{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4117647058823529,
  "eval_steps": 1000,
  "global_step": 17500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.3529411764705884e-05,
      "grad_norm": 1.90625,
      "learning_rate": 1.9607843137254903e-05,
      "loss": 3.9568,
      "step": 1
    },
    {
      "epoch": 4.705882352941177e-05,
      "grad_norm": 2.234375,
      "learning_rate": 3.9215686274509805e-05,
      "loss": 3.9789,
      "step": 2
    },
    {
      "epoch": 7.058823529411764e-05,
      "grad_norm": 2.28125,
      "learning_rate": 5.882352941176471e-05,
      "loss": 3.7759,
      "step": 3
    },
    {
      "epoch": 9.411764705882353e-05,
      "grad_norm": 3.21875,
      "learning_rate": 7.843137254901961e-05,
      "loss": 3.8232,
      "step": 4
    },
    {
      "epoch": 0.00011764705882352942,
      "grad_norm": 2.25,
      "learning_rate": 9.80392156862745e-05,
      "loss": 4.1094,
      "step": 5
    },
    {
      "epoch": 0.00014117647058823528,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00011764705882352942,
      "loss": 3.7314,
      "step": 6
    },
    {
      "epoch": 0.0001647058823529412,
      "grad_norm": 2.625,
      "learning_rate": 0.0001372549019607843,
      "loss": 4.0436,
      "step": 7
    },
    {
      "epoch": 0.00018823529411764707,
      "grad_norm": 2.140625,
      "learning_rate": 0.00015686274509803922,
      "loss": 3.7547,
      "step": 8
    },
    {
      "epoch": 0.00021176470588235295,
      "grad_norm": 1.953125,
      "learning_rate": 0.00017647058823529413,
      "loss": 3.8313,
      "step": 9
    },
    {
      "epoch": 0.00023529411764705883,
      "grad_norm": 2.125,
      "learning_rate": 0.000196078431372549,
      "loss": 3.7643,
      "step": 10
    },
    {
      "epoch": 0.0002588235294117647,
      "grad_norm": 2.390625,
      "learning_rate": 0.00021568627450980392,
      "loss": 3.7204,
      "step": 11
    },
    {
      "epoch": 0.00028235294117647056,
      "grad_norm": 2.53125,
      "learning_rate": 0.00023529411764705883,
      "loss": 4.0806,
      "step": 12
    },
    {
      "epoch": 0.00030588235294117644,
      "grad_norm": 7.53125,
      "learning_rate": 0.00025490196078431374,
      "loss": 4.1665,
      "step": 13
    },
    {
      "epoch": 0.0003294117647058824,
      "grad_norm": 2.0625,
      "learning_rate": 0.0002745098039215686,
      "loss": 3.5025,
      "step": 14
    },
    {
      "epoch": 0.00035294117647058826,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00029411764705882356,
      "loss": 3.551,
      "step": 15
    },
    {
      "epoch": 0.00037647058823529414,
      "grad_norm": 2.375,
      "learning_rate": 0.00031372549019607844,
      "loss": 3.9256,
      "step": 16
    },
    {
      "epoch": 0.0004,
      "grad_norm": 2.21875,
      "learning_rate": 0.0003333333333333333,
      "loss": 3.4917,
      "step": 17
    },
    {
      "epoch": 0.0004235294117647059,
      "grad_norm": 3.25,
      "learning_rate": 0.00035294117647058826,
      "loss": 3.7274,
      "step": 18
    },
    {
      "epoch": 0.0004470588235294118,
      "grad_norm": 3.015625,
      "learning_rate": 0.00037254901960784314,
      "loss": 3.9154,
      "step": 19
    },
    {
      "epoch": 0.00047058823529411766,
      "grad_norm": 7.5,
      "learning_rate": 0.000392156862745098,
      "loss": 3.8858,
      "step": 20
    },
    {
      "epoch": 0.0004941176470588235,
      "grad_norm": 9.375,
      "learning_rate": 0.00041176470588235296,
      "loss": 3.5193,
      "step": 21
    },
    {
      "epoch": 0.0005176470588235294,
      "grad_norm": 8.375,
      "learning_rate": 0.00043137254901960784,
      "loss": 3.8833,
      "step": 22
    },
    {
      "epoch": 0.0005411764705882352,
      "grad_norm": 6.1875,
      "learning_rate": 0.0004509803921568627,
      "loss": 3.4274,
      "step": 23
    },
    {
      "epoch": 0.0005647058823529411,
      "grad_norm": 8.875,
      "learning_rate": 0.00047058823529411766,
      "loss": 3.9125,
      "step": 24
    },
    {
      "epoch": 0.000588235294117647,
      "grad_norm": 5.09375,
      "learning_rate": 0.0004901960784313725,
      "loss": 3.642,
      "step": 25
    },
    {
      "epoch": 0.0006117647058823529,
      "grad_norm": 3.5625,
      "learning_rate": 0.0005098039215686275,
      "loss": 3.5938,
      "step": 26
    },
    {
      "epoch": 0.0006352941176470589,
      "grad_norm": 2.5,
      "learning_rate": 0.0005294117647058823,
      "loss": 3.5931,
      "step": 27
    },
    {
      "epoch": 0.0006588235294117648,
      "grad_norm": 2.71875,
      "learning_rate": 0.0005490196078431372,
      "loss": 3.4867,
      "step": 28
    },
    {
      "epoch": 0.0006823529411764706,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005686274509803922,
      "loss": 3.6586,
      "step": 29
    },
    {
      "epoch": 0.0007058823529411765,
      "grad_norm": 1.484375,
      "learning_rate": 0.0005882352941176471,
      "loss": 3.7071,
      "step": 30
    },
    {
      "epoch": 0.0007294117647058824,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006078431372549019,
      "loss": 3.6543,
      "step": 31
    },
    {
      "epoch": 0.0007529411764705883,
      "grad_norm": 1.578125,
      "learning_rate": 0.0006274509803921569,
      "loss": 3.7594,
      "step": 32
    },
    {
      "epoch": 0.0007764705882352942,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006470588235294118,
      "loss": 3.5442,
      "step": 33
    },
    {
      "epoch": 0.0008,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006666666666666666,
      "loss": 3.7202,
      "step": 34
    },
    {
      "epoch": 0.0008235294117647059,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006862745098039217,
      "loss": 3.7258,
      "step": 35
    },
    {
      "epoch": 0.0008470588235294118,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007058823529411765,
      "loss": 3.699,
      "step": 36
    },
    {
      "epoch": 0.0008705882352941177,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007254901960784315,
      "loss": 3.8727,
      "step": 37
    },
    {
      "epoch": 0.0008941176470588236,
      "grad_norm": 1.25,
      "learning_rate": 0.0007450980392156863,
      "loss": 3.5435,
      "step": 38
    },
    {
      "epoch": 0.0009176470588235294,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007647058823529412,
      "loss": 3.5568,
      "step": 39
    },
    {
      "epoch": 0.0009411764705882353,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000784313725490196,
      "loss": 3.6674,
      "step": 40
    },
    {
      "epoch": 0.0009647058823529412,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008039215686274511,
      "loss": 3.464,
      "step": 41
    },
    {
      "epoch": 0.000988235294117647,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008235294117647059,
      "loss": 3.0771,
      "step": 42
    },
    {
      "epoch": 0.0010117647058823529,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008431372549019609,
      "loss": 3.4763,
      "step": 43
    },
    {
      "epoch": 0.0010352941176470587,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008627450980392157,
      "loss": 3.699,
      "step": 44
    },
    {
      "epoch": 0.0010588235294117646,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008823529411764706,
      "loss": 3.6161,
      "step": 45
    },
    {
      "epoch": 0.0010823529411764705,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009019607843137254,
      "loss": 3.8047,
      "step": 46
    },
    {
      "epoch": 0.0011058823529411764,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009215686274509805,
      "loss": 3.3518,
      "step": 47
    },
    {
      "epoch": 0.0011294117647058823,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009411764705882353,
      "loss": 3.5424,
      "step": 48
    },
    {
      "epoch": 0.0011529411764705881,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009607843137254903,
      "loss": 3.2536,
      "step": 49
    },
    {
      "epoch": 0.001176470588235294,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000980392156862745,
      "loss": 3.776,
      "step": 50
    },
    {
      "epoch": 0.0012,
      "grad_norm": 1.1015625,
      "learning_rate": 0.001,
      "loss": 3.7646,
      "step": 51
    },
    {
      "epoch": 0.0012235294117647058,
      "grad_norm": 0.9765625,
      "learning_rate": 0.001019607843137255,
      "loss": 3.6662,
      "step": 52
    },
    {
      "epoch": 0.0012470588235294119,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00103921568627451,
      "loss": 3.3299,
      "step": 53
    },
    {
      "epoch": 0.0012705882352941178,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0010588235294117646,
      "loss": 3.5746,
      "step": 54
    },
    {
      "epoch": 0.0012941176470588236,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0010784313725490198,
      "loss": 3.0822,
      "step": 55
    },
    {
      "epoch": 0.0013176470588235295,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0010980392156862745,
      "loss": 3.0851,
      "step": 56
    },
    {
      "epoch": 0.0013411764705882354,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0011176470588235294,
      "loss": 3.4332,
      "step": 57
    },
    {
      "epoch": 0.0013647058823529413,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0011372549019607844,
      "loss": 3.3422,
      "step": 58
    },
    {
      "epoch": 0.0013882352941176472,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0011568627450980393,
      "loss": 3.1687,
      "step": 59
    },
    {
      "epoch": 0.001411764705882353,
      "grad_norm": 0.828125,
      "learning_rate": 0.0011764705882352942,
      "loss": 3.3994,
      "step": 60
    },
    {
      "epoch": 0.001435294117647059,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0011960784313725492,
      "loss": 3.3964,
      "step": 61
    },
    {
      "epoch": 0.0014588235294117648,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012156862745098039,
      "loss": 3.201,
      "step": 62
    },
    {
      "epoch": 0.0014823529411764707,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0012352941176470588,
      "loss": 3.5362,
      "step": 63
    },
    {
      "epoch": 0.0015058823529411766,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0012549019607843138,
      "loss": 3.1059,
      "step": 64
    },
    {
      "epoch": 0.0015294117647058824,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012745098039215685,
      "loss": 3.3991,
      "step": 65
    },
    {
      "epoch": 0.0015529411764705883,
      "grad_norm": 0.890625,
      "learning_rate": 0.0012941176470588236,
      "loss": 3.4027,
      "step": 66
    },
    {
      "epoch": 0.0015764705882352942,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013137254901960786,
      "loss": 3.1496,
      "step": 67
    },
    {
      "epoch": 0.0016,
      "grad_norm": 0.90625,
      "learning_rate": 0.0013333333333333333,
      "loss": 3.3433,
      "step": 68
    },
    {
      "epoch": 0.001623529411764706,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0013529411764705882,
      "loss": 3.1302,
      "step": 69
    },
    {
      "epoch": 0.0016470588235294118,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0013725490196078434,
      "loss": 3.2126,
      "step": 70
    },
    {
      "epoch": 0.0016705882352941177,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001392156862745098,
      "loss": 3.2959,
      "step": 71
    },
    {
      "epoch": 0.0016941176470588236,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001411764705882353,
      "loss": 2.9175,
      "step": 72
    },
    {
      "epoch": 0.0017176470588235295,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014313725490196078,
      "loss": 3.3352,
      "step": 73
    },
    {
      "epoch": 0.0017411764705882354,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001450980392156863,
      "loss": 3.3455,
      "step": 74
    },
    {
      "epoch": 0.0017647058823529412,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0014705882352941176,
      "loss": 3.3827,
      "step": 75
    },
    {
      "epoch": 0.0017882352941176471,
      "grad_norm": 0.609375,
      "learning_rate": 0.0014901960784313726,
      "loss": 3.2903,
      "step": 76
    },
    {
      "epoch": 0.001811764705882353,
      "grad_norm": 0.484375,
      "learning_rate": 0.0015098039215686275,
      "loss": 2.8939,
      "step": 77
    },
    {
      "epoch": 0.0018352941176470589,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0015294117647058824,
      "loss": 3.1135,
      "step": 78
    },
    {
      "epoch": 0.0018588235294117648,
      "grad_norm": 0.5,
      "learning_rate": 0.0015490196078431374,
      "loss": 3.0559,
      "step": 79
    },
    {
      "epoch": 0.0018823529411764706,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001568627450980392,
      "loss": 3.079,
      "step": 80
    },
    {
      "epoch": 0.0019058823529411765,
      "grad_norm": 0.578125,
      "learning_rate": 0.001588235294117647,
      "loss": 3.2016,
      "step": 81
    },
    {
      "epoch": 0.0019294117647058824,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0016078431372549022,
      "loss": 3.1778,
      "step": 82
    },
    {
      "epoch": 0.0019529411764705883,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001627450980392157,
      "loss": 2.9344,
      "step": 83
    },
    {
      "epoch": 0.001976470588235294,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0016470588235294118,
      "loss": 3.1187,
      "step": 84
    },
    {
      "epoch": 0.002,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0016666666666666666,
      "loss": 2.7453,
      "step": 85
    },
    {
      "epoch": 0.0020235294117647057,
      "grad_norm": 3.796875,
      "learning_rate": 0.0016862745098039217,
      "loss": 3.1954,
      "step": 86
    },
    {
      "epoch": 0.002047058823529412,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0017058823529411766,
      "loss": 2.8589,
      "step": 87
    },
    {
      "epoch": 0.0020705882352941175,
      "grad_norm": 0.515625,
      "learning_rate": 0.0017254901960784314,
      "loss": 3.0293,
      "step": 88
    },
    {
      "epoch": 0.0020941176470588236,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0017450980392156863,
      "loss": 2.9419,
      "step": 89
    },
    {
      "epoch": 0.0021176470588235292,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0017647058823529412,
      "loss": 3.217,
      "step": 90
    },
    {
      "epoch": 0.0021411764705882353,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0017843137254901962,
      "loss": 2.8917,
      "step": 91
    },
    {
      "epoch": 0.002164705882352941,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001803921568627451,
      "loss": 2.9018,
      "step": 92
    },
    {
      "epoch": 0.002188235294117647,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0018235294117647058,
      "loss": 2.9433,
      "step": 93
    },
    {
      "epoch": 0.0022117647058823527,
      "grad_norm": 0.53125,
      "learning_rate": 0.001843137254901961,
      "loss": 3.0387,
      "step": 94
    },
    {
      "epoch": 0.002235294117647059,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0018627450980392157,
      "loss": 3.0548,
      "step": 95
    },
    {
      "epoch": 0.0022588235294117645,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0018823529411764706,
      "loss": 3.1088,
      "step": 96
    },
    {
      "epoch": 0.0022823529411764706,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0019019607843137254,
      "loss": 2.9728,
      "step": 97
    },
    {
      "epoch": 0.0023058823529411763,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0019215686274509805,
      "loss": 2.9939,
      "step": 98
    },
    {
      "epoch": 0.0023294117647058824,
      "grad_norm": 0.4375,
      "learning_rate": 0.0019411764705882354,
      "loss": 2.9188,
      "step": 99
    },
    {
      "epoch": 0.002352941176470588,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00196078431372549,
      "loss": 2.7921,
      "step": 100
    },
    {
      "epoch": 0.002376470588235294,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001980392156862745,
      "loss": 2.7044,
      "step": 101
    },
    {
      "epoch": 0.0024,
      "grad_norm": 0.4140625,
      "learning_rate": 0.002,
      "loss": 2.804,
      "step": 102
    },
    {
      "epoch": 0.002423529411764706,
      "grad_norm": 0.39453125,
      "learning_rate": 0.002019607843137255,
      "loss": 2.9178,
      "step": 103
    },
    {
      "epoch": 0.0024470588235294116,
      "grad_norm": 0.453125,
      "learning_rate": 0.00203921568627451,
      "loss": 2.9348,
      "step": 104
    },
    {
      "epoch": 0.0024705882352941176,
      "grad_norm": 0.390625,
      "learning_rate": 0.002058823529411765,
      "loss": 2.7403,
      "step": 105
    },
    {
      "epoch": 0.0024941176470588237,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00207843137254902,
      "loss": 2.7532,
      "step": 106
    },
    {
      "epoch": 0.0025176470588235294,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0020980392156862747,
      "loss": 2.6425,
      "step": 107
    },
    {
      "epoch": 0.0025411764705882355,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0021176470588235292,
      "loss": 2.7924,
      "step": 108
    },
    {
      "epoch": 0.002564705882352941,
      "grad_norm": 0.37890625,
      "learning_rate": 0.002137254901960784,
      "loss": 2.8035,
      "step": 109
    },
    {
      "epoch": 0.0025882352941176473,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0021568627450980395,
      "loss": 3.0268,
      "step": 110
    },
    {
      "epoch": 0.002611764705882353,
      "grad_norm": 0.447265625,
      "learning_rate": 0.002176470588235294,
      "loss": 2.5819,
      "step": 111
    },
    {
      "epoch": 0.002635294117647059,
      "grad_norm": 0.357421875,
      "learning_rate": 0.002196078431372549,
      "loss": 2.4956,
      "step": 112
    },
    {
      "epoch": 0.0026588235294117647,
      "grad_norm": 0.416015625,
      "learning_rate": 0.002215686274509804,
      "loss": 2.9831,
      "step": 113
    },
    {
      "epoch": 0.002682352941176471,
      "grad_norm": 0.41015625,
      "learning_rate": 0.002235294117647059,
      "loss": 2.6408,
      "step": 114
    },
    {
      "epoch": 0.0027058823529411765,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0022549019607843138,
      "loss": 2.5874,
      "step": 115
    },
    {
      "epoch": 0.0027294117647058825,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0022745098039215687,
      "loss": 2.7487,
      "step": 116
    },
    {
      "epoch": 0.002752941176470588,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0022941176470588237,
      "loss": 2.6498,
      "step": 117
    },
    {
      "epoch": 0.0027764705882352943,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0023137254901960786,
      "loss": 2.9084,
      "step": 118
    },
    {
      "epoch": 0.0028,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0023333333333333335,
      "loss": 2.8356,
      "step": 119
    },
    {
      "epoch": 0.002823529411764706,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0023529411764705885,
      "loss": 2.6952,
      "step": 120
    },
    {
      "epoch": 0.0028470588235294117,
      "grad_norm": 0.33984375,
      "learning_rate": 0.002372549019607843,
      "loss": 2.6401,
      "step": 121
    },
    {
      "epoch": 0.002870588235294118,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0023921568627450983,
      "loss": 2.6351,
      "step": 122
    },
    {
      "epoch": 0.0028941176470588235,
      "grad_norm": 0.375,
      "learning_rate": 0.002411764705882353,
      "loss": 2.7059,
      "step": 123
    },
    {
      "epoch": 0.0029176470588235296,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0024313725490196078,
      "loss": 2.7688,
      "step": 124
    },
    {
      "epoch": 0.0029411764705882353,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0024509803921568627,
      "loss": 2.5525,
      "step": 125
    },
    {
      "epoch": 0.0029647058823529414,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0024705882352941176,
      "loss": 2.5386,
      "step": 126
    },
    {
      "epoch": 0.002988235294117647,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0024901960784313726,
      "loss": 2.768,
      "step": 127
    },
    {
      "epoch": 0.003011764705882353,
      "grad_norm": 0.34375,
      "learning_rate": 0.0025098039215686275,
      "loss": 2.6744,
      "step": 128
    },
    {
      "epoch": 0.0030352941176470588,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0025294117647058825,
      "loss": 2.6531,
      "step": 129
    },
    {
      "epoch": 0.003058823529411765,
      "grad_norm": 0.31640625,
      "learning_rate": 0.002549019607843137,
      "loss": 2.3649,
      "step": 130
    },
    {
      "epoch": 0.0030823529411764705,
      "grad_norm": 0.349609375,
      "learning_rate": 0.002568627450980392,
      "loss": 2.6988,
      "step": 131
    },
    {
      "epoch": 0.0031058823529411766,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0025882352941176473,
      "loss": 2.3315,
      "step": 132
    },
    {
      "epoch": 0.0031294117647058823,
      "grad_norm": 0.318359375,
      "learning_rate": 0.002607843137254902,
      "loss": 2.4722,
      "step": 133
    },
    {
      "epoch": 0.0031529411764705884,
      "grad_norm": 0.349609375,
      "learning_rate": 0.002627450980392157,
      "loss": 2.5248,
      "step": 134
    },
    {
      "epoch": 0.003176470588235294,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0026470588235294116,
      "loss": 2.551,
      "step": 135
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0026666666666666666,
      "loss": 2.5369,
      "step": 136
    },
    {
      "epoch": 0.003223529411764706,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0026862745098039215,
      "loss": 2.8454,
      "step": 137
    },
    {
      "epoch": 0.003247058823529412,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0027058823529411765,
      "loss": 2.4885,
      "step": 138
    },
    {
      "epoch": 0.0032705882352941176,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0027254901960784314,
      "loss": 2.7896,
      "step": 139
    },
    {
      "epoch": 0.0032941176470588237,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0027450980392156868,
      "loss": 2.5455,
      "step": 140
    },
    {
      "epoch": 0.0033176470588235293,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0027647058823529413,
      "loss": 2.5221,
      "step": 141
    },
    {
      "epoch": 0.0033411764705882354,
      "grad_norm": 0.296875,
      "learning_rate": 0.002784313725490196,
      "loss": 2.6032,
      "step": 142
    },
    {
      "epoch": 0.003364705882352941,
      "grad_norm": 0.306640625,
      "learning_rate": 0.002803921568627451,
      "loss": 2.5689,
      "step": 143
    },
    {
      "epoch": 0.003388235294117647,
      "grad_norm": 0.3125,
      "learning_rate": 0.002823529411764706,
      "loss": 2.6036,
      "step": 144
    },
    {
      "epoch": 0.003411764705882353,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0028431372549019606,
      "loss": 2.5662,
      "step": 145
    },
    {
      "epoch": 0.003435294117647059,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0028627450980392155,
      "loss": 2.4867,
      "step": 146
    },
    {
      "epoch": 0.0034588235294117646,
      "grad_norm": 0.28125,
      "learning_rate": 0.0028823529411764704,
      "loss": 2.3913,
      "step": 147
    },
    {
      "epoch": 0.0034823529411764707,
      "grad_norm": 0.283203125,
      "learning_rate": 0.002901960784313726,
      "loss": 2.5443,
      "step": 148
    },
    {
      "epoch": 0.0035058823529411764,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0029215686274509808,
      "loss": 2.3521,
      "step": 149
    },
    {
      "epoch": 0.0035294117647058825,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0029411764705882353,
      "loss": 2.6995,
      "step": 150
    },
    {
      "epoch": 0.003552941176470588,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00296078431372549,
      "loss": 2.3555,
      "step": 151
    },
    {
      "epoch": 0.0035764705882352942,
      "grad_norm": 0.275390625,
      "learning_rate": 0.002980392156862745,
      "loss": 2.6314,
      "step": 152
    },
    {
      "epoch": 0.0036,
      "grad_norm": 0.271484375,
      "learning_rate": 0.003,
      "loss": 2.5629,
      "step": 153
    },
    {
      "epoch": 0.003623529411764706,
      "grad_norm": 0.2734375,
      "learning_rate": 0.003019607843137255,
      "loss": 2.5639,
      "step": 154
    },
    {
      "epoch": 0.0036470588235294117,
      "grad_norm": 0.28125,
      "learning_rate": 0.0030392156862745095,
      "loss": 2.4456,
      "step": 155
    },
    {
      "epoch": 0.0036705882352941178,
      "grad_norm": 0.26953125,
      "learning_rate": 0.003058823529411765,
      "loss": 2.5811,
      "step": 156
    },
    {
      "epoch": 0.0036941176470588234,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00307843137254902,
      "loss": 2.3388,
      "step": 157
    },
    {
      "epoch": 0.0037176470588235295,
      "grad_norm": 0.265625,
      "learning_rate": 0.0030980392156862747,
      "loss": 2.6341,
      "step": 158
    },
    {
      "epoch": 0.003741176470588235,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0031176470588235297,
      "loss": 2.6175,
      "step": 159
    },
    {
      "epoch": 0.0037647058823529413,
      "grad_norm": 0.265625,
      "learning_rate": 0.003137254901960784,
      "loss": 2.379,
      "step": 160
    },
    {
      "epoch": 0.003788235294117647,
      "grad_norm": 0.26171875,
      "learning_rate": 0.003156862745098039,
      "loss": 2.3883,
      "step": 161
    },
    {
      "epoch": 0.003811764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.003176470588235294,
      "loss": 2.4857,
      "step": 162
    },
    {
      "epoch": 0.0038352941176470587,
      "grad_norm": 0.26171875,
      "learning_rate": 0.003196078431372549,
      "loss": 2.4179,
      "step": 163
    },
    {
      "epoch": 0.003858823529411765,
      "grad_norm": 0.234375,
      "learning_rate": 0.0032156862745098044,
      "loss": 2.338,
      "step": 164
    },
    {
      "epoch": 0.0038823529411764705,
      "grad_norm": 0.248046875,
      "learning_rate": 0.003235294117647059,
      "loss": 2.3605,
      "step": 165
    },
    {
      "epoch": 0.0039058823529411766,
      "grad_norm": 0.255859375,
      "learning_rate": 0.003254901960784314,
      "loss": 2.437,
      "step": 166
    },
    {
      "epoch": 0.003929411764705883,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0032745098039215687,
      "loss": 2.378,
      "step": 167
    },
    {
      "epoch": 0.003952941176470588,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0032941176470588237,
      "loss": 2.3987,
      "step": 168
    },
    {
      "epoch": 0.003976470588235294,
      "grad_norm": 0.255859375,
      "learning_rate": 0.003313725490196078,
      "loss": 2.5146,
      "step": 169
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.003333333333333333,
      "loss": 2.3232,
      "step": 170
    },
    {
      "epoch": 0.004023529411764706,
      "grad_norm": 0.263671875,
      "learning_rate": 0.003352941176470588,
      "loss": 2.3866,
      "step": 171
    },
    {
      "epoch": 0.004047058823529411,
      "grad_norm": 0.25,
      "learning_rate": 0.0033725490196078434,
      "loss": 2.3696,
      "step": 172
    },
    {
      "epoch": 0.0040705882352941175,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0033921568627450984,
      "loss": 2.3369,
      "step": 173
    },
    {
      "epoch": 0.004094117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0034117647058823533,
      "loss": 2.4021,
      "step": 174
    },
    {
      "epoch": 0.00411764705882353,
      "grad_norm": 0.23828125,
      "learning_rate": 0.003431372549019608,
      "loss": 2.5087,
      "step": 175
    },
    {
      "epoch": 0.004141176470588235,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0034509803921568627,
      "loss": 2.5508,
      "step": 176
    },
    {
      "epoch": 0.004164705882352941,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0034705882352941177,
      "loss": 2.3707,
      "step": 177
    },
    {
      "epoch": 0.004188235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0034901960784313726,
      "loss": 2.4821,
      "step": 178
    },
    {
      "epoch": 0.004211764705882353,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.003509803921568627,
      "loss": 2.3921,
      "step": 179
    },
    {
      "epoch": 0.0042352941176470585,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0035294117647058825,
      "loss": 2.4739,
      "step": 180
    },
    {
      "epoch": 0.0042588235294117645,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0035490196078431374,
      "loss": 2.4282,
      "step": 181
    },
    {
      "epoch": 0.004282352941176471,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0035686274509803924,
      "loss": 2.4806,
      "step": 182
    },
    {
      "epoch": 0.004305882352941177,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0035882352941176473,
      "loss": 2.4293,
      "step": 183
    },
    {
      "epoch": 0.004329411764705882,
      "grad_norm": 0.2578125,
      "learning_rate": 0.003607843137254902,
      "loss": 2.3393,
      "step": 184
    },
    {
      "epoch": 0.004352941176470588,
      "grad_norm": 0.25,
      "learning_rate": 0.0036274509803921567,
      "loss": 2.362,
      "step": 185
    },
    {
      "epoch": 0.004376470588235294,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0036470588235294117,
      "loss": 2.514,
      "step": 186
    },
    {
      "epoch": 0.0044,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0036666666666666666,
      "loss": 2.2548,
      "step": 187
    },
    {
      "epoch": 0.0044235294117647055,
      "grad_norm": 0.224609375,
      "learning_rate": 0.003686274509803922,
      "loss": 2.3184,
      "step": 188
    },
    {
      "epoch": 0.004447058823529412,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.003705882352941177,
      "loss": 2.2887,
      "step": 189
    },
    {
      "epoch": 0.004470588235294118,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0037254901960784314,
      "loss": 2.4153,
      "step": 190
    },
    {
      "epoch": 0.004494117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0037450980392156863,
      "loss": 2.3524,
      "step": 191
    },
    {
      "epoch": 0.004517647058823529,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0037647058823529413,
      "loss": 2.3794,
      "step": 192
    },
    {
      "epoch": 0.004541176470588235,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.003784313725490196,
      "loss": 2.3653,
      "step": 193
    },
    {
      "epoch": 0.004564705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0038039215686274507,
      "loss": 2.3105,
      "step": 194
    },
    {
      "epoch": 0.004588235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0038235294117647057,
      "loss": 2.4153,
      "step": 195
    },
    {
      "epoch": 0.0046117647058823525,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.003843137254901961,
      "loss": 2.3851,
      "step": 196
    },
    {
      "epoch": 0.004635294117647059,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.003862745098039216,
      "loss": 2.1889,
      "step": 197
    },
    {
      "epoch": 0.004658823529411765,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.003882352941176471,
      "loss": 2.4044,
      "step": 198
    },
    {
      "epoch": 0.004682352941176471,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0039019607843137254,
      "loss": 2.3428,
      "step": 199
    },
    {
      "epoch": 0.004705882352941176,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00392156862745098,
      "loss": 2.3869,
      "step": 200
    },
    {
      "epoch": 0.004729411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.003941176470588235,
      "loss": 2.3408,
      "step": 201
    },
    {
      "epoch": 0.004752941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00396078431372549,
      "loss": 2.2928,
      "step": 202
    },
    {
      "epoch": 0.004776470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.003980392156862745,
      "loss": 2.4375,
      "step": 203
    },
    {
      "epoch": 0.0048,
      "grad_norm": 0.21484375,
      "learning_rate": 0.004,
      "loss": 2.1251,
      "step": 204
    },
    {
      "epoch": 0.004823529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.004019607843137255,
      "loss": 2.565,
      "step": 205
    },
    {
      "epoch": 0.004847058823529412,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00403921568627451,
      "loss": 2.3847,
      "step": 206
    },
    {
      "epoch": 0.004870588235294118,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.004058823529411765,
      "loss": 2.2639,
      "step": 207
    },
    {
      "epoch": 0.004894117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00407843137254902,
      "loss": 2.2048,
      "step": 208
    },
    {
      "epoch": 0.004917647058823529,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.004098039215686275,
      "loss": 2.2819,
      "step": 209
    },
    {
      "epoch": 0.004941176470588235,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00411764705882353,
      "loss": 2.212,
      "step": 210
    },
    {
      "epoch": 0.004964705882352941,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.004137254901960784,
      "loss": 2.3711,
      "step": 211
    },
    {
      "epoch": 0.0049882352941176475,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00415686274509804,
      "loss": 2.4142,
      "step": 212
    },
    {
      "epoch": 0.005011764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0041764705882352945,
      "loss": 2.2355,
      "step": 213
    },
    {
      "epoch": 0.005035294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0041960784313725494,
      "loss": 2.2607,
      "step": 214
    },
    {
      "epoch": 0.005058823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.004215686274509804,
      "loss": 2.2491,
      "step": 215
    },
    {
      "epoch": 0.005082352941176471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0042352941176470585,
      "loss": 2.1752,
      "step": 216
    },
    {
      "epoch": 0.005105882352941176,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.004254901960784313,
      "loss": 2.3886,
      "step": 217
    },
    {
      "epoch": 0.005129411764705882,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.004274509803921568,
      "loss": 2.2018,
      "step": 218
    },
    {
      "epoch": 0.0051529411764705884,
      "grad_norm": 0.220703125,
      "learning_rate": 0.004294117647058823,
      "loss": 2.1634,
      "step": 219
    },
    {
      "epoch": 0.0051764705882352945,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.004313725490196079,
      "loss": 2.1835,
      "step": 220
    },
    {
      "epoch": 0.0052,
      "grad_norm": 0.21875,
      "learning_rate": 0.004333333333333334,
      "loss": 2.227,
      "step": 221
    },
    {
      "epoch": 0.005223529411764706,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.004352941176470588,
      "loss": 2.3413,
      "step": 222
    },
    {
      "epoch": 0.005247058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.004372549019607843,
      "loss": 2.2824,
      "step": 223
    },
    {
      "epoch": 0.005270588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.004392156862745098,
      "loss": 2.3083,
      "step": 224
    },
    {
      "epoch": 0.005294117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.004411764705882353,
      "loss": 2.2657,
      "step": 225
    },
    {
      "epoch": 0.005317647058823529,
      "grad_norm": 0.22265625,
      "learning_rate": 0.004431372549019608,
      "loss": 2.3261,
      "step": 226
    },
    {
      "epoch": 0.0053411764705882355,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.004450980392156863,
      "loss": 2.2492,
      "step": 227
    },
    {
      "epoch": 0.005364705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.004470588235294118,
      "loss": 2.2707,
      "step": 228
    },
    {
      "epoch": 0.005388235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.004490196078431373,
      "loss": 2.22,
      "step": 229
    },
    {
      "epoch": 0.005411764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0045098039215686276,
      "loss": 2.2627,
      "step": 230
    },
    {
      "epoch": 0.005435294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0045294117647058825,
      "loss": 2.2264,
      "step": 231
    },
    {
      "epoch": 0.005458823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0045490196078431374,
      "loss": 2.4973,
      "step": 232
    },
    {
      "epoch": 0.00548235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.004568627450980392,
      "loss": 2.2939,
      "step": 233
    },
    {
      "epoch": 0.005505882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.004588235294117647,
      "loss": 2.3832,
      "step": 234
    },
    {
      "epoch": 0.0055294117647058825,
      "grad_norm": 0.232421875,
      "learning_rate": 0.004607843137254901,
      "loss": 2.3509,
      "step": 235
    },
    {
      "epoch": 0.005552941176470589,
      "grad_norm": 0.228515625,
      "learning_rate": 0.004627450980392157,
      "loss": 2.4068,
      "step": 236
    },
    {
      "epoch": 0.005576470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.004647058823529412,
      "loss": 2.2272,
      "step": 237
    },
    {
      "epoch": 0.0056,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.004666666666666667,
      "loss": 2.2508,
      "step": 238
    },
    {
      "epoch": 0.005623529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.004686274509803922,
      "loss": 2.3443,
      "step": 239
    },
    {
      "epoch": 0.005647058823529412,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.004705882352941177,
      "loss": 2.2683,
      "step": 240
    },
    {
      "epoch": 0.005670588235294117,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.004725490196078431,
      "loss": 2.2034,
      "step": 241
    },
    {
      "epoch": 0.0056941176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.004745098039215686,
      "loss": 2.2424,
      "step": 242
    },
    {
      "epoch": 0.0057176470588235296,
      "grad_norm": 0.22265625,
      "learning_rate": 0.004764705882352941,
      "loss": 2.2204,
      "step": 243
    },
    {
      "epoch": 0.005741176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.004784313725490197,
      "loss": 2.2765,
      "step": 244
    },
    {
      "epoch": 0.005764705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.004803921568627452,
      "loss": 2.3654,
      "step": 245
    },
    {
      "epoch": 0.005788235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.004823529411764706,
      "loss": 2.2633,
      "step": 246
    },
    {
      "epoch": 0.005811764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.004843137254901961,
      "loss": 2.3213,
      "step": 247
    },
    {
      "epoch": 0.005835294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0048627450980392155,
      "loss": 2.3858,
      "step": 248
    },
    {
      "epoch": 0.005858823529411764,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0048823529411764705,
      "loss": 2.1756,
      "step": 249
    },
    {
      "epoch": 0.0058823529411764705,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.004901960784313725,
      "loss": 2.2957,
      "step": 250
    },
    {
      "epoch": 0.005905882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00492156862745098,
      "loss": 2.068,
      "step": 251
    },
    {
      "epoch": 0.005929411764705883,
      "grad_norm": 0.248046875,
      "learning_rate": 0.004941176470588235,
      "loss": 2.4052,
      "step": 252
    },
    {
      "epoch": 0.005952941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00496078431372549,
      "loss": 2.3512,
      "step": 253
    },
    {
      "epoch": 0.005976470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.004980392156862745,
      "loss": 2.2266,
      "step": 254
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.19921875,
      "learning_rate": 0.005,
      "loss": 2.267,
      "step": 255
    },
    {
      "epoch": 0.006023529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.005019607843137255,
      "loss": 2.1559,
      "step": 256
    },
    {
      "epoch": 0.0060470588235294115,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00503921568627451,
      "loss": 2.1198,
      "step": 257
    },
    {
      "epoch": 0.0060705882352941175,
      "grad_norm": 0.197265625,
      "learning_rate": 0.005058823529411765,
      "loss": 2.2206,
      "step": 258
    },
    {
      "epoch": 0.006094117647058824,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00507843137254902,
      "loss": 2.2863,
      "step": 259
    },
    {
      "epoch": 0.00611764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.005098039215686274,
      "loss": 2.2522,
      "step": 260
    },
    {
      "epoch": 0.006141176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.005117647058823529,
      "loss": 2.0909,
      "step": 261
    },
    {
      "epoch": 0.006164705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.005137254901960784,
      "loss": 2.0993,
      "step": 262
    },
    {
      "epoch": 0.006188235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00515686274509804,
      "loss": 2.1303,
      "step": 263
    },
    {
      "epoch": 0.006211764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0051764705882352945,
      "loss": 2.1485,
      "step": 264
    },
    {
      "epoch": 0.0062352941176470585,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0051960784313725495,
      "loss": 2.2073,
      "step": 265
    },
    {
      "epoch": 0.006258823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.005215686274509804,
      "loss": 2.1251,
      "step": 266
    },
    {
      "epoch": 0.006282352941176471,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.005235294117647059,
      "loss": 2.2986,
      "step": 267
    },
    {
      "epoch": 0.006305882352941177,
      "grad_norm": 0.19921875,
      "learning_rate": 0.005254901960784314,
      "loss": 2.3288,
      "step": 268
    },
    {
      "epoch": 0.006329411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.005274509803921569,
      "loss": 2.2303,
      "step": 269
    },
    {
      "epoch": 0.006352941176470588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.005294117647058823,
      "loss": 2.2022,
      "step": 270
    },
    {
      "epoch": 0.006376470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.005313725490196078,
      "loss": 2.1487,
      "step": 271
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.2109375,
      "learning_rate": 0.005333333333333333,
      "loss": 2.3544,
      "step": 272
    },
    {
      "epoch": 0.0064235294117647055,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.005352941176470588,
      "loss": 2.1653,
      "step": 273
    },
    {
      "epoch": 0.006447058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.005372549019607843,
      "loss": 2.3084,
      "step": 274
    },
    {
      "epoch": 0.006470588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.005392156862745098,
      "loss": 2.3033,
      "step": 275
    },
    {
      "epoch": 0.006494117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.005411764705882353,
      "loss": 2.1827,
      "step": 276
    },
    {
      "epoch": 0.006517647058823529,
      "grad_norm": 0.212890625,
      "learning_rate": 0.005431372549019608,
      "loss": 2.2299,
      "step": 277
    },
    {
      "epoch": 0.006541176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.005450980392156863,
      "loss": 2.1967,
      "step": 278
    },
    {
      "epoch": 0.006564705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.005470588235294119,
      "loss": 2.1228,
      "step": 279
    },
    {
      "epoch": 0.006588235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0054901960784313735,
      "loss": 2.1909,
      "step": 280
    },
    {
      "epoch": 0.006611764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.005509803921568628,
      "loss": 2.1586,
      "step": 281
    },
    {
      "epoch": 0.006635294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0055294117647058825,
      "loss": 2.3327,
      "step": 282
    },
    {
      "epoch": 0.006658823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0055490196078431375,
      "loss": 2.1708,
      "step": 283
    },
    {
      "epoch": 0.006682352941176471,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.005568627450980392,
      "loss": 2.2662,
      "step": 284
    },
    {
      "epoch": 0.006705882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.005588235294117647,
      "loss": 2.2464,
      "step": 285
    },
    {
      "epoch": 0.006729411764705882,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.005607843137254902,
      "loss": 2.0931,
      "step": 286
    },
    {
      "epoch": 0.006752941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.005627450980392157,
      "loss": 2.1943,
      "step": 287
    },
    {
      "epoch": 0.006776470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.005647058823529412,
      "loss": 2.3344,
      "step": 288
    },
    {
      "epoch": 0.0068,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.005666666666666666,
      "loss": 2.2258,
      "step": 289
    },
    {
      "epoch": 0.006823529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.005686274509803921,
      "loss": 2.2853,
      "step": 290
    },
    {
      "epoch": 0.006847058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.005705882352941176,
      "loss": 2.256,
      "step": 291
    },
    {
      "epoch": 0.006870588235294118,
      "grad_norm": 0.197265625,
      "learning_rate": 0.005725490196078431,
      "loss": 2.1996,
      "step": 292
    },
    {
      "epoch": 0.006894117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.005745098039215686,
      "loss": 2.1479,
      "step": 293
    },
    {
      "epoch": 0.006917647058823529,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.005764705882352941,
      "loss": 2.214,
      "step": 294
    },
    {
      "epoch": 0.006941176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.005784313725490197,
      "loss": 2.2882,
      "step": 295
    },
    {
      "epoch": 0.0069647058823529414,
      "grad_norm": 0.208984375,
      "learning_rate": 0.005803921568627452,
      "loss": 2.207,
      "step": 296
    },
    {
      "epoch": 0.006988235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0058235294117647066,
      "loss": 2.2478,
      "step": 297
    },
    {
      "epoch": 0.007011764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0058431372549019615,
      "loss": 2.2355,
      "step": 298
    },
    {
      "epoch": 0.007035294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0058627450980392164,
      "loss": 2.1455,
      "step": 299
    },
    {
      "epoch": 0.007058823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0058823529411764705,
      "loss": 2.2455,
      "step": 300
    },
    {
      "epoch": 0.00708235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0059019607843137254,
      "loss": 2.0384,
      "step": 301
    },
    {
      "epoch": 0.007105882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00592156862745098,
      "loss": 2.2009,
      "step": 302
    },
    {
      "epoch": 0.007129411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.005941176470588235,
      "loss": 2.0704,
      "step": 303
    },
    {
      "epoch": 0.0071529411764705885,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00596078431372549,
      "loss": 2.0397,
      "step": 304
    },
    {
      "epoch": 0.007176470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.005980392156862745,
      "loss": 2.1747,
      "step": 305
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.006,
      "loss": 2.2924,
      "step": 306
    },
    {
      "epoch": 0.007223529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.006019607843137255,
      "loss": 2.1607,
      "step": 307
    },
    {
      "epoch": 0.007247058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00603921568627451,
      "loss": 2.2654,
      "step": 308
    },
    {
      "epoch": 0.007270588235294117,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.006058823529411764,
      "loss": 1.9992,
      "step": 309
    },
    {
      "epoch": 0.007294117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.006078431372549019,
      "loss": 2.1464,
      "step": 310
    },
    {
      "epoch": 0.007317647058823529,
      "grad_norm": 0.208984375,
      "learning_rate": 0.006098039215686275,
      "loss": 2.3234,
      "step": 311
    },
    {
      "epoch": 0.0073411764705882355,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00611764705882353,
      "loss": 2.2331,
      "step": 312
    },
    {
      "epoch": 0.007364705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.006137254901960785,
      "loss": 2.1579,
      "step": 313
    },
    {
      "epoch": 0.007388235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00615686274509804,
      "loss": 2.2339,
      "step": 314
    },
    {
      "epoch": 0.007411764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0061764705882352946,
      "loss": 2.2379,
      "step": 315
    },
    {
      "epoch": 0.007435294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0061960784313725495,
      "loss": 2.1149,
      "step": 316
    },
    {
      "epoch": 0.007458823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.006215686274509804,
      "loss": 2.267,
      "step": 317
    },
    {
      "epoch": 0.00748235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.006235294117647059,
      "loss": 2.1016,
      "step": 318
    },
    {
      "epoch": 0.0075058823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.006254901960784313,
      "loss": 2.2415,
      "step": 319
    },
    {
      "epoch": 0.0075294117647058826,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.006274509803921568,
      "loss": 2.2038,
      "step": 320
    },
    {
      "epoch": 0.007552941176470589,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.006294117647058823,
      "loss": 2.0857,
      "step": 321
    },
    {
      "epoch": 0.007576470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.006313725490196078,
      "loss": 2.216,
      "step": 322
    },
    {
      "epoch": 0.0076,
      "grad_norm": 0.201171875,
      "learning_rate": 0.006333333333333333,
      "loss": 2.0823,
      "step": 323
    },
    {
      "epoch": 0.007623529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.006352941176470588,
      "loss": 2.2043,
      "step": 324
    },
    {
      "epoch": 0.007647058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.006372549019607843,
      "loss": 2.1453,
      "step": 325
    },
    {
      "epoch": 0.007670588235294117,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.006392156862745098,
      "loss": 1.9719,
      "step": 326
    },
    {
      "epoch": 0.0076941176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.006411764705882354,
      "loss": 2.1663,
      "step": 327
    },
    {
      "epoch": 0.00771764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.006431372549019609,
      "loss": 1.9677,
      "step": 328
    },
    {
      "epoch": 0.007741176470588236,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.006450980392156864,
      "loss": 2.0481,
      "step": 329
    },
    {
      "epoch": 0.007764705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.006470588235294118,
      "loss": 2.2803,
      "step": 330
    },
    {
      "epoch": 0.007788235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.006490196078431373,
      "loss": 2.1173,
      "step": 331
    },
    {
      "epoch": 0.007811764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.006509803921568628,
      "loss": 2.1792,
      "step": 332
    },
    {
      "epoch": 0.007835294117647058,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0065294117647058825,
      "loss": 2.2146,
      "step": 333
    },
    {
      "epoch": 0.007858823529411765,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0065490196078431375,
      "loss": 2.1554,
      "step": 334
    },
    {
      "epoch": 0.00788235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.006568627450980392,
      "loss": 2.2285,
      "step": 335
    },
    {
      "epoch": 0.007905882352941176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.006588235294117647,
      "loss": 2.2655,
      "step": 336
    },
    {
      "epoch": 0.007929411764705883,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.006607843137254902,
      "loss": 2.2599,
      "step": 337
    },
    {
      "epoch": 0.007952941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.006627450980392156,
      "loss": 2.1056,
      "step": 338
    },
    {
      "epoch": 0.007976470588235295,
      "grad_norm": 0.19921875,
      "learning_rate": 0.006647058823529411,
      "loss": 2.0897,
      "step": 339
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.006666666666666666,
      "loss": 2.0884,
      "step": 340
    },
    {
      "epoch": 0.008023529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.006686274509803921,
      "loss": 2.1943,
      "step": 341
    },
    {
      "epoch": 0.008047058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.006705882352941176,
      "loss": 2.1829,
      "step": 342
    },
    {
      "epoch": 0.008070588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.006725490196078432,
      "loss": 2.0665,
      "step": 343
    },
    {
      "epoch": 0.008094117647058823,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.006745098039215687,
      "loss": 2.3008,
      "step": 344
    },
    {
      "epoch": 0.00811764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.006764705882352942,
      "loss": 2.0112,
      "step": 345
    },
    {
      "epoch": 0.008141176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.006784313725490197,
      "loss": 2.2294,
      "step": 346
    },
    {
      "epoch": 0.008164705882352942,
      "grad_norm": 0.1796875,
      "learning_rate": 0.006803921568627452,
      "loss": 2.2759,
      "step": 347
    },
    {
      "epoch": 0.008188235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.006823529411764707,
      "loss": 2.2057,
      "step": 348
    },
    {
      "epoch": 0.008211764705882352,
      "grad_norm": 0.181640625,
      "learning_rate": 0.006843137254901961,
      "loss": 2.09,
      "step": 349
    },
    {
      "epoch": 0.00823529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.006862745098039216,
      "loss": 2.0605,
      "step": 350
    },
    {
      "epoch": 0.008258823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0068823529411764705,
      "loss": 2.2348,
      "step": 351
    },
    {
      "epoch": 0.00828235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0069019607843137255,
      "loss": 2.1495,
      "step": 352
    },
    {
      "epoch": 0.008305882352941177,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00692156862745098,
      "loss": 2.1784,
      "step": 353
    },
    {
      "epoch": 0.008329411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.006941176470588235,
      "loss": 2.0765,
      "step": 354
    },
    {
      "epoch": 0.008352941176470589,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00696078431372549,
      "loss": 2.2231,
      "step": 355
    },
    {
      "epoch": 0.008376470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.006980392156862745,
      "loss": 2.146,
      "step": 356
    },
    {
      "epoch": 0.0084,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.006999999999999999,
      "loss": 2.0444,
      "step": 357
    },
    {
      "epoch": 0.008423529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.007019607843137254,
      "loss": 2.1239,
      "step": 358
    },
    {
      "epoch": 0.008447058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.007039215686274511,
      "loss": 2.1227,
      "step": 359
    },
    {
      "epoch": 0.008470588235294117,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.007058823529411765,
      "loss": 2.1576,
      "step": 360
    },
    {
      "epoch": 0.008494117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00707843137254902,
      "loss": 2.065,
      "step": 361
    },
    {
      "epoch": 0.008517647058823529,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.007098039215686275,
      "loss": 2.1478,
      "step": 362
    },
    {
      "epoch": 0.008541176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00711764705882353,
      "loss": 2.0753,
      "step": 363
    },
    {
      "epoch": 0.008564705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.007137254901960785,
      "loss": 2.1687,
      "step": 364
    },
    {
      "epoch": 0.008588235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00715686274509804,
      "loss": 2.2874,
      "step": 365
    },
    {
      "epoch": 0.008611764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.007176470588235295,
      "loss": 2.225,
      "step": 366
    },
    {
      "epoch": 0.008635294117647059,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0071960784313725495,
      "loss": 2.3942,
      "step": 367
    },
    {
      "epoch": 0.008658823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.007215686274509804,
      "loss": 2.2125,
      "step": 368
    },
    {
      "epoch": 0.008682352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0072352941176470585,
      "loss": 2.0872,
      "step": 369
    },
    {
      "epoch": 0.008705882352941176,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0072549019607843135,
      "loss": 2.141,
      "step": 370
    },
    {
      "epoch": 0.008729411764705883,
      "grad_norm": 0.19140625,
      "learning_rate": 0.007274509803921568,
      "loss": 2.1336,
      "step": 371
    },
    {
      "epoch": 0.008752941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.007294117647058823,
      "loss": 2.1631,
      "step": 372
    },
    {
      "epoch": 0.008776470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.007313725490196078,
      "loss": 2.209,
      "step": 373
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.007333333333333333,
      "loss": 2.1608,
      "step": 374
    },
    {
      "epoch": 0.008823529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.007352941176470589,
      "loss": 2.0835,
      "step": 375
    },
    {
      "epoch": 0.008847058823529411,
      "grad_norm": 0.21875,
      "learning_rate": 0.007372549019607844,
      "loss": 2.1893,
      "step": 376
    },
    {
      "epoch": 0.008870588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.007392156862745099,
      "loss": 2.185,
      "step": 377
    },
    {
      "epoch": 0.008894117647058823,
      "grad_norm": 0.193359375,
      "learning_rate": 0.007411764705882354,
      "loss": 2.3353,
      "step": 378
    },
    {
      "epoch": 0.00891764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.007431372549019608,
      "loss": 2.3497,
      "step": 379
    },
    {
      "epoch": 0.008941176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.007450980392156863,
      "loss": 2.0108,
      "step": 380
    },
    {
      "epoch": 0.00896470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.007470588235294118,
      "loss": 2.1573,
      "step": 381
    },
    {
      "epoch": 0.008988235294117648,
      "grad_norm": 0.185546875,
      "learning_rate": 0.007490196078431373,
      "loss": 2.1313,
      "step": 382
    },
    {
      "epoch": 0.009011764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.007509803921568628,
      "loss": 2.1006,
      "step": 383
    },
    {
      "epoch": 0.009035294117647058,
      "grad_norm": 0.1875,
      "learning_rate": 0.0075294117647058826,
      "loss": 2.1571,
      "step": 384
    },
    {
      "epoch": 0.009058823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0075490196078431375,
      "loss": 2.0449,
      "step": 385
    },
    {
      "epoch": 0.00908235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.007568627450980392,
      "loss": 2.1997,
      "step": 386
    },
    {
      "epoch": 0.009105882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0075882352941176465,
      "loss": 2.1927,
      "step": 387
    },
    {
      "epoch": 0.009129411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0076078431372549014,
      "loss": 2.1963,
      "step": 388
    },
    {
      "epoch": 0.009152941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.007627450980392156,
      "loss": 2.0641,
      "step": 389
    },
    {
      "epoch": 0.009176470588235295,
      "grad_norm": 0.162109375,
      "learning_rate": 0.007647058823529411,
      "loss": 2.2003,
      "step": 390
    },
    {
      "epoch": 0.0092,
      "grad_norm": 0.205078125,
      "learning_rate": 0.007666666666666667,
      "loss": 2.1684,
      "step": 391
    },
    {
      "epoch": 0.009223529411764705,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.007686274509803922,
      "loss": 2.1669,
      "step": 392
    },
    {
      "epoch": 0.009247058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.007705882352941177,
      "loss": 2.1013,
      "step": 393
    },
    {
      "epoch": 0.009270588235294117,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.007725490196078432,
      "loss": 2.2286,
      "step": 394
    },
    {
      "epoch": 0.009294117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.007745098039215687,
      "loss": 2.0893,
      "step": 395
    },
    {
      "epoch": 0.00931764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.007764705882352942,
      "loss": 2.127,
      "step": 396
    },
    {
      "epoch": 0.009341176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.007784313725490197,
      "loss": 2.217,
      "step": 397
    },
    {
      "epoch": 0.009364705882352942,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.007803921568627451,
      "loss": 2.1797,
      "step": 398
    },
    {
      "epoch": 0.009388235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.007823529411764707,
      "loss": 2.1499,
      "step": 399
    },
    {
      "epoch": 0.009411764705882352,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00784313725490196,
      "loss": 2.1353,
      "step": 400
    },
    {
      "epoch": 0.009435294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.007862745098039216,
      "loss": 2.1124,
      "step": 401
    },
    {
      "epoch": 0.009458823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00788235294117647,
      "loss": 2.0926,
      "step": 402
    },
    {
      "epoch": 0.009482352941176471,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.007901960784313725,
      "loss": 2.1036,
      "step": 403
    },
    {
      "epoch": 0.009505882352941177,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00792156862745098,
      "loss": 2.2467,
      "step": 404
    },
    {
      "epoch": 0.009529411764705882,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.007941176470588234,
      "loss": 2.1827,
      "step": 405
    },
    {
      "epoch": 0.009552941176470589,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00796078431372549,
      "loss": 2.1607,
      "step": 406
    },
    {
      "epoch": 0.009576470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.007980392156862746,
      "loss": 2.1403,
      "step": 407
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.008,
      "loss": 2.0227,
      "step": 408
    },
    {
      "epoch": 0.009623529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.008019607843137256,
      "loss": 2.0715,
      "step": 409
    },
    {
      "epoch": 0.009647058823529411,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00803921568627451,
      "loss": 2.1582,
      "step": 410
    },
    {
      "epoch": 0.009670588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.008058823529411766,
      "loss": 2.1764,
      "step": 411
    },
    {
      "epoch": 0.009694117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.00807843137254902,
      "loss": 2.145,
      "step": 412
    },
    {
      "epoch": 0.009717647058823529,
      "grad_norm": 0.205078125,
      "learning_rate": 0.008098039215686274,
      "loss": 1.7898,
      "step": 413
    },
    {
      "epoch": 0.009741176470588236,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00811764705882353,
      "loss": 1.9961,
      "step": 414
    },
    {
      "epoch": 0.009764705882352941,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.008137254901960784,
      "loss": 2.0484,
      "step": 415
    },
    {
      "epoch": 0.009788235294117646,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00815686274509804,
      "loss": 2.1118,
      "step": 416
    },
    {
      "epoch": 0.009811764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.008176470588235294,
      "loss": 2.1203,
      "step": 417
    },
    {
      "epoch": 0.009835294117647058,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00819607843137255,
      "loss": 2.15,
      "step": 418
    },
    {
      "epoch": 0.009858823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.008215686274509804,
      "loss": 2.0821,
      "step": 419
    },
    {
      "epoch": 0.00988235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00823529411764706,
      "loss": 2.1261,
      "step": 420
    },
    {
      "epoch": 0.009905882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.008254901960784313,
      "loss": 2.0952,
      "step": 421
    },
    {
      "epoch": 0.009929411764705883,
      "grad_norm": 0.220703125,
      "learning_rate": 0.008274509803921568,
      "loss": 2.1504,
      "step": 422
    },
    {
      "epoch": 0.009952941176470588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.008294117647058823,
      "loss": 2.0334,
      "step": 423
    },
    {
      "epoch": 0.009976470588235295,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00831372549019608,
      "loss": 2.0421,
      "step": 424
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.008333333333333333,
      "loss": 2.1192,
      "step": 425
    },
    {
      "epoch": 0.010023529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.008352941176470589,
      "loss": 2.1195,
      "step": 426
    },
    {
      "epoch": 0.010047058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.008372549019607843,
      "loss": 2.1695,
      "step": 427
    },
    {
      "epoch": 0.010070588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.008392156862745099,
      "loss": 1.9726,
      "step": 428
    },
    {
      "epoch": 0.010094117647058823,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.008411764705882353,
      "loss": 2.1306,
      "step": 429
    },
    {
      "epoch": 0.01011764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.008431372549019609,
      "loss": 2.0452,
      "step": 430
    },
    {
      "epoch": 0.010141176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.008450980392156863,
      "loss": 2.2273,
      "step": 431
    },
    {
      "epoch": 0.010164705882352942,
      "grad_norm": 0.189453125,
      "learning_rate": 0.008470588235294117,
      "loss": 2.0373,
      "step": 432
    },
    {
      "epoch": 0.010188235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.008490196078431373,
      "loss": 1.8611,
      "step": 433
    },
    {
      "epoch": 0.010211764705882352,
      "grad_norm": 0.19921875,
      "learning_rate": 0.008509803921568627,
      "loss": 2.1983,
      "step": 434
    },
    {
      "epoch": 0.01023529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.008529411764705883,
      "loss": 2.1181,
      "step": 435
    },
    {
      "epoch": 0.010258823529411765,
      "grad_norm": 0.236328125,
      "learning_rate": 0.008549019607843137,
      "loss": 2.1015,
      "step": 436
    },
    {
      "epoch": 0.01028235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.008568627450980392,
      "loss": 2.0905,
      "step": 437
    },
    {
      "epoch": 0.010305882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.008588235294117647,
      "loss": 2.2497,
      "step": 438
    },
    {
      "epoch": 0.010329411764705882,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.008607843137254902,
      "loss": 2.1347,
      "step": 439
    },
    {
      "epoch": 0.010352941176470589,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.008627450980392158,
      "loss": 2.1154,
      "step": 440
    },
    {
      "epoch": 0.010376470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.008647058823529412,
      "loss": 1.9484,
      "step": 441
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.197265625,
      "learning_rate": 0.008666666666666668,
      "loss": 1.9952,
      "step": 442
    },
    {
      "epoch": 0.010423529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.008686274509803922,
      "loss": 2.0503,
      "step": 443
    },
    {
      "epoch": 0.010447058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.008705882352941176,
      "loss": 2.1388,
      "step": 444
    },
    {
      "epoch": 0.010470588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.008725490196078432,
      "loss": 2.0707,
      "step": 445
    },
    {
      "epoch": 0.010494117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.008745098039215686,
      "loss": 2.1499,
      "step": 446
    },
    {
      "epoch": 0.01051764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.008764705882352942,
      "loss": 2.1149,
      "step": 447
    },
    {
      "epoch": 0.010541176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.008784313725490196,
      "loss": 2.1019,
      "step": 448
    },
    {
      "epoch": 0.010564705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.008803921568627452,
      "loss": 2.0561,
      "step": 449
    },
    {
      "epoch": 0.010588235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.008823529411764706,
      "loss": 2.1767,
      "step": 450
    },
    {
      "epoch": 0.010611764705882354,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00884313725490196,
      "loss": 2.1577,
      "step": 451
    },
    {
      "epoch": 0.010635294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.008862745098039216,
      "loss": 2.115,
      "step": 452
    },
    {
      "epoch": 0.010658823529411764,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00888235294117647,
      "loss": 2.1383,
      "step": 453
    },
    {
      "epoch": 0.010682352941176471,
      "grad_norm": 0.234375,
      "learning_rate": 0.008901960784313726,
      "loss": 2.1398,
      "step": 454
    },
    {
      "epoch": 0.010705882352941176,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.008921568627450981,
      "loss": 2.1097,
      "step": 455
    },
    {
      "epoch": 0.010729411764705883,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.008941176470588235,
      "loss": 2.1297,
      "step": 456
    },
    {
      "epoch": 0.010752941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.008960784313725491,
      "loss": 2.1422,
      "step": 457
    },
    {
      "epoch": 0.010776470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.008980392156862745,
      "loss": 2.1612,
      "step": 458
    },
    {
      "epoch": 0.0108,
      "grad_norm": 0.240234375,
      "learning_rate": 0.009000000000000001,
      "loss": 2.1037,
      "step": 459
    },
    {
      "epoch": 0.010823529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.009019607843137255,
      "loss": 1.9064,
      "step": 460
    },
    {
      "epoch": 0.010847058823529411,
      "grad_norm": 0.216796875,
      "learning_rate": 0.009039215686274511,
      "loss": 2.0484,
      "step": 461
    },
    {
      "epoch": 0.010870588235294118,
      "grad_norm": 0.21875,
      "learning_rate": 0.009058823529411765,
      "loss": 2.1581,
      "step": 462
    },
    {
      "epoch": 0.010894117647058823,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.009078431372549019,
      "loss": 2.0081,
      "step": 463
    },
    {
      "epoch": 0.01091764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.009098039215686275,
      "loss": 2.3011,
      "step": 464
    },
    {
      "epoch": 0.010941176470588235,
      "grad_norm": 0.22265625,
      "learning_rate": 0.009117647058823529,
      "loss": 2.1539,
      "step": 465
    },
    {
      "epoch": 0.01096470588235294,
      "grad_norm": 0.23046875,
      "learning_rate": 0.009137254901960785,
      "loss": 2.2157,
      "step": 466
    },
    {
      "epoch": 0.010988235294117648,
      "grad_norm": 0.228515625,
      "learning_rate": 0.009156862745098039,
      "loss": 2.1755,
      "step": 467
    },
    {
      "epoch": 0.011011764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.009176470588235295,
      "loss": 2.108,
      "step": 468
    },
    {
      "epoch": 0.011035294117647058,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.009196078431372549,
      "loss": 2.1063,
      "step": 469
    },
    {
      "epoch": 0.011058823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.009215686274509803,
      "loss": 2.175,
      "step": 470
    },
    {
      "epoch": 0.01108235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00923529411764706,
      "loss": 2.0269,
      "step": 471
    },
    {
      "epoch": 0.011105882352941177,
      "grad_norm": 0.177734375,
      "learning_rate": 0.009254901960784314,
      "loss": 2.0749,
      "step": 472
    },
    {
      "epoch": 0.011129411764705882,
      "grad_norm": 0.20703125,
      "learning_rate": 0.009274509803921568,
      "loss": 2.1214,
      "step": 473
    },
    {
      "epoch": 0.011152941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.009294117647058824,
      "loss": 2.2554,
      "step": 474
    },
    {
      "epoch": 0.011176470588235295,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.009313725490196078,
      "loss": 2.1121,
      "step": 475
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.20703125,
      "learning_rate": 0.009333333333333334,
      "loss": 2.2039,
      "step": 476
    },
    {
      "epoch": 0.011223529411764705,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.009352941176470588,
      "loss": 1.9848,
      "step": 477
    },
    {
      "epoch": 0.011247058823529412,
      "grad_norm": 0.22265625,
      "learning_rate": 0.009372549019607844,
      "loss": 2.0259,
      "step": 478
    },
    {
      "epoch": 0.011270588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.009392156862745098,
      "loss": 2.1038,
      "step": 479
    },
    {
      "epoch": 0.011294117647058824,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.009411764705882354,
      "loss": 2.1103,
      "step": 480
    },
    {
      "epoch": 0.01131764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.009431372549019608,
      "loss": 2.067,
      "step": 481
    },
    {
      "epoch": 0.011341176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.009450980392156862,
      "loss": 2.0848,
      "step": 482
    },
    {
      "epoch": 0.011364705882352942,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.009470588235294118,
      "loss": 2.1269,
      "step": 483
    },
    {
      "epoch": 0.011388235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.009490196078431372,
      "loss": 2.0856,
      "step": 484
    },
    {
      "epoch": 0.011411764705882352,
      "grad_norm": 0.201171875,
      "learning_rate": 0.009509803921568628,
      "loss": 2.0485,
      "step": 485
    },
    {
      "epoch": 0.011435294117647059,
      "grad_norm": 0.201171875,
      "learning_rate": 0.009529411764705882,
      "loss": 2.0014,
      "step": 486
    },
    {
      "epoch": 0.011458823529411764,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.009549019607843138,
      "loss": 2.1104,
      "step": 487
    },
    {
      "epoch": 0.011482352941176471,
      "grad_norm": 0.18359375,
      "learning_rate": 0.009568627450980393,
      "loss": 2.1822,
      "step": 488
    },
    {
      "epoch": 0.011505882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.009588235294117647,
      "loss": 1.8574,
      "step": 489
    },
    {
      "epoch": 0.011529411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.009607843137254903,
      "loss": 2.045,
      "step": 490
    },
    {
      "epoch": 0.011552941176470589,
      "grad_norm": 0.193359375,
      "learning_rate": 0.009627450980392157,
      "loss": 2.1607,
      "step": 491
    },
    {
      "epoch": 0.011576470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.009647058823529411,
      "loss": 2.1248,
      "step": 492
    },
    {
      "epoch": 0.0116,
      "grad_norm": 0.251953125,
      "learning_rate": 0.009666666666666667,
      "loss": 2.2034,
      "step": 493
    },
    {
      "epoch": 0.011623529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.009686274509803921,
      "loss": 2.1445,
      "step": 494
    },
    {
      "epoch": 0.011647058823529411,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.009705882352941177,
      "loss": 2.1803,
      "step": 495
    },
    {
      "epoch": 0.011670588235294118,
      "grad_norm": 0.197265625,
      "learning_rate": 0.009725490196078431,
      "loss": 1.9335,
      "step": 496
    },
    {
      "epoch": 0.011694117647058824,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.009745098039215687,
      "loss": 2.0651,
      "step": 497
    },
    {
      "epoch": 0.011717647058823529,
      "grad_norm": 0.1953125,
      "learning_rate": 0.009764705882352941,
      "loss": 2.0316,
      "step": 498
    },
    {
      "epoch": 0.011741176470588236,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.009784313725490197,
      "loss": 2.1402,
      "step": 499
    },
    {
      "epoch": 0.011764705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00980392156862745,
      "loss": 2.0787,
      "step": 500
    },
    {
      "epoch": 0.011788235294117646,
      "grad_norm": 0.19921875,
      "learning_rate": 0.009823529411764705,
      "loss": 2.0241,
      "step": 501
    },
    {
      "epoch": 0.011811764705882353,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00984313725490196,
      "loss": 2.0326,
      "step": 502
    },
    {
      "epoch": 0.011835294117647058,
      "grad_norm": 0.197265625,
      "learning_rate": 0.009862745098039217,
      "loss": 1.9835,
      "step": 503
    },
    {
      "epoch": 0.011858823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00988235294117647,
      "loss": 2.0104,
      "step": 504
    },
    {
      "epoch": 0.01188235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.009901960784313726,
      "loss": 2.2219,
      "step": 505
    },
    {
      "epoch": 0.011905882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00992156862745098,
      "loss": 2.1326,
      "step": 506
    },
    {
      "epoch": 0.011929411764705883,
      "grad_norm": 0.18359375,
      "learning_rate": 0.009941176470588236,
      "loss": 2.0285,
      "step": 507
    },
    {
      "epoch": 0.011952941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00996078431372549,
      "loss": 2.2001,
      "step": 508
    },
    {
      "epoch": 0.011976470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.009980392156862746,
      "loss": 2.0764,
      "step": 509
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.20703125,
      "learning_rate": 0.01,
      "loss": 2.1242,
      "step": 510
    },
    {
      "epoch": 0.012023529411764705,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.010019607843137254,
      "loss": 2.0826,
      "step": 511
    },
    {
      "epoch": 0.012047058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.01003921568627451,
      "loss": 2.0804,
      "step": 512
    },
    {
      "epoch": 0.012070588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.010058823529411764,
      "loss": 2.091,
      "step": 513
    },
    {
      "epoch": 0.012094117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.01007843137254902,
      "loss": 2.0525,
      "step": 514
    },
    {
      "epoch": 0.01211764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.010098039215686274,
      "loss": 1.9225,
      "step": 515
    },
    {
      "epoch": 0.012141176470588235,
      "grad_norm": 0.1640625,
      "learning_rate": 0.01011764705882353,
      "loss": 2.0344,
      "step": 516
    },
    {
      "epoch": 0.01216470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.010137254901960784,
      "loss": 2.1709,
      "step": 517
    },
    {
      "epoch": 0.012188235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.01015686274509804,
      "loss": 2.0741,
      "step": 518
    },
    {
      "epoch": 0.012211764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.010176470588235294,
      "loss": 2.1261,
      "step": 519
    },
    {
      "epoch": 0.01223529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.010196078431372548,
      "loss": 2.0721,
      "step": 520
    },
    {
      "epoch": 0.012258823529411765,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.010215686274509804,
      "loss": 1.9618,
      "step": 521
    },
    {
      "epoch": 0.01228235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.010235294117647058,
      "loss": 2.1736,
      "step": 522
    },
    {
      "epoch": 0.012305882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.010254901960784314,
      "loss": 2.0893,
      "step": 523
    },
    {
      "epoch": 0.012329411764705882,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.010274509803921568,
      "loss": 1.9437,
      "step": 524
    },
    {
      "epoch": 0.012352941176470587,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.010294117647058823,
      "loss": 2.0714,
      "step": 525
    },
    {
      "epoch": 0.012376470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.01031372549019608,
      "loss": 2.0922,
      "step": 526
    },
    {
      "epoch": 0.0124,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.010333333333333335,
      "loss": 2.2482,
      "step": 527
    },
    {
      "epoch": 0.012423529411764707,
      "grad_norm": 0.322265625,
      "learning_rate": 0.010352941176470589,
      "loss": 2.0776,
      "step": 528
    },
    {
      "epoch": 0.012447058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.010372549019607845,
      "loss": 2.0218,
      "step": 529
    },
    {
      "epoch": 0.012470588235294117,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.010392156862745099,
      "loss": 2.1595,
      "step": 530
    },
    {
      "epoch": 0.012494117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.010411764705882355,
      "loss": 2.0498,
      "step": 531
    },
    {
      "epoch": 0.01251764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.010431372549019609,
      "loss": 2.1099,
      "step": 532
    },
    {
      "epoch": 0.012541176470588236,
      "grad_norm": 0.162109375,
      "learning_rate": 0.010450980392156863,
      "loss": 2.1428,
      "step": 533
    },
    {
      "epoch": 0.012564705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.010470588235294119,
      "loss": 2.0563,
      "step": 534
    },
    {
      "epoch": 0.012588235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.010490196078431373,
      "loss": 2.0932,
      "step": 535
    },
    {
      "epoch": 0.012611764705882354,
      "grad_norm": 0.177734375,
      "learning_rate": 0.010509803921568629,
      "loss": 2.033,
      "step": 536
    },
    {
      "epoch": 0.012635294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.010529411764705883,
      "loss": 2.0231,
      "step": 537
    },
    {
      "epoch": 0.012658823529411764,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.010549019607843138,
      "loss": 2.059,
      "step": 538
    },
    {
      "epoch": 0.012682352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.010568627450980392,
      "loss": 2.1413,
      "step": 539
    },
    {
      "epoch": 0.012705882352941176,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.010588235294117647,
      "loss": 2.204,
      "step": 540
    },
    {
      "epoch": 0.012729411764705883,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.010607843137254902,
      "loss": 2.1414,
      "step": 541
    },
    {
      "epoch": 0.012752941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.010627450980392156,
      "loss": 2.1974,
      "step": 542
    },
    {
      "epoch": 0.012776470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.010647058823529412,
      "loss": 2.0953,
      "step": 543
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.010666666666666666,
      "loss": 2.1846,
      "step": 544
    },
    {
      "epoch": 0.012823529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.010686274509803922,
      "loss": 2.1334,
      "step": 545
    },
    {
      "epoch": 0.012847058823529411,
      "grad_norm": 0.197265625,
      "learning_rate": 0.010705882352941176,
      "loss": 2.2247,
      "step": 546
    },
    {
      "epoch": 0.012870588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.010725490196078432,
      "loss": 2.0523,
      "step": 547
    },
    {
      "epoch": 0.012894117647058823,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.010745098039215686,
      "loss": 2.0604,
      "step": 548
    },
    {
      "epoch": 0.01291764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.01076470588235294,
      "loss": 2.1265,
      "step": 549
    },
    {
      "epoch": 0.012941176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.010784313725490196,
      "loss": 2.017,
      "step": 550
    },
    {
      "epoch": 0.01296470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.01080392156862745,
      "loss": 2.1126,
      "step": 551
    },
    {
      "epoch": 0.012988235294117648,
      "grad_norm": 0.216796875,
      "learning_rate": 0.010823529411764706,
      "loss": 2.139,
      "step": 552
    },
    {
      "epoch": 0.013011764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.01084313725490196,
      "loss": 2.1673,
      "step": 553
    },
    {
      "epoch": 0.013035294117647058,
      "grad_norm": 0.205078125,
      "learning_rate": 0.010862745098039216,
      "loss": 2.0313,
      "step": 554
    },
    {
      "epoch": 0.013058823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.01088235294117647,
      "loss": 2.0061,
      "step": 555
    },
    {
      "epoch": 0.01308235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.010901960784313726,
      "loss": 2.0618,
      "step": 556
    },
    {
      "epoch": 0.013105882352941177,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.01092156862745098,
      "loss": 2.1524,
      "step": 557
    },
    {
      "epoch": 0.013129411764705883,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.010941176470588237,
      "loss": 2.0503,
      "step": 558
    },
    {
      "epoch": 0.013152941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.010960784313725491,
      "loss": 2.0985,
      "step": 559
    },
    {
      "epoch": 0.013176470588235295,
      "grad_norm": 0.234375,
      "learning_rate": 0.010980392156862747,
      "loss": 2.1575,
      "step": 560
    },
    {
      "epoch": 0.0132,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.011000000000000001,
      "loss": 1.8644,
      "step": 561
    },
    {
      "epoch": 0.013223529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.011019607843137255,
      "loss": 2.0956,
      "step": 562
    },
    {
      "epoch": 0.013247058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.011039215686274511,
      "loss": 2.2369,
      "step": 563
    },
    {
      "epoch": 0.013270588235294117,
      "grad_norm": 0.224609375,
      "learning_rate": 0.011058823529411765,
      "loss": 1.9664,
      "step": 564
    },
    {
      "epoch": 0.013294117647058824,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.01107843137254902,
      "loss": 2.1718,
      "step": 565
    },
    {
      "epoch": 0.01331764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.011098039215686275,
      "loss": 2.1139,
      "step": 566
    },
    {
      "epoch": 0.013341176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.01111764705882353,
      "loss": 2.024,
      "step": 567
    },
    {
      "epoch": 0.013364705882352942,
      "grad_norm": 0.185546875,
      "learning_rate": 0.011137254901960785,
      "loss": 1.9504,
      "step": 568
    },
    {
      "epoch": 0.013388235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.01115686274509804,
      "loss": 1.9977,
      "step": 569
    },
    {
      "epoch": 0.013411764705882352,
      "grad_norm": 0.212890625,
      "learning_rate": 0.011176470588235295,
      "loss": 2.2212,
      "step": 570
    },
    {
      "epoch": 0.01343529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.011196078431372549,
      "loss": 2.1077,
      "step": 571
    },
    {
      "epoch": 0.013458823529411764,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.011215686274509805,
      "loss": 1.9784,
      "step": 572
    },
    {
      "epoch": 0.013482352941176471,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.011235294117647059,
      "loss": 2.1405,
      "step": 573
    },
    {
      "epoch": 0.013505882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.011254901960784314,
      "loss": 1.9762,
      "step": 574
    },
    {
      "epoch": 0.013529411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.011274509803921568,
      "loss": 2.1075,
      "step": 575
    },
    {
      "epoch": 0.013552941176470589,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.011294117647058824,
      "loss": 2.0681,
      "step": 576
    },
    {
      "epoch": 0.013576470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.011313725490196078,
      "loss": 2.3306,
      "step": 577
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.011333333333333332,
      "loss": 2.0423,
      "step": 578
    },
    {
      "epoch": 0.013623529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.011352941176470588,
      "loss": 2.1421,
      "step": 579
    },
    {
      "epoch": 0.013647058823529411,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.011372549019607842,
      "loss": 2.1155,
      "step": 580
    },
    {
      "epoch": 0.013670588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.011392156862745098,
      "loss": 2.028,
      "step": 581
    },
    {
      "epoch": 0.013694117647058824,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.011411764705882352,
      "loss": 2.1249,
      "step": 582
    },
    {
      "epoch": 0.013717647058823529,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.011431372549019608,
      "loss": 2.0044,
      "step": 583
    },
    {
      "epoch": 0.013741176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.011450980392156862,
      "loss": 2.199,
      "step": 584
    },
    {
      "epoch": 0.013764705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.011470588235294118,
      "loss": 2.028,
      "step": 585
    },
    {
      "epoch": 0.013788235294117646,
      "grad_norm": 0.23046875,
      "learning_rate": 0.011490196078431372,
      "loss": 2.1397,
      "step": 586
    },
    {
      "epoch": 0.013811764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.011509803921568626,
      "loss": 2.0652,
      "step": 587
    },
    {
      "epoch": 0.013835294117647058,
      "grad_norm": 0.20703125,
      "learning_rate": 0.011529411764705882,
      "loss": 2.0602,
      "step": 588
    },
    {
      "epoch": 0.013858823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.011549019607843136,
      "loss": 2.0649,
      "step": 589
    },
    {
      "epoch": 0.01388235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.011568627450980393,
      "loss": 2.0808,
      "step": 590
    },
    {
      "epoch": 0.013905882352941176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.011588235294117647,
      "loss": 2.137,
      "step": 591
    },
    {
      "epoch": 0.013929411764705883,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.011607843137254903,
      "loss": 2.1129,
      "step": 592
    },
    {
      "epoch": 0.013952941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.011627450980392157,
      "loss": 2.2098,
      "step": 593
    },
    {
      "epoch": 0.013976470588235293,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.011647058823529413,
      "loss": 2.1169,
      "step": 594
    },
    {
      "epoch": 0.014,
      "grad_norm": 0.228515625,
      "learning_rate": 0.011666666666666667,
      "loss": 2.053,
      "step": 595
    },
    {
      "epoch": 0.014023529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.011686274509803923,
      "loss": 2.0214,
      "step": 596
    },
    {
      "epoch": 0.014047058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.011705882352941177,
      "loss": 2.0643,
      "step": 597
    },
    {
      "epoch": 0.014070588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.011725490196078433,
      "loss": 2.0185,
      "step": 598
    },
    {
      "epoch": 0.014094117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.011745098039215687,
      "loss": 2.1595,
      "step": 599
    },
    {
      "epoch": 0.01411764705882353,
      "grad_norm": 7.375,
      "learning_rate": 0.011764705882352941,
      "loss": 2.2205,
      "step": 600
    },
    {
      "epoch": 0.014141176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.011784313725490197,
      "loss": 1.9963,
      "step": 601
    },
    {
      "epoch": 0.01416470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.011803921568627451,
      "loss": 2.193,
      "step": 602
    },
    {
      "epoch": 0.014188235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.011823529411764707,
      "loss": 2.0049,
      "step": 603
    },
    {
      "epoch": 0.014211764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.01184313725490196,
      "loss": 2.1838,
      "step": 604
    },
    {
      "epoch": 0.01423529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.011862745098039217,
      "loss": 2.0535,
      "step": 605
    },
    {
      "epoch": 0.014258823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.01188235294117647,
      "loss": 1.9982,
      "step": 606
    },
    {
      "epoch": 0.01428235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.011901960784313726,
      "loss": 2.0949,
      "step": 607
    },
    {
      "epoch": 0.014305882352941177,
      "grad_norm": 0.17578125,
      "learning_rate": 0.01192156862745098,
      "loss": 2.0606,
      "step": 608
    },
    {
      "epoch": 0.014329411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.011941176470588235,
      "loss": 2.136,
      "step": 609
    },
    {
      "epoch": 0.014352941176470587,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.01196078431372549,
      "loss": 2.1132,
      "step": 610
    },
    {
      "epoch": 0.014376470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.011980392156862744,
      "loss": 2.0711,
      "step": 611
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.012,
      "loss": 2.1503,
      "step": 612
    },
    {
      "epoch": 0.014423529411764707,
      "grad_norm": 0.203125,
      "learning_rate": 0.012019607843137254,
      "loss": 1.9778,
      "step": 613
    },
    {
      "epoch": 0.014447058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.01203921568627451,
      "loss": 2.1092,
      "step": 614
    },
    {
      "epoch": 0.014470588235294117,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.012058823529411764,
      "loss": 2.0752,
      "step": 615
    },
    {
      "epoch": 0.014494117647058824,
      "grad_norm": 0.169921875,
      "learning_rate": 0.01207843137254902,
      "loss": 2.0316,
      "step": 616
    },
    {
      "epoch": 0.01451764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.012098039215686274,
      "loss": 2.0265,
      "step": 617
    },
    {
      "epoch": 0.014541176470588234,
      "grad_norm": 0.1953125,
      "learning_rate": 0.012117647058823528,
      "loss": 2.0905,
      "step": 618
    },
    {
      "epoch": 0.014564705882352941,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.012137254901960784,
      "loss": 2.0174,
      "step": 619
    },
    {
      "epoch": 0.014588235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.012156862745098038,
      "loss": 2.1529,
      "step": 620
    },
    {
      "epoch": 0.014611764705882354,
      "grad_norm": 0.208984375,
      "learning_rate": 0.012176470588235294,
      "loss": 1.9191,
      "step": 621
    },
    {
      "epoch": 0.014635294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.01219607843137255,
      "loss": 2.1311,
      "step": 622
    },
    {
      "epoch": 0.014658823529411764,
      "grad_norm": 0.189453125,
      "learning_rate": 0.012215686274509805,
      "loss": 2.0927,
      "step": 623
    },
    {
      "epoch": 0.014682352941176471,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.01223529411764706,
      "loss": 1.945,
      "step": 624
    },
    {
      "epoch": 0.014705882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.012254901960784315,
      "loss": 2.0937,
      "step": 625
    },
    {
      "epoch": 0.014729411764705881,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.01227450980392157,
      "loss": 2.0055,
      "step": 626
    },
    {
      "epoch": 0.014752941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.012294117647058825,
      "loss": 2.2028,
      "step": 627
    },
    {
      "epoch": 0.014776470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.01231372549019608,
      "loss": 1.9753,
      "step": 628
    },
    {
      "epoch": 0.0148,
      "grad_norm": 0.201171875,
      "learning_rate": 0.012333333333333335,
      "loss": 2.0311,
      "step": 629
    },
    {
      "epoch": 0.014823529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.012352941176470589,
      "loss": 2.0592,
      "step": 630
    },
    {
      "epoch": 0.014847058823529411,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.012372549019607843,
      "loss": 2.1233,
      "step": 631
    },
    {
      "epoch": 0.014870588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.012392156862745099,
      "loss": 2.084,
      "step": 632
    },
    {
      "epoch": 0.014894117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.012411764705882353,
      "loss": 2.1377,
      "step": 633
    },
    {
      "epoch": 0.01491764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.012431372549019609,
      "loss": 2.0817,
      "step": 634
    },
    {
      "epoch": 0.014941176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.012450980392156863,
      "loss": 1.9436,
      "step": 635
    },
    {
      "epoch": 0.01496470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.012470588235294119,
      "loss": 1.9902,
      "step": 636
    },
    {
      "epoch": 0.014988235294117648,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.012490196078431373,
      "loss": 2.0502,
      "step": 637
    },
    {
      "epoch": 0.015011764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.012509803921568627,
      "loss": 2.1683,
      "step": 638
    },
    {
      "epoch": 0.015035294117647058,
      "grad_norm": 0.201171875,
      "learning_rate": 0.012529411764705883,
      "loss": 2.0576,
      "step": 639
    },
    {
      "epoch": 0.015058823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.012549019607843137,
      "loss": 2.092,
      "step": 640
    },
    {
      "epoch": 0.01508235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.012568627450980393,
      "loss": 2.0635,
      "step": 641
    },
    {
      "epoch": 0.015105882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.012588235294117647,
      "loss": 2.0515,
      "step": 642
    },
    {
      "epoch": 0.015129411764705883,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.012607843137254902,
      "loss": 1.8519,
      "step": 643
    },
    {
      "epoch": 0.015152941176470588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.012627450980392156,
      "loss": 2.0834,
      "step": 644
    },
    {
      "epoch": 0.015176470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.012647058823529412,
      "loss": 2.1503,
      "step": 645
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.208984375,
      "learning_rate": 0.012666666666666666,
      "loss": 2.0831,
      "step": 646
    },
    {
      "epoch": 0.015223529411764705,
      "grad_norm": 0.220703125,
      "learning_rate": 0.01268627450980392,
      "loss": 2.1905,
      "step": 647
    },
    {
      "epoch": 0.015247058823529412,
      "grad_norm": 0.25390625,
      "learning_rate": 0.012705882352941176,
      "loss": 2.036,
      "step": 648
    },
    {
      "epoch": 0.015270588235294117,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.01272549019607843,
      "loss": 2.0934,
      "step": 649
    },
    {
      "epoch": 0.015294117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.012745098039215686,
      "loss": 2.0621,
      "step": 650
    },
    {
      "epoch": 0.01531764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.01276470588235294,
      "loss": 1.9856,
      "step": 651
    },
    {
      "epoch": 0.015341176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.012784313725490196,
      "loss": 2.133,
      "step": 652
    },
    {
      "epoch": 0.015364705882352942,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.01280392156862745,
      "loss": 2.2082,
      "step": 653
    },
    {
      "epoch": 0.015388235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.012823529411764708,
      "loss": 2.0193,
      "step": 654
    },
    {
      "epoch": 0.015411764705882352,
      "grad_norm": 0.193359375,
      "learning_rate": 0.012843137254901962,
      "loss": 1.9874,
      "step": 655
    },
    {
      "epoch": 0.01543529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.012862745098039217,
      "loss": 2.0638,
      "step": 656
    },
    {
      "epoch": 0.015458823529411764,
      "grad_norm": 0.21484375,
      "learning_rate": 0.012882352941176472,
      "loss": 1.9419,
      "step": 657
    },
    {
      "epoch": 0.015482352941176471,
      "grad_norm": 0.21484375,
      "learning_rate": 0.012901960784313727,
      "loss": 2.0068,
      "step": 658
    },
    {
      "epoch": 0.015505882352941177,
      "grad_norm": 0.203125,
      "learning_rate": 0.012921568627450981,
      "loss": 2.0021,
      "step": 659
    },
    {
      "epoch": 0.015529411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.012941176470588235,
      "loss": 2.0206,
      "step": 660
    },
    {
      "epoch": 0.015552941176470589,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.012960784313725491,
      "loss": 2.1862,
      "step": 661
    },
    {
      "epoch": 0.015576470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.012980392156862745,
      "loss": 2.2162,
      "step": 662
    },
    {
      "epoch": 0.0156,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.013000000000000001,
      "loss": 2.1281,
      "step": 663
    },
    {
      "epoch": 0.015623529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.013019607843137255,
      "loss": 2.1647,
      "step": 664
    },
    {
      "epoch": 0.015647058823529413,
      "grad_norm": 0.203125,
      "learning_rate": 0.013039215686274511,
      "loss": 2.1162,
      "step": 665
    },
    {
      "epoch": 0.015670588235294117,
      "grad_norm": 0.2265625,
      "learning_rate": 0.013058823529411765,
      "loss": 2.1944,
      "step": 666
    },
    {
      "epoch": 0.015694117647058824,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.013078431372549021,
      "loss": 2.1525,
      "step": 667
    },
    {
      "epoch": 0.01571764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.013098039215686275,
      "loss": 2.0534,
      "step": 668
    },
    {
      "epoch": 0.015741176470588234,
      "grad_norm": 0.1875,
      "learning_rate": 0.013117647058823529,
      "loss": 1.9543,
      "step": 669
    },
    {
      "epoch": 0.01576470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.013137254901960785,
      "loss": 1.9969,
      "step": 670
    },
    {
      "epoch": 0.015788235294117648,
      "grad_norm": 0.189453125,
      "learning_rate": 0.013156862745098039,
      "loss": 2.0765,
      "step": 671
    },
    {
      "epoch": 0.01581176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.013176470588235295,
      "loss": 2.0456,
      "step": 672
    },
    {
      "epoch": 0.01583529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.013196078431372549,
      "loss": 2.0345,
      "step": 673
    },
    {
      "epoch": 0.015858823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.013215686274509805,
      "loss": 1.9678,
      "step": 674
    },
    {
      "epoch": 0.01588235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.013235294117647059,
      "loss": 2.0594,
      "step": 675
    },
    {
      "epoch": 0.015905882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.013254901960784313,
      "loss": 2.0095,
      "step": 676
    },
    {
      "epoch": 0.015929411764705883,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.013274509803921569,
      "loss": 2.0772,
      "step": 677
    },
    {
      "epoch": 0.01595294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.013294117647058823,
      "loss": 2.0467,
      "step": 678
    },
    {
      "epoch": 0.015976470588235293,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.013313725490196078,
      "loss": 2.1912,
      "step": 679
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.1796875,
      "learning_rate": 0.013333333333333332,
      "loss": 1.9899,
      "step": 680
    },
    {
      "epoch": 0.016023529411764707,
      "grad_norm": 0.208984375,
      "learning_rate": 0.013352941176470588,
      "loss": 2.1917,
      "step": 681
    },
    {
      "epoch": 0.01604705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.013372549019607842,
      "loss": 1.9918,
      "step": 682
    },
    {
      "epoch": 0.016070588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.013392156862745098,
      "loss": 2.0937,
      "step": 683
    },
    {
      "epoch": 0.016094117647058825,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.013411764705882352,
      "loss": 2.0918,
      "step": 684
    },
    {
      "epoch": 0.016117647058823528,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.013431372549019606,
      "loss": 2.0576,
      "step": 685
    },
    {
      "epoch": 0.016141176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.013450980392156864,
      "loss": 1.9445,
      "step": 686
    },
    {
      "epoch": 0.016164705882352942,
      "grad_norm": 0.1953125,
      "learning_rate": 0.01347058823529412,
      "loss": 2.1343,
      "step": 687
    },
    {
      "epoch": 0.016188235294117646,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.013490196078431374,
      "loss": 2.1482,
      "step": 688
    },
    {
      "epoch": 0.016211764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.013509803921568628,
      "loss": 2.0476,
      "step": 689
    },
    {
      "epoch": 0.01623529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.013529411764705884,
      "loss": 2.0645,
      "step": 690
    },
    {
      "epoch": 0.016258823529411763,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.013549019607843138,
      "loss": 2.1038,
      "step": 691
    },
    {
      "epoch": 0.01628235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.013568627450980393,
      "loss": 1.9731,
      "step": 692
    },
    {
      "epoch": 0.016305882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.013588235294117647,
      "loss": 2.1135,
      "step": 693
    },
    {
      "epoch": 0.016329411764705884,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.013607843137254903,
      "loss": 2.0914,
      "step": 694
    },
    {
      "epoch": 0.016352941176470587,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.013627450980392157,
      "loss": 2.0653,
      "step": 695
    },
    {
      "epoch": 0.016376470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.013647058823529413,
      "loss": 2.013,
      "step": 696
    },
    {
      "epoch": 0.0164,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.013666666666666667,
      "loss": 2.0461,
      "step": 697
    },
    {
      "epoch": 0.016423529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.013686274509803921,
      "loss": 2.0954,
      "step": 698
    },
    {
      "epoch": 0.016447058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.013705882352941177,
      "loss": 1.8668,
      "step": 699
    },
    {
      "epoch": 0.01647058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.013725490196078431,
      "loss": 1.9492,
      "step": 700
    },
    {
      "epoch": 0.016494117647058822,
      "grad_norm": 0.2109375,
      "learning_rate": 0.013745098039215687,
      "loss": 1.9823,
      "step": 701
    },
    {
      "epoch": 0.01651764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.013764705882352941,
      "loss": 1.9439,
      "step": 702
    },
    {
      "epoch": 0.016541176470588236,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.013784313725490197,
      "loss": 2.0239,
      "step": 703
    },
    {
      "epoch": 0.01656470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.013803921568627451,
      "loss": 1.982,
      "step": 704
    },
    {
      "epoch": 0.016588235294117647,
      "grad_norm": 0.234375,
      "learning_rate": 0.013823529411764707,
      "loss": 2.0539,
      "step": 705
    },
    {
      "epoch": 0.016611764705882354,
      "grad_norm": 0.224609375,
      "learning_rate": 0.01384313725490196,
      "loss": 2.037,
      "step": 706
    },
    {
      "epoch": 0.016635294117647057,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.013862745098039215,
      "loss": 2.0561,
      "step": 707
    },
    {
      "epoch": 0.016658823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01388235294117647,
      "loss": 2.0332,
      "step": 708
    },
    {
      "epoch": 0.01668235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.013901960784313725,
      "loss": 2.1291,
      "step": 709
    },
    {
      "epoch": 0.016705882352941178,
      "grad_norm": 0.21484375,
      "learning_rate": 0.01392156862745098,
      "loss": 2.2497,
      "step": 710
    },
    {
      "epoch": 0.01672941176470588,
      "grad_norm": 0.23046875,
      "learning_rate": 0.013941176470588235,
      "loss": 1.9734,
      "step": 711
    },
    {
      "epoch": 0.01675294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.01396078431372549,
      "loss": 2.1375,
      "step": 712
    },
    {
      "epoch": 0.016776470588235295,
      "grad_norm": 0.228515625,
      "learning_rate": 0.013980392156862744,
      "loss": 2.0593,
      "step": 713
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.203125,
      "learning_rate": 0.013999999999999999,
      "loss": 2.0199,
      "step": 714
    },
    {
      "epoch": 0.016823529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.014019607843137254,
      "loss": 2.1901,
      "step": 715
    },
    {
      "epoch": 0.016847058823529413,
      "grad_norm": 0.21875,
      "learning_rate": 0.014039215686274508,
      "loss": 2.1708,
      "step": 716
    },
    {
      "epoch": 0.016870588235294116,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.014058823529411764,
      "loss": 1.9762,
      "step": 717
    },
    {
      "epoch": 0.016894117647058823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.014078431372549022,
      "loss": 1.9248,
      "step": 718
    },
    {
      "epoch": 0.01691764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.014098039215686276,
      "loss": 2.0737,
      "step": 719
    },
    {
      "epoch": 0.016941176470588234,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.01411764705882353,
      "loss": 1.925,
      "step": 720
    },
    {
      "epoch": 0.01696470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.014137254901960786,
      "loss": 1.8835,
      "step": 721
    },
    {
      "epoch": 0.016988235294117648,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.01415686274509804,
      "loss": 2.0038,
      "step": 722
    },
    {
      "epoch": 0.01701176470588235,
      "grad_norm": 0.193359375,
      "learning_rate": 0.014176470588235296,
      "loss": 1.9385,
      "step": 723
    },
    {
      "epoch": 0.017035294117647058,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.01419607843137255,
      "loss": 1.9318,
      "step": 724
    },
    {
      "epoch": 0.017058823529411765,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.014215686274509805,
      "loss": 2.0836,
      "step": 725
    },
    {
      "epoch": 0.017082352941176472,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01423529411764706,
      "loss": 2.0464,
      "step": 726
    },
    {
      "epoch": 0.017105882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.014254901960784314,
      "loss": 2.0533,
      "step": 727
    },
    {
      "epoch": 0.017129411764705883,
      "grad_norm": 0.16796875,
      "learning_rate": 0.01427450980392157,
      "loss": 2.1183,
      "step": 728
    },
    {
      "epoch": 0.01715294117647059,
      "grad_norm": 0.1796875,
      "learning_rate": 0.014294117647058823,
      "loss": 1.9216,
      "step": 729
    },
    {
      "epoch": 0.017176470588235293,
      "grad_norm": 0.17578125,
      "learning_rate": 0.01431372549019608,
      "loss": 1.9485,
      "step": 730
    },
    {
      "epoch": 0.0172,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.014333333333333333,
      "loss": 2.0948,
      "step": 731
    },
    {
      "epoch": 0.017223529411764707,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01435294117647059,
      "loss": 2.0591,
      "step": 732
    },
    {
      "epoch": 0.01724705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.014372549019607843,
      "loss": 2.1161,
      "step": 733
    },
    {
      "epoch": 0.017270588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.014392156862745099,
      "loss": 1.9023,
      "step": 734
    },
    {
      "epoch": 0.017294117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.014411764705882353,
      "loss": 1.9241,
      "step": 735
    },
    {
      "epoch": 0.017317647058823528,
      "grad_norm": 0.189453125,
      "learning_rate": 0.014431372549019607,
      "loss": 2.2231,
      "step": 736
    },
    {
      "epoch": 0.017341176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.014450980392156863,
      "loss": 2.0856,
      "step": 737
    },
    {
      "epoch": 0.017364705882352942,
      "grad_norm": 0.17578125,
      "learning_rate": 0.014470588235294117,
      "loss": 1.9643,
      "step": 738
    },
    {
      "epoch": 0.01738823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.014490196078431373,
      "loss": 1.9748,
      "step": 739
    },
    {
      "epoch": 0.017411764705882352,
      "grad_norm": 0.201171875,
      "learning_rate": 0.014509803921568627,
      "loss": 1.9859,
      "step": 740
    },
    {
      "epoch": 0.01743529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.014529411764705883,
      "loss": 2.1208,
      "step": 741
    },
    {
      "epoch": 0.017458823529411766,
      "grad_norm": 0.1875,
      "learning_rate": 0.014549019607843137,
      "loss": 2.0727,
      "step": 742
    },
    {
      "epoch": 0.01748235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.014568627450980393,
      "loss": 1.9939,
      "step": 743
    },
    {
      "epoch": 0.017505882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.014588235294117647,
      "loss": 2.036,
      "step": 744
    },
    {
      "epoch": 0.017529411764705884,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0146078431372549,
      "loss": 1.9722,
      "step": 745
    },
    {
      "epoch": 0.017552941176470587,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.014627450980392157,
      "loss": 2.0815,
      "step": 746
    },
    {
      "epoch": 0.017576470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.01464705882352941,
      "loss": 2.1432,
      "step": 747
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.014666666666666666,
      "loss": 2.0126,
      "step": 748
    },
    {
      "epoch": 0.017623529411764705,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.01468627450980392,
      "loss": 1.8929,
      "step": 749
    },
    {
      "epoch": 0.01764705882352941,
      "grad_norm": 0.255859375,
      "learning_rate": 0.014705882352941178,
      "loss": 2.0487,
      "step": 750
    },
    {
      "epoch": 0.01767058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.014725490196078432,
      "loss": 1.9343,
      "step": 751
    },
    {
      "epoch": 0.017694117647058822,
      "grad_norm": 0.203125,
      "learning_rate": 0.014745098039215688,
      "loss": 1.8906,
      "step": 752
    },
    {
      "epoch": 0.01771764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.014764705882352942,
      "loss": 1.9338,
      "step": 753
    },
    {
      "epoch": 0.017741176470588236,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.014784313725490198,
      "loss": 1.9124,
      "step": 754
    },
    {
      "epoch": 0.017764705882352943,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.014803921568627452,
      "loss": 2.024,
      "step": 755
    },
    {
      "epoch": 0.017788235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.014823529411764708,
      "loss": 1.9727,
      "step": 756
    },
    {
      "epoch": 0.017811764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.014843137254901962,
      "loss": 2.0323,
      "step": 757
    },
    {
      "epoch": 0.01783529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.014862745098039216,
      "loss": 1.8184,
      "step": 758
    },
    {
      "epoch": 0.017858823529411764,
      "grad_norm": 0.21875,
      "learning_rate": 0.014882352941176472,
      "loss": 1.9484,
      "step": 759
    },
    {
      "epoch": 0.01788235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.014901960784313726,
      "loss": 2.0152,
      "step": 760
    },
    {
      "epoch": 0.017905882352941178,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.014921568627450981,
      "loss": 1.8354,
      "step": 761
    },
    {
      "epoch": 0.01792941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.014941176470588236,
      "loss": 2.0598,
      "step": 762
    },
    {
      "epoch": 0.017952941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.014960784313725491,
      "loss": 2.0777,
      "step": 763
    },
    {
      "epoch": 0.017976470588235295,
      "grad_norm": 0.21484375,
      "learning_rate": 0.014980392156862745,
      "loss": 2.1139,
      "step": 764
    },
    {
      "epoch": 0.018,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.015,
      "loss": 2.0121,
      "step": 765
    },
    {
      "epoch": 0.018023529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.015019607843137255,
      "loss": 2.1161,
      "step": 766
    },
    {
      "epoch": 0.018047058823529413,
      "grad_norm": 0.15625,
      "learning_rate": 0.01503921568627451,
      "loss": 2.0359,
      "step": 767
    },
    {
      "epoch": 0.018070588235294116,
      "grad_norm": 0.1875,
      "learning_rate": 0.015058823529411765,
      "loss": 2.1287,
      "step": 768
    },
    {
      "epoch": 0.018094117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.01507843137254902,
      "loss": 1.9984,
      "step": 769
    },
    {
      "epoch": 0.01811764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.015098039215686275,
      "loss": 2.1423,
      "step": 770
    },
    {
      "epoch": 0.018141176470588237,
      "grad_norm": 0.205078125,
      "learning_rate": 0.015117647058823529,
      "loss": 2.1058,
      "step": 771
    },
    {
      "epoch": 0.01816470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.015137254901960785,
      "loss": 2.1351,
      "step": 772
    },
    {
      "epoch": 0.018188235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.015156862745098039,
      "loss": 2.102,
      "step": 773
    },
    {
      "epoch": 0.018211764705882354,
      "grad_norm": 0.216796875,
      "learning_rate": 0.015176470588235293,
      "loss": 2.165,
      "step": 774
    },
    {
      "epoch": 0.018235294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.015196078431372549,
      "loss": 1.9474,
      "step": 775
    },
    {
      "epoch": 0.018258823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.015215686274509803,
      "loss": 1.8934,
      "step": 776
    },
    {
      "epoch": 0.018282352941176472,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.015235294117647059,
      "loss": 2.0634,
      "step": 777
    },
    {
      "epoch": 0.018305882352941175,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.015254901960784313,
      "loss": 2.0288,
      "step": 778
    },
    {
      "epoch": 0.018329411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.015274509803921569,
      "loss": 2.0515,
      "step": 779
    },
    {
      "epoch": 0.01835294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.015294117647058823,
      "loss": 2.0081,
      "step": 780
    },
    {
      "epoch": 0.018376470588235293,
      "grad_norm": 0.212890625,
      "learning_rate": 0.01531372549019608,
      "loss": 2.2536,
      "step": 781
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.015333333333333334,
      "loss": 1.8673,
      "step": 782
    },
    {
      "epoch": 0.018423529411764707,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.01535294117647059,
      "loss": 2.0765,
      "step": 783
    },
    {
      "epoch": 0.01844705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.015372549019607844,
      "loss": 1.8582,
      "step": 784
    },
    {
      "epoch": 0.018470588235294117,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0153921568627451,
      "loss": 2.0207,
      "step": 785
    },
    {
      "epoch": 0.018494117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.015411764705882354,
      "loss": 1.9912,
      "step": 786
    },
    {
      "epoch": 0.01851764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.015431372549019608,
      "loss": 2.1676,
      "step": 787
    },
    {
      "epoch": 0.018541176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.015450980392156864,
      "loss": 2.0695,
      "step": 788
    },
    {
      "epoch": 0.01856470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.015470588235294118,
      "loss": 1.9578,
      "step": 789
    },
    {
      "epoch": 0.01858823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.015490196078431374,
      "loss": 1.9133,
      "step": 790
    },
    {
      "epoch": 0.018611764705882352,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.015509803921568628,
      "loss": 2.139,
      "step": 791
    },
    {
      "epoch": 0.01863529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.015529411764705884,
      "loss": 1.9071,
      "step": 792
    },
    {
      "epoch": 0.018658823529411766,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.015549019607843138,
      "loss": 1.9151,
      "step": 793
    },
    {
      "epoch": 0.01868235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.015568627450980393,
      "loss": 1.9431,
      "step": 794
    },
    {
      "epoch": 0.018705882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.015588235294117648,
      "loss": 1.9113,
      "step": 795
    },
    {
      "epoch": 0.018729411764705883,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.015607843137254902,
      "loss": 1.9021,
      "step": 796
    },
    {
      "epoch": 0.018752941176470587,
      "grad_norm": 0.185546875,
      "learning_rate": 0.015627450980392157,
      "loss": 2.1068,
      "step": 797
    },
    {
      "epoch": 0.018776470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.015647058823529413,
      "loss": 1.9939,
      "step": 798
    },
    {
      "epoch": 0.0188,
      "grad_norm": 0.193359375,
      "learning_rate": 0.015666666666666666,
      "loss": 2.0982,
      "step": 799
    },
    {
      "epoch": 0.018823529411764704,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01568627450980392,
      "loss": 2.0035,
      "step": 800
    },
    {
      "epoch": 0.01884705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.015705882352941177,
      "loss": 1.9293,
      "step": 801
    },
    {
      "epoch": 0.018870588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.015725490196078433,
      "loss": 1.9186,
      "step": 802
    },
    {
      "epoch": 0.018894117647058825,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.015745098039215685,
      "loss": 1.9852,
      "step": 803
    },
    {
      "epoch": 0.01891764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.01576470588235294,
      "loss": 1.9696,
      "step": 804
    },
    {
      "epoch": 0.018941176470588236,
      "grad_norm": 0.1796875,
      "learning_rate": 0.015784313725490197,
      "loss": 2.0689,
      "step": 805
    },
    {
      "epoch": 0.018964705882352943,
      "grad_norm": 0.1953125,
      "learning_rate": 0.01580392156862745,
      "loss": 1.9725,
      "step": 806
    },
    {
      "epoch": 0.018988235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.015823529411764705,
      "loss": 1.8992,
      "step": 807
    },
    {
      "epoch": 0.019011764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.01584313725490196,
      "loss": 2.0191,
      "step": 808
    },
    {
      "epoch": 0.01903529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.015862745098039217,
      "loss": 2.0215,
      "step": 809
    },
    {
      "epoch": 0.019058823529411763,
      "grad_norm": 0.21875,
      "learning_rate": 0.01588235294117647,
      "loss": 2.0093,
      "step": 810
    },
    {
      "epoch": 0.01908235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.015901960784313725,
      "loss": 2.0319,
      "step": 811
    },
    {
      "epoch": 0.019105882352941177,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.01592156862745098,
      "loss": 1.8627,
      "step": 812
    },
    {
      "epoch": 0.01912941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.015941176470588236,
      "loss": 1.9777,
      "step": 813
    },
    {
      "epoch": 0.019152941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.015960784313725492,
      "loss": 1.8873,
      "step": 814
    },
    {
      "epoch": 0.019176470588235295,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.015980392156862748,
      "loss": 2.1264,
      "step": 815
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.1875,
      "learning_rate": 0.016,
      "loss": 1.826,
      "step": 816
    },
    {
      "epoch": 0.019223529411764705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.016019607843137256,
      "loss": 2.0196,
      "step": 817
    },
    {
      "epoch": 0.019247058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.016039215686274512,
      "loss": 2.0454,
      "step": 818
    },
    {
      "epoch": 0.01927058823529412,
      "grad_norm": 0.201171875,
      "learning_rate": 0.016058823529411764,
      "loss": 1.9986,
      "step": 819
    },
    {
      "epoch": 0.019294117647058823,
      "grad_norm": 0.21875,
      "learning_rate": 0.01607843137254902,
      "loss": 2.1162,
      "step": 820
    },
    {
      "epoch": 0.01931764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.016098039215686276,
      "loss": 2.0113,
      "step": 821
    },
    {
      "epoch": 0.019341176470588237,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.01611764705882353,
      "loss": 1.9895,
      "step": 822
    },
    {
      "epoch": 0.01936470588235294,
      "grad_norm": 0.244140625,
      "learning_rate": 0.016137254901960784,
      "loss": 1.94,
      "step": 823
    },
    {
      "epoch": 0.019388235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.01615686274509804,
      "loss": 2.1669,
      "step": 824
    },
    {
      "epoch": 0.019411764705882354,
      "grad_norm": 0.201171875,
      "learning_rate": 0.016176470588235296,
      "loss": 2.0231,
      "step": 825
    },
    {
      "epoch": 0.019435294117647058,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.016196078431372548,
      "loss": 1.9223,
      "step": 826
    },
    {
      "epoch": 0.019458823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.016215686274509804,
      "loss": 1.9286,
      "step": 827
    },
    {
      "epoch": 0.01948235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.01623529411764706,
      "loss": 2.079,
      "step": 828
    },
    {
      "epoch": 0.019505882352941175,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.016254901960784315,
      "loss": 2.05,
      "step": 829
    },
    {
      "epoch": 0.019529411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.016274509803921568,
      "loss": 2.0234,
      "step": 830
    },
    {
      "epoch": 0.01955294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.016294117647058824,
      "loss": 1.8585,
      "step": 831
    },
    {
      "epoch": 0.019576470588235292,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.01631372549019608,
      "loss": 2.0067,
      "step": 832
    },
    {
      "epoch": 0.0196,
      "grad_norm": 0.203125,
      "learning_rate": 0.01633333333333333,
      "loss": 2.1679,
      "step": 833
    },
    {
      "epoch": 0.019623529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.016352941176470587,
      "loss": 1.9871,
      "step": 834
    },
    {
      "epoch": 0.019647058823529413,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.016372549019607843,
      "loss": 2.1213,
      "step": 835
    },
    {
      "epoch": 0.019670588235294117,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0163921568627451,
      "loss": 2.056,
      "step": 836
    },
    {
      "epoch": 0.019694117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.01641176470588235,
      "loss": 1.9997,
      "step": 837
    },
    {
      "epoch": 0.01971764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.016431372549019607,
      "loss": 1.9433,
      "step": 838
    },
    {
      "epoch": 0.019741176470588234,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.016450980392156863,
      "loss": 1.7692,
      "step": 839
    },
    {
      "epoch": 0.01976470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.01647058823529412,
      "loss": 2.013,
      "step": 840
    },
    {
      "epoch": 0.019788235294117648,
      "grad_norm": 0.19140625,
      "learning_rate": 0.01649019607843137,
      "loss": 2.0949,
      "step": 841
    },
    {
      "epoch": 0.01981176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.016509803921568627,
      "loss": 1.9545,
      "step": 842
    },
    {
      "epoch": 0.01983529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.016529411764705883,
      "loss": 2.0266,
      "step": 843
    },
    {
      "epoch": 0.019858823529411766,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.016549019607843135,
      "loss": 1.8543,
      "step": 844
    },
    {
      "epoch": 0.01988235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.016568627450980394,
      "loss": 1.9835,
      "step": 845
    },
    {
      "epoch": 0.019905882352941176,
      "grad_norm": 0.185546875,
      "learning_rate": 0.016588235294117647,
      "loss": 2.0109,
      "step": 846
    },
    {
      "epoch": 0.019929411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.016607843137254902,
      "loss": 2.1187,
      "step": 847
    },
    {
      "epoch": 0.01995294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.01662745098039216,
      "loss": 2.0659,
      "step": 848
    },
    {
      "epoch": 0.019976470588235293,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.016647058823529414,
      "loss": 1.9427,
      "step": 849
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.016666666666666666,
      "loss": 1.9498,
      "step": 850
    },
    {
      "epoch": 0.020023529411764707,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.016686274509803922,
      "loss": 2.0393,
      "step": 851
    },
    {
      "epoch": 0.02004705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.016705882352941178,
      "loss": 2.0923,
      "step": 852
    },
    {
      "epoch": 0.020070588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.016725490196078434,
      "loss": 1.9783,
      "step": 853
    },
    {
      "epoch": 0.020094117647058825,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.016745098039215686,
      "loss": 1.7334,
      "step": 854
    },
    {
      "epoch": 0.02011764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.016764705882352942,
      "loss": 2.0357,
      "step": 855
    },
    {
      "epoch": 0.020141176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.016784313725490198,
      "loss": 2.0396,
      "step": 856
    },
    {
      "epoch": 0.020164705882352942,
      "grad_norm": 0.2109375,
      "learning_rate": 0.01680392156862745,
      "loss": 1.9563,
      "step": 857
    },
    {
      "epoch": 0.020188235294117646,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.016823529411764706,
      "loss": 2.1334,
      "step": 858
    },
    {
      "epoch": 0.020211764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.01684313725490196,
      "loss": 2.0088,
      "step": 859
    },
    {
      "epoch": 0.02023529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.016862745098039218,
      "loss": 1.9871,
      "step": 860
    },
    {
      "epoch": 0.020258823529411763,
      "grad_norm": 0.208984375,
      "learning_rate": 0.01688235294117647,
      "loss": 2.1158,
      "step": 861
    },
    {
      "epoch": 0.02028235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.016901960784313726,
      "loss": 2.0962,
      "step": 862
    },
    {
      "epoch": 0.020305882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.01692156862745098,
      "loss": 1.9923,
      "step": 863
    },
    {
      "epoch": 0.020329411764705884,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.016941176470588234,
      "loss": 1.8632,
      "step": 864
    },
    {
      "epoch": 0.020352941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.01696078431372549,
      "loss": 2.0502,
      "step": 865
    },
    {
      "epoch": 0.020376470588235295,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.016980392156862745,
      "loss": 2.0106,
      "step": 866
    },
    {
      "epoch": 0.0204,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.017,
      "loss": 1.8715,
      "step": 867
    },
    {
      "epoch": 0.020423529411764705,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.017019607843137254,
      "loss": 1.8703,
      "step": 868
    },
    {
      "epoch": 0.020447058823529412,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.01703921568627451,
      "loss": 1.9605,
      "step": 869
    },
    {
      "epoch": 0.02047058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.017058823529411765,
      "loss": 2.0035,
      "step": 870
    },
    {
      "epoch": 0.020494117647058822,
      "grad_norm": 0.20703125,
      "learning_rate": 0.01707843137254902,
      "loss": 1.9216,
      "step": 871
    },
    {
      "epoch": 0.02051764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.017098039215686273,
      "loss": 2.0194,
      "step": 872
    },
    {
      "epoch": 0.020541176470588236,
      "grad_norm": 0.173828125,
      "learning_rate": 0.01711764705882353,
      "loss": 1.8921,
      "step": 873
    },
    {
      "epoch": 0.02056470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.017137254901960785,
      "loss": 1.8747,
      "step": 874
    },
    {
      "epoch": 0.020588235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.017156862745098037,
      "loss": 2.1082,
      "step": 875
    },
    {
      "epoch": 0.020611764705882354,
      "grad_norm": 0.240234375,
      "learning_rate": 0.017176470588235293,
      "loss": 2.1249,
      "step": 876
    },
    {
      "epoch": 0.020635294117647057,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.01719607843137255,
      "loss": 2.0714,
      "step": 877
    },
    {
      "epoch": 0.020658823529411764,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.017215686274509805,
      "loss": 2.0462,
      "step": 878
    },
    {
      "epoch": 0.02068235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.01723529411764706,
      "loss": 2.0486,
      "step": 879
    },
    {
      "epoch": 0.020705882352941178,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.017254901960784316,
      "loss": 2.0468,
      "step": 880
    },
    {
      "epoch": 0.02072941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.01727450980392157,
      "loss": 1.8853,
      "step": 881
    },
    {
      "epoch": 0.02075294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.017294117647058824,
      "loss": 1.9967,
      "step": 882
    },
    {
      "epoch": 0.020776470588235296,
      "grad_norm": 0.212890625,
      "learning_rate": 0.01731372549019608,
      "loss": 2.0732,
      "step": 883
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.185546875,
      "learning_rate": 0.017333333333333336,
      "loss": 1.9702,
      "step": 884
    },
    {
      "epoch": 0.020823529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.01735294117647059,
      "loss": 2.0623,
      "step": 885
    },
    {
      "epoch": 0.020847058823529413,
      "grad_norm": 0.208984375,
      "learning_rate": 0.017372549019607844,
      "loss": 2.1191,
      "step": 886
    },
    {
      "epoch": 0.020870588235294116,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0173921568627451,
      "loss": 1.9991,
      "step": 887
    },
    {
      "epoch": 0.020894117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.017411764705882352,
      "loss": 2.0862,
      "step": 888
    },
    {
      "epoch": 0.02091764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.017431372549019608,
      "loss": 1.906,
      "step": 889
    },
    {
      "epoch": 0.020941176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.017450980392156864,
      "loss": 1.8141,
      "step": 890
    },
    {
      "epoch": 0.02096470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.01747058823529412,
      "loss": 1.9398,
      "step": 891
    },
    {
      "epoch": 0.020988235294117648,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.017490196078431372,
      "loss": 1.8402,
      "step": 892
    },
    {
      "epoch": 0.02101176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.017509803921568628,
      "loss": 2.1111,
      "step": 893
    },
    {
      "epoch": 0.02103529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.017529411764705884,
      "loss": 1.9251,
      "step": 894
    },
    {
      "epoch": 0.021058823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.017549019607843136,
      "loss": 2.0184,
      "step": 895
    },
    {
      "epoch": 0.021082352941176472,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.017568627450980392,
      "loss": 2.0534,
      "step": 896
    },
    {
      "epoch": 0.021105882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.017588235294117648,
      "loss": 1.8513,
      "step": 897
    },
    {
      "epoch": 0.021129411764705883,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.017607843137254903,
      "loss": 1.9517,
      "step": 898
    },
    {
      "epoch": 0.02115294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.017627450980392156,
      "loss": 2.099,
      "step": 899
    },
    {
      "epoch": 0.021176470588235293,
      "grad_norm": 0.23046875,
      "learning_rate": 0.01764705882352941,
      "loss": 1.9641,
      "step": 900
    },
    {
      "epoch": 0.0212,
      "grad_norm": 0.1875,
      "learning_rate": 0.017666666666666667,
      "loss": 2.0201,
      "step": 901
    },
    {
      "epoch": 0.021223529411764707,
      "grad_norm": 0.189453125,
      "learning_rate": 0.01768627450980392,
      "loss": 1.9719,
      "step": 902
    },
    {
      "epoch": 0.02124705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.017705882352941175,
      "loss": 1.9524,
      "step": 903
    },
    {
      "epoch": 0.021270588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.01772549019607843,
      "loss": 2.0481,
      "step": 904
    },
    {
      "epoch": 0.021294117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.017745098039215687,
      "loss": 2.0472,
      "step": 905
    },
    {
      "epoch": 0.021317647058823528,
      "grad_norm": 0.197265625,
      "learning_rate": 0.01776470588235294,
      "loss": 1.8897,
      "step": 906
    },
    {
      "epoch": 0.021341176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.017784313725490195,
      "loss": 2.0546,
      "step": 907
    },
    {
      "epoch": 0.021364705882352942,
      "grad_norm": 0.2109375,
      "learning_rate": 0.01780392156862745,
      "loss": 1.8933,
      "step": 908
    },
    {
      "epoch": 0.021388235294117645,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.017823529411764707,
      "loss": 1.9716,
      "step": 909
    },
    {
      "epoch": 0.021411764705882352,
      "grad_norm": 0.212890625,
      "learning_rate": 0.017843137254901963,
      "loss": 1.8304,
      "step": 910
    },
    {
      "epoch": 0.02143529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.01786274509803922,
      "loss": 2.0366,
      "step": 911
    },
    {
      "epoch": 0.021458823529411766,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01788235294117647,
      "loss": 2.0673,
      "step": 912
    },
    {
      "epoch": 0.02148235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.017901960784313727,
      "loss": 1.9314,
      "step": 913
    },
    {
      "epoch": 0.021505882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.017921568627450982,
      "loss": 1.9498,
      "step": 914
    },
    {
      "epoch": 0.021529411764705884,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.017941176470588235,
      "loss": 2.0704,
      "step": 915
    },
    {
      "epoch": 0.021552941176470587,
      "grad_norm": 0.197265625,
      "learning_rate": 0.01796078431372549,
      "loss": 1.84,
      "step": 916
    },
    {
      "epoch": 0.021576470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.017980392156862746,
      "loss": 1.902,
      "step": 917
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.19140625,
      "learning_rate": 0.018000000000000002,
      "loss": 1.9741,
      "step": 918
    },
    {
      "epoch": 0.021623529411764705,
      "grad_norm": 0.20703125,
      "learning_rate": 0.018019607843137254,
      "loss": 2.0209,
      "step": 919
    },
    {
      "epoch": 0.02164705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.01803921568627451,
      "loss": 1.9545,
      "step": 920
    },
    {
      "epoch": 0.02167058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.018058823529411766,
      "loss": 1.9845,
      "step": 921
    },
    {
      "epoch": 0.021694117647058822,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.018078431372549022,
      "loss": 1.9355,
      "step": 922
    },
    {
      "epoch": 0.02171764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.018098039215686274,
      "loss": 1.8894,
      "step": 923
    },
    {
      "epoch": 0.021741176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.01811764705882353,
      "loss": 1.8536,
      "step": 924
    },
    {
      "epoch": 0.02176470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.018137254901960786,
      "loss": 1.9452,
      "step": 925
    },
    {
      "epoch": 0.021788235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.018156862745098038,
      "loss": 1.9626,
      "step": 926
    },
    {
      "epoch": 0.021811764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.018176470588235294,
      "loss": 1.9735,
      "step": 927
    },
    {
      "epoch": 0.02183529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.01819607843137255,
      "loss": 1.9846,
      "step": 928
    },
    {
      "epoch": 0.021858823529411764,
      "grad_norm": 0.224609375,
      "learning_rate": 0.018215686274509806,
      "loss": 1.9758,
      "step": 929
    },
    {
      "epoch": 0.02188235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.018235294117647058,
      "loss": 1.7638,
      "step": 930
    },
    {
      "epoch": 0.021905882352941178,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.018254901960784314,
      "loss": 2.0394,
      "step": 931
    },
    {
      "epoch": 0.02192941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.01827450980392157,
      "loss": 1.9298,
      "step": 932
    },
    {
      "epoch": 0.021952941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.018294117647058822,
      "loss": 1.9837,
      "step": 933
    },
    {
      "epoch": 0.021976470588235295,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.018313725490196078,
      "loss": 2.0315,
      "step": 934
    },
    {
      "epoch": 0.022,
      "grad_norm": 0.22265625,
      "learning_rate": 0.018333333333333333,
      "loss": 2.0284,
      "step": 935
    },
    {
      "epoch": 0.022023529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.01835294117647059,
      "loss": 1.8585,
      "step": 936
    },
    {
      "epoch": 0.022047058823529413,
      "grad_norm": 0.197265625,
      "learning_rate": 0.01837254901960784,
      "loss": 2.0577,
      "step": 937
    },
    {
      "epoch": 0.022070588235294116,
      "grad_norm": 0.201171875,
      "learning_rate": 0.018392156862745097,
      "loss": 1.8081,
      "step": 938
    },
    {
      "epoch": 0.022094117647058823,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.018411764705882353,
      "loss": 2.0614,
      "step": 939
    },
    {
      "epoch": 0.02211764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.018431372549019606,
      "loss": 2.0538,
      "step": 940
    },
    {
      "epoch": 0.022141176470588234,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.018450980392156865,
      "loss": 1.9271,
      "step": 941
    },
    {
      "epoch": 0.02216470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.01847058823529412,
      "loss": 2.002,
      "step": 942
    },
    {
      "epoch": 0.022188235294117648,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.018490196078431373,
      "loss": 1.9416,
      "step": 943
    },
    {
      "epoch": 0.022211764705882354,
      "grad_norm": 0.193359375,
      "learning_rate": 0.01850980392156863,
      "loss": 1.8982,
      "step": 944
    },
    {
      "epoch": 0.022235294117647058,
      "grad_norm": 0.20703125,
      "learning_rate": 0.018529411764705885,
      "loss": 1.8526,
      "step": 945
    },
    {
      "epoch": 0.022258823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.018549019607843137,
      "loss": 2.0232,
      "step": 946
    },
    {
      "epoch": 0.022282352941176472,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.018568627450980393,
      "loss": 1.9455,
      "step": 947
    },
    {
      "epoch": 0.022305882352941175,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.01858823529411765,
      "loss": 1.9882,
      "step": 948
    },
    {
      "epoch": 0.022329411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.018607843137254904,
      "loss": 1.8928,
      "step": 949
    },
    {
      "epoch": 0.02235294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.018627450980392157,
      "loss": 1.8783,
      "step": 950
    },
    {
      "epoch": 0.022376470588235293,
      "grad_norm": 0.21484375,
      "learning_rate": 0.018647058823529412,
      "loss": 1.9617,
      "step": 951
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.224609375,
      "learning_rate": 0.018666666666666668,
      "loss": 2.0043,
      "step": 952
    },
    {
      "epoch": 0.022423529411764707,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.01868627450980392,
      "loss": 2.0664,
      "step": 953
    },
    {
      "epoch": 0.02244705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.018705882352941176,
      "loss": 2.0095,
      "step": 954
    },
    {
      "epoch": 0.022470588235294117,
      "grad_norm": 0.220703125,
      "learning_rate": 0.018725490196078432,
      "loss": 2.1061,
      "step": 955
    },
    {
      "epoch": 0.022494117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.018745098039215688,
      "loss": 1.8346,
      "step": 956
    },
    {
      "epoch": 0.02251764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.01876470588235294,
      "loss": 1.9552,
      "step": 957
    },
    {
      "epoch": 0.022541176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.018784313725490196,
      "loss": 1.8551,
      "step": 958
    },
    {
      "epoch": 0.02256470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.018803921568627452,
      "loss": 1.8887,
      "step": 959
    },
    {
      "epoch": 0.02258823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.018823529411764708,
      "loss": 2.1027,
      "step": 960
    },
    {
      "epoch": 0.022611764705882352,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.01884313725490196,
      "loss": 2.0738,
      "step": 961
    },
    {
      "epoch": 0.02263529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.018862745098039216,
      "loss": 1.9634,
      "step": 962
    },
    {
      "epoch": 0.022658823529411766,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.01888235294117647,
      "loss": 1.9743,
      "step": 963
    },
    {
      "epoch": 0.02268235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.018901960784313724,
      "loss": 1.8804,
      "step": 964
    },
    {
      "epoch": 0.022705882352941176,
      "grad_norm": 0.212890625,
      "learning_rate": 0.01892156862745098,
      "loss": 2.1187,
      "step": 965
    },
    {
      "epoch": 0.022729411764705883,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.018941176470588236,
      "loss": 1.9367,
      "step": 966
    },
    {
      "epoch": 0.022752941176470587,
      "grad_norm": 0.22265625,
      "learning_rate": 0.01896078431372549,
      "loss": 2.1467,
      "step": 967
    },
    {
      "epoch": 0.022776470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.018980392156862744,
      "loss": 1.9395,
      "step": 968
    },
    {
      "epoch": 0.0228,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.019,
      "loss": 1.8324,
      "step": 969
    },
    {
      "epoch": 0.022823529411764704,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.019019607843137255,
      "loss": 1.8566,
      "step": 970
    },
    {
      "epoch": 0.02284705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.019039215686274508,
      "loss": 1.9015,
      "step": 971
    },
    {
      "epoch": 0.022870588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.019058823529411763,
      "loss": 1.9128,
      "step": 972
    },
    {
      "epoch": 0.022894117647058825,
      "grad_norm": 0.22265625,
      "learning_rate": 0.019078431372549023,
      "loss": 1.9646,
      "step": 973
    },
    {
      "epoch": 0.02291764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.019098039215686275,
      "loss": 2.0401,
      "step": 974
    },
    {
      "epoch": 0.022941176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.01911764705882353,
      "loss": 1.9694,
      "step": 975
    },
    {
      "epoch": 0.022964705882352943,
      "grad_norm": 0.18359375,
      "learning_rate": 0.019137254901960787,
      "loss": 2.0224,
      "step": 976
    },
    {
      "epoch": 0.022988235294117646,
      "grad_norm": 0.21484375,
      "learning_rate": 0.01915686274509804,
      "loss": 2.0175,
      "step": 977
    },
    {
      "epoch": 0.023011764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.019176470588235295,
      "loss": 2.0317,
      "step": 978
    },
    {
      "epoch": 0.02303529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.01919607843137255,
      "loss": 1.8926,
      "step": 979
    },
    {
      "epoch": 0.023058823529411764,
      "grad_norm": 0.18359375,
      "learning_rate": 0.019215686274509806,
      "loss": 2.0854,
      "step": 980
    },
    {
      "epoch": 0.02308235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.01923529411764706,
      "loss": 1.9954,
      "step": 981
    },
    {
      "epoch": 0.023105882352941177,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.019254901960784315,
      "loss": 1.8605,
      "step": 982
    },
    {
      "epoch": 0.02312941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.01927450980392157,
      "loss": 1.8874,
      "step": 983
    },
    {
      "epoch": 0.023152941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.019294117647058823,
      "loss": 1.9408,
      "step": 984
    },
    {
      "epoch": 0.023176470588235295,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.01931372549019608,
      "loss": 2.1186,
      "step": 985
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.216796875,
      "learning_rate": 0.019333333333333334,
      "loss": 1.8886,
      "step": 986
    },
    {
      "epoch": 0.023223529411764705,
      "grad_norm": 0.216796875,
      "learning_rate": 0.01935294117647059,
      "loss": 1.9675,
      "step": 987
    },
    {
      "epoch": 0.023247058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.019372549019607842,
      "loss": 2.0989,
      "step": 988
    },
    {
      "epoch": 0.02327058823529412,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.019392156862745098,
      "loss": 1.9019,
      "step": 989
    },
    {
      "epoch": 0.023294117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.019411764705882354,
      "loss": 2.0025,
      "step": 990
    },
    {
      "epoch": 0.02331764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.019431372549019606,
      "loss": 2.0342,
      "step": 991
    },
    {
      "epoch": 0.023341176470588237,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.019450980392156862,
      "loss": 1.9791,
      "step": 992
    },
    {
      "epoch": 0.02336470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.019470588235294118,
      "loss": 1.9265,
      "step": 993
    },
    {
      "epoch": 0.023388235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.019490196078431374,
      "loss": 2.108,
      "step": 994
    },
    {
      "epoch": 0.023411764705882354,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.019509803921568626,
      "loss": 1.7812,
      "step": 995
    },
    {
      "epoch": 0.023435294117647058,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.019529411764705882,
      "loss": 2.0016,
      "step": 996
    },
    {
      "epoch": 0.023458823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.019549019607843138,
      "loss": 1.9207,
      "step": 997
    },
    {
      "epoch": 0.02348235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.019568627450980394,
      "loss": 1.9535,
      "step": 998
    },
    {
      "epoch": 0.023505882352941175,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.019588235294117646,
      "loss": 1.9885,
      "step": 999
    },
    {
      "epoch": 0.023529411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0196078431372549,
      "loss": 1.9422,
      "step": 1000
    },
    {
      "epoch": 0.023529411764705882,
      "eval_loss": 2.3058552742004395,
      "eval_runtime": 572.1549,
      "eval_samples_per_second": 14.856,
      "eval_steps_per_second": 3.714,
      "step": 1000
    },
    {
      "epoch": 0.02355294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.019627450980392157,
      "loss": 1.857,
      "step": 1001
    },
    {
      "epoch": 0.023576470588235292,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.01964705882352941,
      "loss": 1.9268,
      "step": 1002
    },
    {
      "epoch": 0.0236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.019666666666666666,
      "loss": 1.9579,
      "step": 1003
    },
    {
      "epoch": 0.023623529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.01968627450980392,
      "loss": 1.8477,
      "step": 1004
    },
    {
      "epoch": 0.023647058823529413,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.019705882352941177,
      "loss": 2.0614,
      "step": 1005
    },
    {
      "epoch": 0.023670588235294117,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.019725490196078433,
      "loss": 2.0462,
      "step": 1006
    },
    {
      "epoch": 0.023694117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.01974509803921569,
      "loss": 1.7971,
      "step": 1007
    },
    {
      "epoch": 0.02371764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.01976470588235294,
      "loss": 2.1321,
      "step": 1008
    },
    {
      "epoch": 0.023741176470588234,
      "grad_norm": 0.201171875,
      "learning_rate": 0.019784313725490197,
      "loss": 1.9477,
      "step": 1009
    },
    {
      "epoch": 0.02376470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.019803921568627453,
      "loss": 1.9775,
      "step": 1010
    },
    {
      "epoch": 0.023788235294117648,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.01982352941176471,
      "loss": 1.9135,
      "step": 1011
    },
    {
      "epoch": 0.02381176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.01984313725490196,
      "loss": 1.9539,
      "step": 1012
    },
    {
      "epoch": 0.02383529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.019862745098039217,
      "loss": 1.9487,
      "step": 1013
    },
    {
      "epoch": 0.023858823529411766,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.019882352941176473,
      "loss": 1.8243,
      "step": 1014
    },
    {
      "epoch": 0.02388235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.019901960784313725,
      "loss": 1.9229,
      "step": 1015
    },
    {
      "epoch": 0.023905882352941176,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.01992156862745098,
      "loss": 1.9341,
      "step": 1016
    },
    {
      "epoch": 0.023929411764705883,
      "grad_norm": 0.177734375,
      "learning_rate": 0.019941176470588236,
      "loss": 2.0764,
      "step": 1017
    },
    {
      "epoch": 0.023952941176470587,
      "grad_norm": 0.21484375,
      "learning_rate": 0.019960784313725492,
      "loss": 1.7523,
      "step": 1018
    },
    {
      "epoch": 0.023976470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.019980392156862745,
      "loss": 2.1098,
      "step": 1019
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7288,
      "step": 1020
    },
    {
      "epoch": 0.024023529411764707,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.9493,
      "step": 1021
    },
    {
      "epoch": 0.02404705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.9629,
      "step": 1022
    },
    {
      "epoch": 0.024070588235294118,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.9951,
      "step": 1023
    },
    {
      "epoch": 0.024094117647058825,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.9186,
      "step": 1024
    },
    {
      "epoch": 0.02411764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.9262,
      "step": 1025
    },
    {
      "epoch": 0.024141176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.8377,
      "step": 1026
    },
    {
      "epoch": 0.024164705882352942,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.9567,
      "step": 1027
    },
    {
      "epoch": 0.024188235294117646,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.889,
      "step": 1028
    },
    {
      "epoch": 0.024211764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.8378,
      "step": 1029
    },
    {
      "epoch": 0.02423529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.8462,
      "step": 1030
    },
    {
      "epoch": 0.024258823529411763,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8657,
      "step": 1031
    },
    {
      "epoch": 0.02428235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 2.0112,
      "step": 1032
    },
    {
      "epoch": 0.024305882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 2.0628,
      "step": 1033
    },
    {
      "epoch": 0.02432941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9651,
      "step": 1034
    },
    {
      "epoch": 0.024352941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.9476,
      "step": 1035
    },
    {
      "epoch": 0.024376470588235295,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 2.0951,
      "step": 1036
    },
    {
      "epoch": 0.0244,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 2.0272,
      "step": 1037
    },
    {
      "epoch": 0.024423529411764705,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 2.0416,
      "step": 1038
    },
    {
      "epoch": 0.024447058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.9444,
      "step": 1039
    },
    {
      "epoch": 0.02447058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8359,
      "step": 1040
    },
    {
      "epoch": 0.024494117647058822,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 2.031,
      "step": 1041
    },
    {
      "epoch": 0.02451764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.9973,
      "step": 1042
    },
    {
      "epoch": 0.024541176470588236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.8839,
      "step": 1043
    },
    {
      "epoch": 0.02456470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.9332,
      "step": 1044
    },
    {
      "epoch": 0.024588235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.8953,
      "step": 1045
    },
    {
      "epoch": 0.024611764705882354,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.9026,
      "step": 1046
    },
    {
      "epoch": 0.024635294117647057,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 2.0087,
      "step": 1047
    },
    {
      "epoch": 0.024658823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.9301,
      "step": 1048
    },
    {
      "epoch": 0.02468235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.941,
      "step": 1049
    },
    {
      "epoch": 0.024705882352941175,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.9793,
      "step": 1050
    },
    {
      "epoch": 0.02472941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 2.0626,
      "step": 1051
    },
    {
      "epoch": 0.02475294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 2.0155,
      "step": 1052
    },
    {
      "epoch": 0.024776470588235296,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.9646,
      "step": 1053
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.9844,
      "step": 1054
    },
    {
      "epoch": 0.024823529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 2.016,
      "step": 1055
    },
    {
      "epoch": 0.024847058823529413,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.8959,
      "step": 1056
    },
    {
      "epoch": 0.024870588235294117,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9972,
      "step": 1057
    },
    {
      "epoch": 0.024894117647058824,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 2.0685,
      "step": 1058
    },
    {
      "epoch": 0.02491764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.979,
      "step": 1059
    },
    {
      "epoch": 0.024941176470588234,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.9986,
      "step": 1060
    },
    {
      "epoch": 0.02496470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.0373,
      "step": 1061
    },
    {
      "epoch": 0.024988235294117648,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.1247,
      "step": 1062
    },
    {
      "epoch": 0.02501176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 2.1275,
      "step": 1063
    },
    {
      "epoch": 0.02503529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.8895,
      "step": 1064
    },
    {
      "epoch": 0.025058823529411765,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 2.0197,
      "step": 1065
    },
    {
      "epoch": 0.025082352941176472,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.8136,
      "step": 1066
    },
    {
      "epoch": 0.025105882352941176,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 2.0453,
      "step": 1067
    },
    {
      "epoch": 0.025129411764705883,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9473,
      "step": 1068
    },
    {
      "epoch": 0.02515294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.8901,
      "step": 1069
    },
    {
      "epoch": 0.025176470588235293,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.9596,
      "step": 1070
    },
    {
      "epoch": 0.0252,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 2.1105,
      "step": 1071
    },
    {
      "epoch": 0.025223529411764707,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 2.0538,
      "step": 1072
    },
    {
      "epoch": 0.02524705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 2.0291,
      "step": 1073
    },
    {
      "epoch": 0.025270588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.9061,
      "step": 1074
    },
    {
      "epoch": 0.025294117647058825,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8564,
      "step": 1075
    },
    {
      "epoch": 0.025317647058823528,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.947,
      "step": 1076
    },
    {
      "epoch": 0.025341176470588235,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 2.0139,
      "step": 1077
    },
    {
      "epoch": 0.025364705882352942,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9677,
      "step": 1078
    },
    {
      "epoch": 0.025388235294117645,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8774,
      "step": 1079
    },
    {
      "epoch": 0.025411764705882352,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.857,
      "step": 1080
    },
    {
      "epoch": 0.02543529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.8704,
      "step": 1081
    },
    {
      "epoch": 0.025458823529411766,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.9408,
      "step": 1082
    },
    {
      "epoch": 0.02548235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.936,
      "step": 1083
    },
    {
      "epoch": 0.025505882352941177,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.8778,
      "step": 1084
    },
    {
      "epoch": 0.025529411764705884,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7381,
      "step": 1085
    },
    {
      "epoch": 0.025552941176470587,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.9207,
      "step": 1086
    },
    {
      "epoch": 0.025576470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.9872,
      "step": 1087
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.9436,
      "step": 1088
    },
    {
      "epoch": 0.025623529411764705,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.9544,
      "step": 1089
    },
    {
      "epoch": 0.02564705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.9332,
      "step": 1090
    },
    {
      "epoch": 0.02567058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9302,
      "step": 1091
    },
    {
      "epoch": 0.025694117647058822,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.9461,
      "step": 1092
    },
    {
      "epoch": 0.02571764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.9969,
      "step": 1093
    },
    {
      "epoch": 0.025741176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.9057,
      "step": 1094
    },
    {
      "epoch": 0.02576470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.8942,
      "step": 1095
    },
    {
      "epoch": 0.025788235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.9731,
      "step": 1096
    },
    {
      "epoch": 0.025811764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 2.0203,
      "step": 1097
    },
    {
      "epoch": 0.02583529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.8863,
      "step": 1098
    },
    {
      "epoch": 0.025858823529411764,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9604,
      "step": 1099
    },
    {
      "epoch": 0.02588235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9243,
      "step": 1100
    },
    {
      "epoch": 0.025905882352941178,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.916,
      "step": 1101
    },
    {
      "epoch": 0.02592941176470588,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 2.0526,
      "step": 1102
    },
    {
      "epoch": 0.02595294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8843,
      "step": 1103
    },
    {
      "epoch": 0.025976470588235295,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8811,
      "step": 1104
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8291,
      "step": 1105
    },
    {
      "epoch": 0.026023529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.8268,
      "step": 1106
    },
    {
      "epoch": 0.026047058823529413,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 2.0167,
      "step": 1107
    },
    {
      "epoch": 0.026070588235294116,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 2.0091,
      "step": 1108
    },
    {
      "epoch": 0.026094117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7639,
      "step": 1109
    },
    {
      "epoch": 0.02611764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.9038,
      "step": 1110
    },
    {
      "epoch": 0.026141176470588234,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.9152,
      "step": 1111
    },
    {
      "epoch": 0.02616470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.9819,
      "step": 1112
    },
    {
      "epoch": 0.026188235294117648,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.858,
      "step": 1113
    },
    {
      "epoch": 0.026211764705882355,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8218,
      "step": 1114
    },
    {
      "epoch": 0.026235294117647058,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9914,
      "step": 1115
    },
    {
      "epoch": 0.026258823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 2.011,
      "step": 1116
    },
    {
      "epoch": 0.026282352941176472,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7053,
      "step": 1117
    },
    {
      "epoch": 0.026305882352941175,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.91,
      "step": 1118
    },
    {
      "epoch": 0.026329411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 2.0429,
      "step": 1119
    },
    {
      "epoch": 0.02635294117647059,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.913,
      "step": 1120
    },
    {
      "epoch": 0.026376470588235293,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 2.056,
      "step": 1121
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8456,
      "step": 1122
    },
    {
      "epoch": 0.026423529411764707,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 2.0101,
      "step": 1123
    },
    {
      "epoch": 0.02644705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.8263,
      "step": 1124
    },
    {
      "epoch": 0.026470588235294117,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.9263,
      "step": 1125
    },
    {
      "epoch": 0.026494117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9009,
      "step": 1126
    },
    {
      "epoch": 0.026517647058823528,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9271,
      "step": 1127
    },
    {
      "epoch": 0.026541176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.832,
      "step": 1128
    },
    {
      "epoch": 0.02656470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.9479,
      "step": 1129
    },
    {
      "epoch": 0.02658823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 2.1429,
      "step": 1130
    },
    {
      "epoch": 0.026611764705882352,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.922,
      "step": 1131
    },
    {
      "epoch": 0.02663529411764706,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.9165,
      "step": 1132
    },
    {
      "epoch": 0.026658823529411766,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.9407,
      "step": 1133
    },
    {
      "epoch": 0.02668235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 2.0616,
      "step": 1134
    },
    {
      "epoch": 0.026705882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.955,
      "step": 1135
    },
    {
      "epoch": 0.026729411764705883,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9398,
      "step": 1136
    },
    {
      "epoch": 0.026752941176470587,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9259,
      "step": 1137
    },
    {
      "epoch": 0.026776470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8154,
      "step": 1138
    },
    {
      "epoch": 0.0268,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8929,
      "step": 1139
    },
    {
      "epoch": 0.026823529411764704,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 2.0221,
      "step": 1140
    },
    {
      "epoch": 0.02684705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9407,
      "step": 1141
    },
    {
      "epoch": 0.02687058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9032,
      "step": 1142
    },
    {
      "epoch": 0.026894117647058822,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.0695,
      "step": 1143
    },
    {
      "epoch": 0.02691764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.7735,
      "step": 1144
    },
    {
      "epoch": 0.026941176470588236,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.9316,
      "step": 1145
    },
    {
      "epoch": 0.026964705882352943,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8993,
      "step": 1146
    },
    {
      "epoch": 0.026988235294117646,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9745,
      "step": 1147
    },
    {
      "epoch": 0.027011764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8822,
      "step": 1148
    },
    {
      "epoch": 0.02703529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7975,
      "step": 1149
    },
    {
      "epoch": 0.027058823529411764,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.9406,
      "step": 1150
    },
    {
      "epoch": 0.02708235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8658,
      "step": 1151
    },
    {
      "epoch": 0.027105882352941178,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 2.0293,
      "step": 1152
    },
    {
      "epoch": 0.02712941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.958,
      "step": 1153
    },
    {
      "epoch": 0.027152941176470588,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.8089,
      "step": 1154
    },
    {
      "epoch": 0.027176470588235295,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.9032,
      "step": 1155
    },
    {
      "epoch": 0.0272,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9676,
      "step": 1156
    },
    {
      "epoch": 0.027223529411764705,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.9631,
      "step": 1157
    },
    {
      "epoch": 0.027247058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.954,
      "step": 1158
    },
    {
      "epoch": 0.02727058823529412,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.7312,
      "step": 1159
    },
    {
      "epoch": 0.027294117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 2.0528,
      "step": 1160
    },
    {
      "epoch": 0.02731764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8244,
      "step": 1161
    },
    {
      "epoch": 0.027341176470588237,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9033,
      "step": 1162
    },
    {
      "epoch": 0.02736470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9466,
      "step": 1163
    },
    {
      "epoch": 0.027388235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 2.0559,
      "step": 1164
    },
    {
      "epoch": 0.027411764705882354,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.9012,
      "step": 1165
    },
    {
      "epoch": 0.027435294117647058,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.8892,
      "step": 1166
    },
    {
      "epoch": 0.027458823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 2.0467,
      "step": 1167
    },
    {
      "epoch": 0.02748235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.9126,
      "step": 1168
    },
    {
      "epoch": 0.027505882352941175,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.8699,
      "step": 1169
    },
    {
      "epoch": 0.027529411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8969,
      "step": 1170
    },
    {
      "epoch": 0.02755294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 2.0053,
      "step": 1171
    },
    {
      "epoch": 0.027576470588235293,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8532,
      "step": 1172
    },
    {
      "epoch": 0.0276,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 2.0508,
      "step": 1173
    },
    {
      "epoch": 0.027623529411764706,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.9737,
      "step": 1174
    },
    {
      "epoch": 0.027647058823529413,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8795,
      "step": 1175
    },
    {
      "epoch": 0.027670588235294117,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.9353,
      "step": 1176
    },
    {
      "epoch": 0.027694117647058824,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 2.0001,
      "step": 1177
    },
    {
      "epoch": 0.02771764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8145,
      "step": 1178
    },
    {
      "epoch": 0.027741176470588234,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.093,
      "step": 1179
    },
    {
      "epoch": 0.02776470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 2.0559,
      "step": 1180
    },
    {
      "epoch": 0.02778823529411765,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 2.0231,
      "step": 1181
    },
    {
      "epoch": 0.027811764705882352,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 2.0635,
      "step": 1182
    },
    {
      "epoch": 0.02783529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8673,
      "step": 1183
    },
    {
      "epoch": 0.027858823529411766,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.983,
      "step": 1184
    },
    {
      "epoch": 0.02788235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9078,
      "step": 1185
    },
    {
      "epoch": 0.027905882352941176,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 2.0545,
      "step": 1186
    },
    {
      "epoch": 0.027929411764705883,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 2.0036,
      "step": 1187
    },
    {
      "epoch": 0.027952941176470587,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.8972,
      "step": 1188
    },
    {
      "epoch": 0.027976470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.9474,
      "step": 1189
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.9386,
      "step": 1190
    },
    {
      "epoch": 0.028023529411764708,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9375,
      "step": 1191
    },
    {
      "epoch": 0.02804705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8775,
      "step": 1192
    },
    {
      "epoch": 0.028070588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8315,
      "step": 1193
    },
    {
      "epoch": 0.028094117647058825,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8974,
      "step": 1194
    },
    {
      "epoch": 0.02811764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 2.0015,
      "step": 1195
    },
    {
      "epoch": 0.028141176470588235,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.8078,
      "step": 1196
    },
    {
      "epoch": 0.028164705882352942,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 2.0218,
      "step": 1197
    },
    {
      "epoch": 0.028188235294117646,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9564,
      "step": 1198
    },
    {
      "epoch": 0.028211764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.0103,
      "step": 1199
    },
    {
      "epoch": 0.02823529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.8691,
      "step": 1200
    },
    {
      "epoch": 0.028258823529411763,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.8882,
      "step": 1201
    },
    {
      "epoch": 0.02828235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9827,
      "step": 1202
    },
    {
      "epoch": 0.028305882352941177,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 2.0295,
      "step": 1203
    },
    {
      "epoch": 0.02832941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8286,
      "step": 1204
    },
    {
      "epoch": 0.028352941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8466,
      "step": 1205
    },
    {
      "epoch": 0.028376470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8541,
      "step": 1206
    },
    {
      "epoch": 0.0284,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.9738,
      "step": 1207
    },
    {
      "epoch": 0.028423529411764705,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8387,
      "step": 1208
    },
    {
      "epoch": 0.028447058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8891,
      "step": 1209
    },
    {
      "epoch": 0.02847058823529412,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 2.0786,
      "step": 1210
    },
    {
      "epoch": 0.028494117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8114,
      "step": 1211
    },
    {
      "epoch": 0.02851764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 2.0073,
      "step": 1212
    },
    {
      "epoch": 0.028541176470588236,
      "grad_norm": 0.310546875,
      "learning_rate": 0.02,
      "loss": 1.7755,
      "step": 1213
    },
    {
      "epoch": 0.02856470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 2.0919,
      "step": 1214
    },
    {
      "epoch": 0.028588235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.9492,
      "step": 1215
    },
    {
      "epoch": 0.028611764705882354,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8933,
      "step": 1216
    },
    {
      "epoch": 0.028635294117647057,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.9423,
      "step": 1217
    },
    {
      "epoch": 0.028658823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8912,
      "step": 1218
    },
    {
      "epoch": 0.02868235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.915,
      "step": 1219
    },
    {
      "epoch": 0.028705882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.9718,
      "step": 1220
    },
    {
      "epoch": 0.028729411764705882,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 2.0172,
      "step": 1221
    },
    {
      "epoch": 0.02875294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8054,
      "step": 1222
    },
    {
      "epoch": 0.028776470588235296,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 2.0373,
      "step": 1223
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8637,
      "step": 1224
    },
    {
      "epoch": 0.028823529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.9214,
      "step": 1225
    },
    {
      "epoch": 0.028847058823529413,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.8096,
      "step": 1226
    },
    {
      "epoch": 0.028870588235294117,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.8689,
      "step": 1227
    },
    {
      "epoch": 0.028894117647058824,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8975,
      "step": 1228
    },
    {
      "epoch": 0.02891764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8934,
      "step": 1229
    },
    {
      "epoch": 0.028941176470588234,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.773,
      "step": 1230
    },
    {
      "epoch": 0.02896470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.7809,
      "step": 1231
    },
    {
      "epoch": 0.028988235294117648,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 2.0221,
      "step": 1232
    },
    {
      "epoch": 0.02901176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8286,
      "step": 1233
    },
    {
      "epoch": 0.02903529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 2.0086,
      "step": 1234
    },
    {
      "epoch": 0.029058823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.9378,
      "step": 1235
    },
    {
      "epoch": 0.02908235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9897,
      "step": 1236
    },
    {
      "epoch": 0.029105882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7497,
      "step": 1237
    },
    {
      "epoch": 0.029129411764705883,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 2.0394,
      "step": 1238
    },
    {
      "epoch": 0.02915294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.774,
      "step": 1239
    },
    {
      "epoch": 0.029176470588235293,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.9507,
      "step": 1240
    },
    {
      "epoch": 0.0292,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.7825,
      "step": 1241
    },
    {
      "epoch": 0.029223529411764707,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.7701,
      "step": 1242
    },
    {
      "epoch": 0.02924705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8572,
      "step": 1243
    },
    {
      "epoch": 0.029270588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.652,
      "step": 1244
    },
    {
      "epoch": 0.029294117647058825,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9165,
      "step": 1245
    },
    {
      "epoch": 0.029317647058823528,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6806,
      "step": 1246
    },
    {
      "epoch": 0.029341176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.9528,
      "step": 1247
    },
    {
      "epoch": 0.029364705882352942,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.9356,
      "step": 1248
    },
    {
      "epoch": 0.029388235294117646,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8374,
      "step": 1249
    },
    {
      "epoch": 0.029411764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.998,
      "step": 1250
    },
    {
      "epoch": 0.02943529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8688,
      "step": 1251
    },
    {
      "epoch": 0.029458823529411763,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 2.0496,
      "step": 1252
    },
    {
      "epoch": 0.02948235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.8791,
      "step": 1253
    },
    {
      "epoch": 0.029505882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.7936,
      "step": 1254
    },
    {
      "epoch": 0.029529411764705884,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.7743,
      "step": 1255
    },
    {
      "epoch": 0.029552941176470587,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.826,
      "step": 1256
    },
    {
      "epoch": 0.029576470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.896,
      "step": 1257
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 2.071,
      "step": 1258
    },
    {
      "epoch": 0.029623529411764705,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7952,
      "step": 1259
    },
    {
      "epoch": 0.029647058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.8289,
      "step": 1260
    },
    {
      "epoch": 0.02967058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 2.0351,
      "step": 1261
    },
    {
      "epoch": 0.029694117647058822,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.908,
      "step": 1262
    },
    {
      "epoch": 0.02971764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.8617,
      "step": 1263
    },
    {
      "epoch": 0.029741176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7765,
      "step": 1264
    },
    {
      "epoch": 0.02976470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.9335,
      "step": 1265
    },
    {
      "epoch": 0.029788235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9285,
      "step": 1266
    },
    {
      "epoch": 0.029811764705882354,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.9569,
      "step": 1267
    },
    {
      "epoch": 0.02983529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.9513,
      "step": 1268
    },
    {
      "epoch": 0.029858823529411764,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.851,
      "step": 1269
    },
    {
      "epoch": 0.02988235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.8625,
      "step": 1270
    },
    {
      "epoch": 0.029905882352941178,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.9501,
      "step": 1271
    },
    {
      "epoch": 0.02992941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9853,
      "step": 1272
    },
    {
      "epoch": 0.02995294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.9222,
      "step": 1273
    },
    {
      "epoch": 0.029976470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9675,
      "step": 1274
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6688,
      "step": 1275
    },
    {
      "epoch": 0.030023529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.8341,
      "step": 1276
    },
    {
      "epoch": 0.030047058823529413,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9166,
      "step": 1277
    },
    {
      "epoch": 0.030070588235294116,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8891,
      "step": 1278
    },
    {
      "epoch": 0.030094117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9042,
      "step": 1279
    },
    {
      "epoch": 0.03011764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.847,
      "step": 1280
    },
    {
      "epoch": 0.030141176470588234,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.9794,
      "step": 1281
    },
    {
      "epoch": 0.03016470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8156,
      "step": 1282
    },
    {
      "epoch": 0.030188235294117648,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.9576,
      "step": 1283
    },
    {
      "epoch": 0.030211764705882355,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9686,
      "step": 1284
    },
    {
      "epoch": 0.030235294117647058,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7986,
      "step": 1285
    },
    {
      "epoch": 0.030258823529411765,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.9225,
      "step": 1286
    },
    {
      "epoch": 0.030282352941176472,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.9082,
      "step": 1287
    },
    {
      "epoch": 0.030305882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.8693,
      "step": 1288
    },
    {
      "epoch": 0.030329411764705883,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8447,
      "step": 1289
    },
    {
      "epoch": 0.03035294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.8463,
      "step": 1290
    },
    {
      "epoch": 0.030376470588235293,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7309,
      "step": 1291
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.8608,
      "step": 1292
    },
    {
      "epoch": 0.030423529411764707,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8943,
      "step": 1293
    },
    {
      "epoch": 0.03044705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.7828,
      "step": 1294
    },
    {
      "epoch": 0.030470588235294117,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9869,
      "step": 1295
    },
    {
      "epoch": 0.030494117647058824,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8906,
      "step": 1296
    },
    {
      "epoch": 0.030517647058823528,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.8894,
      "step": 1297
    },
    {
      "epoch": 0.030541176470588235,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.9362,
      "step": 1298
    },
    {
      "epoch": 0.030564705882352942,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8148,
      "step": 1299
    },
    {
      "epoch": 0.03058823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.8871,
      "step": 1300
    },
    {
      "epoch": 0.030611764705882352,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.8993,
      "step": 1301
    },
    {
      "epoch": 0.03063529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7538,
      "step": 1302
    },
    {
      "epoch": 0.030658823529411766,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9199,
      "step": 1303
    },
    {
      "epoch": 0.03068235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.9213,
      "step": 1304
    },
    {
      "epoch": 0.030705882352941177,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9403,
      "step": 1305
    },
    {
      "epoch": 0.030729411764705884,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6331,
      "step": 1306
    },
    {
      "epoch": 0.030752941176470587,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.7785,
      "step": 1307
    },
    {
      "epoch": 0.030776470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8978,
      "step": 1308
    },
    {
      "epoch": 0.0308,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7674,
      "step": 1309
    },
    {
      "epoch": 0.030823529411764704,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7592,
      "step": 1310
    },
    {
      "epoch": 0.03084705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 2.1095,
      "step": 1311
    },
    {
      "epoch": 0.03087058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.78,
      "step": 1312
    },
    {
      "epoch": 0.030894117647058822,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 2.053,
      "step": 1313
    },
    {
      "epoch": 0.03091764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.8532,
      "step": 1314
    },
    {
      "epoch": 0.030941176470588236,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8371,
      "step": 1315
    },
    {
      "epoch": 0.030964705882352943,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.9257,
      "step": 1316
    },
    {
      "epoch": 0.030988235294117646,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 2.0202,
      "step": 1317
    },
    {
      "epoch": 0.031011764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.9338,
      "step": 1318
    },
    {
      "epoch": 0.03103529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 2.0083,
      "step": 1319
    },
    {
      "epoch": 0.031058823529411764,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8316,
      "step": 1320
    },
    {
      "epoch": 0.03108235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8893,
      "step": 1321
    },
    {
      "epoch": 0.031105882352941178,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.9612,
      "step": 1322
    },
    {
      "epoch": 0.03112941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8145,
      "step": 1323
    },
    {
      "epoch": 0.031152941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.8743,
      "step": 1324
    },
    {
      "epoch": 0.031176470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8329,
      "step": 1325
    },
    {
      "epoch": 0.0312,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.9633,
      "step": 1326
    },
    {
      "epoch": 0.031223529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.9646,
      "step": 1327
    },
    {
      "epoch": 0.031247058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.7206,
      "step": 1328
    },
    {
      "epoch": 0.03127058823529412,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.8229,
      "step": 1329
    },
    {
      "epoch": 0.031294117647058826,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.5176,
      "step": 1330
    },
    {
      "epoch": 0.031317647058823526,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.753,
      "step": 1331
    },
    {
      "epoch": 0.03134117647058823,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9995,
      "step": 1332
    },
    {
      "epoch": 0.03136470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8198,
      "step": 1333
    },
    {
      "epoch": 0.03138823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7953,
      "step": 1334
    },
    {
      "epoch": 0.031411764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9692,
      "step": 1335
    },
    {
      "epoch": 0.03143529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8083,
      "step": 1336
    },
    {
      "epoch": 0.03145882352941176,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.8179,
      "step": 1337
    },
    {
      "epoch": 0.03148235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7661,
      "step": 1338
    },
    {
      "epoch": 0.031505882352941175,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.959,
      "step": 1339
    },
    {
      "epoch": 0.03152941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7562,
      "step": 1340
    },
    {
      "epoch": 0.03155294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8364,
      "step": 1341
    },
    {
      "epoch": 0.031576470588235296,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.9643,
      "step": 1342
    },
    {
      "epoch": 0.0316,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7297,
      "step": 1343
    },
    {
      "epoch": 0.0316235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7377,
      "step": 1344
    },
    {
      "epoch": 0.03164705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.9377,
      "step": 1345
    },
    {
      "epoch": 0.03167058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9264,
      "step": 1346
    },
    {
      "epoch": 0.031694117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.7272,
      "step": 1347
    },
    {
      "epoch": 0.03171764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.9769,
      "step": 1348
    },
    {
      "epoch": 0.03174117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.8958,
      "step": 1349
    },
    {
      "epoch": 0.03176470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 2.0399,
      "step": 1350
    },
    {
      "epoch": 0.031788235294117645,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8783,
      "step": 1351
    },
    {
      "epoch": 0.03181176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.8479,
      "step": 1352
    },
    {
      "epoch": 0.03183529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.7601,
      "step": 1353
    },
    {
      "epoch": 0.031858823529411766,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.763,
      "step": 1354
    },
    {
      "epoch": 0.03188235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 2.0047,
      "step": 1355
    },
    {
      "epoch": 0.03190588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9238,
      "step": 1356
    },
    {
      "epoch": 0.03192941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.9403,
      "step": 1357
    },
    {
      "epoch": 0.03195294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8969,
      "step": 1358
    },
    {
      "epoch": 0.031976470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.9566,
      "step": 1359
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.8877,
      "step": 1360
    },
    {
      "epoch": 0.03202352941176471,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8933,
      "step": 1361
    },
    {
      "epoch": 0.032047058823529415,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7583,
      "step": 1362
    },
    {
      "epoch": 0.032070588235294115,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8998,
      "step": 1363
    },
    {
      "epoch": 0.03209411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.9069,
      "step": 1364
    },
    {
      "epoch": 0.03211764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.9893,
      "step": 1365
    },
    {
      "epoch": 0.032141176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.8113,
      "step": 1366
    },
    {
      "epoch": 0.03216470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.9584,
      "step": 1367
    },
    {
      "epoch": 0.03218823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.8457,
      "step": 1368
    },
    {
      "epoch": 0.032211764705882356,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8257,
      "step": 1369
    },
    {
      "epoch": 0.032235294117647056,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.9569,
      "step": 1370
    },
    {
      "epoch": 0.03225882352941176,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8158,
      "step": 1371
    },
    {
      "epoch": 0.03228235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.9479,
      "step": 1372
    },
    {
      "epoch": 0.03230588235294118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 2.0768,
      "step": 1373
    },
    {
      "epoch": 0.032329411764705884,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9014,
      "step": 1374
    },
    {
      "epoch": 0.03235294117647059,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.9558,
      "step": 1375
    },
    {
      "epoch": 0.03237647058823529,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.808,
      "step": 1376
    },
    {
      "epoch": 0.0324,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.9386,
      "step": 1377
    },
    {
      "epoch": 0.032423529411764705,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8793,
      "step": 1378
    },
    {
      "epoch": 0.03244705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.8397,
      "step": 1379
    },
    {
      "epoch": 0.03247058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 2.0215,
      "step": 1380
    },
    {
      "epoch": 0.032494117647058826,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.7505,
      "step": 1381
    },
    {
      "epoch": 0.032517647058823526,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8556,
      "step": 1382
    },
    {
      "epoch": 0.03254117647058823,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.8629,
      "step": 1383
    },
    {
      "epoch": 0.03256470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7135,
      "step": 1384
    },
    {
      "epoch": 0.03258823529411765,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.8359,
      "step": 1385
    },
    {
      "epoch": 0.032611764705882354,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8277,
      "step": 1386
    },
    {
      "epoch": 0.03263529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8116,
      "step": 1387
    },
    {
      "epoch": 0.03265882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8104,
      "step": 1388
    },
    {
      "epoch": 0.03268235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7837,
      "step": 1389
    },
    {
      "epoch": 0.032705882352941175,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.841,
      "step": 1390
    },
    {
      "epoch": 0.03272941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.8835,
      "step": 1391
    },
    {
      "epoch": 0.03275294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.9048,
      "step": 1392
    },
    {
      "epoch": 0.032776470588235296,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8586,
      "step": 1393
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.8034,
      "step": 1394
    },
    {
      "epoch": 0.0328235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5946,
      "step": 1395
    },
    {
      "epoch": 0.03284705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 2.0251,
      "step": 1396
    },
    {
      "epoch": 0.03287058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8204,
      "step": 1397
    },
    {
      "epoch": 0.032894117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7593,
      "step": 1398
    },
    {
      "epoch": 0.03291764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9034,
      "step": 1399
    },
    {
      "epoch": 0.03294117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.8431,
      "step": 1400
    },
    {
      "epoch": 0.032964705882352945,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.8735,
      "step": 1401
    },
    {
      "epoch": 0.032988235294117645,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 2.0146,
      "step": 1402
    },
    {
      "epoch": 0.03301176470588235,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.886,
      "step": 1403
    },
    {
      "epoch": 0.03303529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8649,
      "step": 1404
    },
    {
      "epoch": 0.033058823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7449,
      "step": 1405
    },
    {
      "epoch": 0.03308235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.9267,
      "step": 1406
    },
    {
      "epoch": 0.03310588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9877,
      "step": 1407
    },
    {
      "epoch": 0.03312941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8087,
      "step": 1408
    },
    {
      "epoch": 0.033152941176470586,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8419,
      "step": 1409
    },
    {
      "epoch": 0.03317647058823529,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8992,
      "step": 1410
    },
    {
      "epoch": 0.0332,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 2.0818,
      "step": 1411
    },
    {
      "epoch": 0.03322352941176471,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8694,
      "step": 1412
    },
    {
      "epoch": 0.033247058823529414,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7578,
      "step": 1413
    },
    {
      "epoch": 0.033270588235294114,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8499,
      "step": 1414
    },
    {
      "epoch": 0.03329411764705882,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.747,
      "step": 1415
    },
    {
      "epoch": 0.03331764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.7318,
      "step": 1416
    },
    {
      "epoch": 0.033341176470588235,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8938,
      "step": 1417
    },
    {
      "epoch": 0.03336470588235294,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.8873,
      "step": 1418
    },
    {
      "epoch": 0.03338823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9225,
      "step": 1419
    },
    {
      "epoch": 0.033411764705882356,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.9327,
      "step": 1420
    },
    {
      "epoch": 0.033435294117647056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9287,
      "step": 1421
    },
    {
      "epoch": 0.03345882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8321,
      "step": 1422
    },
    {
      "epoch": 0.03348235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9743,
      "step": 1423
    },
    {
      "epoch": 0.03350588235294118,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7627,
      "step": 1424
    },
    {
      "epoch": 0.033529411764705884,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8592,
      "step": 1425
    },
    {
      "epoch": 0.03355294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.6683,
      "step": 1426
    },
    {
      "epoch": 0.03357647058823529,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.833,
      "step": 1427
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8692,
      "step": 1428
    },
    {
      "epoch": 0.033623529411764705,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.8034,
      "step": 1429
    },
    {
      "epoch": 0.03364705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.9014,
      "step": 1430
    },
    {
      "epoch": 0.03367058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7485,
      "step": 1431
    },
    {
      "epoch": 0.033694117647058826,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9786,
      "step": 1432
    },
    {
      "epoch": 0.03371764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.9134,
      "step": 1433
    },
    {
      "epoch": 0.03374117647058823,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.889,
      "step": 1434
    },
    {
      "epoch": 0.03376470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8892,
      "step": 1435
    },
    {
      "epoch": 0.03378823529411765,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.9979,
      "step": 1436
    },
    {
      "epoch": 0.033811764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9049,
      "step": 1437
    },
    {
      "epoch": 0.03383529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.877,
      "step": 1438
    },
    {
      "epoch": 0.03385882352941177,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.907,
      "step": 1439
    },
    {
      "epoch": 0.03388235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7624,
      "step": 1440
    },
    {
      "epoch": 0.033905882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6559,
      "step": 1441
    },
    {
      "epoch": 0.03392941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.8333,
      "step": 1442
    },
    {
      "epoch": 0.03395294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8827,
      "step": 1443
    },
    {
      "epoch": 0.033976470588235295,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.9542,
      "step": 1444
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8911,
      "step": 1445
    },
    {
      "epoch": 0.0340235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.8066,
      "step": 1446
    },
    {
      "epoch": 0.03404705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.7652,
      "step": 1447
    },
    {
      "epoch": 0.034070588235294116,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.8634,
      "step": 1448
    },
    {
      "epoch": 0.03409411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.8792,
      "step": 1449
    },
    {
      "epoch": 0.03411764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7992,
      "step": 1450
    },
    {
      "epoch": 0.03414117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6901,
      "step": 1451
    },
    {
      "epoch": 0.034164705882352944,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9741,
      "step": 1452
    },
    {
      "epoch": 0.034188235294117644,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7054,
      "step": 1453
    },
    {
      "epoch": 0.03421176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6302,
      "step": 1454
    },
    {
      "epoch": 0.03423529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.7145,
      "step": 1455
    },
    {
      "epoch": 0.034258823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8461,
      "step": 1456
    },
    {
      "epoch": 0.03428235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8102,
      "step": 1457
    },
    {
      "epoch": 0.03430588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8036,
      "step": 1458
    },
    {
      "epoch": 0.03432941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.7898,
      "step": 1459
    },
    {
      "epoch": 0.034352941176470586,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.7085,
      "step": 1460
    },
    {
      "epoch": 0.03437647058823529,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.8041,
      "step": 1461
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.9066,
      "step": 1462
    },
    {
      "epoch": 0.03442352941176471,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.9179,
      "step": 1463
    },
    {
      "epoch": 0.034447058823529414,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.8681,
      "step": 1464
    },
    {
      "epoch": 0.03447058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7115,
      "step": 1465
    },
    {
      "epoch": 0.03449411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9169,
      "step": 1466
    },
    {
      "epoch": 0.03451764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.9165,
      "step": 1467
    },
    {
      "epoch": 0.034541176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8628,
      "step": 1468
    },
    {
      "epoch": 0.03456470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8766,
      "step": 1469
    },
    {
      "epoch": 0.03458823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9333,
      "step": 1470
    },
    {
      "epoch": 0.034611764705882356,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8325,
      "step": 1471
    },
    {
      "epoch": 0.034635294117647056,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9774,
      "step": 1472
    },
    {
      "epoch": 0.03465882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8906,
      "step": 1473
    },
    {
      "epoch": 0.03468235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.7487,
      "step": 1474
    },
    {
      "epoch": 0.03470588235294118,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.8411,
      "step": 1475
    },
    {
      "epoch": 0.034729411764705884,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7697,
      "step": 1476
    },
    {
      "epoch": 0.03475294117647059,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.8043,
      "step": 1477
    },
    {
      "epoch": 0.0347764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.839,
      "step": 1478
    },
    {
      "epoch": 0.0348,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.7517,
      "step": 1479
    },
    {
      "epoch": 0.034823529411764705,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.851,
      "step": 1480
    },
    {
      "epoch": 0.03484705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8858,
      "step": 1481
    },
    {
      "epoch": 0.03487058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8501,
      "step": 1482
    },
    {
      "epoch": 0.034894117647058825,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8402,
      "step": 1483
    },
    {
      "epoch": 0.03491764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8463,
      "step": 1484
    },
    {
      "epoch": 0.03494117647058823,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.9314,
      "step": 1485
    },
    {
      "epoch": 0.03496470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.9673,
      "step": 1486
    },
    {
      "epoch": 0.034988235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7734,
      "step": 1487
    },
    {
      "epoch": 0.03501176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7383,
      "step": 1488
    },
    {
      "epoch": 0.03503529411764706,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.9822,
      "step": 1489
    },
    {
      "epoch": 0.03505882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8156,
      "step": 1490
    },
    {
      "epoch": 0.03508235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.8592,
      "step": 1491
    },
    {
      "epoch": 0.035105882352941174,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.709,
      "step": 1492
    },
    {
      "epoch": 0.03512941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.678,
      "step": 1493
    },
    {
      "epoch": 0.03515294117647059,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.7386,
      "step": 1494
    },
    {
      "epoch": 0.035176470588235295,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.9468,
      "step": 1495
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6516,
      "step": 1496
    },
    {
      "epoch": 0.03522352941176471,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.7478,
      "step": 1497
    },
    {
      "epoch": 0.03524705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.7643,
      "step": 1498
    },
    {
      "epoch": 0.035270588235294116,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8045,
      "step": 1499
    },
    {
      "epoch": 0.03529411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8625,
      "step": 1500
    },
    {
      "epoch": 0.03531764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.7011,
      "step": 1501
    },
    {
      "epoch": 0.03534117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.9256,
      "step": 1502
    },
    {
      "epoch": 0.035364705882352944,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.841,
      "step": 1503
    },
    {
      "epoch": 0.035388235294117644,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8116,
      "step": 1504
    },
    {
      "epoch": 0.03541176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8596,
      "step": 1505
    },
    {
      "epoch": 0.03543529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.782,
      "step": 1506
    },
    {
      "epoch": 0.035458823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8814,
      "step": 1507
    },
    {
      "epoch": 0.03548235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9823,
      "step": 1508
    },
    {
      "epoch": 0.03550588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.817,
      "step": 1509
    },
    {
      "epoch": 0.035529411764705886,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7486,
      "step": 1510
    },
    {
      "epoch": 0.035552941176470586,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.7581,
      "step": 1511
    },
    {
      "epoch": 0.03557647058823529,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7578,
      "step": 1512
    },
    {
      "epoch": 0.0356,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6568,
      "step": 1513
    },
    {
      "epoch": 0.03562352941176471,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9138,
      "step": 1514
    },
    {
      "epoch": 0.035647058823529414,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8923,
      "step": 1515
    },
    {
      "epoch": 0.03567058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8502,
      "step": 1516
    },
    {
      "epoch": 0.03569411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7559,
      "step": 1517
    },
    {
      "epoch": 0.03571764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.8343,
      "step": 1518
    },
    {
      "epoch": 0.035741176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.787,
      "step": 1519
    },
    {
      "epoch": 0.03576470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7017,
      "step": 1520
    },
    {
      "epoch": 0.03578823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 2.0124,
      "step": 1521
    },
    {
      "epoch": 0.035811764705882355,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8259,
      "step": 1522
    },
    {
      "epoch": 0.035835294117647055,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7354,
      "step": 1523
    },
    {
      "epoch": 0.03585882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7633,
      "step": 1524
    },
    {
      "epoch": 0.03588235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7913,
      "step": 1525
    },
    {
      "epoch": 0.035905882352941176,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.8911,
      "step": 1526
    },
    {
      "epoch": 0.03592941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7096,
      "step": 1527
    },
    {
      "epoch": 0.03595294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8543,
      "step": 1528
    },
    {
      "epoch": 0.0359764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.7425,
      "step": 1529
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8122,
      "step": 1530
    },
    {
      "epoch": 0.036023529411764704,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8803,
      "step": 1531
    },
    {
      "epoch": 0.03604705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8157,
      "step": 1532
    },
    {
      "epoch": 0.03607058823529412,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8255,
      "step": 1533
    },
    {
      "epoch": 0.036094117647058825,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.9734,
      "step": 1534
    },
    {
      "epoch": 0.03611764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6169,
      "step": 1535
    },
    {
      "epoch": 0.03614117647058823,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.8097,
      "step": 1536
    },
    {
      "epoch": 0.03616470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7141,
      "step": 1537
    },
    {
      "epoch": 0.036188235294117646,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.6655,
      "step": 1538
    },
    {
      "epoch": 0.03621176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8344,
      "step": 1539
    },
    {
      "epoch": 0.03623529411764706,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.9494,
      "step": 1540
    },
    {
      "epoch": 0.03625882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.9202,
      "step": 1541
    },
    {
      "epoch": 0.036282352941176474,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9247,
      "step": 1542
    },
    {
      "epoch": 0.036305882352941174,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.8813,
      "step": 1543
    },
    {
      "epoch": 0.03632941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8104,
      "step": 1544
    },
    {
      "epoch": 0.03635294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6938,
      "step": 1545
    },
    {
      "epoch": 0.036376470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8015,
      "step": 1546
    },
    {
      "epoch": 0.0364,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6279,
      "step": 1547
    },
    {
      "epoch": 0.03642352941176471,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.912,
      "step": 1548
    },
    {
      "epoch": 0.03644705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8825,
      "step": 1549
    },
    {
      "epoch": 0.036470588235294116,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8035,
      "step": 1550
    },
    {
      "epoch": 0.03649411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8851,
      "step": 1551
    },
    {
      "epoch": 0.03651764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8019,
      "step": 1552
    },
    {
      "epoch": 0.03654117647058824,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 2.0053,
      "step": 1553
    },
    {
      "epoch": 0.036564705882352944,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8381,
      "step": 1554
    },
    {
      "epoch": 0.036588235294117644,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.8019,
      "step": 1555
    },
    {
      "epoch": 0.03661176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8835,
      "step": 1556
    },
    {
      "epoch": 0.03663529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6789,
      "step": 1557
    },
    {
      "epoch": 0.036658823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6519,
      "step": 1558
    },
    {
      "epoch": 0.03668235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8392,
      "step": 1559
    },
    {
      "epoch": 0.03670588235294118,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7167,
      "step": 1560
    },
    {
      "epoch": 0.036729411764705885,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.9657,
      "step": 1561
    },
    {
      "epoch": 0.036752941176470585,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.9614,
      "step": 1562
    },
    {
      "epoch": 0.03677647058823529,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8661,
      "step": 1563
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6673,
      "step": 1564
    },
    {
      "epoch": 0.036823529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.619,
      "step": 1565
    },
    {
      "epoch": 0.03684705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.9296,
      "step": 1566
    },
    {
      "epoch": 0.03687058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6427,
      "step": 1567
    },
    {
      "epoch": 0.03689411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7524,
      "step": 1568
    },
    {
      "epoch": 0.03691764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6458,
      "step": 1569
    },
    {
      "epoch": 0.036941176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6635,
      "step": 1570
    },
    {
      "epoch": 0.03696470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6952,
      "step": 1571
    },
    {
      "epoch": 0.03698823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.9219,
      "step": 1572
    },
    {
      "epoch": 0.037011764705882355,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7922,
      "step": 1573
    },
    {
      "epoch": 0.03703529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6855,
      "step": 1574
    },
    {
      "epoch": 0.03705882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7392,
      "step": 1575
    },
    {
      "epoch": 0.03708235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.778,
      "step": 1576
    },
    {
      "epoch": 0.037105882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7095,
      "step": 1577
    },
    {
      "epoch": 0.03712941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8812,
      "step": 1578
    },
    {
      "epoch": 0.03715294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6939,
      "step": 1579
    },
    {
      "epoch": 0.0371764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8686,
      "step": 1580
    },
    {
      "epoch": 0.0372,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.9545,
      "step": 1581
    },
    {
      "epoch": 0.037223529411764704,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8037,
      "step": 1582
    },
    {
      "epoch": 0.03724705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.756,
      "step": 1583
    },
    {
      "epoch": 0.03727058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7626,
      "step": 1584
    },
    {
      "epoch": 0.037294117647058825,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8703,
      "step": 1585
    },
    {
      "epoch": 0.03731764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7699,
      "step": 1586
    },
    {
      "epoch": 0.03734117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5686,
      "step": 1587
    },
    {
      "epoch": 0.03736470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.683,
      "step": 1588
    },
    {
      "epoch": 0.037388235294117646,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8856,
      "step": 1589
    },
    {
      "epoch": 0.03741176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.829,
      "step": 1590
    },
    {
      "epoch": 0.03743529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6599,
      "step": 1591
    },
    {
      "epoch": 0.03745882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 2.0123,
      "step": 1592
    },
    {
      "epoch": 0.037482352941176474,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5786,
      "step": 1593
    },
    {
      "epoch": 0.037505882352941174,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6509,
      "step": 1594
    },
    {
      "epoch": 0.03752941176470588,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.755,
      "step": 1595
    },
    {
      "epoch": 0.03755294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.719,
      "step": 1596
    },
    {
      "epoch": 0.037576470588235295,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.7484,
      "step": 1597
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.8189,
      "step": 1598
    },
    {
      "epoch": 0.03762352941176471,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.7057,
      "step": 1599
    },
    {
      "epoch": 0.03764705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6573,
      "step": 1600
    },
    {
      "epoch": 0.037670588235294115,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.851,
      "step": 1601
    },
    {
      "epoch": 0.03769411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6904,
      "step": 1602
    },
    {
      "epoch": 0.03771764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.7872,
      "step": 1603
    },
    {
      "epoch": 0.037741176470588236,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.6256,
      "step": 1604
    },
    {
      "epoch": 0.03776470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5533,
      "step": 1605
    },
    {
      "epoch": 0.03778823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.9465,
      "step": 1606
    },
    {
      "epoch": 0.03781176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.5709,
      "step": 1607
    },
    {
      "epoch": 0.03783529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6877,
      "step": 1608
    },
    {
      "epoch": 0.037858823529411764,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7238,
      "step": 1609
    },
    {
      "epoch": 0.03788235294117647,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.7625,
      "step": 1610
    },
    {
      "epoch": 0.03790588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8294,
      "step": 1611
    },
    {
      "epoch": 0.037929411764705885,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8414,
      "step": 1612
    },
    {
      "epoch": 0.037952941176470585,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.9132,
      "step": 1613
    },
    {
      "epoch": 0.03797647058823529,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8155,
      "step": 1614
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7573,
      "step": 1615
    },
    {
      "epoch": 0.038023529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.762,
      "step": 1616
    },
    {
      "epoch": 0.03804705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7399,
      "step": 1617
    },
    {
      "epoch": 0.03807058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6245,
      "step": 1618
    },
    {
      "epoch": 0.03809411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6463,
      "step": 1619
    },
    {
      "epoch": 0.03811764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7623,
      "step": 1620
    },
    {
      "epoch": 0.038141176470588234,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7834,
      "step": 1621
    },
    {
      "epoch": 0.03816470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.9126,
      "step": 1622
    },
    {
      "epoch": 0.03818823529411765,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.817,
      "step": 1623
    },
    {
      "epoch": 0.038211764705882355,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8729,
      "step": 1624
    },
    {
      "epoch": 0.03823529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7729,
      "step": 1625
    },
    {
      "epoch": 0.03825882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7579,
      "step": 1626
    },
    {
      "epoch": 0.03828235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7094,
      "step": 1627
    },
    {
      "epoch": 0.038305882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8712,
      "step": 1628
    },
    {
      "epoch": 0.03832941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7274,
      "step": 1629
    },
    {
      "epoch": 0.03835294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.7634,
      "step": 1630
    },
    {
      "epoch": 0.0383764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7714,
      "step": 1631
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.834,
      "step": 1632
    },
    {
      "epoch": 0.038423529411764704,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.7503,
      "step": 1633
    },
    {
      "epoch": 0.03844705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.9602,
      "step": 1634
    },
    {
      "epoch": 0.03847058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7399,
      "step": 1635
    },
    {
      "epoch": 0.038494117647058824,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6096,
      "step": 1636
    },
    {
      "epoch": 0.03851764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.917,
      "step": 1637
    },
    {
      "epoch": 0.03854117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7724,
      "step": 1638
    },
    {
      "epoch": 0.03856470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.849,
      "step": 1639
    },
    {
      "epoch": 0.038588235294117645,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7389,
      "step": 1640
    },
    {
      "epoch": 0.03861176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.7321,
      "step": 1641
    },
    {
      "epoch": 0.03863529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.9412,
      "step": 1642
    },
    {
      "epoch": 0.038658823529411766,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7599,
      "step": 1643
    },
    {
      "epoch": 0.03868235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.752,
      "step": 1644
    },
    {
      "epoch": 0.03870588235294117,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9771,
      "step": 1645
    },
    {
      "epoch": 0.03872941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7148,
      "step": 1646
    },
    {
      "epoch": 0.03875294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7313,
      "step": 1647
    },
    {
      "epoch": 0.038776470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.8099,
      "step": 1648
    },
    {
      "epoch": 0.0388,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8597,
      "step": 1649
    },
    {
      "epoch": 0.03882352941176471,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5945,
      "step": 1650
    },
    {
      "epoch": 0.038847058823529415,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6459,
      "step": 1651
    },
    {
      "epoch": 0.038870588235294115,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6166,
      "step": 1652
    },
    {
      "epoch": 0.03889411764705882,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8168,
      "step": 1653
    },
    {
      "epoch": 0.03891764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8002,
      "step": 1654
    },
    {
      "epoch": 0.038941176470588236,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7018,
      "step": 1655
    },
    {
      "epoch": 0.03896470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7522,
      "step": 1656
    },
    {
      "epoch": 0.03898823529411765,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8522,
      "step": 1657
    },
    {
      "epoch": 0.03901176470588235,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.6938,
      "step": 1658
    },
    {
      "epoch": 0.03903529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8438,
      "step": 1659
    },
    {
      "epoch": 0.039058823529411764,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8874,
      "step": 1660
    },
    {
      "epoch": 0.03908235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.697,
      "step": 1661
    },
    {
      "epoch": 0.03910588235294118,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.6171,
      "step": 1662
    },
    {
      "epoch": 0.039129411764705885,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.9464,
      "step": 1663
    },
    {
      "epoch": 0.039152941176470585,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8553,
      "step": 1664
    },
    {
      "epoch": 0.03917647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.845,
      "step": 1665
    },
    {
      "epoch": 0.0392,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6808,
      "step": 1666
    },
    {
      "epoch": 0.039223529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7192,
      "step": 1667
    },
    {
      "epoch": 0.03924705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8622,
      "step": 1668
    },
    {
      "epoch": 0.03927058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7091,
      "step": 1669
    },
    {
      "epoch": 0.03929411764705883,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.6621,
      "step": 1670
    },
    {
      "epoch": 0.03931764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.64,
      "step": 1671
    },
    {
      "epoch": 0.039341176470588234,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7286,
      "step": 1672
    },
    {
      "epoch": 0.03936470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.8421,
      "step": 1673
    },
    {
      "epoch": 0.03938823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.8589,
      "step": 1674
    },
    {
      "epoch": 0.039411764705882354,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8814,
      "step": 1675
    },
    {
      "epoch": 0.03943529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.603,
      "step": 1676
    },
    {
      "epoch": 0.03945882352941176,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.856,
      "step": 1677
    },
    {
      "epoch": 0.03948235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7699,
      "step": 1678
    },
    {
      "epoch": 0.039505882352941175,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6377,
      "step": 1679
    },
    {
      "epoch": 0.03952941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.9513,
      "step": 1680
    },
    {
      "epoch": 0.03955294117647059,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.7575,
      "step": 1681
    },
    {
      "epoch": 0.039576470588235296,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7494,
      "step": 1682
    },
    {
      "epoch": 0.0396,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 2.0159,
      "step": 1683
    },
    {
      "epoch": 0.0396235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8192,
      "step": 1684
    },
    {
      "epoch": 0.03964705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6582,
      "step": 1685
    },
    {
      "epoch": 0.03967058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7685,
      "step": 1686
    },
    {
      "epoch": 0.039694117647058824,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7971,
      "step": 1687
    },
    {
      "epoch": 0.03971764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7382,
      "step": 1688
    },
    {
      "epoch": 0.03974117647058824,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.72,
      "step": 1689
    },
    {
      "epoch": 0.03976470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.7329,
      "step": 1690
    },
    {
      "epoch": 0.039788235294117645,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7099,
      "step": 1691
    },
    {
      "epoch": 0.03981176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8635,
      "step": 1692
    },
    {
      "epoch": 0.03983529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.7404,
      "step": 1693
    },
    {
      "epoch": 0.039858823529411766,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6337,
      "step": 1694
    },
    {
      "epoch": 0.03988235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7272,
      "step": 1695
    },
    {
      "epoch": 0.03990588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5986,
      "step": 1696
    },
    {
      "epoch": 0.03992941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9215,
      "step": 1697
    },
    {
      "epoch": 0.03995294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7491,
      "step": 1698
    },
    {
      "epoch": 0.039976470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7596,
      "step": 1699
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.6718,
      "step": 1700
    },
    {
      "epoch": 0.04002352941176471,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7229,
      "step": 1701
    },
    {
      "epoch": 0.040047058823529415,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8041,
      "step": 1702
    },
    {
      "epoch": 0.040070588235294115,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.695,
      "step": 1703
    },
    {
      "epoch": 0.04009411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.6359,
      "step": 1704
    },
    {
      "epoch": 0.04011764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8555,
      "step": 1705
    },
    {
      "epoch": 0.040141176470588236,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7762,
      "step": 1706
    },
    {
      "epoch": 0.04016470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7366,
      "step": 1707
    },
    {
      "epoch": 0.04018823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.7424,
      "step": 1708
    },
    {
      "epoch": 0.04021176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7558,
      "step": 1709
    },
    {
      "epoch": 0.04023529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7456,
      "step": 1710
    },
    {
      "epoch": 0.040258823529411764,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.7561,
      "step": 1711
    },
    {
      "epoch": 0.04028235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7751,
      "step": 1712
    },
    {
      "epoch": 0.04030588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.718,
      "step": 1713
    },
    {
      "epoch": 0.040329411764705884,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.608,
      "step": 1714
    },
    {
      "epoch": 0.04035294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7623,
      "step": 1715
    },
    {
      "epoch": 0.04037647058823529,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.8498,
      "step": 1716
    },
    {
      "epoch": 0.0404,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8028,
      "step": 1717
    },
    {
      "epoch": 0.040423529411764705,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7073,
      "step": 1718
    },
    {
      "epoch": 0.04044705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8818,
      "step": 1719
    },
    {
      "epoch": 0.04047058823529412,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.8928,
      "step": 1720
    },
    {
      "epoch": 0.040494117647058826,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7961,
      "step": 1721
    },
    {
      "epoch": 0.040517647058823526,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8332,
      "step": 1722
    },
    {
      "epoch": 0.04054117647058823,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8113,
      "step": 1723
    },
    {
      "epoch": 0.04056470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7196,
      "step": 1724
    },
    {
      "epoch": 0.04058823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.8356,
      "step": 1725
    },
    {
      "epoch": 0.040611764705882354,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7765,
      "step": 1726
    },
    {
      "epoch": 0.04063529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7946,
      "step": 1727
    },
    {
      "epoch": 0.04065882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7002,
      "step": 1728
    },
    {
      "epoch": 0.04068235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6871,
      "step": 1729
    },
    {
      "epoch": 0.040705882352941175,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8279,
      "step": 1730
    },
    {
      "epoch": 0.04072941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6798,
      "step": 1731
    },
    {
      "epoch": 0.04075294117647059,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.7944,
      "step": 1732
    },
    {
      "epoch": 0.040776470588235296,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7252,
      "step": 1733
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.8188,
      "step": 1734
    },
    {
      "epoch": 0.0408235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.669,
      "step": 1735
    },
    {
      "epoch": 0.04084705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.5971,
      "step": 1736
    },
    {
      "epoch": 0.04087058823529412,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.7152,
      "step": 1737
    },
    {
      "epoch": 0.040894117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6852,
      "step": 1738
    },
    {
      "epoch": 0.04091764705882353,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.9235,
      "step": 1739
    },
    {
      "epoch": 0.04094117647058824,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.8056,
      "step": 1740
    },
    {
      "epoch": 0.04096470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.9228,
      "step": 1741
    },
    {
      "epoch": 0.040988235294117645,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6506,
      "step": 1742
    },
    {
      "epoch": 0.04101176470588235,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.9238,
      "step": 1743
    },
    {
      "epoch": 0.04103529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.8973,
      "step": 1744
    },
    {
      "epoch": 0.041058823529411766,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8516,
      "step": 1745
    },
    {
      "epoch": 0.04108235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.7167,
      "step": 1746
    },
    {
      "epoch": 0.04110588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7077,
      "step": 1747
    },
    {
      "epoch": 0.04112941176470588,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 2.0164,
      "step": 1748
    },
    {
      "epoch": 0.04115294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8412,
      "step": 1749
    },
    {
      "epoch": 0.041176470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.6344,
      "step": 1750
    },
    {
      "epoch": 0.0412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.7844,
      "step": 1751
    },
    {
      "epoch": 0.04122352941176471,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 2.0221,
      "step": 1752
    },
    {
      "epoch": 0.041247058823529414,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.7453,
      "step": 1753
    },
    {
      "epoch": 0.041270588235294114,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7363,
      "step": 1754
    },
    {
      "epoch": 0.04129411764705882,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.9846,
      "step": 1755
    },
    {
      "epoch": 0.04131764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6048,
      "step": 1756
    },
    {
      "epoch": 0.041341176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7354,
      "step": 1757
    },
    {
      "epoch": 0.04136470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.7062,
      "step": 1758
    },
    {
      "epoch": 0.04138823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.6621,
      "step": 1759
    },
    {
      "epoch": 0.041411764705882356,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.557,
      "step": 1760
    },
    {
      "epoch": 0.041435294117647056,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7535,
      "step": 1761
    },
    {
      "epoch": 0.04145882352941176,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.8961,
      "step": 1762
    },
    {
      "epoch": 0.04148235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7942,
      "step": 1763
    },
    {
      "epoch": 0.04150588235294118,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.7048,
      "step": 1764
    },
    {
      "epoch": 0.041529411764705884,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6849,
      "step": 1765
    },
    {
      "epoch": 0.04155294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6511,
      "step": 1766
    },
    {
      "epoch": 0.04157647058823529,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7932,
      "step": 1767
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8086,
      "step": 1768
    },
    {
      "epoch": 0.041623529411764705,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.9317,
      "step": 1769
    },
    {
      "epoch": 0.04164705882352941,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.7543,
      "step": 1770
    },
    {
      "epoch": 0.04167058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7038,
      "step": 1771
    },
    {
      "epoch": 0.041694117647058826,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7436,
      "step": 1772
    },
    {
      "epoch": 0.041717647058823526,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6342,
      "step": 1773
    },
    {
      "epoch": 0.04174117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6748,
      "step": 1774
    },
    {
      "epoch": 0.04176470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8443,
      "step": 1775
    },
    {
      "epoch": 0.04178823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8179,
      "step": 1776
    },
    {
      "epoch": 0.041811764705882354,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8081,
      "step": 1777
    },
    {
      "epoch": 0.04183529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6166,
      "step": 1778
    },
    {
      "epoch": 0.04185882352941177,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.8144,
      "step": 1779
    },
    {
      "epoch": 0.04188235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5059,
      "step": 1780
    },
    {
      "epoch": 0.041905882352941175,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.5508,
      "step": 1781
    },
    {
      "epoch": 0.04192941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.6734,
      "step": 1782
    },
    {
      "epoch": 0.04195294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5736,
      "step": 1783
    },
    {
      "epoch": 0.041976470588235296,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.7523,
      "step": 1784
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6223,
      "step": 1785
    },
    {
      "epoch": 0.0420235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8251,
      "step": 1786
    },
    {
      "epoch": 0.04204705882352941,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.896,
      "step": 1787
    },
    {
      "epoch": 0.04207058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.787,
      "step": 1788
    },
    {
      "epoch": 0.042094117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.793,
      "step": 1789
    },
    {
      "epoch": 0.04211764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5486,
      "step": 1790
    },
    {
      "epoch": 0.04214117647058824,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.8334,
      "step": 1791
    },
    {
      "epoch": 0.042164705882352944,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9185,
      "step": 1792
    },
    {
      "epoch": 0.042188235294117644,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7309,
      "step": 1793
    },
    {
      "epoch": 0.04221176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8251,
      "step": 1794
    },
    {
      "epoch": 0.04223529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.7706,
      "step": 1795
    },
    {
      "epoch": 0.042258823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.572,
      "step": 1796
    },
    {
      "epoch": 0.04228235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7351,
      "step": 1797
    },
    {
      "epoch": 0.04230588235294118,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.6562,
      "step": 1798
    },
    {
      "epoch": 0.04232941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6811,
      "step": 1799
    },
    {
      "epoch": 0.042352941176470586,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.8185,
      "step": 1800
    },
    {
      "epoch": 0.04237647058823529,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6458,
      "step": 1801
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7266,
      "step": 1802
    },
    {
      "epoch": 0.04242352941176471,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.7895,
      "step": 1803
    },
    {
      "epoch": 0.042447058823529414,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7364,
      "step": 1804
    },
    {
      "epoch": 0.04247058823529412,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.8823,
      "step": 1805
    },
    {
      "epoch": 0.04249411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5452,
      "step": 1806
    },
    {
      "epoch": 0.04251764705882353,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.02,
      "loss": 1.8889,
      "step": 1807
    },
    {
      "epoch": 0.042541176470588235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.9327,
      "step": 1808
    },
    {
      "epoch": 0.04256470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6519,
      "step": 1809
    },
    {
      "epoch": 0.04258823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6755,
      "step": 1810
    },
    {
      "epoch": 0.042611764705882356,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4529,
      "step": 1811
    },
    {
      "epoch": 0.042635294117647056,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7593,
      "step": 1812
    },
    {
      "epoch": 0.04265882352941176,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.9076,
      "step": 1813
    },
    {
      "epoch": 0.04268235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.7374,
      "step": 1814
    },
    {
      "epoch": 0.04270588235294118,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.6895,
      "step": 1815
    },
    {
      "epoch": 0.042729411764705884,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.555,
      "step": 1816
    },
    {
      "epoch": 0.04275294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6041,
      "step": 1817
    },
    {
      "epoch": 0.04277647058823529,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.5219,
      "step": 1818
    },
    {
      "epoch": 0.0428,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7913,
      "step": 1819
    },
    {
      "epoch": 0.042823529411764705,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8167,
      "step": 1820
    },
    {
      "epoch": 0.04284705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7688,
      "step": 1821
    },
    {
      "epoch": 0.04287058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7961,
      "step": 1822
    },
    {
      "epoch": 0.042894117647058826,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7308,
      "step": 1823
    },
    {
      "epoch": 0.04291764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6021,
      "step": 1824
    },
    {
      "epoch": 0.04294117647058823,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5424,
      "step": 1825
    },
    {
      "epoch": 0.04296470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.712,
      "step": 1826
    },
    {
      "epoch": 0.04298823529411765,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.6622,
      "step": 1827
    },
    {
      "epoch": 0.043011764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7609,
      "step": 1828
    },
    {
      "epoch": 0.04303529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7132,
      "step": 1829
    },
    {
      "epoch": 0.04305882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.634,
      "step": 1830
    },
    {
      "epoch": 0.04308235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7223,
      "step": 1831
    },
    {
      "epoch": 0.043105882352941174,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6728,
      "step": 1832
    },
    {
      "epoch": 0.04312941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6125,
      "step": 1833
    },
    {
      "epoch": 0.04315294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6886,
      "step": 1834
    },
    {
      "epoch": 0.043176470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7982,
      "step": 1835
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.827,
      "step": 1836
    },
    {
      "epoch": 0.04322352941176471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.9562,
      "step": 1837
    },
    {
      "epoch": 0.04324705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.7239,
      "step": 1838
    },
    {
      "epoch": 0.043270588235294116,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8785,
      "step": 1839
    },
    {
      "epoch": 0.04329411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8146,
      "step": 1840
    },
    {
      "epoch": 0.04331764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8746,
      "step": 1841
    },
    {
      "epoch": 0.04334117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.6913,
      "step": 1842
    },
    {
      "epoch": 0.043364705882352944,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.7639,
      "step": 1843
    },
    {
      "epoch": 0.043388235294117644,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.8306,
      "step": 1844
    },
    {
      "epoch": 0.04341176470588235,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.7732,
      "step": 1845
    },
    {
      "epoch": 0.04343529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7088,
      "step": 1846
    },
    {
      "epoch": 0.043458823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8005,
      "step": 1847
    },
    {
      "epoch": 0.04348235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6405,
      "step": 1848
    },
    {
      "epoch": 0.04350588235294118,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.6081,
      "step": 1849
    },
    {
      "epoch": 0.04352941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.7342,
      "step": 1850
    },
    {
      "epoch": 0.043552941176470586,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6723,
      "step": 1851
    },
    {
      "epoch": 0.04357647058823529,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.75,
      "step": 1852
    },
    {
      "epoch": 0.0436,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5895,
      "step": 1853
    },
    {
      "epoch": 0.04362352941176471,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.9268,
      "step": 1854
    },
    {
      "epoch": 0.043647058823529414,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6062,
      "step": 1855
    },
    {
      "epoch": 0.04367058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.6705,
      "step": 1856
    },
    {
      "epoch": 0.04369411764705882,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.6328,
      "step": 1857
    },
    {
      "epoch": 0.04371764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.6232,
      "step": 1858
    },
    {
      "epoch": 0.043741176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7068,
      "step": 1859
    },
    {
      "epoch": 0.04376470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5679,
      "step": 1860
    },
    {
      "epoch": 0.04378823529411765,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.7174,
      "step": 1861
    },
    {
      "epoch": 0.043811764705882356,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8167,
      "step": 1862
    },
    {
      "epoch": 0.043835294117647056,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.6732,
      "step": 1863
    },
    {
      "epoch": 0.04385882352941176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6085,
      "step": 1864
    },
    {
      "epoch": 0.04388235294117647,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.7985,
      "step": 1865
    },
    {
      "epoch": 0.043905882352941177,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.827,
      "step": 1866
    },
    {
      "epoch": 0.043929411764705883,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.8385,
      "step": 1867
    },
    {
      "epoch": 0.04395294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.9128,
      "step": 1868
    },
    {
      "epoch": 0.0439764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5851,
      "step": 1869
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.8933,
      "step": 1870
    },
    {
      "epoch": 0.044023529411764704,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6939,
      "step": 1871
    },
    {
      "epoch": 0.04404705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7504,
      "step": 1872
    },
    {
      "epoch": 0.04407058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6518,
      "step": 1873
    },
    {
      "epoch": 0.044094117647058825,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7169,
      "step": 1874
    },
    {
      "epoch": 0.04411764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.815,
      "step": 1875
    },
    {
      "epoch": 0.04414117647058823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.8283,
      "step": 1876
    },
    {
      "epoch": 0.04416470588235294,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.9374,
      "step": 1877
    },
    {
      "epoch": 0.044188235294117646,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.518,
      "step": 1878
    },
    {
      "epoch": 0.04421176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7628,
      "step": 1879
    },
    {
      "epoch": 0.04423529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6043,
      "step": 1880
    },
    {
      "epoch": 0.04425882352941177,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8122,
      "step": 1881
    },
    {
      "epoch": 0.04428235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6895,
      "step": 1882
    },
    {
      "epoch": 0.044305882352941174,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.6294,
      "step": 1883
    },
    {
      "epoch": 0.04432941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.8311,
      "step": 1884
    },
    {
      "epoch": 0.04435294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7962,
      "step": 1885
    },
    {
      "epoch": 0.044376470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7985,
      "step": 1886
    },
    {
      "epoch": 0.0444,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.8256,
      "step": 1887
    },
    {
      "epoch": 0.04442352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7376,
      "step": 1888
    },
    {
      "epoch": 0.04444705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7068,
      "step": 1889
    },
    {
      "epoch": 0.044470588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8372,
      "step": 1890
    },
    {
      "epoch": 0.04449411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.8333,
      "step": 1891
    },
    {
      "epoch": 0.04451764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6861,
      "step": 1892
    },
    {
      "epoch": 0.04454117647058824,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7761,
      "step": 1893
    },
    {
      "epoch": 0.044564705882352944,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8767,
      "step": 1894
    },
    {
      "epoch": 0.044588235294117644,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4273,
      "step": 1895
    },
    {
      "epoch": 0.04461176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6447,
      "step": 1896
    },
    {
      "epoch": 0.04463529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6893,
      "step": 1897
    },
    {
      "epoch": 0.044658823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3945,
      "step": 1898
    },
    {
      "epoch": 0.04468235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6837,
      "step": 1899
    },
    {
      "epoch": 0.04470588235294118,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8064,
      "step": 1900
    },
    {
      "epoch": 0.044729411764705886,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8655,
      "step": 1901
    },
    {
      "epoch": 0.044752941176470586,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.8871,
      "step": 1902
    },
    {
      "epoch": 0.04477647058823529,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.9252,
      "step": 1903
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4706,
      "step": 1904
    },
    {
      "epoch": 0.044823529411764707,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5738,
      "step": 1905
    },
    {
      "epoch": 0.044847058823529413,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.7938,
      "step": 1906
    },
    {
      "epoch": 0.04487058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.9,
      "step": 1907
    },
    {
      "epoch": 0.04489411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.9591,
      "step": 1908
    },
    {
      "epoch": 0.04491764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7051,
      "step": 1909
    },
    {
      "epoch": 0.044941176470588234,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7355,
      "step": 1910
    },
    {
      "epoch": 0.04496470588235294,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.6996,
      "step": 1911
    },
    {
      "epoch": 0.04498823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7551,
      "step": 1912
    },
    {
      "epoch": 0.045011764705882355,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4988,
      "step": 1913
    },
    {
      "epoch": 0.04503529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7678,
      "step": 1914
    },
    {
      "epoch": 0.04505882352941176,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.7003,
      "step": 1915
    },
    {
      "epoch": 0.04508235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7063,
      "step": 1916
    },
    {
      "epoch": 0.045105882352941176,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.6948,
      "step": 1917
    },
    {
      "epoch": 0.04512941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7859,
      "step": 1918
    },
    {
      "epoch": 0.04515294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6116,
      "step": 1919
    },
    {
      "epoch": 0.0451764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.6542,
      "step": 1920
    },
    {
      "epoch": 0.0452,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7784,
      "step": 1921
    },
    {
      "epoch": 0.045223529411764704,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7128,
      "step": 1922
    },
    {
      "epoch": 0.04524705882352941,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.8106,
      "step": 1923
    },
    {
      "epoch": 0.04527058823529412,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.8393,
      "step": 1924
    },
    {
      "epoch": 0.045294117647058825,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5892,
      "step": 1925
    },
    {
      "epoch": 0.04531764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5395,
      "step": 1926
    },
    {
      "epoch": 0.04534117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4699,
      "step": 1927
    },
    {
      "epoch": 0.04536470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.7973,
      "step": 1928
    },
    {
      "epoch": 0.045388235294117646,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4265,
      "step": 1929
    },
    {
      "epoch": 0.04541176470588235,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.646,
      "step": 1930
    },
    {
      "epoch": 0.04543529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6654,
      "step": 1931
    },
    {
      "epoch": 0.04545882352941177,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4828,
      "step": 1932
    },
    {
      "epoch": 0.045482352941176474,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7838,
      "step": 1933
    },
    {
      "epoch": 0.045505882352941174,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6654,
      "step": 1934
    },
    {
      "epoch": 0.04552941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6904,
      "step": 1935
    },
    {
      "epoch": 0.04555294117647059,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6971,
      "step": 1936
    },
    {
      "epoch": 0.045576470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7965,
      "step": 1937
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.8097,
      "step": 1938
    },
    {
      "epoch": 0.04562352941176471,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.8018,
      "step": 1939
    },
    {
      "epoch": 0.04564705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8416,
      "step": 1940
    },
    {
      "epoch": 0.045670588235294116,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7306,
      "step": 1941
    },
    {
      "epoch": 0.04569411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6124,
      "step": 1942
    },
    {
      "epoch": 0.04571764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7455,
      "step": 1943
    },
    {
      "epoch": 0.045741176470588236,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7733,
      "step": 1944
    },
    {
      "epoch": 0.04576470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.8008,
      "step": 1945
    },
    {
      "epoch": 0.04578823529411765,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6879,
      "step": 1946
    },
    {
      "epoch": 0.04581176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.8412,
      "step": 1947
    },
    {
      "epoch": 0.04583529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6983,
      "step": 1948
    },
    {
      "epoch": 0.045858823529411764,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.8291,
      "step": 1949
    },
    {
      "epoch": 0.04588235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4111,
      "step": 1950
    },
    {
      "epoch": 0.04590588235294118,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7158,
      "step": 1951
    },
    {
      "epoch": 0.045929411764705885,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.8756,
      "step": 1952
    },
    {
      "epoch": 0.045952941176470585,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5916,
      "step": 1953
    },
    {
      "epoch": 0.04597647058823529,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.8903,
      "step": 1954
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5145,
      "step": 1955
    },
    {
      "epoch": 0.046023529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6289,
      "step": 1956
    },
    {
      "epoch": 0.04604705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8445,
      "step": 1957
    },
    {
      "epoch": 0.04607058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6467,
      "step": 1958
    },
    {
      "epoch": 0.04609411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5227,
      "step": 1959
    },
    {
      "epoch": 0.04611764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.7938,
      "step": 1960
    },
    {
      "epoch": 0.046141176470588234,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7994,
      "step": 1961
    },
    {
      "epoch": 0.04616470588235294,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.7311,
      "step": 1962
    },
    {
      "epoch": 0.04618823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7886,
      "step": 1963
    },
    {
      "epoch": 0.046211764705882355,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.6314,
      "step": 1964
    },
    {
      "epoch": 0.04623529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6972,
      "step": 1965
    },
    {
      "epoch": 0.04625882352941176,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.849,
      "step": 1966
    },
    {
      "epoch": 0.04628235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6085,
      "step": 1967
    },
    {
      "epoch": 0.046305882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.604,
      "step": 1968
    },
    {
      "epoch": 0.04632941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6546,
      "step": 1969
    },
    {
      "epoch": 0.04635294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6,
      "step": 1970
    },
    {
      "epoch": 0.0463764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6286,
      "step": 1971
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5962,
      "step": 1972
    },
    {
      "epoch": 0.046423529411764704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6415,
      "step": 1973
    },
    {
      "epoch": 0.04644705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6281,
      "step": 1974
    },
    {
      "epoch": 0.04647058823529412,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7965,
      "step": 1975
    },
    {
      "epoch": 0.046494117647058825,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6771,
      "step": 1976
    },
    {
      "epoch": 0.04651764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6782,
      "step": 1977
    },
    {
      "epoch": 0.04654117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6681,
      "step": 1978
    },
    {
      "epoch": 0.04656470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7492,
      "step": 1979
    },
    {
      "epoch": 0.046588235294117646,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.9723,
      "step": 1980
    },
    {
      "epoch": 0.04661176470588235,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3911,
      "step": 1981
    },
    {
      "epoch": 0.04663529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4312,
      "step": 1982
    },
    {
      "epoch": 0.046658823529411766,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.6826,
      "step": 1983
    },
    {
      "epoch": 0.04668235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4577,
      "step": 1984
    },
    {
      "epoch": 0.04670588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.8454,
      "step": 1985
    },
    {
      "epoch": 0.04672941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7134,
      "step": 1986
    },
    {
      "epoch": 0.04675294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6398,
      "step": 1987
    },
    {
      "epoch": 0.046776470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.874,
      "step": 1988
    },
    {
      "epoch": 0.0468,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7384,
      "step": 1989
    },
    {
      "epoch": 0.04682352941176471,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.7503,
      "step": 1990
    },
    {
      "epoch": 0.04684705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.7857,
      "step": 1991
    },
    {
      "epoch": 0.046870588235294115,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6339,
      "step": 1992
    },
    {
      "epoch": 0.04689411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.6722,
      "step": 1993
    },
    {
      "epoch": 0.04691764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.8102,
      "step": 1994
    },
    {
      "epoch": 0.046941176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.6712,
      "step": 1995
    },
    {
      "epoch": 0.04696470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6506,
      "step": 1996
    },
    {
      "epoch": 0.04698823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.708,
      "step": 1997
    },
    {
      "epoch": 0.04701176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7189,
      "step": 1998
    },
    {
      "epoch": 0.04703529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8909,
      "step": 1999
    },
    {
      "epoch": 0.047058823529411764,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6441,
      "step": 2000
    },
    {
      "epoch": 0.047058823529411764,
      "eval_loss": 2.263021469116211,
      "eval_runtime": 676.2315,
      "eval_samples_per_second": 12.57,
      "eval_steps_per_second": 3.142,
      "step": 2000
    },
    {
      "epoch": 0.04708235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7192,
      "step": 2001
    },
    {
      "epoch": 0.04710588235294118,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5818,
      "step": 2002
    },
    {
      "epoch": 0.047129411764705885,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.7346,
      "step": 2003
    },
    {
      "epoch": 0.047152941176470585,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6061,
      "step": 2004
    },
    {
      "epoch": 0.04717647058823529,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5946,
      "step": 2005
    },
    {
      "epoch": 0.0472,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.8624,
      "step": 2006
    },
    {
      "epoch": 0.047223529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7715,
      "step": 2007
    },
    {
      "epoch": 0.04724705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.8046,
      "step": 2008
    },
    {
      "epoch": 0.04727058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.8892,
      "step": 2009
    },
    {
      "epoch": 0.04729411764705883,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7292,
      "step": 2010
    },
    {
      "epoch": 0.04731764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.8077,
      "step": 2011
    },
    {
      "epoch": 0.047341176470588234,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.789,
      "step": 2012
    },
    {
      "epoch": 0.04736470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.79,
      "step": 2013
    },
    {
      "epoch": 0.04738823529411765,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.771,
      "step": 2014
    },
    {
      "epoch": 0.047411764705882355,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5452,
      "step": 2015
    },
    {
      "epoch": 0.04743529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.7471,
      "step": 2016
    },
    {
      "epoch": 0.04745882352941176,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6742,
      "step": 2017
    },
    {
      "epoch": 0.04748235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6303,
      "step": 2018
    },
    {
      "epoch": 0.047505882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4539,
      "step": 2019
    },
    {
      "epoch": 0.04752941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6728,
      "step": 2020
    },
    {
      "epoch": 0.04755294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6091,
      "step": 2021
    },
    {
      "epoch": 0.047576470588235296,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6893,
      "step": 2022
    },
    {
      "epoch": 0.0476,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6719,
      "step": 2023
    },
    {
      "epoch": 0.0476235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7912,
      "step": 2024
    },
    {
      "epoch": 0.04764705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7486,
      "step": 2025
    },
    {
      "epoch": 0.04767058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5003,
      "step": 2026
    },
    {
      "epoch": 0.047694117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6424,
      "step": 2027
    },
    {
      "epoch": 0.04771764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6649,
      "step": 2028
    },
    {
      "epoch": 0.04774117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5891,
      "step": 2029
    },
    {
      "epoch": 0.04776470588235294,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.7431,
      "step": 2030
    },
    {
      "epoch": 0.047788235294117645,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.8176,
      "step": 2031
    },
    {
      "epoch": 0.04781176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7544,
      "step": 2032
    },
    {
      "epoch": 0.04783529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5835,
      "step": 2033
    },
    {
      "epoch": 0.047858823529411766,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5895,
      "step": 2034
    },
    {
      "epoch": 0.04788235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6684,
      "step": 2035
    },
    {
      "epoch": 0.04790588235294117,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7845,
      "step": 2036
    },
    {
      "epoch": 0.04792941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6105,
      "step": 2037
    },
    {
      "epoch": 0.04795294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.7949,
      "step": 2038
    },
    {
      "epoch": 0.047976470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.6844,
      "step": 2039
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.6095,
      "step": 2040
    },
    {
      "epoch": 0.04802352941176471,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.7058,
      "step": 2041
    },
    {
      "epoch": 0.048047058823529415,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7747,
      "step": 2042
    },
    {
      "epoch": 0.048070588235294115,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.4196,
      "step": 2043
    },
    {
      "epoch": 0.04809411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6599,
      "step": 2044
    },
    {
      "epoch": 0.04811764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7213,
      "step": 2045
    },
    {
      "epoch": 0.048141176470588236,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7273,
      "step": 2046
    },
    {
      "epoch": 0.04816470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7196,
      "step": 2047
    },
    {
      "epoch": 0.04818823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7889,
      "step": 2048
    },
    {
      "epoch": 0.04821176470588235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.7086,
      "step": 2049
    },
    {
      "epoch": 0.04823529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6154,
      "step": 2050
    },
    {
      "epoch": 0.048258823529411764,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7005,
      "step": 2051
    },
    {
      "epoch": 0.04828235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.654,
      "step": 2052
    },
    {
      "epoch": 0.04830588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8394,
      "step": 2053
    },
    {
      "epoch": 0.048329411764705885,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6701,
      "step": 2054
    },
    {
      "epoch": 0.04835294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6679,
      "step": 2055
    },
    {
      "epoch": 0.04837647058823529,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7906,
      "step": 2056
    },
    {
      "epoch": 0.0484,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6368,
      "step": 2057
    },
    {
      "epoch": 0.048423529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.681,
      "step": 2058
    },
    {
      "epoch": 0.04844705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.7783,
      "step": 2059
    },
    {
      "epoch": 0.04847058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7206,
      "step": 2060
    },
    {
      "epoch": 0.048494117647058826,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.9258,
      "step": 2061
    },
    {
      "epoch": 0.048517647058823526,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6462,
      "step": 2062
    },
    {
      "epoch": 0.04854117647058823,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.6584,
      "step": 2063
    },
    {
      "epoch": 0.04856470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6514,
      "step": 2064
    },
    {
      "epoch": 0.04858823529411765,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.6638,
      "step": 2065
    },
    {
      "epoch": 0.048611764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6079,
      "step": 2066
    },
    {
      "epoch": 0.04863529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6951,
      "step": 2067
    },
    {
      "epoch": 0.04865882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5287,
      "step": 2068
    },
    {
      "epoch": 0.04868235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6251,
      "step": 2069
    },
    {
      "epoch": 0.048705882352941175,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7078,
      "step": 2070
    },
    {
      "epoch": 0.04872941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6178,
      "step": 2071
    },
    {
      "epoch": 0.04875294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7471,
      "step": 2072
    },
    {
      "epoch": 0.048776470588235296,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6949,
      "step": 2073
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8773,
      "step": 2074
    },
    {
      "epoch": 0.0488235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.7017,
      "step": 2075
    },
    {
      "epoch": 0.04884705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5856,
      "step": 2076
    },
    {
      "epoch": 0.04887058823529412,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.6496,
      "step": 2077
    },
    {
      "epoch": 0.048894117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6763,
      "step": 2078
    },
    {
      "epoch": 0.04891764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6183,
      "step": 2079
    },
    {
      "epoch": 0.04894117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7232,
      "step": 2080
    },
    {
      "epoch": 0.04896470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.7327,
      "step": 2081
    },
    {
      "epoch": 0.048988235294117645,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5099,
      "step": 2082
    },
    {
      "epoch": 0.04901176470588235,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6583,
      "step": 2083
    },
    {
      "epoch": 0.04903529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6489,
      "step": 2084
    },
    {
      "epoch": 0.049058823529411766,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.6374,
      "step": 2085
    },
    {
      "epoch": 0.04908235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4352,
      "step": 2086
    },
    {
      "epoch": 0.04910588235294118,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6706,
      "step": 2087
    },
    {
      "epoch": 0.04912941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6737,
      "step": 2088
    },
    {
      "epoch": 0.04915294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.7219,
      "step": 2089
    },
    {
      "epoch": 0.049176470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6221,
      "step": 2090
    },
    {
      "epoch": 0.0492,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.631,
      "step": 2091
    },
    {
      "epoch": 0.04922352941176471,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6467,
      "step": 2092
    },
    {
      "epoch": 0.049247058823529415,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5579,
      "step": 2093
    },
    {
      "epoch": 0.049270588235294115,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6748,
      "step": 2094
    },
    {
      "epoch": 0.04929411764705882,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5768,
      "step": 2095
    },
    {
      "epoch": 0.04931764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.6541,
      "step": 2096
    },
    {
      "epoch": 0.049341176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7136,
      "step": 2097
    },
    {
      "epoch": 0.04936470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4011,
      "step": 2098
    },
    {
      "epoch": 0.04938823529411765,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3921,
      "step": 2099
    },
    {
      "epoch": 0.04941176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5924,
      "step": 2100
    },
    {
      "epoch": 0.049435294117647056,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7148,
      "step": 2101
    },
    {
      "epoch": 0.04945882352941176,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.7045,
      "step": 2102
    },
    {
      "epoch": 0.04948235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6938,
      "step": 2103
    },
    {
      "epoch": 0.04950588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6452,
      "step": 2104
    },
    {
      "epoch": 0.049529411764705884,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7483,
      "step": 2105
    },
    {
      "epoch": 0.04955294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.542,
      "step": 2106
    },
    {
      "epoch": 0.04957647058823529,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6839,
      "step": 2107
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7769,
      "step": 2108
    },
    {
      "epoch": 0.049623529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6568,
      "step": 2109
    },
    {
      "epoch": 0.04964705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7362,
      "step": 2110
    },
    {
      "epoch": 0.04967058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.6851,
      "step": 2111
    },
    {
      "epoch": 0.049694117647058826,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6514,
      "step": 2112
    },
    {
      "epoch": 0.049717647058823526,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.6557,
      "step": 2113
    },
    {
      "epoch": 0.04974117647058823,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7111,
      "step": 2114
    },
    {
      "epoch": 0.04976470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8645,
      "step": 2115
    },
    {
      "epoch": 0.04978823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5223,
      "step": 2116
    },
    {
      "epoch": 0.049811764705882354,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.8329,
      "step": 2117
    },
    {
      "epoch": 0.04983529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7049,
      "step": 2118
    },
    {
      "epoch": 0.04985882352941177,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6561,
      "step": 2119
    },
    {
      "epoch": 0.04988235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5797,
      "step": 2120
    },
    {
      "epoch": 0.049905882352941175,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6962,
      "step": 2121
    },
    {
      "epoch": 0.04992941176470588,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8252,
      "step": 2122
    },
    {
      "epoch": 0.04995294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5319,
      "step": 2123
    },
    {
      "epoch": 0.049976470588235296,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5843,
      "step": 2124
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6235,
      "step": 2125
    },
    {
      "epoch": 0.0500235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5048,
      "step": 2126
    },
    {
      "epoch": 0.05004705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6297,
      "step": 2127
    },
    {
      "epoch": 0.05007058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.385,
      "step": 2128
    },
    {
      "epoch": 0.050094117647058824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8255,
      "step": 2129
    },
    {
      "epoch": 0.05011764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.8841,
      "step": 2130
    },
    {
      "epoch": 0.05014117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7885,
      "step": 2131
    },
    {
      "epoch": 0.050164705882352945,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5494,
      "step": 2132
    },
    {
      "epoch": 0.050188235294117645,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4071,
      "step": 2133
    },
    {
      "epoch": 0.05021176470588235,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6295,
      "step": 2134
    },
    {
      "epoch": 0.05023529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5653,
      "step": 2135
    },
    {
      "epoch": 0.050258823529411766,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.727,
      "step": 2136
    },
    {
      "epoch": 0.05028235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.6208,
      "step": 2137
    },
    {
      "epoch": 0.05030588235294118,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.48,
      "step": 2138
    },
    {
      "epoch": 0.05032941176470588,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.5317,
      "step": 2139
    },
    {
      "epoch": 0.050352941176470586,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6503,
      "step": 2140
    },
    {
      "epoch": 0.05037647058823529,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8597,
      "step": 2141
    },
    {
      "epoch": 0.0504,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7032,
      "step": 2142
    },
    {
      "epoch": 0.05042352941176471,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.86,
      "step": 2143
    },
    {
      "epoch": 0.050447058823529414,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4363,
      "step": 2144
    },
    {
      "epoch": 0.050470588235294114,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7457,
      "step": 2145
    },
    {
      "epoch": 0.05049411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5571,
      "step": 2146
    },
    {
      "epoch": 0.05051764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6995,
      "step": 2147
    },
    {
      "epoch": 0.050541176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6069,
      "step": 2148
    },
    {
      "epoch": 0.05056470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5446,
      "step": 2149
    },
    {
      "epoch": 0.05058823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6093,
      "step": 2150
    },
    {
      "epoch": 0.050611764705882356,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4714,
      "step": 2151
    },
    {
      "epoch": 0.050635294117647056,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.6334,
      "step": 2152
    },
    {
      "epoch": 0.05065882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5897,
      "step": 2153
    },
    {
      "epoch": 0.05068235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.7626,
      "step": 2154
    },
    {
      "epoch": 0.05070588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.7158,
      "step": 2155
    },
    {
      "epoch": 0.050729411764705884,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.7515,
      "step": 2156
    },
    {
      "epoch": 0.05075294117647059,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2999,
      "step": 2157
    },
    {
      "epoch": 0.05077647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.7856,
      "step": 2158
    },
    {
      "epoch": 0.0508,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.719,
      "step": 2159
    },
    {
      "epoch": 0.050823529411764705,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7783,
      "step": 2160
    },
    {
      "epoch": 0.05084705882352941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.8757,
      "step": 2161
    },
    {
      "epoch": 0.05087058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.698,
      "step": 2162
    },
    {
      "epoch": 0.050894117647058826,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7593,
      "step": 2163
    },
    {
      "epoch": 0.05091764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.5855,
      "step": 2164
    },
    {
      "epoch": 0.05094117647058823,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7019,
      "step": 2165
    },
    {
      "epoch": 0.05096470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.503,
      "step": 2166
    },
    {
      "epoch": 0.05098823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6532,
      "step": 2167
    },
    {
      "epoch": 0.051011764705882354,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6467,
      "step": 2168
    },
    {
      "epoch": 0.05103529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5472,
      "step": 2169
    },
    {
      "epoch": 0.05105882352941177,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7615,
      "step": 2170
    },
    {
      "epoch": 0.05108235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4389,
      "step": 2171
    },
    {
      "epoch": 0.051105882352941175,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.6574,
      "step": 2172
    },
    {
      "epoch": 0.05112941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7905,
      "step": 2173
    },
    {
      "epoch": 0.05115294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6638,
      "step": 2174
    },
    {
      "epoch": 0.051176470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5326,
      "step": 2175
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6883,
      "step": 2176
    },
    {
      "epoch": 0.0512235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7306,
      "step": 2177
    },
    {
      "epoch": 0.05124705882352941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.6769,
      "step": 2178
    },
    {
      "epoch": 0.051270588235294116,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5943,
      "step": 2179
    },
    {
      "epoch": 0.05129411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.511,
      "step": 2180
    },
    {
      "epoch": 0.05131764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.6215,
      "step": 2181
    },
    {
      "epoch": 0.05134117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4711,
      "step": 2182
    },
    {
      "epoch": 0.051364705882352944,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6441,
      "step": 2183
    },
    {
      "epoch": 0.051388235294117644,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.8125,
      "step": 2184
    },
    {
      "epoch": 0.05141176470588235,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6089,
      "step": 2185
    },
    {
      "epoch": 0.05143529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.629,
      "step": 2186
    },
    {
      "epoch": 0.051458823529411765,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.445,
      "step": 2187
    },
    {
      "epoch": 0.05148235294117647,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.706,
      "step": 2188
    },
    {
      "epoch": 0.05150588235294118,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.8115,
      "step": 2189
    },
    {
      "epoch": 0.05152941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5523,
      "step": 2190
    },
    {
      "epoch": 0.051552941176470586,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.6836,
      "step": 2191
    },
    {
      "epoch": 0.05157647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5895,
      "step": 2192
    },
    {
      "epoch": 0.0516,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.581,
      "step": 2193
    },
    {
      "epoch": 0.05162352941176471,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6909,
      "step": 2194
    },
    {
      "epoch": 0.051647058823529414,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.8205,
      "step": 2195
    },
    {
      "epoch": 0.05167058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7581,
      "step": 2196
    },
    {
      "epoch": 0.05169411764705882,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.5263,
      "step": 2197
    },
    {
      "epoch": 0.05171764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.4769,
      "step": 2198
    },
    {
      "epoch": 0.051741176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5687,
      "step": 2199
    },
    {
      "epoch": 0.05176470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.6618,
      "step": 2200
    },
    {
      "epoch": 0.05178823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.5429,
      "step": 2201
    },
    {
      "epoch": 0.051811764705882356,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.6152,
      "step": 2202
    },
    {
      "epoch": 0.051835294117647056,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.7711,
      "step": 2203
    },
    {
      "epoch": 0.05185882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6872,
      "step": 2204
    },
    {
      "epoch": 0.05188235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6615,
      "step": 2205
    },
    {
      "epoch": 0.05190588235294118,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4917,
      "step": 2206
    },
    {
      "epoch": 0.051929411764705884,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6449,
      "step": 2207
    },
    {
      "epoch": 0.05195294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7936,
      "step": 2208
    },
    {
      "epoch": 0.05197647058823529,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5404,
      "step": 2209
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8003,
      "step": 2210
    },
    {
      "epoch": 0.052023529411764705,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6294,
      "step": 2211
    },
    {
      "epoch": 0.05204705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5158,
      "step": 2212
    },
    {
      "epoch": 0.05207058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.8349,
      "step": 2213
    },
    {
      "epoch": 0.052094117647058825,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6518,
      "step": 2214
    },
    {
      "epoch": 0.05211764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7245,
      "step": 2215
    },
    {
      "epoch": 0.05214117647058823,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.637,
      "step": 2216
    },
    {
      "epoch": 0.05216470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7291,
      "step": 2217
    },
    {
      "epoch": 0.052188235294117646,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6507,
      "step": 2218
    },
    {
      "epoch": 0.05221176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5237,
      "step": 2219
    },
    {
      "epoch": 0.05223529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8264,
      "step": 2220
    },
    {
      "epoch": 0.05225882352941177,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7701,
      "step": 2221
    },
    {
      "epoch": 0.05228235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5588,
      "step": 2222
    },
    {
      "epoch": 0.052305882352941174,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.606,
      "step": 2223
    },
    {
      "epoch": 0.05232941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.582,
      "step": 2224
    },
    {
      "epoch": 0.05235294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6646,
      "step": 2225
    },
    {
      "epoch": 0.052376470588235295,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6875,
      "step": 2226
    },
    {
      "epoch": 0.0524,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.3057,
      "step": 2227
    },
    {
      "epoch": 0.05242352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.638,
      "step": 2228
    },
    {
      "epoch": 0.05244705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7163,
      "step": 2229
    },
    {
      "epoch": 0.052470588235294116,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5944,
      "step": 2230
    },
    {
      "epoch": 0.05249411764705882,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5185,
      "step": 2231
    },
    {
      "epoch": 0.05251764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6441,
      "step": 2232
    },
    {
      "epoch": 0.05254117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4904,
      "step": 2233
    },
    {
      "epoch": 0.052564705882352944,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7279,
      "step": 2234
    },
    {
      "epoch": 0.052588235294117644,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5105,
      "step": 2235
    },
    {
      "epoch": 0.05261176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5254,
      "step": 2236
    },
    {
      "epoch": 0.05263529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.7533,
      "step": 2237
    },
    {
      "epoch": 0.052658823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6801,
      "step": 2238
    },
    {
      "epoch": 0.05268235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6431,
      "step": 2239
    },
    {
      "epoch": 0.05270588235294118,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.48,
      "step": 2240
    },
    {
      "epoch": 0.052729411764705886,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.5848,
      "step": 2241
    },
    {
      "epoch": 0.052752941176470586,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4242,
      "step": 2242
    },
    {
      "epoch": 0.05277647058823529,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4496,
      "step": 2243
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7428,
      "step": 2244
    },
    {
      "epoch": 0.05282352941176471,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4896,
      "step": 2245
    },
    {
      "epoch": 0.052847058823529414,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4211,
      "step": 2246
    },
    {
      "epoch": 0.05287058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.7671,
      "step": 2247
    },
    {
      "epoch": 0.05289411764705882,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3942,
      "step": 2248
    },
    {
      "epoch": 0.05291764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5028,
      "step": 2249
    },
    {
      "epoch": 0.052941176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4805,
      "step": 2250
    },
    {
      "epoch": 0.05296470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5707,
      "step": 2251
    },
    {
      "epoch": 0.05298823529411765,
      "grad_norm": 0.33203125,
      "learning_rate": 0.02,
      "loss": 1.7902,
      "step": 2252
    },
    {
      "epoch": 0.053011764705882355,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7185,
      "step": 2253
    },
    {
      "epoch": 0.053035294117647055,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.759,
      "step": 2254
    },
    {
      "epoch": 0.05305882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6259,
      "step": 2255
    },
    {
      "epoch": 0.05308235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.5491,
      "step": 2256
    },
    {
      "epoch": 0.053105882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6749,
      "step": 2257
    },
    {
      "epoch": 0.05312941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6161,
      "step": 2258
    },
    {
      "epoch": 0.05315294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6414,
      "step": 2259
    },
    {
      "epoch": 0.0531764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7299,
      "step": 2260
    },
    {
      "epoch": 0.0532,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3872,
      "step": 2261
    },
    {
      "epoch": 0.053223529411764704,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.9144,
      "step": 2262
    },
    {
      "epoch": 0.05324705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.527,
      "step": 2263
    },
    {
      "epoch": 0.05327058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6669,
      "step": 2264
    },
    {
      "epoch": 0.053294117647058825,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6082,
      "step": 2265
    },
    {
      "epoch": 0.05331764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5571,
      "step": 2266
    },
    {
      "epoch": 0.05334117647058823,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.6269,
      "step": 2267
    },
    {
      "epoch": 0.05336470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4604,
      "step": 2268
    },
    {
      "epoch": 0.053388235294117646,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7286,
      "step": 2269
    },
    {
      "epoch": 0.05341176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6916,
      "step": 2270
    },
    {
      "epoch": 0.05343529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.7427,
      "step": 2271
    },
    {
      "epoch": 0.05345882352941177,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7295,
      "step": 2272
    },
    {
      "epoch": 0.053482352941176474,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6067,
      "step": 2273
    },
    {
      "epoch": 0.053505882352941174,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4387,
      "step": 2274
    },
    {
      "epoch": 0.05352941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6617,
      "step": 2275
    },
    {
      "epoch": 0.05355294117647059,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.5391,
      "step": 2276
    },
    {
      "epoch": 0.053576470588235295,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4529,
      "step": 2277
    },
    {
      "epoch": 0.0536,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4277,
      "step": 2278
    },
    {
      "epoch": 0.05362352941176471,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.6615,
      "step": 2279
    },
    {
      "epoch": 0.05364705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.431,
      "step": 2280
    },
    {
      "epoch": 0.053670588235294116,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.7013,
      "step": 2281
    },
    {
      "epoch": 0.05369411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.6214,
      "step": 2282
    },
    {
      "epoch": 0.05371764705882353,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.4532,
      "step": 2283
    },
    {
      "epoch": 0.05374117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6699,
      "step": 2284
    },
    {
      "epoch": 0.053764705882352944,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.5944,
      "step": 2285
    },
    {
      "epoch": 0.053788235294117644,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.619,
      "step": 2286
    },
    {
      "epoch": 0.05381176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6427,
      "step": 2287
    },
    {
      "epoch": 0.05383529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4588,
      "step": 2288
    },
    {
      "epoch": 0.053858823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6636,
      "step": 2289
    },
    {
      "epoch": 0.05388235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.6914,
      "step": 2290
    },
    {
      "epoch": 0.05390588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.8605,
      "step": 2291
    },
    {
      "epoch": 0.053929411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7274,
      "step": 2292
    },
    {
      "epoch": 0.053952941176470585,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4878,
      "step": 2293
    },
    {
      "epoch": 0.05397647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.7203,
      "step": 2294
    },
    {
      "epoch": 0.054,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6421,
      "step": 2295
    },
    {
      "epoch": 0.054023529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.643,
      "step": 2296
    },
    {
      "epoch": 0.05404705882352941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.5895,
      "step": 2297
    },
    {
      "epoch": 0.05407058823529412,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7337,
      "step": 2298
    },
    {
      "epoch": 0.05409411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6695,
      "step": 2299
    },
    {
      "epoch": 0.05411764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3718,
      "step": 2300
    },
    {
      "epoch": 0.054141176470588234,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4991,
      "step": 2301
    },
    {
      "epoch": 0.05416470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4232,
      "step": 2302
    },
    {
      "epoch": 0.05418823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6758,
      "step": 2303
    },
    {
      "epoch": 0.054211764705882355,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6331,
      "step": 2304
    },
    {
      "epoch": 0.05423529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7037,
      "step": 2305
    },
    {
      "epoch": 0.05425882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.8062,
      "step": 2306
    },
    {
      "epoch": 0.05428235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5874,
      "step": 2307
    },
    {
      "epoch": 0.054305882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6445,
      "step": 2308
    },
    {
      "epoch": 0.05432941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4705,
      "step": 2309
    },
    {
      "epoch": 0.05435294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.8071,
      "step": 2310
    },
    {
      "epoch": 0.0543764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.4212,
      "step": 2311
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4857,
      "step": 2312
    },
    {
      "epoch": 0.054423529411764704,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.7939,
      "step": 2313
    },
    {
      "epoch": 0.05444705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4887,
      "step": 2314
    },
    {
      "epoch": 0.05447058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5795,
      "step": 2315
    },
    {
      "epoch": 0.054494117647058825,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.7796,
      "step": 2316
    },
    {
      "epoch": 0.05451764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.7945,
      "step": 2317
    },
    {
      "epoch": 0.05454117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6666,
      "step": 2318
    },
    {
      "epoch": 0.05456470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.4297,
      "step": 2319
    },
    {
      "epoch": 0.054588235294117646,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4365,
      "step": 2320
    },
    {
      "epoch": 0.05461176470588235,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.5105,
      "step": 2321
    },
    {
      "epoch": 0.05463529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.6015,
      "step": 2322
    },
    {
      "epoch": 0.05465882352941177,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6264,
      "step": 2323
    },
    {
      "epoch": 0.054682352941176474,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6533,
      "step": 2324
    },
    {
      "epoch": 0.054705882352941174,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5586,
      "step": 2325
    },
    {
      "epoch": 0.05472941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6738,
      "step": 2326
    },
    {
      "epoch": 0.05475294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.4458,
      "step": 2327
    },
    {
      "epoch": 0.054776470588235295,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.8443,
      "step": 2328
    },
    {
      "epoch": 0.0548,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6764,
      "step": 2329
    },
    {
      "epoch": 0.05482352941176471,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5252,
      "step": 2330
    },
    {
      "epoch": 0.05484705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.586,
      "step": 2331
    },
    {
      "epoch": 0.054870588235294115,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.8323,
      "step": 2332
    },
    {
      "epoch": 0.05489411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5322,
      "step": 2333
    },
    {
      "epoch": 0.05491764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6051,
      "step": 2334
    },
    {
      "epoch": 0.054941176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.5107,
      "step": 2335
    },
    {
      "epoch": 0.05496470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.6336,
      "step": 2336
    },
    {
      "epoch": 0.05498823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.6934,
      "step": 2337
    },
    {
      "epoch": 0.05501176470588235,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4177,
      "step": 2338
    },
    {
      "epoch": 0.05503529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6242,
      "step": 2339
    },
    {
      "epoch": 0.055058823529411764,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.6815,
      "step": 2340
    },
    {
      "epoch": 0.05508235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6318,
      "step": 2341
    },
    {
      "epoch": 0.05510588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.7333,
      "step": 2342
    },
    {
      "epoch": 0.055129411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.732,
      "step": 2343
    },
    {
      "epoch": 0.055152941176470585,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4659,
      "step": 2344
    },
    {
      "epoch": 0.05517647058823529,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.737,
      "step": 2345
    },
    {
      "epoch": 0.0552,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3961,
      "step": 2346
    },
    {
      "epoch": 0.055223529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8172,
      "step": 2347
    },
    {
      "epoch": 0.05524705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5983,
      "step": 2348
    },
    {
      "epoch": 0.05527058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.498,
      "step": 2349
    },
    {
      "epoch": 0.05529411764705883,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.6584,
      "step": 2350
    },
    {
      "epoch": 0.05531764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5367,
      "step": 2351
    },
    {
      "epoch": 0.055341176470588234,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6522,
      "step": 2352
    },
    {
      "epoch": 0.05536470588235294,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.7597,
      "step": 2353
    },
    {
      "epoch": 0.05538823529411765,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4553,
      "step": 2354
    },
    {
      "epoch": 0.055411764705882355,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6742,
      "step": 2355
    },
    {
      "epoch": 0.05543529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6798,
      "step": 2356
    },
    {
      "epoch": 0.05545882352941176,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.5784,
      "step": 2357
    },
    {
      "epoch": 0.05548235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5269,
      "step": 2358
    },
    {
      "epoch": 0.055505882352941176,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4862,
      "step": 2359
    },
    {
      "epoch": 0.05552941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6926,
      "step": 2360
    },
    {
      "epoch": 0.05555294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5118,
      "step": 2361
    },
    {
      "epoch": 0.0555764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.623,
      "step": 2362
    },
    {
      "epoch": 0.0556,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6114,
      "step": 2363
    },
    {
      "epoch": 0.055623529411764704,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6732,
      "step": 2364
    },
    {
      "epoch": 0.05564705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7916,
      "step": 2365
    },
    {
      "epoch": 0.05567058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6332,
      "step": 2366
    },
    {
      "epoch": 0.055694117647058825,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7017,
      "step": 2367
    },
    {
      "epoch": 0.05571764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5623,
      "step": 2368
    },
    {
      "epoch": 0.05574117647058824,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5589,
      "step": 2369
    },
    {
      "epoch": 0.05576470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4929,
      "step": 2370
    },
    {
      "epoch": 0.055788235294117645,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6689,
      "step": 2371
    },
    {
      "epoch": 0.05581176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4186,
      "step": 2372
    },
    {
      "epoch": 0.05583529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5398,
      "step": 2373
    },
    {
      "epoch": 0.055858823529411766,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7499,
      "step": 2374
    },
    {
      "epoch": 0.05588235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5923,
      "step": 2375
    },
    {
      "epoch": 0.05590588235294117,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6023,
      "step": 2376
    },
    {
      "epoch": 0.05592941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6987,
      "step": 2377
    },
    {
      "epoch": 0.05595294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5529,
      "step": 2378
    },
    {
      "epoch": 0.055976470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7554,
      "step": 2379
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6689,
      "step": 2380
    },
    {
      "epoch": 0.05602352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.6583,
      "step": 2381
    },
    {
      "epoch": 0.056047058823529415,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5373,
      "step": 2382
    },
    {
      "epoch": 0.056070588235294115,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5862,
      "step": 2383
    },
    {
      "epoch": 0.05609411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6866,
      "step": 2384
    },
    {
      "epoch": 0.05611764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5624,
      "step": 2385
    },
    {
      "epoch": 0.056141176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5747,
      "step": 2386
    },
    {
      "epoch": 0.05616470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.5599,
      "step": 2387
    },
    {
      "epoch": 0.05618823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7842,
      "step": 2388
    },
    {
      "epoch": 0.05621176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7524,
      "step": 2389
    },
    {
      "epoch": 0.05623529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.7304,
      "step": 2390
    },
    {
      "epoch": 0.056258823529411764,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5746,
      "step": 2391
    },
    {
      "epoch": 0.05628235294117647,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4327,
      "step": 2392
    },
    {
      "epoch": 0.05630588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6659,
      "step": 2393
    },
    {
      "epoch": 0.056329411764705885,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5571,
      "step": 2394
    },
    {
      "epoch": 0.056352941176470585,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5965,
      "step": 2395
    },
    {
      "epoch": 0.05637647058823529,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7415,
      "step": 2396
    },
    {
      "epoch": 0.0564,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5526,
      "step": 2397
    },
    {
      "epoch": 0.056423529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5094,
      "step": 2398
    },
    {
      "epoch": 0.05644705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6507,
      "step": 2399
    },
    {
      "epoch": 0.05647058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6117,
      "step": 2400
    },
    {
      "epoch": 0.05649411764705883,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5353,
      "step": 2401
    },
    {
      "epoch": 0.05651764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7111,
      "step": 2402
    },
    {
      "epoch": 0.056541176470588234,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5165,
      "step": 2403
    },
    {
      "epoch": 0.05656470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5512,
      "step": 2404
    },
    {
      "epoch": 0.05658823529411765,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.318,
      "step": 2405
    },
    {
      "epoch": 0.056611764705882354,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6636,
      "step": 2406
    },
    {
      "epoch": 0.05663529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4985,
      "step": 2407
    },
    {
      "epoch": 0.05665882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5431,
      "step": 2408
    },
    {
      "epoch": 0.05668235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6306,
      "step": 2409
    },
    {
      "epoch": 0.056705882352941175,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7726,
      "step": 2410
    },
    {
      "epoch": 0.05672941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6789,
      "step": 2411
    },
    {
      "epoch": 0.05675294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6654,
      "step": 2412
    },
    {
      "epoch": 0.056776470588235296,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5526,
      "step": 2413
    },
    {
      "epoch": 0.0568,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6662,
      "step": 2414
    },
    {
      "epoch": 0.0568235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6571,
      "step": 2415
    },
    {
      "epoch": 0.05684705882352941,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.6011,
      "step": 2416
    },
    {
      "epoch": 0.05687058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7143,
      "step": 2417
    },
    {
      "epoch": 0.056894117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7267,
      "step": 2418
    },
    {
      "epoch": 0.05691764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5993,
      "step": 2419
    },
    {
      "epoch": 0.05694117647058824,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5161,
      "step": 2420
    },
    {
      "epoch": 0.05696470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4724,
      "step": 2421
    },
    {
      "epoch": 0.056988235294117645,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6034,
      "step": 2422
    },
    {
      "epoch": 0.05701176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3007,
      "step": 2423
    },
    {
      "epoch": 0.05703529411764706,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.515,
      "step": 2424
    },
    {
      "epoch": 0.057058823529411766,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6909,
      "step": 2425
    },
    {
      "epoch": 0.05708235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5961,
      "step": 2426
    },
    {
      "epoch": 0.05710588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.6075,
      "step": 2427
    },
    {
      "epoch": 0.05712941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.611,
      "step": 2428
    },
    {
      "epoch": 0.05715294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4243,
      "step": 2429
    },
    {
      "epoch": 0.057176470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.5541,
      "step": 2430
    },
    {
      "epoch": 0.0572,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.4424,
      "step": 2431
    },
    {
      "epoch": 0.05722352941176471,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6903,
      "step": 2432
    },
    {
      "epoch": 0.057247058823529415,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6325,
      "step": 2433
    },
    {
      "epoch": 0.057270588235294115,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5195,
      "step": 2434
    },
    {
      "epoch": 0.05729411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6607,
      "step": 2435
    },
    {
      "epoch": 0.05731764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.7293,
      "step": 2436
    },
    {
      "epoch": 0.057341176470588236,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5116,
      "step": 2437
    },
    {
      "epoch": 0.05736470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4582,
      "step": 2438
    },
    {
      "epoch": 0.05738823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5973,
      "step": 2439
    },
    {
      "epoch": 0.05741176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.607,
      "step": 2440
    },
    {
      "epoch": 0.05743529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4216,
      "step": 2441
    },
    {
      "epoch": 0.057458823529411764,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4716,
      "step": 2442
    },
    {
      "epoch": 0.05748235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5696,
      "step": 2443
    },
    {
      "epoch": 0.05750588235294118,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5873,
      "step": 2444
    },
    {
      "epoch": 0.057529411764705884,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5658,
      "step": 2445
    },
    {
      "epoch": 0.05755294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6513,
      "step": 2446
    },
    {
      "epoch": 0.05757647058823529,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.546,
      "step": 2447
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5221,
      "step": 2448
    },
    {
      "epoch": 0.057623529411764705,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4991,
      "step": 2449
    },
    {
      "epoch": 0.05764705882352941,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.877,
      "step": 2450
    },
    {
      "epoch": 0.05767058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5293,
      "step": 2451
    },
    {
      "epoch": 0.057694117647058826,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.465,
      "step": 2452
    },
    {
      "epoch": 0.057717647058823526,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6887,
      "step": 2453
    },
    {
      "epoch": 0.05774117647058823,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.607,
      "step": 2454
    },
    {
      "epoch": 0.05776470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4599,
      "step": 2455
    },
    {
      "epoch": 0.05778823529411765,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.5624,
      "step": 2456
    },
    {
      "epoch": 0.057811764705882354,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5395,
      "step": 2457
    },
    {
      "epoch": 0.05783529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5799,
      "step": 2458
    },
    {
      "epoch": 0.05785882352941177,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.647,
      "step": 2459
    },
    {
      "epoch": 0.05788235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6623,
      "step": 2460
    },
    {
      "epoch": 0.057905882352941175,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5695,
      "step": 2461
    },
    {
      "epoch": 0.05792941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6033,
      "step": 2462
    },
    {
      "epoch": 0.05795294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3443,
      "step": 2463
    },
    {
      "epoch": 0.057976470588235296,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5344,
      "step": 2464
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5716,
      "step": 2465
    },
    {
      "epoch": 0.0580235294117647,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.8171,
      "step": 2466
    },
    {
      "epoch": 0.05804705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4429,
      "step": 2467
    },
    {
      "epoch": 0.05807058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3993,
      "step": 2468
    },
    {
      "epoch": 0.058094117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6655,
      "step": 2469
    },
    {
      "epoch": 0.05811764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.8674,
      "step": 2470
    },
    {
      "epoch": 0.05814117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6874,
      "step": 2471
    },
    {
      "epoch": 0.05816470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6654,
      "step": 2472
    },
    {
      "epoch": 0.058188235294117645,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3484,
      "step": 2473
    },
    {
      "epoch": 0.05821176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5456,
      "step": 2474
    },
    {
      "epoch": 0.05823529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.575,
      "step": 2475
    },
    {
      "epoch": 0.058258823529411766,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5584,
      "step": 2476
    },
    {
      "epoch": 0.05828235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6619,
      "step": 2477
    },
    {
      "epoch": 0.05830588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8014,
      "step": 2478
    },
    {
      "epoch": 0.05832941176470588,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4826,
      "step": 2479
    },
    {
      "epoch": 0.05835294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.6729,
      "step": 2480
    },
    {
      "epoch": 0.058376470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.695,
      "step": 2481
    },
    {
      "epoch": 0.0584,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4471,
      "step": 2482
    },
    {
      "epoch": 0.05842352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.5236,
      "step": 2483
    },
    {
      "epoch": 0.058447058823529414,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5219,
      "step": 2484
    },
    {
      "epoch": 0.058470588235294114,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6314,
      "step": 2485
    },
    {
      "epoch": 0.05849411764705882,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.485,
      "step": 2486
    },
    {
      "epoch": 0.05851764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5114,
      "step": 2487
    },
    {
      "epoch": 0.058541176470588235,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.5278,
      "step": 2488
    },
    {
      "epoch": 0.05856470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6885,
      "step": 2489
    },
    {
      "epoch": 0.05858823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7687,
      "step": 2490
    },
    {
      "epoch": 0.058611764705882356,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5782,
      "step": 2491
    },
    {
      "epoch": 0.058635294117647056,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4009,
      "step": 2492
    },
    {
      "epoch": 0.05865882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5771,
      "step": 2493
    },
    {
      "epoch": 0.05868235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.7405,
      "step": 2494
    },
    {
      "epoch": 0.05870588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7873,
      "step": 2495
    },
    {
      "epoch": 0.058729411764705884,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5693,
      "step": 2496
    },
    {
      "epoch": 0.05875294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.5194,
      "step": 2497
    },
    {
      "epoch": 0.05877647058823529,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5604,
      "step": 2498
    },
    {
      "epoch": 0.0588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.7258,
      "step": 2499
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6687,
      "step": 2500
    },
    {
      "epoch": 0.05884705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5841,
      "step": 2501
    },
    {
      "epoch": 0.05887058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4378,
      "step": 2502
    },
    {
      "epoch": 0.058894117647058826,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4911,
      "step": 2503
    },
    {
      "epoch": 0.058917647058823526,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.6673,
      "step": 2504
    },
    {
      "epoch": 0.05894117647058823,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4535,
      "step": 2505
    },
    {
      "epoch": 0.05896470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5691,
      "step": 2506
    },
    {
      "epoch": 0.05898823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6715,
      "step": 2507
    },
    {
      "epoch": 0.059011764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.487,
      "step": 2508
    },
    {
      "epoch": 0.05903529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5877,
      "step": 2509
    },
    {
      "epoch": 0.05905882352941177,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.738,
      "step": 2510
    },
    {
      "epoch": 0.05908235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5079,
      "step": 2511
    },
    {
      "epoch": 0.059105882352941175,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5544,
      "step": 2512
    },
    {
      "epoch": 0.05912941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7977,
      "step": 2513
    },
    {
      "epoch": 0.05915294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5177,
      "step": 2514
    },
    {
      "epoch": 0.059176470588235296,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4568,
      "step": 2515
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.3152,
      "step": 2516
    },
    {
      "epoch": 0.0592235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6707,
      "step": 2517
    },
    {
      "epoch": 0.05924705882352941,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3648,
      "step": 2518
    },
    {
      "epoch": 0.05927058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3791,
      "step": 2519
    },
    {
      "epoch": 0.059294117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4239,
      "step": 2520
    },
    {
      "epoch": 0.05931764705882353,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4816,
      "step": 2521
    },
    {
      "epoch": 0.05934117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5519,
      "step": 2522
    },
    {
      "epoch": 0.059364705882352944,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.439,
      "step": 2523
    },
    {
      "epoch": 0.059388235294117644,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.6217,
      "step": 2524
    },
    {
      "epoch": 0.05941176470588235,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5328,
      "step": 2525
    },
    {
      "epoch": 0.05943529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6589,
      "step": 2526
    },
    {
      "epoch": 0.059458823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6958,
      "step": 2527
    },
    {
      "epoch": 0.05948235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5472,
      "step": 2528
    },
    {
      "epoch": 0.05950588235294118,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5418,
      "step": 2529
    },
    {
      "epoch": 0.05952941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5874,
      "step": 2530
    },
    {
      "epoch": 0.059552941176470586,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6284,
      "step": 2531
    },
    {
      "epoch": 0.05957647058823529,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7194,
      "step": 2532
    },
    {
      "epoch": 0.0596,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6991,
      "step": 2533
    },
    {
      "epoch": 0.05962352941176471,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.5672,
      "step": 2534
    },
    {
      "epoch": 0.059647058823529414,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6455,
      "step": 2535
    },
    {
      "epoch": 0.05967058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.6142,
      "step": 2536
    },
    {
      "epoch": 0.05969411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6525,
      "step": 2537
    },
    {
      "epoch": 0.05971764705882353,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 2538
    },
    {
      "epoch": 0.059741176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4627,
      "step": 2539
    },
    {
      "epoch": 0.05976470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7075,
      "step": 2540
    },
    {
      "epoch": 0.05978823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.7067,
      "step": 2541
    },
    {
      "epoch": 0.059811764705882356,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6277,
      "step": 2542
    },
    {
      "epoch": 0.059835294117647056,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6632,
      "step": 2543
    },
    {
      "epoch": 0.05985882352941176,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5214,
      "step": 2544
    },
    {
      "epoch": 0.05988235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.7595,
      "step": 2545
    },
    {
      "epoch": 0.05990588235294118,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6297,
      "step": 2546
    },
    {
      "epoch": 0.059929411764705884,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4846,
      "step": 2547
    },
    {
      "epoch": 0.05995294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6373,
      "step": 2548
    },
    {
      "epoch": 0.05997647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.7806,
      "step": 2549
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.7355,
      "step": 2550
    },
    {
      "epoch": 0.060023529411764705,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.783,
      "step": 2551
    },
    {
      "epoch": 0.06004705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6245,
      "step": 2552
    },
    {
      "epoch": 0.06007058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4924,
      "step": 2553
    },
    {
      "epoch": 0.060094117647058826,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.5727,
      "step": 2554
    },
    {
      "epoch": 0.06011764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5285,
      "step": 2555
    },
    {
      "epoch": 0.06014117647058823,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5453,
      "step": 2556
    },
    {
      "epoch": 0.06016470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6484,
      "step": 2557
    },
    {
      "epoch": 0.06018823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6009,
      "step": 2558
    },
    {
      "epoch": 0.060211764705882354,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5682,
      "step": 2559
    },
    {
      "epoch": 0.06023529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4061,
      "step": 2560
    },
    {
      "epoch": 0.06025882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5784,
      "step": 2561
    },
    {
      "epoch": 0.06028235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4453,
      "step": 2562
    },
    {
      "epoch": 0.060305882352941174,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4144,
      "step": 2563
    },
    {
      "epoch": 0.06032941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4871,
      "step": 2564
    },
    {
      "epoch": 0.06035294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5515,
      "step": 2565
    },
    {
      "epoch": 0.060376470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6429,
      "step": 2566
    },
    {
      "epoch": 0.0604,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5646,
      "step": 2567
    },
    {
      "epoch": 0.06042352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5855,
      "step": 2568
    },
    {
      "epoch": 0.06044705882352941,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.481,
      "step": 2569
    },
    {
      "epoch": 0.060470588235294116,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3148,
      "step": 2570
    },
    {
      "epoch": 0.06049411764705882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3313,
      "step": 2571
    },
    {
      "epoch": 0.06051764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5118,
      "step": 2572
    },
    {
      "epoch": 0.06054117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.7791,
      "step": 2573
    },
    {
      "epoch": 0.060564705882352944,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6725,
      "step": 2574
    },
    {
      "epoch": 0.060588235294117644,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.555,
      "step": 2575
    },
    {
      "epoch": 0.06061176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5988,
      "step": 2576
    },
    {
      "epoch": 0.06063529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4562,
      "step": 2577
    },
    {
      "epoch": 0.060658823529411765,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.285,
      "step": 2578
    },
    {
      "epoch": 0.06068235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6174,
      "step": 2579
    },
    {
      "epoch": 0.06070588235294118,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7247,
      "step": 2580
    },
    {
      "epoch": 0.06072941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5737,
      "step": 2581
    },
    {
      "epoch": 0.060752941176470586,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5579,
      "step": 2582
    },
    {
      "epoch": 0.06077647058823529,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.6271,
      "step": 2583
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.5431,
      "step": 2584
    },
    {
      "epoch": 0.06082352941176471,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6452,
      "step": 2585
    },
    {
      "epoch": 0.060847058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6787,
      "step": 2586
    },
    {
      "epoch": 0.06087058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5321,
      "step": 2587
    },
    {
      "epoch": 0.06089411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.7363,
      "step": 2588
    },
    {
      "epoch": 0.06091764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5892,
      "step": 2589
    },
    {
      "epoch": 0.060941176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4349,
      "step": 2590
    },
    {
      "epoch": 0.06096470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6533,
      "step": 2591
    },
    {
      "epoch": 0.06098823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5564,
      "step": 2592
    },
    {
      "epoch": 0.061011764705882356,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4826,
      "step": 2593
    },
    {
      "epoch": 0.061035294117647056,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5485,
      "step": 2594
    },
    {
      "epoch": 0.06105882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5274,
      "step": 2595
    },
    {
      "epoch": 0.06108235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.5798,
      "step": 2596
    },
    {
      "epoch": 0.06110588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.6287,
      "step": 2597
    },
    {
      "epoch": 0.061129411764705884,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6075,
      "step": 2598
    },
    {
      "epoch": 0.06115294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6299,
      "step": 2599
    },
    {
      "epoch": 0.0611764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7792,
      "step": 2600
    },
    {
      "epoch": 0.0612,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.5354,
      "step": 2601
    },
    {
      "epoch": 0.061223529411764704,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.141,
      "step": 2602
    },
    {
      "epoch": 0.06124705882352941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.7231,
      "step": 2603
    },
    {
      "epoch": 0.06127058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5162,
      "step": 2604
    },
    {
      "epoch": 0.061294117647058825,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.4662,
      "step": 2605
    },
    {
      "epoch": 0.06131764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3146,
      "step": 2606
    },
    {
      "epoch": 0.06134117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5693,
      "step": 2607
    },
    {
      "epoch": 0.06136470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4902,
      "step": 2608
    },
    {
      "epoch": 0.061388235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.8654,
      "step": 2609
    },
    {
      "epoch": 0.06141176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6584,
      "step": 2610
    },
    {
      "epoch": 0.06143529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6807,
      "step": 2611
    },
    {
      "epoch": 0.06145882352941177,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4579,
      "step": 2612
    },
    {
      "epoch": 0.06148235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5117,
      "step": 2613
    },
    {
      "epoch": 0.061505882352941174,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3024,
      "step": 2614
    },
    {
      "epoch": 0.06152941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4379,
      "step": 2615
    },
    {
      "epoch": 0.06155294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6063,
      "step": 2616
    },
    {
      "epoch": 0.061576470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5104,
      "step": 2617
    },
    {
      "epoch": 0.0616,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2378,
      "step": 2618
    },
    {
      "epoch": 0.06162352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5852,
      "step": 2619
    },
    {
      "epoch": 0.06164705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.6974,
      "step": 2620
    },
    {
      "epoch": 0.061670588235294116,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.214,
      "step": 2621
    },
    {
      "epoch": 0.06169411764705882,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.6071,
      "step": 2622
    },
    {
      "epoch": 0.06171764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6047,
      "step": 2623
    },
    {
      "epoch": 0.06174117647058824,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6185,
      "step": 2624
    },
    {
      "epoch": 0.061764705882352944,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5325,
      "step": 2625
    },
    {
      "epoch": 0.061788235294117644,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.741,
      "step": 2626
    },
    {
      "epoch": 0.06181176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.634,
      "step": 2627
    },
    {
      "epoch": 0.06183529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3323,
      "step": 2628
    },
    {
      "epoch": 0.061858823529411765,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3913,
      "step": 2629
    },
    {
      "epoch": 0.06188235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4772,
      "step": 2630
    },
    {
      "epoch": 0.06190588235294118,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.544,
      "step": 2631
    },
    {
      "epoch": 0.061929411764705886,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5045,
      "step": 2632
    },
    {
      "epoch": 0.061952941176470586,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.5407,
      "step": 2633
    },
    {
      "epoch": 0.06197647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7285,
      "step": 2634
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6805,
      "step": 2635
    },
    {
      "epoch": 0.062023529411764707,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5162,
      "step": 2636
    },
    {
      "epoch": 0.062047058823529413,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3843,
      "step": 2637
    },
    {
      "epoch": 0.06207058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4651,
      "step": 2638
    },
    {
      "epoch": 0.06209411764705882,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.6672,
      "step": 2639
    },
    {
      "epoch": 0.06211764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5478,
      "step": 2640
    },
    {
      "epoch": 0.062141176470588234,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.575,
      "step": 2641
    },
    {
      "epoch": 0.06216470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6691,
      "step": 2642
    },
    {
      "epoch": 0.06218823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5624,
      "step": 2643
    },
    {
      "epoch": 0.062211764705882355,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7001,
      "step": 2644
    },
    {
      "epoch": 0.06223529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5598,
      "step": 2645
    },
    {
      "epoch": 0.06225882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6877,
      "step": 2646
    },
    {
      "epoch": 0.06228235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.4521,
      "step": 2647
    },
    {
      "epoch": 0.062305882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6457,
      "step": 2648
    },
    {
      "epoch": 0.06232941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6294,
      "step": 2649
    },
    {
      "epoch": 0.06235294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4222,
      "step": 2650
    },
    {
      "epoch": 0.0623764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.765,
      "step": 2651
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4729,
      "step": 2652
    },
    {
      "epoch": 0.062423529411764704,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3119,
      "step": 2653
    },
    {
      "epoch": 0.06244705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4495,
      "step": 2654
    },
    {
      "epoch": 0.06247058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6124,
      "step": 2655
    },
    {
      "epoch": 0.062494117647058825,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.5538,
      "step": 2656
    },
    {
      "epoch": 0.06251764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.7565,
      "step": 2657
    },
    {
      "epoch": 0.06254117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5642,
      "step": 2658
    },
    {
      "epoch": 0.06256470588235294,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.5675,
      "step": 2659
    },
    {
      "epoch": 0.06258823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6479,
      "step": 2660
    },
    {
      "epoch": 0.06261176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.491,
      "step": 2661
    },
    {
      "epoch": 0.06263529411764705,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7817,
      "step": 2662
    },
    {
      "epoch": 0.06265882352941177,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.7575,
      "step": 2663
    },
    {
      "epoch": 0.06268235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7065,
      "step": 2664
    },
    {
      "epoch": 0.06270588235294118,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6727,
      "step": 2665
    },
    {
      "epoch": 0.06272941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5561,
      "step": 2666
    },
    {
      "epoch": 0.0627529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5246,
      "step": 2667
    },
    {
      "epoch": 0.0627764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.7576,
      "step": 2668
    },
    {
      "epoch": 0.0628,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.643,
      "step": 2669
    },
    {
      "epoch": 0.06282352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5686,
      "step": 2670
    },
    {
      "epoch": 0.06284705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.5111,
      "step": 2671
    },
    {
      "epoch": 0.06287058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.474,
      "step": 2672
    },
    {
      "epoch": 0.06289411764705882,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4876,
      "step": 2673
    },
    {
      "epoch": 0.06291764705882352,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2953,
      "step": 2674
    },
    {
      "epoch": 0.06294117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5756,
      "step": 2675
    },
    {
      "epoch": 0.06296470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6455,
      "step": 2676
    },
    {
      "epoch": 0.06298823529411765,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3846,
      "step": 2677
    },
    {
      "epoch": 0.06301176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3216,
      "step": 2678
    },
    {
      "epoch": 0.06303529411764706,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2938,
      "step": 2679
    },
    {
      "epoch": 0.06305882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5632,
      "step": 2680
    },
    {
      "epoch": 0.06308235294117646,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5272,
      "step": 2681
    },
    {
      "epoch": 0.06310588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4838,
      "step": 2682
    },
    {
      "epoch": 0.06312941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4668,
      "step": 2683
    },
    {
      "epoch": 0.06315294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5628,
      "step": 2684
    },
    {
      "epoch": 0.06317647058823529,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4812,
      "step": 2685
    },
    {
      "epoch": 0.0632,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5839,
      "step": 2686
    },
    {
      "epoch": 0.0632235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5182,
      "step": 2687
    },
    {
      "epoch": 0.0632470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4096,
      "step": 2688
    },
    {
      "epoch": 0.06327058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.68,
      "step": 2689
    },
    {
      "epoch": 0.06329411764705882,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.7542,
      "step": 2690
    },
    {
      "epoch": 0.06331764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5164,
      "step": 2691
    },
    {
      "epoch": 0.06334117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6466,
      "step": 2692
    },
    {
      "epoch": 0.06336470588235295,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.777,
      "step": 2693
    },
    {
      "epoch": 0.06338823529411765,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2895,
      "step": 2694
    },
    {
      "epoch": 0.06341176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.437,
      "step": 2695
    },
    {
      "epoch": 0.06343529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.5095,
      "step": 2696
    },
    {
      "epoch": 0.06345882352941176,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3927,
      "step": 2697
    },
    {
      "epoch": 0.06348235294117648,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.7286,
      "step": 2698
    },
    {
      "epoch": 0.06350588235294118,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.3344,
      "step": 2699
    },
    {
      "epoch": 0.06352941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4605,
      "step": 2700
    },
    {
      "epoch": 0.06355294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.368,
      "step": 2701
    },
    {
      "epoch": 0.06357647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4981,
      "step": 2702
    },
    {
      "epoch": 0.0636,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.7345,
      "step": 2703
    },
    {
      "epoch": 0.0636235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4194,
      "step": 2704
    },
    {
      "epoch": 0.06364705882352942,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3701,
      "step": 2705
    },
    {
      "epoch": 0.06367058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.4201,
      "step": 2706
    },
    {
      "epoch": 0.06369411764705882,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.5464,
      "step": 2707
    },
    {
      "epoch": 0.06371764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.7028,
      "step": 2708
    },
    {
      "epoch": 0.06374117647058823,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.492,
      "step": 2709
    },
    {
      "epoch": 0.06376470588235295,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.5573,
      "step": 2710
    },
    {
      "epoch": 0.06378823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2392,
      "step": 2711
    },
    {
      "epoch": 0.06381176470588236,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5485,
      "step": 2712
    },
    {
      "epoch": 0.06383529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.376,
      "step": 2713
    },
    {
      "epoch": 0.06385882352941176,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.296,
      "step": 2714
    },
    {
      "epoch": 0.06388235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4744,
      "step": 2715
    },
    {
      "epoch": 0.06390588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6621,
      "step": 2716
    },
    {
      "epoch": 0.06392941176470589,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5699,
      "step": 2717
    },
    {
      "epoch": 0.06395294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.7623,
      "step": 2718
    },
    {
      "epoch": 0.06397647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6936,
      "step": 2719
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3839,
      "step": 2720
    },
    {
      "epoch": 0.0640235294117647,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 2721
    },
    {
      "epoch": 0.06404705882352942,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5679,
      "step": 2722
    },
    {
      "epoch": 0.06407058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4283,
      "step": 2723
    },
    {
      "epoch": 0.06409411764705883,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.627,
      "step": 2724
    },
    {
      "epoch": 0.06411764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.7428,
      "step": 2725
    },
    {
      "epoch": 0.06414117647058823,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.3165,
      "step": 2726
    },
    {
      "epoch": 0.06416470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.588,
      "step": 2727
    },
    {
      "epoch": 0.06418823529411764,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5037,
      "step": 2728
    },
    {
      "epoch": 0.06421176470588236,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4249,
      "step": 2729
    },
    {
      "epoch": 0.06423529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5163,
      "step": 2730
    },
    {
      "epoch": 0.06425882352941177,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5199,
      "step": 2731
    },
    {
      "epoch": 0.06428235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3652,
      "step": 2732
    },
    {
      "epoch": 0.06430588235294117,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.4198,
      "step": 2733
    },
    {
      "epoch": 0.06432941176470588,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5926,
      "step": 2734
    },
    {
      "epoch": 0.06435294117647058,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6474,
      "step": 2735
    },
    {
      "epoch": 0.0643764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3128,
      "step": 2736
    },
    {
      "epoch": 0.0644,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.6937,
      "step": 2737
    },
    {
      "epoch": 0.06442352941176471,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4744,
      "step": 2738
    },
    {
      "epoch": 0.06444705882352941,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2669,
      "step": 2739
    },
    {
      "epoch": 0.06447058823529411,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5496,
      "step": 2740
    },
    {
      "epoch": 0.06449411764705883,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6446,
      "step": 2741
    },
    {
      "epoch": 0.06451764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.6975,
      "step": 2742
    },
    {
      "epoch": 0.06454117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2008,
      "step": 2743
    },
    {
      "epoch": 0.06456470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3526,
      "step": 2744
    },
    {
      "epoch": 0.06458823529411764,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6442,
      "step": 2745
    },
    {
      "epoch": 0.06461176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6286,
      "step": 2746
    },
    {
      "epoch": 0.06463529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6873,
      "step": 2747
    },
    {
      "epoch": 0.06465882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6463,
      "step": 2748
    },
    {
      "epoch": 0.06468235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6913,
      "step": 2749
    },
    {
      "epoch": 0.06470588235294118,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.7035,
      "step": 2750
    },
    {
      "epoch": 0.06472941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6871,
      "step": 2751
    },
    {
      "epoch": 0.06475294117647058,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6801,
      "step": 2752
    },
    {
      "epoch": 0.0647764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.7352,
      "step": 2753
    },
    {
      "epoch": 0.0648,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 2754
    },
    {
      "epoch": 0.06482352941176471,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6423,
      "step": 2755
    },
    {
      "epoch": 0.06484705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5841,
      "step": 2756
    },
    {
      "epoch": 0.06487058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4588,
      "step": 2757
    },
    {
      "epoch": 0.06489411764705882,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5653,
      "step": 2758
    },
    {
      "epoch": 0.06491764705882352,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5463,
      "step": 2759
    },
    {
      "epoch": 0.06494117647058824,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1798,
      "step": 2760
    },
    {
      "epoch": 0.06496470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6164,
      "step": 2761
    },
    {
      "epoch": 0.06498823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5263,
      "step": 2762
    },
    {
      "epoch": 0.06501176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6462,
      "step": 2763
    },
    {
      "epoch": 0.06503529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5398,
      "step": 2764
    },
    {
      "epoch": 0.06505882352941177,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1499,
      "step": 2765
    },
    {
      "epoch": 0.06508235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7306,
      "step": 2766
    },
    {
      "epoch": 0.06510588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.557,
      "step": 2767
    },
    {
      "epoch": 0.06512941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3936,
      "step": 2768
    },
    {
      "epoch": 0.0651529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.6835,
      "step": 2769
    },
    {
      "epoch": 0.0651764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6842,
      "step": 2770
    },
    {
      "epoch": 0.0652,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.5701,
      "step": 2771
    },
    {
      "epoch": 0.06522352941176471,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5466,
      "step": 2772
    },
    {
      "epoch": 0.06524705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.397,
      "step": 2773
    },
    {
      "epoch": 0.06527058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4545,
      "step": 2774
    },
    {
      "epoch": 0.06529411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4159,
      "step": 2775
    },
    {
      "epoch": 0.06531764705882354,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.6606,
      "step": 2776
    },
    {
      "epoch": 0.06534117647058824,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5999,
      "step": 2777
    },
    {
      "epoch": 0.06536470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4554,
      "step": 2778
    },
    {
      "epoch": 0.06538823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5316,
      "step": 2779
    },
    {
      "epoch": 0.06541176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.325,
      "step": 2780
    },
    {
      "epoch": 0.06543529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5485,
      "step": 2781
    },
    {
      "epoch": 0.06545882352941176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5017,
      "step": 2782
    },
    {
      "epoch": 0.06548235294117646,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6012,
      "step": 2783
    },
    {
      "epoch": 0.06550588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4511,
      "step": 2784
    },
    {
      "epoch": 0.06552941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4149,
      "step": 2785
    },
    {
      "epoch": 0.06555294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4801,
      "step": 2786
    },
    {
      "epoch": 0.06557647058823529,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4539,
      "step": 2787
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.691,
      "step": 2788
    },
    {
      "epoch": 0.0656235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.6134,
      "step": 2789
    },
    {
      "epoch": 0.0656470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2866,
      "step": 2790
    },
    {
      "epoch": 0.06567058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.4517,
      "step": 2791
    },
    {
      "epoch": 0.06569411764705882,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4414,
      "step": 2792
    },
    {
      "epoch": 0.06571764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5224,
      "step": 2793
    },
    {
      "epoch": 0.06574117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3954,
      "step": 2794
    },
    {
      "epoch": 0.06576470588235295,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2115,
      "step": 2795
    },
    {
      "epoch": 0.06578823529411765,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.5572,
      "step": 2796
    },
    {
      "epoch": 0.06581176470588235,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4641,
      "step": 2797
    },
    {
      "epoch": 0.06583529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.6752,
      "step": 2798
    },
    {
      "epoch": 0.06585882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2997,
      "step": 2799
    },
    {
      "epoch": 0.06588235294117648,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.3056,
      "step": 2800
    },
    {
      "epoch": 0.06590588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5866,
      "step": 2801
    },
    {
      "epoch": 0.06592941176470589,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6488,
      "step": 2802
    },
    {
      "epoch": 0.06595294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5204,
      "step": 2803
    },
    {
      "epoch": 0.06597647058823529,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.294,
      "step": 2804
    },
    {
      "epoch": 0.066,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6448,
      "step": 2805
    },
    {
      "epoch": 0.0660235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3678,
      "step": 2806
    },
    {
      "epoch": 0.06604705882352942,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.5577,
      "step": 2807
    },
    {
      "epoch": 0.06607058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6876,
      "step": 2808
    },
    {
      "epoch": 0.06609411764705882,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4724,
      "step": 2809
    },
    {
      "epoch": 0.06611764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5268,
      "step": 2810
    },
    {
      "epoch": 0.06614117647058823,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.5552,
      "step": 2811
    },
    {
      "epoch": 0.06616470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5522,
      "step": 2812
    },
    {
      "epoch": 0.06618823529411764,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6023,
      "step": 2813
    },
    {
      "epoch": 0.06621176470588236,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4787,
      "step": 2814
    },
    {
      "epoch": 0.06623529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2282,
      "step": 2815
    },
    {
      "epoch": 0.06625882352941176,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7715,
      "step": 2816
    },
    {
      "epoch": 0.06628235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4698,
      "step": 2817
    },
    {
      "epoch": 0.06630588235294117,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.5989,
      "step": 2818
    },
    {
      "epoch": 0.06632941176470589,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4924,
      "step": 2819
    },
    {
      "epoch": 0.06635294117647059,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.5726,
      "step": 2820
    },
    {
      "epoch": 0.0663764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.461,
      "step": 2821
    },
    {
      "epoch": 0.0664,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6378,
      "step": 2822
    },
    {
      "epoch": 0.0664235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4247,
      "step": 2823
    },
    {
      "epoch": 0.06644705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4932,
      "step": 2824
    },
    {
      "epoch": 0.06647058823529411,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.4266,
      "step": 2825
    },
    {
      "epoch": 0.06649411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6682,
      "step": 2826
    },
    {
      "epoch": 0.06651764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.5453,
      "step": 2827
    },
    {
      "epoch": 0.06654117647058823,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5969,
      "step": 2828
    },
    {
      "epoch": 0.06656470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5847,
      "step": 2829
    },
    {
      "epoch": 0.06658823529411764,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3014,
      "step": 2830
    },
    {
      "epoch": 0.06661176470588236,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3876,
      "step": 2831
    },
    {
      "epoch": 0.06663529411764706,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2189,
      "step": 2832
    },
    {
      "epoch": 0.06665882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5789,
      "step": 2833
    },
    {
      "epoch": 0.06668235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4844,
      "step": 2834
    },
    {
      "epoch": 0.06670588235294117,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.5893,
      "step": 2835
    },
    {
      "epoch": 0.06672941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3325,
      "step": 2836
    },
    {
      "epoch": 0.06675294117647058,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5734,
      "step": 2837
    },
    {
      "epoch": 0.0667764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5122,
      "step": 2838
    },
    {
      "epoch": 0.0668,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4356,
      "step": 2839
    },
    {
      "epoch": 0.06682352941176471,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.7035,
      "step": 2840
    },
    {
      "epoch": 0.06684705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4805,
      "step": 2841
    },
    {
      "epoch": 0.06687058823529411,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3347,
      "step": 2842
    },
    {
      "epoch": 0.06689411764705883,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4986,
      "step": 2843
    },
    {
      "epoch": 0.06691764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6632,
      "step": 2844
    },
    {
      "epoch": 0.06694117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6858,
      "step": 2845
    },
    {
      "epoch": 0.06696470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5431,
      "step": 2846
    },
    {
      "epoch": 0.06698823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5602,
      "step": 2847
    },
    {
      "epoch": 0.06701176470588235,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2786,
      "step": 2848
    },
    {
      "epoch": 0.06703529411764705,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5192,
      "step": 2849
    },
    {
      "epoch": 0.06705882352941177,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4538,
      "step": 2850
    },
    {
      "epoch": 0.06708235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5964,
      "step": 2851
    },
    {
      "epoch": 0.06710588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5778,
      "step": 2852
    },
    {
      "epoch": 0.06712941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4485,
      "step": 2853
    },
    {
      "epoch": 0.06715294117647058,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4472,
      "step": 2854
    },
    {
      "epoch": 0.0671764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7192,
      "step": 2855
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4563,
      "step": 2856
    },
    {
      "epoch": 0.06722352941176471,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5294,
      "step": 2857
    },
    {
      "epoch": 0.06724705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6018,
      "step": 2858
    },
    {
      "epoch": 0.06727058823529412,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.6316,
      "step": 2859
    },
    {
      "epoch": 0.06729411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3925,
      "step": 2860
    },
    {
      "epoch": 0.06731764705882352,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.8493,
      "step": 2861
    },
    {
      "epoch": 0.06734117647058824,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.7608,
      "step": 2862
    },
    {
      "epoch": 0.06736470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5374,
      "step": 2863
    },
    {
      "epoch": 0.06738823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2909,
      "step": 2864
    },
    {
      "epoch": 0.06741176470588235,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 2865
    },
    {
      "epoch": 0.06743529411764707,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6358,
      "step": 2866
    },
    {
      "epoch": 0.06745882352941177,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4892,
      "step": 2867
    },
    {
      "epoch": 0.06748235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.571,
      "step": 2868
    },
    {
      "epoch": 0.06750588235294118,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3077,
      "step": 2869
    },
    {
      "epoch": 0.06752941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.553,
      "step": 2870
    },
    {
      "epoch": 0.0675529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.7406,
      "step": 2871
    },
    {
      "epoch": 0.0675764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6444,
      "step": 2872
    },
    {
      "epoch": 0.0676,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.6593,
      "step": 2873
    },
    {
      "epoch": 0.06762352941176471,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3586,
      "step": 2874
    },
    {
      "epoch": 0.06764705882352941,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.3166,
      "step": 2875
    },
    {
      "epoch": 0.06767058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4238,
      "step": 2876
    },
    {
      "epoch": 0.06769411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7171,
      "step": 2877
    },
    {
      "epoch": 0.06771764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4733,
      "step": 2878
    },
    {
      "epoch": 0.06774117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5071,
      "step": 2879
    },
    {
      "epoch": 0.06776470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6977,
      "step": 2880
    },
    {
      "epoch": 0.06778823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6856,
      "step": 2881
    },
    {
      "epoch": 0.06781176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5112,
      "step": 2882
    },
    {
      "epoch": 0.06783529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5905,
      "step": 2883
    },
    {
      "epoch": 0.06785882352941176,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.468,
      "step": 2884
    },
    {
      "epoch": 0.06788235294117648,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.3427,
      "step": 2885
    },
    {
      "epoch": 0.06790588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5308,
      "step": 2886
    },
    {
      "epoch": 0.06792941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3681,
      "step": 2887
    },
    {
      "epoch": 0.06795294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4546,
      "step": 2888
    },
    {
      "epoch": 0.06797647058823529,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4498,
      "step": 2889
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4427,
      "step": 2890
    },
    {
      "epoch": 0.0680235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6328,
      "step": 2891
    },
    {
      "epoch": 0.0680470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4556,
      "step": 2892
    },
    {
      "epoch": 0.06807058823529412,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4788,
      "step": 2893
    },
    {
      "epoch": 0.06809411764705882,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.4449,
      "step": 2894
    },
    {
      "epoch": 0.06811764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5493,
      "step": 2895
    },
    {
      "epoch": 0.06814117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5079,
      "step": 2896
    },
    {
      "epoch": 0.06816470588235295,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3412,
      "step": 2897
    },
    {
      "epoch": 0.06818823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5458,
      "step": 2898
    },
    {
      "epoch": 0.06821176470588235,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.52,
      "step": 2899
    },
    {
      "epoch": 0.06823529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.677,
      "step": 2900
    },
    {
      "epoch": 0.06825882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5168,
      "step": 2901
    },
    {
      "epoch": 0.06828235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4775,
      "step": 2902
    },
    {
      "epoch": 0.06830588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4171,
      "step": 2903
    },
    {
      "epoch": 0.06832941176470589,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5137,
      "step": 2904
    },
    {
      "epoch": 0.06835294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6161,
      "step": 2905
    },
    {
      "epoch": 0.06837647058823529,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5345,
      "step": 2906
    },
    {
      "epoch": 0.0684,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5462,
      "step": 2907
    },
    {
      "epoch": 0.0684235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.576,
      "step": 2908
    },
    {
      "epoch": 0.06844705882352942,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3865,
      "step": 2909
    },
    {
      "epoch": 0.06847058823529412,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5879,
      "step": 2910
    },
    {
      "epoch": 0.06849411764705883,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.6034,
      "step": 2911
    },
    {
      "epoch": 0.06851764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6637,
      "step": 2912
    },
    {
      "epoch": 0.06854117647058823,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3453,
      "step": 2913
    },
    {
      "epoch": 0.06856470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4615,
      "step": 2914
    },
    {
      "epoch": 0.06858823529411764,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.6289,
      "step": 2915
    },
    {
      "epoch": 0.06861176470588236,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.62,
      "step": 2916
    },
    {
      "epoch": 0.06863529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5374,
      "step": 2917
    },
    {
      "epoch": 0.06865882352941176,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5686,
      "step": 2918
    },
    {
      "epoch": 0.06868235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4515,
      "step": 2919
    },
    {
      "epoch": 0.06870588235294117,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.6666,
      "step": 2920
    },
    {
      "epoch": 0.06872941176470589,
      "grad_norm": 0.37109375,
      "learning_rate": 0.02,
      "loss": 1.3161,
      "step": 2921
    },
    {
      "epoch": 0.06875294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5861,
      "step": 2922
    },
    {
      "epoch": 0.0687764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.705,
      "step": 2923
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5582,
      "step": 2924
    },
    {
      "epoch": 0.0688235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4775,
      "step": 2925
    },
    {
      "epoch": 0.06884705882352941,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4143,
      "step": 2926
    },
    {
      "epoch": 0.06887058823529411,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.382,
      "step": 2927
    },
    {
      "epoch": 0.06889411764705883,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.5062,
      "step": 2928
    },
    {
      "epoch": 0.06891764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.33,
      "step": 2929
    },
    {
      "epoch": 0.06894117647058824,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.5108,
      "step": 2930
    },
    {
      "epoch": 0.06896470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5501,
      "step": 2931
    },
    {
      "epoch": 0.06898823529411764,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.233,
      "step": 2932
    },
    {
      "epoch": 0.06901176470588236,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3459,
      "step": 2933
    },
    {
      "epoch": 0.06903529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.7025,
      "step": 2934
    },
    {
      "epoch": 0.06905882352941177,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.7169,
      "step": 2935
    },
    {
      "epoch": 0.06908235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.5306,
      "step": 2936
    },
    {
      "epoch": 0.06910588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3446,
      "step": 2937
    },
    {
      "epoch": 0.06912941176470588,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.3988,
      "step": 2938
    },
    {
      "epoch": 0.06915294117647058,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.6346,
      "step": 2939
    },
    {
      "epoch": 0.0691764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.478,
      "step": 2940
    },
    {
      "epoch": 0.0692,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.5236,
      "step": 2941
    },
    {
      "epoch": 0.06922352941176471,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5266,
      "step": 2942
    },
    {
      "epoch": 0.06924705882352941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5402,
      "step": 2943
    },
    {
      "epoch": 0.06927058823529411,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.5968,
      "step": 2944
    },
    {
      "epoch": 0.06929411764705883,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4532,
      "step": 2945
    },
    {
      "epoch": 0.06931764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6082,
      "step": 2946
    },
    {
      "epoch": 0.06934117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2997,
      "step": 2947
    },
    {
      "epoch": 0.06936470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.7442,
      "step": 2948
    },
    {
      "epoch": 0.06938823529411765,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3408,
      "step": 2949
    },
    {
      "epoch": 0.06941176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5361,
      "step": 2950
    },
    {
      "epoch": 0.06943529411764705,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.769,
      "step": 2951
    },
    {
      "epoch": 0.06945882352941177,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.634,
      "step": 2952
    },
    {
      "epoch": 0.06948235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6556,
      "step": 2953
    },
    {
      "epoch": 0.06950588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6923,
      "step": 2954
    },
    {
      "epoch": 0.06952941176470588,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4009,
      "step": 2955
    },
    {
      "epoch": 0.0695529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7097,
      "step": 2956
    },
    {
      "epoch": 0.0695764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3485,
      "step": 2957
    },
    {
      "epoch": 0.0696,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2648,
      "step": 2958
    },
    {
      "epoch": 0.06962352941176471,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.436,
      "step": 2959
    },
    {
      "epoch": 0.06964705882352941,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3278,
      "step": 2960
    },
    {
      "epoch": 0.06967058823529412,
      "grad_norm": 0.4375,
      "learning_rate": 0.02,
      "loss": 1.6172,
      "step": 2961
    },
    {
      "epoch": 0.06969411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5453,
      "step": 2962
    },
    {
      "epoch": 0.06971764705882352,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4988,
      "step": 2963
    },
    {
      "epoch": 0.06974117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.5962,
      "step": 2964
    },
    {
      "epoch": 0.06976470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5296,
      "step": 2965
    },
    {
      "epoch": 0.06978823529411765,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.5864,
      "step": 2966
    },
    {
      "epoch": 0.06981176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.501,
      "step": 2967
    },
    {
      "epoch": 0.06983529411764706,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.5445,
      "step": 2968
    },
    {
      "epoch": 0.06985882352941176,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2118,
      "step": 2969
    },
    {
      "epoch": 0.06988235294117646,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.6086,
      "step": 2970
    },
    {
      "epoch": 0.06990588235294118,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3997,
      "step": 2971
    },
    {
      "epoch": 0.06992941176470588,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5826,
      "step": 2972
    },
    {
      "epoch": 0.06995294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3975,
      "step": 2973
    },
    {
      "epoch": 0.06997647058823529,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.485,
      "step": 2974
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6995,
      "step": 2975
    },
    {
      "epoch": 0.0700235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4854,
      "step": 2976
    },
    {
      "epoch": 0.0700470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7225,
      "step": 2977
    },
    {
      "epoch": 0.07007058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4405,
      "step": 2978
    },
    {
      "epoch": 0.07009411764705882,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5537,
      "step": 2979
    },
    {
      "epoch": 0.07011764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2317,
      "step": 2980
    },
    {
      "epoch": 0.07014117647058823,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4486,
      "step": 2981
    },
    {
      "epoch": 0.07016470588235293,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.456,
      "step": 2982
    },
    {
      "epoch": 0.07018823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.533,
      "step": 2983
    },
    {
      "epoch": 0.07021176470588235,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2994,
      "step": 2984
    },
    {
      "epoch": 0.07023529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5392,
      "step": 2985
    },
    {
      "epoch": 0.07025882352941176,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.3845,
      "step": 2986
    },
    {
      "epoch": 0.07028235294117648,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4638,
      "step": 2987
    },
    {
      "epoch": 0.07030588235294118,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.6061,
      "step": 2988
    },
    {
      "epoch": 0.07032941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5678,
      "step": 2989
    },
    {
      "epoch": 0.07035294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.6513,
      "step": 2990
    },
    {
      "epoch": 0.07037647058823529,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4113,
      "step": 2991
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5177,
      "step": 2992
    },
    {
      "epoch": 0.0704235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4709,
      "step": 2993
    },
    {
      "epoch": 0.07044705882352942,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2854,
      "step": 2994
    },
    {
      "epoch": 0.07047058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2966,
      "step": 2995
    },
    {
      "epoch": 0.07049411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3667,
      "step": 2996
    },
    {
      "epoch": 0.07051764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6361,
      "step": 2997
    },
    {
      "epoch": 0.07054117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5696,
      "step": 2998
    },
    {
      "epoch": 0.07056470588235295,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.8359,
      "step": 2999
    },
    {
      "epoch": 0.07058823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6019,
      "step": 3000
    },
    {
      "epoch": 0.07058823529411765,
      "eval_loss": 2.2543089389801025,
      "eval_runtime": 677.5565,
      "eval_samples_per_second": 12.545,
      "eval_steps_per_second": 3.136,
      "step": 3000
    },
    {
      "epoch": 0.07061176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4226,
      "step": 3001
    },
    {
      "epoch": 0.07063529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6905,
      "step": 3002
    },
    {
      "epoch": 0.07065882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4736,
      "step": 3003
    },
    {
      "epoch": 0.07068235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.4373,
      "step": 3004
    },
    {
      "epoch": 0.07070588235294117,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6568,
      "step": 3005
    },
    {
      "epoch": 0.07072941176470589,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.6615,
      "step": 3006
    },
    {
      "epoch": 0.07075294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5764,
      "step": 3007
    },
    {
      "epoch": 0.07077647058823529,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.5196,
      "step": 3008
    },
    {
      "epoch": 0.0708,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.6577,
      "step": 3009
    },
    {
      "epoch": 0.0708235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3937,
      "step": 3010
    },
    {
      "epoch": 0.07084705882352942,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3001,
      "step": 3011
    },
    {
      "epoch": 0.07087058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5784,
      "step": 3012
    },
    {
      "epoch": 0.07089411764705883,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4928,
      "step": 3013
    },
    {
      "epoch": 0.07091764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.5267,
      "step": 3014
    },
    {
      "epoch": 0.07094117647058823,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4248,
      "step": 3015
    },
    {
      "epoch": 0.07096470588235294,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3291,
      "step": 3016
    },
    {
      "epoch": 0.07098823529411764,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3293,
      "step": 3017
    },
    {
      "epoch": 0.07101176470588236,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.6009,
      "step": 3018
    },
    {
      "epoch": 0.07103529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5609,
      "step": 3019
    },
    {
      "epoch": 0.07105882352941177,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3903,
      "step": 3020
    },
    {
      "epoch": 0.07108235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5069,
      "step": 3021
    },
    {
      "epoch": 0.07110588235294117,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5436,
      "step": 3022
    },
    {
      "epoch": 0.07112941176470589,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4315,
      "step": 3023
    },
    {
      "epoch": 0.07115294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.635,
      "step": 3024
    },
    {
      "epoch": 0.0711764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5003,
      "step": 3025
    },
    {
      "epoch": 0.0712,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5362,
      "step": 3026
    },
    {
      "epoch": 0.0712235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.535,
      "step": 3027
    },
    {
      "epoch": 0.07124705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4863,
      "step": 3028
    },
    {
      "epoch": 0.07127058823529411,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3953,
      "step": 3029
    },
    {
      "epoch": 0.07129411764705883,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1818,
      "step": 3030
    },
    {
      "epoch": 0.07131764705882353,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.1104,
      "step": 3031
    },
    {
      "epoch": 0.07134117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5388,
      "step": 3032
    },
    {
      "epoch": 0.07136470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.6368,
      "step": 3033
    },
    {
      "epoch": 0.07138823529411764,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4729,
      "step": 3034
    },
    {
      "epoch": 0.07141176470588236,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.5555,
      "step": 3035
    },
    {
      "epoch": 0.07143529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5434,
      "step": 3036
    },
    {
      "epoch": 0.07145882352941177,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3799,
      "step": 3037
    },
    {
      "epoch": 0.07148235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3448,
      "step": 3038
    },
    {
      "epoch": 0.07150588235294118,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.624,
      "step": 3039
    },
    {
      "epoch": 0.07152941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5355,
      "step": 3040
    },
    {
      "epoch": 0.07155294117647058,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4633,
      "step": 3041
    },
    {
      "epoch": 0.0715764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5704,
      "step": 3042
    },
    {
      "epoch": 0.0716,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.45,
      "step": 3043
    },
    {
      "epoch": 0.07162352941176471,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.3567,
      "step": 3044
    },
    {
      "epoch": 0.07164705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4621,
      "step": 3045
    },
    {
      "epoch": 0.07167058823529411,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5654,
      "step": 3046
    },
    {
      "epoch": 0.07169411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3935,
      "step": 3047
    },
    {
      "epoch": 0.07171764705882352,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4701,
      "step": 3048
    },
    {
      "epoch": 0.07174117647058824,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.7536,
      "step": 3049
    },
    {
      "epoch": 0.07176470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.5196,
      "step": 3050
    },
    {
      "epoch": 0.07178823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3376,
      "step": 3051
    },
    {
      "epoch": 0.07181176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3834,
      "step": 3052
    },
    {
      "epoch": 0.07183529411764705,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.6447,
      "step": 3053
    },
    {
      "epoch": 0.07185882352941177,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.479,
      "step": 3054
    },
    {
      "epoch": 0.07188235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3962,
      "step": 3055
    },
    {
      "epoch": 0.07190588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5119,
      "step": 3056
    },
    {
      "epoch": 0.07192941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5368,
      "step": 3057
    },
    {
      "epoch": 0.0719529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4431,
      "step": 3058
    },
    {
      "epoch": 0.0719764705882353,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.4289,
      "step": 3059
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6237,
      "step": 3060
    },
    {
      "epoch": 0.07202352941176471,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6027,
      "step": 3061
    },
    {
      "epoch": 0.07204705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4917,
      "step": 3062
    },
    {
      "epoch": 0.07207058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3922,
      "step": 3063
    },
    {
      "epoch": 0.07209411764705882,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.6006,
      "step": 3064
    },
    {
      "epoch": 0.07211764705882354,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4909,
      "step": 3065
    },
    {
      "epoch": 0.07214117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4703,
      "step": 3066
    },
    {
      "epoch": 0.07216470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.5021,
      "step": 3067
    },
    {
      "epoch": 0.07218823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5925,
      "step": 3068
    },
    {
      "epoch": 0.07221176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3903,
      "step": 3069
    },
    {
      "epoch": 0.07223529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3428,
      "step": 3070
    },
    {
      "epoch": 0.07225882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4716,
      "step": 3071
    },
    {
      "epoch": 0.07228235294117646,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5302,
      "step": 3072
    },
    {
      "epoch": 0.07230588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.5268,
      "step": 3073
    },
    {
      "epoch": 0.07232941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6536,
      "step": 3074
    },
    {
      "epoch": 0.07235294117647059,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1692,
      "step": 3075
    },
    {
      "epoch": 0.07237647058823529,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9233,
      "step": 3076
    },
    {
      "epoch": 0.0724,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5731,
      "step": 3077
    },
    {
      "epoch": 0.0724235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 3078
    },
    {
      "epoch": 0.0724470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6858,
      "step": 3079
    },
    {
      "epoch": 0.07247058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4447,
      "step": 3080
    },
    {
      "epoch": 0.07249411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4539,
      "step": 3081
    },
    {
      "epoch": 0.07251764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7181,
      "step": 3082
    },
    {
      "epoch": 0.07254117647058823,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2294,
      "step": 3083
    },
    {
      "epoch": 0.07256470588235295,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.42,
      "step": 3084
    },
    {
      "epoch": 0.07258823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4517,
      "step": 3085
    },
    {
      "epoch": 0.07261176470588235,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5752,
      "step": 3086
    },
    {
      "epoch": 0.07263529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4507,
      "step": 3087
    },
    {
      "epoch": 0.07265882352941176,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 1.2776,
      "step": 3088
    },
    {
      "epoch": 0.07268235294117648,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4956,
      "step": 3089
    },
    {
      "epoch": 0.07270588235294118,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6403,
      "step": 3090
    },
    {
      "epoch": 0.07272941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3938,
      "step": 3091
    },
    {
      "epoch": 0.07275294117647059,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3442,
      "step": 3092
    },
    {
      "epoch": 0.07277647058823529,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4066,
      "step": 3093
    },
    {
      "epoch": 0.0728,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.4942,
      "step": 3094
    },
    {
      "epoch": 0.0728235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4299,
      "step": 3095
    },
    {
      "epoch": 0.07284705882352942,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.7208,
      "step": 3096
    },
    {
      "epoch": 0.07287058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3378,
      "step": 3097
    },
    {
      "epoch": 0.07289411764705882,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6165,
      "step": 3098
    },
    {
      "epoch": 0.07291764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4866,
      "step": 3099
    },
    {
      "epoch": 0.07294117647058823,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3398,
      "step": 3100
    },
    {
      "epoch": 0.07296470588235295,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3307,
      "step": 3101
    },
    {
      "epoch": 0.07298823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4773,
      "step": 3102
    },
    {
      "epoch": 0.07301176470588236,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3931,
      "step": 3103
    },
    {
      "epoch": 0.07303529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2927,
      "step": 3104
    },
    {
      "epoch": 0.07305882352941176,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3631,
      "step": 3105
    },
    {
      "epoch": 0.07308235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.6272,
      "step": 3106
    },
    {
      "epoch": 0.07310588235294117,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5896,
      "step": 3107
    },
    {
      "epoch": 0.07312941176470589,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.327,
      "step": 3108
    },
    {
      "epoch": 0.07315294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3867,
      "step": 3109
    },
    {
      "epoch": 0.07317647058823529,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3185,
      "step": 3110
    },
    {
      "epoch": 0.0732,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1905,
      "step": 3111
    },
    {
      "epoch": 0.0732235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2968,
      "step": 3112
    },
    {
      "epoch": 0.07324705882352942,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7025,
      "step": 3113
    },
    {
      "epoch": 0.07327058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.5129,
      "step": 3114
    },
    {
      "epoch": 0.07329411764705883,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.317,
      "step": 3115
    },
    {
      "epoch": 0.07331764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.4017,
      "step": 3116
    },
    {
      "epoch": 0.07334117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2855,
      "step": 3117
    },
    {
      "epoch": 0.07336470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.461,
      "step": 3118
    },
    {
      "epoch": 0.07338823529411764,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6821,
      "step": 3119
    },
    {
      "epoch": 0.07341176470588236,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.401,
      "step": 3120
    },
    {
      "epoch": 0.07343529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5732,
      "step": 3121
    },
    {
      "epoch": 0.07345882352941177,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.4949,
      "step": 3122
    },
    {
      "epoch": 0.07348235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.697,
      "step": 3123
    },
    {
      "epoch": 0.07350588235294117,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.5023,
      "step": 3124
    },
    {
      "epoch": 0.07352941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5814,
      "step": 3125
    },
    {
      "epoch": 0.07355294117647058,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.6569,
      "step": 3126
    },
    {
      "epoch": 0.0735764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4316,
      "step": 3127
    },
    {
      "epoch": 0.0736,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.413,
      "step": 3128
    },
    {
      "epoch": 0.07362352941176471,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.4943,
      "step": 3129
    },
    {
      "epoch": 0.07364705882352941,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4439,
      "step": 3130
    },
    {
      "epoch": 0.07367058823529411,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6934,
      "step": 3131
    },
    {
      "epoch": 0.07369411764705883,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4579,
      "step": 3132
    },
    {
      "epoch": 0.07371764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4235,
      "step": 3133
    },
    {
      "epoch": 0.07374117647058824,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3512,
      "step": 3134
    },
    {
      "epoch": 0.07376470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5311,
      "step": 3135
    },
    {
      "epoch": 0.07378823529411764,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4756,
      "step": 3136
    },
    {
      "epoch": 0.07381176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4748,
      "step": 3137
    },
    {
      "epoch": 0.07383529411764705,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.4289,
      "step": 3138
    },
    {
      "epoch": 0.07385882352941177,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5612,
      "step": 3139
    },
    {
      "epoch": 0.07388235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.5199,
      "step": 3140
    },
    {
      "epoch": 0.07390588235294118,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 3141
    },
    {
      "epoch": 0.07392941176470588,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0363,
      "step": 3142
    },
    {
      "epoch": 0.07395294117647058,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5577,
      "step": 3143
    },
    {
      "epoch": 0.0739764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4101,
      "step": 3144
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3591,
      "step": 3145
    },
    {
      "epoch": 0.07402352941176471,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.305,
      "step": 3146
    },
    {
      "epoch": 0.07404705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3554,
      "step": 3147
    },
    {
      "epoch": 0.07407058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4261,
      "step": 3148
    },
    {
      "epoch": 0.07409411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6002,
      "step": 3149
    },
    {
      "epoch": 0.07411764705882352,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5022,
      "step": 3150
    },
    {
      "epoch": 0.07414117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4715,
      "step": 3151
    },
    {
      "epoch": 0.07416470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1711,
      "step": 3152
    },
    {
      "epoch": 0.07418823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6272,
      "step": 3153
    },
    {
      "epoch": 0.07421176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.314,
      "step": 3154
    },
    {
      "epoch": 0.07423529411764705,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.634,
      "step": 3155
    },
    {
      "epoch": 0.07425882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6458,
      "step": 3156
    },
    {
      "epoch": 0.07428235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5754,
      "step": 3157
    },
    {
      "epoch": 0.07430588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4831,
      "step": 3158
    },
    {
      "epoch": 0.07432941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4453,
      "step": 3159
    },
    {
      "epoch": 0.0743529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.5052,
      "step": 3160
    },
    {
      "epoch": 0.0743764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.6214,
      "step": 3161
    },
    {
      "epoch": 0.0744,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6689,
      "step": 3162
    },
    {
      "epoch": 0.07442352941176471,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.5626,
      "step": 3163
    },
    {
      "epoch": 0.07444705882352941,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3058,
      "step": 3164
    },
    {
      "epoch": 0.07447058823529412,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2675,
      "step": 3165
    },
    {
      "epoch": 0.07449411764705882,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5955,
      "step": 3166
    },
    {
      "epoch": 0.07451764705882354,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.4064,
      "step": 3167
    },
    {
      "epoch": 0.07454117647058824,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3706,
      "step": 3168
    },
    {
      "epoch": 0.07456470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3683,
      "step": 3169
    },
    {
      "epoch": 0.07458823529411765,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3849,
      "step": 3170
    },
    {
      "epoch": 0.07461176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6331,
      "step": 3171
    },
    {
      "epoch": 0.07463529411764706,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.4577,
      "step": 3172
    },
    {
      "epoch": 0.07465882352941176,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3686,
      "step": 3173
    },
    {
      "epoch": 0.07468235294117648,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4439,
      "step": 3174
    },
    {
      "epoch": 0.07470588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6298,
      "step": 3175
    },
    {
      "epoch": 0.07472941176470588,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.5029,
      "step": 3176
    },
    {
      "epoch": 0.07475294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.6037,
      "step": 3177
    },
    {
      "epoch": 0.07477647058823529,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4732,
      "step": 3178
    },
    {
      "epoch": 0.0748,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3337,
      "step": 3179
    },
    {
      "epoch": 0.0748235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.5022,
      "step": 3180
    },
    {
      "epoch": 0.0748470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2928,
      "step": 3181
    },
    {
      "epoch": 0.07487058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2945,
      "step": 3182
    },
    {
      "epoch": 0.07489411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.621,
      "step": 3183
    },
    {
      "epoch": 0.07491764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1935,
      "step": 3184
    },
    {
      "epoch": 0.07494117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5839,
      "step": 3185
    },
    {
      "epoch": 0.07496470588235295,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.4223,
      "step": 3186
    },
    {
      "epoch": 0.07498823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5692,
      "step": 3187
    },
    {
      "epoch": 0.07501176470588235,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5474,
      "step": 3188
    },
    {
      "epoch": 0.07503529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4981,
      "step": 3189
    },
    {
      "epoch": 0.07505882352941176,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2305,
      "step": 3190
    },
    {
      "epoch": 0.07508235294117648,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4718,
      "step": 3191
    },
    {
      "epoch": 0.07510588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2643,
      "step": 3192
    },
    {
      "epoch": 0.07512941176470589,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.3164,
      "step": 3193
    },
    {
      "epoch": 0.07515294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.575,
      "step": 3194
    },
    {
      "epoch": 0.07517647058823529,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.5499,
      "step": 3195
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3992,
      "step": 3196
    },
    {
      "epoch": 0.0752235294117647,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4521,
      "step": 3197
    },
    {
      "epoch": 0.07524705882352942,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4311,
      "step": 3198
    },
    {
      "epoch": 0.07527058823529412,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.5552,
      "step": 3199
    },
    {
      "epoch": 0.07529411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3523,
      "step": 3200
    },
    {
      "epoch": 0.07531764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4386,
      "step": 3201
    },
    {
      "epoch": 0.07534117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3983,
      "step": 3202
    },
    {
      "epoch": 0.07536470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4747,
      "step": 3203
    },
    {
      "epoch": 0.07538823529411764,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.8639,
      "step": 3204
    },
    {
      "epoch": 0.07541176470588236,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.413,
      "step": 3205
    },
    {
      "epoch": 0.07543529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.428,
      "step": 3206
    },
    {
      "epoch": 0.07545882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5913,
      "step": 3207
    },
    {
      "epoch": 0.07548235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7188,
      "step": 3208
    },
    {
      "epoch": 0.07550588235294117,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4363,
      "step": 3209
    },
    {
      "epoch": 0.07552941176470589,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3953,
      "step": 3210
    },
    {
      "epoch": 0.07555294117647059,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5278,
      "step": 3211
    },
    {
      "epoch": 0.0755764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4783,
      "step": 3212
    },
    {
      "epoch": 0.0756,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2233,
      "step": 3213
    },
    {
      "epoch": 0.0756235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4532,
      "step": 3214
    },
    {
      "epoch": 0.07564705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.6873,
      "step": 3215
    },
    {
      "epoch": 0.07567058823529411,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6463,
      "step": 3216
    },
    {
      "epoch": 0.07569411764705883,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.529,
      "step": 3217
    },
    {
      "epoch": 0.07571764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.5214,
      "step": 3218
    },
    {
      "epoch": 0.07574117647058823,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2503,
      "step": 3219
    },
    {
      "epoch": 0.07576470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5059,
      "step": 3220
    },
    {
      "epoch": 0.07578823529411764,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.578,
      "step": 3221
    },
    {
      "epoch": 0.07581176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4335,
      "step": 3222
    },
    {
      "epoch": 0.07583529411764706,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3474,
      "step": 3223
    },
    {
      "epoch": 0.07585882352941177,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6597,
      "step": 3224
    },
    {
      "epoch": 0.07588235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1983,
      "step": 3225
    },
    {
      "epoch": 0.07590588235294117,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3343,
      "step": 3226
    },
    {
      "epoch": 0.07592941176470588,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2876,
      "step": 3227
    },
    {
      "epoch": 0.07595294117647058,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5323,
      "step": 3228
    },
    {
      "epoch": 0.0759764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.329,
      "step": 3229
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4034,
      "step": 3230
    },
    {
      "epoch": 0.07602352941176471,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3337,
      "step": 3231
    },
    {
      "epoch": 0.07604705882352941,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3032,
      "step": 3232
    },
    {
      "epoch": 0.07607058823529411,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.4817,
      "step": 3233
    },
    {
      "epoch": 0.07609411764705883,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4649,
      "step": 3234
    },
    {
      "epoch": 0.07611764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.6682,
      "step": 3235
    },
    {
      "epoch": 0.07614117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3806,
      "step": 3236
    },
    {
      "epoch": 0.07616470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4833,
      "step": 3237
    },
    {
      "epoch": 0.07618823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5406,
      "step": 3238
    },
    {
      "epoch": 0.07621176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4517,
      "step": 3239
    },
    {
      "epoch": 0.07623529411764705,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.5196,
      "step": 3240
    },
    {
      "epoch": 0.07625882352941177,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 3241
    },
    {
      "epoch": 0.07628235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2592,
      "step": 3242
    },
    {
      "epoch": 0.07630588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4065,
      "step": 3243
    },
    {
      "epoch": 0.07632941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.6052,
      "step": 3244
    },
    {
      "epoch": 0.07635294117647058,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.277,
      "step": 3245
    },
    {
      "epoch": 0.0763764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5152,
      "step": 3246
    },
    {
      "epoch": 0.0764,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4709,
      "step": 3247
    },
    {
      "epoch": 0.07642352941176471,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4639,
      "step": 3248
    },
    {
      "epoch": 0.07644705882352941,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1088,
      "step": 3249
    },
    {
      "epoch": 0.07647058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2599,
      "step": 3250
    },
    {
      "epoch": 0.07649411764705882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.4614,
      "step": 3251
    },
    {
      "epoch": 0.07651764705882352,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2749,
      "step": 3252
    },
    {
      "epoch": 0.07654117647058824,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.4031,
      "step": 3253
    },
    {
      "epoch": 0.07656470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4591,
      "step": 3254
    },
    {
      "epoch": 0.07658823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4163,
      "step": 3255
    },
    {
      "epoch": 0.07661176470588235,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4553,
      "step": 3256
    },
    {
      "epoch": 0.07663529411764707,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.603,
      "step": 3257
    },
    {
      "epoch": 0.07665882352941177,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2078,
      "step": 3258
    },
    {
      "epoch": 0.07668235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.4169,
      "step": 3259
    },
    {
      "epoch": 0.07670588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.5216,
      "step": 3260
    },
    {
      "epoch": 0.07672941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3995,
      "step": 3261
    },
    {
      "epoch": 0.0767529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5833,
      "step": 3262
    },
    {
      "epoch": 0.0767764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.568,
      "step": 3263
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3163,
      "step": 3264
    },
    {
      "epoch": 0.07682352941176471,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6406,
      "step": 3265
    },
    {
      "epoch": 0.07684705882352941,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2362,
      "step": 3266
    },
    {
      "epoch": 0.07687058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4111,
      "step": 3267
    },
    {
      "epoch": 0.07689411764705882,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1575,
      "step": 3268
    },
    {
      "epoch": 0.07691764705882354,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3639,
      "step": 3269
    },
    {
      "epoch": 0.07694117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6111,
      "step": 3270
    },
    {
      "epoch": 0.07696470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.288,
      "step": 3271
    },
    {
      "epoch": 0.07698823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4855,
      "step": 3272
    },
    {
      "epoch": 0.07701176470588235,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2514,
      "step": 3273
    },
    {
      "epoch": 0.07703529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5455,
      "step": 3274
    },
    {
      "epoch": 0.07705882352941176,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3237,
      "step": 3275
    },
    {
      "epoch": 0.07708235294117648,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6192,
      "step": 3276
    },
    {
      "epoch": 0.07710588235294118,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3695,
      "step": 3277
    },
    {
      "epoch": 0.07712941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4347,
      "step": 3278
    },
    {
      "epoch": 0.07715294117647059,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2175,
      "step": 3279
    },
    {
      "epoch": 0.07717647058823529,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4102,
      "step": 3280
    },
    {
      "epoch": 0.0772,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.5096,
      "step": 3281
    },
    {
      "epoch": 0.0772235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5301,
      "step": 3282
    },
    {
      "epoch": 0.07724705882352942,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2901,
      "step": 3283
    },
    {
      "epoch": 0.07727058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5366,
      "step": 3284
    },
    {
      "epoch": 0.07729411764705882,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.5858,
      "step": 3285
    },
    {
      "epoch": 0.07731764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.7137,
      "step": 3286
    },
    {
      "epoch": 0.07734117647058823,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.4781,
      "step": 3287
    },
    {
      "epoch": 0.07736470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5425,
      "step": 3288
    },
    {
      "epoch": 0.07738823529411765,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5168,
      "step": 3289
    },
    {
      "epoch": 0.07741176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4691,
      "step": 3290
    },
    {
      "epoch": 0.07743529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5786,
      "step": 3291
    },
    {
      "epoch": 0.07745882352941176,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.7252,
      "step": 3292
    },
    {
      "epoch": 0.07748235294117647,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1714,
      "step": 3293
    },
    {
      "epoch": 0.07750588235294117,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5537,
      "step": 3294
    },
    {
      "epoch": 0.07752941176470589,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2212,
      "step": 3295
    },
    {
      "epoch": 0.07755294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6727,
      "step": 3296
    },
    {
      "epoch": 0.07757647058823529,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3651,
      "step": 3297
    },
    {
      "epoch": 0.0776,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4156,
      "step": 3298
    },
    {
      "epoch": 0.0776235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5379,
      "step": 3299
    },
    {
      "epoch": 0.07764705882352942,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4056,
      "step": 3300
    },
    {
      "epoch": 0.07767058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3458,
      "step": 3301
    },
    {
      "epoch": 0.07769411764705883,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5515,
      "step": 3302
    },
    {
      "epoch": 0.07771764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3971,
      "step": 3303
    },
    {
      "epoch": 0.07774117647058823,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4251,
      "step": 3304
    },
    {
      "epoch": 0.07776470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3081,
      "step": 3305
    },
    {
      "epoch": 0.07778823529411764,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5508,
      "step": 3306
    },
    {
      "epoch": 0.07781176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5419,
      "step": 3307
    },
    {
      "epoch": 0.07783529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.734,
      "step": 3308
    },
    {
      "epoch": 0.07785882352941176,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.158,
      "step": 3309
    },
    {
      "epoch": 0.07788235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5018,
      "step": 3310
    },
    {
      "epoch": 0.07790588235294117,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1853,
      "step": 3311
    },
    {
      "epoch": 0.07792941176470589,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.564,
      "step": 3312
    },
    {
      "epoch": 0.07795294117647059,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.4411,
      "step": 3313
    },
    {
      "epoch": 0.0779764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5521,
      "step": 3314
    },
    {
      "epoch": 0.078,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6477,
      "step": 3315
    },
    {
      "epoch": 0.0780235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5502,
      "step": 3316
    },
    {
      "epoch": 0.07804705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.508,
      "step": 3317
    },
    {
      "epoch": 0.07807058823529411,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4499,
      "step": 3318
    },
    {
      "epoch": 0.07809411764705883,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.645,
      "step": 3319
    },
    {
      "epoch": 0.07811764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.4081,
      "step": 3320
    },
    {
      "epoch": 0.07814117647058824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3305,
      "step": 3321
    },
    {
      "epoch": 0.07816470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3893,
      "step": 3322
    },
    {
      "epoch": 0.07818823529411764,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2064,
      "step": 3323
    },
    {
      "epoch": 0.07821176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4192,
      "step": 3324
    },
    {
      "epoch": 0.07823529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4862,
      "step": 3325
    },
    {
      "epoch": 0.07825882352941177,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4738,
      "step": 3326
    },
    {
      "epoch": 0.07828235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3746,
      "step": 3327
    },
    {
      "epoch": 0.07830588235294117,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4167,
      "step": 3328
    },
    {
      "epoch": 0.07832941176470588,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 3329
    },
    {
      "epoch": 0.07835294117647058,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4746,
      "step": 3330
    },
    {
      "epoch": 0.0783764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6707,
      "step": 3331
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3245,
      "step": 3332
    },
    {
      "epoch": 0.07842352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4762,
      "step": 3333
    },
    {
      "epoch": 0.07844705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.488,
      "step": 3334
    },
    {
      "epoch": 0.07847058823529411,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2288,
      "step": 3335
    },
    {
      "epoch": 0.07849411764705883,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4675,
      "step": 3336
    },
    {
      "epoch": 0.07851764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 3337
    },
    {
      "epoch": 0.07854117647058824,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3775,
      "step": 3338
    },
    {
      "epoch": 0.07856470588235294,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2256,
      "step": 3339
    },
    {
      "epoch": 0.07858823529411765,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2303,
      "step": 3340
    },
    {
      "epoch": 0.07861176470588235,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3092,
      "step": 3341
    },
    {
      "epoch": 0.07863529411764705,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3118,
      "step": 3342
    },
    {
      "epoch": 0.07865882352941177,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3685,
      "step": 3343
    },
    {
      "epoch": 0.07868235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5654,
      "step": 3344
    },
    {
      "epoch": 0.07870588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.6085,
      "step": 3345
    },
    {
      "epoch": 0.07872941176470588,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2372,
      "step": 3346
    },
    {
      "epoch": 0.0787529411764706,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.2622,
      "step": 3347
    },
    {
      "epoch": 0.0787764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.4992,
      "step": 3348
    },
    {
      "epoch": 0.0788,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2159,
      "step": 3349
    },
    {
      "epoch": 0.07882352941176471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4387,
      "step": 3350
    },
    {
      "epoch": 0.07884705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.614,
      "step": 3351
    },
    {
      "epoch": 0.07887058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1573,
      "step": 3352
    },
    {
      "epoch": 0.07889411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3232,
      "step": 3353
    },
    {
      "epoch": 0.07891764705882352,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2347,
      "step": 3354
    },
    {
      "epoch": 0.07894117647058824,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2938,
      "step": 3355
    },
    {
      "epoch": 0.07896470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.6805,
      "step": 3356
    },
    {
      "epoch": 0.07898823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6099,
      "step": 3357
    },
    {
      "epoch": 0.07901176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5185,
      "step": 3358
    },
    {
      "epoch": 0.07903529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4069,
      "step": 3359
    },
    {
      "epoch": 0.07905882352941176,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3253,
      "step": 3360
    },
    {
      "epoch": 0.07908235294117646,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5556,
      "step": 3361
    },
    {
      "epoch": 0.07910588235294118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5311,
      "step": 3362
    },
    {
      "epoch": 0.07912941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6211,
      "step": 3363
    },
    {
      "epoch": 0.07915294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5252,
      "step": 3364
    },
    {
      "epoch": 0.07917647058823529,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4884,
      "step": 3365
    },
    {
      "epoch": 0.0792,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1427,
      "step": 3366
    },
    {
      "epoch": 0.0792235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2267,
      "step": 3367
    },
    {
      "epoch": 0.0792470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4096,
      "step": 3368
    },
    {
      "epoch": 0.07927058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4654,
      "step": 3369
    },
    {
      "epoch": 0.07929411764705882,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 3370
    },
    {
      "epoch": 0.07931764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 3371
    },
    {
      "epoch": 0.07934117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5476,
      "step": 3372
    },
    {
      "epoch": 0.07936470588235293,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2308,
      "step": 3373
    },
    {
      "epoch": 0.07938823529411765,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2498,
      "step": 3374
    },
    {
      "epoch": 0.07941176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5682,
      "step": 3375
    },
    {
      "epoch": 0.07943529411764706,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 1.0912,
      "step": 3376
    },
    {
      "epoch": 0.07945882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6998,
      "step": 3377
    },
    {
      "epoch": 0.07948235294117648,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3134,
      "step": 3378
    },
    {
      "epoch": 0.07950588235294118,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3327,
      "step": 3379
    },
    {
      "epoch": 0.07952941176470588,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2661,
      "step": 3380
    },
    {
      "epoch": 0.07955294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3834,
      "step": 3381
    },
    {
      "epoch": 0.07957647058823529,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.38,
      "step": 3382
    },
    {
      "epoch": 0.0796,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4526,
      "step": 3383
    },
    {
      "epoch": 0.0796235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3758,
      "step": 3384
    },
    {
      "epoch": 0.07964705882352942,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5298,
      "step": 3385
    },
    {
      "epoch": 0.07967058823529412,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2365,
      "step": 3386
    },
    {
      "epoch": 0.07969411764705882,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2323,
      "step": 3387
    },
    {
      "epoch": 0.07971764705882353,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2493,
      "step": 3388
    },
    {
      "epoch": 0.07974117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.615,
      "step": 3389
    },
    {
      "epoch": 0.07976470588235295,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4763,
      "step": 3390
    },
    {
      "epoch": 0.07978823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4885,
      "step": 3391
    },
    {
      "epoch": 0.07981176470588236,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4427,
      "step": 3392
    },
    {
      "epoch": 0.07983529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5823,
      "step": 3393
    },
    {
      "epoch": 0.07985882352941176,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3037,
      "step": 3394
    },
    {
      "epoch": 0.07988235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3441,
      "step": 3395
    },
    {
      "epoch": 0.07990588235294117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3713,
      "step": 3396
    },
    {
      "epoch": 0.07992941176470589,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5041,
      "step": 3397
    },
    {
      "epoch": 0.07995294117647059,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1925,
      "step": 3398
    },
    {
      "epoch": 0.07997647058823529,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4141,
      "step": 3399
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4185,
      "step": 3400
    },
    {
      "epoch": 0.0800235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3265,
      "step": 3401
    },
    {
      "epoch": 0.08004705882352942,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4447,
      "step": 3402
    },
    {
      "epoch": 0.08007058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4633,
      "step": 3403
    },
    {
      "epoch": 0.08009411764705883,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3229,
      "step": 3404
    },
    {
      "epoch": 0.08011764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5838,
      "step": 3405
    },
    {
      "epoch": 0.08014117647058823,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5638,
      "step": 3406
    },
    {
      "epoch": 0.08016470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3843,
      "step": 3407
    },
    {
      "epoch": 0.08018823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2729,
      "step": 3408
    },
    {
      "epoch": 0.08021176470588236,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3752,
      "step": 3409
    },
    {
      "epoch": 0.08023529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3325,
      "step": 3410
    },
    {
      "epoch": 0.08025882352941177,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4778,
      "step": 3411
    },
    {
      "epoch": 0.08028235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3312,
      "step": 3412
    },
    {
      "epoch": 0.08030588235294117,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2947,
      "step": 3413
    },
    {
      "epoch": 0.08032941176470589,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3187,
      "step": 3414
    },
    {
      "epoch": 0.08035294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.499,
      "step": 3415
    },
    {
      "epoch": 0.0803764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1749,
      "step": 3416
    },
    {
      "epoch": 0.0804,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2988,
      "step": 3417
    },
    {
      "epoch": 0.0804235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.4304,
      "step": 3418
    },
    {
      "epoch": 0.08044705882352941,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3887,
      "step": 3419
    },
    {
      "epoch": 0.08047058823529411,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.3284,
      "step": 3420
    },
    {
      "epoch": 0.08049411764705883,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2542,
      "step": 3421
    },
    {
      "epoch": 0.08051764705882353,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 1.1075,
      "step": 3422
    },
    {
      "epoch": 0.08054117647058824,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2947,
      "step": 3423
    },
    {
      "epoch": 0.08056470588235294,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.188,
      "step": 3424
    },
    {
      "epoch": 0.08058823529411764,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.6481,
      "step": 3425
    },
    {
      "epoch": 0.08061176470588235,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5723,
      "step": 3426
    },
    {
      "epoch": 0.08063529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5134,
      "step": 3427
    },
    {
      "epoch": 0.08065882352941177,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.5026,
      "step": 3428
    },
    {
      "epoch": 0.08068235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3747,
      "step": 3429
    },
    {
      "epoch": 0.08070588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4649,
      "step": 3430
    },
    {
      "epoch": 0.08072941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4892,
      "step": 3431
    },
    {
      "epoch": 0.08075294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5444,
      "step": 3432
    },
    {
      "epoch": 0.0807764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4931,
      "step": 3433
    },
    {
      "epoch": 0.0808,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3487,
      "step": 3434
    },
    {
      "epoch": 0.08082352941176471,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6582,
      "step": 3435
    },
    {
      "epoch": 0.08084705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4487,
      "step": 3436
    },
    {
      "epoch": 0.08087058823529411,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1374,
      "step": 3437
    },
    {
      "epoch": 0.08089411764705882,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3924,
      "step": 3438
    },
    {
      "epoch": 0.08091764705882352,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4411,
      "step": 3439
    },
    {
      "epoch": 0.08094117647058824,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.5068,
      "step": 3440
    },
    {
      "epoch": 0.08096470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.241,
      "step": 3441
    },
    {
      "epoch": 0.08098823529411765,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3238,
      "step": 3442
    },
    {
      "epoch": 0.08101176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3589,
      "step": 3443
    },
    {
      "epoch": 0.08103529411764705,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.5133,
      "step": 3444
    },
    {
      "epoch": 0.08105882352941177,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 3445
    },
    {
      "epoch": 0.08108235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4423,
      "step": 3446
    },
    {
      "epoch": 0.08110588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4647,
      "step": 3447
    },
    {
      "epoch": 0.08112941176470588,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.2797,
      "step": 3448
    },
    {
      "epoch": 0.0811529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6465,
      "step": 3449
    },
    {
      "epoch": 0.0811764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2745,
      "step": 3450
    },
    {
      "epoch": 0.0812,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.539,
      "step": 3451
    },
    {
      "epoch": 0.08122352941176471,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2488,
      "step": 3452
    },
    {
      "epoch": 0.08124705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5127,
      "step": 3453
    },
    {
      "epoch": 0.08127058823529412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3055,
      "step": 3454
    },
    {
      "epoch": 0.08129411764705882,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 3455
    },
    {
      "epoch": 0.08131764705882354,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4522,
      "step": 3456
    },
    {
      "epoch": 0.08134117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 3457
    },
    {
      "epoch": 0.08136470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5098,
      "step": 3458
    },
    {
      "epoch": 0.08138823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.577,
      "step": 3459
    },
    {
      "epoch": 0.08141176470588235,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3266,
      "step": 3460
    },
    {
      "epoch": 0.08143529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4336,
      "step": 3461
    },
    {
      "epoch": 0.08145882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4515,
      "step": 3462
    },
    {
      "epoch": 0.08148235294117646,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2026,
      "step": 3463
    },
    {
      "epoch": 0.08150588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5929,
      "step": 3464
    },
    {
      "epoch": 0.08152941176470588,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1113,
      "step": 3465
    },
    {
      "epoch": 0.08155294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3304,
      "step": 3466
    },
    {
      "epoch": 0.08157647058823529,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.4042,
      "step": 3467
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4713,
      "step": 3468
    },
    {
      "epoch": 0.0816235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5133,
      "step": 3469
    },
    {
      "epoch": 0.0816470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3913,
      "step": 3470
    },
    {
      "epoch": 0.08167058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.526,
      "step": 3471
    },
    {
      "epoch": 0.08169411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.4421,
      "step": 3472
    },
    {
      "epoch": 0.08171764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5372,
      "step": 3473
    },
    {
      "epoch": 0.08174117647058823,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1965,
      "step": 3474
    },
    {
      "epoch": 0.08176470588235295,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3059,
      "step": 3475
    },
    {
      "epoch": 0.08178823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4995,
      "step": 3476
    },
    {
      "epoch": 0.08181176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4826,
      "step": 3477
    },
    {
      "epoch": 0.08183529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2651,
      "step": 3478
    },
    {
      "epoch": 0.08185882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5776,
      "step": 3479
    },
    {
      "epoch": 0.08188235294117648,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4566,
      "step": 3480
    },
    {
      "epoch": 0.08190588235294118,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1908,
      "step": 3481
    },
    {
      "epoch": 0.08192941176470588,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3692,
      "step": 3482
    },
    {
      "epoch": 0.08195294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3464,
      "step": 3483
    },
    {
      "epoch": 0.08197647058823529,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3471,
      "step": 3484
    },
    {
      "epoch": 0.082,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6082,
      "step": 3485
    },
    {
      "epoch": 0.0820235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3567,
      "step": 3486
    },
    {
      "epoch": 0.08204705882352942,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5277,
      "step": 3487
    },
    {
      "epoch": 0.08207058823529412,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3093,
      "step": 3488
    },
    {
      "epoch": 0.08209411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5058,
      "step": 3489
    },
    {
      "epoch": 0.08211764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.432,
      "step": 3490
    },
    {
      "epoch": 0.08214117647058823,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3303,
      "step": 3491
    },
    {
      "epoch": 0.08216470588235295,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2651,
      "step": 3492
    },
    {
      "epoch": 0.08218823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5387,
      "step": 3493
    },
    {
      "epoch": 0.08221176470588236,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2075,
      "step": 3494
    },
    {
      "epoch": 0.08223529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5436,
      "step": 3495
    },
    {
      "epoch": 0.08225882352941176,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2598,
      "step": 3496
    },
    {
      "epoch": 0.08228235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5012,
      "step": 3497
    },
    {
      "epoch": 0.08230588235294117,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.464,
      "step": 3498
    },
    {
      "epoch": 0.08232941176470589,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3667,
      "step": 3499
    },
    {
      "epoch": 0.08235294117647059,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5279,
      "step": 3500
    },
    {
      "epoch": 0.0823764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3635,
      "step": 3501
    },
    {
      "epoch": 0.0824,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4032,
      "step": 3502
    },
    {
      "epoch": 0.0824235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3973,
      "step": 3503
    },
    {
      "epoch": 0.08244705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3753,
      "step": 3504
    },
    {
      "epoch": 0.08247058823529411,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3696,
      "step": 3505
    },
    {
      "epoch": 0.08249411764705883,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4164,
      "step": 3506
    },
    {
      "epoch": 0.08251764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2223,
      "step": 3507
    },
    {
      "epoch": 0.08254117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5089,
      "step": 3508
    },
    {
      "epoch": 0.08256470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1958,
      "step": 3509
    },
    {
      "epoch": 0.08258823529411764,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4334,
      "step": 3510
    },
    {
      "epoch": 0.08261176470588236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4658,
      "step": 3511
    },
    {
      "epoch": 0.08263529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3201,
      "step": 3512
    },
    {
      "epoch": 0.08265882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5541,
      "step": 3513
    },
    {
      "epoch": 0.08268235294117647,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1412,
      "step": 3514
    },
    {
      "epoch": 0.08270588235294117,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.554,
      "step": 3515
    },
    {
      "epoch": 0.08272941176470588,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.4457,
      "step": 3516
    },
    {
      "epoch": 0.08275294117647058,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5521,
      "step": 3517
    },
    {
      "epoch": 0.0827764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.557,
      "step": 3518
    },
    {
      "epoch": 0.0828,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4411,
      "step": 3519
    },
    {
      "epoch": 0.08282352941176471,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4297,
      "step": 3520
    },
    {
      "epoch": 0.08284705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3112,
      "step": 3521
    },
    {
      "epoch": 0.08287058823529411,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2156,
      "step": 3522
    },
    {
      "epoch": 0.08289411764705883,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3849,
      "step": 3523
    },
    {
      "epoch": 0.08291764705882353,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0023,
      "step": 3524
    },
    {
      "epoch": 0.08294117647058824,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4033,
      "step": 3525
    },
    {
      "epoch": 0.08296470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3995,
      "step": 3526
    },
    {
      "epoch": 0.08298823529411764,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4203,
      "step": 3527
    },
    {
      "epoch": 0.08301176470588235,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.1742,
      "step": 3528
    },
    {
      "epoch": 0.08303529411764705,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4062,
      "step": 3529
    },
    {
      "epoch": 0.08305882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5096,
      "step": 3530
    },
    {
      "epoch": 0.08308235294117647,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.5208,
      "step": 3531
    },
    {
      "epoch": 0.08310588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5993,
      "step": 3532
    },
    {
      "epoch": 0.08312941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2063,
      "step": 3533
    },
    {
      "epoch": 0.08315294117647058,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.3356,
      "step": 3534
    },
    {
      "epoch": 0.0831764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.6359,
      "step": 3535
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.7033,
      "step": 3536
    },
    {
      "epoch": 0.08322352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4026,
      "step": 3537
    },
    {
      "epoch": 0.08324705882352941,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.291,
      "step": 3538
    },
    {
      "epoch": 0.08327058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.428,
      "step": 3539
    },
    {
      "epoch": 0.08329411764705882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.4304,
      "step": 3540
    },
    {
      "epoch": 0.08331764705882352,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6129,
      "step": 3541
    },
    {
      "epoch": 0.08334117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2646,
      "step": 3542
    },
    {
      "epoch": 0.08336470588235294,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1103,
      "step": 3543
    },
    {
      "epoch": 0.08338823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3992,
      "step": 3544
    },
    {
      "epoch": 0.08341176470588235,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1899,
      "step": 3545
    },
    {
      "epoch": 0.08343529411764705,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4984,
      "step": 3546
    },
    {
      "epoch": 0.08345882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.624,
      "step": 3547
    },
    {
      "epoch": 0.08348235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3134,
      "step": 3548
    },
    {
      "epoch": 0.08350588235294118,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.4133,
      "step": 3549
    },
    {
      "epoch": 0.08352941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3931,
      "step": 3550
    },
    {
      "epoch": 0.0835529411764706,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.4287,
      "step": 3551
    },
    {
      "epoch": 0.0835764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.6502,
      "step": 3552
    },
    {
      "epoch": 0.0836,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2099,
      "step": 3553
    },
    {
      "epoch": 0.08362352941176471,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.4693,
      "step": 3554
    },
    {
      "epoch": 0.08364705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5794,
      "step": 3555
    },
    {
      "epoch": 0.08367058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.4317,
      "step": 3556
    },
    {
      "epoch": 0.08369411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.552,
      "step": 3557
    },
    {
      "epoch": 0.08371764705882354,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.4623,
      "step": 3558
    },
    {
      "epoch": 0.08374117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5024,
      "step": 3559
    },
    {
      "epoch": 0.08376470588235294,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2536,
      "step": 3560
    },
    {
      "epoch": 0.08378823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2871,
      "step": 3561
    },
    {
      "epoch": 0.08381176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.336,
      "step": 3562
    },
    {
      "epoch": 0.08383529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.5385,
      "step": 3563
    },
    {
      "epoch": 0.08385882352941176,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2554,
      "step": 3564
    },
    {
      "epoch": 0.08388235294117648,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2181,
      "step": 3565
    },
    {
      "epoch": 0.08390588235294118,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.3536,
      "step": 3566
    },
    {
      "epoch": 0.08392941176470588,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.9547,
      "step": 3567
    },
    {
      "epoch": 0.08395294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6346,
      "step": 3568
    },
    {
      "epoch": 0.08397647058823529,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3974,
      "step": 3569
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 1.1059,
      "step": 3570
    },
    {
      "epoch": 0.0840235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4769,
      "step": 3571
    },
    {
      "epoch": 0.0840470588235294,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.4089,
      "step": 3572
    },
    {
      "epoch": 0.08407058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.6489,
      "step": 3573
    },
    {
      "epoch": 0.08409411764705882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2774,
      "step": 3574
    },
    {
      "epoch": 0.08411764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.6142,
      "step": 3575
    },
    {
      "epoch": 0.08414117647058823,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4495,
      "step": 3576
    },
    {
      "epoch": 0.08416470588235295,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.4377,
      "step": 3577
    },
    {
      "epoch": 0.08418823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3709,
      "step": 3578
    },
    {
      "epoch": 0.08421176470588235,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3378,
      "step": 3579
    },
    {
      "epoch": 0.08423529411764706,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 1.0326,
      "step": 3580
    },
    {
      "epoch": 0.08425882352941176,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1697,
      "step": 3581
    },
    {
      "epoch": 0.08428235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4047,
      "step": 3582
    },
    {
      "epoch": 0.08430588235294117,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1314,
      "step": 3583
    },
    {
      "epoch": 0.08432941176470589,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.5646,
      "step": 3584
    },
    {
      "epoch": 0.08435294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2827,
      "step": 3585
    },
    {
      "epoch": 0.08437647058823529,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.255,
      "step": 3586
    },
    {
      "epoch": 0.0844,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3843,
      "step": 3587
    },
    {
      "epoch": 0.0844235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3397,
      "step": 3588
    },
    {
      "epoch": 0.08444705882352942,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5174,
      "step": 3589
    },
    {
      "epoch": 0.08447058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.242,
      "step": 3590
    },
    {
      "epoch": 0.08449411764705882,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4984,
      "step": 3591
    },
    {
      "epoch": 0.08451764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6232,
      "step": 3592
    },
    {
      "epoch": 0.08454117647058823,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.4057,
      "step": 3593
    },
    {
      "epoch": 0.08456470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3029,
      "step": 3594
    },
    {
      "epoch": 0.08458823529411764,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1939,
      "step": 3595
    },
    {
      "epoch": 0.08461176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2118,
      "step": 3596
    },
    {
      "epoch": 0.08463529411764706,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9865,
      "step": 3597
    },
    {
      "epoch": 0.08465882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3213,
      "step": 3598
    },
    {
      "epoch": 0.08468235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.5676,
      "step": 3599
    },
    {
      "epoch": 0.08470588235294117,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.6395,
      "step": 3600
    },
    {
      "epoch": 0.08472941176470589,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 1.2301,
      "step": 3601
    },
    {
      "epoch": 0.08475294117647059,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.5236,
      "step": 3602
    },
    {
      "epoch": 0.0847764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.5189,
      "step": 3603
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.6573,
      "step": 3604
    },
    {
      "epoch": 0.0848235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.556,
      "step": 3605
    },
    {
      "epoch": 0.08484705882352941,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4704,
      "step": 3606
    },
    {
      "epoch": 0.08487058823529411,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2078,
      "step": 3607
    },
    {
      "epoch": 0.08489411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4991,
      "step": 3608
    },
    {
      "epoch": 0.08491764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4952,
      "step": 3609
    },
    {
      "epoch": 0.08494117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.308,
      "step": 3610
    },
    {
      "epoch": 0.08496470588235294,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3282,
      "step": 3611
    },
    {
      "epoch": 0.08498823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2656,
      "step": 3612
    },
    {
      "epoch": 0.08501176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2974,
      "step": 3613
    },
    {
      "epoch": 0.08503529411764706,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3512,
      "step": 3614
    },
    {
      "epoch": 0.08505882352941177,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2643,
      "step": 3615
    },
    {
      "epoch": 0.08508235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6791,
      "step": 3616
    },
    {
      "epoch": 0.08510588235294117,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4217,
      "step": 3617
    },
    {
      "epoch": 0.08512941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3035,
      "step": 3618
    },
    {
      "epoch": 0.08515294117647058,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3178,
      "step": 3619
    },
    {
      "epoch": 0.0851764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3735,
      "step": 3620
    },
    {
      "epoch": 0.0852,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5329,
      "step": 3621
    },
    {
      "epoch": 0.08522352941176471,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3239,
      "step": 3622
    },
    {
      "epoch": 0.08524705882352941,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3464,
      "step": 3623
    },
    {
      "epoch": 0.08527058823529411,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2629,
      "step": 3624
    },
    {
      "epoch": 0.08529411764705883,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.554,
      "step": 3625
    },
    {
      "epoch": 0.08531764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.3162,
      "step": 3626
    },
    {
      "epoch": 0.08534117647058824,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3825,
      "step": 3627
    },
    {
      "epoch": 0.08536470588235294,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.229,
      "step": 3628
    },
    {
      "epoch": 0.08538823529411765,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3048,
      "step": 3629
    },
    {
      "epoch": 0.08541176470588235,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3599,
      "step": 3630
    },
    {
      "epoch": 0.08543529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2856,
      "step": 3631
    },
    {
      "epoch": 0.08545882352941177,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3892,
      "step": 3632
    },
    {
      "epoch": 0.08548235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3175,
      "step": 3633
    },
    {
      "epoch": 0.08550588235294118,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5825,
      "step": 3634
    },
    {
      "epoch": 0.08552941176470588,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.4156,
      "step": 3635
    },
    {
      "epoch": 0.08555294117647058,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.5903,
      "step": 3636
    },
    {
      "epoch": 0.0855764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.535,
      "step": 3637
    },
    {
      "epoch": 0.0856,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3623,
      "step": 3638
    },
    {
      "epoch": 0.08562352941176471,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.5087,
      "step": 3639
    },
    {
      "epoch": 0.08564705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4316,
      "step": 3640
    },
    {
      "epoch": 0.08567058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.7218,
      "step": 3641
    },
    {
      "epoch": 0.08569411764705882,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0369,
      "step": 3642
    },
    {
      "epoch": 0.08571764705882352,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3101,
      "step": 3643
    },
    {
      "epoch": 0.08574117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4646,
      "step": 3644
    },
    {
      "epoch": 0.08576470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4663,
      "step": 3645
    },
    {
      "epoch": 0.08578823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5607,
      "step": 3646
    },
    {
      "epoch": 0.08581176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4514,
      "step": 3647
    },
    {
      "epoch": 0.08583529411764707,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.5513,
      "step": 3648
    },
    {
      "epoch": 0.08585882352941177,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4534,
      "step": 3649
    },
    {
      "epoch": 0.08588235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3119,
      "step": 3650
    },
    {
      "epoch": 0.08590588235294118,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6107,
      "step": 3651
    },
    {
      "epoch": 0.08592941176470588,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2649,
      "step": 3652
    },
    {
      "epoch": 0.0859529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5635,
      "step": 3653
    },
    {
      "epoch": 0.0859764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2513,
      "step": 3654
    },
    {
      "epoch": 0.086,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4216,
      "step": 3655
    },
    {
      "epoch": 0.08602352941176471,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.7263,
      "step": 3656
    },
    {
      "epoch": 0.08604705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3139,
      "step": 3657
    },
    {
      "epoch": 0.08607058823529412,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.5184,
      "step": 3658
    },
    {
      "epoch": 0.08609411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3665,
      "step": 3659
    },
    {
      "epoch": 0.08611764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5659,
      "step": 3660
    },
    {
      "epoch": 0.08614117647058823,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.528,
      "step": 3661
    },
    {
      "epoch": 0.08616470588235293,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1063,
      "step": 3662
    },
    {
      "epoch": 0.08618823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2178,
      "step": 3663
    },
    {
      "epoch": 0.08621176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4432,
      "step": 3664
    },
    {
      "epoch": 0.08623529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3425,
      "step": 3665
    },
    {
      "epoch": 0.08625882352941176,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.2113,
      "step": 3666
    },
    {
      "epoch": 0.08628235294117648,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4296,
      "step": 3667
    },
    {
      "epoch": 0.08630588235294118,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3742,
      "step": 3668
    },
    {
      "epoch": 0.08632941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2978,
      "step": 3669
    },
    {
      "epoch": 0.08635294117647059,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4006,
      "step": 3670
    },
    {
      "epoch": 0.08637647058823529,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2136,
      "step": 3671
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3638,
      "step": 3672
    },
    {
      "epoch": 0.0864235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2514,
      "step": 3673
    },
    {
      "epoch": 0.08644705882352942,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3827,
      "step": 3674
    },
    {
      "epoch": 0.08647058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4242,
      "step": 3675
    },
    {
      "epoch": 0.08649411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2825,
      "step": 3676
    },
    {
      "epoch": 0.08651764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2722,
      "step": 3677
    },
    {
      "epoch": 0.08654117647058823,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4401,
      "step": 3678
    },
    {
      "epoch": 0.08656470588235295,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.6128,
      "step": 3679
    },
    {
      "epoch": 0.08658823529411765,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1999,
      "step": 3680
    },
    {
      "epoch": 0.08661176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2274,
      "step": 3681
    },
    {
      "epoch": 0.08663529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.5795,
      "step": 3682
    },
    {
      "epoch": 0.08665882352941176,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2653,
      "step": 3683
    },
    {
      "epoch": 0.08668235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4854,
      "step": 3684
    },
    {
      "epoch": 0.08670588235294117,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3511,
      "step": 3685
    },
    {
      "epoch": 0.08672941176470589,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3818,
      "step": 3686
    },
    {
      "epoch": 0.08675294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2704,
      "step": 3687
    },
    {
      "epoch": 0.08677647058823529,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2636,
      "step": 3688
    },
    {
      "epoch": 0.0868,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5298,
      "step": 3689
    },
    {
      "epoch": 0.0868235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4205,
      "step": 3690
    },
    {
      "epoch": 0.08684705882352942,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4033,
      "step": 3691
    },
    {
      "epoch": 0.08687058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3322,
      "step": 3692
    },
    {
      "epoch": 0.08689411764705883,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.322,
      "step": 3693
    },
    {
      "epoch": 0.08691764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3522,
      "step": 3694
    },
    {
      "epoch": 0.08694117647058823,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5582,
      "step": 3695
    },
    {
      "epoch": 0.08696470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.7104,
      "step": 3696
    },
    {
      "epoch": 0.08698823529411764,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3979,
      "step": 3697
    },
    {
      "epoch": 0.08701176470588236,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2897,
      "step": 3698
    },
    {
      "epoch": 0.08703529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3673,
      "step": 3699
    },
    {
      "epoch": 0.08705882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3366,
      "step": 3700
    },
    {
      "epoch": 0.08708235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.554,
      "step": 3701
    },
    {
      "epoch": 0.08710588235294117,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2916,
      "step": 3702
    },
    {
      "epoch": 0.08712941176470589,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.4955,
      "step": 3703
    },
    {
      "epoch": 0.08715294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.714,
      "step": 3704
    },
    {
      "epoch": 0.0871764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2497,
      "step": 3705
    },
    {
      "epoch": 0.0872,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.5399,
      "step": 3706
    },
    {
      "epoch": 0.0872235294117647,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1702,
      "step": 3707
    },
    {
      "epoch": 0.08724705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4175,
      "step": 3708
    },
    {
      "epoch": 0.08727058823529411,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3609,
      "step": 3709
    },
    {
      "epoch": 0.08729411764705883,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.302,
      "step": 3710
    },
    {
      "epoch": 0.08731764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.6191,
      "step": 3711
    },
    {
      "epoch": 0.08734117647058824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4343,
      "step": 3712
    },
    {
      "epoch": 0.08736470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5301,
      "step": 3713
    },
    {
      "epoch": 0.08738823529411764,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3866,
      "step": 3714
    },
    {
      "epoch": 0.08741176470588236,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.414,
      "step": 3715
    },
    {
      "epoch": 0.08743529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3252,
      "step": 3716
    },
    {
      "epoch": 0.08745882352941177,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3797,
      "step": 3717
    },
    {
      "epoch": 0.08748235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.5651,
      "step": 3718
    },
    {
      "epoch": 0.08750588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2904,
      "step": 3719
    },
    {
      "epoch": 0.08752941176470588,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.2368,
      "step": 3720
    },
    {
      "epoch": 0.08755294117647058,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2669,
      "step": 3721
    },
    {
      "epoch": 0.0875764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5924,
      "step": 3722
    },
    {
      "epoch": 0.0876,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4391,
      "step": 3723
    },
    {
      "epoch": 0.08762352941176471,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3289,
      "step": 3724
    },
    {
      "epoch": 0.08764705882352941,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1213,
      "step": 3725
    },
    {
      "epoch": 0.08767058823529411,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.816,
      "step": 3726
    },
    {
      "epoch": 0.08769411764705883,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3568,
      "step": 3727
    },
    {
      "epoch": 0.08771764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4442,
      "step": 3728
    },
    {
      "epoch": 0.08774117647058824,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2856,
      "step": 3729
    },
    {
      "epoch": 0.08776470588235294,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3621,
      "step": 3730
    },
    {
      "epoch": 0.08778823529411765,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2362,
      "step": 3731
    },
    {
      "epoch": 0.08781176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.34,
      "step": 3732
    },
    {
      "epoch": 0.08783529411764705,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2459,
      "step": 3733
    },
    {
      "epoch": 0.08785882352941177,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.4392,
      "step": 3734
    },
    {
      "epoch": 0.08788235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3632,
      "step": 3735
    },
    {
      "epoch": 0.08790588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.372,
      "step": 3736
    },
    {
      "epoch": 0.08792941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4577,
      "step": 3737
    },
    {
      "epoch": 0.0879529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4487,
      "step": 3738
    },
    {
      "epoch": 0.0879764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1823,
      "step": 3739
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3954,
      "step": 3740
    },
    {
      "epoch": 0.08802352941176471,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.6559,
      "step": 3741
    },
    {
      "epoch": 0.08804705882352941,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3858,
      "step": 3742
    },
    {
      "epoch": 0.08807058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5021,
      "step": 3743
    },
    {
      "epoch": 0.08809411764705882,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5009,
      "step": 3744
    },
    {
      "epoch": 0.08811764705882352,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4344,
      "step": 3745
    },
    {
      "epoch": 0.08814117647058824,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3021,
      "step": 3746
    },
    {
      "epoch": 0.08816470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4081,
      "step": 3747
    },
    {
      "epoch": 0.08818823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.5671,
      "step": 3748
    },
    {
      "epoch": 0.08821176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4049,
      "step": 3749
    },
    {
      "epoch": 0.08823529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4711,
      "step": 3750
    },
    {
      "epoch": 0.08825882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3224,
      "step": 3751
    },
    {
      "epoch": 0.08828235294117646,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.437,
      "step": 3752
    },
    {
      "epoch": 0.08830588235294118,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5199,
      "step": 3753
    },
    {
      "epoch": 0.08832941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3902,
      "step": 3754
    },
    {
      "epoch": 0.08835294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5328,
      "step": 3755
    },
    {
      "epoch": 0.08837647058823529,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4534,
      "step": 3756
    },
    {
      "epoch": 0.0884,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6264,
      "step": 3757
    },
    {
      "epoch": 0.0884235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9784,
      "step": 3758
    },
    {
      "epoch": 0.0884470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5538,
      "step": 3759
    },
    {
      "epoch": 0.08847058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5603,
      "step": 3760
    },
    {
      "epoch": 0.08849411764705882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.486,
      "step": 3761
    },
    {
      "epoch": 0.08851764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4039,
      "step": 3762
    },
    {
      "epoch": 0.08854117647058823,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3645,
      "step": 3763
    },
    {
      "epoch": 0.08856470588235293,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2865,
      "step": 3764
    },
    {
      "epoch": 0.08858823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4585,
      "step": 3765
    },
    {
      "epoch": 0.08861176470588235,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9395,
      "step": 3766
    },
    {
      "epoch": 0.08863529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3285,
      "step": 3767
    },
    {
      "epoch": 0.08865882352941176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3679,
      "step": 3768
    },
    {
      "epoch": 0.08868235294117648,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4953,
      "step": 3769
    },
    {
      "epoch": 0.08870588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4511,
      "step": 3770
    },
    {
      "epoch": 0.08872941176470588,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.471,
      "step": 3771
    },
    {
      "epoch": 0.08875294117647059,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3205,
      "step": 3772
    },
    {
      "epoch": 0.08877647058823529,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.456,
      "step": 3773
    },
    {
      "epoch": 0.0888,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1986,
      "step": 3774
    },
    {
      "epoch": 0.0888235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.337,
      "step": 3775
    },
    {
      "epoch": 0.08884705882352942,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6449,
      "step": 3776
    },
    {
      "epoch": 0.08887058823529412,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4609,
      "step": 3777
    },
    {
      "epoch": 0.08889411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2045,
      "step": 3778
    },
    {
      "epoch": 0.08891764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5196,
      "step": 3779
    },
    {
      "epoch": 0.08894117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2701,
      "step": 3780
    },
    {
      "epoch": 0.08896470588235295,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.357,
      "step": 3781
    },
    {
      "epoch": 0.08898823529411765,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.332,
      "step": 3782
    },
    {
      "epoch": 0.08901176470588236,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4961,
      "step": 3783
    },
    {
      "epoch": 0.08903529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.5313,
      "step": 3784
    },
    {
      "epoch": 0.08905882352941176,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6119,
      "step": 3785
    },
    {
      "epoch": 0.08908235294117647,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 1.1529,
      "step": 3786
    },
    {
      "epoch": 0.08910588235294117,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2681,
      "step": 3787
    },
    {
      "epoch": 0.08912941176470589,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.5864,
      "step": 3788
    },
    {
      "epoch": 0.08915294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.515,
      "step": 3789
    },
    {
      "epoch": 0.08917647058823529,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2141,
      "step": 3790
    },
    {
      "epoch": 0.0892,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2412,
      "step": 3791
    },
    {
      "epoch": 0.0892235294117647,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.2786,
      "step": 3792
    },
    {
      "epoch": 0.08924705882352942,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3499,
      "step": 3793
    },
    {
      "epoch": 0.08927058823529412,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 3794
    },
    {
      "epoch": 0.08929411764705883,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4281,
      "step": 3795
    },
    {
      "epoch": 0.08931764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3766,
      "step": 3796
    },
    {
      "epoch": 0.08934117647058823,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.276,
      "step": 3797
    },
    {
      "epoch": 0.08936470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.265,
      "step": 3798
    },
    {
      "epoch": 0.08938823529411764,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4199,
      "step": 3799
    },
    {
      "epoch": 0.08941176470588236,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3001,
      "step": 3800
    },
    {
      "epoch": 0.08943529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3975,
      "step": 3801
    },
    {
      "epoch": 0.08945882352941177,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2631,
      "step": 3802
    },
    {
      "epoch": 0.08948235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2153,
      "step": 3803
    },
    {
      "epoch": 0.08950588235294117,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2662,
      "step": 3804
    },
    {
      "epoch": 0.08952941176470589,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4943,
      "step": 3805
    },
    {
      "epoch": 0.08955294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4267,
      "step": 3806
    },
    {
      "epoch": 0.0895764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1844,
      "step": 3807
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3844,
      "step": 3808
    },
    {
      "epoch": 0.0896235294117647,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 3809
    },
    {
      "epoch": 0.08964705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.535,
      "step": 3810
    },
    {
      "epoch": 0.08967058823529411,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2301,
      "step": 3811
    },
    {
      "epoch": 0.08969411764705883,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3694,
      "step": 3812
    },
    {
      "epoch": 0.08971764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 3813
    },
    {
      "epoch": 0.08974117647058824,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.509,
      "step": 3814
    },
    {
      "epoch": 0.08976470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4254,
      "step": 3815
    },
    {
      "epoch": 0.08978823529411764,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4947,
      "step": 3816
    },
    {
      "epoch": 0.08981176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3094,
      "step": 3817
    },
    {
      "epoch": 0.08983529411764705,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.6437,
      "step": 3818
    },
    {
      "epoch": 0.08985882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4658,
      "step": 3819
    },
    {
      "epoch": 0.08988235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1684,
      "step": 3820
    },
    {
      "epoch": 0.08990588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5021,
      "step": 3821
    },
    {
      "epoch": 0.08992941176470588,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2012,
      "step": 3822
    },
    {
      "epoch": 0.08995294117647058,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4313,
      "step": 3823
    },
    {
      "epoch": 0.0899764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4812,
      "step": 3824
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2596,
      "step": 3825
    },
    {
      "epoch": 0.09002352941176471,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4374,
      "step": 3826
    },
    {
      "epoch": 0.09004705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.5625,
      "step": 3827
    },
    {
      "epoch": 0.09007058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4448,
      "step": 3828
    },
    {
      "epoch": 0.09009411764705882,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3517,
      "step": 3829
    },
    {
      "epoch": 0.09011764705882352,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1787,
      "step": 3830
    },
    {
      "epoch": 0.09014117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3842,
      "step": 3831
    },
    {
      "epoch": 0.09016470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2655,
      "step": 3832
    },
    {
      "epoch": 0.09018823529411765,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3425,
      "step": 3833
    },
    {
      "epoch": 0.09021176470588235,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4611,
      "step": 3834
    },
    {
      "epoch": 0.09023529411764705,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.543,
      "step": 3835
    },
    {
      "epoch": 0.09025882352941177,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3613,
      "step": 3836
    },
    {
      "epoch": 0.09028235294117647,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9595,
      "step": 3837
    },
    {
      "epoch": 0.09030588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5275,
      "step": 3838
    },
    {
      "epoch": 0.09032941176470588,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.4609,
      "step": 3839
    },
    {
      "epoch": 0.0903529411764706,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.1802,
      "step": 3840
    },
    {
      "epoch": 0.0903764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.458,
      "step": 3841
    },
    {
      "epoch": 0.0904,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3129,
      "step": 3842
    },
    {
      "epoch": 0.09042352941176471,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1454,
      "step": 3843
    },
    {
      "epoch": 0.09044705882352941,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3313,
      "step": 3844
    },
    {
      "epoch": 0.09047058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3864,
      "step": 3845
    },
    {
      "epoch": 0.09049411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.6103,
      "step": 3846
    },
    {
      "epoch": 0.09051764705882354,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 3847
    },
    {
      "epoch": 0.09054117647058824,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4944,
      "step": 3848
    },
    {
      "epoch": 0.09056470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3955,
      "step": 3849
    },
    {
      "epoch": 0.09058823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 3850
    },
    {
      "epoch": 0.09061176470588235,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1077,
      "step": 3851
    },
    {
      "epoch": 0.09063529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5096,
      "step": 3852
    },
    {
      "epoch": 0.09065882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4267,
      "step": 3853
    },
    {
      "epoch": 0.09068235294117646,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0713,
      "step": 3854
    },
    {
      "epoch": 0.09070588235294118,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2315,
      "step": 3855
    },
    {
      "epoch": 0.09072941176470588,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2419,
      "step": 3856
    },
    {
      "epoch": 0.09075294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.26,
      "step": 3857
    },
    {
      "epoch": 0.09077647058823529,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1648,
      "step": 3858
    },
    {
      "epoch": 0.0908,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0613,
      "step": 3859
    },
    {
      "epoch": 0.0908235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4599,
      "step": 3860
    },
    {
      "epoch": 0.0908470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2106,
      "step": 3861
    },
    {
      "epoch": 0.09087058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4289,
      "step": 3862
    },
    {
      "epoch": 0.09089411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.222,
      "step": 3863
    },
    {
      "epoch": 0.09091764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4522,
      "step": 3864
    },
    {
      "epoch": 0.09094117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3264,
      "step": 3865
    },
    {
      "epoch": 0.09096470588235295,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.22,
      "step": 3866
    },
    {
      "epoch": 0.09098823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4252,
      "step": 3867
    },
    {
      "epoch": 0.09101176470588235,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 3868
    },
    {
      "epoch": 0.09103529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5433,
      "step": 3869
    },
    {
      "epoch": 0.09105882352941176,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 3870
    },
    {
      "epoch": 0.09108235294117648,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.4615,
      "step": 3871
    },
    {
      "epoch": 0.09110588235294118,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.7398,
      "step": 3872
    },
    {
      "epoch": 0.09112941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4197,
      "step": 3873
    },
    {
      "epoch": 0.09115294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3514,
      "step": 3874
    },
    {
      "epoch": 0.09117647058823529,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4335,
      "step": 3875
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2339,
      "step": 3876
    },
    {
      "epoch": 0.0912235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4352,
      "step": 3877
    },
    {
      "epoch": 0.09124705882352942,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.5982,
      "step": 3878
    },
    {
      "epoch": 0.09127058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.5283,
      "step": 3879
    },
    {
      "epoch": 0.09129411764705882,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4564,
      "step": 3880
    },
    {
      "epoch": 0.09131764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5872,
      "step": 3881
    },
    {
      "epoch": 0.09134117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1904,
      "step": 3882
    },
    {
      "epoch": 0.09136470588235295,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3184,
      "step": 3883
    },
    {
      "epoch": 0.09138823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5574,
      "step": 3884
    },
    {
      "epoch": 0.09141176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3523,
      "step": 3885
    },
    {
      "epoch": 0.09143529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3583,
      "step": 3886
    },
    {
      "epoch": 0.09145882352941176,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0027,
      "step": 3887
    },
    {
      "epoch": 0.09148235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2863,
      "step": 3888
    },
    {
      "epoch": 0.09150588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4723,
      "step": 3889
    },
    {
      "epoch": 0.09152941176470589,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3283,
      "step": 3890
    },
    {
      "epoch": 0.09155294117647059,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0342,
      "step": 3891
    },
    {
      "epoch": 0.0915764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4876,
      "step": 3892
    },
    {
      "epoch": 0.0916,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3764,
      "step": 3893
    },
    {
      "epoch": 0.0916235294117647,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.4191,
      "step": 3894
    },
    {
      "epoch": 0.09164705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.6584,
      "step": 3895
    },
    {
      "epoch": 0.09167058823529411,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1857,
      "step": 3896
    },
    {
      "epoch": 0.09169411764705883,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2393,
      "step": 3897
    },
    {
      "epoch": 0.09171764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5783,
      "step": 3898
    },
    {
      "epoch": 0.09174117647058823,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3782,
      "step": 3899
    },
    {
      "epoch": 0.09176470588235294,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1542,
      "step": 3900
    },
    {
      "epoch": 0.09178823529411764,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5784,
      "step": 3901
    },
    {
      "epoch": 0.09181176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4398,
      "step": 3902
    },
    {
      "epoch": 0.09183529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 3903
    },
    {
      "epoch": 0.09185882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5388,
      "step": 3904
    },
    {
      "epoch": 0.09188235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4182,
      "step": 3905
    },
    {
      "epoch": 0.09190588235294117,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4183,
      "step": 3906
    },
    {
      "epoch": 0.09192941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5792,
      "step": 3907
    },
    {
      "epoch": 0.09195294117647058,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.6703,
      "step": 3908
    },
    {
      "epoch": 0.0919764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 3909
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4609,
      "step": 3910
    },
    {
      "epoch": 0.09202352941176471,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2879,
      "step": 3911
    },
    {
      "epoch": 0.09204705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3605,
      "step": 3912
    },
    {
      "epoch": 0.09207058823529411,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2382,
      "step": 3913
    },
    {
      "epoch": 0.09209411764705883,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.5373,
      "step": 3914
    },
    {
      "epoch": 0.09211764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3352,
      "step": 3915
    },
    {
      "epoch": 0.09214117647058824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2594,
      "step": 3916
    },
    {
      "epoch": 0.09216470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4461,
      "step": 3917
    },
    {
      "epoch": 0.09218823529411764,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4243,
      "step": 3918
    },
    {
      "epoch": 0.09221176470588235,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3827,
      "step": 3919
    },
    {
      "epoch": 0.09223529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2781,
      "step": 3920
    },
    {
      "epoch": 0.09225882352941177,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.166,
      "step": 3921
    },
    {
      "epoch": 0.09228235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5131,
      "step": 3922
    },
    {
      "epoch": 0.09230588235294118,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.5831,
      "step": 3923
    },
    {
      "epoch": 0.09232941176470588,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2425,
      "step": 3924
    },
    {
      "epoch": 0.09235294117647058,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3255,
      "step": 3925
    },
    {
      "epoch": 0.0923764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3644,
      "step": 3926
    },
    {
      "epoch": 0.0924,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1866,
      "step": 3927
    },
    {
      "epoch": 0.09242352941176471,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2378,
      "step": 3928
    },
    {
      "epoch": 0.09244705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4971,
      "step": 3929
    },
    {
      "epoch": 0.09247058823529412,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9662,
      "step": 3930
    },
    {
      "epoch": 0.09249411764705882,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0032,
      "step": 3931
    },
    {
      "epoch": 0.09251764705882352,
      "grad_norm": 1.53125,
      "learning_rate": 0.02,
      "loss": 1.1461,
      "step": 3932
    },
    {
      "epoch": 0.09254117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2306,
      "step": 3933
    },
    {
      "epoch": 0.09256470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.424,
      "step": 3934
    },
    {
      "epoch": 0.09258823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4104,
      "step": 3935
    },
    {
      "epoch": 0.09261176470588235,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6269,
      "step": 3936
    },
    {
      "epoch": 0.09263529411764707,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2919,
      "step": 3937
    },
    {
      "epoch": 0.09265882352941177,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5958,
      "step": 3938
    },
    {
      "epoch": 0.09268235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3573,
      "step": 3939
    },
    {
      "epoch": 0.09270588235294118,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2882,
      "step": 3940
    },
    {
      "epoch": 0.09272941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1957,
      "step": 3941
    },
    {
      "epoch": 0.0927529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3671,
      "step": 3942
    },
    {
      "epoch": 0.0927764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3928,
      "step": 3943
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.4069,
      "step": 3944
    },
    {
      "epoch": 0.09282352941176471,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3734,
      "step": 3945
    },
    {
      "epoch": 0.09284705882352941,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2902,
      "step": 3946
    },
    {
      "epoch": 0.09287058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9306,
      "step": 3947
    },
    {
      "epoch": 0.09289411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5928,
      "step": 3948
    },
    {
      "epoch": 0.09291764705882354,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1842,
      "step": 3949
    },
    {
      "epoch": 0.09294117647058824,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.0463,
      "step": 3950
    },
    {
      "epoch": 0.09296470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4844,
      "step": 3951
    },
    {
      "epoch": 0.09298823529411765,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.075,
      "step": 3952
    },
    {
      "epoch": 0.09301176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3898,
      "step": 3953
    },
    {
      "epoch": 0.09303529411764706,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2784,
      "step": 3954
    },
    {
      "epoch": 0.09305882352941176,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2699,
      "step": 3955
    },
    {
      "epoch": 0.09308235294117648,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4345,
      "step": 3956
    },
    {
      "epoch": 0.09310588235294118,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3781,
      "step": 3957
    },
    {
      "epoch": 0.09312941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3658,
      "step": 3958
    },
    {
      "epoch": 0.09315294117647059,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2143,
      "step": 3959
    },
    {
      "epoch": 0.09317647058823529,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3713,
      "step": 3960
    },
    {
      "epoch": 0.0932,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3148,
      "step": 3961
    },
    {
      "epoch": 0.0932235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.5304,
      "step": 3962
    },
    {
      "epoch": 0.0932470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3132,
      "step": 3963
    },
    {
      "epoch": 0.09327058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.375,
      "step": 3964
    },
    {
      "epoch": 0.09329411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4422,
      "step": 3965
    },
    {
      "epoch": 0.09331764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2464,
      "step": 3966
    },
    {
      "epoch": 0.09334117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3312,
      "step": 3967
    },
    {
      "epoch": 0.09336470588235295,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4311,
      "step": 3968
    },
    {
      "epoch": 0.09338823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1349,
      "step": 3969
    },
    {
      "epoch": 0.09341176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3736,
      "step": 3970
    },
    {
      "epoch": 0.09343529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2293,
      "step": 3971
    },
    {
      "epoch": 0.09345882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4389,
      "step": 3972
    },
    {
      "epoch": 0.09348235294117647,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2227,
      "step": 3973
    },
    {
      "epoch": 0.09350588235294117,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3618,
      "step": 3974
    },
    {
      "epoch": 0.09352941176470589,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4226,
      "step": 3975
    },
    {
      "epoch": 0.09355294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5262,
      "step": 3976
    },
    {
      "epoch": 0.09357647058823529,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.4641,
      "step": 3977
    },
    {
      "epoch": 0.0936,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2983,
      "step": 3978
    },
    {
      "epoch": 0.0936235294117647,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.4609,
      "step": 3979
    },
    {
      "epoch": 0.09364705882352942,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3486,
      "step": 3980
    },
    {
      "epoch": 0.09367058823529412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1808,
      "step": 3981
    },
    {
      "epoch": 0.09369411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3083,
      "step": 3982
    },
    {
      "epoch": 0.09371764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.355,
      "step": 3983
    },
    {
      "epoch": 0.09374117647058823,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3296,
      "step": 3984
    },
    {
      "epoch": 0.09376470588235294,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3607,
      "step": 3985
    },
    {
      "epoch": 0.09378823529411764,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3918,
      "step": 3986
    },
    {
      "epoch": 0.09381176470588236,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9987,
      "step": 3987
    },
    {
      "epoch": 0.09383529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.408,
      "step": 3988
    },
    {
      "epoch": 0.09385882352941176,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4606,
      "step": 3989
    },
    {
      "epoch": 0.09388235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3409,
      "step": 3990
    },
    {
      "epoch": 0.09390588235294117,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4761,
      "step": 3991
    },
    {
      "epoch": 0.09392941176470589,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4,
      "step": 3992
    },
    {
      "epoch": 0.09395294117647059,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.4677,
      "step": 3993
    },
    {
      "epoch": 0.0939764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4106,
      "step": 3994
    },
    {
      "epoch": 0.094,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.3188,
      "step": 3995
    },
    {
      "epoch": 0.0940235294117647,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0789,
      "step": 3996
    },
    {
      "epoch": 0.09404705882352941,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.4762,
      "step": 3997
    },
    {
      "epoch": 0.09407058823529411,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5691,
      "step": 3998
    },
    {
      "epoch": 0.09409411764705883,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3529,
      "step": 3999
    },
    {
      "epoch": 0.09411764705882353,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2231,
      "step": 4000
    },
    {
      "epoch": 0.09411764705882353,
      "eval_loss": 2.247562885284424,
      "eval_runtime": 679.3602,
      "eval_samples_per_second": 12.512,
      "eval_steps_per_second": 3.128,
      "step": 4000
    },
    {
      "epoch": 0.09414117647058824,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3159,
      "step": 4001
    },
    {
      "epoch": 0.09416470588235294,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.2281,
      "step": 4002
    },
    {
      "epoch": 0.09418823529411764,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5315,
      "step": 4003
    },
    {
      "epoch": 0.09421176470588236,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 4004
    },
    {
      "epoch": 0.09423529411764706,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1955,
      "step": 4005
    },
    {
      "epoch": 0.09425882352941177,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3713,
      "step": 4006
    },
    {
      "epoch": 0.09428235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3337,
      "step": 4007
    },
    {
      "epoch": 0.09430588235294117,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.223,
      "step": 4008
    },
    {
      "epoch": 0.09432941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4469,
      "step": 4009
    },
    {
      "epoch": 0.09435294117647058,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1236,
      "step": 4010
    },
    {
      "epoch": 0.0943764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.3553,
      "step": 4011
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2207,
      "step": 4012
    },
    {
      "epoch": 0.09442352941176471,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5981,
      "step": 4013
    },
    {
      "epoch": 0.09444705882352941,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1513,
      "step": 4014
    },
    {
      "epoch": 0.09447058823529411,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3182,
      "step": 4015
    },
    {
      "epoch": 0.09449411764705883,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3779,
      "step": 4016
    },
    {
      "epoch": 0.09451764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4293,
      "step": 4017
    },
    {
      "epoch": 0.09454117647058824,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3111,
      "step": 4018
    },
    {
      "epoch": 0.09456470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2927,
      "step": 4019
    },
    {
      "epoch": 0.09458823529411765,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4257,
      "step": 4020
    },
    {
      "epoch": 0.09461176470588235,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2225,
      "step": 4021
    },
    {
      "epoch": 0.09463529411764705,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4639,
      "step": 4022
    },
    {
      "epoch": 0.09465882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.6524,
      "step": 4023
    },
    {
      "epoch": 0.09468235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.6222,
      "step": 4024
    },
    {
      "epoch": 0.09470588235294118,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2166,
      "step": 4025
    },
    {
      "epoch": 0.09472941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.6217,
      "step": 4026
    },
    {
      "epoch": 0.09475294117647058,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3302,
      "step": 4027
    },
    {
      "epoch": 0.0947764705882353,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2594,
      "step": 4028
    },
    {
      "epoch": 0.0948,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4964,
      "step": 4029
    },
    {
      "epoch": 0.09482352941176471,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3473,
      "step": 4030
    },
    {
      "epoch": 0.09484705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3116,
      "step": 4031
    },
    {
      "epoch": 0.09487058823529412,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0456,
      "step": 4032
    },
    {
      "epoch": 0.09489411764705882,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4366,
      "step": 4033
    },
    {
      "epoch": 0.09491764705882352,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3584,
      "step": 4034
    },
    {
      "epoch": 0.09494117647058824,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4148,
      "step": 4035
    },
    {
      "epoch": 0.09496470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.5063,
      "step": 4036
    },
    {
      "epoch": 0.09498823529411765,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.3153,
      "step": 4037
    },
    {
      "epoch": 0.09501176470588235,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2266,
      "step": 4038
    },
    {
      "epoch": 0.09503529411764707,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4304,
      "step": 4039
    },
    {
      "epoch": 0.09505882352941177,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5222,
      "step": 4040
    },
    {
      "epoch": 0.09508235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2399,
      "step": 4041
    },
    {
      "epoch": 0.09510588235294118,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3338,
      "step": 4042
    },
    {
      "epoch": 0.09512941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3359,
      "step": 4043
    },
    {
      "epoch": 0.09515294117647059,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1339,
      "step": 4044
    },
    {
      "epoch": 0.09517647058823529,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9399,
      "step": 4045
    },
    {
      "epoch": 0.0952,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4809,
      "step": 4046
    },
    {
      "epoch": 0.0952235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.5217,
      "step": 4047
    },
    {
      "epoch": 0.0952470588235294,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1601,
      "step": 4048
    },
    {
      "epoch": 0.09527058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3725,
      "step": 4049
    },
    {
      "epoch": 0.09529411764705882,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3222,
      "step": 4050
    },
    {
      "epoch": 0.09531764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.298,
      "step": 4051
    },
    {
      "epoch": 0.09534117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4012,
      "step": 4052
    },
    {
      "epoch": 0.09536470588235293,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.301,
      "step": 4053
    },
    {
      "epoch": 0.09538823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.417,
      "step": 4054
    },
    {
      "epoch": 0.09541176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4214,
      "step": 4055
    },
    {
      "epoch": 0.09543529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4102,
      "step": 4056
    },
    {
      "epoch": 0.09545882352941176,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2373,
      "step": 4057
    },
    {
      "epoch": 0.09548235294117648,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0895,
      "step": 4058
    },
    {
      "epoch": 0.09550588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3389,
      "step": 4059
    },
    {
      "epoch": 0.09552941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3462,
      "step": 4060
    },
    {
      "epoch": 0.09555294117647059,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4124,
      "step": 4061
    },
    {
      "epoch": 0.09557647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.6887,
      "step": 4062
    },
    {
      "epoch": 0.0956,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3054,
      "step": 4063
    },
    {
      "epoch": 0.0956235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.375,
      "step": 4064
    },
    {
      "epoch": 0.09564705882352942,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1082,
      "step": 4065
    },
    {
      "epoch": 0.09567058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 4066
    },
    {
      "epoch": 0.09569411764705882,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0962,
      "step": 4067
    },
    {
      "epoch": 0.09571764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2274,
      "step": 4068
    },
    {
      "epoch": 0.09574117647058823,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4446,
      "step": 4069
    },
    {
      "epoch": 0.09576470588235295,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3554,
      "step": 4070
    },
    {
      "epoch": 0.09578823529411765,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.276,
      "step": 4071
    },
    {
      "epoch": 0.09581176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 4072
    },
    {
      "epoch": 0.09583529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5874,
      "step": 4073
    },
    {
      "epoch": 0.09585882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5454,
      "step": 4074
    },
    {
      "epoch": 0.09588235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6291,
      "step": 4075
    },
    {
      "epoch": 0.09590588235294117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2361,
      "step": 4076
    },
    {
      "epoch": 0.09592941176470589,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.0681,
      "step": 4077
    },
    {
      "epoch": 0.09595294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4149,
      "step": 4078
    },
    {
      "epoch": 0.09597647058823529,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.9431,
      "step": 4079
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.5397,
      "step": 4080
    },
    {
      "epoch": 0.0960235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3475,
      "step": 4081
    },
    {
      "epoch": 0.09604705882352942,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3702,
      "step": 4082
    },
    {
      "epoch": 0.09607058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3336,
      "step": 4083
    },
    {
      "epoch": 0.09609411764705883,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.6716,
      "step": 4084
    },
    {
      "epoch": 0.09611764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4174,
      "step": 4085
    },
    {
      "epoch": 0.09614117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.5,
      "step": 4086
    },
    {
      "epoch": 0.09616470588235294,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.31,
      "step": 4087
    },
    {
      "epoch": 0.09618823529411764,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.2562,
      "step": 4088
    },
    {
      "epoch": 0.09621176470588236,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4994,
      "step": 4089
    },
    {
      "epoch": 0.09623529411764706,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3037,
      "step": 4090
    },
    {
      "epoch": 0.09625882352941176,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2163,
      "step": 4091
    },
    {
      "epoch": 0.09628235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.491,
      "step": 4092
    },
    {
      "epoch": 0.09630588235294117,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3689,
      "step": 4093
    },
    {
      "epoch": 0.09632941176470589,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4795,
      "step": 4094
    },
    {
      "epoch": 0.09635294117647059,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 4095
    },
    {
      "epoch": 0.0963764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.206,
      "step": 4096
    },
    {
      "epoch": 0.0964,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3925,
      "step": 4097
    },
    {
      "epoch": 0.0964235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3982,
      "step": 4098
    },
    {
      "epoch": 0.09644705882352941,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.5652,
      "step": 4099
    },
    {
      "epoch": 0.09647058823529411,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2988,
      "step": 4100
    },
    {
      "epoch": 0.09649411764705883,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.2942,
      "step": 4101
    },
    {
      "epoch": 0.09651764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4106,
      "step": 4102
    },
    {
      "epoch": 0.09654117647058824,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3547,
      "step": 4103
    },
    {
      "epoch": 0.09656470588235294,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3491,
      "step": 4104
    },
    {
      "epoch": 0.09658823529411764,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3587,
      "step": 4105
    },
    {
      "epoch": 0.09661176470588236,
      "grad_norm": 0.306640625,
      "learning_rate": 0.02,
      "loss": 1.1712,
      "step": 4106
    },
    {
      "epoch": 0.09663529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3446,
      "step": 4107
    },
    {
      "epoch": 0.09665882352941177,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.3509,
      "step": 4108
    },
    {
      "epoch": 0.09668235294117647,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3018,
      "step": 4109
    },
    {
      "epoch": 0.09670588235294118,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 4110
    },
    {
      "epoch": 0.09672941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2709,
      "step": 4111
    },
    {
      "epoch": 0.09675294117647058,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 1.2372,
      "step": 4112
    },
    {
      "epoch": 0.0967764705882353,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.136,
      "step": 4113
    },
    {
      "epoch": 0.0968,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5731,
      "step": 4114
    },
    {
      "epoch": 0.09682352941176471,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.3941,
      "step": 4115
    },
    {
      "epoch": 0.09684705882352941,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3892,
      "step": 4116
    },
    {
      "epoch": 0.09687058823529411,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2828,
      "step": 4117
    },
    {
      "epoch": 0.09689411764705883,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.42,
      "step": 4118
    },
    {
      "epoch": 0.09691764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4016,
      "step": 4119
    },
    {
      "epoch": 0.09694117647058824,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.4791,
      "step": 4120
    },
    {
      "epoch": 0.09696470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5257,
      "step": 4121
    },
    {
      "epoch": 0.09698823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.344,
      "step": 4122
    },
    {
      "epoch": 0.09701176470588235,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4108,
      "step": 4123
    },
    {
      "epoch": 0.09703529411764705,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3017,
      "step": 4124
    },
    {
      "epoch": 0.09705882352941177,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.285,
      "step": 4125
    },
    {
      "epoch": 0.09708235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1896,
      "step": 4126
    },
    {
      "epoch": 0.09710588235294118,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.5443,
      "step": 4127
    },
    {
      "epoch": 0.09712941176470588,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.4068,
      "step": 4128
    },
    {
      "epoch": 0.0971529411764706,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3169,
      "step": 4129
    },
    {
      "epoch": 0.0971764705882353,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.3153,
      "step": 4130
    },
    {
      "epoch": 0.0972,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.13,
      "step": 4131
    },
    {
      "epoch": 0.09722352941176471,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.6509,
      "step": 4132
    },
    {
      "epoch": 0.09724705882352941,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0419,
      "step": 4133
    },
    {
      "epoch": 0.09727058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1775,
      "step": 4134
    },
    {
      "epoch": 0.09729411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2701,
      "step": 4135
    },
    {
      "epoch": 0.09731764705882352,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 4136
    },
    {
      "epoch": 0.09734117647058824,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2364,
      "step": 4137
    },
    {
      "epoch": 0.09736470588235294,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3594,
      "step": 4138
    },
    {
      "epoch": 0.09738823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.559,
      "step": 4139
    },
    {
      "epoch": 0.09741176470588235,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.4089,
      "step": 4140
    },
    {
      "epoch": 0.09743529411764706,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 4141
    },
    {
      "epoch": 0.09745882352941176,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3991,
      "step": 4142
    },
    {
      "epoch": 0.09748235294117646,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.476,
      "step": 4143
    },
    {
      "epoch": 0.09750588235294118,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2846,
      "step": 4144
    },
    {
      "epoch": 0.09752941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3981,
      "step": 4145
    },
    {
      "epoch": 0.09755294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2151,
      "step": 4146
    },
    {
      "epoch": 0.09757647058823529,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4391,
      "step": 4147
    },
    {
      "epoch": 0.0976,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.3659,
      "step": 4148
    },
    {
      "epoch": 0.0976235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3545,
      "step": 4149
    },
    {
      "epoch": 0.0976470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2945,
      "step": 4150
    },
    {
      "epoch": 0.09767058823529412,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1247,
      "step": 4151
    },
    {
      "epoch": 0.09769411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3705,
      "step": 4152
    },
    {
      "epoch": 0.09771764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3348,
      "step": 4153
    },
    {
      "epoch": 0.09774117647058823,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1307,
      "step": 4154
    },
    {
      "epoch": 0.09776470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.6142,
      "step": 4155
    },
    {
      "epoch": 0.09778823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4196,
      "step": 4156
    },
    {
      "epoch": 0.09781176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5039,
      "step": 4157
    },
    {
      "epoch": 0.09783529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.6139,
      "step": 4158
    },
    {
      "epoch": 0.09785882352941176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.526,
      "step": 4159
    },
    {
      "epoch": 0.09788235294117648,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2597,
      "step": 4160
    },
    {
      "epoch": 0.09790588235294118,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4365,
      "step": 4161
    },
    {
      "epoch": 0.09792941176470588,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1199,
      "step": 4162
    },
    {
      "epoch": 0.09795294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4135,
      "step": 4163
    },
    {
      "epoch": 0.09797647058823529,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4363,
      "step": 4164
    },
    {
      "epoch": 0.098,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3182,
      "step": 4165
    },
    {
      "epoch": 0.0980235294117647,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.184,
      "step": 4166
    },
    {
      "epoch": 0.09804705882352942,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.5492,
      "step": 4167
    },
    {
      "epoch": 0.09807058823529412,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2288,
      "step": 4168
    },
    {
      "epoch": 0.09809411764705882,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0529,
      "step": 4169
    },
    {
      "epoch": 0.09811764705882353,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3534,
      "step": 4170
    },
    {
      "epoch": 0.09814117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3413,
      "step": 4171
    },
    {
      "epoch": 0.09816470588235295,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0964,
      "step": 4172
    },
    {
      "epoch": 0.09818823529411765,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3139,
      "step": 4173
    },
    {
      "epoch": 0.09821176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4365,
      "step": 4174
    },
    {
      "epoch": 0.09823529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2565,
      "step": 4175
    },
    {
      "epoch": 0.09825882352941176,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4684,
      "step": 4176
    },
    {
      "epoch": 0.09828235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.353,
      "step": 4177
    },
    {
      "epoch": 0.09830588235294117,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.4059,
      "step": 4178
    },
    {
      "epoch": 0.09832941176470589,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2813,
      "step": 4179
    },
    {
      "epoch": 0.09835294117647059,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3442,
      "step": 4180
    },
    {
      "epoch": 0.09837647058823529,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3912,
      "step": 4181
    },
    {
      "epoch": 0.0984,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.433,
      "step": 4182
    },
    {
      "epoch": 0.0984235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2719,
      "step": 4183
    },
    {
      "epoch": 0.09844705882352942,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0693,
      "step": 4184
    },
    {
      "epoch": 0.09847058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3665,
      "step": 4185
    },
    {
      "epoch": 0.09849411764705883,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1917,
      "step": 4186
    },
    {
      "epoch": 0.09851764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4511,
      "step": 4187
    },
    {
      "epoch": 0.09854117647058823,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.7665,
      "step": 4188
    },
    {
      "epoch": 0.09856470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4962,
      "step": 4189
    },
    {
      "epoch": 0.09858823529411764,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5639,
      "step": 4190
    },
    {
      "epoch": 0.09861176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3159,
      "step": 4191
    },
    {
      "epoch": 0.09863529411764706,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4123,
      "step": 4192
    },
    {
      "epoch": 0.09865882352941177,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4653,
      "step": 4193
    },
    {
      "epoch": 0.09868235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.6195,
      "step": 4194
    },
    {
      "epoch": 0.09870588235294117,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1732,
      "step": 4195
    },
    {
      "epoch": 0.09872941176470588,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.4168,
      "step": 4196
    },
    {
      "epoch": 0.09875294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0854,
      "step": 4197
    },
    {
      "epoch": 0.0987764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4475,
      "step": 4198
    },
    {
      "epoch": 0.0988,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5135,
      "step": 4199
    },
    {
      "epoch": 0.0988235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3736,
      "step": 4200
    },
    {
      "epoch": 0.09884705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4432,
      "step": 4201
    },
    {
      "epoch": 0.09887058823529411,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.085,
      "step": 4202
    },
    {
      "epoch": 0.09889411764705883,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4077,
      "step": 4203
    },
    {
      "epoch": 0.09891764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3615,
      "step": 4204
    },
    {
      "epoch": 0.09894117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4109,
      "step": 4205
    },
    {
      "epoch": 0.09896470588235294,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3615,
      "step": 4206
    },
    {
      "epoch": 0.09898823529411764,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 4207
    },
    {
      "epoch": 0.09901176470588235,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3878,
      "step": 4208
    },
    {
      "epoch": 0.09903529411764705,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5344,
      "step": 4209
    },
    {
      "epoch": 0.09905882352941177,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.0728,
      "step": 4210
    },
    {
      "epoch": 0.09908235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5686,
      "step": 4211
    },
    {
      "epoch": 0.09910588235294118,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1008,
      "step": 4212
    },
    {
      "epoch": 0.09912941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4804,
      "step": 4213
    },
    {
      "epoch": 0.09915294117647058,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2866,
      "step": 4214
    },
    {
      "epoch": 0.0991764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.5086,
      "step": 4215
    },
    {
      "epoch": 0.0992,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.4923,
      "step": 4216
    },
    {
      "epoch": 0.09922352941176471,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1072,
      "step": 4217
    },
    {
      "epoch": 0.09924705882352941,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.381,
      "step": 4218
    },
    {
      "epoch": 0.09927058823529412,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3739,
      "step": 4219
    },
    {
      "epoch": 0.09929411764705882,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.331,
      "step": 4220
    },
    {
      "epoch": 0.09931764705882352,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3788,
      "step": 4221
    },
    {
      "epoch": 0.09934117647058824,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3415,
      "step": 4222
    },
    {
      "epoch": 0.09936470588235294,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1256,
      "step": 4223
    },
    {
      "epoch": 0.09938823529411765,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3981,
      "step": 4224
    },
    {
      "epoch": 0.09941176470588235,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4704,
      "step": 4225
    },
    {
      "epoch": 0.09943529411764705,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4885,
      "step": 4226
    },
    {
      "epoch": 0.09945882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1565,
      "step": 4227
    },
    {
      "epoch": 0.09948235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3259,
      "step": 4228
    },
    {
      "epoch": 0.09950588235294118,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3698,
      "step": 4229
    },
    {
      "epoch": 0.09952941176470588,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3253,
      "step": 4230
    },
    {
      "epoch": 0.0995529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3207,
      "step": 4231
    },
    {
      "epoch": 0.0995764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1027,
      "step": 4232
    },
    {
      "epoch": 0.0996,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5623,
      "step": 4233
    },
    {
      "epoch": 0.09962352941176471,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4372,
      "step": 4234
    },
    {
      "epoch": 0.09964705882352941,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.444,
      "step": 4235
    },
    {
      "epoch": 0.09967058823529412,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3384,
      "step": 4236
    },
    {
      "epoch": 0.09969411764705882,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.575,
      "step": 4237
    },
    {
      "epoch": 0.09971764705882354,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2381,
      "step": 4238
    },
    {
      "epoch": 0.09974117647058824,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.2656,
      "step": 4239
    },
    {
      "epoch": 0.09976470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2503,
      "step": 4240
    },
    {
      "epoch": 0.09978823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.7421,
      "step": 4241
    },
    {
      "epoch": 0.09981176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.259,
      "step": 4242
    },
    {
      "epoch": 0.09983529411764706,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.2935,
      "step": 4243
    },
    {
      "epoch": 0.09985882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4361,
      "step": 4244
    },
    {
      "epoch": 0.09988235294117646,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2712,
      "step": 4245
    },
    {
      "epoch": 0.09990588235294118,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3412,
      "step": 4246
    },
    {
      "epoch": 0.09992941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2231,
      "step": 4247
    },
    {
      "epoch": 0.09995294117647059,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.209,
      "step": 4248
    },
    {
      "epoch": 0.09997647058823529,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2109,
      "step": 4249
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 4250
    },
    {
      "epoch": 0.1000235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.205,
      "step": 4251
    },
    {
      "epoch": 0.1000470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1823,
      "step": 4252
    },
    {
      "epoch": 0.10007058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1726,
      "step": 4253
    },
    {
      "epoch": 0.10009411764705882,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3867,
      "step": 4254
    },
    {
      "epoch": 0.10011764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2761,
      "step": 4255
    },
    {
      "epoch": 0.10014117647058823,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 4256
    },
    {
      "epoch": 0.10016470588235295,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3948,
      "step": 4257
    },
    {
      "epoch": 0.10018823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1602,
      "step": 4258
    },
    {
      "epoch": 0.10021176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3003,
      "step": 4259
    },
    {
      "epoch": 0.10023529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4466,
      "step": 4260
    },
    {
      "epoch": 0.10025882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4948,
      "step": 4261
    },
    {
      "epoch": 0.10028235294117648,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.175,
      "step": 4262
    },
    {
      "epoch": 0.10030588235294118,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.278,
      "step": 4263
    },
    {
      "epoch": 0.10032941176470589,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2682,
      "step": 4264
    },
    {
      "epoch": 0.10035294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5596,
      "step": 4265
    },
    {
      "epoch": 0.10037647058823529,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3917,
      "step": 4266
    },
    {
      "epoch": 0.1004,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1045,
      "step": 4267
    },
    {
      "epoch": 0.1004235294117647,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1815,
      "step": 4268
    },
    {
      "epoch": 0.10044705882352942,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1567,
      "step": 4269
    },
    {
      "epoch": 0.10047058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4569,
      "step": 4270
    },
    {
      "epoch": 0.10049411764705882,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2674,
      "step": 4271
    },
    {
      "epoch": 0.10051764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3429,
      "step": 4272
    },
    {
      "epoch": 0.10054117647058823,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1682,
      "step": 4273
    },
    {
      "epoch": 0.10056470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4472,
      "step": 4274
    },
    {
      "epoch": 0.10058823529411764,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.146,
      "step": 4275
    },
    {
      "epoch": 0.10061176470588236,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0837,
      "step": 4276
    },
    {
      "epoch": 0.10063529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5192,
      "step": 4277
    },
    {
      "epoch": 0.10065882352941176,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2742,
      "step": 4278
    },
    {
      "epoch": 0.10068235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0814,
      "step": 4279
    },
    {
      "epoch": 0.10070588235294117,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1528,
      "step": 4280
    },
    {
      "epoch": 0.10072941176470589,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3644,
      "step": 4281
    },
    {
      "epoch": 0.10075294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3237,
      "step": 4282
    },
    {
      "epoch": 0.1007764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.6213,
      "step": 4283
    },
    {
      "epoch": 0.1008,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4278,
      "step": 4284
    },
    {
      "epoch": 0.1008235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3208,
      "step": 4285
    },
    {
      "epoch": 0.10084705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5712,
      "step": 4286
    },
    {
      "epoch": 0.10087058823529411,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3085,
      "step": 4287
    },
    {
      "epoch": 0.10089411764705883,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3719,
      "step": 4288
    },
    {
      "epoch": 0.10091764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2344,
      "step": 4289
    },
    {
      "epoch": 0.10094117647058823,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4281,
      "step": 4290
    },
    {
      "epoch": 0.10096470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.381,
      "step": 4291
    },
    {
      "epoch": 0.10098823529411764,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4332,
      "step": 4292
    },
    {
      "epoch": 0.10101176470588236,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3343,
      "step": 4293
    },
    {
      "epoch": 0.10103529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2892,
      "step": 4294
    },
    {
      "epoch": 0.10105882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4921,
      "step": 4295
    },
    {
      "epoch": 0.10108235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2762,
      "step": 4296
    },
    {
      "epoch": 0.10110588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4553,
      "step": 4297
    },
    {
      "epoch": 0.10112941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4487,
      "step": 4298
    },
    {
      "epoch": 0.10115294117647058,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5143,
      "step": 4299
    },
    {
      "epoch": 0.1011764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3101,
      "step": 4300
    },
    {
      "epoch": 0.1012,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2974,
      "step": 4301
    },
    {
      "epoch": 0.10122352941176471,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4353,
      "step": 4302
    },
    {
      "epoch": 0.10124705882352941,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.5445,
      "step": 4303
    },
    {
      "epoch": 0.10127058823529411,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0365,
      "step": 4304
    },
    {
      "epoch": 0.10129411764705883,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3369,
      "step": 4305
    },
    {
      "epoch": 0.10131764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4975,
      "step": 4306
    },
    {
      "epoch": 0.10134117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3822,
      "step": 4307
    },
    {
      "epoch": 0.10136470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5116,
      "step": 4308
    },
    {
      "epoch": 0.10138823529411764,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.0353,
      "step": 4309
    },
    {
      "epoch": 0.10141176470588235,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.182,
      "step": 4310
    },
    {
      "epoch": 0.10143529411764705,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.2221,
      "step": 4311
    },
    {
      "epoch": 0.10145882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.5932,
      "step": 4312
    },
    {
      "epoch": 0.10148235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3415,
      "step": 4313
    },
    {
      "epoch": 0.10150588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.5099,
      "step": 4314
    },
    {
      "epoch": 0.10152941176470588,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 4315
    },
    {
      "epoch": 0.10155294117647058,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2695,
      "step": 4316
    },
    {
      "epoch": 0.1015764705882353,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.138,
      "step": 4317
    },
    {
      "epoch": 0.1016,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0121,
      "step": 4318
    },
    {
      "epoch": 0.10162352941176471,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2049,
      "step": 4319
    },
    {
      "epoch": 0.10164705882352941,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3445,
      "step": 4320
    },
    {
      "epoch": 0.10167058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9156,
      "step": 4321
    },
    {
      "epoch": 0.10169411764705882,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3136,
      "step": 4322
    },
    {
      "epoch": 0.10171764705882352,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1196,
      "step": 4323
    },
    {
      "epoch": 0.10174117647058824,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3184,
      "step": 4324
    },
    {
      "epoch": 0.10176470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0092,
      "step": 4325
    },
    {
      "epoch": 0.10178823529411765,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.346,
      "step": 4326
    },
    {
      "epoch": 0.10181176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.5397,
      "step": 4327
    },
    {
      "epoch": 0.10183529411764707,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2705,
      "step": 4328
    },
    {
      "epoch": 0.10185882352941177,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2411,
      "step": 4329
    },
    {
      "epoch": 0.10188235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 4330
    },
    {
      "epoch": 0.10190588235294118,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2774,
      "step": 4331
    },
    {
      "epoch": 0.10192941176470588,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.213,
      "step": 4332
    },
    {
      "epoch": 0.1019529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3613,
      "step": 4333
    },
    {
      "epoch": 0.1019764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.4506,
      "step": 4334
    },
    {
      "epoch": 0.102,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1584,
      "step": 4335
    },
    {
      "epoch": 0.10202352941176471,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1953,
      "step": 4336
    },
    {
      "epoch": 0.10204705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2031,
      "step": 4337
    },
    {
      "epoch": 0.10207058823529412,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.1958,
      "step": 4338
    },
    {
      "epoch": 0.10209411764705882,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.544,
      "step": 4339
    },
    {
      "epoch": 0.10211764705882354,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1877,
      "step": 4340
    },
    {
      "epoch": 0.10214117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4394,
      "step": 4341
    },
    {
      "epoch": 0.10216470588235294,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3034,
      "step": 4342
    },
    {
      "epoch": 0.10218823529411765,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3666,
      "step": 4343
    },
    {
      "epoch": 0.10221176470588235,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.1542,
      "step": 4344
    },
    {
      "epoch": 0.10223529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2259,
      "step": 4345
    },
    {
      "epoch": 0.10225882352941176,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2913,
      "step": 4346
    },
    {
      "epoch": 0.10228235294117648,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1626,
      "step": 4347
    },
    {
      "epoch": 0.10230588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4988,
      "step": 4348
    },
    {
      "epoch": 0.10232941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.49,
      "step": 4349
    },
    {
      "epoch": 0.10235294117647059,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2996,
      "step": 4350
    },
    {
      "epoch": 0.10237647058823529,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2543,
      "step": 4351
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1678,
      "step": 4352
    },
    {
      "epoch": 0.1024235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3418,
      "step": 4353
    },
    {
      "epoch": 0.1024470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 4354
    },
    {
      "epoch": 0.10247058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4256,
      "step": 4355
    },
    {
      "epoch": 0.10249411764705882,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1916,
      "step": 4356
    },
    {
      "epoch": 0.10251764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2116,
      "step": 4357
    },
    {
      "epoch": 0.10254117647058823,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4695,
      "step": 4358
    },
    {
      "epoch": 0.10256470588235295,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.5098,
      "step": 4359
    },
    {
      "epoch": 0.10258823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4266,
      "step": 4360
    },
    {
      "epoch": 0.10261176470588235,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2897,
      "step": 4361
    },
    {
      "epoch": 0.10263529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3982,
      "step": 4362
    },
    {
      "epoch": 0.10265882352941176,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3191,
      "step": 4363
    },
    {
      "epoch": 0.10268235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2455,
      "step": 4364
    },
    {
      "epoch": 0.10270588235294117,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3807,
      "step": 4365
    },
    {
      "epoch": 0.10272941176470589,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3484,
      "step": 4366
    },
    {
      "epoch": 0.10275294117647059,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1734,
      "step": 4367
    },
    {
      "epoch": 0.10277647058823529,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.958,
      "step": 4368
    },
    {
      "epoch": 0.1028,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9601,
      "step": 4369
    },
    {
      "epoch": 0.1028235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1799,
      "step": 4370
    },
    {
      "epoch": 0.10284705882352942,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3232,
      "step": 4371
    },
    {
      "epoch": 0.10287058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4815,
      "step": 4372
    },
    {
      "epoch": 0.10289411764705883,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.218,
      "step": 4373
    },
    {
      "epoch": 0.10291764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4827,
      "step": 4374
    },
    {
      "epoch": 0.10294117647058823,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3456,
      "step": 4375
    },
    {
      "epoch": 0.10296470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5334,
      "step": 4376
    },
    {
      "epoch": 0.10298823529411764,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2032,
      "step": 4377
    },
    {
      "epoch": 0.10301176470588236,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3008,
      "step": 4378
    },
    {
      "epoch": 0.10303529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4611,
      "step": 4379
    },
    {
      "epoch": 0.10305882352941176,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1467,
      "step": 4380
    },
    {
      "epoch": 0.10308235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4603,
      "step": 4381
    },
    {
      "epoch": 0.10310588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.6028,
      "step": 4382
    },
    {
      "epoch": 0.10312941176470589,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1702,
      "step": 4383
    },
    {
      "epoch": 0.10315294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3555,
      "step": 4384
    },
    {
      "epoch": 0.1031764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3558,
      "step": 4385
    },
    {
      "epoch": 0.1032,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0948,
      "step": 4386
    },
    {
      "epoch": 0.1032235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4257,
      "step": 4387
    },
    {
      "epoch": 0.10324705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3803,
      "step": 4388
    },
    {
      "epoch": 0.10327058823529411,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1041,
      "step": 4389
    },
    {
      "epoch": 0.10329411764705883,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0502,
      "step": 4390
    },
    {
      "epoch": 0.10331764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4975,
      "step": 4391
    },
    {
      "epoch": 0.10334117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.197,
      "step": 4392
    },
    {
      "epoch": 0.10336470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.5084,
      "step": 4393
    },
    {
      "epoch": 0.10338823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4208,
      "step": 4394
    },
    {
      "epoch": 0.10341176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4037,
      "step": 4395
    },
    {
      "epoch": 0.10343529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.99,
      "step": 4396
    },
    {
      "epoch": 0.10345882352941177,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2229,
      "step": 4397
    },
    {
      "epoch": 0.10348235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3444,
      "step": 4398
    },
    {
      "epoch": 0.10350588235294117,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5114,
      "step": 4399
    },
    {
      "epoch": 0.10352941176470588,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.413,
      "step": 4400
    },
    {
      "epoch": 0.10355294117647058,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.1768,
      "step": 4401
    },
    {
      "epoch": 0.1035764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.372,
      "step": 4402
    },
    {
      "epoch": 0.1036,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2448,
      "step": 4403
    },
    {
      "epoch": 0.10362352941176471,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5577,
      "step": 4404
    },
    {
      "epoch": 0.10364705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.4488,
      "step": 4405
    },
    {
      "epoch": 0.10367058823529411,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.438,
      "step": 4406
    },
    {
      "epoch": 0.10369411764705883,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1093,
      "step": 4407
    },
    {
      "epoch": 0.10371764705882353,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1499,
      "step": 4408
    },
    {
      "epoch": 0.10374117647058824,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1876,
      "step": 4409
    },
    {
      "epoch": 0.10376470588235294,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1524,
      "step": 4410
    },
    {
      "epoch": 0.10378823529411765,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.5474,
      "step": 4411
    },
    {
      "epoch": 0.10381176470588235,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.3953,
      "step": 4412
    },
    {
      "epoch": 0.10383529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1791,
      "step": 4413
    },
    {
      "epoch": 0.10385882352941177,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.247,
      "step": 4414
    },
    {
      "epoch": 0.10388235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3392,
      "step": 4415
    },
    {
      "epoch": 0.10390588235294118,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3482,
      "step": 4416
    },
    {
      "epoch": 0.10392941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3489,
      "step": 4417
    },
    {
      "epoch": 0.10395294117647058,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0674,
      "step": 4418
    },
    {
      "epoch": 0.1039764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 4419
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0801,
      "step": 4420
    },
    {
      "epoch": 0.10402352941176471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2109,
      "step": 4421
    },
    {
      "epoch": 0.10404705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.3584,
      "step": 4422
    },
    {
      "epoch": 0.10407058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4881,
      "step": 4423
    },
    {
      "epoch": 0.10409411764705882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3518,
      "step": 4424
    },
    {
      "epoch": 0.10411764705882352,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 4425
    },
    {
      "epoch": 0.10414117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2364,
      "step": 4426
    },
    {
      "epoch": 0.10416470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.5965,
      "step": 4427
    },
    {
      "epoch": 0.10418823529411765,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1459,
      "step": 4428
    },
    {
      "epoch": 0.10421176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.5427,
      "step": 4429
    },
    {
      "epoch": 0.10423529411764706,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0608,
      "step": 4430
    },
    {
      "epoch": 0.10425882352941176,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.29,
      "step": 4431
    },
    {
      "epoch": 0.10428235294117646,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1609,
      "step": 4432
    },
    {
      "epoch": 0.10430588235294118,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3149,
      "step": 4433
    },
    {
      "epoch": 0.10432941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5091,
      "step": 4434
    },
    {
      "epoch": 0.10435294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.349,
      "step": 4435
    },
    {
      "epoch": 0.10437647058823529,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.0931,
      "step": 4436
    },
    {
      "epoch": 0.1044,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 4437
    },
    {
      "epoch": 0.1044235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3339,
      "step": 4438
    },
    {
      "epoch": 0.1044470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0013,
      "step": 4439
    },
    {
      "epoch": 0.10447058823529412,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5192,
      "step": 4440
    },
    {
      "epoch": 0.10449411764705882,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4428,
      "step": 4441
    },
    {
      "epoch": 0.10451764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3095,
      "step": 4442
    },
    {
      "epoch": 0.10454117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4516,
      "step": 4443
    },
    {
      "epoch": 0.10456470588235293,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9806,
      "step": 4444
    },
    {
      "epoch": 0.10458823529411765,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4297,
      "step": 4445
    },
    {
      "epoch": 0.10461176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4033,
      "step": 4446
    },
    {
      "epoch": 0.10463529411764706,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2437,
      "step": 4447
    },
    {
      "epoch": 0.10465882352941176,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.2465,
      "step": 4448
    },
    {
      "epoch": 0.10468235294117648,
      "grad_norm": 0.30078125,
      "learning_rate": 0.02,
      "loss": 1.0181,
      "step": 4449
    },
    {
      "epoch": 0.10470588235294118,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2566,
      "step": 4450
    },
    {
      "epoch": 0.10472941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3373,
      "step": 4451
    },
    {
      "epoch": 0.10475294117647059,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2337,
      "step": 4452
    },
    {
      "epoch": 0.10477647058823529,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2057,
      "step": 4453
    },
    {
      "epoch": 0.1048,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2028,
      "step": 4454
    },
    {
      "epoch": 0.1048235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3223,
      "step": 4455
    },
    {
      "epoch": 0.10484705882352942,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2338,
      "step": 4456
    },
    {
      "epoch": 0.10487058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3944,
      "step": 4457
    },
    {
      "epoch": 0.10489411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3734,
      "step": 4458
    },
    {
      "epoch": 0.10491764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2767,
      "step": 4459
    },
    {
      "epoch": 0.10494117647058823,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3083,
      "step": 4460
    },
    {
      "epoch": 0.10496470588235295,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4956,
      "step": 4461
    },
    {
      "epoch": 0.10498823529411765,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4943,
      "step": 4462
    },
    {
      "epoch": 0.10501176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4855,
      "step": 4463
    },
    {
      "epoch": 0.10503529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3245,
      "step": 4464
    },
    {
      "epoch": 0.10505882352941176,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4529,
      "step": 4465
    },
    {
      "epoch": 0.10508235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2923,
      "step": 4466
    },
    {
      "epoch": 0.10510588235294117,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.9987,
      "step": 4467
    },
    {
      "epoch": 0.10512941176470589,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2438,
      "step": 4468
    },
    {
      "epoch": 0.10515294117647059,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3249,
      "step": 4469
    },
    {
      "epoch": 0.10517647058823529,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2665,
      "step": 4470
    },
    {
      "epoch": 0.1052,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3144,
      "step": 4471
    },
    {
      "epoch": 0.1052235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3039,
      "step": 4472
    },
    {
      "epoch": 0.10524705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 4473
    },
    {
      "epoch": 0.10527058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2067,
      "step": 4474
    },
    {
      "epoch": 0.10529411764705883,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2524,
      "step": 4475
    },
    {
      "epoch": 0.10531764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1342,
      "step": 4476
    },
    {
      "epoch": 0.10534117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.379,
      "step": 4477
    },
    {
      "epoch": 0.10536470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2117,
      "step": 4478
    },
    {
      "epoch": 0.10538823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3415,
      "step": 4479
    },
    {
      "epoch": 0.10541176470588236,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2777,
      "step": 4480
    },
    {
      "epoch": 0.10543529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2167,
      "step": 4481
    },
    {
      "epoch": 0.10545882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5508,
      "step": 4482
    },
    {
      "epoch": 0.10548235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3201,
      "step": 4483
    },
    {
      "epoch": 0.10550588235294117,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3851,
      "step": 4484
    },
    {
      "epoch": 0.10552941176470589,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 0.8748,
      "step": 4485
    },
    {
      "epoch": 0.10555294117647059,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2465,
      "step": 4486
    },
    {
      "epoch": 0.1055764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4111,
      "step": 4487
    },
    {
      "epoch": 0.1056,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3672,
      "step": 4488
    },
    {
      "epoch": 0.1056235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5435,
      "step": 4489
    },
    {
      "epoch": 0.10564705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4836,
      "step": 4490
    },
    {
      "epoch": 0.10567058823529411,
      "grad_norm": 0.3046875,
      "learning_rate": 0.02,
      "loss": 0.8126,
      "step": 4491
    },
    {
      "epoch": 0.10569411764705883,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1486,
      "step": 4492
    },
    {
      "epoch": 0.10571764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3582,
      "step": 4493
    },
    {
      "epoch": 0.10574117647058824,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0778,
      "step": 4494
    },
    {
      "epoch": 0.10576470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.389,
      "step": 4495
    },
    {
      "epoch": 0.10578823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5841,
      "step": 4496
    },
    {
      "epoch": 0.10581176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3946,
      "step": 4497
    },
    {
      "epoch": 0.10583529411764706,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0633,
      "step": 4498
    },
    {
      "epoch": 0.10585882352941177,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1892,
      "step": 4499
    },
    {
      "epoch": 0.10588235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2309,
      "step": 4500
    },
    {
      "epoch": 0.10590588235294118,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1148,
      "step": 4501
    },
    {
      "epoch": 0.10592941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.41,
      "step": 4502
    },
    {
      "epoch": 0.10595294117647058,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2874,
      "step": 4503
    },
    {
      "epoch": 0.1059764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2861,
      "step": 4504
    },
    {
      "epoch": 0.106,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.138,
      "step": 4505
    },
    {
      "epoch": 0.10602352941176471,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 4506
    },
    {
      "epoch": 0.10604705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.6451,
      "step": 4507
    },
    {
      "epoch": 0.10607058823529411,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1784,
      "step": 4508
    },
    {
      "epoch": 0.10609411764705882,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8376,
      "step": 4509
    },
    {
      "epoch": 0.10611764705882352,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3758,
      "step": 4510
    },
    {
      "epoch": 0.10614117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3967,
      "step": 4511
    },
    {
      "epoch": 0.10616470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9266,
      "step": 4512
    },
    {
      "epoch": 0.10618823529411765,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.0922,
      "step": 4513
    },
    {
      "epoch": 0.10621176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2045,
      "step": 4514
    },
    {
      "epoch": 0.10623529411764705,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2113,
      "step": 4515
    },
    {
      "epoch": 0.10625882352941177,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5162,
      "step": 4516
    },
    {
      "epoch": 0.10628235294117647,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 1.0409,
      "step": 4517
    },
    {
      "epoch": 0.10630588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2739,
      "step": 4518
    },
    {
      "epoch": 0.10632941176470588,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.2268,
      "step": 4519
    },
    {
      "epoch": 0.1063529411764706,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3361,
      "step": 4520
    },
    {
      "epoch": 0.1063764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1509,
      "step": 4521
    },
    {
      "epoch": 0.1064,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3491,
      "step": 4522
    },
    {
      "epoch": 0.10642352941176471,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.279,
      "step": 4523
    },
    {
      "epoch": 0.10644705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3261,
      "step": 4524
    },
    {
      "epoch": 0.10647058823529412,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3044,
      "step": 4525
    },
    {
      "epoch": 0.10649411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3944,
      "step": 4526
    },
    {
      "epoch": 0.10651764705882354,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3169,
      "step": 4527
    },
    {
      "epoch": 0.10654117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3841,
      "step": 4528
    },
    {
      "epoch": 0.10656470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1691,
      "step": 4529
    },
    {
      "epoch": 0.10658823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3026,
      "step": 4530
    },
    {
      "epoch": 0.10661176470588235,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1224,
      "step": 4531
    },
    {
      "epoch": 0.10663529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4961,
      "step": 4532
    },
    {
      "epoch": 0.10665882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3958,
      "step": 4533
    },
    {
      "epoch": 0.10668235294117646,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4851,
      "step": 4534
    },
    {
      "epoch": 0.10670588235294118,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4215,
      "step": 4535
    },
    {
      "epoch": 0.10672941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2619,
      "step": 4536
    },
    {
      "epoch": 0.10675294117647059,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.226,
      "step": 4537
    },
    {
      "epoch": 0.10677647058823529,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4722,
      "step": 4538
    },
    {
      "epoch": 0.1068,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.006,
      "step": 4539
    },
    {
      "epoch": 0.1068235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3293,
      "step": 4540
    },
    {
      "epoch": 0.1068470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1897,
      "step": 4541
    },
    {
      "epoch": 0.10687058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2792,
      "step": 4542
    },
    {
      "epoch": 0.10689411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4154,
      "step": 4543
    },
    {
      "epoch": 0.10691764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3732,
      "step": 4544
    },
    {
      "epoch": 0.10694117647058823,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2429,
      "step": 4545
    },
    {
      "epoch": 0.10696470588235295,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.311,
      "step": 4546
    },
    {
      "epoch": 0.10698823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2238,
      "step": 4547
    },
    {
      "epoch": 0.10701176470588235,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.47,
      "step": 4548
    },
    {
      "epoch": 0.10703529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1993,
      "step": 4549
    },
    {
      "epoch": 0.10705882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2255,
      "step": 4550
    },
    {
      "epoch": 0.10708235294117648,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2584,
      "step": 4551
    },
    {
      "epoch": 0.10710588235294118,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1403,
      "step": 4552
    },
    {
      "epoch": 0.10712941176470588,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2511,
      "step": 4553
    },
    {
      "epoch": 0.10715294117647059,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9762,
      "step": 4554
    },
    {
      "epoch": 0.10717647058823529,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4594,
      "step": 4555
    },
    {
      "epoch": 0.1072,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5063,
      "step": 4556
    },
    {
      "epoch": 0.1072235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0082,
      "step": 4557
    },
    {
      "epoch": 0.10724705882352942,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3375,
      "step": 4558
    },
    {
      "epoch": 0.10727058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1539,
      "step": 4559
    },
    {
      "epoch": 0.10729411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3549,
      "step": 4560
    },
    {
      "epoch": 0.10731764705882353,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1882,
      "step": 4561
    },
    {
      "epoch": 0.10734117647058823,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4391,
      "step": 4562
    },
    {
      "epoch": 0.10736470588235295,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.4178,
      "step": 4563
    },
    {
      "epoch": 0.10738823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2617,
      "step": 4564
    },
    {
      "epoch": 0.10741176470588236,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0696,
      "step": 4565
    },
    {
      "epoch": 0.10743529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.235,
      "step": 4566
    },
    {
      "epoch": 0.10745882352941176,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0562,
      "step": 4567
    },
    {
      "epoch": 0.10748235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3405,
      "step": 4568
    },
    {
      "epoch": 0.10750588235294117,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3195,
      "step": 4569
    },
    {
      "epoch": 0.10752941176470589,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3109,
      "step": 4570
    },
    {
      "epoch": 0.10755294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1357,
      "step": 4571
    },
    {
      "epoch": 0.10757647058823529,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 4572
    },
    {
      "epoch": 0.1076,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3391,
      "step": 4573
    },
    {
      "epoch": 0.1076235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4354,
      "step": 4574
    },
    {
      "epoch": 0.10764705882352942,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4608,
      "step": 4575
    },
    {
      "epoch": 0.10767058823529412,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2602,
      "step": 4576
    },
    {
      "epoch": 0.10769411764705883,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0194,
      "step": 4577
    },
    {
      "epoch": 0.10771764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2927,
      "step": 4578
    },
    {
      "epoch": 0.10774117647058823,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3458,
      "step": 4579
    },
    {
      "epoch": 0.10776470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3084,
      "step": 4580
    },
    {
      "epoch": 0.10778823529411764,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2541,
      "step": 4581
    },
    {
      "epoch": 0.10781176470588236,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1576,
      "step": 4582
    },
    {
      "epoch": 0.10783529411764706,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0971,
      "step": 4583
    },
    {
      "epoch": 0.10785882352941177,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2608,
      "step": 4584
    },
    {
      "epoch": 0.10788235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3873,
      "step": 4585
    },
    {
      "epoch": 0.10790588235294117,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2434,
      "step": 4586
    },
    {
      "epoch": 0.10792941176470588,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1802,
      "step": 4587
    },
    {
      "epoch": 0.10795294117647058,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0869,
      "step": 4588
    },
    {
      "epoch": 0.1079764705882353,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0216,
      "step": 4589
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3886,
      "step": 4590
    },
    {
      "epoch": 0.10802352941176471,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.293,
      "step": 4591
    },
    {
      "epoch": 0.10804705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.155,
      "step": 4592
    },
    {
      "epoch": 0.10807058823529411,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.2809,
      "step": 4593
    },
    {
      "epoch": 0.10809411764705883,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0681,
      "step": 4594
    },
    {
      "epoch": 0.10811764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.3243,
      "step": 4595
    },
    {
      "epoch": 0.10814117647058824,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2631,
      "step": 4596
    },
    {
      "epoch": 0.10816470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2672,
      "step": 4597
    },
    {
      "epoch": 0.10818823529411764,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2509,
      "step": 4598
    },
    {
      "epoch": 0.10821176470588235,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.2629,
      "step": 4599
    },
    {
      "epoch": 0.10823529411764705,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.4851,
      "step": 4600
    },
    {
      "epoch": 0.10825882352941177,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.376,
      "step": 4601
    },
    {
      "epoch": 0.10828235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.329,
      "step": 4602
    },
    {
      "epoch": 0.10830588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4483,
      "step": 4603
    },
    {
      "epoch": 0.10832941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4359,
      "step": 4604
    },
    {
      "epoch": 0.10835294117647058,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.0864,
      "step": 4605
    },
    {
      "epoch": 0.1083764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.356,
      "step": 4606
    },
    {
      "epoch": 0.1084,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3145,
      "step": 4607
    },
    {
      "epoch": 0.10842352941176471,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1337,
      "step": 4608
    },
    {
      "epoch": 0.10844705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.277,
      "step": 4609
    },
    {
      "epoch": 0.10847058823529412,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.209,
      "step": 4610
    },
    {
      "epoch": 0.10849411764705882,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2323,
      "step": 4611
    },
    {
      "epoch": 0.10851764705882352,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3221,
      "step": 4612
    },
    {
      "epoch": 0.10854117647058824,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.7745,
      "step": 4613
    },
    {
      "epoch": 0.10856470588235294,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2684,
      "step": 4614
    },
    {
      "epoch": 0.10858823529411765,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1934,
      "step": 4615
    },
    {
      "epoch": 0.10861176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3548,
      "step": 4616
    },
    {
      "epoch": 0.10863529411764705,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1408,
      "step": 4617
    },
    {
      "epoch": 0.10865882352941177,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2381,
      "step": 4618
    },
    {
      "epoch": 0.10868235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3089,
      "step": 4619
    },
    {
      "epoch": 0.10870588235294118,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 4620
    },
    {
      "epoch": 0.10872941176470588,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1143,
      "step": 4621
    },
    {
      "epoch": 0.1087529411764706,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0678,
      "step": 4622
    },
    {
      "epoch": 0.1087764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5258,
      "step": 4623
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3115,
      "step": 4624
    },
    {
      "epoch": 0.10882352941176471,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 4625
    },
    {
      "epoch": 0.10884705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2448,
      "step": 4626
    },
    {
      "epoch": 0.10887058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3128,
      "step": 4627
    },
    {
      "epoch": 0.10889411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3507,
      "step": 4628
    },
    {
      "epoch": 0.10891764705882354,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1485,
      "step": 4629
    },
    {
      "epoch": 0.10894117647058824,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3127,
      "step": 4630
    },
    {
      "epoch": 0.10896470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1869,
      "step": 4631
    },
    {
      "epoch": 0.10898823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3146,
      "step": 4632
    },
    {
      "epoch": 0.10901176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3009,
      "step": 4633
    },
    {
      "epoch": 0.10903529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.557,
      "step": 4634
    },
    {
      "epoch": 0.10905882352941176,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3607,
      "step": 4635
    },
    {
      "epoch": 0.10908235294117648,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3484,
      "step": 4636
    },
    {
      "epoch": 0.10910588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1739,
      "step": 4637
    },
    {
      "epoch": 0.10912941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4161,
      "step": 4638
    },
    {
      "epoch": 0.10915294117647059,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1017,
      "step": 4639
    },
    {
      "epoch": 0.10917647058823529,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2912,
      "step": 4640
    },
    {
      "epoch": 0.1092,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2625,
      "step": 4641
    },
    {
      "epoch": 0.1092235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1747,
      "step": 4642
    },
    {
      "epoch": 0.1092470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4382,
      "step": 4643
    },
    {
      "epoch": 0.10927058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1772,
      "step": 4644
    },
    {
      "epoch": 0.10929411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3385,
      "step": 4645
    },
    {
      "epoch": 0.10931764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 4646
    },
    {
      "epoch": 0.10934117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4023,
      "step": 4647
    },
    {
      "epoch": 0.10936470588235295,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9918,
      "step": 4648
    },
    {
      "epoch": 0.10938823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1291,
      "step": 4649
    },
    {
      "epoch": 0.10941176470588235,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0697,
      "step": 4650
    },
    {
      "epoch": 0.10943529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4895,
      "step": 4651
    },
    {
      "epoch": 0.10945882352941176,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.132,
      "step": 4652
    },
    {
      "epoch": 0.10948235294117648,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1291,
      "step": 4653
    },
    {
      "epoch": 0.10950588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.403,
      "step": 4654
    },
    {
      "epoch": 0.10952941176470589,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1188,
      "step": 4655
    },
    {
      "epoch": 0.10955294117647059,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1735,
      "step": 4656
    },
    {
      "epoch": 0.10957647058823529,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3072,
      "step": 4657
    },
    {
      "epoch": 0.1096,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.364,
      "step": 4658
    },
    {
      "epoch": 0.1096235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1688,
      "step": 4659
    },
    {
      "epoch": 0.10964705882352942,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2369,
      "step": 4660
    },
    {
      "epoch": 0.10967058823529412,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.4147,
      "step": 4661
    },
    {
      "epoch": 0.10969411764705882,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.2313,
      "step": 4662
    },
    {
      "epoch": 0.10971764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4934,
      "step": 4663
    },
    {
      "epoch": 0.10974117647058823,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.1079,
      "step": 4664
    },
    {
      "epoch": 0.10976470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2685,
      "step": 4665
    },
    {
      "epoch": 0.10978823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.487,
      "step": 4666
    },
    {
      "epoch": 0.10981176470588236,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.4113,
      "step": 4667
    },
    {
      "epoch": 0.10983529411764706,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2824,
      "step": 4668
    },
    {
      "epoch": 0.10985882352941176,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1249,
      "step": 4669
    },
    {
      "epoch": 0.10988235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3102,
      "step": 4670
    },
    {
      "epoch": 0.10990588235294117,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.0227,
      "step": 4671
    },
    {
      "epoch": 0.10992941176470589,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4039,
      "step": 4672
    },
    {
      "epoch": 0.10995294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3117,
      "step": 4673
    },
    {
      "epoch": 0.1099764705882353,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.345,
      "step": 4674
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.7072,
      "step": 4675
    },
    {
      "epoch": 0.1100235294117647,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2617,
      "step": 4676
    },
    {
      "epoch": 0.11004705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2462,
      "step": 4677
    },
    {
      "epoch": 0.11007058823529411,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2901,
      "step": 4678
    },
    {
      "epoch": 0.11009411764705883,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4903,
      "step": 4679
    },
    {
      "epoch": 0.11011764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.3365,
      "step": 4680
    },
    {
      "epoch": 0.11014117647058823,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1133,
      "step": 4681
    },
    {
      "epoch": 0.11016470588235294,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.26,
      "step": 4682
    },
    {
      "epoch": 0.11018823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.361,
      "step": 4683
    },
    {
      "epoch": 0.11021176470588236,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4524,
      "step": 4684
    },
    {
      "epoch": 0.11023529411764706,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.413,
      "step": 4685
    },
    {
      "epoch": 0.11025882352941177,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4775,
      "step": 4686
    },
    {
      "epoch": 0.11028235294117647,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.9563,
      "step": 4687
    },
    {
      "epoch": 0.11030588235294117,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2397,
      "step": 4688
    },
    {
      "epoch": 0.11032941176470588,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3318,
      "step": 4689
    },
    {
      "epoch": 0.11035294117647058,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4206,
      "step": 4690
    },
    {
      "epoch": 0.1103764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 4691
    },
    {
      "epoch": 0.1104,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3588,
      "step": 4692
    },
    {
      "epoch": 0.11042352941176471,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.4884,
      "step": 4693
    },
    {
      "epoch": 0.11044705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.4329,
      "step": 4694
    },
    {
      "epoch": 0.11047058823529411,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.7069,
      "step": 4695
    },
    {
      "epoch": 0.11049411764705883,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 1.0771,
      "step": 4696
    },
    {
      "epoch": 0.11051764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3204,
      "step": 4697
    },
    {
      "epoch": 0.11054117647058824,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2635,
      "step": 4698
    },
    {
      "epoch": 0.11056470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1974,
      "step": 4699
    },
    {
      "epoch": 0.11058823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.7533,
      "step": 4700
    },
    {
      "epoch": 0.11061176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4941,
      "step": 4701
    },
    {
      "epoch": 0.11063529411764705,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 4702
    },
    {
      "epoch": 0.11065882352941177,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1579,
      "step": 4703
    },
    {
      "epoch": 0.11068235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1096,
      "step": 4704
    },
    {
      "epoch": 0.11070588235294118,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2452,
      "step": 4705
    },
    {
      "epoch": 0.11072941176470588,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1635,
      "step": 4706
    },
    {
      "epoch": 0.11075294117647058,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4796,
      "step": 4707
    },
    {
      "epoch": 0.1107764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1961,
      "step": 4708
    },
    {
      "epoch": 0.1108,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.33,
      "step": 4709
    },
    {
      "epoch": 0.11082352941176471,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4983,
      "step": 4710
    },
    {
      "epoch": 0.11084705882352941,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2743,
      "step": 4711
    },
    {
      "epoch": 0.11087058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3198,
      "step": 4712
    },
    {
      "epoch": 0.11089411764705882,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2101,
      "step": 4713
    },
    {
      "epoch": 0.11091764705882352,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9631,
      "step": 4714
    },
    {
      "epoch": 0.11094117647058824,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2818,
      "step": 4715
    },
    {
      "epoch": 0.11096470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.2853,
      "step": 4716
    },
    {
      "epoch": 0.11098823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5435,
      "step": 4717
    },
    {
      "epoch": 0.11101176470588235,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2686,
      "step": 4718
    },
    {
      "epoch": 0.11103529411764707,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3162,
      "step": 4719
    },
    {
      "epoch": 0.11105882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4349,
      "step": 4720
    },
    {
      "epoch": 0.11108235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3978,
      "step": 4721
    },
    {
      "epoch": 0.11110588235294118,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0583,
      "step": 4722
    },
    {
      "epoch": 0.11112941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2481,
      "step": 4723
    },
    {
      "epoch": 0.1111529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2307,
      "step": 4724
    },
    {
      "epoch": 0.1111764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3233,
      "step": 4725
    },
    {
      "epoch": 0.1112,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0013,
      "step": 4726
    },
    {
      "epoch": 0.11122352941176471,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2754,
      "step": 4727
    },
    {
      "epoch": 0.11124705882352941,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2082,
      "step": 4728
    },
    {
      "epoch": 0.11127058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.34,
      "step": 4729
    },
    {
      "epoch": 0.11129411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3502,
      "step": 4730
    },
    {
      "epoch": 0.11131764705882354,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2404,
      "step": 4731
    },
    {
      "epoch": 0.11134117647058824,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1675,
      "step": 4732
    },
    {
      "epoch": 0.11136470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2721,
      "step": 4733
    },
    {
      "epoch": 0.11138823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4399,
      "step": 4734
    },
    {
      "epoch": 0.11141176470588235,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2736,
      "step": 4735
    },
    {
      "epoch": 0.11143529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2418,
      "step": 4736
    },
    {
      "epoch": 0.11145882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2376,
      "step": 4737
    },
    {
      "epoch": 0.11148235294117648,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2675,
      "step": 4738
    },
    {
      "epoch": 0.11150588235294118,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1795,
      "step": 4739
    },
    {
      "epoch": 0.11152941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3667,
      "step": 4740
    },
    {
      "epoch": 0.11155294117647059,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4782,
      "step": 4741
    },
    {
      "epoch": 0.11157647058823529,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1476,
      "step": 4742
    },
    {
      "epoch": 0.1116,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5402,
      "step": 4743
    },
    {
      "epoch": 0.1116235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 4744
    },
    {
      "epoch": 0.11164705882352942,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3253,
      "step": 4745
    },
    {
      "epoch": 0.11167058823529412,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1198,
      "step": 4746
    },
    {
      "epoch": 0.11169411764705882,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1961,
      "step": 4747
    },
    {
      "epoch": 0.11171764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2841,
      "step": 4748
    },
    {
      "epoch": 0.11174117647058823,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4069,
      "step": 4749
    },
    {
      "epoch": 0.11176470588235295,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3216,
      "step": 4750
    },
    {
      "epoch": 0.11178823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.282,
      "step": 4751
    },
    {
      "epoch": 0.11181176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1545,
      "step": 4752
    },
    {
      "epoch": 0.11183529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2879,
      "step": 4753
    },
    {
      "epoch": 0.11185882352941176,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.5987,
      "step": 4754
    },
    {
      "epoch": 0.11188235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0774,
      "step": 4755
    },
    {
      "epoch": 0.11190588235294117,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1561,
      "step": 4756
    },
    {
      "epoch": 0.11192941176470589,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2394,
      "step": 4757
    },
    {
      "epoch": 0.11195294117647059,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.4412,
      "step": 4758
    },
    {
      "epoch": 0.11197647058823529,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0471,
      "step": 4759
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.9937,
      "step": 4760
    },
    {
      "epoch": 0.1120235294117647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0634,
      "step": 4761
    },
    {
      "epoch": 0.11204705882352942,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5823,
      "step": 4762
    },
    {
      "epoch": 0.11207058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 4763
    },
    {
      "epoch": 0.11209411764705883,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1584,
      "step": 4764
    },
    {
      "epoch": 0.11211764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4859,
      "step": 4765
    },
    {
      "epoch": 0.11214117647058823,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2154,
      "step": 4766
    },
    {
      "epoch": 0.11216470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.234,
      "step": 4767
    },
    {
      "epoch": 0.11218823529411764,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3698,
      "step": 4768
    },
    {
      "epoch": 0.11221176470588236,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3879,
      "step": 4769
    },
    {
      "epoch": 0.11223529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4226,
      "step": 4770
    },
    {
      "epoch": 0.11225882352941176,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2971,
      "step": 4771
    },
    {
      "epoch": 0.11228235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2371,
      "step": 4772
    },
    {
      "epoch": 0.11230588235294117,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.164,
      "step": 4773
    },
    {
      "epoch": 0.11232941176470589,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1492,
      "step": 4774
    },
    {
      "epoch": 0.11235294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2764,
      "step": 4775
    },
    {
      "epoch": 0.1123764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3504,
      "step": 4776
    },
    {
      "epoch": 0.1124,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.131,
      "step": 4777
    },
    {
      "epoch": 0.1124235294117647,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2555,
      "step": 4778
    },
    {
      "epoch": 0.11244705882352941,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2485,
      "step": 4779
    },
    {
      "epoch": 0.11247058823529411,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.3705,
      "step": 4780
    },
    {
      "epoch": 0.11249411764705883,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3711,
      "step": 4781
    },
    {
      "epoch": 0.11251764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5381,
      "step": 4782
    },
    {
      "epoch": 0.11254117647058824,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.5151,
      "step": 4783
    },
    {
      "epoch": 0.11256470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3558,
      "step": 4784
    },
    {
      "epoch": 0.11258823529411764,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3066,
      "step": 4785
    },
    {
      "epoch": 0.11261176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.265,
      "step": 4786
    },
    {
      "epoch": 0.11263529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2029,
      "step": 4787
    },
    {
      "epoch": 0.11265882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.5934,
      "step": 4788
    },
    {
      "epoch": 0.11268235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1925,
      "step": 4789
    },
    {
      "epoch": 0.11270588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3723,
      "step": 4790
    },
    {
      "epoch": 0.11272941176470588,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2248,
      "step": 4791
    },
    {
      "epoch": 0.11275294117647058,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.9182,
      "step": 4792
    },
    {
      "epoch": 0.1127764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1733,
      "step": 4793
    },
    {
      "epoch": 0.1128,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.2633,
      "step": 4794
    },
    {
      "epoch": 0.11282352941176471,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9218,
      "step": 4795
    },
    {
      "epoch": 0.11284705882352941,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.0455,
      "step": 4796
    },
    {
      "epoch": 0.11287058823529411,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4985,
      "step": 4797
    },
    {
      "epoch": 0.11289411764705883,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3855,
      "step": 4798
    },
    {
      "epoch": 0.11291764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2855,
      "step": 4799
    },
    {
      "epoch": 0.11294117647058824,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1828,
      "step": 4800
    },
    {
      "epoch": 0.11296470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0393,
      "step": 4801
    },
    {
      "epoch": 0.11298823529411765,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.3818,
      "step": 4802
    },
    {
      "epoch": 0.11301176470588235,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3176,
      "step": 4803
    },
    {
      "epoch": 0.11303529411764705,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3184,
      "step": 4804
    },
    {
      "epoch": 0.11305882352941177,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.3687,
      "step": 4805
    },
    {
      "epoch": 0.11308235294117647,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1672,
      "step": 4806
    },
    {
      "epoch": 0.11310588235294118,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2493,
      "step": 4807
    },
    {
      "epoch": 0.11312941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4686,
      "step": 4808
    },
    {
      "epoch": 0.1131529411764706,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 1.0683,
      "step": 4809
    },
    {
      "epoch": 0.1131764705882353,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9387,
      "step": 4810
    },
    {
      "epoch": 0.1132,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.483,
      "step": 4811
    },
    {
      "epoch": 0.11322352941176471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.338,
      "step": 4812
    },
    {
      "epoch": 0.11324705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.639,
      "step": 4813
    },
    {
      "epoch": 0.11327058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2637,
      "step": 4814
    },
    {
      "epoch": 0.11329411764705882,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9796,
      "step": 4815
    },
    {
      "epoch": 0.11331764705882352,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.391,
      "step": 4816
    },
    {
      "epoch": 0.11334117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2249,
      "step": 4817
    },
    {
      "epoch": 0.11336470588235294,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.326,
      "step": 4818
    },
    {
      "epoch": 0.11338823529411765,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9492,
      "step": 4819
    },
    {
      "epoch": 0.11341176470588235,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0318,
      "step": 4820
    },
    {
      "epoch": 0.11343529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4656,
      "step": 4821
    },
    {
      "epoch": 0.11345882352941176,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5516,
      "step": 4822
    },
    {
      "epoch": 0.11348235294117646,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0324,
      "step": 4823
    },
    {
      "epoch": 0.11350588235294118,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 4824
    },
    {
      "epoch": 0.11352941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4121,
      "step": 4825
    },
    {
      "epoch": 0.11355294117647059,
      "grad_norm": 0.302734375,
      "learning_rate": 0.02,
      "loss": 1.0748,
      "step": 4826
    },
    {
      "epoch": 0.11357647058823529,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2057,
      "step": 4827
    },
    {
      "epoch": 0.1136,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2471,
      "step": 4828
    },
    {
      "epoch": 0.1136235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3131,
      "step": 4829
    },
    {
      "epoch": 0.1136470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2357,
      "step": 4830
    },
    {
      "epoch": 0.11367058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3338,
      "step": 4831
    },
    {
      "epoch": 0.11369411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.388,
      "step": 4832
    },
    {
      "epoch": 0.11371764705882353,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.158,
      "step": 4833
    },
    {
      "epoch": 0.11374117647058823,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2156,
      "step": 4834
    },
    {
      "epoch": 0.11376470588235293,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1378,
      "step": 4835
    },
    {
      "epoch": 0.11378823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3101,
      "step": 4836
    },
    {
      "epoch": 0.11381176470588235,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.957,
      "step": 4837
    },
    {
      "epoch": 0.11383529411764706,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1315,
      "step": 4838
    },
    {
      "epoch": 0.11385882352941176,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1974,
      "step": 4839
    },
    {
      "epoch": 0.11388235294117648,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3491,
      "step": 4840
    },
    {
      "epoch": 0.11390588235294118,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1801,
      "step": 4841
    },
    {
      "epoch": 0.11392941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3679,
      "step": 4842
    },
    {
      "epoch": 0.11395294117647059,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1539,
      "step": 4843
    },
    {
      "epoch": 0.11397647058823529,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3248,
      "step": 4844
    },
    {
      "epoch": 0.114,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1297,
      "step": 4845
    },
    {
      "epoch": 0.1140235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1599,
      "step": 4846
    },
    {
      "epoch": 0.11404705882352942,
      "grad_norm": 0.314453125,
      "learning_rate": 0.02,
      "loss": 0.9385,
      "step": 4847
    },
    {
      "epoch": 0.11407058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.5653,
      "step": 4848
    },
    {
      "epoch": 0.11409411764705882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0489,
      "step": 4849
    },
    {
      "epoch": 0.11411764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3029,
      "step": 4850
    },
    {
      "epoch": 0.11414117647058823,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2033,
      "step": 4851
    },
    {
      "epoch": 0.11416470588235295,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2037,
      "step": 4852
    },
    {
      "epoch": 0.11418823529411765,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2909,
      "step": 4853
    },
    {
      "epoch": 0.11421176470588236,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0958,
      "step": 4854
    },
    {
      "epoch": 0.11423529411764706,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5299,
      "step": 4855
    },
    {
      "epoch": 0.11425882352941176,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 4856
    },
    {
      "epoch": 0.11428235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5178,
      "step": 4857
    },
    {
      "epoch": 0.11430588235294117,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0802,
      "step": 4858
    },
    {
      "epoch": 0.11432941176470589,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2356,
      "step": 4859
    },
    {
      "epoch": 0.11435294117647059,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 4860
    },
    {
      "epoch": 0.11437647058823529,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2818,
      "step": 4861
    },
    {
      "epoch": 0.1144,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2873,
      "step": 4862
    },
    {
      "epoch": 0.1144235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2561,
      "step": 4863
    },
    {
      "epoch": 0.11444705882352942,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1614,
      "step": 4864
    },
    {
      "epoch": 0.11447058823529412,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1735,
      "step": 4865
    },
    {
      "epoch": 0.11449411764705883,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.4788,
      "step": 4866
    },
    {
      "epoch": 0.11451764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9505,
      "step": 4867
    },
    {
      "epoch": 0.11454117647058823,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.298,
      "step": 4868
    },
    {
      "epoch": 0.11456470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3962,
      "step": 4869
    },
    {
      "epoch": 0.11458823529411764,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3798,
      "step": 4870
    },
    {
      "epoch": 0.11461176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2764,
      "step": 4871
    },
    {
      "epoch": 0.11463529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3512,
      "step": 4872
    },
    {
      "epoch": 0.11465882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3772,
      "step": 4873
    },
    {
      "epoch": 0.11468235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 4874
    },
    {
      "epoch": 0.11470588235294117,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2835,
      "step": 4875
    },
    {
      "epoch": 0.11472941176470589,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3855,
      "step": 4876
    },
    {
      "epoch": 0.11475294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4842,
      "step": 4877
    },
    {
      "epoch": 0.1147764705882353,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1332,
      "step": 4878
    },
    {
      "epoch": 0.1148,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.3108,
      "step": 4879
    },
    {
      "epoch": 0.1148235294117647,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 4880
    },
    {
      "epoch": 0.11484705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3053,
      "step": 4881
    },
    {
      "epoch": 0.11487058823529411,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4271,
      "step": 4882
    },
    {
      "epoch": 0.11489411764705883,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4267,
      "step": 4883
    },
    {
      "epoch": 0.11491764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3337,
      "step": 4884
    },
    {
      "epoch": 0.11494117647058824,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0238,
      "step": 4885
    },
    {
      "epoch": 0.11496470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.5905,
      "step": 4886
    },
    {
      "epoch": 0.11498823529411764,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 4887
    },
    {
      "epoch": 0.11501176470588236,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.128,
      "step": 4888
    },
    {
      "epoch": 0.11503529411764706,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2596,
      "step": 4889
    },
    {
      "epoch": 0.11505882352941177,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1201,
      "step": 4890
    },
    {
      "epoch": 0.11508235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1035,
      "step": 4891
    },
    {
      "epoch": 0.11510588235294118,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1907,
      "step": 4892
    },
    {
      "epoch": 0.11512941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2375,
      "step": 4893
    },
    {
      "epoch": 0.11515294117647058,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.5076,
      "step": 4894
    },
    {
      "epoch": 0.1151764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1678,
      "step": 4895
    },
    {
      "epoch": 0.1152,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2153,
      "step": 4896
    },
    {
      "epoch": 0.11522352941176471,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3636,
      "step": 4897
    },
    {
      "epoch": 0.11524705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2063,
      "step": 4898
    },
    {
      "epoch": 0.11527058823529411,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.378,
      "step": 4899
    },
    {
      "epoch": 0.11529411764705882,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0356,
      "step": 4900
    },
    {
      "epoch": 0.11531764705882352,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1817,
      "step": 4901
    },
    {
      "epoch": 0.11534117647058824,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4248,
      "step": 4902
    },
    {
      "epoch": 0.11536470588235294,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2352,
      "step": 4903
    },
    {
      "epoch": 0.11538823529411765,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0391,
      "step": 4904
    },
    {
      "epoch": 0.11541176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3592,
      "step": 4905
    },
    {
      "epoch": 0.11543529411764705,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2196,
      "step": 4906
    },
    {
      "epoch": 0.11545882352941177,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5717,
      "step": 4907
    },
    {
      "epoch": 0.11548235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1955,
      "step": 4908
    },
    {
      "epoch": 0.11550588235294118,
      "grad_norm": 0.314453125,
      "learning_rate": 0.02,
      "loss": 0.8435,
      "step": 4909
    },
    {
      "epoch": 0.11552941176470588,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1624,
      "step": 4910
    },
    {
      "epoch": 0.1155529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0956,
      "step": 4911
    },
    {
      "epoch": 0.1155764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0893,
      "step": 4912
    },
    {
      "epoch": 0.1156,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2266,
      "step": 4913
    },
    {
      "epoch": 0.11562352941176471,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4319,
      "step": 4914
    },
    {
      "epoch": 0.11564705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3728,
      "step": 4915
    },
    {
      "epoch": 0.11567058823529412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0122,
      "step": 4916
    },
    {
      "epoch": 0.11569411764705882,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9912,
      "step": 4917
    },
    {
      "epoch": 0.11571764705882354,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1211,
      "step": 4918
    },
    {
      "epoch": 0.11574117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.4081,
      "step": 4919
    },
    {
      "epoch": 0.11576470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3496,
      "step": 4920
    },
    {
      "epoch": 0.11578823529411765,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1774,
      "step": 4921
    },
    {
      "epoch": 0.11581176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 4922
    },
    {
      "epoch": 0.11583529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4476,
      "step": 4923
    },
    {
      "epoch": 0.11585882352941176,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2622,
      "step": 4924
    },
    {
      "epoch": 0.11588235294117646,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.4074,
      "step": 4925
    },
    {
      "epoch": 0.11590588235294118,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0206,
      "step": 4926
    },
    {
      "epoch": 0.11592941176470588,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2582,
      "step": 4927
    },
    {
      "epoch": 0.11595294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4626,
      "step": 4928
    },
    {
      "epoch": 0.11597647058823529,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3433,
      "step": 4929
    },
    {
      "epoch": 0.116,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2652,
      "step": 4930
    },
    {
      "epoch": 0.1160235294117647,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.1055,
      "step": 4931
    },
    {
      "epoch": 0.1160470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3271,
      "step": 4932
    },
    {
      "epoch": 0.11607058823529412,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1049,
      "step": 4933
    },
    {
      "epoch": 0.11609411764705882,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2839,
      "step": 4934
    },
    {
      "epoch": 0.11611764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3155,
      "step": 4935
    },
    {
      "epoch": 0.11614117647058823,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.4207,
      "step": 4936
    },
    {
      "epoch": 0.11616470588235295,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.5089,
      "step": 4937
    },
    {
      "epoch": 0.11618823529411765,
      "grad_norm": 0.3046875,
      "learning_rate": 0.02,
      "loss": 0.9728,
      "step": 4938
    },
    {
      "epoch": 0.11621176470588235,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.2174,
      "step": 4939
    },
    {
      "epoch": 0.11623529411764706,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.4475,
      "step": 4940
    },
    {
      "epoch": 0.11625882352941176,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3881,
      "step": 4941
    },
    {
      "epoch": 0.11628235294117648,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3187,
      "step": 4942
    },
    {
      "epoch": 0.11630588235294118,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1624,
      "step": 4943
    },
    {
      "epoch": 0.11632941176470588,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.014,
      "step": 4944
    },
    {
      "epoch": 0.11635294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1287,
      "step": 4945
    },
    {
      "epoch": 0.11637647058823529,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.4001,
      "step": 4946
    },
    {
      "epoch": 0.1164,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0769,
      "step": 4947
    },
    {
      "epoch": 0.1164235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.5206,
      "step": 4948
    },
    {
      "epoch": 0.11644705882352942,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3189,
      "step": 4949
    },
    {
      "epoch": 0.11647058823529412,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2412,
      "step": 4950
    },
    {
      "epoch": 0.11649411764705882,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.3439,
      "step": 4951
    },
    {
      "epoch": 0.11651764705882353,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 4952
    },
    {
      "epoch": 0.11654117647058823,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 1.0044,
      "step": 4953
    },
    {
      "epoch": 0.11656470588235295,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1804,
      "step": 4954
    },
    {
      "epoch": 0.11658823529411765,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2348,
      "step": 4955
    },
    {
      "epoch": 0.11661176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4127,
      "step": 4956
    },
    {
      "epoch": 0.11663529411764706,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3169,
      "step": 4957
    },
    {
      "epoch": 0.11665882352941176,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0392,
      "step": 4958
    },
    {
      "epoch": 0.11668235294117647,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.263,
      "step": 4959
    },
    {
      "epoch": 0.11670588235294117,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 4960
    },
    {
      "epoch": 0.11672941176470589,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.398,
      "step": 4961
    },
    {
      "epoch": 0.11675294117647059,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2931,
      "step": 4962
    },
    {
      "epoch": 0.1167764705882353,
      "grad_norm": 0.310546875,
      "learning_rate": 0.02,
      "loss": 0.9807,
      "step": 4963
    },
    {
      "epoch": 0.1168,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3823,
      "step": 4964
    },
    {
      "epoch": 0.1168235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3904,
      "step": 4965
    },
    {
      "epoch": 0.11684705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3519,
      "step": 4966
    },
    {
      "epoch": 0.11687058823529411,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3678,
      "step": 4967
    },
    {
      "epoch": 0.11689411764705883,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1331,
      "step": 4968
    },
    {
      "epoch": 0.11691764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.296,
      "step": 4969
    },
    {
      "epoch": 0.11694117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1538,
      "step": 4970
    },
    {
      "epoch": 0.11696470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.6094,
      "step": 4971
    },
    {
      "epoch": 0.11698823529411764,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1928,
      "step": 4972
    },
    {
      "epoch": 0.11701176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2955,
      "step": 4973
    },
    {
      "epoch": 0.11703529411764706,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1989,
      "step": 4974
    },
    {
      "epoch": 0.11705882352941177,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 0.8988,
      "step": 4975
    },
    {
      "epoch": 0.11708235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1341,
      "step": 4976
    },
    {
      "epoch": 0.11710588235294117,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2968,
      "step": 4977
    },
    {
      "epoch": 0.11712941176470588,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1488,
      "step": 4978
    },
    {
      "epoch": 0.11715294117647058,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3191,
      "step": 4979
    },
    {
      "epoch": 0.1171764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1764,
      "step": 4980
    },
    {
      "epoch": 0.1172,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2372,
      "step": 4981
    },
    {
      "epoch": 0.11722352941176471,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3056,
      "step": 4982
    },
    {
      "epoch": 0.11724705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1058,
      "step": 4983
    },
    {
      "epoch": 0.11727058823529411,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1855,
      "step": 4984
    },
    {
      "epoch": 0.11729411764705883,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2688,
      "step": 4985
    },
    {
      "epoch": 0.11731764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1982,
      "step": 4986
    },
    {
      "epoch": 0.11734117647058824,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.188,
      "step": 4987
    },
    {
      "epoch": 0.11736470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5343,
      "step": 4988
    },
    {
      "epoch": 0.11738823529411764,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2409,
      "step": 4989
    },
    {
      "epoch": 0.11741176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3499,
      "step": 4990
    },
    {
      "epoch": 0.11743529411764705,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1927,
      "step": 4991
    },
    {
      "epoch": 0.11745882352941177,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2077,
      "step": 4992
    },
    {
      "epoch": 0.11748235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4638,
      "step": 4993
    },
    {
      "epoch": 0.11750588235294118,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2783,
      "step": 4994
    },
    {
      "epoch": 0.11752941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4301,
      "step": 4995
    },
    {
      "epoch": 0.11755294117647058,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.501,
      "step": 4996
    },
    {
      "epoch": 0.1175764705882353,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9273,
      "step": 4997
    },
    {
      "epoch": 0.1176,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3094,
      "step": 4998
    },
    {
      "epoch": 0.11762352941176471,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4136,
      "step": 4999
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2358,
      "step": 5000
    },
    {
      "epoch": 0.11764705882352941,
      "eval_loss": 2.2437431812286377,
      "eval_runtime": 683.2742,
      "eval_samples_per_second": 12.44,
      "eval_steps_per_second": 3.11,
      "step": 5000
    },
    {
      "epoch": 0.11767058823529412,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2597,
      "step": 5001
    },
    {
      "epoch": 0.11769411764705882,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.8423,
      "step": 5002
    },
    {
      "epoch": 0.11771764705882352,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 5003
    },
    {
      "epoch": 0.11774117647058824,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9196,
      "step": 5004
    },
    {
      "epoch": 0.11776470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1877,
      "step": 5005
    },
    {
      "epoch": 0.11778823529411765,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0179,
      "step": 5006
    },
    {
      "epoch": 0.11781176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2971,
      "step": 5007
    },
    {
      "epoch": 0.11783529411764705,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3667,
      "step": 5008
    },
    {
      "epoch": 0.11785882352941177,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5043,
      "step": 5009
    },
    {
      "epoch": 0.11788235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2029,
      "step": 5010
    },
    {
      "epoch": 0.11790588235294118,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0568,
      "step": 5011
    },
    {
      "epoch": 0.11792941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2814,
      "step": 5012
    },
    {
      "epoch": 0.1179529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1265,
      "step": 5013
    },
    {
      "epoch": 0.1179764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.5833,
      "step": 5014
    },
    {
      "epoch": 0.118,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 5015
    },
    {
      "epoch": 0.11802352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3748,
      "step": 5016
    },
    {
      "epoch": 0.11804705882352941,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3151,
      "step": 5017
    },
    {
      "epoch": 0.11807058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3463,
      "step": 5018
    },
    {
      "epoch": 0.11809411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 5019
    },
    {
      "epoch": 0.11811764705882354,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1303,
      "step": 5020
    },
    {
      "epoch": 0.11814117647058824,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0238,
      "step": 5021
    },
    {
      "epoch": 0.11816470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0182,
      "step": 5022
    },
    {
      "epoch": 0.11818823529411765,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0626,
      "step": 5023
    },
    {
      "epoch": 0.11821176470588235,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.033,
      "step": 5024
    },
    {
      "epoch": 0.11823529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2053,
      "step": 5025
    },
    {
      "epoch": 0.11825882352941176,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0523,
      "step": 5026
    },
    {
      "epoch": 0.11828235294117648,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2847,
      "step": 5027
    },
    {
      "epoch": 0.11830588235294118,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3618,
      "step": 5028
    },
    {
      "epoch": 0.11832941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1278,
      "step": 5029
    },
    {
      "epoch": 0.11835294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6021,
      "step": 5030
    },
    {
      "epoch": 0.11837647058823529,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0449,
      "step": 5031
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2792,
      "step": 5032
    },
    {
      "epoch": 0.1184235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1723,
      "step": 5033
    },
    {
      "epoch": 0.1184470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2522,
      "step": 5034
    },
    {
      "epoch": 0.11847058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3111,
      "step": 5035
    },
    {
      "epoch": 0.11849411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3048,
      "step": 5036
    },
    {
      "epoch": 0.11851764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2625,
      "step": 5037
    },
    {
      "epoch": 0.11854117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3193,
      "step": 5038
    },
    {
      "epoch": 0.11856470588235295,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.411,
      "step": 5039
    },
    {
      "epoch": 0.11858823529411765,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1793,
      "step": 5040
    },
    {
      "epoch": 0.11861176470588235,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1418,
      "step": 5041
    },
    {
      "epoch": 0.11863529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2283,
      "step": 5042
    },
    {
      "epoch": 0.11865882352941176,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 5043
    },
    {
      "epoch": 0.11868235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3935,
      "step": 5044
    },
    {
      "epoch": 0.11870588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3853,
      "step": 5045
    },
    {
      "epoch": 0.11872941176470589,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9595,
      "step": 5046
    },
    {
      "epoch": 0.11875294117647059,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.024,
      "step": 5047
    },
    {
      "epoch": 0.11877647058823529,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.337,
      "step": 5048
    },
    {
      "epoch": 0.1188,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.5503,
      "step": 5049
    },
    {
      "epoch": 0.1188235294117647,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1929,
      "step": 5050
    },
    {
      "epoch": 0.11884705882352942,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0402,
      "step": 5051
    },
    {
      "epoch": 0.11887058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1622,
      "step": 5052
    },
    {
      "epoch": 0.11889411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1884,
      "step": 5053
    },
    {
      "epoch": 0.11891764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4038,
      "step": 5054
    },
    {
      "epoch": 0.11894117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1992,
      "step": 5055
    },
    {
      "epoch": 0.11896470588235294,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8754,
      "step": 5056
    },
    {
      "epoch": 0.11898823529411764,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3388,
      "step": 5057
    },
    {
      "epoch": 0.11901176470588236,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4189,
      "step": 5058
    },
    {
      "epoch": 0.11903529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0094,
      "step": 5059
    },
    {
      "epoch": 0.11905882352941176,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0164,
      "step": 5060
    },
    {
      "epoch": 0.11908235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2842,
      "step": 5061
    },
    {
      "epoch": 0.11910588235294117,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.4449,
      "step": 5062
    },
    {
      "epoch": 0.11912941176470589,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1842,
      "step": 5063
    },
    {
      "epoch": 0.11915294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4438,
      "step": 5064
    },
    {
      "epoch": 0.1191764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3254,
      "step": 5065
    },
    {
      "epoch": 0.1192,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2349,
      "step": 5066
    },
    {
      "epoch": 0.1192235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3326,
      "step": 5067
    },
    {
      "epoch": 0.11924705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2825,
      "step": 5068
    },
    {
      "epoch": 0.11927058823529411,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3975,
      "step": 5069
    },
    {
      "epoch": 0.11929411764705883,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6341,
      "step": 5070
    },
    {
      "epoch": 0.11931764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.3715,
      "step": 5071
    },
    {
      "epoch": 0.11934117647058824,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3576,
      "step": 5072
    },
    {
      "epoch": 0.11936470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2828,
      "step": 5073
    },
    {
      "epoch": 0.11938823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 5074
    },
    {
      "epoch": 0.11941176470588236,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1952,
      "step": 5075
    },
    {
      "epoch": 0.11943529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.455,
      "step": 5076
    },
    {
      "epoch": 0.11945882352941177,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.309,
      "step": 5077
    },
    {
      "epoch": 0.11948235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.5287,
      "step": 5078
    },
    {
      "epoch": 0.11950588235294117,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3976,
      "step": 5079
    },
    {
      "epoch": 0.11952941176470588,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2918,
      "step": 5080
    },
    {
      "epoch": 0.11955294117647058,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2928,
      "step": 5081
    },
    {
      "epoch": 0.1195764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.5671,
      "step": 5082
    },
    {
      "epoch": 0.1196,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.504,
      "step": 5083
    },
    {
      "epoch": 0.11962352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4039,
      "step": 5084
    },
    {
      "epoch": 0.11964705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3902,
      "step": 5085
    },
    {
      "epoch": 0.11967058823529411,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.319,
      "step": 5086
    },
    {
      "epoch": 0.11969411764705883,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1345,
      "step": 5087
    },
    {
      "epoch": 0.11971764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 5088
    },
    {
      "epoch": 0.11974117647058824,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.125,
      "step": 5089
    },
    {
      "epoch": 0.11976470588235294,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0807,
      "step": 5090
    },
    {
      "epoch": 0.11978823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3457,
      "step": 5091
    },
    {
      "epoch": 0.11981176470588235,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1406,
      "step": 5092
    },
    {
      "epoch": 0.11983529411764705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3503,
      "step": 5093
    },
    {
      "epoch": 0.11985882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1622,
      "step": 5094
    },
    {
      "epoch": 0.11988235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1092,
      "step": 5095
    },
    {
      "epoch": 0.11990588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.376,
      "step": 5096
    },
    {
      "epoch": 0.11992941176470588,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2308,
      "step": 5097
    },
    {
      "epoch": 0.11995294117647058,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9896,
      "step": 5098
    },
    {
      "epoch": 0.1199764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1433,
      "step": 5099
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3859,
      "step": 5100
    },
    {
      "epoch": 0.12002352941176471,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3425,
      "step": 5101
    },
    {
      "epoch": 0.12004705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2871,
      "step": 5102
    },
    {
      "epoch": 0.12007058823529412,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 5103
    },
    {
      "epoch": 0.12009411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0963,
      "step": 5104
    },
    {
      "epoch": 0.12011764705882352,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3749,
      "step": 5105
    },
    {
      "epoch": 0.12014117647058824,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.012,
      "step": 5106
    },
    {
      "epoch": 0.12016470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4512,
      "step": 5107
    },
    {
      "epoch": 0.12018823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.5418,
      "step": 5108
    },
    {
      "epoch": 0.12021176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4481,
      "step": 5109
    },
    {
      "epoch": 0.12023529411764707,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1572,
      "step": 5110
    },
    {
      "epoch": 0.12025882352941177,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2004,
      "step": 5111
    },
    {
      "epoch": 0.12028235294117647,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.0206,
      "step": 5112
    },
    {
      "epoch": 0.12030588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1011,
      "step": 5113
    },
    {
      "epoch": 0.12032941176470588,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3691,
      "step": 5114
    },
    {
      "epoch": 0.1203529411764706,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1361,
      "step": 5115
    },
    {
      "epoch": 0.1203764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.5173,
      "step": 5116
    },
    {
      "epoch": 0.1204,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1105,
      "step": 5117
    },
    {
      "epoch": 0.12042352941176471,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.423,
      "step": 5118
    },
    {
      "epoch": 0.12044705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3983,
      "step": 5119
    },
    {
      "epoch": 0.12047058823529412,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3349,
      "step": 5120
    },
    {
      "epoch": 0.12049411764705882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.271,
      "step": 5121
    },
    {
      "epoch": 0.12051764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1631,
      "step": 5122
    },
    {
      "epoch": 0.12054117647058823,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9336,
      "step": 5123
    },
    {
      "epoch": 0.12056470588235293,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2167,
      "step": 5124
    },
    {
      "epoch": 0.12058823529411765,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0479,
      "step": 5125
    },
    {
      "epoch": 0.12061176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3492,
      "step": 5126
    },
    {
      "epoch": 0.12063529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.237,
      "step": 5127
    },
    {
      "epoch": 0.12065882352941176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3247,
      "step": 5128
    },
    {
      "epoch": 0.12068235294117648,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1049,
      "step": 5129
    },
    {
      "epoch": 0.12070588235294118,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.172,
      "step": 5130
    },
    {
      "epoch": 0.12072941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2485,
      "step": 5131
    },
    {
      "epoch": 0.12075294117647059,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0128,
      "step": 5132
    },
    {
      "epoch": 0.12077647058823529,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0727,
      "step": 5133
    },
    {
      "epoch": 0.1208,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3275,
      "step": 5134
    },
    {
      "epoch": 0.1208235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1625,
      "step": 5135
    },
    {
      "epoch": 0.12084705882352942,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2644,
      "step": 5136
    },
    {
      "epoch": 0.12087058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2693,
      "step": 5137
    },
    {
      "epoch": 0.12089411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3035,
      "step": 5138
    },
    {
      "epoch": 0.12091764705882353,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.9806,
      "step": 5139
    },
    {
      "epoch": 0.12094117647058823,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0988,
      "step": 5140
    },
    {
      "epoch": 0.12096470588235295,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2412,
      "step": 5141
    },
    {
      "epoch": 0.12098823529411765,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1291,
      "step": 5142
    },
    {
      "epoch": 0.12101176470588235,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9404,
      "step": 5143
    },
    {
      "epoch": 0.12103529411764706,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0731,
      "step": 5144
    },
    {
      "epoch": 0.12105882352941176,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1886,
      "step": 5145
    },
    {
      "epoch": 0.12108235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.16,
      "step": 5146
    },
    {
      "epoch": 0.12110588235294117,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 5147
    },
    {
      "epoch": 0.12112941176470589,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9572,
      "step": 5148
    },
    {
      "epoch": 0.12115294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.502,
      "step": 5149
    },
    {
      "epoch": 0.12117647058823529,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1676,
      "step": 5150
    },
    {
      "epoch": 0.1212,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4295,
      "step": 5151
    },
    {
      "epoch": 0.1212235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5941,
      "step": 5152
    },
    {
      "epoch": 0.12124705882352942,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3161,
      "step": 5153
    },
    {
      "epoch": 0.12127058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.359,
      "step": 5154
    },
    {
      "epoch": 0.12129411764705883,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9044,
      "step": 5155
    },
    {
      "epoch": 0.12131764705882353,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 0.8632,
      "step": 5156
    },
    {
      "epoch": 0.12134117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2829,
      "step": 5157
    },
    {
      "epoch": 0.12136470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3904,
      "step": 5158
    },
    {
      "epoch": 0.12138823529411764,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.4473,
      "step": 5159
    },
    {
      "epoch": 0.12141176470588236,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 5160
    },
    {
      "epoch": 0.12143529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2612,
      "step": 5161
    },
    {
      "epoch": 0.12145882352941176,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2805,
      "step": 5162
    },
    {
      "epoch": 0.12148235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.283,
      "step": 5163
    },
    {
      "epoch": 0.12150588235294117,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.4405,
      "step": 5164
    },
    {
      "epoch": 0.12152941176470589,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0571,
      "step": 5165
    },
    {
      "epoch": 0.12155294117647059,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2294,
      "step": 5166
    },
    {
      "epoch": 0.1215764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3353,
      "step": 5167
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3152,
      "step": 5168
    },
    {
      "epoch": 0.1216235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2312,
      "step": 5169
    },
    {
      "epoch": 0.12164705882352941,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3393,
      "step": 5170
    },
    {
      "epoch": 0.12167058823529411,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3515,
      "step": 5171
    },
    {
      "epoch": 0.12169411764705883,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1896,
      "step": 5172
    },
    {
      "epoch": 0.12171764705882353,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1475,
      "step": 5173
    },
    {
      "epoch": 0.12174117647058824,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 5174
    },
    {
      "epoch": 0.12176470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9838,
      "step": 5175
    },
    {
      "epoch": 0.12178823529411764,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9764,
      "step": 5176
    },
    {
      "epoch": 0.12181176470588236,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3073,
      "step": 5177
    },
    {
      "epoch": 0.12183529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5472,
      "step": 5178
    },
    {
      "epoch": 0.12185882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.524,
      "step": 5179
    },
    {
      "epoch": 0.12188235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2591,
      "step": 5180
    },
    {
      "epoch": 0.12190588235294118,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1609,
      "step": 5181
    },
    {
      "epoch": 0.12192941176470588,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2175,
      "step": 5182
    },
    {
      "epoch": 0.12195294117647058,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0747,
      "step": 5183
    },
    {
      "epoch": 0.1219764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.5241,
      "step": 5184
    },
    {
      "epoch": 0.122,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2949,
      "step": 5185
    },
    {
      "epoch": 0.12202352941176471,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.6126,
      "step": 5186
    },
    {
      "epoch": 0.12204705882352941,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2827,
      "step": 5187
    },
    {
      "epoch": 0.12207058823529411,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3436,
      "step": 5188
    },
    {
      "epoch": 0.12209411764705883,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2765,
      "step": 5189
    },
    {
      "epoch": 0.12211764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3521,
      "step": 5190
    },
    {
      "epoch": 0.12214117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.45,
      "step": 5191
    },
    {
      "epoch": 0.12216470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2283,
      "step": 5192
    },
    {
      "epoch": 0.12218823529411765,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1028,
      "step": 5193
    },
    {
      "epoch": 0.12221176470588235,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2771,
      "step": 5194
    },
    {
      "epoch": 0.12223529411764705,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 5195
    },
    {
      "epoch": 0.12225882352941177,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4634,
      "step": 5196
    },
    {
      "epoch": 0.12228235294117647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0816,
      "step": 5197
    },
    {
      "epoch": 0.12230588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.463,
      "step": 5198
    },
    {
      "epoch": 0.12232941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.4577,
      "step": 5199
    },
    {
      "epoch": 0.1223529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9513,
      "step": 5200
    },
    {
      "epoch": 0.1223764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1225,
      "step": 5201
    },
    {
      "epoch": 0.1224,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1865,
      "step": 5202
    },
    {
      "epoch": 0.12242352941176471,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4553,
      "step": 5203
    },
    {
      "epoch": 0.12244705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3879,
      "step": 5204
    },
    {
      "epoch": 0.12247058823529412,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2169,
      "step": 5205
    },
    {
      "epoch": 0.12249411764705882,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9994,
      "step": 5206
    },
    {
      "epoch": 0.12251764705882352,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0708,
      "step": 5207
    },
    {
      "epoch": 0.12254117647058824,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0624,
      "step": 5208
    },
    {
      "epoch": 0.12256470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3834,
      "step": 5209
    },
    {
      "epoch": 0.12258823529411765,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0841,
      "step": 5210
    },
    {
      "epoch": 0.12261176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2584,
      "step": 5211
    },
    {
      "epoch": 0.12263529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2479,
      "step": 5212
    },
    {
      "epoch": 0.12265882352941176,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2214,
      "step": 5213
    },
    {
      "epoch": 0.12268235294117646,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4619,
      "step": 5214
    },
    {
      "epoch": 0.12270588235294118,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1145,
      "step": 5215
    },
    {
      "epoch": 0.12272941176470588,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3101,
      "step": 5216
    },
    {
      "epoch": 0.12275294117647059,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3432,
      "step": 5217
    },
    {
      "epoch": 0.12277647058823529,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0912,
      "step": 5218
    },
    {
      "epoch": 0.1228,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3248,
      "step": 5219
    },
    {
      "epoch": 0.1228235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2596,
      "step": 5220
    },
    {
      "epoch": 0.1228470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0491,
      "step": 5221
    },
    {
      "epoch": 0.12287058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 5222
    },
    {
      "epoch": 0.12289411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3598,
      "step": 5223
    },
    {
      "epoch": 0.12291764705882353,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.205,
      "step": 5224
    },
    {
      "epoch": 0.12294117647058823,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4471,
      "step": 5225
    },
    {
      "epoch": 0.12296470588235293,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2942,
      "step": 5226
    },
    {
      "epoch": 0.12298823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.411,
      "step": 5227
    },
    {
      "epoch": 0.12301176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5092,
      "step": 5228
    },
    {
      "epoch": 0.12303529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 5229
    },
    {
      "epoch": 0.12305882352941176,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.8436,
      "step": 5230
    },
    {
      "epoch": 0.12308235294117648,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2015,
      "step": 5231
    },
    {
      "epoch": 0.12310588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.499,
      "step": 5232
    },
    {
      "epoch": 0.12312941176470588,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2667,
      "step": 5233
    },
    {
      "epoch": 0.12315294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3079,
      "step": 5234
    },
    {
      "epoch": 0.12317647058823529,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2218,
      "step": 5235
    },
    {
      "epoch": 0.1232,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3326,
      "step": 5236
    },
    {
      "epoch": 0.1232235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2836,
      "step": 5237
    },
    {
      "epoch": 0.12324705882352942,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3853,
      "step": 5238
    },
    {
      "epoch": 0.12327058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1754,
      "step": 5239
    },
    {
      "epoch": 0.12329411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1121,
      "step": 5240
    },
    {
      "epoch": 0.12331764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9751,
      "step": 5241
    },
    {
      "epoch": 0.12334117647058823,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2691,
      "step": 5242
    },
    {
      "epoch": 0.12336470588235295,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1899,
      "step": 5243
    },
    {
      "epoch": 0.12338823529411765,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2897,
      "step": 5244
    },
    {
      "epoch": 0.12341176470588236,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3562,
      "step": 5245
    },
    {
      "epoch": 0.12343529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3918,
      "step": 5246
    },
    {
      "epoch": 0.12345882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2737,
      "step": 5247
    },
    {
      "epoch": 0.12348235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4087,
      "step": 5248
    },
    {
      "epoch": 0.12350588235294117,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9594,
      "step": 5249
    },
    {
      "epoch": 0.12352941176470589,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1839,
      "step": 5250
    },
    {
      "epoch": 0.12355294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2881,
      "step": 5251
    },
    {
      "epoch": 0.12357647058823529,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3181,
      "step": 5252
    },
    {
      "epoch": 0.1236,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.0288,
      "step": 5253
    },
    {
      "epoch": 0.1236235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2764,
      "step": 5254
    },
    {
      "epoch": 0.12364705882352942,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1266,
      "step": 5255
    },
    {
      "epoch": 0.12367058823529412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2393,
      "step": 5256
    },
    {
      "epoch": 0.12369411764705883,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3332,
      "step": 5257
    },
    {
      "epoch": 0.12371764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4464,
      "step": 5258
    },
    {
      "epoch": 0.12374117647058823,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0068,
      "step": 5259
    },
    {
      "epoch": 0.12376470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1999,
      "step": 5260
    },
    {
      "epoch": 0.12378823529411764,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3199,
      "step": 5261
    },
    {
      "epoch": 0.12381176470588236,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.6028,
      "step": 5262
    },
    {
      "epoch": 0.12383529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2001,
      "step": 5263
    },
    {
      "epoch": 0.12385882352941177,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2744,
      "step": 5264
    },
    {
      "epoch": 0.12388235294117647,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 5265
    },
    {
      "epoch": 0.12390588235294117,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.158,
      "step": 5266
    },
    {
      "epoch": 0.12392941176470589,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1931,
      "step": 5267
    },
    {
      "epoch": 0.12395294117647059,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0653,
      "step": 5268
    },
    {
      "epoch": 0.1239764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2986,
      "step": 5269
    },
    {
      "epoch": 0.124,
      "grad_norm": 0.3046875,
      "learning_rate": 0.02,
      "loss": 0.9983,
      "step": 5270
    },
    {
      "epoch": 0.1240235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.4704,
      "step": 5271
    },
    {
      "epoch": 0.12404705882352941,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 5272
    },
    {
      "epoch": 0.12407058823529411,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3353,
      "step": 5273
    },
    {
      "epoch": 0.12409411764705883,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2492,
      "step": 5274
    },
    {
      "epoch": 0.12411764705882353,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.2538,
      "step": 5275
    },
    {
      "epoch": 0.12414117647058824,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2459,
      "step": 5276
    },
    {
      "epoch": 0.12416470588235294,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1833,
      "step": 5277
    },
    {
      "epoch": 0.12418823529411764,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.4902,
      "step": 5278
    },
    {
      "epoch": 0.12421176470588235,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3297,
      "step": 5279
    },
    {
      "epoch": 0.12423529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3526,
      "step": 5280
    },
    {
      "epoch": 0.12425882352941177,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1667,
      "step": 5281
    },
    {
      "epoch": 0.12428235294117647,
      "grad_norm": 0.328125,
      "learning_rate": 0.02,
      "loss": 0.9828,
      "step": 5282
    },
    {
      "epoch": 0.12430588235294118,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3181,
      "step": 5283
    },
    {
      "epoch": 0.12432941176470588,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.3456,
      "step": 5284
    },
    {
      "epoch": 0.12435294117647058,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3489,
      "step": 5285
    },
    {
      "epoch": 0.1243764705882353,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9643,
      "step": 5286
    },
    {
      "epoch": 0.1244,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.5592,
      "step": 5287
    },
    {
      "epoch": 0.12442352941176471,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2258,
      "step": 5288
    },
    {
      "epoch": 0.12444705882352941,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.3121,
      "step": 5289
    },
    {
      "epoch": 0.12447058823529412,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2818,
      "step": 5290
    },
    {
      "epoch": 0.12449411764705882,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1195,
      "step": 5291
    },
    {
      "epoch": 0.12451764705882352,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.0531,
      "step": 5292
    },
    {
      "epoch": 0.12454117647058824,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1663,
      "step": 5293
    },
    {
      "epoch": 0.12456470588235294,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.1577,
      "step": 5294
    },
    {
      "epoch": 0.12458823529411765,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.07,
      "step": 5295
    },
    {
      "epoch": 0.12461176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3055,
      "step": 5296
    },
    {
      "epoch": 0.12463529411764705,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.5451,
      "step": 5297
    },
    {
      "epoch": 0.12465882352941177,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 0.9829,
      "step": 5298
    },
    {
      "epoch": 0.12468235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3154,
      "step": 5299
    },
    {
      "epoch": 0.12470588235294118,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 5300
    },
    {
      "epoch": 0.12472941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.439,
      "step": 5301
    },
    {
      "epoch": 0.1247529411764706,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1329,
      "step": 5302
    },
    {
      "epoch": 0.1247764705882353,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8392,
      "step": 5303
    },
    {
      "epoch": 0.1248,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 1.1631,
      "step": 5304
    },
    {
      "epoch": 0.12482352941176471,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2749,
      "step": 5305
    },
    {
      "epoch": 0.12484705882352941,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3197,
      "step": 5306
    },
    {
      "epoch": 0.12487058823529412,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.961,
      "step": 5307
    },
    {
      "epoch": 0.12489411764705882,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2203,
      "step": 5308
    },
    {
      "epoch": 0.12491764705882354,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0807,
      "step": 5309
    },
    {
      "epoch": 0.12494117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4383,
      "step": 5310
    },
    {
      "epoch": 0.12496470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0512,
      "step": 5311
    },
    {
      "epoch": 0.12498823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.6075,
      "step": 5312
    },
    {
      "epoch": 0.12501176470588235,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 5313
    },
    {
      "epoch": 0.12503529411764705,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.502,
      "step": 5314
    },
    {
      "epoch": 0.12505882352941178,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2968,
      "step": 5315
    },
    {
      "epoch": 0.12508235294117648,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1686,
      "step": 5316
    },
    {
      "epoch": 0.12510588235294118,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1766,
      "step": 5317
    },
    {
      "epoch": 0.12512941176470588,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.1343,
      "step": 5318
    },
    {
      "epoch": 0.12515294117647058,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3529,
      "step": 5319
    },
    {
      "epoch": 0.1251764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2766,
      "step": 5320
    },
    {
      "epoch": 0.1252,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.318,
      "step": 5321
    },
    {
      "epoch": 0.1252235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3046,
      "step": 5322
    },
    {
      "epoch": 0.1252470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 5323
    },
    {
      "epoch": 0.1252705882352941,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.4344,
      "step": 5324
    },
    {
      "epoch": 0.12529411764705883,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1931,
      "step": 5325
    },
    {
      "epoch": 0.12531764705882353,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0169,
      "step": 5326
    },
    {
      "epoch": 0.12534117647058823,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2584,
      "step": 5327
    },
    {
      "epoch": 0.12536470588235293,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3039,
      "step": 5328
    },
    {
      "epoch": 0.12538823529411763,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3582,
      "step": 5329
    },
    {
      "epoch": 0.12541176470588236,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.3237,
      "step": 5330
    },
    {
      "epoch": 0.12543529411764706,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1095,
      "step": 5331
    },
    {
      "epoch": 0.12545882352941176,
      "grad_norm": 0.310546875,
      "learning_rate": 0.02,
      "loss": 1.0651,
      "step": 5332
    },
    {
      "epoch": 0.12548235294117646,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 5333
    },
    {
      "epoch": 0.1255058823529412,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.8362,
      "step": 5334
    },
    {
      "epoch": 0.1255294117647059,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2577,
      "step": 5335
    },
    {
      "epoch": 0.1255529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5078,
      "step": 5336
    },
    {
      "epoch": 0.1255764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1156,
      "step": 5337
    },
    {
      "epoch": 0.1256,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4164,
      "step": 5338
    },
    {
      "epoch": 0.12562352941176472,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.3022,
      "step": 5339
    },
    {
      "epoch": 0.12564705882352942,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4748,
      "step": 5340
    },
    {
      "epoch": 0.12567058823529412,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1374,
      "step": 5341
    },
    {
      "epoch": 0.12569411764705882,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1461,
      "step": 5342
    },
    {
      "epoch": 0.12571764705882352,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.607,
      "step": 5343
    },
    {
      "epoch": 0.12574117647058825,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 5344
    },
    {
      "epoch": 0.12576470588235295,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2252,
      "step": 5345
    },
    {
      "epoch": 0.12578823529411765,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 1.01,
      "step": 5346
    },
    {
      "epoch": 0.12581176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2146,
      "step": 5347
    },
    {
      "epoch": 0.12583529411764705,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0818,
      "step": 5348
    },
    {
      "epoch": 0.12585882352941177,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1135,
      "step": 5349
    },
    {
      "epoch": 0.12588235294117647,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1362,
      "step": 5350
    },
    {
      "epoch": 0.12590588235294117,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2045,
      "step": 5351
    },
    {
      "epoch": 0.12592941176470587,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3792,
      "step": 5352
    },
    {
      "epoch": 0.1259529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2634,
      "step": 5353
    },
    {
      "epoch": 0.1259764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1538,
      "step": 5354
    },
    {
      "epoch": 0.126,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.149,
      "step": 5355
    },
    {
      "epoch": 0.1260235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3188,
      "step": 5356
    },
    {
      "epoch": 0.1260470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3908,
      "step": 5357
    },
    {
      "epoch": 0.12607058823529413,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0795,
      "step": 5358
    },
    {
      "epoch": 0.12609411764705883,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4172,
      "step": 5359
    },
    {
      "epoch": 0.12611764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1204,
      "step": 5360
    },
    {
      "epoch": 0.12614117647058823,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2062,
      "step": 5361
    },
    {
      "epoch": 0.12616470588235293,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2205,
      "step": 5362
    },
    {
      "epoch": 0.12618823529411766,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3004,
      "step": 5363
    },
    {
      "epoch": 0.12621176470588236,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4129,
      "step": 5364
    },
    {
      "epoch": 0.12623529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3171,
      "step": 5365
    },
    {
      "epoch": 0.12625882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4894,
      "step": 5366
    },
    {
      "epoch": 0.12628235294117648,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.5065,
      "step": 5367
    },
    {
      "epoch": 0.12630588235294118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.5586,
      "step": 5368
    },
    {
      "epoch": 0.12632941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2644,
      "step": 5369
    },
    {
      "epoch": 0.12635294117647058,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 5370
    },
    {
      "epoch": 0.12637647058823528,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3036,
      "step": 5371
    },
    {
      "epoch": 0.1264,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2162,
      "step": 5372
    },
    {
      "epoch": 0.1264235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2742,
      "step": 5373
    },
    {
      "epoch": 0.1264470588235294,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2996,
      "step": 5374
    },
    {
      "epoch": 0.1264705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2839,
      "step": 5375
    },
    {
      "epoch": 0.1264941176470588,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0503,
      "step": 5376
    },
    {
      "epoch": 0.12651764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4807,
      "step": 5377
    },
    {
      "epoch": 0.12654117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1805,
      "step": 5378
    },
    {
      "epoch": 0.12656470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3304,
      "step": 5379
    },
    {
      "epoch": 0.12658823529411764,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3996,
      "step": 5380
    },
    {
      "epoch": 0.12661176470588234,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3488,
      "step": 5381
    },
    {
      "epoch": 0.12663529411764707,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2633,
      "step": 5382
    },
    {
      "epoch": 0.12665882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2075,
      "step": 5383
    },
    {
      "epoch": 0.12668235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4204,
      "step": 5384
    },
    {
      "epoch": 0.12670588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2721,
      "step": 5385
    },
    {
      "epoch": 0.1267294117647059,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2707,
      "step": 5386
    },
    {
      "epoch": 0.1267529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0953,
      "step": 5387
    },
    {
      "epoch": 0.1267764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.6061,
      "step": 5388
    },
    {
      "epoch": 0.1268,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 5389
    },
    {
      "epoch": 0.1268235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.302,
      "step": 5390
    },
    {
      "epoch": 0.12684705882352942,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0943,
      "step": 5391
    },
    {
      "epoch": 0.12687058823529412,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2166,
      "step": 5392
    },
    {
      "epoch": 0.12689411764705882,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.1236,
      "step": 5393
    },
    {
      "epoch": 0.12691764705882352,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.354,
      "step": 5394
    },
    {
      "epoch": 0.12694117647058822,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2743,
      "step": 5395
    },
    {
      "epoch": 0.12696470588235295,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3578,
      "step": 5396
    },
    {
      "epoch": 0.12698823529411765,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3118,
      "step": 5397
    },
    {
      "epoch": 0.12701176470588235,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0721,
      "step": 5398
    },
    {
      "epoch": 0.12703529411764705,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2588,
      "step": 5399
    },
    {
      "epoch": 0.12705882352941175,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0442,
      "step": 5400
    },
    {
      "epoch": 0.12708235294117648,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2988,
      "step": 5401
    },
    {
      "epoch": 0.12710588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4935,
      "step": 5402
    },
    {
      "epoch": 0.12712941176470588,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 5403
    },
    {
      "epoch": 0.12715294117647058,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4698,
      "step": 5404
    },
    {
      "epoch": 0.1271764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 5405
    },
    {
      "epoch": 0.1272,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0249,
      "step": 5406
    },
    {
      "epoch": 0.1272235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1395,
      "step": 5407
    },
    {
      "epoch": 0.1272470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2614,
      "step": 5408
    },
    {
      "epoch": 0.1272705882352941,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1905,
      "step": 5409
    },
    {
      "epoch": 0.12729411764705884,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2717,
      "step": 5410
    },
    {
      "epoch": 0.12731764705882354,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1359,
      "step": 5411
    },
    {
      "epoch": 0.12734117647058824,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2035,
      "step": 5412
    },
    {
      "epoch": 0.12736470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2253,
      "step": 5413
    },
    {
      "epoch": 0.12738823529411764,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3765,
      "step": 5414
    },
    {
      "epoch": 0.12741176470588236,
      "grad_norm": 0.33984375,
      "learning_rate": 0.02,
      "loss": 0.9812,
      "step": 5415
    },
    {
      "epoch": 0.12743529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3994,
      "step": 5416
    },
    {
      "epoch": 0.12745882352941176,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2525,
      "step": 5417
    },
    {
      "epoch": 0.12748235294117646,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 5418
    },
    {
      "epoch": 0.12750588235294116,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9666,
      "step": 5419
    },
    {
      "epoch": 0.1275294117647059,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.267,
      "step": 5420
    },
    {
      "epoch": 0.1275529411764706,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.2517,
      "step": 5421
    },
    {
      "epoch": 0.1275764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2511,
      "step": 5422
    },
    {
      "epoch": 0.1276,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1995,
      "step": 5423
    },
    {
      "epoch": 0.12762352941176472,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.452,
      "step": 5424
    },
    {
      "epoch": 0.12764705882352942,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2119,
      "step": 5425
    },
    {
      "epoch": 0.12767058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.085,
      "step": 5426
    },
    {
      "epoch": 0.12769411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8983,
      "step": 5427
    },
    {
      "epoch": 0.12771764705882352,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2,
      "step": 5428
    },
    {
      "epoch": 0.12774117647058825,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0822,
      "step": 5429
    },
    {
      "epoch": 0.12776470588235295,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2784,
      "step": 5430
    },
    {
      "epoch": 0.12778823529411765,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0452,
      "step": 5431
    },
    {
      "epoch": 0.12781176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2624,
      "step": 5432
    },
    {
      "epoch": 0.12783529411764705,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2862,
      "step": 5433
    },
    {
      "epoch": 0.12785882352941177,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9386,
      "step": 5434
    },
    {
      "epoch": 0.12788235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3692,
      "step": 5435
    },
    {
      "epoch": 0.12790588235294117,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0929,
      "step": 5436
    },
    {
      "epoch": 0.12792941176470587,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.9816,
      "step": 5437
    },
    {
      "epoch": 0.12795294117647057,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4385,
      "step": 5438
    },
    {
      "epoch": 0.1279764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3361,
      "step": 5439
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1848,
      "step": 5440
    },
    {
      "epoch": 0.1280235294117647,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2168,
      "step": 5441
    },
    {
      "epoch": 0.1280470588235294,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8653,
      "step": 5442
    },
    {
      "epoch": 0.12807058823529413,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.6001,
      "step": 5443
    },
    {
      "epoch": 0.12809411764705883,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.8564,
      "step": 5444
    },
    {
      "epoch": 0.12811764705882353,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.064,
      "step": 5445
    },
    {
      "epoch": 0.12814117647058823,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.016,
      "step": 5446
    },
    {
      "epoch": 0.12816470588235293,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.5564,
      "step": 5447
    },
    {
      "epoch": 0.12818823529411766,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3762,
      "step": 5448
    },
    {
      "epoch": 0.12821176470588236,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3134,
      "step": 5449
    },
    {
      "epoch": 0.12823529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.4463,
      "step": 5450
    },
    {
      "epoch": 0.12825882352941176,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2274,
      "step": 5451
    },
    {
      "epoch": 0.12828235294117646,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2763,
      "step": 5452
    },
    {
      "epoch": 0.1283058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4464,
      "step": 5453
    },
    {
      "epoch": 0.1283294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2647,
      "step": 5454
    },
    {
      "epoch": 0.1283529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2888,
      "step": 5455
    },
    {
      "epoch": 0.1283764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 5456
    },
    {
      "epoch": 0.1284,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.8313,
      "step": 5457
    },
    {
      "epoch": 0.12842352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3737,
      "step": 5458
    },
    {
      "epoch": 0.12844705882352941,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9644,
      "step": 5459
    },
    {
      "epoch": 0.12847058823529411,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2154,
      "step": 5460
    },
    {
      "epoch": 0.12849411764705881,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2433,
      "step": 5461
    },
    {
      "epoch": 0.12851764705882354,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3678,
      "step": 5462
    },
    {
      "epoch": 0.12854117647058824,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1425,
      "step": 5463
    },
    {
      "epoch": 0.12856470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3268,
      "step": 5464
    },
    {
      "epoch": 0.12858823529411764,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.221,
      "step": 5465
    },
    {
      "epoch": 0.12861176470588234,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2719,
      "step": 5466
    },
    {
      "epoch": 0.12863529411764707,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2447,
      "step": 5467
    },
    {
      "epoch": 0.12865882352941177,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0692,
      "step": 5468
    },
    {
      "epoch": 0.12868235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.2137,
      "step": 5469
    },
    {
      "epoch": 0.12870588235294117,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1239,
      "step": 5470
    },
    {
      "epoch": 0.12872941176470587,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.307,
      "step": 5471
    },
    {
      "epoch": 0.1287529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3965,
      "step": 5472
    },
    {
      "epoch": 0.1287764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.3162,
      "step": 5473
    },
    {
      "epoch": 0.1288,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0996,
      "step": 5474
    },
    {
      "epoch": 0.1288235294117647,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1287,
      "step": 5475
    },
    {
      "epoch": 0.12884705882352943,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2168,
      "step": 5476
    },
    {
      "epoch": 0.12887058823529413,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 1.0129,
      "step": 5477
    },
    {
      "epoch": 0.12889411764705883,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3328,
      "step": 5478
    },
    {
      "epoch": 0.12891764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2225,
      "step": 5479
    },
    {
      "epoch": 0.12894117647058823,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4812,
      "step": 5480
    },
    {
      "epoch": 0.12896470588235295,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2728,
      "step": 5481
    },
    {
      "epoch": 0.12898823529411765,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2932,
      "step": 5482
    },
    {
      "epoch": 0.12901176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2478,
      "step": 5483
    },
    {
      "epoch": 0.12903529411764705,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3765,
      "step": 5484
    },
    {
      "epoch": 0.12905882352941175,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1449,
      "step": 5485
    },
    {
      "epoch": 0.12908235294117648,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 5486
    },
    {
      "epoch": 0.12910588235294118,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 5487
    },
    {
      "epoch": 0.12912941176470588,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1908,
      "step": 5488
    },
    {
      "epoch": 0.12915294117647058,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3021,
      "step": 5489
    },
    {
      "epoch": 0.12917647058823528,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4339,
      "step": 5490
    },
    {
      "epoch": 0.1292,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1016,
      "step": 5491
    },
    {
      "epoch": 0.1292235294117647,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1637,
      "step": 5492
    },
    {
      "epoch": 0.1292470588235294,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0438,
      "step": 5493
    },
    {
      "epoch": 0.1292705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2533,
      "step": 5494
    },
    {
      "epoch": 0.12929411764705884,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1077,
      "step": 5495
    },
    {
      "epoch": 0.12931764705882354,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0613,
      "step": 5496
    },
    {
      "epoch": 0.12934117647058824,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2704,
      "step": 5497
    },
    {
      "epoch": 0.12936470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1788,
      "step": 5498
    },
    {
      "epoch": 0.12938823529411764,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9519,
      "step": 5499
    },
    {
      "epoch": 0.12941176470588237,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.7766,
      "step": 5500
    },
    {
      "epoch": 0.12943529411764707,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.456,
      "step": 5501
    },
    {
      "epoch": 0.12945882352941177,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2663,
      "step": 5502
    },
    {
      "epoch": 0.12948235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3836,
      "step": 5503
    },
    {
      "epoch": 0.12950588235294117,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0913,
      "step": 5504
    },
    {
      "epoch": 0.1295294117647059,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1108,
      "step": 5505
    },
    {
      "epoch": 0.1295529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 5506
    },
    {
      "epoch": 0.1295764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1765,
      "step": 5507
    },
    {
      "epoch": 0.1296,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2763,
      "step": 5508
    },
    {
      "epoch": 0.1296235294117647,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.156,
      "step": 5509
    },
    {
      "epoch": 0.12964705882352942,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.311,
      "step": 5510
    },
    {
      "epoch": 0.12967058823529412,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1209,
      "step": 5511
    },
    {
      "epoch": 0.12969411764705882,
      "grad_norm": 0.30859375,
      "learning_rate": 0.02,
      "loss": 0.6357,
      "step": 5512
    },
    {
      "epoch": 0.12971764705882352,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4466,
      "step": 5513
    },
    {
      "epoch": 0.12974117647058825,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3218,
      "step": 5514
    },
    {
      "epoch": 0.12976470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3552,
      "step": 5515
    },
    {
      "epoch": 0.12978823529411765,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3083,
      "step": 5516
    },
    {
      "epoch": 0.12981176470588235,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1205,
      "step": 5517
    },
    {
      "epoch": 0.12983529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3599,
      "step": 5518
    },
    {
      "epoch": 0.12985882352941178,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.194,
      "step": 5519
    },
    {
      "epoch": 0.12988235294117648,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2703,
      "step": 5520
    },
    {
      "epoch": 0.12990588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2413,
      "step": 5521
    },
    {
      "epoch": 0.12992941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4571,
      "step": 5522
    },
    {
      "epoch": 0.12995294117647058,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1979,
      "step": 5523
    },
    {
      "epoch": 0.1299764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0645,
      "step": 5524
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.105,
      "step": 5525
    },
    {
      "epoch": 0.1300235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5698,
      "step": 5526
    },
    {
      "epoch": 0.1300470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2058,
      "step": 5527
    },
    {
      "epoch": 0.1300705882352941,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1417,
      "step": 5528
    },
    {
      "epoch": 0.13009411764705883,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3075,
      "step": 5529
    },
    {
      "epoch": 0.13011764705882353,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3593,
      "step": 5530
    },
    {
      "epoch": 0.13014117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1307,
      "step": 5531
    },
    {
      "epoch": 0.13016470588235293,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0861,
      "step": 5532
    },
    {
      "epoch": 0.13018823529411766,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9956,
      "step": 5533
    },
    {
      "epoch": 0.13021176470588236,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2407,
      "step": 5534
    },
    {
      "epoch": 0.13023529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2945,
      "step": 5535
    },
    {
      "epoch": 0.13025882352941176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1553,
      "step": 5536
    },
    {
      "epoch": 0.13028235294117646,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.197,
      "step": 5537
    },
    {
      "epoch": 0.1303058823529412,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0782,
      "step": 5538
    },
    {
      "epoch": 0.1303294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4038,
      "step": 5539
    },
    {
      "epoch": 0.1303529411764706,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 1.0198,
      "step": 5540
    },
    {
      "epoch": 0.1303764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4384,
      "step": 5541
    },
    {
      "epoch": 0.1304,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2902,
      "step": 5542
    },
    {
      "epoch": 0.13042352941176472,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4787,
      "step": 5543
    },
    {
      "epoch": 0.13044705882352942,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4065,
      "step": 5544
    },
    {
      "epoch": 0.13047058823529412,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 5545
    },
    {
      "epoch": 0.13049411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3438,
      "step": 5546
    },
    {
      "epoch": 0.13051764705882352,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1754,
      "step": 5547
    },
    {
      "epoch": 0.13054117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2252,
      "step": 5548
    },
    {
      "epoch": 0.13056470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2835,
      "step": 5549
    },
    {
      "epoch": 0.13058823529411764,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3248,
      "step": 5550
    },
    {
      "epoch": 0.13061176470588234,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2128,
      "step": 5551
    },
    {
      "epoch": 0.13063529411764707,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0824,
      "step": 5552
    },
    {
      "epoch": 0.13065882352941177,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.0457,
      "step": 5553
    },
    {
      "epoch": 0.13068235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2992,
      "step": 5554
    },
    {
      "epoch": 0.13070588235294117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 5555
    },
    {
      "epoch": 0.13072941176470587,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2488,
      "step": 5556
    },
    {
      "epoch": 0.1307529411764706,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9084,
      "step": 5557
    },
    {
      "epoch": 0.1307764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2372,
      "step": 5558
    },
    {
      "epoch": 0.1308,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 5559
    },
    {
      "epoch": 0.1308235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1584,
      "step": 5560
    },
    {
      "epoch": 0.1308470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.285,
      "step": 5561
    },
    {
      "epoch": 0.13087058823529413,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.4107,
      "step": 5562
    },
    {
      "epoch": 0.13089411764705883,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4522,
      "step": 5563
    },
    {
      "epoch": 0.13091764705882353,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.8484,
      "step": 5564
    },
    {
      "epoch": 0.13094117647058823,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2338,
      "step": 5565
    },
    {
      "epoch": 0.13096470588235293,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.096,
      "step": 5566
    },
    {
      "epoch": 0.13098823529411766,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 0.981,
      "step": 5567
    },
    {
      "epoch": 0.13101176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3083,
      "step": 5568
    },
    {
      "epoch": 0.13103529411764706,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2807,
      "step": 5569
    },
    {
      "epoch": 0.13105882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3131,
      "step": 5570
    },
    {
      "epoch": 0.13108235294117648,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3096,
      "step": 5571
    },
    {
      "epoch": 0.13110588235294118,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0212,
      "step": 5572
    },
    {
      "epoch": 0.13112941176470588,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3786,
      "step": 5573
    },
    {
      "epoch": 0.13115294117647058,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0499,
      "step": 5574
    },
    {
      "epoch": 0.13117647058823528,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4578,
      "step": 5575
    },
    {
      "epoch": 0.1312,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2703,
      "step": 5576
    },
    {
      "epoch": 0.1312235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0792,
      "step": 5577
    },
    {
      "epoch": 0.1312470588235294,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 5578
    },
    {
      "epoch": 0.1312705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3593,
      "step": 5579
    },
    {
      "epoch": 0.1312941176470588,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2498,
      "step": 5580
    },
    {
      "epoch": 0.13131764705882354,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0325,
      "step": 5581
    },
    {
      "epoch": 0.13134117647058824,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1798,
      "step": 5582
    },
    {
      "epoch": 0.13136470588235294,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.3262,
      "step": 5583
    },
    {
      "epoch": 0.13138823529411764,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2424,
      "step": 5584
    },
    {
      "epoch": 0.13141176470588237,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0609,
      "step": 5585
    },
    {
      "epoch": 0.13143529411764707,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1952,
      "step": 5586
    },
    {
      "epoch": 0.13145882352941177,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3507,
      "step": 5587
    },
    {
      "epoch": 0.13148235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 5588
    },
    {
      "epoch": 0.13150588235294117,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.4356,
      "step": 5589
    },
    {
      "epoch": 0.1315294117647059,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.993,
      "step": 5590
    },
    {
      "epoch": 0.1315529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2432,
      "step": 5591
    },
    {
      "epoch": 0.1315764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.4725,
      "step": 5592
    },
    {
      "epoch": 0.1316,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.243,
      "step": 5593
    },
    {
      "epoch": 0.1316235294117647,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.173,
      "step": 5594
    },
    {
      "epoch": 0.13164705882352942,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.139,
      "step": 5595
    },
    {
      "epoch": 0.13167058823529412,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 5596
    },
    {
      "epoch": 0.13169411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3018,
      "step": 5597
    },
    {
      "epoch": 0.13171764705882352,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1727,
      "step": 5598
    },
    {
      "epoch": 0.13174117647058822,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4797,
      "step": 5599
    },
    {
      "epoch": 0.13176470588235295,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2318,
      "step": 5600
    },
    {
      "epoch": 0.13178823529411765,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9513,
      "step": 5601
    },
    {
      "epoch": 0.13181176470588235,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2822,
      "step": 5602
    },
    {
      "epoch": 0.13183529411764705,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9552,
      "step": 5603
    },
    {
      "epoch": 0.13185882352941178,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3519,
      "step": 5604
    },
    {
      "epoch": 0.13188235294117648,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3852,
      "step": 5605
    },
    {
      "epoch": 0.13190588235294118,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2721,
      "step": 5606
    },
    {
      "epoch": 0.13192941176470588,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2059,
      "step": 5607
    },
    {
      "epoch": 0.13195294117647058,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2395,
      "step": 5608
    },
    {
      "epoch": 0.1319764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3762,
      "step": 5609
    },
    {
      "epoch": 0.132,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.5343,
      "step": 5610
    },
    {
      "epoch": 0.1320235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 5611
    },
    {
      "epoch": 0.1320470588235294,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0211,
      "step": 5612
    },
    {
      "epoch": 0.1320705882352941,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2934,
      "step": 5613
    },
    {
      "epoch": 0.13209411764705883,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3416,
      "step": 5614
    },
    {
      "epoch": 0.13211764705882353,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1831,
      "step": 5615
    },
    {
      "epoch": 0.13214117647058823,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1526,
      "step": 5616
    },
    {
      "epoch": 0.13216470588235293,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1571,
      "step": 5617
    },
    {
      "epoch": 0.13218823529411763,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9325,
      "step": 5618
    },
    {
      "epoch": 0.13221176470588236,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4921,
      "step": 5619
    },
    {
      "epoch": 0.13223529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9569,
      "step": 5620
    },
    {
      "epoch": 0.13225882352941176,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.9134,
      "step": 5621
    },
    {
      "epoch": 0.13228235294117646,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3364,
      "step": 5622
    },
    {
      "epoch": 0.1323058823529412,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 5623
    },
    {
      "epoch": 0.1323294117647059,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2364,
      "step": 5624
    },
    {
      "epoch": 0.1323529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2932,
      "step": 5625
    },
    {
      "epoch": 0.1323764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1351,
      "step": 5626
    },
    {
      "epoch": 0.1324,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2472,
      "step": 5627
    },
    {
      "epoch": 0.13242352941176472,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2022,
      "step": 5628
    },
    {
      "epoch": 0.13244705882352942,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.1136,
      "step": 5629
    },
    {
      "epoch": 0.13247058823529412,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0473,
      "step": 5630
    },
    {
      "epoch": 0.13249411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.4675,
      "step": 5631
    },
    {
      "epoch": 0.13251764705882352,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.401,
      "step": 5632
    },
    {
      "epoch": 0.13254117647058825,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3069,
      "step": 5633
    },
    {
      "epoch": 0.13256470588235295,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3807,
      "step": 5634
    },
    {
      "epoch": 0.13258823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4643,
      "step": 5635
    },
    {
      "epoch": 0.13261176470588235,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1391,
      "step": 5636
    },
    {
      "epoch": 0.13263529411764705,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0613,
      "step": 5637
    },
    {
      "epoch": 0.13265882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3741,
      "step": 5638
    },
    {
      "epoch": 0.13268235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2687,
      "step": 5639
    },
    {
      "epoch": 0.13270588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3832,
      "step": 5640
    },
    {
      "epoch": 0.13272941176470587,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3622,
      "step": 5641
    },
    {
      "epoch": 0.1327529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0984,
      "step": 5642
    },
    {
      "epoch": 0.1327764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1077,
      "step": 5643
    },
    {
      "epoch": 0.1328,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2254,
      "step": 5644
    },
    {
      "epoch": 0.1328235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 5645
    },
    {
      "epoch": 0.1328470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2861,
      "step": 5646
    },
    {
      "epoch": 0.13287058823529413,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.858,
      "step": 5647
    },
    {
      "epoch": 0.13289411764705883,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 5648
    },
    {
      "epoch": 0.13291764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1919,
      "step": 5649
    },
    {
      "epoch": 0.13294117647058823,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2885,
      "step": 5650
    },
    {
      "epoch": 0.13296470588235293,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3508,
      "step": 5651
    },
    {
      "epoch": 0.13298823529411766,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3111,
      "step": 5652
    },
    {
      "epoch": 0.13301176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.101,
      "step": 5653
    },
    {
      "epoch": 0.13303529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3274,
      "step": 5654
    },
    {
      "epoch": 0.13305882352941176,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0133,
      "step": 5655
    },
    {
      "epoch": 0.13308235294117646,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3864,
      "step": 5656
    },
    {
      "epoch": 0.13310588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4322,
      "step": 5657
    },
    {
      "epoch": 0.13312941176470589,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 5658
    },
    {
      "epoch": 0.13315294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.387,
      "step": 5659
    },
    {
      "epoch": 0.13317647058823529,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1107,
      "step": 5660
    },
    {
      "epoch": 0.1332,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.53,
      "step": 5661
    },
    {
      "epoch": 0.1332235294117647,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.168,
      "step": 5662
    },
    {
      "epoch": 0.1332470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2857,
      "step": 5663
    },
    {
      "epoch": 0.1332705882352941,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0653,
      "step": 5664
    },
    {
      "epoch": 0.1332941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0472,
      "step": 5665
    },
    {
      "epoch": 0.13331764705882354,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9358,
      "step": 5666
    },
    {
      "epoch": 0.13334117647058824,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2589,
      "step": 5667
    },
    {
      "epoch": 0.13336470588235294,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2203,
      "step": 5668
    },
    {
      "epoch": 0.13338823529411764,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0501,
      "step": 5669
    },
    {
      "epoch": 0.13341176470588234,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1873,
      "step": 5670
    },
    {
      "epoch": 0.13343529411764707,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 5671
    },
    {
      "epoch": 0.13345882352941177,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1445,
      "step": 5672
    },
    {
      "epoch": 0.13348235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2194,
      "step": 5673
    },
    {
      "epoch": 0.13350588235294117,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3068,
      "step": 5674
    },
    {
      "epoch": 0.13352941176470587,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0122,
      "step": 5675
    },
    {
      "epoch": 0.1335529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.351,
      "step": 5676
    },
    {
      "epoch": 0.1335764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 5677
    },
    {
      "epoch": 0.1336,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1246,
      "step": 5678
    },
    {
      "epoch": 0.1336235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0751,
      "step": 5679
    },
    {
      "epoch": 0.13364705882352942,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.907,
      "step": 5680
    },
    {
      "epoch": 0.13367058823529412,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2501,
      "step": 5681
    },
    {
      "epoch": 0.13369411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4822,
      "step": 5682
    },
    {
      "epoch": 0.13371764705882352,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.6988,
      "step": 5683
    },
    {
      "epoch": 0.13374117647058822,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3112,
      "step": 5684
    },
    {
      "epoch": 0.13376470588235295,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3201,
      "step": 5685
    },
    {
      "epoch": 0.13378823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2664,
      "step": 5686
    },
    {
      "epoch": 0.13381176470588235,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1641,
      "step": 5687
    },
    {
      "epoch": 0.13383529411764705,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1731,
      "step": 5688
    },
    {
      "epoch": 0.13385882352941175,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2111,
      "step": 5689
    },
    {
      "epoch": 0.13388235294117648,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1826,
      "step": 5690
    },
    {
      "epoch": 0.13390588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2382,
      "step": 5691
    },
    {
      "epoch": 0.13392941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2541,
      "step": 5692
    },
    {
      "epoch": 0.13395294117647058,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0247,
      "step": 5693
    },
    {
      "epoch": 0.1339764705882353,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.0261,
      "step": 5694
    },
    {
      "epoch": 0.134,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2083,
      "step": 5695
    },
    {
      "epoch": 0.1340235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2646,
      "step": 5696
    },
    {
      "epoch": 0.1340470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4552,
      "step": 5697
    },
    {
      "epoch": 0.1340705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2084,
      "step": 5698
    },
    {
      "epoch": 0.13409411764705884,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4268,
      "step": 5699
    },
    {
      "epoch": 0.13411764705882354,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1019,
      "step": 5700
    },
    {
      "epoch": 0.13414117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2838,
      "step": 5701
    },
    {
      "epoch": 0.13416470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3518,
      "step": 5702
    },
    {
      "epoch": 0.13418823529411764,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1067,
      "step": 5703
    },
    {
      "epoch": 0.13421176470588236,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.5227,
      "step": 5704
    },
    {
      "epoch": 0.13423529411764706,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2473,
      "step": 5705
    },
    {
      "epoch": 0.13425882352941176,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.288,
      "step": 5706
    },
    {
      "epoch": 0.13428235294117646,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2217,
      "step": 5707
    },
    {
      "epoch": 0.13430588235294116,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9589,
      "step": 5708
    },
    {
      "epoch": 0.1343294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1925,
      "step": 5709
    },
    {
      "epoch": 0.1343529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1981,
      "step": 5710
    },
    {
      "epoch": 0.1343764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.063,
      "step": 5711
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3901,
      "step": 5712
    },
    {
      "epoch": 0.13442352941176472,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.403,
      "step": 5713
    },
    {
      "epoch": 0.13444705882352942,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0468,
      "step": 5714
    },
    {
      "epoch": 0.13447058823529412,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1245,
      "step": 5715
    },
    {
      "epoch": 0.13449411764705882,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3645,
      "step": 5716
    },
    {
      "epoch": 0.13451764705882352,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1853,
      "step": 5717
    },
    {
      "epoch": 0.13454117647058825,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3311,
      "step": 5718
    },
    {
      "epoch": 0.13456470588235295,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.7808,
      "step": 5719
    },
    {
      "epoch": 0.13458823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2523,
      "step": 5720
    },
    {
      "epoch": 0.13461176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3962,
      "step": 5721
    },
    {
      "epoch": 0.13463529411764705,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2229,
      "step": 5722
    },
    {
      "epoch": 0.13465882352941178,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3343,
      "step": 5723
    },
    {
      "epoch": 0.13468235294117648,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 5724
    },
    {
      "epoch": 0.13470588235294118,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1364,
      "step": 5725
    },
    {
      "epoch": 0.13472941176470588,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2499,
      "step": 5726
    },
    {
      "epoch": 0.13475294117647058,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2106,
      "step": 5727
    },
    {
      "epoch": 0.1347764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0065,
      "step": 5728
    },
    {
      "epoch": 0.1348,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.312,
      "step": 5729
    },
    {
      "epoch": 0.1348235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1785,
      "step": 5730
    },
    {
      "epoch": 0.1348470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.2172,
      "step": 5731
    },
    {
      "epoch": 0.13487058823529413,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2066,
      "step": 5732
    },
    {
      "epoch": 0.13489411764705883,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2815,
      "step": 5733
    },
    {
      "epoch": 0.13491764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 5734
    },
    {
      "epoch": 0.13494117647058823,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1115,
      "step": 5735
    },
    {
      "epoch": 0.13496470588235293,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 5736
    },
    {
      "epoch": 0.13498823529411766,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9697,
      "step": 5737
    },
    {
      "epoch": 0.13501176470588236,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.208,
      "step": 5738
    },
    {
      "epoch": 0.13503529411764706,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.8912,
      "step": 5739
    },
    {
      "epoch": 0.13505882352941176,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0325,
      "step": 5740
    },
    {
      "epoch": 0.13508235294117646,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1692,
      "step": 5741
    },
    {
      "epoch": 0.1351058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3344,
      "step": 5742
    },
    {
      "epoch": 0.1351294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4497,
      "step": 5743
    },
    {
      "epoch": 0.1351529411764706,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.1317,
      "step": 5744
    },
    {
      "epoch": 0.1351764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 5745
    },
    {
      "epoch": 0.1352,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1314,
      "step": 5746
    },
    {
      "epoch": 0.13522352941176471,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 5747
    },
    {
      "epoch": 0.13524705882352941,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1823,
      "step": 5748
    },
    {
      "epoch": 0.13527058823529411,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1754,
      "step": 5749
    },
    {
      "epoch": 0.13529411764705881,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0559,
      "step": 5750
    },
    {
      "epoch": 0.13531764705882354,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2868,
      "step": 5751
    },
    {
      "epoch": 0.13534117647058824,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 5752
    },
    {
      "epoch": 0.13536470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1839,
      "step": 5753
    },
    {
      "epoch": 0.13538823529411764,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.8123,
      "step": 5754
    },
    {
      "epoch": 0.13541176470588234,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2007,
      "step": 5755
    },
    {
      "epoch": 0.13543529411764707,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1415,
      "step": 5756
    },
    {
      "epoch": 0.13545882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4848,
      "step": 5757
    },
    {
      "epoch": 0.13548235294117647,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.105,
      "step": 5758
    },
    {
      "epoch": 0.13550588235294117,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3465,
      "step": 5759
    },
    {
      "epoch": 0.13552941176470587,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1321,
      "step": 5760
    },
    {
      "epoch": 0.1355529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 5761
    },
    {
      "epoch": 0.1355764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.07,
      "step": 5762
    },
    {
      "epoch": 0.1356,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9904,
      "step": 5763
    },
    {
      "epoch": 0.1356235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.005,
      "step": 5764
    },
    {
      "epoch": 0.1356470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 5765
    },
    {
      "epoch": 0.13567058823529413,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 5766
    },
    {
      "epoch": 0.13569411764705883,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9635,
      "step": 5767
    },
    {
      "epoch": 0.13571764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0908,
      "step": 5768
    },
    {
      "epoch": 0.13574117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4056,
      "step": 5769
    },
    {
      "epoch": 0.13576470588235295,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3464,
      "step": 5770
    },
    {
      "epoch": 0.13578823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0282,
      "step": 5771
    },
    {
      "epoch": 0.13581176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3307,
      "step": 5772
    },
    {
      "epoch": 0.13583529411764705,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9673,
      "step": 5773
    },
    {
      "epoch": 0.13585882352941175,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9651,
      "step": 5774
    },
    {
      "epoch": 0.13588235294117648,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.917,
      "step": 5775
    },
    {
      "epoch": 0.13590588235294118,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0536,
      "step": 5776
    },
    {
      "epoch": 0.13592941176470588,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2208,
      "step": 5777
    },
    {
      "epoch": 0.13595294117647058,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3335,
      "step": 5778
    },
    {
      "epoch": 0.13597647058823528,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3057,
      "step": 5779
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2967,
      "step": 5780
    },
    {
      "epoch": 0.1360235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0242,
      "step": 5781
    },
    {
      "epoch": 0.1360470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3554,
      "step": 5782
    },
    {
      "epoch": 0.1360705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.4438,
      "step": 5783
    },
    {
      "epoch": 0.1360941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2907,
      "step": 5784
    },
    {
      "epoch": 0.13611764705882354,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9465,
      "step": 5785
    },
    {
      "epoch": 0.13614117647058824,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2307,
      "step": 5786
    },
    {
      "epoch": 0.13616470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4145,
      "step": 5787
    },
    {
      "epoch": 0.13618823529411764,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2797,
      "step": 5788
    },
    {
      "epoch": 0.13621176470588237,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1442,
      "step": 5789
    },
    {
      "epoch": 0.13623529411764707,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.0611,
      "step": 5790
    },
    {
      "epoch": 0.13625882352941177,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1738,
      "step": 5791
    },
    {
      "epoch": 0.13628235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2303,
      "step": 5792
    },
    {
      "epoch": 0.13630588235294117,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.3023,
      "step": 5793
    },
    {
      "epoch": 0.1363294117647059,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1626,
      "step": 5794
    },
    {
      "epoch": 0.1363529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3158,
      "step": 5795
    },
    {
      "epoch": 0.1363764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1535,
      "step": 5796
    },
    {
      "epoch": 0.1364,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1435,
      "step": 5797
    },
    {
      "epoch": 0.1364235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1771,
      "step": 5798
    },
    {
      "epoch": 0.13644705882352942,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2519,
      "step": 5799
    },
    {
      "epoch": 0.13647058823529412,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.2149,
      "step": 5800
    },
    {
      "epoch": 0.13649411764705882,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 0.8256,
      "step": 5801
    },
    {
      "epoch": 0.13651764705882352,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 5802
    },
    {
      "epoch": 0.13654117647058825,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2719,
      "step": 5803
    },
    {
      "epoch": 0.13656470588235295,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.2058,
      "step": 5804
    },
    {
      "epoch": 0.13658823529411765,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.9896,
      "step": 5805
    },
    {
      "epoch": 0.13661176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.5795,
      "step": 5806
    },
    {
      "epoch": 0.13663529411764705,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0162,
      "step": 5807
    },
    {
      "epoch": 0.13665882352941178,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4632,
      "step": 5808
    },
    {
      "epoch": 0.13668235294117648,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1348,
      "step": 5809
    },
    {
      "epoch": 0.13670588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3849,
      "step": 5810
    },
    {
      "epoch": 0.13672941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4714,
      "step": 5811
    },
    {
      "epoch": 0.13675294117647058,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1689,
      "step": 5812
    },
    {
      "epoch": 0.1367764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3837,
      "step": 5813
    },
    {
      "epoch": 0.1368,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2594,
      "step": 5814
    },
    {
      "epoch": 0.1368235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3215,
      "step": 5815
    },
    {
      "epoch": 0.1368470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0772,
      "step": 5816
    },
    {
      "epoch": 0.1368705882352941,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 1.0778,
      "step": 5817
    },
    {
      "epoch": 0.13689411764705883,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0438,
      "step": 5818
    },
    {
      "epoch": 0.13691764705882353,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1369,
      "step": 5819
    },
    {
      "epoch": 0.13694117647058823,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2283,
      "step": 5820
    },
    {
      "epoch": 0.13696470588235293,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2801,
      "step": 5821
    },
    {
      "epoch": 0.13698823529411766,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.1139,
      "step": 5822
    },
    {
      "epoch": 0.13701176470588236,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0831,
      "step": 5823
    },
    {
      "epoch": 0.13703529411764706,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8397,
      "step": 5824
    },
    {
      "epoch": 0.13705882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2546,
      "step": 5825
    },
    {
      "epoch": 0.13708235294117646,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.194,
      "step": 5826
    },
    {
      "epoch": 0.1371058823529412,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0918,
      "step": 5827
    },
    {
      "epoch": 0.1371294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2322,
      "step": 5828
    },
    {
      "epoch": 0.1371529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4063,
      "step": 5829
    },
    {
      "epoch": 0.1371764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2542,
      "step": 5830
    },
    {
      "epoch": 0.1372,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0264,
      "step": 5831
    },
    {
      "epoch": 0.13722352941176472,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0302,
      "step": 5832
    },
    {
      "epoch": 0.13724705882352942,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8846,
      "step": 5833
    },
    {
      "epoch": 0.13727058823529412,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1416,
      "step": 5834
    },
    {
      "epoch": 0.13729411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2598,
      "step": 5835
    },
    {
      "epoch": 0.13731764705882352,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1479,
      "step": 5836
    },
    {
      "epoch": 0.13734117647058824,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.097,
      "step": 5837
    },
    {
      "epoch": 0.13736470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.818,
      "step": 5838
    },
    {
      "epoch": 0.13738823529411764,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2454,
      "step": 5839
    },
    {
      "epoch": 0.13741176470588234,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0606,
      "step": 5840
    },
    {
      "epoch": 0.13743529411764707,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2163,
      "step": 5841
    },
    {
      "epoch": 0.13745882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4938,
      "step": 5842
    },
    {
      "epoch": 0.13748235294117647,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0047,
      "step": 5843
    },
    {
      "epoch": 0.13750588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1951,
      "step": 5844
    },
    {
      "epoch": 0.13752941176470587,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1741,
      "step": 5845
    },
    {
      "epoch": 0.1375529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 5846
    },
    {
      "epoch": 0.1375764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2335,
      "step": 5847
    },
    {
      "epoch": 0.1376,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3507,
      "step": 5848
    },
    {
      "epoch": 0.1376235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4746,
      "step": 5849
    },
    {
      "epoch": 0.1376470588235294,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1843,
      "step": 5850
    },
    {
      "epoch": 0.13767058823529413,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2831,
      "step": 5851
    },
    {
      "epoch": 0.13769411764705883,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1039,
      "step": 5852
    },
    {
      "epoch": 0.13771764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 5853
    },
    {
      "epoch": 0.13774117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3135,
      "step": 5854
    },
    {
      "epoch": 0.13776470588235293,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3037,
      "step": 5855
    },
    {
      "epoch": 0.13778823529411766,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8553,
      "step": 5856
    },
    {
      "epoch": 0.13781176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3302,
      "step": 5857
    },
    {
      "epoch": 0.13783529411764706,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 5858
    },
    {
      "epoch": 0.13785882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.501,
      "step": 5859
    },
    {
      "epoch": 0.13788235294117648,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3338,
      "step": 5860
    },
    {
      "epoch": 0.13790588235294118,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.9933,
      "step": 5861
    },
    {
      "epoch": 0.13792941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1316,
      "step": 5862
    },
    {
      "epoch": 0.13795294117647058,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2656,
      "step": 5863
    },
    {
      "epoch": 0.13797647058823528,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2634,
      "step": 5864
    },
    {
      "epoch": 0.138,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2082,
      "step": 5865
    },
    {
      "epoch": 0.1380235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9912,
      "step": 5866
    },
    {
      "epoch": 0.1380470588235294,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.8676,
      "step": 5867
    },
    {
      "epoch": 0.1380705882352941,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8849,
      "step": 5868
    },
    {
      "epoch": 0.1380941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2814,
      "step": 5869
    },
    {
      "epoch": 0.13811764705882354,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9895,
      "step": 5870
    },
    {
      "epoch": 0.13814117647058824,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.4009,
      "step": 5871
    },
    {
      "epoch": 0.13816470588235294,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0541,
      "step": 5872
    },
    {
      "epoch": 0.13818823529411764,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.3017,
      "step": 5873
    },
    {
      "epoch": 0.13821176470588234,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4144,
      "step": 5874
    },
    {
      "epoch": 0.13823529411764707,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 5875
    },
    {
      "epoch": 0.13825882352941177,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2372,
      "step": 5876
    },
    {
      "epoch": 0.13828235294117647,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 0.9008,
      "step": 5877
    },
    {
      "epoch": 0.13830588235294117,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.4399,
      "step": 5878
    },
    {
      "epoch": 0.1383294117647059,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.045,
      "step": 5879
    },
    {
      "epoch": 0.1383529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.2177,
      "step": 5880
    },
    {
      "epoch": 0.1383764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2109,
      "step": 5881
    },
    {
      "epoch": 0.1384,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0499,
      "step": 5882
    },
    {
      "epoch": 0.1384235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1738,
      "step": 5883
    },
    {
      "epoch": 0.13844705882352942,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2028,
      "step": 5884
    },
    {
      "epoch": 0.13847058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3131,
      "step": 5885
    },
    {
      "epoch": 0.13849411764705882,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4604,
      "step": 5886
    },
    {
      "epoch": 0.13851764705882352,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2206,
      "step": 5887
    },
    {
      "epoch": 0.13854117647058822,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 5888
    },
    {
      "epoch": 0.13856470588235295,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 5889
    },
    {
      "epoch": 0.13858823529411765,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2785,
      "step": 5890
    },
    {
      "epoch": 0.13861176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2574,
      "step": 5891
    },
    {
      "epoch": 0.13863529411764705,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2808,
      "step": 5892
    },
    {
      "epoch": 0.13865882352941175,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4669,
      "step": 5893
    },
    {
      "epoch": 0.13868235294117648,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9762,
      "step": 5894
    },
    {
      "epoch": 0.13870588235294118,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1151,
      "step": 5895
    },
    {
      "epoch": 0.13872941176470588,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2466,
      "step": 5896
    },
    {
      "epoch": 0.13875294117647058,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0197,
      "step": 5897
    },
    {
      "epoch": 0.1387764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9393,
      "step": 5898
    },
    {
      "epoch": 0.1388,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 5899
    },
    {
      "epoch": 0.1388235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 5900
    },
    {
      "epoch": 0.1388470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3388,
      "step": 5901
    },
    {
      "epoch": 0.1388705882352941,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0266,
      "step": 5902
    },
    {
      "epoch": 0.13889411764705883,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1732,
      "step": 5903
    },
    {
      "epoch": 0.13891764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0944,
      "step": 5904
    },
    {
      "epoch": 0.13894117647058823,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1503,
      "step": 5905
    },
    {
      "epoch": 0.13896470588235293,
      "grad_norm": 0.302734375,
      "learning_rate": 0.02,
      "loss": 0.8935,
      "step": 5906
    },
    {
      "epoch": 0.13898823529411763,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2306,
      "step": 5907
    },
    {
      "epoch": 0.13901176470588236,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1543,
      "step": 5908
    },
    {
      "epoch": 0.13903529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.4487,
      "step": 5909
    },
    {
      "epoch": 0.13905882352941176,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.077,
      "step": 5910
    },
    {
      "epoch": 0.13908235294117646,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 5911
    },
    {
      "epoch": 0.1391058823529412,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2536,
      "step": 5912
    },
    {
      "epoch": 0.1391294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1938,
      "step": 5913
    },
    {
      "epoch": 0.1391529411764706,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.0138,
      "step": 5914
    },
    {
      "epoch": 0.1391764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3553,
      "step": 5915
    },
    {
      "epoch": 0.1392,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9105,
      "step": 5916
    },
    {
      "epoch": 0.13922352941176472,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 5917
    },
    {
      "epoch": 0.13924705882352942,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.998,
      "step": 5918
    },
    {
      "epoch": 0.13927058823529412,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.8598,
      "step": 5919
    },
    {
      "epoch": 0.13929411764705882,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.275,
      "step": 5920
    },
    {
      "epoch": 0.13931764705882352,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0371,
      "step": 5921
    },
    {
      "epoch": 0.13934117647058825,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4162,
      "step": 5922
    },
    {
      "epoch": 0.13936470588235295,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.027,
      "step": 5923
    },
    {
      "epoch": 0.13938823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3936,
      "step": 5924
    },
    {
      "epoch": 0.13941176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3843,
      "step": 5925
    },
    {
      "epoch": 0.13943529411764705,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 5926
    },
    {
      "epoch": 0.13945882352941177,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2523,
      "step": 5927
    },
    {
      "epoch": 0.13948235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2297,
      "step": 5928
    },
    {
      "epoch": 0.13950588235294117,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1108,
      "step": 5929
    },
    {
      "epoch": 0.13952941176470587,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1478,
      "step": 5930
    },
    {
      "epoch": 0.1395529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1111,
      "step": 5931
    },
    {
      "epoch": 0.1395764705882353,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.2302,
      "step": 5932
    },
    {
      "epoch": 0.1396,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0175,
      "step": 5933
    },
    {
      "epoch": 0.1396235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1553,
      "step": 5934
    },
    {
      "epoch": 0.1396470588235294,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1449,
      "step": 5935
    },
    {
      "epoch": 0.13967058823529413,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.2609,
      "step": 5936
    },
    {
      "epoch": 0.13969411764705883,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1984,
      "step": 5937
    },
    {
      "epoch": 0.13971764705882353,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.8172,
      "step": 5938
    },
    {
      "epoch": 0.13974117647058823,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0526,
      "step": 5939
    },
    {
      "epoch": 0.13976470588235293,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.4027,
      "step": 5940
    },
    {
      "epoch": 0.13978823529411766,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2213,
      "step": 5941
    },
    {
      "epoch": 0.13981176470588236,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2573,
      "step": 5942
    },
    {
      "epoch": 0.13983529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4724,
      "step": 5943
    },
    {
      "epoch": 0.13985882352941176,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1576,
      "step": 5944
    },
    {
      "epoch": 0.13988235294117646,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5667,
      "step": 5945
    },
    {
      "epoch": 0.13990588235294119,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1533,
      "step": 5946
    },
    {
      "epoch": 0.13992941176470589,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2866,
      "step": 5947
    },
    {
      "epoch": 0.13995294117647059,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9699,
      "step": 5948
    },
    {
      "epoch": 0.13997647058823529,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1808,
      "step": 5949
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2977,
      "step": 5950
    },
    {
      "epoch": 0.1400235294117647,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0474,
      "step": 5951
    },
    {
      "epoch": 0.1400470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2169,
      "step": 5952
    },
    {
      "epoch": 0.1400705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0253,
      "step": 5953
    },
    {
      "epoch": 0.1400941176470588,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2976,
      "step": 5954
    },
    {
      "epoch": 0.14011764705882354,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.2444,
      "step": 5955
    },
    {
      "epoch": 0.14014117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2468,
      "step": 5956
    },
    {
      "epoch": 0.14016470588235294,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3177,
      "step": 5957
    },
    {
      "epoch": 0.14018823529411764,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1987,
      "step": 5958
    },
    {
      "epoch": 0.14021176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2379,
      "step": 5959
    },
    {
      "epoch": 0.14023529411764707,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2272,
      "step": 5960
    },
    {
      "epoch": 0.14025882352941177,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2351,
      "step": 5961
    },
    {
      "epoch": 0.14028235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0522,
      "step": 5962
    },
    {
      "epoch": 0.14030588235294117,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4882,
      "step": 5963
    },
    {
      "epoch": 0.14032941176470587,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 5964
    },
    {
      "epoch": 0.1403529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 5965
    },
    {
      "epoch": 0.1403764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.108,
      "step": 5966
    },
    {
      "epoch": 0.1404,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0431,
      "step": 5967
    },
    {
      "epoch": 0.1404235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1852,
      "step": 5968
    },
    {
      "epoch": 0.14044705882352942,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1335,
      "step": 5969
    },
    {
      "epoch": 0.14047058823529412,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1374,
      "step": 5970
    },
    {
      "epoch": 0.14049411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2469,
      "step": 5971
    },
    {
      "epoch": 0.14051764705882352,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9478,
      "step": 5972
    },
    {
      "epoch": 0.14054117647058822,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8063,
      "step": 5973
    },
    {
      "epoch": 0.14056470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.5331,
      "step": 5974
    },
    {
      "epoch": 0.14058823529411765,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3015,
      "step": 5975
    },
    {
      "epoch": 0.14061176470588235,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1005,
      "step": 5976
    },
    {
      "epoch": 0.14063529411764705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2829,
      "step": 5977
    },
    {
      "epoch": 0.14065882352941175,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1581,
      "step": 5978
    },
    {
      "epoch": 0.14068235294117648,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4231,
      "step": 5979
    },
    {
      "epoch": 0.14070588235294118,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2459,
      "step": 5980
    },
    {
      "epoch": 0.14072941176470588,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4044,
      "step": 5981
    },
    {
      "epoch": 0.14075294117647058,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3992,
      "step": 5982
    },
    {
      "epoch": 0.14077647058823528,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1157,
      "step": 5983
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0807,
      "step": 5984
    },
    {
      "epoch": 0.1408235294117647,
      "grad_norm": 0.30078125,
      "learning_rate": 0.02,
      "loss": 0.6182,
      "step": 5985
    },
    {
      "epoch": 0.1408470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3203,
      "step": 5986
    },
    {
      "epoch": 0.1408705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3039,
      "step": 5987
    },
    {
      "epoch": 0.14089411764705884,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4387,
      "step": 5988
    },
    {
      "epoch": 0.14091764705882354,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0614,
      "step": 5989
    },
    {
      "epoch": 0.14094117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3242,
      "step": 5990
    },
    {
      "epoch": 0.14096470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.118,
      "step": 5991
    },
    {
      "epoch": 0.14098823529411764,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2449,
      "step": 5992
    },
    {
      "epoch": 0.14101176470588236,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1576,
      "step": 5993
    },
    {
      "epoch": 0.14103529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4822,
      "step": 5994
    },
    {
      "epoch": 0.14105882352941176,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1714,
      "step": 5995
    },
    {
      "epoch": 0.14108235294117646,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1963,
      "step": 5996
    },
    {
      "epoch": 0.14110588235294116,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.6587,
      "step": 5997
    },
    {
      "epoch": 0.1411294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2898,
      "step": 5998
    },
    {
      "epoch": 0.1411529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 5999
    },
    {
      "epoch": 0.1411764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9342,
      "step": 6000
    },
    {
      "epoch": 0.1411764705882353,
      "eval_loss": 2.2358450889587402,
      "eval_runtime": 681.3738,
      "eval_samples_per_second": 12.475,
      "eval_steps_per_second": 3.119,
      "step": 6000
    },
    {
      "epoch": 0.1412,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2199,
      "step": 6001
    },
    {
      "epoch": 0.1412235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9824,
      "step": 6002
    },
    {
      "epoch": 0.14124705882352942,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3711,
      "step": 6003
    },
    {
      "epoch": 0.14127058823529412,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9074,
      "step": 6004
    },
    {
      "epoch": 0.14129411764705882,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8751,
      "step": 6005
    },
    {
      "epoch": 0.14131764705882352,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1636,
      "step": 6006
    },
    {
      "epoch": 0.14134117647058825,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2235,
      "step": 6007
    },
    {
      "epoch": 0.14136470588235295,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.343,
      "step": 6008
    },
    {
      "epoch": 0.14138823529411765,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.325,
      "step": 6009
    },
    {
      "epoch": 0.14141176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4879,
      "step": 6010
    },
    {
      "epoch": 0.14143529411764705,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 6011
    },
    {
      "epoch": 0.14145882352941178,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1575,
      "step": 6012
    },
    {
      "epoch": 0.14148235294117648,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1314,
      "step": 6013
    },
    {
      "epoch": 0.14150588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0706,
      "step": 6014
    },
    {
      "epoch": 0.14152941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2803,
      "step": 6015
    },
    {
      "epoch": 0.14155294117647058,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0212,
      "step": 6016
    },
    {
      "epoch": 0.1415764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8992,
      "step": 6017
    },
    {
      "epoch": 0.1416,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9778,
      "step": 6018
    },
    {
      "epoch": 0.1416235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.112,
      "step": 6019
    },
    {
      "epoch": 0.1416470588235294,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2504,
      "step": 6020
    },
    {
      "epoch": 0.14167058823529413,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.549,
      "step": 6021
    },
    {
      "epoch": 0.14169411764705883,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3433,
      "step": 6022
    },
    {
      "epoch": 0.14171764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9879,
      "step": 6023
    },
    {
      "epoch": 0.14174117647058823,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 6024
    },
    {
      "epoch": 0.14176470588235293,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.4187,
      "step": 6025
    },
    {
      "epoch": 0.14178823529411766,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3208,
      "step": 6026
    },
    {
      "epoch": 0.14181176470588236,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8172,
      "step": 6027
    },
    {
      "epoch": 0.14183529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2057,
      "step": 6028
    },
    {
      "epoch": 0.14185882352941176,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4078,
      "step": 6029
    },
    {
      "epoch": 0.14188235294117646,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3668,
      "step": 6030
    },
    {
      "epoch": 0.1419058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.5308,
      "step": 6031
    },
    {
      "epoch": 0.1419294117647059,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0101,
      "step": 6032
    },
    {
      "epoch": 0.1419529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1042,
      "step": 6033
    },
    {
      "epoch": 0.1419764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1274,
      "step": 6034
    },
    {
      "epoch": 0.142,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.277,
      "step": 6035
    },
    {
      "epoch": 0.14202352941176472,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.076,
      "step": 6036
    },
    {
      "epoch": 0.14204705882352942,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2526,
      "step": 6037
    },
    {
      "epoch": 0.14207058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2965,
      "step": 6038
    },
    {
      "epoch": 0.14209411764705882,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1951,
      "step": 6039
    },
    {
      "epoch": 0.14211764705882354,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2977,
      "step": 6040
    },
    {
      "epoch": 0.14214117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4423,
      "step": 6041
    },
    {
      "epoch": 0.14216470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3425,
      "step": 6042
    },
    {
      "epoch": 0.14218823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.032,
      "step": 6043
    },
    {
      "epoch": 0.14221176470588234,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1362,
      "step": 6044
    },
    {
      "epoch": 0.14223529411764707,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2346,
      "step": 6045
    },
    {
      "epoch": 0.14225882352941177,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9355,
      "step": 6046
    },
    {
      "epoch": 0.14228235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2876,
      "step": 6047
    },
    {
      "epoch": 0.14230588235294117,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3334,
      "step": 6048
    },
    {
      "epoch": 0.14232941176470587,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2725,
      "step": 6049
    },
    {
      "epoch": 0.1423529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4234,
      "step": 6050
    },
    {
      "epoch": 0.1423764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2791,
      "step": 6051
    },
    {
      "epoch": 0.1424,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2367,
      "step": 6052
    },
    {
      "epoch": 0.1424235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3301,
      "step": 6053
    },
    {
      "epoch": 0.1424470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3402,
      "step": 6054
    },
    {
      "epoch": 0.14247058823529413,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3998,
      "step": 6055
    },
    {
      "epoch": 0.14249411764705883,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1493,
      "step": 6056
    },
    {
      "epoch": 0.14251764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0177,
      "step": 6057
    },
    {
      "epoch": 0.14254117647058823,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.4125,
      "step": 6058
    },
    {
      "epoch": 0.14256470588235295,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9963,
      "step": 6059
    },
    {
      "epoch": 0.14258823529411765,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0222,
      "step": 6060
    },
    {
      "epoch": 0.14261176470588235,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1753,
      "step": 6061
    },
    {
      "epoch": 0.14263529411764705,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1362,
      "step": 6062
    },
    {
      "epoch": 0.14265882352941175,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3403,
      "step": 6063
    },
    {
      "epoch": 0.14268235294117648,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0733,
      "step": 6064
    },
    {
      "epoch": 0.14270588235294118,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.249,
      "step": 6065
    },
    {
      "epoch": 0.14272941176470588,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1416,
      "step": 6066
    },
    {
      "epoch": 0.14275294117647058,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1728,
      "step": 6067
    },
    {
      "epoch": 0.14277647058823528,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2097,
      "step": 6068
    },
    {
      "epoch": 0.1428,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2913,
      "step": 6069
    },
    {
      "epoch": 0.1428235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3833,
      "step": 6070
    },
    {
      "epoch": 0.1428470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8831,
      "step": 6071
    },
    {
      "epoch": 0.1428705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1733,
      "step": 6072
    },
    {
      "epoch": 0.1428941176470588,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.4938,
      "step": 6073
    },
    {
      "epoch": 0.14291764705882354,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.8995,
      "step": 6074
    },
    {
      "epoch": 0.14294117647058824,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0834,
      "step": 6075
    },
    {
      "epoch": 0.14296470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1892,
      "step": 6076
    },
    {
      "epoch": 0.14298823529411764,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9974,
      "step": 6077
    },
    {
      "epoch": 0.14301176470588237,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.4003,
      "step": 6078
    },
    {
      "epoch": 0.14303529411764707,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.8289,
      "step": 6079
    },
    {
      "epoch": 0.14305882352941177,
      "grad_norm": 0.30078125,
      "learning_rate": 0.02,
      "loss": 0.7318,
      "step": 6080
    },
    {
      "epoch": 0.14308235294117647,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 1.1668,
      "step": 6081
    },
    {
      "epoch": 0.14310588235294117,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1864,
      "step": 6082
    },
    {
      "epoch": 0.1431294117647059,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1945,
      "step": 6083
    },
    {
      "epoch": 0.1431529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 6084
    },
    {
      "epoch": 0.1431764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3073,
      "step": 6085
    },
    {
      "epoch": 0.1432,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1257,
      "step": 6086
    },
    {
      "epoch": 0.1432235294117647,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.672,
      "step": 6087
    },
    {
      "epoch": 0.14324705882352942,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.125,
      "step": 6088
    },
    {
      "epoch": 0.14327058823529412,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 0.8496,
      "step": 6089
    },
    {
      "epoch": 0.14329411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1752,
      "step": 6090
    },
    {
      "epoch": 0.14331764705882352,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.235,
      "step": 6091
    },
    {
      "epoch": 0.14334117647058822,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 6092
    },
    {
      "epoch": 0.14336470588235295,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2581,
      "step": 6093
    },
    {
      "epoch": 0.14338823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.5268,
      "step": 6094
    },
    {
      "epoch": 0.14341176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9688,
      "step": 6095
    },
    {
      "epoch": 0.14343529411764705,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.169,
      "step": 6096
    },
    {
      "epoch": 0.14345882352941178,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1326,
      "step": 6097
    },
    {
      "epoch": 0.14348235294117648,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0029,
      "step": 6098
    },
    {
      "epoch": 0.14350588235294118,
      "grad_norm": 0.30078125,
      "learning_rate": 0.02,
      "loss": 0.7574,
      "step": 6099
    },
    {
      "epoch": 0.14352941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3714,
      "step": 6100
    },
    {
      "epoch": 0.14355294117647058,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9753,
      "step": 6101
    },
    {
      "epoch": 0.1435764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2063,
      "step": 6102
    },
    {
      "epoch": 0.1436,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0754,
      "step": 6103
    },
    {
      "epoch": 0.1436235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3239,
      "step": 6104
    },
    {
      "epoch": 0.1436470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2136,
      "step": 6105
    },
    {
      "epoch": 0.1436705882352941,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1087,
      "step": 6106
    },
    {
      "epoch": 0.14369411764705883,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2716,
      "step": 6107
    },
    {
      "epoch": 0.14371764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2404,
      "step": 6108
    },
    {
      "epoch": 0.14374117647058823,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 6109
    },
    {
      "epoch": 0.14376470588235293,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3895,
      "step": 6110
    },
    {
      "epoch": 0.14378823529411763,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 0.9111,
      "step": 6111
    },
    {
      "epoch": 0.14381176470588236,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2597,
      "step": 6112
    },
    {
      "epoch": 0.14383529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2684,
      "step": 6113
    },
    {
      "epoch": 0.14385882352941176,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1307,
      "step": 6114
    },
    {
      "epoch": 0.14388235294117646,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.2243,
      "step": 6115
    },
    {
      "epoch": 0.1439058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.146,
      "step": 6116
    },
    {
      "epoch": 0.1439294117647059,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.1269,
      "step": 6117
    },
    {
      "epoch": 0.1439529411764706,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1666,
      "step": 6118
    },
    {
      "epoch": 0.1439764705882353,
      "grad_norm": 0.32421875,
      "learning_rate": 0.02,
      "loss": 0.7756,
      "step": 6119
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2106,
      "step": 6120
    },
    {
      "epoch": 0.14402352941176472,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3836,
      "step": 6121
    },
    {
      "epoch": 0.14404705882352942,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2296,
      "step": 6122
    },
    {
      "epoch": 0.14407058823529412,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1137,
      "step": 6123
    },
    {
      "epoch": 0.14409411764705882,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.7328,
      "step": 6124
    },
    {
      "epoch": 0.14411764705882352,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1823,
      "step": 6125
    },
    {
      "epoch": 0.14414117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.637,
      "step": 6126
    },
    {
      "epoch": 0.14416470588235294,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.2907,
      "step": 6127
    },
    {
      "epoch": 0.14418823529411764,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.3423,
      "step": 6128
    },
    {
      "epoch": 0.14421176470588234,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 0.8475,
      "step": 6129
    },
    {
      "epoch": 0.14423529411764707,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5942,
      "step": 6130
    },
    {
      "epoch": 0.14425882352941177,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2126,
      "step": 6131
    },
    {
      "epoch": 0.14428235294117647,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0715,
      "step": 6132
    },
    {
      "epoch": 0.14430588235294117,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2044,
      "step": 6133
    },
    {
      "epoch": 0.14432941176470587,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.899,
      "step": 6134
    },
    {
      "epoch": 0.1443529411764706,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 6135
    },
    {
      "epoch": 0.1443764705882353,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1698,
      "step": 6136
    },
    {
      "epoch": 0.1444,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0403,
      "step": 6137
    },
    {
      "epoch": 0.1444235294117647,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1676,
      "step": 6138
    },
    {
      "epoch": 0.1444470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1711,
      "step": 6139
    },
    {
      "epoch": 0.14447058823529413,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3488,
      "step": 6140
    },
    {
      "epoch": 0.14449411764705883,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0352,
      "step": 6141
    },
    {
      "epoch": 0.14451764705882353,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 0.9516,
      "step": 6142
    },
    {
      "epoch": 0.14454117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3653,
      "step": 6143
    },
    {
      "epoch": 0.14456470588235293,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0466,
      "step": 6144
    },
    {
      "epoch": 0.14458823529411766,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0865,
      "step": 6145
    },
    {
      "epoch": 0.14461176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1837,
      "step": 6146
    },
    {
      "epoch": 0.14463529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 6147
    },
    {
      "epoch": 0.14465882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0982,
      "step": 6148
    },
    {
      "epoch": 0.14468235294117648,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 0.6924,
      "step": 6149
    },
    {
      "epoch": 0.14470588235294118,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3486,
      "step": 6150
    },
    {
      "epoch": 0.14472941176470588,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3128,
      "step": 6151
    },
    {
      "epoch": 0.14475294117647058,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0833,
      "step": 6152
    },
    {
      "epoch": 0.14477647058823528,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1531,
      "step": 6153
    },
    {
      "epoch": 0.1448,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3148,
      "step": 6154
    },
    {
      "epoch": 0.1448235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4343,
      "step": 6155
    },
    {
      "epoch": 0.1448470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3117,
      "step": 6156
    },
    {
      "epoch": 0.1448705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2386,
      "step": 6157
    },
    {
      "epoch": 0.1448941176470588,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2124,
      "step": 6158
    },
    {
      "epoch": 0.14491764705882354,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.2417,
      "step": 6159
    },
    {
      "epoch": 0.14494117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1795,
      "step": 6160
    },
    {
      "epoch": 0.14496470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.363,
      "step": 6161
    },
    {
      "epoch": 0.14498823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 6162
    },
    {
      "epoch": 0.14501176470588234,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2243,
      "step": 6163
    },
    {
      "epoch": 0.14503529411764707,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 0.8262,
      "step": 6164
    },
    {
      "epoch": 0.14505882352941177,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0502,
      "step": 6165
    },
    {
      "epoch": 0.14508235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2145,
      "step": 6166
    },
    {
      "epoch": 0.14510588235294117,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 6167
    },
    {
      "epoch": 0.1451294117647059,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0499,
      "step": 6168
    },
    {
      "epoch": 0.1451529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2593,
      "step": 6169
    },
    {
      "epoch": 0.1451764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.5985,
      "step": 6170
    },
    {
      "epoch": 0.1452,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3335,
      "step": 6171
    },
    {
      "epoch": 0.1452235294117647,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0042,
      "step": 6172
    },
    {
      "epoch": 0.14524705882352942,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.8719,
      "step": 6173
    },
    {
      "epoch": 0.14527058823529412,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1324,
      "step": 6174
    },
    {
      "epoch": 0.14529411764705882,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9926,
      "step": 6175
    },
    {
      "epoch": 0.14531764705882352,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2137,
      "step": 6176
    },
    {
      "epoch": 0.14534117647058822,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1588,
      "step": 6177
    },
    {
      "epoch": 0.14536470588235295,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1716,
      "step": 6178
    },
    {
      "epoch": 0.14538823529411765,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0806,
      "step": 6179
    },
    {
      "epoch": 0.14541176470588235,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0874,
      "step": 6180
    },
    {
      "epoch": 0.14543529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2075,
      "step": 6181
    },
    {
      "epoch": 0.14545882352941175,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2247,
      "step": 6182
    },
    {
      "epoch": 0.14548235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1257,
      "step": 6183
    },
    {
      "epoch": 0.14550588235294118,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9638,
      "step": 6184
    },
    {
      "epoch": 0.14552941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0045,
      "step": 6185
    },
    {
      "epoch": 0.14555294117647058,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1206,
      "step": 6186
    },
    {
      "epoch": 0.1455764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2854,
      "step": 6187
    },
    {
      "epoch": 0.1456,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.3064,
      "step": 6188
    },
    {
      "epoch": 0.1456235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0279,
      "step": 6189
    },
    {
      "epoch": 0.1456470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9609,
      "step": 6190
    },
    {
      "epoch": 0.1456705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2209,
      "step": 6191
    },
    {
      "epoch": 0.14569411764705884,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2612,
      "step": 6192
    },
    {
      "epoch": 0.14571764705882354,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.2061,
      "step": 6193
    },
    {
      "epoch": 0.14574117647058824,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1119,
      "step": 6194
    },
    {
      "epoch": 0.14576470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1623,
      "step": 6195
    },
    {
      "epoch": 0.14578823529411764,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2718,
      "step": 6196
    },
    {
      "epoch": 0.14581176470588236,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 6197
    },
    {
      "epoch": 0.14583529411764706,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0698,
      "step": 6198
    },
    {
      "epoch": 0.14585882352941176,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.801,
      "step": 6199
    },
    {
      "epoch": 0.14588235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3062,
      "step": 6200
    },
    {
      "epoch": 0.14590588235294116,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3743,
      "step": 6201
    },
    {
      "epoch": 0.1459294117647059,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 0.7144,
      "step": 6202
    },
    {
      "epoch": 0.1459529411764706,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9698,
      "step": 6203
    },
    {
      "epoch": 0.1459764705882353,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.8837,
      "step": 6204
    },
    {
      "epoch": 0.146,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2078,
      "step": 6205
    },
    {
      "epoch": 0.14602352941176472,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.215,
      "step": 6206
    },
    {
      "epoch": 0.14604705882352942,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8998,
      "step": 6207
    },
    {
      "epoch": 0.14607058823529412,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2473,
      "step": 6208
    },
    {
      "epoch": 0.14609411764705882,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 6209
    },
    {
      "epoch": 0.14611764705882352,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3392,
      "step": 6210
    },
    {
      "epoch": 0.14614117647058825,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1336,
      "step": 6211
    },
    {
      "epoch": 0.14616470588235295,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3755,
      "step": 6212
    },
    {
      "epoch": 0.14618823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1716,
      "step": 6213
    },
    {
      "epoch": 0.14621176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3169,
      "step": 6214
    },
    {
      "epoch": 0.14623529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2995,
      "step": 6215
    },
    {
      "epoch": 0.14625882352941177,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 6216
    },
    {
      "epoch": 0.14628235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0635,
      "step": 6217
    },
    {
      "epoch": 0.14630588235294117,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2179,
      "step": 6218
    },
    {
      "epoch": 0.14632941176470587,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 6219
    },
    {
      "epoch": 0.14635294117647057,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.326,
      "step": 6220
    },
    {
      "epoch": 0.1463764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.318,
      "step": 6221
    },
    {
      "epoch": 0.1464,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 6222
    },
    {
      "epoch": 0.1464235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.186,
      "step": 6223
    },
    {
      "epoch": 0.1464470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2152,
      "step": 6224
    },
    {
      "epoch": 0.14647058823529413,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0601,
      "step": 6225
    },
    {
      "epoch": 0.14649411764705883,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 6226
    },
    {
      "epoch": 0.14651764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3055,
      "step": 6227
    },
    {
      "epoch": 0.14654117647058823,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3699,
      "step": 6228
    },
    {
      "epoch": 0.14656470588235293,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.845,
      "step": 6229
    },
    {
      "epoch": 0.14658823529411766,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9382,
      "step": 6230
    },
    {
      "epoch": 0.14661176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2088,
      "step": 6231
    },
    {
      "epoch": 0.14663529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4049,
      "step": 6232
    },
    {
      "epoch": 0.14665882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2562,
      "step": 6233
    },
    {
      "epoch": 0.14668235294117646,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2859,
      "step": 6234
    },
    {
      "epoch": 0.14670588235294119,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3587,
      "step": 6235
    },
    {
      "epoch": 0.14672941176470589,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0094,
      "step": 6236
    },
    {
      "epoch": 0.14675294117647059,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3251,
      "step": 6237
    },
    {
      "epoch": 0.14677647058823529,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.287,
      "step": 6238
    },
    {
      "epoch": 0.1468,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2528,
      "step": 6239
    },
    {
      "epoch": 0.1468235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1825,
      "step": 6240
    },
    {
      "epoch": 0.1468470588235294,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9012,
      "step": 6241
    },
    {
      "epoch": 0.14687058823529411,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1557,
      "step": 6242
    },
    {
      "epoch": 0.14689411764705881,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 6243
    },
    {
      "epoch": 0.14691764705882354,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8469,
      "step": 6244
    },
    {
      "epoch": 0.14694117647058824,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1684,
      "step": 6245
    },
    {
      "epoch": 0.14696470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1992,
      "step": 6246
    },
    {
      "epoch": 0.14698823529411764,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2617,
      "step": 6247
    },
    {
      "epoch": 0.14701176470588234,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.2148,
      "step": 6248
    },
    {
      "epoch": 0.14703529411764707,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.3715,
      "step": 6249
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3849,
      "step": 6250
    },
    {
      "epoch": 0.14708235294117647,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1344,
      "step": 6251
    },
    {
      "epoch": 0.14710588235294117,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.484,
      "step": 6252
    },
    {
      "epoch": 0.14712941176470587,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2152,
      "step": 6253
    },
    {
      "epoch": 0.1471529411764706,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0181,
      "step": 6254
    },
    {
      "epoch": 0.1471764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1115,
      "step": 6255
    },
    {
      "epoch": 0.1472,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2623,
      "step": 6256
    },
    {
      "epoch": 0.1472235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1809,
      "step": 6257
    },
    {
      "epoch": 0.14724705882352943,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2329,
      "step": 6258
    },
    {
      "epoch": 0.14727058823529413,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0157,
      "step": 6259
    },
    {
      "epoch": 0.14729411764705883,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.1651,
      "step": 6260
    },
    {
      "epoch": 0.14731764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2591,
      "step": 6261
    },
    {
      "epoch": 0.14734117647058823,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1705,
      "step": 6262
    },
    {
      "epoch": 0.14736470588235295,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0433,
      "step": 6263
    },
    {
      "epoch": 0.14738823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1832,
      "step": 6264
    },
    {
      "epoch": 0.14741176470588235,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0624,
      "step": 6265
    },
    {
      "epoch": 0.14743529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1097,
      "step": 6266
    },
    {
      "epoch": 0.14745882352941175,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1384,
      "step": 6267
    },
    {
      "epoch": 0.14748235294117648,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3905,
      "step": 6268
    },
    {
      "epoch": 0.14750588235294118,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3389,
      "step": 6269
    },
    {
      "epoch": 0.14752941176470588,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.7343,
      "step": 6270
    },
    {
      "epoch": 0.14755294117647058,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.9475,
      "step": 6271
    },
    {
      "epoch": 0.14757647058823528,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.387,
      "step": 6272
    },
    {
      "epoch": 0.1476,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0985,
      "step": 6273
    },
    {
      "epoch": 0.1476235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9705,
      "step": 6274
    },
    {
      "epoch": 0.1476470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4433,
      "step": 6275
    },
    {
      "epoch": 0.1476705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8691,
      "step": 6276
    },
    {
      "epoch": 0.14769411764705884,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1485,
      "step": 6277
    },
    {
      "epoch": 0.14771764705882354,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.137,
      "step": 6278
    },
    {
      "epoch": 0.14774117647058824,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0074,
      "step": 6279
    },
    {
      "epoch": 0.14776470588235294,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.1557,
      "step": 6280
    },
    {
      "epoch": 0.14778823529411764,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1639,
      "step": 6281
    },
    {
      "epoch": 0.14781176470588236,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3959,
      "step": 6282
    },
    {
      "epoch": 0.14783529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2194,
      "step": 6283
    },
    {
      "epoch": 0.14785882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1603,
      "step": 6284
    },
    {
      "epoch": 0.14788235294117646,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1364,
      "step": 6285
    },
    {
      "epoch": 0.14790588235294116,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0263,
      "step": 6286
    },
    {
      "epoch": 0.1479294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 6287
    },
    {
      "epoch": 0.1479529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.4946,
      "step": 6288
    },
    {
      "epoch": 0.1479764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0307,
      "step": 6289
    },
    {
      "epoch": 0.148,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 6290
    },
    {
      "epoch": 0.1480235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1853,
      "step": 6291
    },
    {
      "epoch": 0.14804705882352942,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1021,
      "step": 6292
    },
    {
      "epoch": 0.14807058823529412,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0739,
      "step": 6293
    },
    {
      "epoch": 0.14809411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1273,
      "step": 6294
    },
    {
      "epoch": 0.14811764705882352,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9413,
      "step": 6295
    },
    {
      "epoch": 0.14814117647058825,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.374,
      "step": 6296
    },
    {
      "epoch": 0.14816470588235295,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.5013,
      "step": 6297
    },
    {
      "epoch": 0.14818823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1092,
      "step": 6298
    },
    {
      "epoch": 0.14821176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2311,
      "step": 6299
    },
    {
      "epoch": 0.14823529411764705,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2842,
      "step": 6300
    },
    {
      "epoch": 0.14825882352941178,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2574,
      "step": 6301
    },
    {
      "epoch": 0.14828235294117648,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1606,
      "step": 6302
    },
    {
      "epoch": 0.14830588235294118,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.399,
      "step": 6303
    },
    {
      "epoch": 0.14832941176470588,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2428,
      "step": 6304
    },
    {
      "epoch": 0.14835294117647058,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4689,
      "step": 6305
    },
    {
      "epoch": 0.1483764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3203,
      "step": 6306
    },
    {
      "epoch": 0.1484,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3386,
      "step": 6307
    },
    {
      "epoch": 0.1484235294117647,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 6308
    },
    {
      "epoch": 0.1484470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4964,
      "step": 6309
    },
    {
      "epoch": 0.1484705882352941,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3878,
      "step": 6310
    },
    {
      "epoch": 0.14849411764705883,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9871,
      "step": 6311
    },
    {
      "epoch": 0.14851764705882353,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9144,
      "step": 6312
    },
    {
      "epoch": 0.14854117647058823,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.351,
      "step": 6313
    },
    {
      "epoch": 0.14856470588235293,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.295,
      "step": 6314
    },
    {
      "epoch": 0.14858823529411766,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 6315
    },
    {
      "epoch": 0.14861176470588236,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2825,
      "step": 6316
    },
    {
      "epoch": 0.14863529411764706,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.7013,
      "step": 6317
    },
    {
      "epoch": 0.14865882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 6318
    },
    {
      "epoch": 0.14868235294117646,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.9408,
      "step": 6319
    },
    {
      "epoch": 0.1487058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1246,
      "step": 6320
    },
    {
      "epoch": 0.1487294117647059,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9901,
      "step": 6321
    },
    {
      "epoch": 0.1487529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1904,
      "step": 6322
    },
    {
      "epoch": 0.1487764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.9664,
      "step": 6323
    },
    {
      "epoch": 0.1488,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2761,
      "step": 6324
    },
    {
      "epoch": 0.14882352941176472,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0955,
      "step": 6325
    },
    {
      "epoch": 0.14884705882352942,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2318,
      "step": 6326
    },
    {
      "epoch": 0.14887058823529412,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2094,
      "step": 6327
    },
    {
      "epoch": 0.14889411764705882,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2565,
      "step": 6328
    },
    {
      "epoch": 0.14891764705882352,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2914,
      "step": 6329
    },
    {
      "epoch": 0.14894117647058824,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.6143,
      "step": 6330
    },
    {
      "epoch": 0.14896470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0632,
      "step": 6331
    },
    {
      "epoch": 0.14898823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2155,
      "step": 6332
    },
    {
      "epoch": 0.14901176470588234,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.7396,
      "step": 6333
    },
    {
      "epoch": 0.14903529411764707,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.284,
      "step": 6334
    },
    {
      "epoch": 0.14905882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3136,
      "step": 6335
    },
    {
      "epoch": 0.14908235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2573,
      "step": 6336
    },
    {
      "epoch": 0.14910588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.295,
      "step": 6337
    },
    {
      "epoch": 0.14912941176470587,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9919,
      "step": 6338
    },
    {
      "epoch": 0.1491529411764706,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9455,
      "step": 6339
    },
    {
      "epoch": 0.1491764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2752,
      "step": 6340
    },
    {
      "epoch": 0.1492,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.8838,
      "step": 6341
    },
    {
      "epoch": 0.1492235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0556,
      "step": 6342
    },
    {
      "epoch": 0.1492470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3072,
      "step": 6343
    },
    {
      "epoch": 0.14927058823529413,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1902,
      "step": 6344
    },
    {
      "epoch": 0.14929411764705883,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.211,
      "step": 6345
    },
    {
      "epoch": 0.14931764705882353,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.6828,
      "step": 6346
    },
    {
      "epoch": 0.14934117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4312,
      "step": 6347
    },
    {
      "epoch": 0.14936470588235295,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9873,
      "step": 6348
    },
    {
      "epoch": 0.14938823529411766,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1047,
      "step": 6349
    },
    {
      "epoch": 0.14941176470588236,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2936,
      "step": 6350
    },
    {
      "epoch": 0.14943529411764706,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9469,
      "step": 6351
    },
    {
      "epoch": 0.14945882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3606,
      "step": 6352
    },
    {
      "epoch": 0.14948235294117648,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0087,
      "step": 6353
    },
    {
      "epoch": 0.14950588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1239,
      "step": 6354
    },
    {
      "epoch": 0.14952941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1759,
      "step": 6355
    },
    {
      "epoch": 0.14955294117647058,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.219,
      "step": 6356
    },
    {
      "epoch": 0.14957647058823528,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.4283,
      "step": 6357
    },
    {
      "epoch": 0.1496,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1713,
      "step": 6358
    },
    {
      "epoch": 0.1496235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2364,
      "step": 6359
    },
    {
      "epoch": 0.1496470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.26,
      "step": 6360
    },
    {
      "epoch": 0.1496705882352941,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1742,
      "step": 6361
    },
    {
      "epoch": 0.1496941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.237,
      "step": 6362
    },
    {
      "epoch": 0.14971764705882354,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 0.7401,
      "step": 6363
    },
    {
      "epoch": 0.14974117647058824,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0022,
      "step": 6364
    },
    {
      "epoch": 0.14976470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1388,
      "step": 6365
    },
    {
      "epoch": 0.14978823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4466,
      "step": 6366
    },
    {
      "epoch": 0.14981176470588237,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.1262,
      "step": 6367
    },
    {
      "epoch": 0.14983529411764707,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1447,
      "step": 6368
    },
    {
      "epoch": 0.14985882352941177,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9908,
      "step": 6369
    },
    {
      "epoch": 0.14988235294117647,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.5918,
      "step": 6370
    },
    {
      "epoch": 0.14990588235294117,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2042,
      "step": 6371
    },
    {
      "epoch": 0.1499294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4266,
      "step": 6372
    },
    {
      "epoch": 0.1499529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2716,
      "step": 6373
    },
    {
      "epoch": 0.1499764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1638,
      "step": 6374
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4797,
      "step": 6375
    },
    {
      "epoch": 0.1500235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3379,
      "step": 6376
    },
    {
      "epoch": 0.15004705882352942,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 6377
    },
    {
      "epoch": 0.15007058823529412,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9346,
      "step": 6378
    },
    {
      "epoch": 0.15009411764705882,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.8034,
      "step": 6379
    },
    {
      "epoch": 0.15011764705882352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4213,
      "step": 6380
    },
    {
      "epoch": 0.15014117647058822,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2796,
      "step": 6381
    },
    {
      "epoch": 0.15016470588235295,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1301,
      "step": 6382
    },
    {
      "epoch": 0.15018823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2609,
      "step": 6383
    },
    {
      "epoch": 0.15021176470588235,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8219,
      "step": 6384
    },
    {
      "epoch": 0.15023529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0454,
      "step": 6385
    },
    {
      "epoch": 0.15025882352941178,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9607,
      "step": 6386
    },
    {
      "epoch": 0.15028235294117648,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2643,
      "step": 6387
    },
    {
      "epoch": 0.15030588235294118,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0974,
      "step": 6388
    },
    {
      "epoch": 0.15032941176470588,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2098,
      "step": 6389
    },
    {
      "epoch": 0.15035294117647058,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1228,
      "step": 6390
    },
    {
      "epoch": 0.1503764705882353,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 1.0929,
      "step": 6391
    },
    {
      "epoch": 0.1504,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2775,
      "step": 6392
    },
    {
      "epoch": 0.1504235294117647,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 1.016,
      "step": 6393
    },
    {
      "epoch": 0.1504470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3783,
      "step": 6394
    },
    {
      "epoch": 0.1504705882352941,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1624,
      "step": 6395
    },
    {
      "epoch": 0.15049411764705883,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0345,
      "step": 6396
    },
    {
      "epoch": 0.15051764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3568,
      "step": 6397
    },
    {
      "epoch": 0.15054117647058823,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2861,
      "step": 6398
    },
    {
      "epoch": 0.15056470588235293,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0691,
      "step": 6399
    },
    {
      "epoch": 0.15058823529411763,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2533,
      "step": 6400
    },
    {
      "epoch": 0.15061176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4907,
      "step": 6401
    },
    {
      "epoch": 0.15063529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2907,
      "step": 6402
    },
    {
      "epoch": 0.15065882352941176,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0732,
      "step": 6403
    },
    {
      "epoch": 0.15068235294117646,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0629,
      "step": 6404
    },
    {
      "epoch": 0.1507058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2391,
      "step": 6405
    },
    {
      "epoch": 0.1507294117647059,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2772,
      "step": 6406
    },
    {
      "epoch": 0.1507529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 6407
    },
    {
      "epoch": 0.1507764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3852,
      "step": 6408
    },
    {
      "epoch": 0.1508,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1549,
      "step": 6409
    },
    {
      "epoch": 0.15082352941176472,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2513,
      "step": 6410
    },
    {
      "epoch": 0.15084705882352942,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2723,
      "step": 6411
    },
    {
      "epoch": 0.15087058823529412,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9707,
      "step": 6412
    },
    {
      "epoch": 0.15089411764705882,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.102,
      "step": 6413
    },
    {
      "epoch": 0.15091764705882352,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9784,
      "step": 6414
    },
    {
      "epoch": 0.15094117647058825,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 6415
    },
    {
      "epoch": 0.15096470588235295,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3353,
      "step": 6416
    },
    {
      "epoch": 0.15098823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4344,
      "step": 6417
    },
    {
      "epoch": 0.15101176470588235,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 6418
    },
    {
      "epoch": 0.15103529411764705,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.084,
      "step": 6419
    },
    {
      "epoch": 0.15105882352941177,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3593,
      "step": 6420
    },
    {
      "epoch": 0.15108235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1742,
      "step": 6421
    },
    {
      "epoch": 0.15110588235294117,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3645,
      "step": 6422
    },
    {
      "epoch": 0.15112941176470587,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.898,
      "step": 6423
    },
    {
      "epoch": 0.1511529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3725,
      "step": 6424
    },
    {
      "epoch": 0.1511764705882353,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.171,
      "step": 6425
    },
    {
      "epoch": 0.1512,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.885,
      "step": 6426
    },
    {
      "epoch": 0.1512235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4877,
      "step": 6427
    },
    {
      "epoch": 0.1512470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0154,
      "step": 6428
    },
    {
      "epoch": 0.15127058823529413,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.2466,
      "step": 6429
    },
    {
      "epoch": 0.15129411764705883,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0591,
      "step": 6430
    },
    {
      "epoch": 0.15131764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8697,
      "step": 6431
    },
    {
      "epoch": 0.15134117647058823,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2493,
      "step": 6432
    },
    {
      "epoch": 0.15136470588235293,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3146,
      "step": 6433
    },
    {
      "epoch": 0.15138823529411766,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1814,
      "step": 6434
    },
    {
      "epoch": 0.15141176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2511,
      "step": 6435
    },
    {
      "epoch": 0.15143529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3585,
      "step": 6436
    },
    {
      "epoch": 0.15145882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1617,
      "step": 6437
    },
    {
      "epoch": 0.15148235294117646,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4076,
      "step": 6438
    },
    {
      "epoch": 0.15150588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.4349,
      "step": 6439
    },
    {
      "epoch": 0.15152941176470588,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2537,
      "step": 6440
    },
    {
      "epoch": 0.15155294117647058,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0762,
      "step": 6441
    },
    {
      "epoch": 0.15157647058823528,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8812,
      "step": 6442
    },
    {
      "epoch": 0.1516,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1946,
      "step": 6443
    },
    {
      "epoch": 0.1516235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1936,
      "step": 6444
    },
    {
      "epoch": 0.1516470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4556,
      "step": 6445
    },
    {
      "epoch": 0.1516705882352941,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0163,
      "step": 6446
    },
    {
      "epoch": 0.1516941176470588,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 6447
    },
    {
      "epoch": 0.15171764705882354,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8842,
      "step": 6448
    },
    {
      "epoch": 0.15174117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3534,
      "step": 6449
    },
    {
      "epoch": 0.15176470588235294,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9097,
      "step": 6450
    },
    {
      "epoch": 0.15178823529411764,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.08,
      "step": 6451
    },
    {
      "epoch": 0.15181176470588234,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1221,
      "step": 6452
    },
    {
      "epoch": 0.15183529411764707,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.9074,
      "step": 6453
    },
    {
      "epoch": 0.15185882352941177,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9496,
      "step": 6454
    },
    {
      "epoch": 0.15188235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3469,
      "step": 6455
    },
    {
      "epoch": 0.15190588235294117,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0947,
      "step": 6456
    },
    {
      "epoch": 0.1519294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2935,
      "step": 6457
    },
    {
      "epoch": 0.1519529411764706,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9667,
      "step": 6458
    },
    {
      "epoch": 0.1519764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2827,
      "step": 6459
    },
    {
      "epoch": 0.152,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2417,
      "step": 6460
    },
    {
      "epoch": 0.1520235294117647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1877,
      "step": 6461
    },
    {
      "epoch": 0.15204705882352942,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2101,
      "step": 6462
    },
    {
      "epoch": 0.15207058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4218,
      "step": 6463
    },
    {
      "epoch": 0.15209411764705882,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9357,
      "step": 6464
    },
    {
      "epoch": 0.15211764705882352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3265,
      "step": 6465
    },
    {
      "epoch": 0.15214117647058822,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.207,
      "step": 6466
    },
    {
      "epoch": 0.15216470588235295,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.8173,
      "step": 6467
    },
    {
      "epoch": 0.15218823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2147,
      "step": 6468
    },
    {
      "epoch": 0.15221176470588235,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1307,
      "step": 6469
    },
    {
      "epoch": 0.15223529411764705,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9866,
      "step": 6470
    },
    {
      "epoch": 0.15225882352941175,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2178,
      "step": 6471
    },
    {
      "epoch": 0.15228235294117648,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1974,
      "step": 6472
    },
    {
      "epoch": 0.15230588235294118,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.119,
      "step": 6473
    },
    {
      "epoch": 0.15232941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4574,
      "step": 6474
    },
    {
      "epoch": 0.15235294117647058,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.9053,
      "step": 6475
    },
    {
      "epoch": 0.1523764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1354,
      "step": 6476
    },
    {
      "epoch": 0.1524,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.237,
      "step": 6477
    },
    {
      "epoch": 0.1524235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3424,
      "step": 6478
    },
    {
      "epoch": 0.1524470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2147,
      "step": 6479
    },
    {
      "epoch": 0.1524705882352941,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9131,
      "step": 6480
    },
    {
      "epoch": 0.15249411764705884,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0456,
      "step": 6481
    },
    {
      "epoch": 0.15251764705882354,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 6482
    },
    {
      "epoch": 0.15254117647058824,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.3574,
      "step": 6483
    },
    {
      "epoch": 0.15256470588235294,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2836,
      "step": 6484
    },
    {
      "epoch": 0.15258823529411764,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 6485
    },
    {
      "epoch": 0.15261176470588236,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2085,
      "step": 6486
    },
    {
      "epoch": 0.15263529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9122,
      "step": 6487
    },
    {
      "epoch": 0.15265882352941176,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9266,
      "step": 6488
    },
    {
      "epoch": 0.15268235294117646,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3598,
      "step": 6489
    },
    {
      "epoch": 0.15270588235294116,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1897,
      "step": 6490
    },
    {
      "epoch": 0.1527294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1446,
      "step": 6491
    },
    {
      "epoch": 0.1527529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.263,
      "step": 6492
    },
    {
      "epoch": 0.1527764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1603,
      "step": 6493
    },
    {
      "epoch": 0.1528,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2004,
      "step": 6494
    },
    {
      "epoch": 0.15282352941176472,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9181,
      "step": 6495
    },
    {
      "epoch": 0.15284705882352942,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9563,
      "step": 6496
    },
    {
      "epoch": 0.15287058823529412,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2554,
      "step": 6497
    },
    {
      "epoch": 0.15289411764705882,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0245,
      "step": 6498
    },
    {
      "epoch": 0.15291764705882352,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3338,
      "step": 6499
    },
    {
      "epoch": 0.15294117647058825,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0738,
      "step": 6500
    },
    {
      "epoch": 0.15296470588235295,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2909,
      "step": 6501
    },
    {
      "epoch": 0.15298823529411765,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1166,
      "step": 6502
    },
    {
      "epoch": 0.15301176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2969,
      "step": 6503
    },
    {
      "epoch": 0.15303529411764705,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.804,
      "step": 6504
    },
    {
      "epoch": 0.15305882352941177,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9439,
      "step": 6505
    },
    {
      "epoch": 0.15308235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4003,
      "step": 6506
    },
    {
      "epoch": 0.15310588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4535,
      "step": 6507
    },
    {
      "epoch": 0.15312941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9711,
      "step": 6508
    },
    {
      "epoch": 0.15315294117647058,
      "grad_norm": 0.302734375,
      "learning_rate": 0.02,
      "loss": 0.9065,
      "step": 6509
    },
    {
      "epoch": 0.1531764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.318,
      "step": 6510
    },
    {
      "epoch": 0.1532,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3432,
      "step": 6511
    },
    {
      "epoch": 0.1532235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3437,
      "step": 6512
    },
    {
      "epoch": 0.1532470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2476,
      "step": 6513
    },
    {
      "epoch": 0.15327058823529413,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2449,
      "step": 6514
    },
    {
      "epoch": 0.15329411764705883,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 6515
    },
    {
      "epoch": 0.15331764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3273,
      "step": 6516
    },
    {
      "epoch": 0.15334117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1972,
      "step": 6517
    },
    {
      "epoch": 0.15336470588235293,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2887,
      "step": 6518
    },
    {
      "epoch": 0.15338823529411766,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0685,
      "step": 6519
    },
    {
      "epoch": 0.15341176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.286,
      "step": 6520
    },
    {
      "epoch": 0.15343529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2357,
      "step": 6521
    },
    {
      "epoch": 0.15345882352941176,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 6522
    },
    {
      "epoch": 0.15348235294117646,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0602,
      "step": 6523
    },
    {
      "epoch": 0.1535058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0542,
      "step": 6524
    },
    {
      "epoch": 0.1535294117647059,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1746,
      "step": 6525
    },
    {
      "epoch": 0.1535529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 6526
    },
    {
      "epoch": 0.1535764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0766,
      "step": 6527
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1279,
      "step": 6528
    },
    {
      "epoch": 0.15362352941176471,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.077,
      "step": 6529
    },
    {
      "epoch": 0.15364705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3445,
      "step": 6530
    },
    {
      "epoch": 0.15367058823529411,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3008,
      "step": 6531
    },
    {
      "epoch": 0.15369411764705881,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0738,
      "step": 6532
    },
    {
      "epoch": 0.15371764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1286,
      "step": 6533
    },
    {
      "epoch": 0.15374117647058824,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1922,
      "step": 6534
    },
    {
      "epoch": 0.15376470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1821,
      "step": 6535
    },
    {
      "epoch": 0.15378823529411764,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.333,
      "step": 6536
    },
    {
      "epoch": 0.15381176470588234,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1739,
      "step": 6537
    },
    {
      "epoch": 0.15383529411764707,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.246,
      "step": 6538
    },
    {
      "epoch": 0.15385882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2934,
      "step": 6539
    },
    {
      "epoch": 0.15388235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3574,
      "step": 6540
    },
    {
      "epoch": 0.15390588235294117,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0625,
      "step": 6541
    },
    {
      "epoch": 0.15392941176470587,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9866,
      "step": 6542
    },
    {
      "epoch": 0.1539529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3887,
      "step": 6543
    },
    {
      "epoch": 0.1539764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9068,
      "step": 6544
    },
    {
      "epoch": 0.154,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0817,
      "step": 6545
    },
    {
      "epoch": 0.1540235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4021,
      "step": 6546
    },
    {
      "epoch": 0.1540470588235294,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1202,
      "step": 6547
    },
    {
      "epoch": 0.15407058823529413,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 6548
    },
    {
      "epoch": 0.15409411764705883,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 6549
    },
    {
      "epoch": 0.15411764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2774,
      "step": 6550
    },
    {
      "epoch": 0.15414117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.068,
      "step": 6551
    },
    {
      "epoch": 0.15416470588235295,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3023,
      "step": 6552
    },
    {
      "epoch": 0.15418823529411765,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1442,
      "step": 6553
    },
    {
      "epoch": 0.15421176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2145,
      "step": 6554
    },
    {
      "epoch": 0.15423529411764705,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9588,
      "step": 6555
    },
    {
      "epoch": 0.15425882352941175,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2586,
      "step": 6556
    },
    {
      "epoch": 0.15428235294117648,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8996,
      "step": 6557
    },
    {
      "epoch": 0.15430588235294118,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.974,
      "step": 6558
    },
    {
      "epoch": 0.15432941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 6559
    },
    {
      "epoch": 0.15435294117647058,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2979,
      "step": 6560
    },
    {
      "epoch": 0.15437647058823528,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.088,
      "step": 6561
    },
    {
      "epoch": 0.1544,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9334,
      "step": 6562
    },
    {
      "epoch": 0.1544235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2422,
      "step": 6563
    },
    {
      "epoch": 0.1544470588235294,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1117,
      "step": 6564
    },
    {
      "epoch": 0.1544705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3434,
      "step": 6565
    },
    {
      "epoch": 0.15449411764705884,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9089,
      "step": 6566
    },
    {
      "epoch": 0.15451764705882354,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3121,
      "step": 6567
    },
    {
      "epoch": 0.15454117647058824,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9113,
      "step": 6568
    },
    {
      "epoch": 0.15456470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2311,
      "step": 6569
    },
    {
      "epoch": 0.15458823529411764,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.279,
      "step": 6570
    },
    {
      "epoch": 0.15461176470588237,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2219,
      "step": 6571
    },
    {
      "epoch": 0.15463529411764707,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0584,
      "step": 6572
    },
    {
      "epoch": 0.15465882352941177,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1904,
      "step": 6573
    },
    {
      "epoch": 0.15468235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.321,
      "step": 6574
    },
    {
      "epoch": 0.15470588235294117,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8298,
      "step": 6575
    },
    {
      "epoch": 0.1547294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2609,
      "step": 6576
    },
    {
      "epoch": 0.1547529411764706,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9099,
      "step": 6577
    },
    {
      "epoch": 0.1547764705882353,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1726,
      "step": 6578
    },
    {
      "epoch": 0.1548,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2171,
      "step": 6579
    },
    {
      "epoch": 0.1548235294117647,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.3379,
      "step": 6580
    },
    {
      "epoch": 0.15484705882352942,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9058,
      "step": 6581
    },
    {
      "epoch": 0.15487058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2383,
      "step": 6582
    },
    {
      "epoch": 0.15489411764705882,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 6583
    },
    {
      "epoch": 0.15491764705882352,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0572,
      "step": 6584
    },
    {
      "epoch": 0.15494117647058825,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2187,
      "step": 6585
    },
    {
      "epoch": 0.15496470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4573,
      "step": 6586
    },
    {
      "epoch": 0.15498823529411765,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9413,
      "step": 6587
    },
    {
      "epoch": 0.15501176470588235,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1159,
      "step": 6588
    },
    {
      "epoch": 0.15503529411764705,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1723,
      "step": 6589
    },
    {
      "epoch": 0.15505882352941178,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2682,
      "step": 6590
    },
    {
      "epoch": 0.15508235294117648,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.7345,
      "step": 6591
    },
    {
      "epoch": 0.15510588235294118,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1835,
      "step": 6592
    },
    {
      "epoch": 0.15512941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4889,
      "step": 6593
    },
    {
      "epoch": 0.15515294117647058,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3346,
      "step": 6594
    },
    {
      "epoch": 0.1551764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1686,
      "step": 6595
    },
    {
      "epoch": 0.1552,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.4405,
      "step": 6596
    },
    {
      "epoch": 0.1552235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4471,
      "step": 6597
    },
    {
      "epoch": 0.1552470588235294,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2613,
      "step": 6598
    },
    {
      "epoch": 0.1552705882352941,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1336,
      "step": 6599
    },
    {
      "epoch": 0.15529411764705883,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1757,
      "step": 6600
    },
    {
      "epoch": 0.15531764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2537,
      "step": 6601
    },
    {
      "epoch": 0.15534117647058823,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8881,
      "step": 6602
    },
    {
      "epoch": 0.15536470588235293,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1788,
      "step": 6603
    },
    {
      "epoch": 0.15538823529411766,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2875,
      "step": 6604
    },
    {
      "epoch": 0.15541176470588236,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.6426,
      "step": 6605
    },
    {
      "epoch": 0.15543529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9154,
      "step": 6606
    },
    {
      "epoch": 0.15545882352941176,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 6607
    },
    {
      "epoch": 0.15548235294117646,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0627,
      "step": 6608
    },
    {
      "epoch": 0.1555058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1206,
      "step": 6609
    },
    {
      "epoch": 0.1555294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3458,
      "step": 6610
    },
    {
      "epoch": 0.1555529411764706,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 6611
    },
    {
      "epoch": 0.1555764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.8872,
      "step": 6612
    },
    {
      "epoch": 0.1556,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.035,
      "step": 6613
    },
    {
      "epoch": 0.15562352941176472,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3583,
      "step": 6614
    },
    {
      "epoch": 0.15564705882352942,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 6615
    },
    {
      "epoch": 0.15567058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3076,
      "step": 6616
    },
    {
      "epoch": 0.15569411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0996,
      "step": 6617
    },
    {
      "epoch": 0.15571764705882352,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.4651,
      "step": 6618
    },
    {
      "epoch": 0.15574117647058824,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 6619
    },
    {
      "epoch": 0.15576470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4143,
      "step": 6620
    },
    {
      "epoch": 0.15578823529411764,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1975,
      "step": 6621
    },
    {
      "epoch": 0.15581176470588234,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3747,
      "step": 6622
    },
    {
      "epoch": 0.15583529411764707,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3834,
      "step": 6623
    },
    {
      "epoch": 0.15585882352941177,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3364,
      "step": 6624
    },
    {
      "epoch": 0.15588235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1494,
      "step": 6625
    },
    {
      "epoch": 0.15590588235294117,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2206,
      "step": 6626
    },
    {
      "epoch": 0.15592941176470587,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0727,
      "step": 6627
    },
    {
      "epoch": 0.1559529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.9227,
      "step": 6628
    },
    {
      "epoch": 0.1559764705882353,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 1.0449,
      "step": 6629
    },
    {
      "epoch": 0.156,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1209,
      "step": 6630
    },
    {
      "epoch": 0.1560235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3087,
      "step": 6631
    },
    {
      "epoch": 0.1560470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2142,
      "step": 6632
    },
    {
      "epoch": 0.15607058823529413,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8028,
      "step": 6633
    },
    {
      "epoch": 0.15609411764705883,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9463,
      "step": 6634
    },
    {
      "epoch": 0.15611764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1492,
      "step": 6635
    },
    {
      "epoch": 0.15614117647058823,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9201,
      "step": 6636
    },
    {
      "epoch": 0.15616470588235293,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.5295,
      "step": 6637
    },
    {
      "epoch": 0.15618823529411766,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.304,
      "step": 6638
    },
    {
      "epoch": 0.15621176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.3347,
      "step": 6639
    },
    {
      "epoch": 0.15623529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2571,
      "step": 6640
    },
    {
      "epoch": 0.15625882352941176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2023,
      "step": 6641
    },
    {
      "epoch": 0.15628235294117648,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1413,
      "step": 6642
    },
    {
      "epoch": 0.15630588235294118,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0525,
      "step": 6643
    },
    {
      "epoch": 0.15632941176470588,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1478,
      "step": 6644
    },
    {
      "epoch": 0.15635294117647058,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.055,
      "step": 6645
    },
    {
      "epoch": 0.15637647058823528,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1737,
      "step": 6646
    },
    {
      "epoch": 0.1564,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.251,
      "step": 6647
    },
    {
      "epoch": 0.1564235294117647,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0268,
      "step": 6648
    },
    {
      "epoch": 0.1564470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.0053,
      "step": 6649
    },
    {
      "epoch": 0.1564705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3649,
      "step": 6650
    },
    {
      "epoch": 0.1564941176470588,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1379,
      "step": 6651
    },
    {
      "epoch": 0.15651764705882354,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9883,
      "step": 6652
    },
    {
      "epoch": 0.15654117647058824,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1182,
      "step": 6653
    },
    {
      "epoch": 0.15656470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3976,
      "step": 6654
    },
    {
      "epoch": 0.15658823529411764,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1181,
      "step": 6655
    },
    {
      "epoch": 0.15661176470588234,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.292,
      "step": 6656
    },
    {
      "epoch": 0.15663529411764707,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1699,
      "step": 6657
    },
    {
      "epoch": 0.15665882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 6658
    },
    {
      "epoch": 0.15668235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0456,
      "step": 6659
    },
    {
      "epoch": 0.15670588235294117,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3124,
      "step": 6660
    },
    {
      "epoch": 0.1567294117647059,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1448,
      "step": 6661
    },
    {
      "epoch": 0.1567529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2222,
      "step": 6662
    },
    {
      "epoch": 0.1567764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0035,
      "step": 6663
    },
    {
      "epoch": 0.1568,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3377,
      "step": 6664
    },
    {
      "epoch": 0.1568235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0088,
      "step": 6665
    },
    {
      "epoch": 0.15684705882352942,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2907,
      "step": 6666
    },
    {
      "epoch": 0.15687058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3158,
      "step": 6667
    },
    {
      "epoch": 0.15689411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2061,
      "step": 6668
    },
    {
      "epoch": 0.15691764705882352,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3179,
      "step": 6669
    },
    {
      "epoch": 0.15694117647058822,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3624,
      "step": 6670
    },
    {
      "epoch": 0.15696470588235295,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9948,
      "step": 6671
    },
    {
      "epoch": 0.15698823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2831,
      "step": 6672
    },
    {
      "epoch": 0.15701176470588235,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1809,
      "step": 6673
    },
    {
      "epoch": 0.15703529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3031,
      "step": 6674
    },
    {
      "epoch": 0.15705882352941178,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1512,
      "step": 6675
    },
    {
      "epoch": 0.15708235294117648,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.206,
      "step": 6676
    },
    {
      "epoch": 0.15710588235294118,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2827,
      "step": 6677
    },
    {
      "epoch": 0.15712941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 6678
    },
    {
      "epoch": 0.15715294117647058,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1973,
      "step": 6679
    },
    {
      "epoch": 0.1571764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8947,
      "step": 6680
    },
    {
      "epoch": 0.1572,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2653,
      "step": 6681
    },
    {
      "epoch": 0.1572235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9976,
      "step": 6682
    },
    {
      "epoch": 0.1572470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.3685,
      "step": 6683
    },
    {
      "epoch": 0.1572705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3078,
      "step": 6684
    },
    {
      "epoch": 0.15729411764705883,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1501,
      "step": 6685
    },
    {
      "epoch": 0.15731764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1962,
      "step": 6686
    },
    {
      "epoch": 0.15734117647058823,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.2391,
      "step": 6687
    },
    {
      "epoch": 0.15736470588235293,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9742,
      "step": 6688
    },
    {
      "epoch": 0.15738823529411763,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0917,
      "step": 6689
    },
    {
      "epoch": 0.15741176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1902,
      "step": 6690
    },
    {
      "epoch": 0.15743529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3805,
      "step": 6691
    },
    {
      "epoch": 0.15745882352941176,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0433,
      "step": 6692
    },
    {
      "epoch": 0.15748235294117646,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.002,
      "step": 6693
    },
    {
      "epoch": 0.1575058823529412,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9166,
      "step": 6694
    },
    {
      "epoch": 0.1575294117647059,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2482,
      "step": 6695
    },
    {
      "epoch": 0.1575529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3956,
      "step": 6696
    },
    {
      "epoch": 0.1575764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1924,
      "step": 6697
    },
    {
      "epoch": 0.1576,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3954,
      "step": 6698
    },
    {
      "epoch": 0.15762352941176472,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2466,
      "step": 6699
    },
    {
      "epoch": 0.15764705882352942,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9835,
      "step": 6700
    },
    {
      "epoch": 0.15767058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3232,
      "step": 6701
    },
    {
      "epoch": 0.15769411764705882,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2114,
      "step": 6702
    },
    {
      "epoch": 0.15771764705882352,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9919,
      "step": 6703
    },
    {
      "epoch": 0.15774117647058825,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2374,
      "step": 6704
    },
    {
      "epoch": 0.15776470588235295,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3663,
      "step": 6705
    },
    {
      "epoch": 0.15778823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1805,
      "step": 6706
    },
    {
      "epoch": 0.15781176470588235,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1791,
      "step": 6707
    },
    {
      "epoch": 0.15783529411764705,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1346,
      "step": 6708
    },
    {
      "epoch": 0.15785882352941177,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3411,
      "step": 6709
    },
    {
      "epoch": 0.15788235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0253,
      "step": 6710
    },
    {
      "epoch": 0.15790588235294117,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2239,
      "step": 6711
    },
    {
      "epoch": 0.15792941176470587,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2859,
      "step": 6712
    },
    {
      "epoch": 0.1579529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.198,
      "step": 6713
    },
    {
      "epoch": 0.1579764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2353,
      "step": 6714
    },
    {
      "epoch": 0.158,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3663,
      "step": 6715
    },
    {
      "epoch": 0.1580235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.309,
      "step": 6716
    },
    {
      "epoch": 0.1580470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9659,
      "step": 6717
    },
    {
      "epoch": 0.15807058823529413,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1544,
      "step": 6718
    },
    {
      "epoch": 0.15809411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4318,
      "step": 6719
    },
    {
      "epoch": 0.15811764705882353,
      "grad_norm": 0.298828125,
      "learning_rate": 0.02,
      "loss": 0.8909,
      "step": 6720
    },
    {
      "epoch": 0.15814117647058823,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1711,
      "step": 6721
    },
    {
      "epoch": 0.15816470588235293,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0805,
      "step": 6722
    },
    {
      "epoch": 0.15818823529411766,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1525,
      "step": 6723
    },
    {
      "epoch": 0.15821176470588236,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.5135,
      "step": 6724
    },
    {
      "epoch": 0.15823529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 6725
    },
    {
      "epoch": 0.15825882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1213,
      "step": 6726
    },
    {
      "epoch": 0.15828235294117646,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0379,
      "step": 6727
    },
    {
      "epoch": 0.15830588235294119,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1422,
      "step": 6728
    },
    {
      "epoch": 0.15832941176470589,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0858,
      "step": 6729
    },
    {
      "epoch": 0.15835294117647059,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2259,
      "step": 6730
    },
    {
      "epoch": 0.15837647058823529,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1599,
      "step": 6731
    },
    {
      "epoch": 0.1584,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.835,
      "step": 6732
    },
    {
      "epoch": 0.1584235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2792,
      "step": 6733
    },
    {
      "epoch": 0.1584470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1915,
      "step": 6734
    },
    {
      "epoch": 0.1584705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0527,
      "step": 6735
    },
    {
      "epoch": 0.1584941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.961,
      "step": 6736
    },
    {
      "epoch": 0.15851764705882354,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 6737
    },
    {
      "epoch": 0.15854117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1955,
      "step": 6738
    },
    {
      "epoch": 0.15856470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1459,
      "step": 6739
    },
    {
      "epoch": 0.15858823529411764,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3568,
      "step": 6740
    },
    {
      "epoch": 0.15861176470588234,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2867,
      "step": 6741
    },
    {
      "epoch": 0.15863529411764707,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0826,
      "step": 6742
    },
    {
      "epoch": 0.15865882352941177,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0029,
      "step": 6743
    },
    {
      "epoch": 0.15868235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2887,
      "step": 6744
    },
    {
      "epoch": 0.15870588235294117,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9057,
      "step": 6745
    },
    {
      "epoch": 0.15872941176470587,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2181,
      "step": 6746
    },
    {
      "epoch": 0.1587529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2179,
      "step": 6747
    },
    {
      "epoch": 0.1587764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7322,
      "step": 6748
    },
    {
      "epoch": 0.1588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.3079,
      "step": 6749
    },
    {
      "epoch": 0.1588235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0574,
      "step": 6750
    },
    {
      "epoch": 0.15884705882352942,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.5714,
      "step": 6751
    },
    {
      "epoch": 0.15887058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4509,
      "step": 6752
    },
    {
      "epoch": 0.15889411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4116,
      "step": 6753
    },
    {
      "epoch": 0.15891764705882352,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3797,
      "step": 6754
    },
    {
      "epoch": 0.15894117647058822,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0919,
      "step": 6755
    },
    {
      "epoch": 0.15896470588235295,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 6756
    },
    {
      "epoch": 0.15898823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1685,
      "step": 6757
    },
    {
      "epoch": 0.15901176470588235,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0487,
      "step": 6758
    },
    {
      "epoch": 0.15903529411764705,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2013,
      "step": 6759
    },
    {
      "epoch": 0.15905882352941175,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.3113,
      "step": 6760
    },
    {
      "epoch": 0.15908235294117648,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0827,
      "step": 6761
    },
    {
      "epoch": 0.15910588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.4478,
      "step": 6762
    },
    {
      "epoch": 0.15912941176470588,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 6763
    },
    {
      "epoch": 0.15915294117647058,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.3212,
      "step": 6764
    },
    {
      "epoch": 0.15917647058823528,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2459,
      "step": 6765
    },
    {
      "epoch": 0.1592,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2789,
      "step": 6766
    },
    {
      "epoch": 0.1592235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1624,
      "step": 6767
    },
    {
      "epoch": 0.1592470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1541,
      "step": 6768
    },
    {
      "epoch": 0.1592705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0681,
      "step": 6769
    },
    {
      "epoch": 0.15929411764705884,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1628,
      "step": 6770
    },
    {
      "epoch": 0.15931764705882354,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1224,
      "step": 6771
    },
    {
      "epoch": 0.15934117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1863,
      "step": 6772
    },
    {
      "epoch": 0.15936470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1963,
      "step": 6773
    },
    {
      "epoch": 0.15938823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1692,
      "step": 6774
    },
    {
      "epoch": 0.15941176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0676,
      "step": 6775
    },
    {
      "epoch": 0.15943529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1637,
      "step": 6776
    },
    {
      "epoch": 0.15945882352941176,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0351,
      "step": 6777
    },
    {
      "epoch": 0.15948235294117646,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0467,
      "step": 6778
    },
    {
      "epoch": 0.15950588235294116,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0273,
      "step": 6779
    },
    {
      "epoch": 0.1595294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1564,
      "step": 6780
    },
    {
      "epoch": 0.1595529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2795,
      "step": 6781
    },
    {
      "epoch": 0.1595764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9694,
      "step": 6782
    },
    {
      "epoch": 0.1596,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.2389,
      "step": 6783
    },
    {
      "epoch": 0.15962352941176472,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0066,
      "step": 6784
    },
    {
      "epoch": 0.15964705882352942,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1985,
      "step": 6785
    },
    {
      "epoch": 0.15967058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4615,
      "step": 6786
    },
    {
      "epoch": 0.15969411764705882,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0387,
      "step": 6787
    },
    {
      "epoch": 0.15971764705882352,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0153,
      "step": 6788
    },
    {
      "epoch": 0.15974117647058825,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9095,
      "step": 6789
    },
    {
      "epoch": 0.15976470588235295,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.308,
      "step": 6790
    },
    {
      "epoch": 0.15978823529411765,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2829,
      "step": 6791
    },
    {
      "epoch": 0.15981176470588235,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0833,
      "step": 6792
    },
    {
      "epoch": 0.15983529411764705,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9381,
      "step": 6793
    },
    {
      "epoch": 0.15985882352941178,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9999,
      "step": 6794
    },
    {
      "epoch": 0.15988235294117648,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0092,
      "step": 6795
    },
    {
      "epoch": 0.15990588235294118,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1481,
      "step": 6796
    },
    {
      "epoch": 0.15992941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1051,
      "step": 6797
    },
    {
      "epoch": 0.15995294117647058,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.064,
      "step": 6798
    },
    {
      "epoch": 0.1599764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 6799
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4323,
      "step": 6800
    },
    {
      "epoch": 0.1600235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 6801
    },
    {
      "epoch": 0.1600470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1311,
      "step": 6802
    },
    {
      "epoch": 0.16007058823529413,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2904,
      "step": 6803
    },
    {
      "epoch": 0.16009411764705883,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.2149,
      "step": 6804
    },
    {
      "epoch": 0.16011764705882353,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9334,
      "step": 6805
    },
    {
      "epoch": 0.16014117647058823,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 6806
    },
    {
      "epoch": 0.16016470588235293,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2453,
      "step": 6807
    },
    {
      "epoch": 0.16018823529411766,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8196,
      "step": 6808
    },
    {
      "epoch": 0.16021176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1355,
      "step": 6809
    },
    {
      "epoch": 0.16023529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1044,
      "step": 6810
    },
    {
      "epoch": 0.16025882352941176,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0229,
      "step": 6811
    },
    {
      "epoch": 0.16028235294117646,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1028,
      "step": 6812
    },
    {
      "epoch": 0.1603058823529412,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.9048,
      "step": 6813
    },
    {
      "epoch": 0.1603294117647059,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2444,
      "step": 6814
    },
    {
      "epoch": 0.1603529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3918,
      "step": 6815
    },
    {
      "epoch": 0.1603764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1473,
      "step": 6816
    },
    {
      "epoch": 0.1604,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2856,
      "step": 6817
    },
    {
      "epoch": 0.16042352941176471,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.215,
      "step": 6818
    },
    {
      "epoch": 0.16044705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2519,
      "step": 6819
    },
    {
      "epoch": 0.16047058823529411,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3681,
      "step": 6820
    },
    {
      "epoch": 0.16049411764705881,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0584,
      "step": 6821
    },
    {
      "epoch": 0.16051764705882354,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1016,
      "step": 6822
    },
    {
      "epoch": 0.16054117647058824,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2026,
      "step": 6823
    },
    {
      "epoch": 0.16056470588235294,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8452,
      "step": 6824
    },
    {
      "epoch": 0.16058823529411764,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3715,
      "step": 6825
    },
    {
      "epoch": 0.16061176470588234,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2273,
      "step": 6826
    },
    {
      "epoch": 0.16063529411764707,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.916,
      "step": 6827
    },
    {
      "epoch": 0.16065882352941177,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2699,
      "step": 6828
    },
    {
      "epoch": 0.16068235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9824,
      "step": 6829
    },
    {
      "epoch": 0.16070588235294117,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.17,
      "step": 6830
    },
    {
      "epoch": 0.16072941176470587,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.799,
      "step": 6831
    },
    {
      "epoch": 0.1607529411764706,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8836,
      "step": 6832
    },
    {
      "epoch": 0.1607764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1621,
      "step": 6833
    },
    {
      "epoch": 0.1608,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9403,
      "step": 6834
    },
    {
      "epoch": 0.1608235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9337,
      "step": 6835
    },
    {
      "epoch": 0.1608470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 6836
    },
    {
      "epoch": 0.16087058823529413,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1259,
      "step": 6837
    },
    {
      "epoch": 0.16089411764705883,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4044,
      "step": 6838
    },
    {
      "epoch": 0.16091764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.2678,
      "step": 6839
    },
    {
      "epoch": 0.16094117647058823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.4302,
      "step": 6840
    },
    {
      "epoch": 0.16096470588235295,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3452,
      "step": 6841
    },
    {
      "epoch": 0.16098823529411765,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2041,
      "step": 6842
    },
    {
      "epoch": 0.16101176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3276,
      "step": 6843
    },
    {
      "epoch": 0.16103529411764705,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1034,
      "step": 6844
    },
    {
      "epoch": 0.16105882352941175,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3832,
      "step": 6845
    },
    {
      "epoch": 0.16108235294117648,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4847,
      "step": 6846
    },
    {
      "epoch": 0.16110588235294118,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 6847
    },
    {
      "epoch": 0.16112941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3695,
      "step": 6848
    },
    {
      "epoch": 0.16115294117647058,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.4533,
      "step": 6849
    },
    {
      "epoch": 0.16117647058823528,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.469,
      "step": 6850
    },
    {
      "epoch": 0.1612,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.0998,
      "step": 6851
    },
    {
      "epoch": 0.1612235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.274,
      "step": 6852
    },
    {
      "epoch": 0.1612470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2167,
      "step": 6853
    },
    {
      "epoch": 0.1612705882352941,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2941,
      "step": 6854
    },
    {
      "epoch": 0.1612941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2702,
      "step": 6855
    },
    {
      "epoch": 0.16131764705882354,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8625,
      "step": 6856
    },
    {
      "epoch": 0.16134117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 6857
    },
    {
      "epoch": 0.16136470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 6858
    },
    {
      "epoch": 0.16138823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0539,
      "step": 6859
    },
    {
      "epoch": 0.16141176470588237,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.3668,
      "step": 6860
    },
    {
      "epoch": 0.16143529411764707,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 6861
    },
    {
      "epoch": 0.16145882352941177,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.084,
      "step": 6862
    },
    {
      "epoch": 0.16148235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.1066,
      "step": 6863
    },
    {
      "epoch": 0.16150588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3312,
      "step": 6864
    },
    {
      "epoch": 0.1615294117647059,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9029,
      "step": 6865
    },
    {
      "epoch": 0.1615529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0976,
      "step": 6866
    },
    {
      "epoch": 0.1615764705882353,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8095,
      "step": 6867
    },
    {
      "epoch": 0.1616,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1983,
      "step": 6868
    },
    {
      "epoch": 0.1616235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3423,
      "step": 6869
    },
    {
      "epoch": 0.16164705882352942,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.856,
      "step": 6870
    },
    {
      "epoch": 0.16167058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 6871
    },
    {
      "epoch": 0.16169411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9936,
      "step": 6872
    },
    {
      "epoch": 0.16171764705882352,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3656,
      "step": 6873
    },
    {
      "epoch": 0.16174117647058822,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2138,
      "step": 6874
    },
    {
      "epoch": 0.16176470588235295,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1078,
      "step": 6875
    },
    {
      "epoch": 0.16178823529411765,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0561,
      "step": 6876
    },
    {
      "epoch": 0.16181176470588235,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9899,
      "step": 6877
    },
    {
      "epoch": 0.16183529411764705,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1463,
      "step": 6878
    },
    {
      "epoch": 0.16185882352941178,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.3272,
      "step": 6879
    },
    {
      "epoch": 0.16188235294117648,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9689,
      "step": 6880
    },
    {
      "epoch": 0.16190588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.3427,
      "step": 6881
    },
    {
      "epoch": 0.16192941176470588,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0128,
      "step": 6882
    },
    {
      "epoch": 0.16195294117647058,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9298,
      "step": 6883
    },
    {
      "epoch": 0.1619764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2573,
      "step": 6884
    },
    {
      "epoch": 0.162,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0072,
      "step": 6885
    },
    {
      "epoch": 0.1620235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1356,
      "step": 6886
    },
    {
      "epoch": 0.1620470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3614,
      "step": 6887
    },
    {
      "epoch": 0.1620705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3085,
      "step": 6888
    },
    {
      "epoch": 0.16209411764705883,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 6889
    },
    {
      "epoch": 0.16211764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4004,
      "step": 6890
    },
    {
      "epoch": 0.16214117647058823,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.922,
      "step": 6891
    },
    {
      "epoch": 0.16216470588235293,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4473,
      "step": 6892
    },
    {
      "epoch": 0.16218823529411766,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1071,
      "step": 6893
    },
    {
      "epoch": 0.16221176470588236,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1813,
      "step": 6894
    },
    {
      "epoch": 0.16223529411764706,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0949,
      "step": 6895
    },
    {
      "epoch": 0.16225882352941176,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0145,
      "step": 6896
    },
    {
      "epoch": 0.16228235294117646,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 6897
    },
    {
      "epoch": 0.1623058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1305,
      "step": 6898
    },
    {
      "epoch": 0.1623294117647059,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.106,
      "step": 6899
    },
    {
      "epoch": 0.1623529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.068,
      "step": 6900
    },
    {
      "epoch": 0.1623764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2726,
      "step": 6901
    },
    {
      "epoch": 0.1624,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0506,
      "step": 6902
    },
    {
      "epoch": 0.16242352941176472,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 6903
    },
    {
      "epoch": 0.16244705882352942,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.547,
      "step": 6904
    },
    {
      "epoch": 0.16247058823529412,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2309,
      "step": 6905
    },
    {
      "epoch": 0.16249411764705882,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.218,
      "step": 6906
    },
    {
      "epoch": 0.16251764705882352,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.083,
      "step": 6907
    },
    {
      "epoch": 0.16254117647058824,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.242,
      "step": 6908
    },
    {
      "epoch": 0.16256470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.4019,
      "step": 6909
    },
    {
      "epoch": 0.16258823529411764,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7976,
      "step": 6910
    },
    {
      "epoch": 0.16261176470588234,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3751,
      "step": 6911
    },
    {
      "epoch": 0.16263529411764707,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2847,
      "step": 6912
    },
    {
      "epoch": 0.16265882352941177,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0642,
      "step": 6913
    },
    {
      "epoch": 0.16268235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0693,
      "step": 6914
    },
    {
      "epoch": 0.16270588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1743,
      "step": 6915
    },
    {
      "epoch": 0.16272941176470587,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8942,
      "step": 6916
    },
    {
      "epoch": 0.1627529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2366,
      "step": 6917
    },
    {
      "epoch": 0.1627764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1716,
      "step": 6918
    },
    {
      "epoch": 0.1628,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0319,
      "step": 6919
    },
    {
      "epoch": 0.1628235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3661,
      "step": 6920
    },
    {
      "epoch": 0.1628470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.4276,
      "step": 6921
    },
    {
      "epoch": 0.16287058823529413,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.843,
      "step": 6922
    },
    {
      "epoch": 0.16289411764705883,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3242,
      "step": 6923
    },
    {
      "epoch": 0.16291764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.373,
      "step": 6924
    },
    {
      "epoch": 0.16294117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0557,
      "step": 6925
    },
    {
      "epoch": 0.16296470588235293,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1081,
      "step": 6926
    },
    {
      "epoch": 0.16298823529411766,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0997,
      "step": 6927
    },
    {
      "epoch": 0.16301176470588236,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8116,
      "step": 6928
    },
    {
      "epoch": 0.16303529411764706,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1801,
      "step": 6929
    },
    {
      "epoch": 0.16305882352941176,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0132,
      "step": 6930
    },
    {
      "epoch": 0.16308235294117648,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4384,
      "step": 6931
    },
    {
      "epoch": 0.16310588235294118,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.97,
      "step": 6932
    },
    {
      "epoch": 0.16312941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2952,
      "step": 6933
    },
    {
      "epoch": 0.16315294117647058,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7362,
      "step": 6934
    },
    {
      "epoch": 0.16317647058823528,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 6935
    },
    {
      "epoch": 0.1632,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2542,
      "step": 6936
    },
    {
      "epoch": 0.1632235294117647,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.3112,
      "step": 6937
    },
    {
      "epoch": 0.1632470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2733,
      "step": 6938
    },
    {
      "epoch": 0.1632705882352941,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7287,
      "step": 6939
    },
    {
      "epoch": 0.1632941176470588,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1821,
      "step": 6940
    },
    {
      "epoch": 0.16331764705882354,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2496,
      "step": 6941
    },
    {
      "epoch": 0.16334117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 6942
    },
    {
      "epoch": 0.16336470588235294,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9708,
      "step": 6943
    },
    {
      "epoch": 0.16338823529411764,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1013,
      "step": 6944
    },
    {
      "epoch": 0.16341176470588234,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9773,
      "step": 6945
    },
    {
      "epoch": 0.16343529411764707,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.1851,
      "step": 6946
    },
    {
      "epoch": 0.16345882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1387,
      "step": 6947
    },
    {
      "epoch": 0.16348235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.8762,
      "step": 6948
    },
    {
      "epoch": 0.16350588235294117,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.7392,
      "step": 6949
    },
    {
      "epoch": 0.1635294117647059,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1672,
      "step": 6950
    },
    {
      "epoch": 0.1635529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1439,
      "step": 6951
    },
    {
      "epoch": 0.1635764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2928,
      "step": 6952
    },
    {
      "epoch": 0.1636,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.879,
      "step": 6953
    },
    {
      "epoch": 0.1636235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.109,
      "step": 6954
    },
    {
      "epoch": 0.16364705882352942,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2787,
      "step": 6955
    },
    {
      "epoch": 0.16367058823529412,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.2388,
      "step": 6956
    },
    {
      "epoch": 0.16369411764705882,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9249,
      "step": 6957
    },
    {
      "epoch": 0.16371764705882352,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0658,
      "step": 6958
    },
    {
      "epoch": 0.16374117647058822,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3108,
      "step": 6959
    },
    {
      "epoch": 0.16376470588235295,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.7276,
      "step": 6960
    },
    {
      "epoch": 0.16378823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 6961
    },
    {
      "epoch": 0.16381176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.367,
      "step": 6962
    },
    {
      "epoch": 0.16383529411764705,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2583,
      "step": 6963
    },
    {
      "epoch": 0.16385882352941175,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 6964
    },
    {
      "epoch": 0.16388235294117648,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3918,
      "step": 6965
    },
    {
      "epoch": 0.16390588235294118,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.7416,
      "step": 6966
    },
    {
      "epoch": 0.16392941176470588,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9227,
      "step": 6967
    },
    {
      "epoch": 0.16395294117647058,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1578,
      "step": 6968
    },
    {
      "epoch": 0.1639764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1839,
      "step": 6969
    },
    {
      "epoch": 0.164,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9295,
      "step": 6970
    },
    {
      "epoch": 0.1640235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.259,
      "step": 6971
    },
    {
      "epoch": 0.1640470588235294,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9238,
      "step": 6972
    },
    {
      "epoch": 0.1640705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 6973
    },
    {
      "epoch": 0.16409411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2019,
      "step": 6974
    },
    {
      "epoch": 0.16411764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9897,
      "step": 6975
    },
    {
      "epoch": 0.16414117647058823,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 6976
    },
    {
      "epoch": 0.16416470588235293,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1508,
      "step": 6977
    },
    {
      "epoch": 0.16418823529411763,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1353,
      "step": 6978
    },
    {
      "epoch": 0.16421176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2414,
      "step": 6979
    },
    {
      "epoch": 0.16423529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0913,
      "step": 6980
    },
    {
      "epoch": 0.16425882352941176,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1969,
      "step": 6981
    },
    {
      "epoch": 0.16428235294117646,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.231,
      "step": 6982
    },
    {
      "epoch": 0.16430588235294116,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0033,
      "step": 6983
    },
    {
      "epoch": 0.1643294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4841,
      "step": 6984
    },
    {
      "epoch": 0.1643529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1022,
      "step": 6985
    },
    {
      "epoch": 0.1643764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8902,
      "step": 6986
    },
    {
      "epoch": 0.1644,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0164,
      "step": 6987
    },
    {
      "epoch": 0.16442352941176472,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2235,
      "step": 6988
    },
    {
      "epoch": 0.16444705882352942,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0272,
      "step": 6989
    },
    {
      "epoch": 0.16447058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0799,
      "step": 6990
    },
    {
      "epoch": 0.16449411764705882,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0815,
      "step": 6991
    },
    {
      "epoch": 0.16451764705882352,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2796,
      "step": 6992
    },
    {
      "epoch": 0.16454117647058825,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1723,
      "step": 6993
    },
    {
      "epoch": 0.16456470588235295,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1146,
      "step": 6994
    },
    {
      "epoch": 0.16458823529411765,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9575,
      "step": 6995
    },
    {
      "epoch": 0.16461176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8878,
      "step": 6996
    },
    {
      "epoch": 0.16463529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1555,
      "step": 6997
    },
    {
      "epoch": 0.16465882352941177,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8369,
      "step": 6998
    },
    {
      "epoch": 0.16468235294117647,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8084,
      "step": 6999
    },
    {
      "epoch": 0.16470588235294117,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1206,
      "step": 7000
    },
    {
      "epoch": 0.16470588235294117,
      "eval_loss": 2.23186993598938,
      "eval_runtime": 689.5698,
      "eval_samples_per_second": 12.327,
      "eval_steps_per_second": 3.082,
      "step": 7000
    },
    {
      "epoch": 0.16472941176470587,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9277,
      "step": 7001
    },
    {
      "epoch": 0.1647529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2097,
      "step": 7002
    },
    {
      "epoch": 0.1647764705882353,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.719,
      "step": 7003
    },
    {
      "epoch": 0.1648,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.143,
      "step": 7004
    },
    {
      "epoch": 0.1648235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.338,
      "step": 7005
    },
    {
      "epoch": 0.1648470588235294,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.7349,
      "step": 7006
    },
    {
      "epoch": 0.16487058823529413,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1482,
      "step": 7007
    },
    {
      "epoch": 0.16489411764705883,
      "grad_norm": 0.310546875,
      "learning_rate": 0.02,
      "loss": 0.7177,
      "step": 7008
    },
    {
      "epoch": 0.16491764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.7084,
      "step": 7009
    },
    {
      "epoch": 0.16494117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1474,
      "step": 7010
    },
    {
      "epoch": 0.16496470588235293,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8675,
      "step": 7011
    },
    {
      "epoch": 0.16498823529411766,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.169,
      "step": 7012
    },
    {
      "epoch": 0.16501176470588236,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3363,
      "step": 7013
    },
    {
      "epoch": 0.16503529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 7014
    },
    {
      "epoch": 0.16505882352941176,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.874,
      "step": 7015
    },
    {
      "epoch": 0.16508235294117646,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3682,
      "step": 7016
    },
    {
      "epoch": 0.16510588235294119,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9103,
      "step": 7017
    },
    {
      "epoch": 0.16512941176470589,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9262,
      "step": 7018
    },
    {
      "epoch": 0.16515294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3153,
      "step": 7019
    },
    {
      "epoch": 0.16517647058823529,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0581,
      "step": 7020
    },
    {
      "epoch": 0.1652,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9754,
      "step": 7021
    },
    {
      "epoch": 0.1652235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 7022
    },
    {
      "epoch": 0.1652470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3339,
      "step": 7023
    },
    {
      "epoch": 0.1652705882352941,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9872,
      "step": 7024
    },
    {
      "epoch": 0.1652941176470588,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8388,
      "step": 7025
    },
    {
      "epoch": 0.16531764705882354,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.934,
      "step": 7026
    },
    {
      "epoch": 0.16534117647058824,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2253,
      "step": 7027
    },
    {
      "epoch": 0.16536470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1632,
      "step": 7028
    },
    {
      "epoch": 0.16538823529411764,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.7497,
      "step": 7029
    },
    {
      "epoch": 0.16541176470588234,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.378,
      "step": 7030
    },
    {
      "epoch": 0.16543529411764707,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.3533,
      "step": 7031
    },
    {
      "epoch": 0.16545882352941177,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1216,
      "step": 7032
    },
    {
      "epoch": 0.16548235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.4115,
      "step": 7033
    },
    {
      "epoch": 0.16550588235294117,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0526,
      "step": 7034
    },
    {
      "epoch": 0.16552941176470587,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.302,
      "step": 7035
    },
    {
      "epoch": 0.1655529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2736,
      "step": 7036
    },
    {
      "epoch": 0.1655764705882353,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.0391,
      "step": 7037
    },
    {
      "epoch": 0.1656,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3811,
      "step": 7038
    },
    {
      "epoch": 0.1656235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2945,
      "step": 7039
    },
    {
      "epoch": 0.16564705882352943,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1966,
      "step": 7040
    },
    {
      "epoch": 0.16567058823529413,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1941,
      "step": 7041
    },
    {
      "epoch": 0.16569411764705883,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.8376,
      "step": 7042
    },
    {
      "epoch": 0.16571764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2695,
      "step": 7043
    },
    {
      "epoch": 0.16574117647058823,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2844,
      "step": 7044
    },
    {
      "epoch": 0.16576470588235295,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1262,
      "step": 7045
    },
    {
      "epoch": 0.16578823529411765,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0431,
      "step": 7046
    },
    {
      "epoch": 0.16581176470588235,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9415,
      "step": 7047
    },
    {
      "epoch": 0.16583529411764705,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.3083,
      "step": 7048
    },
    {
      "epoch": 0.16585882352941175,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.1465,
      "step": 7049
    },
    {
      "epoch": 0.16588235294117648,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.3658,
      "step": 7050
    },
    {
      "epoch": 0.16590588235294118,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 1.0649,
      "step": 7051
    },
    {
      "epoch": 0.16592941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1304,
      "step": 7052
    },
    {
      "epoch": 0.16595294117647058,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1564,
      "step": 7053
    },
    {
      "epoch": 0.16597647058823528,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2324,
      "step": 7054
    },
    {
      "epoch": 0.166,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0473,
      "step": 7055
    },
    {
      "epoch": 0.1660235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4079,
      "step": 7056
    },
    {
      "epoch": 0.1660470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.219,
      "step": 7057
    },
    {
      "epoch": 0.1660705882352941,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1901,
      "step": 7058
    },
    {
      "epoch": 0.16609411764705884,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2337,
      "step": 7059
    },
    {
      "epoch": 0.16611764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1496,
      "step": 7060
    },
    {
      "epoch": 0.16614117647058824,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.973,
      "step": 7061
    },
    {
      "epoch": 0.16616470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0444,
      "step": 7062
    },
    {
      "epoch": 0.16618823529411764,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0398,
      "step": 7063
    },
    {
      "epoch": 0.16621176470588236,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.2994,
      "step": 7064
    },
    {
      "epoch": 0.16623529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3147,
      "step": 7065
    },
    {
      "epoch": 0.16625882352941176,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.08,
      "step": 7066
    },
    {
      "epoch": 0.16628235294117646,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9083,
      "step": 7067
    },
    {
      "epoch": 0.16630588235294116,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9336,
      "step": 7068
    },
    {
      "epoch": 0.1663294117647059,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1888,
      "step": 7069
    },
    {
      "epoch": 0.1663529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1264,
      "step": 7070
    },
    {
      "epoch": 0.1663764705882353,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.2961,
      "step": 7071
    },
    {
      "epoch": 0.1664,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.2211,
      "step": 7072
    },
    {
      "epoch": 0.1664235294117647,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0713,
      "step": 7073
    },
    {
      "epoch": 0.16644705882352942,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.193,
      "step": 7074
    },
    {
      "epoch": 0.16647058823529412,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.168,
      "step": 7075
    },
    {
      "epoch": 0.16649411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4382,
      "step": 7076
    },
    {
      "epoch": 0.16651764705882352,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1312,
      "step": 7077
    },
    {
      "epoch": 0.16654117647058825,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1189,
      "step": 7078
    },
    {
      "epoch": 0.16656470588235295,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1915,
      "step": 7079
    },
    {
      "epoch": 0.16658823529411765,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1734,
      "step": 7080
    },
    {
      "epoch": 0.16661176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2907,
      "step": 7081
    },
    {
      "epoch": 0.16663529411764705,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3221,
      "step": 7082
    },
    {
      "epoch": 0.16665882352941178,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 0.7238,
      "step": 7083
    },
    {
      "epoch": 0.16668235294117648,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.3058,
      "step": 7084
    },
    {
      "epoch": 0.16670588235294118,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0111,
      "step": 7085
    },
    {
      "epoch": 0.16672941176470588,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0184,
      "step": 7086
    },
    {
      "epoch": 0.16675294117647058,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2124,
      "step": 7087
    },
    {
      "epoch": 0.1667764705882353,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8875,
      "step": 7088
    },
    {
      "epoch": 0.1668,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2754,
      "step": 7089
    },
    {
      "epoch": 0.1668235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4699,
      "step": 7090
    },
    {
      "epoch": 0.1668470588235294,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0411,
      "step": 7091
    },
    {
      "epoch": 0.1668705882352941,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8907,
      "step": 7092
    },
    {
      "epoch": 0.16689411764705883,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1257,
      "step": 7093
    },
    {
      "epoch": 0.16691764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2271,
      "step": 7094
    },
    {
      "epoch": 0.16694117647058823,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1062,
      "step": 7095
    },
    {
      "epoch": 0.16696470588235293,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8357,
      "step": 7096
    },
    {
      "epoch": 0.16698823529411766,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2912,
      "step": 7097
    },
    {
      "epoch": 0.16701176470588236,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9643,
      "step": 7098
    },
    {
      "epoch": 0.16703529411764706,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2351,
      "step": 7099
    },
    {
      "epoch": 0.16705882352941176,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8487,
      "step": 7100
    },
    {
      "epoch": 0.16708235294117646,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.8015,
      "step": 7101
    },
    {
      "epoch": 0.1671058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0987,
      "step": 7102
    },
    {
      "epoch": 0.1671294117647059,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1239,
      "step": 7103
    },
    {
      "epoch": 0.1671529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1558,
      "step": 7104
    },
    {
      "epoch": 0.1671764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1278,
      "step": 7105
    },
    {
      "epoch": 0.1672,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8098,
      "step": 7106
    },
    {
      "epoch": 0.16722352941176472,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3765,
      "step": 7107
    },
    {
      "epoch": 0.16724705882352942,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8643,
      "step": 7108
    },
    {
      "epoch": 0.16727058823529412,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3317,
      "step": 7109
    },
    {
      "epoch": 0.16729411764705882,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3535,
      "step": 7110
    },
    {
      "epoch": 0.16731764705882354,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2154,
      "step": 7111
    },
    {
      "epoch": 0.16734117647058824,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 7112
    },
    {
      "epoch": 0.16736470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2672,
      "step": 7113
    },
    {
      "epoch": 0.16738823529411764,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0174,
      "step": 7114
    },
    {
      "epoch": 0.16741176470588234,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1141,
      "step": 7115
    },
    {
      "epoch": 0.16743529411764707,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1652,
      "step": 7116
    },
    {
      "epoch": 0.16745882352941177,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.9864,
      "step": 7117
    },
    {
      "epoch": 0.16748235294117647,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.9577,
      "step": 7118
    },
    {
      "epoch": 0.16750588235294117,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 7119
    },
    {
      "epoch": 0.16752941176470587,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.269,
      "step": 7120
    },
    {
      "epoch": 0.1675529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.254,
      "step": 7121
    },
    {
      "epoch": 0.1675764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.4411,
      "step": 7122
    },
    {
      "epoch": 0.1676,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.339,
      "step": 7123
    },
    {
      "epoch": 0.1676235294117647,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.9108,
      "step": 7124
    },
    {
      "epoch": 0.1676470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1078,
      "step": 7125
    },
    {
      "epoch": 0.16767058823529413,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0037,
      "step": 7126
    },
    {
      "epoch": 0.16769411764705883,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.5537,
      "step": 7127
    },
    {
      "epoch": 0.16771764705882353,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.7684,
      "step": 7128
    },
    {
      "epoch": 0.16774117647058823,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1253,
      "step": 7129
    },
    {
      "epoch": 0.16776470588235295,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.8843,
      "step": 7130
    },
    {
      "epoch": 0.16778823529411765,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 7131
    },
    {
      "epoch": 0.16781176470588235,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0734,
      "step": 7132
    },
    {
      "epoch": 0.16783529411764705,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0921,
      "step": 7133
    },
    {
      "epoch": 0.16785882352941175,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3641,
      "step": 7134
    },
    {
      "epoch": 0.16788235294117648,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2918,
      "step": 7135
    },
    {
      "epoch": 0.16790588235294118,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0264,
      "step": 7136
    },
    {
      "epoch": 0.16792941176470588,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1372,
      "step": 7137
    },
    {
      "epoch": 0.16795294117647058,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3331,
      "step": 7138
    },
    {
      "epoch": 0.16797647058823528,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3494,
      "step": 7139
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 7140
    },
    {
      "epoch": 0.1680235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0927,
      "step": 7141
    },
    {
      "epoch": 0.1680470588235294,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2105,
      "step": 7142
    },
    {
      "epoch": 0.1680705882352941,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8473,
      "step": 7143
    },
    {
      "epoch": 0.1680941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2018,
      "step": 7144
    },
    {
      "epoch": 0.16811764705882354,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4299,
      "step": 7145
    },
    {
      "epoch": 0.16814117647058824,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7958,
      "step": 7146
    },
    {
      "epoch": 0.16816470588235294,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.2055,
      "step": 7147
    },
    {
      "epoch": 0.16818823529411764,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.977,
      "step": 7148
    },
    {
      "epoch": 0.16821176470588237,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0899,
      "step": 7149
    },
    {
      "epoch": 0.16823529411764707,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0589,
      "step": 7150
    },
    {
      "epoch": 0.16825882352941177,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1225,
      "step": 7151
    },
    {
      "epoch": 0.16828235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 7152
    },
    {
      "epoch": 0.16830588235294117,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.7176,
      "step": 7153
    },
    {
      "epoch": 0.1683294117647059,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2512,
      "step": 7154
    },
    {
      "epoch": 0.1683529411764706,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8349,
      "step": 7155
    },
    {
      "epoch": 0.1683764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2255,
      "step": 7156
    },
    {
      "epoch": 0.1684,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1244,
      "step": 7157
    },
    {
      "epoch": 0.1684235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4553,
      "step": 7158
    },
    {
      "epoch": 0.16844705882352942,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1581,
      "step": 7159
    },
    {
      "epoch": 0.16847058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2628,
      "step": 7160
    },
    {
      "epoch": 0.16849411764705882,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0461,
      "step": 7161
    },
    {
      "epoch": 0.16851764705882352,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1047,
      "step": 7162
    },
    {
      "epoch": 0.16854117647058822,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1494,
      "step": 7163
    },
    {
      "epoch": 0.16856470588235295,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1978,
      "step": 7164
    },
    {
      "epoch": 0.16858823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1866,
      "step": 7165
    },
    {
      "epoch": 0.16861176470588235,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.0419,
      "step": 7166
    },
    {
      "epoch": 0.16863529411764705,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 7167
    },
    {
      "epoch": 0.16865882352941178,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3646,
      "step": 7168
    },
    {
      "epoch": 0.16868235294117648,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.308,
      "step": 7169
    },
    {
      "epoch": 0.16870588235294118,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9944,
      "step": 7170
    },
    {
      "epoch": 0.16872941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3496,
      "step": 7171
    },
    {
      "epoch": 0.16875294117647058,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2747,
      "step": 7172
    },
    {
      "epoch": 0.1687764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2533,
      "step": 7173
    },
    {
      "epoch": 0.1688,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0418,
      "step": 7174
    },
    {
      "epoch": 0.1688235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1893,
      "step": 7175
    },
    {
      "epoch": 0.1688470588235294,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 7176
    },
    {
      "epoch": 0.1688705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0219,
      "step": 7177
    },
    {
      "epoch": 0.16889411764705883,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1875,
      "step": 7178
    },
    {
      "epoch": 0.16891764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2832,
      "step": 7179
    },
    {
      "epoch": 0.16894117647058823,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 7180
    },
    {
      "epoch": 0.16896470588235293,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 1.1182,
      "step": 7181
    },
    {
      "epoch": 0.16898823529411763,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1469,
      "step": 7182
    },
    {
      "epoch": 0.16901176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2189,
      "step": 7183
    },
    {
      "epoch": 0.16903529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.305,
      "step": 7184
    },
    {
      "epoch": 0.16905882352941176,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1309,
      "step": 7185
    },
    {
      "epoch": 0.16908235294117646,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.412,
      "step": 7186
    },
    {
      "epoch": 0.1691058823529412,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4415,
      "step": 7187
    },
    {
      "epoch": 0.1691294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 7188
    },
    {
      "epoch": 0.1691529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8884,
      "step": 7189
    },
    {
      "epoch": 0.1691764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2037,
      "step": 7190
    },
    {
      "epoch": 0.1692,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2943,
      "step": 7191
    },
    {
      "epoch": 0.16922352941176472,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0864,
      "step": 7192
    },
    {
      "epoch": 0.16924705882352942,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.4383,
      "step": 7193
    },
    {
      "epoch": 0.16927058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1815,
      "step": 7194
    },
    {
      "epoch": 0.16929411764705882,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9809,
      "step": 7195
    },
    {
      "epoch": 0.16931764705882352,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2758,
      "step": 7196
    },
    {
      "epoch": 0.16934117647058824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2229,
      "step": 7197
    },
    {
      "epoch": 0.16936470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1841,
      "step": 7198
    },
    {
      "epoch": 0.16938823529411765,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.102,
      "step": 7199
    },
    {
      "epoch": 0.16941176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3943,
      "step": 7200
    },
    {
      "epoch": 0.16943529411764705,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2624,
      "step": 7201
    },
    {
      "epoch": 0.16945882352941177,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.983,
      "step": 7202
    },
    {
      "epoch": 0.16948235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.1154,
      "step": 7203
    },
    {
      "epoch": 0.16950588235294117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4067,
      "step": 7204
    },
    {
      "epoch": 0.16952941176470587,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4568,
      "step": 7205
    },
    {
      "epoch": 0.1695529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1097,
      "step": 7206
    },
    {
      "epoch": 0.1695764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2726,
      "step": 7207
    },
    {
      "epoch": 0.1696,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1857,
      "step": 7208
    },
    {
      "epoch": 0.1696235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1045,
      "step": 7209
    },
    {
      "epoch": 0.1696470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0562,
      "step": 7210
    },
    {
      "epoch": 0.16967058823529413,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1347,
      "step": 7211
    },
    {
      "epoch": 0.16969411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4048,
      "step": 7212
    },
    {
      "epoch": 0.16971764705882353,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9333,
      "step": 7213
    },
    {
      "epoch": 0.16974117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.5146,
      "step": 7214
    },
    {
      "epoch": 0.16976470588235293,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.925,
      "step": 7215
    },
    {
      "epoch": 0.16978823529411766,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0815,
      "step": 7216
    },
    {
      "epoch": 0.16981176470588236,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.4096,
      "step": 7217
    },
    {
      "epoch": 0.16983529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8819,
      "step": 7218
    },
    {
      "epoch": 0.16985882352941176,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.074,
      "step": 7219
    },
    {
      "epoch": 0.16988235294117648,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8801,
      "step": 7220
    },
    {
      "epoch": 0.16990588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2881,
      "step": 7221
    },
    {
      "epoch": 0.16992941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.481,
      "step": 7222
    },
    {
      "epoch": 0.16995294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2386,
      "step": 7223
    },
    {
      "epoch": 0.16997647058823528,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1293,
      "step": 7224
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9941,
      "step": 7225
    },
    {
      "epoch": 0.1700235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.8692,
      "step": 7226
    },
    {
      "epoch": 0.1700470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3361,
      "step": 7227
    },
    {
      "epoch": 0.1700705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.105,
      "step": 7228
    },
    {
      "epoch": 0.1700941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1134,
      "step": 7229
    },
    {
      "epoch": 0.17011764705882354,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3221,
      "step": 7230
    },
    {
      "epoch": 0.17014117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1916,
      "step": 7231
    },
    {
      "epoch": 0.17016470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9199,
      "step": 7232
    },
    {
      "epoch": 0.17018823529411764,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3432,
      "step": 7233
    },
    {
      "epoch": 0.17021176470588234,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2594,
      "step": 7234
    },
    {
      "epoch": 0.17023529411764707,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2333,
      "step": 7235
    },
    {
      "epoch": 0.17025882352941177,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.3158,
      "step": 7236
    },
    {
      "epoch": 0.17028235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0557,
      "step": 7237
    },
    {
      "epoch": 0.17030588235294117,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4296,
      "step": 7238
    },
    {
      "epoch": 0.1703294117647059,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0761,
      "step": 7239
    },
    {
      "epoch": 0.1703529411764706,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 7240
    },
    {
      "epoch": 0.1703764705882353,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0258,
      "step": 7241
    },
    {
      "epoch": 0.1704,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 7242
    },
    {
      "epoch": 0.1704235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1848,
      "step": 7243
    },
    {
      "epoch": 0.17044705882352942,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0894,
      "step": 7244
    },
    {
      "epoch": 0.17047058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2562,
      "step": 7245
    },
    {
      "epoch": 0.17049411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2904,
      "step": 7246
    },
    {
      "epoch": 0.17051764705882352,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 7247
    },
    {
      "epoch": 0.17054117647058822,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0399,
      "step": 7248
    },
    {
      "epoch": 0.17056470588235295,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1922,
      "step": 7249
    },
    {
      "epoch": 0.17058823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3169,
      "step": 7250
    },
    {
      "epoch": 0.17061176470588235,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 7251
    },
    {
      "epoch": 0.17063529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3736,
      "step": 7252
    },
    {
      "epoch": 0.17065882352941175,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3218,
      "step": 7253
    },
    {
      "epoch": 0.17068235294117648,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3179,
      "step": 7254
    },
    {
      "epoch": 0.17070588235294118,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.14,
      "step": 7255
    },
    {
      "epoch": 0.17072941176470588,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8642,
      "step": 7256
    },
    {
      "epoch": 0.17075294117647058,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9128,
      "step": 7257
    },
    {
      "epoch": 0.1707764705882353,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2184,
      "step": 7258
    },
    {
      "epoch": 0.1708,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.4645,
      "step": 7259
    },
    {
      "epoch": 0.1708235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.1671,
      "step": 7260
    },
    {
      "epoch": 0.1708470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.3043,
      "step": 7261
    },
    {
      "epoch": 0.1708705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2002,
      "step": 7262
    },
    {
      "epoch": 0.17089411764705884,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4231,
      "step": 7263
    },
    {
      "epoch": 0.17091764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4367,
      "step": 7264
    },
    {
      "epoch": 0.17094117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3872,
      "step": 7265
    },
    {
      "epoch": 0.17096470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0828,
      "step": 7266
    },
    {
      "epoch": 0.17098823529411764,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1992,
      "step": 7267
    },
    {
      "epoch": 0.17101176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.154,
      "step": 7268
    },
    {
      "epoch": 0.17103529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1306,
      "step": 7269
    },
    {
      "epoch": 0.17105882352941176,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1009,
      "step": 7270
    },
    {
      "epoch": 0.17108235294117646,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8483,
      "step": 7271
    },
    {
      "epoch": 0.17110588235294116,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.164,
      "step": 7272
    },
    {
      "epoch": 0.1711294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3487,
      "step": 7273
    },
    {
      "epoch": 0.1711529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0207,
      "step": 7274
    },
    {
      "epoch": 0.1711764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.446,
      "step": 7275
    },
    {
      "epoch": 0.1712,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2289,
      "step": 7276
    },
    {
      "epoch": 0.17122352941176472,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9321,
      "step": 7277
    },
    {
      "epoch": 0.17124705882352942,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9423,
      "step": 7278
    },
    {
      "epoch": 0.17127058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 7279
    },
    {
      "epoch": 0.17129411764705882,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1417,
      "step": 7280
    },
    {
      "epoch": 0.17131764705882352,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.192,
      "step": 7281
    },
    {
      "epoch": 0.17134117647058825,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3115,
      "step": 7282
    },
    {
      "epoch": 0.17136470588235295,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0396,
      "step": 7283
    },
    {
      "epoch": 0.17138823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 7284
    },
    {
      "epoch": 0.17141176470588235,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8093,
      "step": 7285
    },
    {
      "epoch": 0.17143529411764705,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0948,
      "step": 7286
    },
    {
      "epoch": 0.17145882352941177,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.9776,
      "step": 7287
    },
    {
      "epoch": 0.17148235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1379,
      "step": 7288
    },
    {
      "epoch": 0.17150588235294117,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.973,
      "step": 7289
    },
    {
      "epoch": 0.17152941176470587,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 7290
    },
    {
      "epoch": 0.17155294117647057,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2182,
      "step": 7291
    },
    {
      "epoch": 0.1715764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4351,
      "step": 7292
    },
    {
      "epoch": 0.1716,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2426,
      "step": 7293
    },
    {
      "epoch": 0.1716235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.1962,
      "step": 7294
    },
    {
      "epoch": 0.1716470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2089,
      "step": 7295
    },
    {
      "epoch": 0.17167058823529413,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0395,
      "step": 7296
    },
    {
      "epoch": 0.17169411764705883,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2981,
      "step": 7297
    },
    {
      "epoch": 0.17171764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2689,
      "step": 7298
    },
    {
      "epoch": 0.17174117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1204,
      "step": 7299
    },
    {
      "epoch": 0.17176470588235293,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.258,
      "step": 7300
    },
    {
      "epoch": 0.17178823529411766,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0504,
      "step": 7301
    },
    {
      "epoch": 0.17181176470588236,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1107,
      "step": 7302
    },
    {
      "epoch": 0.17183529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2584,
      "step": 7303
    },
    {
      "epoch": 0.17185882352941176,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.1988,
      "step": 7304
    },
    {
      "epoch": 0.17188235294117646,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0254,
      "step": 7305
    },
    {
      "epoch": 0.1719058823529412,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.011,
      "step": 7306
    },
    {
      "epoch": 0.1719294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2146,
      "step": 7307
    },
    {
      "epoch": 0.1719529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3823,
      "step": 7308
    },
    {
      "epoch": 0.1719764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1442,
      "step": 7309
    },
    {
      "epoch": 0.172,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1827,
      "step": 7310
    },
    {
      "epoch": 0.17202352941176471,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1527,
      "step": 7311
    },
    {
      "epoch": 0.17204705882352941,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1538,
      "step": 7312
    },
    {
      "epoch": 0.17207058823529411,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9701,
      "step": 7313
    },
    {
      "epoch": 0.17209411764705881,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1472,
      "step": 7314
    },
    {
      "epoch": 0.17211764705882354,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1993,
      "step": 7315
    },
    {
      "epoch": 0.17214117647058824,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2187,
      "step": 7316
    },
    {
      "epoch": 0.17216470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2799,
      "step": 7317
    },
    {
      "epoch": 0.17218823529411764,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.227,
      "step": 7318
    },
    {
      "epoch": 0.17221176470588234,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2952,
      "step": 7319
    },
    {
      "epoch": 0.17223529411764707,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.3056,
      "step": 7320
    },
    {
      "epoch": 0.17225882352941177,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2942,
      "step": 7321
    },
    {
      "epoch": 0.17228235294117647,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.8576,
      "step": 7322
    },
    {
      "epoch": 0.17230588235294117,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3094,
      "step": 7323
    },
    {
      "epoch": 0.17232941176470587,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1085,
      "step": 7324
    },
    {
      "epoch": 0.1723529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4201,
      "step": 7325
    },
    {
      "epoch": 0.1723764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 7326
    },
    {
      "epoch": 0.1724,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2898,
      "step": 7327
    },
    {
      "epoch": 0.1724235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2324,
      "step": 7328
    },
    {
      "epoch": 0.17244705882352943,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1797,
      "step": 7329
    },
    {
      "epoch": 0.17247058823529413,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2222,
      "step": 7330
    },
    {
      "epoch": 0.17249411764705883,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1101,
      "step": 7331
    },
    {
      "epoch": 0.17251764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3244,
      "step": 7332
    },
    {
      "epoch": 0.17254117647058823,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2446,
      "step": 7333
    },
    {
      "epoch": 0.17256470588235295,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.8783,
      "step": 7334
    },
    {
      "epoch": 0.17258823529411765,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0056,
      "step": 7335
    },
    {
      "epoch": 0.17261176470588235,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.1484,
      "step": 7336
    },
    {
      "epoch": 0.17263529411764705,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 7337
    },
    {
      "epoch": 0.17265882352941175,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9896,
      "step": 7338
    },
    {
      "epoch": 0.17268235294117648,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2103,
      "step": 7339
    },
    {
      "epoch": 0.17270588235294118,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.8599,
      "step": 7340
    },
    {
      "epoch": 0.17272941176470588,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0786,
      "step": 7341
    },
    {
      "epoch": 0.17275294117647058,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2441,
      "step": 7342
    },
    {
      "epoch": 0.17277647058823528,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.021,
      "step": 7343
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.3129,
      "step": 7344
    },
    {
      "epoch": 0.1728235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2463,
      "step": 7345
    },
    {
      "epoch": 0.1728470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2933,
      "step": 7346
    },
    {
      "epoch": 0.1728705882352941,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9635,
      "step": 7347
    },
    {
      "epoch": 0.17289411764705884,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.6048,
      "step": 7348
    },
    {
      "epoch": 0.17291764705882354,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0303,
      "step": 7349
    },
    {
      "epoch": 0.17294117647058824,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.3796,
      "step": 7350
    },
    {
      "epoch": 0.17296470588235294,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.813,
      "step": 7351
    },
    {
      "epoch": 0.17298823529411764,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7419,
      "step": 7352
    },
    {
      "epoch": 0.17301176470588236,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3601,
      "step": 7353
    },
    {
      "epoch": 0.17303529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1764,
      "step": 7354
    },
    {
      "epoch": 0.17305882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1161,
      "step": 7355
    },
    {
      "epoch": 0.17308235294117646,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0272,
      "step": 7356
    },
    {
      "epoch": 0.17310588235294116,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1898,
      "step": 7357
    },
    {
      "epoch": 0.1731294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0847,
      "step": 7358
    },
    {
      "epoch": 0.1731529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9133,
      "step": 7359
    },
    {
      "epoch": 0.1731764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2832,
      "step": 7360
    },
    {
      "epoch": 0.1732,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1193,
      "step": 7361
    },
    {
      "epoch": 0.1732235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4155,
      "step": 7362
    },
    {
      "epoch": 0.17324705882352942,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 7363
    },
    {
      "epoch": 0.17327058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2381,
      "step": 7364
    },
    {
      "epoch": 0.17329411764705882,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9967,
      "step": 7365
    },
    {
      "epoch": 0.17331764705882352,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0048,
      "step": 7366
    },
    {
      "epoch": 0.17334117647058825,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.06,
      "step": 7367
    },
    {
      "epoch": 0.17336470588235295,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2457,
      "step": 7368
    },
    {
      "epoch": 0.17338823529411765,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8375,
      "step": 7369
    },
    {
      "epoch": 0.17341176470588235,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1506,
      "step": 7370
    },
    {
      "epoch": 0.17343529411764705,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2735,
      "step": 7371
    },
    {
      "epoch": 0.17345882352941178,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.203,
      "step": 7372
    },
    {
      "epoch": 0.17348235294117648,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.6027,
      "step": 7373
    },
    {
      "epoch": 0.17350588235294118,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 7374
    },
    {
      "epoch": 0.17352941176470588,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.06,
      "step": 7375
    },
    {
      "epoch": 0.17355294117647058,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.5007,
      "step": 7376
    },
    {
      "epoch": 0.1735764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 7377
    },
    {
      "epoch": 0.1736,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3514,
      "step": 7378
    },
    {
      "epoch": 0.1736235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2075,
      "step": 7379
    },
    {
      "epoch": 0.1736470588235294,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0632,
      "step": 7380
    },
    {
      "epoch": 0.1736705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1025,
      "step": 7381
    },
    {
      "epoch": 0.17369411764705883,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0891,
      "step": 7382
    },
    {
      "epoch": 0.17371764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9017,
      "step": 7383
    },
    {
      "epoch": 0.17374117647058823,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9279,
      "step": 7384
    },
    {
      "epoch": 0.17376470588235293,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.12,
      "step": 7385
    },
    {
      "epoch": 0.17378823529411766,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.1172,
      "step": 7386
    },
    {
      "epoch": 0.17381176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4402,
      "step": 7387
    },
    {
      "epoch": 0.17383529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.179,
      "step": 7388
    },
    {
      "epoch": 0.17385882352941176,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9127,
      "step": 7389
    },
    {
      "epoch": 0.17388235294117646,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7546,
      "step": 7390
    },
    {
      "epoch": 0.1739058823529412,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0923,
      "step": 7391
    },
    {
      "epoch": 0.1739294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2038,
      "step": 7392
    },
    {
      "epoch": 0.1739529411764706,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 7393
    },
    {
      "epoch": 0.1739764705882353,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0082,
      "step": 7394
    },
    {
      "epoch": 0.174,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0561,
      "step": 7395
    },
    {
      "epoch": 0.17402352941176472,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.5703,
      "step": 7396
    },
    {
      "epoch": 0.17404705882352942,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.046,
      "step": 7397
    },
    {
      "epoch": 0.17407058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3511,
      "step": 7398
    },
    {
      "epoch": 0.17409411764705882,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0686,
      "step": 7399
    },
    {
      "epoch": 0.17411764705882352,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0038,
      "step": 7400
    },
    {
      "epoch": 0.17414117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9739,
      "step": 7401
    },
    {
      "epoch": 0.17416470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9474,
      "step": 7402
    },
    {
      "epoch": 0.17418823529411764,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0201,
      "step": 7403
    },
    {
      "epoch": 0.17421176470588234,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.7672,
      "step": 7404
    },
    {
      "epoch": 0.17423529411764707,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3086,
      "step": 7405
    },
    {
      "epoch": 0.17425882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0221,
      "step": 7406
    },
    {
      "epoch": 0.17428235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2576,
      "step": 7407
    },
    {
      "epoch": 0.17430588235294117,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8656,
      "step": 7408
    },
    {
      "epoch": 0.17432941176470587,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0751,
      "step": 7409
    },
    {
      "epoch": 0.1743529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0625,
      "step": 7410
    },
    {
      "epoch": 0.1743764705882353,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9182,
      "step": 7411
    },
    {
      "epoch": 0.1744,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1973,
      "step": 7412
    },
    {
      "epoch": 0.1744235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.4026,
      "step": 7413
    },
    {
      "epoch": 0.1744470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0768,
      "step": 7414
    },
    {
      "epoch": 0.17447058823529413,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2845,
      "step": 7415
    },
    {
      "epoch": 0.17449411764705883,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.357,
      "step": 7416
    },
    {
      "epoch": 0.17451764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 7417
    },
    {
      "epoch": 0.17454117647058823,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3051,
      "step": 7418
    },
    {
      "epoch": 0.17456470588235293,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2908,
      "step": 7419
    },
    {
      "epoch": 0.17458823529411766,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0973,
      "step": 7420
    },
    {
      "epoch": 0.17461176470588236,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1564,
      "step": 7421
    },
    {
      "epoch": 0.17463529411764706,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9247,
      "step": 7422
    },
    {
      "epoch": 0.17465882352941176,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8758,
      "step": 7423
    },
    {
      "epoch": 0.17468235294117648,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2131,
      "step": 7424
    },
    {
      "epoch": 0.17470588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2188,
      "step": 7425
    },
    {
      "epoch": 0.17472941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9448,
      "step": 7426
    },
    {
      "epoch": 0.17475294117647058,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0064,
      "step": 7427
    },
    {
      "epoch": 0.17477647058823528,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3311,
      "step": 7428
    },
    {
      "epoch": 0.1748,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2602,
      "step": 7429
    },
    {
      "epoch": 0.1748235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 7430
    },
    {
      "epoch": 0.1748470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0185,
      "step": 7431
    },
    {
      "epoch": 0.1748705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2818,
      "step": 7432
    },
    {
      "epoch": 0.1748941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0606,
      "step": 7433
    },
    {
      "epoch": 0.17491764705882354,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1921,
      "step": 7434
    },
    {
      "epoch": 0.17494117647058824,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1099,
      "step": 7435
    },
    {
      "epoch": 0.17496470588235294,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.1435,
      "step": 7436
    },
    {
      "epoch": 0.17498823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0557,
      "step": 7437
    },
    {
      "epoch": 0.17501176470588237,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.007,
      "step": 7438
    },
    {
      "epoch": 0.17503529411764707,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.8944,
      "step": 7439
    },
    {
      "epoch": 0.17505882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.4559,
      "step": 7440
    },
    {
      "epoch": 0.17508235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4576,
      "step": 7441
    },
    {
      "epoch": 0.17510588235294117,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1404,
      "step": 7442
    },
    {
      "epoch": 0.1751294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.4808,
      "step": 7443
    },
    {
      "epoch": 0.1751529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.23,
      "step": 7444
    },
    {
      "epoch": 0.1751764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9948,
      "step": 7445
    },
    {
      "epoch": 0.1752,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.784,
      "step": 7446
    },
    {
      "epoch": 0.1752235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9295,
      "step": 7447
    },
    {
      "epoch": 0.17524705882352942,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 1.1357,
      "step": 7448
    },
    {
      "epoch": 0.17527058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2872,
      "step": 7449
    },
    {
      "epoch": 0.17529411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4127,
      "step": 7450
    },
    {
      "epoch": 0.17531764705882352,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0031,
      "step": 7451
    },
    {
      "epoch": 0.17534117647058822,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.2003,
      "step": 7452
    },
    {
      "epoch": 0.17536470588235295,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2244,
      "step": 7453
    },
    {
      "epoch": 0.17538823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.179,
      "step": 7454
    },
    {
      "epoch": 0.17541176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0721,
      "step": 7455
    },
    {
      "epoch": 0.17543529411764705,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8452,
      "step": 7456
    },
    {
      "epoch": 0.17545882352941178,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.309,
      "step": 7457
    },
    {
      "epoch": 0.17548235294117648,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.134,
      "step": 7458
    },
    {
      "epoch": 0.17550588235294118,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9683,
      "step": 7459
    },
    {
      "epoch": 0.17552941176470588,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0041,
      "step": 7460
    },
    {
      "epoch": 0.17555294117647058,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1653,
      "step": 7461
    },
    {
      "epoch": 0.1755764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2986,
      "step": 7462
    },
    {
      "epoch": 0.1756,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0316,
      "step": 7463
    },
    {
      "epoch": 0.1756235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3852,
      "step": 7464
    },
    {
      "epoch": 0.1756470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.962,
      "step": 7465
    },
    {
      "epoch": 0.1756705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3607,
      "step": 7466
    },
    {
      "epoch": 0.17569411764705883,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.5702,
      "step": 7467
    },
    {
      "epoch": 0.17571764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8315,
      "step": 7468
    },
    {
      "epoch": 0.17574117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0193,
      "step": 7469
    },
    {
      "epoch": 0.17576470588235293,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8239,
      "step": 7470
    },
    {
      "epoch": 0.17578823529411763,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8216,
      "step": 7471
    },
    {
      "epoch": 0.17581176470588236,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2868,
      "step": 7472
    },
    {
      "epoch": 0.17583529411764706,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8589,
      "step": 7473
    },
    {
      "epoch": 0.17585882352941176,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.0057,
      "step": 7474
    },
    {
      "epoch": 0.17588235294117646,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 7475
    },
    {
      "epoch": 0.1759058823529412,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.4713,
      "step": 7476
    },
    {
      "epoch": 0.1759294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3797,
      "step": 7477
    },
    {
      "epoch": 0.1759529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1666,
      "step": 7478
    },
    {
      "epoch": 0.1759764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1452,
      "step": 7479
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.347,
      "step": 7480
    },
    {
      "epoch": 0.17602352941176472,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2497,
      "step": 7481
    },
    {
      "epoch": 0.17604705882352942,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9117,
      "step": 7482
    },
    {
      "epoch": 0.17607058823529412,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.7489,
      "step": 7483
    },
    {
      "epoch": 0.17609411764705882,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2276,
      "step": 7484
    },
    {
      "epoch": 0.17611764705882352,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2726,
      "step": 7485
    },
    {
      "epoch": 0.17614117647058825,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1904,
      "step": 7486
    },
    {
      "epoch": 0.17616470588235295,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.159,
      "step": 7487
    },
    {
      "epoch": 0.17618823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 7488
    },
    {
      "epoch": 0.17621176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.203,
      "step": 7489
    },
    {
      "epoch": 0.17623529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2689,
      "step": 7490
    },
    {
      "epoch": 0.17625882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3363,
      "step": 7491
    },
    {
      "epoch": 0.17628235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9754,
      "step": 7492
    },
    {
      "epoch": 0.17630588235294117,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2838,
      "step": 7493
    },
    {
      "epoch": 0.17632941176470587,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2438,
      "step": 7494
    },
    {
      "epoch": 0.1763529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0824,
      "step": 7495
    },
    {
      "epoch": 0.1763764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.931,
      "step": 7496
    },
    {
      "epoch": 0.1764,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 7497
    },
    {
      "epoch": 0.1764235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0716,
      "step": 7498
    },
    {
      "epoch": 0.1764470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3131,
      "step": 7499
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0804,
      "step": 7500
    },
    {
      "epoch": 0.17649411764705883,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.8271,
      "step": 7501
    },
    {
      "epoch": 0.17651764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1481,
      "step": 7502
    },
    {
      "epoch": 0.17654117647058823,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1654,
      "step": 7503
    },
    {
      "epoch": 0.17656470588235293,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9137,
      "step": 7504
    },
    {
      "epoch": 0.17658823529411766,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1325,
      "step": 7505
    },
    {
      "epoch": 0.17661176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0395,
      "step": 7506
    },
    {
      "epoch": 0.17663529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3163,
      "step": 7507
    },
    {
      "epoch": 0.17665882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.934,
      "step": 7508
    },
    {
      "epoch": 0.17668235294117646,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2995,
      "step": 7509
    },
    {
      "epoch": 0.17670588235294118,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.4271,
      "step": 7510
    },
    {
      "epoch": 0.17672941176470588,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9484,
      "step": 7511
    },
    {
      "epoch": 0.17675294117647058,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2875,
      "step": 7512
    },
    {
      "epoch": 0.17677647058823528,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9892,
      "step": 7513
    },
    {
      "epoch": 0.1768,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9336,
      "step": 7514
    },
    {
      "epoch": 0.1768235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3705,
      "step": 7515
    },
    {
      "epoch": 0.1768470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.4165,
      "step": 7516
    },
    {
      "epoch": 0.1768705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 7517
    },
    {
      "epoch": 0.1768941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3761,
      "step": 7518
    },
    {
      "epoch": 0.17691764705882354,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2201,
      "step": 7519
    },
    {
      "epoch": 0.17694117647058824,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1712,
      "step": 7520
    },
    {
      "epoch": 0.17696470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3057,
      "step": 7521
    },
    {
      "epoch": 0.17698823529411764,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9048,
      "step": 7522
    },
    {
      "epoch": 0.17701176470588234,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 7523
    },
    {
      "epoch": 0.17703529411764707,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2443,
      "step": 7524
    },
    {
      "epoch": 0.17705882352941177,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 7525
    },
    {
      "epoch": 0.17708235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9555,
      "step": 7526
    },
    {
      "epoch": 0.17710588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3082,
      "step": 7527
    },
    {
      "epoch": 0.17712941176470587,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1365,
      "step": 7528
    },
    {
      "epoch": 0.1771529411764706,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.916,
      "step": 7529
    },
    {
      "epoch": 0.1771764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2529,
      "step": 7530
    },
    {
      "epoch": 0.1772,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1958,
      "step": 7531
    },
    {
      "epoch": 0.1772235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.233,
      "step": 7532
    },
    {
      "epoch": 0.17724705882352942,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2959,
      "step": 7533
    },
    {
      "epoch": 0.17727058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1141,
      "step": 7534
    },
    {
      "epoch": 0.17729411764705882,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1732,
      "step": 7535
    },
    {
      "epoch": 0.17731764705882352,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1916,
      "step": 7536
    },
    {
      "epoch": 0.17734117647058822,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2336,
      "step": 7537
    },
    {
      "epoch": 0.17736470588235295,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2094,
      "step": 7538
    },
    {
      "epoch": 0.17738823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0088,
      "step": 7539
    },
    {
      "epoch": 0.17741176470588235,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8424,
      "step": 7540
    },
    {
      "epoch": 0.17743529411764705,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0556,
      "step": 7541
    },
    {
      "epoch": 0.17745882352941175,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.7421,
      "step": 7542
    },
    {
      "epoch": 0.17748235294117648,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1784,
      "step": 7543
    },
    {
      "epoch": 0.17750588235294118,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8953,
      "step": 7544
    },
    {
      "epoch": 0.17752941176470588,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9518,
      "step": 7545
    },
    {
      "epoch": 0.17755294117647058,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0565,
      "step": 7546
    },
    {
      "epoch": 0.1775764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.332,
      "step": 7547
    },
    {
      "epoch": 0.1776,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.2021,
      "step": 7548
    },
    {
      "epoch": 0.1776235294117647,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9391,
      "step": 7549
    },
    {
      "epoch": 0.1776470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1974,
      "step": 7550
    },
    {
      "epoch": 0.1776705882352941,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 7551
    },
    {
      "epoch": 0.17769411764705884,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 7552
    },
    {
      "epoch": 0.17771764705882354,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4704,
      "step": 7553
    },
    {
      "epoch": 0.17774117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.4583,
      "step": 7554
    },
    {
      "epoch": 0.17776470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0345,
      "step": 7555
    },
    {
      "epoch": 0.17778823529411764,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9682,
      "step": 7556
    },
    {
      "epoch": 0.17781176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.4511,
      "step": 7557
    },
    {
      "epoch": 0.17783529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 7558
    },
    {
      "epoch": 0.17785882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2462,
      "step": 7559
    },
    {
      "epoch": 0.17788235294117646,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1419,
      "step": 7560
    },
    {
      "epoch": 0.17790588235294116,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0345,
      "step": 7561
    },
    {
      "epoch": 0.1779294117647059,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 7562
    },
    {
      "epoch": 0.1779529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2402,
      "step": 7563
    },
    {
      "epoch": 0.1779764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 7564
    },
    {
      "epoch": 0.178,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2159,
      "step": 7565
    },
    {
      "epoch": 0.17802352941176472,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2704,
      "step": 7566
    },
    {
      "epoch": 0.17804705882352942,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2491,
      "step": 7567
    },
    {
      "epoch": 0.17807058823529412,
      "grad_norm": 0.3984375,
      "learning_rate": 0.02,
      "loss": 1.4345,
      "step": 7568
    },
    {
      "epoch": 0.17809411764705882,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 7569
    },
    {
      "epoch": 0.17811764705882352,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7926,
      "step": 7570
    },
    {
      "epoch": 0.17814117647058825,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0148,
      "step": 7571
    },
    {
      "epoch": 0.17816470588235295,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0292,
      "step": 7572
    },
    {
      "epoch": 0.17818823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.3507,
      "step": 7573
    },
    {
      "epoch": 0.17821176470588235,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9701,
      "step": 7574
    },
    {
      "epoch": 0.17823529411764705,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1518,
      "step": 7575
    },
    {
      "epoch": 0.17825882352941178,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2578,
      "step": 7576
    },
    {
      "epoch": 0.17828235294117648,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2884,
      "step": 7577
    },
    {
      "epoch": 0.17830588235294118,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9148,
      "step": 7578
    },
    {
      "epoch": 0.17832941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1754,
      "step": 7579
    },
    {
      "epoch": 0.17835294117647058,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0135,
      "step": 7580
    },
    {
      "epoch": 0.1783764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2693,
      "step": 7581
    },
    {
      "epoch": 0.1784,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2699,
      "step": 7582
    },
    {
      "epoch": 0.1784235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9633,
      "step": 7583
    },
    {
      "epoch": 0.1784470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1266,
      "step": 7584
    },
    {
      "epoch": 0.17847058823529413,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0757,
      "step": 7585
    },
    {
      "epoch": 0.17849411764705883,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 7586
    },
    {
      "epoch": 0.17851764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0297,
      "step": 7587
    },
    {
      "epoch": 0.17854117647058823,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.91,
      "step": 7588
    },
    {
      "epoch": 0.17856470588235293,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.168,
      "step": 7589
    },
    {
      "epoch": 0.17858823529411766,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9293,
      "step": 7590
    },
    {
      "epoch": 0.17861176470588236,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.2401,
      "step": 7591
    },
    {
      "epoch": 0.17863529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1601,
      "step": 7592
    },
    {
      "epoch": 0.17865882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0682,
      "step": 7593
    },
    {
      "epoch": 0.17868235294117646,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.1258,
      "step": 7594
    },
    {
      "epoch": 0.1787058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4466,
      "step": 7595
    },
    {
      "epoch": 0.1787294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2431,
      "step": 7596
    },
    {
      "epoch": 0.1787529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0777,
      "step": 7597
    },
    {
      "epoch": 0.1787764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9297,
      "step": 7598
    },
    {
      "epoch": 0.1788,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0187,
      "step": 7599
    },
    {
      "epoch": 0.17882352941176471,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1345,
      "step": 7600
    },
    {
      "epoch": 0.17884705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3087,
      "step": 7601
    },
    {
      "epoch": 0.17887058823529411,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2774,
      "step": 7602
    },
    {
      "epoch": 0.17889411764705881,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 7603
    },
    {
      "epoch": 0.17891764705882354,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2249,
      "step": 7604
    },
    {
      "epoch": 0.17894117647058824,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9524,
      "step": 7605
    },
    {
      "epoch": 0.17896470588235294,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 7606
    },
    {
      "epoch": 0.17898823529411764,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1037,
      "step": 7607
    },
    {
      "epoch": 0.17901176470588234,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 1.003,
      "step": 7608
    },
    {
      "epoch": 0.17903529411764707,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2255,
      "step": 7609
    },
    {
      "epoch": 0.17905882352941177,
      "grad_norm": 0.2890625,
      "learning_rate": 0.02,
      "loss": 0.8339,
      "step": 7610
    },
    {
      "epoch": 0.17908235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2992,
      "step": 7611
    },
    {
      "epoch": 0.17910588235294117,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0538,
      "step": 7612
    },
    {
      "epoch": 0.17912941176470587,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0796,
      "step": 7613
    },
    {
      "epoch": 0.1791529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9659,
      "step": 7614
    },
    {
      "epoch": 0.1791764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0834,
      "step": 7615
    },
    {
      "epoch": 0.1792,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.9294,
      "step": 7616
    },
    {
      "epoch": 0.1792235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1115,
      "step": 7617
    },
    {
      "epoch": 0.1792470588235294,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0272,
      "step": 7618
    },
    {
      "epoch": 0.17927058823529413,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.223,
      "step": 7619
    },
    {
      "epoch": 0.17929411764705883,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1095,
      "step": 7620
    },
    {
      "epoch": 0.17931764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3545,
      "step": 7621
    },
    {
      "epoch": 0.17934117647058823,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3973,
      "step": 7622
    },
    {
      "epoch": 0.17936470588235295,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 7623
    },
    {
      "epoch": 0.17938823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.235,
      "step": 7624
    },
    {
      "epoch": 0.17941176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0815,
      "step": 7625
    },
    {
      "epoch": 0.17943529411764705,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.5665,
      "step": 7626
    },
    {
      "epoch": 0.17945882352941175,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3757,
      "step": 7627
    },
    {
      "epoch": 0.17948235294117648,
      "grad_norm": 0.27734375,
      "learning_rate": 0.02,
      "loss": 0.9,
      "step": 7628
    },
    {
      "epoch": 0.17950588235294118,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2628,
      "step": 7629
    },
    {
      "epoch": 0.17952941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.094,
      "step": 7630
    },
    {
      "epoch": 0.17955294117647058,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2668,
      "step": 7631
    },
    {
      "epoch": 0.17957647058823528,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7604,
      "step": 7632
    },
    {
      "epoch": 0.1796,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1627,
      "step": 7633
    },
    {
      "epoch": 0.1796235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.283,
      "step": 7634
    },
    {
      "epoch": 0.1796470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2438,
      "step": 7635
    },
    {
      "epoch": 0.1796705882352941,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9677,
      "step": 7636
    },
    {
      "epoch": 0.1796941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2182,
      "step": 7637
    },
    {
      "epoch": 0.17971764705882354,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9102,
      "step": 7638
    },
    {
      "epoch": 0.17974117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2189,
      "step": 7639
    },
    {
      "epoch": 0.17976470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3009,
      "step": 7640
    },
    {
      "epoch": 0.17978823529411764,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.0676,
      "step": 7641
    },
    {
      "epoch": 0.17981176470588237,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1564,
      "step": 7642
    },
    {
      "epoch": 0.17983529411764707,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.2043,
      "step": 7643
    },
    {
      "epoch": 0.17985882352941177,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0503,
      "step": 7644
    },
    {
      "epoch": 0.17988235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2235,
      "step": 7645
    },
    {
      "epoch": 0.17990588235294117,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0333,
      "step": 7646
    },
    {
      "epoch": 0.1799294117647059,
      "grad_norm": 0.28515625,
      "learning_rate": 0.02,
      "loss": 0.6261,
      "step": 7647
    },
    {
      "epoch": 0.1799529411764706,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0508,
      "step": 7648
    },
    {
      "epoch": 0.1799764705882353,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.8493,
      "step": 7649
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1014,
      "step": 7650
    },
    {
      "epoch": 0.1800235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0625,
      "step": 7651
    },
    {
      "epoch": 0.18004705882352942,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.7912,
      "step": 7652
    },
    {
      "epoch": 0.18007058823529412,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0721,
      "step": 7653
    },
    {
      "epoch": 0.18009411764705882,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0504,
      "step": 7654
    },
    {
      "epoch": 0.18011764705882352,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1085,
      "step": 7655
    },
    {
      "epoch": 0.18014117647058825,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1726,
      "step": 7656
    },
    {
      "epoch": 0.18016470588235295,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3185,
      "step": 7657
    },
    {
      "epoch": 0.18018823529411765,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0527,
      "step": 7658
    },
    {
      "epoch": 0.18021176470588235,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.142,
      "step": 7659
    },
    {
      "epoch": 0.18023529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2355,
      "step": 7660
    },
    {
      "epoch": 0.18025882352941178,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8339,
      "step": 7661
    },
    {
      "epoch": 0.18028235294117648,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2018,
      "step": 7662
    },
    {
      "epoch": 0.18030588235294118,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0802,
      "step": 7663
    },
    {
      "epoch": 0.18032941176470588,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8887,
      "step": 7664
    },
    {
      "epoch": 0.18035294117647058,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9913,
      "step": 7665
    },
    {
      "epoch": 0.1803764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1396,
      "step": 7666
    },
    {
      "epoch": 0.1804,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2329,
      "step": 7667
    },
    {
      "epoch": 0.1804235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0076,
      "step": 7668
    },
    {
      "epoch": 0.1804470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2688,
      "step": 7669
    },
    {
      "epoch": 0.1804705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 7670
    },
    {
      "epoch": 0.18049411764705883,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3419,
      "step": 7671
    },
    {
      "epoch": 0.18051764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1175,
      "step": 7672
    },
    {
      "epoch": 0.18054117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0072,
      "step": 7673
    },
    {
      "epoch": 0.18056470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3413,
      "step": 7674
    },
    {
      "epoch": 0.18058823529411766,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.1827,
      "step": 7675
    },
    {
      "epoch": 0.18061176470588236,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.005,
      "step": 7676
    },
    {
      "epoch": 0.18063529411764706,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 7677
    },
    {
      "epoch": 0.18065882352941176,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8523,
      "step": 7678
    },
    {
      "epoch": 0.18068235294117646,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2206,
      "step": 7679
    },
    {
      "epoch": 0.1807058823529412,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9366,
      "step": 7680
    },
    {
      "epoch": 0.1807294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0909,
      "step": 7681
    },
    {
      "epoch": 0.1807529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.8955,
      "step": 7682
    },
    {
      "epoch": 0.1807764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1786,
      "step": 7683
    },
    {
      "epoch": 0.1808,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8438,
      "step": 7684
    },
    {
      "epoch": 0.18082352941176472,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0545,
      "step": 7685
    },
    {
      "epoch": 0.18084705882352942,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9157,
      "step": 7686
    },
    {
      "epoch": 0.18087058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 7687
    },
    {
      "epoch": 0.18089411764705882,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7661,
      "step": 7688
    },
    {
      "epoch": 0.18091764705882352,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1256,
      "step": 7689
    },
    {
      "epoch": 0.18094117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.3029,
      "step": 7690
    },
    {
      "epoch": 0.18096470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9663,
      "step": 7691
    },
    {
      "epoch": 0.18098823529411764,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.288,
      "step": 7692
    },
    {
      "epoch": 0.18101176470588234,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9799,
      "step": 7693
    },
    {
      "epoch": 0.18103529411764707,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0887,
      "step": 7694
    },
    {
      "epoch": 0.18105882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2539,
      "step": 7695
    },
    {
      "epoch": 0.18108235294117647,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 7696
    },
    {
      "epoch": 0.18110588235294117,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3497,
      "step": 7697
    },
    {
      "epoch": 0.18112941176470587,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2578,
      "step": 7698
    },
    {
      "epoch": 0.1811529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.335,
      "step": 7699
    },
    {
      "epoch": 0.1811764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 7700
    },
    {
      "epoch": 0.1812,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1957,
      "step": 7701
    },
    {
      "epoch": 0.1812235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1545,
      "step": 7702
    },
    {
      "epoch": 0.1812470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1893,
      "step": 7703
    },
    {
      "epoch": 0.18127058823529413,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.8587,
      "step": 7704
    },
    {
      "epoch": 0.18129411764705883,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3059,
      "step": 7705
    },
    {
      "epoch": 0.18131764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.25,
      "step": 7706
    },
    {
      "epoch": 0.18134117647058823,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.8441,
      "step": 7707
    },
    {
      "epoch": 0.18136470588235293,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9792,
      "step": 7708
    },
    {
      "epoch": 0.18138823529411766,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0817,
      "step": 7709
    },
    {
      "epoch": 0.18141176470588236,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2328,
      "step": 7710
    },
    {
      "epoch": 0.18143529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3538,
      "step": 7711
    },
    {
      "epoch": 0.18145882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.216,
      "step": 7712
    },
    {
      "epoch": 0.18148235294117648,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3917,
      "step": 7713
    },
    {
      "epoch": 0.18150588235294118,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1715,
      "step": 7714
    },
    {
      "epoch": 0.18152941176470588,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0426,
      "step": 7715
    },
    {
      "epoch": 0.18155294117647058,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8303,
      "step": 7716
    },
    {
      "epoch": 0.18157647058823528,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0972,
      "step": 7717
    },
    {
      "epoch": 0.1816,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9383,
      "step": 7718
    },
    {
      "epoch": 0.1816235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2059,
      "step": 7719
    },
    {
      "epoch": 0.1816470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0905,
      "step": 7720
    },
    {
      "epoch": 0.1816705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 7721
    },
    {
      "epoch": 0.1816941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0945,
      "step": 7722
    },
    {
      "epoch": 0.18171764705882354,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 7723
    },
    {
      "epoch": 0.18174117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2525,
      "step": 7724
    },
    {
      "epoch": 0.18176470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 7725
    },
    {
      "epoch": 0.18178823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2811,
      "step": 7726
    },
    {
      "epoch": 0.18181176470588234,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.177,
      "step": 7727
    },
    {
      "epoch": 0.18183529411764707,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0491,
      "step": 7728
    },
    {
      "epoch": 0.18185882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3752,
      "step": 7729
    },
    {
      "epoch": 0.18188235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2543,
      "step": 7730
    },
    {
      "epoch": 0.18190588235294117,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0131,
      "step": 7731
    },
    {
      "epoch": 0.1819294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2727,
      "step": 7732
    },
    {
      "epoch": 0.1819529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2109,
      "step": 7733
    },
    {
      "epoch": 0.1819764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 7734
    },
    {
      "epoch": 0.182,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4288,
      "step": 7735
    },
    {
      "epoch": 0.1820235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2972,
      "step": 7736
    },
    {
      "epoch": 0.18204705882352942,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.311,
      "step": 7737
    },
    {
      "epoch": 0.18207058823529412,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.8686,
      "step": 7738
    },
    {
      "epoch": 0.18209411764705882,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2979,
      "step": 7739
    },
    {
      "epoch": 0.18211764705882352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0604,
      "step": 7740
    },
    {
      "epoch": 0.18214117647058822,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 7741
    },
    {
      "epoch": 0.18216470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.4422,
      "step": 7742
    },
    {
      "epoch": 0.18218823529411765,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.7166,
      "step": 7743
    },
    {
      "epoch": 0.18221176470588235,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.8445,
      "step": 7744
    },
    {
      "epoch": 0.18223529411764705,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.988,
      "step": 7745
    },
    {
      "epoch": 0.18225882352941175,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0987,
      "step": 7746
    },
    {
      "epoch": 0.18228235294117648,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3583,
      "step": 7747
    },
    {
      "epoch": 0.18230588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2366,
      "step": 7748
    },
    {
      "epoch": 0.18232941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3922,
      "step": 7749
    },
    {
      "epoch": 0.18235294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2731,
      "step": 7750
    },
    {
      "epoch": 0.1823764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.4536,
      "step": 7751
    },
    {
      "epoch": 0.1824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0253,
      "step": 7752
    },
    {
      "epoch": 0.1824235294117647,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.6429,
      "step": 7753
    },
    {
      "epoch": 0.1824470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.2438,
      "step": 7754
    },
    {
      "epoch": 0.1824705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3374,
      "step": 7755
    },
    {
      "epoch": 0.18249411764705883,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0682,
      "step": 7756
    },
    {
      "epoch": 0.18251764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.111,
      "step": 7757
    },
    {
      "epoch": 0.18254117647058823,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9966,
      "step": 7758
    },
    {
      "epoch": 0.18256470588235293,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.3275,
      "step": 7759
    },
    {
      "epoch": 0.18258823529411763,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.2337,
      "step": 7760
    },
    {
      "epoch": 0.18261176470588236,
      "grad_norm": 0.310546875,
      "learning_rate": 0.02,
      "loss": 0.6701,
      "step": 7761
    },
    {
      "epoch": 0.18263529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 7762
    },
    {
      "epoch": 0.18265882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2292,
      "step": 7763
    },
    {
      "epoch": 0.18268235294117646,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2634,
      "step": 7764
    },
    {
      "epoch": 0.1827058823529412,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0114,
      "step": 7765
    },
    {
      "epoch": 0.1827294117647059,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9907,
      "step": 7766
    },
    {
      "epoch": 0.1827529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2009,
      "step": 7767
    },
    {
      "epoch": 0.1827764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2067,
      "step": 7768
    },
    {
      "epoch": 0.1828,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3838,
      "step": 7769
    },
    {
      "epoch": 0.18282352941176472,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3671,
      "step": 7770
    },
    {
      "epoch": 0.18284705882352942,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2585,
      "step": 7771
    },
    {
      "epoch": 0.18287058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2154,
      "step": 7772
    },
    {
      "epoch": 0.18289411764705882,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 7773
    },
    {
      "epoch": 0.18291764705882352,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1461,
      "step": 7774
    },
    {
      "epoch": 0.18294117647058825,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9473,
      "step": 7775
    },
    {
      "epoch": 0.18296470588235295,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3001,
      "step": 7776
    },
    {
      "epoch": 0.18298823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1804,
      "step": 7777
    },
    {
      "epoch": 0.18301176470588235,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0077,
      "step": 7778
    },
    {
      "epoch": 0.18303529411764705,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7563,
      "step": 7779
    },
    {
      "epoch": 0.18305882352941177,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0146,
      "step": 7780
    },
    {
      "epoch": 0.18308235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0767,
      "step": 7781
    },
    {
      "epoch": 0.18310588235294117,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 1.0168,
      "step": 7782
    },
    {
      "epoch": 0.18312941176470587,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0065,
      "step": 7783
    },
    {
      "epoch": 0.1831529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0829,
      "step": 7784
    },
    {
      "epoch": 0.1831764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9874,
      "step": 7785
    },
    {
      "epoch": 0.1832,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0839,
      "step": 7786
    },
    {
      "epoch": 0.1832235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1589,
      "step": 7787
    },
    {
      "epoch": 0.1832470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3139,
      "step": 7788
    },
    {
      "epoch": 0.18327058823529413,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.7273,
      "step": 7789
    },
    {
      "epoch": 0.18329411764705883,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0367,
      "step": 7790
    },
    {
      "epoch": 0.18331764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3332,
      "step": 7791
    },
    {
      "epoch": 0.18334117647058823,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.114,
      "step": 7792
    },
    {
      "epoch": 0.18336470588235293,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0531,
      "step": 7793
    },
    {
      "epoch": 0.18338823529411766,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1379,
      "step": 7794
    },
    {
      "epoch": 0.18341176470588236,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.311,
      "step": 7795
    },
    {
      "epoch": 0.18343529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3922,
      "step": 7796
    },
    {
      "epoch": 0.18345882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.141,
      "step": 7797
    },
    {
      "epoch": 0.18348235294117646,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9049,
      "step": 7798
    },
    {
      "epoch": 0.18350588235294119,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2187,
      "step": 7799
    },
    {
      "epoch": 0.18352941176470589,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2022,
      "step": 7800
    },
    {
      "epoch": 0.18355294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2497,
      "step": 7801
    },
    {
      "epoch": 0.18357647058823529,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9281,
      "step": 7802
    },
    {
      "epoch": 0.1836,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0983,
      "step": 7803
    },
    {
      "epoch": 0.1836235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 7804
    },
    {
      "epoch": 0.1836470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3129,
      "step": 7805
    },
    {
      "epoch": 0.1836705882352941,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.982,
      "step": 7806
    },
    {
      "epoch": 0.1836941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1861,
      "step": 7807
    },
    {
      "epoch": 0.18371764705882354,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1695,
      "step": 7808
    },
    {
      "epoch": 0.18374117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.089,
      "step": 7809
    },
    {
      "epoch": 0.18376470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2376,
      "step": 7810
    },
    {
      "epoch": 0.18378823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1459,
      "step": 7811
    },
    {
      "epoch": 0.18381176470588234,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2227,
      "step": 7812
    },
    {
      "epoch": 0.18383529411764707,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0208,
      "step": 7813
    },
    {
      "epoch": 0.18385882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 7814
    },
    {
      "epoch": 0.18388235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2044,
      "step": 7815
    },
    {
      "epoch": 0.18390588235294117,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1993,
      "step": 7816
    },
    {
      "epoch": 0.18392941176470587,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.6076,
      "step": 7817
    },
    {
      "epoch": 0.1839529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2801,
      "step": 7818
    },
    {
      "epoch": 0.1839764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 7819
    },
    {
      "epoch": 0.184,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0316,
      "step": 7820
    },
    {
      "epoch": 0.1840235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3641,
      "step": 7821
    },
    {
      "epoch": 0.18404705882352942,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1456,
      "step": 7822
    },
    {
      "epoch": 0.18407058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1761,
      "step": 7823
    },
    {
      "epoch": 0.18409411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3959,
      "step": 7824
    },
    {
      "epoch": 0.18411764705882352,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1507,
      "step": 7825
    },
    {
      "epoch": 0.18414117647058822,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0214,
      "step": 7826
    },
    {
      "epoch": 0.18416470588235295,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1518,
      "step": 7827
    },
    {
      "epoch": 0.18418823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3701,
      "step": 7828
    },
    {
      "epoch": 0.18421176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8368,
      "step": 7829
    },
    {
      "epoch": 0.18423529411764705,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.7909,
      "step": 7830
    },
    {
      "epoch": 0.18425882352941175,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0279,
      "step": 7831
    },
    {
      "epoch": 0.18428235294117648,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 7832
    },
    {
      "epoch": 0.18430588235294118,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2338,
      "step": 7833
    },
    {
      "epoch": 0.18432941176470588,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 7834
    },
    {
      "epoch": 0.18435294117647058,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9388,
      "step": 7835
    },
    {
      "epoch": 0.18437647058823528,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.989,
      "step": 7836
    },
    {
      "epoch": 0.1844,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4619,
      "step": 7837
    },
    {
      "epoch": 0.1844235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1724,
      "step": 7838
    },
    {
      "epoch": 0.1844470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1984,
      "step": 7839
    },
    {
      "epoch": 0.1844705882352941,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8647,
      "step": 7840
    },
    {
      "epoch": 0.18449411764705884,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2102,
      "step": 7841
    },
    {
      "epoch": 0.18451764705882354,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2823,
      "step": 7842
    },
    {
      "epoch": 0.18454117647058824,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0657,
      "step": 7843
    },
    {
      "epoch": 0.18456470588235294,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0179,
      "step": 7844
    },
    {
      "epoch": 0.18458823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3597,
      "step": 7845
    },
    {
      "epoch": 0.18461176470588236,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0419,
      "step": 7846
    },
    {
      "epoch": 0.18463529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9274,
      "step": 7847
    },
    {
      "epoch": 0.18465882352941176,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0106,
      "step": 7848
    },
    {
      "epoch": 0.18468235294117646,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9949,
      "step": 7849
    },
    {
      "epoch": 0.18470588235294116,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2592,
      "step": 7850
    },
    {
      "epoch": 0.1847294117647059,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0254,
      "step": 7851
    },
    {
      "epoch": 0.1847529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.333,
      "step": 7852
    },
    {
      "epoch": 0.1847764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 7853
    },
    {
      "epoch": 0.1848,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9995,
      "step": 7854
    },
    {
      "epoch": 0.1848235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2508,
      "step": 7855
    },
    {
      "epoch": 0.18484705882352942,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7847,
      "step": 7856
    },
    {
      "epoch": 0.18487058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1032,
      "step": 7857
    },
    {
      "epoch": 0.18489411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.896,
      "step": 7858
    },
    {
      "epoch": 0.18491764705882352,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1308,
      "step": 7859
    },
    {
      "epoch": 0.18494117647058825,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 7860
    },
    {
      "epoch": 0.18496470588235295,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9318,
      "step": 7861
    },
    {
      "epoch": 0.18498823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2778,
      "step": 7862
    },
    {
      "epoch": 0.18501176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 7863
    },
    {
      "epoch": 0.18503529411764705,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0728,
      "step": 7864
    },
    {
      "epoch": 0.18505882352941178,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1116,
      "step": 7865
    },
    {
      "epoch": 0.18508235294117648,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8735,
      "step": 7866
    },
    {
      "epoch": 0.18510588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3753,
      "step": 7867
    },
    {
      "epoch": 0.18512941176470588,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3535,
      "step": 7868
    },
    {
      "epoch": 0.18515294117647058,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1626,
      "step": 7869
    },
    {
      "epoch": 0.1851764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.169,
      "step": 7870
    },
    {
      "epoch": 0.1852,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1625,
      "step": 7871
    },
    {
      "epoch": 0.1852235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9511,
      "step": 7872
    },
    {
      "epoch": 0.1852470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.5829,
      "step": 7873
    },
    {
      "epoch": 0.18527058823529413,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3286,
      "step": 7874
    },
    {
      "epoch": 0.18529411764705883,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 7875
    },
    {
      "epoch": 0.18531764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3016,
      "step": 7876
    },
    {
      "epoch": 0.18534117647058823,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9326,
      "step": 7877
    },
    {
      "epoch": 0.18536470588235293,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3932,
      "step": 7878
    },
    {
      "epoch": 0.18538823529411766,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9714,
      "step": 7879
    },
    {
      "epoch": 0.18541176470588236,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0106,
      "step": 7880
    },
    {
      "epoch": 0.18543529411764706,
      "grad_norm": 0.287109375,
      "learning_rate": 0.02,
      "loss": 0.6407,
      "step": 7881
    },
    {
      "epoch": 0.18545882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.245,
      "step": 7882
    },
    {
      "epoch": 0.18548235294117646,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 7883
    },
    {
      "epoch": 0.1855058823529412,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.9586,
      "step": 7884
    },
    {
      "epoch": 0.1855294117647059,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0471,
      "step": 7885
    },
    {
      "epoch": 0.1855529411764706,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0153,
      "step": 7886
    },
    {
      "epoch": 0.1855764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0588,
      "step": 7887
    },
    {
      "epoch": 0.1856,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0217,
      "step": 7888
    },
    {
      "epoch": 0.18562352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0216,
      "step": 7889
    },
    {
      "epoch": 0.18564705882352942,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.3042,
      "step": 7890
    },
    {
      "epoch": 0.18567058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0901,
      "step": 7891
    },
    {
      "epoch": 0.18569411764705882,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1352,
      "step": 7892
    },
    {
      "epoch": 0.18571764705882354,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1644,
      "step": 7893
    },
    {
      "epoch": 0.18574117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1812,
      "step": 7894
    },
    {
      "epoch": 0.18576470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1722,
      "step": 7895
    },
    {
      "epoch": 0.18578823529411764,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.8431,
      "step": 7896
    },
    {
      "epoch": 0.18581176470588234,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 7897
    },
    {
      "epoch": 0.18583529411764707,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.7216,
      "step": 7898
    },
    {
      "epoch": 0.18585882352941177,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9513,
      "step": 7899
    },
    {
      "epoch": 0.18588235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.359,
      "step": 7900
    },
    {
      "epoch": 0.18590588235294117,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9264,
      "step": 7901
    },
    {
      "epoch": 0.18592941176470587,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1886,
      "step": 7902
    },
    {
      "epoch": 0.1859529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1964,
      "step": 7903
    },
    {
      "epoch": 0.1859764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2367,
      "step": 7904
    },
    {
      "epoch": 0.186,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9468,
      "step": 7905
    },
    {
      "epoch": 0.1860235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9275,
      "step": 7906
    },
    {
      "epoch": 0.1860470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 7907
    },
    {
      "epoch": 0.18607058823529413,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0511,
      "step": 7908
    },
    {
      "epoch": 0.18609411764705883,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1056,
      "step": 7909
    },
    {
      "epoch": 0.18611764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9773,
      "step": 7910
    },
    {
      "epoch": 0.18614117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2441,
      "step": 7911
    },
    {
      "epoch": 0.18616470588235295,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7528,
      "step": 7912
    },
    {
      "epoch": 0.18618823529411765,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0006,
      "step": 7913
    },
    {
      "epoch": 0.18621176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0981,
      "step": 7914
    },
    {
      "epoch": 0.18623529411764705,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.082,
      "step": 7915
    },
    {
      "epoch": 0.18625882352941175,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3688,
      "step": 7916
    },
    {
      "epoch": 0.18628235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1577,
      "step": 7917
    },
    {
      "epoch": 0.18630588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8395,
      "step": 7918
    },
    {
      "epoch": 0.18632941176470588,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0321,
      "step": 7919
    },
    {
      "epoch": 0.18635294117647058,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.984,
      "step": 7920
    },
    {
      "epoch": 0.18637647058823528,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0744,
      "step": 7921
    },
    {
      "epoch": 0.1864,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2812,
      "step": 7922
    },
    {
      "epoch": 0.1864235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.399,
      "step": 7923
    },
    {
      "epoch": 0.1864470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1503,
      "step": 7924
    },
    {
      "epoch": 0.1864705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9542,
      "step": 7925
    },
    {
      "epoch": 0.1864941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2761,
      "step": 7926
    },
    {
      "epoch": 0.18651764705882354,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8009,
      "step": 7927
    },
    {
      "epoch": 0.18654117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2735,
      "step": 7928
    },
    {
      "epoch": 0.18656470588235294,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9512,
      "step": 7929
    },
    {
      "epoch": 0.18658823529411764,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9632,
      "step": 7930
    },
    {
      "epoch": 0.18661176470588237,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1288,
      "step": 7931
    },
    {
      "epoch": 0.18663529411764707,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0019,
      "step": 7932
    },
    {
      "epoch": 0.18665882352941177,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0086,
      "step": 7933
    },
    {
      "epoch": 0.18668235294117647,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9612,
      "step": 7934
    },
    {
      "epoch": 0.18670588235294117,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8931,
      "step": 7935
    },
    {
      "epoch": 0.1867294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.4024,
      "step": 7936
    },
    {
      "epoch": 0.1867529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8947,
      "step": 7937
    },
    {
      "epoch": 0.1867764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.744,
      "step": 7938
    },
    {
      "epoch": 0.1868,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2215,
      "step": 7939
    },
    {
      "epoch": 0.1868235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1887,
      "step": 7940
    },
    {
      "epoch": 0.18684705882352942,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1105,
      "step": 7941
    },
    {
      "epoch": 0.18687058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3665,
      "step": 7942
    },
    {
      "epoch": 0.18689411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.4248,
      "step": 7943
    },
    {
      "epoch": 0.18691764705882352,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2927,
      "step": 7944
    },
    {
      "epoch": 0.18694117647058822,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8137,
      "step": 7945
    },
    {
      "epoch": 0.18696470588235295,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4314,
      "step": 7946
    },
    {
      "epoch": 0.18698823529411765,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 1.0531,
      "step": 7947
    },
    {
      "epoch": 0.18701176470588235,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 7948
    },
    {
      "epoch": 0.18703529411764705,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9382,
      "step": 7949
    },
    {
      "epoch": 0.18705882352941178,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9036,
      "step": 7950
    },
    {
      "epoch": 0.18708235294117648,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0504,
      "step": 7951
    },
    {
      "epoch": 0.18710588235294118,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8212,
      "step": 7952
    },
    {
      "epoch": 0.18712941176470588,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8858,
      "step": 7953
    },
    {
      "epoch": 0.18715294117647058,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2185,
      "step": 7954
    },
    {
      "epoch": 0.1871764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.27,
      "step": 7955
    },
    {
      "epoch": 0.1872,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0501,
      "step": 7956
    },
    {
      "epoch": 0.1872235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 7957
    },
    {
      "epoch": 0.1872470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8291,
      "step": 7958
    },
    {
      "epoch": 0.1872705882352941,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0123,
      "step": 7959
    },
    {
      "epoch": 0.18729411764705883,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 7960
    },
    {
      "epoch": 0.18731764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8727,
      "step": 7961
    },
    {
      "epoch": 0.18734117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1924,
      "step": 7962
    },
    {
      "epoch": 0.18736470588235293,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8738,
      "step": 7963
    },
    {
      "epoch": 0.18738823529411763,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.6559,
      "step": 7964
    },
    {
      "epoch": 0.18741176470588236,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.019,
      "step": 7965
    },
    {
      "epoch": 0.18743529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 7966
    },
    {
      "epoch": 0.18745882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1842,
      "step": 7967
    },
    {
      "epoch": 0.18748235294117646,
      "grad_norm": 0.294921875,
      "learning_rate": 0.02,
      "loss": 0.5633,
      "step": 7968
    },
    {
      "epoch": 0.1875058823529412,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.6617,
      "step": 7969
    },
    {
      "epoch": 0.1875294117647059,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0548,
      "step": 7970
    },
    {
      "epoch": 0.1875529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1211,
      "step": 7971
    },
    {
      "epoch": 0.1875764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0153,
      "step": 7972
    },
    {
      "epoch": 0.1876,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1804,
      "step": 7973
    },
    {
      "epoch": 0.18762352941176472,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9217,
      "step": 7974
    },
    {
      "epoch": 0.18764705882352942,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0376,
      "step": 7975
    },
    {
      "epoch": 0.18767058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3113,
      "step": 7976
    },
    {
      "epoch": 0.18769411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 7977
    },
    {
      "epoch": 0.18771764705882352,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.107,
      "step": 7978
    },
    {
      "epoch": 0.18774117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8167,
      "step": 7979
    },
    {
      "epoch": 0.18776470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2496,
      "step": 7980
    },
    {
      "epoch": 0.18778823529411764,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.952,
      "step": 7981
    },
    {
      "epoch": 0.18781176470588234,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0707,
      "step": 7982
    },
    {
      "epoch": 0.18783529411764707,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.7749,
      "step": 7983
    },
    {
      "epoch": 0.18785882352941177,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2702,
      "step": 7984
    },
    {
      "epoch": 0.18788235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1994,
      "step": 7985
    },
    {
      "epoch": 0.18790588235294117,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.7469,
      "step": 7986
    },
    {
      "epoch": 0.18792941176470587,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8918,
      "step": 7987
    },
    {
      "epoch": 0.1879529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 7988
    },
    {
      "epoch": 0.1879764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.7947,
      "step": 7989
    },
    {
      "epoch": 0.188,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2851,
      "step": 7990
    },
    {
      "epoch": 0.1880235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8793,
      "step": 7991
    },
    {
      "epoch": 0.1880470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1074,
      "step": 7992
    },
    {
      "epoch": 0.18807058823529413,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0252,
      "step": 7993
    },
    {
      "epoch": 0.18809411764705883,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9613,
      "step": 7994
    },
    {
      "epoch": 0.18811764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1617,
      "step": 7995
    },
    {
      "epoch": 0.18814117647058823,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.7004,
      "step": 7996
    },
    {
      "epoch": 0.18816470588235293,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1697,
      "step": 7997
    },
    {
      "epoch": 0.18818823529411766,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8933,
      "step": 7998
    },
    {
      "epoch": 0.18821176470588236,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8488,
      "step": 7999
    },
    {
      "epoch": 0.18823529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3995,
      "step": 8000
    },
    {
      "epoch": 0.18823529411764706,
      "eval_loss": 2.22733736038208,
      "eval_runtime": 681.8209,
      "eval_samples_per_second": 12.467,
      "eval_steps_per_second": 3.117,
      "step": 8000
    },
    {
      "epoch": 0.18825882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2597,
      "step": 8001
    },
    {
      "epoch": 0.18828235294117648,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0184,
      "step": 8002
    },
    {
      "epoch": 0.18830588235294118,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0399,
      "step": 8003
    },
    {
      "epoch": 0.18832941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.319,
      "step": 8004
    },
    {
      "epoch": 0.18835294117647058,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0124,
      "step": 8005
    },
    {
      "epoch": 0.18837647058823528,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1241,
      "step": 8006
    },
    {
      "epoch": 0.1884,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1945,
      "step": 8007
    },
    {
      "epoch": 0.1884235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3156,
      "step": 8008
    },
    {
      "epoch": 0.1884470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2599,
      "step": 8009
    },
    {
      "epoch": 0.1884705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.006,
      "step": 8010
    },
    {
      "epoch": 0.1884941176470588,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.09,
      "step": 8011
    },
    {
      "epoch": 0.18851764705882354,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0945,
      "step": 8012
    },
    {
      "epoch": 0.18854117647058824,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.7707,
      "step": 8013
    },
    {
      "epoch": 0.18856470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 8014
    },
    {
      "epoch": 0.18858823529411764,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4409,
      "step": 8015
    },
    {
      "epoch": 0.18861176470588234,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2759,
      "step": 8016
    },
    {
      "epoch": 0.18863529411764707,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3461,
      "step": 8017
    },
    {
      "epoch": 0.18865882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.3604,
      "step": 8018
    },
    {
      "epoch": 0.18868235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0227,
      "step": 8019
    },
    {
      "epoch": 0.18870588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4038,
      "step": 8020
    },
    {
      "epoch": 0.1887294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2434,
      "step": 8021
    },
    {
      "epoch": 0.1887529411764706,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9571,
      "step": 8022
    },
    {
      "epoch": 0.1887764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1479,
      "step": 8023
    },
    {
      "epoch": 0.1888,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2509,
      "step": 8024
    },
    {
      "epoch": 0.1888235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3035,
      "step": 8025
    },
    {
      "epoch": 0.18884705882352942,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9743,
      "step": 8026
    },
    {
      "epoch": 0.18887058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0636,
      "step": 8027
    },
    {
      "epoch": 0.18889411764705882,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 1.0824,
      "step": 8028
    },
    {
      "epoch": 0.18891764705882352,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.7221,
      "step": 8029
    },
    {
      "epoch": 0.18894117647058822,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 1.0579,
      "step": 8030
    },
    {
      "epoch": 0.18896470588235295,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8576,
      "step": 8031
    },
    {
      "epoch": 0.18898823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 8032
    },
    {
      "epoch": 0.18901176470588235,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9538,
      "step": 8033
    },
    {
      "epoch": 0.18903529411764705,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1808,
      "step": 8034
    },
    {
      "epoch": 0.18905882352941175,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0883,
      "step": 8035
    },
    {
      "epoch": 0.18908235294117648,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.0312,
      "step": 8036
    },
    {
      "epoch": 0.18910588235294118,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0952,
      "step": 8037
    },
    {
      "epoch": 0.18912941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.045,
      "step": 8038
    },
    {
      "epoch": 0.18915294117647058,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 8039
    },
    {
      "epoch": 0.1891764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2727,
      "step": 8040
    },
    {
      "epoch": 0.1892,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3448,
      "step": 8041
    },
    {
      "epoch": 0.1892235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9929,
      "step": 8042
    },
    {
      "epoch": 0.1892470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3053,
      "step": 8043
    },
    {
      "epoch": 0.1892705882352941,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8974,
      "step": 8044
    },
    {
      "epoch": 0.18929411764705883,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1886,
      "step": 8045
    },
    {
      "epoch": 0.18931764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1067,
      "step": 8046
    },
    {
      "epoch": 0.18934117647058823,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.99,
      "step": 8047
    },
    {
      "epoch": 0.18936470588235293,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 8048
    },
    {
      "epoch": 0.18938823529411764,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.874,
      "step": 8049
    },
    {
      "epoch": 0.18941176470588236,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2379,
      "step": 8050
    },
    {
      "epoch": 0.18943529411764706,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.078,
      "step": 8051
    },
    {
      "epoch": 0.18945882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 8052
    },
    {
      "epoch": 0.18948235294117646,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0383,
      "step": 8053
    },
    {
      "epoch": 0.18950588235294116,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3891,
      "step": 8054
    },
    {
      "epoch": 0.1895294117647059,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8437,
      "step": 8055
    },
    {
      "epoch": 0.1895529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1486,
      "step": 8056
    },
    {
      "epoch": 0.1895764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4342,
      "step": 8057
    },
    {
      "epoch": 0.1896,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8377,
      "step": 8058
    },
    {
      "epoch": 0.18962352941176472,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8619,
      "step": 8059
    },
    {
      "epoch": 0.18964705882352942,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0163,
      "step": 8060
    },
    {
      "epoch": 0.18967058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2349,
      "step": 8061
    },
    {
      "epoch": 0.18969411764705882,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.6922,
      "step": 8062
    },
    {
      "epoch": 0.18971764705882352,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8937,
      "step": 8063
    },
    {
      "epoch": 0.18974117647058825,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1955,
      "step": 8064
    },
    {
      "epoch": 0.18976470588235295,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9719,
      "step": 8065
    },
    {
      "epoch": 0.18978823529411765,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9838,
      "step": 8066
    },
    {
      "epoch": 0.18981176470588235,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.6091,
      "step": 8067
    },
    {
      "epoch": 0.18983529411764705,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.5902,
      "step": 8068
    },
    {
      "epoch": 0.18985882352941177,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1364,
      "step": 8069
    },
    {
      "epoch": 0.18988235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8778,
      "step": 8070
    },
    {
      "epoch": 0.18990588235294117,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0188,
      "step": 8071
    },
    {
      "epoch": 0.18992941176470587,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.25,
      "step": 8072
    },
    {
      "epoch": 0.18995294117647057,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0975,
      "step": 8073
    },
    {
      "epoch": 0.1899764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0536,
      "step": 8074
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0065,
      "step": 8075
    },
    {
      "epoch": 0.1900235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8838,
      "step": 8076
    },
    {
      "epoch": 0.1900470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0142,
      "step": 8077
    },
    {
      "epoch": 0.19007058823529413,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0166,
      "step": 8078
    },
    {
      "epoch": 0.19009411764705883,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3846,
      "step": 8079
    },
    {
      "epoch": 0.19011764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3329,
      "step": 8080
    },
    {
      "epoch": 0.19014117647058823,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2006,
      "step": 8081
    },
    {
      "epoch": 0.19016470588235293,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8083,
      "step": 8082
    },
    {
      "epoch": 0.19018823529411766,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0877,
      "step": 8083
    },
    {
      "epoch": 0.19021176470588236,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9526,
      "step": 8084
    },
    {
      "epoch": 0.19023529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8309,
      "step": 8085
    },
    {
      "epoch": 0.19025882352941176,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1652,
      "step": 8086
    },
    {
      "epoch": 0.19028235294117646,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 1.0343,
      "step": 8087
    },
    {
      "epoch": 0.19030588235294119,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0233,
      "step": 8088
    },
    {
      "epoch": 0.19032941176470589,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0157,
      "step": 8089
    },
    {
      "epoch": 0.19035294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3246,
      "step": 8090
    },
    {
      "epoch": 0.19037647058823529,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.7191,
      "step": 8091
    },
    {
      "epoch": 0.1904,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1991,
      "step": 8092
    },
    {
      "epoch": 0.1904235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8977,
      "step": 8093
    },
    {
      "epoch": 0.1904470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9458,
      "step": 8094
    },
    {
      "epoch": 0.1904705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2208,
      "step": 8095
    },
    {
      "epoch": 0.1904941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2957,
      "step": 8096
    },
    {
      "epoch": 0.19051764705882354,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2653,
      "step": 8097
    },
    {
      "epoch": 0.19054117647058824,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2209,
      "step": 8098
    },
    {
      "epoch": 0.19056470588235294,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1087,
      "step": 8099
    },
    {
      "epoch": 0.19058823529411764,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9293,
      "step": 8100
    },
    {
      "epoch": 0.19061176470588234,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.7736,
      "step": 8101
    },
    {
      "epoch": 0.19063529411764707,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9209,
      "step": 8102
    },
    {
      "epoch": 0.19065882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1053,
      "step": 8103
    },
    {
      "epoch": 0.19068235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1805,
      "step": 8104
    },
    {
      "epoch": 0.19070588235294117,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1816,
      "step": 8105
    },
    {
      "epoch": 0.19072941176470587,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9992,
      "step": 8106
    },
    {
      "epoch": 0.1907529411764706,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.7047,
      "step": 8107
    },
    {
      "epoch": 0.1907764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0111,
      "step": 8108
    },
    {
      "epoch": 0.1908,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1701,
      "step": 8109
    },
    {
      "epoch": 0.1908235294117647,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9782,
      "step": 8110
    },
    {
      "epoch": 0.19084705882352943,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9385,
      "step": 8111
    },
    {
      "epoch": 0.19087058823529413,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.186,
      "step": 8112
    },
    {
      "epoch": 0.19089411764705883,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9031,
      "step": 8113
    },
    {
      "epoch": 0.19091764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2493,
      "step": 8114
    },
    {
      "epoch": 0.19094117647058823,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.9319,
      "step": 8115
    },
    {
      "epoch": 0.19096470588235295,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9559,
      "step": 8116
    },
    {
      "epoch": 0.19098823529411765,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.972,
      "step": 8117
    },
    {
      "epoch": 0.19101176470588235,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.984,
      "step": 8118
    },
    {
      "epoch": 0.19103529411764705,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9919,
      "step": 8119
    },
    {
      "epoch": 0.19105882352941175,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1572,
      "step": 8120
    },
    {
      "epoch": 0.19108235294117648,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1487,
      "step": 8121
    },
    {
      "epoch": 0.19110588235294118,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9236,
      "step": 8122
    },
    {
      "epoch": 0.19112941176470588,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.1432,
      "step": 8123
    },
    {
      "epoch": 0.19115294117647058,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4656,
      "step": 8124
    },
    {
      "epoch": 0.19117647058823528,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9645,
      "step": 8125
    },
    {
      "epoch": 0.1912,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8586,
      "step": 8126
    },
    {
      "epoch": 0.1912235294117647,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.7093,
      "step": 8127
    },
    {
      "epoch": 0.1912470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7702,
      "step": 8128
    },
    {
      "epoch": 0.1912705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0878,
      "step": 8129
    },
    {
      "epoch": 0.19129411764705884,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 8130
    },
    {
      "epoch": 0.19131764705882354,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8681,
      "step": 8131
    },
    {
      "epoch": 0.19134117647058824,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 8132
    },
    {
      "epoch": 0.19136470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.2005,
      "step": 8133
    },
    {
      "epoch": 0.19138823529411764,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9732,
      "step": 8134
    },
    {
      "epoch": 0.19141176470588236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 8135
    },
    {
      "epoch": 0.19143529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.102,
      "step": 8136
    },
    {
      "epoch": 0.19145882352941176,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 8137
    },
    {
      "epoch": 0.19148235294117646,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2448,
      "step": 8138
    },
    {
      "epoch": 0.19150588235294116,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3409,
      "step": 8139
    },
    {
      "epoch": 0.1915294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2118,
      "step": 8140
    },
    {
      "epoch": 0.1915529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2916,
      "step": 8141
    },
    {
      "epoch": 0.1915764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9286,
      "step": 8142
    },
    {
      "epoch": 0.1916,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9703,
      "step": 8143
    },
    {
      "epoch": 0.1916235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1924,
      "step": 8144
    },
    {
      "epoch": 0.19164705882352942,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1294,
      "step": 8145
    },
    {
      "epoch": 0.19167058823529412,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.088,
      "step": 8146
    },
    {
      "epoch": 0.19169411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.986,
      "step": 8147
    },
    {
      "epoch": 0.19171764705882352,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0622,
      "step": 8148
    },
    {
      "epoch": 0.19174117647058825,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0893,
      "step": 8149
    },
    {
      "epoch": 0.19176470588235295,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0714,
      "step": 8150
    },
    {
      "epoch": 0.19178823529411765,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9096,
      "step": 8151
    },
    {
      "epoch": 0.19181176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1486,
      "step": 8152
    },
    {
      "epoch": 0.19183529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2282,
      "step": 8153
    },
    {
      "epoch": 0.19185882352941178,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1113,
      "step": 8154
    },
    {
      "epoch": 0.19188235294117648,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1924,
      "step": 8155
    },
    {
      "epoch": 0.19190588235294118,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9288,
      "step": 8156
    },
    {
      "epoch": 0.19192941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1976,
      "step": 8157
    },
    {
      "epoch": 0.19195294117647058,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.3604,
      "step": 8158
    },
    {
      "epoch": 0.1919764705882353,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0061,
      "step": 8159
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1792,
      "step": 8160
    },
    {
      "epoch": 0.1920235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0817,
      "step": 8161
    },
    {
      "epoch": 0.1920470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2655,
      "step": 8162
    },
    {
      "epoch": 0.1920705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2974,
      "step": 8163
    },
    {
      "epoch": 0.19209411764705883,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0848,
      "step": 8164
    },
    {
      "epoch": 0.19211764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2897,
      "step": 8165
    },
    {
      "epoch": 0.19214117647058823,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 8166
    },
    {
      "epoch": 0.19216470588235293,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.7649,
      "step": 8167
    },
    {
      "epoch": 0.19218823529411766,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.089,
      "step": 8168
    },
    {
      "epoch": 0.19221176470588236,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8488,
      "step": 8169
    },
    {
      "epoch": 0.19223529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2486,
      "step": 8170
    },
    {
      "epoch": 0.19225882352941176,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.883,
      "step": 8171
    },
    {
      "epoch": 0.19228235294117646,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.027,
      "step": 8172
    },
    {
      "epoch": 0.1923058823529412,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9004,
      "step": 8173
    },
    {
      "epoch": 0.1923294117647059,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2755,
      "step": 8174
    },
    {
      "epoch": 0.1923529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0658,
      "step": 8175
    },
    {
      "epoch": 0.1923764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2392,
      "step": 8176
    },
    {
      "epoch": 0.1924,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.022,
      "step": 8177
    },
    {
      "epoch": 0.19242352941176472,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.6949,
      "step": 8178
    },
    {
      "epoch": 0.19244705882352942,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9009,
      "step": 8179
    },
    {
      "epoch": 0.19247058823529412,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7058,
      "step": 8180
    },
    {
      "epoch": 0.19249411764705882,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8356,
      "step": 8181
    },
    {
      "epoch": 0.19251764705882352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2078,
      "step": 8182
    },
    {
      "epoch": 0.19254117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0027,
      "step": 8183
    },
    {
      "epoch": 0.19256470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.033,
      "step": 8184
    },
    {
      "epoch": 0.19258823529411764,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0616,
      "step": 8185
    },
    {
      "epoch": 0.19261176470588234,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2486,
      "step": 8186
    },
    {
      "epoch": 0.19263529411764707,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.077,
      "step": 8187
    },
    {
      "epoch": 0.19265882352941177,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3421,
      "step": 8188
    },
    {
      "epoch": 0.19268235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.196,
      "step": 8189
    },
    {
      "epoch": 0.19270588235294117,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0043,
      "step": 8190
    },
    {
      "epoch": 0.19272941176470587,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9528,
      "step": 8191
    },
    {
      "epoch": 0.1927529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3219,
      "step": 8192
    },
    {
      "epoch": 0.1927764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 8193
    },
    {
      "epoch": 0.1928,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2454,
      "step": 8194
    },
    {
      "epoch": 0.1928235294117647,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 1.0941,
      "step": 8195
    },
    {
      "epoch": 0.1928470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9767,
      "step": 8196
    },
    {
      "epoch": 0.19287058823529413,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2082,
      "step": 8197
    },
    {
      "epoch": 0.19289411764705883,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0928,
      "step": 8198
    },
    {
      "epoch": 0.19291764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8322,
      "step": 8199
    },
    {
      "epoch": 0.19294117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1832,
      "step": 8200
    },
    {
      "epoch": 0.19296470588235295,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0631,
      "step": 8201
    },
    {
      "epoch": 0.19298823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0872,
      "step": 8202
    },
    {
      "epoch": 0.19301176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2335,
      "step": 8203
    },
    {
      "epoch": 0.19303529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.034,
      "step": 8204
    },
    {
      "epoch": 0.19305882352941175,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.2831,
      "step": 8205
    },
    {
      "epoch": 0.19308235294117648,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0648,
      "step": 8206
    },
    {
      "epoch": 0.19310588235294118,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0964,
      "step": 8207
    },
    {
      "epoch": 0.19312941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1949,
      "step": 8208
    },
    {
      "epoch": 0.19315294117647058,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 8209
    },
    {
      "epoch": 0.19317647058823528,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.99,
      "step": 8210
    },
    {
      "epoch": 0.1932,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1455,
      "step": 8211
    },
    {
      "epoch": 0.1932235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1598,
      "step": 8212
    },
    {
      "epoch": 0.1932470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8904,
      "step": 8213
    },
    {
      "epoch": 0.1932705882352941,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9657,
      "step": 8214
    },
    {
      "epoch": 0.1932941176470588,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1341,
      "step": 8215
    },
    {
      "epoch": 0.19331764705882354,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0986,
      "step": 8216
    },
    {
      "epoch": 0.19334117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 8217
    },
    {
      "epoch": 0.19336470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2119,
      "step": 8218
    },
    {
      "epoch": 0.19338823529411764,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2751,
      "step": 8219
    },
    {
      "epoch": 0.19341176470588237,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1596,
      "step": 8220
    },
    {
      "epoch": 0.19343529411764707,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.8838,
      "step": 8221
    },
    {
      "epoch": 0.19345882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2593,
      "step": 8222
    },
    {
      "epoch": 0.19348235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9002,
      "step": 8223
    },
    {
      "epoch": 0.19350588235294117,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0457,
      "step": 8224
    },
    {
      "epoch": 0.1935294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2734,
      "step": 8225
    },
    {
      "epoch": 0.1935529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1151,
      "step": 8226
    },
    {
      "epoch": 0.1935764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9305,
      "step": 8227
    },
    {
      "epoch": 0.1936,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2119,
      "step": 8228
    },
    {
      "epoch": 0.1936235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.3197,
      "step": 8229
    },
    {
      "epoch": 0.19364705882352942,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3422,
      "step": 8230
    },
    {
      "epoch": 0.19367058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2663,
      "step": 8231
    },
    {
      "epoch": 0.19369411764705882,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.8338,
      "step": 8232
    },
    {
      "epoch": 0.19371764705882352,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2955,
      "step": 8233
    },
    {
      "epoch": 0.19374117647058822,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 8234
    },
    {
      "epoch": 0.19376470588235295,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0717,
      "step": 8235
    },
    {
      "epoch": 0.19378823529411765,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.9326,
      "step": 8236
    },
    {
      "epoch": 0.19381176470588235,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2354,
      "step": 8237
    },
    {
      "epoch": 0.19383529411764705,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8408,
      "step": 8238
    },
    {
      "epoch": 0.19385882352941178,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.8612,
      "step": 8239
    },
    {
      "epoch": 0.19388235294117648,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1886,
      "step": 8240
    },
    {
      "epoch": 0.19390588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 8241
    },
    {
      "epoch": 0.19392941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1638,
      "step": 8242
    },
    {
      "epoch": 0.19395294117647058,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8925,
      "step": 8243
    },
    {
      "epoch": 0.1939764705882353,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.0214,
      "step": 8244
    },
    {
      "epoch": 0.194,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0522,
      "step": 8245
    },
    {
      "epoch": 0.1940235294117647,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.834,
      "step": 8246
    },
    {
      "epoch": 0.1940470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1239,
      "step": 8247
    },
    {
      "epoch": 0.1940705882352941,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 8248
    },
    {
      "epoch": 0.19409411764705883,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2092,
      "step": 8249
    },
    {
      "epoch": 0.19411764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.3683,
      "step": 8250
    },
    {
      "epoch": 0.19414117647058823,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.9401,
      "step": 8251
    },
    {
      "epoch": 0.19416470588235293,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0998,
      "step": 8252
    },
    {
      "epoch": 0.19418823529411763,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8028,
      "step": 8253
    },
    {
      "epoch": 0.19421176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7968,
      "step": 8254
    },
    {
      "epoch": 0.19423529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4368,
      "step": 8255
    },
    {
      "epoch": 0.19425882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9751,
      "step": 8256
    },
    {
      "epoch": 0.19428235294117646,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8873,
      "step": 8257
    },
    {
      "epoch": 0.1943058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0654,
      "step": 8258
    },
    {
      "epoch": 0.1943294117647059,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9828,
      "step": 8259
    },
    {
      "epoch": 0.1943529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 8260
    },
    {
      "epoch": 0.1943764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3081,
      "step": 8261
    },
    {
      "epoch": 0.1944,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9736,
      "step": 8262
    },
    {
      "epoch": 0.19442352941176472,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0239,
      "step": 8263
    },
    {
      "epoch": 0.19444705882352942,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8443,
      "step": 8264
    },
    {
      "epoch": 0.19447058823529412,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0683,
      "step": 8265
    },
    {
      "epoch": 0.19449411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2954,
      "step": 8266
    },
    {
      "epoch": 0.19451764705882352,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1478,
      "step": 8267
    },
    {
      "epoch": 0.19454117647058825,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.251,
      "step": 8268
    },
    {
      "epoch": 0.19456470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.3727,
      "step": 8269
    },
    {
      "epoch": 0.19458823529411765,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2823,
      "step": 8270
    },
    {
      "epoch": 0.19461176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1636,
      "step": 8271
    },
    {
      "epoch": 0.19463529411764705,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0214,
      "step": 8272
    },
    {
      "epoch": 0.19465882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2088,
      "step": 8273
    },
    {
      "epoch": 0.19468235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.169,
      "step": 8274
    },
    {
      "epoch": 0.19470588235294117,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0433,
      "step": 8275
    },
    {
      "epoch": 0.19472941176470587,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3899,
      "step": 8276
    },
    {
      "epoch": 0.1947529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9599,
      "step": 8277
    },
    {
      "epoch": 0.1947764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9273,
      "step": 8278
    },
    {
      "epoch": 0.1948,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8659,
      "step": 8279
    },
    {
      "epoch": 0.1948235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0085,
      "step": 8280
    },
    {
      "epoch": 0.1948470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1842,
      "step": 8281
    },
    {
      "epoch": 0.19487058823529413,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0585,
      "step": 8282
    },
    {
      "epoch": 0.19489411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3603,
      "step": 8283
    },
    {
      "epoch": 0.19491764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3021,
      "step": 8284
    },
    {
      "epoch": 0.19494117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2672,
      "step": 8285
    },
    {
      "epoch": 0.19496470588235293,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 8286
    },
    {
      "epoch": 0.19498823529411766,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1846,
      "step": 8287
    },
    {
      "epoch": 0.19501176470588236,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9475,
      "step": 8288
    },
    {
      "epoch": 0.19503529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.235,
      "step": 8289
    },
    {
      "epoch": 0.19505882352941176,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1268,
      "step": 8290
    },
    {
      "epoch": 0.19508235294117646,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0718,
      "step": 8291
    },
    {
      "epoch": 0.19510588235294118,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9939,
      "step": 8292
    },
    {
      "epoch": 0.19512941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8601,
      "step": 8293
    },
    {
      "epoch": 0.19515294117647058,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3067,
      "step": 8294
    },
    {
      "epoch": 0.19517647058823528,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1274,
      "step": 8295
    },
    {
      "epoch": 0.1952,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.2507,
      "step": 8296
    },
    {
      "epoch": 0.1952235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2884,
      "step": 8297
    },
    {
      "epoch": 0.1952470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0626,
      "step": 8298
    },
    {
      "epoch": 0.1952705882352941,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8626,
      "step": 8299
    },
    {
      "epoch": 0.1952941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1646,
      "step": 8300
    },
    {
      "epoch": 0.19531764705882354,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0149,
      "step": 8301
    },
    {
      "epoch": 0.19534117647058824,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0554,
      "step": 8302
    },
    {
      "epoch": 0.19536470588235294,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0719,
      "step": 8303
    },
    {
      "epoch": 0.19538823529411764,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9937,
      "step": 8304
    },
    {
      "epoch": 0.19541176470588234,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0073,
      "step": 8305
    },
    {
      "epoch": 0.19543529411764707,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1979,
      "step": 8306
    },
    {
      "epoch": 0.19545882352941177,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0054,
      "step": 8307
    },
    {
      "epoch": 0.19548235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1269,
      "step": 8308
    },
    {
      "epoch": 0.19550588235294117,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0943,
      "step": 8309
    },
    {
      "epoch": 0.1955294117647059,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.9784,
      "step": 8310
    },
    {
      "epoch": 0.1955529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4211,
      "step": 8311
    },
    {
      "epoch": 0.1955764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0992,
      "step": 8312
    },
    {
      "epoch": 0.1956,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0939,
      "step": 8313
    },
    {
      "epoch": 0.1956235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9934,
      "step": 8314
    },
    {
      "epoch": 0.19564705882352942,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1314,
      "step": 8315
    },
    {
      "epoch": 0.19567058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2345,
      "step": 8316
    },
    {
      "epoch": 0.19569411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1107,
      "step": 8317
    },
    {
      "epoch": 0.19571764705882352,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3135,
      "step": 8318
    },
    {
      "epoch": 0.19574117647058822,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2427,
      "step": 8319
    },
    {
      "epoch": 0.19576470588235295,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.86,
      "step": 8320
    },
    {
      "epoch": 0.19578823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 8321
    },
    {
      "epoch": 0.19581176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1191,
      "step": 8322
    },
    {
      "epoch": 0.19583529411764705,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4185,
      "step": 8323
    },
    {
      "epoch": 0.19585882352941175,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8748,
      "step": 8324
    },
    {
      "epoch": 0.19588235294117648,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1734,
      "step": 8325
    },
    {
      "epoch": 0.19590588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1135,
      "step": 8326
    },
    {
      "epoch": 0.19592941176470588,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8535,
      "step": 8327
    },
    {
      "epoch": 0.19595294117647058,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.7795,
      "step": 8328
    },
    {
      "epoch": 0.1959764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1908,
      "step": 8329
    },
    {
      "epoch": 0.196,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9374,
      "step": 8330
    },
    {
      "epoch": 0.1960235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 8331
    },
    {
      "epoch": 0.1960470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 8332
    },
    {
      "epoch": 0.1960705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2028,
      "step": 8333
    },
    {
      "epoch": 0.19609411764705884,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.515,
      "step": 8334
    },
    {
      "epoch": 0.19611764705882354,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1074,
      "step": 8335
    },
    {
      "epoch": 0.19614117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1989,
      "step": 8336
    },
    {
      "epoch": 0.19616470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3785,
      "step": 8337
    },
    {
      "epoch": 0.19618823529411764,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9166,
      "step": 8338
    },
    {
      "epoch": 0.19621176470588236,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.2003,
      "step": 8339
    },
    {
      "epoch": 0.19623529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.4451,
      "step": 8340
    },
    {
      "epoch": 0.19625882352941176,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 0.4881,
      "step": 8341
    },
    {
      "epoch": 0.19628235294117646,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2585,
      "step": 8342
    },
    {
      "epoch": 0.19630588235294116,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 8343
    },
    {
      "epoch": 0.1963294117647059,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9591,
      "step": 8344
    },
    {
      "epoch": 0.1963529411764706,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8791,
      "step": 8345
    },
    {
      "epoch": 0.1963764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3245,
      "step": 8346
    },
    {
      "epoch": 0.1964,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.905,
      "step": 8347
    },
    {
      "epoch": 0.19642352941176472,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1582,
      "step": 8348
    },
    {
      "epoch": 0.19644705882352942,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9353,
      "step": 8349
    },
    {
      "epoch": 0.19647058823529412,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.3468,
      "step": 8350
    },
    {
      "epoch": 0.19649411764705882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8055,
      "step": 8351
    },
    {
      "epoch": 0.19651764705882352,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.073,
      "step": 8352
    },
    {
      "epoch": 0.19654117647058825,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.6452,
      "step": 8353
    },
    {
      "epoch": 0.19656470588235295,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.8222,
      "step": 8354
    },
    {
      "epoch": 0.19658823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2178,
      "step": 8355
    },
    {
      "epoch": 0.19661176470588235,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 8356
    },
    {
      "epoch": 0.19663529411764705,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9649,
      "step": 8357
    },
    {
      "epoch": 0.19665882352941177,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0383,
      "step": 8358
    },
    {
      "epoch": 0.19668235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2407,
      "step": 8359
    },
    {
      "epoch": 0.19670588235294117,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 8360
    },
    {
      "epoch": 0.19672941176470587,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2905,
      "step": 8361
    },
    {
      "epoch": 0.19675294117647057,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2571,
      "step": 8362
    },
    {
      "epoch": 0.1967764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 8363
    },
    {
      "epoch": 0.1968,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.1615,
      "step": 8364
    },
    {
      "epoch": 0.1968235294117647,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.7459,
      "step": 8365
    },
    {
      "epoch": 0.1968470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2035,
      "step": 8366
    },
    {
      "epoch": 0.19687058823529413,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0862,
      "step": 8367
    },
    {
      "epoch": 0.19689411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.3154,
      "step": 8368
    },
    {
      "epoch": 0.19691764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2844,
      "step": 8369
    },
    {
      "epoch": 0.19694117647058823,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7885,
      "step": 8370
    },
    {
      "epoch": 0.19696470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1473,
      "step": 8371
    },
    {
      "epoch": 0.19698823529411766,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1777,
      "step": 8372
    },
    {
      "epoch": 0.19701176470588236,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2859,
      "step": 8373
    },
    {
      "epoch": 0.19703529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0504,
      "step": 8374
    },
    {
      "epoch": 0.19705882352941176,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 8375
    },
    {
      "epoch": 0.19708235294117646,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1706,
      "step": 8376
    },
    {
      "epoch": 0.1971058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1997,
      "step": 8377
    },
    {
      "epoch": 0.1971294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3038,
      "step": 8378
    },
    {
      "epoch": 0.1971529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1997,
      "step": 8379
    },
    {
      "epoch": 0.1971764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1099,
      "step": 8380
    },
    {
      "epoch": 0.1972,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2461,
      "step": 8381
    },
    {
      "epoch": 0.19722352941176471,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9782,
      "step": 8382
    },
    {
      "epoch": 0.19724705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9965,
      "step": 8383
    },
    {
      "epoch": 0.19727058823529411,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8424,
      "step": 8384
    },
    {
      "epoch": 0.19729411764705881,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1094,
      "step": 8385
    },
    {
      "epoch": 0.19731764705882354,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7968,
      "step": 8386
    },
    {
      "epoch": 0.19734117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3414,
      "step": 8387
    },
    {
      "epoch": 0.19736470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.8648,
      "step": 8388
    },
    {
      "epoch": 0.19738823529411764,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9577,
      "step": 8389
    },
    {
      "epoch": 0.19741176470588234,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1136,
      "step": 8390
    },
    {
      "epoch": 0.19743529411764707,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1081,
      "step": 8391
    },
    {
      "epoch": 0.19745882352941177,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1393,
      "step": 8392
    },
    {
      "epoch": 0.19748235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.928,
      "step": 8393
    },
    {
      "epoch": 0.19750588235294117,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2073,
      "step": 8394
    },
    {
      "epoch": 0.19752941176470587,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1072,
      "step": 8395
    },
    {
      "epoch": 0.1975529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9423,
      "step": 8396
    },
    {
      "epoch": 0.1975764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8725,
      "step": 8397
    },
    {
      "epoch": 0.1976,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2008,
      "step": 8398
    },
    {
      "epoch": 0.1976235294117647,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.9026,
      "step": 8399
    },
    {
      "epoch": 0.1976470588235294,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.604,
      "step": 8400
    },
    {
      "epoch": 0.19767058823529413,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0825,
      "step": 8401
    },
    {
      "epoch": 0.19769411764705883,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0005,
      "step": 8402
    },
    {
      "epoch": 0.19771764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8833,
      "step": 8403
    },
    {
      "epoch": 0.19774117647058823,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2573,
      "step": 8404
    },
    {
      "epoch": 0.19776470588235295,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 8405
    },
    {
      "epoch": 0.19778823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1577,
      "step": 8406
    },
    {
      "epoch": 0.19781176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 8407
    },
    {
      "epoch": 0.19783529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1328,
      "step": 8408
    },
    {
      "epoch": 0.19785882352941175,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.933,
      "step": 8409
    },
    {
      "epoch": 0.19788235294117648,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.116,
      "step": 8410
    },
    {
      "epoch": 0.19790588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3224,
      "step": 8411
    },
    {
      "epoch": 0.19792941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1207,
      "step": 8412
    },
    {
      "epoch": 0.19795294117647058,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9746,
      "step": 8413
    },
    {
      "epoch": 0.19797647058823528,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2503,
      "step": 8414
    },
    {
      "epoch": 0.198,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2752,
      "step": 8415
    },
    {
      "epoch": 0.1980235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3337,
      "step": 8416
    },
    {
      "epoch": 0.1980470588235294,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9092,
      "step": 8417
    },
    {
      "epoch": 0.1980705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2773,
      "step": 8418
    },
    {
      "epoch": 0.19809411764705884,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8392,
      "step": 8419
    },
    {
      "epoch": 0.19811764705882354,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.333,
      "step": 8420
    },
    {
      "epoch": 0.19814117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1007,
      "step": 8421
    },
    {
      "epoch": 0.19816470588235294,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.8153,
      "step": 8422
    },
    {
      "epoch": 0.19818823529411764,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9628,
      "step": 8423
    },
    {
      "epoch": 0.19821176470588237,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0275,
      "step": 8424
    },
    {
      "epoch": 0.19823529411764707,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2946,
      "step": 8425
    },
    {
      "epoch": 0.19825882352941177,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1736,
      "step": 8426
    },
    {
      "epoch": 0.19828235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.7179,
      "step": 8427
    },
    {
      "epoch": 0.19830588235294117,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3291,
      "step": 8428
    },
    {
      "epoch": 0.1983294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3016,
      "step": 8429
    },
    {
      "epoch": 0.1983529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0836,
      "step": 8430
    },
    {
      "epoch": 0.1983764705882353,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.955,
      "step": 8431
    },
    {
      "epoch": 0.1984,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1745,
      "step": 8432
    },
    {
      "epoch": 0.1984235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7948,
      "step": 8433
    },
    {
      "epoch": 0.19844705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 8434
    },
    {
      "epoch": 0.19847058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9145,
      "step": 8435
    },
    {
      "epoch": 0.19849411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.4656,
      "step": 8436
    },
    {
      "epoch": 0.19851764705882352,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0938,
      "step": 8437
    },
    {
      "epoch": 0.19854117647058825,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.7603,
      "step": 8438
    },
    {
      "epoch": 0.19856470588235295,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0605,
      "step": 8439
    },
    {
      "epoch": 0.19858823529411765,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8105,
      "step": 8440
    },
    {
      "epoch": 0.19861176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1263,
      "step": 8441
    },
    {
      "epoch": 0.19863529411764705,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 8442
    },
    {
      "epoch": 0.19865882352941178,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9156,
      "step": 8443
    },
    {
      "epoch": 0.19868235294117648,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.3072,
      "step": 8444
    },
    {
      "epoch": 0.19870588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1078,
      "step": 8445
    },
    {
      "epoch": 0.19872941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2107,
      "step": 8446
    },
    {
      "epoch": 0.19875294117647058,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.377,
      "step": 8447
    },
    {
      "epoch": 0.1987764705882353,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9414,
      "step": 8448
    },
    {
      "epoch": 0.1988,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1255,
      "step": 8449
    },
    {
      "epoch": 0.1988235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1365,
      "step": 8450
    },
    {
      "epoch": 0.1988470588235294,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8297,
      "step": 8451
    },
    {
      "epoch": 0.1988705882352941,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9559,
      "step": 8452
    },
    {
      "epoch": 0.19889411764705883,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.6551,
      "step": 8453
    },
    {
      "epoch": 0.19891764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2864,
      "step": 8454
    },
    {
      "epoch": 0.19894117647058823,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2027,
      "step": 8455
    },
    {
      "epoch": 0.19896470588235293,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7678,
      "step": 8456
    },
    {
      "epoch": 0.19898823529411766,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8205,
      "step": 8457
    },
    {
      "epoch": 0.19901176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8638,
      "step": 8458
    },
    {
      "epoch": 0.19903529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 8459
    },
    {
      "epoch": 0.19905882352941176,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7978,
      "step": 8460
    },
    {
      "epoch": 0.19908235294117646,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1731,
      "step": 8461
    },
    {
      "epoch": 0.1991058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4978,
      "step": 8462
    },
    {
      "epoch": 0.1991294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0675,
      "step": 8463
    },
    {
      "epoch": 0.1991529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0717,
      "step": 8464
    },
    {
      "epoch": 0.1991764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 8465
    },
    {
      "epoch": 0.1992,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1192,
      "step": 8466
    },
    {
      "epoch": 0.19922352941176472,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1995,
      "step": 8467
    },
    {
      "epoch": 0.19924705882352942,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8711,
      "step": 8468
    },
    {
      "epoch": 0.19927058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0467,
      "step": 8469
    },
    {
      "epoch": 0.19929411764705882,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8727,
      "step": 8470
    },
    {
      "epoch": 0.19931764705882352,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.853,
      "step": 8471
    },
    {
      "epoch": 0.19934117647058824,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1618,
      "step": 8472
    },
    {
      "epoch": 0.19936470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.084,
      "step": 8473
    },
    {
      "epoch": 0.19938823529411764,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4249,
      "step": 8474
    },
    {
      "epoch": 0.19941176470588234,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.241,
      "step": 8475
    },
    {
      "epoch": 0.19943529411764707,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1717,
      "step": 8476
    },
    {
      "epoch": 0.19945882352941177,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9401,
      "step": 8477
    },
    {
      "epoch": 0.19948235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9725,
      "step": 8478
    },
    {
      "epoch": 0.19950588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0964,
      "step": 8479
    },
    {
      "epoch": 0.19952941176470587,
      "grad_norm": 0.296875,
      "learning_rate": 0.02,
      "loss": 0.5611,
      "step": 8480
    },
    {
      "epoch": 0.1995529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8922,
      "step": 8481
    },
    {
      "epoch": 0.1995764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0237,
      "step": 8482
    },
    {
      "epoch": 0.1996,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 8483
    },
    {
      "epoch": 0.1996235294117647,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.763,
      "step": 8484
    },
    {
      "epoch": 0.1996470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0463,
      "step": 8485
    },
    {
      "epoch": 0.19967058823529413,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4891,
      "step": 8486
    },
    {
      "epoch": 0.19969411764705883,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9122,
      "step": 8487
    },
    {
      "epoch": 0.19971764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2153,
      "step": 8488
    },
    {
      "epoch": 0.19974117647058823,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2024,
      "step": 8489
    },
    {
      "epoch": 0.19976470588235293,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9376,
      "step": 8490
    },
    {
      "epoch": 0.19978823529411766,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0013,
      "step": 8491
    },
    {
      "epoch": 0.19981176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1189,
      "step": 8492
    },
    {
      "epoch": 0.19983529411764706,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7523,
      "step": 8493
    },
    {
      "epoch": 0.19985882352941176,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0788,
      "step": 8494
    },
    {
      "epoch": 0.19988235294117648,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9722,
      "step": 8495
    },
    {
      "epoch": 0.19990588235294118,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.8548,
      "step": 8496
    },
    {
      "epoch": 0.19992941176470588,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9533,
      "step": 8497
    },
    {
      "epoch": 0.19995294117647058,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2787,
      "step": 8498
    },
    {
      "epoch": 0.19997647058823528,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7569,
      "step": 8499
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1074,
      "step": 8500
    },
    {
      "epoch": 0.2000235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.132,
      "step": 8501
    },
    {
      "epoch": 0.2000470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2849,
      "step": 8502
    },
    {
      "epoch": 0.2000705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1642,
      "step": 8503
    },
    {
      "epoch": 0.2000941176470588,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0052,
      "step": 8504
    },
    {
      "epoch": 0.20011764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1029,
      "step": 8505
    },
    {
      "epoch": 0.20014117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2282,
      "step": 8506
    },
    {
      "epoch": 0.20016470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9467,
      "step": 8507
    },
    {
      "epoch": 0.20018823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9031,
      "step": 8508
    },
    {
      "epoch": 0.20021176470588234,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 8509
    },
    {
      "epoch": 0.20023529411764707,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0887,
      "step": 8510
    },
    {
      "epoch": 0.20025882352941177,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0487,
      "step": 8511
    },
    {
      "epoch": 0.20028235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3359,
      "step": 8512
    },
    {
      "epoch": 0.20030588235294117,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.829,
      "step": 8513
    },
    {
      "epoch": 0.2003294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2734,
      "step": 8514
    },
    {
      "epoch": 0.2003529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3897,
      "step": 8515
    },
    {
      "epoch": 0.2003764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9262,
      "step": 8516
    },
    {
      "epoch": 0.2004,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9899,
      "step": 8517
    },
    {
      "epoch": 0.2004235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3199,
      "step": 8518
    },
    {
      "epoch": 0.20044705882352942,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.8523,
      "step": 8519
    },
    {
      "epoch": 0.20047058823529412,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.059,
      "step": 8520
    },
    {
      "epoch": 0.20049411764705882,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0127,
      "step": 8521
    },
    {
      "epoch": 0.20051764705882352,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9006,
      "step": 8522
    },
    {
      "epoch": 0.20054117647058822,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0944,
      "step": 8523
    },
    {
      "epoch": 0.20056470588235295,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9632,
      "step": 8524
    },
    {
      "epoch": 0.20058823529411765,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.016,
      "step": 8525
    },
    {
      "epoch": 0.20061176470588235,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9925,
      "step": 8526
    },
    {
      "epoch": 0.20063529411764705,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1261,
      "step": 8527
    },
    {
      "epoch": 0.20065882352941178,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1717,
      "step": 8528
    },
    {
      "epoch": 0.20068235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 8529
    },
    {
      "epoch": 0.20070588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1927,
      "step": 8530
    },
    {
      "epoch": 0.20072941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.239,
      "step": 8531
    },
    {
      "epoch": 0.20075294117647058,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.9109,
      "step": 8532
    },
    {
      "epoch": 0.2007764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3335,
      "step": 8533
    },
    {
      "epoch": 0.2008,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3315,
      "step": 8534
    },
    {
      "epoch": 0.2008235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2506,
      "step": 8535
    },
    {
      "epoch": 0.2008470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 8536
    },
    {
      "epoch": 0.2008705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2272,
      "step": 8537
    },
    {
      "epoch": 0.20089411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9937,
      "step": 8538
    },
    {
      "epoch": 0.20091764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2012,
      "step": 8539
    },
    {
      "epoch": 0.20094117647058823,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 8540
    },
    {
      "epoch": 0.20096470588235293,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0777,
      "step": 8541
    },
    {
      "epoch": 0.20098823529411763,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.007,
      "step": 8542
    },
    {
      "epoch": 0.20101176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 8543
    },
    {
      "epoch": 0.20103529411764706,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.985,
      "step": 8544
    },
    {
      "epoch": 0.20105882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2399,
      "step": 8545
    },
    {
      "epoch": 0.20108235294117646,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.6363,
      "step": 8546
    },
    {
      "epoch": 0.2011058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9737,
      "step": 8547
    },
    {
      "epoch": 0.2011294117647059,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.2329,
      "step": 8548
    },
    {
      "epoch": 0.2011529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 8549
    },
    {
      "epoch": 0.2011764705882353,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 1.0544,
      "step": 8550
    },
    {
      "epoch": 0.2012,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2995,
      "step": 8551
    },
    {
      "epoch": 0.20122352941176472,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9379,
      "step": 8552
    },
    {
      "epoch": 0.20124705882352942,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0545,
      "step": 8553
    },
    {
      "epoch": 0.20127058823529412,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9782,
      "step": 8554
    },
    {
      "epoch": 0.20129411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0696,
      "step": 8555
    },
    {
      "epoch": 0.20131764705882352,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3143,
      "step": 8556
    },
    {
      "epoch": 0.20134117647058825,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2936,
      "step": 8557
    },
    {
      "epoch": 0.20136470588235295,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8658,
      "step": 8558
    },
    {
      "epoch": 0.20138823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3344,
      "step": 8559
    },
    {
      "epoch": 0.20141176470588235,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8786,
      "step": 8560
    },
    {
      "epoch": 0.20143529411764705,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1311,
      "step": 8561
    },
    {
      "epoch": 0.20145882352941177,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9643,
      "step": 8562
    },
    {
      "epoch": 0.20148235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1594,
      "step": 8563
    },
    {
      "epoch": 0.20150588235294117,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3444,
      "step": 8564
    },
    {
      "epoch": 0.20152941176470587,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9977,
      "step": 8565
    },
    {
      "epoch": 0.2015529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9604,
      "step": 8566
    },
    {
      "epoch": 0.2015764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0666,
      "step": 8567
    },
    {
      "epoch": 0.2016,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0151,
      "step": 8568
    },
    {
      "epoch": 0.2016235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2834,
      "step": 8569
    },
    {
      "epoch": 0.2016470588235294,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 1.0739,
      "step": 8570
    },
    {
      "epoch": 0.20167058823529413,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2855,
      "step": 8571
    },
    {
      "epoch": 0.20169411764705883,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3473,
      "step": 8572
    },
    {
      "epoch": 0.20171764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9456,
      "step": 8573
    },
    {
      "epoch": 0.20174117647058823,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9733,
      "step": 8574
    },
    {
      "epoch": 0.20176470588235293,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.057,
      "step": 8575
    },
    {
      "epoch": 0.20178823529411766,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9437,
      "step": 8576
    },
    {
      "epoch": 0.20181176470588236,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9517,
      "step": 8577
    },
    {
      "epoch": 0.20183529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1527,
      "step": 8578
    },
    {
      "epoch": 0.20185882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1127,
      "step": 8579
    },
    {
      "epoch": 0.20188235294117646,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1688,
      "step": 8580
    },
    {
      "epoch": 0.20190588235294119,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9546,
      "step": 8581
    },
    {
      "epoch": 0.20192941176470589,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9429,
      "step": 8582
    },
    {
      "epoch": 0.20195294117647059,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9212,
      "step": 8583
    },
    {
      "epoch": 0.20197647058823529,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.417,
      "step": 8584
    },
    {
      "epoch": 0.202,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0577,
      "step": 8585
    },
    {
      "epoch": 0.2020235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0251,
      "step": 8586
    },
    {
      "epoch": 0.2020470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 8587
    },
    {
      "epoch": 0.2020705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9662,
      "step": 8588
    },
    {
      "epoch": 0.2020941176470588,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0426,
      "step": 8589
    },
    {
      "epoch": 0.20211764705882354,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2161,
      "step": 8590
    },
    {
      "epoch": 0.20214117647058824,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9048,
      "step": 8591
    },
    {
      "epoch": 0.20216470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2305,
      "step": 8592
    },
    {
      "epoch": 0.20218823529411764,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2467,
      "step": 8593
    },
    {
      "epoch": 0.20221176470588234,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 1.1667,
      "step": 8594
    },
    {
      "epoch": 0.20223529411764707,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.106,
      "step": 8595
    },
    {
      "epoch": 0.20225882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 8596
    },
    {
      "epoch": 0.20228235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7398,
      "step": 8597
    },
    {
      "epoch": 0.20230588235294117,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0331,
      "step": 8598
    },
    {
      "epoch": 0.20232941176470587,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.7802,
      "step": 8599
    },
    {
      "epoch": 0.2023529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8528,
      "step": 8600
    },
    {
      "epoch": 0.2023764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2741,
      "step": 8601
    },
    {
      "epoch": 0.2024,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8723,
      "step": 8602
    },
    {
      "epoch": 0.2024235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.2725,
      "step": 8603
    },
    {
      "epoch": 0.20244705882352942,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 1.0683,
      "step": 8604
    },
    {
      "epoch": 0.20247058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3115,
      "step": 8605
    },
    {
      "epoch": 0.20249411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.048,
      "step": 8606
    },
    {
      "epoch": 0.20251764705882352,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0231,
      "step": 8607
    },
    {
      "epoch": 0.20254117647058822,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3895,
      "step": 8608
    },
    {
      "epoch": 0.20256470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1504,
      "step": 8609
    },
    {
      "epoch": 0.20258823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2355,
      "step": 8610
    },
    {
      "epoch": 0.20261176470588235,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0981,
      "step": 8611
    },
    {
      "epoch": 0.20263529411764705,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 8612
    },
    {
      "epoch": 0.20265882352941175,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1644,
      "step": 8613
    },
    {
      "epoch": 0.20268235294117648,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.8753,
      "step": 8614
    },
    {
      "epoch": 0.20270588235294118,
      "grad_norm": 0.291015625,
      "learning_rate": 0.02,
      "loss": 0.7616,
      "step": 8615
    },
    {
      "epoch": 0.20272941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3538,
      "step": 8616
    },
    {
      "epoch": 0.20275294117647058,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9203,
      "step": 8617
    },
    {
      "epoch": 0.20277647058823528,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.7222,
      "step": 8618
    },
    {
      "epoch": 0.2028,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7635,
      "step": 8619
    },
    {
      "epoch": 0.2028235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8657,
      "step": 8620
    },
    {
      "epoch": 0.2028470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9755,
      "step": 8621
    },
    {
      "epoch": 0.2028705882352941,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0459,
      "step": 8622
    },
    {
      "epoch": 0.20289411764705884,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1517,
      "step": 8623
    },
    {
      "epoch": 0.20291764705882354,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.8143,
      "step": 8624
    },
    {
      "epoch": 0.20294117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9773,
      "step": 8625
    },
    {
      "epoch": 0.20296470588235294,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.8742,
      "step": 8626
    },
    {
      "epoch": 0.20298823529411764,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9892,
      "step": 8627
    },
    {
      "epoch": 0.20301176470588236,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.4202,
      "step": 8628
    },
    {
      "epoch": 0.20303529411764706,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7414,
      "step": 8629
    },
    {
      "epoch": 0.20305882352941176,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.6095,
      "step": 8630
    },
    {
      "epoch": 0.20308235294117646,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2248,
      "step": 8631
    },
    {
      "epoch": 0.20310588235294116,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9551,
      "step": 8632
    },
    {
      "epoch": 0.2031294117647059,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1558,
      "step": 8633
    },
    {
      "epoch": 0.2031529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9235,
      "step": 8634
    },
    {
      "epoch": 0.2031764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0354,
      "step": 8635
    },
    {
      "epoch": 0.2032,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1786,
      "step": 8636
    },
    {
      "epoch": 0.20322352941176472,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2824,
      "step": 8637
    },
    {
      "epoch": 0.20324705882352942,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9417,
      "step": 8638
    },
    {
      "epoch": 0.20327058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2423,
      "step": 8639
    },
    {
      "epoch": 0.20329411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2519,
      "step": 8640
    },
    {
      "epoch": 0.20331764705882352,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 8641
    },
    {
      "epoch": 0.20334117647058825,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2235,
      "step": 8642
    },
    {
      "epoch": 0.20336470588235295,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2701,
      "step": 8643
    },
    {
      "epoch": 0.20338823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1227,
      "step": 8644
    },
    {
      "epoch": 0.20341176470588235,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8672,
      "step": 8645
    },
    {
      "epoch": 0.20343529411764705,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0042,
      "step": 8646
    },
    {
      "epoch": 0.20345882352941178,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1583,
      "step": 8647
    },
    {
      "epoch": 0.20348235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.101,
      "step": 8648
    },
    {
      "epoch": 0.20350588235294118,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.7577,
      "step": 8649
    },
    {
      "epoch": 0.20352941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 8650
    },
    {
      "epoch": 0.20355294117647058,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.5878,
      "step": 8651
    },
    {
      "epoch": 0.2035764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.4672,
      "step": 8652
    },
    {
      "epoch": 0.2036,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.7637,
      "step": 8653
    },
    {
      "epoch": 0.2036235294117647,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.9135,
      "step": 8654
    },
    {
      "epoch": 0.2036470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 8655
    },
    {
      "epoch": 0.20367058823529413,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2611,
      "step": 8656
    },
    {
      "epoch": 0.20369411764705883,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9751,
      "step": 8657
    },
    {
      "epoch": 0.20371764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9793,
      "step": 8658
    },
    {
      "epoch": 0.20374117647058823,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9066,
      "step": 8659
    },
    {
      "epoch": 0.20376470588235293,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 8660
    },
    {
      "epoch": 0.20378823529411766,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9199,
      "step": 8661
    },
    {
      "epoch": 0.20381176470588236,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0357,
      "step": 8662
    },
    {
      "epoch": 0.20383529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8556,
      "step": 8663
    },
    {
      "epoch": 0.20385882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.219,
      "step": 8664
    },
    {
      "epoch": 0.20388235294117646,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.4921,
      "step": 8665
    },
    {
      "epoch": 0.2039058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3257,
      "step": 8666
    },
    {
      "epoch": 0.2039294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4558,
      "step": 8667
    },
    {
      "epoch": 0.2039529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9069,
      "step": 8668
    },
    {
      "epoch": 0.2039764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0929,
      "step": 8669
    },
    {
      "epoch": 0.204,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 8670
    },
    {
      "epoch": 0.20402352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1657,
      "step": 8671
    },
    {
      "epoch": 0.20404705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.177,
      "step": 8672
    },
    {
      "epoch": 0.20407058823529411,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1234,
      "step": 8673
    },
    {
      "epoch": 0.20409411764705881,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2053,
      "step": 8674
    },
    {
      "epoch": 0.20411764705882354,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0922,
      "step": 8675
    },
    {
      "epoch": 0.20414117647058824,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8879,
      "step": 8676
    },
    {
      "epoch": 0.20416470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0704,
      "step": 8677
    },
    {
      "epoch": 0.20418823529411764,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1352,
      "step": 8678
    },
    {
      "epoch": 0.20421176470588234,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.6791,
      "step": 8679
    },
    {
      "epoch": 0.20423529411764707,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2886,
      "step": 8680
    },
    {
      "epoch": 0.20425882352941177,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9337,
      "step": 8681
    },
    {
      "epoch": 0.20428235294117647,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9384,
      "step": 8682
    },
    {
      "epoch": 0.20430588235294117,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0412,
      "step": 8683
    },
    {
      "epoch": 0.20432941176470587,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0555,
      "step": 8684
    },
    {
      "epoch": 0.2043529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2636,
      "step": 8685
    },
    {
      "epoch": 0.2043764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.6577,
      "step": 8686
    },
    {
      "epoch": 0.2044,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0509,
      "step": 8687
    },
    {
      "epoch": 0.2044235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3304,
      "step": 8688
    },
    {
      "epoch": 0.2044470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.7981,
      "step": 8689
    },
    {
      "epoch": 0.20447058823529413,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 1.2482,
      "step": 8690
    },
    {
      "epoch": 0.20449411764705883,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.974,
      "step": 8691
    },
    {
      "epoch": 0.20451764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1599,
      "step": 8692
    },
    {
      "epoch": 0.20454117647058823,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.3173,
      "step": 8693
    },
    {
      "epoch": 0.20456470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2798,
      "step": 8694
    },
    {
      "epoch": 0.20458823529411765,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.963,
      "step": 8695
    },
    {
      "epoch": 0.20461176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0439,
      "step": 8696
    },
    {
      "epoch": 0.20463529411764705,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9804,
      "step": 8697
    },
    {
      "epoch": 0.20465882352941175,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9954,
      "step": 8698
    },
    {
      "epoch": 0.20468235294117648,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.1186,
      "step": 8699
    },
    {
      "epoch": 0.20470588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1937,
      "step": 8700
    },
    {
      "epoch": 0.20472941176470588,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8717,
      "step": 8701
    },
    {
      "epoch": 0.20475294117647058,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9363,
      "step": 8702
    },
    {
      "epoch": 0.20477647058823528,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.976,
      "step": 8703
    },
    {
      "epoch": 0.2048,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3201,
      "step": 8704
    },
    {
      "epoch": 0.2048235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0987,
      "step": 8705
    },
    {
      "epoch": 0.2048470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0964,
      "step": 8706
    },
    {
      "epoch": 0.2048705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9002,
      "step": 8707
    },
    {
      "epoch": 0.2048941176470588,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0275,
      "step": 8708
    },
    {
      "epoch": 0.20491764705882354,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9412,
      "step": 8709
    },
    {
      "epoch": 0.20494117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0371,
      "step": 8710
    },
    {
      "epoch": 0.20496470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2682,
      "step": 8711
    },
    {
      "epoch": 0.20498823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0288,
      "step": 8712
    },
    {
      "epoch": 0.20501176470588237,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0048,
      "step": 8713
    },
    {
      "epoch": 0.20503529411764707,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1903,
      "step": 8714
    },
    {
      "epoch": 0.20505882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1218,
      "step": 8715
    },
    {
      "epoch": 0.20508235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.2969,
      "step": 8716
    },
    {
      "epoch": 0.20510588235294117,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1101,
      "step": 8717
    },
    {
      "epoch": 0.2051294117647059,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9268,
      "step": 8718
    },
    {
      "epoch": 0.2051529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2163,
      "step": 8719
    },
    {
      "epoch": 0.2051764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0584,
      "step": 8720
    },
    {
      "epoch": 0.2052,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0789,
      "step": 8721
    },
    {
      "epoch": 0.2052235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1729,
      "step": 8722
    },
    {
      "epoch": 0.20524705882352942,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.7927,
      "step": 8723
    },
    {
      "epoch": 0.20527058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0981,
      "step": 8724
    },
    {
      "epoch": 0.20529411764705882,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2156,
      "step": 8725
    },
    {
      "epoch": 0.20531764705882352,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0959,
      "step": 8726
    },
    {
      "epoch": 0.20534117647058822,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 8727
    },
    {
      "epoch": 0.20536470588235295,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.322,
      "step": 8728
    },
    {
      "epoch": 0.20538823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0138,
      "step": 8729
    },
    {
      "epoch": 0.20541176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0996,
      "step": 8730
    },
    {
      "epoch": 0.20543529411764705,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1239,
      "step": 8731
    },
    {
      "epoch": 0.20545882352941178,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1629,
      "step": 8732
    },
    {
      "epoch": 0.20548235294117648,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1387,
      "step": 8733
    },
    {
      "epoch": 0.20550588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0324,
      "step": 8734
    },
    {
      "epoch": 0.20552941176470588,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2211,
      "step": 8735
    },
    {
      "epoch": 0.20555294117647058,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.7409,
      "step": 8736
    },
    {
      "epoch": 0.2055764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4255,
      "step": 8737
    },
    {
      "epoch": 0.2056,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1962,
      "step": 8738
    },
    {
      "epoch": 0.2056235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7666,
      "step": 8739
    },
    {
      "epoch": 0.2056470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0391,
      "step": 8740
    },
    {
      "epoch": 0.2056705882352941,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0441,
      "step": 8741
    },
    {
      "epoch": 0.20569411764705883,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 8742
    },
    {
      "epoch": 0.20571764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0848,
      "step": 8743
    },
    {
      "epoch": 0.20574117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8471,
      "step": 8744
    },
    {
      "epoch": 0.20576470588235293,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.8993,
      "step": 8745
    },
    {
      "epoch": 0.20578823529411766,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1671,
      "step": 8746
    },
    {
      "epoch": 0.20581176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2025,
      "step": 8747
    },
    {
      "epoch": 0.20583529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1537,
      "step": 8748
    },
    {
      "epoch": 0.20585882352941176,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.263,
      "step": 8749
    },
    {
      "epoch": 0.20588235294117646,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0427,
      "step": 8750
    },
    {
      "epoch": 0.2059058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0908,
      "step": 8751
    },
    {
      "epoch": 0.2059294117647059,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0714,
      "step": 8752
    },
    {
      "epoch": 0.2059529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.172,
      "step": 8753
    },
    {
      "epoch": 0.2059764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 8754
    },
    {
      "epoch": 0.206,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.4003,
      "step": 8755
    },
    {
      "epoch": 0.20602352941176472,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0609,
      "step": 8756
    },
    {
      "epoch": 0.20604705882352942,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9205,
      "step": 8757
    },
    {
      "epoch": 0.20607058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9317,
      "step": 8758
    },
    {
      "epoch": 0.20609411764705882,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1727,
      "step": 8759
    },
    {
      "epoch": 0.20611764705882352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1831,
      "step": 8760
    },
    {
      "epoch": 0.20614117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0418,
      "step": 8761
    },
    {
      "epoch": 0.20616470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1535,
      "step": 8762
    },
    {
      "epoch": 0.20618823529411764,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2365,
      "step": 8763
    },
    {
      "epoch": 0.20621176470588234,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 8764
    },
    {
      "epoch": 0.20623529411764707,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1781,
      "step": 8765
    },
    {
      "epoch": 0.20625882352941177,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.219,
      "step": 8766
    },
    {
      "epoch": 0.20628235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2512,
      "step": 8767
    },
    {
      "epoch": 0.20630588235294117,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1055,
      "step": 8768
    },
    {
      "epoch": 0.20632941176470587,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2252,
      "step": 8769
    },
    {
      "epoch": 0.2063529411764706,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8794,
      "step": 8770
    },
    {
      "epoch": 0.2063764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.265,
      "step": 8771
    },
    {
      "epoch": 0.2064,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.777,
      "step": 8772
    },
    {
      "epoch": 0.2064235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9944,
      "step": 8773
    },
    {
      "epoch": 0.2064470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0264,
      "step": 8774
    },
    {
      "epoch": 0.20647058823529413,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2705,
      "step": 8775
    },
    {
      "epoch": 0.20649411764705883,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0233,
      "step": 8776
    },
    {
      "epoch": 0.20651764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1271,
      "step": 8777
    },
    {
      "epoch": 0.20654117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3198,
      "step": 8778
    },
    {
      "epoch": 0.20656470588235293,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1662,
      "step": 8779
    },
    {
      "epoch": 0.20658823529411766,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1731,
      "step": 8780
    },
    {
      "epoch": 0.20661176470588236,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1356,
      "step": 8781
    },
    {
      "epoch": 0.20663529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.2679,
      "step": 8782
    },
    {
      "epoch": 0.20665882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4591,
      "step": 8783
    },
    {
      "epoch": 0.20668235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2164,
      "step": 8784
    },
    {
      "epoch": 0.20670588235294118,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9196,
      "step": 8785
    },
    {
      "epoch": 0.20672941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 8786
    },
    {
      "epoch": 0.20675294117647058,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 8787
    },
    {
      "epoch": 0.20677647058823528,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0083,
      "step": 8788
    },
    {
      "epoch": 0.2068,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0614,
      "step": 8789
    },
    {
      "epoch": 0.2068235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.158,
      "step": 8790
    },
    {
      "epoch": 0.2068470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1191,
      "step": 8791
    },
    {
      "epoch": 0.2068705882352941,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1427,
      "step": 8792
    },
    {
      "epoch": 0.2068941176470588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1956,
      "step": 8793
    },
    {
      "epoch": 0.20691764705882354,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.733,
      "step": 8794
    },
    {
      "epoch": 0.20694117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 8795
    },
    {
      "epoch": 0.20696470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1199,
      "step": 8796
    },
    {
      "epoch": 0.20698823529411764,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.2613,
      "step": 8797
    },
    {
      "epoch": 0.20701176470588234,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8953,
      "step": 8798
    },
    {
      "epoch": 0.20703529411764707,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2964,
      "step": 8799
    },
    {
      "epoch": 0.20705882352941177,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0135,
      "step": 8800
    },
    {
      "epoch": 0.20708235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3661,
      "step": 8801
    },
    {
      "epoch": 0.20710588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3478,
      "step": 8802
    },
    {
      "epoch": 0.2071294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.161,
      "step": 8803
    },
    {
      "epoch": 0.2071529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1453,
      "step": 8804
    },
    {
      "epoch": 0.2071764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 8805
    },
    {
      "epoch": 0.2072,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1573,
      "step": 8806
    },
    {
      "epoch": 0.2072235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4064,
      "step": 8807
    },
    {
      "epoch": 0.20724705882352942,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.924,
      "step": 8808
    },
    {
      "epoch": 0.20727058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2007,
      "step": 8809
    },
    {
      "epoch": 0.20729411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1963,
      "step": 8810
    },
    {
      "epoch": 0.20731764705882352,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9652,
      "step": 8811
    },
    {
      "epoch": 0.20734117647058822,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.7669,
      "step": 8812
    },
    {
      "epoch": 0.20736470588235295,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1785,
      "step": 8813
    },
    {
      "epoch": 0.20738823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.2016,
      "step": 8814
    },
    {
      "epoch": 0.20741176470588235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2502,
      "step": 8815
    },
    {
      "epoch": 0.20743529411764705,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8759,
      "step": 8816
    },
    {
      "epoch": 0.20745882352941175,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.6684,
      "step": 8817
    },
    {
      "epoch": 0.20748235294117648,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9715,
      "step": 8818
    },
    {
      "epoch": 0.20750588235294118,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9607,
      "step": 8819
    },
    {
      "epoch": 0.20752941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1867,
      "step": 8820
    },
    {
      "epoch": 0.20755294117647058,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.4644,
      "step": 8821
    },
    {
      "epoch": 0.2075764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0796,
      "step": 8822
    },
    {
      "epoch": 0.2076,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9844,
      "step": 8823
    },
    {
      "epoch": 0.2076235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8066,
      "step": 8824
    },
    {
      "epoch": 0.2076470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8921,
      "step": 8825
    },
    {
      "epoch": 0.2076705882352941,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9105,
      "step": 8826
    },
    {
      "epoch": 0.20769411764705883,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1452,
      "step": 8827
    },
    {
      "epoch": 0.20771764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3046,
      "step": 8828
    },
    {
      "epoch": 0.20774117647058823,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0985,
      "step": 8829
    },
    {
      "epoch": 0.20776470588235293,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3017,
      "step": 8830
    },
    {
      "epoch": 0.20778823529411763,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0868,
      "step": 8831
    },
    {
      "epoch": 0.20781176470588236,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2127,
      "step": 8832
    },
    {
      "epoch": 0.20783529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1454,
      "step": 8833
    },
    {
      "epoch": 0.20785882352941176,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0414,
      "step": 8834
    },
    {
      "epoch": 0.20788235294117646,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8131,
      "step": 8835
    },
    {
      "epoch": 0.20790588235294116,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2984,
      "step": 8836
    },
    {
      "epoch": 0.2079294117647059,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1807,
      "step": 8837
    },
    {
      "epoch": 0.2079529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.283,
      "step": 8838
    },
    {
      "epoch": 0.2079764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2815,
      "step": 8839
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0399,
      "step": 8840
    },
    {
      "epoch": 0.20802352941176472,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3791,
      "step": 8841
    },
    {
      "epoch": 0.20804705882352942,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8579,
      "step": 8842
    },
    {
      "epoch": 0.20807058823529412,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9669,
      "step": 8843
    },
    {
      "epoch": 0.20809411764705882,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0367,
      "step": 8844
    },
    {
      "epoch": 0.20811764705882352,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1884,
      "step": 8845
    },
    {
      "epoch": 0.20814117647058825,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7967,
      "step": 8846
    },
    {
      "epoch": 0.20816470588235295,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.6028,
      "step": 8847
    },
    {
      "epoch": 0.20818823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3627,
      "step": 8848
    },
    {
      "epoch": 0.20821176470588235,
      "grad_norm": 0.279296875,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 8849
    },
    {
      "epoch": 0.20823529411764705,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2211,
      "step": 8850
    },
    {
      "epoch": 0.20825882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 8851
    },
    {
      "epoch": 0.20828235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0969,
      "step": 8852
    },
    {
      "epoch": 0.20830588235294117,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0059,
      "step": 8853
    },
    {
      "epoch": 0.20832941176470587,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 8854
    },
    {
      "epoch": 0.2083529411764706,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.5895,
      "step": 8855
    },
    {
      "epoch": 0.2083764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1753,
      "step": 8856
    },
    {
      "epoch": 0.2084,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3442,
      "step": 8857
    },
    {
      "epoch": 0.2084235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.6698,
      "step": 8858
    },
    {
      "epoch": 0.2084470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3263,
      "step": 8859
    },
    {
      "epoch": 0.20847058823529413,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0534,
      "step": 8860
    },
    {
      "epoch": 0.20849411764705883,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.927,
      "step": 8861
    },
    {
      "epoch": 0.20851764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1289,
      "step": 8862
    },
    {
      "epoch": 0.20854117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2415,
      "step": 8863
    },
    {
      "epoch": 0.20856470588235293,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0719,
      "step": 8864
    },
    {
      "epoch": 0.20858823529411766,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8405,
      "step": 8865
    },
    {
      "epoch": 0.20861176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8783,
      "step": 8866
    },
    {
      "epoch": 0.20863529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.075,
      "step": 8867
    },
    {
      "epoch": 0.20865882352941176,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.255,
      "step": 8868
    },
    {
      "epoch": 0.20868235294117646,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9357,
      "step": 8869
    },
    {
      "epoch": 0.20870588235294119,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2368,
      "step": 8870
    },
    {
      "epoch": 0.20872941176470589,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9524,
      "step": 8871
    },
    {
      "epoch": 0.20875294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0475,
      "step": 8872
    },
    {
      "epoch": 0.20877647058823529,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8775,
      "step": 8873
    },
    {
      "epoch": 0.2088,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.227,
      "step": 8874
    },
    {
      "epoch": 0.2088235294117647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.9021,
      "step": 8875
    },
    {
      "epoch": 0.2088470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1488,
      "step": 8876
    },
    {
      "epoch": 0.2088705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.168,
      "step": 8877
    },
    {
      "epoch": 0.2088941176470588,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 8878
    },
    {
      "epoch": 0.20891764705882354,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4346,
      "step": 8879
    },
    {
      "epoch": 0.20894117647058824,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7658,
      "step": 8880
    },
    {
      "epoch": 0.20896470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.948,
      "step": 8881
    },
    {
      "epoch": 0.20898823529411764,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3373,
      "step": 8882
    },
    {
      "epoch": 0.20901176470588234,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0495,
      "step": 8883
    },
    {
      "epoch": 0.20903529411764707,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.8676,
      "step": 8884
    },
    {
      "epoch": 0.20905882352941177,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3283,
      "step": 8885
    },
    {
      "epoch": 0.20908235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1108,
      "step": 8886
    },
    {
      "epoch": 0.20910588235294117,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8802,
      "step": 8887
    },
    {
      "epoch": 0.20912941176470587,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.167,
      "step": 8888
    },
    {
      "epoch": 0.2091529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1399,
      "step": 8889
    },
    {
      "epoch": 0.2091764705882353,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.5803,
      "step": 8890
    },
    {
      "epoch": 0.2092,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1951,
      "step": 8891
    },
    {
      "epoch": 0.2092235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0772,
      "step": 8892
    },
    {
      "epoch": 0.20924705882352942,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0103,
      "step": 8893
    },
    {
      "epoch": 0.20927058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2134,
      "step": 8894
    },
    {
      "epoch": 0.20929411764705882,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3332,
      "step": 8895
    },
    {
      "epoch": 0.20931764705882352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0845,
      "step": 8896
    },
    {
      "epoch": 0.20934117647058822,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1305,
      "step": 8897
    },
    {
      "epoch": 0.20936470588235295,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 8898
    },
    {
      "epoch": 0.20938823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0891,
      "step": 8899
    },
    {
      "epoch": 0.20941176470588235,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9471,
      "step": 8900
    },
    {
      "epoch": 0.20943529411764705,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.7835,
      "step": 8901
    },
    {
      "epoch": 0.20945882352941175,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2105,
      "step": 8902
    },
    {
      "epoch": 0.20948235294117648,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2327,
      "step": 8903
    },
    {
      "epoch": 0.20950588235294118,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.4634,
      "step": 8904
    },
    {
      "epoch": 0.20952941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1032,
      "step": 8905
    },
    {
      "epoch": 0.20955294117647058,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8081,
      "step": 8906
    },
    {
      "epoch": 0.20957647058823528,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4385,
      "step": 8907
    },
    {
      "epoch": 0.2096,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9022,
      "step": 8908
    },
    {
      "epoch": 0.2096235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.4351,
      "step": 8909
    },
    {
      "epoch": 0.2096470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8547,
      "step": 8910
    },
    {
      "epoch": 0.2096705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8483,
      "step": 8911
    },
    {
      "epoch": 0.20969411764705884,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7469,
      "step": 8912
    },
    {
      "epoch": 0.20971764705882354,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.063,
      "step": 8913
    },
    {
      "epoch": 0.20974117647058824,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9878,
      "step": 8914
    },
    {
      "epoch": 0.20976470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3497,
      "step": 8915
    },
    {
      "epoch": 0.20978823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.02,
      "step": 8916
    },
    {
      "epoch": 0.20981176470588236,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8043,
      "step": 8917
    },
    {
      "epoch": 0.20983529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0084,
      "step": 8918
    },
    {
      "epoch": 0.20985882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1669,
      "step": 8919
    },
    {
      "epoch": 0.20988235294117646,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0651,
      "step": 8920
    },
    {
      "epoch": 0.20990588235294116,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9741,
      "step": 8921
    },
    {
      "epoch": 0.2099294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1942,
      "step": 8922
    },
    {
      "epoch": 0.2099529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3397,
      "step": 8923
    },
    {
      "epoch": 0.2099764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1973,
      "step": 8924
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.5264,
      "step": 8925
    },
    {
      "epoch": 0.2100235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1762,
      "step": 8926
    },
    {
      "epoch": 0.21004705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.074,
      "step": 8927
    },
    {
      "epoch": 0.21007058823529412,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0906,
      "step": 8928
    },
    {
      "epoch": 0.21009411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.2113,
      "step": 8929
    },
    {
      "epoch": 0.21011764705882352,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3633,
      "step": 8930
    },
    {
      "epoch": 0.21014117647058825,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1175,
      "step": 8931
    },
    {
      "epoch": 0.21016470588235295,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3556,
      "step": 8932
    },
    {
      "epoch": 0.21018823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2121,
      "step": 8933
    },
    {
      "epoch": 0.21021176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.302,
      "step": 8934
    },
    {
      "epoch": 0.21023529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0257,
      "step": 8935
    },
    {
      "epoch": 0.21025882352941178,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0745,
      "step": 8936
    },
    {
      "epoch": 0.21028235294117648,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1435,
      "step": 8937
    },
    {
      "epoch": 0.21030588235294118,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9561,
      "step": 8938
    },
    {
      "epoch": 0.21032941176470588,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8801,
      "step": 8939
    },
    {
      "epoch": 0.21035294117647058,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 8940
    },
    {
      "epoch": 0.2103764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1354,
      "step": 8941
    },
    {
      "epoch": 0.2104,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1232,
      "step": 8942
    },
    {
      "epoch": 0.2104235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 8943
    },
    {
      "epoch": 0.2104470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0925,
      "step": 8944
    },
    {
      "epoch": 0.21047058823529413,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9417,
      "step": 8945
    },
    {
      "epoch": 0.21049411764705883,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.339,
      "step": 8946
    },
    {
      "epoch": 0.21051764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2419,
      "step": 8947
    },
    {
      "epoch": 0.21054117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9875,
      "step": 8948
    },
    {
      "epoch": 0.21056470588235293,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0048,
      "step": 8949
    },
    {
      "epoch": 0.21058823529411766,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2114,
      "step": 8950
    },
    {
      "epoch": 0.21061176470588236,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8354,
      "step": 8951
    },
    {
      "epoch": 0.21063529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1232,
      "step": 8952
    },
    {
      "epoch": 0.21065882352941176,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.7746,
      "step": 8953
    },
    {
      "epoch": 0.21068235294117646,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.7292,
      "step": 8954
    },
    {
      "epoch": 0.2107058823529412,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.9008,
      "step": 8955
    },
    {
      "epoch": 0.2107294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0723,
      "step": 8956
    },
    {
      "epoch": 0.2107529411764706,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9547,
      "step": 8957
    },
    {
      "epoch": 0.2107764705882353,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.6663,
      "step": 8958
    },
    {
      "epoch": 0.2108,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2815,
      "step": 8959
    },
    {
      "epoch": 0.21082352941176472,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3554,
      "step": 8960
    },
    {
      "epoch": 0.21084705882352942,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0368,
      "step": 8961
    },
    {
      "epoch": 0.21087058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0963,
      "step": 8962
    },
    {
      "epoch": 0.21089411764705882,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.9848,
      "step": 8963
    },
    {
      "epoch": 0.21091764705882354,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9713,
      "step": 8964
    },
    {
      "epoch": 0.21094117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4931,
      "step": 8965
    },
    {
      "epoch": 0.21096470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 8966
    },
    {
      "epoch": 0.21098823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9954,
      "step": 8967
    },
    {
      "epoch": 0.21101176470588234,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2801,
      "step": 8968
    },
    {
      "epoch": 0.21103529411764707,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.0853,
      "step": 8969
    },
    {
      "epoch": 0.21105882352941177,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0062,
      "step": 8970
    },
    {
      "epoch": 0.21108235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.044,
      "step": 8971
    },
    {
      "epoch": 0.21110588235294117,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.2782,
      "step": 8972
    },
    {
      "epoch": 0.21112941176470587,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.073,
      "step": 8973
    },
    {
      "epoch": 0.2111529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3492,
      "step": 8974
    },
    {
      "epoch": 0.2111764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9296,
      "step": 8975
    },
    {
      "epoch": 0.2112,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.155,
      "step": 8976
    },
    {
      "epoch": 0.2112235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.2563,
      "step": 8977
    },
    {
      "epoch": 0.2112470588235294,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.9921,
      "step": 8978
    },
    {
      "epoch": 0.21127058823529413,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9065,
      "step": 8979
    },
    {
      "epoch": 0.21129411764705883,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8302,
      "step": 8980
    },
    {
      "epoch": 0.21131764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.844,
      "step": 8981
    },
    {
      "epoch": 0.21134117647058823,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1409,
      "step": 8982
    },
    {
      "epoch": 0.21136470588235295,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.975,
      "step": 8983
    },
    {
      "epoch": 0.21138823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 8984
    },
    {
      "epoch": 0.21141176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.4401,
      "step": 8985
    },
    {
      "epoch": 0.21143529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.3922,
      "step": 8986
    },
    {
      "epoch": 0.21145882352941175,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1437,
      "step": 8987
    },
    {
      "epoch": 0.21148235294117648,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0032,
      "step": 8988
    },
    {
      "epoch": 0.21150588235294118,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1251,
      "step": 8989
    },
    {
      "epoch": 0.21152941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3978,
      "step": 8990
    },
    {
      "epoch": 0.21155294117647058,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0865,
      "step": 8991
    },
    {
      "epoch": 0.21157647058823528,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9566,
      "step": 8992
    },
    {
      "epoch": 0.2116,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2578,
      "step": 8993
    },
    {
      "epoch": 0.2116235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0342,
      "step": 8994
    },
    {
      "epoch": 0.2116470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 8995
    },
    {
      "epoch": 0.2116705882352941,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.8127,
      "step": 8996
    },
    {
      "epoch": 0.2116941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0843,
      "step": 8997
    },
    {
      "epoch": 0.21171764705882354,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1439,
      "step": 8998
    },
    {
      "epoch": 0.21174117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3062,
      "step": 8999
    },
    {
      "epoch": 0.21176470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8355,
      "step": 9000
    },
    {
      "epoch": 0.21176470588235294,
      "eval_loss": 2.2205469608306885,
      "eval_runtime": 684.3701,
      "eval_samples_per_second": 12.42,
      "eval_steps_per_second": 3.105,
      "step": 9000
    },
    {
      "epoch": 0.21178823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 9001
    },
    {
      "epoch": 0.21181176470588237,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 9002
    },
    {
      "epoch": 0.21183529411764707,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9065,
      "step": 9003
    },
    {
      "epoch": 0.21185882352941177,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7212,
      "step": 9004
    },
    {
      "epoch": 0.21188235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2669,
      "step": 9005
    },
    {
      "epoch": 0.21190588235294117,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9723,
      "step": 9006
    },
    {
      "epoch": 0.2119294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4144,
      "step": 9007
    },
    {
      "epoch": 0.2119529411764706,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.7712,
      "step": 9008
    },
    {
      "epoch": 0.2119764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9941,
      "step": 9009
    },
    {
      "epoch": 0.212,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1954,
      "step": 9010
    },
    {
      "epoch": 0.2120235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1142,
      "step": 9011
    },
    {
      "epoch": 0.21204705882352942,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8414,
      "step": 9012
    },
    {
      "epoch": 0.21207058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2087,
      "step": 9013
    },
    {
      "epoch": 0.21209411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9266,
      "step": 9014
    },
    {
      "epoch": 0.21211764705882352,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2768,
      "step": 9015
    },
    {
      "epoch": 0.21214117647058822,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0287,
      "step": 9016
    },
    {
      "epoch": 0.21216470588235295,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.213,
      "step": 9017
    },
    {
      "epoch": 0.21218823529411765,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9164,
      "step": 9018
    },
    {
      "epoch": 0.21221176470588235,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9066,
      "step": 9019
    },
    {
      "epoch": 0.21223529411764705,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8303,
      "step": 9020
    },
    {
      "epoch": 0.21225882352941178,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8534,
      "step": 9021
    },
    {
      "epoch": 0.21228235294117648,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 9022
    },
    {
      "epoch": 0.21230588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.4002,
      "step": 9023
    },
    {
      "epoch": 0.21232941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9728,
      "step": 9024
    },
    {
      "epoch": 0.21235294117647058,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2317,
      "step": 9025
    },
    {
      "epoch": 0.2123764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0442,
      "step": 9026
    },
    {
      "epoch": 0.2124,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.6182,
      "step": 9027
    },
    {
      "epoch": 0.2124235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0772,
      "step": 9028
    },
    {
      "epoch": 0.2124470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1884,
      "step": 9029
    },
    {
      "epoch": 0.2124705882352941,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8473,
      "step": 9030
    },
    {
      "epoch": 0.21249411764705883,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7719,
      "step": 9031
    },
    {
      "epoch": 0.21251764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0357,
      "step": 9032
    },
    {
      "epoch": 0.21254117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1446,
      "step": 9033
    },
    {
      "epoch": 0.21256470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1556,
      "step": 9034
    },
    {
      "epoch": 0.21258823529411763,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0299,
      "step": 9035
    },
    {
      "epoch": 0.21261176470588236,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.131,
      "step": 9036
    },
    {
      "epoch": 0.21263529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3549,
      "step": 9037
    },
    {
      "epoch": 0.21265882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 9038
    },
    {
      "epoch": 0.21268235294117646,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.111,
      "step": 9039
    },
    {
      "epoch": 0.2127058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.159,
      "step": 9040
    },
    {
      "epoch": 0.2127294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1186,
      "step": 9041
    },
    {
      "epoch": 0.2127529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9502,
      "step": 9042
    },
    {
      "epoch": 0.2127764705882353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7067,
      "step": 9043
    },
    {
      "epoch": 0.2128,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0693,
      "step": 9044
    },
    {
      "epoch": 0.21282352941176472,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9778,
      "step": 9045
    },
    {
      "epoch": 0.21284705882352942,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2791,
      "step": 9046
    },
    {
      "epoch": 0.21287058823529412,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.0044,
      "step": 9047
    },
    {
      "epoch": 0.21289411764705882,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9809,
      "step": 9048
    },
    {
      "epoch": 0.21291764705882352,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8485,
      "step": 9049
    },
    {
      "epoch": 0.21294117647058824,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2667,
      "step": 9050
    },
    {
      "epoch": 0.21296470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1268,
      "step": 9051
    },
    {
      "epoch": 0.21298823529411764,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.8848,
      "step": 9052
    },
    {
      "epoch": 0.21301176470588234,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2545,
      "step": 9053
    },
    {
      "epoch": 0.21303529411764707,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2333,
      "step": 9054
    },
    {
      "epoch": 0.21305882352941177,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8964,
      "step": 9055
    },
    {
      "epoch": 0.21308235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1942,
      "step": 9056
    },
    {
      "epoch": 0.21310588235294117,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.1014,
      "step": 9057
    },
    {
      "epoch": 0.21312941176470587,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1494,
      "step": 9058
    },
    {
      "epoch": 0.2131529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.4789,
      "step": 9059
    },
    {
      "epoch": 0.2131764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 9060
    },
    {
      "epoch": 0.2132,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.4996,
      "step": 9061
    },
    {
      "epoch": 0.2132235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9326,
      "step": 9062
    },
    {
      "epoch": 0.2132470588235294,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9126,
      "step": 9063
    },
    {
      "epoch": 0.21327058823529413,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.827,
      "step": 9064
    },
    {
      "epoch": 0.21329411764705883,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0252,
      "step": 9065
    },
    {
      "epoch": 0.21331764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0711,
      "step": 9066
    },
    {
      "epoch": 0.21334117647058823,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0493,
      "step": 9067
    },
    {
      "epoch": 0.21336470588235293,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0318,
      "step": 9068
    },
    {
      "epoch": 0.21338823529411766,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1592,
      "step": 9069
    },
    {
      "epoch": 0.21341176470588236,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7906,
      "step": 9070
    },
    {
      "epoch": 0.21343529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3108,
      "step": 9071
    },
    {
      "epoch": 0.21345882352941176,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9405,
      "step": 9072
    },
    {
      "epoch": 0.21348235294117648,
      "grad_norm": 0.26171875,
      "learning_rate": 0.02,
      "loss": 0.8822,
      "step": 9073
    },
    {
      "epoch": 0.21350588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0129,
      "step": 9074
    },
    {
      "epoch": 0.21352941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0875,
      "step": 9075
    },
    {
      "epoch": 0.21355294117647058,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.6713,
      "step": 9076
    },
    {
      "epoch": 0.21357647058823528,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 9077
    },
    {
      "epoch": 0.2136,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2064,
      "step": 9078
    },
    {
      "epoch": 0.2136235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3389,
      "step": 9079
    },
    {
      "epoch": 0.2136470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 9080
    },
    {
      "epoch": 0.2136705882352941,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.1028,
      "step": 9081
    },
    {
      "epoch": 0.2136941176470588,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9034,
      "step": 9082
    },
    {
      "epoch": 0.21371764705882354,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2171,
      "step": 9083
    },
    {
      "epoch": 0.21374117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2301,
      "step": 9084
    },
    {
      "epoch": 0.21376470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1881,
      "step": 9085
    },
    {
      "epoch": 0.21378823529411764,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 9086
    },
    {
      "epoch": 0.21381176470588234,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9961,
      "step": 9087
    },
    {
      "epoch": 0.21383529411764707,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0934,
      "step": 9088
    },
    {
      "epoch": 0.21385882352941177,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 1.0561,
      "step": 9089
    },
    {
      "epoch": 0.21388235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1491,
      "step": 9090
    },
    {
      "epoch": 0.21390588235294117,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 9091
    },
    {
      "epoch": 0.2139294117647059,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9354,
      "step": 9092
    },
    {
      "epoch": 0.2139529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1258,
      "step": 9093
    },
    {
      "epoch": 0.2139764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1477,
      "step": 9094
    },
    {
      "epoch": 0.214,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8742,
      "step": 9095
    },
    {
      "epoch": 0.2140235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2198,
      "step": 9096
    },
    {
      "epoch": 0.21404705882352942,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3219,
      "step": 9097
    },
    {
      "epoch": 0.21407058823529412,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7725,
      "step": 9098
    },
    {
      "epoch": 0.21409411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9896,
      "step": 9099
    },
    {
      "epoch": 0.21411764705882352,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7865,
      "step": 9100
    },
    {
      "epoch": 0.21414117647058822,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1434,
      "step": 9101
    },
    {
      "epoch": 0.21416470588235295,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2792,
      "step": 9102
    },
    {
      "epoch": 0.21418823529411765,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.6237,
      "step": 9103
    },
    {
      "epoch": 0.21421176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1401,
      "step": 9104
    },
    {
      "epoch": 0.21423529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0031,
      "step": 9105
    },
    {
      "epoch": 0.21425882352941175,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.968,
      "step": 9106
    },
    {
      "epoch": 0.21428235294117648,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2564,
      "step": 9107
    },
    {
      "epoch": 0.21430588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2621,
      "step": 9108
    },
    {
      "epoch": 0.21432941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 9109
    },
    {
      "epoch": 0.21435294117647058,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0496,
      "step": 9110
    },
    {
      "epoch": 0.2143764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2764,
      "step": 9111
    },
    {
      "epoch": 0.2144,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0852,
      "step": 9112
    },
    {
      "epoch": 0.2144235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3038,
      "step": 9113
    },
    {
      "epoch": 0.2144470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3852,
      "step": 9114
    },
    {
      "epoch": 0.2144705882352941,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.7755,
      "step": 9115
    },
    {
      "epoch": 0.21449411764705884,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1028,
      "step": 9116
    },
    {
      "epoch": 0.21451764705882354,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1619,
      "step": 9117
    },
    {
      "epoch": 0.21454117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8583,
      "step": 9118
    },
    {
      "epoch": 0.21456470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3557,
      "step": 9119
    },
    {
      "epoch": 0.21458823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.3776,
      "step": 9120
    },
    {
      "epoch": 0.21461176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1156,
      "step": 9121
    },
    {
      "epoch": 0.21463529411764706,
      "grad_norm": 0.255859375,
      "learning_rate": 0.02,
      "loss": 0.646,
      "step": 9122
    },
    {
      "epoch": 0.21465882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9828,
      "step": 9123
    },
    {
      "epoch": 0.21468235294117646,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.8186,
      "step": 9124
    },
    {
      "epoch": 0.21470588235294116,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3838,
      "step": 9125
    },
    {
      "epoch": 0.2147294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8439,
      "step": 9126
    },
    {
      "epoch": 0.2147529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0525,
      "step": 9127
    },
    {
      "epoch": 0.2147764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3521,
      "step": 9128
    },
    {
      "epoch": 0.2148,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0403,
      "step": 9129
    },
    {
      "epoch": 0.21482352941176472,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4644,
      "step": 9130
    },
    {
      "epoch": 0.21484705882352942,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7302,
      "step": 9131
    },
    {
      "epoch": 0.21487058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2459,
      "step": 9132
    },
    {
      "epoch": 0.21489411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0112,
      "step": 9133
    },
    {
      "epoch": 0.21491764705882352,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9542,
      "step": 9134
    },
    {
      "epoch": 0.21494117647058825,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1079,
      "step": 9135
    },
    {
      "epoch": 0.21496470588235295,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1859,
      "step": 9136
    },
    {
      "epoch": 0.21498823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.4782,
      "step": 9137
    },
    {
      "epoch": 0.21501176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0303,
      "step": 9138
    },
    {
      "epoch": 0.21503529411764705,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1893,
      "step": 9139
    },
    {
      "epoch": 0.21505882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2385,
      "step": 9140
    },
    {
      "epoch": 0.21508235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2854,
      "step": 9141
    },
    {
      "epoch": 0.21510588235294117,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0066,
      "step": 9142
    },
    {
      "epoch": 0.21512941176470587,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1938,
      "step": 9143
    },
    {
      "epoch": 0.21515294117647057,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.988,
      "step": 9144
    },
    {
      "epoch": 0.2151764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.3414,
      "step": 9145
    },
    {
      "epoch": 0.2152,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.086,
      "step": 9146
    },
    {
      "epoch": 0.2152235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0099,
      "step": 9147
    },
    {
      "epoch": 0.2152470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1775,
      "step": 9148
    },
    {
      "epoch": 0.21527058823529413,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.234,
      "step": 9149
    },
    {
      "epoch": 0.21529411764705883,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2045,
      "step": 9150
    },
    {
      "epoch": 0.21531764705882353,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.6611,
      "step": 9151
    },
    {
      "epoch": 0.21534117647058823,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8029,
      "step": 9152
    },
    {
      "epoch": 0.21536470588235293,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9144,
      "step": 9153
    },
    {
      "epoch": 0.21538823529411766,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2526,
      "step": 9154
    },
    {
      "epoch": 0.21541176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8964,
      "step": 9155
    },
    {
      "epoch": 0.21543529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8967,
      "step": 9156
    },
    {
      "epoch": 0.21545882352941176,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 9157
    },
    {
      "epoch": 0.21548235294117646,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0567,
      "step": 9158
    },
    {
      "epoch": 0.21550588235294119,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3549,
      "step": 9159
    },
    {
      "epoch": 0.21552941176470589,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 9160
    },
    {
      "epoch": 0.21555294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.4007,
      "step": 9161
    },
    {
      "epoch": 0.21557647058823529,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9955,
      "step": 9162
    },
    {
      "epoch": 0.2156,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1651,
      "step": 9163
    },
    {
      "epoch": 0.21562352941176471,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1193,
      "step": 9164
    },
    {
      "epoch": 0.21564705882352941,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0123,
      "step": 9165
    },
    {
      "epoch": 0.21567058823529411,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2642,
      "step": 9166
    },
    {
      "epoch": 0.21569411764705881,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2058,
      "step": 9167
    },
    {
      "epoch": 0.21571764705882354,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2314,
      "step": 9168
    },
    {
      "epoch": 0.21574117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1136,
      "step": 9169
    },
    {
      "epoch": 0.21576470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0013,
      "step": 9170
    },
    {
      "epoch": 0.21578823529411764,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.257,
      "step": 9171
    },
    {
      "epoch": 0.21581176470588234,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0053,
      "step": 9172
    },
    {
      "epoch": 0.21583529411764707,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2426,
      "step": 9173
    },
    {
      "epoch": 0.21585882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1962,
      "step": 9174
    },
    {
      "epoch": 0.21588235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9307,
      "step": 9175
    },
    {
      "epoch": 0.21590588235294117,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.7943,
      "step": 9176
    },
    {
      "epoch": 0.21592941176470587,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9187,
      "step": 9177
    },
    {
      "epoch": 0.2159529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1569,
      "step": 9178
    },
    {
      "epoch": 0.2159764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0502,
      "step": 9179
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.16,
      "step": 9180
    },
    {
      "epoch": 0.2160235294117647,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.8819,
      "step": 9181
    },
    {
      "epoch": 0.21604705882352943,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0187,
      "step": 9182
    },
    {
      "epoch": 0.21607058823529413,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0117,
      "step": 9183
    },
    {
      "epoch": 0.21609411764705883,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8919,
      "step": 9184
    },
    {
      "epoch": 0.21611764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 9185
    },
    {
      "epoch": 0.21614117647058823,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2352,
      "step": 9186
    },
    {
      "epoch": 0.21616470588235295,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.6301,
      "step": 9187
    },
    {
      "epoch": 0.21618823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2732,
      "step": 9188
    },
    {
      "epoch": 0.21621176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1234,
      "step": 9189
    },
    {
      "epoch": 0.21623529411764705,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1634,
      "step": 9190
    },
    {
      "epoch": 0.21625882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.337,
      "step": 9191
    },
    {
      "epoch": 0.21628235294117648,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.3491,
      "step": 9192
    },
    {
      "epoch": 0.21630588235294118,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9323,
      "step": 9193
    },
    {
      "epoch": 0.21632941176470588,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0445,
      "step": 9194
    },
    {
      "epoch": 0.21635294117647058,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.026,
      "step": 9195
    },
    {
      "epoch": 0.21637647058823528,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.067,
      "step": 9196
    },
    {
      "epoch": 0.2164,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9915,
      "step": 9197
    },
    {
      "epoch": 0.2164235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8879,
      "step": 9198
    },
    {
      "epoch": 0.2164470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3929,
      "step": 9199
    },
    {
      "epoch": 0.2164705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0445,
      "step": 9200
    },
    {
      "epoch": 0.21649411764705884,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2606,
      "step": 9201
    },
    {
      "epoch": 0.21651764705882354,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1579,
      "step": 9202
    },
    {
      "epoch": 0.21654117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2508,
      "step": 9203
    },
    {
      "epoch": 0.21656470588235294,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 9204
    },
    {
      "epoch": 0.21658823529411764,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1121,
      "step": 9205
    },
    {
      "epoch": 0.21661176470588236,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 1.1288,
      "step": 9206
    },
    {
      "epoch": 0.21663529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0249,
      "step": 9207
    },
    {
      "epoch": 0.21665882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9204,
      "step": 9208
    },
    {
      "epoch": 0.21668235294117646,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.7536,
      "step": 9209
    },
    {
      "epoch": 0.21670588235294116,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2795,
      "step": 9210
    },
    {
      "epoch": 0.2167294117647059,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.7205,
      "step": 9211
    },
    {
      "epoch": 0.2167529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0761,
      "step": 9212
    },
    {
      "epoch": 0.2167764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1302,
      "step": 9213
    },
    {
      "epoch": 0.2168,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1695,
      "step": 9214
    },
    {
      "epoch": 0.2168235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.4987,
      "step": 9215
    },
    {
      "epoch": 0.21684705882352942,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1195,
      "step": 9216
    },
    {
      "epoch": 0.21687058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.4232,
      "step": 9217
    },
    {
      "epoch": 0.21689411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1683,
      "step": 9218
    },
    {
      "epoch": 0.21691764705882352,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2093,
      "step": 9219
    },
    {
      "epoch": 0.21694117647058825,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3462,
      "step": 9220
    },
    {
      "epoch": 0.21696470588235295,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1064,
      "step": 9221
    },
    {
      "epoch": 0.21698823529411765,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.9447,
      "step": 9222
    },
    {
      "epoch": 0.21701176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.4982,
      "step": 9223
    },
    {
      "epoch": 0.21703529411764705,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9354,
      "step": 9224
    },
    {
      "epoch": 0.21705882352941178,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9877,
      "step": 9225
    },
    {
      "epoch": 0.21708235294117648,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7535,
      "step": 9226
    },
    {
      "epoch": 0.21710588235294118,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9688,
      "step": 9227
    },
    {
      "epoch": 0.21712941176470588,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.257,
      "step": 9228
    },
    {
      "epoch": 0.21715294117647058,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.5144,
      "step": 9229
    },
    {
      "epoch": 0.2171764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.6733,
      "step": 9230
    },
    {
      "epoch": 0.2172,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1698,
      "step": 9231
    },
    {
      "epoch": 0.2172235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1993,
      "step": 9232
    },
    {
      "epoch": 0.2172470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0444,
      "step": 9233
    },
    {
      "epoch": 0.2172705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0086,
      "step": 9234
    },
    {
      "epoch": 0.21729411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9816,
      "step": 9235
    },
    {
      "epoch": 0.21731764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0054,
      "step": 9236
    },
    {
      "epoch": 0.21734117647058823,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.5881,
      "step": 9237
    },
    {
      "epoch": 0.21736470588235293,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8857,
      "step": 9238
    },
    {
      "epoch": 0.21738823529411766,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1806,
      "step": 9239
    },
    {
      "epoch": 0.21741176470588236,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.5653,
      "step": 9240
    },
    {
      "epoch": 0.21743529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.901,
      "step": 9241
    },
    {
      "epoch": 0.21745882352941176,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.6611,
      "step": 9242
    },
    {
      "epoch": 0.21748235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 9243
    },
    {
      "epoch": 0.2175058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0259,
      "step": 9244
    },
    {
      "epoch": 0.2175294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8543,
      "step": 9245
    },
    {
      "epoch": 0.2175529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1661,
      "step": 9246
    },
    {
      "epoch": 0.2175764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1829,
      "step": 9247
    },
    {
      "epoch": 0.2176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9121,
      "step": 9248
    },
    {
      "epoch": 0.21762352941176472,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8226,
      "step": 9249
    },
    {
      "epoch": 0.21764705882352942,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2469,
      "step": 9250
    },
    {
      "epoch": 0.21767058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1621,
      "step": 9251
    },
    {
      "epoch": 0.21769411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2747,
      "step": 9252
    },
    {
      "epoch": 0.21771764705882352,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3465,
      "step": 9253
    },
    {
      "epoch": 0.21774117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2195,
      "step": 9254
    },
    {
      "epoch": 0.21776470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0028,
      "step": 9255
    },
    {
      "epoch": 0.21778823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1328,
      "step": 9256
    },
    {
      "epoch": 0.21781176470588234,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3608,
      "step": 9257
    },
    {
      "epoch": 0.21783529411764707,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.117,
      "step": 9258
    },
    {
      "epoch": 0.21785882352941177,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1027,
      "step": 9259
    },
    {
      "epoch": 0.21788235294117647,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.96,
      "step": 9260
    },
    {
      "epoch": 0.21790588235294117,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.921,
      "step": 9261
    },
    {
      "epoch": 0.21792941176470587,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1315,
      "step": 9262
    },
    {
      "epoch": 0.2179529411764706,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.7815,
      "step": 9263
    },
    {
      "epoch": 0.2179764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1371,
      "step": 9264
    },
    {
      "epoch": 0.218,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2206,
      "step": 9265
    },
    {
      "epoch": 0.2180235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7049,
      "step": 9266
    },
    {
      "epoch": 0.2180470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7949,
      "step": 9267
    },
    {
      "epoch": 0.21807058823529413,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.741,
      "step": 9268
    },
    {
      "epoch": 0.21809411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1292,
      "step": 9269
    },
    {
      "epoch": 0.21811764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.104,
      "step": 9270
    },
    {
      "epoch": 0.21814117647058823,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0567,
      "step": 9271
    },
    {
      "epoch": 0.21816470588235296,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0369,
      "step": 9272
    },
    {
      "epoch": 0.21818823529411766,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.36,
      "step": 9273
    },
    {
      "epoch": 0.21821176470588236,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 9274
    },
    {
      "epoch": 0.21823529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2045,
      "step": 9275
    },
    {
      "epoch": 0.21825882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4081,
      "step": 9276
    },
    {
      "epoch": 0.21828235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2963,
      "step": 9277
    },
    {
      "epoch": 0.21830588235294118,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9506,
      "step": 9278
    },
    {
      "epoch": 0.21832941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0903,
      "step": 9279
    },
    {
      "epoch": 0.21835294117647058,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8988,
      "step": 9280
    },
    {
      "epoch": 0.21837647058823528,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0014,
      "step": 9281
    },
    {
      "epoch": 0.2184,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.8238,
      "step": 9282
    },
    {
      "epoch": 0.2184235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8159,
      "step": 9283
    },
    {
      "epoch": 0.2184470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9208,
      "step": 9284
    },
    {
      "epoch": 0.2184705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2843,
      "step": 9285
    },
    {
      "epoch": 0.2184941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9771,
      "step": 9286
    },
    {
      "epoch": 0.21851764705882354,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0282,
      "step": 9287
    },
    {
      "epoch": 0.21854117647058824,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.862,
      "step": 9288
    },
    {
      "epoch": 0.21856470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 9289
    },
    {
      "epoch": 0.21858823529411764,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2043,
      "step": 9290
    },
    {
      "epoch": 0.21861176470588237,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.27,
      "step": 9291
    },
    {
      "epoch": 0.21863529411764707,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8847,
      "step": 9292
    },
    {
      "epoch": 0.21865882352941177,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1136,
      "step": 9293
    },
    {
      "epoch": 0.21868235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.188,
      "step": 9294
    },
    {
      "epoch": 0.21870588235294117,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1895,
      "step": 9295
    },
    {
      "epoch": 0.2187294117647059,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 9296
    },
    {
      "epoch": 0.2187529411764706,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8739,
      "step": 9297
    },
    {
      "epoch": 0.2187764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.262,
      "step": 9298
    },
    {
      "epoch": 0.2188,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2773,
      "step": 9299
    },
    {
      "epoch": 0.2188235294117647,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.8777,
      "step": 9300
    },
    {
      "epoch": 0.21884705882352942,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.067,
      "step": 9301
    },
    {
      "epoch": 0.21887058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1785,
      "step": 9302
    },
    {
      "epoch": 0.21889411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1482,
      "step": 9303
    },
    {
      "epoch": 0.21891764705882352,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2342,
      "step": 9304
    },
    {
      "epoch": 0.21894117647058822,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0107,
      "step": 9305
    },
    {
      "epoch": 0.21896470588235295,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9825,
      "step": 9306
    },
    {
      "epoch": 0.21898823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.015,
      "step": 9307
    },
    {
      "epoch": 0.21901176470588235,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1015,
      "step": 9308
    },
    {
      "epoch": 0.21903529411764705,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 1.1714,
      "step": 9309
    },
    {
      "epoch": 0.21905882352941178,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.929,
      "step": 9310
    },
    {
      "epoch": 0.21908235294117648,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.8427,
      "step": 9311
    },
    {
      "epoch": 0.21910588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2282,
      "step": 9312
    },
    {
      "epoch": 0.21912941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2407,
      "step": 9313
    },
    {
      "epoch": 0.21915294117647058,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9565,
      "step": 9314
    },
    {
      "epoch": 0.2191764705882353,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.8884,
      "step": 9315
    },
    {
      "epoch": 0.2192,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0952,
      "step": 9316
    },
    {
      "epoch": 0.2192235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 9317
    },
    {
      "epoch": 0.2192470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3507,
      "step": 9318
    },
    {
      "epoch": 0.2192705882352941,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0285,
      "step": 9319
    },
    {
      "epoch": 0.21929411764705883,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9809,
      "step": 9320
    },
    {
      "epoch": 0.21931764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.986,
      "step": 9321
    },
    {
      "epoch": 0.21934117647058823,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1586,
      "step": 9322
    },
    {
      "epoch": 0.21936470588235293,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8712,
      "step": 9323
    },
    {
      "epoch": 0.21938823529411763,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.223,
      "step": 9324
    },
    {
      "epoch": 0.21941176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1611,
      "step": 9325
    },
    {
      "epoch": 0.21943529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0415,
      "step": 9326
    },
    {
      "epoch": 0.21945882352941176,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0543,
      "step": 9327
    },
    {
      "epoch": 0.21948235294117646,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1237,
      "step": 9328
    },
    {
      "epoch": 0.2195058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8777,
      "step": 9329
    },
    {
      "epoch": 0.2195294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1386,
      "step": 9330
    },
    {
      "epoch": 0.2195529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.025,
      "step": 9331
    },
    {
      "epoch": 0.2195764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2428,
      "step": 9332
    },
    {
      "epoch": 0.2196,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1107,
      "step": 9333
    },
    {
      "epoch": 0.21962352941176472,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.1388,
      "step": 9334
    },
    {
      "epoch": 0.21964705882352942,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 1.0159,
      "step": 9335
    },
    {
      "epoch": 0.21967058823529412,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7712,
      "step": 9336
    },
    {
      "epoch": 0.21969411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2104,
      "step": 9337
    },
    {
      "epoch": 0.21971764705882352,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0258,
      "step": 9338
    },
    {
      "epoch": 0.21974117647058825,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2145,
      "step": 9339
    },
    {
      "epoch": 0.21976470588235295,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9733,
      "step": 9340
    },
    {
      "epoch": 0.21978823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1541,
      "step": 9341
    },
    {
      "epoch": 0.21981176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7926,
      "step": 9342
    },
    {
      "epoch": 0.21983529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2093,
      "step": 9343
    },
    {
      "epoch": 0.21985882352941177,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4173,
      "step": 9344
    },
    {
      "epoch": 0.21988235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9956,
      "step": 9345
    },
    {
      "epoch": 0.21990588235294117,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1664,
      "step": 9346
    },
    {
      "epoch": 0.21992941176470587,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.103,
      "step": 9347
    },
    {
      "epoch": 0.2199529411764706,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9963,
      "step": 9348
    },
    {
      "epoch": 0.2199764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8012,
      "step": 9349
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2382,
      "step": 9350
    },
    {
      "epoch": 0.2200235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1453,
      "step": 9351
    },
    {
      "epoch": 0.2200470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.772,
      "step": 9352
    },
    {
      "epoch": 0.22007058823529413,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1687,
      "step": 9353
    },
    {
      "epoch": 0.22009411764705883,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0399,
      "step": 9354
    },
    {
      "epoch": 0.22011764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2129,
      "step": 9355
    },
    {
      "epoch": 0.22014117647058823,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9408,
      "step": 9356
    },
    {
      "epoch": 0.22016470588235293,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1045,
      "step": 9357
    },
    {
      "epoch": 0.22018823529411766,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.3186,
      "step": 9358
    },
    {
      "epoch": 0.22021176470588236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0882,
      "step": 9359
    },
    {
      "epoch": 0.22023529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.919,
      "step": 9360
    },
    {
      "epoch": 0.22025882352941176,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9913,
      "step": 9361
    },
    {
      "epoch": 0.22028235294117646,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.7093,
      "step": 9362
    },
    {
      "epoch": 0.22030588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8263,
      "step": 9363
    },
    {
      "epoch": 0.22032941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2814,
      "step": 9364
    },
    {
      "epoch": 0.22035294117647058,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9801,
      "step": 9365
    },
    {
      "epoch": 0.22037647058823528,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2411,
      "step": 9366
    },
    {
      "epoch": 0.2204,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1627,
      "step": 9367
    },
    {
      "epoch": 0.2204235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0494,
      "step": 9368
    },
    {
      "epoch": 0.2204470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0152,
      "step": 9369
    },
    {
      "epoch": 0.2204705882352941,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9238,
      "step": 9370
    },
    {
      "epoch": 0.2204941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1487,
      "step": 9371
    },
    {
      "epoch": 0.22051764705882354,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9143,
      "step": 9372
    },
    {
      "epoch": 0.22054117647058824,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0922,
      "step": 9373
    },
    {
      "epoch": 0.22056470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 1.0301,
      "step": 9374
    },
    {
      "epoch": 0.22058823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8222,
      "step": 9375
    },
    {
      "epoch": 0.22061176470588234,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9282,
      "step": 9376
    },
    {
      "epoch": 0.22063529411764707,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.7291,
      "step": 9377
    },
    {
      "epoch": 0.22065882352941177,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.4147,
      "step": 9378
    },
    {
      "epoch": 0.22068235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2029,
      "step": 9379
    },
    {
      "epoch": 0.22070588235294117,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.6804,
      "step": 9380
    },
    {
      "epoch": 0.2207294117647059,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.6924,
      "step": 9381
    },
    {
      "epoch": 0.2207529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1414,
      "step": 9382
    },
    {
      "epoch": 0.2207764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0913,
      "step": 9383
    },
    {
      "epoch": 0.2208,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.8755,
      "step": 9384
    },
    {
      "epoch": 0.2208235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1743,
      "step": 9385
    },
    {
      "epoch": 0.22084705882352942,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.7582,
      "step": 9386
    },
    {
      "epoch": 0.22087058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1152,
      "step": 9387
    },
    {
      "epoch": 0.22089411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2274,
      "step": 9388
    },
    {
      "epoch": 0.22091764705882352,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8996,
      "step": 9389
    },
    {
      "epoch": 0.22094117647058822,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9183,
      "step": 9390
    },
    {
      "epoch": 0.22096470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2823,
      "step": 9391
    },
    {
      "epoch": 0.22098823529411765,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7035,
      "step": 9392
    },
    {
      "epoch": 0.22101176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9844,
      "step": 9393
    },
    {
      "epoch": 0.22103529411764705,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7538,
      "step": 9394
    },
    {
      "epoch": 0.22105882352941175,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0323,
      "step": 9395
    },
    {
      "epoch": 0.22108235294117648,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1356,
      "step": 9396
    },
    {
      "epoch": 0.22110588235294118,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 9397
    },
    {
      "epoch": 0.22112941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2331,
      "step": 9398
    },
    {
      "epoch": 0.22115294117647058,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9795,
      "step": 9399
    },
    {
      "epoch": 0.2211764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2002,
      "step": 9400
    },
    {
      "epoch": 0.2212,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 1.1828,
      "step": 9401
    },
    {
      "epoch": 0.2212235294117647,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8472,
      "step": 9402
    },
    {
      "epoch": 0.2212470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.2235,
      "step": 9403
    },
    {
      "epoch": 0.2212705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1086,
      "step": 9404
    },
    {
      "epoch": 0.22129411764705884,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7661,
      "step": 9405
    },
    {
      "epoch": 0.22131764705882354,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1041,
      "step": 9406
    },
    {
      "epoch": 0.22134117647058824,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0027,
      "step": 9407
    },
    {
      "epoch": 0.22136470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9724,
      "step": 9408
    },
    {
      "epoch": 0.22138823529411764,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8016,
      "step": 9409
    },
    {
      "epoch": 0.22141176470588236,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0477,
      "step": 9410
    },
    {
      "epoch": 0.22143529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2564,
      "step": 9411
    },
    {
      "epoch": 0.22145882352941176,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0584,
      "step": 9412
    },
    {
      "epoch": 0.22148235294117646,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0484,
      "step": 9413
    },
    {
      "epoch": 0.22150588235294116,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8073,
      "step": 9414
    },
    {
      "epoch": 0.2215294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0515,
      "step": 9415
    },
    {
      "epoch": 0.2215529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4785,
      "step": 9416
    },
    {
      "epoch": 0.2215764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1325,
      "step": 9417
    },
    {
      "epoch": 0.2216,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0493,
      "step": 9418
    },
    {
      "epoch": 0.22162352941176472,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1056,
      "step": 9419
    },
    {
      "epoch": 0.22164705882352942,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0665,
      "step": 9420
    },
    {
      "epoch": 0.22167058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0012,
      "step": 9421
    },
    {
      "epoch": 0.22169411764705882,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.025,
      "step": 9422
    },
    {
      "epoch": 0.22171764705882352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2651,
      "step": 9423
    },
    {
      "epoch": 0.22174117647058825,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1171,
      "step": 9424
    },
    {
      "epoch": 0.22176470588235295,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.951,
      "step": 9425
    },
    {
      "epoch": 0.22178823529411765,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0804,
      "step": 9426
    },
    {
      "epoch": 0.22181176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0569,
      "step": 9427
    },
    {
      "epoch": 0.22183529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0408,
      "step": 9428
    },
    {
      "epoch": 0.22185882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0597,
      "step": 9429
    },
    {
      "epoch": 0.22188235294117648,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8211,
      "step": 9430
    },
    {
      "epoch": 0.22190588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2056,
      "step": 9431
    },
    {
      "epoch": 0.22192941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.4765,
      "step": 9432
    },
    {
      "epoch": 0.22195294117647058,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1503,
      "step": 9433
    },
    {
      "epoch": 0.2219764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9868,
      "step": 9434
    },
    {
      "epoch": 0.222,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3362,
      "step": 9435
    },
    {
      "epoch": 0.2220235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0017,
      "step": 9436
    },
    {
      "epoch": 0.2220470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2034,
      "step": 9437
    },
    {
      "epoch": 0.22207058823529413,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.6048,
      "step": 9438
    },
    {
      "epoch": 0.22209411764705883,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.7914,
      "step": 9439
    },
    {
      "epoch": 0.22211764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9848,
      "step": 9440
    },
    {
      "epoch": 0.22214117647058823,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1614,
      "step": 9441
    },
    {
      "epoch": 0.22216470588235293,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2135,
      "step": 9442
    },
    {
      "epoch": 0.22218823529411766,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.207,
      "step": 9443
    },
    {
      "epoch": 0.22221176470588236,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9409,
      "step": 9444
    },
    {
      "epoch": 0.22223529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.023,
      "step": 9445
    },
    {
      "epoch": 0.22225882352941176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9329,
      "step": 9446
    },
    {
      "epoch": 0.22228235294117646,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0079,
      "step": 9447
    },
    {
      "epoch": 0.2223058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0453,
      "step": 9448
    },
    {
      "epoch": 0.2223294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2846,
      "step": 9449
    },
    {
      "epoch": 0.2223529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0978,
      "step": 9450
    },
    {
      "epoch": 0.2223764705882353,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 9451
    },
    {
      "epoch": 0.2224,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1746,
      "step": 9452
    },
    {
      "epoch": 0.22242352941176471,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0826,
      "step": 9453
    },
    {
      "epoch": 0.22244705882352941,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.704,
      "step": 9454
    },
    {
      "epoch": 0.22247058823529411,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3452,
      "step": 9455
    },
    {
      "epoch": 0.22249411764705881,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.7491,
      "step": 9456
    },
    {
      "epoch": 0.22251764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9804,
      "step": 9457
    },
    {
      "epoch": 0.22254117647058824,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.007,
      "step": 9458
    },
    {
      "epoch": 0.22256470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2742,
      "step": 9459
    },
    {
      "epoch": 0.22258823529411764,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0084,
      "step": 9460
    },
    {
      "epoch": 0.22261176470588234,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.7521,
      "step": 9461
    },
    {
      "epoch": 0.22263529411764707,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.2254,
      "step": 9462
    },
    {
      "epoch": 0.22265882352941177,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.124,
      "step": 9463
    },
    {
      "epoch": 0.22268235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0135,
      "step": 9464
    },
    {
      "epoch": 0.22270588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1898,
      "step": 9465
    },
    {
      "epoch": 0.22272941176470587,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1426,
      "step": 9466
    },
    {
      "epoch": 0.2227529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0351,
      "step": 9467
    },
    {
      "epoch": 0.2227764705882353,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9134,
      "step": 9468
    },
    {
      "epoch": 0.2228,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1082,
      "step": 9469
    },
    {
      "epoch": 0.2228235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2315,
      "step": 9470
    },
    {
      "epoch": 0.2228470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2554,
      "step": 9471
    },
    {
      "epoch": 0.22287058823529413,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3795,
      "step": 9472
    },
    {
      "epoch": 0.22289411764705883,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0222,
      "step": 9473
    },
    {
      "epoch": 0.22291764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2942,
      "step": 9474
    },
    {
      "epoch": 0.22294117647058823,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1639,
      "step": 9475
    },
    {
      "epoch": 0.22296470588235295,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2695,
      "step": 9476
    },
    {
      "epoch": 0.22298823529411765,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.085,
      "step": 9477
    },
    {
      "epoch": 0.22301176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1051,
      "step": 9478
    },
    {
      "epoch": 0.22303529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.087,
      "step": 9479
    },
    {
      "epoch": 0.22305882352941175,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0576,
      "step": 9480
    },
    {
      "epoch": 0.22308235294117648,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2591,
      "step": 9481
    },
    {
      "epoch": 0.22310588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9446,
      "step": 9482
    },
    {
      "epoch": 0.22312941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1189,
      "step": 9483
    },
    {
      "epoch": 0.22315294117647058,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0506,
      "step": 9484
    },
    {
      "epoch": 0.22317647058823528,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.0555,
      "step": 9485
    },
    {
      "epoch": 0.2232,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0044,
      "step": 9486
    },
    {
      "epoch": 0.2232235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0315,
      "step": 9487
    },
    {
      "epoch": 0.2232470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9502,
      "step": 9488
    },
    {
      "epoch": 0.2232705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2223,
      "step": 9489
    },
    {
      "epoch": 0.22329411764705884,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 9490
    },
    {
      "epoch": 0.22331764705882354,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1209,
      "step": 9491
    },
    {
      "epoch": 0.22334117647058824,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9792,
      "step": 9492
    },
    {
      "epoch": 0.22336470588235294,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.8282,
      "step": 9493
    },
    {
      "epoch": 0.22338823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8556,
      "step": 9494
    },
    {
      "epoch": 0.22341176470588237,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 1.2858,
      "step": 9495
    },
    {
      "epoch": 0.22343529411764707,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2289,
      "step": 9496
    },
    {
      "epoch": 0.22345882352941177,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1563,
      "step": 9497
    },
    {
      "epoch": 0.22348235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.877,
      "step": 9498
    },
    {
      "epoch": 0.22350588235294117,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2462,
      "step": 9499
    },
    {
      "epoch": 0.2235294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1178,
      "step": 9500
    },
    {
      "epoch": 0.2235529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8725,
      "step": 9501
    },
    {
      "epoch": 0.2235764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7885,
      "step": 9502
    },
    {
      "epoch": 0.2236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9314,
      "step": 9503
    },
    {
      "epoch": 0.2236235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3542,
      "step": 9504
    },
    {
      "epoch": 0.22364705882352942,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.7218,
      "step": 9505
    },
    {
      "epoch": 0.22367058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0571,
      "step": 9506
    },
    {
      "epoch": 0.22369411764705882,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.417,
      "step": 9507
    },
    {
      "epoch": 0.22371764705882352,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9224,
      "step": 9508
    },
    {
      "epoch": 0.22374117647058825,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8703,
      "step": 9509
    },
    {
      "epoch": 0.22376470588235295,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 9510
    },
    {
      "epoch": 0.22378823529411765,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.846,
      "step": 9511
    },
    {
      "epoch": 0.22381176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.089,
      "step": 9512
    },
    {
      "epoch": 0.22383529411764705,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.4186,
      "step": 9513
    },
    {
      "epoch": 0.22385882352941178,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8464,
      "step": 9514
    },
    {
      "epoch": 0.22388235294117648,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9253,
      "step": 9515
    },
    {
      "epoch": 0.22390588235294118,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1556,
      "step": 9516
    },
    {
      "epoch": 0.22392941176470588,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0257,
      "step": 9517
    },
    {
      "epoch": 0.22395294117647058,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8929,
      "step": 9518
    },
    {
      "epoch": 0.2239764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.132,
      "step": 9519
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 9520
    },
    {
      "epoch": 0.2240235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0364,
      "step": 9521
    },
    {
      "epoch": 0.2240470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8836,
      "step": 9522
    },
    {
      "epoch": 0.2240705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2854,
      "step": 9523
    },
    {
      "epoch": 0.22409411764705883,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9036,
      "step": 9524
    },
    {
      "epoch": 0.22411764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1524,
      "step": 9525
    },
    {
      "epoch": 0.22414117647058823,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9888,
      "step": 9526
    },
    {
      "epoch": 0.22416470588235293,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9349,
      "step": 9527
    },
    {
      "epoch": 0.22418823529411766,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8934,
      "step": 9528
    },
    {
      "epoch": 0.22421176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1548,
      "step": 9529
    },
    {
      "epoch": 0.22423529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1941,
      "step": 9530
    },
    {
      "epoch": 0.22425882352941176,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.821,
      "step": 9531
    },
    {
      "epoch": 0.22428235294117646,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3558,
      "step": 9532
    },
    {
      "epoch": 0.2243058823529412,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.9054,
      "step": 9533
    },
    {
      "epoch": 0.2243294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0996,
      "step": 9534
    },
    {
      "epoch": 0.2243529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0326,
      "step": 9535
    },
    {
      "epoch": 0.2243764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4701,
      "step": 9536
    },
    {
      "epoch": 0.2244,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2855,
      "step": 9537
    },
    {
      "epoch": 0.22442352941176472,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8987,
      "step": 9538
    },
    {
      "epoch": 0.22444705882352942,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.8864,
      "step": 9539
    },
    {
      "epoch": 0.22447058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 9540
    },
    {
      "epoch": 0.22449411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 9541
    },
    {
      "epoch": 0.22451764705882352,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2114,
      "step": 9542
    },
    {
      "epoch": 0.22454117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 9543
    },
    {
      "epoch": 0.22456470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0177,
      "step": 9544
    },
    {
      "epoch": 0.22458823529411764,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8352,
      "step": 9545
    },
    {
      "epoch": 0.22461176470588234,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.08,
      "step": 9546
    },
    {
      "epoch": 0.22463529411764707,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.045,
      "step": 9547
    },
    {
      "epoch": 0.22465882352941177,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9504,
      "step": 9548
    },
    {
      "epoch": 0.22468235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1973,
      "step": 9549
    },
    {
      "epoch": 0.22470588235294117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2117,
      "step": 9550
    },
    {
      "epoch": 0.22472941176470587,
      "grad_norm": 0.341796875,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 9551
    },
    {
      "epoch": 0.2247529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2812,
      "step": 9552
    },
    {
      "epoch": 0.2247764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4493,
      "step": 9553
    },
    {
      "epoch": 0.2248,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8727,
      "step": 9554
    },
    {
      "epoch": 0.2248235294117647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.7385,
      "step": 9555
    },
    {
      "epoch": 0.2248470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.4614,
      "step": 9556
    },
    {
      "epoch": 0.22487058823529413,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 9557
    },
    {
      "epoch": 0.22489411764705883,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1643,
      "step": 9558
    },
    {
      "epoch": 0.22491764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9348,
      "step": 9559
    },
    {
      "epoch": 0.22494117647058823,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9072,
      "step": 9560
    },
    {
      "epoch": 0.22496470588235293,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0906,
      "step": 9561
    },
    {
      "epoch": 0.22498823529411766,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9718,
      "step": 9562
    },
    {
      "epoch": 0.22501176470588236,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0068,
      "step": 9563
    },
    {
      "epoch": 0.22503529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2094,
      "step": 9564
    },
    {
      "epoch": 0.22505882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9448,
      "step": 9565
    },
    {
      "epoch": 0.22508235294117648,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9989,
      "step": 9566
    },
    {
      "epoch": 0.22510588235294118,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0034,
      "step": 9567
    },
    {
      "epoch": 0.22512941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1026,
      "step": 9568
    },
    {
      "epoch": 0.22515294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0583,
      "step": 9569
    },
    {
      "epoch": 0.22517647058823528,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1497,
      "step": 9570
    },
    {
      "epoch": 0.2252,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9386,
      "step": 9571
    },
    {
      "epoch": 0.2252235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1425,
      "step": 9572
    },
    {
      "epoch": 0.2252470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1832,
      "step": 9573
    },
    {
      "epoch": 0.2252705882352941,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8417,
      "step": 9574
    },
    {
      "epoch": 0.2252941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1566,
      "step": 9575
    },
    {
      "epoch": 0.22531764705882354,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2204,
      "step": 9576
    },
    {
      "epoch": 0.22534117647058824,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1273,
      "step": 9577
    },
    {
      "epoch": 0.22536470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0047,
      "step": 9578
    },
    {
      "epoch": 0.22538823529411764,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.852,
      "step": 9579
    },
    {
      "epoch": 0.22541176470588234,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8656,
      "step": 9580
    },
    {
      "epoch": 0.22543529411764707,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0186,
      "step": 9581
    },
    {
      "epoch": 0.22545882352941177,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9383,
      "step": 9582
    },
    {
      "epoch": 0.22548235294117647,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.3007,
      "step": 9583
    },
    {
      "epoch": 0.22550588235294117,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1177,
      "step": 9584
    },
    {
      "epoch": 0.2255294117647059,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8887,
      "step": 9585
    },
    {
      "epoch": 0.2255529411764706,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9044,
      "step": 9586
    },
    {
      "epoch": 0.2255764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.4088,
      "step": 9587
    },
    {
      "epoch": 0.2256,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 9588
    },
    {
      "epoch": 0.2256235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3443,
      "step": 9589
    },
    {
      "epoch": 0.22564705882352942,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7962,
      "step": 9590
    },
    {
      "epoch": 0.22567058823529412,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9481,
      "step": 9591
    },
    {
      "epoch": 0.22569411764705882,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0422,
      "step": 9592
    },
    {
      "epoch": 0.22571764705882352,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.4547,
      "step": 9593
    },
    {
      "epoch": 0.22574117647058822,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 9594
    },
    {
      "epoch": 0.22576470588235295,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1314,
      "step": 9595
    },
    {
      "epoch": 0.22578823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1988,
      "step": 9596
    },
    {
      "epoch": 0.22581176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2321,
      "step": 9597
    },
    {
      "epoch": 0.22583529411764705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1821,
      "step": 9598
    },
    {
      "epoch": 0.22585882352941178,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1266,
      "step": 9599
    },
    {
      "epoch": 0.22588235294117648,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9563,
      "step": 9600
    },
    {
      "epoch": 0.22590588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8752,
      "step": 9601
    },
    {
      "epoch": 0.22592941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3495,
      "step": 9602
    },
    {
      "epoch": 0.22595294117647058,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8432,
      "step": 9603
    },
    {
      "epoch": 0.2259764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1349,
      "step": 9604
    },
    {
      "epoch": 0.226,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0084,
      "step": 9605
    },
    {
      "epoch": 0.2260235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0102,
      "step": 9606
    },
    {
      "epoch": 0.2260470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8182,
      "step": 9607
    },
    {
      "epoch": 0.2260705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1928,
      "step": 9608
    },
    {
      "epoch": 0.22609411764705883,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1009,
      "step": 9609
    },
    {
      "epoch": 0.22611764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9131,
      "step": 9610
    },
    {
      "epoch": 0.22614117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1273,
      "step": 9611
    },
    {
      "epoch": 0.22616470588235293,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1052,
      "step": 9612
    },
    {
      "epoch": 0.22618823529411763,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.094,
      "step": 9613
    },
    {
      "epoch": 0.22621176470588236,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1954,
      "step": 9614
    },
    {
      "epoch": 0.22623529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4209,
      "step": 9615
    },
    {
      "epoch": 0.22625882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.057,
      "step": 9616
    },
    {
      "epoch": 0.22628235294117646,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 9617
    },
    {
      "epoch": 0.2263058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.048,
      "step": 9618
    },
    {
      "epoch": 0.2263294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0801,
      "step": 9619
    },
    {
      "epoch": 0.2263529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1836,
      "step": 9620
    },
    {
      "epoch": 0.2263764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2434,
      "step": 9621
    },
    {
      "epoch": 0.2264,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4463,
      "step": 9622
    },
    {
      "epoch": 0.22642352941176472,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1575,
      "step": 9623
    },
    {
      "epoch": 0.22644705882352942,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2469,
      "step": 9624
    },
    {
      "epoch": 0.22647058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 9625
    },
    {
      "epoch": 0.22649411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8978,
      "step": 9626
    },
    {
      "epoch": 0.22651764705882352,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9209,
      "step": 9627
    },
    {
      "epoch": 0.22654117647058825,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 9628
    },
    {
      "epoch": 0.22656470588235295,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1302,
      "step": 9629
    },
    {
      "epoch": 0.22658823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0884,
      "step": 9630
    },
    {
      "epoch": 0.22661176470588235,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9269,
      "step": 9631
    },
    {
      "epoch": 0.22663529411764705,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.266,
      "step": 9632
    },
    {
      "epoch": 0.22665882352941177,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9375,
      "step": 9633
    },
    {
      "epoch": 0.22668235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1933,
      "step": 9634
    },
    {
      "epoch": 0.22670588235294117,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0611,
      "step": 9635
    },
    {
      "epoch": 0.22672941176470587,
      "grad_norm": 0.28125,
      "learning_rate": 0.02,
      "loss": 0.4584,
      "step": 9636
    },
    {
      "epoch": 0.2267529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 9637
    },
    {
      "epoch": 0.2267764705882353,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.9438,
      "step": 9638
    },
    {
      "epoch": 0.2268,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0996,
      "step": 9639
    },
    {
      "epoch": 0.2268235294117647,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9574,
      "step": 9640
    },
    {
      "epoch": 0.2268470588235294,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.9047,
      "step": 9641
    },
    {
      "epoch": 0.22687058823529413,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1955,
      "step": 9642
    },
    {
      "epoch": 0.22689411764705883,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1414,
      "step": 9643
    },
    {
      "epoch": 0.22691764705882353,
      "grad_norm": 0.275390625,
      "learning_rate": 0.02,
      "loss": 0.6246,
      "step": 9644
    },
    {
      "epoch": 0.22694117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.2877,
      "step": 9645
    },
    {
      "epoch": 0.22696470588235293,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.791,
      "step": 9646
    },
    {
      "epoch": 0.22698823529411766,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.7053,
      "step": 9647
    },
    {
      "epoch": 0.22701176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1625,
      "step": 9648
    },
    {
      "epoch": 0.22703529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9267,
      "step": 9649
    },
    {
      "epoch": 0.22705882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8917,
      "step": 9650
    },
    {
      "epoch": 0.22708235294117646,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7772,
      "step": 9651
    },
    {
      "epoch": 0.22710588235294119,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.5119,
      "step": 9652
    },
    {
      "epoch": 0.22712941176470589,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0712,
      "step": 9653
    },
    {
      "epoch": 0.22715294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1583,
      "step": 9654
    },
    {
      "epoch": 0.22717647058823529,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1536,
      "step": 9655
    },
    {
      "epoch": 0.2272,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9658,
      "step": 9656
    },
    {
      "epoch": 0.2272235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1986,
      "step": 9657
    },
    {
      "epoch": 0.2272470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0934,
      "step": 9658
    },
    {
      "epoch": 0.2272705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2422,
      "step": 9659
    },
    {
      "epoch": 0.2272941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8406,
      "step": 9660
    },
    {
      "epoch": 0.22731764705882354,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1178,
      "step": 9661
    },
    {
      "epoch": 0.22734117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 9662
    },
    {
      "epoch": 0.22736470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8163,
      "step": 9663
    },
    {
      "epoch": 0.22738823529411764,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.234,
      "step": 9664
    },
    {
      "epoch": 0.22741176470588234,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.5318,
      "step": 9665
    },
    {
      "epoch": 0.22743529411764707,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9779,
      "step": 9666
    },
    {
      "epoch": 0.22745882352941177,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0122,
      "step": 9667
    },
    {
      "epoch": 0.22748235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0736,
      "step": 9668
    },
    {
      "epoch": 0.22750588235294117,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1712,
      "step": 9669
    },
    {
      "epoch": 0.22752941176470587,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8926,
      "step": 9670
    },
    {
      "epoch": 0.2275529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0835,
      "step": 9671
    },
    {
      "epoch": 0.2275764705882353,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9152,
      "step": 9672
    },
    {
      "epoch": 0.2276,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0508,
      "step": 9673
    },
    {
      "epoch": 0.2276235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0137,
      "step": 9674
    },
    {
      "epoch": 0.22764705882352942,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3392,
      "step": 9675
    },
    {
      "epoch": 0.22767058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.087,
      "step": 9676
    },
    {
      "epoch": 0.22769411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8847,
      "step": 9677
    },
    {
      "epoch": 0.22771764705882352,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9053,
      "step": 9678
    },
    {
      "epoch": 0.22774117647058822,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3222,
      "step": 9679
    },
    {
      "epoch": 0.22776470588235295,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1273,
      "step": 9680
    },
    {
      "epoch": 0.22778823529411765,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0512,
      "step": 9681
    },
    {
      "epoch": 0.22781176470588235,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0874,
      "step": 9682
    },
    {
      "epoch": 0.22783529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0745,
      "step": 9683
    },
    {
      "epoch": 0.22785882352941175,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0533,
      "step": 9684
    },
    {
      "epoch": 0.22788235294117648,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1026,
      "step": 9685
    },
    {
      "epoch": 0.22790588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.4309,
      "step": 9686
    },
    {
      "epoch": 0.22792941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1092,
      "step": 9687
    },
    {
      "epoch": 0.22795294117647058,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9914,
      "step": 9688
    },
    {
      "epoch": 0.22797647058823528,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.7291,
      "step": 9689
    },
    {
      "epoch": 0.228,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9878,
      "step": 9690
    },
    {
      "epoch": 0.2280235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9147,
      "step": 9691
    },
    {
      "epoch": 0.2280470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0497,
      "step": 9692
    },
    {
      "epoch": 0.2280705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1639,
      "step": 9693
    },
    {
      "epoch": 0.22809411764705884,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9825,
      "step": 9694
    },
    {
      "epoch": 0.22811764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.25,
      "step": 9695
    },
    {
      "epoch": 0.22814117647058824,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2007,
      "step": 9696
    },
    {
      "epoch": 0.22816470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9286,
      "step": 9697
    },
    {
      "epoch": 0.22818823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0255,
      "step": 9698
    },
    {
      "epoch": 0.22821176470588236,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9471,
      "step": 9699
    },
    {
      "epoch": 0.22823529411764706,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8772,
      "step": 9700
    },
    {
      "epoch": 0.22825882352941176,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0414,
      "step": 9701
    },
    {
      "epoch": 0.22828235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1643,
      "step": 9702
    },
    {
      "epoch": 0.22830588235294116,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0116,
      "step": 9703
    },
    {
      "epoch": 0.2283294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2301,
      "step": 9704
    },
    {
      "epoch": 0.2283529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8185,
      "step": 9705
    },
    {
      "epoch": 0.2283764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1609,
      "step": 9706
    },
    {
      "epoch": 0.2284,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8493,
      "step": 9707
    },
    {
      "epoch": 0.22842352941176472,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0752,
      "step": 9708
    },
    {
      "epoch": 0.22844705882352942,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.3674,
      "step": 9709
    },
    {
      "epoch": 0.22847058823529412,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8831,
      "step": 9710
    },
    {
      "epoch": 0.22849411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.105,
      "step": 9711
    },
    {
      "epoch": 0.22851764705882352,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2297,
      "step": 9712
    },
    {
      "epoch": 0.22854117647058825,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1796,
      "step": 9713
    },
    {
      "epoch": 0.22856470588235295,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.4221,
      "step": 9714
    },
    {
      "epoch": 0.22858823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0472,
      "step": 9715
    },
    {
      "epoch": 0.22861176470588235,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.6437,
      "step": 9716
    },
    {
      "epoch": 0.22863529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1745,
      "step": 9717
    },
    {
      "epoch": 0.22865882352941178,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.9212,
      "step": 9718
    },
    {
      "epoch": 0.22868235294117648,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8111,
      "step": 9719
    },
    {
      "epoch": 0.22870588235294118,
      "grad_norm": 0.26953125,
      "learning_rate": 0.02,
      "loss": 0.6161,
      "step": 9720
    },
    {
      "epoch": 0.22872941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0121,
      "step": 9721
    },
    {
      "epoch": 0.22875294117647058,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.029,
      "step": 9722
    },
    {
      "epoch": 0.2287764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9402,
      "step": 9723
    },
    {
      "epoch": 0.2288,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2312,
      "step": 9724
    },
    {
      "epoch": 0.2288235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1681,
      "step": 9725
    },
    {
      "epoch": 0.2288470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9721,
      "step": 9726
    },
    {
      "epoch": 0.22887058823529413,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2538,
      "step": 9727
    },
    {
      "epoch": 0.22889411764705883,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8597,
      "step": 9728
    },
    {
      "epoch": 0.22891764705882353,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9369,
      "step": 9729
    },
    {
      "epoch": 0.22894117647058823,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 9730
    },
    {
      "epoch": 0.22896470588235293,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.0047,
      "step": 9731
    },
    {
      "epoch": 0.22898823529411766,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9728,
      "step": 9732
    },
    {
      "epoch": 0.22901176470588236,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8809,
      "step": 9733
    },
    {
      "epoch": 0.22903529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9816,
      "step": 9734
    },
    {
      "epoch": 0.22905882352941176,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.059,
      "step": 9735
    },
    {
      "epoch": 0.22908235294117646,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3391,
      "step": 9736
    },
    {
      "epoch": 0.2291058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3049,
      "step": 9737
    },
    {
      "epoch": 0.2291294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3391,
      "step": 9738
    },
    {
      "epoch": 0.2291529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.095,
      "step": 9739
    },
    {
      "epoch": 0.2291764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0899,
      "step": 9740
    },
    {
      "epoch": 0.2292,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0688,
      "step": 9741
    },
    {
      "epoch": 0.22922352941176471,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9894,
      "step": 9742
    },
    {
      "epoch": 0.22924705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4684,
      "step": 9743
    },
    {
      "epoch": 0.22927058823529411,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.797,
      "step": 9744
    },
    {
      "epoch": 0.22929411764705881,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1919,
      "step": 9745
    },
    {
      "epoch": 0.22931764705882354,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0609,
      "step": 9746
    },
    {
      "epoch": 0.22934117647058824,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9592,
      "step": 9747
    },
    {
      "epoch": 0.22936470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1304,
      "step": 9748
    },
    {
      "epoch": 0.22938823529411764,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1162,
      "step": 9749
    },
    {
      "epoch": 0.22941176470588234,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1125,
      "step": 9750
    },
    {
      "epoch": 0.22943529411764707,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1187,
      "step": 9751
    },
    {
      "epoch": 0.22945882352941177,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1484,
      "step": 9752
    },
    {
      "epoch": 0.22948235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1411,
      "step": 9753
    },
    {
      "epoch": 0.22950588235294117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9202,
      "step": 9754
    },
    {
      "epoch": 0.22952941176470587,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9634,
      "step": 9755
    },
    {
      "epoch": 0.2295529411764706,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8829,
      "step": 9756
    },
    {
      "epoch": 0.2295764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1189,
      "step": 9757
    },
    {
      "epoch": 0.2296,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0688,
      "step": 9758
    },
    {
      "epoch": 0.2296235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0253,
      "step": 9759
    },
    {
      "epoch": 0.2296470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0619,
      "step": 9760
    },
    {
      "epoch": 0.22967058823529413,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1076,
      "step": 9761
    },
    {
      "epoch": 0.22969411764705883,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.267,
      "step": 9762
    },
    {
      "epoch": 0.22971764705882353,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.7449,
      "step": 9763
    },
    {
      "epoch": 0.22974117647058823,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8964,
      "step": 9764
    },
    {
      "epoch": 0.22976470588235295,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0568,
      "step": 9765
    },
    {
      "epoch": 0.22978823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0999,
      "step": 9766
    },
    {
      "epoch": 0.22981176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0696,
      "step": 9767
    },
    {
      "epoch": 0.22983529411764705,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2619,
      "step": 9768
    },
    {
      "epoch": 0.22985882352941175,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9435,
      "step": 9769
    },
    {
      "epoch": 0.22988235294117648,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0371,
      "step": 9770
    },
    {
      "epoch": 0.22990588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0,
      "step": 9771
    },
    {
      "epoch": 0.22992941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1233,
      "step": 9772
    },
    {
      "epoch": 0.22995294117647058,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9804,
      "step": 9773
    },
    {
      "epoch": 0.22997647058823528,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9091,
      "step": 9774
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2872,
      "step": 9775
    },
    {
      "epoch": 0.2300235294117647,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7121,
      "step": 9776
    },
    {
      "epoch": 0.2300470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1715,
      "step": 9777
    },
    {
      "epoch": 0.2300705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1383,
      "step": 9778
    },
    {
      "epoch": 0.2300941176470588,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0795,
      "step": 9779
    },
    {
      "epoch": 0.23011764705882354,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.087,
      "step": 9780
    },
    {
      "epoch": 0.23014117647058824,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3045,
      "step": 9781
    },
    {
      "epoch": 0.23016470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0141,
      "step": 9782
    },
    {
      "epoch": 0.23018823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 9783
    },
    {
      "epoch": 0.23021176470588237,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2903,
      "step": 9784
    },
    {
      "epoch": 0.23023529411764707,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2829,
      "step": 9785
    },
    {
      "epoch": 0.23025882352941177,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1255,
      "step": 9786
    },
    {
      "epoch": 0.23028235294117647,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9775,
      "step": 9787
    },
    {
      "epoch": 0.23030588235294117,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9345,
      "step": 9788
    },
    {
      "epoch": 0.2303294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1514,
      "step": 9789
    },
    {
      "epoch": 0.2303529411764706,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.1511,
      "step": 9790
    },
    {
      "epoch": 0.2303764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.1412,
      "step": 9791
    },
    {
      "epoch": 0.2304,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8529,
      "step": 9792
    },
    {
      "epoch": 0.2304235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2183,
      "step": 9793
    },
    {
      "epoch": 0.23044705882352942,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8109,
      "step": 9794
    },
    {
      "epoch": 0.23047058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.833,
      "step": 9795
    },
    {
      "epoch": 0.23049411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9283,
      "step": 9796
    },
    {
      "epoch": 0.23051764705882352,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.1363,
      "step": 9797
    },
    {
      "epoch": 0.23054117647058822,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9152,
      "step": 9798
    },
    {
      "epoch": 0.23056470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0624,
      "step": 9799
    },
    {
      "epoch": 0.23058823529411765,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8116,
      "step": 9800
    },
    {
      "epoch": 0.23061176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3378,
      "step": 9801
    },
    {
      "epoch": 0.23063529411764705,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9798,
      "step": 9802
    },
    {
      "epoch": 0.23065882352941178,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7927,
      "step": 9803
    },
    {
      "epoch": 0.23068235294117648,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0023,
      "step": 9804
    },
    {
      "epoch": 0.23070588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1095,
      "step": 9805
    },
    {
      "epoch": 0.23072941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2475,
      "step": 9806
    },
    {
      "epoch": 0.23075294117647058,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0582,
      "step": 9807
    },
    {
      "epoch": 0.2307764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1795,
      "step": 9808
    },
    {
      "epoch": 0.2308,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1294,
      "step": 9809
    },
    {
      "epoch": 0.2308235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 9810
    },
    {
      "epoch": 0.2308470588235294,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3494,
      "step": 9811
    },
    {
      "epoch": 0.2308705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1218,
      "step": 9812
    },
    {
      "epoch": 0.23089411764705883,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0982,
      "step": 9813
    },
    {
      "epoch": 0.23091764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1632,
      "step": 9814
    },
    {
      "epoch": 0.23094117647058823,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1003,
      "step": 9815
    },
    {
      "epoch": 0.23096470588235293,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0038,
      "step": 9816
    },
    {
      "epoch": 0.23098823529411766,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0844,
      "step": 9817
    },
    {
      "epoch": 0.23101176470588236,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2348,
      "step": 9818
    },
    {
      "epoch": 0.23103529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 9819
    },
    {
      "epoch": 0.23105882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0743,
      "step": 9820
    },
    {
      "epoch": 0.23108235294117646,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8309,
      "step": 9821
    },
    {
      "epoch": 0.2311058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9534,
      "step": 9822
    },
    {
      "epoch": 0.2311294117647059,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7438,
      "step": 9823
    },
    {
      "epoch": 0.2311529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.027,
      "step": 9824
    },
    {
      "epoch": 0.2311764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 9825
    },
    {
      "epoch": 0.2312,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0887,
      "step": 9826
    },
    {
      "epoch": 0.23122352941176472,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.011,
      "step": 9827
    },
    {
      "epoch": 0.23124705882352942,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1812,
      "step": 9828
    },
    {
      "epoch": 0.23127058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4715,
      "step": 9829
    },
    {
      "epoch": 0.23129411764705882,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9928,
      "step": 9830
    },
    {
      "epoch": 0.23131764705882352,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9146,
      "step": 9831
    },
    {
      "epoch": 0.23134117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 9832
    },
    {
      "epoch": 0.23136470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9017,
      "step": 9833
    },
    {
      "epoch": 0.23138823529411764,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8479,
      "step": 9834
    },
    {
      "epoch": 0.23141176470588234,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.306,
      "step": 9835
    },
    {
      "epoch": 0.23143529411764707,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8086,
      "step": 9836
    },
    {
      "epoch": 0.23145882352941177,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7283,
      "step": 9837
    },
    {
      "epoch": 0.23148235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8436,
      "step": 9838
    },
    {
      "epoch": 0.23150588235294117,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2472,
      "step": 9839
    },
    {
      "epoch": 0.23152941176470587,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8508,
      "step": 9840
    },
    {
      "epoch": 0.2315529411764706,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1153,
      "step": 9841
    },
    {
      "epoch": 0.2315764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1808,
      "step": 9842
    },
    {
      "epoch": 0.2316,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.1568,
      "step": 9843
    },
    {
      "epoch": 0.2316235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7922,
      "step": 9844
    },
    {
      "epoch": 0.2316470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.999,
      "step": 9845
    },
    {
      "epoch": 0.23167058823529413,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.725,
      "step": 9846
    },
    {
      "epoch": 0.23169411764705883,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2678,
      "step": 9847
    },
    {
      "epoch": 0.23171764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.261,
      "step": 9848
    },
    {
      "epoch": 0.23174117647058823,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8523,
      "step": 9849
    },
    {
      "epoch": 0.23176470588235293,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 9850
    },
    {
      "epoch": 0.23178823529411766,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1614,
      "step": 9851
    },
    {
      "epoch": 0.23181176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.888,
      "step": 9852
    },
    {
      "epoch": 0.23183529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2198,
      "step": 9853
    },
    {
      "epoch": 0.23185882352941176,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1733,
      "step": 9854
    },
    {
      "epoch": 0.23188235294117648,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8637,
      "step": 9855
    },
    {
      "epoch": 0.23190588235294118,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8232,
      "step": 9856
    },
    {
      "epoch": 0.23192941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2227,
      "step": 9857
    },
    {
      "epoch": 0.23195294117647058,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.294,
      "step": 9858
    },
    {
      "epoch": 0.23197647058823528,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.7728,
      "step": 9859
    },
    {
      "epoch": 0.232,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2025,
      "step": 9860
    },
    {
      "epoch": 0.2320235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2676,
      "step": 9861
    },
    {
      "epoch": 0.2320470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9897,
      "step": 9862
    },
    {
      "epoch": 0.2320705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.905,
      "step": 9863
    },
    {
      "epoch": 0.2320941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0739,
      "step": 9864
    },
    {
      "epoch": 0.23211764705882354,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9391,
      "step": 9865
    },
    {
      "epoch": 0.23214117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2479,
      "step": 9866
    },
    {
      "epoch": 0.23216470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1789,
      "step": 9867
    },
    {
      "epoch": 0.23218823529411764,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3257,
      "step": 9868
    },
    {
      "epoch": 0.23221176470588234,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2,
      "step": 9869
    },
    {
      "epoch": 0.23223529411764707,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 9870
    },
    {
      "epoch": 0.23225882352941177,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1424,
      "step": 9871
    },
    {
      "epoch": 0.23228235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0012,
      "step": 9872
    },
    {
      "epoch": 0.23230588235294117,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2628,
      "step": 9873
    },
    {
      "epoch": 0.2323294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 9874
    },
    {
      "epoch": 0.2323529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9427,
      "step": 9875
    },
    {
      "epoch": 0.2323764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0184,
      "step": 9876
    },
    {
      "epoch": 0.2324,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 9877
    },
    {
      "epoch": 0.2324235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.915,
      "step": 9878
    },
    {
      "epoch": 0.23244705882352942,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1709,
      "step": 9879
    },
    {
      "epoch": 0.23247058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9979,
      "step": 9880
    },
    {
      "epoch": 0.23249411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1681,
      "step": 9881
    },
    {
      "epoch": 0.23251764705882352,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8545,
      "step": 9882
    },
    {
      "epoch": 0.23254117647058822,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7363,
      "step": 9883
    },
    {
      "epoch": 0.23256470588235295,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7538,
      "step": 9884
    },
    {
      "epoch": 0.23258823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.155,
      "step": 9885
    },
    {
      "epoch": 0.23261176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3154,
      "step": 9886
    },
    {
      "epoch": 0.23263529411764705,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0902,
      "step": 9887
    },
    {
      "epoch": 0.23265882352941175,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1924,
      "step": 9888
    },
    {
      "epoch": 0.23268235294117648,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0134,
      "step": 9889
    },
    {
      "epoch": 0.23270588235294118,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1213,
      "step": 9890
    },
    {
      "epoch": 0.23272941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0755,
      "step": 9891
    },
    {
      "epoch": 0.23275294117647058,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0949,
      "step": 9892
    },
    {
      "epoch": 0.2327764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0877,
      "step": 9893
    },
    {
      "epoch": 0.2328,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8685,
      "step": 9894
    },
    {
      "epoch": 0.2328235294117647,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.958,
      "step": 9895
    },
    {
      "epoch": 0.2328470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0598,
      "step": 9896
    },
    {
      "epoch": 0.2328705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9658,
      "step": 9897
    },
    {
      "epoch": 0.23289411764705883,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7611,
      "step": 9898
    },
    {
      "epoch": 0.23291764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2272,
      "step": 9899
    },
    {
      "epoch": 0.23294117647058823,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8833,
      "step": 9900
    },
    {
      "epoch": 0.23296470588235293,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1347,
      "step": 9901
    },
    {
      "epoch": 0.23298823529411763,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1685,
      "step": 9902
    },
    {
      "epoch": 0.23301176470588236,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.6405,
      "step": 9903
    },
    {
      "epoch": 0.23303529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.893,
      "step": 9904
    },
    {
      "epoch": 0.23305882352941176,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2433,
      "step": 9905
    },
    {
      "epoch": 0.23308235294117646,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9052,
      "step": 9906
    },
    {
      "epoch": 0.23310588235294116,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9619,
      "step": 9907
    },
    {
      "epoch": 0.2331294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9356,
      "step": 9908
    },
    {
      "epoch": 0.2331529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1102,
      "step": 9909
    },
    {
      "epoch": 0.2331764705882353,
      "grad_norm": 0.29296875,
      "learning_rate": 0.02,
      "loss": 0.5685,
      "step": 9910
    },
    {
      "epoch": 0.2332,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0063,
      "step": 9911
    },
    {
      "epoch": 0.23322352941176472,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8673,
      "step": 9912
    },
    {
      "epoch": 0.23324705882352942,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.345,
      "step": 9913
    },
    {
      "epoch": 0.23327058823529412,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 9914
    },
    {
      "epoch": 0.23329411764705882,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0831,
      "step": 9915
    },
    {
      "epoch": 0.23331764705882352,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0369,
      "step": 9916
    },
    {
      "epoch": 0.23334117647058825,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0503,
      "step": 9917
    },
    {
      "epoch": 0.23336470588235295,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.002,
      "step": 9918
    },
    {
      "epoch": 0.23338823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0656,
      "step": 9919
    },
    {
      "epoch": 0.23341176470588235,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8851,
      "step": 9920
    },
    {
      "epoch": 0.23343529411764705,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 9921
    },
    {
      "epoch": 0.23345882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1131,
      "step": 9922
    },
    {
      "epoch": 0.23348235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3199,
      "step": 9923
    },
    {
      "epoch": 0.23350588235294117,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2069,
      "step": 9924
    },
    {
      "epoch": 0.23352941176470587,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9041,
      "step": 9925
    },
    {
      "epoch": 0.2335529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3487,
      "step": 9926
    },
    {
      "epoch": 0.2335764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1502,
      "step": 9927
    },
    {
      "epoch": 0.2336,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.812,
      "step": 9928
    },
    {
      "epoch": 0.2336235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1039,
      "step": 9929
    },
    {
      "epoch": 0.2336470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1085,
      "step": 9930
    },
    {
      "epoch": 0.23367058823529413,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.9534,
      "step": 9931
    },
    {
      "epoch": 0.23369411764705883,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9402,
      "step": 9932
    },
    {
      "epoch": 0.23371764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8916,
      "step": 9933
    },
    {
      "epoch": 0.23374117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9585,
      "step": 9934
    },
    {
      "epoch": 0.23376470588235293,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9508,
      "step": 9935
    },
    {
      "epoch": 0.23378823529411766,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2254,
      "step": 9936
    },
    {
      "epoch": 0.23381176470588236,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.9101,
      "step": 9937
    },
    {
      "epoch": 0.23383529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1386,
      "step": 9938
    },
    {
      "epoch": 0.23385882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3668,
      "step": 9939
    },
    {
      "epoch": 0.23388235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1159,
      "step": 9940
    },
    {
      "epoch": 0.23390588235294119,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3898,
      "step": 9941
    },
    {
      "epoch": 0.23392941176470589,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1457,
      "step": 9942
    },
    {
      "epoch": 0.23395294117647059,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1715,
      "step": 9943
    },
    {
      "epoch": 0.23397647058823529,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9164,
      "step": 9944
    },
    {
      "epoch": 0.234,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1386,
      "step": 9945
    },
    {
      "epoch": 0.2340235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2398,
      "step": 9946
    },
    {
      "epoch": 0.2340470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.6258,
      "step": 9947
    },
    {
      "epoch": 0.2340705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1764,
      "step": 9948
    },
    {
      "epoch": 0.2340941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0764,
      "step": 9949
    },
    {
      "epoch": 0.23411764705882354,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0386,
      "step": 9950
    },
    {
      "epoch": 0.23414117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.953,
      "step": 9951
    },
    {
      "epoch": 0.23416470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1522,
      "step": 9952
    },
    {
      "epoch": 0.23418823529411764,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0257,
      "step": 9953
    },
    {
      "epoch": 0.23421176470588234,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9966,
      "step": 9954
    },
    {
      "epoch": 0.23423529411764707,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9285,
      "step": 9955
    },
    {
      "epoch": 0.23425882352941177,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0026,
      "step": 9956
    },
    {
      "epoch": 0.23428235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2884,
      "step": 9957
    },
    {
      "epoch": 0.23430588235294117,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0853,
      "step": 9958
    },
    {
      "epoch": 0.23432941176470587,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 9959
    },
    {
      "epoch": 0.2343529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 9960
    },
    {
      "epoch": 0.2343764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1042,
      "step": 9961
    },
    {
      "epoch": 0.2344,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2287,
      "step": 9962
    },
    {
      "epoch": 0.2344235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9065,
      "step": 9963
    },
    {
      "epoch": 0.23444705882352943,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 1.0019,
      "step": 9964
    },
    {
      "epoch": 0.23447058823529413,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7555,
      "step": 9965
    },
    {
      "epoch": 0.23449411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9981,
      "step": 9966
    },
    {
      "epoch": 0.23451764705882353,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.7459,
      "step": 9967
    },
    {
      "epoch": 0.23454117647058823,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.803,
      "step": 9968
    },
    {
      "epoch": 0.23456470588235295,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2011,
      "step": 9969
    },
    {
      "epoch": 0.23458823529411765,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.4982,
      "step": 9970
    },
    {
      "epoch": 0.23461176470588235,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0867,
      "step": 9971
    },
    {
      "epoch": 0.23463529411764705,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.892,
      "step": 9972
    },
    {
      "epoch": 0.23465882352941175,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.84,
      "step": 9973
    },
    {
      "epoch": 0.23468235294117648,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1401,
      "step": 9974
    },
    {
      "epoch": 0.23470588235294118,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8585,
      "step": 9975
    },
    {
      "epoch": 0.23472941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2936,
      "step": 9976
    },
    {
      "epoch": 0.23475294117647058,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2545,
      "step": 9977
    },
    {
      "epoch": 0.23477647058823528,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8979,
      "step": 9978
    },
    {
      "epoch": 0.2348,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0879,
      "step": 9979
    },
    {
      "epoch": 0.2348235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2426,
      "step": 9980
    },
    {
      "epoch": 0.2348470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9225,
      "step": 9981
    },
    {
      "epoch": 0.2348705882352941,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0187,
      "step": 9982
    },
    {
      "epoch": 0.23489411764705884,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0448,
      "step": 9983
    },
    {
      "epoch": 0.23491764705882354,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1514,
      "step": 9984
    },
    {
      "epoch": 0.23494117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 9985
    },
    {
      "epoch": 0.23496470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8804,
      "step": 9986
    },
    {
      "epoch": 0.23498823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0604,
      "step": 9987
    },
    {
      "epoch": 0.23501176470588236,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 1.0615,
      "step": 9988
    },
    {
      "epoch": 0.23503529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0984,
      "step": 9989
    },
    {
      "epoch": 0.23505882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1008,
      "step": 9990
    },
    {
      "epoch": 0.23508235294117646,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.8487,
      "step": 9991
    },
    {
      "epoch": 0.23510588235294116,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9274,
      "step": 9992
    },
    {
      "epoch": 0.2351294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1463,
      "step": 9993
    },
    {
      "epoch": 0.2351529411764706,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9231,
      "step": 9994
    },
    {
      "epoch": 0.2351764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.1027,
      "step": 9995
    },
    {
      "epoch": 0.2352,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9043,
      "step": 9996
    },
    {
      "epoch": 0.2352235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.042,
      "step": 9997
    },
    {
      "epoch": 0.23524705882352942,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2801,
      "step": 9998
    },
    {
      "epoch": 0.23527058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.6812,
      "step": 9999
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.124,
      "step": 10000
    },
    {
      "epoch": 0.23529411764705882,
      "eval_loss": 2.219313383102417,
      "eval_runtime": 686.0292,
      "eval_samples_per_second": 12.39,
      "eval_steps_per_second": 3.098,
      "step": 10000
    },
    {
      "epoch": 0.23531764705882352,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1444,
      "step": 10001
    },
    {
      "epoch": 0.23534117647058825,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.7349,
      "step": 10002
    },
    {
      "epoch": 0.23536470588235295,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 10003
    },
    {
      "epoch": 0.23538823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7281,
      "step": 10004
    },
    {
      "epoch": 0.23541176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1431,
      "step": 10005
    },
    {
      "epoch": 0.23543529411764705,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0678,
      "step": 10006
    },
    {
      "epoch": 0.23545882352941178,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1104,
      "step": 10007
    },
    {
      "epoch": 0.23548235294117648,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.994,
      "step": 10008
    },
    {
      "epoch": 0.23550588235294118,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0719,
      "step": 10009
    },
    {
      "epoch": 0.23552941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9659,
      "step": 10010
    },
    {
      "epoch": 0.23555294117647058,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1642,
      "step": 10011
    },
    {
      "epoch": 0.2355764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3394,
      "step": 10012
    },
    {
      "epoch": 0.2356,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9419,
      "step": 10013
    },
    {
      "epoch": 0.2356235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1303,
      "step": 10014
    },
    {
      "epoch": 0.2356470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9233,
      "step": 10015
    },
    {
      "epoch": 0.2356705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.946,
      "step": 10016
    },
    {
      "epoch": 0.23569411764705883,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8979,
      "step": 10017
    },
    {
      "epoch": 0.23571764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0079,
      "step": 10018
    },
    {
      "epoch": 0.23574117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2172,
      "step": 10019
    },
    {
      "epoch": 0.23576470588235293,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2519,
      "step": 10020
    },
    {
      "epoch": 0.23578823529411766,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3118,
      "step": 10021
    },
    {
      "epoch": 0.23581176470588236,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0633,
      "step": 10022
    },
    {
      "epoch": 0.23583529411764706,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0856,
      "step": 10023
    },
    {
      "epoch": 0.23585882352941176,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8719,
      "step": 10024
    },
    {
      "epoch": 0.23588235294117646,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2692,
      "step": 10025
    },
    {
      "epoch": 0.2359058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0278,
      "step": 10026
    },
    {
      "epoch": 0.2359294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.875,
      "step": 10027
    },
    {
      "epoch": 0.2359529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.154,
      "step": 10028
    },
    {
      "epoch": 0.2359764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 10029
    },
    {
      "epoch": 0.236,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.099,
      "step": 10030
    },
    {
      "epoch": 0.23602352941176472,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9711,
      "step": 10031
    },
    {
      "epoch": 0.23604705882352942,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8237,
      "step": 10032
    },
    {
      "epoch": 0.23607058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.6936,
      "step": 10033
    },
    {
      "epoch": 0.23609411764705882,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.6024,
      "step": 10034
    },
    {
      "epoch": 0.23611764705882354,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.6447,
      "step": 10035
    },
    {
      "epoch": 0.23614117647058824,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.066,
      "step": 10036
    },
    {
      "epoch": 0.23616470588235294,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.7706,
      "step": 10037
    },
    {
      "epoch": 0.23618823529411764,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8848,
      "step": 10038
    },
    {
      "epoch": 0.23621176470588234,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.148,
      "step": 10039
    },
    {
      "epoch": 0.23623529411764707,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0541,
      "step": 10040
    },
    {
      "epoch": 0.23625882352941177,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2567,
      "step": 10041
    },
    {
      "epoch": 0.23628235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.866,
      "step": 10042
    },
    {
      "epoch": 0.23630588235294117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8984,
      "step": 10043
    },
    {
      "epoch": 0.23632941176470587,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1894,
      "step": 10044
    },
    {
      "epoch": 0.2363529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1564,
      "step": 10045
    },
    {
      "epoch": 0.2363764705882353,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8633,
      "step": 10046
    },
    {
      "epoch": 0.2364,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8222,
      "step": 10047
    },
    {
      "epoch": 0.2364235294117647,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9093,
      "step": 10048
    },
    {
      "epoch": 0.2364470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9137,
      "step": 10049
    },
    {
      "epoch": 0.23647058823529413,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.502,
      "step": 10050
    },
    {
      "epoch": 0.23649411764705883,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0863,
      "step": 10051
    },
    {
      "epoch": 0.23651764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8318,
      "step": 10052
    },
    {
      "epoch": 0.23654117647058823,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1379,
      "step": 10053
    },
    {
      "epoch": 0.23656470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 10054
    },
    {
      "epoch": 0.23658823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9023,
      "step": 10055
    },
    {
      "epoch": 0.23661176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0259,
      "step": 10056
    },
    {
      "epoch": 0.23663529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1064,
      "step": 10057
    },
    {
      "epoch": 0.23665882352941175,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.842,
      "step": 10058
    },
    {
      "epoch": 0.23668235294117648,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0536,
      "step": 10059
    },
    {
      "epoch": 0.23670588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 10060
    },
    {
      "epoch": 0.23672941176470588,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.4084,
      "step": 10061
    },
    {
      "epoch": 0.23675294117647058,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9434,
      "step": 10062
    },
    {
      "epoch": 0.23677647058823528,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2385,
      "step": 10063
    },
    {
      "epoch": 0.2368,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.4367,
      "step": 10064
    },
    {
      "epoch": 0.2368235294117647,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.7498,
      "step": 10065
    },
    {
      "epoch": 0.2368470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1096,
      "step": 10066
    },
    {
      "epoch": 0.2368705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8823,
      "step": 10067
    },
    {
      "epoch": 0.2368941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2572,
      "step": 10068
    },
    {
      "epoch": 0.23691764705882354,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1494,
      "step": 10069
    },
    {
      "epoch": 0.23694117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.2993,
      "step": 10070
    },
    {
      "epoch": 0.23696470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.1739,
      "step": 10071
    },
    {
      "epoch": 0.23698823529411764,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.0034,
      "step": 10072
    },
    {
      "epoch": 0.23701176470588237,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9201,
      "step": 10073
    },
    {
      "epoch": 0.23703529411764707,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.3181,
      "step": 10074
    },
    {
      "epoch": 0.23705882352941177,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.8128,
      "step": 10075
    },
    {
      "epoch": 0.23708235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0901,
      "step": 10076
    },
    {
      "epoch": 0.23710588235294117,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1598,
      "step": 10077
    },
    {
      "epoch": 0.2371294117647059,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9845,
      "step": 10078
    },
    {
      "epoch": 0.2371529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0436,
      "step": 10079
    },
    {
      "epoch": 0.2371764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0752,
      "step": 10080
    },
    {
      "epoch": 0.2372,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.5281,
      "step": 10081
    },
    {
      "epoch": 0.2372235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2373,
      "step": 10082
    },
    {
      "epoch": 0.23724705882352942,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1431,
      "step": 10083
    },
    {
      "epoch": 0.23727058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0812,
      "step": 10084
    },
    {
      "epoch": 0.23729411764705882,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9188,
      "step": 10085
    },
    {
      "epoch": 0.23731764705882352,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8779,
      "step": 10086
    },
    {
      "epoch": 0.23734117647058822,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 10087
    },
    {
      "epoch": 0.23736470588235295,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0992,
      "step": 10088
    },
    {
      "epoch": 0.23738823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0207,
      "step": 10089
    },
    {
      "epoch": 0.23741176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1369,
      "step": 10090
    },
    {
      "epoch": 0.23743529411764705,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.5127,
      "step": 10091
    },
    {
      "epoch": 0.23745882352941178,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9429,
      "step": 10092
    },
    {
      "epoch": 0.23748235294117648,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.209,
      "step": 10093
    },
    {
      "epoch": 0.23750588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2912,
      "step": 10094
    },
    {
      "epoch": 0.23752941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.7612,
      "step": 10095
    },
    {
      "epoch": 0.23755294117647058,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2739,
      "step": 10096
    },
    {
      "epoch": 0.2375764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2039,
      "step": 10097
    },
    {
      "epoch": 0.2376,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0507,
      "step": 10098
    },
    {
      "epoch": 0.2376235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.998,
      "step": 10099
    },
    {
      "epoch": 0.2376470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2465,
      "step": 10100
    },
    {
      "epoch": 0.2376705882352941,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3149,
      "step": 10101
    },
    {
      "epoch": 0.23769411764705883,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2602,
      "step": 10102
    },
    {
      "epoch": 0.23771764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9521,
      "step": 10103
    },
    {
      "epoch": 0.23774117647058823,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9657,
      "step": 10104
    },
    {
      "epoch": 0.23776470588235293,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.6626,
      "step": 10105
    },
    {
      "epoch": 0.23778823529411763,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.3021,
      "step": 10106
    },
    {
      "epoch": 0.23781176470588236,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1935,
      "step": 10107
    },
    {
      "epoch": 0.23783529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1544,
      "step": 10108
    },
    {
      "epoch": 0.23785882352941176,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.976,
      "step": 10109
    },
    {
      "epoch": 0.23788235294117646,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7743,
      "step": 10110
    },
    {
      "epoch": 0.2379058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9156,
      "step": 10111
    },
    {
      "epoch": 0.2379294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 10112
    },
    {
      "epoch": 0.2379529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9167,
      "step": 10113
    },
    {
      "epoch": 0.2379764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1141,
      "step": 10114
    },
    {
      "epoch": 0.238,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9518,
      "step": 10115
    },
    {
      "epoch": 0.23802352941176472,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1921,
      "step": 10116
    },
    {
      "epoch": 0.23804705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 10117
    },
    {
      "epoch": 0.23807058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2956,
      "step": 10118
    },
    {
      "epoch": 0.23809411764705882,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7299,
      "step": 10119
    },
    {
      "epoch": 0.23811764705882352,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1392,
      "step": 10120
    },
    {
      "epoch": 0.23814117647058825,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0897,
      "step": 10121
    },
    {
      "epoch": 0.23816470588235295,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.937,
      "step": 10122
    },
    {
      "epoch": 0.23818823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.224,
      "step": 10123
    },
    {
      "epoch": 0.23821176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.237,
      "step": 10124
    },
    {
      "epoch": 0.23823529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2174,
      "step": 10125
    },
    {
      "epoch": 0.23825882352941177,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9535,
      "step": 10126
    },
    {
      "epoch": 0.23828235294117647,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8366,
      "step": 10127
    },
    {
      "epoch": 0.23830588235294117,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2294,
      "step": 10128
    },
    {
      "epoch": 0.23832941176470587,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0402,
      "step": 10129
    },
    {
      "epoch": 0.2383529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2471,
      "step": 10130
    },
    {
      "epoch": 0.2383764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3236,
      "step": 10131
    },
    {
      "epoch": 0.2384,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8326,
      "step": 10132
    },
    {
      "epoch": 0.2384235294117647,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7933,
      "step": 10133
    },
    {
      "epoch": 0.2384470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9218,
      "step": 10134
    },
    {
      "epoch": 0.23847058823529413,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9343,
      "step": 10135
    },
    {
      "epoch": 0.23849411764705883,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0363,
      "step": 10136
    },
    {
      "epoch": 0.23851764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.2446,
      "step": 10137
    },
    {
      "epoch": 0.23854117647058823,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0706,
      "step": 10138
    },
    {
      "epoch": 0.23856470588235293,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9,
      "step": 10139
    },
    {
      "epoch": 0.23858823529411766,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2824,
      "step": 10140
    },
    {
      "epoch": 0.23861176470588236,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.5726,
      "step": 10141
    },
    {
      "epoch": 0.23863529411764706,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0896,
      "step": 10142
    },
    {
      "epoch": 0.23865882352941176,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 10143
    },
    {
      "epoch": 0.23868235294117648,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2483,
      "step": 10144
    },
    {
      "epoch": 0.23870588235294118,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8831,
      "step": 10145
    },
    {
      "epoch": 0.23872941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0695,
      "step": 10146
    },
    {
      "epoch": 0.23875294117647058,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9467,
      "step": 10147
    },
    {
      "epoch": 0.23877647058823528,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8174,
      "step": 10148
    },
    {
      "epoch": 0.2388,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1532,
      "step": 10149
    },
    {
      "epoch": 0.2388235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2997,
      "step": 10150
    },
    {
      "epoch": 0.2388470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0526,
      "step": 10151
    },
    {
      "epoch": 0.2388705882352941,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2757,
      "step": 10152
    },
    {
      "epoch": 0.2388941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1197,
      "step": 10153
    },
    {
      "epoch": 0.23891764705882354,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0337,
      "step": 10154
    },
    {
      "epoch": 0.23894117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9824,
      "step": 10155
    },
    {
      "epoch": 0.23896470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8328,
      "step": 10156
    },
    {
      "epoch": 0.23898823529411764,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0513,
      "step": 10157
    },
    {
      "epoch": 0.23901176470588234,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0251,
      "step": 10158
    },
    {
      "epoch": 0.23903529411764707,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2163,
      "step": 10159
    },
    {
      "epoch": 0.23905882352941177,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9782,
      "step": 10160
    },
    {
      "epoch": 0.23908235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8792,
      "step": 10161
    },
    {
      "epoch": 0.23910588235294117,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1973,
      "step": 10162
    },
    {
      "epoch": 0.2391294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9698,
      "step": 10163
    },
    {
      "epoch": 0.2391529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0646,
      "step": 10164
    },
    {
      "epoch": 0.2391764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.853,
      "step": 10165
    },
    {
      "epoch": 0.2392,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8241,
      "step": 10166
    },
    {
      "epoch": 0.2392235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8062,
      "step": 10167
    },
    {
      "epoch": 0.23924705882352942,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0658,
      "step": 10168
    },
    {
      "epoch": 0.23927058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0849,
      "step": 10169
    },
    {
      "epoch": 0.23929411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2617,
      "step": 10170
    },
    {
      "epoch": 0.23931764705882352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8358,
      "step": 10171
    },
    {
      "epoch": 0.23934117647058822,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9266,
      "step": 10172
    },
    {
      "epoch": 0.23936470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1039,
      "step": 10173
    },
    {
      "epoch": 0.23938823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2775,
      "step": 10174
    },
    {
      "epoch": 0.23941176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1437,
      "step": 10175
    },
    {
      "epoch": 0.23943529411764705,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 10176
    },
    {
      "epoch": 0.23945882352941175,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 10177
    },
    {
      "epoch": 0.23948235294117648,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1516,
      "step": 10178
    },
    {
      "epoch": 0.23950588235294118,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8137,
      "step": 10179
    },
    {
      "epoch": 0.23952941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0786,
      "step": 10180
    },
    {
      "epoch": 0.23955294117647058,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1603,
      "step": 10181
    },
    {
      "epoch": 0.2395764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0769,
      "step": 10182
    },
    {
      "epoch": 0.2396,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1799,
      "step": 10183
    },
    {
      "epoch": 0.2396235294117647,
      "grad_norm": 0.267578125,
      "learning_rate": 0.02,
      "loss": 0.6649,
      "step": 10184
    },
    {
      "epoch": 0.2396470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2472,
      "step": 10185
    },
    {
      "epoch": 0.2396705882352941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1914,
      "step": 10186
    },
    {
      "epoch": 0.23969411764705884,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7508,
      "step": 10187
    },
    {
      "epoch": 0.23971764705882354,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9607,
      "step": 10188
    },
    {
      "epoch": 0.23974117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.866,
      "step": 10189
    },
    {
      "epoch": 0.23976470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9559,
      "step": 10190
    },
    {
      "epoch": 0.23978823529411764,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1446,
      "step": 10191
    },
    {
      "epoch": 0.23981176470588236,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9559,
      "step": 10192
    },
    {
      "epoch": 0.23983529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3304,
      "step": 10193
    },
    {
      "epoch": 0.23985882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9741,
      "step": 10194
    },
    {
      "epoch": 0.23988235294117646,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.8891,
      "step": 10195
    },
    {
      "epoch": 0.23990588235294116,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0372,
      "step": 10196
    },
    {
      "epoch": 0.2399294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0343,
      "step": 10197
    },
    {
      "epoch": 0.2399529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1818,
      "step": 10198
    },
    {
      "epoch": 0.2399764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1971,
      "step": 10199
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9902,
      "step": 10200
    },
    {
      "epoch": 0.24002352941176472,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0894,
      "step": 10201
    },
    {
      "epoch": 0.24004705882352942,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.957,
      "step": 10202
    },
    {
      "epoch": 0.24007058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0634,
      "step": 10203
    },
    {
      "epoch": 0.24009411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.855,
      "step": 10204
    },
    {
      "epoch": 0.24011764705882352,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9504,
      "step": 10205
    },
    {
      "epoch": 0.24014117647058825,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1976,
      "step": 10206
    },
    {
      "epoch": 0.24016470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1222,
      "step": 10207
    },
    {
      "epoch": 0.24018823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1169,
      "step": 10208
    },
    {
      "epoch": 0.24021176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2257,
      "step": 10209
    },
    {
      "epoch": 0.24023529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0849,
      "step": 10210
    },
    {
      "epoch": 0.24025882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0776,
      "step": 10211
    },
    {
      "epoch": 0.24028235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0536,
      "step": 10212
    },
    {
      "epoch": 0.24030588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 10213
    },
    {
      "epoch": 0.24032941176470587,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1266,
      "step": 10214
    },
    {
      "epoch": 0.24035294117647057,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.952,
      "step": 10215
    },
    {
      "epoch": 0.2403764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0214,
      "step": 10216
    },
    {
      "epoch": 0.2404,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3648,
      "step": 10217
    },
    {
      "epoch": 0.2404235294117647,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7125,
      "step": 10218
    },
    {
      "epoch": 0.2404470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0872,
      "step": 10219
    },
    {
      "epoch": 0.24047058823529413,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0876,
      "step": 10220
    },
    {
      "epoch": 0.24049411764705883,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1042,
      "step": 10221
    },
    {
      "epoch": 0.24051764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.131,
      "step": 10222
    },
    {
      "epoch": 0.24054117647058823,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1064,
      "step": 10223
    },
    {
      "epoch": 0.24056470588235293,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0098,
      "step": 10224
    },
    {
      "epoch": 0.24058823529411766,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.681,
      "step": 10225
    },
    {
      "epoch": 0.24061176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7995,
      "step": 10226
    },
    {
      "epoch": 0.24063529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1803,
      "step": 10227
    },
    {
      "epoch": 0.24065882352941176,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2007,
      "step": 10228
    },
    {
      "epoch": 0.24068235294117646,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.051,
      "step": 10229
    },
    {
      "epoch": 0.2407058823529412,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1456,
      "step": 10230
    },
    {
      "epoch": 0.2407294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.966,
      "step": 10231
    },
    {
      "epoch": 0.2407529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9693,
      "step": 10232
    },
    {
      "epoch": 0.2407764705882353,
      "grad_norm": 0.2734375,
      "learning_rate": 0.02,
      "loss": 0.732,
      "step": 10233
    },
    {
      "epoch": 0.2408,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 10234
    },
    {
      "epoch": 0.24082352941176471,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.279,
      "step": 10235
    },
    {
      "epoch": 0.24084705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1312,
      "step": 10236
    },
    {
      "epoch": 0.24087058823529411,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1619,
      "step": 10237
    },
    {
      "epoch": 0.24089411764705881,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.257,
      "step": 10238
    },
    {
      "epoch": 0.24091764705882354,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3822,
      "step": 10239
    },
    {
      "epoch": 0.24094117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0847,
      "step": 10240
    },
    {
      "epoch": 0.24096470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9847,
      "step": 10241
    },
    {
      "epoch": 0.24098823529411764,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8214,
      "step": 10242
    },
    {
      "epoch": 0.24101176470588234,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2248,
      "step": 10243
    },
    {
      "epoch": 0.24103529411764707,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.1914,
      "step": 10244
    },
    {
      "epoch": 0.24105882352941177,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9818,
      "step": 10245
    },
    {
      "epoch": 0.24108235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.9177,
      "step": 10246
    },
    {
      "epoch": 0.24110588235294117,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3808,
      "step": 10247
    },
    {
      "epoch": 0.24112941176470587,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0026,
      "step": 10248
    },
    {
      "epoch": 0.2411529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0572,
      "step": 10249
    },
    {
      "epoch": 0.2411764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7708,
      "step": 10250
    },
    {
      "epoch": 0.2412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2706,
      "step": 10251
    },
    {
      "epoch": 0.2412235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.985,
      "step": 10252
    },
    {
      "epoch": 0.24124705882352943,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.067,
      "step": 10253
    },
    {
      "epoch": 0.24127058823529413,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1119,
      "step": 10254
    },
    {
      "epoch": 0.24129411764705883,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2153,
      "step": 10255
    },
    {
      "epoch": 0.24131764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8084,
      "step": 10256
    },
    {
      "epoch": 0.24134117647058823,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0933,
      "step": 10257
    },
    {
      "epoch": 0.24136470588235295,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 1.0401,
      "step": 10258
    },
    {
      "epoch": 0.24138823529411765,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8056,
      "step": 10259
    },
    {
      "epoch": 0.24141176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0446,
      "step": 10260
    },
    {
      "epoch": 0.24143529411764705,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.7925,
      "step": 10261
    },
    {
      "epoch": 0.24145882352941175,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3668,
      "step": 10262
    },
    {
      "epoch": 0.24148235294117648,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1206,
      "step": 10263
    },
    {
      "epoch": 0.24150588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1851,
      "step": 10264
    },
    {
      "epoch": 0.24152941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0727,
      "step": 10265
    },
    {
      "epoch": 0.24155294117647058,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2565,
      "step": 10266
    },
    {
      "epoch": 0.24157647058823528,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 1.0677,
      "step": 10267
    },
    {
      "epoch": 0.2416,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0278,
      "step": 10268
    },
    {
      "epoch": 0.2416235294117647,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.7137,
      "step": 10269
    },
    {
      "epoch": 0.2416470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9215,
      "step": 10270
    },
    {
      "epoch": 0.2416705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0754,
      "step": 10271
    },
    {
      "epoch": 0.24169411764705884,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9659,
      "step": 10272
    },
    {
      "epoch": 0.24171764705882354,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.757,
      "step": 10273
    },
    {
      "epoch": 0.24174117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 10274
    },
    {
      "epoch": 0.24176470588235294,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.8049,
      "step": 10275
    },
    {
      "epoch": 0.24178823529411764,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.4005,
      "step": 10276
    },
    {
      "epoch": 0.24181176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0495,
      "step": 10277
    },
    {
      "epoch": 0.24183529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1636,
      "step": 10278
    },
    {
      "epoch": 0.24185882352941176,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 10279
    },
    {
      "epoch": 0.24188235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1289,
      "step": 10280
    },
    {
      "epoch": 0.24190588235294117,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8134,
      "step": 10281
    },
    {
      "epoch": 0.2419294117647059,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.928,
      "step": 10282
    },
    {
      "epoch": 0.2419529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1291,
      "step": 10283
    },
    {
      "epoch": 0.2419764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9256,
      "step": 10284
    },
    {
      "epoch": 0.242,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.1124,
      "step": 10285
    },
    {
      "epoch": 0.2420235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.4903,
      "step": 10286
    },
    {
      "epoch": 0.24204705882352942,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9871,
      "step": 10287
    },
    {
      "epoch": 0.24207058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4001,
      "step": 10288
    },
    {
      "epoch": 0.24209411764705882,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.5076,
      "step": 10289
    },
    {
      "epoch": 0.24211764705882352,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.119,
      "step": 10290
    },
    {
      "epoch": 0.24214117647058825,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 10291
    },
    {
      "epoch": 0.24216470588235295,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8591,
      "step": 10292
    },
    {
      "epoch": 0.24218823529411765,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9143,
      "step": 10293
    },
    {
      "epoch": 0.24221176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2946,
      "step": 10294
    },
    {
      "epoch": 0.24223529411764705,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.1156,
      "step": 10295
    },
    {
      "epoch": 0.24225882352941178,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.7892,
      "step": 10296
    },
    {
      "epoch": 0.24228235294117648,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9591,
      "step": 10297
    },
    {
      "epoch": 0.24230588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0675,
      "step": 10298
    },
    {
      "epoch": 0.24232941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8665,
      "step": 10299
    },
    {
      "epoch": 0.24235294117647058,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0876,
      "step": 10300
    },
    {
      "epoch": 0.2423764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8393,
      "step": 10301
    },
    {
      "epoch": 0.2424,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1047,
      "step": 10302
    },
    {
      "epoch": 0.2424235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8806,
      "step": 10303
    },
    {
      "epoch": 0.2424470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.378,
      "step": 10304
    },
    {
      "epoch": 0.2424705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0437,
      "step": 10305
    },
    {
      "epoch": 0.24249411764705883,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8022,
      "step": 10306
    },
    {
      "epoch": 0.24251764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9865,
      "step": 10307
    },
    {
      "epoch": 0.24254117647058823,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0582,
      "step": 10308
    },
    {
      "epoch": 0.24256470588235293,
      "grad_norm": 0.25390625,
      "learning_rate": 0.02,
      "loss": 0.8693,
      "step": 10309
    },
    {
      "epoch": 0.24258823529411766,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2981,
      "step": 10310
    },
    {
      "epoch": 0.24261176470588236,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9647,
      "step": 10311
    },
    {
      "epoch": 0.24263529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1906,
      "step": 10312
    },
    {
      "epoch": 0.24265882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.226,
      "step": 10313
    },
    {
      "epoch": 0.24268235294117646,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 10314
    },
    {
      "epoch": 0.2427058823529412,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.829,
      "step": 10315
    },
    {
      "epoch": 0.2427294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9931,
      "step": 10316
    },
    {
      "epoch": 0.2427529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0577,
      "step": 10317
    },
    {
      "epoch": 0.2427764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4823,
      "step": 10318
    },
    {
      "epoch": 0.2428,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1502,
      "step": 10319
    },
    {
      "epoch": 0.24282352941176472,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2421,
      "step": 10320
    },
    {
      "epoch": 0.24284705882352942,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9538,
      "step": 10321
    },
    {
      "epoch": 0.24287058823529412,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0162,
      "step": 10322
    },
    {
      "epoch": 0.24289411764705882,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7795,
      "step": 10323
    },
    {
      "epoch": 0.24291764705882352,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1545,
      "step": 10324
    },
    {
      "epoch": 0.24294117647058824,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2521,
      "step": 10325
    },
    {
      "epoch": 0.24296470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8557,
      "step": 10326
    },
    {
      "epoch": 0.24298823529411764,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.8402,
      "step": 10327
    },
    {
      "epoch": 0.24301176470588234,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.966,
      "step": 10328
    },
    {
      "epoch": 0.24303529411764707,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 10329
    },
    {
      "epoch": 0.24305882352941177,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1424,
      "step": 10330
    },
    {
      "epoch": 0.24308235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2292,
      "step": 10331
    },
    {
      "epoch": 0.24310588235294117,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8159,
      "step": 10332
    },
    {
      "epoch": 0.24312941176470587,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.3828,
      "step": 10333
    },
    {
      "epoch": 0.2431529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9883,
      "step": 10334
    },
    {
      "epoch": 0.2431764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1348,
      "step": 10335
    },
    {
      "epoch": 0.2432,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 1.0611,
      "step": 10336
    },
    {
      "epoch": 0.2432235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.934,
      "step": 10337
    },
    {
      "epoch": 0.2432470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0727,
      "step": 10338
    },
    {
      "epoch": 0.24327058823529413,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3728,
      "step": 10339
    },
    {
      "epoch": 0.24329411764705883,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0902,
      "step": 10340
    },
    {
      "epoch": 0.24331764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9735,
      "step": 10341
    },
    {
      "epoch": 0.24334117647058823,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3739,
      "step": 10342
    },
    {
      "epoch": 0.24336470588235293,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.789,
      "step": 10343
    },
    {
      "epoch": 0.24338823529411766,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8828,
      "step": 10344
    },
    {
      "epoch": 0.24341176470588236,
      "grad_norm": 1.390625,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 10345
    },
    {
      "epoch": 0.24343529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.169,
      "step": 10346
    },
    {
      "epoch": 0.24345882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.043,
      "step": 10347
    },
    {
      "epoch": 0.24348235294117648,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.5737,
      "step": 10348
    },
    {
      "epoch": 0.24350588235294118,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9372,
      "step": 10349
    },
    {
      "epoch": 0.24352941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8261,
      "step": 10350
    },
    {
      "epoch": 0.24355294117647058,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 10351
    },
    {
      "epoch": 0.24357647058823528,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0772,
      "step": 10352
    },
    {
      "epoch": 0.2436,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0193,
      "step": 10353
    },
    {
      "epoch": 0.2436235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2876,
      "step": 10354
    },
    {
      "epoch": 0.2436470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9893,
      "step": 10355
    },
    {
      "epoch": 0.2436705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.895,
      "step": 10356
    },
    {
      "epoch": 0.2436941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9596,
      "step": 10357
    },
    {
      "epoch": 0.24371764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.7406,
      "step": 10358
    },
    {
      "epoch": 0.24374117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0596,
      "step": 10359
    },
    {
      "epoch": 0.24376470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1505,
      "step": 10360
    },
    {
      "epoch": 0.24378823529411764,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0043,
      "step": 10361
    },
    {
      "epoch": 0.24381176470588237,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1614,
      "step": 10362
    },
    {
      "epoch": 0.24383529411764707,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.283,
      "step": 10363
    },
    {
      "epoch": 0.24385882352941177,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.1628,
      "step": 10364
    },
    {
      "epoch": 0.24388235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.4055,
      "step": 10365
    },
    {
      "epoch": 0.24390588235294117,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1613,
      "step": 10366
    },
    {
      "epoch": 0.2439294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1474,
      "step": 10367
    },
    {
      "epoch": 0.2439529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.7212,
      "step": 10368
    },
    {
      "epoch": 0.2439764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3401,
      "step": 10369
    },
    {
      "epoch": 0.244,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2592,
      "step": 10370
    },
    {
      "epoch": 0.2440235294117647,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.7469,
      "step": 10371
    },
    {
      "epoch": 0.24404705882352942,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0088,
      "step": 10372
    },
    {
      "epoch": 0.24407058823529412,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7891,
      "step": 10373
    },
    {
      "epoch": 0.24409411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2273,
      "step": 10374
    },
    {
      "epoch": 0.24411764705882352,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 1.0185,
      "step": 10375
    },
    {
      "epoch": 0.24414117647058822,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9441,
      "step": 10376
    },
    {
      "epoch": 0.24416470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.4959,
      "step": 10377
    },
    {
      "epoch": 0.24418823529411765,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.1398,
      "step": 10378
    },
    {
      "epoch": 0.24421176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0265,
      "step": 10379
    },
    {
      "epoch": 0.24423529411764705,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9963,
      "step": 10380
    },
    {
      "epoch": 0.24425882352941178,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8433,
      "step": 10381
    },
    {
      "epoch": 0.24428235294117648,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2978,
      "step": 10382
    },
    {
      "epoch": 0.24430588235294118,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.9047,
      "step": 10383
    },
    {
      "epoch": 0.24432941176470588,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0805,
      "step": 10384
    },
    {
      "epoch": 0.24435294117647058,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1688,
      "step": 10385
    },
    {
      "epoch": 0.2443764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1365,
      "step": 10386
    },
    {
      "epoch": 0.2444,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7918,
      "step": 10387
    },
    {
      "epoch": 0.2444235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.345,
      "step": 10388
    },
    {
      "epoch": 0.2444470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1773,
      "step": 10389
    },
    {
      "epoch": 0.2444705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0961,
      "step": 10390
    },
    {
      "epoch": 0.24449411764705883,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2521,
      "step": 10391
    },
    {
      "epoch": 0.24451764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8902,
      "step": 10392
    },
    {
      "epoch": 0.24454117647058823,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1358,
      "step": 10393
    },
    {
      "epoch": 0.24456470588235293,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0905,
      "step": 10394
    },
    {
      "epoch": 0.24458823529411763,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3109,
      "step": 10395
    },
    {
      "epoch": 0.24461176470588236,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 10396
    },
    {
      "epoch": 0.24463529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8765,
      "step": 10397
    },
    {
      "epoch": 0.24465882352941176,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0215,
      "step": 10398
    },
    {
      "epoch": 0.24468235294117646,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.459,
      "step": 10399
    },
    {
      "epoch": 0.2447058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.228,
      "step": 10400
    },
    {
      "epoch": 0.2447294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2112,
      "step": 10401
    },
    {
      "epoch": 0.2447529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 10402
    },
    {
      "epoch": 0.2447764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0304,
      "step": 10403
    },
    {
      "epoch": 0.2448,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2905,
      "step": 10404
    },
    {
      "epoch": 0.24482352941176472,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.893,
      "step": 10405
    },
    {
      "epoch": 0.24484705882352942,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3331,
      "step": 10406
    },
    {
      "epoch": 0.24487058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1619,
      "step": 10407
    },
    {
      "epoch": 0.24489411764705882,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3827,
      "step": 10408
    },
    {
      "epoch": 0.24491764705882352,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8646,
      "step": 10409
    },
    {
      "epoch": 0.24494117647058825,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1105,
      "step": 10410
    },
    {
      "epoch": 0.24496470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 10411
    },
    {
      "epoch": 0.24498823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2071,
      "step": 10412
    },
    {
      "epoch": 0.24501176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0514,
      "step": 10413
    },
    {
      "epoch": 0.24503529411764705,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9394,
      "step": 10414
    },
    {
      "epoch": 0.24505882352941177,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8493,
      "step": 10415
    },
    {
      "epoch": 0.24508235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.1626,
      "step": 10416
    },
    {
      "epoch": 0.24510588235294117,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.052,
      "step": 10417
    },
    {
      "epoch": 0.24512941176470587,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7791,
      "step": 10418
    },
    {
      "epoch": 0.2451529411764706,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3723,
      "step": 10419
    },
    {
      "epoch": 0.2451764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1052,
      "step": 10420
    },
    {
      "epoch": 0.2452,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8298,
      "step": 10421
    },
    {
      "epoch": 0.2452235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1508,
      "step": 10422
    },
    {
      "epoch": 0.2452470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1613,
      "step": 10423
    },
    {
      "epoch": 0.24527058823529413,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 10424
    },
    {
      "epoch": 0.24529411764705883,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8093,
      "step": 10425
    },
    {
      "epoch": 0.24531764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0295,
      "step": 10426
    },
    {
      "epoch": 0.24534117647058823,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 10427
    },
    {
      "epoch": 0.24536470588235293,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2409,
      "step": 10428
    },
    {
      "epoch": 0.24538823529411766,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.935,
      "step": 10429
    },
    {
      "epoch": 0.24541176470588236,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2969,
      "step": 10430
    },
    {
      "epoch": 0.24543529411764706,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1421,
      "step": 10431
    },
    {
      "epoch": 0.24545882352941176,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0634,
      "step": 10432
    },
    {
      "epoch": 0.24548235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1427,
      "step": 10433
    },
    {
      "epoch": 0.24550588235294118,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2265,
      "step": 10434
    },
    {
      "epoch": 0.24552941176470588,
      "grad_norm": 0.271484375,
      "learning_rate": 0.02,
      "loss": 0.7041,
      "step": 10435
    },
    {
      "epoch": 0.24555294117647058,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0161,
      "step": 10436
    },
    {
      "epoch": 0.24557647058823528,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1297,
      "step": 10437
    },
    {
      "epoch": 0.2456,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3037,
      "step": 10438
    },
    {
      "epoch": 0.2456235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0548,
      "step": 10439
    },
    {
      "epoch": 0.2456470588235294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.6518,
      "step": 10440
    },
    {
      "epoch": 0.2456705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0355,
      "step": 10441
    },
    {
      "epoch": 0.2456941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9076,
      "step": 10442
    },
    {
      "epoch": 0.24571764705882354,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2699,
      "step": 10443
    },
    {
      "epoch": 0.24574117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1726,
      "step": 10444
    },
    {
      "epoch": 0.24576470588235294,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 1.2877,
      "step": 10445
    },
    {
      "epoch": 0.24578823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0045,
      "step": 10446
    },
    {
      "epoch": 0.24581176470588234,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2001,
      "step": 10447
    },
    {
      "epoch": 0.24583529411764707,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9224,
      "step": 10448
    },
    {
      "epoch": 0.24585882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0601,
      "step": 10449
    },
    {
      "epoch": 0.24588235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0846,
      "step": 10450
    },
    {
      "epoch": 0.24590588235294117,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2192,
      "step": 10451
    },
    {
      "epoch": 0.24592941176470587,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1537,
      "step": 10452
    },
    {
      "epoch": 0.2459529411764706,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 10453
    },
    {
      "epoch": 0.2459764705882353,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8905,
      "step": 10454
    },
    {
      "epoch": 0.246,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0928,
      "step": 10455
    },
    {
      "epoch": 0.2460235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0043,
      "step": 10456
    },
    {
      "epoch": 0.24604705882352942,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2126,
      "step": 10457
    },
    {
      "epoch": 0.24607058823529412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.5398,
      "step": 10458
    },
    {
      "epoch": 0.24609411764705882,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7249,
      "step": 10459
    },
    {
      "epoch": 0.24611764705882352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 10460
    },
    {
      "epoch": 0.24614117647058822,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.6625,
      "step": 10461
    },
    {
      "epoch": 0.24616470588235295,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1082,
      "step": 10462
    },
    {
      "epoch": 0.24618823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 10463
    },
    {
      "epoch": 0.24621176470588235,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3133,
      "step": 10464
    },
    {
      "epoch": 0.24623529411764705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2104,
      "step": 10465
    },
    {
      "epoch": 0.24625882352941175,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1882,
      "step": 10466
    },
    {
      "epoch": 0.24628235294117648,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1003,
      "step": 10467
    },
    {
      "epoch": 0.24630588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2404,
      "step": 10468
    },
    {
      "epoch": 0.24632941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.018,
      "step": 10469
    },
    {
      "epoch": 0.24635294117647058,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.7723,
      "step": 10470
    },
    {
      "epoch": 0.2463764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0854,
      "step": 10471
    },
    {
      "epoch": 0.2464,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0111,
      "step": 10472
    },
    {
      "epoch": 0.2464235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1268,
      "step": 10473
    },
    {
      "epoch": 0.2464470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0791,
      "step": 10474
    },
    {
      "epoch": 0.2464705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1914,
      "step": 10475
    },
    {
      "epoch": 0.24649411764705884,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0754,
      "step": 10476
    },
    {
      "epoch": 0.24651764705882354,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1745,
      "step": 10477
    },
    {
      "epoch": 0.24654117647058824,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9446,
      "step": 10478
    },
    {
      "epoch": 0.24656470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0155,
      "step": 10479
    },
    {
      "epoch": 0.24658823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2461,
      "step": 10480
    },
    {
      "epoch": 0.24661176470588236,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0657,
      "step": 10481
    },
    {
      "epoch": 0.24663529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0036,
      "step": 10482
    },
    {
      "epoch": 0.24665882352941176,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9547,
      "step": 10483
    },
    {
      "epoch": 0.24668235294117646,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1117,
      "step": 10484
    },
    {
      "epoch": 0.24670588235294116,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9082,
      "step": 10485
    },
    {
      "epoch": 0.2467294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2461,
      "step": 10486
    },
    {
      "epoch": 0.2467529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.4223,
      "step": 10487
    },
    {
      "epoch": 0.2467764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0612,
      "step": 10488
    },
    {
      "epoch": 0.2468,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.121,
      "step": 10489
    },
    {
      "epoch": 0.24682352941176472,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2525,
      "step": 10490
    },
    {
      "epoch": 0.24684705882352942,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9422,
      "step": 10491
    },
    {
      "epoch": 0.24687058823529412,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0968,
      "step": 10492
    },
    {
      "epoch": 0.24689411764705882,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 10493
    },
    {
      "epoch": 0.24691764705882352,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3286,
      "step": 10494
    },
    {
      "epoch": 0.24694117647058825,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8676,
      "step": 10495
    },
    {
      "epoch": 0.24696470588235295,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.6758,
      "step": 10496
    },
    {
      "epoch": 0.24698823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.2523,
      "step": 10497
    },
    {
      "epoch": 0.24701176470588235,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.7694,
      "step": 10498
    },
    {
      "epoch": 0.24703529411764705,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2462,
      "step": 10499
    },
    {
      "epoch": 0.24705882352941178,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0833,
      "step": 10500
    },
    {
      "epoch": 0.24708235294117648,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.793,
      "step": 10501
    },
    {
      "epoch": 0.24710588235294118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1605,
      "step": 10502
    },
    {
      "epoch": 0.24712941176470588,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9382,
      "step": 10503
    },
    {
      "epoch": 0.24715294117647058,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1953,
      "step": 10504
    },
    {
      "epoch": 0.2471764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9405,
      "step": 10505
    },
    {
      "epoch": 0.2472,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0312,
      "step": 10506
    },
    {
      "epoch": 0.2472235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2782,
      "step": 10507
    },
    {
      "epoch": 0.2472470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3997,
      "step": 10508
    },
    {
      "epoch": 0.24727058823529413,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0798,
      "step": 10509
    },
    {
      "epoch": 0.24729411764705883,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1761,
      "step": 10510
    },
    {
      "epoch": 0.24731764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.159,
      "step": 10511
    },
    {
      "epoch": 0.24734117647058823,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3197,
      "step": 10512
    },
    {
      "epoch": 0.24736470588235293,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2455,
      "step": 10513
    },
    {
      "epoch": 0.24738823529411766,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0077,
      "step": 10514
    },
    {
      "epoch": 0.24741176470588236,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.763,
      "step": 10515
    },
    {
      "epoch": 0.24743529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7818,
      "step": 10516
    },
    {
      "epoch": 0.24745882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 10517
    },
    {
      "epoch": 0.24748235294117646,
      "grad_norm": 0.263671875,
      "learning_rate": 0.02,
      "loss": 0.7656,
      "step": 10518
    },
    {
      "epoch": 0.2475058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0424,
      "step": 10519
    },
    {
      "epoch": 0.2475294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 10520
    },
    {
      "epoch": 0.2475529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 10521
    },
    {
      "epoch": 0.2475764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1221,
      "step": 10522
    },
    {
      "epoch": 0.2476,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1871,
      "step": 10523
    },
    {
      "epoch": 0.24762352941176471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.733,
      "step": 10524
    },
    {
      "epoch": 0.24764705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1173,
      "step": 10525
    },
    {
      "epoch": 0.24767058823529411,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3685,
      "step": 10526
    },
    {
      "epoch": 0.24769411764705881,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0038,
      "step": 10527
    },
    {
      "epoch": 0.24771764705882354,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8395,
      "step": 10528
    },
    {
      "epoch": 0.24774117647058824,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8252,
      "step": 10529
    },
    {
      "epoch": 0.24776470588235294,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 1.0428,
      "step": 10530
    },
    {
      "epoch": 0.24778823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9601,
      "step": 10531
    },
    {
      "epoch": 0.24781176470588234,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9715,
      "step": 10532
    },
    {
      "epoch": 0.24783529411764707,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1471,
      "step": 10533
    },
    {
      "epoch": 0.24785882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2908,
      "step": 10534
    },
    {
      "epoch": 0.24788235294117647,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.903,
      "step": 10535
    },
    {
      "epoch": 0.24790588235294117,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1113,
      "step": 10536
    },
    {
      "epoch": 0.24792941176470587,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.6634,
      "step": 10537
    },
    {
      "epoch": 0.2479529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2362,
      "step": 10538
    },
    {
      "epoch": 0.2479764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 10539
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0228,
      "step": 10540
    },
    {
      "epoch": 0.2480235294117647,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.6826,
      "step": 10541
    },
    {
      "epoch": 0.2480470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0425,
      "step": 10542
    },
    {
      "epoch": 0.24807058823529413,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.2269,
      "step": 10543
    },
    {
      "epoch": 0.24809411764705883,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.9441,
      "step": 10544
    },
    {
      "epoch": 0.24811764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2294,
      "step": 10545
    },
    {
      "epoch": 0.24814117647058823,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.39,
      "step": 10546
    },
    {
      "epoch": 0.24816470588235295,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9487,
      "step": 10547
    },
    {
      "epoch": 0.24818823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8742,
      "step": 10548
    },
    {
      "epoch": 0.24821176470588235,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8963,
      "step": 10549
    },
    {
      "epoch": 0.24823529411764705,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.6955,
      "step": 10550
    },
    {
      "epoch": 0.24825882352941175,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3553,
      "step": 10551
    },
    {
      "epoch": 0.24828235294117648,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.0166,
      "step": 10552
    },
    {
      "epoch": 0.24830588235294118,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2463,
      "step": 10553
    },
    {
      "epoch": 0.24832941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.3122,
      "step": 10554
    },
    {
      "epoch": 0.24835294117647058,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8628,
      "step": 10555
    },
    {
      "epoch": 0.24837647058823528,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0414,
      "step": 10556
    },
    {
      "epoch": 0.2484,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0841,
      "step": 10557
    },
    {
      "epoch": 0.2484235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.07,
      "step": 10558
    },
    {
      "epoch": 0.2484470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0176,
      "step": 10559
    },
    {
      "epoch": 0.2484705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9014,
      "step": 10560
    },
    {
      "epoch": 0.2484941176470588,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9218,
      "step": 10561
    },
    {
      "epoch": 0.24851764705882354,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2679,
      "step": 10562
    },
    {
      "epoch": 0.24854117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.3171,
      "step": 10563
    },
    {
      "epoch": 0.24856470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2413,
      "step": 10564
    },
    {
      "epoch": 0.24858823529411764,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 10565
    },
    {
      "epoch": 0.24861176470588237,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3043,
      "step": 10566
    },
    {
      "epoch": 0.24863529411764707,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1796,
      "step": 10567
    },
    {
      "epoch": 0.24865882352941177,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1218,
      "step": 10568
    },
    {
      "epoch": 0.24868235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2605,
      "step": 10569
    },
    {
      "epoch": 0.24870588235294117,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9563,
      "step": 10570
    },
    {
      "epoch": 0.2487294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1383,
      "step": 10571
    },
    {
      "epoch": 0.2487529411764706,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7943,
      "step": 10572
    },
    {
      "epoch": 0.2487764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2418,
      "step": 10573
    },
    {
      "epoch": 0.2488,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1479,
      "step": 10574
    },
    {
      "epoch": 0.2488235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0275,
      "step": 10575
    },
    {
      "epoch": 0.24884705882352942,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8622,
      "step": 10576
    },
    {
      "epoch": 0.24887058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.093,
      "step": 10577
    },
    {
      "epoch": 0.24889411764705882,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.7628,
      "step": 10578
    },
    {
      "epoch": 0.24891764705882352,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 10579
    },
    {
      "epoch": 0.24894117647058825,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.185,
      "step": 10580
    },
    {
      "epoch": 0.24896470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0973,
      "step": 10581
    },
    {
      "epoch": 0.24898823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 10582
    },
    {
      "epoch": 0.24901176470588235,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8322,
      "step": 10583
    },
    {
      "epoch": 0.24903529411764705,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0526,
      "step": 10584
    },
    {
      "epoch": 0.24905882352941178,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9569,
      "step": 10585
    },
    {
      "epoch": 0.24908235294117648,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3676,
      "step": 10586
    },
    {
      "epoch": 0.24910588235294118,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9202,
      "step": 10587
    },
    {
      "epoch": 0.24912941176470588,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9029,
      "step": 10588
    },
    {
      "epoch": 0.24915294117647058,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3134,
      "step": 10589
    },
    {
      "epoch": 0.2491764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.113,
      "step": 10590
    },
    {
      "epoch": 0.2492,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.3193,
      "step": 10591
    },
    {
      "epoch": 0.2492235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0197,
      "step": 10592
    },
    {
      "epoch": 0.2492470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 10593
    },
    {
      "epoch": 0.2492705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1152,
      "step": 10594
    },
    {
      "epoch": 0.24929411764705883,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0468,
      "step": 10595
    },
    {
      "epoch": 0.24931764705882353,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7924,
      "step": 10596
    },
    {
      "epoch": 0.24934117647058823,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9098,
      "step": 10597
    },
    {
      "epoch": 0.24936470588235293,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.9518,
      "step": 10598
    },
    {
      "epoch": 0.24938823529411766,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0926,
      "step": 10599
    },
    {
      "epoch": 0.24941176470588236,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.157,
      "step": 10600
    },
    {
      "epoch": 0.24943529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1661,
      "step": 10601
    },
    {
      "epoch": 0.24945882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1549,
      "step": 10602
    },
    {
      "epoch": 0.24948235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1536,
      "step": 10603
    },
    {
      "epoch": 0.2495058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9593,
      "step": 10604
    },
    {
      "epoch": 0.2495294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1946,
      "step": 10605
    },
    {
      "epoch": 0.2495529411764706,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9395,
      "step": 10606
    },
    {
      "epoch": 0.2495764705882353,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8818,
      "step": 10607
    },
    {
      "epoch": 0.2496,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9652,
      "step": 10608
    },
    {
      "epoch": 0.24962352941176472,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2233,
      "step": 10609
    },
    {
      "epoch": 0.24964705882352942,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9423,
      "step": 10610
    },
    {
      "epoch": 0.24967058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 10611
    },
    {
      "epoch": 0.24969411764705882,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0297,
      "step": 10612
    },
    {
      "epoch": 0.24971764705882352,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.9839,
      "step": 10613
    },
    {
      "epoch": 0.24974117647058824,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0643,
      "step": 10614
    },
    {
      "epoch": 0.24976470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9746,
      "step": 10615
    },
    {
      "epoch": 0.24978823529411764,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7433,
      "step": 10616
    },
    {
      "epoch": 0.24981176470588234,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0186,
      "step": 10617
    },
    {
      "epoch": 0.24983529411764707,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1672,
      "step": 10618
    },
    {
      "epoch": 0.24985882352941177,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7784,
      "step": 10619
    },
    {
      "epoch": 0.24988235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1676,
      "step": 10620
    },
    {
      "epoch": 0.24990588235294117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0617,
      "step": 10621
    },
    {
      "epoch": 0.24992941176470587,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2353,
      "step": 10622
    },
    {
      "epoch": 0.2499529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 10623
    },
    {
      "epoch": 0.2499764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.7374,
      "step": 10624
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 10625
    },
    {
      "epoch": 0.2500235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.3186,
      "step": 10626
    },
    {
      "epoch": 0.2500470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9668,
      "step": 10627
    },
    {
      "epoch": 0.2500705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 10628
    },
    {
      "epoch": 0.2500941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0794,
      "step": 10629
    },
    {
      "epoch": 0.25011764705882356,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8439,
      "step": 10630
    },
    {
      "epoch": 0.25014117647058826,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9831,
      "step": 10631
    },
    {
      "epoch": 0.25016470588235296,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1543,
      "step": 10632
    },
    {
      "epoch": 0.25018823529411766,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8325,
      "step": 10633
    },
    {
      "epoch": 0.25021176470588236,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8755,
      "step": 10634
    },
    {
      "epoch": 0.25023529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.951,
      "step": 10635
    },
    {
      "epoch": 0.25025882352941176,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1409,
      "step": 10636
    },
    {
      "epoch": 0.25028235294117646,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9692,
      "step": 10637
    },
    {
      "epoch": 0.25030588235294116,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0981,
      "step": 10638
    },
    {
      "epoch": 0.25032941176470586,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0117,
      "step": 10639
    },
    {
      "epoch": 0.2503529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 1.0096,
      "step": 10640
    },
    {
      "epoch": 0.2503764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1625,
      "step": 10641
    },
    {
      "epoch": 0.2504,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.202,
      "step": 10642
    },
    {
      "epoch": 0.2504235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9742,
      "step": 10643
    },
    {
      "epoch": 0.2504470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0017,
      "step": 10644
    },
    {
      "epoch": 0.2504705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2307,
      "step": 10645
    },
    {
      "epoch": 0.2504941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3988,
      "step": 10646
    },
    {
      "epoch": 0.2505176470588235,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.183,
      "step": 10647
    },
    {
      "epoch": 0.2505411764705882,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1727,
      "step": 10648
    },
    {
      "epoch": 0.25056470588235297,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2374,
      "step": 10649
    },
    {
      "epoch": 0.25058823529411767,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 10650
    },
    {
      "epoch": 0.25061176470588237,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0419,
      "step": 10651
    },
    {
      "epoch": 0.25063529411764707,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0845,
      "step": 10652
    },
    {
      "epoch": 0.25065882352941177,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0079,
      "step": 10653
    },
    {
      "epoch": 0.25068235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 10654
    },
    {
      "epoch": 0.25070588235294117,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8438,
      "step": 10655
    },
    {
      "epoch": 0.25072941176470587,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0775,
      "step": 10656
    },
    {
      "epoch": 0.25075294117647057,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9846,
      "step": 10657
    },
    {
      "epoch": 0.25077647058823527,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9754,
      "step": 10658
    },
    {
      "epoch": 0.2508,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1652,
      "step": 10659
    },
    {
      "epoch": 0.2508235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9916,
      "step": 10660
    },
    {
      "epoch": 0.2508470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9918,
      "step": 10661
    },
    {
      "epoch": 0.2508705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9745,
      "step": 10662
    },
    {
      "epoch": 0.2508941176470588,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8228,
      "step": 10663
    },
    {
      "epoch": 0.2509176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2722,
      "step": 10664
    },
    {
      "epoch": 0.2509411764705882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.7521,
      "step": 10665
    },
    {
      "epoch": 0.2509647058823529,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1293,
      "step": 10666
    },
    {
      "epoch": 0.2509882352941176,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0423,
      "step": 10667
    },
    {
      "epoch": 0.2510117647058824,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2179,
      "step": 10668
    },
    {
      "epoch": 0.2510352941176471,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9011,
      "step": 10669
    },
    {
      "epoch": 0.2510588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9234,
      "step": 10670
    },
    {
      "epoch": 0.2510823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0011,
      "step": 10671
    },
    {
      "epoch": 0.2511058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3013,
      "step": 10672
    },
    {
      "epoch": 0.2511294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2432,
      "step": 10673
    },
    {
      "epoch": 0.2511529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0818,
      "step": 10674
    },
    {
      "epoch": 0.2511764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9342,
      "step": 10675
    },
    {
      "epoch": 0.2512,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9534,
      "step": 10676
    },
    {
      "epoch": 0.2512235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2306,
      "step": 10677
    },
    {
      "epoch": 0.25124705882352943,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8177,
      "step": 10678
    },
    {
      "epoch": 0.25127058823529413,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1092,
      "step": 10679
    },
    {
      "epoch": 0.25129411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0421,
      "step": 10680
    },
    {
      "epoch": 0.25131764705882353,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7222,
      "step": 10681
    },
    {
      "epoch": 0.25134117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9779,
      "step": 10682
    },
    {
      "epoch": 0.25136470588235293,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9756,
      "step": 10683
    },
    {
      "epoch": 0.25138823529411763,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0835,
      "step": 10684
    },
    {
      "epoch": 0.25141176470588233,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7254,
      "step": 10685
    },
    {
      "epoch": 0.25143529411764703,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0189,
      "step": 10686
    },
    {
      "epoch": 0.2514588235294118,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7556,
      "step": 10687
    },
    {
      "epoch": 0.2514823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1306,
      "step": 10688
    },
    {
      "epoch": 0.2515058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2478,
      "step": 10689
    },
    {
      "epoch": 0.2515294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0363,
      "step": 10690
    },
    {
      "epoch": 0.2515529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9744,
      "step": 10691
    },
    {
      "epoch": 0.2515764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1108,
      "step": 10692
    },
    {
      "epoch": 0.2516,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.6121,
      "step": 10693
    },
    {
      "epoch": 0.2516235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0954,
      "step": 10694
    },
    {
      "epoch": 0.2516470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2826,
      "step": 10695
    },
    {
      "epoch": 0.2516705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2497,
      "step": 10696
    },
    {
      "epoch": 0.25169411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0378,
      "step": 10697
    },
    {
      "epoch": 0.25171764705882355,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 10698
    },
    {
      "epoch": 0.25174117647058825,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1441,
      "step": 10699
    },
    {
      "epoch": 0.25176470588235295,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1865,
      "step": 10700
    },
    {
      "epoch": 0.25178823529411765,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1944,
      "step": 10701
    },
    {
      "epoch": 0.25181176470588235,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7733,
      "step": 10702
    },
    {
      "epoch": 0.25183529411764705,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8673,
      "step": 10703
    },
    {
      "epoch": 0.25185882352941175,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8411,
      "step": 10704
    },
    {
      "epoch": 0.25188235294117645,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9393,
      "step": 10705
    },
    {
      "epoch": 0.2519058823529412,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3215,
      "step": 10706
    },
    {
      "epoch": 0.2519294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0087,
      "step": 10707
    },
    {
      "epoch": 0.2519529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8242,
      "step": 10708
    },
    {
      "epoch": 0.2519764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9798,
      "step": 10709
    },
    {
      "epoch": 0.252,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3032,
      "step": 10710
    },
    {
      "epoch": 0.2520235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3997,
      "step": 10711
    },
    {
      "epoch": 0.2520470588235294,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3985,
      "step": 10712
    },
    {
      "epoch": 0.2520705882352941,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 1.0404,
      "step": 10713
    },
    {
      "epoch": 0.2520941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0019,
      "step": 10714
    },
    {
      "epoch": 0.2521176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1826,
      "step": 10715
    },
    {
      "epoch": 0.25214117647058826,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.869,
      "step": 10716
    },
    {
      "epoch": 0.25216470588235296,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.823,
      "step": 10717
    },
    {
      "epoch": 0.25218823529411766,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.7177,
      "step": 10718
    },
    {
      "epoch": 0.25221176470588236,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1415,
      "step": 10719
    },
    {
      "epoch": 0.25223529411764706,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2672,
      "step": 10720
    },
    {
      "epoch": 0.25225882352941176,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8938,
      "step": 10721
    },
    {
      "epoch": 0.25228235294117646,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2456,
      "step": 10722
    },
    {
      "epoch": 0.25230588235294116,
      "grad_norm": 0.24609375,
      "learning_rate": 0.02,
      "loss": 0.728,
      "step": 10723
    },
    {
      "epoch": 0.25232941176470586,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2323,
      "step": 10724
    },
    {
      "epoch": 0.2523529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2671,
      "step": 10725
    },
    {
      "epoch": 0.2523764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 10726
    },
    {
      "epoch": 0.2524,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8447,
      "step": 10727
    },
    {
      "epoch": 0.2524235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9508,
      "step": 10728
    },
    {
      "epoch": 0.2524470588235294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.716,
      "step": 10729
    },
    {
      "epoch": 0.2524705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0217,
      "step": 10730
    },
    {
      "epoch": 0.2524941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.933,
      "step": 10731
    },
    {
      "epoch": 0.2525176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8584,
      "step": 10732
    },
    {
      "epoch": 0.2525411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2331,
      "step": 10733
    },
    {
      "epoch": 0.25256470588235297,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1237,
      "step": 10734
    },
    {
      "epoch": 0.25258823529411767,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0603,
      "step": 10735
    },
    {
      "epoch": 0.25261176470588237,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7937,
      "step": 10736
    },
    {
      "epoch": 0.25263529411764707,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9763,
      "step": 10737
    },
    {
      "epoch": 0.25265882352941177,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1493,
      "step": 10738
    },
    {
      "epoch": 0.25268235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.4689,
      "step": 10739
    },
    {
      "epoch": 0.25270588235294117,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8577,
      "step": 10740
    },
    {
      "epoch": 0.25272941176470587,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2818,
      "step": 10741
    },
    {
      "epoch": 0.25275294117647057,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 10742
    },
    {
      "epoch": 0.25277647058823527,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9753,
      "step": 10743
    },
    {
      "epoch": 0.2528,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.879,
      "step": 10744
    },
    {
      "epoch": 0.2528235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.843,
      "step": 10745
    },
    {
      "epoch": 0.2528470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.6525,
      "step": 10746
    },
    {
      "epoch": 0.2528705882352941,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.9111,
      "step": 10747
    },
    {
      "epoch": 0.2528941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8552,
      "step": 10748
    },
    {
      "epoch": 0.2529176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2322,
      "step": 10749
    },
    {
      "epoch": 0.2529411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9341,
      "step": 10750
    },
    {
      "epoch": 0.2529647058823529,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9254,
      "step": 10751
    },
    {
      "epoch": 0.2529882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0008,
      "step": 10752
    },
    {
      "epoch": 0.2530117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0854,
      "step": 10753
    },
    {
      "epoch": 0.2530352941176471,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1918,
      "step": 10754
    },
    {
      "epoch": 0.2530588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 10755
    },
    {
      "epoch": 0.2530823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0208,
      "step": 10756
    },
    {
      "epoch": 0.2531058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1337,
      "step": 10757
    },
    {
      "epoch": 0.2531294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 10758
    },
    {
      "epoch": 0.2531529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9482,
      "step": 10759
    },
    {
      "epoch": 0.2531764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8861,
      "step": 10760
    },
    {
      "epoch": 0.2532,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2492,
      "step": 10761
    },
    {
      "epoch": 0.2532235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1729,
      "step": 10762
    },
    {
      "epoch": 0.25324705882352944,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2542,
      "step": 10763
    },
    {
      "epoch": 0.25327058823529414,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1383,
      "step": 10764
    },
    {
      "epoch": 0.25329411764705884,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7289,
      "step": 10765
    },
    {
      "epoch": 0.25331764705882354,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0473,
      "step": 10766
    },
    {
      "epoch": 0.25334117647058824,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8771,
      "step": 10767
    },
    {
      "epoch": 0.25336470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.362,
      "step": 10768
    },
    {
      "epoch": 0.25338823529411764,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.9949,
      "step": 10769
    },
    {
      "epoch": 0.25341176470588234,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2296,
      "step": 10770
    },
    {
      "epoch": 0.25343529411764704,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2152,
      "step": 10771
    },
    {
      "epoch": 0.2534588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9989,
      "step": 10772
    },
    {
      "epoch": 0.2534823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1042,
      "step": 10773
    },
    {
      "epoch": 0.2535058823529412,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3174,
      "step": 10774
    },
    {
      "epoch": 0.2535294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1335,
      "step": 10775
    },
    {
      "epoch": 0.2535529411764706,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8,
      "step": 10776
    },
    {
      "epoch": 0.2535764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0763,
      "step": 10777
    },
    {
      "epoch": 0.2536,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0579,
      "step": 10778
    },
    {
      "epoch": 0.2536235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0839,
      "step": 10779
    },
    {
      "epoch": 0.2536470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2521,
      "step": 10780
    },
    {
      "epoch": 0.2536705882352941,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.6556,
      "step": 10781
    },
    {
      "epoch": 0.25369411764705885,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 10782
    },
    {
      "epoch": 0.25371764705882355,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0395,
      "step": 10783
    },
    {
      "epoch": 0.25374117647058825,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3215,
      "step": 10784
    },
    {
      "epoch": 0.25376470588235295,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0871,
      "step": 10785
    },
    {
      "epoch": 0.25378823529411765,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8771,
      "step": 10786
    },
    {
      "epoch": 0.25381176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.4243,
      "step": 10787
    },
    {
      "epoch": 0.25383529411764705,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0722,
      "step": 10788
    },
    {
      "epoch": 0.25385882352941175,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2034,
      "step": 10789
    },
    {
      "epoch": 0.25388235294117645,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9015,
      "step": 10790
    },
    {
      "epoch": 0.2539058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 10791
    },
    {
      "epoch": 0.2539294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0381,
      "step": 10792
    },
    {
      "epoch": 0.2539529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1502,
      "step": 10793
    },
    {
      "epoch": 0.2539764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3092,
      "step": 10794
    },
    {
      "epoch": 0.254,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 10795
    },
    {
      "epoch": 0.2540235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3321,
      "step": 10796
    },
    {
      "epoch": 0.2540470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1172,
      "step": 10797
    },
    {
      "epoch": 0.2540705882352941,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9851,
      "step": 10798
    },
    {
      "epoch": 0.2540941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8593,
      "step": 10799
    },
    {
      "epoch": 0.2541176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0084,
      "step": 10800
    },
    {
      "epoch": 0.25414117647058826,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0879,
      "step": 10801
    },
    {
      "epoch": 0.25416470588235296,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1143,
      "step": 10802
    },
    {
      "epoch": 0.25418823529411766,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2659,
      "step": 10803
    },
    {
      "epoch": 0.25421176470588236,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 10804
    },
    {
      "epoch": 0.25423529411764706,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7689,
      "step": 10805
    },
    {
      "epoch": 0.25425882352941176,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9819,
      "step": 10806
    },
    {
      "epoch": 0.25428235294117646,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7888,
      "step": 10807
    },
    {
      "epoch": 0.25430588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2579,
      "step": 10808
    },
    {
      "epoch": 0.25432941176470586,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0094,
      "step": 10809
    },
    {
      "epoch": 0.2543529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 10810
    },
    {
      "epoch": 0.2543764705882353,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.8656,
      "step": 10811
    },
    {
      "epoch": 0.2544,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0145,
      "step": 10812
    },
    {
      "epoch": 0.2544235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1828,
      "step": 10813
    },
    {
      "epoch": 0.2544470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 10814
    },
    {
      "epoch": 0.2544705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2001,
      "step": 10815
    },
    {
      "epoch": 0.2544941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 10816
    },
    {
      "epoch": 0.2545176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1629,
      "step": 10817
    },
    {
      "epoch": 0.2545411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8602,
      "step": 10818
    },
    {
      "epoch": 0.2545647058823529,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.5614,
      "step": 10819
    },
    {
      "epoch": 0.25458823529411767,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1763,
      "step": 10820
    },
    {
      "epoch": 0.25461176470588237,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0049,
      "step": 10821
    },
    {
      "epoch": 0.25463529411764707,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1152,
      "step": 10822
    },
    {
      "epoch": 0.25465882352941177,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0917,
      "step": 10823
    },
    {
      "epoch": 0.25468235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0293,
      "step": 10824
    },
    {
      "epoch": 0.25470588235294117,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0945,
      "step": 10825
    },
    {
      "epoch": 0.25472941176470587,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1404,
      "step": 10826
    },
    {
      "epoch": 0.25475294117647057,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2743,
      "step": 10827
    },
    {
      "epoch": 0.25477647058823527,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.149,
      "step": 10828
    },
    {
      "epoch": 0.2548,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.4384,
      "step": 10829
    },
    {
      "epoch": 0.2548235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 10830
    },
    {
      "epoch": 0.2548470588235294,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7063,
      "step": 10831
    },
    {
      "epoch": 0.2548705882352941,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0242,
      "step": 10832
    },
    {
      "epoch": 0.2548941176470588,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9312,
      "step": 10833
    },
    {
      "epoch": 0.2549176470588235,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8667,
      "step": 10834
    },
    {
      "epoch": 0.2549411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1141,
      "step": 10835
    },
    {
      "epoch": 0.2549647058823529,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7678,
      "step": 10836
    },
    {
      "epoch": 0.2549882352941176,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.124,
      "step": 10837
    },
    {
      "epoch": 0.2550117647058823,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8433,
      "step": 10838
    },
    {
      "epoch": 0.2550352941176471,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8565,
      "step": 10839
    },
    {
      "epoch": 0.2550588235294118,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8414,
      "step": 10840
    },
    {
      "epoch": 0.2550823529411765,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1215,
      "step": 10841
    },
    {
      "epoch": 0.2551058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 10842
    },
    {
      "epoch": 0.2551294117647059,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2017,
      "step": 10843
    },
    {
      "epoch": 0.2551529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9514,
      "step": 10844
    },
    {
      "epoch": 0.2551764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0938,
      "step": 10845
    },
    {
      "epoch": 0.2552,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0513,
      "step": 10846
    },
    {
      "epoch": 0.2552235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9436,
      "step": 10847
    },
    {
      "epoch": 0.25524705882352944,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9341,
      "step": 10848
    },
    {
      "epoch": 0.25527058823529414,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0454,
      "step": 10849
    },
    {
      "epoch": 0.25529411764705884,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.18,
      "step": 10850
    },
    {
      "epoch": 0.25531764705882354,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9887,
      "step": 10851
    },
    {
      "epoch": 0.25534117647058824,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2517,
      "step": 10852
    },
    {
      "epoch": 0.25536470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2076,
      "step": 10853
    },
    {
      "epoch": 0.25538823529411764,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0978,
      "step": 10854
    },
    {
      "epoch": 0.25541176470588234,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0316,
      "step": 10855
    },
    {
      "epoch": 0.25543529411764704,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9979,
      "step": 10856
    },
    {
      "epoch": 0.25545882352941174,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 1.3681,
      "step": 10857
    },
    {
      "epoch": 0.2554823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1569,
      "step": 10858
    },
    {
      "epoch": 0.2555058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1541,
      "step": 10859
    },
    {
      "epoch": 0.2555294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2626,
      "step": 10860
    },
    {
      "epoch": 0.2555529411764706,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.7709,
      "step": 10861
    },
    {
      "epoch": 0.2555764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.23,
      "step": 10862
    },
    {
      "epoch": 0.2556,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8267,
      "step": 10863
    },
    {
      "epoch": 0.2556235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0188,
      "step": 10864
    },
    {
      "epoch": 0.2556470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8042,
      "step": 10865
    },
    {
      "epoch": 0.2556705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0132,
      "step": 10866
    },
    {
      "epoch": 0.25569411764705885,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0065,
      "step": 10867
    },
    {
      "epoch": 0.25571764705882355,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0165,
      "step": 10868
    },
    {
      "epoch": 0.25574117647058825,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.941,
      "step": 10869
    },
    {
      "epoch": 0.25576470588235295,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1638,
      "step": 10870
    },
    {
      "epoch": 0.25578823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8651,
      "step": 10871
    },
    {
      "epoch": 0.25581176470588235,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.805,
      "step": 10872
    },
    {
      "epoch": 0.25583529411764705,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3143,
      "step": 10873
    },
    {
      "epoch": 0.25585882352941175,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8777,
      "step": 10874
    },
    {
      "epoch": 0.25588235294117645,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3674,
      "step": 10875
    },
    {
      "epoch": 0.25590588235294115,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0787,
      "step": 10876
    },
    {
      "epoch": 0.2559294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1306,
      "step": 10877
    },
    {
      "epoch": 0.2559529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.357,
      "step": 10878
    },
    {
      "epoch": 0.2559764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0719,
      "step": 10879
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.042,
      "step": 10880
    },
    {
      "epoch": 0.2560235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2409,
      "step": 10881
    },
    {
      "epoch": 0.2560470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1293,
      "step": 10882
    },
    {
      "epoch": 0.2560705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8498,
      "step": 10883
    },
    {
      "epoch": 0.2560941176470588,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.891,
      "step": 10884
    },
    {
      "epoch": 0.2561176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1433,
      "step": 10885
    },
    {
      "epoch": 0.25614117647058826,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8249,
      "step": 10886
    },
    {
      "epoch": 0.25616470588235296,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0073,
      "step": 10887
    },
    {
      "epoch": 0.25618823529411766,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.4256,
      "step": 10888
    },
    {
      "epoch": 0.25621176470588236,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 10889
    },
    {
      "epoch": 0.25623529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0134,
      "step": 10890
    },
    {
      "epoch": 0.25625882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.1217,
      "step": 10891
    },
    {
      "epoch": 0.25628235294117646,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9182,
      "step": 10892
    },
    {
      "epoch": 0.25630588235294116,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0703,
      "step": 10893
    },
    {
      "epoch": 0.25632941176470586,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1503,
      "step": 10894
    },
    {
      "epoch": 0.25635294117647056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2059,
      "step": 10895
    },
    {
      "epoch": 0.2563764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3329,
      "step": 10896
    },
    {
      "epoch": 0.2564,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0442,
      "step": 10897
    },
    {
      "epoch": 0.2564235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0902,
      "step": 10898
    },
    {
      "epoch": 0.2564470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1962,
      "step": 10899
    },
    {
      "epoch": 0.2564705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0358,
      "step": 10900
    },
    {
      "epoch": 0.2564941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2769,
      "step": 10901
    },
    {
      "epoch": 0.2565176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2719,
      "step": 10902
    },
    {
      "epoch": 0.2565411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8874,
      "step": 10903
    },
    {
      "epoch": 0.2565647058823529,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.673,
      "step": 10904
    },
    {
      "epoch": 0.2565882352941177,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9465,
      "step": 10905
    },
    {
      "epoch": 0.2566117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0034,
      "step": 10906
    },
    {
      "epoch": 0.2566352941176471,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0608,
      "step": 10907
    },
    {
      "epoch": 0.2566588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8862,
      "step": 10908
    },
    {
      "epoch": 0.2566823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0016,
      "step": 10909
    },
    {
      "epoch": 0.2567058823529412,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8954,
      "step": 10910
    },
    {
      "epoch": 0.2567294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0863,
      "step": 10911
    },
    {
      "epoch": 0.2567529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0319,
      "step": 10912
    },
    {
      "epoch": 0.2567764705882353,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7088,
      "step": 10913
    },
    {
      "epoch": 0.2568,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1767,
      "step": 10914
    },
    {
      "epoch": 0.25682352941176473,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.6796,
      "step": 10915
    },
    {
      "epoch": 0.25684705882352943,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.7248,
      "step": 10916
    },
    {
      "epoch": 0.25687058823529413,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2145,
      "step": 10917
    },
    {
      "epoch": 0.25689411764705883,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2707,
      "step": 10918
    },
    {
      "epoch": 0.25691764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2021,
      "step": 10919
    },
    {
      "epoch": 0.25694117647058823,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2667,
      "step": 10920
    },
    {
      "epoch": 0.25696470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1454,
      "step": 10921
    },
    {
      "epoch": 0.25698823529411763,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1779,
      "step": 10922
    },
    {
      "epoch": 0.25701176470588233,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0172,
      "step": 10923
    },
    {
      "epoch": 0.2570352941176471,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2114,
      "step": 10924
    },
    {
      "epoch": 0.2570588235294118,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0719,
      "step": 10925
    },
    {
      "epoch": 0.2570823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1689,
      "step": 10926
    },
    {
      "epoch": 0.2571058823529412,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.5942,
      "step": 10927
    },
    {
      "epoch": 0.2571294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2858,
      "step": 10928
    },
    {
      "epoch": 0.2571529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8975,
      "step": 10929
    },
    {
      "epoch": 0.2571764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0612,
      "step": 10930
    },
    {
      "epoch": 0.2572,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0803,
      "step": 10931
    },
    {
      "epoch": 0.2572235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9059,
      "step": 10932
    },
    {
      "epoch": 0.2572470588235294,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9919,
      "step": 10933
    },
    {
      "epoch": 0.25727058823529414,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.982,
      "step": 10934
    },
    {
      "epoch": 0.25729411764705884,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 10935
    },
    {
      "epoch": 0.25731764705882354,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2742,
      "step": 10936
    },
    {
      "epoch": 0.25734117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6821,
      "step": 10937
    },
    {
      "epoch": 0.25736470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.161,
      "step": 10938
    },
    {
      "epoch": 0.25738823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.98,
      "step": 10939
    },
    {
      "epoch": 0.25741176470588234,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1111,
      "step": 10940
    },
    {
      "epoch": 0.25743529411764704,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9067,
      "step": 10941
    },
    {
      "epoch": 0.25745882352941174,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.049,
      "step": 10942
    },
    {
      "epoch": 0.2574823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 10943
    },
    {
      "epoch": 0.2575058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8402,
      "step": 10944
    },
    {
      "epoch": 0.2575294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 10945
    },
    {
      "epoch": 0.2575529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0483,
      "step": 10946
    },
    {
      "epoch": 0.2575764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3646,
      "step": 10947
    },
    {
      "epoch": 0.2576,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 10948
    },
    {
      "epoch": 0.2576235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0307,
      "step": 10949
    },
    {
      "epoch": 0.2576470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.153,
      "step": 10950
    },
    {
      "epoch": 0.2576705882352941,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2728,
      "step": 10951
    },
    {
      "epoch": 0.25769411764705885,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2588,
      "step": 10952
    },
    {
      "epoch": 0.25771764705882355,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2038,
      "step": 10953
    },
    {
      "epoch": 0.25774117647058825,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3351,
      "step": 10954
    },
    {
      "epoch": 0.25776470588235295,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7961,
      "step": 10955
    },
    {
      "epoch": 0.25778823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1812,
      "step": 10956
    },
    {
      "epoch": 0.25781176470588235,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7757,
      "step": 10957
    },
    {
      "epoch": 0.25783529411764705,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9232,
      "step": 10958
    },
    {
      "epoch": 0.25785882352941175,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1575,
      "step": 10959
    },
    {
      "epoch": 0.25788235294117645,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1588,
      "step": 10960
    },
    {
      "epoch": 0.25790588235294115,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9091,
      "step": 10961
    },
    {
      "epoch": 0.2579294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2523,
      "step": 10962
    },
    {
      "epoch": 0.2579529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2095,
      "step": 10963
    },
    {
      "epoch": 0.2579764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9216,
      "step": 10964
    },
    {
      "epoch": 0.258,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 1.0276,
      "step": 10965
    },
    {
      "epoch": 0.2580235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0405,
      "step": 10966
    },
    {
      "epoch": 0.2580470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0962,
      "step": 10967
    },
    {
      "epoch": 0.2580705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6758,
      "step": 10968
    },
    {
      "epoch": 0.2580941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1927,
      "step": 10969
    },
    {
      "epoch": 0.2581176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0358,
      "step": 10970
    },
    {
      "epoch": 0.25814117647058826,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8347,
      "step": 10971
    },
    {
      "epoch": 0.25816470588235296,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1509,
      "step": 10972
    },
    {
      "epoch": 0.25818823529411766,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.5738,
      "step": 10973
    },
    {
      "epoch": 0.25821176470588236,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0838,
      "step": 10974
    },
    {
      "epoch": 0.25823529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 10975
    },
    {
      "epoch": 0.25825882352941176,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.8327,
      "step": 10976
    },
    {
      "epoch": 0.25828235294117646,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1561,
      "step": 10977
    },
    {
      "epoch": 0.25830588235294116,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9271,
      "step": 10978
    },
    {
      "epoch": 0.25832941176470586,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1135,
      "step": 10979
    },
    {
      "epoch": 0.25835294117647056,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.058,
      "step": 10980
    },
    {
      "epoch": 0.2583764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.063,
      "step": 10981
    },
    {
      "epoch": 0.2584,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9887,
      "step": 10982
    },
    {
      "epoch": 0.2584235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0547,
      "step": 10983
    },
    {
      "epoch": 0.2584470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0131,
      "step": 10984
    },
    {
      "epoch": 0.2584705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0574,
      "step": 10985
    },
    {
      "epoch": 0.2584941176470588,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0511,
      "step": 10986
    },
    {
      "epoch": 0.2585176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2816,
      "step": 10987
    },
    {
      "epoch": 0.2585411764705882,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8454,
      "step": 10988
    },
    {
      "epoch": 0.2585647058823529,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1219,
      "step": 10989
    },
    {
      "epoch": 0.2585882352941177,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.4536,
      "step": 10990
    },
    {
      "epoch": 0.2586117647058824,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.906,
      "step": 10991
    },
    {
      "epoch": 0.2586352941176471,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1301,
      "step": 10992
    },
    {
      "epoch": 0.2586588235294118,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1139,
      "step": 10993
    },
    {
      "epoch": 0.2586823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1283,
      "step": 10994
    },
    {
      "epoch": 0.2587058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1903,
      "step": 10995
    },
    {
      "epoch": 0.2587294117647059,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.9714,
      "step": 10996
    },
    {
      "epoch": 0.2587529411764706,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8174,
      "step": 10997
    },
    {
      "epoch": 0.2587764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0736,
      "step": 10998
    },
    {
      "epoch": 0.2588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1086,
      "step": 10999
    },
    {
      "epoch": 0.25882352941176473,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 11000
    },
    {
      "epoch": 0.25882352941176473,
      "eval_loss": 2.2128984928131104,
      "eval_runtime": 680.7307,
      "eval_samples_per_second": 12.487,
      "eval_steps_per_second": 3.122,
      "step": 11000
    },
    {
      "epoch": 0.25884705882352943,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 11001
    },
    {
      "epoch": 0.25887058823529413,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9866,
      "step": 11002
    },
    {
      "epoch": 0.25889411764705883,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3266,
      "step": 11003
    },
    {
      "epoch": 0.25891764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.3377,
      "step": 11004
    },
    {
      "epoch": 0.25894117647058823,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1361,
      "step": 11005
    },
    {
      "epoch": 0.25896470588235293,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9552,
      "step": 11006
    },
    {
      "epoch": 0.25898823529411763,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2066,
      "step": 11007
    },
    {
      "epoch": 0.25901176470588233,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8306,
      "step": 11008
    },
    {
      "epoch": 0.2590352941176471,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0703,
      "step": 11009
    },
    {
      "epoch": 0.2590588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.32,
      "step": 11010
    },
    {
      "epoch": 0.2590823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9174,
      "step": 11011
    },
    {
      "epoch": 0.2591058823529412,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.7084,
      "step": 11012
    },
    {
      "epoch": 0.2591294117647059,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.6489,
      "step": 11013
    },
    {
      "epoch": 0.2591529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0152,
      "step": 11014
    },
    {
      "epoch": 0.2591764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8665,
      "step": 11015
    },
    {
      "epoch": 0.2592,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8922,
      "step": 11016
    },
    {
      "epoch": 0.2592235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9809,
      "step": 11017
    },
    {
      "epoch": 0.2592470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.3177,
      "step": 11018
    },
    {
      "epoch": 0.25927058823529414,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6035,
      "step": 11019
    },
    {
      "epoch": 0.25929411764705884,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1138,
      "step": 11020
    },
    {
      "epoch": 0.25931764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.962,
      "step": 11021
    },
    {
      "epoch": 0.25934117647058824,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7569,
      "step": 11022
    },
    {
      "epoch": 0.25936470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9401,
      "step": 11023
    },
    {
      "epoch": 0.25938823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9442,
      "step": 11024
    },
    {
      "epoch": 0.25941176470588234,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0301,
      "step": 11025
    },
    {
      "epoch": 0.25943529411764704,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3259,
      "step": 11026
    },
    {
      "epoch": 0.25945882352941174,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8663,
      "step": 11027
    },
    {
      "epoch": 0.2594823529411765,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7449,
      "step": 11028
    },
    {
      "epoch": 0.2595058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 11029
    },
    {
      "epoch": 0.2595294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1018,
      "step": 11030
    },
    {
      "epoch": 0.2595529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1789,
      "step": 11031
    },
    {
      "epoch": 0.2595764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2712,
      "step": 11032
    },
    {
      "epoch": 0.2596,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1911,
      "step": 11033
    },
    {
      "epoch": 0.2596235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3445,
      "step": 11034
    },
    {
      "epoch": 0.2596470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1779,
      "step": 11035
    },
    {
      "epoch": 0.2596705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9904,
      "step": 11036
    },
    {
      "epoch": 0.2596941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8994,
      "step": 11037
    },
    {
      "epoch": 0.25971764705882355,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.118,
      "step": 11038
    },
    {
      "epoch": 0.25974117647058825,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3124,
      "step": 11039
    },
    {
      "epoch": 0.25976470588235295,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0241,
      "step": 11040
    },
    {
      "epoch": 0.25978823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2153,
      "step": 11041
    },
    {
      "epoch": 0.25981176470588235,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.7235,
      "step": 11042
    },
    {
      "epoch": 0.25983529411764705,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 11043
    },
    {
      "epoch": 0.25985882352941175,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9282,
      "step": 11044
    },
    {
      "epoch": 0.25988235294117645,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.962,
      "step": 11045
    },
    {
      "epoch": 0.25990588235294115,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0191,
      "step": 11046
    },
    {
      "epoch": 0.2599294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2136,
      "step": 11047
    },
    {
      "epoch": 0.2599529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.957,
      "step": 11048
    },
    {
      "epoch": 0.2599764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0915,
      "step": 11049
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2507,
      "step": 11050
    },
    {
      "epoch": 0.2600235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9992,
      "step": 11051
    },
    {
      "epoch": 0.2600470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2842,
      "step": 11052
    },
    {
      "epoch": 0.2600705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1631,
      "step": 11053
    },
    {
      "epoch": 0.2600941176470588,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3751,
      "step": 11054
    },
    {
      "epoch": 0.2601176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9645,
      "step": 11055
    },
    {
      "epoch": 0.2601411764705882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3282,
      "step": 11056
    },
    {
      "epoch": 0.26016470588235296,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9706,
      "step": 11057
    },
    {
      "epoch": 0.26018823529411766,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1229,
      "step": 11058
    },
    {
      "epoch": 0.26021176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0418,
      "step": 11059
    },
    {
      "epoch": 0.26023529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.075,
      "step": 11060
    },
    {
      "epoch": 0.26025882352941176,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1614,
      "step": 11061
    },
    {
      "epoch": 0.26028235294117646,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.738,
      "step": 11062
    },
    {
      "epoch": 0.26030588235294116,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9689,
      "step": 11063
    },
    {
      "epoch": 0.26032941176470586,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9149,
      "step": 11064
    },
    {
      "epoch": 0.26035294117647056,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 11065
    },
    {
      "epoch": 0.2603764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.054,
      "step": 11066
    },
    {
      "epoch": 0.2604,
      "grad_norm": 0.248046875,
      "learning_rate": 0.02,
      "loss": 0.7348,
      "step": 11067
    },
    {
      "epoch": 0.2604235294117647,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1911,
      "step": 11068
    },
    {
      "epoch": 0.2604470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2739,
      "step": 11069
    },
    {
      "epoch": 0.2604705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1018,
      "step": 11070
    },
    {
      "epoch": 0.2604941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9493,
      "step": 11071
    },
    {
      "epoch": 0.2605176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0,
      "step": 11072
    },
    {
      "epoch": 0.2605411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9072,
      "step": 11073
    },
    {
      "epoch": 0.2605647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0349,
      "step": 11074
    },
    {
      "epoch": 0.2605882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1829,
      "step": 11075
    },
    {
      "epoch": 0.2606117647058824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1934,
      "step": 11076
    },
    {
      "epoch": 0.2606352941176471,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9023,
      "step": 11077
    },
    {
      "epoch": 0.2606588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8155,
      "step": 11078
    },
    {
      "epoch": 0.2606823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2031,
      "step": 11079
    },
    {
      "epoch": 0.2607058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2596,
      "step": 11080
    },
    {
      "epoch": 0.2607294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 1.1313,
      "step": 11081
    },
    {
      "epoch": 0.2607529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7833,
      "step": 11082
    },
    {
      "epoch": 0.2607764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3638,
      "step": 11083
    },
    {
      "epoch": 0.2608,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8728,
      "step": 11084
    },
    {
      "epoch": 0.26082352941176473,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.128,
      "step": 11085
    },
    {
      "epoch": 0.26084705882352943,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1615,
      "step": 11086
    },
    {
      "epoch": 0.26087058823529413,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3125,
      "step": 11087
    },
    {
      "epoch": 0.26089411764705883,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0749,
      "step": 11088
    },
    {
      "epoch": 0.26091764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0479,
      "step": 11089
    },
    {
      "epoch": 0.26094117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1151,
      "step": 11090
    },
    {
      "epoch": 0.26096470588235293,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 11091
    },
    {
      "epoch": 0.26098823529411763,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2575,
      "step": 11092
    },
    {
      "epoch": 0.26101176470588233,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0773,
      "step": 11093
    },
    {
      "epoch": 0.26103529411764703,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 1.0518,
      "step": 11094
    },
    {
      "epoch": 0.2610588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.055,
      "step": 11095
    },
    {
      "epoch": 0.2610823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.3439,
      "step": 11096
    },
    {
      "epoch": 0.2611058823529412,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1518,
      "step": 11097
    },
    {
      "epoch": 0.2611294117647059,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9228,
      "step": 11098
    },
    {
      "epoch": 0.2611529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1402,
      "step": 11099
    },
    {
      "epoch": 0.2611764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8374,
      "step": 11100
    },
    {
      "epoch": 0.2612,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1198,
      "step": 11101
    },
    {
      "epoch": 0.2612235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1012,
      "step": 11102
    },
    {
      "epoch": 0.2612470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0288,
      "step": 11103
    },
    {
      "epoch": 0.26127058823529414,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2389,
      "step": 11104
    },
    {
      "epoch": 0.26129411764705884,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 11105
    },
    {
      "epoch": 0.26131764705882354,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 1.0019,
      "step": 11106
    },
    {
      "epoch": 0.26134117647058824,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0997,
      "step": 11107
    },
    {
      "epoch": 0.26136470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.144,
      "step": 11108
    },
    {
      "epoch": 0.26138823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1046,
      "step": 11109
    },
    {
      "epoch": 0.26141176470588234,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3628,
      "step": 11110
    },
    {
      "epoch": 0.26143529411764704,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2166,
      "step": 11111
    },
    {
      "epoch": 0.26145882352941174,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9059,
      "step": 11112
    },
    {
      "epoch": 0.26148235294117644,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9907,
      "step": 11113
    },
    {
      "epoch": 0.2615058823529412,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8165,
      "step": 11114
    },
    {
      "epoch": 0.2615294117647059,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9803,
      "step": 11115
    },
    {
      "epoch": 0.2615529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1353,
      "step": 11116
    },
    {
      "epoch": 0.2615764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7684,
      "step": 11117
    },
    {
      "epoch": 0.2616,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8649,
      "step": 11118
    },
    {
      "epoch": 0.2616235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1548,
      "step": 11119
    },
    {
      "epoch": 0.2616470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.943,
      "step": 11120
    },
    {
      "epoch": 0.2616705882352941,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1613,
      "step": 11121
    },
    {
      "epoch": 0.2616941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9612,
      "step": 11122
    },
    {
      "epoch": 0.26171764705882355,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1377,
      "step": 11123
    },
    {
      "epoch": 0.26174117647058825,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 11124
    },
    {
      "epoch": 0.26176470588235295,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3709,
      "step": 11125
    },
    {
      "epoch": 0.26178823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9157,
      "step": 11126
    },
    {
      "epoch": 0.26181176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1901,
      "step": 11127
    },
    {
      "epoch": 0.26183529411764705,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.839,
      "step": 11128
    },
    {
      "epoch": 0.26185882352941175,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9424,
      "step": 11129
    },
    {
      "epoch": 0.26188235294117646,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1357,
      "step": 11130
    },
    {
      "epoch": 0.26190588235294116,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9195,
      "step": 11131
    },
    {
      "epoch": 0.26192941176470586,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0672,
      "step": 11132
    },
    {
      "epoch": 0.2619529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.992,
      "step": 11133
    },
    {
      "epoch": 0.2619764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1994,
      "step": 11134
    },
    {
      "epoch": 0.262,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0352,
      "step": 11135
    },
    {
      "epoch": 0.2620235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0841,
      "step": 11136
    },
    {
      "epoch": 0.2620470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8869,
      "step": 11137
    },
    {
      "epoch": 0.2620705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9263,
      "step": 11138
    },
    {
      "epoch": 0.2620941176470588,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8811,
      "step": 11139
    },
    {
      "epoch": 0.2621176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0926,
      "step": 11140
    },
    {
      "epoch": 0.2621411764705882,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9423,
      "step": 11141
    },
    {
      "epoch": 0.26216470588235297,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1023,
      "step": 11142
    },
    {
      "epoch": 0.26218823529411767,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1447,
      "step": 11143
    },
    {
      "epoch": 0.26221176470588237,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1461,
      "step": 11144
    },
    {
      "epoch": 0.26223529411764707,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3643,
      "step": 11145
    },
    {
      "epoch": 0.26225882352941177,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8166,
      "step": 11146
    },
    {
      "epoch": 0.26228235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.4393,
      "step": 11147
    },
    {
      "epoch": 0.26230588235294117,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8239,
      "step": 11148
    },
    {
      "epoch": 0.26232941176470587,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0422,
      "step": 11149
    },
    {
      "epoch": 0.26235294117647057,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2713,
      "step": 11150
    },
    {
      "epoch": 0.26237647058823527,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.188,
      "step": 11151
    },
    {
      "epoch": 0.2624,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0717,
      "step": 11152
    },
    {
      "epoch": 0.2624235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 11153
    },
    {
      "epoch": 0.2624470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9868,
      "step": 11154
    },
    {
      "epoch": 0.2624705882352941,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.629,
      "step": 11155
    },
    {
      "epoch": 0.2624941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9827,
      "step": 11156
    },
    {
      "epoch": 0.2625176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1009,
      "step": 11157
    },
    {
      "epoch": 0.2625411764705882,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.5852,
      "step": 11158
    },
    {
      "epoch": 0.2625647058823529,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 11159
    },
    {
      "epoch": 0.2625882352941176,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9549,
      "step": 11160
    },
    {
      "epoch": 0.2626117647058824,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8629,
      "step": 11161
    },
    {
      "epoch": 0.2626352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 11162
    },
    {
      "epoch": 0.2626588235294118,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.619,
      "step": 11163
    },
    {
      "epoch": 0.2626823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.3194,
      "step": 11164
    },
    {
      "epoch": 0.2627058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2111,
      "step": 11165
    },
    {
      "epoch": 0.2627294117647059,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9215,
      "step": 11166
    },
    {
      "epoch": 0.2627529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.3458,
      "step": 11167
    },
    {
      "epoch": 0.2627764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.6366,
      "step": 11168
    },
    {
      "epoch": 0.2628,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1408,
      "step": 11169
    },
    {
      "epoch": 0.26282352941176473,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2342,
      "step": 11170
    },
    {
      "epoch": 0.26284705882352943,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 11171
    },
    {
      "epoch": 0.26287058823529413,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.821,
      "step": 11172
    },
    {
      "epoch": 0.26289411764705883,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1538,
      "step": 11173
    },
    {
      "epoch": 0.26291764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0208,
      "step": 11174
    },
    {
      "epoch": 0.26294117647058823,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0384,
      "step": 11175
    },
    {
      "epoch": 0.26296470588235293,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.6907,
      "step": 11176
    },
    {
      "epoch": 0.26298823529411763,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2474,
      "step": 11177
    },
    {
      "epoch": 0.26301176470588233,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1528,
      "step": 11178
    },
    {
      "epoch": 0.26303529411764703,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9308,
      "step": 11179
    },
    {
      "epoch": 0.2630588235294118,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9137,
      "step": 11180
    },
    {
      "epoch": 0.2630823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8745,
      "step": 11181
    },
    {
      "epoch": 0.2631058823529412,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2452,
      "step": 11182
    },
    {
      "epoch": 0.2631294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1674,
      "step": 11183
    },
    {
      "epoch": 0.2631529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2802,
      "step": 11184
    },
    {
      "epoch": 0.2631764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7606,
      "step": 11185
    },
    {
      "epoch": 0.2632,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3191,
      "step": 11186
    },
    {
      "epoch": 0.2632235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 11187
    },
    {
      "epoch": 0.2632470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3212,
      "step": 11188
    },
    {
      "epoch": 0.26327058823529415,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8891,
      "step": 11189
    },
    {
      "epoch": 0.26329411764705885,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1487,
      "step": 11190
    },
    {
      "epoch": 0.26331764705882355,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2081,
      "step": 11191
    },
    {
      "epoch": 0.26334117647058825,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9174,
      "step": 11192
    },
    {
      "epoch": 0.26336470588235295,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0103,
      "step": 11193
    },
    {
      "epoch": 0.26338823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.816,
      "step": 11194
    },
    {
      "epoch": 0.26341176470588235,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.5499,
      "step": 11195
    },
    {
      "epoch": 0.26343529411764705,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1263,
      "step": 11196
    },
    {
      "epoch": 0.26345882352941175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2184,
      "step": 11197
    },
    {
      "epoch": 0.26348235294117645,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9624,
      "step": 11198
    },
    {
      "epoch": 0.2635058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2993,
      "step": 11199
    },
    {
      "epoch": 0.2635294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2427,
      "step": 11200
    },
    {
      "epoch": 0.2635529411764706,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7482,
      "step": 11201
    },
    {
      "epoch": 0.2635764705882353,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7017,
      "step": 11202
    },
    {
      "epoch": 0.2636,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.969,
      "step": 11203
    },
    {
      "epoch": 0.2636235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0664,
      "step": 11204
    },
    {
      "epoch": 0.2636470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1159,
      "step": 11205
    },
    {
      "epoch": 0.2636705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.942,
      "step": 11206
    },
    {
      "epoch": 0.2636941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2697,
      "step": 11207
    },
    {
      "epoch": 0.26371764705882356,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9057,
      "step": 11208
    },
    {
      "epoch": 0.26374117647058826,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9498,
      "step": 11209
    },
    {
      "epoch": 0.26376470588235296,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9903,
      "step": 11210
    },
    {
      "epoch": 0.26378823529411766,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1848,
      "step": 11211
    },
    {
      "epoch": 0.26381176470588236,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2978,
      "step": 11212
    },
    {
      "epoch": 0.26383529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1132,
      "step": 11213
    },
    {
      "epoch": 0.26385882352941176,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.9096,
      "step": 11214
    },
    {
      "epoch": 0.26388235294117646,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8255,
      "step": 11215
    },
    {
      "epoch": 0.26390588235294116,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.5531,
      "step": 11216
    },
    {
      "epoch": 0.26392941176470586,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1772,
      "step": 11217
    },
    {
      "epoch": 0.2639529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3751,
      "step": 11218
    },
    {
      "epoch": 0.2639764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0813,
      "step": 11219
    },
    {
      "epoch": 0.264,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.3744,
      "step": 11220
    },
    {
      "epoch": 0.2640235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0397,
      "step": 11221
    },
    {
      "epoch": 0.2640470588235294,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.6078,
      "step": 11222
    },
    {
      "epoch": 0.2640705882352941,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7858,
      "step": 11223
    },
    {
      "epoch": 0.2640941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3689,
      "step": 11224
    },
    {
      "epoch": 0.2641176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2083,
      "step": 11225
    },
    {
      "epoch": 0.2641411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9285,
      "step": 11226
    },
    {
      "epoch": 0.26416470588235297,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9186,
      "step": 11227
    },
    {
      "epoch": 0.26418823529411767,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7852,
      "step": 11228
    },
    {
      "epoch": 0.26421176470588237,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0423,
      "step": 11229
    },
    {
      "epoch": 0.26423529411764707,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1139,
      "step": 11230
    },
    {
      "epoch": 0.26425882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9077,
      "step": 11231
    },
    {
      "epoch": 0.26428235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9167,
      "step": 11232
    },
    {
      "epoch": 0.26430588235294117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 11233
    },
    {
      "epoch": 0.26432941176470587,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0625,
      "step": 11234
    },
    {
      "epoch": 0.26435294117647057,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0379,
      "step": 11235
    },
    {
      "epoch": 0.26437647058823527,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.4356,
      "step": 11236
    },
    {
      "epoch": 0.2644,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1259,
      "step": 11237
    },
    {
      "epoch": 0.2644235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1499,
      "step": 11238
    },
    {
      "epoch": 0.2644470588235294,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.9318,
      "step": 11239
    },
    {
      "epoch": 0.2644705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.146,
      "step": 11240
    },
    {
      "epoch": 0.2644941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0216,
      "step": 11241
    },
    {
      "epoch": 0.2645176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.989,
      "step": 11242
    },
    {
      "epoch": 0.2645411764705882,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3618,
      "step": 11243
    },
    {
      "epoch": 0.2645647058823529,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8553,
      "step": 11244
    },
    {
      "epoch": 0.2645882352941176,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3232,
      "step": 11245
    },
    {
      "epoch": 0.2646117647058824,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 11246
    },
    {
      "epoch": 0.2646352941176471,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9078,
      "step": 11247
    },
    {
      "epoch": 0.2646588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0677,
      "step": 11248
    },
    {
      "epoch": 0.2646823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2186,
      "step": 11249
    },
    {
      "epoch": 0.2647058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1978,
      "step": 11250
    },
    {
      "epoch": 0.2647294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9359,
      "step": 11251
    },
    {
      "epoch": 0.2647529411764706,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0172,
      "step": 11252
    },
    {
      "epoch": 0.2647764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.225,
      "step": 11253
    },
    {
      "epoch": 0.2648,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1368,
      "step": 11254
    },
    {
      "epoch": 0.2648235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0642,
      "step": 11255
    },
    {
      "epoch": 0.26484705882352944,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2373,
      "step": 11256
    },
    {
      "epoch": 0.26487058823529414,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8762,
      "step": 11257
    },
    {
      "epoch": 0.26489411764705884,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0431,
      "step": 11258
    },
    {
      "epoch": 0.26491764705882354,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.013,
      "step": 11259
    },
    {
      "epoch": 0.26494117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9224,
      "step": 11260
    },
    {
      "epoch": 0.26496470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9402,
      "step": 11261
    },
    {
      "epoch": 0.26498823529411764,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8993,
      "step": 11262
    },
    {
      "epoch": 0.26501176470588234,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1854,
      "step": 11263
    },
    {
      "epoch": 0.26503529411764704,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8947,
      "step": 11264
    },
    {
      "epoch": 0.2650588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1594,
      "step": 11265
    },
    {
      "epoch": 0.2650823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.867,
      "step": 11266
    },
    {
      "epoch": 0.2651058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2641,
      "step": 11267
    },
    {
      "epoch": 0.2651294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8658,
      "step": 11268
    },
    {
      "epoch": 0.2651529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1422,
      "step": 11269
    },
    {
      "epoch": 0.2651764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2658,
      "step": 11270
    },
    {
      "epoch": 0.2652,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8308,
      "step": 11271
    },
    {
      "epoch": 0.2652235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0782,
      "step": 11272
    },
    {
      "epoch": 0.2652470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.247,
      "step": 11273
    },
    {
      "epoch": 0.2652705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1682,
      "step": 11274
    },
    {
      "epoch": 0.26529411764705885,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1111,
      "step": 11275
    },
    {
      "epoch": 0.26531764705882355,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3736,
      "step": 11276
    },
    {
      "epoch": 0.26534117647058825,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0137,
      "step": 11277
    },
    {
      "epoch": 0.26536470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.275,
      "step": 11278
    },
    {
      "epoch": 0.26538823529411765,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.7857,
      "step": 11279
    },
    {
      "epoch": 0.26541176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1766,
      "step": 11280
    },
    {
      "epoch": 0.26543529411764705,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8594,
      "step": 11281
    },
    {
      "epoch": 0.26545882352941175,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1871,
      "step": 11282
    },
    {
      "epoch": 0.26548235294117645,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3866,
      "step": 11283
    },
    {
      "epoch": 0.2655058823529412,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9893,
      "step": 11284
    },
    {
      "epoch": 0.2655294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2744,
      "step": 11285
    },
    {
      "epoch": 0.2655529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7681,
      "step": 11286
    },
    {
      "epoch": 0.2655764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 11287
    },
    {
      "epoch": 0.2656,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.256,
      "step": 11288
    },
    {
      "epoch": 0.2656235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8745,
      "step": 11289
    },
    {
      "epoch": 0.2656470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1692,
      "step": 11290
    },
    {
      "epoch": 0.2656705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2934,
      "step": 11291
    },
    {
      "epoch": 0.2656941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9761,
      "step": 11292
    },
    {
      "epoch": 0.2657176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8384,
      "step": 11293
    },
    {
      "epoch": 0.26574117647058826,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0146,
      "step": 11294
    },
    {
      "epoch": 0.26576470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0566,
      "step": 11295
    },
    {
      "epoch": 0.26578823529411766,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 11296
    },
    {
      "epoch": 0.26581176470588236,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 11297
    },
    {
      "epoch": 0.26583529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9708,
      "step": 11298
    },
    {
      "epoch": 0.26585882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1767,
      "step": 11299
    },
    {
      "epoch": 0.26588235294117646,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9335,
      "step": 11300
    },
    {
      "epoch": 0.26590588235294116,
      "grad_norm": 0.265625,
      "learning_rate": 0.02,
      "loss": 0.7235,
      "step": 11301
    },
    {
      "epoch": 0.26592941176470586,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.8998,
      "step": 11302
    },
    {
      "epoch": 0.2659529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1069,
      "step": 11303
    },
    {
      "epoch": 0.2659764705882353,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8211,
      "step": 11304
    },
    {
      "epoch": 0.266,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.6308,
      "step": 11305
    },
    {
      "epoch": 0.2660235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8755,
      "step": 11306
    },
    {
      "epoch": 0.2660470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1366,
      "step": 11307
    },
    {
      "epoch": 0.2660705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 11308
    },
    {
      "epoch": 0.2660941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.097,
      "step": 11309
    },
    {
      "epoch": 0.2661176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9582,
      "step": 11310
    },
    {
      "epoch": 0.2661411764705882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0592,
      "step": 11311
    },
    {
      "epoch": 0.2661647058823529,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.1434,
      "step": 11312
    },
    {
      "epoch": 0.26618823529411767,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8243,
      "step": 11313
    },
    {
      "epoch": 0.26621176470588237,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9868,
      "step": 11314
    },
    {
      "epoch": 0.26623529411764707,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 11315
    },
    {
      "epoch": 0.26625882352941177,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1599,
      "step": 11316
    },
    {
      "epoch": 0.26628235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 11317
    },
    {
      "epoch": 0.26630588235294117,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2413,
      "step": 11318
    },
    {
      "epoch": 0.26632941176470587,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0093,
      "step": 11319
    },
    {
      "epoch": 0.26635294117647057,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0169,
      "step": 11320
    },
    {
      "epoch": 0.26637647058823527,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8621,
      "step": 11321
    },
    {
      "epoch": 0.2664,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.6646,
      "step": 11322
    },
    {
      "epoch": 0.2664235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0431,
      "step": 11323
    },
    {
      "epoch": 0.2664470588235294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9006,
      "step": 11324
    },
    {
      "epoch": 0.2664705882352941,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9024,
      "step": 11325
    },
    {
      "epoch": 0.2664941176470588,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9743,
      "step": 11326
    },
    {
      "epoch": 0.2665176470588235,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0598,
      "step": 11327
    },
    {
      "epoch": 0.2665411764705882,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0459,
      "step": 11328
    },
    {
      "epoch": 0.2665647058823529,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1201,
      "step": 11329
    },
    {
      "epoch": 0.2665882352941176,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0099,
      "step": 11330
    },
    {
      "epoch": 0.2666117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8616,
      "step": 11331
    },
    {
      "epoch": 0.2666352941176471,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1544,
      "step": 11332
    },
    {
      "epoch": 0.2666588235294118,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0269,
      "step": 11333
    },
    {
      "epoch": 0.2666823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0535,
      "step": 11334
    },
    {
      "epoch": 0.2667058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7548,
      "step": 11335
    },
    {
      "epoch": 0.2667294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8712,
      "step": 11336
    },
    {
      "epoch": 0.2667529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0438,
      "step": 11337
    },
    {
      "epoch": 0.2667764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8883,
      "step": 11338
    },
    {
      "epoch": 0.2668,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.982,
      "step": 11339
    },
    {
      "epoch": 0.2668235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 11340
    },
    {
      "epoch": 0.26684705882352944,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7856,
      "step": 11341
    },
    {
      "epoch": 0.26687058823529414,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0582,
      "step": 11342
    },
    {
      "epoch": 0.26689411764705884,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.178,
      "step": 11343
    },
    {
      "epoch": 0.26691764705882354,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9993,
      "step": 11344
    },
    {
      "epoch": 0.26694117647058824,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9468,
      "step": 11345
    },
    {
      "epoch": 0.26696470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0519,
      "step": 11346
    },
    {
      "epoch": 0.26698823529411764,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.6991,
      "step": 11347
    },
    {
      "epoch": 0.26701176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.048,
      "step": 11348
    },
    {
      "epoch": 0.26703529411764704,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9432,
      "step": 11349
    },
    {
      "epoch": 0.26705882352941174,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9127,
      "step": 11350
    },
    {
      "epoch": 0.2670823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 11351
    },
    {
      "epoch": 0.2671058823529412,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 11352
    },
    {
      "epoch": 0.2671294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0639,
      "step": 11353
    },
    {
      "epoch": 0.2671529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8173,
      "step": 11354
    },
    {
      "epoch": 0.2671764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8241,
      "step": 11355
    },
    {
      "epoch": 0.2672,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 11356
    },
    {
      "epoch": 0.2672235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8791,
      "step": 11357
    },
    {
      "epoch": 0.2672470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1235,
      "step": 11358
    },
    {
      "epoch": 0.2672705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2476,
      "step": 11359
    },
    {
      "epoch": 0.26729411764705885,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0053,
      "step": 11360
    },
    {
      "epoch": 0.26731764705882355,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.161,
      "step": 11361
    },
    {
      "epoch": 0.26734117647058825,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1589,
      "step": 11362
    },
    {
      "epoch": 0.26736470588235295,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1312,
      "step": 11363
    },
    {
      "epoch": 0.26738823529411765,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0289,
      "step": 11364
    },
    {
      "epoch": 0.26741176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1901,
      "step": 11365
    },
    {
      "epoch": 0.26743529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2061,
      "step": 11366
    },
    {
      "epoch": 0.26745882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8971,
      "step": 11367
    },
    {
      "epoch": 0.26748235294117645,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9604,
      "step": 11368
    },
    {
      "epoch": 0.26750588235294115,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0573,
      "step": 11369
    },
    {
      "epoch": 0.2675294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9839,
      "step": 11370
    },
    {
      "epoch": 0.2675529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7956,
      "step": 11371
    },
    {
      "epoch": 0.2675764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7788,
      "step": 11372
    },
    {
      "epoch": 0.2676,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1612,
      "step": 11373
    },
    {
      "epoch": 0.2676235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9005,
      "step": 11374
    },
    {
      "epoch": 0.2676470588235294,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7847,
      "step": 11375
    },
    {
      "epoch": 0.2676705882352941,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2014,
      "step": 11376
    },
    {
      "epoch": 0.2676941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9073,
      "step": 11377
    },
    {
      "epoch": 0.2677176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0289,
      "step": 11378
    },
    {
      "epoch": 0.26774117647058826,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9428,
      "step": 11379
    },
    {
      "epoch": 0.26776470588235296,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.7471,
      "step": 11380
    },
    {
      "epoch": 0.26778823529411766,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3687,
      "step": 11381
    },
    {
      "epoch": 0.26781176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9536,
      "step": 11382
    },
    {
      "epoch": 0.26783529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 11383
    },
    {
      "epoch": 0.26785882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0596,
      "step": 11384
    },
    {
      "epoch": 0.26788235294117646,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2381,
      "step": 11385
    },
    {
      "epoch": 0.26790588235294116,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9722,
      "step": 11386
    },
    {
      "epoch": 0.26792941176470586,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0026,
      "step": 11387
    },
    {
      "epoch": 0.2679529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 11388
    },
    {
      "epoch": 0.2679764705882353,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 11389
    },
    {
      "epoch": 0.268,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1224,
      "step": 11390
    },
    {
      "epoch": 0.2680235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1596,
      "step": 11391
    },
    {
      "epoch": 0.2680470588235294,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.745,
      "step": 11392
    },
    {
      "epoch": 0.2680705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8948,
      "step": 11393
    },
    {
      "epoch": 0.2680941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8544,
      "step": 11394
    },
    {
      "epoch": 0.2681176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1796,
      "step": 11395
    },
    {
      "epoch": 0.2681411764705882,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8036,
      "step": 11396
    },
    {
      "epoch": 0.2681647058823529,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2128,
      "step": 11397
    },
    {
      "epoch": 0.26818823529411767,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0138,
      "step": 11398
    },
    {
      "epoch": 0.26821176470588237,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7035,
      "step": 11399
    },
    {
      "epoch": 0.26823529411764707,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8617,
      "step": 11400
    },
    {
      "epoch": 0.26825882352941177,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9805,
      "step": 11401
    },
    {
      "epoch": 0.26828235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2599,
      "step": 11402
    },
    {
      "epoch": 0.26830588235294117,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7585,
      "step": 11403
    },
    {
      "epoch": 0.26832941176470587,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0852,
      "step": 11404
    },
    {
      "epoch": 0.26835294117647057,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.157,
      "step": 11405
    },
    {
      "epoch": 0.26837647058823527,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9471,
      "step": 11406
    },
    {
      "epoch": 0.2684,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7588,
      "step": 11407
    },
    {
      "epoch": 0.2684235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 11408
    },
    {
      "epoch": 0.2684470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2664,
      "step": 11409
    },
    {
      "epoch": 0.2684705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0537,
      "step": 11410
    },
    {
      "epoch": 0.2684941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.955,
      "step": 11411
    },
    {
      "epoch": 0.2685176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8265,
      "step": 11412
    },
    {
      "epoch": 0.2685411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7227,
      "step": 11413
    },
    {
      "epoch": 0.2685647058823529,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9959,
      "step": 11414
    },
    {
      "epoch": 0.2685882352941176,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2192,
      "step": 11415
    },
    {
      "epoch": 0.2686117647058823,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8721,
      "step": 11416
    },
    {
      "epoch": 0.2686352941176471,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0374,
      "step": 11417
    },
    {
      "epoch": 0.2686588235294118,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1158,
      "step": 11418
    },
    {
      "epoch": 0.2686823529411765,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2848,
      "step": 11419
    },
    {
      "epoch": 0.2687058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.3181,
      "step": 11420
    },
    {
      "epoch": 0.2687294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2834,
      "step": 11421
    },
    {
      "epoch": 0.2687529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0922,
      "step": 11422
    },
    {
      "epoch": 0.2687764705882353,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.7115,
      "step": 11423
    },
    {
      "epoch": 0.2688,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.3314,
      "step": 11424
    },
    {
      "epoch": 0.2688235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8489,
      "step": 11425
    },
    {
      "epoch": 0.26884705882352944,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.2442,
      "step": 11426
    },
    {
      "epoch": 0.26887058823529414,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9171,
      "step": 11427
    },
    {
      "epoch": 0.26889411764705884,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0225,
      "step": 11428
    },
    {
      "epoch": 0.26891764705882354,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0745,
      "step": 11429
    },
    {
      "epoch": 0.26894117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9269,
      "step": 11430
    },
    {
      "epoch": 0.26896470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1231,
      "step": 11431
    },
    {
      "epoch": 0.26898823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1357,
      "step": 11432
    },
    {
      "epoch": 0.26901176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1208,
      "step": 11433
    },
    {
      "epoch": 0.26903529411764704,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1399,
      "step": 11434
    },
    {
      "epoch": 0.26905882352941174,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0477,
      "step": 11435
    },
    {
      "epoch": 0.2690823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9729,
      "step": 11436
    },
    {
      "epoch": 0.2691058823529412,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.345,
      "step": 11437
    },
    {
      "epoch": 0.2691294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1892,
      "step": 11438
    },
    {
      "epoch": 0.2691529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1074,
      "step": 11439
    },
    {
      "epoch": 0.2691764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.96,
      "step": 11440
    },
    {
      "epoch": 0.2692,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.902,
      "step": 11441
    },
    {
      "epoch": 0.2692235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1162,
      "step": 11442
    },
    {
      "epoch": 0.2692470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9941,
      "step": 11443
    },
    {
      "epoch": 0.2692705882352941,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9168,
      "step": 11444
    },
    {
      "epoch": 0.26929411764705885,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.1482,
      "step": 11445
    },
    {
      "epoch": 0.26931764705882355,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9162,
      "step": 11446
    },
    {
      "epoch": 0.26934117647058825,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.984,
      "step": 11447
    },
    {
      "epoch": 0.26936470588235295,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2177,
      "step": 11448
    },
    {
      "epoch": 0.26938823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9734,
      "step": 11449
    },
    {
      "epoch": 0.26941176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1006,
      "step": 11450
    },
    {
      "epoch": 0.26943529411764705,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1555,
      "step": 11451
    },
    {
      "epoch": 0.26945882352941175,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7259,
      "step": 11452
    },
    {
      "epoch": 0.26948235294117645,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0489,
      "step": 11453
    },
    {
      "epoch": 0.26950588235294115,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1205,
      "step": 11454
    },
    {
      "epoch": 0.2695294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.168,
      "step": 11455
    },
    {
      "epoch": 0.2695529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9073,
      "step": 11456
    },
    {
      "epoch": 0.2695764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9469,
      "step": 11457
    },
    {
      "epoch": 0.2696,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7847,
      "step": 11458
    },
    {
      "epoch": 0.2696235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1429,
      "step": 11459
    },
    {
      "epoch": 0.2696470588235294,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9126,
      "step": 11460
    },
    {
      "epoch": 0.2696705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0869,
      "step": 11461
    },
    {
      "epoch": 0.2696941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2113,
      "step": 11462
    },
    {
      "epoch": 0.2697176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2556,
      "step": 11463
    },
    {
      "epoch": 0.26974117647058826,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.3164,
      "step": 11464
    },
    {
      "epoch": 0.26976470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1684,
      "step": 11465
    },
    {
      "epoch": 0.26978823529411766,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 0.6541,
      "step": 11466
    },
    {
      "epoch": 0.26981176470588236,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0955,
      "step": 11467
    },
    {
      "epoch": 0.26983529411764706,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.5225,
      "step": 11468
    },
    {
      "epoch": 0.26985882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1363,
      "step": 11469
    },
    {
      "epoch": 0.26988235294117646,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0684,
      "step": 11470
    },
    {
      "epoch": 0.26990588235294116,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7523,
      "step": 11471
    },
    {
      "epoch": 0.26992941176470586,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8035,
      "step": 11472
    },
    {
      "epoch": 0.26995294117647056,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2701,
      "step": 11473
    },
    {
      "epoch": 0.2699764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 11474
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.013,
      "step": 11475
    },
    {
      "epoch": 0.2700235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2899,
      "step": 11476
    },
    {
      "epoch": 0.2700470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.792,
      "step": 11477
    },
    {
      "epoch": 0.2700705882352941,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2324,
      "step": 11478
    },
    {
      "epoch": 0.2700941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.7748,
      "step": 11479
    },
    {
      "epoch": 0.2701176470588235,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.5307,
      "step": 11480
    },
    {
      "epoch": 0.2701411764705882,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9984,
      "step": 11481
    },
    {
      "epoch": 0.2701647058823529,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2434,
      "step": 11482
    },
    {
      "epoch": 0.2701882352941177,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3396,
      "step": 11483
    },
    {
      "epoch": 0.2702117647058824,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.251,
      "step": 11484
    },
    {
      "epoch": 0.2702352941176471,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.462,
      "step": 11485
    },
    {
      "epoch": 0.2702588235294118,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.95,
      "step": 11486
    },
    {
      "epoch": 0.2702823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0374,
      "step": 11487
    },
    {
      "epoch": 0.2703058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 11488
    },
    {
      "epoch": 0.2703294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0673,
      "step": 11489
    },
    {
      "epoch": 0.2703529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0376,
      "step": 11490
    },
    {
      "epoch": 0.2703764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8872,
      "step": 11491
    },
    {
      "epoch": 0.2704,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1263,
      "step": 11492
    },
    {
      "epoch": 0.27042352941176473,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8277,
      "step": 11493
    },
    {
      "epoch": 0.27044705882352943,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0716,
      "step": 11494
    },
    {
      "epoch": 0.27047058823529413,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3206,
      "step": 11495
    },
    {
      "epoch": 0.27049411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7337,
      "step": 11496
    },
    {
      "epoch": 0.27051764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.5774,
      "step": 11497
    },
    {
      "epoch": 0.27054117647058823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 11498
    },
    {
      "epoch": 0.27056470588235293,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2337,
      "step": 11499
    },
    {
      "epoch": 0.27058823529411763,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.7478,
      "step": 11500
    },
    {
      "epoch": 0.27061176470588233,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9298,
      "step": 11501
    },
    {
      "epoch": 0.2706352941176471,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2377,
      "step": 11502
    },
    {
      "epoch": 0.2706588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9466,
      "step": 11503
    },
    {
      "epoch": 0.2706823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9732,
      "step": 11504
    },
    {
      "epoch": 0.2707058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2239,
      "step": 11505
    },
    {
      "epoch": 0.2707294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1537,
      "step": 11506
    },
    {
      "epoch": 0.2707529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.178,
      "step": 11507
    },
    {
      "epoch": 0.2707764705882353,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.5365,
      "step": 11508
    },
    {
      "epoch": 0.2708,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9436,
      "step": 11509
    },
    {
      "epoch": 0.2708235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9113,
      "step": 11510
    },
    {
      "epoch": 0.2708470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9294,
      "step": 11511
    },
    {
      "epoch": 0.27087058823529414,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 11512
    },
    {
      "epoch": 0.27089411764705884,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9256,
      "step": 11513
    },
    {
      "epoch": 0.27091764705882354,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0877,
      "step": 11514
    },
    {
      "epoch": 0.27094117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9695,
      "step": 11515
    },
    {
      "epoch": 0.27096470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.895,
      "step": 11516
    },
    {
      "epoch": 0.27098823529411764,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8987,
      "step": 11517
    },
    {
      "epoch": 0.27101176470588234,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.146,
      "step": 11518
    },
    {
      "epoch": 0.27103529411764704,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 11519
    },
    {
      "epoch": 0.27105882352941174,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 1.0068,
      "step": 11520
    },
    {
      "epoch": 0.2710823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9281,
      "step": 11521
    },
    {
      "epoch": 0.2711058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7241,
      "step": 11522
    },
    {
      "epoch": 0.2711294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9394,
      "step": 11523
    },
    {
      "epoch": 0.2711529411764706,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.262,
      "step": 11524
    },
    {
      "epoch": 0.2711764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0561,
      "step": 11525
    },
    {
      "epoch": 0.2712,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0241,
      "step": 11526
    },
    {
      "epoch": 0.2712235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3369,
      "step": 11527
    },
    {
      "epoch": 0.2712470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1096,
      "step": 11528
    },
    {
      "epoch": 0.2712705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 11529
    },
    {
      "epoch": 0.2712941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9478,
      "step": 11530
    },
    {
      "epoch": 0.27131764705882355,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0133,
      "step": 11531
    },
    {
      "epoch": 0.27134117647058825,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2551,
      "step": 11532
    },
    {
      "epoch": 0.27136470588235295,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1693,
      "step": 11533
    },
    {
      "epoch": 0.27138823529411765,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8999,
      "step": 11534
    },
    {
      "epoch": 0.27141176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.16,
      "step": 11535
    },
    {
      "epoch": 0.27143529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2632,
      "step": 11536
    },
    {
      "epoch": 0.27145882352941175,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2023,
      "step": 11537
    },
    {
      "epoch": 0.27148235294117645,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9446,
      "step": 11538
    },
    {
      "epoch": 0.27150588235294115,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.778,
      "step": 11539
    },
    {
      "epoch": 0.2715294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1704,
      "step": 11540
    },
    {
      "epoch": 0.2715529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.142,
      "step": 11541
    },
    {
      "epoch": 0.2715764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8319,
      "step": 11542
    },
    {
      "epoch": 0.2716,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9816,
      "step": 11543
    },
    {
      "epoch": 0.2716235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7186,
      "step": 11544
    },
    {
      "epoch": 0.2716470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9368,
      "step": 11545
    },
    {
      "epoch": 0.2716705882352941,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8218,
      "step": 11546
    },
    {
      "epoch": 0.2716941176470588,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.4663,
      "step": 11547
    },
    {
      "epoch": 0.2717176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2956,
      "step": 11548
    },
    {
      "epoch": 0.2717411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.146,
      "step": 11549
    },
    {
      "epoch": 0.27176470588235296,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1846,
      "step": 11550
    },
    {
      "epoch": 0.27178823529411766,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0111,
      "step": 11551
    },
    {
      "epoch": 0.27181176470588236,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3045,
      "step": 11552
    },
    {
      "epoch": 0.27183529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7517,
      "step": 11553
    },
    {
      "epoch": 0.27185882352941176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8918,
      "step": 11554
    },
    {
      "epoch": 0.27188235294117646,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2934,
      "step": 11555
    },
    {
      "epoch": 0.27190588235294116,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1859,
      "step": 11556
    },
    {
      "epoch": 0.27192941176470586,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2673,
      "step": 11557
    },
    {
      "epoch": 0.27195294117647056,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.987,
      "step": 11558
    },
    {
      "epoch": 0.2719764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8428,
      "step": 11559
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8809,
      "step": 11560
    },
    {
      "epoch": 0.2720235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1242,
      "step": 11561
    },
    {
      "epoch": 0.2720470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2332,
      "step": 11562
    },
    {
      "epoch": 0.2720705882352941,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0204,
      "step": 11563
    },
    {
      "epoch": 0.2720941176470588,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.8424,
      "step": 11564
    },
    {
      "epoch": 0.2721176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0968,
      "step": 11565
    },
    {
      "epoch": 0.2721411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0327,
      "step": 11566
    },
    {
      "epoch": 0.2721647058823529,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0215,
      "step": 11567
    },
    {
      "epoch": 0.2721882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0129,
      "step": 11568
    },
    {
      "epoch": 0.2722117647058824,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0813,
      "step": 11569
    },
    {
      "epoch": 0.2722352941176471,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9239,
      "step": 11570
    },
    {
      "epoch": 0.2722588235294118,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9896,
      "step": 11571
    },
    {
      "epoch": 0.2722823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9388,
      "step": 11572
    },
    {
      "epoch": 0.2723058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0634,
      "step": 11573
    },
    {
      "epoch": 0.2723294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9546,
      "step": 11574
    },
    {
      "epoch": 0.2723529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3395,
      "step": 11575
    },
    {
      "epoch": 0.2723764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1694,
      "step": 11576
    },
    {
      "epoch": 0.2724,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9947,
      "step": 11577
    },
    {
      "epoch": 0.27242352941176473,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.6846,
      "step": 11578
    },
    {
      "epoch": 0.27244705882352943,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0699,
      "step": 11579
    },
    {
      "epoch": 0.27247058823529413,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.5458,
      "step": 11580
    },
    {
      "epoch": 0.27249411764705883,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8817,
      "step": 11581
    },
    {
      "epoch": 0.27251764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2843,
      "step": 11582
    },
    {
      "epoch": 0.27254117647058823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9361,
      "step": 11583
    },
    {
      "epoch": 0.27256470588235293,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 1.0085,
      "step": 11584
    },
    {
      "epoch": 0.27258823529411763,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0339,
      "step": 11585
    },
    {
      "epoch": 0.27261176470588233,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1165,
      "step": 11586
    },
    {
      "epoch": 0.27263529411764703,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9429,
      "step": 11587
    },
    {
      "epoch": 0.2726588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0569,
      "step": 11588
    },
    {
      "epoch": 0.2726823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9708,
      "step": 11589
    },
    {
      "epoch": 0.2727058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1425,
      "step": 11590
    },
    {
      "epoch": 0.2727294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9431,
      "step": 11591
    },
    {
      "epoch": 0.2727529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0194,
      "step": 11592
    },
    {
      "epoch": 0.2727764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.588,
      "step": 11593
    },
    {
      "epoch": 0.2728,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1999,
      "step": 11594
    },
    {
      "epoch": 0.2728235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1689,
      "step": 11595
    },
    {
      "epoch": 0.2728470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.121,
      "step": 11596
    },
    {
      "epoch": 0.27287058823529414,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.044,
      "step": 11597
    },
    {
      "epoch": 0.27289411764705884,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3233,
      "step": 11598
    },
    {
      "epoch": 0.27291764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2225,
      "step": 11599
    },
    {
      "epoch": 0.27294117647058824,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1698,
      "step": 11600
    },
    {
      "epoch": 0.27296470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9677,
      "step": 11601
    },
    {
      "epoch": 0.27298823529411764,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1981,
      "step": 11602
    },
    {
      "epoch": 0.27301176470588234,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0384,
      "step": 11603
    },
    {
      "epoch": 0.27303529411764704,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.6626,
      "step": 11604
    },
    {
      "epoch": 0.27305882352941174,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1482,
      "step": 11605
    },
    {
      "epoch": 0.2730823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 11606
    },
    {
      "epoch": 0.2731058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8063,
      "step": 11607
    },
    {
      "epoch": 0.2731294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8338,
      "step": 11608
    },
    {
      "epoch": 0.2731529411764706,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3598,
      "step": 11609
    },
    {
      "epoch": 0.2731764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.2244,
      "step": 11610
    },
    {
      "epoch": 0.2732,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.706,
      "step": 11611
    },
    {
      "epoch": 0.2732235294117647,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.7318,
      "step": 11612
    },
    {
      "epoch": 0.2732470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2053,
      "step": 11613
    },
    {
      "epoch": 0.2732705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1171,
      "step": 11614
    },
    {
      "epoch": 0.2732941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1898,
      "step": 11615
    },
    {
      "epoch": 0.27331764705882355,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8408,
      "step": 11616
    },
    {
      "epoch": 0.27334117647058825,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0953,
      "step": 11617
    },
    {
      "epoch": 0.27336470588235295,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3185,
      "step": 11618
    },
    {
      "epoch": 0.27338823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2178,
      "step": 11619
    },
    {
      "epoch": 0.27341176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1323,
      "step": 11620
    },
    {
      "epoch": 0.27343529411764705,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.4145,
      "step": 11621
    },
    {
      "epoch": 0.27345882352941175,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 11622
    },
    {
      "epoch": 0.27348235294117645,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0705,
      "step": 11623
    },
    {
      "epoch": 0.27350588235294115,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2433,
      "step": 11624
    },
    {
      "epoch": 0.2735294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0119,
      "step": 11625
    },
    {
      "epoch": 0.2735529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1422,
      "step": 11626
    },
    {
      "epoch": 0.2735764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8321,
      "step": 11627
    },
    {
      "epoch": 0.2736,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.057,
      "step": 11628
    },
    {
      "epoch": 0.2736235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1331,
      "step": 11629
    },
    {
      "epoch": 0.2736470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2641,
      "step": 11630
    },
    {
      "epoch": 0.2736705882352941,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9808,
      "step": 11631
    },
    {
      "epoch": 0.2736941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1829,
      "step": 11632
    },
    {
      "epoch": 0.2737176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1312,
      "step": 11633
    },
    {
      "epoch": 0.2737411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9572,
      "step": 11634
    },
    {
      "epoch": 0.27376470588235297,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.7593,
      "step": 11635
    },
    {
      "epoch": 0.27378823529411767,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 11636
    },
    {
      "epoch": 0.27381176470588237,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.8518,
      "step": 11637
    },
    {
      "epoch": 0.27383529411764707,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7162,
      "step": 11638
    },
    {
      "epoch": 0.27385882352941177,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.127,
      "step": 11639
    },
    {
      "epoch": 0.27388235294117647,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9638,
      "step": 11640
    },
    {
      "epoch": 0.27390588235294117,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.726,
      "step": 11641
    },
    {
      "epoch": 0.27392941176470587,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3323,
      "step": 11642
    },
    {
      "epoch": 0.27395294117647057,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9334,
      "step": 11643
    },
    {
      "epoch": 0.2739764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.6978,
      "step": 11644
    },
    {
      "epoch": 0.274,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1106,
      "step": 11645
    },
    {
      "epoch": 0.2740235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6714,
      "step": 11646
    },
    {
      "epoch": 0.2740470588235294,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7414,
      "step": 11647
    },
    {
      "epoch": 0.2740705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2415,
      "step": 11648
    },
    {
      "epoch": 0.2740941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7395,
      "step": 11649
    },
    {
      "epoch": 0.2741176470588235,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3503,
      "step": 11650
    },
    {
      "epoch": 0.2741411764705882,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8213,
      "step": 11651
    },
    {
      "epoch": 0.2741647058823529,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0262,
      "step": 11652
    },
    {
      "epoch": 0.2741882352941176,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2077,
      "step": 11653
    },
    {
      "epoch": 0.2742117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.107,
      "step": 11654
    },
    {
      "epoch": 0.2742352941176471,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1869,
      "step": 11655
    },
    {
      "epoch": 0.2742588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9131,
      "step": 11656
    },
    {
      "epoch": 0.2742823529411765,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3635,
      "step": 11657
    },
    {
      "epoch": 0.2743058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.2263,
      "step": 11658
    },
    {
      "epoch": 0.2743294117647059,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.407,
      "step": 11659
    },
    {
      "epoch": 0.2743529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0905,
      "step": 11660
    },
    {
      "epoch": 0.2743764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7986,
      "step": 11661
    },
    {
      "epoch": 0.2744,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1262,
      "step": 11662
    },
    {
      "epoch": 0.27442352941176473,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.7419,
      "step": 11663
    },
    {
      "epoch": 0.27444705882352943,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0386,
      "step": 11664
    },
    {
      "epoch": 0.27447058823529413,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1122,
      "step": 11665
    },
    {
      "epoch": 0.27449411764705883,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2997,
      "step": 11666
    },
    {
      "epoch": 0.27451764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3715,
      "step": 11667
    },
    {
      "epoch": 0.27454117647058823,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0226,
      "step": 11668
    },
    {
      "epoch": 0.27456470588235293,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.292,
      "step": 11669
    },
    {
      "epoch": 0.27458823529411763,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 11670
    },
    {
      "epoch": 0.27461176470588233,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.987,
      "step": 11671
    },
    {
      "epoch": 0.27463529411764703,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8645,
      "step": 11672
    },
    {
      "epoch": 0.2746588235294118,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2259,
      "step": 11673
    },
    {
      "epoch": 0.2746823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.331,
      "step": 11674
    },
    {
      "epoch": 0.2747058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.009,
      "step": 11675
    },
    {
      "epoch": 0.2747294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 11676
    },
    {
      "epoch": 0.2747529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9428,
      "step": 11677
    },
    {
      "epoch": 0.2747764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1406,
      "step": 11678
    },
    {
      "epoch": 0.2748,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2708,
      "step": 11679
    },
    {
      "epoch": 0.2748235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0442,
      "step": 11680
    },
    {
      "epoch": 0.2748470588235294,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7919,
      "step": 11681
    },
    {
      "epoch": 0.27487058823529414,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 11682
    },
    {
      "epoch": 0.27489411764705884,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9011,
      "step": 11683
    },
    {
      "epoch": 0.27491764705882354,
      "grad_norm": 0.259765625,
      "learning_rate": 0.02,
      "loss": 0.7846,
      "step": 11684
    },
    {
      "epoch": 0.27494117647058824,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8397,
      "step": 11685
    },
    {
      "epoch": 0.27496470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.306,
      "step": 11686
    },
    {
      "epoch": 0.27498823529411764,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9364,
      "step": 11687
    },
    {
      "epoch": 0.27501176470588234,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2143,
      "step": 11688
    },
    {
      "epoch": 0.27503529411764704,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.5873,
      "step": 11689
    },
    {
      "epoch": 0.27505882352941174,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8496,
      "step": 11690
    },
    {
      "epoch": 0.27508235294117644,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0846,
      "step": 11691
    },
    {
      "epoch": 0.2751058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9347,
      "step": 11692
    },
    {
      "epoch": 0.2751294117647059,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.7268,
      "step": 11693
    },
    {
      "epoch": 0.2751529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2551,
      "step": 11694
    },
    {
      "epoch": 0.2751764705882353,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2714,
      "step": 11695
    },
    {
      "epoch": 0.2752,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0063,
      "step": 11696
    },
    {
      "epoch": 0.2752235294117647,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9754,
      "step": 11697
    },
    {
      "epoch": 0.2752470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1967,
      "step": 11698
    },
    {
      "epoch": 0.2752705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2019,
      "step": 11699
    },
    {
      "epoch": 0.2752941176470588,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9422,
      "step": 11700
    },
    {
      "epoch": 0.27531764705882356,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0384,
      "step": 11701
    },
    {
      "epoch": 0.27534117647058826,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9922,
      "step": 11702
    },
    {
      "epoch": 0.27536470588235296,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.6817,
      "step": 11703
    },
    {
      "epoch": 0.27538823529411766,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9571,
      "step": 11704
    },
    {
      "epoch": 0.27541176470588236,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0666,
      "step": 11705
    },
    {
      "epoch": 0.27543529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0163,
      "step": 11706
    },
    {
      "epoch": 0.27545882352941176,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1279,
      "step": 11707
    },
    {
      "epoch": 0.27548235294117646,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0728,
      "step": 11708
    },
    {
      "epoch": 0.27550588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1782,
      "step": 11709
    },
    {
      "epoch": 0.27552941176470586,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1789,
      "step": 11710
    },
    {
      "epoch": 0.2755529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9352,
      "step": 11711
    },
    {
      "epoch": 0.2755764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8881,
      "step": 11712
    },
    {
      "epoch": 0.2756,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2719,
      "step": 11713
    },
    {
      "epoch": 0.2756235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.176,
      "step": 11714
    },
    {
      "epoch": 0.2756470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.963,
      "step": 11715
    },
    {
      "epoch": 0.2756705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7237,
      "step": 11716
    },
    {
      "epoch": 0.2756941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1581,
      "step": 11717
    },
    {
      "epoch": 0.2757176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3708,
      "step": 11718
    },
    {
      "epoch": 0.2757411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0352,
      "step": 11719
    },
    {
      "epoch": 0.27576470588235297,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 11720
    },
    {
      "epoch": 0.27578823529411767,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8101,
      "step": 11721
    },
    {
      "epoch": 0.27581176470588237,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.4292,
      "step": 11722
    },
    {
      "epoch": 0.27583529411764707,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1172,
      "step": 11723
    },
    {
      "epoch": 0.27585882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0598,
      "step": 11724
    },
    {
      "epoch": 0.27588235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9851,
      "step": 11725
    },
    {
      "epoch": 0.27590588235294117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0622,
      "step": 11726
    },
    {
      "epoch": 0.27592941176470587,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9857,
      "step": 11727
    },
    {
      "epoch": 0.27595294117647057,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.8301,
      "step": 11728
    },
    {
      "epoch": 0.27597647058823527,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9431,
      "step": 11729
    },
    {
      "epoch": 0.276,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0595,
      "step": 11730
    },
    {
      "epoch": 0.2760235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.6929,
      "step": 11731
    },
    {
      "epoch": 0.2760470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8581,
      "step": 11732
    },
    {
      "epoch": 0.2760705882352941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9353,
      "step": 11733
    },
    {
      "epoch": 0.2760941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0915,
      "step": 11734
    },
    {
      "epoch": 0.2761176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9283,
      "step": 11735
    },
    {
      "epoch": 0.2761411764705882,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8632,
      "step": 11736
    },
    {
      "epoch": 0.2761647058823529,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7861,
      "step": 11737
    },
    {
      "epoch": 0.2761882352941176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1116,
      "step": 11738
    },
    {
      "epoch": 0.2762117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0977,
      "step": 11739
    },
    {
      "epoch": 0.2762352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 11740
    },
    {
      "epoch": 0.2762588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0902,
      "step": 11741
    },
    {
      "epoch": 0.2762823529411765,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7634,
      "step": 11742
    },
    {
      "epoch": 0.2763058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8784,
      "step": 11743
    },
    {
      "epoch": 0.2763294117647059,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7279,
      "step": 11744
    },
    {
      "epoch": 0.2763529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.356,
      "step": 11745
    },
    {
      "epoch": 0.2763764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9991,
      "step": 11746
    },
    {
      "epoch": 0.2764,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 11747
    },
    {
      "epoch": 0.2764235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9894,
      "step": 11748
    },
    {
      "epoch": 0.27644705882352943,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2154,
      "step": 11749
    },
    {
      "epoch": 0.27647058823529413,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8499,
      "step": 11750
    },
    {
      "epoch": 0.27649411764705883,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1061,
      "step": 11751
    },
    {
      "epoch": 0.27651764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.309,
      "step": 11752
    },
    {
      "epoch": 0.27654117647058823,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.8169,
      "step": 11753
    },
    {
      "epoch": 0.27656470588235293,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.1287,
      "step": 11754
    },
    {
      "epoch": 0.27658823529411763,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 11755
    },
    {
      "epoch": 0.27661176470588233,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0656,
      "step": 11756
    },
    {
      "epoch": 0.27663529411764703,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7234,
      "step": 11757
    },
    {
      "epoch": 0.2766588235294118,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1649,
      "step": 11758
    },
    {
      "epoch": 0.2766823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9265,
      "step": 11759
    },
    {
      "epoch": 0.2767058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2406,
      "step": 11760
    },
    {
      "epoch": 0.2767294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2155,
      "step": 11761
    },
    {
      "epoch": 0.2767529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8698,
      "step": 11762
    },
    {
      "epoch": 0.2767764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8546,
      "step": 11763
    },
    {
      "epoch": 0.2768,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 11764
    },
    {
      "epoch": 0.2768235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.101,
      "step": 11765
    },
    {
      "epoch": 0.2768470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 11766
    },
    {
      "epoch": 0.2768705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0632,
      "step": 11767
    },
    {
      "epoch": 0.27689411764705885,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1409,
      "step": 11768
    },
    {
      "epoch": 0.27691764705882355,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1145,
      "step": 11769
    },
    {
      "epoch": 0.27694117647058825,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 11770
    },
    {
      "epoch": 0.27696470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9089,
      "step": 11771
    },
    {
      "epoch": 0.27698823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0405,
      "step": 11772
    },
    {
      "epoch": 0.27701176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0955,
      "step": 11773
    },
    {
      "epoch": 0.27703529411764705,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8421,
      "step": 11774
    },
    {
      "epoch": 0.27705882352941175,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0114,
      "step": 11775
    },
    {
      "epoch": 0.27708235294117645,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0234,
      "step": 11776
    },
    {
      "epoch": 0.2771058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.059,
      "step": 11777
    },
    {
      "epoch": 0.2771294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0979,
      "step": 11778
    },
    {
      "epoch": 0.2771529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1224,
      "step": 11779
    },
    {
      "epoch": 0.2771764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.022,
      "step": 11780
    },
    {
      "epoch": 0.2772,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3697,
      "step": 11781
    },
    {
      "epoch": 0.2772235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0152,
      "step": 11782
    },
    {
      "epoch": 0.2772470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2471,
      "step": 11783
    },
    {
      "epoch": 0.2772705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0954,
      "step": 11784
    },
    {
      "epoch": 0.2772941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8407,
      "step": 11785
    },
    {
      "epoch": 0.2773176470588235,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.8578,
      "step": 11786
    },
    {
      "epoch": 0.27734117647058826,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.6582,
      "step": 11787
    },
    {
      "epoch": 0.27736470588235296,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2049,
      "step": 11788
    },
    {
      "epoch": 0.27738823529411766,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2137,
      "step": 11789
    },
    {
      "epoch": 0.27741176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9056,
      "step": 11790
    },
    {
      "epoch": 0.27743529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0641,
      "step": 11791
    },
    {
      "epoch": 0.27745882352941176,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2699,
      "step": 11792
    },
    {
      "epoch": 0.27748235294117646,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.11,
      "step": 11793
    },
    {
      "epoch": 0.27750588235294116,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1767,
      "step": 11794
    },
    {
      "epoch": 0.27752941176470586,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9111,
      "step": 11795
    },
    {
      "epoch": 0.2775529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2299,
      "step": 11796
    },
    {
      "epoch": 0.2775764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1307,
      "step": 11797
    },
    {
      "epoch": 0.2776,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0868,
      "step": 11798
    },
    {
      "epoch": 0.2776235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.096,
      "step": 11799
    },
    {
      "epoch": 0.2776470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.941,
      "step": 11800
    },
    {
      "epoch": 0.2776705882352941,
      "grad_norm": 0.25,
      "learning_rate": 0.02,
      "loss": 0.8265,
      "step": 11801
    },
    {
      "epoch": 0.2776941176470588,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1547,
      "step": 11802
    },
    {
      "epoch": 0.2777176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7752,
      "step": 11803
    },
    {
      "epoch": 0.2777411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 11804
    },
    {
      "epoch": 0.2777647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3444,
      "step": 11805
    },
    {
      "epoch": 0.27778823529411767,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 11806
    },
    {
      "epoch": 0.27781176470588237,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9818,
      "step": 11807
    },
    {
      "epoch": 0.27783529411764707,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8749,
      "step": 11808
    },
    {
      "epoch": 0.27785882352941177,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1341,
      "step": 11809
    },
    {
      "epoch": 0.27788235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8329,
      "step": 11810
    },
    {
      "epoch": 0.27790588235294117,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.094,
      "step": 11811
    },
    {
      "epoch": 0.27792941176470587,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.9581,
      "step": 11812
    },
    {
      "epoch": 0.27795294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 11813
    },
    {
      "epoch": 0.27797647058823527,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0597,
      "step": 11814
    },
    {
      "epoch": 0.278,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.6962,
      "step": 11815
    },
    {
      "epoch": 0.2780235294117647,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.6799,
      "step": 11816
    },
    {
      "epoch": 0.2780470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1245,
      "step": 11817
    },
    {
      "epoch": 0.2780705882352941,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8384,
      "step": 11818
    },
    {
      "epoch": 0.2780941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1989,
      "step": 11819
    },
    {
      "epoch": 0.2781176470588235,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.765,
      "step": 11820
    },
    {
      "epoch": 0.2781411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9349,
      "step": 11821
    },
    {
      "epoch": 0.2781647058823529,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1193,
      "step": 11822
    },
    {
      "epoch": 0.2781882352941176,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8643,
      "step": 11823
    },
    {
      "epoch": 0.2782117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9229,
      "step": 11824
    },
    {
      "epoch": 0.2782352941176471,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.218,
      "step": 11825
    },
    {
      "epoch": 0.2782588235294118,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9474,
      "step": 11826
    },
    {
      "epoch": 0.2782823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3033,
      "step": 11827
    },
    {
      "epoch": 0.2783058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8453,
      "step": 11828
    },
    {
      "epoch": 0.2783294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9192,
      "step": 11829
    },
    {
      "epoch": 0.2783529411764706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2817,
      "step": 11830
    },
    {
      "epoch": 0.2783764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.5551,
      "step": 11831
    },
    {
      "epoch": 0.2784,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6864,
      "step": 11832
    },
    {
      "epoch": 0.2784235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1051,
      "step": 11833
    },
    {
      "epoch": 0.27844705882352944,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8965,
      "step": 11834
    },
    {
      "epoch": 0.27847058823529414,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.8113,
      "step": 11835
    },
    {
      "epoch": 0.27849411764705884,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1343,
      "step": 11836
    },
    {
      "epoch": 0.27851764705882354,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1484,
      "step": 11837
    },
    {
      "epoch": 0.27854117647058824,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1825,
      "step": 11838
    },
    {
      "epoch": 0.27856470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.17,
      "step": 11839
    },
    {
      "epoch": 0.27858823529411764,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 11840
    },
    {
      "epoch": 0.27861176470588234,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.447,
      "step": 11841
    },
    {
      "epoch": 0.27863529411764704,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1676,
      "step": 11842
    },
    {
      "epoch": 0.2786588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2495,
      "step": 11843
    },
    {
      "epoch": 0.2786823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1032,
      "step": 11844
    },
    {
      "epoch": 0.2787058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9686,
      "step": 11845
    },
    {
      "epoch": 0.2787294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8827,
      "step": 11846
    },
    {
      "epoch": 0.2787529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3716,
      "step": 11847
    },
    {
      "epoch": 0.2787764705882353,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1951,
      "step": 11848
    },
    {
      "epoch": 0.2788,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0112,
      "step": 11849
    },
    {
      "epoch": 0.2788235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8311,
      "step": 11850
    },
    {
      "epoch": 0.2788470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2064,
      "step": 11851
    },
    {
      "epoch": 0.2788705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.272,
      "step": 11852
    },
    {
      "epoch": 0.27889411764705885,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.773,
      "step": 11853
    },
    {
      "epoch": 0.27891764705882355,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0724,
      "step": 11854
    },
    {
      "epoch": 0.27894117647058825,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8476,
      "step": 11855
    },
    {
      "epoch": 0.27896470588235295,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.979,
      "step": 11856
    },
    {
      "epoch": 0.27898823529411765,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.4387,
      "step": 11857
    },
    {
      "epoch": 0.27901176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2725,
      "step": 11858
    },
    {
      "epoch": 0.27903529411764705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3242,
      "step": 11859
    },
    {
      "epoch": 0.27905882352941175,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 11860
    },
    {
      "epoch": 0.27908235294117645,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.9029,
      "step": 11861
    },
    {
      "epoch": 0.2791058823529412,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9077,
      "step": 11862
    },
    {
      "epoch": 0.2791294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0451,
      "step": 11863
    },
    {
      "epoch": 0.2791529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 11864
    },
    {
      "epoch": 0.2791764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9921,
      "step": 11865
    },
    {
      "epoch": 0.2792,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0848,
      "step": 11866
    },
    {
      "epoch": 0.2792235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0801,
      "step": 11867
    },
    {
      "epoch": 0.2792470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.12,
      "step": 11868
    },
    {
      "epoch": 0.2792705882352941,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7505,
      "step": 11869
    },
    {
      "epoch": 0.2792941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0343,
      "step": 11870
    },
    {
      "epoch": 0.2793176470588235,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8491,
      "step": 11871
    },
    {
      "epoch": 0.27934117647058826,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0251,
      "step": 11872
    },
    {
      "epoch": 0.27936470588235296,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1722,
      "step": 11873
    },
    {
      "epoch": 0.27938823529411766,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8942,
      "step": 11874
    },
    {
      "epoch": 0.27941176470588236,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1248,
      "step": 11875
    },
    {
      "epoch": 0.27943529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 11876
    },
    {
      "epoch": 0.27945882352941176,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0468,
      "step": 11877
    },
    {
      "epoch": 0.27948235294117646,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.6832,
      "step": 11878
    },
    {
      "epoch": 0.27950588235294116,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 11879
    },
    {
      "epoch": 0.27952941176470586,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1181,
      "step": 11880
    },
    {
      "epoch": 0.2795529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1236,
      "step": 11881
    },
    {
      "epoch": 0.2795764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3408,
      "step": 11882
    },
    {
      "epoch": 0.2796,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2737,
      "step": 11883
    },
    {
      "epoch": 0.2796235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1258,
      "step": 11884
    },
    {
      "epoch": 0.2796470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 11885
    },
    {
      "epoch": 0.2796705882352941,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.853,
      "step": 11886
    },
    {
      "epoch": 0.2796941176470588,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.5148,
      "step": 11887
    },
    {
      "epoch": 0.2797176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1018,
      "step": 11888
    },
    {
      "epoch": 0.2797411764705882,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2525,
      "step": 11889
    },
    {
      "epoch": 0.2797647058823529,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0585,
      "step": 11890
    },
    {
      "epoch": 0.27978823529411767,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1915,
      "step": 11891
    },
    {
      "epoch": 0.27981176470588237,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1926,
      "step": 11892
    },
    {
      "epoch": 0.27983529411764707,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8683,
      "step": 11893
    },
    {
      "epoch": 0.27985882352941177,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9376,
      "step": 11894
    },
    {
      "epoch": 0.27988235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.984,
      "step": 11895
    },
    {
      "epoch": 0.27990588235294117,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 11896
    },
    {
      "epoch": 0.27992941176470587,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 11897
    },
    {
      "epoch": 0.27995294117647057,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0867,
      "step": 11898
    },
    {
      "epoch": 0.27997647058823527,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.223,
      "step": 11899
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1023,
      "step": 11900
    },
    {
      "epoch": 0.2800235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1389,
      "step": 11901
    },
    {
      "epoch": 0.2800470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.3451,
      "step": 11902
    },
    {
      "epoch": 0.2800705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9787,
      "step": 11903
    },
    {
      "epoch": 0.2800941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1785,
      "step": 11904
    },
    {
      "epoch": 0.2801176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0411,
      "step": 11905
    },
    {
      "epoch": 0.2801411764705882,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0385,
      "step": 11906
    },
    {
      "epoch": 0.2801647058823529,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0898,
      "step": 11907
    },
    {
      "epoch": 0.2801882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8951,
      "step": 11908
    },
    {
      "epoch": 0.2802117647058823,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.5363,
      "step": 11909
    },
    {
      "epoch": 0.2802352941176471,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1594,
      "step": 11910
    },
    {
      "epoch": 0.2802588235294118,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0256,
      "step": 11911
    },
    {
      "epoch": 0.2802823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1636,
      "step": 11912
    },
    {
      "epoch": 0.2803058823529412,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0363,
      "step": 11913
    },
    {
      "epoch": 0.2803294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2509,
      "step": 11914
    },
    {
      "epoch": 0.2803529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9525,
      "step": 11915
    },
    {
      "epoch": 0.2803764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1566,
      "step": 11916
    },
    {
      "epoch": 0.2804,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8125,
      "step": 11917
    },
    {
      "epoch": 0.2804235294117647,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7389,
      "step": 11918
    },
    {
      "epoch": 0.28044705882352944,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0574,
      "step": 11919
    },
    {
      "epoch": 0.28047058823529414,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8187,
      "step": 11920
    },
    {
      "epoch": 0.28049411764705884,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9149,
      "step": 11921
    },
    {
      "epoch": 0.28051764705882354,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 11922
    },
    {
      "epoch": 0.28054117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2352,
      "step": 11923
    },
    {
      "epoch": 0.28056470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9803,
      "step": 11924
    },
    {
      "epoch": 0.28058823529411764,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9178,
      "step": 11925
    },
    {
      "epoch": 0.28061176470588234,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0539,
      "step": 11926
    },
    {
      "epoch": 0.28063529411764704,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.916,
      "step": 11927
    },
    {
      "epoch": 0.28065882352941174,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8427,
      "step": 11928
    },
    {
      "epoch": 0.2806823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1607,
      "step": 11929
    },
    {
      "epoch": 0.2807058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0763,
      "step": 11930
    },
    {
      "epoch": 0.2807294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1135,
      "step": 11931
    },
    {
      "epoch": 0.2807529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9578,
      "step": 11932
    },
    {
      "epoch": 0.2807764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9113,
      "step": 11933
    },
    {
      "epoch": 0.2808,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.497,
      "step": 11934
    },
    {
      "epoch": 0.2808235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7562,
      "step": 11935
    },
    {
      "epoch": 0.2808470588235294,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6681,
      "step": 11936
    },
    {
      "epoch": 0.2808705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1694,
      "step": 11937
    },
    {
      "epoch": 0.28089411764705885,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0879,
      "step": 11938
    },
    {
      "epoch": 0.28091764705882355,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8378,
      "step": 11939
    },
    {
      "epoch": 0.28094117647058825,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1683,
      "step": 11940
    },
    {
      "epoch": 0.28096470588235295,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2407,
      "step": 11941
    },
    {
      "epoch": 0.28098823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.902,
      "step": 11942
    },
    {
      "epoch": 0.28101176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2735,
      "step": 11943
    },
    {
      "epoch": 0.28103529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0824,
      "step": 11944
    },
    {
      "epoch": 0.28105882352941175,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0779,
      "step": 11945
    },
    {
      "epoch": 0.28108235294117645,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2251,
      "step": 11946
    },
    {
      "epoch": 0.28110588235294115,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.147,
      "step": 11947
    },
    {
      "epoch": 0.2811294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1456,
      "step": 11948
    },
    {
      "epoch": 0.2811529411764706,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.02,
      "loss": 0.936,
      "step": 11949
    },
    {
      "epoch": 0.2811764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9677,
      "step": 11950
    },
    {
      "epoch": 0.2812,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9541,
      "step": 11951
    },
    {
      "epoch": 0.2812235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3846,
      "step": 11952
    },
    {
      "epoch": 0.2812470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2864,
      "step": 11953
    },
    {
      "epoch": 0.2812705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0389,
      "step": 11954
    },
    {
      "epoch": 0.2812941176470588,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2869,
      "step": 11955
    },
    {
      "epoch": 0.2813176470588235,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.4307,
      "step": 11956
    },
    {
      "epoch": 0.28134117647058826,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 11957
    },
    {
      "epoch": 0.28136470588235296,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0361,
      "step": 11958
    },
    {
      "epoch": 0.28138823529411766,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9206,
      "step": 11959
    },
    {
      "epoch": 0.28141176470588236,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.8328,
      "step": 11960
    },
    {
      "epoch": 0.28143529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0756,
      "step": 11961
    },
    {
      "epoch": 0.28145882352941176,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.6494,
      "step": 11962
    },
    {
      "epoch": 0.28148235294117646,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1651,
      "step": 11963
    },
    {
      "epoch": 0.28150588235294116,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1662,
      "step": 11964
    },
    {
      "epoch": 0.28152941176470586,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9722,
      "step": 11965
    },
    {
      "epoch": 0.28155294117647056,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7802,
      "step": 11966
    },
    {
      "epoch": 0.2815764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8241,
      "step": 11967
    },
    {
      "epoch": 0.2816,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8531,
      "step": 11968
    },
    {
      "epoch": 0.2816235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9964,
      "step": 11969
    },
    {
      "epoch": 0.2816470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0809,
      "step": 11970
    },
    {
      "epoch": 0.2816705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9895,
      "step": 11971
    },
    {
      "epoch": 0.2816941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.014,
      "step": 11972
    },
    {
      "epoch": 0.2817176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.004,
      "step": 11973
    },
    {
      "epoch": 0.2817411764705882,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2395,
      "step": 11974
    },
    {
      "epoch": 0.2817647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8247,
      "step": 11975
    },
    {
      "epoch": 0.2817882352941177,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9587,
      "step": 11976
    },
    {
      "epoch": 0.2818117647058824,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9539,
      "step": 11977
    },
    {
      "epoch": 0.2818352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1304,
      "step": 11978
    },
    {
      "epoch": 0.2818588235294118,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6783,
      "step": 11979
    },
    {
      "epoch": 0.2818823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.284,
      "step": 11980
    },
    {
      "epoch": 0.2819058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 11981
    },
    {
      "epoch": 0.2819294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 11982
    },
    {
      "epoch": 0.2819529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7952,
      "step": 11983
    },
    {
      "epoch": 0.2819764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 1.0132,
      "step": 11984
    },
    {
      "epoch": 0.282,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.5107,
      "step": 11985
    },
    {
      "epoch": 0.28202352941176473,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1549,
      "step": 11986
    },
    {
      "epoch": 0.28204705882352943,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1934,
      "step": 11987
    },
    {
      "epoch": 0.28207058823529413,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9363,
      "step": 11988
    },
    {
      "epoch": 0.28209411764705883,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.06,
      "step": 11989
    },
    {
      "epoch": 0.28211764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2003,
      "step": 11990
    },
    {
      "epoch": 0.28214117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0317,
      "step": 11991
    },
    {
      "epoch": 0.28216470588235293,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8635,
      "step": 11992
    },
    {
      "epoch": 0.28218823529411763,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0304,
      "step": 11993
    },
    {
      "epoch": 0.28221176470588233,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.143,
      "step": 11994
    },
    {
      "epoch": 0.2822352941176471,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9634,
      "step": 11995
    },
    {
      "epoch": 0.2822588235294118,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9196,
      "step": 11996
    },
    {
      "epoch": 0.2822823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9848,
      "step": 11997
    },
    {
      "epoch": 0.2823058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2279,
      "step": 11998
    },
    {
      "epoch": 0.2823294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.014,
      "step": 11999
    },
    {
      "epoch": 0.2823529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1547,
      "step": 12000
    },
    {
      "epoch": 0.2823529411764706,
      "eval_loss": 2.211031675338745,
      "eval_runtime": 684.4193,
      "eval_samples_per_second": 12.419,
      "eval_steps_per_second": 3.105,
      "step": 12000
    },
    {
      "epoch": 0.2823764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.688,
      "step": 12001
    },
    {
      "epoch": 0.2824,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0545,
      "step": 12002
    },
    {
      "epoch": 0.2824235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6519,
      "step": 12003
    },
    {
      "epoch": 0.2824470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.033,
      "step": 12004
    },
    {
      "epoch": 0.28247058823529414,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2179,
      "step": 12005
    },
    {
      "epoch": 0.28249411764705884,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8302,
      "step": 12006
    },
    {
      "epoch": 0.28251764705882354,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0665,
      "step": 12007
    },
    {
      "epoch": 0.28254117647058824,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8198,
      "step": 12008
    },
    {
      "epoch": 0.28256470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9093,
      "step": 12009
    },
    {
      "epoch": 0.28258823529411764,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0075,
      "step": 12010
    },
    {
      "epoch": 0.28261176470588234,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1493,
      "step": 12011
    },
    {
      "epoch": 0.28263529411764704,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9909,
      "step": 12012
    },
    {
      "epoch": 0.28265882352941174,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0264,
      "step": 12013
    },
    {
      "epoch": 0.2826823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0935,
      "step": 12014
    },
    {
      "epoch": 0.2827058823529412,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8191,
      "step": 12015
    },
    {
      "epoch": 0.2827294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0133,
      "step": 12016
    },
    {
      "epoch": 0.2827529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.01,
      "step": 12017
    },
    {
      "epoch": 0.2827764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1794,
      "step": 12018
    },
    {
      "epoch": 0.2828,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1812,
      "step": 12019
    },
    {
      "epoch": 0.2828235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9427,
      "step": 12020
    },
    {
      "epoch": 0.2828470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1919,
      "step": 12021
    },
    {
      "epoch": 0.2828705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9036,
      "step": 12022
    },
    {
      "epoch": 0.2828941176470588,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2165,
      "step": 12023
    },
    {
      "epoch": 0.28291764705882355,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7866,
      "step": 12024
    },
    {
      "epoch": 0.28294117647058825,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1939,
      "step": 12025
    },
    {
      "epoch": 0.28296470588235295,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3418,
      "step": 12026
    },
    {
      "epoch": 0.28298823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9618,
      "step": 12027
    },
    {
      "epoch": 0.28301176470588235,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3104,
      "step": 12028
    },
    {
      "epoch": 0.28303529411764705,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1104,
      "step": 12029
    },
    {
      "epoch": 0.28305882352941175,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2939,
      "step": 12030
    },
    {
      "epoch": 0.28308235294117645,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3414,
      "step": 12031
    },
    {
      "epoch": 0.28310588235294115,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8649,
      "step": 12032
    },
    {
      "epoch": 0.2831294117647059,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0274,
      "step": 12033
    },
    {
      "epoch": 0.2831529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 12034
    },
    {
      "epoch": 0.2831764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1682,
      "step": 12035
    },
    {
      "epoch": 0.2832,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8304,
      "step": 12036
    },
    {
      "epoch": 0.2832235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.933,
      "step": 12037
    },
    {
      "epoch": 0.2832470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 12038
    },
    {
      "epoch": 0.2832705882352941,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8563,
      "step": 12039
    },
    {
      "epoch": 0.2832941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9821,
      "step": 12040
    },
    {
      "epoch": 0.2833176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9275,
      "step": 12041
    },
    {
      "epoch": 0.28334117647058826,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8675,
      "step": 12042
    },
    {
      "epoch": 0.28336470588235296,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9205,
      "step": 12043
    },
    {
      "epoch": 0.28338823529411766,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3119,
      "step": 12044
    },
    {
      "epoch": 0.28341176470588236,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1735,
      "step": 12045
    },
    {
      "epoch": 0.28343529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.2037,
      "step": 12046
    },
    {
      "epoch": 0.28345882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9997,
      "step": 12047
    },
    {
      "epoch": 0.28348235294117646,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8932,
      "step": 12048
    },
    {
      "epoch": 0.28350588235294116,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.6335,
      "step": 12049
    },
    {
      "epoch": 0.28352941176470586,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 12050
    },
    {
      "epoch": 0.28355294117647056,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 12051
    },
    {
      "epoch": 0.2835764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2678,
      "step": 12052
    },
    {
      "epoch": 0.2836,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8731,
      "step": 12053
    },
    {
      "epoch": 0.2836235294117647,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7146,
      "step": 12054
    },
    {
      "epoch": 0.2836470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8602,
      "step": 12055
    },
    {
      "epoch": 0.2836705882352941,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7953,
      "step": 12056
    },
    {
      "epoch": 0.2836941176470588,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9727,
      "step": 12057
    },
    {
      "epoch": 0.2837176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9712,
      "step": 12058
    },
    {
      "epoch": 0.2837411764705882,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9097,
      "step": 12059
    },
    {
      "epoch": 0.2837647058823529,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0435,
      "step": 12060
    },
    {
      "epoch": 0.2837882352941177,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 12061
    },
    {
      "epoch": 0.2838117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0624,
      "step": 12062
    },
    {
      "epoch": 0.2838352941176471,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1343,
      "step": 12063
    },
    {
      "epoch": 0.2838588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8375,
      "step": 12064
    },
    {
      "epoch": 0.2838823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.733,
      "step": 12065
    },
    {
      "epoch": 0.2839058823529412,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8805,
      "step": 12066
    },
    {
      "epoch": 0.2839294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1933,
      "step": 12067
    },
    {
      "epoch": 0.2839529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2418,
      "step": 12068
    },
    {
      "epoch": 0.2839764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0653,
      "step": 12069
    },
    {
      "epoch": 0.284,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.068,
      "step": 12070
    },
    {
      "epoch": 0.28402352941176473,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9846,
      "step": 12071
    },
    {
      "epoch": 0.28404705882352943,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.908,
      "step": 12072
    },
    {
      "epoch": 0.28407058823529413,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3292,
      "step": 12073
    },
    {
      "epoch": 0.28409411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0613,
      "step": 12074
    },
    {
      "epoch": 0.28411764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 12075
    },
    {
      "epoch": 0.28414117647058823,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.188,
      "step": 12076
    },
    {
      "epoch": 0.28416470588235293,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0095,
      "step": 12077
    },
    {
      "epoch": 0.28418823529411763,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2856,
      "step": 12078
    },
    {
      "epoch": 0.28421176470588233,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.6578,
      "step": 12079
    },
    {
      "epoch": 0.2842352941176471,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9046,
      "step": 12080
    },
    {
      "epoch": 0.2842588235294118,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9117,
      "step": 12081
    },
    {
      "epoch": 0.2842823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9802,
      "step": 12082
    },
    {
      "epoch": 0.2843058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8819,
      "step": 12083
    },
    {
      "epoch": 0.2843294117647059,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2141,
      "step": 12084
    },
    {
      "epoch": 0.2843529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9859,
      "step": 12085
    },
    {
      "epoch": 0.2843764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3251,
      "step": 12086
    },
    {
      "epoch": 0.2844,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1208,
      "step": 12087
    },
    {
      "epoch": 0.2844235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7107,
      "step": 12088
    },
    {
      "epoch": 0.2844470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1872,
      "step": 12089
    },
    {
      "epoch": 0.28447058823529414,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7772,
      "step": 12090
    },
    {
      "epoch": 0.28449411764705884,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9533,
      "step": 12091
    },
    {
      "epoch": 0.28451764705882354,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.7817,
      "step": 12092
    },
    {
      "epoch": 0.28454117647058824,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.995,
      "step": 12093
    },
    {
      "epoch": 0.28456470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0792,
      "step": 12094
    },
    {
      "epoch": 0.28458823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1191,
      "step": 12095
    },
    {
      "epoch": 0.28461176470588234,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.903,
      "step": 12096
    },
    {
      "epoch": 0.28463529411764704,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2053,
      "step": 12097
    },
    {
      "epoch": 0.28465882352941174,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 12098
    },
    {
      "epoch": 0.2846823529411765,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1691,
      "step": 12099
    },
    {
      "epoch": 0.2847058823529412,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7997,
      "step": 12100
    },
    {
      "epoch": 0.2847294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.058,
      "step": 12101
    },
    {
      "epoch": 0.2847529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9522,
      "step": 12102
    },
    {
      "epoch": 0.2847764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9314,
      "step": 12103
    },
    {
      "epoch": 0.2848,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0643,
      "step": 12104
    },
    {
      "epoch": 0.2848235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8493,
      "step": 12105
    },
    {
      "epoch": 0.2848470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.051,
      "step": 12106
    },
    {
      "epoch": 0.2848705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0539,
      "step": 12107
    },
    {
      "epoch": 0.2848941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1448,
      "step": 12108
    },
    {
      "epoch": 0.28491764705882355,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2064,
      "step": 12109
    },
    {
      "epoch": 0.28494117647058825,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8822,
      "step": 12110
    },
    {
      "epoch": 0.28496470588235295,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9775,
      "step": 12111
    },
    {
      "epoch": 0.28498823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1464,
      "step": 12112
    },
    {
      "epoch": 0.28501176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0064,
      "step": 12113
    },
    {
      "epoch": 0.28503529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7958,
      "step": 12114
    },
    {
      "epoch": 0.28505882352941175,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.355,
      "step": 12115
    },
    {
      "epoch": 0.28508235294117645,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8607,
      "step": 12116
    },
    {
      "epoch": 0.28510588235294115,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2417,
      "step": 12117
    },
    {
      "epoch": 0.2851294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9265,
      "step": 12118
    },
    {
      "epoch": 0.2851529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0574,
      "step": 12119
    },
    {
      "epoch": 0.2851764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0734,
      "step": 12120
    },
    {
      "epoch": 0.2852,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3955,
      "step": 12121
    },
    {
      "epoch": 0.2852235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0608,
      "step": 12122
    },
    {
      "epoch": 0.2852470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2022,
      "step": 12123
    },
    {
      "epoch": 0.2852705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1659,
      "step": 12124
    },
    {
      "epoch": 0.2852941176470588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7589,
      "step": 12125
    },
    {
      "epoch": 0.2853176470588235,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0835,
      "step": 12126
    },
    {
      "epoch": 0.2853411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 12127
    },
    {
      "epoch": 0.28536470588235296,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.0373,
      "step": 12128
    },
    {
      "epoch": 0.28538823529411766,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.9698,
      "step": 12129
    },
    {
      "epoch": 0.28541176470588236,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.2395,
      "step": 12130
    },
    {
      "epoch": 0.28543529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1019,
      "step": 12131
    },
    {
      "epoch": 0.28545882352941176,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.389,
      "step": 12132
    },
    {
      "epoch": 0.28548235294117646,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1126,
      "step": 12133
    },
    {
      "epoch": 0.28550588235294116,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0944,
      "step": 12134
    },
    {
      "epoch": 0.28552941176470586,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.015,
      "step": 12135
    },
    {
      "epoch": 0.28555294117647056,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0905,
      "step": 12136
    },
    {
      "epoch": 0.2855764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0576,
      "step": 12137
    },
    {
      "epoch": 0.2856,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.275,
      "step": 12138
    },
    {
      "epoch": 0.2856235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.1506,
      "step": 12139
    },
    {
      "epoch": 0.2856470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0756,
      "step": 12140
    },
    {
      "epoch": 0.2856705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1196,
      "step": 12141
    },
    {
      "epoch": 0.2856941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0661,
      "step": 12142
    },
    {
      "epoch": 0.2857176470588235,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9359,
      "step": 12143
    },
    {
      "epoch": 0.2857411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9272,
      "step": 12144
    },
    {
      "epoch": 0.2857647058823529,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9719,
      "step": 12145
    },
    {
      "epoch": 0.2857882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1084,
      "step": 12146
    },
    {
      "epoch": 0.2858117647058824,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9776,
      "step": 12147
    },
    {
      "epoch": 0.2858352941176471,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9273,
      "step": 12148
    },
    {
      "epoch": 0.2858588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2901,
      "step": 12149
    },
    {
      "epoch": 0.2858823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0413,
      "step": 12150
    },
    {
      "epoch": 0.2859058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2328,
      "step": 12151
    },
    {
      "epoch": 0.2859294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1718,
      "step": 12152
    },
    {
      "epoch": 0.2859529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2999,
      "step": 12153
    },
    {
      "epoch": 0.2859764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0591,
      "step": 12154
    },
    {
      "epoch": 0.286,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1034,
      "step": 12155
    },
    {
      "epoch": 0.28602352941176473,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8244,
      "step": 12156
    },
    {
      "epoch": 0.28604705882352943,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6039,
      "step": 12157
    },
    {
      "epoch": 0.28607058823529413,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9288,
      "step": 12158
    },
    {
      "epoch": 0.28609411764705883,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3678,
      "step": 12159
    },
    {
      "epoch": 0.28611764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0314,
      "step": 12160
    },
    {
      "epoch": 0.28614117647058823,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1125,
      "step": 12161
    },
    {
      "epoch": 0.28616470588235293,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 12162
    },
    {
      "epoch": 0.28618823529411763,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.8237,
      "step": 12163
    },
    {
      "epoch": 0.28621176470588233,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1966,
      "step": 12164
    },
    {
      "epoch": 0.28623529411764703,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0078,
      "step": 12165
    },
    {
      "epoch": 0.2862588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 12166
    },
    {
      "epoch": 0.2862823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0663,
      "step": 12167
    },
    {
      "epoch": 0.2863058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 12168
    },
    {
      "epoch": 0.2863294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9783,
      "step": 12169
    },
    {
      "epoch": 0.2863529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0896,
      "step": 12170
    },
    {
      "epoch": 0.2863764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 12171
    },
    {
      "epoch": 0.2864,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2332,
      "step": 12172
    },
    {
      "epoch": 0.2864235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.1209,
      "step": 12173
    },
    {
      "epoch": 0.2864470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1765,
      "step": 12174
    },
    {
      "epoch": 0.28647058823529414,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.6665,
      "step": 12175
    },
    {
      "epoch": 0.28649411764705884,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2078,
      "step": 12176
    },
    {
      "epoch": 0.28651764705882354,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1556,
      "step": 12177
    },
    {
      "epoch": 0.28654117647058824,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9443,
      "step": 12178
    },
    {
      "epoch": 0.28656470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9209,
      "step": 12179
    },
    {
      "epoch": 0.28658823529411764,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3804,
      "step": 12180
    },
    {
      "epoch": 0.28661176470588234,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8314,
      "step": 12181
    },
    {
      "epoch": 0.28663529411764704,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8995,
      "step": 12182
    },
    {
      "epoch": 0.28665882352941174,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1153,
      "step": 12183
    },
    {
      "epoch": 0.28668235294117644,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2098,
      "step": 12184
    },
    {
      "epoch": 0.2867058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1679,
      "step": 12185
    },
    {
      "epoch": 0.2867294117647059,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8668,
      "step": 12186
    },
    {
      "epoch": 0.2867529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1899,
      "step": 12187
    },
    {
      "epoch": 0.2867764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9385,
      "step": 12188
    },
    {
      "epoch": 0.2868,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9514,
      "step": 12189
    },
    {
      "epoch": 0.2868235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7637,
      "step": 12190
    },
    {
      "epoch": 0.2868470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1752,
      "step": 12191
    },
    {
      "epoch": 0.2868705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.2911,
      "step": 12192
    },
    {
      "epoch": 0.2868941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0863,
      "step": 12193
    },
    {
      "epoch": 0.28691764705882356,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1438,
      "step": 12194
    },
    {
      "epoch": 0.28694117647058826,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.6906,
      "step": 12195
    },
    {
      "epoch": 0.28696470588235296,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6674,
      "step": 12196
    },
    {
      "epoch": 0.28698823529411766,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7751,
      "step": 12197
    },
    {
      "epoch": 0.28701176470588236,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1305,
      "step": 12198
    },
    {
      "epoch": 0.28703529411764706,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.02,
      "loss": 0.7515,
      "step": 12199
    },
    {
      "epoch": 0.28705882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9199,
      "step": 12200
    },
    {
      "epoch": 0.28708235294117646,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.9068,
      "step": 12201
    },
    {
      "epoch": 0.28710588235294116,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0076,
      "step": 12202
    },
    {
      "epoch": 0.28712941176470586,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2412,
      "step": 12203
    },
    {
      "epoch": 0.2871529411764706,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.5998,
      "step": 12204
    },
    {
      "epoch": 0.2871764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8519,
      "step": 12205
    },
    {
      "epoch": 0.2872,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9379,
      "step": 12206
    },
    {
      "epoch": 0.2872235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0573,
      "step": 12207
    },
    {
      "epoch": 0.2872470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0903,
      "step": 12208
    },
    {
      "epoch": 0.2872705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.102,
      "step": 12209
    },
    {
      "epoch": 0.2872941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9912,
      "step": 12210
    },
    {
      "epoch": 0.2873176470588235,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2587,
      "step": 12211
    },
    {
      "epoch": 0.2873411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0103,
      "step": 12212
    },
    {
      "epoch": 0.28736470588235297,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0511,
      "step": 12213
    },
    {
      "epoch": 0.28738823529411767,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9317,
      "step": 12214
    },
    {
      "epoch": 0.28741176470588237,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9969,
      "step": 12215
    },
    {
      "epoch": 0.28743529411764707,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9876,
      "step": 12216
    },
    {
      "epoch": 0.28745882352941177,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0313,
      "step": 12217
    },
    {
      "epoch": 0.28748235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9555,
      "step": 12218
    },
    {
      "epoch": 0.28750588235294117,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0354,
      "step": 12219
    },
    {
      "epoch": 0.28752941176470587,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 12220
    },
    {
      "epoch": 0.28755294117647057,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0376,
      "step": 12221
    },
    {
      "epoch": 0.28757647058823527,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9967,
      "step": 12222
    },
    {
      "epoch": 0.2876,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8502,
      "step": 12223
    },
    {
      "epoch": 0.2876235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1719,
      "step": 12224
    },
    {
      "epoch": 0.2876470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1521,
      "step": 12225
    },
    {
      "epoch": 0.2876705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9712,
      "step": 12226
    },
    {
      "epoch": 0.2876941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9765,
      "step": 12227
    },
    {
      "epoch": 0.2877176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0709,
      "step": 12228
    },
    {
      "epoch": 0.2877411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0047,
      "step": 12229
    },
    {
      "epoch": 0.2877647058823529,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1227,
      "step": 12230
    },
    {
      "epoch": 0.2877882352941176,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7815,
      "step": 12231
    },
    {
      "epoch": 0.2878117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9356,
      "step": 12232
    },
    {
      "epoch": 0.2878352941176471,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.202,
      "step": 12233
    },
    {
      "epoch": 0.2878588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.052,
      "step": 12234
    },
    {
      "epoch": 0.2878823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1946,
      "step": 12235
    },
    {
      "epoch": 0.2879058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2284,
      "step": 12236
    },
    {
      "epoch": 0.2879294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2789,
      "step": 12237
    },
    {
      "epoch": 0.2879529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0884,
      "step": 12238
    },
    {
      "epoch": 0.2879764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.007,
      "step": 12239
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8075,
      "step": 12240
    },
    {
      "epoch": 0.2880235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.144,
      "step": 12241
    },
    {
      "epoch": 0.28804705882352943,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2526,
      "step": 12242
    },
    {
      "epoch": 0.28807058823529413,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0968,
      "step": 12243
    },
    {
      "epoch": 0.28809411764705883,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1899,
      "step": 12244
    },
    {
      "epoch": 0.28811764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8568,
      "step": 12245
    },
    {
      "epoch": 0.28814117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0498,
      "step": 12246
    },
    {
      "epoch": 0.28816470588235293,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7501,
      "step": 12247
    },
    {
      "epoch": 0.28818823529411763,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0726,
      "step": 12248
    },
    {
      "epoch": 0.28821176470588233,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2196,
      "step": 12249
    },
    {
      "epoch": 0.28823529411764703,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0276,
      "step": 12250
    },
    {
      "epoch": 0.2882588235294118,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9507,
      "step": 12251
    },
    {
      "epoch": 0.2882823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2826,
      "step": 12252
    },
    {
      "epoch": 0.2883058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0263,
      "step": 12253
    },
    {
      "epoch": 0.2883294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1363,
      "step": 12254
    },
    {
      "epoch": 0.2883529411764706,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.411,
      "step": 12255
    },
    {
      "epoch": 0.2883764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.385,
      "step": 12256
    },
    {
      "epoch": 0.2884,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1202,
      "step": 12257
    },
    {
      "epoch": 0.2884235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9302,
      "step": 12258
    },
    {
      "epoch": 0.2884470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0134,
      "step": 12259
    },
    {
      "epoch": 0.28847058823529415,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2365,
      "step": 12260
    },
    {
      "epoch": 0.28849411764705885,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3643,
      "step": 12261
    },
    {
      "epoch": 0.28851764705882355,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0149,
      "step": 12262
    },
    {
      "epoch": 0.28854117647058825,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.4163,
      "step": 12263
    },
    {
      "epoch": 0.28856470588235295,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9031,
      "step": 12264
    },
    {
      "epoch": 0.28858823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.182,
      "step": 12265
    },
    {
      "epoch": 0.28861176470588235,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8118,
      "step": 12266
    },
    {
      "epoch": 0.28863529411764705,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9363,
      "step": 12267
    },
    {
      "epoch": 0.28865882352941175,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0042,
      "step": 12268
    },
    {
      "epoch": 0.28868235294117645,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1282,
      "step": 12269
    },
    {
      "epoch": 0.2887058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0894,
      "step": 12270
    },
    {
      "epoch": 0.2887294117647059,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 12271
    },
    {
      "epoch": 0.2887529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 12272
    },
    {
      "epoch": 0.2887764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9805,
      "step": 12273
    },
    {
      "epoch": 0.2888,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9588,
      "step": 12274
    },
    {
      "epoch": 0.2888235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7407,
      "step": 12275
    },
    {
      "epoch": 0.2888470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0759,
      "step": 12276
    },
    {
      "epoch": 0.2888705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 12277
    },
    {
      "epoch": 0.2888941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1521,
      "step": 12278
    },
    {
      "epoch": 0.28891764705882356,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8025,
      "step": 12279
    },
    {
      "epoch": 0.28894117647058826,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0983,
      "step": 12280
    },
    {
      "epoch": 0.28896470588235296,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8017,
      "step": 12281
    },
    {
      "epoch": 0.28898823529411766,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.8168,
      "step": 12282
    },
    {
      "epoch": 0.28901176470588236,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1679,
      "step": 12283
    },
    {
      "epoch": 0.28903529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9229,
      "step": 12284
    },
    {
      "epoch": 0.28905882352941176,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1193,
      "step": 12285
    },
    {
      "epoch": 0.28908235294117646,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0152,
      "step": 12286
    },
    {
      "epoch": 0.28910588235294116,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8738,
      "step": 12287
    },
    {
      "epoch": 0.28912941176470586,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2589,
      "step": 12288
    },
    {
      "epoch": 0.2891529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1646,
      "step": 12289
    },
    {
      "epoch": 0.2891764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1247,
      "step": 12290
    },
    {
      "epoch": 0.2892,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.751,
      "step": 12291
    },
    {
      "epoch": 0.2892235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8917,
      "step": 12292
    },
    {
      "epoch": 0.2892470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8573,
      "step": 12293
    },
    {
      "epoch": 0.2892705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6787,
      "step": 12294
    },
    {
      "epoch": 0.2892941176470588,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.723,
      "step": 12295
    },
    {
      "epoch": 0.2893176470588235,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2165,
      "step": 12296
    },
    {
      "epoch": 0.2893411764705882,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 12297
    },
    {
      "epoch": 0.28936470588235297,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1854,
      "step": 12298
    },
    {
      "epoch": 0.28938823529411767,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9277,
      "step": 12299
    },
    {
      "epoch": 0.28941176470588237,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9749,
      "step": 12300
    },
    {
      "epoch": 0.28943529411764707,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2843,
      "step": 12301
    },
    {
      "epoch": 0.28945882352941177,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 12302
    },
    {
      "epoch": 0.28948235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0964,
      "step": 12303
    },
    {
      "epoch": 0.28950588235294117,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1199,
      "step": 12304
    },
    {
      "epoch": 0.28952941176470587,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.1202,
      "step": 12305
    },
    {
      "epoch": 0.28955294117647057,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8188,
      "step": 12306
    },
    {
      "epoch": 0.28957647058823527,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1758,
      "step": 12307
    },
    {
      "epoch": 0.2896,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0286,
      "step": 12308
    },
    {
      "epoch": 0.2896235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2065,
      "step": 12309
    },
    {
      "epoch": 0.2896470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9928,
      "step": 12310
    },
    {
      "epoch": 0.2896705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8696,
      "step": 12311
    },
    {
      "epoch": 0.2896941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1406,
      "step": 12312
    },
    {
      "epoch": 0.2897176470588235,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7713,
      "step": 12313
    },
    {
      "epoch": 0.2897411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1815,
      "step": 12314
    },
    {
      "epoch": 0.2897647058823529,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 12315
    },
    {
      "epoch": 0.2897882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.5819,
      "step": 12316
    },
    {
      "epoch": 0.2898117647058824,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1405,
      "step": 12317
    },
    {
      "epoch": 0.2898352941176471,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0634,
      "step": 12318
    },
    {
      "epoch": 0.2898588235294118,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2529,
      "step": 12319
    },
    {
      "epoch": 0.2898823529411765,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3229,
      "step": 12320
    },
    {
      "epoch": 0.2899058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2732,
      "step": 12321
    },
    {
      "epoch": 0.2899294117647059,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3675,
      "step": 12322
    },
    {
      "epoch": 0.2899529411764706,
      "grad_norm": 0.4140625,
      "learning_rate": 0.02,
      "loss": 0.8223,
      "step": 12323
    },
    {
      "epoch": 0.2899764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8112,
      "step": 12324
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0177,
      "step": 12325
    },
    {
      "epoch": 0.2900235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9814,
      "step": 12326
    },
    {
      "epoch": 0.29004705882352944,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2121,
      "step": 12327
    },
    {
      "epoch": 0.29007058823529414,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.3721,
      "step": 12328
    },
    {
      "epoch": 0.29009411764705884,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9812,
      "step": 12329
    },
    {
      "epoch": 0.29011764705882354,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6315,
      "step": 12330
    },
    {
      "epoch": 0.29014117647058824,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2983,
      "step": 12331
    },
    {
      "epoch": 0.29016470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3071,
      "step": 12332
    },
    {
      "epoch": 0.29018823529411764,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0653,
      "step": 12333
    },
    {
      "epoch": 0.29021176470588234,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1906,
      "step": 12334
    },
    {
      "epoch": 0.29023529411764704,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9242,
      "step": 12335
    },
    {
      "epoch": 0.2902588235294118,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1759,
      "step": 12336
    },
    {
      "epoch": 0.2902823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.019,
      "step": 12337
    },
    {
      "epoch": 0.2903058823529412,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1033,
      "step": 12338
    },
    {
      "epoch": 0.2903294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8698,
      "step": 12339
    },
    {
      "epoch": 0.2903529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.2124,
      "step": 12340
    },
    {
      "epoch": 0.2903764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9436,
      "step": 12341
    },
    {
      "epoch": 0.2904,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0893,
      "step": 12342
    },
    {
      "epoch": 0.2904235294117647,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2293,
      "step": 12343
    },
    {
      "epoch": 0.2904470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 12344
    },
    {
      "epoch": 0.2904705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9542,
      "step": 12345
    },
    {
      "epoch": 0.29049411764705885,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2059,
      "step": 12346
    },
    {
      "epoch": 0.29051764705882355,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 12347
    },
    {
      "epoch": 0.29054117647058825,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0397,
      "step": 12348
    },
    {
      "epoch": 0.29056470588235295,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9964,
      "step": 12349
    },
    {
      "epoch": 0.29058823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8912,
      "step": 12350
    },
    {
      "epoch": 0.29061176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8639,
      "step": 12351
    },
    {
      "epoch": 0.29063529411764705,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9681,
      "step": 12352
    },
    {
      "epoch": 0.29065882352941175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0172,
      "step": 12353
    },
    {
      "epoch": 0.29068235294117645,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2143,
      "step": 12354
    },
    {
      "epoch": 0.2907058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.914,
      "step": 12355
    },
    {
      "epoch": 0.2907294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7516,
      "step": 12356
    },
    {
      "epoch": 0.2907529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2038,
      "step": 12357
    },
    {
      "epoch": 0.2907764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1837,
      "step": 12358
    },
    {
      "epoch": 0.2908,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.6541,
      "step": 12359
    },
    {
      "epoch": 0.2908235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7577,
      "step": 12360
    },
    {
      "epoch": 0.2908470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0442,
      "step": 12361
    },
    {
      "epoch": 0.2908705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 12362
    },
    {
      "epoch": 0.2908941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9192,
      "step": 12363
    },
    {
      "epoch": 0.2909176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9625,
      "step": 12364
    },
    {
      "epoch": 0.29094117647058826,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0295,
      "step": 12365
    },
    {
      "epoch": 0.29096470588235296,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0543,
      "step": 12366
    },
    {
      "epoch": 0.29098823529411766,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9326,
      "step": 12367
    },
    {
      "epoch": 0.29101176470588236,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0862,
      "step": 12368
    },
    {
      "epoch": 0.29103529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7681,
      "step": 12369
    },
    {
      "epoch": 0.29105882352941176,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2069,
      "step": 12370
    },
    {
      "epoch": 0.29108235294117646,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1731,
      "step": 12371
    },
    {
      "epoch": 0.29110588235294116,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2388,
      "step": 12372
    },
    {
      "epoch": 0.29112941176470586,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.106,
      "step": 12373
    },
    {
      "epoch": 0.2911529411764706,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.736,
      "step": 12374
    },
    {
      "epoch": 0.2911764705882353,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8983,
      "step": 12375
    },
    {
      "epoch": 0.2912,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.945,
      "step": 12376
    },
    {
      "epoch": 0.2912235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8128,
      "step": 12377
    },
    {
      "epoch": 0.2912470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3458,
      "step": 12378
    },
    {
      "epoch": 0.2912705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8808,
      "step": 12379
    },
    {
      "epoch": 0.2912941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8283,
      "step": 12380
    },
    {
      "epoch": 0.2913176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 12381
    },
    {
      "epoch": 0.2913411764705882,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.6652,
      "step": 12382
    },
    {
      "epoch": 0.2913647058823529,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0809,
      "step": 12383
    },
    {
      "epoch": 0.29138823529411767,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2213,
      "step": 12384
    },
    {
      "epoch": 0.29141176470588237,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0493,
      "step": 12385
    },
    {
      "epoch": 0.29143529411764707,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.184,
      "step": 12386
    },
    {
      "epoch": 0.29145882352941177,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2059,
      "step": 12387
    },
    {
      "epoch": 0.29148235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8253,
      "step": 12388
    },
    {
      "epoch": 0.29150588235294117,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1115,
      "step": 12389
    },
    {
      "epoch": 0.29152941176470587,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7953,
      "step": 12390
    },
    {
      "epoch": 0.29155294117647057,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1282,
      "step": 12391
    },
    {
      "epoch": 0.29157647058823527,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1222,
      "step": 12392
    },
    {
      "epoch": 0.2916,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2732,
      "step": 12393
    },
    {
      "epoch": 0.2916235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7489,
      "step": 12394
    },
    {
      "epoch": 0.2916470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3503,
      "step": 12395
    },
    {
      "epoch": 0.2916705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0436,
      "step": 12396
    },
    {
      "epoch": 0.2916941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8266,
      "step": 12397
    },
    {
      "epoch": 0.2917176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 12398
    },
    {
      "epoch": 0.2917411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9719,
      "step": 12399
    },
    {
      "epoch": 0.2917647058823529,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2353,
      "step": 12400
    },
    {
      "epoch": 0.2917882352941176,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0574,
      "step": 12401
    },
    {
      "epoch": 0.2918117647058823,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 12402
    },
    {
      "epoch": 0.2918352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.027,
      "step": 12403
    },
    {
      "epoch": 0.2918588235294118,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.225,
      "step": 12404
    },
    {
      "epoch": 0.2918823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.024,
      "step": 12405
    },
    {
      "epoch": 0.2919058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.036,
      "step": 12406
    },
    {
      "epoch": 0.2919294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9963,
      "step": 12407
    },
    {
      "epoch": 0.2919529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9053,
      "step": 12408
    },
    {
      "epoch": 0.2919764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.219,
      "step": 12409
    },
    {
      "epoch": 0.292,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1839,
      "step": 12410
    },
    {
      "epoch": 0.2920235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9203,
      "step": 12411
    },
    {
      "epoch": 0.29204705882352944,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0957,
      "step": 12412
    },
    {
      "epoch": 0.29207058823529414,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3409,
      "step": 12413
    },
    {
      "epoch": 0.29209411764705884,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2408,
      "step": 12414
    },
    {
      "epoch": 0.29211764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1802,
      "step": 12415
    },
    {
      "epoch": 0.29214117647058824,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 12416
    },
    {
      "epoch": 0.29216470588235294,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7679,
      "step": 12417
    },
    {
      "epoch": 0.29218823529411764,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2676,
      "step": 12418
    },
    {
      "epoch": 0.29221176470588234,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1066,
      "step": 12419
    },
    {
      "epoch": 0.29223529411764704,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.6794,
      "step": 12420
    },
    {
      "epoch": 0.29225882352941174,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1771,
      "step": 12421
    },
    {
      "epoch": 0.2922823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1185,
      "step": 12422
    },
    {
      "epoch": 0.2923058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0672,
      "step": 12423
    },
    {
      "epoch": 0.2923294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0308,
      "step": 12424
    },
    {
      "epoch": 0.2923529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2716,
      "step": 12425
    },
    {
      "epoch": 0.2923764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1744,
      "step": 12426
    },
    {
      "epoch": 0.2924,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 12427
    },
    {
      "epoch": 0.2924235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.893,
      "step": 12428
    },
    {
      "epoch": 0.2924470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0107,
      "step": 12429
    },
    {
      "epoch": 0.2924705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3133,
      "step": 12430
    },
    {
      "epoch": 0.29249411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9277,
      "step": 12431
    },
    {
      "epoch": 0.29251764705882355,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9915,
      "step": 12432
    },
    {
      "epoch": 0.29254117647058825,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.125,
      "step": 12433
    },
    {
      "epoch": 0.29256470588235295,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2128,
      "step": 12434
    },
    {
      "epoch": 0.29258823529411765,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9999,
      "step": 12435
    },
    {
      "epoch": 0.29261176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7606,
      "step": 12436
    },
    {
      "epoch": 0.29263529411764705,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1841,
      "step": 12437
    },
    {
      "epoch": 0.29265882352941175,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8709,
      "step": 12438
    },
    {
      "epoch": 0.29268235294117645,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.215,
      "step": 12439
    },
    {
      "epoch": 0.29270588235294115,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.276,
      "step": 12440
    },
    {
      "epoch": 0.2927294117647059,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8173,
      "step": 12441
    },
    {
      "epoch": 0.2927529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9886,
      "step": 12442
    },
    {
      "epoch": 0.2927764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1619,
      "step": 12443
    },
    {
      "epoch": 0.2928,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0763,
      "step": 12444
    },
    {
      "epoch": 0.2928235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2196,
      "step": 12445
    },
    {
      "epoch": 0.2928470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0104,
      "step": 12446
    },
    {
      "epoch": 0.2928705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1378,
      "step": 12447
    },
    {
      "epoch": 0.2928941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8561,
      "step": 12448
    },
    {
      "epoch": 0.2929176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9845,
      "step": 12449
    },
    {
      "epoch": 0.29294117647058826,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.6845,
      "step": 12450
    },
    {
      "epoch": 0.29296470588235296,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2529,
      "step": 12451
    },
    {
      "epoch": 0.29298823529411766,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2175,
      "step": 12452
    },
    {
      "epoch": 0.29301176470588236,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9212,
      "step": 12453
    },
    {
      "epoch": 0.29303529411764706,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9746,
      "step": 12454
    },
    {
      "epoch": 0.29305882352941176,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.8787,
      "step": 12455
    },
    {
      "epoch": 0.29308235294117646,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0095,
      "step": 12456
    },
    {
      "epoch": 0.29310588235294116,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0957,
      "step": 12457
    },
    {
      "epoch": 0.29312941176470586,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9589,
      "step": 12458
    },
    {
      "epoch": 0.29315294117647056,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.114,
      "step": 12459
    },
    {
      "epoch": 0.2931764705882353,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9325,
      "step": 12460
    },
    {
      "epoch": 0.2932,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0886,
      "step": 12461
    },
    {
      "epoch": 0.2932235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9897,
      "step": 12462
    },
    {
      "epoch": 0.2932470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0298,
      "step": 12463
    },
    {
      "epoch": 0.2932705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9277,
      "step": 12464
    },
    {
      "epoch": 0.2932941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9677,
      "step": 12465
    },
    {
      "epoch": 0.2933176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8544,
      "step": 12466
    },
    {
      "epoch": 0.2933411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3074,
      "step": 12467
    },
    {
      "epoch": 0.2933647058823529,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9203,
      "step": 12468
    },
    {
      "epoch": 0.29338823529411767,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3309,
      "step": 12469
    },
    {
      "epoch": 0.29341176470588237,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9765,
      "step": 12470
    },
    {
      "epoch": 0.29343529411764707,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0373,
      "step": 12471
    },
    {
      "epoch": 0.29345882352941177,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0057,
      "step": 12472
    },
    {
      "epoch": 0.29348235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.908,
      "step": 12473
    },
    {
      "epoch": 0.29350588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8793,
      "step": 12474
    },
    {
      "epoch": 0.29352941176470587,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0447,
      "step": 12475
    },
    {
      "epoch": 0.29355294117647057,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7194,
      "step": 12476
    },
    {
      "epoch": 0.29357647058823527,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0002,
      "step": 12477
    },
    {
      "epoch": 0.2936,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.7503,
      "step": 12478
    },
    {
      "epoch": 0.2936235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9589,
      "step": 12479
    },
    {
      "epoch": 0.2936470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.17,
      "step": 12480
    },
    {
      "epoch": 0.2936705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0197,
      "step": 12481
    },
    {
      "epoch": 0.2936941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0089,
      "step": 12482
    },
    {
      "epoch": 0.2937176470588235,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 12483
    },
    {
      "epoch": 0.29374117647058823,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8949,
      "step": 12484
    },
    {
      "epoch": 0.29376470588235293,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2632,
      "step": 12485
    },
    {
      "epoch": 0.29378823529411763,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8828,
      "step": 12486
    },
    {
      "epoch": 0.29381176470588233,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0163,
      "step": 12487
    },
    {
      "epoch": 0.2938352941176471,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0153,
      "step": 12488
    },
    {
      "epoch": 0.2938588235294118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.3392,
      "step": 12489
    },
    {
      "epoch": 0.2938823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3271,
      "step": 12490
    },
    {
      "epoch": 0.2939058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1211,
      "step": 12491
    },
    {
      "epoch": 0.2939294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1735,
      "step": 12492
    },
    {
      "epoch": 0.2939529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0934,
      "step": 12493
    },
    {
      "epoch": 0.2939764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8701,
      "step": 12494
    },
    {
      "epoch": 0.294,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8031,
      "step": 12495
    },
    {
      "epoch": 0.2940235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2926,
      "step": 12496
    },
    {
      "epoch": 0.29404705882352944,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7837,
      "step": 12497
    },
    {
      "epoch": 0.29407058823529414,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1248,
      "step": 12498
    },
    {
      "epoch": 0.29409411764705884,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.782,
      "step": 12499
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1619,
      "step": 12500
    },
    {
      "epoch": 0.29414117647058824,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2076,
      "step": 12501
    },
    {
      "epoch": 0.29416470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1421,
      "step": 12502
    },
    {
      "epoch": 0.29418823529411764,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0451,
      "step": 12503
    },
    {
      "epoch": 0.29421176470588234,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2051,
      "step": 12504
    },
    {
      "epoch": 0.29423529411764704,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7692,
      "step": 12505
    },
    {
      "epoch": 0.29425882352941174,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 12506
    },
    {
      "epoch": 0.2942823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8525,
      "step": 12507
    },
    {
      "epoch": 0.2943058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1413,
      "step": 12508
    },
    {
      "epoch": 0.2943294117647059,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9015,
      "step": 12509
    },
    {
      "epoch": 0.2943529411764706,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8604,
      "step": 12510
    },
    {
      "epoch": 0.2943764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.927,
      "step": 12511
    },
    {
      "epoch": 0.2944,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9927,
      "step": 12512
    },
    {
      "epoch": 0.2944235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 12513
    },
    {
      "epoch": 0.2944470588235294,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3205,
      "step": 12514
    },
    {
      "epoch": 0.2944705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1035,
      "step": 12515
    },
    {
      "epoch": 0.29449411764705885,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8411,
      "step": 12516
    },
    {
      "epoch": 0.29451764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2409,
      "step": 12517
    },
    {
      "epoch": 0.29454117647058825,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1613,
      "step": 12518
    },
    {
      "epoch": 0.29456470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9361,
      "step": 12519
    },
    {
      "epoch": 0.29458823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1056,
      "step": 12520
    },
    {
      "epoch": 0.29461176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9381,
      "step": 12521
    },
    {
      "epoch": 0.29463529411764705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9928,
      "step": 12522
    },
    {
      "epoch": 0.29465882352941175,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7759,
      "step": 12523
    },
    {
      "epoch": 0.29468235294117645,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0891,
      "step": 12524
    },
    {
      "epoch": 0.29470588235294115,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2129,
      "step": 12525
    },
    {
      "epoch": 0.2947294117647059,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1682,
      "step": 12526
    },
    {
      "epoch": 0.2947529411764706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.3902,
      "step": 12527
    },
    {
      "epoch": 0.2947764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1595,
      "step": 12528
    },
    {
      "epoch": 0.2948,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3675,
      "step": 12529
    },
    {
      "epoch": 0.2948235294117647,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8112,
      "step": 12530
    },
    {
      "epoch": 0.2948470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2527,
      "step": 12531
    },
    {
      "epoch": 0.2948705882352941,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.909,
      "step": 12532
    },
    {
      "epoch": 0.2948941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1803,
      "step": 12533
    },
    {
      "epoch": 0.2949176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0811,
      "step": 12534
    },
    {
      "epoch": 0.29494117647058826,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.02,
      "loss": 0.7381,
      "step": 12535
    },
    {
      "epoch": 0.29496470588235296,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0036,
      "step": 12536
    },
    {
      "epoch": 0.29498823529411766,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0457,
      "step": 12537
    },
    {
      "epoch": 0.29501176470588236,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8259,
      "step": 12538
    },
    {
      "epoch": 0.29503529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9997,
      "step": 12539
    },
    {
      "epoch": 0.29505882352941176,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8812,
      "step": 12540
    },
    {
      "epoch": 0.29508235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1708,
      "step": 12541
    },
    {
      "epoch": 0.29510588235294116,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7391,
      "step": 12542
    },
    {
      "epoch": 0.29512941176470586,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2958,
      "step": 12543
    },
    {
      "epoch": 0.29515294117647056,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.157,
      "step": 12544
    },
    {
      "epoch": 0.2951764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1781,
      "step": 12545
    },
    {
      "epoch": 0.2952,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9456,
      "step": 12546
    },
    {
      "epoch": 0.2952235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0076,
      "step": 12547
    },
    {
      "epoch": 0.2952470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2092,
      "step": 12548
    },
    {
      "epoch": 0.2952705882352941,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0302,
      "step": 12549
    },
    {
      "epoch": 0.2952941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0603,
      "step": 12550
    },
    {
      "epoch": 0.2953176470588235,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.757,
      "step": 12551
    },
    {
      "epoch": 0.2953411764705882,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0436,
      "step": 12552
    },
    {
      "epoch": 0.2953647058823529,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1749,
      "step": 12553
    },
    {
      "epoch": 0.2953882352941177,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2222,
      "step": 12554
    },
    {
      "epoch": 0.2954117647058824,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0102,
      "step": 12555
    },
    {
      "epoch": 0.2954352941176471,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8893,
      "step": 12556
    },
    {
      "epoch": 0.2954588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0778,
      "step": 12557
    },
    {
      "epoch": 0.2954823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0435,
      "step": 12558
    },
    {
      "epoch": 0.2955058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0829,
      "step": 12559
    },
    {
      "epoch": 0.2955294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.6628,
      "step": 12560
    },
    {
      "epoch": 0.2955529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9373,
      "step": 12561
    },
    {
      "epoch": 0.2955764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7515,
      "step": 12562
    },
    {
      "epoch": 0.2956,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7887,
      "step": 12563
    },
    {
      "epoch": 0.29562352941176473,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 12564
    },
    {
      "epoch": 0.29564705882352943,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.3598,
      "step": 12565
    },
    {
      "epoch": 0.29567058823529413,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0831,
      "step": 12566
    },
    {
      "epoch": 0.29569411764705883,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 12567
    },
    {
      "epoch": 0.29571764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1067,
      "step": 12568
    },
    {
      "epoch": 0.29574117647058823,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.311,
      "step": 12569
    },
    {
      "epoch": 0.29576470588235293,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.24,
      "step": 12570
    },
    {
      "epoch": 0.29578823529411763,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9348,
      "step": 12571
    },
    {
      "epoch": 0.29581176470588233,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7675,
      "step": 12572
    },
    {
      "epoch": 0.2958352941176471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3119,
      "step": 12573
    },
    {
      "epoch": 0.2958588235294118,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.4373,
      "step": 12574
    },
    {
      "epoch": 0.2958823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1025,
      "step": 12575
    },
    {
      "epoch": 0.2959058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1875,
      "step": 12576
    },
    {
      "epoch": 0.2959294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0142,
      "step": 12577
    },
    {
      "epoch": 0.2959529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9854,
      "step": 12578
    },
    {
      "epoch": 0.2959764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2403,
      "step": 12579
    },
    {
      "epoch": 0.296,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8116,
      "step": 12580
    },
    {
      "epoch": 0.2960235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0969,
      "step": 12581
    },
    {
      "epoch": 0.2960470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9752,
      "step": 12582
    },
    {
      "epoch": 0.29607058823529414,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1015,
      "step": 12583
    },
    {
      "epoch": 0.29609411764705884,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1405,
      "step": 12584
    },
    {
      "epoch": 0.29611764705882354,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1304,
      "step": 12585
    },
    {
      "epoch": 0.29614117647058824,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8713,
      "step": 12586
    },
    {
      "epoch": 0.29616470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2149,
      "step": 12587
    },
    {
      "epoch": 0.29618823529411764,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.7891,
      "step": 12588
    },
    {
      "epoch": 0.29621176470588234,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2736,
      "step": 12589
    },
    {
      "epoch": 0.29623529411764704,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1291,
      "step": 12590
    },
    {
      "epoch": 0.29625882352941174,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0859,
      "step": 12591
    },
    {
      "epoch": 0.2962823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0277,
      "step": 12592
    },
    {
      "epoch": 0.2963058823529412,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3273,
      "step": 12593
    },
    {
      "epoch": 0.2963294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.1316,
      "step": 12594
    },
    {
      "epoch": 0.2963529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9965,
      "step": 12595
    },
    {
      "epoch": 0.2963764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0847,
      "step": 12596
    },
    {
      "epoch": 0.2964,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9692,
      "step": 12597
    },
    {
      "epoch": 0.2964235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0681,
      "step": 12598
    },
    {
      "epoch": 0.2964470588235294,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.3652,
      "step": 12599
    },
    {
      "epoch": 0.2964705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2209,
      "step": 12600
    },
    {
      "epoch": 0.2964941176470588,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3104,
      "step": 12601
    },
    {
      "epoch": 0.29651764705882355,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1341,
      "step": 12602
    },
    {
      "epoch": 0.29654117647058825,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.7885,
      "step": 12603
    },
    {
      "epoch": 0.29656470588235295,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2884,
      "step": 12604
    },
    {
      "epoch": 0.29658823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3417,
      "step": 12605
    },
    {
      "epoch": 0.29661176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1411,
      "step": 12606
    },
    {
      "epoch": 0.29663529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9866,
      "step": 12607
    },
    {
      "epoch": 0.29665882352941175,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1413,
      "step": 12608
    },
    {
      "epoch": 0.29668235294117645,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9627,
      "step": 12609
    },
    {
      "epoch": 0.29670588235294115,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7635,
      "step": 12610
    },
    {
      "epoch": 0.2967294117647059,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.7916,
      "step": 12611
    },
    {
      "epoch": 0.2967529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2338,
      "step": 12612
    },
    {
      "epoch": 0.2967764705882353,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3646,
      "step": 12613
    },
    {
      "epoch": 0.2968,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9146,
      "step": 12614
    },
    {
      "epoch": 0.2968235294117647,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.6586,
      "step": 12615
    },
    {
      "epoch": 0.2968470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9319,
      "step": 12616
    },
    {
      "epoch": 0.2968705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1382,
      "step": 12617
    },
    {
      "epoch": 0.2968941176470588,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 0.9474,
      "step": 12618
    },
    {
      "epoch": 0.2969176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9052,
      "step": 12619
    },
    {
      "epoch": 0.2969411764705882,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4343,
      "step": 12620
    },
    {
      "epoch": 0.29696470588235296,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8215,
      "step": 12621
    },
    {
      "epoch": 0.29698823529411766,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8909,
      "step": 12622
    },
    {
      "epoch": 0.29701176470588236,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3505,
      "step": 12623
    },
    {
      "epoch": 0.29703529411764706,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.4423,
      "step": 12624
    },
    {
      "epoch": 0.29705882352941176,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 12625
    },
    {
      "epoch": 0.29708235294117646,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0411,
      "step": 12626
    },
    {
      "epoch": 0.29710588235294116,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 12627
    },
    {
      "epoch": 0.29712941176470586,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9672,
      "step": 12628
    },
    {
      "epoch": 0.29715294117647056,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.33,
      "step": 12629
    },
    {
      "epoch": 0.2971764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0739,
      "step": 12630
    },
    {
      "epoch": 0.2972,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9746,
      "step": 12631
    },
    {
      "epoch": 0.2972235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1342,
      "step": 12632
    },
    {
      "epoch": 0.2972470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6695,
      "step": 12633
    },
    {
      "epoch": 0.2972705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1157,
      "step": 12634
    },
    {
      "epoch": 0.2972941176470588,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.4458,
      "step": 12635
    },
    {
      "epoch": 0.2973176470588235,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.4219,
      "step": 12636
    },
    {
      "epoch": 0.2973411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0537,
      "step": 12637
    },
    {
      "epoch": 0.2973647058823529,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0448,
      "step": 12638
    },
    {
      "epoch": 0.2973882352941176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7173,
      "step": 12639
    },
    {
      "epoch": 0.2974117647058824,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 12640
    },
    {
      "epoch": 0.2974352941176471,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7361,
      "step": 12641
    },
    {
      "epoch": 0.2974588235294118,
      "grad_norm": 0.240234375,
      "learning_rate": 0.02,
      "loss": 0.6219,
      "step": 12642
    },
    {
      "epoch": 0.2974823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2042,
      "step": 12643
    },
    {
      "epoch": 0.2975058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0051,
      "step": 12644
    },
    {
      "epoch": 0.2975294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1664,
      "step": 12645
    },
    {
      "epoch": 0.2975529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0204,
      "step": 12646
    },
    {
      "epoch": 0.2975764705882353,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.8307,
      "step": 12647
    },
    {
      "epoch": 0.2976,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0796,
      "step": 12648
    },
    {
      "epoch": 0.29762352941176473,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9855,
      "step": 12649
    },
    {
      "epoch": 0.29764705882352943,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8695,
      "step": 12650
    },
    {
      "epoch": 0.29767058823529413,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0745,
      "step": 12651
    },
    {
      "epoch": 0.29769411764705883,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 12652
    },
    {
      "epoch": 0.29771764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 12653
    },
    {
      "epoch": 0.29774117647058823,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8405,
      "step": 12654
    },
    {
      "epoch": 0.29776470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8545,
      "step": 12655
    },
    {
      "epoch": 0.29778823529411763,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9316,
      "step": 12656
    },
    {
      "epoch": 0.29781176470588233,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1233,
      "step": 12657
    },
    {
      "epoch": 0.29783529411764703,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9676,
      "step": 12658
    },
    {
      "epoch": 0.2978588235294118,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1659,
      "step": 12659
    },
    {
      "epoch": 0.2978823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0076,
      "step": 12660
    },
    {
      "epoch": 0.2979058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.099,
      "step": 12661
    },
    {
      "epoch": 0.2979294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0749,
      "step": 12662
    },
    {
      "epoch": 0.2979529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1355,
      "step": 12663
    },
    {
      "epoch": 0.2979764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1049,
      "step": 12664
    },
    {
      "epoch": 0.298,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2347,
      "step": 12665
    },
    {
      "epoch": 0.2980235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8968,
      "step": 12666
    },
    {
      "epoch": 0.2980470588235294,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.02,
      "loss": 0.6143,
      "step": 12667
    },
    {
      "epoch": 0.29807058823529414,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1801,
      "step": 12668
    },
    {
      "epoch": 0.29809411764705884,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0886,
      "step": 12669
    },
    {
      "epoch": 0.29811764705882354,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2529,
      "step": 12670
    },
    {
      "epoch": 0.29814117647058824,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.786,
      "step": 12671
    },
    {
      "epoch": 0.29816470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2013,
      "step": 12672
    },
    {
      "epoch": 0.29818823529411764,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1673,
      "step": 12673
    },
    {
      "epoch": 0.29821176470588234,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7493,
      "step": 12674
    },
    {
      "epoch": 0.29823529411764704,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1709,
      "step": 12675
    },
    {
      "epoch": 0.29825882352941174,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1222,
      "step": 12676
    },
    {
      "epoch": 0.29828235294117644,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8277,
      "step": 12677
    },
    {
      "epoch": 0.2983058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7993,
      "step": 12678
    },
    {
      "epoch": 0.2983294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0757,
      "step": 12679
    },
    {
      "epoch": 0.2983529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7837,
      "step": 12680
    },
    {
      "epoch": 0.2983764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0864,
      "step": 12681
    },
    {
      "epoch": 0.2984,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0245,
      "step": 12682
    },
    {
      "epoch": 0.2984235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 12683
    },
    {
      "epoch": 0.2984470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8513,
      "step": 12684
    },
    {
      "epoch": 0.2984705882352941,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9504,
      "step": 12685
    },
    {
      "epoch": 0.2984941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8807,
      "step": 12686
    },
    {
      "epoch": 0.29851764705882355,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9352,
      "step": 12687
    },
    {
      "epoch": 0.29854117647058825,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0035,
      "step": 12688
    },
    {
      "epoch": 0.29856470588235295,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3235,
      "step": 12689
    },
    {
      "epoch": 0.29858823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0478,
      "step": 12690
    },
    {
      "epoch": 0.29861176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 12691
    },
    {
      "epoch": 0.29863529411764705,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0812,
      "step": 12692
    },
    {
      "epoch": 0.29865882352941175,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.9193,
      "step": 12693
    },
    {
      "epoch": 0.29868235294117645,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9988,
      "step": 12694
    },
    {
      "epoch": 0.29870588235294115,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8045,
      "step": 12695
    },
    {
      "epoch": 0.2987294117647059,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7607,
      "step": 12696
    },
    {
      "epoch": 0.2987529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8065,
      "step": 12697
    },
    {
      "epoch": 0.2987764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0168,
      "step": 12698
    },
    {
      "epoch": 0.2988,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9818,
      "step": 12699
    },
    {
      "epoch": 0.2988235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.945,
      "step": 12700
    },
    {
      "epoch": 0.2988470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 12701
    },
    {
      "epoch": 0.2988705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1567,
      "step": 12702
    },
    {
      "epoch": 0.2988941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1466,
      "step": 12703
    },
    {
      "epoch": 0.2989176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8312,
      "step": 12704
    },
    {
      "epoch": 0.2989411764705882,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0854,
      "step": 12705
    },
    {
      "epoch": 0.29896470588235297,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1437,
      "step": 12706
    },
    {
      "epoch": 0.29898823529411767,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1388,
      "step": 12707
    },
    {
      "epoch": 0.29901176470588237,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1981,
      "step": 12708
    },
    {
      "epoch": 0.29903529411764707,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9152,
      "step": 12709
    },
    {
      "epoch": 0.29905882352941177,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8435,
      "step": 12710
    },
    {
      "epoch": 0.29908235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8927,
      "step": 12711
    },
    {
      "epoch": 0.29910588235294117,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1131,
      "step": 12712
    },
    {
      "epoch": 0.29912941176470587,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0861,
      "step": 12713
    },
    {
      "epoch": 0.29915294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 12714
    },
    {
      "epoch": 0.2991764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8351,
      "step": 12715
    },
    {
      "epoch": 0.2992,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9153,
      "step": 12716
    },
    {
      "epoch": 0.2992235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2173,
      "step": 12717
    },
    {
      "epoch": 0.2992470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9145,
      "step": 12718
    },
    {
      "epoch": 0.2992705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1635,
      "step": 12719
    },
    {
      "epoch": 0.2992941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 12720
    },
    {
      "epoch": 0.2993176470588235,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2485,
      "step": 12721
    },
    {
      "epoch": 0.2993411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0174,
      "step": 12722
    },
    {
      "epoch": 0.2993647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9554,
      "step": 12723
    },
    {
      "epoch": 0.2993882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1496,
      "step": 12724
    },
    {
      "epoch": 0.2994117647058824,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3835,
      "step": 12725
    },
    {
      "epoch": 0.2994352941176471,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.4169,
      "step": 12726
    },
    {
      "epoch": 0.2994588235294118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0855,
      "step": 12727
    },
    {
      "epoch": 0.2994823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.3339,
      "step": 12728
    },
    {
      "epoch": 0.2995058823529412,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9893,
      "step": 12729
    },
    {
      "epoch": 0.2995294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0747,
      "step": 12730
    },
    {
      "epoch": 0.2995529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9925,
      "step": 12731
    },
    {
      "epoch": 0.2995764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2306,
      "step": 12732
    },
    {
      "epoch": 0.2996,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0668,
      "step": 12733
    },
    {
      "epoch": 0.29962352941176473,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9197,
      "step": 12734
    },
    {
      "epoch": 0.29964705882352943,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1657,
      "step": 12735
    },
    {
      "epoch": 0.29967058823529413,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9527,
      "step": 12736
    },
    {
      "epoch": 0.29969411764705883,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0839,
      "step": 12737
    },
    {
      "epoch": 0.29971764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.1209,
      "step": 12738
    },
    {
      "epoch": 0.29974117647058823,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.8072,
      "step": 12739
    },
    {
      "epoch": 0.29976470588235293,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 12740
    },
    {
      "epoch": 0.29978823529411763,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 12741
    },
    {
      "epoch": 0.29981176470588233,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9428,
      "step": 12742
    },
    {
      "epoch": 0.29983529411764703,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1236,
      "step": 12743
    },
    {
      "epoch": 0.2998588235294118,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7654,
      "step": 12744
    },
    {
      "epoch": 0.2998823529411765,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2339,
      "step": 12745
    },
    {
      "epoch": 0.2999058823529412,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9004,
      "step": 12746
    },
    {
      "epoch": 0.2999294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1987,
      "step": 12747
    },
    {
      "epoch": 0.2999529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.13,
      "step": 12748
    },
    {
      "epoch": 0.2999764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0722,
      "step": 12749
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2344,
      "step": 12750
    },
    {
      "epoch": 0.3000235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0928,
      "step": 12751
    },
    {
      "epoch": 0.3000470588235294,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.6486,
      "step": 12752
    },
    {
      "epoch": 0.30007058823529414,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 12753
    },
    {
      "epoch": 0.30009411764705884,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.017,
      "step": 12754
    },
    {
      "epoch": 0.30011764705882354,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.009,
      "step": 12755
    },
    {
      "epoch": 0.30014117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1891,
      "step": 12756
    },
    {
      "epoch": 0.30016470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 12757
    },
    {
      "epoch": 0.30018823529411764,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2248,
      "step": 12758
    },
    {
      "epoch": 0.30021176470588234,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8149,
      "step": 12759
    },
    {
      "epoch": 0.30023529411764704,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0074,
      "step": 12760
    },
    {
      "epoch": 0.30025882352941174,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1368,
      "step": 12761
    },
    {
      "epoch": 0.30028235294117644,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1027,
      "step": 12762
    },
    {
      "epoch": 0.3003058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1295,
      "step": 12763
    },
    {
      "epoch": 0.3003294117647059,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9216,
      "step": 12764
    },
    {
      "epoch": 0.3003529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0449,
      "step": 12765
    },
    {
      "epoch": 0.3003764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1425,
      "step": 12766
    },
    {
      "epoch": 0.3004,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2637,
      "step": 12767
    },
    {
      "epoch": 0.3004235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9298,
      "step": 12768
    },
    {
      "epoch": 0.3004470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8575,
      "step": 12769
    },
    {
      "epoch": 0.3004705882352941,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.1553,
      "step": 12770
    },
    {
      "epoch": 0.3004941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0104,
      "step": 12771
    },
    {
      "epoch": 0.30051764705882356,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8313,
      "step": 12772
    },
    {
      "epoch": 0.30054117647058826,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1574,
      "step": 12773
    },
    {
      "epoch": 0.30056470588235296,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1284,
      "step": 12774
    },
    {
      "epoch": 0.30058823529411766,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3196,
      "step": 12775
    },
    {
      "epoch": 0.30061176470588236,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1997,
      "step": 12776
    },
    {
      "epoch": 0.30063529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1671,
      "step": 12777
    },
    {
      "epoch": 0.30065882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0324,
      "step": 12778
    },
    {
      "epoch": 0.30068235294117646,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9964,
      "step": 12779
    },
    {
      "epoch": 0.30070588235294116,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.253,
      "step": 12780
    },
    {
      "epoch": 0.30072941176470586,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 12781
    },
    {
      "epoch": 0.3007529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.708,
      "step": 12782
    },
    {
      "epoch": 0.3007764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1922,
      "step": 12783
    },
    {
      "epoch": 0.3008,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2388,
      "step": 12784
    },
    {
      "epoch": 0.3008235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7525,
      "step": 12785
    },
    {
      "epoch": 0.3008470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7374,
      "step": 12786
    },
    {
      "epoch": 0.3008705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0474,
      "step": 12787
    },
    {
      "epoch": 0.3008941176470588,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9288,
      "step": 12788
    },
    {
      "epoch": 0.3009176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1333,
      "step": 12789
    },
    {
      "epoch": 0.3009411764705882,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.7867,
      "step": 12790
    },
    {
      "epoch": 0.30096470588235297,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7812,
      "step": 12791
    },
    {
      "epoch": 0.30098823529411767,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 12792
    },
    {
      "epoch": 0.30101176470588237,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1411,
      "step": 12793
    },
    {
      "epoch": 0.30103529411764707,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1555,
      "step": 12794
    },
    {
      "epoch": 0.30105882352941177,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0532,
      "step": 12795
    },
    {
      "epoch": 0.30108235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1116,
      "step": 12796
    },
    {
      "epoch": 0.30110588235294117,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1505,
      "step": 12797
    },
    {
      "epoch": 0.30112941176470587,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0874,
      "step": 12798
    },
    {
      "epoch": 0.30115294117647057,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1654,
      "step": 12799
    },
    {
      "epoch": 0.30117647058823527,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0587,
      "step": 12800
    },
    {
      "epoch": 0.3012,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9263,
      "step": 12801
    },
    {
      "epoch": 0.3012235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1296,
      "step": 12802
    },
    {
      "epoch": 0.3012470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9365,
      "step": 12803
    },
    {
      "epoch": 0.3012705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9988,
      "step": 12804
    },
    {
      "epoch": 0.3012941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 12805
    },
    {
      "epoch": 0.3013176470588235,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.6182,
      "step": 12806
    },
    {
      "epoch": 0.3013411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9431,
      "step": 12807
    },
    {
      "epoch": 0.3013647058823529,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0009,
      "step": 12808
    },
    {
      "epoch": 0.3013882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.209,
      "step": 12809
    },
    {
      "epoch": 0.3014117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1799,
      "step": 12810
    },
    {
      "epoch": 0.3014352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0581,
      "step": 12811
    },
    {
      "epoch": 0.3014588235294118,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9028,
      "step": 12812
    },
    {
      "epoch": 0.3014823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9018,
      "step": 12813
    },
    {
      "epoch": 0.3015058823529412,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9045,
      "step": 12814
    },
    {
      "epoch": 0.3015294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7909,
      "step": 12815
    },
    {
      "epoch": 0.3015529411764706,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2275,
      "step": 12816
    },
    {
      "epoch": 0.3015764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1755,
      "step": 12817
    },
    {
      "epoch": 0.3016,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.6684,
      "step": 12818
    },
    {
      "epoch": 0.3016235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0597,
      "step": 12819
    },
    {
      "epoch": 0.30164705882352943,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0066,
      "step": 12820
    },
    {
      "epoch": 0.30167058823529413,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7811,
      "step": 12821
    },
    {
      "epoch": 0.30169411764705883,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0488,
      "step": 12822
    },
    {
      "epoch": 0.30171764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0833,
      "step": 12823
    },
    {
      "epoch": 0.30174117647058823,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0037,
      "step": 12824
    },
    {
      "epoch": 0.30176470588235293,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9945,
      "step": 12825
    },
    {
      "epoch": 0.30178823529411763,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1818,
      "step": 12826
    },
    {
      "epoch": 0.30181176470588233,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0585,
      "step": 12827
    },
    {
      "epoch": 0.30183529411764703,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 12828
    },
    {
      "epoch": 0.3018588235294118,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0388,
      "step": 12829
    },
    {
      "epoch": 0.3018823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7887,
      "step": 12830
    },
    {
      "epoch": 0.3019058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1352,
      "step": 12831
    },
    {
      "epoch": 0.3019294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8335,
      "step": 12832
    },
    {
      "epoch": 0.3019529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0581,
      "step": 12833
    },
    {
      "epoch": 0.3019764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.6882,
      "step": 12834
    },
    {
      "epoch": 0.302,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1549,
      "step": 12835
    },
    {
      "epoch": 0.3020235294117647,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 12836
    },
    {
      "epoch": 0.3020470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 12837
    },
    {
      "epoch": 0.3020705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9462,
      "step": 12838
    },
    {
      "epoch": 0.30209411764705885,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9344,
      "step": 12839
    },
    {
      "epoch": 0.30211764705882355,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1213,
      "step": 12840
    },
    {
      "epoch": 0.30214117647058825,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9948,
      "step": 12841
    },
    {
      "epoch": 0.30216470588235295,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 12842
    },
    {
      "epoch": 0.30218823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8866,
      "step": 12843
    },
    {
      "epoch": 0.30221176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0935,
      "step": 12844
    },
    {
      "epoch": 0.30223529411764705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1247,
      "step": 12845
    },
    {
      "epoch": 0.30225882352941175,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8731,
      "step": 12846
    },
    {
      "epoch": 0.30228235294117645,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0159,
      "step": 12847
    },
    {
      "epoch": 0.3023058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 12848
    },
    {
      "epoch": 0.3023294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.088,
      "step": 12849
    },
    {
      "epoch": 0.3023529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8519,
      "step": 12850
    },
    {
      "epoch": 0.3023764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1753,
      "step": 12851
    },
    {
      "epoch": 0.3024,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3031,
      "step": 12852
    },
    {
      "epoch": 0.3024235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0034,
      "step": 12853
    },
    {
      "epoch": 0.3024470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0861,
      "step": 12854
    },
    {
      "epoch": 0.3024705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6043,
      "step": 12855
    },
    {
      "epoch": 0.3024941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0452,
      "step": 12856
    },
    {
      "epoch": 0.3025176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0671,
      "step": 12857
    },
    {
      "epoch": 0.30254117647058826,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0845,
      "step": 12858
    },
    {
      "epoch": 0.30256470588235296,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0571,
      "step": 12859
    },
    {
      "epoch": 0.30258823529411766,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1166,
      "step": 12860
    },
    {
      "epoch": 0.30261176470588236,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0555,
      "step": 12861
    },
    {
      "epoch": 0.30263529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9904,
      "step": 12862
    },
    {
      "epoch": 0.30265882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1295,
      "step": 12863
    },
    {
      "epoch": 0.30268235294117646,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1225,
      "step": 12864
    },
    {
      "epoch": 0.30270588235294116,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9674,
      "step": 12865
    },
    {
      "epoch": 0.30272941176470586,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7986,
      "step": 12866
    },
    {
      "epoch": 0.3027529411764706,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1425,
      "step": 12867
    },
    {
      "epoch": 0.3027764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1646,
      "step": 12868
    },
    {
      "epoch": 0.3028,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1516,
      "step": 12869
    },
    {
      "epoch": 0.3028235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3338,
      "step": 12870
    },
    {
      "epoch": 0.3028470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1221,
      "step": 12871
    },
    {
      "epoch": 0.3028705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.059,
      "step": 12872
    },
    {
      "epoch": 0.3028941176470588,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.212,
      "step": 12873
    },
    {
      "epoch": 0.3029176470588235,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0321,
      "step": 12874
    },
    {
      "epoch": 0.3029411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.3094,
      "step": 12875
    },
    {
      "epoch": 0.3029647058823529,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8874,
      "step": 12876
    },
    {
      "epoch": 0.30298823529411767,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9053,
      "step": 12877
    },
    {
      "epoch": 0.30301176470588237,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9573,
      "step": 12878
    },
    {
      "epoch": 0.30303529411764707,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9433,
      "step": 12879
    },
    {
      "epoch": 0.30305882352941177,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.2222,
      "step": 12880
    },
    {
      "epoch": 0.30308235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8712,
      "step": 12881
    },
    {
      "epoch": 0.30310588235294117,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.6966,
      "step": 12882
    },
    {
      "epoch": 0.30312941176470587,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0769,
      "step": 12883
    },
    {
      "epoch": 0.30315294117647057,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8697,
      "step": 12884
    },
    {
      "epoch": 0.30317647058823527,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9447,
      "step": 12885
    },
    {
      "epoch": 0.3032,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8258,
      "step": 12886
    },
    {
      "epoch": 0.3032235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9236,
      "step": 12887
    },
    {
      "epoch": 0.3032470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0462,
      "step": 12888
    },
    {
      "epoch": 0.3032705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0321,
      "step": 12889
    },
    {
      "epoch": 0.3032941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.142,
      "step": 12890
    },
    {
      "epoch": 0.3033176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2341,
      "step": 12891
    },
    {
      "epoch": 0.3033411764705882,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8538,
      "step": 12892
    },
    {
      "epoch": 0.3033647058823529,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8425,
      "step": 12893
    },
    {
      "epoch": 0.3033882352941176,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2962,
      "step": 12894
    },
    {
      "epoch": 0.3034117647058823,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 12895
    },
    {
      "epoch": 0.3034352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0322,
      "step": 12896
    },
    {
      "epoch": 0.3034588235294118,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9285,
      "step": 12897
    },
    {
      "epoch": 0.3034823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0254,
      "step": 12898
    },
    {
      "epoch": 0.3035058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 12899
    },
    {
      "epoch": 0.3035294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.2658,
      "step": 12900
    },
    {
      "epoch": 0.3035529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.4188,
      "step": 12901
    },
    {
      "epoch": 0.3035764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9979,
      "step": 12902
    },
    {
      "epoch": 0.3036,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7479,
      "step": 12903
    },
    {
      "epoch": 0.3036235294117647,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.7941,
      "step": 12904
    },
    {
      "epoch": 0.30364705882352944,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9072,
      "step": 12905
    },
    {
      "epoch": 0.30367058823529414,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2693,
      "step": 12906
    },
    {
      "epoch": 0.30369411764705884,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8604,
      "step": 12907
    },
    {
      "epoch": 0.30371764705882354,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.028,
      "step": 12908
    },
    {
      "epoch": 0.30374117647058824,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9622,
      "step": 12909
    },
    {
      "epoch": 0.30376470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2345,
      "step": 12910
    },
    {
      "epoch": 0.30378823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.862,
      "step": 12911
    },
    {
      "epoch": 0.30381176470588234,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1481,
      "step": 12912
    },
    {
      "epoch": 0.30383529411764704,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9647,
      "step": 12913
    },
    {
      "epoch": 0.3038588235294118,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8229,
      "step": 12914
    },
    {
      "epoch": 0.3038823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2039,
      "step": 12915
    },
    {
      "epoch": 0.3039058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7789,
      "step": 12916
    },
    {
      "epoch": 0.3039294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1384,
      "step": 12917
    },
    {
      "epoch": 0.3039529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9627,
      "step": 12918
    },
    {
      "epoch": 0.3039764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 12919
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.258,
      "step": 12920
    },
    {
      "epoch": 0.3040235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1879,
      "step": 12921
    },
    {
      "epoch": 0.3040470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2966,
      "step": 12922
    },
    {
      "epoch": 0.3040705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8565,
      "step": 12923
    },
    {
      "epoch": 0.30409411764705885,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9711,
      "step": 12924
    },
    {
      "epoch": 0.30411764705882355,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6664,
      "step": 12925
    },
    {
      "epoch": 0.30414117647058825,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0046,
      "step": 12926
    },
    {
      "epoch": 0.30416470588235295,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9313,
      "step": 12927
    },
    {
      "epoch": 0.30418823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 12928
    },
    {
      "epoch": 0.30421176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.2592,
      "step": 12929
    },
    {
      "epoch": 0.30423529411764705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 1.0044,
      "step": 12930
    },
    {
      "epoch": 0.30425882352941175,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3356,
      "step": 12931
    },
    {
      "epoch": 0.30428235294117645,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.0077,
      "step": 12932
    },
    {
      "epoch": 0.3043058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9642,
      "step": 12933
    },
    {
      "epoch": 0.3043294117647059,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9539,
      "step": 12934
    },
    {
      "epoch": 0.3043529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.941,
      "step": 12935
    },
    {
      "epoch": 0.3043764705882353,
      "grad_norm": 0.318359375,
      "learning_rate": 0.02,
      "loss": 1.1025,
      "step": 12936
    },
    {
      "epoch": 0.3044,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7394,
      "step": 12937
    },
    {
      "epoch": 0.3044235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0828,
      "step": 12938
    },
    {
      "epoch": 0.3044470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1255,
      "step": 12939
    },
    {
      "epoch": 0.3044705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8771,
      "step": 12940
    },
    {
      "epoch": 0.3044941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0988,
      "step": 12941
    },
    {
      "epoch": 0.3045176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.119,
      "step": 12942
    },
    {
      "epoch": 0.30454117647058826,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8716,
      "step": 12943
    },
    {
      "epoch": 0.30456470588235296,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 1.0113,
      "step": 12944
    },
    {
      "epoch": 0.30458823529411766,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9853,
      "step": 12945
    },
    {
      "epoch": 0.30461176470588236,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8352,
      "step": 12946
    },
    {
      "epoch": 0.30463529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.175,
      "step": 12947
    },
    {
      "epoch": 0.30465882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1495,
      "step": 12948
    },
    {
      "epoch": 0.30468235294117646,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.5224,
      "step": 12949
    },
    {
      "epoch": 0.30470588235294116,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7359,
      "step": 12950
    },
    {
      "epoch": 0.30472941176470586,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 12951
    },
    {
      "epoch": 0.3047529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8613,
      "step": 12952
    },
    {
      "epoch": 0.3047764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 1.0751,
      "step": 12953
    },
    {
      "epoch": 0.3048,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0271,
      "step": 12954
    },
    {
      "epoch": 0.3048235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 12955
    },
    {
      "epoch": 0.3048470588235294,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3074,
      "step": 12956
    },
    {
      "epoch": 0.3048705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9372,
      "step": 12957
    },
    {
      "epoch": 0.3048941176470588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.709,
      "step": 12958
    },
    {
      "epoch": 0.3049176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8804,
      "step": 12959
    },
    {
      "epoch": 0.3049411764705882,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9747,
      "step": 12960
    },
    {
      "epoch": 0.3049647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.4235,
      "step": 12961
    },
    {
      "epoch": 0.30498823529411767,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.8439,
      "step": 12962
    },
    {
      "epoch": 0.30501176470588237,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.9661,
      "step": 12963
    },
    {
      "epoch": 0.30503529411764707,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.8228,
      "step": 12964
    },
    {
      "epoch": 0.30505882352941177,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.3562,
      "step": 12965
    },
    {
      "epoch": 0.30508235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1825,
      "step": 12966
    },
    {
      "epoch": 0.30510588235294117,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9521,
      "step": 12967
    },
    {
      "epoch": 0.30512941176470587,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.6476,
      "step": 12968
    },
    {
      "epoch": 0.30515294117647057,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6908,
      "step": 12969
    },
    {
      "epoch": 0.30517647058823527,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.271,
      "step": 12970
    },
    {
      "epoch": 0.3052,
      "grad_norm": 0.73828125,
      "learning_rate": 0.02,
      "loss": 0.866,
      "step": 12971
    },
    {
      "epoch": 0.3052235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 12972
    },
    {
      "epoch": 0.3052470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1322,
      "step": 12973
    },
    {
      "epoch": 0.3052705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.257,
      "step": 12974
    },
    {
      "epoch": 0.3052941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8166,
      "step": 12975
    },
    {
      "epoch": 0.3053176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1144,
      "step": 12976
    },
    {
      "epoch": 0.3053411764705882,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9485,
      "step": 12977
    },
    {
      "epoch": 0.3053647058823529,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.914,
      "step": 12978
    },
    {
      "epoch": 0.3053882352941176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0844,
      "step": 12979
    },
    {
      "epoch": 0.3054117647058823,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9871,
      "step": 12980
    },
    {
      "epoch": 0.3054352941176471,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.925,
      "step": 12981
    },
    {
      "epoch": 0.3054588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0536,
      "step": 12982
    },
    {
      "epoch": 0.3054823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0729,
      "step": 12983
    },
    {
      "epoch": 0.3055058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3307,
      "step": 12984
    },
    {
      "epoch": 0.3055294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1362,
      "step": 12985
    },
    {
      "epoch": 0.3055529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7814,
      "step": 12986
    },
    {
      "epoch": 0.3055764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1344,
      "step": 12987
    },
    {
      "epoch": 0.3056,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.5962,
      "step": 12988
    },
    {
      "epoch": 0.3056235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.3194,
      "step": 12989
    },
    {
      "epoch": 0.30564705882352944,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9537,
      "step": 12990
    },
    {
      "epoch": 0.30567058823529414,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9602,
      "step": 12991
    },
    {
      "epoch": 0.30569411764705884,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.834,
      "step": 12992
    },
    {
      "epoch": 0.30571764705882354,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.197,
      "step": 12993
    },
    {
      "epoch": 0.30574117647058824,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3172,
      "step": 12994
    },
    {
      "epoch": 0.30576470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9606,
      "step": 12995
    },
    {
      "epoch": 0.30578823529411764,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1514,
      "step": 12996
    },
    {
      "epoch": 0.30581176470588234,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7119,
      "step": 12997
    },
    {
      "epoch": 0.30583529411764704,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1373,
      "step": 12998
    },
    {
      "epoch": 0.30585882352941174,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2613,
      "step": 12999
    },
    {
      "epoch": 0.3058823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9309,
      "step": 13000
    },
    {
      "epoch": 0.3058823529411765,
      "eval_loss": 2.2015132904052734,
      "eval_runtime": 683.9391,
      "eval_samples_per_second": 12.428,
      "eval_steps_per_second": 3.107,
      "step": 13000
    },
    {
      "epoch": 0.3059058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1553,
      "step": 13001
    },
    {
      "epoch": 0.3059294117647059,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.288,
      "step": 13002
    },
    {
      "epoch": 0.3059529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0806,
      "step": 13003
    },
    {
      "epoch": 0.3059764705882353,
      "grad_norm": 0.2578125,
      "learning_rate": 0.02,
      "loss": 0.8829,
      "step": 13004
    },
    {
      "epoch": 0.306,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.9516,
      "step": 13005
    },
    {
      "epoch": 0.3060235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.486,
      "step": 13006
    },
    {
      "epoch": 0.3060470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2632,
      "step": 13007
    },
    {
      "epoch": 0.3060705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0298,
      "step": 13008
    },
    {
      "epoch": 0.30609411764705885,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0789,
      "step": 13009
    },
    {
      "epoch": 0.30611764705882355,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9361,
      "step": 13010
    },
    {
      "epoch": 0.30614117647058825,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0789,
      "step": 13011
    },
    {
      "epoch": 0.30616470588235295,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7847,
      "step": 13012
    },
    {
      "epoch": 0.30618823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2502,
      "step": 13013
    },
    {
      "epoch": 0.30621176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2146,
      "step": 13014
    },
    {
      "epoch": 0.30623529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 1.0331,
      "step": 13015
    },
    {
      "epoch": 0.30625882352941175,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1615,
      "step": 13016
    },
    {
      "epoch": 0.30628235294117645,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1422,
      "step": 13017
    },
    {
      "epoch": 0.30630588235294115,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1882,
      "step": 13018
    },
    {
      "epoch": 0.3063294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.983,
      "step": 13019
    },
    {
      "epoch": 0.3063529411764706,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1149,
      "step": 13020
    },
    {
      "epoch": 0.3063764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.078,
      "step": 13021
    },
    {
      "epoch": 0.3064,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1242,
      "step": 13022
    },
    {
      "epoch": 0.3064235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0373,
      "step": 13023
    },
    {
      "epoch": 0.3064470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 13024
    },
    {
      "epoch": 0.3064705882352941,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0175,
      "step": 13025
    },
    {
      "epoch": 0.3064941176470588,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.8885,
      "step": 13026
    },
    {
      "epoch": 0.3065176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0572,
      "step": 13027
    },
    {
      "epoch": 0.30654117647058826,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9364,
      "step": 13028
    },
    {
      "epoch": 0.30656470588235296,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8199,
      "step": 13029
    },
    {
      "epoch": 0.30658823529411766,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8845,
      "step": 13030
    },
    {
      "epoch": 0.30661176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9722,
      "step": 13031
    },
    {
      "epoch": 0.30663529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9512,
      "step": 13032
    },
    {
      "epoch": 0.30665882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1062,
      "step": 13033
    },
    {
      "epoch": 0.30668235294117646,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9714,
      "step": 13034
    },
    {
      "epoch": 0.30670588235294116,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7098,
      "step": 13035
    },
    {
      "epoch": 0.30672941176470586,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2746,
      "step": 13036
    },
    {
      "epoch": 0.30675294117647056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0616,
      "step": 13037
    },
    {
      "epoch": 0.3067764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9803,
      "step": 13038
    },
    {
      "epoch": 0.3068,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2482,
      "step": 13039
    },
    {
      "epoch": 0.3068235294117647,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3368,
      "step": 13040
    },
    {
      "epoch": 0.3068470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9432,
      "step": 13041
    },
    {
      "epoch": 0.3068705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7637,
      "step": 13042
    },
    {
      "epoch": 0.3068941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.011,
      "step": 13043
    },
    {
      "epoch": 0.3069176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 13044
    },
    {
      "epoch": 0.3069411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1128,
      "step": 13045
    },
    {
      "epoch": 0.3069647058823529,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1213,
      "step": 13046
    },
    {
      "epoch": 0.3069882352941177,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.023,
      "step": 13047
    },
    {
      "epoch": 0.3070117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9715,
      "step": 13048
    },
    {
      "epoch": 0.3070352941176471,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.3262,
      "step": 13049
    },
    {
      "epoch": 0.3070588235294118,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8977,
      "step": 13050
    },
    {
      "epoch": 0.3070823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1536,
      "step": 13051
    },
    {
      "epoch": 0.3071058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0293,
      "step": 13052
    },
    {
      "epoch": 0.3071294117647059,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8304,
      "step": 13053
    },
    {
      "epoch": 0.3071529411764706,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7295,
      "step": 13054
    },
    {
      "epoch": 0.3071764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0334,
      "step": 13055
    },
    {
      "epoch": 0.3072,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7655,
      "step": 13056
    },
    {
      "epoch": 0.30722352941176473,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1156,
      "step": 13057
    },
    {
      "epoch": 0.30724705882352943,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0849,
      "step": 13058
    },
    {
      "epoch": 0.30727058823529413,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0885,
      "step": 13059
    },
    {
      "epoch": 0.30729411764705883,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3606,
      "step": 13060
    },
    {
      "epoch": 0.30731764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.1864,
      "step": 13061
    },
    {
      "epoch": 0.30734117647058823,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.2902,
      "step": 13062
    },
    {
      "epoch": 0.30736470588235293,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.5823,
      "step": 13063
    },
    {
      "epoch": 0.30738823529411763,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.4154,
      "step": 13064
    },
    {
      "epoch": 0.30741176470588233,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2003,
      "step": 13065
    },
    {
      "epoch": 0.3074352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0414,
      "step": 13066
    },
    {
      "epoch": 0.3074588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2341,
      "step": 13067
    },
    {
      "epoch": 0.3074823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0832,
      "step": 13068
    },
    {
      "epoch": 0.3075058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0764,
      "step": 13069
    },
    {
      "epoch": 0.3075294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9011,
      "step": 13070
    },
    {
      "epoch": 0.3075529411764706,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3503,
      "step": 13071
    },
    {
      "epoch": 0.3075764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9237,
      "step": 13072
    },
    {
      "epoch": 0.3076,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1905,
      "step": 13073
    },
    {
      "epoch": 0.3076235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8552,
      "step": 13074
    },
    {
      "epoch": 0.3076470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9966,
      "step": 13075
    },
    {
      "epoch": 0.30767058823529414,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0083,
      "step": 13076
    },
    {
      "epoch": 0.30769411764705884,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1953,
      "step": 13077
    },
    {
      "epoch": 0.30771764705882354,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9882,
      "step": 13078
    },
    {
      "epoch": 0.30774117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0353,
      "step": 13079
    },
    {
      "epoch": 0.30776470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8728,
      "step": 13080
    },
    {
      "epoch": 0.30778823529411764,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 13081
    },
    {
      "epoch": 0.30781176470588234,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7724,
      "step": 13082
    },
    {
      "epoch": 0.30783529411764704,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 13083
    },
    {
      "epoch": 0.30785882352941174,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1804,
      "step": 13084
    },
    {
      "epoch": 0.3078823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7728,
      "step": 13085
    },
    {
      "epoch": 0.3079058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8628,
      "step": 13086
    },
    {
      "epoch": 0.3079294117647059,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0312,
      "step": 13087
    },
    {
      "epoch": 0.3079529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8633,
      "step": 13088
    },
    {
      "epoch": 0.3079764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.069,
      "step": 13089
    },
    {
      "epoch": 0.308,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.94,
      "step": 13090
    },
    {
      "epoch": 0.3080235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1423,
      "step": 13091
    },
    {
      "epoch": 0.3080470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2016,
      "step": 13092
    },
    {
      "epoch": 0.3080705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1838,
      "step": 13093
    },
    {
      "epoch": 0.3080941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9923,
      "step": 13094
    },
    {
      "epoch": 0.30811764705882355,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.4392,
      "step": 13095
    },
    {
      "epoch": 0.30814117647058825,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0927,
      "step": 13096
    },
    {
      "epoch": 0.30816470588235295,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.02,
      "loss": 0.5387,
      "step": 13097
    },
    {
      "epoch": 0.30818823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8917,
      "step": 13098
    },
    {
      "epoch": 0.30821176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8235,
      "step": 13099
    },
    {
      "epoch": 0.30823529411764705,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0479,
      "step": 13100
    },
    {
      "epoch": 0.30825882352941175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2776,
      "step": 13101
    },
    {
      "epoch": 0.30828235294117645,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8668,
      "step": 13102
    },
    {
      "epoch": 0.30830588235294115,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2911,
      "step": 13103
    },
    {
      "epoch": 0.3083294117647059,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.4841,
      "step": 13104
    },
    {
      "epoch": 0.3083529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2049,
      "step": 13105
    },
    {
      "epoch": 0.3083764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2981,
      "step": 13106
    },
    {
      "epoch": 0.3084,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9013,
      "step": 13107
    },
    {
      "epoch": 0.3084235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9572,
      "step": 13108
    },
    {
      "epoch": 0.3084470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9223,
      "step": 13109
    },
    {
      "epoch": 0.3084705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9922,
      "step": 13110
    },
    {
      "epoch": 0.3084941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0035,
      "step": 13111
    },
    {
      "epoch": 0.3085176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 13112
    },
    {
      "epoch": 0.3085411764705882,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.063,
      "step": 13113
    },
    {
      "epoch": 0.30856470588235296,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 13114
    },
    {
      "epoch": 0.30858823529411766,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8297,
      "step": 13115
    },
    {
      "epoch": 0.30861176470588236,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.76,
      "step": 13116
    },
    {
      "epoch": 0.30863529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0367,
      "step": 13117
    },
    {
      "epoch": 0.30865882352941176,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2925,
      "step": 13118
    },
    {
      "epoch": 0.30868235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1094,
      "step": 13119
    },
    {
      "epoch": 0.30870588235294116,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.849,
      "step": 13120
    },
    {
      "epoch": 0.30872941176470586,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2473,
      "step": 13121
    },
    {
      "epoch": 0.30875294117647056,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0777,
      "step": 13122
    },
    {
      "epoch": 0.3087764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1735,
      "step": 13123
    },
    {
      "epoch": 0.3088,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0684,
      "step": 13124
    },
    {
      "epoch": 0.3088235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9676,
      "step": 13125
    },
    {
      "epoch": 0.3088470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1676,
      "step": 13126
    },
    {
      "epoch": 0.3088705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1387,
      "step": 13127
    },
    {
      "epoch": 0.3088941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1179,
      "step": 13128
    },
    {
      "epoch": 0.3089176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0349,
      "step": 13129
    },
    {
      "epoch": 0.3089411764705882,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0863,
      "step": 13130
    },
    {
      "epoch": 0.3089647058823529,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8528,
      "step": 13131
    },
    {
      "epoch": 0.3089882352941177,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8104,
      "step": 13132
    },
    {
      "epoch": 0.3090117647058824,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9374,
      "step": 13133
    },
    {
      "epoch": 0.3090352941176471,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0462,
      "step": 13134
    },
    {
      "epoch": 0.3090588235294118,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1546,
      "step": 13135
    },
    {
      "epoch": 0.3090823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.152,
      "step": 13136
    },
    {
      "epoch": 0.3091058823529412,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.7054,
      "step": 13137
    },
    {
      "epoch": 0.3091294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1254,
      "step": 13138
    },
    {
      "epoch": 0.3091529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9398,
      "step": 13139
    },
    {
      "epoch": 0.3091764705882353,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9853,
      "step": 13140
    },
    {
      "epoch": 0.3092,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9558,
      "step": 13141
    },
    {
      "epoch": 0.30922352941176473,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0909,
      "step": 13142
    },
    {
      "epoch": 0.30924705882352943,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.4037,
      "step": 13143
    },
    {
      "epoch": 0.30927058823529413,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0558,
      "step": 13144
    },
    {
      "epoch": 0.30929411764705883,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1198,
      "step": 13145
    },
    {
      "epoch": 0.30931764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 13146
    },
    {
      "epoch": 0.30934117647058823,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7695,
      "step": 13147
    },
    {
      "epoch": 0.30936470588235293,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1484,
      "step": 13148
    },
    {
      "epoch": 0.30938823529411763,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9115,
      "step": 13149
    },
    {
      "epoch": 0.30941176470588233,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2006,
      "step": 13150
    },
    {
      "epoch": 0.3094352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1606,
      "step": 13151
    },
    {
      "epoch": 0.3094588235294118,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 13152
    },
    {
      "epoch": 0.3094823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0734,
      "step": 13153
    },
    {
      "epoch": 0.3095058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0402,
      "step": 13154
    },
    {
      "epoch": 0.3095294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0396,
      "step": 13155
    },
    {
      "epoch": 0.3095529411764706,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.172,
      "step": 13156
    },
    {
      "epoch": 0.3095764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1587,
      "step": 13157
    },
    {
      "epoch": 0.3096,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9332,
      "step": 13158
    },
    {
      "epoch": 0.3096235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7946,
      "step": 13159
    },
    {
      "epoch": 0.3096470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0156,
      "step": 13160
    },
    {
      "epoch": 0.30967058823529414,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 13161
    },
    {
      "epoch": 0.30969411764705884,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.4091,
      "step": 13162
    },
    {
      "epoch": 0.30971764705882354,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9842,
      "step": 13163
    },
    {
      "epoch": 0.30974117647058824,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7658,
      "step": 13164
    },
    {
      "epoch": 0.30976470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0604,
      "step": 13165
    },
    {
      "epoch": 0.30978823529411764,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0707,
      "step": 13166
    },
    {
      "epoch": 0.30981176470588234,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8709,
      "step": 13167
    },
    {
      "epoch": 0.30983529411764704,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7656,
      "step": 13168
    },
    {
      "epoch": 0.30985882352941174,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7802,
      "step": 13169
    },
    {
      "epoch": 0.3098823529411765,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1565,
      "step": 13170
    },
    {
      "epoch": 0.3099058823529412,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.2038,
      "step": 13171
    },
    {
      "epoch": 0.3099294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.171,
      "step": 13172
    },
    {
      "epoch": 0.3099529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9294,
      "step": 13173
    },
    {
      "epoch": 0.3099764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0416,
      "step": 13174
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8927,
      "step": 13175
    },
    {
      "epoch": 0.3100235294117647,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.02,
      "loss": 0.7533,
      "step": 13176
    },
    {
      "epoch": 0.3100470588235294,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.4117,
      "step": 13177
    },
    {
      "epoch": 0.3100705882352941,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2692,
      "step": 13178
    },
    {
      "epoch": 0.3100941176470588,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2114,
      "step": 13179
    },
    {
      "epoch": 0.31011764705882355,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1601,
      "step": 13180
    },
    {
      "epoch": 0.31014117647058825,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3297,
      "step": 13181
    },
    {
      "epoch": 0.31016470588235295,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2502,
      "step": 13182
    },
    {
      "epoch": 0.31018823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1382,
      "step": 13183
    },
    {
      "epoch": 0.31021176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0707,
      "step": 13184
    },
    {
      "epoch": 0.31023529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.125,
      "step": 13185
    },
    {
      "epoch": 0.31025882352941175,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.118,
      "step": 13186
    },
    {
      "epoch": 0.31028235294117645,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7866,
      "step": 13187
    },
    {
      "epoch": 0.31030588235294115,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.035,
      "step": 13188
    },
    {
      "epoch": 0.3103294117647059,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.4753,
      "step": 13189
    },
    {
      "epoch": 0.3103529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.5335,
      "step": 13190
    },
    {
      "epoch": 0.3103764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9127,
      "step": 13191
    },
    {
      "epoch": 0.3104,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0155,
      "step": 13192
    },
    {
      "epoch": 0.3104235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1762,
      "step": 13193
    },
    {
      "epoch": 0.3104470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.259,
      "step": 13194
    },
    {
      "epoch": 0.3104705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9541,
      "step": 13195
    },
    {
      "epoch": 0.3104941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7768,
      "step": 13196
    },
    {
      "epoch": 0.3105176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0196,
      "step": 13197
    },
    {
      "epoch": 0.3105411764705882,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8232,
      "step": 13198
    },
    {
      "epoch": 0.31056470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2654,
      "step": 13199
    },
    {
      "epoch": 0.31058823529411766,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9218,
      "step": 13200
    },
    {
      "epoch": 0.31061176470588236,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.265,
      "step": 13201
    },
    {
      "epoch": 0.31063529411764707,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0482,
      "step": 13202
    },
    {
      "epoch": 0.31065882352941177,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.9103,
      "step": 13203
    },
    {
      "epoch": 0.31068235294117647,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8915,
      "step": 13204
    },
    {
      "epoch": 0.31070588235294117,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7059,
      "step": 13205
    },
    {
      "epoch": 0.31072941176470587,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.6999,
      "step": 13206
    },
    {
      "epoch": 0.31075294117647057,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7528,
      "step": 13207
    },
    {
      "epoch": 0.3107764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.6809,
      "step": 13208
    },
    {
      "epoch": 0.3108,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1788,
      "step": 13209
    },
    {
      "epoch": 0.3108235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9201,
      "step": 13210
    },
    {
      "epoch": 0.3108470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9315,
      "step": 13211
    },
    {
      "epoch": 0.3108705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8683,
      "step": 13212
    },
    {
      "epoch": 0.3108941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9399,
      "step": 13213
    },
    {
      "epoch": 0.3109176470588235,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.885,
      "step": 13214
    },
    {
      "epoch": 0.3109411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0441,
      "step": 13215
    },
    {
      "epoch": 0.3109647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.068,
      "step": 13216
    },
    {
      "epoch": 0.3109882352941176,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2377,
      "step": 13217
    },
    {
      "epoch": 0.3110117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8378,
      "step": 13218
    },
    {
      "epoch": 0.3110352941176471,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.201,
      "step": 13219
    },
    {
      "epoch": 0.3110588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9358,
      "step": 13220
    },
    {
      "epoch": 0.3110823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9968,
      "step": 13221
    },
    {
      "epoch": 0.3111058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0043,
      "step": 13222
    },
    {
      "epoch": 0.3111294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9869,
      "step": 13223
    },
    {
      "epoch": 0.3111529411764706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.4398,
      "step": 13224
    },
    {
      "epoch": 0.3111764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6539,
      "step": 13225
    },
    {
      "epoch": 0.3112,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1876,
      "step": 13226
    },
    {
      "epoch": 0.31122352941176473,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6521,
      "step": 13227
    },
    {
      "epoch": 0.31124705882352943,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9833,
      "step": 13228
    },
    {
      "epoch": 0.31127058823529413,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.273,
      "step": 13229
    },
    {
      "epoch": 0.31129411764705883,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9859,
      "step": 13230
    },
    {
      "epoch": 0.31131764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1098,
      "step": 13231
    },
    {
      "epoch": 0.31134117647058823,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2308,
      "step": 13232
    },
    {
      "epoch": 0.31136470588235293,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9922,
      "step": 13233
    },
    {
      "epoch": 0.31138823529411763,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2749,
      "step": 13234
    },
    {
      "epoch": 0.31141176470588233,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9742,
      "step": 13235
    },
    {
      "epoch": 0.31143529411764703,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8216,
      "step": 13236
    },
    {
      "epoch": 0.3114588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0488,
      "step": 13237
    },
    {
      "epoch": 0.3114823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0359,
      "step": 13238
    },
    {
      "epoch": 0.3115058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9291,
      "step": 13239
    },
    {
      "epoch": 0.3115294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1928,
      "step": 13240
    },
    {
      "epoch": 0.3115529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8007,
      "step": 13241
    },
    {
      "epoch": 0.3115764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 13242
    },
    {
      "epoch": 0.3116,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9943,
      "step": 13243
    },
    {
      "epoch": 0.3116235294117647,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.7025,
      "step": 13244
    },
    {
      "epoch": 0.3116470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.039,
      "step": 13245
    },
    {
      "epoch": 0.31167058823529414,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8477,
      "step": 13246
    },
    {
      "epoch": 0.31169411764705884,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8008,
      "step": 13247
    },
    {
      "epoch": 0.31171764705882354,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9895,
      "step": 13248
    },
    {
      "epoch": 0.31174117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8288,
      "step": 13249
    },
    {
      "epoch": 0.31176470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1297,
      "step": 13250
    },
    {
      "epoch": 0.31178823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0568,
      "step": 13251
    },
    {
      "epoch": 0.31181176470588234,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1551,
      "step": 13252
    },
    {
      "epoch": 0.31183529411764704,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1945,
      "step": 13253
    },
    {
      "epoch": 0.31185882352941174,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0649,
      "step": 13254
    },
    {
      "epoch": 0.31188235294117644,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8219,
      "step": 13255
    },
    {
      "epoch": 0.3119058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0565,
      "step": 13256
    },
    {
      "epoch": 0.3119294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8148,
      "step": 13257
    },
    {
      "epoch": 0.3119529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6697,
      "step": 13258
    },
    {
      "epoch": 0.3119764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0861,
      "step": 13259
    },
    {
      "epoch": 0.312,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8142,
      "step": 13260
    },
    {
      "epoch": 0.3120235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8151,
      "step": 13261
    },
    {
      "epoch": 0.3120470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.867,
      "step": 13262
    },
    {
      "epoch": 0.3120705882352941,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.5407,
      "step": 13263
    },
    {
      "epoch": 0.3120941176470588,
      "grad_norm": 0.244140625,
      "learning_rate": 0.02,
      "loss": 0.4139,
      "step": 13264
    },
    {
      "epoch": 0.31211764705882356,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0088,
      "step": 13265
    },
    {
      "epoch": 0.31214117647058826,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9926,
      "step": 13266
    },
    {
      "epoch": 0.31216470588235296,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1606,
      "step": 13267
    },
    {
      "epoch": 0.31218823529411766,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.02,
      "loss": 0.5015,
      "step": 13268
    },
    {
      "epoch": 0.31221176470588236,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.178,
      "step": 13269
    },
    {
      "epoch": 0.31223529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9525,
      "step": 13270
    },
    {
      "epoch": 0.31225882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 13271
    },
    {
      "epoch": 0.31228235294117646,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 13272
    },
    {
      "epoch": 0.31230588235294116,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.6313,
      "step": 13273
    },
    {
      "epoch": 0.31232941176470586,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.4587,
      "step": 13274
    },
    {
      "epoch": 0.3123529411764706,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1541,
      "step": 13275
    },
    {
      "epoch": 0.3123764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0848,
      "step": 13276
    },
    {
      "epoch": 0.3124,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1828,
      "step": 13277
    },
    {
      "epoch": 0.3124235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0369,
      "step": 13278
    },
    {
      "epoch": 0.3124470588235294,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.72,
      "step": 13279
    },
    {
      "epoch": 0.3124705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0112,
      "step": 13280
    },
    {
      "epoch": 0.3124941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2145,
      "step": 13281
    },
    {
      "epoch": 0.3125176470588235,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7677,
      "step": 13282
    },
    {
      "epoch": 0.3125411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0474,
      "step": 13283
    },
    {
      "epoch": 0.31256470588235297,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0165,
      "step": 13284
    },
    {
      "epoch": 0.31258823529411767,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1405,
      "step": 13285
    },
    {
      "epoch": 0.31261176470588237,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 13286
    },
    {
      "epoch": 0.31263529411764707,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9334,
      "step": 13287
    },
    {
      "epoch": 0.31265882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1265,
      "step": 13288
    },
    {
      "epoch": 0.31268235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9606,
      "step": 13289
    },
    {
      "epoch": 0.31270588235294117,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0683,
      "step": 13290
    },
    {
      "epoch": 0.31272941176470587,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1086,
      "step": 13291
    },
    {
      "epoch": 0.31275294117647057,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0097,
      "step": 13292
    },
    {
      "epoch": 0.31277647058823527,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8621,
      "step": 13293
    },
    {
      "epoch": 0.3128,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9618,
      "step": 13294
    },
    {
      "epoch": 0.3128235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8502,
      "step": 13295
    },
    {
      "epoch": 0.3128470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8487,
      "step": 13296
    },
    {
      "epoch": 0.3128705882352941,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0404,
      "step": 13297
    },
    {
      "epoch": 0.3128941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1697,
      "step": 13298
    },
    {
      "epoch": 0.3129176470588235,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2618,
      "step": 13299
    },
    {
      "epoch": 0.3129411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0511,
      "step": 13300
    },
    {
      "epoch": 0.3129647058823529,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0346,
      "step": 13301
    },
    {
      "epoch": 0.3129882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9136,
      "step": 13302
    },
    {
      "epoch": 0.3130117647058824,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2505,
      "step": 13303
    },
    {
      "epoch": 0.3130352941176471,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.004,
      "step": 13304
    },
    {
      "epoch": 0.3130588235294118,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.5187,
      "step": 13305
    },
    {
      "epoch": 0.3130823529411765,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8123,
      "step": 13306
    },
    {
      "epoch": 0.3131058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0164,
      "step": 13307
    },
    {
      "epoch": 0.3131294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9642,
      "step": 13308
    },
    {
      "epoch": 0.3131529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0939,
      "step": 13309
    },
    {
      "epoch": 0.3131764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0636,
      "step": 13310
    },
    {
      "epoch": 0.3132,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0137,
      "step": 13311
    },
    {
      "epoch": 0.3132235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1926,
      "step": 13312
    },
    {
      "epoch": 0.31324705882352943,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1332,
      "step": 13313
    },
    {
      "epoch": 0.31327058823529413,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3047,
      "step": 13314
    },
    {
      "epoch": 0.31329411764705883,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.08,
      "step": 13315
    },
    {
      "epoch": 0.31331764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9147,
      "step": 13316
    },
    {
      "epoch": 0.31334117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0917,
      "step": 13317
    },
    {
      "epoch": 0.31336470588235293,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2479,
      "step": 13318
    },
    {
      "epoch": 0.31338823529411763,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1692,
      "step": 13319
    },
    {
      "epoch": 0.31341176470588233,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.807,
      "step": 13320
    },
    {
      "epoch": 0.31343529411764703,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9576,
      "step": 13321
    },
    {
      "epoch": 0.3134588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8545,
      "step": 13322
    },
    {
      "epoch": 0.3134823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0656,
      "step": 13323
    },
    {
      "epoch": 0.3135058823529412,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2396,
      "step": 13324
    },
    {
      "epoch": 0.3135294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1082,
      "step": 13325
    },
    {
      "epoch": 0.3135529411764706,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1818,
      "step": 13326
    },
    {
      "epoch": 0.3135764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9697,
      "step": 13327
    },
    {
      "epoch": 0.3136,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1736,
      "step": 13328
    },
    {
      "epoch": 0.3136235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4204,
      "step": 13329
    },
    {
      "epoch": 0.3136470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1757,
      "step": 13330
    },
    {
      "epoch": 0.3136705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1512,
      "step": 13331
    },
    {
      "epoch": 0.31369411764705885,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.203,
      "step": 13332
    },
    {
      "epoch": 0.31371764705882355,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1877,
      "step": 13333
    },
    {
      "epoch": 0.31374117647058825,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7569,
      "step": 13334
    },
    {
      "epoch": 0.31376470588235295,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8801,
      "step": 13335
    },
    {
      "epoch": 0.31378823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2919,
      "step": 13336
    },
    {
      "epoch": 0.31381176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9754,
      "step": 13337
    },
    {
      "epoch": 0.31383529411764705,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 13338
    },
    {
      "epoch": 0.31385882352941175,
      "grad_norm": 0.251953125,
      "learning_rate": 0.02,
      "loss": 1.0398,
      "step": 13339
    },
    {
      "epoch": 0.31388235294117645,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0016,
      "step": 13340
    },
    {
      "epoch": 0.3139058823529412,
      "grad_norm": 0.2421875,
      "learning_rate": 0.02,
      "loss": 0.2903,
      "step": 13341
    },
    {
      "epoch": 0.3139294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 13342
    },
    {
      "epoch": 0.3139529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.207,
      "step": 13343
    },
    {
      "epoch": 0.3139764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1333,
      "step": 13344
    },
    {
      "epoch": 0.314,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9613,
      "step": 13345
    },
    {
      "epoch": 0.3140235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0073,
      "step": 13346
    },
    {
      "epoch": 0.3140470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0509,
      "step": 13347
    },
    {
      "epoch": 0.3140705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0485,
      "step": 13348
    },
    {
      "epoch": 0.3140941176470588,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0016,
      "step": 13349
    },
    {
      "epoch": 0.31411764705882356,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6472,
      "step": 13350
    },
    {
      "epoch": 0.31414117647058826,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1033,
      "step": 13351
    },
    {
      "epoch": 0.31416470588235296,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6671,
      "step": 13352
    },
    {
      "epoch": 0.31418823529411766,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7475,
      "step": 13353
    },
    {
      "epoch": 0.31421176470588236,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9157,
      "step": 13354
    },
    {
      "epoch": 0.31423529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9717,
      "step": 13355
    },
    {
      "epoch": 0.31425882352941176,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0209,
      "step": 13356
    },
    {
      "epoch": 0.31428235294117646,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.047,
      "step": 13357
    },
    {
      "epoch": 0.31430588235294116,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0838,
      "step": 13358
    },
    {
      "epoch": 0.31432941176470586,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0693,
      "step": 13359
    },
    {
      "epoch": 0.3143529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1275,
      "step": 13360
    },
    {
      "epoch": 0.3143764705882353,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.6648,
      "step": 13361
    },
    {
      "epoch": 0.3144,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0967,
      "step": 13362
    },
    {
      "epoch": 0.3144235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8835,
      "step": 13363
    },
    {
      "epoch": 0.3144470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7232,
      "step": 13364
    },
    {
      "epoch": 0.3144705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8561,
      "step": 13365
    },
    {
      "epoch": 0.3144941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0754,
      "step": 13366
    },
    {
      "epoch": 0.3145176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 13367
    },
    {
      "epoch": 0.3145411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9672,
      "step": 13368
    },
    {
      "epoch": 0.31456470588235297,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9772,
      "step": 13369
    },
    {
      "epoch": 0.31458823529411767,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9901,
      "step": 13370
    },
    {
      "epoch": 0.31461176470588237,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0819,
      "step": 13371
    },
    {
      "epoch": 0.31463529411764707,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.038,
      "step": 13372
    },
    {
      "epoch": 0.31465882352941177,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0351,
      "step": 13373
    },
    {
      "epoch": 0.31468235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2095,
      "step": 13374
    },
    {
      "epoch": 0.31470588235294117,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1744,
      "step": 13375
    },
    {
      "epoch": 0.31472941176470587,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.202,
      "step": 13376
    },
    {
      "epoch": 0.31475294117647057,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9923,
      "step": 13377
    },
    {
      "epoch": 0.31477647058823527,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0641,
      "step": 13378
    },
    {
      "epoch": 0.3148,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8372,
      "step": 13379
    },
    {
      "epoch": 0.3148235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0452,
      "step": 13380
    },
    {
      "epoch": 0.3148470588235294,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2681,
      "step": 13381
    },
    {
      "epoch": 0.3148705882352941,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1782,
      "step": 13382
    },
    {
      "epoch": 0.3148941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1974,
      "step": 13383
    },
    {
      "epoch": 0.3149176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1177,
      "step": 13384
    },
    {
      "epoch": 0.3149411764705882,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.9994,
      "step": 13385
    },
    {
      "epoch": 0.3149647058823529,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2316,
      "step": 13386
    },
    {
      "epoch": 0.3149882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9118,
      "step": 13387
    },
    {
      "epoch": 0.3150117647058824,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.4985,
      "step": 13388
    },
    {
      "epoch": 0.3150352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 13389
    },
    {
      "epoch": 0.3150588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0003,
      "step": 13390
    },
    {
      "epoch": 0.3150823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.2243,
      "step": 13391
    },
    {
      "epoch": 0.3151058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 13392
    },
    {
      "epoch": 0.3151294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9996,
      "step": 13393
    },
    {
      "epoch": 0.3151529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9638,
      "step": 13394
    },
    {
      "epoch": 0.3151764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 1.0144,
      "step": 13395
    },
    {
      "epoch": 0.3152,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3949,
      "step": 13396
    },
    {
      "epoch": 0.3152235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9661,
      "step": 13397
    },
    {
      "epoch": 0.31524705882352944,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9076,
      "step": 13398
    },
    {
      "epoch": 0.31527058823529414,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2666,
      "step": 13399
    },
    {
      "epoch": 0.31529411764705884,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0561,
      "step": 13400
    },
    {
      "epoch": 0.31531764705882354,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.116,
      "step": 13401
    },
    {
      "epoch": 0.31534117647058824,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1664,
      "step": 13402
    },
    {
      "epoch": 0.31536470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8376,
      "step": 13403
    },
    {
      "epoch": 0.31538823529411764,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.454,
      "step": 13404
    },
    {
      "epoch": 0.31541176470588234,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.069,
      "step": 13405
    },
    {
      "epoch": 0.31543529411764704,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0263,
      "step": 13406
    },
    {
      "epoch": 0.3154588235294118,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.322,
      "step": 13407
    },
    {
      "epoch": 0.3154823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9602,
      "step": 13408
    },
    {
      "epoch": 0.3155058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9383,
      "step": 13409
    },
    {
      "epoch": 0.3155294117647059,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2292,
      "step": 13410
    },
    {
      "epoch": 0.3155529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7494,
      "step": 13411
    },
    {
      "epoch": 0.3155764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6918,
      "step": 13412
    },
    {
      "epoch": 0.3156,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8421,
      "step": 13413
    },
    {
      "epoch": 0.3156235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3438,
      "step": 13414
    },
    {
      "epoch": 0.3156470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0485,
      "step": 13415
    },
    {
      "epoch": 0.3156705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1717,
      "step": 13416
    },
    {
      "epoch": 0.31569411764705885,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8763,
      "step": 13417
    },
    {
      "epoch": 0.31571764705882355,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7509,
      "step": 13418
    },
    {
      "epoch": 0.31574117647058825,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 13419
    },
    {
      "epoch": 0.31576470588235295,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9188,
      "step": 13420
    },
    {
      "epoch": 0.31578823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1426,
      "step": 13421
    },
    {
      "epoch": 0.31581176470588235,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.531,
      "step": 13422
    },
    {
      "epoch": 0.31583529411764705,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2642,
      "step": 13423
    },
    {
      "epoch": 0.31585882352941175,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 13424
    },
    {
      "epoch": 0.31588235294117645,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1673,
      "step": 13425
    },
    {
      "epoch": 0.3159058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8642,
      "step": 13426
    },
    {
      "epoch": 0.3159294117647059,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1959,
      "step": 13427
    },
    {
      "epoch": 0.3159529411764706,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6548,
      "step": 13428
    },
    {
      "epoch": 0.3159764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8746,
      "step": 13429
    },
    {
      "epoch": 0.316,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8468,
      "step": 13430
    },
    {
      "epoch": 0.3160235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7963,
      "step": 13431
    },
    {
      "epoch": 0.3160470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0554,
      "step": 13432
    },
    {
      "epoch": 0.3160705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2766,
      "step": 13433
    },
    {
      "epoch": 0.3160941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1355,
      "step": 13434
    },
    {
      "epoch": 0.3161176470588235,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.6717,
      "step": 13435
    },
    {
      "epoch": 0.31614117647058826,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8483,
      "step": 13436
    },
    {
      "epoch": 0.31616470588235296,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9434,
      "step": 13437
    },
    {
      "epoch": 0.31618823529411766,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 13438
    },
    {
      "epoch": 0.31621176470588236,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9823,
      "step": 13439
    },
    {
      "epoch": 0.31623529411764706,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.6727,
      "step": 13440
    },
    {
      "epoch": 0.31625882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7532,
      "step": 13441
    },
    {
      "epoch": 0.31628235294117646,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9895,
      "step": 13442
    },
    {
      "epoch": 0.31630588235294116,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0838,
      "step": 13443
    },
    {
      "epoch": 0.31632941176470586,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8871,
      "step": 13444
    },
    {
      "epoch": 0.3163529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0216,
      "step": 13445
    },
    {
      "epoch": 0.3163764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0663,
      "step": 13446
    },
    {
      "epoch": 0.3164,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2339,
      "step": 13447
    },
    {
      "epoch": 0.3164235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0029,
      "step": 13448
    },
    {
      "epoch": 0.3164470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1646,
      "step": 13449
    },
    {
      "epoch": 0.3164705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1623,
      "step": 13450
    },
    {
      "epoch": 0.3164941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1697,
      "step": 13451
    },
    {
      "epoch": 0.3165176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8962,
      "step": 13452
    },
    {
      "epoch": 0.3165411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 13453
    },
    {
      "epoch": 0.3165647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1164,
      "step": 13454
    },
    {
      "epoch": 0.31658823529411767,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6782,
      "step": 13455
    },
    {
      "epoch": 0.31661176470588237,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7759,
      "step": 13456
    },
    {
      "epoch": 0.31663529411764707,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0538,
      "step": 13457
    },
    {
      "epoch": 0.31665882352941177,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9676,
      "step": 13458
    },
    {
      "epoch": 0.31668235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2673,
      "step": 13459
    },
    {
      "epoch": 0.31670588235294117,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0915,
      "step": 13460
    },
    {
      "epoch": 0.31672941176470587,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.0754,
      "step": 13461
    },
    {
      "epoch": 0.31675294117647057,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 13462
    },
    {
      "epoch": 0.31677647058823527,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9711,
      "step": 13463
    },
    {
      "epoch": 0.3168,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1632,
      "step": 13464
    },
    {
      "epoch": 0.3168235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3409,
      "step": 13465
    },
    {
      "epoch": 0.3168470588235294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7656,
      "step": 13466
    },
    {
      "epoch": 0.3168705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9807,
      "step": 13467
    },
    {
      "epoch": 0.3168941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9099,
      "step": 13468
    },
    {
      "epoch": 0.3169176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1941,
      "step": 13469
    },
    {
      "epoch": 0.3169411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8477,
      "step": 13470
    },
    {
      "epoch": 0.3169647058823529,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2741,
      "step": 13471
    },
    {
      "epoch": 0.3169882352941176,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8261,
      "step": 13472
    },
    {
      "epoch": 0.3170117647058823,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1443,
      "step": 13473
    },
    {
      "epoch": 0.3170352941176471,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.046,
      "step": 13474
    },
    {
      "epoch": 0.3170588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.7519,
      "step": 13475
    },
    {
      "epoch": 0.3170823529411765,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9106,
      "step": 13476
    },
    {
      "epoch": 0.3171058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8868,
      "step": 13477
    },
    {
      "epoch": 0.3171294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.934,
      "step": 13478
    },
    {
      "epoch": 0.3171529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0813,
      "step": 13479
    },
    {
      "epoch": 0.3171764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2549,
      "step": 13480
    },
    {
      "epoch": 0.3172,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1326,
      "step": 13481
    },
    {
      "epoch": 0.3172235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3126,
      "step": 13482
    },
    {
      "epoch": 0.31724705882352944,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0605,
      "step": 13483
    },
    {
      "epoch": 0.31727058823529414,
      "grad_norm": 4.9375,
      "learning_rate": 0.02,
      "loss": 1.1221,
      "step": 13484
    },
    {
      "epoch": 0.31729411764705884,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.4508,
      "step": 13485
    },
    {
      "epoch": 0.31731764705882354,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9209,
      "step": 13486
    },
    {
      "epoch": 0.31734117647058824,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0151,
      "step": 13487
    },
    {
      "epoch": 0.31736470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2049,
      "step": 13488
    },
    {
      "epoch": 0.31738823529411764,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9426,
      "step": 13489
    },
    {
      "epoch": 0.31741176470588234,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0663,
      "step": 13490
    },
    {
      "epoch": 0.31743529411764704,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0796,
      "step": 13491
    },
    {
      "epoch": 0.31745882352941174,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2293,
      "step": 13492
    },
    {
      "epoch": 0.3174823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2003,
      "step": 13493
    },
    {
      "epoch": 0.3175058823529412,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1475,
      "step": 13494
    },
    {
      "epoch": 0.3175294117647059,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.008,
      "step": 13495
    },
    {
      "epoch": 0.3175529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7832,
      "step": 13496
    },
    {
      "epoch": 0.3175764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.6665,
      "step": 13497
    },
    {
      "epoch": 0.3176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.93,
      "step": 13498
    },
    {
      "epoch": 0.3176235294117647,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.3379,
      "step": 13499
    },
    {
      "epoch": 0.3176470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0335,
      "step": 13500
    },
    {
      "epoch": 0.3176705882352941,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8366,
      "step": 13501
    },
    {
      "epoch": 0.31769411764705885,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2458,
      "step": 13502
    },
    {
      "epoch": 0.31771764705882355,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9567,
      "step": 13503
    },
    {
      "epoch": 0.31774117647058825,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1251,
      "step": 13504
    },
    {
      "epoch": 0.31776470588235295,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.013,
      "step": 13505
    },
    {
      "epoch": 0.31778823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0504,
      "step": 13506
    },
    {
      "epoch": 0.31781176470588235,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 13507
    },
    {
      "epoch": 0.31783529411764705,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9618,
      "step": 13508
    },
    {
      "epoch": 0.31785882352941175,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.649,
      "step": 13509
    },
    {
      "epoch": 0.31788235294117645,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0249,
      "step": 13510
    },
    {
      "epoch": 0.31790588235294115,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0308,
      "step": 13511
    },
    {
      "epoch": 0.3179294117647059,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2121,
      "step": 13512
    },
    {
      "epoch": 0.3179529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0001,
      "step": 13513
    },
    {
      "epoch": 0.3179764705882353,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.3885,
      "step": 13514
    },
    {
      "epoch": 0.318,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2863,
      "step": 13515
    },
    {
      "epoch": 0.3180235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0389,
      "step": 13516
    },
    {
      "epoch": 0.3180470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8869,
      "step": 13517
    },
    {
      "epoch": 0.3180705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.129,
      "step": 13518
    },
    {
      "epoch": 0.3180941176470588,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.233,
      "step": 13519
    },
    {
      "epoch": 0.3181176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9534,
      "step": 13520
    },
    {
      "epoch": 0.31814117647058826,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1412,
      "step": 13521
    },
    {
      "epoch": 0.31816470588235296,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.762,
      "step": 13522
    },
    {
      "epoch": 0.31818823529411766,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3369,
      "step": 13523
    },
    {
      "epoch": 0.31821176470588236,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1807,
      "step": 13524
    },
    {
      "epoch": 0.31823529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8719,
      "step": 13525
    },
    {
      "epoch": 0.31825882352941176,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9958,
      "step": 13526
    },
    {
      "epoch": 0.31828235294117646,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.994,
      "step": 13527
    },
    {
      "epoch": 0.31830588235294116,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8732,
      "step": 13528
    },
    {
      "epoch": 0.31832941176470586,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3488,
      "step": 13529
    },
    {
      "epoch": 0.31835294117647056,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.4585,
      "step": 13530
    },
    {
      "epoch": 0.3183764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 13531
    },
    {
      "epoch": 0.3184,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4478,
      "step": 13532
    },
    {
      "epoch": 0.3184235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0832,
      "step": 13533
    },
    {
      "epoch": 0.3184470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0691,
      "step": 13534
    },
    {
      "epoch": 0.3184705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7402,
      "step": 13535
    },
    {
      "epoch": 0.3184941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3144,
      "step": 13536
    },
    {
      "epoch": 0.3185176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8331,
      "step": 13537
    },
    {
      "epoch": 0.3185411764705882,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9518,
      "step": 13538
    },
    {
      "epoch": 0.3185647058823529,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 13539
    },
    {
      "epoch": 0.31858823529411767,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.8405,
      "step": 13540
    },
    {
      "epoch": 0.31861176470588237,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 1.0774,
      "step": 13541
    },
    {
      "epoch": 0.31863529411764707,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.8371,
      "step": 13542
    },
    {
      "epoch": 0.31865882352941177,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8476,
      "step": 13543
    },
    {
      "epoch": 0.31868235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1699,
      "step": 13544
    },
    {
      "epoch": 0.31870588235294117,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0118,
      "step": 13545
    },
    {
      "epoch": 0.31872941176470587,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2835,
      "step": 13546
    },
    {
      "epoch": 0.31875294117647057,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8984,
      "step": 13547
    },
    {
      "epoch": 0.3187764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1854,
      "step": 13548
    },
    {
      "epoch": 0.3188,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0907,
      "step": 13549
    },
    {
      "epoch": 0.31882352941176473,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1878,
      "step": 13550
    },
    {
      "epoch": 0.31884705882352943,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9062,
      "step": 13551
    },
    {
      "epoch": 0.31887058823529413,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2817,
      "step": 13552
    },
    {
      "epoch": 0.31889411764705883,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9942,
      "step": 13553
    },
    {
      "epoch": 0.31891764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8897,
      "step": 13554
    },
    {
      "epoch": 0.31894117647058823,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9876,
      "step": 13555
    },
    {
      "epoch": 0.31896470588235293,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0472,
      "step": 13556
    },
    {
      "epoch": 0.31898823529411763,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7428,
      "step": 13557
    },
    {
      "epoch": 0.31901176470588233,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9032,
      "step": 13558
    },
    {
      "epoch": 0.3190352941176471,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1118,
      "step": 13559
    },
    {
      "epoch": 0.3190588235294118,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7985,
      "step": 13560
    },
    {
      "epoch": 0.3190823529411765,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.6941,
      "step": 13561
    },
    {
      "epoch": 0.3191058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8438,
      "step": 13562
    },
    {
      "epoch": 0.3191294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8341,
      "step": 13563
    },
    {
      "epoch": 0.3191529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9913,
      "step": 13564
    },
    {
      "epoch": 0.3191764705882353,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.7387,
      "step": 13565
    },
    {
      "epoch": 0.3192,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3856,
      "step": 13566
    },
    {
      "epoch": 0.3192235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1867,
      "step": 13567
    },
    {
      "epoch": 0.31924705882352944,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8568,
      "step": 13568
    },
    {
      "epoch": 0.31927058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 13569
    },
    {
      "epoch": 0.31929411764705884,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8485,
      "step": 13570
    },
    {
      "epoch": 0.31931764705882354,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.137,
      "step": 13571
    },
    {
      "epoch": 0.31934117647058824,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0005,
      "step": 13572
    },
    {
      "epoch": 0.31936470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7491,
      "step": 13573
    },
    {
      "epoch": 0.31938823529411764,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2516,
      "step": 13574
    },
    {
      "epoch": 0.31941176470588234,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8478,
      "step": 13575
    },
    {
      "epoch": 0.31943529411764704,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2181,
      "step": 13576
    },
    {
      "epoch": 0.31945882352941174,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7902,
      "step": 13577
    },
    {
      "epoch": 0.3194823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0845,
      "step": 13578
    },
    {
      "epoch": 0.3195058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2464,
      "step": 13579
    },
    {
      "epoch": 0.3195294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7894,
      "step": 13580
    },
    {
      "epoch": 0.3195529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0104,
      "step": 13581
    },
    {
      "epoch": 0.3195764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.2164,
      "step": 13582
    },
    {
      "epoch": 0.3196,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7547,
      "step": 13583
    },
    {
      "epoch": 0.3196235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7093,
      "step": 13584
    },
    {
      "epoch": 0.3196470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8706,
      "step": 13585
    },
    {
      "epoch": 0.3196705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1532,
      "step": 13586
    },
    {
      "epoch": 0.31969411764705885,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3812,
      "step": 13587
    },
    {
      "epoch": 0.31971764705882355,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9059,
      "step": 13588
    },
    {
      "epoch": 0.31974117647058825,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2763,
      "step": 13589
    },
    {
      "epoch": 0.31976470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7772,
      "step": 13590
    },
    {
      "epoch": 0.31978823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0088,
      "step": 13591
    },
    {
      "epoch": 0.31981176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.072,
      "step": 13592
    },
    {
      "epoch": 0.31983529411764705,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9559,
      "step": 13593
    },
    {
      "epoch": 0.31985882352941175,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0141,
      "step": 13594
    },
    {
      "epoch": 0.31988235294117645,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9213,
      "step": 13595
    },
    {
      "epoch": 0.31990588235294115,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.081,
      "step": 13596
    },
    {
      "epoch": 0.3199294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7794,
      "step": 13597
    },
    {
      "epoch": 0.3199529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0003,
      "step": 13598
    },
    {
      "epoch": 0.3199764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.205,
      "step": 13599
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1681,
      "step": 13600
    },
    {
      "epoch": 0.3200235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2264,
      "step": 13601
    },
    {
      "epoch": 0.3200470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.214,
      "step": 13602
    },
    {
      "epoch": 0.3200705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0568,
      "step": 13603
    },
    {
      "epoch": 0.3200941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0655,
      "step": 13604
    },
    {
      "epoch": 0.3201176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8505,
      "step": 13605
    },
    {
      "epoch": 0.32014117647058826,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9703,
      "step": 13606
    },
    {
      "epoch": 0.32016470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7744,
      "step": 13607
    },
    {
      "epoch": 0.32018823529411766,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7033,
      "step": 13608
    },
    {
      "epoch": 0.32021176470588236,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.891,
      "step": 13609
    },
    {
      "epoch": 0.32023529411764706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 13610
    },
    {
      "epoch": 0.32025882352941176,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3586,
      "step": 13611
    },
    {
      "epoch": 0.32028235294117646,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.6609,
      "step": 13612
    },
    {
      "epoch": 0.32030588235294116,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9717,
      "step": 13613
    },
    {
      "epoch": 0.32032941176470586,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9038,
      "step": 13614
    },
    {
      "epoch": 0.32035294117647056,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1344,
      "step": 13615
    },
    {
      "epoch": 0.3203764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0006,
      "step": 13616
    },
    {
      "epoch": 0.3204,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7438,
      "step": 13617
    },
    {
      "epoch": 0.3204235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.043,
      "step": 13618
    },
    {
      "epoch": 0.3204470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.24,
      "step": 13619
    },
    {
      "epoch": 0.3204705882352941,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6767,
      "step": 13620
    },
    {
      "epoch": 0.3204941176470588,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1088,
      "step": 13621
    },
    {
      "epoch": 0.3205176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9959,
      "step": 13622
    },
    {
      "epoch": 0.3205411764705882,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8794,
      "step": 13623
    },
    {
      "epoch": 0.3205647058823529,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9387,
      "step": 13624
    },
    {
      "epoch": 0.3205882352941177,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1727,
      "step": 13625
    },
    {
      "epoch": 0.3206117647058824,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.758,
      "step": 13626
    },
    {
      "epoch": 0.3206352941176471,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1309,
      "step": 13627
    },
    {
      "epoch": 0.3206588235294118,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2539,
      "step": 13628
    },
    {
      "epoch": 0.3206823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 13629
    },
    {
      "epoch": 0.3207058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.09,
      "step": 13630
    },
    {
      "epoch": 0.3207294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 13631
    },
    {
      "epoch": 0.3207529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0488,
      "step": 13632
    },
    {
      "epoch": 0.3207764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1945,
      "step": 13633
    },
    {
      "epoch": 0.3208,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2164,
      "step": 13634
    },
    {
      "epoch": 0.32082352941176473,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1882,
      "step": 13635
    },
    {
      "epoch": 0.32084705882352943,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3934,
      "step": 13636
    },
    {
      "epoch": 0.32087058823529413,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2357,
      "step": 13637
    },
    {
      "epoch": 0.32089411764705883,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7631,
      "step": 13638
    },
    {
      "epoch": 0.32091764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1826,
      "step": 13639
    },
    {
      "epoch": 0.32094117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0361,
      "step": 13640
    },
    {
      "epoch": 0.32096470588235293,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9105,
      "step": 13641
    },
    {
      "epoch": 0.32098823529411763,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9653,
      "step": 13642
    },
    {
      "epoch": 0.32101176470588233,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1732,
      "step": 13643
    },
    {
      "epoch": 0.3210352941176471,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1026,
      "step": 13644
    },
    {
      "epoch": 0.3210588235294118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9077,
      "step": 13645
    },
    {
      "epoch": 0.3210823529411765,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7371,
      "step": 13646
    },
    {
      "epoch": 0.3211058823529412,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.237,
      "step": 13647
    },
    {
      "epoch": 0.3211294117647059,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0575,
      "step": 13648
    },
    {
      "epoch": 0.3211529411764706,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.8348,
      "step": 13649
    },
    {
      "epoch": 0.3211764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8924,
      "step": 13650
    },
    {
      "epoch": 0.3212,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6874,
      "step": 13651
    },
    {
      "epoch": 0.3212235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0873,
      "step": 13652
    },
    {
      "epoch": 0.3212470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1721,
      "step": 13653
    },
    {
      "epoch": 0.32127058823529414,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8552,
      "step": 13654
    },
    {
      "epoch": 0.32129411764705884,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3781,
      "step": 13655
    },
    {
      "epoch": 0.32131764705882354,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3569,
      "step": 13656
    },
    {
      "epoch": 0.32134117647058824,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1034,
      "step": 13657
    },
    {
      "epoch": 0.32136470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9429,
      "step": 13658
    },
    {
      "epoch": 0.32138823529411764,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.5361,
      "step": 13659
    },
    {
      "epoch": 0.32141176470588234,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 13660
    },
    {
      "epoch": 0.32143529411764704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0812,
      "step": 13661
    },
    {
      "epoch": 0.32145882352941174,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.878,
      "step": 13662
    },
    {
      "epoch": 0.3214823529411765,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.6903,
      "step": 13663
    },
    {
      "epoch": 0.3215058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9942,
      "step": 13664
    },
    {
      "epoch": 0.3215294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2752,
      "step": 13665
    },
    {
      "epoch": 0.3215529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.108,
      "step": 13666
    },
    {
      "epoch": 0.3215764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9112,
      "step": 13667
    },
    {
      "epoch": 0.3216,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7464,
      "step": 13668
    },
    {
      "epoch": 0.3216235294117647,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8373,
      "step": 13669
    },
    {
      "epoch": 0.3216470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0914,
      "step": 13670
    },
    {
      "epoch": 0.3216705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1572,
      "step": 13671
    },
    {
      "epoch": 0.3216941176470588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0296,
      "step": 13672
    },
    {
      "epoch": 0.32171764705882355,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0607,
      "step": 13673
    },
    {
      "epoch": 0.32174117647058825,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8652,
      "step": 13674
    },
    {
      "epoch": 0.32176470588235295,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9526,
      "step": 13675
    },
    {
      "epoch": 0.32178823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.97,
      "step": 13676
    },
    {
      "epoch": 0.32181176470588235,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0496,
      "step": 13677
    },
    {
      "epoch": 0.32183529411764705,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7572,
      "step": 13678
    },
    {
      "epoch": 0.32185882352941175,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2396,
      "step": 13679
    },
    {
      "epoch": 0.32188235294117645,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9263,
      "step": 13680
    },
    {
      "epoch": 0.32190588235294115,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9945,
      "step": 13681
    },
    {
      "epoch": 0.3219294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.59,
      "step": 13682
    },
    {
      "epoch": 0.3219529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8792,
      "step": 13683
    },
    {
      "epoch": 0.3219764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.5969,
      "step": 13684
    },
    {
      "epoch": 0.322,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9267,
      "step": 13685
    },
    {
      "epoch": 0.3220235294117647,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.6274,
      "step": 13686
    },
    {
      "epoch": 0.3220470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9587,
      "step": 13687
    },
    {
      "epoch": 0.3220705882352941,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2867,
      "step": 13688
    },
    {
      "epoch": 0.3220941176470588,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.4354,
      "step": 13689
    },
    {
      "epoch": 0.3221176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9362,
      "step": 13690
    },
    {
      "epoch": 0.3221411764705882,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0503,
      "step": 13691
    },
    {
      "epoch": 0.32216470588235296,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1795,
      "step": 13692
    },
    {
      "epoch": 0.32218823529411766,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0459,
      "step": 13693
    },
    {
      "epoch": 0.32221176470588236,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 13694
    },
    {
      "epoch": 0.32223529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2051,
      "step": 13695
    },
    {
      "epoch": 0.32225882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8556,
      "step": 13696
    },
    {
      "epoch": 0.32228235294117646,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1227,
      "step": 13697
    },
    {
      "epoch": 0.32230588235294116,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1394,
      "step": 13698
    },
    {
      "epoch": 0.32232941176470586,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8643,
      "step": 13699
    },
    {
      "epoch": 0.32235294117647056,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1059,
      "step": 13700
    },
    {
      "epoch": 0.3223764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9923,
      "step": 13701
    },
    {
      "epoch": 0.3224,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8661,
      "step": 13702
    },
    {
      "epoch": 0.3224235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8367,
      "step": 13703
    },
    {
      "epoch": 0.3224470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0639,
      "step": 13704
    },
    {
      "epoch": 0.3224705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8573,
      "step": 13705
    },
    {
      "epoch": 0.3224941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9094,
      "step": 13706
    },
    {
      "epoch": 0.3225176470588235,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.6792,
      "step": 13707
    },
    {
      "epoch": 0.3225411764705882,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.051,
      "step": 13708
    },
    {
      "epoch": 0.3225647058823529,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0788,
      "step": 13709
    },
    {
      "epoch": 0.3225882352941176,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9953,
      "step": 13710
    },
    {
      "epoch": 0.3226117647058824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9529,
      "step": 13711
    },
    {
      "epoch": 0.3226352941176471,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3267,
      "step": 13712
    },
    {
      "epoch": 0.3226588235294118,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8007,
      "step": 13713
    },
    {
      "epoch": 0.3226823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0162,
      "step": 13714
    },
    {
      "epoch": 0.3227058823529412,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.096,
      "step": 13715
    },
    {
      "epoch": 0.3227294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9084,
      "step": 13716
    },
    {
      "epoch": 0.3227529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.1571,
      "step": 13717
    },
    {
      "epoch": 0.3227764705882353,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8785,
      "step": 13718
    },
    {
      "epoch": 0.3228,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.6539,
      "step": 13719
    },
    {
      "epoch": 0.32282352941176473,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0655,
      "step": 13720
    },
    {
      "epoch": 0.32284705882352943,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1916,
      "step": 13721
    },
    {
      "epoch": 0.32287058823529413,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9535,
      "step": 13722
    },
    {
      "epoch": 0.32289411764705883,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1999,
      "step": 13723
    },
    {
      "epoch": 0.32291764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8585,
      "step": 13724
    },
    {
      "epoch": 0.32294117647058823,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9389,
      "step": 13725
    },
    {
      "epoch": 0.32296470588235293,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7364,
      "step": 13726
    },
    {
      "epoch": 0.32298823529411763,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1822,
      "step": 13727
    },
    {
      "epoch": 0.32301176470588233,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0167,
      "step": 13728
    },
    {
      "epoch": 0.32303529411764703,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9884,
      "step": 13729
    },
    {
      "epoch": 0.3230588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0332,
      "step": 13730
    },
    {
      "epoch": 0.3230823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9703,
      "step": 13731
    },
    {
      "epoch": 0.3231058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0208,
      "step": 13732
    },
    {
      "epoch": 0.3231294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1844,
      "step": 13733
    },
    {
      "epoch": 0.3231529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7648,
      "step": 13734
    },
    {
      "epoch": 0.3231764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7906,
      "step": 13735
    },
    {
      "epoch": 0.3232,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1052,
      "step": 13736
    },
    {
      "epoch": 0.3232235294117647,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1525,
      "step": 13737
    },
    {
      "epoch": 0.3232470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2578,
      "step": 13738
    },
    {
      "epoch": 0.32327058823529414,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1667,
      "step": 13739
    },
    {
      "epoch": 0.32329411764705884,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9869,
      "step": 13740
    },
    {
      "epoch": 0.32331764705882354,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7588,
      "step": 13741
    },
    {
      "epoch": 0.32334117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8626,
      "step": 13742
    },
    {
      "epoch": 0.32336470588235294,
      "grad_norm": 0.236328125,
      "learning_rate": 0.02,
      "loss": 0.5378,
      "step": 13743
    },
    {
      "epoch": 0.32338823529411764,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1753,
      "step": 13744
    },
    {
      "epoch": 0.32341176470588234,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0452,
      "step": 13745
    },
    {
      "epoch": 0.32343529411764704,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7923,
      "step": 13746
    },
    {
      "epoch": 0.32345882352941174,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9558,
      "step": 13747
    },
    {
      "epoch": 0.32348235294117644,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8322,
      "step": 13748
    },
    {
      "epoch": 0.3235058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1624,
      "step": 13749
    },
    {
      "epoch": 0.3235294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 13750
    },
    {
      "epoch": 0.3235529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0805,
      "step": 13751
    },
    {
      "epoch": 0.3235764705882353,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.3494,
      "step": 13752
    },
    {
      "epoch": 0.3236,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0148,
      "step": 13753
    },
    {
      "epoch": 0.3236235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7484,
      "step": 13754
    },
    {
      "epoch": 0.3236470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0018,
      "step": 13755
    },
    {
      "epoch": 0.3236705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9828,
      "step": 13756
    },
    {
      "epoch": 0.3236941176470588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 13757
    },
    {
      "epoch": 0.32371764705882355,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8206,
      "step": 13758
    },
    {
      "epoch": 0.32374117647058825,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.847,
      "step": 13759
    },
    {
      "epoch": 0.32376470588235295,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.5032,
      "step": 13760
    },
    {
      "epoch": 0.32378823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1757,
      "step": 13761
    },
    {
      "epoch": 0.32381176470588235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0853,
      "step": 13762
    },
    {
      "epoch": 0.32383529411764705,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9684,
      "step": 13763
    },
    {
      "epoch": 0.32385882352941175,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8859,
      "step": 13764
    },
    {
      "epoch": 0.32388235294117645,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9732,
      "step": 13765
    },
    {
      "epoch": 0.32390588235294115,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9974,
      "step": 13766
    },
    {
      "epoch": 0.32392941176470585,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0821,
      "step": 13767
    },
    {
      "epoch": 0.3239529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9551,
      "step": 13768
    },
    {
      "epoch": 0.3239764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1453,
      "step": 13769
    },
    {
      "epoch": 0.324,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2377,
      "step": 13770
    },
    {
      "epoch": 0.3240235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.328,
      "step": 13771
    },
    {
      "epoch": 0.3240470588235294,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.93,
      "step": 13772
    },
    {
      "epoch": 0.3240705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8425,
      "step": 13773
    },
    {
      "epoch": 0.3240941176470588,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0109,
      "step": 13774
    },
    {
      "epoch": 0.3241176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.9295,
      "step": 13775
    },
    {
      "epoch": 0.3241411764705882,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6719,
      "step": 13776
    },
    {
      "epoch": 0.32416470588235297,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1707,
      "step": 13777
    },
    {
      "epoch": 0.32418823529411767,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.842,
      "step": 13778
    },
    {
      "epoch": 0.32421176470588237,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8026,
      "step": 13779
    },
    {
      "epoch": 0.32423529411764707,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9647,
      "step": 13780
    },
    {
      "epoch": 0.32425882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9513,
      "step": 13781
    },
    {
      "epoch": 0.32428235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0915,
      "step": 13782
    },
    {
      "epoch": 0.32430588235294117,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1566,
      "step": 13783
    },
    {
      "epoch": 0.32432941176470587,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9996,
      "step": 13784
    },
    {
      "epoch": 0.32435294117647057,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9625,
      "step": 13785
    },
    {
      "epoch": 0.3243764705882353,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.4839,
      "step": 13786
    },
    {
      "epoch": 0.3244,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 13787
    },
    {
      "epoch": 0.3244235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0927,
      "step": 13788
    },
    {
      "epoch": 0.3244470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0482,
      "step": 13789
    },
    {
      "epoch": 0.3244705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8389,
      "step": 13790
    },
    {
      "epoch": 0.3244941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.934,
      "step": 13791
    },
    {
      "epoch": 0.3245176470588235,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.2108,
      "step": 13792
    },
    {
      "epoch": 0.3245411764705882,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2087,
      "step": 13793
    },
    {
      "epoch": 0.3245647058823529,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8985,
      "step": 13794
    },
    {
      "epoch": 0.3245882352941176,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1602,
      "step": 13795
    },
    {
      "epoch": 0.3246117647058824,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.618,
      "step": 13796
    },
    {
      "epoch": 0.3246352941176471,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 13797
    },
    {
      "epoch": 0.3246588235294118,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.6162,
      "step": 13798
    },
    {
      "epoch": 0.3246823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8619,
      "step": 13799
    },
    {
      "epoch": 0.3247058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1926,
      "step": 13800
    },
    {
      "epoch": 0.3247294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0641,
      "step": 13801
    },
    {
      "epoch": 0.3247529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0014,
      "step": 13802
    },
    {
      "epoch": 0.3247764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0265,
      "step": 13803
    },
    {
      "epoch": 0.3248,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1396,
      "step": 13804
    },
    {
      "epoch": 0.32482352941176473,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2397,
      "step": 13805
    },
    {
      "epoch": 0.32484705882352943,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8879,
      "step": 13806
    },
    {
      "epoch": 0.32487058823529413,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8126,
      "step": 13807
    },
    {
      "epoch": 0.32489411764705883,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0333,
      "step": 13808
    },
    {
      "epoch": 0.32491764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8182,
      "step": 13809
    },
    {
      "epoch": 0.32494117647058823,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2056,
      "step": 13810
    },
    {
      "epoch": 0.32496470588235293,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1433,
      "step": 13811
    },
    {
      "epoch": 0.32498823529411763,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0066,
      "step": 13812
    },
    {
      "epoch": 0.32501176470588233,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8325,
      "step": 13813
    },
    {
      "epoch": 0.32503529411764703,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2743,
      "step": 13814
    },
    {
      "epoch": 0.3250588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0427,
      "step": 13815
    },
    {
      "epoch": 0.3250823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0267,
      "step": 13816
    },
    {
      "epoch": 0.3251058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0535,
      "step": 13817
    },
    {
      "epoch": 0.3251294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9386,
      "step": 13818
    },
    {
      "epoch": 0.3251529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0009,
      "step": 13819
    },
    {
      "epoch": 0.3251764705882353,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.175,
      "step": 13820
    },
    {
      "epoch": 0.3252,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0224,
      "step": 13821
    },
    {
      "epoch": 0.3252235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0938,
      "step": 13822
    },
    {
      "epoch": 0.3252470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7314,
      "step": 13823
    },
    {
      "epoch": 0.32527058823529414,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0129,
      "step": 13824
    },
    {
      "epoch": 0.32529411764705884,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0512,
      "step": 13825
    },
    {
      "epoch": 0.32531764705882354,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9416,
      "step": 13826
    },
    {
      "epoch": 0.32534117647058824,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 0.9714,
      "step": 13827
    },
    {
      "epoch": 0.32536470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1246,
      "step": 13828
    },
    {
      "epoch": 0.32538823529411764,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0052,
      "step": 13829
    },
    {
      "epoch": 0.32541176470588234,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0103,
      "step": 13830
    },
    {
      "epoch": 0.32543529411764704,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.5828,
      "step": 13831
    },
    {
      "epoch": 0.32545882352941174,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0096,
      "step": 13832
    },
    {
      "epoch": 0.32548235294117644,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7424,
      "step": 13833
    },
    {
      "epoch": 0.3255058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1583,
      "step": 13834
    },
    {
      "epoch": 0.3255294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1228,
      "step": 13835
    },
    {
      "epoch": 0.3255529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0064,
      "step": 13836
    },
    {
      "epoch": 0.3255764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0165,
      "step": 13837
    },
    {
      "epoch": 0.3256,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8447,
      "step": 13838
    },
    {
      "epoch": 0.3256235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9802,
      "step": 13839
    },
    {
      "epoch": 0.3256470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0461,
      "step": 13840
    },
    {
      "epoch": 0.3256705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0447,
      "step": 13841
    },
    {
      "epoch": 0.3256941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1328,
      "step": 13842
    },
    {
      "epoch": 0.32571764705882356,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.22,
      "step": 13843
    },
    {
      "epoch": 0.32574117647058826,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0139,
      "step": 13844
    },
    {
      "epoch": 0.32576470588235296,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8776,
      "step": 13845
    },
    {
      "epoch": 0.32578823529411766,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9389,
      "step": 13846
    },
    {
      "epoch": 0.32581176470588236,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1798,
      "step": 13847
    },
    {
      "epoch": 0.32583529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0315,
      "step": 13848
    },
    {
      "epoch": 0.32585882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8525,
      "step": 13849
    },
    {
      "epoch": 0.32588235294117646,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2775,
      "step": 13850
    },
    {
      "epoch": 0.32590588235294116,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9617,
      "step": 13851
    },
    {
      "epoch": 0.32592941176470586,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8927,
      "step": 13852
    },
    {
      "epoch": 0.3259529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.887,
      "step": 13853
    },
    {
      "epoch": 0.3259764705882353,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1215,
      "step": 13854
    },
    {
      "epoch": 0.326,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.062,
      "step": 13855
    },
    {
      "epoch": 0.3260235294117647,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1756,
      "step": 13856
    },
    {
      "epoch": 0.3260470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1742,
      "step": 13857
    },
    {
      "epoch": 0.3260705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0771,
      "step": 13858
    },
    {
      "epoch": 0.3260941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1532,
      "step": 13859
    },
    {
      "epoch": 0.3261176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0307,
      "step": 13860
    },
    {
      "epoch": 0.3261411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0603,
      "step": 13861
    },
    {
      "epoch": 0.32616470588235297,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0656,
      "step": 13862
    },
    {
      "epoch": 0.32618823529411767,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6908,
      "step": 13863
    },
    {
      "epoch": 0.32621176470588237,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1045,
      "step": 13864
    },
    {
      "epoch": 0.32623529411764707,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9149,
      "step": 13865
    },
    {
      "epoch": 0.32625882352941177,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.577,
      "step": 13866
    },
    {
      "epoch": 0.32628235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1538,
      "step": 13867
    },
    {
      "epoch": 0.32630588235294117,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0541,
      "step": 13868
    },
    {
      "epoch": 0.32632941176470587,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0142,
      "step": 13869
    },
    {
      "epoch": 0.32635294117647057,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1759,
      "step": 13870
    },
    {
      "epoch": 0.32637647058823527,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3515,
      "step": 13871
    },
    {
      "epoch": 0.3264,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8772,
      "step": 13872
    },
    {
      "epoch": 0.3264235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.328,
      "step": 13873
    },
    {
      "epoch": 0.3264470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9275,
      "step": 13874
    },
    {
      "epoch": 0.3264705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.7853,
      "step": 13875
    },
    {
      "epoch": 0.3264941176470588,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.3796,
      "step": 13876
    },
    {
      "epoch": 0.3265176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8318,
      "step": 13877
    },
    {
      "epoch": 0.3265411764705882,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2925,
      "step": 13878
    },
    {
      "epoch": 0.3265647058823529,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0989,
      "step": 13879
    },
    {
      "epoch": 0.3265882352941176,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1146,
      "step": 13880
    },
    {
      "epoch": 0.3266117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0055,
      "step": 13881
    },
    {
      "epoch": 0.3266352941176471,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0631,
      "step": 13882
    },
    {
      "epoch": 0.3266588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 13883
    },
    {
      "epoch": 0.3266823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0922,
      "step": 13884
    },
    {
      "epoch": 0.3267058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1958,
      "step": 13885
    },
    {
      "epoch": 0.3267294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0648,
      "step": 13886
    },
    {
      "epoch": 0.3267529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2299,
      "step": 13887
    },
    {
      "epoch": 0.3267764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0328,
      "step": 13888
    },
    {
      "epoch": 0.3268,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2643,
      "step": 13889
    },
    {
      "epoch": 0.3268235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7249,
      "step": 13890
    },
    {
      "epoch": 0.32684705882352943,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0186,
      "step": 13891
    },
    {
      "epoch": 0.32687058823529413,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2126,
      "step": 13892
    },
    {
      "epoch": 0.32689411764705884,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.9009,
      "step": 13893
    },
    {
      "epoch": 0.32691764705882354,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2195,
      "step": 13894
    },
    {
      "epoch": 0.32694117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9197,
      "step": 13895
    },
    {
      "epoch": 0.32696470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8834,
      "step": 13896
    },
    {
      "epoch": 0.32698823529411764,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7511,
      "step": 13897
    },
    {
      "epoch": 0.32701176470588234,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0468,
      "step": 13898
    },
    {
      "epoch": 0.32703529411764704,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1704,
      "step": 13899
    },
    {
      "epoch": 0.3270588235294118,
      "grad_norm": 0.2265625,
      "learning_rate": 0.02,
      "loss": 0.8028,
      "step": 13900
    },
    {
      "epoch": 0.3270823529411765,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.6665,
      "step": 13901
    },
    {
      "epoch": 0.3271058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0947,
      "step": 13902
    },
    {
      "epoch": 0.3271294117647059,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2838,
      "step": 13903
    },
    {
      "epoch": 0.3271529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2398,
      "step": 13904
    },
    {
      "epoch": 0.3271764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.6469,
      "step": 13905
    },
    {
      "epoch": 0.3272,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7866,
      "step": 13906
    },
    {
      "epoch": 0.3272235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9907,
      "step": 13907
    },
    {
      "epoch": 0.3272470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0838,
      "step": 13908
    },
    {
      "epoch": 0.3272705882352941,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0677,
      "step": 13909
    },
    {
      "epoch": 0.32729411764705885,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9107,
      "step": 13910
    },
    {
      "epoch": 0.32731764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0966,
      "step": 13911
    },
    {
      "epoch": 0.32734117647058825,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0749,
      "step": 13912
    },
    {
      "epoch": 0.32736470588235295,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9583,
      "step": 13913
    },
    {
      "epoch": 0.32738823529411765,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1723,
      "step": 13914
    },
    {
      "epoch": 0.32741176470588235,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8269,
      "step": 13915
    },
    {
      "epoch": 0.32743529411764705,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.7339,
      "step": 13916
    },
    {
      "epoch": 0.32745882352941175,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2305,
      "step": 13917
    },
    {
      "epoch": 0.32748235294117645,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0484,
      "step": 13918
    },
    {
      "epoch": 0.3275058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.979,
      "step": 13919
    },
    {
      "epoch": 0.3275294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9808,
      "step": 13920
    },
    {
      "epoch": 0.3275529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.046,
      "step": 13921
    },
    {
      "epoch": 0.3275764705882353,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7671,
      "step": 13922
    },
    {
      "epoch": 0.3276,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1081,
      "step": 13923
    },
    {
      "epoch": 0.3276235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 13924
    },
    {
      "epoch": 0.3276470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9901,
      "step": 13925
    },
    {
      "epoch": 0.3276705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8885,
      "step": 13926
    },
    {
      "epoch": 0.3276941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0858,
      "step": 13927
    },
    {
      "epoch": 0.3277176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0661,
      "step": 13928
    },
    {
      "epoch": 0.32774117647058826,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.911,
      "step": 13929
    },
    {
      "epoch": 0.32776470588235296,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1558,
      "step": 13930
    },
    {
      "epoch": 0.32778823529411766,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9599,
      "step": 13931
    },
    {
      "epoch": 0.32781176470588236,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.4301,
      "step": 13932
    },
    {
      "epoch": 0.32783529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9045,
      "step": 13933
    },
    {
      "epoch": 0.32785882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0901,
      "step": 13934
    },
    {
      "epoch": 0.32788235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8164,
      "step": 13935
    },
    {
      "epoch": 0.32790588235294116,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9317,
      "step": 13936
    },
    {
      "epoch": 0.32792941176470586,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3086,
      "step": 13937
    },
    {
      "epoch": 0.3279529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 13938
    },
    {
      "epoch": 0.3279764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0084,
      "step": 13939
    },
    {
      "epoch": 0.328,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1318,
      "step": 13940
    },
    {
      "epoch": 0.3280235294117647,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.002,
      "step": 13941
    },
    {
      "epoch": 0.3280470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9798,
      "step": 13942
    },
    {
      "epoch": 0.3280705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.97,
      "step": 13943
    },
    {
      "epoch": 0.3280941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2798,
      "step": 13944
    },
    {
      "epoch": 0.3281176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9603,
      "step": 13945
    },
    {
      "epoch": 0.3281411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8837,
      "step": 13946
    },
    {
      "epoch": 0.3281647058823529,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9841,
      "step": 13947
    },
    {
      "epoch": 0.32818823529411767,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0247,
      "step": 13948
    },
    {
      "epoch": 0.32821176470588237,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.05,
      "step": 13949
    },
    {
      "epoch": 0.32823529411764707,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0307,
      "step": 13950
    },
    {
      "epoch": 0.32825882352941177,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9466,
      "step": 13951
    },
    {
      "epoch": 0.32828235294117647,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3127,
      "step": 13952
    },
    {
      "epoch": 0.32830588235294117,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2369,
      "step": 13953
    },
    {
      "epoch": 0.32832941176470587,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 13954
    },
    {
      "epoch": 0.32835294117647057,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9139,
      "step": 13955
    },
    {
      "epoch": 0.32837647058823527,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9311,
      "step": 13956
    },
    {
      "epoch": 0.3284,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.823,
      "step": 13957
    },
    {
      "epoch": 0.3284235294117647,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.468,
      "step": 13958
    },
    {
      "epoch": 0.3284470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 13959
    },
    {
      "epoch": 0.3284705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7579,
      "step": 13960
    },
    {
      "epoch": 0.3284941176470588,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2763,
      "step": 13961
    },
    {
      "epoch": 0.3285176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9833,
      "step": 13962
    },
    {
      "epoch": 0.3285411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1667,
      "step": 13963
    },
    {
      "epoch": 0.3285647058823529,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.924,
      "step": 13964
    },
    {
      "epoch": 0.3285882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0708,
      "step": 13965
    },
    {
      "epoch": 0.3286117647058823,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7991,
      "step": 13966
    },
    {
      "epoch": 0.3286352941176471,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3119,
      "step": 13967
    },
    {
      "epoch": 0.3286588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8844,
      "step": 13968
    },
    {
      "epoch": 0.3286823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1308,
      "step": 13969
    },
    {
      "epoch": 0.3287058823529412,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0914,
      "step": 13970
    },
    {
      "epoch": 0.3287294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8215,
      "step": 13971
    },
    {
      "epoch": 0.3287529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8305,
      "step": 13972
    },
    {
      "epoch": 0.3287764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9856,
      "step": 13973
    },
    {
      "epoch": 0.3288,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0704,
      "step": 13974
    },
    {
      "epoch": 0.3288235294117647,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8941,
      "step": 13975
    },
    {
      "epoch": 0.32884705882352944,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.953,
      "step": 13976
    },
    {
      "epoch": 0.32887058823529414,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9316,
      "step": 13977
    },
    {
      "epoch": 0.32889411764705884,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 13978
    },
    {
      "epoch": 0.32891764705882354,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2925,
      "step": 13979
    },
    {
      "epoch": 0.32894117647058824,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7199,
      "step": 13980
    },
    {
      "epoch": 0.32896470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.5844,
      "step": 13981
    },
    {
      "epoch": 0.32898823529411764,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2329,
      "step": 13982
    },
    {
      "epoch": 0.32901176470588234,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0079,
      "step": 13983
    },
    {
      "epoch": 0.32903529411764704,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.121,
      "step": 13984
    },
    {
      "epoch": 0.32905882352941174,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2076,
      "step": 13985
    },
    {
      "epoch": 0.3290823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0326,
      "step": 13986
    },
    {
      "epoch": 0.3291058823529412,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8291,
      "step": 13987
    },
    {
      "epoch": 0.3291294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9564,
      "step": 13988
    },
    {
      "epoch": 0.3291529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1793,
      "step": 13989
    },
    {
      "epoch": 0.3291764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8194,
      "step": 13990
    },
    {
      "epoch": 0.3292,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0507,
      "step": 13991
    },
    {
      "epoch": 0.3292235294117647,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0278,
      "step": 13992
    },
    {
      "epoch": 0.3292470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.018,
      "step": 13993
    },
    {
      "epoch": 0.3292705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.171,
      "step": 13994
    },
    {
      "epoch": 0.32929411764705885,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 13995
    },
    {
      "epoch": 0.32931764705882355,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8668,
      "step": 13996
    },
    {
      "epoch": 0.32934117647058825,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8699,
      "step": 13997
    },
    {
      "epoch": 0.32936470588235295,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8886,
      "step": 13998
    },
    {
      "epoch": 0.32938823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9135,
      "step": 13999
    },
    {
      "epoch": 0.32941176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1461,
      "step": 14000
    },
    {
      "epoch": 0.32941176470588235,
      "eval_loss": 2.200066089630127,
      "eval_runtime": 689.0455,
      "eval_samples_per_second": 12.336,
      "eval_steps_per_second": 3.084,
      "step": 14000
    },
    {
      "epoch": 0.32943529411764705,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0601,
      "step": 14001
    },
    {
      "epoch": 0.32945882352941175,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9208,
      "step": 14002
    },
    {
      "epoch": 0.32948235294117645,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9888,
      "step": 14003
    },
    {
      "epoch": 0.3295058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9937,
      "step": 14004
    },
    {
      "epoch": 0.3295294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0289,
      "step": 14005
    },
    {
      "epoch": 0.3295529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8197,
      "step": 14006
    },
    {
      "epoch": 0.3295764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7114,
      "step": 14007
    },
    {
      "epoch": 0.3296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 14008
    },
    {
      "epoch": 0.3296235294117647,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1542,
      "step": 14009
    },
    {
      "epoch": 0.3296470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.516,
      "step": 14010
    },
    {
      "epoch": 0.3296705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0267,
      "step": 14011
    },
    {
      "epoch": 0.3296941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9892,
      "step": 14012
    },
    {
      "epoch": 0.3297176470588235,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8875,
      "step": 14013
    },
    {
      "epoch": 0.32974117647058826,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8312,
      "step": 14014
    },
    {
      "epoch": 0.32976470588235296,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7962,
      "step": 14015
    },
    {
      "epoch": 0.32978823529411766,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.4697,
      "step": 14016
    },
    {
      "epoch": 0.32981176470588236,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.96,
      "step": 14017
    },
    {
      "epoch": 0.32983529411764706,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7189,
      "step": 14018
    },
    {
      "epoch": 0.32985882352941176,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2392,
      "step": 14019
    },
    {
      "epoch": 0.32988235294117646,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3578,
      "step": 14020
    },
    {
      "epoch": 0.32990588235294116,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6951,
      "step": 14021
    },
    {
      "epoch": 0.32992941176470586,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0871,
      "step": 14022
    },
    {
      "epoch": 0.3299529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1643,
      "step": 14023
    },
    {
      "epoch": 0.3299764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0938,
      "step": 14024
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.4558,
      "step": 14025
    },
    {
      "epoch": 0.3300235294117647,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.3296,
      "step": 14026
    },
    {
      "epoch": 0.3300470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0953,
      "step": 14027
    },
    {
      "epoch": 0.3300705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.6118,
      "step": 14028
    },
    {
      "epoch": 0.3300941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8214,
      "step": 14029
    },
    {
      "epoch": 0.3301176470588235,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8097,
      "step": 14030
    },
    {
      "epoch": 0.3301411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8709,
      "step": 14031
    },
    {
      "epoch": 0.3301647058823529,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2729,
      "step": 14032
    },
    {
      "epoch": 0.33018823529411767,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8182,
      "step": 14033
    },
    {
      "epoch": 0.33021176470588237,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8534,
      "step": 14034
    },
    {
      "epoch": 0.33023529411764707,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1262,
      "step": 14035
    },
    {
      "epoch": 0.33025882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.082,
      "step": 14036
    },
    {
      "epoch": 0.33028235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 14037
    },
    {
      "epoch": 0.33030588235294117,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1224,
      "step": 14038
    },
    {
      "epoch": 0.33032941176470587,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1186,
      "step": 14039
    },
    {
      "epoch": 0.33035294117647057,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7945,
      "step": 14040
    },
    {
      "epoch": 0.33037647058823527,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0275,
      "step": 14041
    },
    {
      "epoch": 0.3304,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1233,
      "step": 14042
    },
    {
      "epoch": 0.3304235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1068,
      "step": 14043
    },
    {
      "epoch": 0.3304470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8049,
      "step": 14044
    },
    {
      "epoch": 0.3304705882352941,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.6965,
      "step": 14045
    },
    {
      "epoch": 0.3304941176470588,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.064,
      "step": 14046
    },
    {
      "epoch": 0.3305176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9165,
      "step": 14047
    },
    {
      "epoch": 0.3305411764705882,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.6463,
      "step": 14048
    },
    {
      "epoch": 0.3305647058823529,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.954,
      "step": 14049
    },
    {
      "epoch": 0.3305882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.931,
      "step": 14050
    },
    {
      "epoch": 0.3306117647058823,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7945,
      "step": 14051
    },
    {
      "epoch": 0.3306352941176471,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9507,
      "step": 14052
    },
    {
      "epoch": 0.3306588235294118,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1197,
      "step": 14053
    },
    {
      "epoch": 0.3306823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1182,
      "step": 14054
    },
    {
      "epoch": 0.3307058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1849,
      "step": 14055
    },
    {
      "epoch": 0.3307294117647059,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9299,
      "step": 14056
    },
    {
      "epoch": 0.3307529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0041,
      "step": 14057
    },
    {
      "epoch": 0.3307764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9793,
      "step": 14058
    },
    {
      "epoch": 0.3308,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8101,
      "step": 14059
    },
    {
      "epoch": 0.3308235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0314,
      "step": 14060
    },
    {
      "epoch": 0.33084705882352944,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0163,
      "step": 14061
    },
    {
      "epoch": 0.33087058823529414,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 14062
    },
    {
      "epoch": 0.33089411764705884,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1742,
      "step": 14063
    },
    {
      "epoch": 0.33091764705882354,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1107,
      "step": 14064
    },
    {
      "epoch": 0.33094117647058824,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1633,
      "step": 14065
    },
    {
      "epoch": 0.33096470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7576,
      "step": 14066
    },
    {
      "epoch": 0.33098823529411764,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9356,
      "step": 14067
    },
    {
      "epoch": 0.33101176470588234,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 14068
    },
    {
      "epoch": 0.33103529411764704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1471,
      "step": 14069
    },
    {
      "epoch": 0.33105882352941174,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0325,
      "step": 14070
    },
    {
      "epoch": 0.3310823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0855,
      "step": 14071
    },
    {
      "epoch": 0.3311058823529412,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8262,
      "step": 14072
    },
    {
      "epoch": 0.3311294117647059,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.6534,
      "step": 14073
    },
    {
      "epoch": 0.3311529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1479,
      "step": 14074
    },
    {
      "epoch": 0.3311764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1535,
      "step": 14075
    },
    {
      "epoch": 0.3312,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1215,
      "step": 14076
    },
    {
      "epoch": 0.3312235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1087,
      "step": 14077
    },
    {
      "epoch": 0.3312470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.3181,
      "step": 14078
    },
    {
      "epoch": 0.3312705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9984,
      "step": 14079
    },
    {
      "epoch": 0.33129411764705885,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.069,
      "step": 14080
    },
    {
      "epoch": 0.33131764705882355,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6299,
      "step": 14081
    },
    {
      "epoch": 0.33134117647058825,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1874,
      "step": 14082
    },
    {
      "epoch": 0.33136470588235295,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8526,
      "step": 14083
    },
    {
      "epoch": 0.33138823529411765,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8655,
      "step": 14084
    },
    {
      "epoch": 0.33141176470588235,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7879,
      "step": 14085
    },
    {
      "epoch": 0.33143529411764705,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0254,
      "step": 14086
    },
    {
      "epoch": 0.33145882352941175,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1419,
      "step": 14087
    },
    {
      "epoch": 0.33148235294117645,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.6273,
      "step": 14088
    },
    {
      "epoch": 0.33150588235294115,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0082,
      "step": 14089
    },
    {
      "epoch": 0.3315294117647059,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.184,
      "step": 14090
    },
    {
      "epoch": 0.3315529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0716,
      "step": 14091
    },
    {
      "epoch": 0.3315764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1922,
      "step": 14092
    },
    {
      "epoch": 0.3316,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0269,
      "step": 14093
    },
    {
      "epoch": 0.3316235294117647,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.4721,
      "step": 14094
    },
    {
      "epoch": 0.3316470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.157,
      "step": 14095
    },
    {
      "epoch": 0.3316705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8651,
      "step": 14096
    },
    {
      "epoch": 0.3316941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1544,
      "step": 14097
    },
    {
      "epoch": 0.3317176470588235,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1091,
      "step": 14098
    },
    {
      "epoch": 0.33174117647058826,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.1347,
      "step": 14099
    },
    {
      "epoch": 0.33176470588235296,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.4505,
      "step": 14100
    },
    {
      "epoch": 0.33178823529411766,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9427,
      "step": 14101
    },
    {
      "epoch": 0.33181176470588236,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9026,
      "step": 14102
    },
    {
      "epoch": 0.33183529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1311,
      "step": 14103
    },
    {
      "epoch": 0.33185882352941176,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 14104
    },
    {
      "epoch": 0.33188235294117646,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2376,
      "step": 14105
    },
    {
      "epoch": 0.33190588235294116,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.998,
      "step": 14106
    },
    {
      "epoch": 0.33192941176470586,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0849,
      "step": 14107
    },
    {
      "epoch": 0.33195294117647056,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.5853,
      "step": 14108
    },
    {
      "epoch": 0.3319764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.7479,
      "step": 14109
    },
    {
      "epoch": 0.332,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1328,
      "step": 14110
    },
    {
      "epoch": 0.3320235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9094,
      "step": 14111
    },
    {
      "epoch": 0.3320470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6293,
      "step": 14112
    },
    {
      "epoch": 0.3320705882352941,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8416,
      "step": 14113
    },
    {
      "epoch": 0.3320941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.6732,
      "step": 14114
    },
    {
      "epoch": 0.3321176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1445,
      "step": 14115
    },
    {
      "epoch": 0.3321411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0498,
      "step": 14116
    },
    {
      "epoch": 0.3321647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.338,
      "step": 14117
    },
    {
      "epoch": 0.3321882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9458,
      "step": 14118
    },
    {
      "epoch": 0.3322117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.066,
      "step": 14119
    },
    {
      "epoch": 0.3322352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.6965,
      "step": 14120
    },
    {
      "epoch": 0.3322588235294118,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3729,
      "step": 14121
    },
    {
      "epoch": 0.3322823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0533,
      "step": 14122
    },
    {
      "epoch": 0.3323058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8576,
      "step": 14123
    },
    {
      "epoch": 0.3323294117647059,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1829,
      "step": 14124
    },
    {
      "epoch": 0.3323529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2594,
      "step": 14125
    },
    {
      "epoch": 0.3323764705882353,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0085,
      "step": 14126
    },
    {
      "epoch": 0.3324,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1577,
      "step": 14127
    },
    {
      "epoch": 0.33242352941176473,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2667,
      "step": 14128
    },
    {
      "epoch": 0.33244705882352943,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8705,
      "step": 14129
    },
    {
      "epoch": 0.33247058823529413,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 14130
    },
    {
      "epoch": 0.33249411764705883,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1368,
      "step": 14131
    },
    {
      "epoch": 0.33251764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1346,
      "step": 14132
    },
    {
      "epoch": 0.33254117647058823,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7813,
      "step": 14133
    },
    {
      "epoch": 0.33256470588235293,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7668,
      "step": 14134
    },
    {
      "epoch": 0.33258823529411763,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.3932,
      "step": 14135
    },
    {
      "epoch": 0.33261176470588233,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.2052,
      "step": 14136
    },
    {
      "epoch": 0.3326352941176471,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9769,
      "step": 14137
    },
    {
      "epoch": 0.3326588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0764,
      "step": 14138
    },
    {
      "epoch": 0.3326823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9094,
      "step": 14139
    },
    {
      "epoch": 0.3327058823529412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.5528,
      "step": 14140
    },
    {
      "epoch": 0.3327294117647059,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6537,
      "step": 14141
    },
    {
      "epoch": 0.3327529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9792,
      "step": 14142
    },
    {
      "epoch": 0.3327764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0961,
      "step": 14143
    },
    {
      "epoch": 0.3328,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9022,
      "step": 14144
    },
    {
      "epoch": 0.3328235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.165,
      "step": 14145
    },
    {
      "epoch": 0.3328470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1436,
      "step": 14146
    },
    {
      "epoch": 0.33287058823529414,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1693,
      "step": 14147
    },
    {
      "epoch": 0.33289411764705884,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1365,
      "step": 14148
    },
    {
      "epoch": 0.33291764705882354,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8836,
      "step": 14149
    },
    {
      "epoch": 0.33294117647058824,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 14150
    },
    {
      "epoch": 0.33296470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.8239,
      "step": 14151
    },
    {
      "epoch": 0.33298823529411764,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0829,
      "step": 14152
    },
    {
      "epoch": 0.33301176470588234,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1384,
      "step": 14153
    },
    {
      "epoch": 0.33303529411764704,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.6436,
      "step": 14154
    },
    {
      "epoch": 0.33305882352941174,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.9761,
      "step": 14155
    },
    {
      "epoch": 0.3330823529411765,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.02,
      "loss": 0.7754,
      "step": 14156
    },
    {
      "epoch": 0.3331058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0864,
      "step": 14157
    },
    {
      "epoch": 0.3331294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9564,
      "step": 14158
    },
    {
      "epoch": 0.3331529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8142,
      "step": 14159
    },
    {
      "epoch": 0.3331764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8455,
      "step": 14160
    },
    {
      "epoch": 0.3332,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8743,
      "step": 14161
    },
    {
      "epoch": 0.3332235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0354,
      "step": 14162
    },
    {
      "epoch": 0.3332470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0955,
      "step": 14163
    },
    {
      "epoch": 0.3332705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2471,
      "step": 14164
    },
    {
      "epoch": 0.3332941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0841,
      "step": 14165
    },
    {
      "epoch": 0.33331764705882355,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2859,
      "step": 14166
    },
    {
      "epoch": 0.33334117647058825,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 0.5054,
      "step": 14167
    },
    {
      "epoch": 0.33336470588235295,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3494,
      "step": 14168
    },
    {
      "epoch": 0.33338823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7812,
      "step": 14169
    },
    {
      "epoch": 0.33341176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8481,
      "step": 14170
    },
    {
      "epoch": 0.33343529411764705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.5911,
      "step": 14171
    },
    {
      "epoch": 0.33345882352941175,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0376,
      "step": 14172
    },
    {
      "epoch": 0.33348235294117645,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0005,
      "step": 14173
    },
    {
      "epoch": 0.33350588235294115,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.4899,
      "step": 14174
    },
    {
      "epoch": 0.3335294117647059,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7859,
      "step": 14175
    },
    {
      "epoch": 0.3335529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.204,
      "step": 14176
    },
    {
      "epoch": 0.3335764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9679,
      "step": 14177
    },
    {
      "epoch": 0.3336,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6424,
      "step": 14178
    },
    {
      "epoch": 0.3336235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1561,
      "step": 14179
    },
    {
      "epoch": 0.3336470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9075,
      "step": 14180
    },
    {
      "epoch": 0.3336705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.973,
      "step": 14181
    },
    {
      "epoch": 0.3336941176470588,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2497,
      "step": 14182
    },
    {
      "epoch": 0.3337176470588235,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.774,
      "step": 14183
    },
    {
      "epoch": 0.3337411764705882,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.013,
      "step": 14184
    },
    {
      "epoch": 0.33376470588235296,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1095,
      "step": 14185
    },
    {
      "epoch": 0.33378823529411766,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0891,
      "step": 14186
    },
    {
      "epoch": 0.33381176470588236,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1143,
      "step": 14187
    },
    {
      "epoch": 0.33383529411764706,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9041,
      "step": 14188
    },
    {
      "epoch": 0.33385882352941176,
      "grad_norm": 0.14453125,
      "learning_rate": 0.02,
      "loss": 1.3618,
      "step": 14189
    },
    {
      "epoch": 0.33388235294117646,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 14190
    },
    {
      "epoch": 0.33390588235294116,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8839,
      "step": 14191
    },
    {
      "epoch": 0.33392941176470586,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0969,
      "step": 14192
    },
    {
      "epoch": 0.33395294117647056,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.195,
      "step": 14193
    },
    {
      "epoch": 0.3339764705882353,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0698,
      "step": 14194
    },
    {
      "epoch": 0.334,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8143,
      "step": 14195
    },
    {
      "epoch": 0.3340235294117647,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0899,
      "step": 14196
    },
    {
      "epoch": 0.3340470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1156,
      "step": 14197
    },
    {
      "epoch": 0.3340705882352941,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2367,
      "step": 14198
    },
    {
      "epoch": 0.3340941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8894,
      "step": 14199
    },
    {
      "epoch": 0.3341176470588235,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0106,
      "step": 14200
    },
    {
      "epoch": 0.3341411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7322,
      "step": 14201
    },
    {
      "epoch": 0.3341647058823529,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 14202
    },
    {
      "epoch": 0.3341882352941177,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0644,
      "step": 14203
    },
    {
      "epoch": 0.3342117647058824,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8323,
      "step": 14204
    },
    {
      "epoch": 0.3342352941176471,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0184,
      "step": 14205
    },
    {
      "epoch": 0.3342588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.112,
      "step": 14206
    },
    {
      "epoch": 0.3342823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 14207
    },
    {
      "epoch": 0.3343058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1022,
      "step": 14208
    },
    {
      "epoch": 0.3343294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0194,
      "step": 14209
    },
    {
      "epoch": 0.3343529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7817,
      "step": 14210
    },
    {
      "epoch": 0.3343764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7747,
      "step": 14211
    },
    {
      "epoch": 0.3344,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.084,
      "step": 14212
    },
    {
      "epoch": 0.33442352941176473,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0943,
      "step": 14213
    },
    {
      "epoch": 0.33444705882352943,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1565,
      "step": 14214
    },
    {
      "epoch": 0.33447058823529413,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9634,
      "step": 14215
    },
    {
      "epoch": 0.33449411764705883,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0044,
      "step": 14216
    },
    {
      "epoch": 0.33451764705882353,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.306,
      "step": 14217
    },
    {
      "epoch": 0.33454117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.243,
      "step": 14218
    },
    {
      "epoch": 0.33456470588235293,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8712,
      "step": 14219
    },
    {
      "epoch": 0.33458823529411763,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.991,
      "step": 14220
    },
    {
      "epoch": 0.33461176470588233,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0198,
      "step": 14221
    },
    {
      "epoch": 0.3346352941176471,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2259,
      "step": 14222
    },
    {
      "epoch": 0.3346588235294118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 14223
    },
    {
      "epoch": 0.3346823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0308,
      "step": 14224
    },
    {
      "epoch": 0.3347058823529412,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0349,
      "step": 14225
    },
    {
      "epoch": 0.3347294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8317,
      "step": 14226
    },
    {
      "epoch": 0.3347529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1271,
      "step": 14227
    },
    {
      "epoch": 0.3347764705882353,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.6339,
      "step": 14228
    },
    {
      "epoch": 0.3348,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7059,
      "step": 14229
    },
    {
      "epoch": 0.3348235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0265,
      "step": 14230
    },
    {
      "epoch": 0.3348470588235294,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2971,
      "step": 14231
    },
    {
      "epoch": 0.33487058823529414,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1031,
      "step": 14232
    },
    {
      "epoch": 0.33489411764705884,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.3655,
      "step": 14233
    },
    {
      "epoch": 0.33491764705882354,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1319,
      "step": 14234
    },
    {
      "epoch": 0.33494117647058824,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.322,
      "step": 14235
    },
    {
      "epoch": 0.33496470588235294,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 14236
    },
    {
      "epoch": 0.33498823529411764,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8855,
      "step": 14237
    },
    {
      "epoch": 0.33501176470588234,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.3535,
      "step": 14238
    },
    {
      "epoch": 0.33503529411764704,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9989,
      "step": 14239
    },
    {
      "epoch": 0.33505882352941174,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0023,
      "step": 14240
    },
    {
      "epoch": 0.3350823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9739,
      "step": 14241
    },
    {
      "epoch": 0.3351058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.4511,
      "step": 14242
    },
    {
      "epoch": 0.3351294117647059,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.998,
      "step": 14243
    },
    {
      "epoch": 0.3351529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0607,
      "step": 14244
    },
    {
      "epoch": 0.3351764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1305,
      "step": 14245
    },
    {
      "epoch": 0.3352,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8227,
      "step": 14246
    },
    {
      "epoch": 0.3352235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0118,
      "step": 14247
    },
    {
      "epoch": 0.3352470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.7601,
      "step": 14248
    },
    {
      "epoch": 0.3352705882352941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0931,
      "step": 14249
    },
    {
      "epoch": 0.3352941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9494,
      "step": 14250
    },
    {
      "epoch": 0.33531764705882355,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9407,
      "step": 14251
    },
    {
      "epoch": 0.33534117647058825,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9208,
      "step": 14252
    },
    {
      "epoch": 0.33536470588235295,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9476,
      "step": 14253
    },
    {
      "epoch": 0.33538823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0132,
      "step": 14254
    },
    {
      "epoch": 0.33541176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9497,
      "step": 14255
    },
    {
      "epoch": 0.33543529411764705,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9566,
      "step": 14256
    },
    {
      "epoch": 0.33545882352941175,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9102,
      "step": 14257
    },
    {
      "epoch": 0.33548235294117645,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 14258
    },
    {
      "epoch": 0.33550588235294115,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8292,
      "step": 14259
    },
    {
      "epoch": 0.3355294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7062,
      "step": 14260
    },
    {
      "epoch": 0.3355529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 14261
    },
    {
      "epoch": 0.3355764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1311,
      "step": 14262
    },
    {
      "epoch": 0.3356,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3153,
      "step": 14263
    },
    {
      "epoch": 0.3356235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1749,
      "step": 14264
    },
    {
      "epoch": 0.3356470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 14265
    },
    {
      "epoch": 0.3356705882352941,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6835,
      "step": 14266
    },
    {
      "epoch": 0.3356941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9264,
      "step": 14267
    },
    {
      "epoch": 0.3357176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7769,
      "step": 14268
    },
    {
      "epoch": 0.3357411764705882,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.1889,
      "step": 14269
    },
    {
      "epoch": 0.33576470588235297,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9435,
      "step": 14270
    },
    {
      "epoch": 0.33578823529411767,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1478,
      "step": 14271
    },
    {
      "epoch": 0.33581176470588237,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1254,
      "step": 14272
    },
    {
      "epoch": 0.33583529411764707,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 14273
    },
    {
      "epoch": 0.33585882352941177,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1039,
      "step": 14274
    },
    {
      "epoch": 0.33588235294117647,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1347,
      "step": 14275
    },
    {
      "epoch": 0.33590588235294117,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9618,
      "step": 14276
    },
    {
      "epoch": 0.33592941176470587,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0342,
      "step": 14277
    },
    {
      "epoch": 0.33595294117647057,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1217,
      "step": 14278
    },
    {
      "epoch": 0.3359764705882353,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.9819,
      "step": 14279
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6559,
      "step": 14280
    },
    {
      "epoch": 0.3360235294117647,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2375,
      "step": 14281
    },
    {
      "epoch": 0.3360470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.281,
      "step": 14282
    },
    {
      "epoch": 0.3360705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0454,
      "step": 14283
    },
    {
      "epoch": 0.3360941176470588,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0542,
      "step": 14284
    },
    {
      "epoch": 0.3361176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7883,
      "step": 14285
    },
    {
      "epoch": 0.3361411764705882,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0348,
      "step": 14286
    },
    {
      "epoch": 0.3361647058823529,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0045,
      "step": 14287
    },
    {
      "epoch": 0.3361882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0714,
      "step": 14288
    },
    {
      "epoch": 0.3362117647058824,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2983,
      "step": 14289
    },
    {
      "epoch": 0.3362352941176471,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.9289,
      "step": 14290
    },
    {
      "epoch": 0.3362588235294118,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.26,
      "step": 14291
    },
    {
      "epoch": 0.3362823529411765,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1233,
      "step": 14292
    },
    {
      "epoch": 0.3363058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2412,
      "step": 14293
    },
    {
      "epoch": 0.3363294117647059,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.8071,
      "step": 14294
    },
    {
      "epoch": 0.3363529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1329,
      "step": 14295
    },
    {
      "epoch": 0.3363764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0729,
      "step": 14296
    },
    {
      "epoch": 0.3364,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9331,
      "step": 14297
    },
    {
      "epoch": 0.33642352941176473,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0687,
      "step": 14298
    },
    {
      "epoch": 0.33644705882352943,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0007,
      "step": 14299
    },
    {
      "epoch": 0.33647058823529413,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8684,
      "step": 14300
    },
    {
      "epoch": 0.33649411764705883,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9844,
      "step": 14301
    },
    {
      "epoch": 0.33651764705882353,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1438,
      "step": 14302
    },
    {
      "epoch": 0.33654117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0018,
      "step": 14303
    },
    {
      "epoch": 0.33656470588235293,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9723,
      "step": 14304
    },
    {
      "epoch": 0.33658823529411763,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8692,
      "step": 14305
    },
    {
      "epoch": 0.33661176470588233,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2632,
      "step": 14306
    },
    {
      "epoch": 0.33663529411764703,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9216,
      "step": 14307
    },
    {
      "epoch": 0.3366588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9348,
      "step": 14308
    },
    {
      "epoch": 0.3366823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8112,
      "step": 14309
    },
    {
      "epoch": 0.3367058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.734,
      "step": 14310
    },
    {
      "epoch": 0.3367294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2815,
      "step": 14311
    },
    {
      "epoch": 0.3367529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0667,
      "step": 14312
    },
    {
      "epoch": 0.3367764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0625,
      "step": 14313
    },
    {
      "epoch": 0.3368,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9562,
      "step": 14314
    },
    {
      "epoch": 0.3368235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0148,
      "step": 14315
    },
    {
      "epoch": 0.3368470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0581,
      "step": 14316
    },
    {
      "epoch": 0.33687058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9665,
      "step": 14317
    },
    {
      "epoch": 0.33689411764705884,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7071,
      "step": 14318
    },
    {
      "epoch": 0.33691764705882354,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0224,
      "step": 14319
    },
    {
      "epoch": 0.33694117647058824,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9076,
      "step": 14320
    },
    {
      "epoch": 0.33696470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9887,
      "step": 14321
    },
    {
      "epoch": 0.33698823529411764,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2225,
      "step": 14322
    },
    {
      "epoch": 0.33701176470588234,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 0.9965,
      "step": 14323
    },
    {
      "epoch": 0.33703529411764704,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9346,
      "step": 14324
    },
    {
      "epoch": 0.33705882352941174,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1811,
      "step": 14325
    },
    {
      "epoch": 0.33708235294117644,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7036,
      "step": 14326
    },
    {
      "epoch": 0.3371058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9271,
      "step": 14327
    },
    {
      "epoch": 0.3371294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7627,
      "step": 14328
    },
    {
      "epoch": 0.3371529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7894,
      "step": 14329
    },
    {
      "epoch": 0.3371764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9371,
      "step": 14330
    },
    {
      "epoch": 0.3372,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8973,
      "step": 14331
    },
    {
      "epoch": 0.3372235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8777,
      "step": 14332
    },
    {
      "epoch": 0.3372470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.811,
      "step": 14333
    },
    {
      "epoch": 0.3372705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9441,
      "step": 14334
    },
    {
      "epoch": 0.3372941176470588,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.5934,
      "step": 14335
    },
    {
      "epoch": 0.33731764705882356,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.2266,
      "step": 14336
    },
    {
      "epoch": 0.33734117647058826,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.2354,
      "step": 14337
    },
    {
      "epoch": 0.33736470588235296,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7188,
      "step": 14338
    },
    {
      "epoch": 0.33738823529411766,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1483,
      "step": 14339
    },
    {
      "epoch": 0.33741176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0333,
      "step": 14340
    },
    {
      "epoch": 0.33743529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8755,
      "step": 14341
    },
    {
      "epoch": 0.33745882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.211,
      "step": 14342
    },
    {
      "epoch": 0.33748235294117646,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8077,
      "step": 14343
    },
    {
      "epoch": 0.33750588235294116,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0082,
      "step": 14344
    },
    {
      "epoch": 0.33752941176470586,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0373,
      "step": 14345
    },
    {
      "epoch": 0.3375529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2386,
      "step": 14346
    },
    {
      "epoch": 0.3375764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0053,
      "step": 14347
    },
    {
      "epoch": 0.3376,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.271,
      "step": 14348
    },
    {
      "epoch": 0.3376235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1332,
      "step": 14349
    },
    {
      "epoch": 0.3376470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2396,
      "step": 14350
    },
    {
      "epoch": 0.3376705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2872,
      "step": 14351
    },
    {
      "epoch": 0.3376941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0697,
      "step": 14352
    },
    {
      "epoch": 0.3377176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7517,
      "step": 14353
    },
    {
      "epoch": 0.3377411764705882,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0052,
      "step": 14354
    },
    {
      "epoch": 0.33776470588235297,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0053,
      "step": 14355
    },
    {
      "epoch": 0.33778823529411767,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9991,
      "step": 14356
    },
    {
      "epoch": 0.33781176470588237,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.12,
      "step": 14357
    },
    {
      "epoch": 0.33783529411764707,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1027,
      "step": 14358
    },
    {
      "epoch": 0.33785882352941177,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.1501,
      "step": 14359
    },
    {
      "epoch": 0.33788235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9794,
      "step": 14360
    },
    {
      "epoch": 0.33790588235294117,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9113,
      "step": 14361
    },
    {
      "epoch": 0.33792941176470587,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1845,
      "step": 14362
    },
    {
      "epoch": 0.33795294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2668,
      "step": 14363
    },
    {
      "epoch": 0.33797647058823527,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0092,
      "step": 14364
    },
    {
      "epoch": 0.338,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0529,
      "step": 14365
    },
    {
      "epoch": 0.3380235294117647,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0361,
      "step": 14366
    },
    {
      "epoch": 0.3380470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.6095,
      "step": 14367
    },
    {
      "epoch": 0.3380705882352941,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2658,
      "step": 14368
    },
    {
      "epoch": 0.3380941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9231,
      "step": 14369
    },
    {
      "epoch": 0.3381176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0554,
      "step": 14370
    },
    {
      "epoch": 0.3381411764705882,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1158,
      "step": 14371
    },
    {
      "epoch": 0.3381647058823529,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3674,
      "step": 14372
    },
    {
      "epoch": 0.3381882352941176,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8224,
      "step": 14373
    },
    {
      "epoch": 0.3382117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8569,
      "step": 14374
    },
    {
      "epoch": 0.3382352941176471,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9283,
      "step": 14375
    },
    {
      "epoch": 0.3382588235294118,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2407,
      "step": 14376
    },
    {
      "epoch": 0.3382823529411765,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.6795,
      "step": 14377
    },
    {
      "epoch": 0.3383058823529412,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7752,
      "step": 14378
    },
    {
      "epoch": 0.3383294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0591,
      "step": 14379
    },
    {
      "epoch": 0.3383529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 14380
    },
    {
      "epoch": 0.3383764705882353,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3297,
      "step": 14381
    },
    {
      "epoch": 0.3384,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1856,
      "step": 14382
    },
    {
      "epoch": 0.3384235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9101,
      "step": 14383
    },
    {
      "epoch": 0.33844705882352943,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2522,
      "step": 14384
    },
    {
      "epoch": 0.33847058823529413,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8105,
      "step": 14385
    },
    {
      "epoch": 0.33849411764705883,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.3702,
      "step": 14386
    },
    {
      "epoch": 0.33851764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7397,
      "step": 14387
    },
    {
      "epoch": 0.33854117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8885,
      "step": 14388
    },
    {
      "epoch": 0.33856470588235293,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0334,
      "step": 14389
    },
    {
      "epoch": 0.33858823529411763,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3499,
      "step": 14390
    },
    {
      "epoch": 0.33861176470588233,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8582,
      "step": 14391
    },
    {
      "epoch": 0.33863529411764703,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0171,
      "step": 14392
    },
    {
      "epoch": 0.3386588235294118,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1197,
      "step": 14393
    },
    {
      "epoch": 0.3386823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0285,
      "step": 14394
    },
    {
      "epoch": 0.3387058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0269,
      "step": 14395
    },
    {
      "epoch": 0.3387294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1155,
      "step": 14396
    },
    {
      "epoch": 0.3387529411764706,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.7453,
      "step": 14397
    },
    {
      "epoch": 0.3387764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1789,
      "step": 14398
    },
    {
      "epoch": 0.3388,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7636,
      "step": 14399
    },
    {
      "epoch": 0.3388235294117647,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.4081,
      "step": 14400
    },
    {
      "epoch": 0.3388470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8928,
      "step": 14401
    },
    {
      "epoch": 0.3388705882352941,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9863,
      "step": 14402
    },
    {
      "epoch": 0.33889411764705885,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.5038,
      "step": 14403
    },
    {
      "epoch": 0.33891764705882355,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8296,
      "step": 14404
    },
    {
      "epoch": 0.33894117647058825,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1047,
      "step": 14405
    },
    {
      "epoch": 0.33896470588235295,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1006,
      "step": 14406
    },
    {
      "epoch": 0.33898823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9348,
      "step": 14407
    },
    {
      "epoch": 0.33901176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1059,
      "step": 14408
    },
    {
      "epoch": 0.33903529411764705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9104,
      "step": 14409
    },
    {
      "epoch": 0.33905882352941175,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2517,
      "step": 14410
    },
    {
      "epoch": 0.33908235294117645,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1189,
      "step": 14411
    },
    {
      "epoch": 0.3391058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0142,
      "step": 14412
    },
    {
      "epoch": 0.3391294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0884,
      "step": 14413
    },
    {
      "epoch": 0.3391529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0243,
      "step": 14414
    },
    {
      "epoch": 0.3391764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2585,
      "step": 14415
    },
    {
      "epoch": 0.3392,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 14416
    },
    {
      "epoch": 0.3392235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8277,
      "step": 14417
    },
    {
      "epoch": 0.3392470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9483,
      "step": 14418
    },
    {
      "epoch": 0.3392705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8518,
      "step": 14419
    },
    {
      "epoch": 0.3392941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2503,
      "step": 14420
    },
    {
      "epoch": 0.33931764705882356,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0705,
      "step": 14421
    },
    {
      "epoch": 0.33934117647058826,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0174,
      "step": 14422
    },
    {
      "epoch": 0.33936470588235296,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 14423
    },
    {
      "epoch": 0.33938823529411766,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9107,
      "step": 14424
    },
    {
      "epoch": 0.33941176470588236,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.6669,
      "step": 14425
    },
    {
      "epoch": 0.33943529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.393,
      "step": 14426
    },
    {
      "epoch": 0.33945882352941176,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9332,
      "step": 14427
    },
    {
      "epoch": 0.33948235294117646,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2103,
      "step": 14428
    },
    {
      "epoch": 0.33950588235294116,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8787,
      "step": 14429
    },
    {
      "epoch": 0.33952941176470586,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1218,
      "step": 14430
    },
    {
      "epoch": 0.3395529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9259,
      "step": 14431
    },
    {
      "epoch": 0.3395764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0809,
      "step": 14432
    },
    {
      "epoch": 0.3396,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0649,
      "step": 14433
    },
    {
      "epoch": 0.3396235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 14434
    },
    {
      "epoch": 0.3396470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0711,
      "step": 14435
    },
    {
      "epoch": 0.3396705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0741,
      "step": 14436
    },
    {
      "epoch": 0.3396941176470588,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8733,
      "step": 14437
    },
    {
      "epoch": 0.3397176470588235,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.02,
      "loss": 1.4291,
      "step": 14438
    },
    {
      "epoch": 0.3397411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1228,
      "step": 14439
    },
    {
      "epoch": 0.33976470588235297,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0908,
      "step": 14440
    },
    {
      "epoch": 0.33978823529411767,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0651,
      "step": 14441
    },
    {
      "epoch": 0.33981176470588237,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.856,
      "step": 14442
    },
    {
      "epoch": 0.33983529411764707,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7634,
      "step": 14443
    },
    {
      "epoch": 0.33985882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9124,
      "step": 14444
    },
    {
      "epoch": 0.33988235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.973,
      "step": 14445
    },
    {
      "epoch": 0.33990588235294117,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1413,
      "step": 14446
    },
    {
      "epoch": 0.33992941176470587,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0899,
      "step": 14447
    },
    {
      "epoch": 0.33995294117647057,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9602,
      "step": 14448
    },
    {
      "epoch": 0.33997647058823527,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9642,
      "step": 14449
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8869,
      "step": 14450
    },
    {
      "epoch": 0.3400235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7378,
      "step": 14451
    },
    {
      "epoch": 0.3400470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9976,
      "step": 14452
    },
    {
      "epoch": 0.3400705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.123,
      "step": 14453
    },
    {
      "epoch": 0.3400941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9082,
      "step": 14454
    },
    {
      "epoch": 0.3401176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8791,
      "step": 14455
    },
    {
      "epoch": 0.3401411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9766,
      "step": 14456
    },
    {
      "epoch": 0.3401647058823529,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.306,
      "step": 14457
    },
    {
      "epoch": 0.3401882352941176,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.733,
      "step": 14458
    },
    {
      "epoch": 0.3402117647058824,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1462,
      "step": 14459
    },
    {
      "epoch": 0.3402352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0799,
      "step": 14460
    },
    {
      "epoch": 0.3402588235294118,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6825,
      "step": 14461
    },
    {
      "epoch": 0.3402823529411765,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.7755,
      "step": 14462
    },
    {
      "epoch": 0.3403058823529412,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 14463
    },
    {
      "epoch": 0.3403294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8217,
      "step": 14464
    },
    {
      "epoch": 0.3403529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1244,
      "step": 14465
    },
    {
      "epoch": 0.3403764705882353,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6905,
      "step": 14466
    },
    {
      "epoch": 0.3404,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.5745,
      "step": 14467
    },
    {
      "epoch": 0.3404235294117647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1741,
      "step": 14468
    },
    {
      "epoch": 0.34044705882352944,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0934,
      "step": 14469
    },
    {
      "epoch": 0.34047058823529414,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0538,
      "step": 14470
    },
    {
      "epoch": 0.34049411764705884,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2323,
      "step": 14471
    },
    {
      "epoch": 0.34051764705882354,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0684,
      "step": 14472
    },
    {
      "epoch": 0.34054117647058824,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1351,
      "step": 14473
    },
    {
      "epoch": 0.34056470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0546,
      "step": 14474
    },
    {
      "epoch": 0.34058823529411764,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2005,
      "step": 14475
    },
    {
      "epoch": 0.34061176470588234,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7301,
      "step": 14476
    },
    {
      "epoch": 0.34063529411764704,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0517,
      "step": 14477
    },
    {
      "epoch": 0.3406588235294118,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.0687,
      "step": 14478
    },
    {
      "epoch": 0.3406823529411765,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3518,
      "step": 14479
    },
    {
      "epoch": 0.3407058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8735,
      "step": 14480
    },
    {
      "epoch": 0.3407294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.3191,
      "step": 14481
    },
    {
      "epoch": 0.3407529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3635,
      "step": 14482
    },
    {
      "epoch": 0.3407764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.9169,
      "step": 14483
    },
    {
      "epoch": 0.3408,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.5187,
      "step": 14484
    },
    {
      "epoch": 0.3408235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7845,
      "step": 14485
    },
    {
      "epoch": 0.3408470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7615,
      "step": 14486
    },
    {
      "epoch": 0.3408705882352941,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2397,
      "step": 14487
    },
    {
      "epoch": 0.34089411764705885,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8518,
      "step": 14488
    },
    {
      "epoch": 0.34091764705882355,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1421,
      "step": 14489
    },
    {
      "epoch": 0.34094117647058825,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2892,
      "step": 14490
    },
    {
      "epoch": 0.34096470588235295,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.3179,
      "step": 14491
    },
    {
      "epoch": 0.34098823529411765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.95,
      "step": 14492
    },
    {
      "epoch": 0.34101176470588235,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.357,
      "step": 14493
    },
    {
      "epoch": 0.34103529411764705,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.1018,
      "step": 14494
    },
    {
      "epoch": 0.34105882352941175,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.7779,
      "step": 14495
    },
    {
      "epoch": 0.34108235294117645,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1637,
      "step": 14496
    },
    {
      "epoch": 0.3411058823529412,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.137,
      "step": 14497
    },
    {
      "epoch": 0.3411294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0408,
      "step": 14498
    },
    {
      "epoch": 0.3411529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9601,
      "step": 14499
    },
    {
      "epoch": 0.3411764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7088,
      "step": 14500
    },
    {
      "epoch": 0.3412,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7907,
      "step": 14501
    },
    {
      "epoch": 0.3412235294117647,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9398,
      "step": 14502
    },
    {
      "epoch": 0.3412470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1187,
      "step": 14503
    },
    {
      "epoch": 0.3412705882352941,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2408,
      "step": 14504
    },
    {
      "epoch": 0.3412941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0762,
      "step": 14505
    },
    {
      "epoch": 0.3413176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.2573,
      "step": 14506
    },
    {
      "epoch": 0.34134117647058826,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8568,
      "step": 14507
    },
    {
      "epoch": 0.34136470588235296,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0798,
      "step": 14508
    },
    {
      "epoch": 0.34138823529411766,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1601,
      "step": 14509
    },
    {
      "epoch": 0.34141176470588236,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0262,
      "step": 14510
    },
    {
      "epoch": 0.34143529411764706,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8317,
      "step": 14511
    },
    {
      "epoch": 0.34145882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9275,
      "step": 14512
    },
    {
      "epoch": 0.34148235294117646,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1475,
      "step": 14513
    },
    {
      "epoch": 0.34150588235294116,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8806,
      "step": 14514
    },
    {
      "epoch": 0.34152941176470586,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9051,
      "step": 14515
    },
    {
      "epoch": 0.3415529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0464,
      "step": 14516
    },
    {
      "epoch": 0.3415764705882353,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2218,
      "step": 14517
    },
    {
      "epoch": 0.3416,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.101,
      "step": 14518
    },
    {
      "epoch": 0.3416235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0686,
      "step": 14519
    },
    {
      "epoch": 0.3416470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7791,
      "step": 14520
    },
    {
      "epoch": 0.3416705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0825,
      "step": 14521
    },
    {
      "epoch": 0.3416941176470588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0356,
      "step": 14522
    },
    {
      "epoch": 0.3417176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1406,
      "step": 14523
    },
    {
      "epoch": 0.3417411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.704,
      "step": 14524
    },
    {
      "epoch": 0.3417647058823529,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8678,
      "step": 14525
    },
    {
      "epoch": 0.34178823529411767,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8983,
      "step": 14526
    },
    {
      "epoch": 0.34181176470588237,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8502,
      "step": 14527
    },
    {
      "epoch": 0.34183529411764707,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7845,
      "step": 14528
    },
    {
      "epoch": 0.34185882352941177,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9466,
      "step": 14529
    },
    {
      "epoch": 0.34188235294117647,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.1867,
      "step": 14530
    },
    {
      "epoch": 0.34190588235294117,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.068,
      "step": 14531
    },
    {
      "epoch": 0.34192941176470587,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0974,
      "step": 14532
    },
    {
      "epoch": 0.34195294117647057,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9561,
      "step": 14533
    },
    {
      "epoch": 0.34197647058823527,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0251,
      "step": 14534
    },
    {
      "epoch": 0.342,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0317,
      "step": 14535
    },
    {
      "epoch": 0.3420235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.5805,
      "step": 14536
    },
    {
      "epoch": 0.3420470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 14537
    },
    {
      "epoch": 0.3420705882352941,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7511,
      "step": 14538
    },
    {
      "epoch": 0.3420941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9729,
      "step": 14539
    },
    {
      "epoch": 0.3421176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8251,
      "step": 14540
    },
    {
      "epoch": 0.3421411764705882,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1588,
      "step": 14541
    },
    {
      "epoch": 0.3421647058823529,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9044,
      "step": 14542
    },
    {
      "epoch": 0.3421882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7665,
      "step": 14543
    },
    {
      "epoch": 0.3422117647058823,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7199,
      "step": 14544
    },
    {
      "epoch": 0.3422352941176471,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9469,
      "step": 14545
    },
    {
      "epoch": 0.3422588235294118,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.994,
      "step": 14546
    },
    {
      "epoch": 0.3422823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1617,
      "step": 14547
    },
    {
      "epoch": 0.3423058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 0.9652,
      "step": 14548
    },
    {
      "epoch": 0.3423294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9657,
      "step": 14549
    },
    {
      "epoch": 0.3423529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.038,
      "step": 14550
    },
    {
      "epoch": 0.3423764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0176,
      "step": 14551
    },
    {
      "epoch": 0.3424,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2786,
      "step": 14552
    },
    {
      "epoch": 0.3424235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.876,
      "step": 14553
    },
    {
      "epoch": 0.34244705882352944,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.7261,
      "step": 14554
    },
    {
      "epoch": 0.34247058823529414,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.232,
      "step": 14555
    },
    {
      "epoch": 0.34249411764705884,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.888,
      "step": 14556
    },
    {
      "epoch": 0.34251764705882354,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.983,
      "step": 14557
    },
    {
      "epoch": 0.34254117647058824,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3497,
      "step": 14558
    },
    {
      "epoch": 0.34256470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0274,
      "step": 14559
    },
    {
      "epoch": 0.34258823529411764,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.9695,
      "step": 14560
    },
    {
      "epoch": 0.34261176470588234,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0703,
      "step": 14561
    },
    {
      "epoch": 0.34263529411764704,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0363,
      "step": 14562
    },
    {
      "epoch": 0.34265882352941174,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1396,
      "step": 14563
    },
    {
      "epoch": 0.3426823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3734,
      "step": 14564
    },
    {
      "epoch": 0.3427058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8198,
      "step": 14565
    },
    {
      "epoch": 0.3427294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0716,
      "step": 14566
    },
    {
      "epoch": 0.3427529411764706,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7571,
      "step": 14567
    },
    {
      "epoch": 0.3427764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8908,
      "step": 14568
    },
    {
      "epoch": 0.3428,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.124,
      "step": 14569
    },
    {
      "epoch": 0.3428235294117647,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.4042,
      "step": 14570
    },
    {
      "epoch": 0.3428470588235294,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2685,
      "step": 14571
    },
    {
      "epoch": 0.3428705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9116,
      "step": 14572
    },
    {
      "epoch": 0.34289411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.081,
      "step": 14573
    },
    {
      "epoch": 0.34291764705882355,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9731,
      "step": 14574
    },
    {
      "epoch": 0.34294117647058825,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8486,
      "step": 14575
    },
    {
      "epoch": 0.34296470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9307,
      "step": 14576
    },
    {
      "epoch": 0.34298823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 14577
    },
    {
      "epoch": 0.34301176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7212,
      "step": 14578
    },
    {
      "epoch": 0.34303529411764705,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8751,
      "step": 14579
    },
    {
      "epoch": 0.34305882352941175,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2005,
      "step": 14580
    },
    {
      "epoch": 0.34308235294117645,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1443,
      "step": 14581
    },
    {
      "epoch": 0.34310588235294115,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9486,
      "step": 14582
    },
    {
      "epoch": 0.3431294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 14583
    },
    {
      "epoch": 0.3431529411764706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.8945,
      "step": 14584
    },
    {
      "epoch": 0.3431764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0056,
      "step": 14585
    },
    {
      "epoch": 0.3432,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3054,
      "step": 14586
    },
    {
      "epoch": 0.3432235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.061,
      "step": 14587
    },
    {
      "epoch": 0.3432470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.763,
      "step": 14588
    },
    {
      "epoch": 0.3432705882352941,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.657,
      "step": 14589
    },
    {
      "epoch": 0.3432941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7981,
      "step": 14590
    },
    {
      "epoch": 0.3433176470588235,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.912,
      "step": 14591
    },
    {
      "epoch": 0.34334117647058826,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7925,
      "step": 14592
    },
    {
      "epoch": 0.34336470588235296,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1285,
      "step": 14593
    },
    {
      "epoch": 0.34338823529411766,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3079,
      "step": 14594
    },
    {
      "epoch": 0.34341176470588236,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9466,
      "step": 14595
    },
    {
      "epoch": 0.34343529411764706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0663,
      "step": 14596
    },
    {
      "epoch": 0.34345882352941176,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9695,
      "step": 14597
    },
    {
      "epoch": 0.34348235294117646,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1096,
      "step": 14598
    },
    {
      "epoch": 0.34350588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1427,
      "step": 14599
    },
    {
      "epoch": 0.34352941176470586,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2424,
      "step": 14600
    },
    {
      "epoch": 0.34355294117647056,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.5366,
      "step": 14601
    },
    {
      "epoch": 0.3435764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1405,
      "step": 14602
    },
    {
      "epoch": 0.3436,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.6496,
      "step": 14603
    },
    {
      "epoch": 0.3436235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0091,
      "step": 14604
    },
    {
      "epoch": 0.3436470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9812,
      "step": 14605
    },
    {
      "epoch": 0.3436705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.047,
      "step": 14606
    },
    {
      "epoch": 0.3436941176470588,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2446,
      "step": 14607
    },
    {
      "epoch": 0.3437176470588235,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9667,
      "step": 14608
    },
    {
      "epoch": 0.3437411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0069,
      "step": 14609
    },
    {
      "epoch": 0.3437647058823529,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1335,
      "step": 14610
    },
    {
      "epoch": 0.34378823529411767,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0452,
      "step": 14611
    },
    {
      "epoch": 0.3438117647058824,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0295,
      "step": 14612
    },
    {
      "epoch": 0.3438352941176471,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0881,
      "step": 14613
    },
    {
      "epoch": 0.3438588235294118,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0171,
      "step": 14614
    },
    {
      "epoch": 0.3438823529411765,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1811,
      "step": 14615
    },
    {
      "epoch": 0.3439058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1416,
      "step": 14616
    },
    {
      "epoch": 0.3439294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1028,
      "step": 14617
    },
    {
      "epoch": 0.3439529411764706,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.933,
      "step": 14618
    },
    {
      "epoch": 0.3439764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2128,
      "step": 14619
    },
    {
      "epoch": 0.344,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0596,
      "step": 14620
    },
    {
      "epoch": 0.34402352941176473,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0448,
      "step": 14621
    },
    {
      "epoch": 0.34404705882352943,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0425,
      "step": 14622
    },
    {
      "epoch": 0.34407058823529413,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3792,
      "step": 14623
    },
    {
      "epoch": 0.34409411764705883,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0806,
      "step": 14624
    },
    {
      "epoch": 0.34411764705882353,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.947,
      "step": 14625
    },
    {
      "epoch": 0.34414117647058823,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1319,
      "step": 14626
    },
    {
      "epoch": 0.34416470588235293,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8667,
      "step": 14627
    },
    {
      "epoch": 0.34418823529411763,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0517,
      "step": 14628
    },
    {
      "epoch": 0.34421176470588233,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.137,
      "step": 14629
    },
    {
      "epoch": 0.3442352941176471,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2278,
      "step": 14630
    },
    {
      "epoch": 0.3442588235294118,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8342,
      "step": 14631
    },
    {
      "epoch": 0.3442823529411765,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0907,
      "step": 14632
    },
    {
      "epoch": 0.3443058823529412,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9802,
      "step": 14633
    },
    {
      "epoch": 0.3443294117647059,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.249,
      "step": 14634
    },
    {
      "epoch": 0.3443529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9922,
      "step": 14635
    },
    {
      "epoch": 0.3443764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0618,
      "step": 14636
    },
    {
      "epoch": 0.3444,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.5592,
      "step": 14637
    },
    {
      "epoch": 0.3444235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 14638
    },
    {
      "epoch": 0.34444705882352944,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6981,
      "step": 14639
    },
    {
      "epoch": 0.34447058823529414,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9189,
      "step": 14640
    },
    {
      "epoch": 0.34449411764705884,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8928,
      "step": 14641
    },
    {
      "epoch": 0.34451764705882354,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.946,
      "step": 14642
    },
    {
      "epoch": 0.34454117647058824,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.743,
      "step": 14643
    },
    {
      "epoch": 0.34456470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.056,
      "step": 14644
    },
    {
      "epoch": 0.34458823529411764,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0017,
      "step": 14645
    },
    {
      "epoch": 0.34461176470588234,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8639,
      "step": 14646
    },
    {
      "epoch": 0.34463529411764704,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8214,
      "step": 14647
    },
    {
      "epoch": 0.34465882352941174,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8291,
      "step": 14648
    },
    {
      "epoch": 0.3446823529411765,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2024,
      "step": 14649
    },
    {
      "epoch": 0.3447058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 14650
    },
    {
      "epoch": 0.3447294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1356,
      "step": 14651
    },
    {
      "epoch": 0.3447529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1017,
      "step": 14652
    },
    {
      "epoch": 0.3447764705882353,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1469,
      "step": 14653
    },
    {
      "epoch": 0.3448,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2793,
      "step": 14654
    },
    {
      "epoch": 0.3448235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9845,
      "step": 14655
    },
    {
      "epoch": 0.3448470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0224,
      "step": 14656
    },
    {
      "epoch": 0.3448705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 14657
    },
    {
      "epoch": 0.34489411764705885,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8561,
      "step": 14658
    },
    {
      "epoch": 0.34491764705882355,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8163,
      "step": 14659
    },
    {
      "epoch": 0.34494117647058825,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2267,
      "step": 14660
    },
    {
      "epoch": 0.34496470588235295,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8916,
      "step": 14661
    },
    {
      "epoch": 0.34498823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1009,
      "step": 14662
    },
    {
      "epoch": 0.34501176470588235,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2426,
      "step": 14663
    },
    {
      "epoch": 0.34503529411764705,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1332,
      "step": 14664
    },
    {
      "epoch": 0.34505882352941175,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 14665
    },
    {
      "epoch": 0.34508235294117645,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8969,
      "step": 14666
    },
    {
      "epoch": 0.34510588235294115,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1278,
      "step": 14667
    },
    {
      "epoch": 0.3451294117647059,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.4362,
      "step": 14668
    },
    {
      "epoch": 0.3451529411764706,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8197,
      "step": 14669
    },
    {
      "epoch": 0.3451764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9808,
      "step": 14670
    },
    {
      "epoch": 0.3452,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.4257,
      "step": 14671
    },
    {
      "epoch": 0.3452235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0949,
      "step": 14672
    },
    {
      "epoch": 0.3452470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1862,
      "step": 14673
    },
    {
      "epoch": 0.3452705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1879,
      "step": 14674
    },
    {
      "epoch": 0.3452941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1509,
      "step": 14675
    },
    {
      "epoch": 0.3453176470588235,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 0.5386,
      "step": 14676
    },
    {
      "epoch": 0.34534117647058826,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0038,
      "step": 14677
    },
    {
      "epoch": 0.34536470588235296,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 14678
    },
    {
      "epoch": 0.34538823529411766,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8302,
      "step": 14679
    },
    {
      "epoch": 0.34541176470588236,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9359,
      "step": 14680
    },
    {
      "epoch": 0.34543529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2418,
      "step": 14681
    },
    {
      "epoch": 0.34545882352941176,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.693,
      "step": 14682
    },
    {
      "epoch": 0.34548235294117646,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8946,
      "step": 14683
    },
    {
      "epoch": 0.34550588235294116,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0956,
      "step": 14684
    },
    {
      "epoch": 0.34552941176470586,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0787,
      "step": 14685
    },
    {
      "epoch": 0.34555294117647056,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2569,
      "step": 14686
    },
    {
      "epoch": 0.3455764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9144,
      "step": 14687
    },
    {
      "epoch": 0.3456,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2016,
      "step": 14688
    },
    {
      "epoch": 0.3456235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1427,
      "step": 14689
    },
    {
      "epoch": 0.3456470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0273,
      "step": 14690
    },
    {
      "epoch": 0.3456705882352941,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1368,
      "step": 14691
    },
    {
      "epoch": 0.3456941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0075,
      "step": 14692
    },
    {
      "epoch": 0.3457176470588235,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0701,
      "step": 14693
    },
    {
      "epoch": 0.3457411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1998,
      "step": 14694
    },
    {
      "epoch": 0.3457647058823529,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.9749,
      "step": 14695
    },
    {
      "epoch": 0.3457882352941177,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.6191,
      "step": 14696
    },
    {
      "epoch": 0.3458117647058824,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2518,
      "step": 14697
    },
    {
      "epoch": 0.3458352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1053,
      "step": 14698
    },
    {
      "epoch": 0.3458588235294118,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2221,
      "step": 14699
    },
    {
      "epoch": 0.3458823529411765,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7109,
      "step": 14700
    },
    {
      "epoch": 0.3459058823529412,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.683,
      "step": 14701
    },
    {
      "epoch": 0.3459294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1703,
      "step": 14702
    },
    {
      "epoch": 0.3459529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.919,
      "step": 14703
    },
    {
      "epoch": 0.3459764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.6978,
      "step": 14704
    },
    {
      "epoch": 0.346,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.093,
      "step": 14705
    },
    {
      "epoch": 0.34602352941176473,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0188,
      "step": 14706
    },
    {
      "epoch": 0.34604705882352943,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3225,
      "step": 14707
    },
    {
      "epoch": 0.34607058823529413,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.4632,
      "step": 14708
    },
    {
      "epoch": 0.34609411764705883,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0379,
      "step": 14709
    },
    {
      "epoch": 0.34611764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1147,
      "step": 14710
    },
    {
      "epoch": 0.34614117647058823,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2522,
      "step": 14711
    },
    {
      "epoch": 0.34616470588235293,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3635,
      "step": 14712
    },
    {
      "epoch": 0.34618823529411763,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.052,
      "step": 14713
    },
    {
      "epoch": 0.34621176470588233,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0446,
      "step": 14714
    },
    {
      "epoch": 0.3462352941176471,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2707,
      "step": 14715
    },
    {
      "epoch": 0.3462588235294118,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0714,
      "step": 14716
    },
    {
      "epoch": 0.3462823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3447,
      "step": 14717
    },
    {
      "epoch": 0.3463058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0054,
      "step": 14718
    },
    {
      "epoch": 0.3463294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8585,
      "step": 14719
    },
    {
      "epoch": 0.3463529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0814,
      "step": 14720
    },
    {
      "epoch": 0.3463764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7201,
      "step": 14721
    },
    {
      "epoch": 0.3464,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0054,
      "step": 14722
    },
    {
      "epoch": 0.3464235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9253,
      "step": 14723
    },
    {
      "epoch": 0.3464470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1329,
      "step": 14724
    },
    {
      "epoch": 0.34647058823529414,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1937,
      "step": 14725
    },
    {
      "epoch": 0.34649411764705884,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.6719,
      "step": 14726
    },
    {
      "epoch": 0.34651764705882354,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8591,
      "step": 14727
    },
    {
      "epoch": 0.34654117647058824,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0923,
      "step": 14728
    },
    {
      "epoch": 0.34656470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8764,
      "step": 14729
    },
    {
      "epoch": 0.34658823529411764,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.7486,
      "step": 14730
    },
    {
      "epoch": 0.34661176470588234,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9726,
      "step": 14731
    },
    {
      "epoch": 0.34663529411764704,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2985,
      "step": 14732
    },
    {
      "epoch": 0.34665882352941174,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8898,
      "step": 14733
    },
    {
      "epoch": 0.3466823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.678,
      "step": 14734
    },
    {
      "epoch": 0.3467058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3306,
      "step": 14735
    },
    {
      "epoch": 0.3467294117647059,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9132,
      "step": 14736
    },
    {
      "epoch": 0.3467529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.509,
      "step": 14737
    },
    {
      "epoch": 0.3467764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0282,
      "step": 14738
    },
    {
      "epoch": 0.3468,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8521,
      "step": 14739
    },
    {
      "epoch": 0.3468235294117647,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.305,
      "step": 14740
    },
    {
      "epoch": 0.3468470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0462,
      "step": 14741
    },
    {
      "epoch": 0.3468705882352941,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1046,
      "step": 14742
    },
    {
      "epoch": 0.3468941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 14743
    },
    {
      "epoch": 0.34691764705882355,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1247,
      "step": 14744
    },
    {
      "epoch": 0.34694117647058825,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0621,
      "step": 14745
    },
    {
      "epoch": 0.34696470588235295,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 14746
    },
    {
      "epoch": 0.34698823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8164,
      "step": 14747
    },
    {
      "epoch": 0.34701176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9106,
      "step": 14748
    },
    {
      "epoch": 0.34703529411764705,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0772,
      "step": 14749
    },
    {
      "epoch": 0.34705882352941175,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1906,
      "step": 14750
    },
    {
      "epoch": 0.34708235294117645,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8949,
      "step": 14751
    },
    {
      "epoch": 0.34710588235294115,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1082,
      "step": 14752
    },
    {
      "epoch": 0.3471294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2494,
      "step": 14753
    },
    {
      "epoch": 0.3471529411764706,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9852,
      "step": 14754
    },
    {
      "epoch": 0.3471764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 0.854,
      "step": 14755
    },
    {
      "epoch": 0.3472,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2139,
      "step": 14756
    },
    {
      "epoch": 0.3472235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9211,
      "step": 14757
    },
    {
      "epoch": 0.3472470588235294,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9735,
      "step": 14758
    },
    {
      "epoch": 0.3472705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.6633,
      "step": 14759
    },
    {
      "epoch": 0.3472941176470588,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.189,
      "step": 14760
    },
    {
      "epoch": 0.3473176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.094,
      "step": 14761
    },
    {
      "epoch": 0.3473411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0271,
      "step": 14762
    },
    {
      "epoch": 0.34736470588235296,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0185,
      "step": 14763
    },
    {
      "epoch": 0.34738823529411766,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.929,
      "step": 14764
    },
    {
      "epoch": 0.34741176470588236,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2649,
      "step": 14765
    },
    {
      "epoch": 0.34743529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2549,
      "step": 14766
    },
    {
      "epoch": 0.34745882352941176,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1218,
      "step": 14767
    },
    {
      "epoch": 0.34748235294117646,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9513,
      "step": 14768
    },
    {
      "epoch": 0.34750588235294116,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9382,
      "step": 14769
    },
    {
      "epoch": 0.34752941176470586,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6983,
      "step": 14770
    },
    {
      "epoch": 0.34755294117647056,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2754,
      "step": 14771
    },
    {
      "epoch": 0.3475764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.6031,
      "step": 14772
    },
    {
      "epoch": 0.3476,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9976,
      "step": 14773
    },
    {
      "epoch": 0.3476235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 14774
    },
    {
      "epoch": 0.3476470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9188,
      "step": 14775
    },
    {
      "epoch": 0.3476705882352941,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3436,
      "step": 14776
    },
    {
      "epoch": 0.3476941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1309,
      "step": 14777
    },
    {
      "epoch": 0.3477176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 14778
    },
    {
      "epoch": 0.3477411764705882,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.127,
      "step": 14779
    },
    {
      "epoch": 0.3477647058823529,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0643,
      "step": 14780
    },
    {
      "epoch": 0.3477882352941176,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0171,
      "step": 14781
    },
    {
      "epoch": 0.3478117647058824,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0969,
      "step": 14782
    },
    {
      "epoch": 0.3478352941176471,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0601,
      "step": 14783
    },
    {
      "epoch": 0.3478588235294118,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.8184,
      "step": 14784
    },
    {
      "epoch": 0.3478823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.135,
      "step": 14785
    },
    {
      "epoch": 0.3479058823529412,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 14786
    },
    {
      "epoch": 0.3479294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9652,
      "step": 14787
    },
    {
      "epoch": 0.3479529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8486,
      "step": 14788
    },
    {
      "epoch": 0.3479764705882353,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.3406,
      "step": 14789
    },
    {
      "epoch": 0.348,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9184,
      "step": 14790
    },
    {
      "epoch": 0.34802352941176473,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1428,
      "step": 14791
    },
    {
      "epoch": 0.34804705882352943,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1012,
      "step": 14792
    },
    {
      "epoch": 0.34807058823529413,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8945,
      "step": 14793
    },
    {
      "epoch": 0.34809411764705883,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.115,
      "step": 14794
    },
    {
      "epoch": 0.34811764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1275,
      "step": 14795
    },
    {
      "epoch": 0.34814117647058823,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2718,
      "step": 14796
    },
    {
      "epoch": 0.34816470588235293,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0154,
      "step": 14797
    },
    {
      "epoch": 0.34818823529411763,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8968,
      "step": 14798
    },
    {
      "epoch": 0.34821176470588233,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2223,
      "step": 14799
    },
    {
      "epoch": 0.34823529411764703,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0379,
      "step": 14800
    },
    {
      "epoch": 0.3482588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9566,
      "step": 14801
    },
    {
      "epoch": 0.3482823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1279,
      "step": 14802
    },
    {
      "epoch": 0.3483058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7977,
      "step": 14803
    },
    {
      "epoch": 0.3483294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1764,
      "step": 14804
    },
    {
      "epoch": 0.3483529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6743,
      "step": 14805
    },
    {
      "epoch": 0.3483764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9875,
      "step": 14806
    },
    {
      "epoch": 0.3484,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.9524,
      "step": 14807
    },
    {
      "epoch": 0.3484235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 14808
    },
    {
      "epoch": 0.3484470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0477,
      "step": 14809
    },
    {
      "epoch": 0.34847058823529414,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6996,
      "step": 14810
    },
    {
      "epoch": 0.34849411764705884,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2967,
      "step": 14811
    },
    {
      "epoch": 0.34851764705882354,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0078,
      "step": 14812
    },
    {
      "epoch": 0.34854117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1103,
      "step": 14813
    },
    {
      "epoch": 0.34856470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2477,
      "step": 14814
    },
    {
      "epoch": 0.34858823529411764,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0655,
      "step": 14815
    },
    {
      "epoch": 0.34861176470588234,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.2098,
      "step": 14816
    },
    {
      "epoch": 0.34863529411764704,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 14817
    },
    {
      "epoch": 0.34865882352941174,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9102,
      "step": 14818
    },
    {
      "epoch": 0.34868235294117644,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8862,
      "step": 14819
    },
    {
      "epoch": 0.3487058823529412,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3166,
      "step": 14820
    },
    {
      "epoch": 0.3487294117647059,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2102,
      "step": 14821
    },
    {
      "epoch": 0.3487529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.116,
      "step": 14822
    },
    {
      "epoch": 0.3487764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 14823
    },
    {
      "epoch": 0.3488,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.386,
      "step": 14824
    },
    {
      "epoch": 0.3488235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.024,
      "step": 14825
    },
    {
      "epoch": 0.3488470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2182,
      "step": 14826
    },
    {
      "epoch": 0.3488705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.086,
      "step": 14827
    },
    {
      "epoch": 0.3488941176470588,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9069,
      "step": 14828
    },
    {
      "epoch": 0.34891764705882355,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0311,
      "step": 14829
    },
    {
      "epoch": 0.34894117647058825,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9018,
      "step": 14830
    },
    {
      "epoch": 0.34896470588235295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 14831
    },
    {
      "epoch": 0.34898823529411765,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.02,
      "loss": 1.5229,
      "step": 14832
    },
    {
      "epoch": 0.34901176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1805,
      "step": 14833
    },
    {
      "epoch": 0.34903529411764705,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1664,
      "step": 14834
    },
    {
      "epoch": 0.34905882352941175,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9009,
      "step": 14835
    },
    {
      "epoch": 0.34908235294117645,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0779,
      "step": 14836
    },
    {
      "epoch": 0.34910588235294115,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9033,
      "step": 14837
    },
    {
      "epoch": 0.34912941176470585,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 14838
    },
    {
      "epoch": 0.3491529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1482,
      "step": 14839
    },
    {
      "epoch": 0.3491764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8617,
      "step": 14840
    },
    {
      "epoch": 0.3492,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.007,
      "step": 14841
    },
    {
      "epoch": 0.3492235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.076,
      "step": 14842
    },
    {
      "epoch": 0.3492470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8552,
      "step": 14843
    },
    {
      "epoch": 0.3492705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1705,
      "step": 14844
    },
    {
      "epoch": 0.3492941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9766,
      "step": 14845
    },
    {
      "epoch": 0.3493176470588235,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3802,
      "step": 14846
    },
    {
      "epoch": 0.3493411764705882,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0793,
      "step": 14847
    },
    {
      "epoch": 0.34936470588235297,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1925,
      "step": 14848
    },
    {
      "epoch": 0.34938823529411767,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1216,
      "step": 14849
    },
    {
      "epoch": 0.34941176470588237,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8725,
      "step": 14850
    },
    {
      "epoch": 0.34943529411764707,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9556,
      "step": 14851
    },
    {
      "epoch": 0.34945882352941177,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0462,
      "step": 14852
    },
    {
      "epoch": 0.34948235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7518,
      "step": 14853
    },
    {
      "epoch": 0.34950588235294117,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.145,
      "step": 14854
    },
    {
      "epoch": 0.34952941176470587,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.0811,
      "step": 14855
    },
    {
      "epoch": 0.34955294117647057,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0011,
      "step": 14856
    },
    {
      "epoch": 0.3495764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9475,
      "step": 14857
    },
    {
      "epoch": 0.3496,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0993,
      "step": 14858
    },
    {
      "epoch": 0.3496235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.872,
      "step": 14859
    },
    {
      "epoch": 0.3496470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7132,
      "step": 14860
    },
    {
      "epoch": 0.3496705882352941,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0482,
      "step": 14861
    },
    {
      "epoch": 0.3496941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9507,
      "step": 14862
    },
    {
      "epoch": 0.3497176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 14863
    },
    {
      "epoch": 0.3497411764705882,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.885,
      "step": 14864
    },
    {
      "epoch": 0.3497647058823529,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8383,
      "step": 14865
    },
    {
      "epoch": 0.3497882352941176,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7516,
      "step": 14866
    },
    {
      "epoch": 0.3498117647058824,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.4214,
      "step": 14867
    },
    {
      "epoch": 0.3498352941176471,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0297,
      "step": 14868
    },
    {
      "epoch": 0.3498588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9269,
      "step": 14869
    },
    {
      "epoch": 0.3498823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9753,
      "step": 14870
    },
    {
      "epoch": 0.3499058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.224,
      "step": 14871
    },
    {
      "epoch": 0.3499294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9698,
      "step": 14872
    },
    {
      "epoch": 0.3499529411764706,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6965,
      "step": 14873
    },
    {
      "epoch": 0.3499764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1807,
      "step": 14874
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0534,
      "step": 14875
    },
    {
      "epoch": 0.35002352941176473,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 14876
    },
    {
      "epoch": 0.35004705882352943,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7168,
      "step": 14877
    },
    {
      "epoch": 0.35007058823529413,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8729,
      "step": 14878
    },
    {
      "epoch": 0.35009411764705883,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.6464,
      "step": 14879
    },
    {
      "epoch": 0.35011764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.083,
      "step": 14880
    },
    {
      "epoch": 0.35014117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9225,
      "step": 14881
    },
    {
      "epoch": 0.35016470588235293,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1109,
      "step": 14882
    },
    {
      "epoch": 0.35018823529411763,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9329,
      "step": 14883
    },
    {
      "epoch": 0.35021176470588233,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9628,
      "step": 14884
    },
    {
      "epoch": 0.35023529411764703,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.236,
      "step": 14885
    },
    {
      "epoch": 0.3502588235294118,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.404,
      "step": 14886
    },
    {
      "epoch": 0.3502823529411765,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2126,
      "step": 14887
    },
    {
      "epoch": 0.3503058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.984,
      "step": 14888
    },
    {
      "epoch": 0.3503294117647059,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1583,
      "step": 14889
    },
    {
      "epoch": 0.3503529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0841,
      "step": 14890
    },
    {
      "epoch": 0.3503764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9189,
      "step": 14891
    },
    {
      "epoch": 0.3504,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9016,
      "step": 14892
    },
    {
      "epoch": 0.3504235294117647,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.3158,
      "step": 14893
    },
    {
      "epoch": 0.3504470588235294,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 14894
    },
    {
      "epoch": 0.35047058823529414,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9763,
      "step": 14895
    },
    {
      "epoch": 0.35049411764705884,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8922,
      "step": 14896
    },
    {
      "epoch": 0.35051764705882354,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.5032,
      "step": 14897
    },
    {
      "epoch": 0.35054117647058824,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3233,
      "step": 14898
    },
    {
      "epoch": 0.35056470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 14899
    },
    {
      "epoch": 0.35058823529411764,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2298,
      "step": 14900
    },
    {
      "epoch": 0.35061176470588234,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.149,
      "step": 14901
    },
    {
      "epoch": 0.35063529411764705,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2519,
      "step": 14902
    },
    {
      "epoch": 0.35065882352941175,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.1933,
      "step": 14903
    },
    {
      "epoch": 0.35068235294117645,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9637,
      "step": 14904
    },
    {
      "epoch": 0.3507058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9738,
      "step": 14905
    },
    {
      "epoch": 0.3507294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7818,
      "step": 14906
    },
    {
      "epoch": 0.3507529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0799,
      "step": 14907
    },
    {
      "epoch": 0.3507764705882353,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2587,
      "step": 14908
    },
    {
      "epoch": 0.3508,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.5152,
      "step": 14909
    },
    {
      "epoch": 0.3508235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.072,
      "step": 14910
    },
    {
      "epoch": 0.3508470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9196,
      "step": 14911
    },
    {
      "epoch": 0.3508705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8131,
      "step": 14912
    },
    {
      "epoch": 0.3508941176470588,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8241,
      "step": 14913
    },
    {
      "epoch": 0.35091764705882356,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9393,
      "step": 14914
    },
    {
      "epoch": 0.35094117647058826,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7808,
      "step": 14915
    },
    {
      "epoch": 0.35096470588235296,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8602,
      "step": 14916
    },
    {
      "epoch": 0.35098823529411766,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7908,
      "step": 14917
    },
    {
      "epoch": 0.35101176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9801,
      "step": 14918
    },
    {
      "epoch": 0.35103529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.025,
      "step": 14919
    },
    {
      "epoch": 0.35105882352941176,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7558,
      "step": 14920
    },
    {
      "epoch": 0.35108235294117646,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8163,
      "step": 14921
    },
    {
      "epoch": 0.35110588235294116,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.992,
      "step": 14922
    },
    {
      "epoch": 0.35112941176470586,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8717,
      "step": 14923
    },
    {
      "epoch": 0.3511529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7961,
      "step": 14924
    },
    {
      "epoch": 0.3511764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7096,
      "step": 14925
    },
    {
      "epoch": 0.3512,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9874,
      "step": 14926
    },
    {
      "epoch": 0.3512235294117647,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 14927
    },
    {
      "epoch": 0.3512470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0856,
      "step": 14928
    },
    {
      "epoch": 0.3512705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0325,
      "step": 14929
    },
    {
      "epoch": 0.3512941176470588,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9077,
      "step": 14930
    },
    {
      "epoch": 0.3513176470588235,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2869,
      "step": 14931
    },
    {
      "epoch": 0.3513411764705882,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1424,
      "step": 14932
    },
    {
      "epoch": 0.35136470588235297,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0267,
      "step": 14933
    },
    {
      "epoch": 0.35138823529411767,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2037,
      "step": 14934
    },
    {
      "epoch": 0.35141176470588237,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 14935
    },
    {
      "epoch": 0.35143529411764707,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9723,
      "step": 14936
    },
    {
      "epoch": 0.35145882352941177,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.9902,
      "step": 14937
    },
    {
      "epoch": 0.35148235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8987,
      "step": 14938
    },
    {
      "epoch": 0.35150588235294117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1945,
      "step": 14939
    },
    {
      "epoch": 0.35152941176470587,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2353,
      "step": 14940
    },
    {
      "epoch": 0.35155294117647057,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 1.2028,
      "step": 14941
    },
    {
      "epoch": 0.35157647058823527,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0349,
      "step": 14942
    },
    {
      "epoch": 0.3516,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8111,
      "step": 14943
    },
    {
      "epoch": 0.3516235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.054,
      "step": 14944
    },
    {
      "epoch": 0.3516470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.6529,
      "step": 14945
    },
    {
      "epoch": 0.3516705882352941,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0445,
      "step": 14946
    },
    {
      "epoch": 0.3516941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9751,
      "step": 14947
    },
    {
      "epoch": 0.3517176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7348,
      "step": 14948
    },
    {
      "epoch": 0.3517411764705882,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.5094,
      "step": 14949
    },
    {
      "epoch": 0.3517647058823529,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0198,
      "step": 14950
    },
    {
      "epoch": 0.3517882352941176,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3446,
      "step": 14951
    },
    {
      "epoch": 0.3518117647058824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8686,
      "step": 14952
    },
    {
      "epoch": 0.3518352941176471,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0063,
      "step": 14953
    },
    {
      "epoch": 0.3518588235294118,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9025,
      "step": 14954
    },
    {
      "epoch": 0.3518823529411765,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.47,
      "step": 14955
    },
    {
      "epoch": 0.3519058823529412,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 0.9338,
      "step": 14956
    },
    {
      "epoch": 0.3519294117647059,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1598,
      "step": 14957
    },
    {
      "epoch": 0.3519529411764706,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2018,
      "step": 14958
    },
    {
      "epoch": 0.3519764705882353,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1263,
      "step": 14959
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9481,
      "step": 14960
    },
    {
      "epoch": 0.3520235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1605,
      "step": 14961
    },
    {
      "epoch": 0.35204705882352944,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0171,
      "step": 14962
    },
    {
      "epoch": 0.35207058823529414,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7999,
      "step": 14963
    },
    {
      "epoch": 0.35209411764705884,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1527,
      "step": 14964
    },
    {
      "epoch": 0.35211764705882354,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.6106,
      "step": 14965
    },
    {
      "epoch": 0.35214117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7767,
      "step": 14966
    },
    {
      "epoch": 0.35216470588235294,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.7777,
      "step": 14967
    },
    {
      "epoch": 0.35218823529411764,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.077,
      "step": 14968
    },
    {
      "epoch": 0.35221176470588234,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8662,
      "step": 14969
    },
    {
      "epoch": 0.35223529411764704,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.175,
      "step": 14970
    },
    {
      "epoch": 0.3522588235294118,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.8095,
      "step": 14971
    },
    {
      "epoch": 0.3522823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0571,
      "step": 14972
    },
    {
      "epoch": 0.3523058823529412,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7044,
      "step": 14973
    },
    {
      "epoch": 0.3523294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9643,
      "step": 14974
    },
    {
      "epoch": 0.3523529411764706,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3138,
      "step": 14975
    },
    {
      "epoch": 0.3523764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7996,
      "step": 14976
    },
    {
      "epoch": 0.3524,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.873,
      "step": 14977
    },
    {
      "epoch": 0.3524235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8844,
      "step": 14978
    },
    {
      "epoch": 0.3524470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0275,
      "step": 14979
    },
    {
      "epoch": 0.3524705882352941,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0646,
      "step": 14980
    },
    {
      "epoch": 0.35249411764705885,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8223,
      "step": 14981
    },
    {
      "epoch": 0.35251764705882355,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2169,
      "step": 14982
    },
    {
      "epoch": 0.35254117647058825,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.199,
      "step": 14983
    },
    {
      "epoch": 0.35256470588235295,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8008,
      "step": 14984
    },
    {
      "epoch": 0.35258823529411765,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0038,
      "step": 14985
    },
    {
      "epoch": 0.35261176470588235,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1557,
      "step": 14986
    },
    {
      "epoch": 0.35263529411764705,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2039,
      "step": 14987
    },
    {
      "epoch": 0.35265882352941175,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.106,
      "step": 14988
    },
    {
      "epoch": 0.35268235294117645,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2241,
      "step": 14989
    },
    {
      "epoch": 0.3527058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8938,
      "step": 14990
    },
    {
      "epoch": 0.3527294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.841,
      "step": 14991
    },
    {
      "epoch": 0.3527529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1678,
      "step": 14992
    },
    {
      "epoch": 0.3527764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1449,
      "step": 14993
    },
    {
      "epoch": 0.3528,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9605,
      "step": 14994
    },
    {
      "epoch": 0.3528235294117647,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1465,
      "step": 14995
    },
    {
      "epoch": 0.3528470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0588,
      "step": 14996
    },
    {
      "epoch": 0.3528705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0308,
      "step": 14997
    },
    {
      "epoch": 0.3528941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7951,
      "step": 14998
    },
    {
      "epoch": 0.3529176470588235,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8711,
      "step": 14999
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1968,
      "step": 15000
    },
    {
      "epoch": 0.35294117647058826,
      "eval_loss": 2.194629430770874,
      "eval_runtime": 681.2593,
      "eval_samples_per_second": 12.477,
      "eval_steps_per_second": 3.119,
      "step": 15000
    },
    {
      "epoch": 0.35296470588235296,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.5946,
      "step": 15001
    },
    {
      "epoch": 0.35298823529411766,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9766,
      "step": 15002
    },
    {
      "epoch": 0.35301176470588236,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1483,
      "step": 15003
    },
    {
      "epoch": 0.35303529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3054,
      "step": 15004
    },
    {
      "epoch": 0.35305882352941176,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2404,
      "step": 15005
    },
    {
      "epoch": 0.35308235294117646,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.5371,
      "step": 15006
    },
    {
      "epoch": 0.35310588235294116,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.8011,
      "step": 15007
    },
    {
      "epoch": 0.35312941176470586,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.896,
      "step": 15008
    },
    {
      "epoch": 0.3531529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3531,
      "step": 15009
    },
    {
      "epoch": 0.3531764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 15010
    },
    {
      "epoch": 0.3532,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3041,
      "step": 15011
    },
    {
      "epoch": 0.3532235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.108,
      "step": 15012
    },
    {
      "epoch": 0.3532470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7981,
      "step": 15013
    },
    {
      "epoch": 0.3532705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7653,
      "step": 15014
    },
    {
      "epoch": 0.3532941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8935,
      "step": 15015
    },
    {
      "epoch": 0.3533176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7329,
      "step": 15016
    },
    {
      "epoch": 0.3533411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0025,
      "step": 15017
    },
    {
      "epoch": 0.3533647058823529,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1722,
      "step": 15018
    },
    {
      "epoch": 0.35338823529411767,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9607,
      "step": 15019
    },
    {
      "epoch": 0.35341176470588237,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.633,
      "step": 15020
    },
    {
      "epoch": 0.35343529411764707,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9984,
      "step": 15021
    },
    {
      "epoch": 0.35345882352941177,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0844,
      "step": 15022
    },
    {
      "epoch": 0.35348235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9873,
      "step": 15023
    },
    {
      "epoch": 0.35350588235294117,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1697,
      "step": 15024
    },
    {
      "epoch": 0.35352941176470587,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.246,
      "step": 15025
    },
    {
      "epoch": 0.35355294117647057,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 15026
    },
    {
      "epoch": 0.35357647058823527,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.02,
      "step": 15027
    },
    {
      "epoch": 0.3536,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2146,
      "step": 15028
    },
    {
      "epoch": 0.3536235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.9368,
      "step": 15029
    },
    {
      "epoch": 0.3536470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.5428,
      "step": 15030
    },
    {
      "epoch": 0.3536705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9907,
      "step": 15031
    },
    {
      "epoch": 0.3536941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0374,
      "step": 15032
    },
    {
      "epoch": 0.3537176470588235,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9562,
      "step": 15033
    },
    {
      "epoch": 0.3537411764705882,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.02,
      "loss": 0.7895,
      "step": 15034
    },
    {
      "epoch": 0.3537647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0079,
      "step": 15035
    },
    {
      "epoch": 0.3537882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.036,
      "step": 15036
    },
    {
      "epoch": 0.3538117647058823,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.21,
      "step": 15037
    },
    {
      "epoch": 0.3538352941176471,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0431,
      "step": 15038
    },
    {
      "epoch": 0.3538588235294118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1807,
      "step": 15039
    },
    {
      "epoch": 0.3538823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.985,
      "step": 15040
    },
    {
      "epoch": 0.3539058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0206,
      "step": 15041
    },
    {
      "epoch": 0.3539294117647059,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8863,
      "step": 15042
    },
    {
      "epoch": 0.3539529411764706,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2395,
      "step": 15043
    },
    {
      "epoch": 0.3539764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3374,
      "step": 15044
    },
    {
      "epoch": 0.354,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.881,
      "step": 15045
    },
    {
      "epoch": 0.3540235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9188,
      "step": 15046
    },
    {
      "epoch": 0.35404705882352944,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2215,
      "step": 15047
    },
    {
      "epoch": 0.35407058823529414,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8985,
      "step": 15048
    },
    {
      "epoch": 0.35409411764705884,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8297,
      "step": 15049
    },
    {
      "epoch": 0.35411764705882354,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.6075,
      "step": 15050
    },
    {
      "epoch": 0.35414117647058824,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2754,
      "step": 15051
    },
    {
      "epoch": 0.35416470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0335,
      "step": 15052
    },
    {
      "epoch": 0.35418823529411764,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1397,
      "step": 15053
    },
    {
      "epoch": 0.35421176470588234,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 15054
    },
    {
      "epoch": 0.35423529411764704,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7231,
      "step": 15055
    },
    {
      "epoch": 0.35425882352941174,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9597,
      "step": 15056
    },
    {
      "epoch": 0.3542823529411765,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1333,
      "step": 15057
    },
    {
      "epoch": 0.3543058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1043,
      "step": 15058
    },
    {
      "epoch": 0.3543294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.953,
      "step": 15059
    },
    {
      "epoch": 0.3543529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6705,
      "step": 15060
    },
    {
      "epoch": 0.3543764705882353,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.3734,
      "step": 15061
    },
    {
      "epoch": 0.3544,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.18,
      "step": 15062
    },
    {
      "epoch": 0.3544235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8419,
      "step": 15063
    },
    {
      "epoch": 0.3544470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.909,
      "step": 15064
    },
    {
      "epoch": 0.3544705882352941,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2948,
      "step": 15065
    },
    {
      "epoch": 0.35449411764705885,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0274,
      "step": 15066
    },
    {
      "epoch": 0.35451764705882355,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.2405,
      "step": 15067
    },
    {
      "epoch": 0.35454117647058825,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.3404,
      "step": 15068
    },
    {
      "epoch": 0.35456470588235295,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8918,
      "step": 15069
    },
    {
      "epoch": 0.35458823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0125,
      "step": 15070
    },
    {
      "epoch": 0.35461176470588235,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7412,
      "step": 15071
    },
    {
      "epoch": 0.35463529411764705,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9821,
      "step": 15072
    },
    {
      "epoch": 0.35465882352941175,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1975,
      "step": 15073
    },
    {
      "epoch": 0.35468235294117645,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0944,
      "step": 15074
    },
    {
      "epoch": 0.3547058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7678,
      "step": 15075
    },
    {
      "epoch": 0.3547294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8675,
      "step": 15076
    },
    {
      "epoch": 0.3547529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1184,
      "step": 15077
    },
    {
      "epoch": 0.3547764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2436,
      "step": 15078
    },
    {
      "epoch": 0.3548,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7653,
      "step": 15079
    },
    {
      "epoch": 0.3548235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2284,
      "step": 15080
    },
    {
      "epoch": 0.3548470588235294,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7149,
      "step": 15081
    },
    {
      "epoch": 0.3548705882352941,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0758,
      "step": 15082
    },
    {
      "epoch": 0.3548941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9795,
      "step": 15083
    },
    {
      "epoch": 0.3549176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9963,
      "step": 15084
    },
    {
      "epoch": 0.35494117647058826,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.49,
      "step": 15085
    },
    {
      "epoch": 0.35496470588235296,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8693,
      "step": 15086
    },
    {
      "epoch": 0.35498823529411766,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 15087
    },
    {
      "epoch": 0.35501176470588236,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.6385,
      "step": 15088
    },
    {
      "epoch": 0.35503529411764706,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6182,
      "step": 15089
    },
    {
      "epoch": 0.35505882352941176,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 15090
    },
    {
      "epoch": 0.35508235294117646,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2079,
      "step": 15091
    },
    {
      "epoch": 0.35510588235294116,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0055,
      "step": 15092
    },
    {
      "epoch": 0.35512941176470586,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3694,
      "step": 15093
    },
    {
      "epoch": 0.3551529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9871,
      "step": 15094
    },
    {
      "epoch": 0.3551764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0159,
      "step": 15095
    },
    {
      "epoch": 0.3552,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7665,
      "step": 15096
    },
    {
      "epoch": 0.3552235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9762,
      "step": 15097
    },
    {
      "epoch": 0.3552470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8926,
      "step": 15098
    },
    {
      "epoch": 0.3552705882352941,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1109,
      "step": 15099
    },
    {
      "epoch": 0.3552941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8573,
      "step": 15100
    },
    {
      "epoch": 0.3553176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0723,
      "step": 15101
    },
    {
      "epoch": 0.3553411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1535,
      "step": 15102
    },
    {
      "epoch": 0.3553647058823529,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9747,
      "step": 15103
    },
    {
      "epoch": 0.35538823529411767,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8114,
      "step": 15104
    },
    {
      "epoch": 0.35541176470588237,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0046,
      "step": 15105
    },
    {
      "epoch": 0.35543529411764707,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8326,
      "step": 15106
    },
    {
      "epoch": 0.35545882352941177,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0298,
      "step": 15107
    },
    {
      "epoch": 0.35548235294117647,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9724,
      "step": 15108
    },
    {
      "epoch": 0.35550588235294117,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3121,
      "step": 15109
    },
    {
      "epoch": 0.35552941176470587,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1791,
      "step": 15110
    },
    {
      "epoch": 0.35555294117647057,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.584,
      "step": 15111
    },
    {
      "epoch": 0.35557647058823527,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9248,
      "step": 15112
    },
    {
      "epoch": 0.3556,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0698,
      "step": 15113
    },
    {
      "epoch": 0.3556235294117647,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6295,
      "step": 15114
    },
    {
      "epoch": 0.3556470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1837,
      "step": 15115
    },
    {
      "epoch": 0.3556705882352941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8008,
      "step": 15116
    },
    {
      "epoch": 0.3556941176470588,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.1412,
      "step": 15117
    },
    {
      "epoch": 0.3557176470588235,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1871,
      "step": 15118
    },
    {
      "epoch": 0.3557411764705882,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.781,
      "step": 15119
    },
    {
      "epoch": 0.3557647058823529,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2544,
      "step": 15120
    },
    {
      "epoch": 0.3557882352941176,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9646,
      "step": 15121
    },
    {
      "epoch": 0.3558117647058823,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8578,
      "step": 15122
    },
    {
      "epoch": 0.3558352941176471,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 15123
    },
    {
      "epoch": 0.3558588235294118,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.02,
      "loss": 0.5843,
      "step": 15124
    },
    {
      "epoch": 0.3558823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2193,
      "step": 15125
    },
    {
      "epoch": 0.3559058823529412,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.4393,
      "step": 15126
    },
    {
      "epoch": 0.3559294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.016,
      "step": 15127
    },
    {
      "epoch": 0.3559529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0113,
      "step": 15128
    },
    {
      "epoch": 0.3559764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9878,
      "step": 15129
    },
    {
      "epoch": 0.356,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7319,
      "step": 15130
    },
    {
      "epoch": 0.3560235294117647,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1387,
      "step": 15131
    },
    {
      "epoch": 0.35604705882352944,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.6292,
      "step": 15132
    },
    {
      "epoch": 0.35607058823529414,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8791,
      "step": 15133
    },
    {
      "epoch": 0.35609411764705884,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8612,
      "step": 15134
    },
    {
      "epoch": 0.35611764705882354,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.8886,
      "step": 15135
    },
    {
      "epoch": 0.35614117647058824,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8827,
      "step": 15136
    },
    {
      "epoch": 0.35616470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9822,
      "step": 15137
    },
    {
      "epoch": 0.35618823529411764,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.4126,
      "step": 15138
    },
    {
      "epoch": 0.35621176470588234,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9525,
      "step": 15139
    },
    {
      "epoch": 0.35623529411764704,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9653,
      "step": 15140
    },
    {
      "epoch": 0.35625882352941174,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.6066,
      "step": 15141
    },
    {
      "epoch": 0.3562823529411765,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0422,
      "step": 15142
    },
    {
      "epoch": 0.3563058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 15143
    },
    {
      "epoch": 0.3563294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 15144
    },
    {
      "epoch": 0.3563529411764706,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2617,
      "step": 15145
    },
    {
      "epoch": 0.3563764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7234,
      "step": 15146
    },
    {
      "epoch": 0.3564,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9767,
      "step": 15147
    },
    {
      "epoch": 0.3564235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9473,
      "step": 15148
    },
    {
      "epoch": 0.3564470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1262,
      "step": 15149
    },
    {
      "epoch": 0.3564705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9749,
      "step": 15150
    },
    {
      "epoch": 0.35649411764705885,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9159,
      "step": 15151
    },
    {
      "epoch": 0.35651764705882355,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7801,
      "step": 15152
    },
    {
      "epoch": 0.35654117647058825,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0509,
      "step": 15153
    },
    {
      "epoch": 0.35656470588235295,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2851,
      "step": 15154
    },
    {
      "epoch": 0.35658823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2336,
      "step": 15155
    },
    {
      "epoch": 0.35661176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0636,
      "step": 15156
    },
    {
      "epoch": 0.35663529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2516,
      "step": 15157
    },
    {
      "epoch": 0.35665882352941175,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7907,
      "step": 15158
    },
    {
      "epoch": 0.35668235294117645,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.0591,
      "step": 15159
    },
    {
      "epoch": 0.35670588235294115,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0888,
      "step": 15160
    },
    {
      "epoch": 0.3567294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0374,
      "step": 15161
    },
    {
      "epoch": 0.3567529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9385,
      "step": 15162
    },
    {
      "epoch": 0.3567764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0156,
      "step": 15163
    },
    {
      "epoch": 0.3568,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2951,
      "step": 15164
    },
    {
      "epoch": 0.3568235294117647,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7035,
      "step": 15165
    },
    {
      "epoch": 0.3568470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 15166
    },
    {
      "epoch": 0.3568705882352941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0698,
      "step": 15167
    },
    {
      "epoch": 0.3568941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1986,
      "step": 15168
    },
    {
      "epoch": 0.3569176470588235,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9722,
      "step": 15169
    },
    {
      "epoch": 0.35694117647058826,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1998,
      "step": 15170
    },
    {
      "epoch": 0.35696470588235296,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3522,
      "step": 15171
    },
    {
      "epoch": 0.35698823529411766,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.888,
      "step": 15172
    },
    {
      "epoch": 0.35701176470588236,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2988,
      "step": 15173
    },
    {
      "epoch": 0.35703529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0903,
      "step": 15174
    },
    {
      "epoch": 0.35705882352941176,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1194,
      "step": 15175
    },
    {
      "epoch": 0.35708235294117646,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9602,
      "step": 15176
    },
    {
      "epoch": 0.35710588235294116,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0945,
      "step": 15177
    },
    {
      "epoch": 0.35712941176470586,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8635,
      "step": 15178
    },
    {
      "epoch": 0.35715294117647056,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9445,
      "step": 15179
    },
    {
      "epoch": 0.3571764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8945,
      "step": 15180
    },
    {
      "epoch": 0.3572,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9407,
      "step": 15181
    },
    {
      "epoch": 0.3572235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0204,
      "step": 15182
    },
    {
      "epoch": 0.3572470588235294,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2743,
      "step": 15183
    },
    {
      "epoch": 0.3572705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8969,
      "step": 15184
    },
    {
      "epoch": 0.3572941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1397,
      "step": 15185
    },
    {
      "epoch": 0.3573176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9767,
      "step": 15186
    },
    {
      "epoch": 0.3573411764705882,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9598,
      "step": 15187
    },
    {
      "epoch": 0.3573647058823529,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3106,
      "step": 15188
    },
    {
      "epoch": 0.3573882352941177,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8418,
      "step": 15189
    },
    {
      "epoch": 0.3574117647058824,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2293,
      "step": 15190
    },
    {
      "epoch": 0.3574352941176471,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 15191
    },
    {
      "epoch": 0.3574588235294118,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1764,
      "step": 15192
    },
    {
      "epoch": 0.3574823529411765,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0108,
      "step": 15193
    },
    {
      "epoch": 0.3575058823529412,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2131,
      "step": 15194
    },
    {
      "epoch": 0.3575294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7687,
      "step": 15195
    },
    {
      "epoch": 0.3575529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0906,
      "step": 15196
    },
    {
      "epoch": 0.3575764705882353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.257,
      "step": 15197
    },
    {
      "epoch": 0.3576,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0881,
      "step": 15198
    },
    {
      "epoch": 0.35762352941176473,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1274,
      "step": 15199
    },
    {
      "epoch": 0.35764705882352943,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.021,
      "step": 15200
    },
    {
      "epoch": 0.35767058823529413,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 15201
    },
    {
      "epoch": 0.35769411764705883,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0594,
      "step": 15202
    },
    {
      "epoch": 0.35771764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9195,
      "step": 15203
    },
    {
      "epoch": 0.35774117647058823,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2118,
      "step": 15204
    },
    {
      "epoch": 0.35776470588235293,
      "grad_norm": 0.283203125,
      "learning_rate": 0.02,
      "loss": 0.7427,
      "step": 15205
    },
    {
      "epoch": 0.35778823529411763,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.956,
      "step": 15206
    },
    {
      "epoch": 0.35781176470588233,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9112,
      "step": 15207
    },
    {
      "epoch": 0.3578352941176471,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 15208
    },
    {
      "epoch": 0.3578588235294118,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.5768,
      "step": 15209
    },
    {
      "epoch": 0.3578823529411765,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0264,
      "step": 15210
    },
    {
      "epoch": 0.3579058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 15211
    },
    {
      "epoch": 0.3579294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.141,
      "step": 15212
    },
    {
      "epoch": 0.3579529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.953,
      "step": 15213
    },
    {
      "epoch": 0.3579764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0597,
      "step": 15214
    },
    {
      "epoch": 0.358,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9644,
      "step": 15215
    },
    {
      "epoch": 0.3580235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9437,
      "step": 15216
    },
    {
      "epoch": 0.3580470588235294,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9108,
      "step": 15217
    },
    {
      "epoch": 0.35807058823529414,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1062,
      "step": 15218
    },
    {
      "epoch": 0.35809411764705884,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.184,
      "step": 15219
    },
    {
      "epoch": 0.35811764705882354,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.6341,
      "step": 15220
    },
    {
      "epoch": 0.35814117647058824,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7964,
      "step": 15221
    },
    {
      "epoch": 0.35816470588235294,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9648,
      "step": 15222
    },
    {
      "epoch": 0.35818823529411764,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0082,
      "step": 15223
    },
    {
      "epoch": 0.35821176470588234,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.9396,
      "step": 15224
    },
    {
      "epoch": 0.35823529411764704,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.712,
      "step": 15225
    },
    {
      "epoch": 0.35825882352941174,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 15226
    },
    {
      "epoch": 0.3582823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0825,
      "step": 15227
    },
    {
      "epoch": 0.3583058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0182,
      "step": 15228
    },
    {
      "epoch": 0.3583294117647059,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1818,
      "step": 15229
    },
    {
      "epoch": 0.3583529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2203,
      "step": 15230
    },
    {
      "epoch": 0.3583764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.037,
      "step": 15231
    },
    {
      "epoch": 0.3584,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.191,
      "step": 15232
    },
    {
      "epoch": 0.3584235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0546,
      "step": 15233
    },
    {
      "epoch": 0.3584470588235294,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9174,
      "step": 15234
    },
    {
      "epoch": 0.3584705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8868,
      "step": 15235
    },
    {
      "epoch": 0.3584941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 15236
    },
    {
      "epoch": 0.35851764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0128,
      "step": 15237
    },
    {
      "epoch": 0.35854117647058825,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8745,
      "step": 15238
    },
    {
      "epoch": 0.35856470588235295,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0909,
      "step": 15239
    },
    {
      "epoch": 0.35858823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0136,
      "step": 15240
    },
    {
      "epoch": 0.35861176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9379,
      "step": 15241
    },
    {
      "epoch": 0.35863529411764705,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8285,
      "step": 15242
    },
    {
      "epoch": 0.35865882352941175,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7728,
      "step": 15243
    },
    {
      "epoch": 0.35868235294117645,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9625,
      "step": 15244
    },
    {
      "epoch": 0.35870588235294115,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8595,
      "step": 15245
    },
    {
      "epoch": 0.3587294117647059,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.4461,
      "step": 15246
    },
    {
      "epoch": 0.3587529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3028,
      "step": 15247
    },
    {
      "epoch": 0.3587764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9115,
      "step": 15248
    },
    {
      "epoch": 0.3588,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1502,
      "step": 15249
    },
    {
      "epoch": 0.3588235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0545,
      "step": 15250
    },
    {
      "epoch": 0.3588470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1245,
      "step": 15251
    },
    {
      "epoch": 0.3588705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9706,
      "step": 15252
    },
    {
      "epoch": 0.3588941176470588,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2477,
      "step": 15253
    },
    {
      "epoch": 0.3589176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0629,
      "step": 15254
    },
    {
      "epoch": 0.3589411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.6081,
      "step": 15255
    },
    {
      "epoch": 0.35896470588235296,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.0565,
      "step": 15256
    },
    {
      "epoch": 0.35898823529411766,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.745,
      "step": 15257
    },
    {
      "epoch": 0.35901176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9325,
      "step": 15258
    },
    {
      "epoch": 0.35903529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0944,
      "step": 15259
    },
    {
      "epoch": 0.35905882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7439,
      "step": 15260
    },
    {
      "epoch": 0.35908235294117646,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2118,
      "step": 15261
    },
    {
      "epoch": 0.35910588235294116,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9611,
      "step": 15262
    },
    {
      "epoch": 0.35912941176470586,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1369,
      "step": 15263
    },
    {
      "epoch": 0.35915294117647056,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0478,
      "step": 15264
    },
    {
      "epoch": 0.3591764705882353,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.8693,
      "step": 15265
    },
    {
      "epoch": 0.3592,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 15266
    },
    {
      "epoch": 0.3592235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0355,
      "step": 15267
    },
    {
      "epoch": 0.3592470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7329,
      "step": 15268
    },
    {
      "epoch": 0.3592705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1007,
      "step": 15269
    },
    {
      "epoch": 0.3592941176470588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1036,
      "step": 15270
    },
    {
      "epoch": 0.3593176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7851,
      "step": 15271
    },
    {
      "epoch": 0.3593411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8382,
      "step": 15272
    },
    {
      "epoch": 0.3593647058823529,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2705,
      "step": 15273
    },
    {
      "epoch": 0.3593882352941176,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7691,
      "step": 15274
    },
    {
      "epoch": 0.3594117647058824,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2046,
      "step": 15275
    },
    {
      "epoch": 0.3594352941176471,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0354,
      "step": 15276
    },
    {
      "epoch": 0.3594588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.2111,
      "step": 15277
    },
    {
      "epoch": 0.3594823529411765,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.193,
      "step": 15278
    },
    {
      "epoch": 0.3595058823529412,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0785,
      "step": 15279
    },
    {
      "epoch": 0.3595294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9847,
      "step": 15280
    },
    {
      "epoch": 0.3595529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0395,
      "step": 15281
    },
    {
      "epoch": 0.3595764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1757,
      "step": 15282
    },
    {
      "epoch": 0.3596,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.1393,
      "step": 15283
    },
    {
      "epoch": 0.35962352941176473,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9932,
      "step": 15284
    },
    {
      "epoch": 0.35964705882352943,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 15285
    },
    {
      "epoch": 0.35967058823529413,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3354,
      "step": 15286
    },
    {
      "epoch": 0.35969411764705883,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9851,
      "step": 15287
    },
    {
      "epoch": 0.35971764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 15288
    },
    {
      "epoch": 0.35974117647058823,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1574,
      "step": 15289
    },
    {
      "epoch": 0.35976470588235293,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2364,
      "step": 15290
    },
    {
      "epoch": 0.35978823529411763,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0184,
      "step": 15291
    },
    {
      "epoch": 0.35981176470588233,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.034,
      "step": 15292
    },
    {
      "epoch": 0.3598352941176471,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0714,
      "step": 15293
    },
    {
      "epoch": 0.3598588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0992,
      "step": 15294
    },
    {
      "epoch": 0.3598823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.845,
      "step": 15295
    },
    {
      "epoch": 0.3599058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1906,
      "step": 15296
    },
    {
      "epoch": 0.3599294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2253,
      "step": 15297
    },
    {
      "epoch": 0.3599529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 15298
    },
    {
      "epoch": 0.3599764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0674,
      "step": 15299
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0464,
      "step": 15300
    },
    {
      "epoch": 0.3600235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.218,
      "step": 15301
    },
    {
      "epoch": 0.3600470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 15302
    },
    {
      "epoch": 0.36007058823529414,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 15303
    },
    {
      "epoch": 0.36009411764705884,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.133,
      "step": 15304
    },
    {
      "epoch": 0.36011764705882354,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.955,
      "step": 15305
    },
    {
      "epoch": 0.36014117647058824,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0702,
      "step": 15306
    },
    {
      "epoch": 0.36016470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9105,
      "step": 15307
    },
    {
      "epoch": 0.36018823529411764,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0661,
      "step": 15308
    },
    {
      "epoch": 0.36021176470588234,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9992,
      "step": 15309
    },
    {
      "epoch": 0.36023529411764704,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8634,
      "step": 15310
    },
    {
      "epoch": 0.36025882352941174,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.5858,
      "step": 15311
    },
    {
      "epoch": 0.3602823529411765,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.265,
      "step": 15312
    },
    {
      "epoch": 0.3603058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9551,
      "step": 15313
    },
    {
      "epoch": 0.3603294117647059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.839,
      "step": 15314
    },
    {
      "epoch": 0.3603529411764706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2956,
      "step": 15315
    },
    {
      "epoch": 0.3603764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9399,
      "step": 15316
    },
    {
      "epoch": 0.3604,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9487,
      "step": 15317
    },
    {
      "epoch": 0.3604235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.041,
      "step": 15318
    },
    {
      "epoch": 0.3604470588235294,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8271,
      "step": 15319
    },
    {
      "epoch": 0.3604705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9581,
      "step": 15320
    },
    {
      "epoch": 0.3604941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0851,
      "step": 15321
    },
    {
      "epoch": 0.36051764705882355,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9214,
      "step": 15322
    },
    {
      "epoch": 0.36054117647058825,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8319,
      "step": 15323
    },
    {
      "epoch": 0.36056470588235295,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0673,
      "step": 15324
    },
    {
      "epoch": 0.36058823529411765,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9584,
      "step": 15325
    },
    {
      "epoch": 0.36061176470588235,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8908,
      "step": 15326
    },
    {
      "epoch": 0.36063529411764705,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0987,
      "step": 15327
    },
    {
      "epoch": 0.36065882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8045,
      "step": 15328
    },
    {
      "epoch": 0.36068235294117645,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 15329
    },
    {
      "epoch": 0.36070588235294115,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0397,
      "step": 15330
    },
    {
      "epoch": 0.3607294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 15331
    },
    {
      "epoch": 0.3607529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.2006,
      "step": 15332
    },
    {
      "epoch": 0.3607764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.4768,
      "step": 15333
    },
    {
      "epoch": 0.3608,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0042,
      "step": 15334
    },
    {
      "epoch": 0.3608235294117647,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 0.9399,
      "step": 15335
    },
    {
      "epoch": 0.3608470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9315,
      "step": 15336
    },
    {
      "epoch": 0.3608705882352941,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.074,
      "step": 15337
    },
    {
      "epoch": 0.3608941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9323,
      "step": 15338
    },
    {
      "epoch": 0.3609176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8235,
      "step": 15339
    },
    {
      "epoch": 0.3609411764705882,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.888,
      "step": 15340
    },
    {
      "epoch": 0.36096470588235297,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1717,
      "step": 15341
    },
    {
      "epoch": 0.36098823529411767,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.683,
      "step": 15342
    },
    {
      "epoch": 0.36101176470588237,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8902,
      "step": 15343
    },
    {
      "epoch": 0.36103529411764707,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3675,
      "step": 15344
    },
    {
      "epoch": 0.36105882352941177,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2081,
      "step": 15345
    },
    {
      "epoch": 0.36108235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8224,
      "step": 15346
    },
    {
      "epoch": 0.36110588235294117,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1111,
      "step": 15347
    },
    {
      "epoch": 0.36112941176470587,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8557,
      "step": 15348
    },
    {
      "epoch": 0.36115294117647057,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0608,
      "step": 15349
    },
    {
      "epoch": 0.3611764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.766,
      "step": 15350
    },
    {
      "epoch": 0.3612,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0158,
      "step": 15351
    },
    {
      "epoch": 0.3612235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8945,
      "step": 15352
    },
    {
      "epoch": 0.3612470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8526,
      "step": 15353
    },
    {
      "epoch": 0.3612705882352941,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8916,
      "step": 15354
    },
    {
      "epoch": 0.3612941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8329,
      "step": 15355
    },
    {
      "epoch": 0.3613176470588235,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3396,
      "step": 15356
    },
    {
      "epoch": 0.3613411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 15357
    },
    {
      "epoch": 0.3613647058823529,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8541,
      "step": 15358
    },
    {
      "epoch": 0.3613882352941176,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2584,
      "step": 15359
    },
    {
      "epoch": 0.3614117647058824,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0,
      "step": 15360
    },
    {
      "epoch": 0.3614352941176471,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8539,
      "step": 15361
    },
    {
      "epoch": 0.3614588235294118,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9298,
      "step": 15362
    },
    {
      "epoch": 0.3614823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7898,
      "step": 15363
    },
    {
      "epoch": 0.3615058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7113,
      "step": 15364
    },
    {
      "epoch": 0.3615294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1825,
      "step": 15365
    },
    {
      "epoch": 0.3615529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.969,
      "step": 15366
    },
    {
      "epoch": 0.3615764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7888,
      "step": 15367
    },
    {
      "epoch": 0.3616,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1363,
      "step": 15368
    },
    {
      "epoch": 0.36162352941176473,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0019,
      "step": 15369
    },
    {
      "epoch": 0.36164705882352943,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.196,
      "step": 15370
    },
    {
      "epoch": 0.36167058823529413,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9817,
      "step": 15371
    },
    {
      "epoch": 0.36169411764705883,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9842,
      "step": 15372
    },
    {
      "epoch": 0.36171764705882353,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.903,
      "step": 15373
    },
    {
      "epoch": 0.36174117647058823,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7158,
      "step": 15374
    },
    {
      "epoch": 0.36176470588235293,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.189,
      "step": 15375
    },
    {
      "epoch": 0.36178823529411763,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.5079,
      "step": 15376
    },
    {
      "epoch": 0.36181176470588233,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1402,
      "step": 15377
    },
    {
      "epoch": 0.36183529411764703,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0314,
      "step": 15378
    },
    {
      "epoch": 0.3618588235294118,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1769,
      "step": 15379
    },
    {
      "epoch": 0.3618823529411765,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 1.0792,
      "step": 15380
    },
    {
      "epoch": 0.3619058823529412,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2468,
      "step": 15381
    },
    {
      "epoch": 0.3619294117647059,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2957,
      "step": 15382
    },
    {
      "epoch": 0.3619529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0221,
      "step": 15383
    },
    {
      "epoch": 0.3619764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.985,
      "step": 15384
    },
    {
      "epoch": 0.362,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.1964,
      "step": 15385
    },
    {
      "epoch": 0.3620235294117647,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0064,
      "step": 15386
    },
    {
      "epoch": 0.3620470588235294,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2036,
      "step": 15387
    },
    {
      "epoch": 0.36207058823529414,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.7616,
      "step": 15388
    },
    {
      "epoch": 0.36209411764705884,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.062,
      "step": 15389
    },
    {
      "epoch": 0.36211764705882354,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.017,
      "step": 15390
    },
    {
      "epoch": 0.36214117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8352,
      "step": 15391
    },
    {
      "epoch": 0.36216470588235294,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1169,
      "step": 15392
    },
    {
      "epoch": 0.36218823529411764,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0106,
      "step": 15393
    },
    {
      "epoch": 0.36221176470588234,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1049,
      "step": 15394
    },
    {
      "epoch": 0.36223529411764704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0695,
      "step": 15395
    },
    {
      "epoch": 0.36225882352941174,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7356,
      "step": 15396
    },
    {
      "epoch": 0.36228235294117644,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.779,
      "step": 15397
    },
    {
      "epoch": 0.3623058823529412,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2742,
      "step": 15398
    },
    {
      "epoch": 0.3623294117647059,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0393,
      "step": 15399
    },
    {
      "epoch": 0.3623529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0839,
      "step": 15400
    },
    {
      "epoch": 0.3623764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.8034,
      "step": 15401
    },
    {
      "epoch": 0.3624,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8861,
      "step": 15402
    },
    {
      "epoch": 0.3624235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0212,
      "step": 15403
    },
    {
      "epoch": 0.3624470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9477,
      "step": 15404
    },
    {
      "epoch": 0.3624705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.8344,
      "step": 15405
    },
    {
      "epoch": 0.3624941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9133,
      "step": 15406
    },
    {
      "epoch": 0.36251764705882356,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9773,
      "step": 15407
    },
    {
      "epoch": 0.36254117647058826,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.2982,
      "step": 15408
    },
    {
      "epoch": 0.36256470588235296,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0297,
      "step": 15409
    },
    {
      "epoch": 0.36258823529411766,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8046,
      "step": 15410
    },
    {
      "epoch": 0.36261176470588236,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.6331,
      "step": 15411
    },
    {
      "epoch": 0.36263529411764706,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1666,
      "step": 15412
    },
    {
      "epoch": 0.36265882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.921,
      "step": 15413
    },
    {
      "epoch": 0.36268235294117646,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.8418,
      "step": 15414
    },
    {
      "epoch": 0.36270588235294116,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0636,
      "step": 15415
    },
    {
      "epoch": 0.36272941176470586,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2144,
      "step": 15416
    },
    {
      "epoch": 0.3627529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0721,
      "step": 15417
    },
    {
      "epoch": 0.3627764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9287,
      "step": 15418
    },
    {
      "epoch": 0.3628,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.02,
      "loss": 0.8338,
      "step": 15419
    },
    {
      "epoch": 0.3628235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.209,
      "step": 15420
    },
    {
      "epoch": 0.3628470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1947,
      "step": 15421
    },
    {
      "epoch": 0.3628705882352941,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1665,
      "step": 15422
    },
    {
      "epoch": 0.3628941176470588,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7256,
      "step": 15423
    },
    {
      "epoch": 0.3629176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 15424
    },
    {
      "epoch": 0.3629411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0819,
      "step": 15425
    },
    {
      "epoch": 0.36296470588235297,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.6985,
      "step": 15426
    },
    {
      "epoch": 0.36298823529411767,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1016,
      "step": 15427
    },
    {
      "epoch": 0.36301176470588237,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9172,
      "step": 15428
    },
    {
      "epoch": 0.36303529411764707,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9089,
      "step": 15429
    },
    {
      "epoch": 0.36305882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9407,
      "step": 15430
    },
    {
      "epoch": 0.36308235294117647,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8916,
      "step": 15431
    },
    {
      "epoch": 0.36310588235294117,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3394,
      "step": 15432
    },
    {
      "epoch": 0.36312941176470587,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2331,
      "step": 15433
    },
    {
      "epoch": 0.36315294117647057,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0331,
      "step": 15434
    },
    {
      "epoch": 0.36317647058823527,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9244,
      "step": 15435
    },
    {
      "epoch": 0.3632,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9851,
      "step": 15436
    },
    {
      "epoch": 0.3632235294117647,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.5708,
      "step": 15437
    },
    {
      "epoch": 0.3632470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0206,
      "step": 15438
    },
    {
      "epoch": 0.3632705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0408,
      "step": 15439
    },
    {
      "epoch": 0.3632941176470588,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2548,
      "step": 15440
    },
    {
      "epoch": 0.3633176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7888,
      "step": 15441
    },
    {
      "epoch": 0.3633411764705882,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0543,
      "step": 15442
    },
    {
      "epoch": 0.3633647058823529,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1476,
      "step": 15443
    },
    {
      "epoch": 0.3633882352941176,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.037,
      "step": 15444
    },
    {
      "epoch": 0.3634117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0809,
      "step": 15445
    },
    {
      "epoch": 0.3634352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2696,
      "step": 15446
    },
    {
      "epoch": 0.3634588235294118,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1891,
      "step": 15447
    },
    {
      "epoch": 0.3634823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9245,
      "step": 15448
    },
    {
      "epoch": 0.3635058823529412,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1745,
      "step": 15449
    },
    {
      "epoch": 0.3635294117647059,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 15450
    },
    {
      "epoch": 0.3635529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.6625,
      "step": 15451
    },
    {
      "epoch": 0.3635764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3225,
      "step": 15452
    },
    {
      "epoch": 0.3636,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0991,
      "step": 15453
    },
    {
      "epoch": 0.3636235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8987,
      "step": 15454
    },
    {
      "epoch": 0.36364705882352943,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7428,
      "step": 15455
    },
    {
      "epoch": 0.36367058823529413,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.92,
      "step": 15456
    },
    {
      "epoch": 0.36369411764705883,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0172,
      "step": 15457
    },
    {
      "epoch": 0.36371764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0187,
      "step": 15458
    },
    {
      "epoch": 0.36374117647058823,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8658,
      "step": 15459
    },
    {
      "epoch": 0.36376470588235293,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2233,
      "step": 15460
    },
    {
      "epoch": 0.36378823529411763,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1313,
      "step": 15461
    },
    {
      "epoch": 0.36381176470588233,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.22,
      "step": 15462
    },
    {
      "epoch": 0.36383529411764703,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9932,
      "step": 15463
    },
    {
      "epoch": 0.3638588235294118,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7501,
      "step": 15464
    },
    {
      "epoch": 0.3638823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0077,
      "step": 15465
    },
    {
      "epoch": 0.3639058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.03,
      "step": 15466
    },
    {
      "epoch": 0.3639294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 15467
    },
    {
      "epoch": 0.3639529411764706,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.9862,
      "step": 15468
    },
    {
      "epoch": 0.3639764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8464,
      "step": 15469
    },
    {
      "epoch": 0.364,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0301,
      "step": 15470
    },
    {
      "epoch": 0.3640235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7126,
      "step": 15471
    },
    {
      "epoch": 0.3640470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8758,
      "step": 15472
    },
    {
      "epoch": 0.3640705882352941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9391,
      "step": 15473
    },
    {
      "epoch": 0.36409411764705885,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1539,
      "step": 15474
    },
    {
      "epoch": 0.36411764705882355,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2864,
      "step": 15475
    },
    {
      "epoch": 0.36414117647058825,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9295,
      "step": 15476
    },
    {
      "epoch": 0.36416470588235295,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9548,
      "step": 15477
    },
    {
      "epoch": 0.36418823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.0319,
      "step": 15478
    },
    {
      "epoch": 0.36421176470588235,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2017,
      "step": 15479
    },
    {
      "epoch": 0.36423529411764705,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7777,
      "step": 15480
    },
    {
      "epoch": 0.36425882352941175,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7599,
      "step": 15481
    },
    {
      "epoch": 0.36428235294117645,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.931,
      "step": 15482
    },
    {
      "epoch": 0.3643058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9929,
      "step": 15483
    },
    {
      "epoch": 0.3643294117647059,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6395,
      "step": 15484
    },
    {
      "epoch": 0.3643529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7839,
      "step": 15485
    },
    {
      "epoch": 0.3643764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7237,
      "step": 15486
    },
    {
      "epoch": 0.3644,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0406,
      "step": 15487
    },
    {
      "epoch": 0.3644235294117647,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.1274,
      "step": 15488
    },
    {
      "epoch": 0.3644470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0032,
      "step": 15489
    },
    {
      "epoch": 0.3644705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.089,
      "step": 15490
    },
    {
      "epoch": 0.3644941176470588,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8091,
      "step": 15491
    },
    {
      "epoch": 0.3645176470588235,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0821,
      "step": 15492
    },
    {
      "epoch": 0.36454117647058826,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.6512,
      "step": 15493
    },
    {
      "epoch": 0.36456470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.6873,
      "step": 15494
    },
    {
      "epoch": 0.36458823529411766,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.3445,
      "step": 15495
    },
    {
      "epoch": 0.36461176470588236,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8374,
      "step": 15496
    },
    {
      "epoch": 0.36463529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.7784,
      "step": 15497
    },
    {
      "epoch": 0.36465882352941176,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2085,
      "step": 15498
    },
    {
      "epoch": 0.36468235294117646,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8824,
      "step": 15499
    },
    {
      "epoch": 0.36470588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8465,
      "step": 15500
    },
    {
      "epoch": 0.36472941176470586,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 15501
    },
    {
      "epoch": 0.3647529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0447,
      "step": 15502
    },
    {
      "epoch": 0.3647764705882353,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0979,
      "step": 15503
    },
    {
      "epoch": 0.3648,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0962,
      "step": 15504
    },
    {
      "epoch": 0.3648235294117647,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9789,
      "step": 15505
    },
    {
      "epoch": 0.3648470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.096,
      "step": 15506
    },
    {
      "epoch": 0.3648705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0291,
      "step": 15507
    },
    {
      "epoch": 0.3648941176470588,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0099,
      "step": 15508
    },
    {
      "epoch": 0.3649176470588235,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.6826,
      "step": 15509
    },
    {
      "epoch": 0.3649411764705882,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9541,
      "step": 15510
    },
    {
      "epoch": 0.36496470588235297,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 15511
    },
    {
      "epoch": 0.36498823529411767,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1176,
      "step": 15512
    },
    {
      "epoch": 0.36501176470588237,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8072,
      "step": 15513
    },
    {
      "epoch": 0.36503529411764707,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3791,
      "step": 15514
    },
    {
      "epoch": 0.36505882352941177,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0503,
      "step": 15515
    },
    {
      "epoch": 0.36508235294117647,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7254,
      "step": 15516
    },
    {
      "epoch": 0.36510588235294117,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8489,
      "step": 15517
    },
    {
      "epoch": 0.36512941176470587,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9642,
      "step": 15518
    },
    {
      "epoch": 0.36515294117647057,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7369,
      "step": 15519
    },
    {
      "epoch": 0.36517647058823527,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9659,
      "step": 15520
    },
    {
      "epoch": 0.3652,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 15521
    },
    {
      "epoch": 0.3652235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1504,
      "step": 15522
    },
    {
      "epoch": 0.3652470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.8783,
      "step": 15523
    },
    {
      "epoch": 0.3652705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 15524
    },
    {
      "epoch": 0.3652941176470588,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0543,
      "step": 15525
    },
    {
      "epoch": 0.3653176470588235,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.6547,
      "step": 15526
    },
    {
      "epoch": 0.3653411764705882,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9581,
      "step": 15527
    },
    {
      "epoch": 0.3653647058823529,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.6954,
      "step": 15528
    },
    {
      "epoch": 0.3653882352941176,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8003,
      "step": 15529
    },
    {
      "epoch": 0.3654117647058824,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1331,
      "step": 15530
    },
    {
      "epoch": 0.3654352941176471,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9974,
      "step": 15531
    },
    {
      "epoch": 0.3654588235294118,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2252,
      "step": 15532
    },
    {
      "epoch": 0.3654823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0943,
      "step": 15533
    },
    {
      "epoch": 0.3655058823529412,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8227,
      "step": 15534
    },
    {
      "epoch": 0.3655294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1346,
      "step": 15535
    },
    {
      "epoch": 0.3655529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9314,
      "step": 15536
    },
    {
      "epoch": 0.3655764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1805,
      "step": 15537
    },
    {
      "epoch": 0.3656,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0676,
      "step": 15538
    },
    {
      "epoch": 0.3656235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1276,
      "step": 15539
    },
    {
      "epoch": 0.36564705882352944,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2842,
      "step": 15540
    },
    {
      "epoch": 0.36567058823529414,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0764,
      "step": 15541
    },
    {
      "epoch": 0.36569411764705884,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 0.9942,
      "step": 15542
    },
    {
      "epoch": 0.36571764705882354,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8289,
      "step": 15543
    },
    {
      "epoch": 0.36574117647058824,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.02,
      "loss": 1.2589,
      "step": 15544
    },
    {
      "epoch": 0.36576470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7955,
      "step": 15545
    },
    {
      "epoch": 0.36578823529411764,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.8774,
      "step": 15546
    },
    {
      "epoch": 0.36581176470588234,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0959,
      "step": 15547
    },
    {
      "epoch": 0.36583529411764704,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8891,
      "step": 15548
    },
    {
      "epoch": 0.3658588235294118,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1704,
      "step": 15549
    },
    {
      "epoch": 0.3658823529411765,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7385,
      "step": 15550
    },
    {
      "epoch": 0.3659058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1599,
      "step": 15551
    },
    {
      "epoch": 0.3659294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0727,
      "step": 15552
    },
    {
      "epoch": 0.3659529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9604,
      "step": 15553
    },
    {
      "epoch": 0.3659764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.6481,
      "step": 15554
    },
    {
      "epoch": 0.366,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.318,
      "step": 15555
    },
    {
      "epoch": 0.3660235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.926,
      "step": 15556
    },
    {
      "epoch": 0.3660470588235294,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.0935,
      "step": 15557
    },
    {
      "epoch": 0.3660705882352941,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1046,
      "step": 15558
    },
    {
      "epoch": 0.36609411764705885,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7905,
      "step": 15559
    },
    {
      "epoch": 0.36611764705882355,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8231,
      "step": 15560
    },
    {
      "epoch": 0.36614117647058825,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1378,
      "step": 15561
    },
    {
      "epoch": 0.36616470588235295,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8722,
      "step": 15562
    },
    {
      "epoch": 0.36618823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9918,
      "step": 15563
    },
    {
      "epoch": 0.36621176470588235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0095,
      "step": 15564
    },
    {
      "epoch": 0.36623529411764705,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3428,
      "step": 15565
    },
    {
      "epoch": 0.36625882352941175,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7878,
      "step": 15566
    },
    {
      "epoch": 0.36628235294117645,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7861,
      "step": 15567
    },
    {
      "epoch": 0.3663058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9705,
      "step": 15568
    },
    {
      "epoch": 0.3663294117647059,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0913,
      "step": 15569
    },
    {
      "epoch": 0.3663529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9071,
      "step": 15570
    },
    {
      "epoch": 0.3663764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9241,
      "step": 15571
    },
    {
      "epoch": 0.3664,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9987,
      "step": 15572
    },
    {
      "epoch": 0.3664235294117647,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1167,
      "step": 15573
    },
    {
      "epoch": 0.3664470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0428,
      "step": 15574
    },
    {
      "epoch": 0.3664705882352941,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3343,
      "step": 15575
    },
    {
      "epoch": 0.3664941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0092,
      "step": 15576
    },
    {
      "epoch": 0.3665176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8218,
      "step": 15577
    },
    {
      "epoch": 0.36654117647058826,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0324,
      "step": 15578
    },
    {
      "epoch": 0.36656470588235296,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3163,
      "step": 15579
    },
    {
      "epoch": 0.36658823529411766,
      "grad_norm": 0.63671875,
      "learning_rate": 0.02,
      "loss": 0.9853,
      "step": 15580
    },
    {
      "epoch": 0.36661176470588236,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9894,
      "step": 15581
    },
    {
      "epoch": 0.36663529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.208,
      "step": 15582
    },
    {
      "epoch": 0.36665882352941176,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2576,
      "step": 15583
    },
    {
      "epoch": 0.36668235294117646,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2177,
      "step": 15584
    },
    {
      "epoch": 0.36670588235294116,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1382,
      "step": 15585
    },
    {
      "epoch": 0.36672941176470586,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9849,
      "step": 15586
    },
    {
      "epoch": 0.3667529411764706,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8897,
      "step": 15587
    },
    {
      "epoch": 0.3667764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9392,
      "step": 15588
    },
    {
      "epoch": 0.3668,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.02,
      "loss": 1.355,
      "step": 15589
    },
    {
      "epoch": 0.3668235294117647,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8756,
      "step": 15590
    },
    {
      "epoch": 0.3668470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7062,
      "step": 15591
    },
    {
      "epoch": 0.3668705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7886,
      "step": 15592
    },
    {
      "epoch": 0.3668941176470588,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 0.994,
      "step": 15593
    },
    {
      "epoch": 0.3669176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9768,
      "step": 15594
    },
    {
      "epoch": 0.3669411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.655,
      "step": 15595
    },
    {
      "epoch": 0.3669647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8589,
      "step": 15596
    },
    {
      "epoch": 0.36698823529411767,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8763,
      "step": 15597
    },
    {
      "epoch": 0.36701176470588237,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9967,
      "step": 15598
    },
    {
      "epoch": 0.36703529411764707,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2423,
      "step": 15599
    },
    {
      "epoch": 0.36705882352941177,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9517,
      "step": 15600
    },
    {
      "epoch": 0.36708235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0618,
      "step": 15601
    },
    {
      "epoch": 0.36710588235294117,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1217,
      "step": 15602
    },
    {
      "epoch": 0.36712941176470587,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2647,
      "step": 15603
    },
    {
      "epoch": 0.36715294117647057,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0655,
      "step": 15604
    },
    {
      "epoch": 0.36717647058823527,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0449,
      "step": 15605
    },
    {
      "epoch": 0.3672,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.976,
      "step": 15606
    },
    {
      "epoch": 0.3672235294117647,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.3854,
      "step": 15607
    },
    {
      "epoch": 0.3672470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 15608
    },
    {
      "epoch": 0.3672705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9551,
      "step": 15609
    },
    {
      "epoch": 0.3672941176470588,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2539,
      "step": 15610
    },
    {
      "epoch": 0.3673176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.93,
      "step": 15611
    },
    {
      "epoch": 0.3673411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.039,
      "step": 15612
    },
    {
      "epoch": 0.3673647058823529,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.3163,
      "step": 15613
    },
    {
      "epoch": 0.3673882352941176,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3202,
      "step": 15614
    },
    {
      "epoch": 0.3674117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8102,
      "step": 15615
    },
    {
      "epoch": 0.3674352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9429,
      "step": 15616
    },
    {
      "epoch": 0.3674588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1683,
      "step": 15617
    },
    {
      "epoch": 0.3674823529411765,
      "grad_norm": 0.234375,
      "learning_rate": 0.02,
      "loss": 0.9455,
      "step": 15618
    },
    {
      "epoch": 0.3675058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.992,
      "step": 15619
    },
    {
      "epoch": 0.3675294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8521,
      "step": 15620
    },
    {
      "epoch": 0.3675529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9118,
      "step": 15621
    },
    {
      "epoch": 0.3675764705882353,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.656,
      "step": 15622
    },
    {
      "epoch": 0.3676,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8074,
      "step": 15623
    },
    {
      "epoch": 0.3676235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0006,
      "step": 15624
    },
    {
      "epoch": 0.36764705882352944,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1766,
      "step": 15625
    },
    {
      "epoch": 0.36767058823529414,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9021,
      "step": 15626
    },
    {
      "epoch": 0.36769411764705884,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9164,
      "step": 15627
    },
    {
      "epoch": 0.36771764705882354,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.768,
      "step": 15628
    },
    {
      "epoch": 0.36774117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1303,
      "step": 15629
    },
    {
      "epoch": 0.36776470588235294,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.7334,
      "step": 15630
    },
    {
      "epoch": 0.36778823529411764,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.4687,
      "step": 15631
    },
    {
      "epoch": 0.36781176470588234,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0543,
      "step": 15632
    },
    {
      "epoch": 0.36783529411764704,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0317,
      "step": 15633
    },
    {
      "epoch": 0.36785882352941174,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0894,
      "step": 15634
    },
    {
      "epoch": 0.3678823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7301,
      "step": 15635
    },
    {
      "epoch": 0.3679058823529412,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8292,
      "step": 15636
    },
    {
      "epoch": 0.3679294117647059,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0458,
      "step": 15637
    },
    {
      "epoch": 0.3679529411764706,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1419,
      "step": 15638
    },
    {
      "epoch": 0.3679764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.991,
      "step": 15639
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8617,
      "step": 15640
    },
    {
      "epoch": 0.3680235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 15641
    },
    {
      "epoch": 0.3680470588235294,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.119,
      "step": 15642
    },
    {
      "epoch": 0.3680705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7774,
      "step": 15643
    },
    {
      "epoch": 0.36809411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8279,
      "step": 15644
    },
    {
      "epoch": 0.36811764705882355,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1301,
      "step": 15645
    },
    {
      "epoch": 0.36814117647058825,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1516,
      "step": 15646
    },
    {
      "epoch": 0.36816470588235295,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.024,
      "step": 15647
    },
    {
      "epoch": 0.36818823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2271,
      "step": 15648
    },
    {
      "epoch": 0.36821176470588235,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.6974,
      "step": 15649
    },
    {
      "epoch": 0.36823529411764705,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8948,
      "step": 15650
    },
    {
      "epoch": 0.36825882352941175,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2158,
      "step": 15651
    },
    {
      "epoch": 0.36828235294117645,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7853,
      "step": 15652
    },
    {
      "epoch": 0.36830588235294115,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1273,
      "step": 15653
    },
    {
      "epoch": 0.3683294117647059,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1206,
      "step": 15654
    },
    {
      "epoch": 0.3683529411764706,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1234,
      "step": 15655
    },
    {
      "epoch": 0.3683764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9254,
      "step": 15656
    },
    {
      "epoch": 0.3684,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.6512,
      "step": 15657
    },
    {
      "epoch": 0.3684235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0061,
      "step": 15658
    },
    {
      "epoch": 0.3684470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2276,
      "step": 15659
    },
    {
      "epoch": 0.3684705882352941,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.2027,
      "step": 15660
    },
    {
      "epoch": 0.3684941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.8463,
      "step": 15661
    },
    {
      "epoch": 0.3685176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0599,
      "step": 15662
    },
    {
      "epoch": 0.36854117647058826,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0664,
      "step": 15663
    },
    {
      "epoch": 0.36856470588235296,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7127,
      "step": 15664
    },
    {
      "epoch": 0.36858823529411766,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1453,
      "step": 15665
    },
    {
      "epoch": 0.36861176470588236,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.039,
      "step": 15666
    },
    {
      "epoch": 0.36863529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0083,
      "step": 15667
    },
    {
      "epoch": 0.36865882352941176,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 15668
    },
    {
      "epoch": 0.36868235294117646,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7498,
      "step": 15669
    },
    {
      "epoch": 0.36870588235294116,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.4807,
      "step": 15670
    },
    {
      "epoch": 0.36872941176470586,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2082,
      "step": 15671
    },
    {
      "epoch": 0.36875294117647056,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8948,
      "step": 15672
    },
    {
      "epoch": 0.3687764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0013,
      "step": 15673
    },
    {
      "epoch": 0.3688,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.6627,
      "step": 15674
    },
    {
      "epoch": 0.3688235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9437,
      "step": 15675
    },
    {
      "epoch": 0.3688470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1322,
      "step": 15676
    },
    {
      "epoch": 0.3688705882352941,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.6614,
      "step": 15677
    },
    {
      "epoch": 0.3688941176470588,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8111,
      "step": 15678
    },
    {
      "epoch": 0.3689176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.942,
      "step": 15679
    },
    {
      "epoch": 0.3689411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0729,
      "step": 15680
    },
    {
      "epoch": 0.3689647058823529,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.932,
      "step": 15681
    },
    {
      "epoch": 0.3689882352941177,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.951,
      "step": 15682
    },
    {
      "epoch": 0.3690117647058824,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.803,
      "step": 15683
    },
    {
      "epoch": 0.3690352941176471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0652,
      "step": 15684
    },
    {
      "epoch": 0.3690588235294118,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7209,
      "step": 15685
    },
    {
      "epoch": 0.3690823529411765,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0455,
      "step": 15686
    },
    {
      "epoch": 0.3691058823529412,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.819,
      "step": 15687
    },
    {
      "epoch": 0.3691294117647059,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.7075,
      "step": 15688
    },
    {
      "epoch": 0.3691529411764706,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.7096,
      "step": 15689
    },
    {
      "epoch": 0.3691764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7889,
      "step": 15690
    },
    {
      "epoch": 0.3692,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7005,
      "step": 15691
    },
    {
      "epoch": 0.36922352941176473,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.296,
      "step": 15692
    },
    {
      "epoch": 0.36924705882352943,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0336,
      "step": 15693
    },
    {
      "epoch": 0.36927058823529413,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.247,
      "step": 15694
    },
    {
      "epoch": 0.36929411764705883,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.02,
      "loss": 0.7248,
      "step": 15695
    },
    {
      "epoch": 0.36931764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0682,
      "step": 15696
    },
    {
      "epoch": 0.36934117647058823,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3207,
      "step": 15697
    },
    {
      "epoch": 0.36936470588235293,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9926,
      "step": 15698
    },
    {
      "epoch": 0.36938823529411763,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9906,
      "step": 15699
    },
    {
      "epoch": 0.36941176470588233,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7504,
      "step": 15700
    },
    {
      "epoch": 0.3694352941176471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.309,
      "step": 15701
    },
    {
      "epoch": 0.3694588235294118,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.1523,
      "step": 15702
    },
    {
      "epoch": 0.3694823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9454,
      "step": 15703
    },
    {
      "epoch": 0.3695058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0966,
      "step": 15704
    },
    {
      "epoch": 0.3695294117647059,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 15705
    },
    {
      "epoch": 0.3695529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0767,
      "step": 15706
    },
    {
      "epoch": 0.3695764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0072,
      "step": 15707
    },
    {
      "epoch": 0.3696,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9186,
      "step": 15708
    },
    {
      "epoch": 0.3696235294117647,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.7586,
      "step": 15709
    },
    {
      "epoch": 0.3696470588235294,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1875,
      "step": 15710
    },
    {
      "epoch": 0.36967058823529414,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 15711
    },
    {
      "epoch": 0.36969411764705884,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1762,
      "step": 15712
    },
    {
      "epoch": 0.36971764705882354,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2439,
      "step": 15713
    },
    {
      "epoch": 0.36974117647058824,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7367,
      "step": 15714
    },
    {
      "epoch": 0.36976470588235294,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.9504,
      "step": 15715
    },
    {
      "epoch": 0.36978823529411764,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3118,
      "step": 15716
    },
    {
      "epoch": 0.36981176470588234,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8864,
      "step": 15717
    },
    {
      "epoch": 0.36983529411764704,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8159,
      "step": 15718
    },
    {
      "epoch": 0.36985882352941174,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.7932,
      "step": 15719
    },
    {
      "epoch": 0.3698823529411765,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.133,
      "step": 15720
    },
    {
      "epoch": 0.3699058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0984,
      "step": 15721
    },
    {
      "epoch": 0.3699294117647059,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.055,
      "step": 15722
    },
    {
      "epoch": 0.3699529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0179,
      "step": 15723
    },
    {
      "epoch": 0.3699764705882353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 0.901,
      "step": 15724
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2249,
      "step": 15725
    },
    {
      "epoch": 0.3700235294117647,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.205,
      "step": 15726
    },
    {
      "epoch": 0.3700470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1842,
      "step": 15727
    },
    {
      "epoch": 0.3700705882352941,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.4097,
      "step": 15728
    },
    {
      "epoch": 0.37009411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8987,
      "step": 15729
    },
    {
      "epoch": 0.37011764705882355,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9658,
      "step": 15730
    },
    {
      "epoch": 0.37014117647058825,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.177,
      "step": 15731
    },
    {
      "epoch": 0.37016470588235295,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0564,
      "step": 15732
    },
    {
      "epoch": 0.37018823529411765,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.0141,
      "step": 15733
    },
    {
      "epoch": 0.37021176470588235,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9693,
      "step": 15734
    },
    {
      "epoch": 0.37023529411764705,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.7662,
      "step": 15735
    },
    {
      "epoch": 0.37025882352941175,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9976,
      "step": 15736
    },
    {
      "epoch": 0.37028235294117645,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9728,
      "step": 15737
    },
    {
      "epoch": 0.37030588235294115,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8133,
      "step": 15738
    },
    {
      "epoch": 0.3703294117647059,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0596,
      "step": 15739
    },
    {
      "epoch": 0.3703529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0984,
      "step": 15740
    },
    {
      "epoch": 0.3703764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.989,
      "step": 15741
    },
    {
      "epoch": 0.3704,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7965,
      "step": 15742
    },
    {
      "epoch": 0.3704235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8985,
      "step": 15743
    },
    {
      "epoch": 0.3704470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0313,
      "step": 15744
    },
    {
      "epoch": 0.3704705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0555,
      "step": 15745
    },
    {
      "epoch": 0.3704941176470588,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1604,
      "step": 15746
    },
    {
      "epoch": 0.3705176470588235,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8895,
      "step": 15747
    },
    {
      "epoch": 0.37054117647058826,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.972,
      "step": 15748
    },
    {
      "epoch": 0.37056470588235296,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.651,
      "step": 15749
    },
    {
      "epoch": 0.37058823529411766,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9212,
      "step": 15750
    },
    {
      "epoch": 0.37061176470588236,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0647,
      "step": 15751
    },
    {
      "epoch": 0.37063529411764706,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8883,
      "step": 15752
    },
    {
      "epoch": 0.37065882352941176,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8883,
      "step": 15753
    },
    {
      "epoch": 0.37068235294117646,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0329,
      "step": 15754
    },
    {
      "epoch": 0.37070588235294116,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8457,
      "step": 15755
    },
    {
      "epoch": 0.37072941176470586,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9125,
      "step": 15756
    },
    {
      "epoch": 0.37075294117647056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0061,
      "step": 15757
    },
    {
      "epoch": 0.3707764705882353,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.8103,
      "step": 15758
    },
    {
      "epoch": 0.3708,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1674,
      "step": 15759
    },
    {
      "epoch": 0.3708235294117647,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2874,
      "step": 15760
    },
    {
      "epoch": 0.3708470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9331,
      "step": 15761
    },
    {
      "epoch": 0.3708705882352941,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.6659,
      "step": 15762
    },
    {
      "epoch": 0.3708941176470588,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.978,
      "step": 15763
    },
    {
      "epoch": 0.3709176470588235,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0992,
      "step": 15764
    },
    {
      "epoch": 0.3709411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0927,
      "step": 15765
    },
    {
      "epoch": 0.3709647058823529,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9188,
      "step": 15766
    },
    {
      "epoch": 0.3709882352941177,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0389,
      "step": 15767
    },
    {
      "epoch": 0.3710117647058824,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9282,
      "step": 15768
    },
    {
      "epoch": 0.3710352941176471,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 15769
    },
    {
      "epoch": 0.3710588235294118,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7023,
      "step": 15770
    },
    {
      "epoch": 0.3710823529411765,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6841,
      "step": 15771
    },
    {
      "epoch": 0.3711058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8803,
      "step": 15772
    },
    {
      "epoch": 0.3711294117647059,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1725,
      "step": 15773
    },
    {
      "epoch": 0.3711529411764706,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.115,
      "step": 15774
    },
    {
      "epoch": 0.3711764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2302,
      "step": 15775
    },
    {
      "epoch": 0.3712,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 15776
    },
    {
      "epoch": 0.37122352941176473,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8072,
      "step": 15777
    },
    {
      "epoch": 0.37124705882352943,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8999,
      "step": 15778
    },
    {
      "epoch": 0.37127058823529413,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7154,
      "step": 15779
    },
    {
      "epoch": 0.37129411764705883,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8722,
      "step": 15780
    },
    {
      "epoch": 0.37131764705882353,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.3541,
      "step": 15781
    },
    {
      "epoch": 0.37134117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.7752,
      "step": 15782
    },
    {
      "epoch": 0.37136470588235293,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2468,
      "step": 15783
    },
    {
      "epoch": 0.37138823529411763,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.163,
      "step": 15784
    },
    {
      "epoch": 0.37141176470588233,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9101,
      "step": 15785
    },
    {
      "epoch": 0.3714352941176471,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8621,
      "step": 15786
    },
    {
      "epoch": 0.3714588235294118,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8286,
      "step": 15787
    },
    {
      "epoch": 0.3714823529411765,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.6638,
      "step": 15788
    },
    {
      "epoch": 0.3715058823529412,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1356,
      "step": 15789
    },
    {
      "epoch": 0.3715294117647059,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 15790
    },
    {
      "epoch": 0.3715529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0789,
      "step": 15791
    },
    {
      "epoch": 0.3715764705882353,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3192,
      "step": 15792
    },
    {
      "epoch": 0.3716,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1536,
      "step": 15793
    },
    {
      "epoch": 0.3716235294117647,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2517,
      "step": 15794
    },
    {
      "epoch": 0.3716470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0427,
      "step": 15795
    },
    {
      "epoch": 0.37167058823529414,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.232,
      "step": 15796
    },
    {
      "epoch": 0.37169411764705884,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0454,
      "step": 15797
    },
    {
      "epoch": 0.37171764705882354,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.368,
      "step": 15798
    },
    {
      "epoch": 0.37174117647058824,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9938,
      "step": 15799
    },
    {
      "epoch": 0.37176470588235294,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1573,
      "step": 15800
    },
    {
      "epoch": 0.37178823529411764,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.1413,
      "step": 15801
    },
    {
      "epoch": 0.37181176470588234,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0646,
      "step": 15802
    },
    {
      "epoch": 0.37183529411764704,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0524,
      "step": 15803
    },
    {
      "epoch": 0.37185882352941174,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0713,
      "step": 15804
    },
    {
      "epoch": 0.3718823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1358,
      "step": 15805
    },
    {
      "epoch": 0.3719058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.1013,
      "step": 15806
    },
    {
      "epoch": 0.3719294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0185,
      "step": 15807
    },
    {
      "epoch": 0.3719529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8478,
      "step": 15808
    },
    {
      "epoch": 0.3719764705882353,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.2285,
      "step": 15809
    },
    {
      "epoch": 0.372,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9153,
      "step": 15810
    },
    {
      "epoch": 0.3720235294117647,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3481,
      "step": 15811
    },
    {
      "epoch": 0.3720470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8719,
      "step": 15812
    },
    {
      "epoch": 0.3720705882352941,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2639,
      "step": 15813
    },
    {
      "epoch": 0.3720941176470588,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7691,
      "step": 15814
    },
    {
      "epoch": 0.37211764705882355,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8695,
      "step": 15815
    },
    {
      "epoch": 0.37214117647058825,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.971,
      "step": 15816
    },
    {
      "epoch": 0.37216470588235295,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1408,
      "step": 15817
    },
    {
      "epoch": 0.37218823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9577,
      "step": 15818
    },
    {
      "epoch": 0.37221176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1631,
      "step": 15819
    },
    {
      "epoch": 0.37223529411764705,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8809,
      "step": 15820
    },
    {
      "epoch": 0.37225882352941175,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.575,
      "step": 15821
    },
    {
      "epoch": 0.37228235294117645,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 15822
    },
    {
      "epoch": 0.37230588235294115,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9249,
      "step": 15823
    },
    {
      "epoch": 0.3723294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9787,
      "step": 15824
    },
    {
      "epoch": 0.3723529411764706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2344,
      "step": 15825
    },
    {
      "epoch": 0.3723764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.214,
      "step": 15826
    },
    {
      "epoch": 0.3724,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8639,
      "step": 15827
    },
    {
      "epoch": 0.3724235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7948,
      "step": 15828
    },
    {
      "epoch": 0.3724470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0114,
      "step": 15829
    },
    {
      "epoch": 0.3724705882352941,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.3721,
      "step": 15830
    },
    {
      "epoch": 0.3724941176470588,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.015,
      "step": 15831
    },
    {
      "epoch": 0.3725176470588235,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.841,
      "step": 15832
    },
    {
      "epoch": 0.3725411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.7359,
      "step": 15833
    },
    {
      "epoch": 0.37256470588235296,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7634,
      "step": 15834
    },
    {
      "epoch": 0.37258823529411766,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7483,
      "step": 15835
    },
    {
      "epoch": 0.37261176470588236,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2974,
      "step": 15836
    },
    {
      "epoch": 0.37263529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.4037,
      "step": 15837
    },
    {
      "epoch": 0.37265882352941176,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0407,
      "step": 15838
    },
    {
      "epoch": 0.37268235294117646,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1853,
      "step": 15839
    },
    {
      "epoch": 0.37270588235294116,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2016,
      "step": 15840
    },
    {
      "epoch": 0.37272941176470586,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.0296,
      "step": 15841
    },
    {
      "epoch": 0.37275294117647056,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9051,
      "step": 15842
    },
    {
      "epoch": 0.3727764705882353,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7652,
      "step": 15843
    },
    {
      "epoch": 0.3728,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 1.1345,
      "step": 15844
    },
    {
      "epoch": 0.3728235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.2866,
      "step": 15845
    },
    {
      "epoch": 0.3728470588235294,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.5107,
      "step": 15846
    },
    {
      "epoch": 0.3728705882352941,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1791,
      "step": 15847
    },
    {
      "epoch": 0.3728941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9008,
      "step": 15848
    },
    {
      "epoch": 0.3729176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9269,
      "step": 15849
    },
    {
      "epoch": 0.3729411764705882,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1251,
      "step": 15850
    },
    {
      "epoch": 0.3729647058823529,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.6432,
      "step": 15851
    },
    {
      "epoch": 0.3729882352941176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1359,
      "step": 15852
    },
    {
      "epoch": 0.3730117647058824,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.4529,
      "step": 15853
    },
    {
      "epoch": 0.3730352941176471,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7722,
      "step": 15854
    },
    {
      "epoch": 0.3730588235294118,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8821,
      "step": 15855
    },
    {
      "epoch": 0.3730823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2219,
      "step": 15856
    },
    {
      "epoch": 0.3731058823529412,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9996,
      "step": 15857
    },
    {
      "epoch": 0.3731294117647059,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8291,
      "step": 15858
    },
    {
      "epoch": 0.3731529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9577,
      "step": 15859
    },
    {
      "epoch": 0.3731764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8244,
      "step": 15860
    },
    {
      "epoch": 0.3732,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2096,
      "step": 15861
    },
    {
      "epoch": 0.37322352941176473,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2566,
      "step": 15862
    },
    {
      "epoch": 0.37324705882352943,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9381,
      "step": 15863
    },
    {
      "epoch": 0.37327058823529413,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1208,
      "step": 15864
    },
    {
      "epoch": 0.37329411764705883,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7325,
      "step": 15865
    },
    {
      "epoch": 0.37331764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9169,
      "step": 15866
    },
    {
      "epoch": 0.37334117647058823,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0093,
      "step": 15867
    },
    {
      "epoch": 0.37336470588235293,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9062,
      "step": 15868
    },
    {
      "epoch": 0.37338823529411763,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9097,
      "step": 15869
    },
    {
      "epoch": 0.37341176470588233,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2216,
      "step": 15870
    },
    {
      "epoch": 0.37343529411764703,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1811,
      "step": 15871
    },
    {
      "epoch": 0.3734588235294118,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2201,
      "step": 15872
    },
    {
      "epoch": 0.3734823529411765,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9584,
      "step": 15873
    },
    {
      "epoch": 0.3735058823529412,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.877,
      "step": 15874
    },
    {
      "epoch": 0.3735294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1718,
      "step": 15875
    },
    {
      "epoch": 0.3735529411764706,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9752,
      "step": 15876
    },
    {
      "epoch": 0.3735764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 15877
    },
    {
      "epoch": 0.3736,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2677,
      "step": 15878
    },
    {
      "epoch": 0.3736235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.3123,
      "step": 15879
    },
    {
      "epoch": 0.3736470588235294,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1797,
      "step": 15880
    },
    {
      "epoch": 0.37367058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9104,
      "step": 15881
    },
    {
      "epoch": 0.37369411764705884,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.713,
      "step": 15882
    },
    {
      "epoch": 0.37371764705882354,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0381,
      "step": 15883
    },
    {
      "epoch": 0.37374117647058824,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8929,
      "step": 15884
    },
    {
      "epoch": 0.37376470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7086,
      "step": 15885
    },
    {
      "epoch": 0.37378823529411764,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1483,
      "step": 15886
    },
    {
      "epoch": 0.37381176470588234,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9238,
      "step": 15887
    },
    {
      "epoch": 0.37383529411764704,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1056,
      "step": 15888
    },
    {
      "epoch": 0.37385882352941174,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1608,
      "step": 15889
    },
    {
      "epoch": 0.37388235294117644,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8884,
      "step": 15890
    },
    {
      "epoch": 0.3739058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.8527,
      "step": 15891
    },
    {
      "epoch": 0.3739294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 1.2635,
      "step": 15892
    },
    {
      "epoch": 0.3739529411764706,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.9908,
      "step": 15893
    },
    {
      "epoch": 0.3739764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0102,
      "step": 15894
    },
    {
      "epoch": 0.374,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.5101,
      "step": 15895
    },
    {
      "epoch": 0.3740235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.2542,
      "step": 15896
    },
    {
      "epoch": 0.3740470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.6424,
      "step": 15897
    },
    {
      "epoch": 0.3740705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7216,
      "step": 15898
    },
    {
      "epoch": 0.3740941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9613,
      "step": 15899
    },
    {
      "epoch": 0.37411764705882355,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.6883,
      "step": 15900
    },
    {
      "epoch": 0.37414117647058825,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2474,
      "step": 15901
    },
    {
      "epoch": 0.37416470588235295,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0768,
      "step": 15902
    },
    {
      "epoch": 0.37418823529411765,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.082,
      "step": 15903
    },
    {
      "epoch": 0.37421176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1066,
      "step": 15904
    },
    {
      "epoch": 0.37423529411764705,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.9214,
      "step": 15905
    },
    {
      "epoch": 0.37425882352941175,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6546,
      "step": 15906
    },
    {
      "epoch": 0.37428235294117645,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2799,
      "step": 15907
    },
    {
      "epoch": 0.37430588235294115,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.02,
      "loss": 0.4886,
      "step": 15908
    },
    {
      "epoch": 0.37432941176470585,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8672,
      "step": 15909
    },
    {
      "epoch": 0.3743529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9782,
      "step": 15910
    },
    {
      "epoch": 0.3743764705882353,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.809,
      "step": 15911
    },
    {
      "epoch": 0.3744,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2159,
      "step": 15912
    },
    {
      "epoch": 0.3744235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.8906,
      "step": 15913
    },
    {
      "epoch": 0.3744470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 15914
    },
    {
      "epoch": 0.3744705882352941,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.045,
      "step": 15915
    },
    {
      "epoch": 0.3744941176470588,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3596,
      "step": 15916
    },
    {
      "epoch": 0.3745176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2401,
      "step": 15917
    },
    {
      "epoch": 0.3745411764705882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0497,
      "step": 15918
    },
    {
      "epoch": 0.37456470588235297,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9461,
      "step": 15919
    },
    {
      "epoch": 0.37458823529411767,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7111,
      "step": 15920
    },
    {
      "epoch": 0.37461176470588237,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9929,
      "step": 15921
    },
    {
      "epoch": 0.37463529411764707,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.34,
      "step": 15922
    },
    {
      "epoch": 0.37465882352941177,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3372,
      "step": 15923
    },
    {
      "epoch": 0.37468235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8565,
      "step": 15924
    },
    {
      "epoch": 0.37470588235294117,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2598,
      "step": 15925
    },
    {
      "epoch": 0.37472941176470587,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3032,
      "step": 15926
    },
    {
      "epoch": 0.37475294117647057,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.927,
      "step": 15927
    },
    {
      "epoch": 0.37477647058823527,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0569,
      "step": 15928
    },
    {
      "epoch": 0.3748,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0016,
      "step": 15929
    },
    {
      "epoch": 0.3748235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9849,
      "step": 15930
    },
    {
      "epoch": 0.3748470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.7602,
      "step": 15931
    },
    {
      "epoch": 0.3748705882352941,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.2079,
      "step": 15932
    },
    {
      "epoch": 0.3748941176470588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7263,
      "step": 15933
    },
    {
      "epoch": 0.3749176470588235,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8252,
      "step": 15934
    },
    {
      "epoch": 0.3749411764705882,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9192,
      "step": 15935
    },
    {
      "epoch": 0.3749647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0548,
      "step": 15936
    },
    {
      "epoch": 0.3749882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1165,
      "step": 15937
    },
    {
      "epoch": 0.3750117647058824,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9929,
      "step": 15938
    },
    {
      "epoch": 0.3750352941176471,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9724,
      "step": 15939
    },
    {
      "epoch": 0.3750588235294118,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1931,
      "step": 15940
    },
    {
      "epoch": 0.3750823529411765,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9947,
      "step": 15941
    },
    {
      "epoch": 0.3751058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8759,
      "step": 15942
    },
    {
      "epoch": 0.3751294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8807,
      "step": 15943
    },
    {
      "epoch": 0.3751529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1647,
      "step": 15944
    },
    {
      "epoch": 0.3751764705882353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2389,
      "step": 15945
    },
    {
      "epoch": 0.3752,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0818,
      "step": 15946
    },
    {
      "epoch": 0.37522352941176473,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8691,
      "step": 15947
    },
    {
      "epoch": 0.37524705882352943,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0219,
      "step": 15948
    },
    {
      "epoch": 0.37527058823529413,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9208,
      "step": 15949
    },
    {
      "epoch": 0.37529411764705883,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1534,
      "step": 15950
    },
    {
      "epoch": 0.37531764705882353,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9318,
      "step": 15951
    },
    {
      "epoch": 0.37534117647058823,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2561,
      "step": 15952
    },
    {
      "epoch": 0.37536470588235293,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0711,
      "step": 15953
    },
    {
      "epoch": 0.37538823529411763,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7229,
      "step": 15954
    },
    {
      "epoch": 0.37541176470588233,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0917,
      "step": 15955
    },
    {
      "epoch": 0.37543529411764703,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9986,
      "step": 15956
    },
    {
      "epoch": 0.3754588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9358,
      "step": 15957
    },
    {
      "epoch": 0.3754823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.3146,
      "step": 15958
    },
    {
      "epoch": 0.3755058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9685,
      "step": 15959
    },
    {
      "epoch": 0.3755294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9082,
      "step": 15960
    },
    {
      "epoch": 0.3755529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1877,
      "step": 15961
    },
    {
      "epoch": 0.3755764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9504,
      "step": 15962
    },
    {
      "epoch": 0.3756,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8263,
      "step": 15963
    },
    {
      "epoch": 0.3756235294117647,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3582,
      "step": 15964
    },
    {
      "epoch": 0.3756470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0895,
      "step": 15965
    },
    {
      "epoch": 0.37567058823529415,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0071,
      "step": 15966
    },
    {
      "epoch": 0.37569411764705885,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9697,
      "step": 15967
    },
    {
      "epoch": 0.37571764705882355,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0207,
      "step": 15968
    },
    {
      "epoch": 0.37574117647058825,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0649,
      "step": 15969
    },
    {
      "epoch": 0.37576470588235295,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0864,
      "step": 15970
    },
    {
      "epoch": 0.37578823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9022,
      "step": 15971
    },
    {
      "epoch": 0.37581176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1542,
      "step": 15972
    },
    {
      "epoch": 0.37583529411764705,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.963,
      "step": 15973
    },
    {
      "epoch": 0.37585882352941175,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0027,
      "step": 15974
    },
    {
      "epoch": 0.37588235294117645,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.1502,
      "step": 15975
    },
    {
      "epoch": 0.3759058823529412,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.252,
      "step": 15976
    },
    {
      "epoch": 0.3759294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8725,
      "step": 15977
    },
    {
      "epoch": 0.3759529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1806,
      "step": 15978
    },
    {
      "epoch": 0.3759764705882353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9197,
      "step": 15979
    },
    {
      "epoch": 0.376,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0364,
      "step": 15980
    },
    {
      "epoch": 0.3760235294117647,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9751,
      "step": 15981
    },
    {
      "epoch": 0.3760470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 15982
    },
    {
      "epoch": 0.3760705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.1015,
      "step": 15983
    },
    {
      "epoch": 0.3760941176470588,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1579,
      "step": 15984
    },
    {
      "epoch": 0.37611764705882356,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0219,
      "step": 15985
    },
    {
      "epoch": 0.37614117647058826,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2535,
      "step": 15986
    },
    {
      "epoch": 0.37616470588235296,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7836,
      "step": 15987
    },
    {
      "epoch": 0.37618823529411766,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1926,
      "step": 15988
    },
    {
      "epoch": 0.37621176470588236,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0737,
      "step": 15989
    },
    {
      "epoch": 0.37623529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0235,
      "step": 15990
    },
    {
      "epoch": 0.37625882352941176,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3884,
      "step": 15991
    },
    {
      "epoch": 0.37628235294117646,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1994,
      "step": 15992
    },
    {
      "epoch": 0.37630588235294116,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7312,
      "step": 15993
    },
    {
      "epoch": 0.37632941176470586,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0224,
      "step": 15994
    },
    {
      "epoch": 0.3763529411764706,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.581,
      "step": 15995
    },
    {
      "epoch": 0.3763764705882353,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.5856,
      "step": 15996
    },
    {
      "epoch": 0.3764,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 15997
    },
    {
      "epoch": 0.3764235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2512,
      "step": 15998
    },
    {
      "epoch": 0.3764470588235294,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.02,
      "loss": 0.6554,
      "step": 15999
    },
    {
      "epoch": 0.3764705882352941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.707,
      "step": 16000
    },
    {
      "epoch": 0.3764705882352941,
      "eval_loss": 2.1887595653533936,
      "eval_runtime": 679.3218,
      "eval_samples_per_second": 12.512,
      "eval_steps_per_second": 3.128,
      "step": 16000
    },
    {
      "epoch": 0.3764941176470588,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.785,
      "step": 16001
    },
    {
      "epoch": 0.3765176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1141,
      "step": 16002
    },
    {
      "epoch": 0.3765411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9296,
      "step": 16003
    },
    {
      "epoch": 0.37656470588235297,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9441,
      "step": 16004
    },
    {
      "epoch": 0.37658823529411767,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0682,
      "step": 16005
    },
    {
      "epoch": 0.37661176470588237,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0076,
      "step": 16006
    },
    {
      "epoch": 0.37663529411764707,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0257,
      "step": 16007
    },
    {
      "epoch": 0.37665882352941177,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0939,
      "step": 16008
    },
    {
      "epoch": 0.37668235294117647,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3631,
      "step": 16009
    },
    {
      "epoch": 0.37670588235294117,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1608,
      "step": 16010
    },
    {
      "epoch": 0.37672941176470587,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7462,
      "step": 16011
    },
    {
      "epoch": 0.37675294117647057,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7852,
      "step": 16012
    },
    {
      "epoch": 0.37677647058823527,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9878,
      "step": 16013
    },
    {
      "epoch": 0.3768,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.8971,
      "step": 16014
    },
    {
      "epoch": 0.3768235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.6234,
      "step": 16015
    },
    {
      "epoch": 0.3768470588235294,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.0817,
      "step": 16016
    },
    {
      "epoch": 0.3768705882352941,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0346,
      "step": 16017
    },
    {
      "epoch": 0.3768941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0991,
      "step": 16018
    },
    {
      "epoch": 0.3769176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9831,
      "step": 16019
    },
    {
      "epoch": 0.3769411764705882,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0149,
      "step": 16020
    },
    {
      "epoch": 0.3769647058823529,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2224,
      "step": 16021
    },
    {
      "epoch": 0.3769882352941176,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0567,
      "step": 16022
    },
    {
      "epoch": 0.3770117647058824,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0021,
      "step": 16023
    },
    {
      "epoch": 0.3770352941176471,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8997,
      "step": 16024
    },
    {
      "epoch": 0.3770588235294118,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3875,
      "step": 16025
    },
    {
      "epoch": 0.3770823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.6998,
      "step": 16026
    },
    {
      "epoch": 0.3771058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1264,
      "step": 16027
    },
    {
      "epoch": 0.3771294117647059,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3073,
      "step": 16028
    },
    {
      "epoch": 0.3771529411764706,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.303,
      "step": 16029
    },
    {
      "epoch": 0.3771764705882353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0954,
      "step": 16030
    },
    {
      "epoch": 0.3772,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8882,
      "step": 16031
    },
    {
      "epoch": 0.3772235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0054,
      "step": 16032
    },
    {
      "epoch": 0.37724705882352944,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.025,
      "step": 16033
    },
    {
      "epoch": 0.37727058823529414,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.3355,
      "step": 16034
    },
    {
      "epoch": 0.37729411764705884,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9477,
      "step": 16035
    },
    {
      "epoch": 0.37731764705882354,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0899,
      "step": 16036
    },
    {
      "epoch": 0.37734117647058824,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0526,
      "step": 16037
    },
    {
      "epoch": 0.37736470588235294,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7737,
      "step": 16038
    },
    {
      "epoch": 0.37738823529411764,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6937,
      "step": 16039
    },
    {
      "epoch": 0.37741176470588234,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8623,
      "step": 16040
    },
    {
      "epoch": 0.37743529411764704,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0242,
      "step": 16041
    },
    {
      "epoch": 0.3774588235294118,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1636,
      "step": 16042
    },
    {
      "epoch": 0.3774823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0398,
      "step": 16043
    },
    {
      "epoch": 0.3775058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9567,
      "step": 16044
    },
    {
      "epoch": 0.3775294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1541,
      "step": 16045
    },
    {
      "epoch": 0.3775529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8141,
      "step": 16046
    },
    {
      "epoch": 0.3775764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1109,
      "step": 16047
    },
    {
      "epoch": 0.3776,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8065,
      "step": 16048
    },
    {
      "epoch": 0.3776235294117647,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8225,
      "step": 16049
    },
    {
      "epoch": 0.3776470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1431,
      "step": 16050
    },
    {
      "epoch": 0.3776705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.956,
      "step": 16051
    },
    {
      "epoch": 0.37769411764705885,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9259,
      "step": 16052
    },
    {
      "epoch": 0.37771764705882355,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9604,
      "step": 16053
    },
    {
      "epoch": 0.37774117647058825,
      "grad_norm": 0.220703125,
      "learning_rate": 0.02,
      "loss": 0.5828,
      "step": 16054
    },
    {
      "epoch": 0.37776470588235295,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0216,
      "step": 16055
    },
    {
      "epoch": 0.37778823529411765,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.943,
      "step": 16056
    },
    {
      "epoch": 0.37781176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0593,
      "step": 16057
    },
    {
      "epoch": 0.37783529411764705,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.02,
      "loss": 0.2771,
      "step": 16058
    },
    {
      "epoch": 0.37785882352941175,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9756,
      "step": 16059
    },
    {
      "epoch": 0.37788235294117645,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2674,
      "step": 16060
    },
    {
      "epoch": 0.3779058823529412,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2226,
      "step": 16061
    },
    {
      "epoch": 0.3779294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9653,
      "step": 16062
    },
    {
      "epoch": 0.3779529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0097,
      "step": 16063
    },
    {
      "epoch": 0.3779764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.8841,
      "step": 16064
    },
    {
      "epoch": 0.378,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1301,
      "step": 16065
    },
    {
      "epoch": 0.3780235294117647,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.82,
      "step": 16066
    },
    {
      "epoch": 0.3780470588235294,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1283,
      "step": 16067
    },
    {
      "epoch": 0.3780705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9523,
      "step": 16068
    },
    {
      "epoch": 0.3780941176470588,
      "grad_norm": 0.14453125,
      "learning_rate": 0.02,
      "loss": 1.2622,
      "step": 16069
    },
    {
      "epoch": 0.3781176470588235,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0812,
      "step": 16070
    },
    {
      "epoch": 0.37814117647058826,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2531,
      "step": 16071
    },
    {
      "epoch": 0.37816470588235296,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8077,
      "step": 16072
    },
    {
      "epoch": 0.37818823529411766,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0842,
      "step": 16073
    },
    {
      "epoch": 0.37821176470588236,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8952,
      "step": 16074
    },
    {
      "epoch": 0.37823529411764706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7889,
      "step": 16075
    },
    {
      "epoch": 0.37825882352941176,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1324,
      "step": 16076
    },
    {
      "epoch": 0.37828235294117646,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1381,
      "step": 16077
    },
    {
      "epoch": 0.37830588235294116,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0805,
      "step": 16078
    },
    {
      "epoch": 0.37832941176470586,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2671,
      "step": 16079
    },
    {
      "epoch": 0.3783529411764706,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0603,
      "step": 16080
    },
    {
      "epoch": 0.3783764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1091,
      "step": 16081
    },
    {
      "epoch": 0.3784,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3307,
      "step": 16082
    },
    {
      "epoch": 0.3784235294117647,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0439,
      "step": 16083
    },
    {
      "epoch": 0.3784470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.999,
      "step": 16084
    },
    {
      "epoch": 0.3784705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.6319,
      "step": 16085
    },
    {
      "epoch": 0.3784941176470588,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.02,
      "loss": 1.3559,
      "step": 16086
    },
    {
      "epoch": 0.3785176470588235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9833,
      "step": 16087
    },
    {
      "epoch": 0.3785411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7256,
      "step": 16088
    },
    {
      "epoch": 0.3785647058823529,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8687,
      "step": 16089
    },
    {
      "epoch": 0.37858823529411767,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8482,
      "step": 16090
    },
    {
      "epoch": 0.37861176470588237,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.808,
      "step": 16091
    },
    {
      "epoch": 0.37863529411764707,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0569,
      "step": 16092
    },
    {
      "epoch": 0.37865882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0379,
      "step": 16093
    },
    {
      "epoch": 0.37868235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8041,
      "step": 16094
    },
    {
      "epoch": 0.37870588235294117,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0198,
      "step": 16095
    },
    {
      "epoch": 0.37872941176470587,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1105,
      "step": 16096
    },
    {
      "epoch": 0.37875294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1798,
      "step": 16097
    },
    {
      "epoch": 0.37877647058823527,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1318,
      "step": 16098
    },
    {
      "epoch": 0.3788,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8416,
      "step": 16099
    },
    {
      "epoch": 0.3788235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 16100
    },
    {
      "epoch": 0.3788470588235294,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1109,
      "step": 16101
    },
    {
      "epoch": 0.3788705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1279,
      "step": 16102
    },
    {
      "epoch": 0.3788941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9346,
      "step": 16103
    },
    {
      "epoch": 0.3789176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0972,
      "step": 16104
    },
    {
      "epoch": 0.3789411764705882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0831,
      "step": 16105
    },
    {
      "epoch": 0.3789647058823529,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8602,
      "step": 16106
    },
    {
      "epoch": 0.3789882352941176,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1212,
      "step": 16107
    },
    {
      "epoch": 0.3790117647058823,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0068,
      "step": 16108
    },
    {
      "epoch": 0.3790352941176471,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.784,
      "step": 16109
    },
    {
      "epoch": 0.3790588235294118,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9411,
      "step": 16110
    },
    {
      "epoch": 0.3790823529411765,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.0761,
      "step": 16111
    },
    {
      "epoch": 0.3791058823529412,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0117,
      "step": 16112
    },
    {
      "epoch": 0.3791294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1803,
      "step": 16113
    },
    {
      "epoch": 0.3791529411764706,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1595,
      "step": 16114
    },
    {
      "epoch": 0.3791764705882353,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8662,
      "step": 16115
    },
    {
      "epoch": 0.3792,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3003,
      "step": 16116
    },
    {
      "epoch": 0.3792235294117647,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 16117
    },
    {
      "epoch": 0.37924705882352944,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2058,
      "step": 16118
    },
    {
      "epoch": 0.37927058823529414,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8915,
      "step": 16119
    },
    {
      "epoch": 0.37929411764705884,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.925,
      "step": 16120
    },
    {
      "epoch": 0.37931764705882354,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2837,
      "step": 16121
    },
    {
      "epoch": 0.37934117647058824,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8921,
      "step": 16122
    },
    {
      "epoch": 0.37936470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9161,
      "step": 16123
    },
    {
      "epoch": 0.37938823529411764,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1593,
      "step": 16124
    },
    {
      "epoch": 0.37941176470588234,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0604,
      "step": 16125
    },
    {
      "epoch": 0.37943529411764704,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9057,
      "step": 16126
    },
    {
      "epoch": 0.37945882352941174,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1926,
      "step": 16127
    },
    {
      "epoch": 0.3794823529411765,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.6366,
      "step": 16128
    },
    {
      "epoch": 0.3795058823529412,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.1226,
      "step": 16129
    },
    {
      "epoch": 0.3795294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8431,
      "step": 16130
    },
    {
      "epoch": 0.3795529411764706,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1293,
      "step": 16131
    },
    {
      "epoch": 0.3795764705882353,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 0.9678,
      "step": 16132
    },
    {
      "epoch": 0.3796,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2529,
      "step": 16133
    },
    {
      "epoch": 0.3796235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.817,
      "step": 16134
    },
    {
      "epoch": 0.3796470588235294,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 16135
    },
    {
      "epoch": 0.3796705882352941,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8659,
      "step": 16136
    },
    {
      "epoch": 0.37969411764705885,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0978,
      "step": 16137
    },
    {
      "epoch": 0.37971764705882355,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.075,
      "step": 16138
    },
    {
      "epoch": 0.37974117647058825,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9245,
      "step": 16139
    },
    {
      "epoch": 0.37976470588235295,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7328,
      "step": 16140
    },
    {
      "epoch": 0.37978823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0932,
      "step": 16141
    },
    {
      "epoch": 0.37981176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 16142
    },
    {
      "epoch": 0.37983529411764705,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9962,
      "step": 16143
    },
    {
      "epoch": 0.37985882352941175,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9535,
      "step": 16144
    },
    {
      "epoch": 0.37988235294117645,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8588,
      "step": 16145
    },
    {
      "epoch": 0.37990588235294115,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9648,
      "step": 16146
    },
    {
      "epoch": 0.3799294117647059,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 0.8463,
      "step": 16147
    },
    {
      "epoch": 0.3799529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8535,
      "step": 16148
    },
    {
      "epoch": 0.3799764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9793,
      "step": 16149
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7205,
      "step": 16150
    },
    {
      "epoch": 0.3800235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0685,
      "step": 16151
    },
    {
      "epoch": 0.3800470588235294,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8222,
      "step": 16152
    },
    {
      "epoch": 0.3800705882352941,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2688,
      "step": 16153
    },
    {
      "epoch": 0.3800941176470588,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8868,
      "step": 16154
    },
    {
      "epoch": 0.3801176470588235,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.3019,
      "step": 16155
    },
    {
      "epoch": 0.38014117647058826,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7545,
      "step": 16156
    },
    {
      "epoch": 0.38016470588235296,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8422,
      "step": 16157
    },
    {
      "epoch": 0.38018823529411766,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0835,
      "step": 16158
    },
    {
      "epoch": 0.38021176470588236,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8952,
      "step": 16159
    },
    {
      "epoch": 0.38023529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8667,
      "step": 16160
    },
    {
      "epoch": 0.38025882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9293,
      "step": 16161
    },
    {
      "epoch": 0.38028235294117646,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9667,
      "step": 16162
    },
    {
      "epoch": 0.38030588235294116,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.6207,
      "step": 16163
    },
    {
      "epoch": 0.38032941176470586,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9642,
      "step": 16164
    },
    {
      "epoch": 0.3803529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2212,
      "step": 16165
    },
    {
      "epoch": 0.3803764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1989,
      "step": 16166
    },
    {
      "epoch": 0.3804,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7733,
      "step": 16167
    },
    {
      "epoch": 0.3804235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0126,
      "step": 16168
    },
    {
      "epoch": 0.3804470588235294,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.066,
      "step": 16169
    },
    {
      "epoch": 0.3804705882352941,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1103,
      "step": 16170
    },
    {
      "epoch": 0.3804941176470588,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0606,
      "step": 16171
    },
    {
      "epoch": 0.3805176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9198,
      "step": 16172
    },
    {
      "epoch": 0.3805411764705882,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.993,
      "step": 16173
    },
    {
      "epoch": 0.3805647058823529,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9957,
      "step": 16174
    },
    {
      "epoch": 0.38058823529411767,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2512,
      "step": 16175
    },
    {
      "epoch": 0.38061176470588237,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0457,
      "step": 16176
    },
    {
      "epoch": 0.38063529411764707,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.777,
      "step": 16177
    },
    {
      "epoch": 0.38065882352941177,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8797,
      "step": 16178
    },
    {
      "epoch": 0.38068235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0654,
      "step": 16179
    },
    {
      "epoch": 0.38070588235294117,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 1.0298,
      "step": 16180
    },
    {
      "epoch": 0.38072941176470587,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3132,
      "step": 16181
    },
    {
      "epoch": 0.38075294117647057,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1276,
      "step": 16182
    },
    {
      "epoch": 0.38077647058823527,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.0466,
      "step": 16183
    },
    {
      "epoch": 0.3808,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.082,
      "step": 16184
    },
    {
      "epoch": 0.3808235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1294,
      "step": 16185
    },
    {
      "epoch": 0.3808470588235294,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8285,
      "step": 16186
    },
    {
      "epoch": 0.3808705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1322,
      "step": 16187
    },
    {
      "epoch": 0.3808941176470588,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0586,
      "step": 16188
    },
    {
      "epoch": 0.3809176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 16189
    },
    {
      "epoch": 0.3809411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9584,
      "step": 16190
    },
    {
      "epoch": 0.3809647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.177,
      "step": 16191
    },
    {
      "epoch": 0.3809882352941176,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.9272,
      "step": 16192
    },
    {
      "epoch": 0.3810117647058823,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9045,
      "step": 16193
    },
    {
      "epoch": 0.3810352941176471,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7305,
      "step": 16194
    },
    {
      "epoch": 0.3810588235294118,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7707,
      "step": 16195
    },
    {
      "epoch": 0.3810823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9826,
      "step": 16196
    },
    {
      "epoch": 0.3811058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0626,
      "step": 16197
    },
    {
      "epoch": 0.3811294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0914,
      "step": 16198
    },
    {
      "epoch": 0.3811529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1346,
      "step": 16199
    },
    {
      "epoch": 0.3811764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9903,
      "step": 16200
    },
    {
      "epoch": 0.3812,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0268,
      "step": 16201
    },
    {
      "epoch": 0.3812235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0401,
      "step": 16202
    },
    {
      "epoch": 0.38124705882352944,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8745,
      "step": 16203
    },
    {
      "epoch": 0.38127058823529414,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1349,
      "step": 16204
    },
    {
      "epoch": 0.38129411764705884,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9087,
      "step": 16205
    },
    {
      "epoch": 0.38131764705882354,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8757,
      "step": 16206
    },
    {
      "epoch": 0.38134117647058824,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2248,
      "step": 16207
    },
    {
      "epoch": 0.38136470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1474,
      "step": 16208
    },
    {
      "epoch": 0.38138823529411764,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0408,
      "step": 16209
    },
    {
      "epoch": 0.38141176470588234,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.0821,
      "step": 16210
    },
    {
      "epoch": 0.38143529411764704,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 16211
    },
    {
      "epoch": 0.38145882352941174,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1644,
      "step": 16212
    },
    {
      "epoch": 0.3814823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8784,
      "step": 16213
    },
    {
      "epoch": 0.3815058823529412,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.6372,
      "step": 16214
    },
    {
      "epoch": 0.3815294117647059,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1443,
      "step": 16215
    },
    {
      "epoch": 0.3815529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9694,
      "step": 16216
    },
    {
      "epoch": 0.3815764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9962,
      "step": 16217
    },
    {
      "epoch": 0.3816,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1745,
      "step": 16218
    },
    {
      "epoch": 0.3816235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9793,
      "step": 16219
    },
    {
      "epoch": 0.3816470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0326,
      "step": 16220
    },
    {
      "epoch": 0.3816705882352941,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 16221
    },
    {
      "epoch": 0.38169411764705885,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8015,
      "step": 16222
    },
    {
      "epoch": 0.38171764705882355,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3449,
      "step": 16223
    },
    {
      "epoch": 0.38174117647058825,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1505,
      "step": 16224
    },
    {
      "epoch": 0.38176470588235295,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1509,
      "step": 16225
    },
    {
      "epoch": 0.38178823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0558,
      "step": 16226
    },
    {
      "epoch": 0.38181176470588235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9403,
      "step": 16227
    },
    {
      "epoch": 0.38183529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1068,
      "step": 16228
    },
    {
      "epoch": 0.38185882352941175,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.7776,
      "step": 16229
    },
    {
      "epoch": 0.38188235294117645,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 1.0085,
      "step": 16230
    },
    {
      "epoch": 0.38190588235294115,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2374,
      "step": 16231
    },
    {
      "epoch": 0.3819294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.91,
      "step": 16232
    },
    {
      "epoch": 0.3819529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1045,
      "step": 16233
    },
    {
      "epoch": 0.3819764705882353,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.4921,
      "step": 16234
    },
    {
      "epoch": 0.382,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3399,
      "step": 16235
    },
    {
      "epoch": 0.3820235294117647,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0893,
      "step": 16236
    },
    {
      "epoch": 0.3820470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.2062,
      "step": 16237
    },
    {
      "epoch": 0.3820705882352941,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.8409,
      "step": 16238
    },
    {
      "epoch": 0.3820941176470588,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.1404,
      "step": 16239
    },
    {
      "epoch": 0.3821176470588235,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1126,
      "step": 16240
    },
    {
      "epoch": 0.38214117647058826,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9709,
      "step": 16241
    },
    {
      "epoch": 0.38216470588235296,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7607,
      "step": 16242
    },
    {
      "epoch": 0.38218823529411766,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3298,
      "step": 16243
    },
    {
      "epoch": 0.38221176470588236,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2013,
      "step": 16244
    },
    {
      "epoch": 0.38223529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0733,
      "step": 16245
    },
    {
      "epoch": 0.38225882352941176,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3167,
      "step": 16246
    },
    {
      "epoch": 0.38228235294117646,
      "grad_norm": 0.224609375,
      "learning_rate": 0.02,
      "loss": 0.5858,
      "step": 16247
    },
    {
      "epoch": 0.38230588235294116,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1202,
      "step": 16248
    },
    {
      "epoch": 0.38232941176470586,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9575,
      "step": 16249
    },
    {
      "epoch": 0.38235294117647056,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.6828,
      "step": 16250
    },
    {
      "epoch": 0.3823764705882353,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.4131,
      "step": 16251
    },
    {
      "epoch": 0.3824,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8131,
      "step": 16252
    },
    {
      "epoch": 0.3824235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9767,
      "step": 16253
    },
    {
      "epoch": 0.3824470588235294,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7442,
      "step": 16254
    },
    {
      "epoch": 0.3824705882352941,
      "grad_norm": 0.23828125,
      "learning_rate": 0.02,
      "loss": 0.7401,
      "step": 16255
    },
    {
      "epoch": 0.3824941176470588,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.2326,
      "step": 16256
    },
    {
      "epoch": 0.3825176470588235,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8588,
      "step": 16257
    },
    {
      "epoch": 0.3825411764705882,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0153,
      "step": 16258
    },
    {
      "epoch": 0.3825647058823529,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0252,
      "step": 16259
    },
    {
      "epoch": 0.3825882352941177,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3578,
      "step": 16260
    },
    {
      "epoch": 0.3826117647058824,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8942,
      "step": 16261
    },
    {
      "epoch": 0.3826352941176471,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1539,
      "step": 16262
    },
    {
      "epoch": 0.3826588235294118,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8186,
      "step": 16263
    },
    {
      "epoch": 0.3826823529411765,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0318,
      "step": 16264
    },
    {
      "epoch": 0.3827058823529412,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.8907,
      "step": 16265
    },
    {
      "epoch": 0.3827294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1557,
      "step": 16266
    },
    {
      "epoch": 0.3827529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1344,
      "step": 16267
    },
    {
      "epoch": 0.3827764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.7121,
      "step": 16268
    },
    {
      "epoch": 0.3828,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7309,
      "step": 16269
    },
    {
      "epoch": 0.38282352941176473,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2473,
      "step": 16270
    },
    {
      "epoch": 0.38284705882352943,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.038,
      "step": 16271
    },
    {
      "epoch": 0.38287058823529413,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1051,
      "step": 16272
    },
    {
      "epoch": 0.38289411764705883,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0847,
      "step": 16273
    },
    {
      "epoch": 0.38291764705882353,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8229,
      "step": 16274
    },
    {
      "epoch": 0.38294117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.6137,
      "step": 16275
    },
    {
      "epoch": 0.38296470588235293,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7055,
      "step": 16276
    },
    {
      "epoch": 0.38298823529411763,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8069,
      "step": 16277
    },
    {
      "epoch": 0.38301176470588233,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0593,
      "step": 16278
    },
    {
      "epoch": 0.3830352941176471,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0725,
      "step": 16279
    },
    {
      "epoch": 0.3830588235294118,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1543,
      "step": 16280
    },
    {
      "epoch": 0.3830823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0165,
      "step": 16281
    },
    {
      "epoch": 0.3831058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0335,
      "step": 16282
    },
    {
      "epoch": 0.3831294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8317,
      "step": 16283
    },
    {
      "epoch": 0.3831529411764706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9482,
      "step": 16284
    },
    {
      "epoch": 0.3831764705882353,
      "grad_norm": 0.140625,
      "learning_rate": 0.02,
      "loss": 1.3405,
      "step": 16285
    },
    {
      "epoch": 0.3832,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1371,
      "step": 16286
    },
    {
      "epoch": 0.3832235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3035,
      "step": 16287
    },
    {
      "epoch": 0.3832470588235294,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3179,
      "step": 16288
    },
    {
      "epoch": 0.38327058823529414,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8122,
      "step": 16289
    },
    {
      "epoch": 0.38329411764705884,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2365,
      "step": 16290
    },
    {
      "epoch": 0.38331764705882354,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8094,
      "step": 16291
    },
    {
      "epoch": 0.38334117647058824,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1129,
      "step": 16292
    },
    {
      "epoch": 0.38336470588235294,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0024,
      "step": 16293
    },
    {
      "epoch": 0.38338823529411764,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.7914,
      "step": 16294
    },
    {
      "epoch": 0.38341176470588234,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2366,
      "step": 16295
    },
    {
      "epoch": 0.38343529411764704,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9998,
      "step": 16296
    },
    {
      "epoch": 0.38345882352941174,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8084,
      "step": 16297
    },
    {
      "epoch": 0.3834823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.2275,
      "step": 16298
    },
    {
      "epoch": 0.3835058823529412,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8016,
      "step": 16299
    },
    {
      "epoch": 0.3835294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8482,
      "step": 16300
    },
    {
      "epoch": 0.3835529411764706,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8791,
      "step": 16301
    },
    {
      "epoch": 0.3835764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 0.9391,
      "step": 16302
    },
    {
      "epoch": 0.3836,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7869,
      "step": 16303
    },
    {
      "epoch": 0.3836235294117647,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7994,
      "step": 16304
    },
    {
      "epoch": 0.3836470588235294,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2169,
      "step": 16305
    },
    {
      "epoch": 0.3836705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.803,
      "step": 16306
    },
    {
      "epoch": 0.3836941176470588,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 16307
    },
    {
      "epoch": 0.38371764705882355,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.8988,
      "step": 16308
    },
    {
      "epoch": 0.38374117647058825,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0296,
      "step": 16309
    },
    {
      "epoch": 0.38376470588235295,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.2276,
      "step": 16310
    },
    {
      "epoch": 0.38378823529411765,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8004,
      "step": 16311
    },
    {
      "epoch": 0.38381176470588235,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2559,
      "step": 16312
    },
    {
      "epoch": 0.38383529411764705,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0309,
      "step": 16313
    },
    {
      "epoch": 0.38385882352941175,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9358,
      "step": 16314
    },
    {
      "epoch": 0.38388235294117645,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7231,
      "step": 16315
    },
    {
      "epoch": 0.38390588235294115,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8824,
      "step": 16316
    },
    {
      "epoch": 0.3839294117647059,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.27,
      "step": 16317
    },
    {
      "epoch": 0.3839529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1324,
      "step": 16318
    },
    {
      "epoch": 0.3839764705882353,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7081,
      "step": 16319
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9098,
      "step": 16320
    },
    {
      "epoch": 0.3840235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0781,
      "step": 16321
    },
    {
      "epoch": 0.3840470588235294,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2048,
      "step": 16322
    },
    {
      "epoch": 0.3840705882352941,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2047,
      "step": 16323
    },
    {
      "epoch": 0.3840941176470588,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1113,
      "step": 16324
    },
    {
      "epoch": 0.3841176470588235,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0232,
      "step": 16325
    },
    {
      "epoch": 0.3841411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8684,
      "step": 16326
    },
    {
      "epoch": 0.38416470588235296,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2304,
      "step": 16327
    },
    {
      "epoch": 0.38418823529411766,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 16328
    },
    {
      "epoch": 0.38421176470588236,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0286,
      "step": 16329
    },
    {
      "epoch": 0.38423529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7924,
      "step": 16330
    },
    {
      "epoch": 0.38425882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.874,
      "step": 16331
    },
    {
      "epoch": 0.38428235294117646,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1492,
      "step": 16332
    },
    {
      "epoch": 0.38430588235294116,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.6757,
      "step": 16333
    },
    {
      "epoch": 0.38432941176470586,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.083,
      "step": 16334
    },
    {
      "epoch": 0.38435294117647056,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8586,
      "step": 16335
    },
    {
      "epoch": 0.3843764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9033,
      "step": 16336
    },
    {
      "epoch": 0.3844,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1198,
      "step": 16337
    },
    {
      "epoch": 0.3844235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.098,
      "step": 16338
    },
    {
      "epoch": 0.3844470588235294,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3585,
      "step": 16339
    },
    {
      "epoch": 0.3844705882352941,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1468,
      "step": 16340
    },
    {
      "epoch": 0.3844941176470588,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.2308,
      "step": 16341
    },
    {
      "epoch": 0.3845176470588235,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.965,
      "step": 16342
    },
    {
      "epoch": 0.3845411764705882,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0482,
      "step": 16343
    },
    {
      "epoch": 0.3845647058823529,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1055,
      "step": 16344
    },
    {
      "epoch": 0.3845882352941176,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1992,
      "step": 16345
    },
    {
      "epoch": 0.3846117647058824,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9773,
      "step": 16346
    },
    {
      "epoch": 0.3846352941176471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1256,
      "step": 16347
    },
    {
      "epoch": 0.3846588235294118,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1683,
      "step": 16348
    },
    {
      "epoch": 0.3846823529411765,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.6862,
      "step": 16349
    },
    {
      "epoch": 0.3847058823529412,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.8372,
      "step": 16350
    },
    {
      "epoch": 0.3847294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8921,
      "step": 16351
    },
    {
      "epoch": 0.3847529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9168,
      "step": 16352
    },
    {
      "epoch": 0.3847764705882353,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9245,
      "step": 16353
    },
    {
      "epoch": 0.3848,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.6051,
      "step": 16354
    },
    {
      "epoch": 0.38482352941176473,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0836,
      "step": 16355
    },
    {
      "epoch": 0.38484705882352943,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8655,
      "step": 16356
    },
    {
      "epoch": 0.38487058823529413,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1325,
      "step": 16357
    },
    {
      "epoch": 0.38489411764705883,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.9186,
      "step": 16358
    },
    {
      "epoch": 0.38491764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0327,
      "step": 16359
    },
    {
      "epoch": 0.38494117647058823,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9943,
      "step": 16360
    },
    {
      "epoch": 0.38496470588235293,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.9529,
      "step": 16361
    },
    {
      "epoch": 0.38498823529411763,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8332,
      "step": 16362
    },
    {
      "epoch": 0.38501176470588233,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8551,
      "step": 16363
    },
    {
      "epoch": 0.38503529411764703,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9212,
      "step": 16364
    },
    {
      "epoch": 0.3850588235294118,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0025,
      "step": 16365
    },
    {
      "epoch": 0.3850823529411765,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0809,
      "step": 16366
    },
    {
      "epoch": 0.3851058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9987,
      "step": 16367
    },
    {
      "epoch": 0.3851294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9054,
      "step": 16368
    },
    {
      "epoch": 0.3851529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7586,
      "step": 16369
    },
    {
      "epoch": 0.3851764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0396,
      "step": 16370
    },
    {
      "epoch": 0.3852,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9724,
      "step": 16371
    },
    {
      "epoch": 0.3852235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9932,
      "step": 16372
    },
    {
      "epoch": 0.3852470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0637,
      "step": 16373
    },
    {
      "epoch": 0.38527058823529414,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0449,
      "step": 16374
    },
    {
      "epoch": 0.38529411764705884,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 16375
    },
    {
      "epoch": 0.38531764705882354,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.5061,
      "step": 16376
    },
    {
      "epoch": 0.38534117647058824,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8025,
      "step": 16377
    },
    {
      "epoch": 0.38536470588235294,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.0877,
      "step": 16378
    },
    {
      "epoch": 0.38538823529411764,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3227,
      "step": 16379
    },
    {
      "epoch": 0.38541176470588234,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1354,
      "step": 16380
    },
    {
      "epoch": 0.38543529411764704,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2112,
      "step": 16381
    },
    {
      "epoch": 0.38545882352941174,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8521,
      "step": 16382
    },
    {
      "epoch": 0.3854823529411765,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0496,
      "step": 16383
    },
    {
      "epoch": 0.3855058823529412,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0056,
      "step": 16384
    },
    {
      "epoch": 0.3855294117647059,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7023,
      "step": 16385
    },
    {
      "epoch": 0.3855529411764706,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6639,
      "step": 16386
    },
    {
      "epoch": 0.3855764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9047,
      "step": 16387
    },
    {
      "epoch": 0.3856,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0882,
      "step": 16388
    },
    {
      "epoch": 0.3856235294117647,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8233,
      "step": 16389
    },
    {
      "epoch": 0.3856470588235294,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.105,
      "step": 16390
    },
    {
      "epoch": 0.3856705882352941,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0034,
      "step": 16391
    },
    {
      "epoch": 0.3856941176470588,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6684,
      "step": 16392
    },
    {
      "epoch": 0.38571764705882355,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7092,
      "step": 16393
    },
    {
      "epoch": 0.38574117647058825,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1173,
      "step": 16394
    },
    {
      "epoch": 0.38576470588235295,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8763,
      "step": 16395
    },
    {
      "epoch": 0.38578823529411765,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1478,
      "step": 16396
    },
    {
      "epoch": 0.38581176470588235,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 0.9711,
      "step": 16397
    },
    {
      "epoch": 0.38583529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9924,
      "step": 16398
    },
    {
      "epoch": 0.38585882352941175,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8533,
      "step": 16399
    },
    {
      "epoch": 0.38588235294117645,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9092,
      "step": 16400
    },
    {
      "epoch": 0.38590588235294115,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.4722,
      "step": 16401
    },
    {
      "epoch": 0.3859294117647059,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.175,
      "step": 16402
    },
    {
      "epoch": 0.3859529411764706,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.0445,
      "step": 16403
    },
    {
      "epoch": 0.3859764705882353,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.6673,
      "step": 16404
    },
    {
      "epoch": 0.386,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.6598,
      "step": 16405
    },
    {
      "epoch": 0.3860235294117647,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.333,
      "step": 16406
    },
    {
      "epoch": 0.3860470588235294,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2971,
      "step": 16407
    },
    {
      "epoch": 0.3860705882352941,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.0359,
      "step": 16408
    },
    {
      "epoch": 0.3860941176470588,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.127,
      "step": 16409
    },
    {
      "epoch": 0.3861176470588235,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7723,
      "step": 16410
    },
    {
      "epoch": 0.3861411764705882,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1386,
      "step": 16411
    },
    {
      "epoch": 0.38616470588235297,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.589,
      "step": 16412
    },
    {
      "epoch": 0.38618823529411767,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.2233,
      "step": 16413
    },
    {
      "epoch": 0.38621176470588237,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1355,
      "step": 16414
    },
    {
      "epoch": 0.38623529411764707,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2813,
      "step": 16415
    },
    {
      "epoch": 0.38625882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.987,
      "step": 16416
    },
    {
      "epoch": 0.38628235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 1.0223,
      "step": 16417
    },
    {
      "epoch": 0.38630588235294117,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.455,
      "step": 16418
    },
    {
      "epoch": 0.38632941176470587,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2431,
      "step": 16419
    },
    {
      "epoch": 0.38635294117647057,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.8345,
      "step": 16420
    },
    {
      "epoch": 0.3863764705882353,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.8123,
      "step": 16421
    },
    {
      "epoch": 0.3864,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0193,
      "step": 16422
    },
    {
      "epoch": 0.3864235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1112,
      "step": 16423
    },
    {
      "epoch": 0.3864470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2774,
      "step": 16424
    },
    {
      "epoch": 0.3864705882352941,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7543,
      "step": 16425
    },
    {
      "epoch": 0.3864941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.216,
      "step": 16426
    },
    {
      "epoch": 0.3865176470588235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8591,
      "step": 16427
    },
    {
      "epoch": 0.3865411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9755,
      "step": 16428
    },
    {
      "epoch": 0.3865647058823529,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1323,
      "step": 16429
    },
    {
      "epoch": 0.3865882352941176,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1977,
      "step": 16430
    },
    {
      "epoch": 0.3866117647058824,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9755,
      "step": 16431
    },
    {
      "epoch": 0.3866352941176471,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8802,
      "step": 16432
    },
    {
      "epoch": 0.3866588235294118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1748,
      "step": 16433
    },
    {
      "epoch": 0.3866823529411765,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8173,
      "step": 16434
    },
    {
      "epoch": 0.3867058823529412,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.864,
      "step": 16435
    },
    {
      "epoch": 0.3867294117647059,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 16436
    },
    {
      "epoch": 0.3867529411764706,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8835,
      "step": 16437
    },
    {
      "epoch": 0.3867764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0679,
      "step": 16438
    },
    {
      "epoch": 0.3868,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9636,
      "step": 16439
    },
    {
      "epoch": 0.38682352941176473,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9268,
      "step": 16440
    },
    {
      "epoch": 0.38684705882352943,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.8573,
      "step": 16441
    },
    {
      "epoch": 0.38687058823529413,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.2429,
      "step": 16442
    },
    {
      "epoch": 0.38689411764705883,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8123,
      "step": 16443
    },
    {
      "epoch": 0.38691764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9236,
      "step": 16444
    },
    {
      "epoch": 0.38694117647058823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8825,
      "step": 16445
    },
    {
      "epoch": 0.38696470588235293,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.946,
      "step": 16446
    },
    {
      "epoch": 0.38698823529411763,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9466,
      "step": 16447
    },
    {
      "epoch": 0.38701176470588233,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7787,
      "step": 16448
    },
    {
      "epoch": 0.38703529411764703,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9074,
      "step": 16449
    },
    {
      "epoch": 0.3870588235294118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 16450
    },
    {
      "epoch": 0.3870823529411765,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8858,
      "step": 16451
    },
    {
      "epoch": 0.3871058823529412,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2844,
      "step": 16452
    },
    {
      "epoch": 0.3871294117647059,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9385,
      "step": 16453
    },
    {
      "epoch": 0.3871529411764706,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9256,
      "step": 16454
    },
    {
      "epoch": 0.3871764705882353,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9833,
      "step": 16455
    },
    {
      "epoch": 0.3872,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.217,
      "step": 16456
    },
    {
      "epoch": 0.3872235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9825,
      "step": 16457
    },
    {
      "epoch": 0.3872470588235294,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.1983,
      "step": 16458
    },
    {
      "epoch": 0.38727058823529414,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1868,
      "step": 16459
    },
    {
      "epoch": 0.38729411764705884,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3447,
      "step": 16460
    },
    {
      "epoch": 0.38731764705882354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9945,
      "step": 16461
    },
    {
      "epoch": 0.38734117647058824,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9723,
      "step": 16462
    },
    {
      "epoch": 0.38736470588235294,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.2676,
      "step": 16463
    },
    {
      "epoch": 0.38738823529411764,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0168,
      "step": 16464
    },
    {
      "epoch": 0.38741176470588234,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.5456,
      "step": 16465
    },
    {
      "epoch": 0.38743529411764704,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9068,
      "step": 16466
    },
    {
      "epoch": 0.38745882352941174,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1215,
      "step": 16467
    },
    {
      "epoch": 0.38748235294117644,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1714,
      "step": 16468
    },
    {
      "epoch": 0.3875058823529412,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0933,
      "step": 16469
    },
    {
      "epoch": 0.3875294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0415,
      "step": 16470
    },
    {
      "epoch": 0.3875529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8946,
      "step": 16471
    },
    {
      "epoch": 0.3875764705882353,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.76,
      "step": 16472
    },
    {
      "epoch": 0.3876,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0525,
      "step": 16473
    },
    {
      "epoch": 0.3876235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0708,
      "step": 16474
    },
    {
      "epoch": 0.3876470588235294,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8866,
      "step": 16475
    },
    {
      "epoch": 0.3876705882352941,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7921,
      "step": 16476
    },
    {
      "epoch": 0.3876941176470588,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.005,
      "step": 16477
    },
    {
      "epoch": 0.38771764705882356,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1647,
      "step": 16478
    },
    {
      "epoch": 0.38774117647058826,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8147,
      "step": 16479
    },
    {
      "epoch": 0.38776470588235296,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.731,
      "step": 16480
    },
    {
      "epoch": 0.38778823529411766,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0595,
      "step": 16481
    },
    {
      "epoch": 0.38781176470588236,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0761,
      "step": 16482
    },
    {
      "epoch": 0.38783529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1093,
      "step": 16483
    },
    {
      "epoch": 0.38785882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8892,
      "step": 16484
    },
    {
      "epoch": 0.38788235294117646,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1305,
      "step": 16485
    },
    {
      "epoch": 0.38790588235294116,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 0.9942,
      "step": 16486
    },
    {
      "epoch": 0.38792941176470586,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.1589,
      "step": 16487
    },
    {
      "epoch": 0.3879529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.728,
      "step": 16488
    },
    {
      "epoch": 0.3879764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8642,
      "step": 16489
    },
    {
      "epoch": 0.388,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1227,
      "step": 16490
    },
    {
      "epoch": 0.3880235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0882,
      "step": 16491
    },
    {
      "epoch": 0.3880470588235294,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.8227,
      "step": 16492
    },
    {
      "epoch": 0.3880705882352941,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1548,
      "step": 16493
    },
    {
      "epoch": 0.3880941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.095,
      "step": 16494
    },
    {
      "epoch": 0.3881176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0696,
      "step": 16495
    },
    {
      "epoch": 0.3881411764705882,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1151,
      "step": 16496
    },
    {
      "epoch": 0.38816470588235297,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1905,
      "step": 16497
    },
    {
      "epoch": 0.38818823529411767,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8779,
      "step": 16498
    },
    {
      "epoch": 0.38821176470588237,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.2162,
      "step": 16499
    },
    {
      "epoch": 0.38823529411764707,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1285,
      "step": 16500
    },
    {
      "epoch": 0.38825882352941177,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8795,
      "step": 16501
    },
    {
      "epoch": 0.38828235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1971,
      "step": 16502
    },
    {
      "epoch": 0.38830588235294117,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.5896,
      "step": 16503
    },
    {
      "epoch": 0.38832941176470587,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0227,
      "step": 16504
    },
    {
      "epoch": 0.38835294117647057,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.016,
      "step": 16505
    },
    {
      "epoch": 0.38837647058823527,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0553,
      "step": 16506
    },
    {
      "epoch": 0.3884,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7392,
      "step": 16507
    },
    {
      "epoch": 0.3884235294117647,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.2052,
      "step": 16508
    },
    {
      "epoch": 0.3884470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8276,
      "step": 16509
    },
    {
      "epoch": 0.3884705882352941,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9473,
      "step": 16510
    },
    {
      "epoch": 0.3884941176470588,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2865,
      "step": 16511
    },
    {
      "epoch": 0.3885176470588235,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9377,
      "step": 16512
    },
    {
      "epoch": 0.3885411764705882,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8557,
      "step": 16513
    },
    {
      "epoch": 0.3885647058823529,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9394,
      "step": 16514
    },
    {
      "epoch": 0.3885882352941176,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0062,
      "step": 16515
    },
    {
      "epoch": 0.3886117647058824,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.6945,
      "step": 16516
    },
    {
      "epoch": 0.3886352941176471,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2094,
      "step": 16517
    },
    {
      "epoch": 0.3886588235294118,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0186,
      "step": 16518
    },
    {
      "epoch": 0.3886823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0496,
      "step": 16519
    },
    {
      "epoch": 0.3887058823529412,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0356,
      "step": 16520
    },
    {
      "epoch": 0.3887294117647059,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.781,
      "step": 16521
    },
    {
      "epoch": 0.3887529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8703,
      "step": 16522
    },
    {
      "epoch": 0.3887764705882353,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8155,
      "step": 16523
    },
    {
      "epoch": 0.3888,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.77,
      "step": 16524
    },
    {
      "epoch": 0.3888235294117647,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.6988,
      "step": 16525
    },
    {
      "epoch": 0.38884705882352943,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.7295,
      "step": 16526
    },
    {
      "epoch": 0.38887058823529413,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9785,
      "step": 16527
    },
    {
      "epoch": 0.38889411764705883,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.087,
      "step": 16528
    },
    {
      "epoch": 0.38891764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8425,
      "step": 16529
    },
    {
      "epoch": 0.38894117647058823,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8525,
      "step": 16530
    },
    {
      "epoch": 0.38896470588235293,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1751,
      "step": 16531
    },
    {
      "epoch": 0.38898823529411763,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 16532
    },
    {
      "epoch": 0.38901176470588233,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.143,
      "step": 16533
    },
    {
      "epoch": 0.38903529411764703,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9703,
      "step": 16534
    },
    {
      "epoch": 0.3890588235294118,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.063,
      "step": 16535
    },
    {
      "epoch": 0.3890823529411765,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0276,
      "step": 16536
    },
    {
      "epoch": 0.3891058823529412,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 16537
    },
    {
      "epoch": 0.3891294117647059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6346,
      "step": 16538
    },
    {
      "epoch": 0.3891529411764706,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0473,
      "step": 16539
    },
    {
      "epoch": 0.3891764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7886,
      "step": 16540
    },
    {
      "epoch": 0.3892,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.8515,
      "step": 16541
    },
    {
      "epoch": 0.3892235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9544,
      "step": 16542
    },
    {
      "epoch": 0.3892470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0497,
      "step": 16543
    },
    {
      "epoch": 0.3892705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0768,
      "step": 16544
    },
    {
      "epoch": 0.38929411764705885,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1452,
      "step": 16545
    },
    {
      "epoch": 0.38931764705882355,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.0338,
      "step": 16546
    },
    {
      "epoch": 0.38934117647058825,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0391,
      "step": 16547
    },
    {
      "epoch": 0.38936470588235295,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1854,
      "step": 16548
    },
    {
      "epoch": 0.38938823529411765,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.9483,
      "step": 16549
    },
    {
      "epoch": 0.38941176470588235,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9914,
      "step": 16550
    },
    {
      "epoch": 0.38943529411764705,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0444,
      "step": 16551
    },
    {
      "epoch": 0.38945882352941175,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.098,
      "step": 16552
    },
    {
      "epoch": 0.38948235294117645,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9406,
      "step": 16553
    },
    {
      "epoch": 0.3895058823529412,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 16554
    },
    {
      "epoch": 0.3895294117647059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.6352,
      "step": 16555
    },
    {
      "epoch": 0.3895529411764706,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1382,
      "step": 16556
    },
    {
      "epoch": 0.3895764705882353,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8018,
      "step": 16557
    },
    {
      "epoch": 0.3896,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0358,
      "step": 16558
    },
    {
      "epoch": 0.3896235294117647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.246,
      "step": 16559
    },
    {
      "epoch": 0.3896470588235294,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7771,
      "step": 16560
    },
    {
      "epoch": 0.3896705882352941,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0438,
      "step": 16561
    },
    {
      "epoch": 0.3896941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9489,
      "step": 16562
    },
    {
      "epoch": 0.3897176470588235,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3187,
      "step": 16563
    },
    {
      "epoch": 0.38974117647058826,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9581,
      "step": 16564
    },
    {
      "epoch": 0.38976470588235296,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9423,
      "step": 16565
    },
    {
      "epoch": 0.38978823529411766,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 16566
    },
    {
      "epoch": 0.38981176470588236,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9656,
      "step": 16567
    },
    {
      "epoch": 0.38983529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1367,
      "step": 16568
    },
    {
      "epoch": 0.38985882352941176,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1961,
      "step": 16569
    },
    {
      "epoch": 0.38988235294117646,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.6735,
      "step": 16570
    },
    {
      "epoch": 0.38990588235294116,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9676,
      "step": 16571
    },
    {
      "epoch": 0.38992941176470586,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2077,
      "step": 16572
    },
    {
      "epoch": 0.3899529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0624,
      "step": 16573
    },
    {
      "epoch": 0.3899764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9891,
      "step": 16574
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.986,
      "step": 16575
    },
    {
      "epoch": 0.3900235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0052,
      "step": 16576
    },
    {
      "epoch": 0.3900470588235294,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0565,
      "step": 16577
    },
    {
      "epoch": 0.3900705882352941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2285,
      "step": 16578
    },
    {
      "epoch": 0.3900941176470588,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0551,
      "step": 16579
    },
    {
      "epoch": 0.3901176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8578,
      "step": 16580
    },
    {
      "epoch": 0.3901411764705882,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.7092,
      "step": 16581
    },
    {
      "epoch": 0.3901647058823529,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1883,
      "step": 16582
    },
    {
      "epoch": 0.39018823529411767,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.009,
      "step": 16583
    },
    {
      "epoch": 0.39021176470588237,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9759,
      "step": 16584
    },
    {
      "epoch": 0.39023529411764707,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3439,
      "step": 16585
    },
    {
      "epoch": 0.39025882352941177,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.6767,
      "step": 16586
    },
    {
      "epoch": 0.39028235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1006,
      "step": 16587
    },
    {
      "epoch": 0.39030588235294117,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 0.9166,
      "step": 16588
    },
    {
      "epoch": 0.39032941176470587,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.893,
      "step": 16589
    },
    {
      "epoch": 0.39035294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0232,
      "step": 16590
    },
    {
      "epoch": 0.39037647058823527,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0377,
      "step": 16591
    },
    {
      "epoch": 0.3904,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7344,
      "step": 16592
    },
    {
      "epoch": 0.3904235294117647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2685,
      "step": 16593
    },
    {
      "epoch": 0.3904470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1446,
      "step": 16594
    },
    {
      "epoch": 0.3904705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0096,
      "step": 16595
    },
    {
      "epoch": 0.3904941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0459,
      "step": 16596
    },
    {
      "epoch": 0.3905176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9786,
      "step": 16597
    },
    {
      "epoch": 0.3905411764705882,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2139,
      "step": 16598
    },
    {
      "epoch": 0.3905647058823529,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 1.1048,
      "step": 16599
    },
    {
      "epoch": 0.3905882352941176,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1369,
      "step": 16600
    },
    {
      "epoch": 0.3906117647058824,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.8104,
      "step": 16601
    },
    {
      "epoch": 0.3906352941176471,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.0845,
      "step": 16602
    },
    {
      "epoch": 0.3906588235294118,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2043,
      "step": 16603
    },
    {
      "epoch": 0.3906823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.7813,
      "step": 16604
    },
    {
      "epoch": 0.3907058823529412,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0814,
      "step": 16605
    },
    {
      "epoch": 0.3907294117647059,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0277,
      "step": 16606
    },
    {
      "epoch": 0.3907529411764706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0756,
      "step": 16607
    },
    {
      "epoch": 0.3907764705882353,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9918,
      "step": 16608
    },
    {
      "epoch": 0.3908,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0501,
      "step": 16609
    },
    {
      "epoch": 0.3908235294117647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1114,
      "step": 16610
    },
    {
      "epoch": 0.39084705882352944,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.8973,
      "step": 16611
    },
    {
      "epoch": 0.39087058823529414,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9492,
      "step": 16612
    },
    {
      "epoch": 0.39089411764705884,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1192,
      "step": 16613
    },
    {
      "epoch": 0.39091764705882354,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.99,
      "step": 16614
    },
    {
      "epoch": 0.39094117647058824,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.281,
      "step": 16615
    },
    {
      "epoch": 0.39096470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9091,
      "step": 16616
    },
    {
      "epoch": 0.39098823529411764,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.823,
      "step": 16617
    },
    {
      "epoch": 0.39101176470588234,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9175,
      "step": 16618
    },
    {
      "epoch": 0.39103529411764704,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1988,
      "step": 16619
    },
    {
      "epoch": 0.3910588235294118,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.0226,
      "step": 16620
    },
    {
      "epoch": 0.3910823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0192,
      "step": 16621
    },
    {
      "epoch": 0.3911058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.853,
      "step": 16622
    },
    {
      "epoch": 0.3911294117647059,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.3041,
      "step": 16623
    },
    {
      "epoch": 0.3911529411764706,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 0.9703,
      "step": 16624
    },
    {
      "epoch": 0.3911764705882353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0062,
      "step": 16625
    },
    {
      "epoch": 0.3912,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9695,
      "step": 16626
    },
    {
      "epoch": 0.3912235294117647,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.5148,
      "step": 16627
    },
    {
      "epoch": 0.3912470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0842,
      "step": 16628
    },
    {
      "epoch": 0.3912705882352941,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8902,
      "step": 16629
    },
    {
      "epoch": 0.39129411764705885,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1575,
      "step": 16630
    },
    {
      "epoch": 0.39131764705882355,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0306,
      "step": 16631
    },
    {
      "epoch": 0.39134117647058825,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7129,
      "step": 16632
    },
    {
      "epoch": 0.39136470588235295,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1904,
      "step": 16633
    },
    {
      "epoch": 0.39138823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.006,
      "step": 16634
    },
    {
      "epoch": 0.39141176470588235,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0695,
      "step": 16635
    },
    {
      "epoch": 0.39143529411764705,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9108,
      "step": 16636
    },
    {
      "epoch": 0.39145882352941175,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9691,
      "step": 16637
    },
    {
      "epoch": 0.39148235294117645,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9046,
      "step": 16638
    },
    {
      "epoch": 0.3915058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1064,
      "step": 16639
    },
    {
      "epoch": 0.3915294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9354,
      "step": 16640
    },
    {
      "epoch": 0.3915529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0733,
      "step": 16641
    },
    {
      "epoch": 0.3915764705882353,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.1897,
      "step": 16642
    },
    {
      "epoch": 0.3916,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 1.0767,
      "step": 16643
    },
    {
      "epoch": 0.3916235294117647,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9396,
      "step": 16644
    },
    {
      "epoch": 0.3916470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1594,
      "step": 16645
    },
    {
      "epoch": 0.3916705882352941,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 16646
    },
    {
      "epoch": 0.3916941176470588,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2195,
      "step": 16647
    },
    {
      "epoch": 0.3917176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.982,
      "step": 16648
    },
    {
      "epoch": 0.39174117647058826,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9938,
      "step": 16649
    },
    {
      "epoch": 0.39176470588235296,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0676,
      "step": 16650
    },
    {
      "epoch": 0.39178823529411766,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 16651
    },
    {
      "epoch": 0.39181176470588236,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0961,
      "step": 16652
    },
    {
      "epoch": 0.39183529411764706,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2341,
      "step": 16653
    },
    {
      "epoch": 0.39185882352941176,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 1.1667,
      "step": 16654
    },
    {
      "epoch": 0.39188235294117646,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 16655
    },
    {
      "epoch": 0.39190588235294116,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2285,
      "step": 16656
    },
    {
      "epoch": 0.39192941176470586,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9547,
      "step": 16657
    },
    {
      "epoch": 0.3919529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.7865,
      "step": 16658
    },
    {
      "epoch": 0.3919764705882353,
      "grad_norm": 0.138671875,
      "learning_rate": 0.02,
      "loss": 1.2568,
      "step": 16659
    },
    {
      "epoch": 0.392,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.103,
      "step": 16660
    },
    {
      "epoch": 0.3920235294117647,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 16661
    },
    {
      "epoch": 0.3920470588235294,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.0869,
      "step": 16662
    },
    {
      "epoch": 0.3920705882352941,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.4,
      "step": 16663
    },
    {
      "epoch": 0.3920941176470588,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1005,
      "step": 16664
    },
    {
      "epoch": 0.3921176470588235,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0942,
      "step": 16665
    },
    {
      "epoch": 0.3921411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0142,
      "step": 16666
    },
    {
      "epoch": 0.3921647058823529,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7945,
      "step": 16667
    },
    {
      "epoch": 0.39218823529411767,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1641,
      "step": 16668
    },
    {
      "epoch": 0.39221176470588237,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1234,
      "step": 16669
    },
    {
      "epoch": 0.39223529411764707,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9367,
      "step": 16670
    },
    {
      "epoch": 0.39225882352941177,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.6296,
      "step": 16671
    },
    {
      "epoch": 0.39228235294117647,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.02,
      "loss": 1.4395,
      "step": 16672
    },
    {
      "epoch": 0.39230588235294117,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0976,
      "step": 16673
    },
    {
      "epoch": 0.39232941176470587,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0391,
      "step": 16674
    },
    {
      "epoch": 0.39235294117647057,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0199,
      "step": 16675
    },
    {
      "epoch": 0.39237647058823527,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 16676
    },
    {
      "epoch": 0.3924,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.4876,
      "step": 16677
    },
    {
      "epoch": 0.3924235294117647,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8537,
      "step": 16678
    },
    {
      "epoch": 0.3924470588235294,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1253,
      "step": 16679
    },
    {
      "epoch": 0.3924705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0204,
      "step": 16680
    },
    {
      "epoch": 0.3924941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9015,
      "step": 16681
    },
    {
      "epoch": 0.3925176470588235,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.2291,
      "step": 16682
    },
    {
      "epoch": 0.3925411764705882,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8513,
      "step": 16683
    },
    {
      "epoch": 0.3925647058823529,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9811,
      "step": 16684
    },
    {
      "epoch": 0.3925882352941176,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.166,
      "step": 16685
    },
    {
      "epoch": 0.3926117647058823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7847,
      "step": 16686
    },
    {
      "epoch": 0.3926352941176471,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.6571,
      "step": 16687
    },
    {
      "epoch": 0.3926588235294118,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7295,
      "step": 16688
    },
    {
      "epoch": 0.3926823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9324,
      "step": 16689
    },
    {
      "epoch": 0.3927058823529412,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.3082,
      "step": 16690
    },
    {
      "epoch": 0.3927294117647059,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 16691
    },
    {
      "epoch": 0.3927529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1183,
      "step": 16692
    },
    {
      "epoch": 0.3927764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1243,
      "step": 16693
    },
    {
      "epoch": 0.3928,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.527,
      "step": 16694
    },
    {
      "epoch": 0.3928235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.224,
      "step": 16695
    },
    {
      "epoch": 0.39284705882352944,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9716,
      "step": 16696
    },
    {
      "epoch": 0.39287058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8304,
      "step": 16697
    },
    {
      "epoch": 0.39289411764705884,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8584,
      "step": 16698
    },
    {
      "epoch": 0.39291764705882354,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1492,
      "step": 16699
    },
    {
      "epoch": 0.39294117647058824,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 0.8667,
      "step": 16700
    },
    {
      "epoch": 0.39296470588235294,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0958,
      "step": 16701
    },
    {
      "epoch": 0.39298823529411764,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8867,
      "step": 16702
    },
    {
      "epoch": 0.39301176470588234,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.634,
      "step": 16703
    },
    {
      "epoch": 0.39303529411764704,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8051,
      "step": 16704
    },
    {
      "epoch": 0.39305882352941174,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7941,
      "step": 16705
    },
    {
      "epoch": 0.3930823529411765,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.4186,
      "step": 16706
    },
    {
      "epoch": 0.3931058823529412,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.5096,
      "step": 16707
    },
    {
      "epoch": 0.3931294117647059,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.77,
      "step": 16708
    },
    {
      "epoch": 0.3931529411764706,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0488,
      "step": 16709
    },
    {
      "epoch": 0.3931764705882353,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1777,
      "step": 16710
    },
    {
      "epoch": 0.3932,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8805,
      "step": 16711
    },
    {
      "epoch": 0.3932235294117647,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2199,
      "step": 16712
    },
    {
      "epoch": 0.3932470588235294,
      "grad_norm": 0.228515625,
      "learning_rate": 0.02,
      "loss": 1.2898,
      "step": 16713
    },
    {
      "epoch": 0.3932705882352941,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9307,
      "step": 16714
    },
    {
      "epoch": 0.39329411764705885,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.02,
      "loss": 0.7767,
      "step": 16715
    },
    {
      "epoch": 0.39331764705882355,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1083,
      "step": 16716
    },
    {
      "epoch": 0.39334117647058825,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9768,
      "step": 16717
    },
    {
      "epoch": 0.39336470588235295,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0393,
      "step": 16718
    },
    {
      "epoch": 0.39338823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1266,
      "step": 16719
    },
    {
      "epoch": 0.39341176470588235,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0249,
      "step": 16720
    },
    {
      "epoch": 0.39343529411764705,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.017,
      "step": 16721
    },
    {
      "epoch": 0.39345882352941175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0347,
      "step": 16722
    },
    {
      "epoch": 0.39348235294117645,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.705,
      "step": 16723
    },
    {
      "epoch": 0.39350588235294115,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1235,
      "step": 16724
    },
    {
      "epoch": 0.3935294117647059,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9472,
      "step": 16725
    },
    {
      "epoch": 0.3935529411764706,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8554,
      "step": 16726
    },
    {
      "epoch": 0.3935764705882353,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9294,
      "step": 16727
    },
    {
      "epoch": 0.3936,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7454,
      "step": 16728
    },
    {
      "epoch": 0.3936235294117647,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.2199,
      "step": 16729
    },
    {
      "epoch": 0.3936470588235294,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8766,
      "step": 16730
    },
    {
      "epoch": 0.3936705882352941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0408,
      "step": 16731
    },
    {
      "epoch": 0.3936941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9915,
      "step": 16732
    },
    {
      "epoch": 0.3937176470588235,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9206,
      "step": 16733
    },
    {
      "epoch": 0.39374117647058826,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9435,
      "step": 16734
    },
    {
      "epoch": 0.39376470588235296,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9441,
      "step": 16735
    },
    {
      "epoch": 0.39378823529411766,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 0.8968,
      "step": 16736
    },
    {
      "epoch": 0.39381176470588236,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0101,
      "step": 16737
    },
    {
      "epoch": 0.39383529411764706,
      "grad_norm": 0.22265625,
      "learning_rate": 0.02,
      "loss": 0.6803,
      "step": 16738
    },
    {
      "epoch": 0.39385882352941176,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3639,
      "step": 16739
    },
    {
      "epoch": 0.39388235294117646,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9405,
      "step": 16740
    },
    {
      "epoch": 0.39390588235294116,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8635,
      "step": 16741
    },
    {
      "epoch": 0.39392941176470586,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7259,
      "step": 16742
    },
    {
      "epoch": 0.39395294117647056,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9405,
      "step": 16743
    },
    {
      "epoch": 0.3939764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1932,
      "step": 16744
    },
    {
      "epoch": 0.394,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9656,
      "step": 16745
    },
    {
      "epoch": 0.3940235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8751,
      "step": 16746
    },
    {
      "epoch": 0.3940470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1783,
      "step": 16747
    },
    {
      "epoch": 0.3940705882352941,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0744,
      "step": 16748
    },
    {
      "epoch": 0.3940941176470588,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0903,
      "step": 16749
    },
    {
      "epoch": 0.3941176470588235,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1715,
      "step": 16750
    },
    {
      "epoch": 0.3941411764705882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1476,
      "step": 16751
    },
    {
      "epoch": 0.3941647058823529,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.6656,
      "step": 16752
    },
    {
      "epoch": 0.3941882352941177,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7119,
      "step": 16753
    },
    {
      "epoch": 0.3942117647058824,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1135,
      "step": 16754
    },
    {
      "epoch": 0.3942352941176471,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1022,
      "step": 16755
    },
    {
      "epoch": 0.3942588235294118,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.868,
      "step": 16756
    },
    {
      "epoch": 0.3942823529411765,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1696,
      "step": 16757
    },
    {
      "epoch": 0.3943058823529412,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1084,
      "step": 16758
    },
    {
      "epoch": 0.3943294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.955,
      "step": 16759
    },
    {
      "epoch": 0.3943529411764706,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.02,
      "loss": 1.4611,
      "step": 16760
    },
    {
      "epoch": 0.3943764705882353,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1038,
      "step": 16761
    },
    {
      "epoch": 0.3944,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3647,
      "step": 16762
    },
    {
      "epoch": 0.39442352941176473,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 1.1023,
      "step": 16763
    },
    {
      "epoch": 0.39444705882352943,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 1.0746,
      "step": 16764
    },
    {
      "epoch": 0.39447058823529413,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9292,
      "step": 16765
    },
    {
      "epoch": 0.39449411764705883,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.02,
      "loss": 1.1776,
      "step": 16766
    },
    {
      "epoch": 0.39451764705882353,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0943,
      "step": 16767
    },
    {
      "epoch": 0.39454117647058823,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9235,
      "step": 16768
    },
    {
      "epoch": 0.39456470588235293,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.071,
      "step": 16769
    },
    {
      "epoch": 0.39458823529411763,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.4844,
      "step": 16770
    },
    {
      "epoch": 0.39461176470588233,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0423,
      "step": 16771
    },
    {
      "epoch": 0.3946352941176471,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.098,
      "step": 16772
    },
    {
      "epoch": 0.3946588235294118,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.0868,
      "step": 16773
    },
    {
      "epoch": 0.3946823529411765,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1293,
      "step": 16774
    },
    {
      "epoch": 0.3947058823529412,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6995,
      "step": 16775
    },
    {
      "epoch": 0.3947294117647059,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9458,
      "step": 16776
    },
    {
      "epoch": 0.3947529411764706,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0999,
      "step": 16777
    },
    {
      "epoch": 0.3947764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1976,
      "step": 16778
    },
    {
      "epoch": 0.3948,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0302,
      "step": 16779
    },
    {
      "epoch": 0.3948235294117647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.1042,
      "step": 16780
    },
    {
      "epoch": 0.3948470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.2562,
      "step": 16781
    },
    {
      "epoch": 0.39487058823529414,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.015,
      "step": 16782
    },
    {
      "epoch": 0.39489411764705884,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0887,
      "step": 16783
    },
    {
      "epoch": 0.39491764705882354,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9233,
      "step": 16784
    },
    {
      "epoch": 0.39494117647058824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1612,
      "step": 16785
    },
    {
      "epoch": 0.39496470588235294,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1164,
      "step": 16786
    },
    {
      "epoch": 0.39498823529411764,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9279,
      "step": 16787
    },
    {
      "epoch": 0.39501176470588234,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8926,
      "step": 16788
    },
    {
      "epoch": 0.39503529411764704,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9748,
      "step": 16789
    },
    {
      "epoch": 0.39505882352941174,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9327,
      "step": 16790
    },
    {
      "epoch": 0.3950823529411765,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2552,
      "step": 16791
    },
    {
      "epoch": 0.3951058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2138,
      "step": 16792
    },
    {
      "epoch": 0.3951294117647059,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.302,
      "step": 16793
    },
    {
      "epoch": 0.3951529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0611,
      "step": 16794
    },
    {
      "epoch": 0.3951764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2136,
      "step": 16795
    },
    {
      "epoch": 0.3952,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8991,
      "step": 16796
    },
    {
      "epoch": 0.3952235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7539,
      "step": 16797
    },
    {
      "epoch": 0.3952470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0778,
      "step": 16798
    },
    {
      "epoch": 0.3952705882352941,
      "grad_norm": 0.2109375,
      "learning_rate": 0.02,
      "loss": 0.7775,
      "step": 16799
    },
    {
      "epoch": 0.3952941176470588,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9496,
      "step": 16800
    },
    {
      "epoch": 0.39531764705882355,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.5422,
      "step": 16801
    },
    {
      "epoch": 0.39534117647058825,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.2315,
      "step": 16802
    },
    {
      "epoch": 0.39536470588235295,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8291,
      "step": 16803
    },
    {
      "epoch": 0.39538823529411765,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9561,
      "step": 16804
    },
    {
      "epoch": 0.39541176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.014,
      "step": 16805
    },
    {
      "epoch": 0.39543529411764705,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2596,
      "step": 16806
    },
    {
      "epoch": 0.39545882352941175,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1063,
      "step": 16807
    },
    {
      "epoch": 0.39548235294117645,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9664,
      "step": 16808
    },
    {
      "epoch": 0.39550588235294115,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9774,
      "step": 16809
    },
    {
      "epoch": 0.3955294117647059,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.6421,
      "step": 16810
    },
    {
      "epoch": 0.3955529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0201,
      "step": 16811
    },
    {
      "epoch": 0.3955764705882353,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.5294,
      "step": 16812
    },
    {
      "epoch": 0.3956,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7685,
      "step": 16813
    },
    {
      "epoch": 0.3956235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8767,
      "step": 16814
    },
    {
      "epoch": 0.3956470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9875,
      "step": 16815
    },
    {
      "epoch": 0.3956705882352941,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7414,
      "step": 16816
    },
    {
      "epoch": 0.3956941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.82,
      "step": 16817
    },
    {
      "epoch": 0.3957176470588235,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7797,
      "step": 16818
    },
    {
      "epoch": 0.39574117647058826,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.2012,
      "step": 16819
    },
    {
      "epoch": 0.39576470588235296,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.1966,
      "step": 16820
    },
    {
      "epoch": 0.39578823529411766,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8167,
      "step": 16821
    },
    {
      "epoch": 0.39581176470588236,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.2303,
      "step": 16822
    },
    {
      "epoch": 0.39583529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.05,
      "step": 16823
    },
    {
      "epoch": 0.39585882352941176,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0204,
      "step": 16824
    },
    {
      "epoch": 0.39588235294117646,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1193,
      "step": 16825
    },
    {
      "epoch": 0.39590588235294116,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0955,
      "step": 16826
    },
    {
      "epoch": 0.39592941176470586,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8946,
      "step": 16827
    },
    {
      "epoch": 0.39595294117647056,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.02,
      "loss": 1.2159,
      "step": 16828
    },
    {
      "epoch": 0.3959764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0261,
      "step": 16829
    },
    {
      "epoch": 0.396,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8921,
      "step": 16830
    },
    {
      "epoch": 0.3960235294117647,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9309,
      "step": 16831
    },
    {
      "epoch": 0.3960470588235294,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.677,
      "step": 16832
    },
    {
      "epoch": 0.3960705882352941,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1003,
      "step": 16833
    },
    {
      "epoch": 0.3960941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0115,
      "step": 16834
    },
    {
      "epoch": 0.3961176470588235,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2238,
      "step": 16835
    },
    {
      "epoch": 0.3961411764705882,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0055,
      "step": 16836
    },
    {
      "epoch": 0.3961647058823529,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.9479,
      "step": 16837
    },
    {
      "epoch": 0.3961882352941177,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.4974,
      "step": 16838
    },
    {
      "epoch": 0.3962117647058824,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.5954,
      "step": 16839
    },
    {
      "epoch": 0.3962352941176471,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.7177,
      "step": 16840
    },
    {
      "epoch": 0.3962588235294118,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1869,
      "step": 16841
    },
    {
      "epoch": 0.3962823529411765,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0413,
      "step": 16842
    },
    {
      "epoch": 0.3963058823529412,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.3459,
      "step": 16843
    },
    {
      "epoch": 0.3963294117647059,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0246,
      "step": 16844
    },
    {
      "epoch": 0.3963529411764706,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9716,
      "step": 16845
    },
    {
      "epoch": 0.3963764705882353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.6374,
      "step": 16846
    },
    {
      "epoch": 0.3964,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0904,
      "step": 16847
    },
    {
      "epoch": 0.39642352941176473,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2224,
      "step": 16848
    },
    {
      "epoch": 0.39644705882352943,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9348,
      "step": 16849
    },
    {
      "epoch": 0.39647058823529413,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 16850
    },
    {
      "epoch": 0.39649411764705883,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8684,
      "step": 16851
    },
    {
      "epoch": 0.39651764705882353,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9163,
      "step": 16852
    },
    {
      "epoch": 0.39654117647058823,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8105,
      "step": 16853
    },
    {
      "epoch": 0.39656470588235293,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8241,
      "step": 16854
    },
    {
      "epoch": 0.39658823529411763,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.7869,
      "step": 16855
    },
    {
      "epoch": 0.39661176470588233,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1588,
      "step": 16856
    },
    {
      "epoch": 0.3966352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1928,
      "step": 16857
    },
    {
      "epoch": 0.3966588235294118,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9702,
      "step": 16858
    },
    {
      "epoch": 0.3966823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8629,
      "step": 16859
    },
    {
      "epoch": 0.3967058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0186,
      "step": 16860
    },
    {
      "epoch": 0.3967294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8385,
      "step": 16861
    },
    {
      "epoch": 0.3967529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0891,
      "step": 16862
    },
    {
      "epoch": 0.3967764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.091,
      "step": 16863
    },
    {
      "epoch": 0.3968,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0762,
      "step": 16864
    },
    {
      "epoch": 0.3968235294117647,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9347,
      "step": 16865
    },
    {
      "epoch": 0.3968470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7403,
      "step": 16866
    },
    {
      "epoch": 0.39687058823529414,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.3184,
      "step": 16867
    },
    {
      "epoch": 0.39689411764705884,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1344,
      "step": 16868
    },
    {
      "epoch": 0.39691764705882354,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1133,
      "step": 16869
    },
    {
      "epoch": 0.39694117647058824,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.5864,
      "step": 16870
    },
    {
      "epoch": 0.39696470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1511,
      "step": 16871
    },
    {
      "epoch": 0.39698823529411764,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0113,
      "step": 16872
    },
    {
      "epoch": 0.39701176470588234,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9237,
      "step": 16873
    },
    {
      "epoch": 0.39703529411764704,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.7704,
      "step": 16874
    },
    {
      "epoch": 0.39705882352941174,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9191,
      "step": 16875
    },
    {
      "epoch": 0.3970823529411765,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0857,
      "step": 16876
    },
    {
      "epoch": 0.3971058823529412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9927,
      "step": 16877
    },
    {
      "epoch": 0.3971294117647059,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0796,
      "step": 16878
    },
    {
      "epoch": 0.3971529411764706,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7357,
      "step": 16879
    },
    {
      "epoch": 0.3971764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.039,
      "step": 16880
    },
    {
      "epoch": 0.3972,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.0116,
      "step": 16881
    },
    {
      "epoch": 0.3972235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9867,
      "step": 16882
    },
    {
      "epoch": 0.3972470588235294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.978,
      "step": 16883
    },
    {
      "epoch": 0.3972705882352941,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.013,
      "step": 16884
    },
    {
      "epoch": 0.3972941176470588,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.94,
      "step": 16885
    },
    {
      "epoch": 0.39731764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9594,
      "step": 16886
    },
    {
      "epoch": 0.39734117647058825,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9086,
      "step": 16887
    },
    {
      "epoch": 0.39736470588235295,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7454,
      "step": 16888
    },
    {
      "epoch": 0.39738823529411765,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 0.9628,
      "step": 16889
    },
    {
      "epoch": 0.39741176470588235,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.6127,
      "step": 16890
    },
    {
      "epoch": 0.39743529411764705,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8859,
      "step": 16891
    },
    {
      "epoch": 0.39745882352941175,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9278,
      "step": 16892
    },
    {
      "epoch": 0.39748235294117645,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.3749,
      "step": 16893
    },
    {
      "epoch": 0.39750588235294115,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0885,
      "step": 16894
    },
    {
      "epoch": 0.3975294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8703,
      "step": 16895
    },
    {
      "epoch": 0.3975529411764706,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1552,
      "step": 16896
    },
    {
      "epoch": 0.3975764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.089,
      "step": 16897
    },
    {
      "epoch": 0.3976,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7545,
      "step": 16898
    },
    {
      "epoch": 0.3976235294117647,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 16899
    },
    {
      "epoch": 0.3976470588235294,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8555,
      "step": 16900
    },
    {
      "epoch": 0.3976705882352941,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.02,
      "loss": 0.9321,
      "step": 16901
    },
    {
      "epoch": 0.3976941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.5453,
      "step": 16902
    },
    {
      "epoch": 0.3977176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.8213,
      "step": 16903
    },
    {
      "epoch": 0.3977411764705882,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0776,
      "step": 16904
    },
    {
      "epoch": 0.39776470588235296,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9464,
      "step": 16905
    },
    {
      "epoch": 0.39778823529411766,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8502,
      "step": 16906
    },
    {
      "epoch": 0.39781176470588236,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9234,
      "step": 16907
    },
    {
      "epoch": 0.39783529411764706,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2406,
      "step": 16908
    },
    {
      "epoch": 0.39785882352941176,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8349,
      "step": 16909
    },
    {
      "epoch": 0.39788235294117646,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9417,
      "step": 16910
    },
    {
      "epoch": 0.39790588235294116,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.1352,
      "step": 16911
    },
    {
      "epoch": 0.39792941176470586,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0122,
      "step": 16912
    },
    {
      "epoch": 0.39795294117647056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.5282,
      "step": 16913
    },
    {
      "epoch": 0.3979764705882353,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.04,
      "step": 16914
    },
    {
      "epoch": 0.398,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.9014,
      "step": 16915
    },
    {
      "epoch": 0.3980235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0466,
      "step": 16916
    },
    {
      "epoch": 0.3980470588235294,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.3018,
      "step": 16917
    },
    {
      "epoch": 0.3980705882352941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0412,
      "step": 16918
    },
    {
      "epoch": 0.3980941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9696,
      "step": 16919
    },
    {
      "epoch": 0.3981176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0017,
      "step": 16920
    },
    {
      "epoch": 0.3981411764705882,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9038,
      "step": 16921
    },
    {
      "epoch": 0.3981647058823529,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0813,
      "step": 16922
    },
    {
      "epoch": 0.3981882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.642,
      "step": 16923
    },
    {
      "epoch": 0.3982117647058824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0506,
      "step": 16924
    },
    {
      "epoch": 0.3982352941176471,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2286,
      "step": 16925
    },
    {
      "epoch": 0.3982588235294118,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2895,
      "step": 16926
    },
    {
      "epoch": 0.3982823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9942,
      "step": 16927
    },
    {
      "epoch": 0.3983058823529412,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9733,
      "step": 16928
    },
    {
      "epoch": 0.3983294117647059,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.7704,
      "step": 16929
    },
    {
      "epoch": 0.3983529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1057,
      "step": 16930
    },
    {
      "epoch": 0.3983764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9586,
      "step": 16931
    },
    {
      "epoch": 0.3984,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.653,
      "step": 16932
    },
    {
      "epoch": 0.39842352941176473,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.0739,
      "step": 16933
    },
    {
      "epoch": 0.39844705882352943,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0122,
      "step": 16934
    },
    {
      "epoch": 0.39847058823529413,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1603,
      "step": 16935
    },
    {
      "epoch": 0.39849411764705883,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8871,
      "step": 16936
    },
    {
      "epoch": 0.39851764705882353,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0518,
      "step": 16937
    },
    {
      "epoch": 0.39854117647058823,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8928,
      "step": 16938
    },
    {
      "epoch": 0.39856470588235293,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0635,
      "step": 16939
    },
    {
      "epoch": 0.39858823529411763,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0414,
      "step": 16940
    },
    {
      "epoch": 0.39861176470588233,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8889,
      "step": 16941
    },
    {
      "epoch": 0.39863529411764703,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.954,
      "step": 16942
    },
    {
      "epoch": 0.3986588235294118,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1089,
      "step": 16943
    },
    {
      "epoch": 0.3986823529411765,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.746,
      "step": 16944
    },
    {
      "epoch": 0.3987058823529412,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1447,
      "step": 16945
    },
    {
      "epoch": 0.3987294117647059,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.1678,
      "step": 16946
    },
    {
      "epoch": 0.3987529411764706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9376,
      "step": 16947
    },
    {
      "epoch": 0.3987764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.713,
      "step": 16948
    },
    {
      "epoch": 0.3988,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.671,
      "step": 16949
    },
    {
      "epoch": 0.3988235294117647,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7236,
      "step": 16950
    },
    {
      "epoch": 0.3988470588235294,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.159,
      "step": 16951
    },
    {
      "epoch": 0.39887058823529414,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.7944,
      "step": 16952
    },
    {
      "epoch": 0.39889411764705884,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.065,
      "step": 16953
    },
    {
      "epoch": 0.39891764705882354,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7546,
      "step": 16954
    },
    {
      "epoch": 0.39894117647058824,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1727,
      "step": 16955
    },
    {
      "epoch": 0.39896470588235294,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9706,
      "step": 16956
    },
    {
      "epoch": 0.39898823529411764,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.8956,
      "step": 16957
    },
    {
      "epoch": 0.39901176470588234,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.78,
      "step": 16958
    },
    {
      "epoch": 0.39903529411764704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2203,
      "step": 16959
    },
    {
      "epoch": 0.39905882352941174,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0078,
      "step": 16960
    },
    {
      "epoch": 0.39908235294117644,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.0711,
      "step": 16961
    },
    {
      "epoch": 0.3991058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8457,
      "step": 16962
    },
    {
      "epoch": 0.3991294117647059,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 1.0285,
      "step": 16963
    },
    {
      "epoch": 0.3991529411764706,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8361,
      "step": 16964
    },
    {
      "epoch": 0.3991764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9356,
      "step": 16965
    },
    {
      "epoch": 0.3992,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.0173,
      "step": 16966
    },
    {
      "epoch": 0.3992235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.6848,
      "step": 16967
    },
    {
      "epoch": 0.3992470588235294,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.6542,
      "step": 16968
    },
    {
      "epoch": 0.3992705882352941,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8394,
      "step": 16969
    },
    {
      "epoch": 0.3992941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.153,
      "step": 16970
    },
    {
      "epoch": 0.39931764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 16971
    },
    {
      "epoch": 0.39934117647058825,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.0377,
      "step": 16972
    },
    {
      "epoch": 0.39936470588235295,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9592,
      "step": 16973
    },
    {
      "epoch": 0.39938823529411766,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9292,
      "step": 16974
    },
    {
      "epoch": 0.39941176470588236,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9771,
      "step": 16975
    },
    {
      "epoch": 0.39943529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9662,
      "step": 16976
    },
    {
      "epoch": 0.39945882352941176,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1901,
      "step": 16977
    },
    {
      "epoch": 0.39948235294117646,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0199,
      "step": 16978
    },
    {
      "epoch": 0.39950588235294116,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.8206,
      "step": 16979
    },
    {
      "epoch": 0.39952941176470586,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8663,
      "step": 16980
    },
    {
      "epoch": 0.3995529411764706,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.3564,
      "step": 16981
    },
    {
      "epoch": 0.3995764705882353,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1081,
      "step": 16982
    },
    {
      "epoch": 0.3996,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.6861,
      "step": 16983
    },
    {
      "epoch": 0.3996235294117647,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9104,
      "step": 16984
    },
    {
      "epoch": 0.3996470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.6132,
      "step": 16985
    },
    {
      "epoch": 0.3996705882352941,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.8048,
      "step": 16986
    },
    {
      "epoch": 0.3996941176470588,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9868,
      "step": 16987
    },
    {
      "epoch": 0.3997176470588235,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8991,
      "step": 16988
    },
    {
      "epoch": 0.3997411764705882,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9899,
      "step": 16989
    },
    {
      "epoch": 0.39976470588235297,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.5737,
      "step": 16990
    },
    {
      "epoch": 0.39978823529411767,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9269,
      "step": 16991
    },
    {
      "epoch": 0.39981176470588237,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 16992
    },
    {
      "epoch": 0.39983529411764707,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9804,
      "step": 16993
    },
    {
      "epoch": 0.39985882352941177,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0818,
      "step": 16994
    },
    {
      "epoch": 0.39988235294117647,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0744,
      "step": 16995
    },
    {
      "epoch": 0.39990588235294117,
      "grad_norm": 0.21875,
      "learning_rate": 0.02,
      "loss": 0.5892,
      "step": 16996
    },
    {
      "epoch": 0.39992941176470587,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.2538,
      "step": 16997
    },
    {
      "epoch": 0.39995294117647057,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 16998
    },
    {
      "epoch": 0.39997647058823527,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7813,
      "step": 16999
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.0842,
      "step": 17000
    },
    {
      "epoch": 0.4,
      "eval_loss": 2.186739206314087,
      "eval_runtime": 685.2807,
      "eval_samples_per_second": 12.404,
      "eval_steps_per_second": 3.101,
      "step": 17000
    },
    {
      "epoch": 0.4000235294117647,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9425,
      "step": 17001
    },
    {
      "epoch": 0.4000470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1127,
      "step": 17002
    },
    {
      "epoch": 0.4000705882352941,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.881,
      "step": 17003
    },
    {
      "epoch": 0.4000941176470588,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0563,
      "step": 17004
    },
    {
      "epoch": 0.4001176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9175,
      "step": 17005
    },
    {
      "epoch": 0.4001411764705882,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0026,
      "step": 17006
    },
    {
      "epoch": 0.4001647058823529,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0117,
      "step": 17007
    },
    {
      "epoch": 0.4001882352941176,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1724,
      "step": 17008
    },
    {
      "epoch": 0.4002117647058824,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7454,
      "step": 17009
    },
    {
      "epoch": 0.4002352941176471,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1664,
      "step": 17010
    },
    {
      "epoch": 0.4002588235294118,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0794,
      "step": 17011
    },
    {
      "epoch": 0.4002823529411765,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8012,
      "step": 17012
    },
    {
      "epoch": 0.4003058823529412,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7609,
      "step": 17013
    },
    {
      "epoch": 0.4003294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.9115,
      "step": 17014
    },
    {
      "epoch": 0.4003529411764706,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 1.1214,
      "step": 17015
    },
    {
      "epoch": 0.4003764705882353,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.3172,
      "step": 17016
    },
    {
      "epoch": 0.4004,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.1545,
      "step": 17017
    },
    {
      "epoch": 0.4004235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8438,
      "step": 17018
    },
    {
      "epoch": 0.40044705882352943,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 0.9248,
      "step": 17019
    },
    {
      "epoch": 0.40047058823529413,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0116,
      "step": 17020
    },
    {
      "epoch": 0.40049411764705883,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.2066,
      "step": 17021
    },
    {
      "epoch": 0.40051764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1136,
      "step": 17022
    },
    {
      "epoch": 0.40054117647058823,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.8762,
      "step": 17023
    },
    {
      "epoch": 0.40056470588235293,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0464,
      "step": 17024
    },
    {
      "epoch": 0.40058823529411763,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0352,
      "step": 17025
    },
    {
      "epoch": 0.40061176470588233,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.3244,
      "step": 17026
    },
    {
      "epoch": 0.40063529411764703,
      "grad_norm": 0.14453125,
      "learning_rate": 0.02,
      "loss": 1.3058,
      "step": 17027
    },
    {
      "epoch": 0.4006588235294118,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9756,
      "step": 17028
    },
    {
      "epoch": 0.4006823529411765,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0493,
      "step": 17029
    },
    {
      "epoch": 0.4007058823529412,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1088,
      "step": 17030
    },
    {
      "epoch": 0.4007294117647059,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7538,
      "step": 17031
    },
    {
      "epoch": 0.4007529411764706,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.7748,
      "step": 17032
    },
    {
      "epoch": 0.4007764705882353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.8317,
      "step": 17033
    },
    {
      "epoch": 0.4008,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2267,
      "step": 17034
    },
    {
      "epoch": 0.4008235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2295,
      "step": 17035
    },
    {
      "epoch": 0.4008470588235294,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9492,
      "step": 17036
    },
    {
      "epoch": 0.40087058823529415,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.0189,
      "step": 17037
    },
    {
      "epoch": 0.40089411764705885,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.6118,
      "step": 17038
    },
    {
      "epoch": 0.40091764705882355,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1571,
      "step": 17039
    },
    {
      "epoch": 0.40094117647058825,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.7591,
      "step": 17040
    },
    {
      "epoch": 0.40096470588235295,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1753,
      "step": 17041
    },
    {
      "epoch": 0.40098823529411765,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.211,
      "step": 17042
    },
    {
      "epoch": 0.40101176470588235,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2165,
      "step": 17043
    },
    {
      "epoch": 0.40103529411764705,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.9459,
      "step": 17044
    },
    {
      "epoch": 0.40105882352941175,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.1515,
      "step": 17045
    },
    {
      "epoch": 0.40108235294117645,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.3869,
      "step": 17046
    },
    {
      "epoch": 0.4011058823529412,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.84,
      "step": 17047
    },
    {
      "epoch": 0.4011294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8394,
      "step": 17048
    },
    {
      "epoch": 0.4011529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0313,
      "step": 17049
    },
    {
      "epoch": 0.4011764705882353,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2015,
      "step": 17050
    },
    {
      "epoch": 0.4012,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.7755,
      "step": 17051
    },
    {
      "epoch": 0.4012235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.272,
      "step": 17052
    },
    {
      "epoch": 0.4012470588235294,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.8693,
      "step": 17053
    },
    {
      "epoch": 0.4012705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.8343,
      "step": 17054
    },
    {
      "epoch": 0.4012941176470588,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1274,
      "step": 17055
    },
    {
      "epoch": 0.40131764705882356,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.162,
      "step": 17056
    },
    {
      "epoch": 0.40134117647058826,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9518,
      "step": 17057
    },
    {
      "epoch": 0.40136470588235296,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.4068,
      "step": 17058
    },
    {
      "epoch": 0.40138823529411766,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.7975,
      "step": 17059
    },
    {
      "epoch": 0.40141176470588236,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.7518,
      "step": 17060
    },
    {
      "epoch": 0.40143529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.1391,
      "step": 17061
    },
    {
      "epoch": 0.40145882352941176,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1011,
      "step": 17062
    },
    {
      "epoch": 0.40148235294117646,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9322,
      "step": 17063
    },
    {
      "epoch": 0.40150588235294116,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1685,
      "step": 17064
    },
    {
      "epoch": 0.40152941176470586,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.0015,
      "step": 17065
    },
    {
      "epoch": 0.4015529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1257,
      "step": 17066
    },
    {
      "epoch": 0.4015764705882353,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.0115,
      "step": 17067
    },
    {
      "epoch": 0.4016,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0303,
      "step": 17068
    },
    {
      "epoch": 0.4016235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7906,
      "step": 17069
    },
    {
      "epoch": 0.4016470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.9961,
      "step": 17070
    },
    {
      "epoch": 0.4016705882352941,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3695,
      "step": 17071
    },
    {
      "epoch": 0.4016941176470588,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0638,
      "step": 17072
    },
    {
      "epoch": 0.4017176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0031,
      "step": 17073
    },
    {
      "epoch": 0.4017411764705882,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.02,
      "loss": 1.3632,
      "step": 17074
    },
    {
      "epoch": 0.40176470588235297,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.1893,
      "step": 17075
    },
    {
      "epoch": 0.40178823529411767,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.8428,
      "step": 17076
    },
    {
      "epoch": 0.40181176470588237,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1203,
      "step": 17077
    },
    {
      "epoch": 0.40183529411764707,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.3239,
      "step": 17078
    },
    {
      "epoch": 0.40185882352941177,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0866,
      "step": 17079
    },
    {
      "epoch": 0.40188235294117647,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0628,
      "step": 17080
    },
    {
      "epoch": 0.40190588235294117,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0908,
      "step": 17081
    },
    {
      "epoch": 0.40192941176470587,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9108,
      "step": 17082
    },
    {
      "epoch": 0.40195294117647057,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.5492,
      "step": 17083
    },
    {
      "epoch": 0.40197647058823527,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0518,
      "step": 17084
    },
    {
      "epoch": 0.402,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1543,
      "step": 17085
    },
    {
      "epoch": 0.4020235294117647,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0257,
      "step": 17086
    },
    {
      "epoch": 0.4020470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0821,
      "step": 17087
    },
    {
      "epoch": 0.4020705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 1.0106,
      "step": 17088
    },
    {
      "epoch": 0.4020941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1693,
      "step": 17089
    },
    {
      "epoch": 0.4021176470588235,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8198,
      "step": 17090
    },
    {
      "epoch": 0.4021411764705882,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8855,
      "step": 17091
    },
    {
      "epoch": 0.4021647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1338,
      "step": 17092
    },
    {
      "epoch": 0.4021882352941176,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8018,
      "step": 17093
    },
    {
      "epoch": 0.4022117647058824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0909,
      "step": 17094
    },
    {
      "epoch": 0.4022352941176471,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1388,
      "step": 17095
    },
    {
      "epoch": 0.4022588235294118,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8887,
      "step": 17096
    },
    {
      "epoch": 0.4022823529411765,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1531,
      "step": 17097
    },
    {
      "epoch": 0.4023058823529412,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.8436,
      "step": 17098
    },
    {
      "epoch": 0.4023294117647059,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1123,
      "step": 17099
    },
    {
      "epoch": 0.4023529411764706,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2892,
      "step": 17100
    },
    {
      "epoch": 0.4023764705882353,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1654,
      "step": 17101
    },
    {
      "epoch": 0.4024,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 1.1655,
      "step": 17102
    },
    {
      "epoch": 0.4024235294117647,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.1253,
      "step": 17103
    },
    {
      "epoch": 0.40244705882352944,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9301,
      "step": 17104
    },
    {
      "epoch": 0.40247058823529414,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.251,
      "step": 17105
    },
    {
      "epoch": 0.40249411764705884,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1,
      "step": 17106
    },
    {
      "epoch": 0.40251764705882354,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.2532,
      "step": 17107
    },
    {
      "epoch": 0.40254117647058824,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9884,
      "step": 17108
    },
    {
      "epoch": 0.40256470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9268,
      "step": 17109
    },
    {
      "epoch": 0.40258823529411764,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7137,
      "step": 17110
    },
    {
      "epoch": 0.40261176470588234,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7353,
      "step": 17111
    },
    {
      "epoch": 0.40263529411764704,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.8077,
      "step": 17112
    },
    {
      "epoch": 0.4026588235294118,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8965,
      "step": 17113
    },
    {
      "epoch": 0.4026823529411765,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.0482,
      "step": 17114
    },
    {
      "epoch": 0.4027058823529412,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8993,
      "step": 17115
    },
    {
      "epoch": 0.4027294117647059,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.8083,
      "step": 17116
    },
    {
      "epoch": 0.4027529411764706,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.3182,
      "step": 17117
    },
    {
      "epoch": 0.4027764705882353,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1977,
      "step": 17118
    },
    {
      "epoch": 0.4028,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.02,
      "loss": 0.6051,
      "step": 17119
    },
    {
      "epoch": 0.4028235294117647,
      "grad_norm": 0.140625,
      "learning_rate": 0.02,
      "loss": 1.3925,
      "step": 17120
    },
    {
      "epoch": 0.4028470588235294,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9893,
      "step": 17121
    },
    {
      "epoch": 0.4028705882352941,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7159,
      "step": 17122
    },
    {
      "epoch": 0.40289411764705885,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3556,
      "step": 17123
    },
    {
      "epoch": 0.40291764705882355,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.8188,
      "step": 17124
    },
    {
      "epoch": 0.40294117647058825,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0958,
      "step": 17125
    },
    {
      "epoch": 0.40296470588235295,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0515,
      "step": 17126
    },
    {
      "epoch": 0.40298823529411765,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.8715,
      "step": 17127
    },
    {
      "epoch": 0.40301176470588235,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.8511,
      "step": 17128
    },
    {
      "epoch": 0.40303529411764705,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8132,
      "step": 17129
    },
    {
      "epoch": 0.40305882352941175,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2194,
      "step": 17130
    },
    {
      "epoch": 0.40308235294117645,
      "grad_norm": 0.216796875,
      "learning_rate": 0.02,
      "loss": 0.6812,
      "step": 17131
    },
    {
      "epoch": 0.4031058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0492,
      "step": 17132
    },
    {
      "epoch": 0.4031294117647059,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9542,
      "step": 17133
    },
    {
      "epoch": 0.4031529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8249,
      "step": 17134
    },
    {
      "epoch": 0.4031764705882353,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1489,
      "step": 17135
    },
    {
      "epoch": 0.4032,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9845,
      "step": 17136
    },
    {
      "epoch": 0.4032235294117647,
      "grad_norm": 0.14453125,
      "learning_rate": 0.02,
      "loss": 1.3071,
      "step": 17137
    },
    {
      "epoch": 0.4032470588235294,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0848,
      "step": 17138
    },
    {
      "epoch": 0.4032705882352941,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.081,
      "step": 17139
    },
    {
      "epoch": 0.4032941176470588,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.6974,
      "step": 17140
    },
    {
      "epoch": 0.4033176470588235,
      "grad_norm": 0.142578125,
      "learning_rate": 0.02,
      "loss": 1.3637,
      "step": 17141
    },
    {
      "epoch": 0.40334117647058826,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9164,
      "step": 17142
    },
    {
      "epoch": 0.40336470588235296,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2162,
      "step": 17143
    },
    {
      "epoch": 0.40338823529411766,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.994,
      "step": 17144
    },
    {
      "epoch": 0.40341176470588236,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0412,
      "step": 17145
    },
    {
      "epoch": 0.40343529411764706,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.2081,
      "step": 17146
    },
    {
      "epoch": 0.40345882352941176,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1625,
      "step": 17147
    },
    {
      "epoch": 0.40348235294117646,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7394,
      "step": 17148
    },
    {
      "epoch": 0.40350588235294116,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9139,
      "step": 17149
    },
    {
      "epoch": 0.40352941176470586,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9313,
      "step": 17150
    },
    {
      "epoch": 0.4035529411764706,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0672,
      "step": 17151
    },
    {
      "epoch": 0.4035764705882353,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.3656,
      "step": 17152
    },
    {
      "epoch": 0.4036,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1254,
      "step": 17153
    },
    {
      "epoch": 0.4036235294117647,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1152,
      "step": 17154
    },
    {
      "epoch": 0.4036470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9297,
      "step": 17155
    },
    {
      "epoch": 0.4036705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9743,
      "step": 17156
    },
    {
      "epoch": 0.4036941176470588,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.204,
      "step": 17157
    },
    {
      "epoch": 0.4037176470588235,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.7067,
      "step": 17158
    },
    {
      "epoch": 0.4037411764705882,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.054,
      "step": 17159
    },
    {
      "epoch": 0.4037647058823529,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.8672,
      "step": 17160
    },
    {
      "epoch": 0.40378823529411767,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.8327,
      "step": 17161
    },
    {
      "epoch": 0.40381176470588237,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9894,
      "step": 17162
    },
    {
      "epoch": 0.40383529411764707,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2237,
      "step": 17163
    },
    {
      "epoch": 0.40385882352941177,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.1976,
      "step": 17164
    },
    {
      "epoch": 0.40388235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.162,
      "step": 17165
    },
    {
      "epoch": 0.40390588235294117,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9842,
      "step": 17166
    },
    {
      "epoch": 0.40392941176470587,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.1607,
      "step": 17167
    },
    {
      "epoch": 0.40395294117647057,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9762,
      "step": 17168
    },
    {
      "epoch": 0.40397647058823527,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0473,
      "step": 17169
    },
    {
      "epoch": 0.404,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0917,
      "step": 17170
    },
    {
      "epoch": 0.4040235294117647,
      "grad_norm": 0.19140625,
      "learning_rate": 0.02,
      "loss": 0.7753,
      "step": 17171
    },
    {
      "epoch": 0.4040470588235294,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9192,
      "step": 17172
    },
    {
      "epoch": 0.4040705882352941,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.1711,
      "step": 17173
    },
    {
      "epoch": 0.4040941176470588,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9857,
      "step": 17174
    },
    {
      "epoch": 0.4041176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9291,
      "step": 17175
    },
    {
      "epoch": 0.4041411764705882,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1163,
      "step": 17176
    },
    {
      "epoch": 0.4041647058823529,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.947,
      "step": 17177
    },
    {
      "epoch": 0.4041882352941176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 1.072,
      "step": 17178
    },
    {
      "epoch": 0.4042117647058823,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.3276,
      "step": 17179
    },
    {
      "epoch": 0.4042352941176471,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 1.1129,
      "step": 17180
    },
    {
      "epoch": 0.4042588235294118,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0923,
      "step": 17181
    },
    {
      "epoch": 0.4042823529411765,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8461,
      "step": 17182
    },
    {
      "epoch": 0.4043058823529412,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8515,
      "step": 17183
    },
    {
      "epoch": 0.4043294117647059,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7813,
      "step": 17184
    },
    {
      "epoch": 0.4043529411764706,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0486,
      "step": 17185
    },
    {
      "epoch": 0.4043764705882353,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0588,
      "step": 17186
    },
    {
      "epoch": 0.4044,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.902,
      "step": 17187
    },
    {
      "epoch": 0.4044235294117647,
      "grad_norm": 0.201171875,
      "learning_rate": 0.02,
      "loss": 0.7027,
      "step": 17188
    },
    {
      "epoch": 0.40444705882352944,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.2354,
      "step": 17189
    },
    {
      "epoch": 0.40447058823529414,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9165,
      "step": 17190
    },
    {
      "epoch": 0.40449411764705884,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9932,
      "step": 17191
    },
    {
      "epoch": 0.40451764705882354,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.2293,
      "step": 17192
    },
    {
      "epoch": 0.40454117647058824,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9673,
      "step": 17193
    },
    {
      "epoch": 0.40456470588235294,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8289,
      "step": 17194
    },
    {
      "epoch": 0.40458823529411764,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0277,
      "step": 17195
    },
    {
      "epoch": 0.40461176470588234,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8185,
      "step": 17196
    },
    {
      "epoch": 0.40463529411764704,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 1.142,
      "step": 17197
    },
    {
      "epoch": 0.40465882352941174,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.272,
      "step": 17198
    },
    {
      "epoch": 0.4046823529411765,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.107,
      "step": 17199
    },
    {
      "epoch": 0.4047058823529412,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.7105,
      "step": 17200
    },
    {
      "epoch": 0.4047294117647059,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.9502,
      "step": 17201
    },
    {
      "epoch": 0.4047529411764706,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.1318,
      "step": 17202
    },
    {
      "epoch": 0.4047764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9866,
      "step": 17203
    },
    {
      "epoch": 0.4048,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.1249,
      "step": 17204
    },
    {
      "epoch": 0.4048235294117647,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.02,
      "loss": 0.9172,
      "step": 17205
    },
    {
      "epoch": 0.4048470588235294,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.171,
      "step": 17206
    },
    {
      "epoch": 0.4048705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.0481,
      "step": 17207
    },
    {
      "epoch": 0.40489411764705885,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1763,
      "step": 17208
    },
    {
      "epoch": 0.40491764705882355,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7625,
      "step": 17209
    },
    {
      "epoch": 0.40494117647058825,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.978,
      "step": 17210
    },
    {
      "epoch": 0.40496470588235295,
      "grad_norm": 0.1953125,
      "learning_rate": 0.02,
      "loss": 0.8986,
      "step": 17211
    },
    {
      "epoch": 0.40498823529411765,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.3412,
      "step": 17212
    },
    {
      "epoch": 0.40501176470588235,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9691,
      "step": 17213
    },
    {
      "epoch": 0.40503529411764705,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.02,
      "loss": 1.1834,
      "step": 17214
    },
    {
      "epoch": 0.40505882352941175,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9825,
      "step": 17215
    },
    {
      "epoch": 0.40508235294117645,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.2318,
      "step": 17216
    },
    {
      "epoch": 0.40510588235294115,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.9551,
      "step": 17217
    },
    {
      "epoch": 0.4051294117647059,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1259,
      "step": 17218
    },
    {
      "epoch": 0.4051529411764706,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.02,
      "loss": 0.5993,
      "step": 17219
    },
    {
      "epoch": 0.4051764705882353,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8359,
      "step": 17220
    },
    {
      "epoch": 0.4052,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.976,
      "step": 17221
    },
    {
      "epoch": 0.4052235294117647,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.1752,
      "step": 17222
    },
    {
      "epoch": 0.4052470588235294,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7416,
      "step": 17223
    },
    {
      "epoch": 0.4052705882352941,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.1178,
      "step": 17224
    },
    {
      "epoch": 0.4052941176470588,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.3074,
      "step": 17225
    },
    {
      "epoch": 0.4053176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9688,
      "step": 17226
    },
    {
      "epoch": 0.40534117647058826,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0882,
      "step": 17227
    },
    {
      "epoch": 0.40536470588235296,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.8716,
      "step": 17228
    },
    {
      "epoch": 0.40538823529411766,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9068,
      "step": 17229
    },
    {
      "epoch": 0.40541176470588236,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0332,
      "step": 17230
    },
    {
      "epoch": 0.40543529411764706,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.6996,
      "step": 17231
    },
    {
      "epoch": 0.40545882352941176,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3087,
      "step": 17232
    },
    {
      "epoch": 0.40548235294117646,
      "grad_norm": 0.203125,
      "learning_rate": 0.02,
      "loss": 0.6656,
      "step": 17233
    },
    {
      "epoch": 0.40550588235294116,
      "grad_norm": 0.15234375,
      "learning_rate": 0.02,
      "loss": 1.2658,
      "step": 17234
    },
    {
      "epoch": 0.40552941176470586,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9038,
      "step": 17235
    },
    {
      "epoch": 0.40555294117647056,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9417,
      "step": 17236
    },
    {
      "epoch": 0.4055764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0006,
      "step": 17237
    },
    {
      "epoch": 0.4056,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.695,
      "step": 17238
    },
    {
      "epoch": 0.4056235294117647,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.7048,
      "step": 17239
    },
    {
      "epoch": 0.4056470588235294,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.2478,
      "step": 17240
    },
    {
      "epoch": 0.4056705882352941,
      "grad_norm": 0.208984375,
      "learning_rate": 0.02,
      "loss": 0.7238,
      "step": 17241
    },
    {
      "epoch": 0.4056941176470588,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8606,
      "step": 17242
    },
    {
      "epoch": 0.4057176470588235,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.1693,
      "step": 17243
    },
    {
      "epoch": 0.4057411764705882,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.154,
      "step": 17244
    },
    {
      "epoch": 0.4057647058823529,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.018,
      "step": 17245
    },
    {
      "epoch": 0.40578823529411767,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.7783,
      "step": 17246
    },
    {
      "epoch": 0.40581176470588237,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.3528,
      "step": 17247
    },
    {
      "epoch": 0.40583529411764707,
      "grad_norm": 0.205078125,
      "learning_rate": 0.02,
      "loss": 0.5329,
      "step": 17248
    },
    {
      "epoch": 0.40585882352941177,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0538,
      "step": 17249
    },
    {
      "epoch": 0.40588235294117647,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.1143,
      "step": 17250
    },
    {
      "epoch": 0.40590588235294117,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9592,
      "step": 17251
    },
    {
      "epoch": 0.40592941176470587,
      "grad_norm": 0.23046875,
      "learning_rate": 0.02,
      "loss": 0.5554,
      "step": 17252
    },
    {
      "epoch": 0.40595294117647057,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0203,
      "step": 17253
    },
    {
      "epoch": 0.40597647058823527,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1132,
      "step": 17254
    },
    {
      "epoch": 0.406,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.02,
      "loss": 0.6278,
      "step": 17255
    },
    {
      "epoch": 0.4060235294117647,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9721,
      "step": 17256
    },
    {
      "epoch": 0.4060470588235294,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9506,
      "step": 17257
    },
    {
      "epoch": 0.4060705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8321,
      "step": 17258
    },
    {
      "epoch": 0.4060941176470588,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.8733,
      "step": 17259
    },
    {
      "epoch": 0.4061176470588235,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.896,
      "step": 17260
    },
    {
      "epoch": 0.4061411764705882,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9384,
      "step": 17261
    },
    {
      "epoch": 0.4061647058823529,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0154,
      "step": 17262
    },
    {
      "epoch": 0.4061882352941176,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8331,
      "step": 17263
    },
    {
      "epoch": 0.40621176470588233,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.0836,
      "step": 17264
    },
    {
      "epoch": 0.4062352941176471,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 0.9461,
      "step": 17265
    },
    {
      "epoch": 0.4062588235294118,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8016,
      "step": 17266
    },
    {
      "epoch": 0.4062823529411765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.108,
      "step": 17267
    },
    {
      "epoch": 0.4063058823529412,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.7708,
      "step": 17268
    },
    {
      "epoch": 0.4063294117647059,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.9839,
      "step": 17269
    },
    {
      "epoch": 0.4063529411764706,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.5937,
      "step": 17270
    },
    {
      "epoch": 0.4063764705882353,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.8992,
      "step": 17271
    },
    {
      "epoch": 0.4064,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.7944,
      "step": 17272
    },
    {
      "epoch": 0.4064235294117647,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.0185,
      "step": 17273
    },
    {
      "epoch": 0.40644705882352944,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1351,
      "step": 17274
    },
    {
      "epoch": 0.40647058823529414,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.7251,
      "step": 17275
    },
    {
      "epoch": 0.40649411764705884,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.9398,
      "step": 17276
    },
    {
      "epoch": 0.40651764705882354,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.2471,
      "step": 17277
    },
    {
      "epoch": 0.40654117647058824,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 0.8147,
      "step": 17278
    },
    {
      "epoch": 0.40656470588235294,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.3041,
      "step": 17279
    },
    {
      "epoch": 0.40658823529411764,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 0.9734,
      "step": 17280
    },
    {
      "epoch": 0.40661176470588234,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8927,
      "step": 17281
    },
    {
      "epoch": 0.40663529411764704,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0506,
      "step": 17282
    },
    {
      "epoch": 0.40665882352941174,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8944,
      "step": 17283
    },
    {
      "epoch": 0.4066823529411765,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 1.0478,
      "step": 17284
    },
    {
      "epoch": 0.4067058823529412,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0063,
      "step": 17285
    },
    {
      "epoch": 0.4067294117647059,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9928,
      "step": 17286
    },
    {
      "epoch": 0.4067529411764706,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0183,
      "step": 17287
    },
    {
      "epoch": 0.4067764705882353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 0.9407,
      "step": 17288
    },
    {
      "epoch": 0.4068,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1332,
      "step": 17289
    },
    {
      "epoch": 0.4068235294117647,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.198,
      "step": 17290
    },
    {
      "epoch": 0.4068470588235294,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.1922,
      "step": 17291
    },
    {
      "epoch": 0.4068705882352941,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.2188,
      "step": 17292
    },
    {
      "epoch": 0.40689411764705885,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.9074,
      "step": 17293
    },
    {
      "epoch": 0.40691764705882355,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.278,
      "step": 17294
    },
    {
      "epoch": 0.40694117647058825,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1944,
      "step": 17295
    },
    {
      "epoch": 0.40696470588235295,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.7776,
      "step": 17296
    },
    {
      "epoch": 0.40698823529411765,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9612,
      "step": 17297
    },
    {
      "epoch": 0.40701176470588235,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.0174,
      "step": 17298
    },
    {
      "epoch": 0.40703529411764705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.02,
      "loss": 0.6054,
      "step": 17299
    },
    {
      "epoch": 0.40705882352941175,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9858,
      "step": 17300
    },
    {
      "epoch": 0.40708235294117645,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.1682,
      "step": 17301
    },
    {
      "epoch": 0.40710588235294115,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.9298,
      "step": 17302
    },
    {
      "epoch": 0.4071294117647059,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.1009,
      "step": 17303
    },
    {
      "epoch": 0.4071529411764706,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1047,
      "step": 17304
    },
    {
      "epoch": 0.4071764705882353,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 1.198,
      "step": 17305
    },
    {
      "epoch": 0.4072,
      "grad_norm": 0.20703125,
      "learning_rate": 0.02,
      "loss": 0.7679,
      "step": 17306
    },
    {
      "epoch": 0.4072235294117647,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8667,
      "step": 17307
    },
    {
      "epoch": 0.4072470588235294,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.855,
      "step": 17308
    },
    {
      "epoch": 0.4072705882352941,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.02,
      "loss": 1.0924,
      "step": 17309
    },
    {
      "epoch": 0.4072941176470588,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.995,
      "step": 17310
    },
    {
      "epoch": 0.4073176470588235,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8187,
      "step": 17311
    },
    {
      "epoch": 0.40734117647058826,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.02,
      "loss": 0.5567,
      "step": 17312
    },
    {
      "epoch": 0.40736470588235296,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.113,
      "step": 17313
    },
    {
      "epoch": 0.40738823529411766,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.1602,
      "step": 17314
    },
    {
      "epoch": 0.40741176470588236,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1762,
      "step": 17315
    },
    {
      "epoch": 0.40743529411764706,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0398,
      "step": 17316
    },
    {
      "epoch": 0.40745882352941176,
      "grad_norm": 0.21484375,
      "learning_rate": 0.02,
      "loss": 0.7318,
      "step": 17317
    },
    {
      "epoch": 0.40748235294117646,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.0366,
      "step": 17318
    },
    {
      "epoch": 0.40750588235294116,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 1.039,
      "step": 17319
    },
    {
      "epoch": 0.40752941176470586,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7627,
      "step": 17320
    },
    {
      "epoch": 0.40755294117647056,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.905,
      "step": 17321
    },
    {
      "epoch": 0.4075764705882353,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7726,
      "step": 17322
    },
    {
      "epoch": 0.4076,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.2421,
      "step": 17323
    },
    {
      "epoch": 0.4076235294117647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 0.9984,
      "step": 17324
    },
    {
      "epoch": 0.4076470588235294,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6548,
      "step": 17325
    },
    {
      "epoch": 0.4076705882352941,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 1.0465,
      "step": 17326
    },
    {
      "epoch": 0.4076941176470588,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.0409,
      "step": 17327
    },
    {
      "epoch": 0.4077176470588235,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.02,
      "loss": 0.8835,
      "step": 17328
    },
    {
      "epoch": 0.4077411764705882,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9497,
      "step": 17329
    },
    {
      "epoch": 0.4077647058823529,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.1592,
      "step": 17330
    },
    {
      "epoch": 0.4077882352941177,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.924,
      "step": 17331
    },
    {
      "epoch": 0.4078117647058824,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8257,
      "step": 17332
    },
    {
      "epoch": 0.4078352941176471,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0285,
      "step": 17333
    },
    {
      "epoch": 0.4078588235294118,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9621,
      "step": 17334
    },
    {
      "epoch": 0.4078823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.0896,
      "step": 17335
    },
    {
      "epoch": 0.4079058823529412,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.6147,
      "step": 17336
    },
    {
      "epoch": 0.4079294117647059,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0192,
      "step": 17337
    },
    {
      "epoch": 0.4079529411764706,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9187,
      "step": 17338
    },
    {
      "epoch": 0.4079764705882353,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.02,
      "loss": 0.5403,
      "step": 17339
    },
    {
      "epoch": 0.408,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.1591,
      "step": 17340
    },
    {
      "epoch": 0.40802352941176473,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.194,
      "step": 17341
    },
    {
      "epoch": 0.40804705882352943,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.0852,
      "step": 17342
    },
    {
      "epoch": 0.40807058823529413,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9111,
      "step": 17343
    },
    {
      "epoch": 0.40809411764705883,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.8542,
      "step": 17344
    },
    {
      "epoch": 0.40811764705882353,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0045,
      "step": 17345
    },
    {
      "epoch": 0.40814117647058823,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9902,
      "step": 17346
    },
    {
      "epoch": 0.40816470588235293,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2307,
      "step": 17347
    },
    {
      "epoch": 0.40818823529411763,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0349,
      "step": 17348
    },
    {
      "epoch": 0.40821176470588233,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0671,
      "step": 17349
    },
    {
      "epoch": 0.4082352941176471,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2955,
      "step": 17350
    },
    {
      "epoch": 0.4082588235294118,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0509,
      "step": 17351
    },
    {
      "epoch": 0.4082823529411765,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0023,
      "step": 17352
    },
    {
      "epoch": 0.4083058823529412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.2119,
      "step": 17353
    },
    {
      "epoch": 0.4083294117647059,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1903,
      "step": 17354
    },
    {
      "epoch": 0.4083529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.073,
      "step": 17355
    },
    {
      "epoch": 0.4083764705882353,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.2084,
      "step": 17356
    },
    {
      "epoch": 0.4084,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.236,
      "step": 17357
    },
    {
      "epoch": 0.4084235294117647,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.02,
      "loss": 0.618,
      "step": 17358
    },
    {
      "epoch": 0.4084470588235294,
      "grad_norm": 0.150390625,
      "learning_rate": 0.02,
      "loss": 1.2224,
      "step": 17359
    },
    {
      "epoch": 0.40847058823529414,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.1705,
      "step": 17360
    },
    {
      "epoch": 0.40849411764705884,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.7593,
      "step": 17361
    },
    {
      "epoch": 0.40851764705882354,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8611,
      "step": 17362
    },
    {
      "epoch": 0.40854117647058824,
      "grad_norm": 0.17578125,
      "learning_rate": 0.02,
      "loss": 1.1258,
      "step": 17363
    },
    {
      "epoch": 0.40856470588235294,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0144,
      "step": 17364
    },
    {
      "epoch": 0.40858823529411764,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1432,
      "step": 17365
    },
    {
      "epoch": 0.40861176470588234,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8927,
      "step": 17366
    },
    {
      "epoch": 0.40863529411764704,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9715,
      "step": 17367
    },
    {
      "epoch": 0.40865882352941174,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.8232,
      "step": 17368
    },
    {
      "epoch": 0.4086823529411765,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 1.1441,
      "step": 17369
    },
    {
      "epoch": 0.4087058823529412,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 1.0752,
      "step": 17370
    },
    {
      "epoch": 0.4087294117647059,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0091,
      "step": 17371
    },
    {
      "epoch": 0.4087529411764706,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.9849,
      "step": 17372
    },
    {
      "epoch": 0.4087764705882353,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.8529,
      "step": 17373
    },
    {
      "epoch": 0.4088,
      "grad_norm": 0.181640625,
      "learning_rate": 0.02,
      "loss": 0.8532,
      "step": 17374
    },
    {
      "epoch": 0.4088235294117647,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.0749,
      "step": 17375
    },
    {
      "epoch": 0.4088470588235294,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.9581,
      "step": 17376
    },
    {
      "epoch": 0.4088705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.7233,
      "step": 17377
    },
    {
      "epoch": 0.4088941176470588,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0892,
      "step": 17378
    },
    {
      "epoch": 0.40891764705882355,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.85,
      "step": 17379
    },
    {
      "epoch": 0.40894117647058825,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 0.9249,
      "step": 17380
    },
    {
      "epoch": 0.40896470588235295,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.6787,
      "step": 17381
    },
    {
      "epoch": 0.40898823529411765,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 1.2041,
      "step": 17382
    },
    {
      "epoch": 0.40901176470588235,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9958,
      "step": 17383
    },
    {
      "epoch": 0.40903529411764705,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1861,
      "step": 17384
    },
    {
      "epoch": 0.40905882352941175,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.3528,
      "step": 17385
    },
    {
      "epoch": 0.40908235294117645,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.7648,
      "step": 17386
    },
    {
      "epoch": 0.40910588235294115,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0604,
      "step": 17387
    },
    {
      "epoch": 0.4091294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.1201,
      "step": 17388
    },
    {
      "epoch": 0.4091529411764706,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.7446,
      "step": 17389
    },
    {
      "epoch": 0.4091764705882353,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 1.0475,
      "step": 17390
    },
    {
      "epoch": 0.4092,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.042,
      "step": 17391
    },
    {
      "epoch": 0.4092235294117647,
      "grad_norm": 0.185546875,
      "learning_rate": 0.02,
      "loss": 0.9977,
      "step": 17392
    },
    {
      "epoch": 0.4092470588235294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.8851,
      "step": 17393
    },
    {
      "epoch": 0.4092705882352941,
      "grad_norm": 0.19921875,
      "learning_rate": 0.02,
      "loss": 0.4205,
      "step": 17394
    },
    {
      "epoch": 0.4092941176470588,
      "grad_norm": 0.154296875,
      "learning_rate": 0.02,
      "loss": 1.1432,
      "step": 17395
    },
    {
      "epoch": 0.4093176470588235,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9345,
      "step": 17396
    },
    {
      "epoch": 0.4093411764705882,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.02,
      "loss": 1.5996,
      "step": 17397
    },
    {
      "epoch": 0.40936470588235296,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.02,
      "loss": 0.9157,
      "step": 17398
    },
    {
      "epoch": 0.40938823529411766,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 1.1186,
      "step": 17399
    },
    {
      "epoch": 0.40941176470588236,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.8571,
      "step": 17400
    },
    {
      "epoch": 0.40943529411764706,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0684,
      "step": 17401
    },
    {
      "epoch": 0.40945882352941176,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 1.0205,
      "step": 17402
    },
    {
      "epoch": 0.40948235294117646,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8055,
      "step": 17403
    },
    {
      "epoch": 0.40950588235294116,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9619,
      "step": 17404
    },
    {
      "epoch": 0.40952941176470586,
      "grad_norm": 0.166015625,
      "learning_rate": 0.02,
      "loss": 0.9278,
      "step": 17405
    },
    {
      "epoch": 0.40955294117647056,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.02,
      "loss": 1.0689,
      "step": 17406
    },
    {
      "epoch": 0.4095764705882353,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.1644,
      "step": 17407
    },
    {
      "epoch": 0.4096,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.0105,
      "step": 17408
    },
    {
      "epoch": 0.4096235294117647,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.05,
      "step": 17409
    },
    {
      "epoch": 0.4096470588235294,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0545,
      "step": 17410
    },
    {
      "epoch": 0.4096705882352941,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 1.0305,
      "step": 17411
    },
    {
      "epoch": 0.4096941176470588,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.9338,
      "step": 17412
    },
    {
      "epoch": 0.4097176470588235,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9499,
      "step": 17413
    },
    {
      "epoch": 0.4097411764705882,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8427,
      "step": 17414
    },
    {
      "epoch": 0.4097647058823529,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.8781,
      "step": 17415
    },
    {
      "epoch": 0.4097882352941176,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9169,
      "step": 17416
    },
    {
      "epoch": 0.4098117647058824,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 0.8384,
      "step": 17417
    },
    {
      "epoch": 0.4098352941176471,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.0911,
      "step": 17418
    },
    {
      "epoch": 0.4098588235294118,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 0.9272,
      "step": 17419
    },
    {
      "epoch": 0.4098823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.2464,
      "step": 17420
    },
    {
      "epoch": 0.4099058823529412,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.0563,
      "step": 17421
    },
    {
      "epoch": 0.4099294117647059,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.02,
      "loss": 1.0287,
      "step": 17422
    },
    {
      "epoch": 0.4099529411764706,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 1.0661,
      "step": 17423
    },
    {
      "epoch": 0.4099764705882353,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 1.3528,
      "step": 17424
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.8789,
      "step": 17425
    },
    {
      "epoch": 0.41002352941176473,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.999,
      "step": 17426
    },
    {
      "epoch": 0.41004705882352943,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.147,
      "step": 17427
    },
    {
      "epoch": 0.41007058823529413,
      "grad_norm": 0.1484375,
      "learning_rate": 0.02,
      "loss": 1.3393,
      "step": 17428
    },
    {
      "epoch": 0.41009411764705883,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 1.0963,
      "step": 17429
    },
    {
      "epoch": 0.41011764705882353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 1.3362,
      "step": 17430
    },
    {
      "epoch": 0.41014117647058823,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.02,
      "loss": 0.907,
      "step": 17431
    },
    {
      "epoch": 0.41016470588235293,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 1.1074,
      "step": 17432
    },
    {
      "epoch": 0.41018823529411763,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.02,
      "loss": 0.8849,
      "step": 17433
    },
    {
      "epoch": 0.41021176470588233,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9648,
      "step": 17434
    },
    {
      "epoch": 0.41023529411764703,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.8564,
      "step": 17435
    },
    {
      "epoch": 0.4102588235294118,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 0.9583,
      "step": 17436
    },
    {
      "epoch": 0.4102823529411765,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.9131,
      "step": 17437
    },
    {
      "epoch": 0.4103058823529412,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.02,
      "loss": 1.1528,
      "step": 17438
    },
    {
      "epoch": 0.4103294117647059,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9926,
      "step": 17439
    },
    {
      "epoch": 0.4103529411764706,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.02,
      "loss": 1.1352,
      "step": 17440
    },
    {
      "epoch": 0.4103764705882353,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9453,
      "step": 17441
    },
    {
      "epoch": 0.4104,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.7089,
      "step": 17442
    },
    {
      "epoch": 0.4104235294117647,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.02,
      "loss": 0.9972,
      "step": 17443
    },
    {
      "epoch": 0.4104470588235294,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 1.1402,
      "step": 17444
    },
    {
      "epoch": 0.41047058823529414,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.8717,
      "step": 17445
    },
    {
      "epoch": 0.41049411764705884,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2409,
      "step": 17446
    },
    {
      "epoch": 0.41051764705882354,
      "grad_norm": 0.212890625,
      "learning_rate": 0.02,
      "loss": 0.6196,
      "step": 17447
    },
    {
      "epoch": 0.41054117647058824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 1.0623,
      "step": 17448
    },
    {
      "epoch": 0.41056470588235294,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.7673,
      "step": 17449
    },
    {
      "epoch": 0.41058823529411764,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8843,
      "step": 17450
    },
    {
      "epoch": 0.41061176470588234,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.1235,
      "step": 17451
    },
    {
      "epoch": 0.41063529411764704,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.7907,
      "step": 17452
    },
    {
      "epoch": 0.41065882352941174,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.9433,
      "step": 17453
    },
    {
      "epoch": 0.41068235294117644,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.8361,
      "step": 17454
    },
    {
      "epoch": 0.4107058823529412,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 1.002,
      "step": 17455
    },
    {
      "epoch": 0.4107294117647059,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.9969,
      "step": 17456
    },
    {
      "epoch": 0.4107529411764706,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.9169,
      "step": 17457
    },
    {
      "epoch": 0.4107764705882353,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7629,
      "step": 17458
    },
    {
      "epoch": 0.4108,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.02,
      "loss": 0.9237,
      "step": 17459
    },
    {
      "epoch": 0.4108235294117647,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.02,
      "loss": 0.825,
      "step": 17460
    },
    {
      "epoch": 0.4108470588235294,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0602,
      "step": 17461
    },
    {
      "epoch": 0.4108705882352941,
      "grad_norm": 0.171875,
      "learning_rate": 0.02,
      "loss": 0.9619,
      "step": 17462
    },
    {
      "epoch": 0.4108941176470588,
      "grad_norm": 0.162109375,
      "learning_rate": 0.02,
      "loss": 1.0673,
      "step": 17463
    },
    {
      "epoch": 0.41091764705882355,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.02,
      "loss": 1.2337,
      "step": 17464
    },
    {
      "epoch": 0.41094117647058825,
      "grad_norm": 0.197265625,
      "learning_rate": 0.02,
      "loss": 0.7072,
      "step": 17465
    },
    {
      "epoch": 0.41096470588235295,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.02,
      "loss": 0.8795,
      "step": 17466
    },
    {
      "epoch": 0.41098823529411765,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 0.9933,
      "step": 17467
    },
    {
      "epoch": 0.41101176470588235,
      "grad_norm": 0.1640625,
      "learning_rate": 0.02,
      "loss": 1.151,
      "step": 17468
    },
    {
      "epoch": 0.41103529411764705,
      "grad_norm": 0.1796875,
      "learning_rate": 0.02,
      "loss": 0.7845,
      "step": 17469
    },
    {
      "epoch": 0.41105882352941175,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 1.0023,
      "step": 17470
    },
    {
      "epoch": 0.41108235294117645,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.02,
      "loss": 0.9398,
      "step": 17471
    },
    {
      "epoch": 0.41110588235294115,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.02,
      "loss": 1.4449,
      "step": 17472
    },
    {
      "epoch": 0.4111294117647059,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.8314,
      "step": 17473
    },
    {
      "epoch": 0.4111529411764706,
      "grad_norm": 0.173828125,
      "learning_rate": 0.02,
      "loss": 0.7942,
      "step": 17474
    },
    {
      "epoch": 0.4111764705882353,
      "grad_norm": 0.16015625,
      "learning_rate": 0.02,
      "loss": 1.2878,
      "step": 17475
    },
    {
      "epoch": 0.4112,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.02,
      "loss": 0.9063,
      "step": 17476
    },
    {
      "epoch": 0.4112235294117647,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.0267,
      "step": 17477
    },
    {
      "epoch": 0.4112470588235294,
      "grad_norm": 0.177734375,
      "learning_rate": 0.02,
      "loss": 0.9543,
      "step": 17478
    },
    {
      "epoch": 0.4112705882352941,
      "grad_norm": 0.1875,
      "learning_rate": 0.02,
      "loss": 1.1084,
      "step": 17479
    },
    {
      "epoch": 0.4112941176470588,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.02,
      "loss": 0.986,
      "step": 17480
    },
    {
      "epoch": 0.4113176470588235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.02,
      "loss": 0.9048,
      "step": 17481
    },
    {
      "epoch": 0.4113411764705882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.02,
      "loss": 0.8176,
      "step": 17482
    },
    {
      "epoch": 0.41136470588235297,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.2305,
      "step": 17483
    },
    {
      "epoch": 0.41138823529411767,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.02,
      "loss": 0.7281,
      "step": 17484
    },
    {
      "epoch": 0.41141176470588237,
      "grad_norm": 0.15625,
      "learning_rate": 0.02,
      "loss": 1.0059,
      "step": 17485
    },
    {
      "epoch": 0.41143529411764707,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.02,
      "loss": 0.9923,
      "step": 17486
    },
    {
      "epoch": 0.41145882352941177,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0094,
      "step": 17487
    },
    {
      "epoch": 0.41148235294117647,
      "grad_norm": 0.169921875,
      "learning_rate": 0.02,
      "loss": 1.0283,
      "step": 17488
    },
    {
      "epoch": 0.41150588235294117,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.02,
      "loss": 0.7556,
      "step": 17489
    },
    {
      "epoch": 0.41152941176470587,
      "grad_norm": 0.158203125,
      "learning_rate": 0.02,
      "loss": 1.0252,
      "step": 17490
    },
    {
      "epoch": 0.41155294117647057,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.02,
      "loss": 1.1781,
      "step": 17491
    },
    {
      "epoch": 0.4115764705882353,
      "grad_norm": 0.232421875,
      "learning_rate": 0.02,
      "loss": 1.0702,
      "step": 17492
    },
    {
      "epoch": 0.4116,
      "grad_norm": 0.146484375,
      "learning_rate": 0.02,
      "loss": 1.4155,
      "step": 17493
    },
    {
      "epoch": 0.4116235294117647,
      "grad_norm": 0.193359375,
      "learning_rate": 0.02,
      "loss": 0.7684,
      "step": 17494
    },
    {
      "epoch": 0.4116470588235294,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.02,
      "loss": 0.7149,
      "step": 17495
    },
    {
      "epoch": 0.4116705882352941,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.02,
      "loss": 0.6803,
      "step": 17496
    },
    {
      "epoch": 0.4116941176470588,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.02,
      "loss": 0.6649,
      "step": 17497
    },
    {
      "epoch": 0.4117176470588235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.02,
      "loss": 1.0341,
      "step": 17498
    },
    {
      "epoch": 0.4117411764705882,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.02,
      "loss": 1.1641,
      "step": 17499
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.02,
      "loss": 0.8341,
      "step": 17500
    }
  ],
  "logging_steps": 1,
  "max_steps": 42500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0620032579532251e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}