05,
      "loss": 0.951,
      "step": 464
    },
    {
      "epoch": 1.648936170212766,
      "grad_norm": 0.3539528727618403,
      "learning_rate": 1.0061923243942947e-05,
      "loss": 0.9668,
      "step": 465
    },
    {
      "epoch": 1.6524822695035462,
      "grad_norm": 0.3630678066448606,
      "learning_rate": 1.0020641198572582e-05,
      "loss": 0.9669,
      "step": 466
    },
    {
      "epoch": 1.6560283687943262,
      "grad_norm": 0.3963323235424878,
      "learning_rate": 9.97935880142742e-06,
      "loss": 0.9728,
      "step": 467
    },
    {
      "epoch": 1.6595744680851063,
      "grad_norm": 0.3708572491271238,
      "learning_rate": 9.938076756057055e-06,
      "loss": 0.9762,
      "step": 468
    },
    {
      "epoch": 1.6631205673758864,
      "grad_norm": 0.3930914103727205,
      "learning_rate": 9.896795766005104e-06,
      "loss": 0.958,
      "step": 469
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.3878977246149519,
      "learning_rate": 9.855516534797187e-06,
      "loss": 0.9612,
      "step": 470
    },
    {
      "epoch": 1.6702127659574468,
      "grad_norm": 0.32502813361802174,
      "learning_rate": 9.81423976592895e-06,
      "loss": 0.954,
      "step": 471
    },
    {
      "epoch": 1.673758865248227,
      "grad_norm": 0.3470600424638187,
      "learning_rate": 9.77296616285408e-06,
      "loss": 0.9468,
      "step": 472
    },
    {
      "epoch": 1.6773049645390072,
      "grad_norm": 0.3998668324871742,
      "learning_rate": 9.731696428972304e-06,
      "loss": 0.983,
      "step": 473
    },
    {
      "epoch": 1.6808510638297873,
      "grad_norm": 0.3853130034264598,
      "learning_rate": 9.690431267617421e-06,
      "loss": 0.9814,
      "step": 474
    },
    {
      "epoch": 1.6843971631205674,
      "grad_norm": 0.3393026991437144,
      "learning_rate": 9.649171382045293e-06,
      "loss": 0.9785,
      "step": 475
    },
    {
      "epoch": 1.6879432624113475,
      "grad_norm": 0.40590543355182296,
      "learning_rate": 9.607917475421868e-06,
      "loss": 0.9443,
      "step": 476
    },
    {
      "epoch": 1.6914893617021276,
      "grad_norm": 0.3536310502781339,
      "learning_rate": 9.566670250811213e-06,
      "loss": 0.9698,
      "step": 477
    },
    {
      "epoch": 1.6950354609929077,
      "grad_norm": 0.3784372371857604,
      "learning_rate": 9.5254304111635e-06,
      "loss": 0.986,
      "step": 478
    },
    {
      "epoch": 1.6985815602836878,
      "grad_norm": 0.3688331189520331,
      "learning_rate": 9.484198659303057e-06,
      "loss": 0.9149,
      "step": 479
    },
    {
      "epoch": 1.702127659574468,
      "grad_norm": 0.34633104551777094,
      "learning_rate": 9.442975697916372e-06,
      "loss": 0.945,
      "step": 480
    },
    {
      "epoch": 1.7056737588652482,
      "grad_norm": 0.3689457615265894,
      "learning_rate": 9.401762229540116e-06,
      "loss": 0.9336,
      "step": 481
    },
    {
      "epoch": 1.7092198581560285,
      "grad_norm": 0.37664918461886804,
      "learning_rate": 9.360558956549193e-06,
      "loss": 0.927,
      "step": 482
    },
    {
      "epoch": 1.7127659574468086,
      "grad_norm": 0.34778225857336614,
      "learning_rate": 9.319366581144736e-06,
      "loss": 0.964,
      "step": 483
    },
    {
      "epoch": 1.7163120567375887,
      "grad_norm": 0.3527437985857996,
      "learning_rate": 9.278185805342171e-06,
      "loss": 0.9792,
      "step": 484
    },
    {
      "epoch": 1.7198581560283688,
      "grad_norm": 0.36930633702754206,
      "learning_rate": 9.237017330959233e-06,
      "loss": 0.9646,
      "step": 485
    },
    {
      "epoch": 1.7234042553191489,
      "grad_norm": 0.3321105431295584,
      "learning_rate": 9.195861859604008e-06,
      "loss": 0.9775,
      "step": 486
    },
    {
      "epoch": 1.726950354609929,
      "grad_norm": 0.336145975690601,
      "learning_rate": 9.154720092662986e-06,
      "loss": 0.9748,
      "step": 487
    },
    {
      "epoch": 1.7304964539007093,
      "grad_norm": 0.34486030423878333,
      "learning_rate": 9.113592731289102e-06,
      "loss": 0.9214,
      "step": 488
    },
    {
      "epoch": 1.7340425531914894,
      "grad_norm": 0.3432838041796281,
      "learning_rate": 9.072480476389782e-06,
      "loss": 0.9536,
      "step": 489
    },
    {
      "epoch": 1.7375886524822695,
      "grad_norm": 0.35613857468820703,
      "learning_rate": 9.031384028615005e-06,
      "loss": 1.0034,
      "step": 490
    },
    {
      "epoch": 1.7411347517730498,
      "grad_norm": 0.3361342737820547,
      "learning_rate": 8.99030408834535e-06,
      "loss": 0.9503,
      "step": 491
    },
    {
      "epoch": 1.7446808510638299,
      "grad_norm": 0.363558506677557,
      "learning_rate": 8.94924135568009e-06,
      "loss": 0.9651,
      "step": 492
    },
    {
      "epoch": 1.74822695035461,
      "grad_norm": 0.33621685011472696,
      "learning_rate": 8.908196530425213e-06,
      "loss": 0.986,
      "step": 493
    },
    {
      "epoch": 1.75177304964539,
      "grad_norm": 0.33671054356465846,
      "learning_rate": 8.867170312081546e-06,
      "loss": 1.0193,
      "step": 494
    },
    {
      "epoch": 1.7553191489361701,
      "grad_norm": 0.35209822799689444,
      "learning_rate": 8.8261633998328e-06,
      "loss": 0.9437,
      "step": 495
    },
    {
      "epoch": 1.7588652482269502,
      "grad_norm": 0.3553864898061415,
      "learning_rate": 8.785176492533656e-06,
      "loss": 0.9782,
      "step": 496
    },
    {
      "epoch": 1.7624113475177305,
      "grad_norm": 0.33174231265756776,
      "learning_rate": 8.744210288697882e-06,
      "loss": 0.9628,
      "step": 497
    },
    {
      "epoch": 1.7659574468085106,
      "grad_norm": 0.3409297743356347,
      "learning_rate": 8.703265486486391e-06,
      "loss": 0.9449,
      "step": 498
    },
    {
      "epoch": 1.7695035460992907,
      "grad_norm": 0.3445185706849665,
      "learning_rate": 8.662342783695381e-06,
      "loss": 0.9841,
      "step": 499
    },
    {
      "epoch": 1.773049645390071,
      "grad_norm": 0.3441733375131873,
      "learning_rate": 8.621442877744409e-06,
      "loss": 0.9841,
      "step": 500
    },
    {
      "epoch": 1.7765957446808511,
      "grad_norm": 0.35173263772752744,
      "learning_rate": 8.58056646566452e-06,
      "loss": 0.9635,
      "step": 501
    },
    {
      "epoch": 1.7801418439716312,
      "grad_norm": 0.3823010790918624,
      "learning_rate": 8.539714244086381e-06,
      "loss": 0.9493,
      "step": 502
    },
    {
      "epoch": 1.7836879432624113,
      "grad_norm": 0.3312383340054398,
      "learning_rate": 8.49888690922838e-06,
      "loss": 0.9869,
      "step": 503
    },
    {
      "epoch": 1.7872340425531914,
      "grad_norm": 0.3560759208074366,
      "learning_rate": 8.458085156884791e-06,
      "loss": 0.9595,
      "step": 504
    },
    {
      "epoch": 1.7907801418439715,
      "grad_norm": 0.36728196100689803,
      "learning_rate": 8.41730968241389e-06,
      "loss": 0.9542,
      "step": 505
    },
    {
      "epoch": 1.7943262411347518,
      "grad_norm": 0.32868509859724265,
      "learning_rate": 8.376561180726115e-06,
      "loss": 0.9237,
      "step": 506
    },
    {
      "epoch": 1.797872340425532,
      "grad_norm": 0.33553779719082805,
      "learning_rate": 8.33584034627224e-06,
      "loss": 0.9668,
      "step": 507
    },
    {
      "epoch": 1.8014184397163122,
      "grad_norm": 0.3592656530700375,
      "learning_rate": 8.295147873031502e-06,
      "loss": 0.9687,
      "step": 508
    },
    {
      "epoch": 1.8049645390070923,
      "grad_norm": 0.3330489518869409,
      "learning_rate": 8.254484454499817e-06,
      "loss": 0.9594,
      "step": 509
    },
    {
      "epoch": 1.8085106382978724,
      "grad_norm": 0.37275609779719715,
      "learning_rate": 8.213850783677926e-06,
      "loss": 0.9727,
      "step": 510
    },
    {
      "epoch": 1.8120567375886525,
      "grad_norm": 0.35997150953099766,
      "learning_rate": 8.1732475530596e-06,
      "loss": 0.9343,
      "step": 511
    },
    {
      "epoch": 1.8156028368794326,
      "grad_norm": 0.3382875860314504,
      "learning_rate": 8.132675454619846e-06,
      "loss": 0.9812,
      "step": 512
    },
    {
      "epoch": 1.8191489361702127,
      "grad_norm": 0.38581953129694097,
      "learning_rate": 8.092135179803091e-06,
      "loss": 0.92,
      "step": 513
    },
    {
      "epoch": 1.8226950354609928,
      "grad_norm": 0.38913155301147206,
      "learning_rate": 8.051627419511429e-06,
      "loss": 0.9934,
      "step": 514
    },
    {
      "epoch": 1.826241134751773,
      "grad_norm": 0.37743270636171417,
      "learning_rate": 8.011152864092812e-06,
      "loss": 0.9969,
      "step": 515
    },
    {
      "epoch": 1.8297872340425532,
      "grad_norm": 0.33196718140374704,
      "learning_rate": 7.97071220332932e-06,
      "loss": 0.9646,
      "step": 516
    },
    {
      "epoch": 1.8333333333333335,
      "grad_norm": 0.3618820601707212,
      "learning_rate": 7.930306126425377e-06,
      "loss": 0.911,
      "step": 517
    },
    {
      "epoch": 1.8368794326241136,
      "grad_norm": 0.36679621653721584,
      "learning_rate": 7.889935321996014e-06,
      "loss": 0.9445,
      "step": 518
    },
    {
      "epoch": 1.8404255319148937,
      "grad_norm": 0.3667701675004188,
      "learning_rate": 7.849600478055145e-06,
      "loss": 0.9695,
      "step": 519
    },
    {
      "epoch": 1.8439716312056738,
      "grad_norm": 0.34841171823590084,
      "learning_rate": 7.809302282003823e-06,
      "loss": 0.934,
      "step": 520
    },
    {
      "epoch": 1.8475177304964538,
      "grad_norm": 0.35865381928135853,
      "learning_rate": 7.769041420618545e-06,
      "loss": 0.9229,
      "step": 521
    },
    {
      "epoch": 1.851063829787234,
      "grad_norm": 0.3606496256848997,
      "learning_rate": 7.72881858003953e-06,
      "loss": 0.9691,
      "step": 522
    },
    {
      "epoch": 1.854609929078014,
      "grad_norm": 0.34646840430668596,
      "learning_rate": 7.688634445759026e-06,
      "loss": 0.9578,
      "step": 523
    },
    {
      "epoch": 1.8581560283687943,
      "grad_norm": 0.3619989219136004,
      "learning_rate": 7.648489702609653e-06,
      "loss": 0.9504,
      "step": 524
    },
    {
      "epoch": 1.8617021276595744,
      "grad_norm": 0.36867918774777586,
      "learning_rate": 7.608385034752696e-06,
      "loss": 0.9504,
      "step": 525
    },
    {
      "epoch": 1.8652482269503547,
      "grad_norm": 0.35376310088793333,
      "learning_rate": 7.56832112566647e-06,
      "loss": 0.9848,
      "step": 526
    },
    {
      "epoch": 1.8687943262411348,
      "grad_norm": 0.3862728350589978,
      "learning_rate": 7.528298658134658e-06,
      "loss": 0.9575,
      "step": 527
    },
    {
      "epoch": 1.872340425531915,
      "grad_norm": 0.322531140818402,
      "learning_rate": 7.48831831423469e-06,
      "loss": 0.9162,
      "step": 528
    },
    {
      "epoch": 1.875886524822695,
      "grad_norm": 0.37085381420014846,
      "learning_rate": 7.448380775326108e-06,
      "loss": 0.9554,
      "step": 529
    },
    {
      "epoch": 1.8794326241134751,
      "grad_norm": 0.35655546944032424,
      "learning_rate": 7.408486722038943e-06,
      "loss": 0.9625,
      "step": 530
    },
    {
      "epoch": 1.8829787234042552,
      "grad_norm": 0.3178457155060676,
      "learning_rate": 7.3686368342621496e-06,
      "loss": 0.9307,
      "step": 531
    },
    {
      "epoch": 1.8865248226950353,
      "grad_norm": 0.3406313695465732,
      "learning_rate": 7.32883179113198e-06,
      "loss": 0.9651,
      "step": 532
    },
    {
      "epoch": 1.8900709219858156,
      "grad_norm": 0.346901645086471,
      "learning_rate": 7.289072271020434e-06,
      "loss": 0.934,
      "step": 533
    },
    {
      "epoch": 1.8936170212765957,
      "grad_norm": 0.3663202173018621,
      "learning_rate": 7.249358951523693e-06,
      "loss": 0.9966,
      "step": 534
    },
    {
      "epoch": 1.897163120567376,
      "grad_norm": 0.33452988005856144,
      "learning_rate": 7.2096925094505695e-06,
      "loss": 0.9684,
      "step": 535
    },
    {
      "epoch": 1.900709219858156,
      "grad_norm": 0.3416578941577139,
      "learning_rate": 7.170073620810975e-06,
      "loss": 0.963,
      "step": 536
    },
    {
      "epoch": 1.9042553191489362,
      "grad_norm": 0.34915133717799923,
      "learning_rate": 7.130502960804397e-06,
      "loss": 0.9705,
      "step": 537
    },
    {
      "epoch": 1.9078014184397163,
      "grad_norm": 0.3873543770637985,
      "learning_rate": 7.090981203808387e-06,
      "loss": 0.9533,
      "step": 538
    },
    {
      "epoch": 1.9113475177304964,
      "grad_norm": 0.35578865044419283,
      "learning_rate": 7.051509023367087e-06,
      "loss": 0.9742,
      "step": 539
    },
    {
      "epoch": 1.9148936170212765,
      "grad_norm": 0.38862412698080284,
      "learning_rate": 7.012087092179725e-06,
      "loss": 0.9822,
      "step": 540
    },
    {
      "epoch": 1.9184397163120568,
      "grad_norm": 0.35817528542327143,
      "learning_rate": 6.972716082089173e-06,
      "loss": 0.9672,
      "step": 541
    },
    {
      "epoch": 1.9219858156028369,
      "grad_norm": 0.33008334862633676,
      "learning_rate": 6.93339666407048e-06,
      "loss": 0.9608,
      "step": 542
    },
    {
      "epoch": 1.925531914893617,
      "grad_norm": 0.3338787358040796,
      "learning_rate": 6.894129508219444e-06,
      "loss": 0.9571,
      "step": 543
    },
    {
      "epoch": 1.9290780141843973,
      "grad_norm": 0.3505646347148005,
      "learning_rate": 6.854915283741198e-06,
      "loss": 0.9578,
      "step": 544
    },
    {
      "epoch": 1.9326241134751774,
      "grad_norm": 0.33261145396470737,
      "learning_rate": 6.81575465893879e-06,
      "loss": 0.9647,
      "step": 545
    },
    {
      "epoch": 1.9361702127659575,
      "grad_norm": 0.3226217535211914,
      "learning_rate": 6.776648301201817e-06,
      "loss": 0.9398,
      "step": 546
    },
    {
      "epoch": 1.9397163120567376,
      "grad_norm": 0.3278245419317454,
      "learning_rate": 6.7375968769950255e-06,
      "loss": 0.9135,
      "step": 547
    },
    {
      "epoch": 1.9432624113475176,
      "grad_norm": 0.3343650792150059,
      "learning_rate": 6.698601051846961e-06,
      "loss": 0.947,
      "step": 548
    },
    {
      "epoch": 1.9468085106382977,
      "grad_norm": 0.32170493330780797,
      "learning_rate": 6.659661490338648e-06,
      "loss": 0.9202,
      "step": 549
    },
    {
      "epoch": 1.950354609929078,
      "grad_norm": 0.32984166937379644,
      "learning_rate": 6.620778856092227e-06,
      "loss": 0.9876,
      "step": 550
    },
    {
      "epoch": 1.9539007092198581,
      "grad_norm": 0.3205825135290648,
      "learning_rate": 6.5819538117596735e-06,
      "loss": 0.9587,
      "step": 551
    },
    {
      "epoch": 1.9574468085106385,
      "grad_norm": 0.3239884548225763,
      "learning_rate": 6.5431870190114925e-06,
      "loss": 0.9502,
      "step": 552
    },
    {
      "epoch": 1.9609929078014185,
      "grad_norm": 0.32314508912258516,
      "learning_rate": 6.504479138525437e-06,
      "loss": 0.9579,
      "step": 553
    },
    {
      "epoch": 1.9645390070921986,
      "grad_norm": 0.3295935728159571,
      "learning_rate": 6.465830829975273e-06,
      "loss": 0.9448,
      "step": 554
    },
    {
      "epoch": 1.9680851063829787,
      "grad_norm": 0.36994353060133206,
      "learning_rate": 6.427242752019499e-06,
      "loss": 0.9572,
      "step": 555
    },
    {
      "epoch": 1.9716312056737588,
      "grad_norm": 0.3574700453216689,
      "learning_rate": 6.388715562290163e-06,
      "loss": 0.9871,
      "step": 556
    },
    {
      "epoch": 1.975177304964539,
      "grad_norm": 0.313649405440105,
      "learning_rate": 6.350249917381619e-06,
      "loss": 0.9548,
      "step": 557
    },
    {
      "epoch": 1.978723404255319,
      "grad_norm": 0.31820131189414436,
      "learning_rate": 6.3118464728393515e-06,
      "loss": 0.9808,
      "step": 558
    },
    {
      "epoch": 1.9822695035460993,
      "grad_norm": 0.3457642013148487,
      "learning_rate": 6.273505883148815e-06,
      "loss": 0.9583,
      "step": 559
    },
    {
      "epoch": 1.9858156028368794,
      "grad_norm": 0.33267823822326276,
      "learning_rate": 6.235228801724254e-06,
      "loss": 0.9509,
      "step": 560
    },
    {
      "epoch": 1.9893617021276597,
      "grad_norm": 0.3497300533666417,
      "learning_rate": 6.197015880897599e-06,
      "loss": 0.9602,
      "step": 561
    },
    {
      "epoch": 1.9929078014184398,
      "grad_norm": 0.34360079386079867,
      "learning_rate": 6.158867771907319e-06,
      "loss": 0.9472,
      "step": 562
    },
    {
      "epoch": 1.99645390070922,
      "grad_norm": 0.32442906842984515,
      "learning_rate": 6.120785124887339e-06,
      "loss": 0.9132,
      "step": 563
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.34622104170979245,
      "learning_rate": 6.082768588855963e-06,
      "loss": 0.9596,
      "step": 564
    },
    {
      "epoch": 2.00354609929078,
      "grad_norm": 0.4566911346970862,
      "learning_rate": 6.0448188117048e-06,
      "loss": 0.8959,
      "step": 565
    },
    {
      "epoch": 2.00709219858156,
      "grad_norm": 0.3617148784050208,
      "learning_rate": 6.006936440187735e-06,
      "loss": 0.9563,
      "step": 566
    },
    {
      "epoch": 2.0106382978723403,
      "grad_norm": 0.3681452353035544,
      "learning_rate": 5.9691221199099e-06,
      "loss": 0.9183,
      "step": 567
    },
    {
      "epoch": 2.0141843971631204,
      "grad_norm": 0.3386502623032431,
      "learning_rate": 5.93137649531667e-06,
      "loss": 0.9249,
      "step": 568
    },
    {
      "epoch": 2.017730496453901,
      "grad_norm": 0.34223363446968597,
      "learning_rate": 5.893700209682691e-06,
      "loss": 0.9212,
      "step": 569
    },
    {
      "epoch": 2.021276595744681,
      "grad_norm": 0.3918409012245659,
      "learning_rate": 5.856093905100899e-06,
      "loss": 0.8968,
      "step": 570
    },
    {
      "epoch": 2.024822695035461,
      "grad_norm": 0.3365985447006349,
      "learning_rate": 5.818558222471603e-06,
      "loss": 0.9246,
      "step": 571
    },
    {
      "epoch": 2.028368794326241,
      "grad_norm": 0.3433680975231668,
      "learning_rate": 5.7810938014915284e-06,
      "loss": 0.9053,
      "step": 572
    },
    {
      "epoch": 2.0319148936170213,
      "grad_norm": 0.3607462051926254,
      "learning_rate": 5.74370128064294e-06,
      "loss": 0.9063,
      "step": 573
    },
    {
      "epoch": 2.0354609929078014,
      "grad_norm": 0.3394654339420203,
      "learning_rate": 5.7063812971827655e-06,
      "loss": 0.9299,
      "step": 574
    },
    {
      "epoch": 2.0390070921985815,
      "grad_norm": 0.32380633576578366,
      "learning_rate": 5.669134487131707e-06,
      "loss": 0.9036,
      "step": 575
    },
    {
      "epoch": 2.0425531914893615,
      "grad_norm": 0.3799496700402083,
      "learning_rate": 5.631961485263436e-06,
      "loss": 0.9188,
      "step": 576
    },
    {
      "epoch": 2.0460992907801416,
      "grad_norm": 0.36059710867622713,
      "learning_rate": 5.594862925093745e-06,
      "loss": 0.9567,
      "step": 577
    },
    {
      "epoch": 2.049645390070922,
      "grad_norm": 0.31267668524991454,
      "learning_rate": 5.557839438869764e-06,
      "loss": 0.8937,
      "step": 578
    },
    {
      "epoch": 2.0531914893617023,
      "grad_norm": 0.34523656875027586,
      "learning_rate": 5.520891657559197e-06,
      "loss": 0.9055,
      "step": 579
    },
    {
      "epoch": 2.0567375886524824,
      "grad_norm": 0.3428830226809694,
      "learning_rate": 5.484020210839547e-06,
      "loss": 0.9121,
      "step": 580
    },
    {
      "epoch": 2.0602836879432624,
      "grad_norm": 0.34407502937578527,
      "learning_rate": 5.447225727087399e-06,
      "loss": 0.8889,
      "step": 581
    },
    {
      "epoch": 2.0638297872340425,
      "grad_norm": 0.3639918638016257,
      "learning_rate": 5.410508833367709e-06,
      "loss": 0.9199,
      "step": 582
    },
    {
      "epoch": 2.0673758865248226,
      "grad_norm": 0.33877778305397416,
      "learning_rate": 5.373870155423106e-06,
      "loss": 0.9459,
      "step": 583
    },
    {
      "epoch": 2.0709219858156027,
      "grad_norm": 0.31092698823821685,
      "learning_rate": 5.337310317663255e-06,
      "loss": 0.9379,
      "step": 584
    },
    {
      "epoch": 2.074468085106383,
      "grad_norm": 0.44030677058791035,
      "learning_rate": 5.300829943154181e-06,
      "loss": 0.8792,
      "step": 585
    },
    {
      "epoch": 2.078014184397163,
      "grad_norm": 0.32870413268141346,
      "learning_rate": 5.264429653607675e-06,
      "loss": 0.9146,
      "step": 586
    },
    {
      "epoch": 2.0815602836879434,
      "grad_norm": 0.406809678755843,
      "learning_rate": 5.228110069370689e-06,
      "loss": 0.9238,
      "step": 587
    },
    {
      "epoch": 2.0851063829787235,
      "grad_norm": 0.3298950250685358,
      "learning_rate": 5.191871809414763e-06,
      "loss": 0.9119,
      "step": 588
    },
    {
      "epoch": 2.0886524822695036,
      "grad_norm": 0.326775177491553,
      "learning_rate": 5.155715491325493e-06,
      "loss": 0.9088,
      "step": 589
    },
    {
      "epoch": 2.0921985815602837,
      "grad_norm": 0.33745417440187936,
      "learning_rate": 5.119641731291971e-06,
      "loss": 0.9006,
      "step": 590
    },
    {
      "epoch": 2.095744680851064,
      "grad_norm": 0.33773499121718675,
      "learning_rate": 5.083651144096323e-06,
      "loss": 0.9447,
      "step": 591
    },
    {
      "epoch": 2.099290780141844,
      "grad_norm": 0.346227680847854,
      "learning_rate": 5.047744343103203e-06,
      "loss": 0.8952,
      "step": 592
    },
    {
      "epoch": 2.102836879432624,
      "grad_norm": 0.3416473020704537,
      "learning_rate": 5.011921940249349e-06,
      "loss": 0.9094,
      "step": 593
    },
    {
      "epoch": 2.106382978723404,
      "grad_norm": 0.3345099331995336,
      "learning_rate": 4.976184546033167e-06,
      "loss": 0.9292,
      "step": 594
    },
    {
      "epoch": 2.1099290780141846,
      "grad_norm": 0.3373034338886128,
      "learning_rate": 4.940532769504297e-06,
      "loss": 0.9096,
      "step": 595
    },
    {
      "epoch": 2.1134751773049647,
      "grad_norm": 0.3429936434310372,
      "learning_rate": 4.904967218253271e-06,
      "loss": 0.9183,
      "step": 596
    },
    {
      "epoch": 2.117021276595745,
      "grad_norm": 0.3173720112615683,
      "learning_rate": 4.869488498401126e-06,
      "loss": 0.8964,
      "step": 597
    },
    {
      "epoch": 2.120567375886525,
      "grad_norm": 0.35213901124678976,
      "learning_rate": 4.834097214589082e-06,
      "loss": 0.9403,
      "step": 598
    },
    {
      "epoch": 2.124113475177305,
      "grad_norm": 0.3493639815881956,
      "learning_rate": 4.7987939699682605e-06,
      "loss": 0.9279,
      "step": 599
    },
    {
      "epoch": 2.127659574468085,
      "grad_norm": 0.34103344147668196,
      "learning_rate": 4.763579366189367e-06,
      "loss": 0.9153,
      "step": 600
    },
    {
      "epoch": 2.131205673758865,
      "grad_norm": 0.33549617227086487,
      "learning_rate": 4.728454003392475e-06,
      "loss": 0.9226,
      "step": 601
    },
    {
      "epoch": 2.1347517730496453,
      "grad_norm": 0.32714622326432746,
      "learning_rate": 4.69341848019677e-06,
      "loss": 0.9311,
      "step": 602
    },
    {
      "epoch": 2.1382978723404253,
      "grad_norm": 0.3311453632580336,
      "learning_rate": 4.658473393690359e-06,
      "loss": 0.9084,
      "step": 603
    },
    {
      "epoch": 2.141843971631206,
      "grad_norm": 0.38675624940083536,
      "learning_rate": 4.623619339420098e-06,
      "loss": 0.8874,
      "step": 604
    },
    {
      "epoch": 2.145390070921986,
      "grad_norm": 0.3279868665844946,
      "learning_rate": 4.588856911381433e-06,
      "loss": 0.9217,
      "step": 605
    },
    {
      "epoch": 2.148936170212766,
      "grad_norm": 0.3217769678798751,
      "learning_rate": 4.554186702008292e-06,
      "loss": 0.931,
      "step": 606
    },
    {
      "epoch": 2.152482269503546,
      "grad_norm": 0.4389700818952468,
      "learning_rate": 4.519609302162972e-06,
      "loss": 0.9032,
      "step": 607
    },
    {
      "epoch": 2.1560283687943262,
      "grad_norm": 0.3658714505599075,
      "learning_rate": 4.485125301126072e-06,
      "loss": 0.9136,
      "step": 608
    },
    {
      "epoch": 2.1595744680851063,
      "grad_norm": 0.35358017573741335,
      "learning_rate": 4.4507352865864686e-06,
      "loss": 0.9179,
      "step": 609
    },
    {
      "epoch": 2.1631205673758864,
      "grad_norm": 0.32146202938676643,
      "learning_rate": 4.416439844631271e-06,
      "loss": 0.9296,
      "step": 610
    },
    {
      "epoch": 2.1666666666666665,
      "grad_norm": 0.31895440604586967,
      "learning_rate": 4.38223955973586e-06,
      "loss": 0.901,
      "step": 611
    },
    {
      "epoch": 2.1702127659574466,
      "grad_norm": 0.6707694969413162,
      "learning_rate": 4.348135014753906e-06,
      "loss": 0.8751,
      "step": 612
    },
    {
      "epoch": 2.173758865248227,
      "grad_norm": 0.3545708295368673,
      "learning_rate": 4.314126790907455e-06,
      "loss": 0.9363,
      "step": 613
    },
    {
      "epoch": 2.1773049645390072,
      "grad_norm": 0.32363120553664343,
      "learning_rate": 4.280215467777005e-06,
      "loss": 0.9259,
      "step": 614
    },
    {
      "epoch": 2.1808510638297873,
      "grad_norm": 0.34309179301070913,
      "learning_rate": 4.246401623291634e-06,
      "loss": 0.8897,
      "step": 615
    },
    {
      "epoch": 2.1843971631205674,
      "grad_norm": 0.3453687929831591,
      "learning_rate": 4.2126858337191675e-06,
      "loss": 0.9135,
      "step": 616
    },
    {
      "epoch": 2.1879432624113475,
      "grad_norm": 0.43729618573609375,
      "learning_rate": 4.179068673656332e-06,
      "loss": 0.9041,
      "step": 617
    },
    {
      "epoch": 2.1914893617021276,
      "grad_norm": 0.3202082582840827,
      "learning_rate": 4.145550716018974e-06,
      "loss": 0.9135,
      "step": 618
    },
    {
      "epoch": 2.1950354609929077,
      "grad_norm": 0.3246145160502331,
      "learning_rate": 4.112132532032302e-06,
      "loss": 0.8856,
      "step": 619
    },
    {
      "epoch": 2.198581560283688,
      "grad_norm": 0.3440105242186509,
      "learning_rate": 4.078814691221139e-06,
      "loss": 0.958,
      "step": 620
    },
    {
      "epoch": 2.202127659574468,
      "grad_norm": 0.32716180743353473,
      "learning_rate": 4.045597761400234e-06,
      "loss": 0.9002,
      "step": 621
    },
    {
      "epoch": 2.2056737588652484,
      "grad_norm": 0.32188979652918004,
      "learning_rate": 4.01248230866456e-06,
      "loss": 0.9512,
      "step": 622
    },
    {
      "epoch": 2.2092198581560285,
      "grad_norm": 0.32072445524392357,
      "learning_rate": 3.979468897379697e-06,
      "loss": 0.9526,
      "step": 623
    },
    {
      "epoch": 2.2127659574468086,
      "grad_norm": 0.3387410378590688,
      "learning_rate": 3.946558090172182e-06,
      "loss": 0.9277,
      "step": 624
    },
    {
      "epoch": 2.2163120567375887,
      "grad_norm": 0.3196709727101075,
      "learning_rate": 3.9137504479199405e-06,
      "loss": 0.8677,
      "step": 625
    },
    {
      "epoch": 2.219858156028369,
      "grad_norm": 0.3255137199624219,
      "learning_rate": 3.881046529742734e-06,
      "loss": 0.9138,
      "step": 626
    },
    {
      "epoch": 2.223404255319149,
      "grad_norm": 0.30874496281243,
      "learning_rate": 3.848446892992604e-06,
      "loss": 0.9102,
      "step": 627
    },
    {
      "epoch": 2.226950354609929,
      "grad_norm": 0.5051339031173951,
      "learning_rate": 3.815952093244405e-06,
      "loss": 0.9271,
      "step": 628
    },
    {
      "epoch": 2.230496453900709,
      "grad_norm": 0.3176852007190203,
      "learning_rate": 3.7835626842863105e-06,
      "loss": 0.9647,
      "step": 629
    },
    {
      "epoch": 2.2340425531914896,
      "grad_norm": 0.3220485429110355,
      "learning_rate": 3.7512792181103872e-06,
      "loss": 0.9188,
      "step": 630
    },
    {
      "epoch": 2.2375886524822697,
      "grad_norm": 0.3441318897854401,
      "learning_rate": 3.7191022449031965e-06,
      "loss": 0.9265,
      "step": 631
    },
    {
      "epoch": 2.2411347517730498,
      "grad_norm": 0.3141082795259724,
      "learning_rate": 3.6870323130363972e-06,
      "loss": 0.9366,
      "step": 632
    },
    {
      "epoch": 2.24468085106383,
      "grad_norm": 0.36501665747304923,
      "learning_rate": 3.6550699690574133e-06,
      "loss": 0.8975,
      "step": 633
    },
    {
      "epoch": 2.24822695035461,
      "grad_norm": 0.3067756636590056,
      "learning_rate": 3.623215757680113e-06,
      "loss": 0.9314,
      "step": 634
    },
    {
      "epoch": 2.25177304964539,
      "grad_norm": 0.325736875227407,
      "learning_rate": 3.591470221775544e-06,
      "loss": 0.8838,
      "step": 635
    },
    {
      "epoch": 2.25531914893617,
      "grad_norm": 0.3363855691116925,
      "learning_rate": 3.5598339023626514e-06,
      "loss": 0.9203,
      "step": 636
    },
    {
      "epoch": 2.2588652482269502,
      "grad_norm": 0.3201591221800834,
      "learning_rate": 3.5283073385990762e-06,
      "loss": 0.8914,
      "step": 637
    },
    {
      "epoch": 2.2624113475177303,
      "grad_norm": 0.3057573202166175,
      "learning_rate": 3.4968910677719746e-06,
      "loss": 0.8962,
      "step": 638
    },
    {
      "epoch": 2.2659574468085104,
      "grad_norm": 0.31394171796231907,
      "learning_rate": 3.4655856252888376e-06,
      "loss": 0.8974,
      "step": 639
    },
    {
      "epoch": 2.269503546099291,
      "grad_norm": 0.3251813869113006,
      "learning_rate": 3.434391544668383e-06,
      "loss": 0.9322,
      "step": 640
    },
    {
      "epoch": 2.273049645390071,
      "grad_norm": 0.3315408177538822,
      "learning_rate": 3.4033093575314645e-06,
      "loss": 0.9369,
      "step": 641
    },
    {
      "epoch": 2.276595744680851,
      "grad_norm": 0.3241398323929304,
      "learning_rate": 3.372339593591998e-06,
      "loss": 0.8951,
      "step": 642
    },
    {
      "epoch": 2.280141843971631,
      "grad_norm": 0.3222392701358019,
      "learning_rate": 3.341482780647952e-06,
      "loss": 0.9324,
      "step": 643
    },
    {
      "epoch": 2.2836879432624113,
      "grad_norm": 0.3078014669602889,
      "learning_rate": 3.310739444572336e-06,
      "loss": 0.9195,
      "step": 644
    },
    {
      "epoch": 2.2872340425531914,
      "grad_norm": 0.30721991786253994,
      "learning_rate": 3.2801101093042444e-06,
      "loss": 0.9166,
      "step": 645
    },
    {
      "epoch": 2.2907801418439715,
      "grad_norm": 0.32609727890595375,
      "learning_rate": 3.249595296839937e-06,
      "loss": 0.9395,
      "step": 646
    },
    {
      "epoch": 2.2943262411347516,
      "grad_norm": 0.31254234076499143,
      "learning_rate": 3.219195527223923e-06,
      "loss": 0.9041,
      "step": 647
    },
    {
      "epoch": 2.297872340425532,
      "grad_norm": 0.3176629795843061,
      "learning_rate": 3.1889113185401223e-06,
      "loss": 0.9306,
      "step": 648
    },
    {
      "epoch": 2.301418439716312,
      "grad_norm": 0.30783904152042746,
      "learning_rate": 3.1587431869030126e-06,
      "loss": 0.9353,
      "step": 649
    },
    {
      "epoch": 2.3049645390070923,
      "grad_norm": 0.30800111746280723,
      "learning_rate": 3.1286916464488506e-06,
      "loss": 0.8775,
      "step": 650
    },
    {
      "epoch": 2.3085106382978724,
      "grad_norm": 0.3137322351937391,
      "learning_rate": 3.098757209326898e-06,
      "loss": 0.9063,
      "step": 651
    },
    {
      "epoch": 2.3120567375886525,
      "grad_norm": 0.31125500054596683,
      "learning_rate": 3.068940385690704e-06,
      "loss": 0.8757,
      "step": 652
    },
    {
      "epoch": 2.3156028368794326,
      "grad_norm": 0.3120079580710984,
      "learning_rate": 3.0392416836894066e-06,
      "loss": 0.855,
      "step": 653
    },
    {
      "epoch": 2.3191489361702127,
      "grad_norm": 0.31258285093042815,
      "learning_rate": 3.0096616094590712e-06,
      "loss": 0.8695,
      "step": 654
    },
    {
      "epoch": 2.3226950354609928,
      "grad_norm": 0.31805496506154607,
      "learning_rate": 2.9802006671140583e-06,
      "loss": 0.9319,
      "step": 655
    },
    {
      "epoch": 2.326241134751773,
      "grad_norm": 0.3270645593794086,
      "learning_rate": 2.950859358738453e-06,
      "loss": 0.9082,
      "step": 656
    },
    {
      "epoch": 2.329787234042553,
      "grad_norm": 0.3121962070049027,
      "learning_rate": 2.9216381843774812e-06,
      "loss": 0.9174,
      "step": 657
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.32503615811718056,
      "learning_rate": 2.8925376420290153e-06,
      "loss": 0.9192,
      "step": 658
    },
    {
      "epoch": 2.3368794326241136,
      "grad_norm": 0.3127522704871516,
      "learning_rate": 2.8635582276350606e-06,
      "loss": 0.9003,
      "step": 659
    },
    {
      "epoch": 2.3404255319148937,
      "grad_norm": 0.3159719914352033,
      "learning_rate": 2.834700435073319e-06,
      "loss": 0.9108,
      "step": 660
    },
    {
      "epoch": 2.3439716312056738,
      "grad_norm": 0.30569977271102494,
      "learning_rate": 2.8059647561487758e-06,
      "loss": 0.9276,
      "step": 661
    },
    {
      "epoch": 2.347517730496454,
      "grad_norm": 0.33735915634007474,
      "learning_rate": 2.777351680585302e-06,
      "loss": 0.9223,
      "step": 662
    },
    {
      "epoch": 2.351063829787234,
      "grad_norm": 0.3168581351513594,
      "learning_rate": 2.7488616960173265e-06,
      "loss": 0.9474,
      "step": 663
    },
    {
      "epoch": 2.354609929078014,
      "grad_norm": 0.3260808135927312,
      "learning_rate": 2.7204952879815096e-06,
      "loss": 0.896,
      "step": 664
    },
    {
      "epoch": 2.3581560283687946,
      "grad_norm": 0.3277475079557903,
      "learning_rate": 2.6922529399084773e-06,
      "loss": 0.909,
      "step": 665
    },
    {
      "epoch": 2.3617021276595747,
      "grad_norm": 0.30002466874093064,
      "learning_rate": 2.6641351331145827e-06,
      "loss": 0.9368,
      "step": 666
    },
    {
      "epoch": 2.3652482269503547,
      "grad_norm": 0.31282975390328294,
      "learning_rate": 2.6361423467936954e-06,
      "loss": 0.9352,
      "step": 667
    },
    {
      "epoch": 2.368794326241135,
      "grad_norm": 0.3141556521218453,
      "learning_rate": 2.608275058009053e-06,
      "loss": 0.9102,
      "step": 668
    },
    {
      "epoch": 2.372340425531915,
      "grad_norm": 0.29970361864896106,
      "learning_rate": 2.580533741685106e-06,
      "loss": 0.9079,
      "step": 669
    },
    {
      "epoch": 2.375886524822695,
      "grad_norm": 0.32369551859579393,
      "learning_rate": 2.55291887059944e-06,
      "loss": 0.9367,
      "step": 670
    },
    {
      "epoch": 2.379432624113475,
      "grad_norm": 0.3201091360816704,
      "learning_rate": 2.52543091537472e-06,
      "loss": 0.9146,
      "step": 671
    },
    {
      "epoch": 2.382978723404255,
      "grad_norm": 0.3014262832146507,
      "learning_rate": 2.4980703444706567e-06,
      "loss": 0.8984,
      "step": 672
    },
    {
      "epoch": 2.3865248226950353,
      "grad_norm": 0.30972681757749293,
      "learning_rate": 2.4708376241760412e-06,
      "loss": 0.9084,
      "step": 673
    },
    {
      "epoch": 2.3900709219858154,
      "grad_norm": 0.3086157439660352,
      "learning_rate": 2.4437332186007768e-06,
      "loss": 0.9541,
      "step": 674
    },
    {
      "epoch": 2.393617021276596,
      "grad_norm": 0.3278071415194472,
      "learning_rate": 2.416757589667985e-06,
      "loss": 0.9274,
      "step": 675
    },
    {
      "epoch": 2.397163120567376,
      "grad_norm": 0.3064069385087261,
      "learning_rate": 2.3899111971061362e-06,
      "loss": 0.9118,
      "step": 676
    },
    {
      "epoch": 2.400709219858156,
      "grad_norm": 0.31634968287824605,
      "learning_rate": 2.3631944984411947e-06,
      "loss": 0.9175,
      "step": 677
    },
    {
      "epoch": 2.404255319148936,
      "grad_norm": 0.31946966423089546,
      "learning_rate": 2.3366079489888473e-06,
      "loss": 0.9104,
      "step": 678
    },
    {
      "epoch": 2.4078014184397163,
      "grad_norm": 0.3073126775907417,
      "learning_rate": 2.3101520018467214e-06,
      "loss": 0.875,
      "step": 679
    },
    {
      "epoch": 2.4113475177304964,
      "grad_norm": 0.36537378058022496,
      "learning_rate": 2.2838271078866714e-06,
      "loss": 0.9268,
      "step": 680
    },
    {
      "epoch": 2.4148936170212765,
      "grad_norm": 0.3106854060545735,
      "learning_rate": 2.2576337157471083e-06,
      "loss": 0.9233,
      "step": 681
    },
    {
      "epoch": 2.4184397163120566,
      "grad_norm": 0.3055447112943066,
      "learning_rate": 2.2315722718253276e-06,
      "loss": 0.9249,
      "step": 682
    },
    {
      "epoch": 2.421985815602837,
      "grad_norm": 1.3811799608320727,
      "learning_rate": 2.205643220269921e-06,
      "loss": 0.9085,
      "step": 683
    },
    {
      "epoch": 2.425531914893617,
      "grad_norm": 0.3002261018646463,
      "learning_rate": 2.179847002973201e-06,
      "loss": 0.9091,
      "step": 684
    },
    {
      "epoch": 2.4290780141843973,
      "grad_norm": 0.30674947932499097,
      "learning_rate": 2.1541840595636678e-06,
      "loss": 0.9179,
      "step": 685
    },
    {
      "epoch": 2.4326241134751774,
      "grad_norm": 0.3278506911887618,
      "learning_rate": 2.1286548273985288e-06,
      "loss": 0.8875,
      "step": 686
    },
    {
      "epoch": 2.4361702127659575,
      "grad_norm": 0.3334426883441487,
      "learning_rate": 2.1032597415562218e-06,
      "loss": 0.9133,
      "step": 687
    },
    {
      "epoch": 2.4397163120567376,
      "grad_norm": 0.3088493635408435,
      "learning_rate": 2.077999234829029e-06,
      "loss": 0.9126,
      "step": 688
    },
    {
      "epoch": 2.4432624113475176,
      "grad_norm": 0.30877224814331133,
      "learning_rate": 2.052873737715675e-06,
      "loss": 0.9509,
      "step": 689
    },
    {
      "epoch": 2.4468085106382977,
      "grad_norm": 0.3178316512514858,
      "learning_rate": 2.0278836784140043e-06,
      "loss": 0.9185,
      "step": 690
    },
    {
      "epoch": 2.450354609929078,
      "grad_norm": 0.3187218318871324,
      "learning_rate": 2.0030294828136867e-06,
      "loss": 0.9007,
      "step": 691
    },
    {
      "epoch": 2.453900709219858,
      "grad_norm": 0.3894554824207546,
      "learning_rate": 1.9783115744889447e-06,
      "loss": 0.9312,
      "step": 692
    },
    {
      "epoch": 2.4574468085106385,
      "grad_norm": 0.29941537548459557,
      "learning_rate": 1.953730374691353e-06,
      "loss": 0.9449,
      "step": 693
    },
    {
      "epoch": 2.4609929078014185,
      "grad_norm": 0.32251616671371774,
      "learning_rate": 1.9292863023426413e-06,
      "loss": 0.8896,
      "step": 694
    },
    {
      "epoch": 2.4645390070921986,
      "grad_norm": 0.3257419849968331,
      "learning_rate": 1.9049797740275655e-06,
      "loss": 0.9188,
      "step": 695
    },
    {
      "epoch": 2.4680851063829787,
      "grad_norm": 0.2981120109547122,
      "learning_rate": 1.8808112039868143e-06,
      "loss": 0.8842,
      "step": 696
    },
    {
      "epoch": 2.471631205673759,
      "grad_norm": 0.3543784363301713,
      "learning_rate": 1.856781004109931e-06,
      "loss": 0.9478,
      "step": 697
    },
    {
      "epoch": 2.475177304964539,
      "grad_norm": 0.293436787335392,
      "learning_rate": 1.832889583928308e-06,
      "loss": 0.9176,
      "step": 698
    },
    {
      "epoch": 2.478723404255319,
      "grad_norm": 0.30922135296942016,
      "learning_rate": 1.8091373506082033e-06,
      "loss": 0.9176,
      "step": 699
    },
    {
      "epoch": 2.482269503546099,
      "grad_norm": 0.3244973967157742,
      "learning_rate": 1.7855247089438022e-06,
      "loss": 0.9226,
      "step": 700
    },
    {
      "epoch": 2.4858156028368796,
      "grad_norm": 0.30699412163526396,
      "learning_rate": 1.7620520613503223e-06,
      "loss": 0.9028,
      "step": 701
    },
    {
      "epoch": 2.4893617021276597,
      "grad_norm": 0.31940538100442684,
      "learning_rate": 1.7387198078571433e-06,
      "loss": 0.9242,
      "step": 702
    },
    {
      "epoch": 2.49290780141844,
      "grad_norm": 0.3040599236363412,
      "learning_rate": 1.7155283461010064e-06,
      "loss": 0.9349,
      "step": 703
    },
    {
      "epoch": 2.49645390070922,
      "grad_norm": 0.3008914360277951,
      "learning_rate": 1.6924780713192235e-06,
      "loss": 0.9238,
      "step": 704
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.3047659130967524,
      "learning_rate": 1.6695693763429433e-06,
      "loss": 0.9035,
      "step": 705
    },
    {
      "epoch": 2.50354609929078,
      "grad_norm": 0.33760543099149587,
      "learning_rate": 1.6468026515904712e-06,
      "loss": 0.9369,
      "step": 706
    },
    {
      "epoch": 2.50709219858156,
      "grad_norm": 0.3019407982427647,
      "learning_rate": 1.6241782850605925e-06,
      "loss": 0.8842,
      "step": 707
    },
    {
      "epoch": 2.5106382978723403,
      "grad_norm": 0.31360360735621046,
      "learning_rate": 1.6016966623259821e-06,
      "loss": 0.9079,
      "step": 708
    },
    {
      "epoch": 2.5141843971631204,
      "grad_norm": 0.31362601497035747,
      "learning_rate": 1.5793581665266145e-06,
      "loss": 0.8965,
      "step": 709
    },
    {
      "epoch": 2.5177304964539005,
      "grad_norm": 0.29493198915188423,
      "learning_rate": 1.5571631783632513e-06,
      "loss": 0.8966,
      "step": 710
    },
    {
      "epoch": 2.521276595744681,
      "grad_norm": 0.3020594965557035,
      "learning_rate": 1.5351120760909388e-06,
      "loss": 0.8975,
      "step": 711
    },
    {
      "epoch": 2.524822695035461,
      "grad_norm": 0.31339820524414264,
      "learning_rate": 1.5132052355125693e-06,
      "loss": 0.9028,
      "step": 712
    },
    {
      "epoch": 2.528368794326241,
      "grad_norm": 0.2972691263629277,
      "learning_rate": 1.4914430299724747e-06,
      "loss": 0.9754,
      "step": 713
    },
    {
      "epoch": 2.5319148936170213,
      "grad_norm": 0.29873853790696653,
      "learning_rate": 1.4698258303500624e-06,
      "loss": 0.9313,
      "step": 714
    },
    {
      "epoch": 2.5354609929078014,
      "grad_norm": 0.3010289927697155,
      "learning_rate": 1.4483540050535038e-06,
      "loss": 0.9079,
      "step": 715
    },
    {
      "epoch": 2.5390070921985815,
      "grad_norm": 0.30300125772170716,
      "learning_rate": 1.4270279200134406e-06,
      "loss": 0.8838,
      "step": 716
    },
    {
      "epoch": 2.5425531914893615,
      "grad_norm": 0.32295056893197593,
      "learning_rate": 1.4058479386767554e-06,
      "loss": 0.9027,
      "step": 717
    },
    {
      "epoch": 2.546099290780142,
      "grad_norm": 0.31767282991785056,
      "learning_rate": 1.3848144220003867e-06,
      "loss": 0.9429,
      "step": 718
    },
    {
      "epoch": 2.549645390070922,
      "grad_norm": 0.3112289718032724,
      "learning_rate": 1.363927728445159e-06,
      "loss": 0.8892,
      "step": 719
    },
    {
      "epoch": 2.5531914893617023,
      "grad_norm": 0.2995342176385735,
      "learning_rate": 1.3431882139696916e-06,
      "loss": 0.8905,
      "step": 720
    },
    {
      "epoch": 2.5567375886524824,
      "grad_norm": 0.3032745086759523,
      "learning_rate": 1.3225962320243225e-06,
      "loss": 0.886,
      "step": 721
    },
    {
      "epoch": 2.5602836879432624,
      "grad_norm": 0.30710485752267236,
      "learning_rate": 1.3021521335450815e-06,
      "loss": 0.8881,
      "step": 722
    },
    {
      "epoch": 2.5638297872340425,
      "grad_norm": 0.2999497506890768,
      "learning_rate": 1.281856266947723e-06,
      "loss": 0.9114,
      "step": 723
    },
    {
      "epoch": 2.5673758865248226,
      "grad_norm": 0.30349869158909365,
      "learning_rate": 1.261708978121773e-06,
      "loss": 0.9228,
      "step": 724
    },
    {
      "epoch": 2.5709219858156027,
      "grad_norm": 0.30950924252855117,
      "learning_rate": 1.241710610424647e-06,
      "loss": 0.9028,
      "step": 725
    },
    {
      "epoch": 2.574468085106383,
      "grad_norm": 0.30623095850995313,
      "learning_rate": 1.2218615046757877e-06,
      "loss": 0.9086,
      "step": 726
    },
    {
      "epoch": 2.578014184397163,
      "grad_norm": 0.29773890784186613,
      "learning_rate": 1.2021619991508614e-06,
      "loss": 0.8969,
      "step": 727
    },
    {
      "epoch": 2.581560283687943,
      "grad_norm": 0.2966683063241434,
      "learning_rate": 1.182612429575999e-06,
      "loss": 0.913,
      "step": 728
    },
    {
      "epoch": 2.5851063829787235,
      "grad_norm": 0.4233006249369692,
      "learning_rate": 1.1632131291220628e-06,
      "loss": 0.9152,
      "step": 729
    },
    {
      "epoch": 2.5886524822695036,
      "grad_norm": 0.3061580358258722,
      "learning_rate": 1.1439644283989749e-06,
      "loss": 0.925,
      "step": 730
    },
    {
      "epoch": 2.5921985815602837,
      "grad_norm": 0.3034994086209508,
      "learning_rate": 1.1248666554500831e-06,
      "loss": 0.9038,
      "step": 731
    },
    {
      "epoch": 2.595744680851064,
      "grad_norm": 0.3022611564679423,
      "learning_rate": 1.1059201357465687e-06,
      "loss": 0.9354,
      "step": 732
    },
    {
      "epoch": 2.599290780141844,
      "grad_norm": 0.29847498771161884,
      "learning_rate": 1.0871251921819027e-06,
      "loss": 0.9059,
      "step": 733
    },
    {
      "epoch": 2.602836879432624,
      "grad_norm": 0.2819425119692849,
      "learning_rate": 1.0684821450663364e-06,
      "loss": 0.9686,
      "step": 734
    },
    {
      "epoch": 2.6063829787234045,
      "grad_norm": 0.28132632342568453,
      "learning_rate": 1.0499913121214511e-06,
      "loss": 0.9344,
      "step": 735
    },
    {
      "epoch": 2.6099290780141846,
      "grad_norm": 0.3084807404737308,
      "learning_rate": 1.0316530084747344e-06,
      "loss": 0.933,
      "step": 736
    },
    {
      "epoch": 2.6134751773049647,
      "grad_norm": 0.31243627686333275,
      "learning_rate": 1.0134675466542166e-06,
      "loss": 0.9304,
      "step": 737
    },
    {
      "epoch": 2.617021276595745,
      "grad_norm": 0.2916454484641493,
      "learning_rate": 9.954352365831455e-07,
      "loss": 0.9209,
      "step": 738
    },
    {
      "epoch": 2.620567375886525,
      "grad_norm": 0.31575534533705746,
      "learning_rate": 9.775563855746918e-07,
      "loss": 0.9391,
      "step": 739
    },
    {
      "epoch": 2.624113475177305,
      "grad_norm": 0.307647066815103,
      "learning_rate": 9.59831298326731e-07,
      "loss": 0.8955,
      "step": 740
    },
    {
      "epoch": 2.627659574468085,
      "grad_norm": 0.30851549554520924,
      "learning_rate": 9.422602769166367e-07,
      "loss": 0.9238,
      "step": 741
    },
    {
      "epoch": 2.631205673758865,
      "grad_norm": 0.30972798605613994,
      "learning_rate": 9.248436207961331e-07,
      "loss": 0.8963,
      "step": 742
    },
    {
      "epoch": 2.6347517730496453,
      "grad_norm": 0.3078284112186838,
      "learning_rate": 9.075816267862036e-07,
      "loss": 0.8868,
      "step": 743
    },
    {
      "epoch": 2.6382978723404253,
      "grad_norm": 0.2988265475360356,
      "learning_rate": 8.904745890720146e-07,
      "loss": 0.9384,
      "step": 744
    },
    {
      "epoch": 2.6418439716312054,
      "grad_norm": 0.298948231596996,
      "learning_rate": 8.735227991979178e-07,
      "loss": 0.8951,
      "step": 745
    },
    {
      "epoch": 2.645390070921986,
      "grad_norm": 0.32214256840915684,
      "learning_rate": 8.567265460624697e-07,
      "loss": 0.9464,
      "step": 746
    },
    {
      "epoch": 2.648936170212766,
      "grad_norm": 0.2929574896449302,
      "learning_rate": 8.400861159135143e-07,
      "loss": 0.907,
      "step": 747
    },
    {
      "epoch": 2.652482269503546,
      "grad_norm": 0.31556050296308524,
      "learning_rate": 8.236017923433104e-07,
      "loss": 0.9504,
      "step": 748
    },
    {
      "epoch": 2.6560283687943262,
      "grad_norm": 0.2905809683490354,
      "learning_rate": 8.072738562836801e-07,
      "loss": 0.8962,
      "step": 749
    },
    {
      "epoch": 2.6595744680851063,
      "grad_norm": 0.29418627161215105,
      "learning_rate": 7.911025860012445e-07,
      "loss": 0.9155,
      "step": 750
    },
    {
      "epoch": 2.6631205673758864,
      "grad_norm": 0.3007094116002144,
      "learning_rate": 7.750882570926621e-07,
      "loss": 0.8947,
      "step": 751
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.29163111870403013,
      "learning_rate": 7.592311424799404e-07,
      "loss": 0.886,
      "step": 752
    },
    {
      "epoch": 2.670212765957447,
      "grad_norm": 0.2849545087615603,
      "learning_rate": 7.435315124057874e-07,
      "loss": 0.9413,
      "step": 753
    },
    {
      "epoch": 2.673758865248227,
      "grad_norm": 0.2957391801845664,
      "learning_rate": 7.279896344289971e-07,
      "loss": 0.8786,
      "step": 754
    },
    {
      "epoch": 2.6773049645390072,
      "grad_norm": 0.3001167850961704,
      "learning_rate": 7.12605773419901e-07,
      "loss": 0.9233,
      "step": 755
    },
    {
      "epoch": 2.6808510638297873,
      "grad_norm": 0.31548360829669203,
      "learning_rate": 6.973801915558432e-07,
      "loss": 0.886,
      "step": 756
    },
    {
      "epoch": 2.6843971631205674,
      "grad_norm": 0.2966682832596965,
      "learning_rate": 6.823131483167167e-07,
      "loss": 0.9072,
      "step": 757
    },
    {
      "epoch": 2.6879432624113475,
      "grad_norm": 0.3047122286238573,
      "learning_rate": 6.674049004805483e-07,
      "loss": 0.8865,
      "step": 758
    },
    {
      "epoch": 2.6914893617021276,
      "grad_norm": 0.3034622193668983,
      "learning_rate": 6.526557021191093e-07,
      "loss": 0.9262,
      "step": 759
    },
    {
      "epoch": 2.6950354609929077,
      "grad_norm": 0.3055046353235313,
      "learning_rate": 6.38065804593595e-07,
      "loss": 0.9175,
      "step": 760
    },
    {
      "epoch": 2.698581560283688,
      "grad_norm": 1.273630558495017,
      "learning_rate": 6.236354565503333e-07,
      "loss": 0.9167,
      "step": 761
    },
    {
      "epoch": 2.702127659574468,
      "grad_norm": 0.3166786008903364,
      "learning_rate": 6.093649039165616e-07,
      "loss": 0.9213,
      "step": 762
    },
    {
      "epoch": 2.705673758865248,
      "grad_norm": 0.297235641807391,
      "learning_rate": 5.95254389896217e-07,
      "loss": 0.9473,
      "step": 763
    },
    {
      "epoch": 2.7092198581560285,
      "grad_norm": 0.293044710631368,
      "learning_rate": 5.813041549658027e-07,
      "loss": 0.939,
      "step": 764
    },
    {
      "epoch": 2.7127659574468086,
      "grad_norm": 0.2933241514093948,
      "learning_rate": 5.675144368702901e-07,
      "loss": 0.9254,
      "step": 765
    },
    {
      "epoch": 2.7163120567375887,
      "grad_norm": 0.31330364961383234,
      "learning_rate": 5.538854706190633e-07,
      "loss": 0.9573,
      "step": 766
    },
    {
      "epoch": 2.719858156028369,
      "grad_norm": 0.31715255291186145,
      "learning_rate": 5.404174884819102e-07,
      "loss": 0.8879,
      "step": 767
    },
    {
      "epoch": 2.723404255319149,
      "grad_norm": 2.600016967668859,
      "learning_rate": 5.271107199850777e-07,
      "loss": 0.9455,
      "step": 768
    },
    {
      "epoch": 2.726950354609929,
      "grad_norm": 0.30313200742910756,
      "learning_rate": 5.139653919073451e-07,
      "loss": 0.9212,
      "step": 769
    },
    {
      "epoch": 2.7304964539007095,
      "grad_norm": 0.29776498237206805,
      "learning_rate": 5.009817282761675e-07,
      "loss": 0.9094,
      "step": 770
    },
    {
      "epoch": 2.7340425531914896,
      "grad_norm": 0.285529807459011,
      "learning_rate": 4.881599503638546e-07,
      "loss": 0.9218,
      "step": 771
    },
    {
      "epoch": 2.7375886524822697,
      "grad_norm": 0.290727814060904,
      "learning_rate": 4.7550027668380127e-07,
      "loss": 0.94,
      "step": 772
    },
    {
      "epoch": 2.7411347517730498,
      "grad_norm": 0.31720136641130625,
      "learning_rate": 4.6300292298676497e-07,
      "loss": 0.9137,
      "step": 773
    },
    {
      "epoch": 2.74468085106383,
      "grad_norm": 0.3103862279410868,
      "learning_rate": 4.506681022571846e-07,
      "loss": 0.9136,
      "step": 774
    },
    {
      "epoch": 2.74822695035461,
      "grad_norm": 0.28157134238787646,
      "learning_rate": 4.384960247095549e-07,
      "loss": 0.8961,
      "step": 775
    },
    {
      "epoch": 2.75177304964539,
      "grad_norm": 0.28882298841598714,
      "learning_rate": 4.2648689778484243e-07,
      "loss": 0.9073,
      "step": 776
    },
    {
      "epoch": 2.75531914893617,
      "grad_norm": 0.2879740747490477,
      "learning_rate": 4.146409261469486e-07,
      "loss": 0.9342,
      "step": 777
    },
    {
      "epoch": 2.7588652482269502,
      "grad_norm": 0.3186520970581419,
      "learning_rate": 4.029583116792235e-07,
      "loss": 0.8772,
      "step": 778
    },
    {
      "epoch": 2.7624113475177303,
      "grad_norm": 0.29547086736383116,
      "learning_rate": 3.914392534810241e-07,
      "loss": 0.9141,
      "step": 779
    },
    {
      "epoch": 2.7659574468085104,
      "grad_norm": 0.2980358066392843,
      "learning_rate": 3.800839478643259e-07,
      "loss": 0.9228,
      "step": 780
    },
    {
      "epoch": 2.7695035460992905,
      "grad_norm": 0.28384720444289635,
      "learning_rate": 3.68892588350368e-07,
      "loss": 0.901,
      "step": 781
    },
    {
      "epoch": 2.773049645390071,
      "grad_norm": 0.36764343853620207,
      "learning_rate": 3.578653656663655e-07,
      "loss": 0.9218,
      "step": 782
    },
    {
      "epoch": 2.776595744680851,
      "grad_norm": 0.29961480527731943,
      "learning_rate": 3.4700246774225124e-07,
      "loss": 0.9248,
      "step": 783
    },
    {
      "epoch": 2.780141843971631,
      "grad_norm": 0.29529547831757363,
      "learning_rate": 3.3630407970747457e-07,
      "loss": 0.9089,
      "step": 784
    },
    {
      "epoch": 2.7836879432624113,
      "grad_norm": 0.2928645193950287,
      "learning_rate": 3.257703838878523e-07,
      "loss": 0.945,
      "step": 785
    },
    {
      "epoch": 2.7872340425531914,
      "grad_norm": 0.3013558839362537,
      "learning_rate": 3.154015598024529e-07,
      "loss": 0.955,
      "step": 786
    },
    {
      "epoch": 2.7907801418439715,
      "grad_norm": 0.3103783207215448,
      "learning_rate": 3.0519778416054023e-07,
      "loss": 0.8506,
      "step": 787
    },
    {
      "epoch": 2.794326241134752,
      "grad_norm": 0.2821659794655092,
      "learning_rate": 2.951592308585671e-07,
      "loss": 0.9653,
      "step": 788
    },
    {
      "epoch": 2.797872340425532,
      "grad_norm": 0.3057307542652081,
      "learning_rate": 2.8528607097719986e-07,
      "loss": 0.9494,
      "step": 789
    },
    {
      "epoch": 2.801418439716312,
      "grad_norm": 0.30527993253257074,
      "learning_rate": 2.755784727784194e-07,
      "loss": 0.9221,
      "step": 790
    },
    {
      "epoch": 2.8049645390070923,
      "grad_norm": 0.29700128829756606,
      "learning_rate": 2.6603660170263725e-07,
      "loss": 0.9206,
      "step": 791
    },
    {
      "epoch": 2.8085106382978724,
      "grad_norm": 0.3190882543220009,
      "learning_rate": 2.5666062036588504e-07,
      "loss": 0.9236,
      "step": 792
    },
    {
      "epoch": 2.8120567375886525,
      "grad_norm": 0.2930659555504179,
      "learning_rate": 2.474506885570416e-07,
      "loss": 0.9211,
      "step": 793
    },
    {
      "epoch": 2.8156028368794326,
      "grad_norm": 0.29895327454226917,
      "learning_rate": 2.3840696323510826e-07,
      "loss": 0.909,
      "step": 794
    },
    {
      "epoch": 2.8191489361702127,
      "grad_norm": 0.3005687202021333,
      "learning_rate": 2.2952959852653888e-07,
      "loss": 0.8934,
      "step": 795
    },
    {
      "epoch": 2.8226950354609928,
      "grad_norm": 0.28501994772756833,
      "learning_rate": 2.208187457226052e-07,
      "loss": 0.9048,
      "step": 796
    },
    {
      "epoch": 2.826241134751773,
      "grad_norm": 0.31036487157590603,
      "learning_rate": 2.1227455327682223e-07,
      "loss": 0.9531,
      "step": 797
    },
    {
      "epoch": 2.829787234042553,
      "grad_norm": 0.28761242138340226,
      "learning_rate": 2.0389716680242368e-07,
      "loss": 0.9218,
      "step": 798
    },
    {
      "epoch": 2.8333333333333335,
      "grad_norm": 0.30278424167169304,
      "learning_rate": 1.956867290698683e-07,
      "loss": 0.9244,
      "step": 799
    },
    {
      "epoch": 2.8368794326241136,
      "grad_norm": 0.29268419217196073,
      "learning_rate": 1.8764338000442083e-07,
      "loss": 0.8955,
      "step": 800
    },
    {
      "epoch": 2.8404255319148937,
      "grad_norm": 0.3006157318897833,
      "learning_rate": 1.7976725668375493e-07,
      "loss": 0.9066,
      "step": 801
    },
    {
      "epoch": 2.8439716312056738,
      "grad_norm": 0.2989253069129778,
      "learning_rate": 1.7205849333562509e-07,
      "loss": 0.9521,
      "step": 802
    },
    {
      "epoch": 2.847517730496454,
      "grad_norm": 0.2948078480861051,
      "learning_rate": 1.6451722133557613e-07,
      "loss": 0.9087,
      "step": 803
    },
    {
      "epoch": 2.851063829787234,
      "grad_norm": 0.28509008741050834,
      "learning_rate": 1.5714356920470297e-07,
      "loss": 0.8979,
      "step": 804
    },
    {
      "epoch": 2.854609929078014,
      "grad_norm": 0.2878937554326602,
      "learning_rate": 1.4993766260746333e-07,
      "loss": 0.9296,
      "step": 805
    },
    {
      "epoch": 2.8581560283687946,
      "grad_norm": 0.3079617762356363,
      "learning_rate": 1.4289962434953508e-07,
      "loss": 0.8937,
      "step": 806
    },
    {
      "epoch": 2.8617021276595747,
      "grad_norm": 0.28625196843148937,
      "learning_rate": 1.3602957437571896e-07,
      "loss": 0.9351,
      "step": 807
    },
    {
      "epoch": 2.8652482269503547,
      "grad_norm": 0.2923705943500138,
      "learning_rate": 1.293276297679058e-07,
      "loss": 0.9291,
      "step": 808
    },
    {
      "epoch": 2.868794326241135,
      "grad_norm": 0.2820391476002473,
      "learning_rate": 1.2279390474306596e-07,
      "loss": 0.9227,
      "step": 809
    },
    {
      "epoch": 2.872340425531915,
      "grad_norm": 0.5592736002471681,
      "learning_rate": 1.1642851065131632e-07,
      "loss": 0.9244,
      "step": 810
    },
    {
      "epoch": 2.875886524822695,
      "grad_norm": 0.30536487775765697,
      "learning_rate": 1.1023155597401191e-07,
      "loss": 0.918,
      "step": 811
    },
    {
      "epoch": 2.879432624113475,
      "grad_norm": 0.3029280843642643,
      "learning_rate": 1.042031463219062e-07,
      "loss": 0.9302,
      "step": 812
    },
    {
      "epoch": 2.882978723404255,
      "grad_norm": 0.2994340315241542,
      "learning_rate": 9.8343384433347e-08,
      "loss": 0.9309,
      "step": 813
    },
    {
      "epoch": 2.8865248226950353,
      "grad_norm": 0.3031555751284906,
      "learning_rate": 9.265237017252127e-08,
      "loss": 0.9233,
      "step": 814
    },
    {
      "epoch": 2.8900709219858154,
      "grad_norm": 0.2859515624839857,
      "learning_rate": 8.7130200527763e-08,
      "loss": 0.9329,
      "step": 815
    },
    {
      "epoch": 2.8936170212765955,
      "grad_norm": 0.30049527098612866,
      "learning_rate": 8.177696960989134e-08,
      "loss": 0.9497,
      "step": 816
    },
    {
      "epoch": 2.897163120567376,
      "grad_norm": 0.29111500301529464,
      "learning_rate": 7.659276865061405e-08,
      "loss": 0.9346,
      "step": 817
    },
    {
      "epoch": 2.900709219858156,
      "grad_norm": 0.31603907957179006,
      "learning_rate": 7.157768600096315e-08,
      "loss": 0.8679,
      "step": 818
    },
    {
      "epoch": 2.904255319148936,
      "grad_norm": 0.2990219556138474,
      "learning_rate": 6.673180712979955e-08,
      "loss": 0.916,
      "step": 819
    },
    {
      "epoch": 2.9078014184397163,
      "grad_norm": 0.30153042995412016,
      "learning_rate": 6.205521462235186e-08,
      "loss": 0.8636,
      "step": 820
    },
    {
      "epoch": 2.9113475177304964,
      "grad_norm": 0.3029791908615506,
      "learning_rate": 5.754798817880547e-08,
      "loss": 0.8801,
      "step": 821
    },
    {
      "epoch": 2.9148936170212765,
      "grad_norm": 0.2984269406531723,
      "learning_rate": 5.321020461295345e-08,
      "loss": 0.8671,
      "step": 822
    },
    {
      "epoch": 2.918439716312057,
      "grad_norm": 0.2904036664928134,
      "learning_rate": 4.904193785087552e-08,
      "loss": 0.9076,
      "step": 823
    },
    {
      "epoch": 2.921985815602837,
      "grad_norm": 0.30300705182935816,
      "learning_rate": 4.504325892968897e-08,
      "loss": 0.9292,
      "step": 824
    },
    {
      "epoch": 2.925531914893617,
      "grad_norm": 0.29507244978624,
      "learning_rate": 4.1214235996331894e-08,
      "loss": 0.9023,
      "step": 825
    },
    {
      "epoch": 2.9290780141843973,
      "grad_norm": 0.293245569012375,
      "learning_rate": 3.755493430640411e-08,
      "loss": 0.9244,
      "step": 826
    },
    {
      "epoch": 2.9326241134751774,
      "grad_norm": 0.3166445042435038,
      "learning_rate": 3.406541622305248e-08,
      "loss": 0.8939,
      "step": 827
    },
    {
      "epoch": 2.9361702127659575,
      "grad_norm": 0.28444259392761967,
      "learning_rate": 3.0745741215912897e-08,
      "loss": 0.9311,
      "step": 828
    },
    {
      "epoch": 2.9397163120567376,
      "grad_norm": 0.29915613230977617,
      "learning_rate": 2.7595965860089947e-08,
      "loss": 0.8967,
      "step": 829
    },
    {
      "epoch": 2.9432624113475176,
      "grad_norm": 0.28446692658506895,
      "learning_rate": 2.4616143835202166e-08,
      "loss": 0.8871,
      "step": 830
    },
    {
      "epoch": 2.9468085106382977,
      "grad_norm": 0.2911373845781085,
      "learning_rate": 2.1806325924457196e-08,
      "loss": 0.8811,
      "step": 831
    },
    {
      "epoch": 2.950354609929078,
      "grad_norm": 0.3061109679698943,
      "learning_rate": 1.9166560013791358e-08,
      "loss": 0.953,
      "step": 832
    },
    {
      "epoch": 2.953900709219858,
      "grad_norm": 0.29911989312390747,
      "learning_rate": 1.669689109105588e-08,
      "loss": 0.9677,
      "step": 833
    },
    {
      "epoch": 2.9574468085106385,
      "grad_norm": 0.2820359915616927,
      "learning_rate": 1.4397361245243046e-08,
      "loss": 0.9245,
      "step": 834
    },
    {
      "epoch": 2.9609929078014185,
      "grad_norm": 0.29965639425928353,
      "learning_rate": 1.2268009665776793e-08,
      "loss": 0.9243,
      "step": 835
    },
    {
      "epoch": 2.9645390070921986,
      "grad_norm": 0.29357391717111103,
      "learning_rate": 1.0308872641839885e-08,
      "loss": 0.9284,
      "step": 836
    },
    {
      "epoch": 2.9680851063829787,
      "grad_norm": 0.29141026412523086,
      "learning_rate": 8.51998356175776e-09,
      "loss": 0.9194,
      "step": 837
    },
    {
      "epoch": 2.971631205673759,
      "grad_norm": 0.29560515229715406,
      "learning_rate": 6.901372912426762e-09,
      "loss": 0.8976,
      "step": 838
    },
    {
      "epoch": 2.975177304964539,
      "grad_norm": 0.29014624788605387,
      "learning_rate": 5.453068278796769e-09,
      "loss": 0.9236,
      "step": 839
    },
    {
      "epoch": 2.978723404255319,
      "grad_norm": 0.3015143386594201,
      "learning_rate": 4.175094343402686e-09,
      "loss": 0.8851,
      "step": 840
    },
    {
      "epoch": 2.9822695035460995,
      "grad_norm": 0.28915706574444056,
      "learning_rate": 3.0674728859392267e-09,
      "loss": 0.8792,
      "step": 841
    },
    {
      "epoch": 2.9858156028368796,
      "grad_norm": 0.2934637584959333,
      "learning_rate": 2.130222782894542e-09,
      "loss": 0.919,
      "step": 842
    },
    {
      "epoch": 2.9893617021276597,
      "grad_norm": 0.28832944418271755,
      "learning_rate": 1.3633600072249232e-09,
      "loss": 0.9196,
      "step": 843
    },
    {
      "epoch": 2.99290780141844,
      "grad_norm": 0.3010464732427794,
      "learning_rate": 7.668976280839069e-10,
      "loss": 0.9249,
      "step": 844
    },
    {
      "epoch": 2.99645390070922,
      "grad_norm": 0.2903066150490175,
      "learning_rate": 3.408458105991219e-10,
      "loss": 0.9185,
      "step": 845
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.2891940920591566,
      "learning_rate": 8.521181570242398e-11,
      "loss": 0.8954,
      "step": 846
    }
  ],
  "logging_steps": 1,
  "max_steps": 846,
  "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": 1043657926901760.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}