{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 6638,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.1291910702757493,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.7898,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5743399453486893,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.7976,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.408140036816484,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.8066,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0369231609977705,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.7742,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4045258377865406,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.8093,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.192854358756726,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.7987,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.179435171496737,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.7806,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7600145093184723,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.7779,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6334162961123107,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.791,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7593573087407774,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.7963,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.022298873333028,
      "learning_rate": 1.1e-06,
      "loss": 0.7786,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8571091808252138,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.7457,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8014325263875204,
      "learning_rate": 1.3e-06,
      "loss": 0.7587,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7217253806056376,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.7474,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3092653863179984,
      "learning_rate": 1.5e-06,
      "loss": 0.7216,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2331588787170795,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.7354,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1616501150831484,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.7348,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0440466305242875,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.7405,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2286286434459004,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.7246,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3461806021811673,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7137,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5919071579098707,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.722,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8871137439671788,
      "learning_rate": 2.2e-06,
      "loss": 0.6969,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7029957541596448,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.7336,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.484120057066233,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.7315,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3229722524380603,
      "learning_rate": 2.5e-06,
      "loss": 0.7426,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0649017852882396,
      "learning_rate": 2.6e-06,
      "loss": 0.7005,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1911261976529834,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.7307,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9351221228696979,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.694,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9705102378917375,
      "learning_rate": 2.9e-06,
      "loss": 0.7313,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9285785868390756,
      "learning_rate": 3e-06,
      "loss": 0.7013,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9121368602563168,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.6738,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8244102961685571,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.695,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8592042543994333,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.6832,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9353271711278436,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.6786,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8573046001620593,
      "learning_rate": 3.5e-06,
      "loss": 0.6824,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8666925084777215,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.6538,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8630078909751495,
      "learning_rate": 3.7e-06,
      "loss": 0.6763,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9282260272410224,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.7216,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8819477248142638,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.7026,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8658728499129157,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7268,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8442691494378844,
      "learning_rate": 4.1e-06,
      "loss": 0.6607,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8096259039104774,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.6463,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.833271560085352,
      "learning_rate": 4.3e-06,
      "loss": 0.6496,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8304641583538493,
      "learning_rate": 4.4e-06,
      "loss": 0.6421,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8720165878371252,
      "learning_rate": 4.5e-06,
      "loss": 0.6646,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8585253022931961,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.6421,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8288683852456689,
      "learning_rate": 4.7e-06,
      "loss": 0.6676,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8832959006124463,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6602,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8521227455713961,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.6488,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0203646273811025,
      "learning_rate": 5e-06,
      "loss": 0.6712,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8365183138742138,
      "learning_rate": 5.1e-06,
      "loss": 0.6533,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9399614172283228,
      "learning_rate": 5.2e-06,
      "loss": 0.6422,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.911888362188921,
      "learning_rate": 5.300000000000001e-06,
      "loss": 0.6321,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9685166547676656,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.6532,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9185758242102973,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.6581,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8066420429747783,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.6282,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8778138005027036,
      "learning_rate": 5.7e-06,
      "loss": 0.6105,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8693052857912392,
      "learning_rate": 5.8e-06,
      "loss": 0.654,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9305910120933834,
      "learning_rate": 5.9e-06,
      "loss": 0.6162,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9176642372764262,
      "learning_rate": 6e-06,
      "loss": 0.6347,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9552276407473728,
      "learning_rate": 6.1e-06,
      "loss": 0.6373,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9277512147682052,
      "learning_rate": 6.200000000000001e-06,
      "loss": 0.6281,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9702714083014237,
      "learning_rate": 6.300000000000001e-06,
      "loss": 0.6369,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0455476276636446,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.6408,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9196492982027352,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.6592,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0086245863210648,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.642,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0810509774431099,
      "learning_rate": 6.700000000000001e-06,
      "loss": 0.6377,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8327359755124601,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.5989,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1166780819825322,
      "learning_rate": 6.9e-06,
      "loss": 0.628,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7983355211085578,
      "learning_rate": 7e-06,
      "loss": 0.6306,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9964382394841694,
      "learning_rate": 7.100000000000001e-06,
      "loss": 0.6189,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8466285450835838,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.648,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9753129568148445,
      "learning_rate": 7.3e-06,
      "loss": 0.6133,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8296347822942614,
      "learning_rate": 7.4e-06,
      "loss": 0.6125,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9702605487029193,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.6522,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9121250803491526,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.6442,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8865887151614031,
      "learning_rate": 7.7e-06,
      "loss": 0.6244,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.889106726051307,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.6253,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8443640850896325,
      "learning_rate": 7.9e-06,
      "loss": 0.6336,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9185854421046048,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6331,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8848093481928698,
      "learning_rate": 8.1e-06,
      "loss": 0.6044,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8094414699257236,
      "learning_rate": 8.2e-06,
      "loss": 0.6117,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9195540107216379,
      "learning_rate": 8.3e-06,
      "loss": 0.6394,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8039538759528256,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.6311,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7572281165070127,
      "learning_rate": 8.5e-06,
      "loss": 0.607,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1240141532027759,
      "learning_rate": 8.6e-06,
      "loss": 0.6356,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8763456350715176,
      "learning_rate": 8.700000000000001e-06,
      "loss": 0.6152,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9152618607340062,
      "learning_rate": 8.8e-06,
      "loss": 0.6094,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.386784435183377,
      "learning_rate": 8.900000000000001e-06,
      "loss": 0.6159,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.141890084193862,
      "learning_rate": 9e-06,
      "loss": 0.627,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9949470146717242,
      "learning_rate": 9.100000000000001e-06,
      "loss": 0.6343,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.030345011422465,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.6255,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0521986623364197,
      "learning_rate": 9.3e-06,
      "loss": 0.6341,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9259852672842965,
      "learning_rate": 9.4e-06,
      "loss": 0.5954,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9101017052814089,
      "learning_rate": 9.5e-06,
      "loss": 0.6384,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0911086244027206,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.6449,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7817880721225067,
      "learning_rate": 9.7e-06,
      "loss": 0.6187,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9449710644430548,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.6297,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0150974726021715,
      "learning_rate": 9.9e-06,
      "loss": 0.629,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8747109445435874,
      "learning_rate": 1e-05,
      "loss": 0.5869,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9871104750127209,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 0.6086,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8886805543491788,
      "learning_rate": 1.02e-05,
      "loss": 0.6339,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9242438488022526,
      "learning_rate": 1.0300000000000001e-05,
      "loss": 0.5894,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9310553792520874,
      "learning_rate": 1.04e-05,
      "loss": 0.6452,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8618850173698189,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 0.6223,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8685307061569937,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.6156,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9775148738619646,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 0.6206,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9184944445092023,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.6138,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9689622899493467,
      "learning_rate": 1.0900000000000002e-05,
      "loss": 0.609,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0705653694907644,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.6468,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.041632309263693,
      "learning_rate": 1.1100000000000002e-05,
      "loss": 0.6184,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9559000015404067,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.6102,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1490617195076391,
      "learning_rate": 1.13e-05,
      "loss": 0.6387,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8173676498187492,
      "learning_rate": 1.14e-05,
      "loss": 0.6315,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9889096866043532,
      "learning_rate": 1.15e-05,
      "loss": 0.6023,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9367491489877903,
      "learning_rate": 1.16e-05,
      "loss": 0.6191,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7836995168660515,
      "learning_rate": 1.17e-05,
      "loss": 0.6064,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7511491104815611,
      "learning_rate": 1.18e-05,
      "loss": 0.6166,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8897609297944588,
      "learning_rate": 1.1900000000000001e-05,
      "loss": 0.6113,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9609169128153736,
      "learning_rate": 1.2e-05,
      "loss": 0.6238,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9078723097173822,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 0.6262,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8345145283516961,
      "learning_rate": 1.22e-05,
      "loss": 0.6049,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8746750598451308,
      "learning_rate": 1.23e-05,
      "loss": 0.6135,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8456177968495505,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.5999,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9750732526219859,
      "learning_rate": 1.25e-05,
      "loss": 0.6221,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1793704234923532,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.6401,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8239583650508742,
      "learning_rate": 1.27e-05,
      "loss": 0.6138,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0551004167174587,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.6137,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8531553842486622,
      "learning_rate": 1.2900000000000002e-05,
      "loss": 0.6049,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1232287815766062,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.6143,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8316745924379109,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 0.6175,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0474498709226914,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.618,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8644277122360153,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 0.6009,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1169211446846186,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.6168,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0024172109827714,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.6246,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0197272393483,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.615,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9351097292191448,
      "learning_rate": 1.3700000000000003e-05,
      "loss": 0.6054,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0114992186987921,
      "learning_rate": 1.38e-05,
      "loss": 0.6081,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9512574666921964,
      "learning_rate": 1.39e-05,
      "loss": 0.5947,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9614826716661977,
      "learning_rate": 1.4e-05,
      "loss": 0.6223,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8828501625350947,
      "learning_rate": 1.41e-05,
      "loss": 0.595,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8605787149828122,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.6244,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8922191045500464,
      "learning_rate": 1.43e-05,
      "loss": 0.6213,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8775581521626897,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.5955,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8830578659008969,
      "learning_rate": 1.45e-05,
      "loss": 0.6157,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9599519435312424,
      "learning_rate": 1.46e-05,
      "loss": 0.6295,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9305165517799335,
      "learning_rate": 1.4700000000000002e-05,
      "loss": 0.6037,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9404766146821452,
      "learning_rate": 1.48e-05,
      "loss": 0.6082,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8749723441054693,
      "learning_rate": 1.4900000000000001e-05,
      "loss": 0.6087,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.836393816636118,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.6059,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8248685225153953,
      "learning_rate": 1.5100000000000001e-05,
      "loss": 0.6215,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9517614470626706,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.6193,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8874182679954177,
      "learning_rate": 1.5300000000000003e-05,
      "loss": 0.6019,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9053566485587093,
      "learning_rate": 1.54e-05,
      "loss": 0.6183,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0047867186514934,
      "learning_rate": 1.55e-05,
      "loss": 0.61,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8632764963763306,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.5931,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8979673120872339,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 0.6058,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9295015677709256,
      "learning_rate": 1.58e-05,
      "loss": 0.6141,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8651756390032996,
      "learning_rate": 1.5900000000000004e-05,
      "loss": 0.6383,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9576764645219543,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.6378,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9562832208584039,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 0.6436,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.842180276759221,
      "learning_rate": 1.62e-05,
      "loss": 0.6134,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8933911914535261,
      "learning_rate": 1.63e-05,
      "loss": 0.6139,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8609596665488398,
      "learning_rate": 1.64e-05,
      "loss": 0.6259,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9038589067930154,
      "learning_rate": 1.65e-05,
      "loss": 0.6006,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9159368720168773,
      "learning_rate": 1.66e-05,
      "loss": 0.6413,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8268613628259516,
      "learning_rate": 1.67e-05,
      "loss": 0.6517,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8707392123640353,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.6112,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9728792869227342,
      "learning_rate": 1.69e-05,
      "loss": 0.5987,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8963016459874905,
      "learning_rate": 1.7e-05,
      "loss": 0.6292,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8511340991468719,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 0.6169,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8032322791373534,
      "learning_rate": 1.72e-05,
      "loss": 0.5992,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8247583874184422,
      "learning_rate": 1.73e-05,
      "loss": 0.6125,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7619856489701343,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.6006,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.85080176741147,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.6276,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7655701337916373,
      "learning_rate": 1.76e-05,
      "loss": 0.6095,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8982536001774453,
      "learning_rate": 1.77e-05,
      "loss": 0.5991,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8582703872444867,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.6127,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8086712317242247,
      "learning_rate": 1.79e-05,
      "loss": 0.6192,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9301182536445927,
      "learning_rate": 1.8e-05,
      "loss": 0.6014,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.946931765446461,
      "learning_rate": 1.8100000000000003e-05,
      "loss": 0.6128,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.808029198674719,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.5991,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9540930800247533,
      "learning_rate": 1.83e-05,
      "loss": 0.6142,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8233501640169226,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.6203,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8710606643872496,
      "learning_rate": 1.8500000000000002e-05,
      "loss": 0.6266,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8698312840584582,
      "learning_rate": 1.86e-05,
      "loss": 0.6221,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8373012004693599,
      "learning_rate": 1.8700000000000004e-05,
      "loss": 0.6413,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8914217213586093,
      "learning_rate": 1.88e-05,
      "loss": 0.6122,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8496316034938707,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.5938,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9059606579462255,
      "learning_rate": 1.9e-05,
      "loss": 0.6183,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7851187238540059,
      "learning_rate": 1.91e-05,
      "loss": 0.6142,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7812436970503833,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.598,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8952159510828972,
      "learning_rate": 1.93e-05,
      "loss": 0.621,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.842071205049164,
      "learning_rate": 1.94e-05,
      "loss": 0.6233,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9231029146421179,
      "learning_rate": 1.95e-05,
      "loss": 0.6314,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8541355245707236,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.6191,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9109607622413463,
      "learning_rate": 1.97e-05,
      "loss": 0.6098,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8034688813017887,
      "learning_rate": 1.98e-05,
      "loss": 0.6143,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0790404874189345,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 0.6184,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8946079782329716,
      "learning_rate": 2e-05,
      "loss": 0.6011,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8667949716375661,
      "learning_rate": 1.9999998809394743e-05,
      "loss": 0.6178,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8289636293502011,
      "learning_rate": 1.999999523757925e-05,
      "loss": 0.6129,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9751641640848245,
      "learning_rate": 1.9999989284554374e-05,
      "loss": 0.6101,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9092990711705473,
      "learning_rate": 1.9999980950321534e-05,
      "loss": 0.6293,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8758042953946426,
      "learning_rate": 1.999997023488271e-05,
      "loss": 0.6069,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9671873016512136,
      "learning_rate": 1.999995713824046e-05,
      "loss": 0.6242,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8992519262617124,
      "learning_rate": 1.9999941660397898e-05,
      "loss": 0.6067,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7739313736113903,
      "learning_rate": 1.9999923801358706e-05,
      "loss": 0.6002,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8121368221745613,
      "learning_rate": 1.9999903561127147e-05,
      "loss": 0.6072,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8447708172988327,
      "learning_rate": 1.9999880939708028e-05,
      "loss": 0.6269,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8427372580421794,
      "learning_rate": 1.9999855937106748e-05,
      "loss": 0.6382,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8792509308316667,
      "learning_rate": 1.9999828553329254e-05,
      "loss": 0.625,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8892040254206709,
      "learning_rate": 1.999979878838207e-05,
      "loss": 0.6159,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.898352510393593,
      "learning_rate": 1.9999766642272278e-05,
      "loss": 0.6073,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8271058179192874,
      "learning_rate": 1.9999732115007537e-05,
      "loss": 0.6277,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8228916759151659,
      "learning_rate": 1.999969520659607e-05,
      "loss": 0.6031,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8830609159980283,
      "learning_rate": 1.999965591704666e-05,
      "loss": 0.6183,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8567489226333039,
      "learning_rate": 1.9999614246368666e-05,
      "loss": 0.6314,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8602864135439753,
      "learning_rate": 1.9999570194572015e-05,
      "loss": 0.6275,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.788337765910359,
      "learning_rate": 1.9999523761667188e-05,
      "loss": 0.6153,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8503289917165229,
      "learning_rate": 1.9999474947665248e-05,
      "loss": 0.6056,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7286762220289179,
      "learning_rate": 1.9999423752577816e-05,
      "loss": 0.6187,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.870251703399336,
      "learning_rate": 1.9999370176417083e-05,
      "loss": 0.6494,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7905739776545552,
      "learning_rate": 1.9999314219195808e-05,
      "loss": 0.6207,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8158200537387095,
      "learning_rate": 1.9999255880927313e-05,
      "loss": 0.6304,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8061028561956776,
      "learning_rate": 1.9999195161625494e-05,
      "loss": 0.6011,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.866270617202648,
      "learning_rate": 1.9999132061304803e-05,
      "loss": 0.6109,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8125673091865734,
      "learning_rate": 1.9999066579980274e-05,
      "loss": 0.616,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7516226097321476,
      "learning_rate": 1.999899871766749e-05,
      "loss": 0.6082,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8191560388696365,
      "learning_rate": 1.9998928474382617e-05,
      "loss": 0.6064,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8626374401966489,
      "learning_rate": 1.999885585014238e-05,
      "loss": 0.6117,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7393257990658336,
      "learning_rate": 1.999878084496407e-05,
      "loss": 0.6198,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7473034506178765,
      "learning_rate": 1.999870345886555e-05,
      "loss": 0.6144,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7696400394202073,
      "learning_rate": 1.9998623691865247e-05,
      "loss": 0.5836,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8789695046428294,
      "learning_rate": 1.9998541543982154e-05,
      "loss": 0.6231,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8277046448447798,
      "learning_rate": 1.9998457015235836e-05,
      "loss": 0.6204,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.804013586893842,
      "learning_rate": 1.9998370105646414e-05,
      "loss": 0.6081,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9798766792492065,
      "learning_rate": 1.999828081523459e-05,
      "loss": 0.6317,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7843464271575187,
      "learning_rate": 1.999818914402162e-05,
      "loss": 0.5687,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8710131989149477,
      "learning_rate": 1.9998095092029334e-05,
      "loss": 0.6188,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7231753848804833,
      "learning_rate": 1.9997998659280134e-05,
      "loss": 0.5976,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1353198334421062,
      "learning_rate": 1.9997899845796975e-05,
      "loss": 0.6158,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7973293969720194,
      "learning_rate": 1.999779865160339e-05,
      "loss": 0.6473,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9326340754214019,
      "learning_rate": 1.9997695076723475e-05,
      "loss": 0.6189,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8819221223754583,
      "learning_rate": 1.9997589121181897e-05,
      "loss": 0.6388,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7644006600409795,
      "learning_rate": 1.9997480785003876e-05,
      "loss": 0.6366,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9262652446711976,
      "learning_rate": 1.999737006821522e-05,
      "loss": 0.6228,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9513398265030417,
      "learning_rate": 1.9997256970842288e-05,
      "loss": 0.6188,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7927602228576611,
      "learning_rate": 1.9997141492912014e-05,
      "loss": 0.6145,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.767421103905615,
      "learning_rate": 1.9997023634451887e-05,
      "loss": 0.6171,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7633445286227557,
      "learning_rate": 1.9996903395489986e-05,
      "loss": 0.5918,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7569859853533771,
      "learning_rate": 1.999678077605493e-05,
      "loss": 0.6153,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8475243774529355,
      "learning_rate": 1.9996655776175925e-05,
      "loss": 0.6209,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9595281983611552,
      "learning_rate": 1.999652839588273e-05,
      "loss": 0.6229,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8165164588608772,
      "learning_rate": 1.999639863520568e-05,
      "loss": 0.6228,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9744701909117809,
      "learning_rate": 1.999626649417567e-05,
      "loss": 0.6331,
      "step": 256
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8254055260121674,
      "learning_rate": 1.9996131972824172e-05,
      "loss": 0.6139,
      "step": 257
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8995321023820985,
      "learning_rate": 1.999599507118322e-05,
      "loss": 0.5898,
      "step": 258
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.888016725157491,
      "learning_rate": 1.9995855789285402e-05,
      "loss": 0.6148,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7696890958944222,
      "learning_rate": 1.999571412716389e-05,
      "loss": 0.6122,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7873486640880475,
      "learning_rate": 1.9995570084852423e-05,
      "loss": 0.6084,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7849376031274812,
      "learning_rate": 1.9995423662385292e-05,
      "loss": 0.6428,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8741378288653926,
      "learning_rate": 1.9995274859797368e-05,
      "loss": 0.6237,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8334121294911215,
      "learning_rate": 1.999512367712408e-05,
      "loss": 0.6088,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9453542389534016,
      "learning_rate": 1.9994970114401427e-05,
      "loss": 0.6416,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7846916985588137,
      "learning_rate": 1.999481417166598e-05,
      "loss": 0.6372,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8073472732630338,
      "learning_rate": 1.9994655848954872e-05,
      "loss": 0.6121,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8346670975973219,
      "learning_rate": 1.9994495146305805e-05,
      "loss": 0.6192,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8262587962506021,
      "learning_rate": 1.9994332063757037e-05,
      "loss": 0.5906,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7409703851240032,
      "learning_rate": 1.9994166601347408e-05,
      "loss": 0.6214,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8303553620740758,
      "learning_rate": 1.9993998759116318e-05,
      "loss": 0.6485,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8166395582688053,
      "learning_rate": 1.999382853710373e-05,
      "loss": 0.6281,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8460451178904631,
      "learning_rate": 1.999365593535018e-05,
      "loss": 0.6451,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8397704700245936,
      "learning_rate": 1.999348095389677e-05,
      "loss": 0.6337,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8336878725335758,
      "learning_rate": 1.9993303592785165e-05,
      "loss": 0.6392,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8004180766733375,
      "learning_rate": 1.9993123852057597e-05,
      "loss": 0.6257,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8307559645663873,
      "learning_rate": 1.9992941731756867e-05,
      "loss": 0.5968,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8827935546115524,
      "learning_rate": 1.9992757231926344e-05,
      "loss": 0.6353,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9461642244348395,
      "learning_rate": 1.9992570352609957e-05,
      "loss": 0.6229,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8510371022674922,
      "learning_rate": 1.9992381093852213e-05,
      "loss": 0.6234,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8595402715350201,
      "learning_rate": 1.9992189455698167e-05,
      "loss": 0.5955,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9090282781288724,
      "learning_rate": 1.999199543819346e-05,
      "loss": 0.6052,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8419607092972856,
      "learning_rate": 1.9991799041384293e-05,
      "loss": 0.6171,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8261359773929668,
      "learning_rate": 1.9991600265317428e-05,
      "loss": 0.6291,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7376957707976967,
      "learning_rate": 1.99913991100402e-05,
      "loss": 0.6116,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9004359221149307,
      "learning_rate": 1.9991195575600506e-05,
      "loss": 0.6297,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8398442378832626,
      "learning_rate": 1.999098966204682e-05,
      "loss": 0.6201,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9938152643462491,
      "learning_rate": 1.9990781369428158e-05,
      "loss": 0.607,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9899494645461415,
      "learning_rate": 1.9990570697794137e-05,
      "loss": 0.612,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7491514276939893,
      "learning_rate": 1.9990357647194907e-05,
      "loss": 0.6259,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9540558193081939,
      "learning_rate": 1.999014221768121e-05,
      "loss": 0.6155,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8954609421735867,
      "learning_rate": 1.9989924409304344e-05,
      "loss": 0.612,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8793122968795871,
      "learning_rate": 1.998970422211617e-05,
      "loss": 0.6322,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8549478098050506,
      "learning_rate": 1.9989481656169116e-05,
      "loss": 0.6177,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8422173563876716,
      "learning_rate": 1.9989256711516188e-05,
      "loss": 0.62,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7877688334987212,
      "learning_rate": 1.998902938821094e-05,
      "loss": 0.6039,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7899569054755726,
      "learning_rate": 1.9988799686307515e-05,
      "loss": 0.6114,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9422795325985925,
      "learning_rate": 1.99885676058606e-05,
      "loss": 0.6059,
      "step": 298
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.758782465012535,
      "learning_rate": 1.9988333146925464e-05,
      "loss": 0.612,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8873080224005514,
      "learning_rate": 1.9988096309557933e-05,
      "loss": 0.6272,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7435883606709703,
      "learning_rate": 1.9987857093814402e-05,
      "loss": 0.6206,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6918239594393029,
      "learning_rate": 1.9987615499751837e-05,
      "loss": 0.6228,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8471154808940652,
      "learning_rate": 1.9987371527427763e-05,
      "loss": 0.6272,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8858395309694191,
      "learning_rate": 1.998712517690028e-05,
      "loss": 0.6256,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7014895777310494,
      "learning_rate": 1.9986876448228044e-05,
      "loss": 0.6129,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7803088308639056,
      "learning_rate": 1.9986625341470285e-05,
      "loss": 0.6281,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9526350643320368,
      "learning_rate": 1.9986371856686795e-05,
      "loss": 0.6598,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7759039632256363,
      "learning_rate": 1.9986115993937936e-05,
      "loss": 0.64,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7251899134741734,
      "learning_rate": 1.9985857753284635e-05,
      "loss": 0.5959,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7814917336202796,
      "learning_rate": 1.9985597134788382e-05,
      "loss": 0.5913,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8916094612293517,
      "learning_rate": 1.998533413851124e-05,
      "loss": 0.6224,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9181990281237873,
      "learning_rate": 1.998506876451583e-05,
      "loss": 0.6001,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7540171882816669,
      "learning_rate": 1.998480101286534e-05,
      "loss": 0.6228,
      "step": 313
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7922177717927621,
      "learning_rate": 1.9984530883623538e-05,
      "loss": 0.6167,
      "step": 314
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8633613549188706,
      "learning_rate": 1.9984258376854736e-05,
      "loss": 0.6335,
      "step": 315
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7745922326995865,
      "learning_rate": 1.9983983492623832e-05,
      "loss": 0.6003,
      "step": 316
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7643629577737578,
      "learning_rate": 1.9983706230996276e-05,
      "loss": 0.6315,
      "step": 317
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6705131953533214,
      "learning_rate": 1.9983426592038097e-05,
      "loss": 0.6294,
      "step": 318
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7904360830247973,
      "learning_rate": 1.9983144575815873e-05,
      "loss": 0.6159,
      "step": 319
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8159941719959191,
      "learning_rate": 1.9982860182396763e-05,
      "loss": 0.5922,
      "step": 320
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8692584463920163,
      "learning_rate": 1.9982573411848487e-05,
      "loss": 0.6147,
      "step": 321
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7645407359717779,
      "learning_rate": 1.9982284264239334e-05,
      "loss": 0.6182,
      "step": 322
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.856755728488203,
      "learning_rate": 1.998199273963815e-05,
      "loss": 0.6298,
      "step": 323
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7114522605685071,
      "learning_rate": 1.9981698838114358e-05,
      "loss": 0.6129,
      "step": 324
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7839064812345653,
      "learning_rate": 1.9981402559737942e-05,
      "loss": 0.62,
      "step": 325
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7464297445278554,
      "learning_rate": 1.998110390457945e-05,
      "loss": 0.6119,
      "step": 326
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7722183769270237,
      "learning_rate": 1.9980802872709994e-05,
      "loss": 0.6304,
      "step": 327
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8392492229668477,
      "learning_rate": 1.998049946420127e-05,
      "loss": 0.6049,
      "step": 328
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8142541978392466,
      "learning_rate": 1.998019367912551e-05,
      "loss": 0.6598,
      "step": 329
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7727157511346325,
      "learning_rate": 1.9979885517555536e-05,
      "loss": 0.6277,
      "step": 330
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7695211732382756,
      "learning_rate": 1.9979574979564726e-05,
      "loss": 0.5953,
      "step": 331
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9291405716319489,
      "learning_rate": 1.997926206522703e-05,
      "loss": 0.6233,
      "step": 332
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6641789142008461,
      "learning_rate": 1.9978946774616948e-05,
      "loss": 0.6413,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8605177262749619,
      "learning_rate": 1.9978629107809572e-05,
      "loss": 0.5888,
      "step": 334
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9089308105048556,
      "learning_rate": 1.997830906488053e-05,
      "loss": 0.614,
      "step": 335
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9259032626742323,
      "learning_rate": 1.997798664590605e-05,
      "loss": 0.624,
      "step": 336
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8071106495391157,
      "learning_rate": 1.997766185096289e-05,
      "loss": 0.6307,
      "step": 337
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7022654394965974,
      "learning_rate": 1.9977334680128394e-05,
      "loss": 0.6059,
      "step": 338
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8486031527575401,
      "learning_rate": 1.9977005133480475e-05,
      "loss": 0.568,
      "step": 339
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8384133960705784,
      "learning_rate": 1.9976673211097596e-05,
      "loss": 0.6164,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7058610773098568,
      "learning_rate": 1.99763389130588e-05,
      "loss": 0.6409,
      "step": 341
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7641788049891635,
      "learning_rate": 1.9976002239443694e-05,
      "loss": 0.5973,
      "step": 342
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8419199532657943,
      "learning_rate": 1.9975663190332438e-05,
      "loss": 0.6312,
      "step": 343
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7447555546823579,
      "learning_rate": 1.9975321765805777e-05,
      "loss": 0.6321,
      "step": 344
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7027083178291814,
      "learning_rate": 1.9974977965945e-05,
      "loss": 0.6105,
      "step": 345
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6400845880934808,
      "learning_rate": 1.997463179083198e-05,
      "loss": 0.6098,
      "step": 346
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7080630913742814,
      "learning_rate": 1.9974283240549147e-05,
      "loss": 0.5983,
      "step": 347
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8559277514601893,
      "learning_rate": 1.9973932315179502e-05,
      "loss": 0.6304,
      "step": 348
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7241947574444814,
      "learning_rate": 1.99735790148066e-05,
      "loss": 0.6206,
      "step": 349
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9092739754226152,
      "learning_rate": 1.9973223339514573e-05,
      "loss": 0.6156,
      "step": 350
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1387181099778687,
      "learning_rate": 1.997286528938812e-05,
      "loss": 0.6036,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7969837829976442,
      "learning_rate": 1.9972504864512492e-05,
      "loss": 0.5943,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8599585096286595,
      "learning_rate": 1.997214206497352e-05,
      "loss": 0.6624,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8558612867615886,
      "learning_rate": 1.9971776890857587e-05,
      "loss": 0.6095,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8479424973604691,
      "learning_rate": 1.9971409342251656e-05,
      "loss": 0.6422,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7682483565760448,
      "learning_rate": 1.9971039419243246e-05,
      "loss": 0.6324,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6989677160538194,
      "learning_rate": 1.9970667121920443e-05,
      "loss": 0.6317,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7152858278629338,
      "learning_rate": 1.9970292450371896e-05,
      "loss": 0.5906,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6981524301642531,
      "learning_rate": 1.9969915404686827e-05,
      "loss": 0.6099,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7916022789662893,
      "learning_rate": 1.9969535984955012e-05,
      "loss": 0.6147,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7093801372309906,
      "learning_rate": 1.9969154191266805e-05,
      "loss": 0.6166,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7833893452723494,
      "learning_rate": 1.9968770023713123e-05,
      "loss": 0.6031,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8120149623997154,
      "learning_rate": 1.9968383482385433e-05,
      "loss": 0.6273,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7486064949330141,
      "learning_rate": 1.9967994567375783e-05,
      "loss": 0.5997,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7614018671112961,
      "learning_rate": 1.9967603278776785e-05,
      "loss": 0.6263,
      "step": 365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7534264395530595,
      "learning_rate": 1.996720961668161e-05,
      "loss": 0.6263,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7956627685316836,
      "learning_rate": 1.9966813581184002e-05,
      "loss": 0.6146,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8492976221042569,
      "learning_rate": 1.9966415172378256e-05,
      "loss": 0.6314,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7656819380296873,
      "learning_rate": 1.996601439035925e-05,
      "loss": 0.6285,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8803750215758357,
      "learning_rate": 1.9965611235222416e-05,
      "loss": 0.648,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8064978906255269,
      "learning_rate": 1.996520570706375e-05,
      "loss": 0.6576,
      "step": 371
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.694115462868371,
      "learning_rate": 1.9964797805979826e-05,
      "loss": 0.5896,
      "step": 372
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7018694537597717,
      "learning_rate": 1.9964387532067767e-05,
      "loss": 0.6286,
      "step": 373
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7439662283803894,
      "learning_rate": 1.9963974885425267e-05,
      "loss": 0.5982,
      "step": 374
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7973481038772918,
      "learning_rate": 1.9963559866150587e-05,
      "loss": 0.6051,
      "step": 375
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8158737530637662,
      "learning_rate": 1.9963142474342554e-05,
      "loss": 0.612,
      "step": 376
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.662796782075047,
      "learning_rate": 1.996272271010056e-05,
      "loss": 0.6054,
      "step": 377
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.839570424273354,
      "learning_rate": 1.9962300573524552e-05,
      "loss": 0.6163,
      "step": 378
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6696374140585627,
      "learning_rate": 1.9961876064715052e-05,
      "loss": 0.6021,
      "step": 379
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7398244507089898,
      "learning_rate": 1.9961449183773148e-05,
      "loss": 0.6312,
      "step": 380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7439564671411125,
      "learning_rate": 1.996101993080049e-05,
      "loss": 0.608,
      "step": 381
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6660003681065855,
      "learning_rate": 1.9960588305899288e-05,
      "loss": 0.6203,
      "step": 382
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7567916428422115,
      "learning_rate": 1.9960154309172325e-05,
      "loss": 0.6038,
      "step": 383
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7650347133076641,
      "learning_rate": 1.9959717940722937e-05,
      "loss": 0.625,
      "step": 384
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7178355739964113,
      "learning_rate": 1.9959279200655044e-05,
      "loss": 0.61,
      "step": 385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7620570807693864,
      "learning_rate": 1.995883808907311e-05,
      "loss": 0.6189,
      "step": 386
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7030804263098365,
      "learning_rate": 1.9958394606082176e-05,
      "loss": 0.6105,
      "step": 387
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7804518252268752,
      "learning_rate": 1.995794875178785e-05,
      "loss": 0.6343,
      "step": 388
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9197293047481968,
      "learning_rate": 1.9957500526296288e-05,
      "loss": 0.6126,
      "step": 389
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7126671443992677,
      "learning_rate": 1.9957049929714228e-05,
      "loss": 0.6298,
      "step": 390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7766309554603201,
      "learning_rate": 1.9956596962148967e-05,
      "loss": 0.5956,
      "step": 391
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8719829678077013,
      "learning_rate": 1.9956141623708373e-05,
      "loss": 0.6127,
      "step": 392
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7546888341468397,
      "learning_rate": 1.9955683914500856e-05,
      "loss": 0.6173,
      "step": 393
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1765250527253128,
      "learning_rate": 1.995522383463542e-05,
      "loss": 0.628,
      "step": 394
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7316738338356245,
      "learning_rate": 1.995476138422161e-05,
      "loss": 0.6308,
      "step": 395
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.179375682880823,
      "learning_rate": 1.995429656336955e-05,
      "loss": 0.69,
      "step": 396
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8002312734875234,
      "learning_rate": 1.9953829372189927e-05,
      "loss": 0.6135,
      "step": 397
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7861816299307243,
      "learning_rate": 1.995335981079398e-05,
      "loss": 0.6286,
      "step": 398
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8096239728102159,
      "learning_rate": 1.995288787929353e-05,
      "loss": 0.6323,
      "step": 399
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6496954037823662,
      "learning_rate": 1.9952413577800948e-05,
      "loss": 0.5861,
      "step": 400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.751898528550087,
      "learning_rate": 1.9951936906429177e-05,
      "loss": 0.6291,
      "step": 401
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7719506410821964,
      "learning_rate": 1.9951457865291724e-05,
      "loss": 0.6071,
      "step": 402
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8300039418444127,
      "learning_rate": 1.995097645450266e-05,
      "loss": 0.6179,
      "step": 403
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7106158259241128,
      "learning_rate": 1.995049267417662e-05,
      "loss": 0.6275,
      "step": 404
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7699885269119395,
      "learning_rate": 1.9950006524428794e-05,
      "loss": 0.6252,
      "step": 405
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.694662805723059,
      "learning_rate": 1.994951800537495e-05,
      "loss": 0.6128,
      "step": 406
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7660491438669541,
      "learning_rate": 1.9949027117131415e-05,
      "loss": 0.6399,
      "step": 407
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.729834079269708,
      "learning_rate": 1.994853385981508e-05,
      "loss": 0.6111,
      "step": 408
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6639500185547685,
      "learning_rate": 1.9948038233543396e-05,
      "loss": 0.6356,
      "step": 409
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7591212028191443,
      "learning_rate": 1.994754023843439e-05,
      "loss": 0.6378,
      "step": 410
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8321955131241071,
      "learning_rate": 1.9947039874606635e-05,
      "loss": 0.6019,
      "step": 411
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.750330813088129,
      "learning_rate": 1.994653714217929e-05,
      "loss": 0.6056,
      "step": 412
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8905307425983428,
      "learning_rate": 1.9946032041272052e-05,
      "loss": 0.6403,
      "step": 413
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7824786724127989,
      "learning_rate": 1.9945524572005208e-05,
      "loss": 0.6087,
      "step": 414
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9075499016716488,
      "learning_rate": 1.9945014734499594e-05,
      "loss": 0.6321,
      "step": 415
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7708285678854071,
      "learning_rate": 1.994450252887661e-05,
      "loss": 0.6156,
      "step": 416
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7983714837028923,
      "learning_rate": 1.9943987955258227e-05,
      "loss": 0.6145,
      "step": 417
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8528161211494226,
      "learning_rate": 1.994347101376697e-05,
      "loss": 0.6168,
      "step": 418
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7962771058299046,
      "learning_rate": 1.994295170452594e-05,
      "loss": 0.6203,
      "step": 419
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5645118105540998,
      "learning_rate": 1.9942430027658793e-05,
      "loss": 0.6272,
      "step": 420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7617339042346049,
      "learning_rate": 1.994190598328975e-05,
      "loss": 0.6349,
      "step": 421
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8475718683110381,
      "learning_rate": 1.9941379571543597e-05,
      "loss": 0.628,
      "step": 422
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7768413203310208,
      "learning_rate": 1.9940850792545687e-05,
      "loss": 0.5946,
      "step": 423
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7369575426657241,
      "learning_rate": 1.994031964642193e-05,
      "loss": 0.6191,
      "step": 424
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8025277507611599,
      "learning_rate": 1.9939786133298805e-05,
      "loss": 0.6039,
      "step": 425
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8606446991419747,
      "learning_rate": 1.9939250253303354e-05,
      "loss": 0.6029,
      "step": 426
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7722210048598722,
      "learning_rate": 1.9938712006563176e-05,
      "loss": 0.6329,
      "step": 427
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6821072781181471,
      "learning_rate": 1.993817139320644e-05,
      "loss": 0.611,
      "step": 428
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8201083638492951,
      "learning_rate": 1.9937628413361884e-05,
      "loss": 0.6142,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8820423418401047,
      "learning_rate": 1.99370830671588e-05,
      "loss": 0.6231,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7695302661503748,
      "learning_rate": 1.9936535354727038e-05,
      "loss": 0.6162,
      "step": 431
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.96501159920692,
      "learning_rate": 1.9935985276197033e-05,
      "loss": 0.6335,
      "step": 432
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7853817103698107,
      "learning_rate": 1.993543283169976e-05,
      "loss": 0.6293,
      "step": 433
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8217907959395888,
      "learning_rate": 1.9934878021366773e-05,
      "loss": 0.6185,
      "step": 434
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7679838570536166,
      "learning_rate": 1.9934320845330185e-05,
      "loss": 0.6069,
      "step": 435
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7831743785063524,
      "learning_rate": 1.9933761303722666e-05,
      "loss": 0.5884,
      "step": 436
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7881310158892244,
      "learning_rate": 1.9933199396677457e-05,
      "loss": 0.6209,
      "step": 437
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4016991003061543,
      "learning_rate": 1.9932635124328365e-05,
      "loss": 0.7139,
      "step": 438
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.713458587170905,
      "learning_rate": 1.9932068486809748e-05,
      "loss": 0.6137,
      "step": 439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7631275905488909,
      "learning_rate": 1.9931499484256537e-05,
      "loss": 0.6159,
      "step": 440
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7899524321202206,
      "learning_rate": 1.9930928116804223e-05,
      "loss": 0.5924,
      "step": 441
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8424663406549892,
      "learning_rate": 1.9930354384588864e-05,
      "loss": 0.622,
      "step": 442
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7805131051492997,
      "learning_rate": 1.9929778287747074e-05,
      "loss": 0.5969,
      "step": 443
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7380329624898354,
      "learning_rate": 1.9929199826416033e-05,
      "loss": 0.6126,
      "step": 444
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7634814754370074,
      "learning_rate": 1.9928619000733486e-05,
      "loss": 0.6037,
      "step": 445
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7476410446745763,
      "learning_rate": 1.992803581083774e-05,
      "loss": 0.5821,
      "step": 446
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7838353871384552,
      "learning_rate": 1.992745025686767e-05,
      "loss": 0.6213,
      "step": 447
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8261841527203132,
      "learning_rate": 1.99268623389627e-05,
      "loss": 0.6586,
      "step": 448
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7238885921338609,
      "learning_rate": 1.9926272057262828e-05,
      "loss": 0.6281,
      "step": 449
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7810615066379085,
      "learning_rate": 1.9925679411908616e-05,
      "loss": 0.5833,
      "step": 450
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.775564839618241,
      "learning_rate": 1.9925084403041183e-05,
      "loss": 0.6241,
      "step": 451
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9270491700244067,
      "learning_rate": 1.9924487030802214e-05,
      "loss": 0.6254,
      "step": 452
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7322965800680568,
      "learning_rate": 1.9923887295333956e-05,
      "loss": 0.6274,
      "step": 453
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9028043759874168,
      "learning_rate": 1.9923285196779217e-05,
      "loss": 0.6117,
      "step": 454
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7779859621074109,
      "learning_rate": 1.992268073528137e-05,
      "loss": 0.614,
      "step": 455
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8376192549016361,
      "learning_rate": 1.992207391098435e-05,
      "loss": 0.6487,
      "step": 456
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7531144459938007,
      "learning_rate": 1.9921464724032654e-05,
      "loss": 0.5927,
      "step": 457
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9032713841743735,
      "learning_rate": 1.9920853174571345e-05,
      "loss": 0.6441,
      "step": 458
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6818167480531648,
      "learning_rate": 1.9920239262746045e-05,
      "loss": 0.6264,
      "step": 459
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7337307322605258,
      "learning_rate": 1.9919622988702936e-05,
      "loss": 0.5907,
      "step": 460
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6958335068075395,
      "learning_rate": 1.9919004352588768e-05,
      "loss": 0.6222,
      "step": 461
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7255287504586757,
      "learning_rate": 1.9918383354550857e-05,
      "loss": 0.6186,
      "step": 462
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7250190587297178,
      "learning_rate": 1.9917759994737064e-05,
      "loss": 0.6197,
      "step": 463
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0841101450178388,
      "learning_rate": 1.9917134273295833e-05,
      "loss": 0.653,
      "step": 464
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6629787035816725,
      "learning_rate": 1.991650619037616e-05,
      "loss": 0.5691,
      "step": 465
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7768922476590833,
      "learning_rate": 1.9915875746127598e-05,
      "loss": 0.615,
      "step": 466
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7482566488952391,
      "learning_rate": 1.9915242940700277e-05,
      "loss": 0.6266,
      "step": 467
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7910336956944817,
      "learning_rate": 1.9914607774244882e-05,
      "loss": 0.6007,
      "step": 468
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8459265397758001,
      "learning_rate": 1.9913970246912653e-05,
      "loss": 0.6447,
      "step": 469
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8600461035685132,
      "learning_rate": 1.99133303588554e-05,
      "loss": 0.6082,
      "step": 470
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8281950292710593,
      "learning_rate": 1.99126881102255e-05,
      "loss": 0.5986,
      "step": 471
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6720905002700042,
      "learning_rate": 1.9912043501175882e-05,
      "loss": 0.6022,
      "step": 472
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7222091262774799,
      "learning_rate": 1.9911396531860037e-05,
      "loss": 0.6172,
      "step": 473
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7719831768637111,
      "learning_rate": 1.991074720243203e-05,
      "loss": 0.6241,
      "step": 474
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7700368249651243,
      "learning_rate": 1.991009551304647e-05,
      "loss": 0.6029,
      "step": 475
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.188632339458289,
      "learning_rate": 1.990944146385855e-05,
      "loss": 0.7134,
      "step": 476
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8320582357723095,
      "learning_rate": 1.9908785055024003e-05,
      "loss": 0.6565,
      "step": 477
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.72244594761386,
      "learning_rate": 1.9908126286699136e-05,
      "loss": 0.5832,
      "step": 478
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7416595574271287,
      "learning_rate": 1.990746515904082e-05,
      "loss": 0.6134,
      "step": 479
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.760897365999147,
      "learning_rate": 1.990680167220648e-05,
      "loss": 0.6614,
      "step": 480
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7080075341693459,
      "learning_rate": 1.9906135826354106e-05,
      "loss": 0.5896,
      "step": 481
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.775509699222339,
      "learning_rate": 1.9905467621642248e-05,
      "loss": 0.6278,
      "step": 482
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7355755647949731,
      "learning_rate": 1.9904797058230027e-05,
      "loss": 0.6117,
      "step": 483
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4330077575816556,
      "learning_rate": 1.990412413627711e-05,
      "loss": 0.6598,
      "step": 484
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.310688777743281,
      "learning_rate": 1.990344885594374e-05,
      "loss": 0.7049,
      "step": 485
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6250572047782004,
      "learning_rate": 1.9902771217390707e-05,
      "loss": 0.6058,
      "step": 486
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3087350572372987,
      "learning_rate": 1.9902091220779384e-05,
      "loss": 0.6522,
      "step": 487
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3913101848989538,
      "learning_rate": 1.990140886627168e-05,
      "loss": 0.6388,
      "step": 488
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9553686281073248,
      "learning_rate": 1.990072415403008e-05,
      "loss": 0.6283,
      "step": 489
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9623306016337845,
      "learning_rate": 1.9900037084217637e-05,
      "loss": 0.6632,
      "step": 490
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8442593847994329,
      "learning_rate": 1.989934765699795e-05,
      "loss": 0.6337,
      "step": 491
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7453844241373188,
      "learning_rate": 1.989865587253519e-05,
      "loss": 0.636,
      "step": 492
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7732323388920992,
      "learning_rate": 1.9897961730994084e-05,
      "loss": 0.6355,
      "step": 493
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2323219754172092,
      "learning_rate": 1.9897265232539917e-05,
      "loss": 0.6582,
      "step": 494
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9857914314799552,
      "learning_rate": 1.9896566377338544e-05,
      "loss": 0.6837,
      "step": 495
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7312808315745551,
      "learning_rate": 1.9895865165556375e-05,
      "loss": 0.6507,
      "step": 496
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8719980334896998,
      "learning_rate": 1.9895161597360386e-05,
      "loss": 0.6518,
      "step": 497
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7116284505330697,
      "learning_rate": 1.9894455672918113e-05,
      "loss": 0.6478,
      "step": 498
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8466536300173176,
      "learning_rate": 1.9893747392397643e-05,
      "loss": 0.6352,
      "step": 499
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8775965271118692,
      "learning_rate": 1.9893036755967645e-05,
      "loss": 0.6272,
      "step": 500
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8487671081783965,
      "learning_rate": 1.989232376379733e-05,
      "loss": 0.6102,
      "step": 501
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7937740513046948,
      "learning_rate": 1.989160841605647e-05,
      "loss": 0.6117,
      "step": 502
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9064712378921003,
      "learning_rate": 1.9890890712915416e-05,
      "loss": 0.6341,
      "step": 503
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8903013955411859,
      "learning_rate": 1.989017065454506e-05,
      "loss": 0.6453,
      "step": 504
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7237962308013638,
      "learning_rate": 1.9889448241116867e-05,
      "loss": 0.6144,
      "step": 505
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7008827524481582,
      "learning_rate": 1.988872347280286e-05,
      "loss": 0.5851,
      "step": 506
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7902375961805486,
      "learning_rate": 1.9887996349775617e-05,
      "loss": 0.6526,
      "step": 507
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6982063272019828,
      "learning_rate": 1.9887266872208283e-05,
      "loss": 0.6183,
      "step": 508
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7053831577220021,
      "learning_rate": 1.9886535040274562e-05,
      "loss": 0.5883,
      "step": 509
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7379172638425741,
      "learning_rate": 1.988580085414872e-05,
      "loss": 0.6214,
      "step": 510
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.317450764107576,
      "learning_rate": 1.988506431400558e-05,
      "loss": 0.6321,
      "step": 511
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8448144710123224,
      "learning_rate": 1.9884325420020535e-05,
      "loss": 0.6251,
      "step": 512
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8211652368016493,
      "learning_rate": 1.988358417236952e-05,
      "loss": 0.5979,
      "step": 513
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.999224564624112,
      "learning_rate": 1.988284057122905e-05,
      "loss": 0.6662,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8422625688830921,
      "learning_rate": 1.9882094616776186e-05,
      "loss": 0.6267,
      "step": 515
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6787524335320186,
      "learning_rate": 1.9881346309188562e-05,
      "loss": 0.6117,
      "step": 516
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.240133589080367,
      "learning_rate": 1.988059564864436e-05,
      "loss": 0.6385,
      "step": 517
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1753762284272034,
      "learning_rate": 1.987984263532233e-05,
      "loss": 0.6391,
      "step": 518
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.690098551318456,
      "learning_rate": 1.9879087269401782e-05,
      "loss": 0.6215,
      "step": 519
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7068060756862515,
      "learning_rate": 1.987832955106258e-05,
      "loss": 0.6207,
      "step": 520
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.918671259507849,
      "learning_rate": 1.987756948048516e-05,
      "loss": 0.6217,
      "step": 521
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6276841189177624,
      "learning_rate": 1.9876807057850505e-05,
      "loss": 0.6317,
      "step": 522
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7051063273974869,
      "learning_rate": 1.987604228334017e-05,
      "loss": 0.6304,
      "step": 523
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7257627379552473,
      "learning_rate": 1.9875275157136254e-05,
      "loss": 0.6195,
      "step": 524
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6929527567258855,
      "learning_rate": 1.9874505679421435e-05,
      "loss": 0.6186,
      "step": 525
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9127466379645698,
      "learning_rate": 1.9873733850378937e-05,
      "loss": 0.6326,
      "step": 526
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7735226787651531,
      "learning_rate": 1.9872959670192548e-05,
      "loss": 0.619,
      "step": 527
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6987343646649199,
      "learning_rate": 1.987218313904662e-05,
      "loss": 0.6225,
      "step": 528
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0040307587756636,
      "learning_rate": 1.9871404257126062e-05,
      "loss": 0.6643,
      "step": 529
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6762205282494035,
      "learning_rate": 1.9870623024616338e-05,
      "loss": 0.6343,
      "step": 530
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7175181399823198,
      "learning_rate": 1.9869839441703478e-05,
      "loss": 0.6777,
      "step": 531
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.615450109513188,
      "learning_rate": 1.9869053508574073e-05,
      "loss": 0.6031,
      "step": 532
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6502611991632854,
      "learning_rate": 1.9868265225415263e-05,
      "loss": 0.6082,
      "step": 533
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8071788445474578,
      "learning_rate": 1.986747459241476e-05,
      "loss": 0.6256,
      "step": 534
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7128543409186496,
      "learning_rate": 1.986668160976083e-05,
      "loss": 0.6273,
      "step": 535
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7253534458575437,
      "learning_rate": 1.9865886277642297e-05,
      "loss": 0.629,
      "step": 536
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6712841191160509,
      "learning_rate": 1.9865088596248546e-05,
      "loss": 0.6192,
      "step": 537
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6644987823452454,
      "learning_rate": 1.9864288565769527e-05,
      "loss": 0.608,
      "step": 538
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0969378668835663,
      "learning_rate": 1.9863486186395738e-05,
      "loss": 0.6538,
      "step": 539
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7961628126385569,
      "learning_rate": 1.9862681458318245e-05,
      "loss": 0.6189,
      "step": 540
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8176566372460758,
      "learning_rate": 1.9861874381728668e-05,
      "loss": 0.6266,
      "step": 541
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7129282195554545,
      "learning_rate": 1.9861064956819195e-05,
      "loss": 0.6186,
      "step": 542
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7727481785576014,
      "learning_rate": 1.986025318378256e-05,
      "loss": 0.63,
      "step": 543
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8756336769336365,
      "learning_rate": 1.985943906281207e-05,
      "loss": 0.6733,
      "step": 544
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.713886794070528,
      "learning_rate": 1.9858622594101574e-05,
      "loss": 0.6114,
      "step": 545
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0528049413543323,
      "learning_rate": 1.9857803777845507e-05,
      "loss": 0.6611,
      "step": 546
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7211652473936138,
      "learning_rate": 1.9856982614238827e-05,
      "loss": 0.5943,
      "step": 547
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8737556438703004,
      "learning_rate": 1.9856159103477085e-05,
      "loss": 0.625,
      "step": 548
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6785963358049218,
      "learning_rate": 1.985533324575637e-05,
      "loss": 0.6027,
      "step": 549
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.751222066398309,
      "learning_rate": 1.985450504127334e-05,
      "loss": 0.5998,
      "step": 550
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.781126792922416,
      "learning_rate": 1.9853674490225202e-05,
      "loss": 0.626,
      "step": 551
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.746320977067496,
      "learning_rate": 1.9852841592809736e-05,
      "loss": 0.6149,
      "step": 552
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6483201634758485,
      "learning_rate": 1.9852006349225262e-05,
      "loss": 0.6181,
      "step": 553
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7580758887158079,
      "learning_rate": 1.9851168759670677e-05,
      "loss": 0.6481,
      "step": 554
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8155247441696265,
      "learning_rate": 1.9850328824345422e-05,
      "loss": 0.6225,
      "step": 555
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7608567095988,
      "learning_rate": 1.9849486543449514e-05,
      "loss": 0.656,
      "step": 556
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7837987467976344,
      "learning_rate": 1.984864191718351e-05,
      "loss": 0.6295,
      "step": 557
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8733451150183376,
      "learning_rate": 1.9847794945748532e-05,
      "loss": 0.6128,
      "step": 558
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8483294163729141,
      "learning_rate": 1.9846945629346264e-05,
      "loss": 0.6235,
      "step": 559
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8457299320487791,
      "learning_rate": 1.9846093968178947e-05,
      "loss": 0.6315,
      "step": 560
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7971057159111398,
      "learning_rate": 1.984523996244938e-05,
      "loss": 0.6167,
      "step": 561
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0774367446050805,
      "learning_rate": 1.9844383612360917e-05,
      "loss": 0.6397,
      "step": 562
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6755331012115962,
      "learning_rate": 1.9843524918117476e-05,
      "loss": 0.6027,
      "step": 563
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.714958648721034,
      "learning_rate": 1.984266387992353e-05,
      "loss": 0.624,
      "step": 564
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6926673783837005,
      "learning_rate": 1.9841800497984107e-05,
      "loss": 0.5833,
      "step": 565
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.680254586119293,
      "learning_rate": 1.98409347725048e-05,
      "loss": 0.5862,
      "step": 566
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8168172293361394,
      "learning_rate": 1.984006670369175e-05,
      "loss": 0.6323,
      "step": 567
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7120685436589537,
      "learning_rate": 1.9839196291751673e-05,
      "loss": 0.6356,
      "step": 568
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7329639325181813,
      "learning_rate": 1.9838323536891826e-05,
      "loss": 0.5966,
      "step": 569
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.782911378778992,
      "learning_rate": 1.9837448439320027e-05,
      "loss": 0.6205,
      "step": 570
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7459628840569184,
      "learning_rate": 1.9836570999244667e-05,
      "loss": 0.6387,
      "step": 571
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7709216076694532,
      "learning_rate": 1.9835691216874667e-05,
      "loss": 0.6046,
      "step": 572
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8033966056268187,
      "learning_rate": 1.9834809092419534e-05,
      "loss": 0.6233,
      "step": 573
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7232733461062751,
      "learning_rate": 1.9833924626089315e-05,
      "loss": 0.5996,
      "step": 574
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8520059712447654,
      "learning_rate": 1.9833037818094625e-05,
      "loss": 0.6286,
      "step": 575
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7321442613964699,
      "learning_rate": 1.9832148668646623e-05,
      "loss": 0.6289,
      "step": 576
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8068649765638398,
      "learning_rate": 1.9831257177957045e-05,
      "loss": 0.6151,
      "step": 577
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6828852697589847,
      "learning_rate": 1.9830363346238163e-05,
      "loss": 0.6109,
      "step": 578
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8080561191569116,
      "learning_rate": 1.9829467173702825e-05,
      "loss": 0.6423,
      "step": 579
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7091896632128074,
      "learning_rate": 1.982856866056442e-05,
      "loss": 0.604,
      "step": 580
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.692620009164206,
      "learning_rate": 1.9827667807036915e-05,
      "loss": 0.618,
      "step": 581
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7370231951031447,
      "learning_rate": 1.9826764613334815e-05,
      "loss": 0.6154,
      "step": 582
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7655496769860747,
      "learning_rate": 1.982585907967319e-05,
      "loss": 0.5981,
      "step": 583
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0732460075258394,
      "learning_rate": 1.9824951206267668e-05,
      "loss": 0.6392,
      "step": 584
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0340057467428665,
      "learning_rate": 1.9824040993334427e-05,
      "loss": 0.6349,
      "step": 585
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.74233824452199,
      "learning_rate": 1.9823128441090215e-05,
      "loss": 0.5985,
      "step": 586
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.860239878598874,
      "learning_rate": 1.9822213549752325e-05,
      "loss": 0.646,
      "step": 587
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7273829855969589,
      "learning_rate": 1.9821296319538617e-05,
      "loss": 0.621,
      "step": 588
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7349641229883287,
      "learning_rate": 1.9820376750667498e-05,
      "loss": 0.6104,
      "step": 589
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.764957869072944,
      "learning_rate": 1.9819454843357942e-05,
      "loss": 0.5871,
      "step": 590
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8452896094106164,
      "learning_rate": 1.981853059782947e-05,
      "loss": 0.6424,
      "step": 591
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.293488157989972,
      "learning_rate": 1.9817604014302164e-05,
      "loss": 0.6254,
      "step": 592
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.755116488423172,
      "learning_rate": 1.9816675092996663e-05,
      "loss": 0.6112,
      "step": 593
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7685013618777834,
      "learning_rate": 1.981574383413417e-05,
      "loss": 0.5945,
      "step": 594
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.825384656492738,
      "learning_rate": 1.981481023793643e-05,
      "loss": 0.6077,
      "step": 595
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7572913662753514,
      "learning_rate": 1.981387430462575e-05,
      "loss": 0.6159,
      "step": 596
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8550996067702907,
      "learning_rate": 1.9812936034425e-05,
      "loss": 0.7185,
      "step": 597
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6286423442983022,
      "learning_rate": 1.9811995427557602e-05,
      "loss": 0.6189,
      "step": 598
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.511840044777551,
      "learning_rate": 1.9811052484247533e-05,
      "loss": 0.6496,
      "step": 599
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7282121279034152,
      "learning_rate": 1.981010720471933e-05,
      "loss": 0.6307,
      "step": 600
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.082825142699373,
      "learning_rate": 1.9809159589198078e-05,
      "loss": 0.6286,
      "step": 601
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1924136448794806,
      "learning_rate": 1.980820963790943e-05,
      "loss": 0.7063,
      "step": 602
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8149988580836437,
      "learning_rate": 1.9807257351079586e-05,
      "loss": 0.6161,
      "step": 603
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7221725143498796,
      "learning_rate": 1.980630272893531e-05,
      "loss": 0.5991,
      "step": 604
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6809092851974986,
      "learning_rate": 1.9805345771703913e-05,
      "loss": 0.6124,
      "step": 605
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0299173385392675,
      "learning_rate": 1.9804386479613268e-05,
      "loss": 0.652,
      "step": 606
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7195334801747982,
      "learning_rate": 1.9803424852891803e-05,
      "loss": 0.6339,
      "step": 607
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7638417340593627,
      "learning_rate": 1.9802460891768504e-05,
      "loss": 0.6465,
      "step": 608
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2821207582770813,
      "learning_rate": 1.9801494596472904e-05,
      "loss": 0.6883,
      "step": 609
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0569275742673798,
      "learning_rate": 1.9800525967235104e-05,
      "loss": 0.6268,
      "step": 610
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.870862845568111,
      "learning_rate": 1.9799555004285752e-05,
      "loss": 0.6727,
      "step": 611
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7385480441474603,
      "learning_rate": 1.9798581707856058e-05,
      "loss": 0.5903,
      "step": 612
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7485435723338446,
      "learning_rate": 1.979760607817778e-05,
      "loss": 0.6327,
      "step": 613
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8259847672620604,
      "learning_rate": 1.979662811548324e-05,
      "loss": 0.6191,
      "step": 614
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7401955559251102,
      "learning_rate": 1.9795647820005308e-05,
      "loss": 0.6138,
      "step": 615
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7098673372424309,
      "learning_rate": 1.979466519197742e-05,
      "loss": 0.6172,
      "step": 616
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0129393016675496,
      "learning_rate": 1.979368023163355e-05,
      "loss": 0.6981,
      "step": 617
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.737383176122977,
      "learning_rate": 1.979269293920824e-05,
      "loss": 0.6308,
      "step": 618
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8824868371859269,
      "learning_rate": 1.979170331493659e-05,
      "loss": 0.6324,
      "step": 619
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6886831782301382,
      "learning_rate": 1.9790711359054254e-05,
      "loss": 0.592,
      "step": 620
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7366007473477361,
      "learning_rate": 1.9789717071797428e-05,
      "loss": 0.6224,
      "step": 621
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7087804276660592,
      "learning_rate": 1.9788720453402876e-05,
      "loss": 0.6147,
      "step": 622
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7433196795879241,
      "learning_rate": 1.9787721504107916e-05,
      "loss": 0.6207,
      "step": 623
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4318537851001878,
      "learning_rate": 1.9786720224150416e-05,
      "loss": 0.6979,
      "step": 624
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8911838377087027,
      "learning_rate": 1.9785716613768805e-05,
      "loss": 0.6663,
      "step": 625
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8403404345640849,
      "learning_rate": 1.978471067320206e-05,
      "loss": 0.6332,
      "step": 626
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6898410319296178,
      "learning_rate": 1.978370240268972e-05,
      "loss": 0.6143,
      "step": 627
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7584695099561513,
      "learning_rate": 1.9782691802471874e-05,
      "loss": 0.6167,
      "step": 628
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.840238749584161,
      "learning_rate": 1.9781678872789166e-05,
      "loss": 0.6241,
      "step": 629
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7857469456848938,
      "learning_rate": 1.97806636138828e-05,
      "loss": 0.6087,
      "step": 630
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6917053182619627,
      "learning_rate": 1.9779646025994525e-05,
      "loss": 0.6187,
      "step": 631
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8261976742162516,
      "learning_rate": 1.9778626109366653e-05,
      "loss": 0.6141,
      "step": 632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1073411046550887,
      "learning_rate": 1.9777603864242048e-05,
      "loss": 0.7079,
      "step": 633
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7646778481063056,
      "learning_rate": 1.9776579290864127e-05,
      "loss": 0.6289,
      "step": 634
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6621276638795526,
      "learning_rate": 1.9775552389476865e-05,
      "loss": 0.5898,
      "step": 635
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7220281782858642,
      "learning_rate": 1.9774523160324785e-05,
      "loss": 0.6334,
      "step": 636
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9944381108726549,
      "learning_rate": 1.977349160365297e-05,
      "loss": 0.6257,
      "step": 637
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7567568056274979,
      "learning_rate": 1.9772457719707053e-05,
      "loss": 0.638,
      "step": 638
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7263124311438787,
      "learning_rate": 1.977142150873323e-05,
      "loss": 0.6178,
      "step": 639
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7982284184849164,
      "learning_rate": 1.9770382970978235e-05,
      "loss": 0.6342,
      "step": 640
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.697154084107242,
      "learning_rate": 1.9769342106689375e-05,
      "loss": 0.6179,
      "step": 641
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.841471247452727,
      "learning_rate": 1.9768298916114498e-05,
      "loss": 0.6016,
      "step": 642
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8292682253198131,
      "learning_rate": 1.976725339950201e-05,
      "loss": 0.6209,
      "step": 643
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7302268599259572,
      "learning_rate": 1.976620555710087e-05,
      "loss": 0.6167,
      "step": 644
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7749710777653257,
      "learning_rate": 1.976515538916059e-05,
      "loss": 0.6553,
      "step": 645
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8567450677152005,
      "learning_rate": 1.976410289593124e-05,
      "loss": 0.6169,
      "step": 646
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7058299272494912,
      "learning_rate": 1.9763048077663436e-05,
      "loss": 0.6398,
      "step": 647
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7959715584411231,
      "learning_rate": 1.9761990934608355e-05,
      "loss": 0.643,
      "step": 648
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7081660502154444,
      "learning_rate": 1.9760931467017728e-05,
      "loss": 0.5991,
      "step": 649
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6784641971394779,
      "learning_rate": 1.9759869675143835e-05,
      "loss": 0.6167,
      "step": 650
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7263515045587797,
      "learning_rate": 1.9758805559239512e-05,
      "loss": 0.6328,
      "step": 651
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7365390083129658,
      "learning_rate": 1.975773911955814e-05,
      "loss": 0.6206,
      "step": 652
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8781559879608727,
      "learning_rate": 1.975667035635367e-05,
      "loss": 0.6346,
      "step": 653
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6798200236113251,
      "learning_rate": 1.9755599269880596e-05,
      "loss": 0.6007,
      "step": 654
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7739525287385178,
      "learning_rate": 1.9754525860393957e-05,
      "loss": 0.6143,
      "step": 655
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7212829342232073,
      "learning_rate": 1.9753450128149366e-05,
      "loss": 0.6062,
      "step": 656
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2212786822161092,
      "learning_rate": 1.975237207340297e-05,
      "loss": 0.6336,
      "step": 657
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8206557761982326,
      "learning_rate": 1.975129169641148e-05,
      "loss": 0.6027,
      "step": 658
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7008621602855821,
      "learning_rate": 1.9750208997432156e-05,
      "loss": 0.5935,
      "step": 659
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7470545760755769,
      "learning_rate": 1.974912397672281e-05,
      "loss": 0.642,
      "step": 660
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7282815571795552,
      "learning_rate": 1.974803663454181e-05,
      "loss": 0.621,
      "step": 661
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7908600073614778,
      "learning_rate": 1.9746946971148073e-05,
      "loss": 0.6346,
      "step": 662
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7714386235008999,
      "learning_rate": 1.9745854986801076e-05,
      "loss": 0.6231,
      "step": 663
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7151092431935803,
      "learning_rate": 1.9744760681760832e-05,
      "loss": 0.6269,
      "step": 664
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7089141470710774,
      "learning_rate": 1.974366405628793e-05,
      "loss": 0.5956,
      "step": 665
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7510881992714112,
      "learning_rate": 1.9742565110643496e-05,
      "loss": 0.6006,
      "step": 666
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7265808290889159,
      "learning_rate": 1.974146384508921e-05,
      "loss": 0.6146,
      "step": 667
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7861620492576713,
      "learning_rate": 1.9740360259887312e-05,
      "loss": 0.6318,
      "step": 668
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7612390065381435,
      "learning_rate": 1.973925435530058e-05,
      "loss": 0.6269,
      "step": 669
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.746680996209593,
      "learning_rate": 1.973814613159236e-05,
      "loss": 0.624,
      "step": 670
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7834929330382654,
      "learning_rate": 1.9737035589026535e-05,
      "loss": 0.6444,
      "step": 671
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7672804105426552,
      "learning_rate": 1.9735922727867564e-05,
      "loss": 0.6248,
      "step": 672
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7485410913788418,
      "learning_rate": 1.9734807548380428e-05,
      "loss": 0.6249,
      "step": 673
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6977454587610082,
      "learning_rate": 1.9733690050830684e-05,
      "loss": 0.6159,
      "step": 674
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8404124487560702,
      "learning_rate": 1.9732570235484422e-05,
      "loss": 0.6011,
      "step": 675
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7800407810963915,
      "learning_rate": 1.9731448102608308e-05,
      "loss": 0.6515,
      "step": 676
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7280617379097433,
      "learning_rate": 1.9730323652469526e-05,
      "loss": 0.6138,
      "step": 677
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7067882807103518,
      "learning_rate": 1.972919688533585e-05,
      "loss": 0.5777,
      "step": 678
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7237005479831743,
      "learning_rate": 1.9728067801475576e-05,
      "loss": 0.6076,
      "step": 679
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7156361134606002,
      "learning_rate": 1.972693640115757e-05,
      "loss": 0.6055,
      "step": 680
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8858488037746489,
      "learning_rate": 1.9725802684651235e-05,
      "loss": 0.6294,
      "step": 681
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.727102616433886,
      "learning_rate": 1.9724666652226535e-05,
      "loss": 0.6216,
      "step": 682
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.853589336252682,
      "learning_rate": 1.9723528304153985e-05,
      "loss": 0.6267,
      "step": 683
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7423905377657909,
      "learning_rate": 1.9722387640704646e-05,
      "loss": 0.6014,
      "step": 684
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9358875818447169,
      "learning_rate": 1.972124466215014e-05,
      "loss": 0.6385,
      "step": 685
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8765213736135155,
      "learning_rate": 1.9720099368762633e-05,
      "loss": 0.5919,
      "step": 686
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8062780100347243,
      "learning_rate": 1.9718951760814837e-05,
      "loss": 0.627,
      "step": 687
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9739378360082319,
      "learning_rate": 1.9717801838580028e-05,
      "loss": 0.6183,
      "step": 688
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7670052624945201,
      "learning_rate": 1.9716649602332025e-05,
      "loss": 0.6316,
      "step": 689
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7760229437609637,
      "learning_rate": 1.9715495052345197e-05,
      "loss": 0.6053,
      "step": 690
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8485873186813222,
      "learning_rate": 1.9714338188894472e-05,
      "loss": 0.5803,
      "step": 691
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.793195019623452,
      "learning_rate": 1.971317901225532e-05,
      "loss": 0.6591,
      "step": 692
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7254035639517777,
      "learning_rate": 1.9712017522703764e-05,
      "loss": 0.6063,
      "step": 693
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8294899989708288,
      "learning_rate": 1.9710853720516382e-05,
      "loss": 0.6461,
      "step": 694
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7633439159924648,
      "learning_rate": 1.97096876059703e-05,
      "loss": 0.6305,
      "step": 695
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8217067709753798,
      "learning_rate": 1.9708519179343192e-05,
      "loss": 0.6285,
      "step": 696
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8275771631330927,
      "learning_rate": 1.970734844091329e-05,
      "loss": 0.6245,
      "step": 697
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7340224530922668,
      "learning_rate": 1.9706175390959362e-05,
      "loss": 0.6104,
      "step": 698
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8226487040081369,
      "learning_rate": 1.9705000029760745e-05,
      "loss": 0.6244,
      "step": 699
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6519047069659184,
      "learning_rate": 1.9703822357597313e-05,
      "loss": 0.5852,
      "step": 700
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8658905227892336,
      "learning_rate": 1.9702642374749495e-05,
      "loss": 0.6098,
      "step": 701
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6643955771654639,
      "learning_rate": 1.970146008149827e-05,
      "loss": 0.5884,
      "step": 702
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8193392649209864,
      "learning_rate": 1.970027547812517e-05,
      "loss": 0.6338,
      "step": 703
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6775722236760872,
      "learning_rate": 1.9699088564912266e-05,
      "loss": 0.6199,
      "step": 704
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7614086567378746,
      "learning_rate": 1.9697899342142195e-05,
      "loss": 0.6222,
      "step": 705
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7472604570309049,
      "learning_rate": 1.9696707810098133e-05,
      "loss": 0.6308,
      "step": 706
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7435528706319602,
      "learning_rate": 1.969551396906381e-05,
      "loss": 0.5971,
      "step": 707
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6647641058039578,
      "learning_rate": 1.96943178193235e-05,
      "loss": 0.5957,
      "step": 708
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6831301080016964,
      "learning_rate": 1.969311936116204e-05,
      "loss": 0.614,
      "step": 709
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.02445824186703,
      "learning_rate": 1.9691918594864798e-05,
      "loss": 0.6253,
      "step": 710
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.821388608637812,
      "learning_rate": 1.969071552071771e-05,
      "loss": 0.6615,
      "step": 711
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6894675046285522,
      "learning_rate": 1.968951013900725e-05,
      "loss": 0.6133,
      "step": 712
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7243528752683165,
      "learning_rate": 1.9688302450020448e-05,
      "loss": 0.629,
      "step": 713
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7736497644475002,
      "learning_rate": 1.968709245404487e-05,
      "loss": 0.6337,
      "step": 714
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.671640329618698,
      "learning_rate": 1.9685880151368653e-05,
      "loss": 0.6052,
      "step": 715
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7888151954912176,
      "learning_rate": 1.9684665542280466e-05,
      "loss": 0.6246,
      "step": 716
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7721634677760203,
      "learning_rate": 1.9683448627069536e-05,
      "loss": 0.6321,
      "step": 717
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7497528861201961,
      "learning_rate": 1.9682229406025635e-05,
      "loss": 0.6165,
      "step": 718
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8129147282066777,
      "learning_rate": 1.9681007879439082e-05,
      "loss": 0.6069,
      "step": 719
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2089104964396444,
      "learning_rate": 1.9679784047600755e-05,
      "loss": 0.5676,
      "step": 720
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7439171560010488,
      "learning_rate": 1.9678557910802064e-05,
      "loss": 0.6363,
      "step": 721
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8386915547441941,
      "learning_rate": 1.967732946933499e-05,
      "loss": 0.6143,
      "step": 722
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7000953038815468,
      "learning_rate": 1.967609872349204e-05,
      "loss": 0.5998,
      "step": 723
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7164936334256558,
      "learning_rate": 1.967486567356629e-05,
      "loss": 0.6038,
      "step": 724
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7659679725457678,
      "learning_rate": 1.967363031985135e-05,
      "loss": 0.6186,
      "step": 725
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7719290098555641,
      "learning_rate": 1.9672392662641382e-05,
      "loss": 0.6383,
      "step": 726
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7867454528124787,
      "learning_rate": 1.96711527022311e-05,
      "loss": 0.638,
      "step": 727
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.824615245540997,
      "learning_rate": 1.9669910438915763e-05,
      "loss": 0.6293,
      "step": 728
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7203709957248383,
      "learning_rate": 1.9668665872991188e-05,
      "loss": 0.6041,
      "step": 729
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7049958876221536,
      "learning_rate": 1.966741900475372e-05,
      "loss": 0.6291,
      "step": 730
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7403611462226473,
      "learning_rate": 1.966616983450028e-05,
      "loss": 0.6229,
      "step": 731
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6458181841486706,
      "learning_rate": 1.9664918362528304e-05,
      "loss": 0.6099,
      "step": 732
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7089927347513715,
      "learning_rate": 1.9663664589135807e-05,
      "loss": 0.6164,
      "step": 733
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6905450619994759,
      "learning_rate": 1.9662408514621334e-05,
      "loss": 0.6085,
      "step": 734
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6844187992943844,
      "learning_rate": 1.966115013928398e-05,
      "loss": 0.6009,
      "step": 735
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7658157537436936,
      "learning_rate": 1.9659889463423402e-05,
      "loss": 0.6083,
      "step": 736
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7903618835310894,
      "learning_rate": 1.965862648733978e-05,
      "loss": 0.6011,
      "step": 737
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6580293715701928,
      "learning_rate": 1.965736121133386e-05,
      "loss": 0.6051,
      "step": 738
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7290082088316266,
      "learning_rate": 1.9656093635706928e-05,
      "loss": 0.6078,
      "step": 739
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6391915117268699,
      "learning_rate": 1.9654823760760832e-05,
      "loss": 0.5906,
      "step": 740
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7701716023142317,
      "learning_rate": 1.965355158679794e-05,
      "loss": 0.6403,
      "step": 741
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7270077695368161,
      "learning_rate": 1.9652277114121198e-05,
      "loss": 0.6113,
      "step": 742
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7112338229872814,
      "learning_rate": 1.9651000343034075e-05,
      "loss": 0.6062,
      "step": 743
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.741138399741411,
      "learning_rate": 1.96497212738406e-05,
      "loss": 0.6122,
      "step": 744
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9444469826009294,
      "learning_rate": 1.9648439906845345e-05,
      "loss": 0.668,
      "step": 745
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7693297540005155,
      "learning_rate": 1.964715624235343e-05,
      "loss": 0.606,
      "step": 746
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.830331222977454,
      "learning_rate": 1.9645870280670532e-05,
      "loss": 0.636,
      "step": 747
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7353864974744931,
      "learning_rate": 1.9644582022102853e-05,
      "loss": 0.6203,
      "step": 748
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7965297144831059,
      "learning_rate": 1.964329146695716e-05,
      "loss": 0.6076,
      "step": 749
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.835335342136412,
      "learning_rate": 1.964199861554076e-05,
      "loss": 0.6024,
      "step": 750
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7753218698646047,
      "learning_rate": 1.9640703468161508e-05,
      "loss": 0.5928,
      "step": 751
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8440937286721967,
      "learning_rate": 1.963940602512781e-05,
      "loss": 0.6459,
      "step": 752
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8603047526785532,
      "learning_rate": 1.963810628674861e-05,
      "loss": 0.6111,
      "step": 753
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8025932795050124,
      "learning_rate": 1.96368042533334e-05,
      "loss": 0.6013,
      "step": 754
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7599076577448644,
      "learning_rate": 1.963549992519223e-05,
      "loss": 0.6282,
      "step": 755
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8000590006261796,
      "learning_rate": 1.9634193302635683e-05,
      "loss": 0.6336,
      "step": 756
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.726488512195959,
      "learning_rate": 1.9632884385974895e-05,
      "loss": 0.6015,
      "step": 757
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7466700250900666,
      "learning_rate": 1.9631573175521547e-05,
      "loss": 0.6007,
      "step": 758
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7772213490759602,
      "learning_rate": 1.9630259671587862e-05,
      "loss": 0.6142,
      "step": 759
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7092640977614442,
      "learning_rate": 1.9628943874486615e-05,
      "loss": 0.5977,
      "step": 760
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7358036666909251,
      "learning_rate": 1.962762578453113e-05,
      "loss": 0.6138,
      "step": 761
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7501947897280842,
      "learning_rate": 1.9626305402035263e-05,
      "loss": 0.6302,
      "step": 762
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7181613862487944,
      "learning_rate": 1.9624982727313433e-05,
      "loss": 0.6141,
      "step": 763
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7475963386060731,
      "learning_rate": 1.962365776068059e-05,
      "loss": 0.6127,
      "step": 764
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8154589300679024,
      "learning_rate": 1.9622330502452242e-05,
      "loss": 0.594,
      "step": 765
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6624407633721697,
      "learning_rate": 1.9621000952944433e-05,
      "loss": 0.5841,
      "step": 766
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6933915455959632,
      "learning_rate": 1.9619669112473758e-05,
      "loss": 0.6076,
      "step": 767
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7244883851004442,
      "learning_rate": 1.9618334981357358e-05,
      "loss": 0.6159,
      "step": 768
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7691256028185538,
      "learning_rate": 1.9616998559912915e-05,
      "loss": 0.6099,
      "step": 769
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7125143519837165,
      "learning_rate": 1.961565984845866e-05,
      "loss": 0.6232,
      "step": 770
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6938283711863175,
      "learning_rate": 1.961431884731337e-05,
      "loss": 0.588,
      "step": 771
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7729144021150318,
      "learning_rate": 1.9612975556796368e-05,
      "loss": 0.6313,
      "step": 772
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8396594862560324,
      "learning_rate": 1.961162997722751e-05,
      "loss": 0.6226,
      "step": 773
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6487913069267633,
      "learning_rate": 1.9610282108927216e-05,
      "loss": 0.6052,
      "step": 774
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7135387071114779,
      "learning_rate": 1.9608931952216438e-05,
      "loss": 0.602,
      "step": 775
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7600858481267894,
      "learning_rate": 1.9607579507416677e-05,
      "loss": 0.6146,
      "step": 776
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6169649326329585,
      "learning_rate": 1.9606224774849983e-05,
      "loss": 0.6138,
      "step": 777
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7295152729810457,
      "learning_rate": 1.960486775483894e-05,
      "loss": 0.6034,
      "step": 778
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7704284757304501,
      "learning_rate": 1.9603508447706682e-05,
      "loss": 0.5908,
      "step": 779
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7808045526605131,
      "learning_rate": 1.9602146853776894e-05,
      "loss": 0.623,
      "step": 780
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6850092294175874,
      "learning_rate": 1.96007829733738e-05,
      "loss": 0.5872,
      "step": 781
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7137054552245067,
      "learning_rate": 1.9599416806822167e-05,
      "loss": 0.6049,
      "step": 782
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7008589804365108,
      "learning_rate": 1.959804835444731e-05,
      "loss": 0.6378,
      "step": 783
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8096048061986084,
      "learning_rate": 1.9596677616575078e-05,
      "loss": 0.6074,
      "step": 784
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7074521798941361,
      "learning_rate": 1.9595304593531886e-05,
      "loss": 0.6477,
      "step": 785
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8217590584241163,
      "learning_rate": 1.9593929285644665e-05,
      "loss": 0.6096,
      "step": 786
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7646995325390206,
      "learning_rate": 1.9592551693240918e-05,
      "loss": 0.6005,
      "step": 787
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8295065489029027,
      "learning_rate": 1.959117181664867e-05,
      "loss": 0.6273,
      "step": 788
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8569816021969652,
      "learning_rate": 1.9589789656196502e-05,
      "loss": 0.6374,
      "step": 789
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7922269293582486,
      "learning_rate": 1.9588405212213538e-05,
      "loss": 0.6242,
      "step": 790
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8577594879326967,
      "learning_rate": 1.958701848502944e-05,
      "loss": 0.6344,
      "step": 791
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6878802576422789,
      "learning_rate": 1.9585629474974413e-05,
      "loss": 0.6091,
      "step": 792
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6837683638365537,
      "learning_rate": 1.958423818237922e-05,
      "loss": 0.6241,
      "step": 793
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7303074191573635,
      "learning_rate": 1.9582844607575146e-05,
      "loss": 0.6159,
      "step": 794
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.761079540839613,
      "learning_rate": 1.9581448750894037e-05,
      "loss": 0.6338,
      "step": 795
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7536235236023798,
      "learning_rate": 1.9580050612668274e-05,
      "loss": 0.6042,
      "step": 796
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.733412719692991,
      "learning_rate": 1.9578650193230783e-05,
      "loss": 0.6243,
      "step": 797
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7209420490295504,
      "learning_rate": 1.9577247492915036e-05,
      "loss": 0.6372,
      "step": 798
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.806505410607875,
      "learning_rate": 1.9575842512055043e-05,
      "loss": 0.6101,
      "step": 799
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8101219168795204,
      "learning_rate": 1.957443525098536e-05,
      "loss": 0.6506,
      "step": 800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7752683355687418,
      "learning_rate": 1.9573025710041084e-05,
      "loss": 0.6148,
      "step": 801
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7840555462377451,
      "learning_rate": 1.9571613889557857e-05,
      "loss": 0.6365,
      "step": 802
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7873894248166782,
      "learning_rate": 1.9570199789871866e-05,
      "loss": 0.5955,
      "step": 803
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.700953512782844,
      "learning_rate": 1.9568783411319834e-05,
      "loss": 0.5821,
      "step": 804
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8080698081700554,
      "learning_rate": 1.9567364754239034e-05,
      "loss": 0.6244,
      "step": 805
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.636176380420077,
      "learning_rate": 1.9565943818967276e-05,
      "loss": 0.6122,
      "step": 806
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9103205634244681,
      "learning_rate": 1.956452060584291e-05,
      "loss": 0.5977,
      "step": 807
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7683179630984855,
      "learning_rate": 1.9563095115204845e-05,
      "loss": 0.6266,
      "step": 808
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.779354579800773,
      "learning_rate": 1.956166734739251e-05,
      "loss": 0.6138,
      "step": 809
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6924051225933391,
      "learning_rate": 1.956023730274589e-05,
      "loss": 0.6091,
      "step": 810
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6460562406332159,
      "learning_rate": 1.9558804981605514e-05,
      "loss": 0.5957,
      "step": 811
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7891108366879648,
      "learning_rate": 1.9557370384312436e-05,
      "loss": 0.6152,
      "step": 812
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7768790971290466,
      "learning_rate": 1.955593351120827e-05,
      "loss": 0.6111,
      "step": 813
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7194369570187407,
      "learning_rate": 1.955449436263517e-05,
      "loss": 0.5928,
      "step": 814
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7187095735840223,
      "learning_rate": 1.955305293893582e-05,
      "loss": 0.6295,
      "step": 815
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7138820474903014,
      "learning_rate": 1.955160924045346e-05,
      "loss": 0.6187,
      "step": 816
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8115213072894557,
      "learning_rate": 1.955016326753186e-05,
      "loss": 0.6024,
      "step": 817
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7347593004415222,
      "learning_rate": 1.954871502051534e-05,
      "loss": 0.6033,
      "step": 818
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7091090160152017,
      "learning_rate": 1.9547264499748758e-05,
      "loss": 0.6233,
      "step": 819
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9119660095422073,
      "learning_rate": 1.954581170557751e-05,
      "loss": 0.5879,
      "step": 820
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7655958697999641,
      "learning_rate": 1.9544356638347538e-05,
      "loss": 0.6005,
      "step": 821
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6750491560799518,
      "learning_rate": 1.9542899298405326e-05,
      "loss": 0.6309,
      "step": 822
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7140730351993304,
      "learning_rate": 1.9541439686097896e-05,
      "loss": 0.6226,
      "step": 823
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6907716940035762,
      "learning_rate": 1.9539977801772812e-05,
      "loss": 0.6072,
      "step": 824
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6732070826875013,
      "learning_rate": 1.9538513645778185e-05,
      "loss": 0.5978,
      "step": 825
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.676529373074725,
      "learning_rate": 1.9537047218462653e-05,
      "loss": 0.6494,
      "step": 826
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6258845289847692,
      "learning_rate": 1.9535578520175408e-05,
      "loss": 0.6095,
      "step": 827
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7095622890506054,
      "learning_rate": 1.9534107551266178e-05,
      "loss": 0.6197,
      "step": 828
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8136827391589732,
      "learning_rate": 1.953263431208523e-05,
      "loss": 0.6322,
      "step": 829
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7298182777296318,
      "learning_rate": 1.9531158802983374e-05,
      "loss": 0.6081,
      "step": 830
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7887786845124409,
      "learning_rate": 1.9529681024311955e-05,
      "loss": 0.6271,
      "step": 831
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.743564358650425,
      "learning_rate": 1.952820097642287e-05,
      "loss": 0.6056,
      "step": 832
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7504357432611688,
      "learning_rate": 1.9526718659668554e-05,
      "loss": 0.6464,
      "step": 833
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6724752732191573,
      "learning_rate": 1.9525234074401965e-05,
      "loss": 0.6026,
      "step": 834
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.800764054028567,
      "learning_rate": 1.952374722097662e-05,
      "loss": 0.6068,
      "step": 835
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7525790174084266,
      "learning_rate": 1.9522258099746572e-05,
      "loss": 0.6189,
      "step": 836
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6770774956721368,
      "learning_rate": 1.952076671106641e-05,
      "loss": 0.6048,
      "step": 837
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7777443833177453,
      "learning_rate": 1.9519273055291266e-05,
      "loss": 0.655,
      "step": 838
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6788451753630809,
      "learning_rate": 1.951777713277681e-05,
      "loss": 0.6078,
      "step": 839
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6557045131967084,
      "learning_rate": 1.9516278943879253e-05,
      "loss": 0.5958,
      "step": 840
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6337514048487866,
      "learning_rate": 1.9514778488955347e-05,
      "loss": 0.6001,
      "step": 841
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7002188947228336,
      "learning_rate": 1.9513275768362377e-05,
      "loss": 0.6225,
      "step": 842
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6895414717466448,
      "learning_rate": 1.951177078245818e-05,
      "loss": 0.6123,
      "step": 843
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7200177251328861,
      "learning_rate": 1.9510263531601117e-05,
      "loss": 0.6064,
      "step": 844
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6540186271671573,
      "learning_rate": 1.95087540161501e-05,
      "loss": 0.5889,
      "step": 845
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6400167222623452,
      "learning_rate": 1.9507242236464575e-05,
      "loss": 0.6075,
      "step": 846
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7233696904266249,
      "learning_rate": 1.9505728192904532e-05,
      "loss": 0.6088,
      "step": 847
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7479001620120134,
      "learning_rate": 1.9504211885830494e-05,
      "loss": 0.6093,
      "step": 848
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6787809627431617,
      "learning_rate": 1.9502693315603524e-05,
      "loss": 0.5959,
      "step": 849
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7059206344890812,
      "learning_rate": 1.950117248258523e-05,
      "loss": 0.5929,
      "step": 850
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7372657656904235,
      "learning_rate": 1.949964938713775e-05,
      "loss": 0.6203,
      "step": 851
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7449259025362394,
      "learning_rate": 1.9498124029623766e-05,
      "loss": 0.6029,
      "step": 852
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7347953398325622,
      "learning_rate": 1.94965964104065e-05,
      "loss": 0.6003,
      "step": 853
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0125356061015005,
      "learning_rate": 1.9495066529849706e-05,
      "loss": 0.6073,
      "step": 854
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7705017285054142,
      "learning_rate": 1.9493534388317686e-05,
      "loss": 0.5984,
      "step": 855
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7433225391790784,
      "learning_rate": 1.9491999986175276e-05,
      "loss": 0.6053,
      "step": 856
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8501568648689679,
      "learning_rate": 1.9490463323787842e-05,
      "loss": 0.6487,
      "step": 857
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7359155423556251,
      "learning_rate": 1.9488924401521303e-05,
      "loss": 0.6308,
      "step": 858
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7447667152929615,
      "learning_rate": 1.94873832197421e-05,
      "loss": 0.6088,
      "step": 859
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8089211443748,
      "learning_rate": 1.9485839778817236e-05,
      "loss": 0.5932,
      "step": 860
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8240109201938625,
      "learning_rate": 1.9484294079114222e-05,
      "loss": 0.6263,
      "step": 861
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8162728271568246,
      "learning_rate": 1.9482746121001125e-05,
      "loss": 0.6298,
      "step": 862
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8586515164729288,
      "learning_rate": 1.948119590484655e-05,
      "loss": 0.6115,
      "step": 863
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7976548159647023,
      "learning_rate": 1.9479643431019634e-05,
      "loss": 0.5817,
      "step": 864
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.700396859618924,
      "learning_rate": 1.9478088699890056e-05,
      "loss": 0.6033,
      "step": 865
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8182681054909283,
      "learning_rate": 1.9476531711828027e-05,
      "loss": 0.6463,
      "step": 866
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.7652789080308775,
      "learning_rate": 1.9474972467204298e-05,
      "loss": 0.6073,
      "step": 867
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7004124200985176,
      "learning_rate": 1.9473410966390162e-05,
      "loss": 0.6033,
      "step": 868
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.840885085696591,
      "learning_rate": 1.9471847209757447e-05,
      "loss": 0.612,
      "step": 869
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6390362260578468,
      "learning_rate": 1.9470281197678506e-05,
      "loss": 0.6022,
      "step": 870
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6616962327256917,
      "learning_rate": 1.946871293052625e-05,
      "loss": 0.6075,
      "step": 871
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7847662778770788,
      "learning_rate": 1.9467142408674112e-05,
      "loss": 0.6284,
      "step": 872
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7183748543543097,
      "learning_rate": 1.9465569632496067e-05,
      "loss": 0.604,
      "step": 873
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.670933535222292,
      "learning_rate": 1.9463994602366622e-05,
      "loss": 0.5899,
      "step": 874
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0798327968433112,
      "learning_rate": 1.9462417318660837e-05,
      "loss": 0.6068,
      "step": 875
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.723740620078199,
      "learning_rate": 1.946083778175428e-05,
      "loss": 0.6217,
      "step": 876
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8075999654647895,
      "learning_rate": 1.9459255992023082e-05,
      "loss": 0.6407,
      "step": 877
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.692250009110226,
      "learning_rate": 1.94576719498439e-05,
      "loss": 0.5997,
      "step": 878
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7503964130568173,
      "learning_rate": 1.9456085655593927e-05,
      "loss": 0.6333,
      "step": 879
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6958935198435149,
      "learning_rate": 1.9454497109650892e-05,
      "loss": 0.6104,
      "step": 880
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7157263306898249,
      "learning_rate": 1.945290631239306e-05,
      "loss": 0.6021,
      "step": 881
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6912742224497352,
      "learning_rate": 1.9451313264199236e-05,
      "loss": 0.5963,
      "step": 882
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7678159998836004,
      "learning_rate": 1.9449717965448757e-05,
      "loss": 0.6297,
      "step": 883
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7119333787253396,
      "learning_rate": 1.9448120416521502e-05,
      "loss": 0.5964,
      "step": 884
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8506826737093864,
      "learning_rate": 1.9446520617797872e-05,
      "loss": 0.6206,
      "step": 885
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7172754676246215,
      "learning_rate": 1.944491856965882e-05,
      "loss": 0.6158,
      "step": 886
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8473146425087763,
      "learning_rate": 1.944331427248582e-05,
      "loss": 0.6368,
      "step": 887
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6680056224151352,
      "learning_rate": 1.9441707726660898e-05,
      "loss": 0.5832,
      "step": 888
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.758244082891134,
      "learning_rate": 1.9440098932566597e-05,
      "loss": 0.6054,
      "step": 889
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8276778249606765,
      "learning_rate": 1.9438487890586014e-05,
      "loss": 0.5626,
      "step": 890
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9500398198447653,
      "learning_rate": 1.9436874601102768e-05,
      "loss": 0.6881,
      "step": 891
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.222565639704216,
      "learning_rate": 1.9435259064501015e-05,
      "loss": 0.6475,
      "step": 892
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7450607851923396,
      "learning_rate": 1.943364128116545e-05,
      "loss": 0.627,
      "step": 893
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7375373073006383,
      "learning_rate": 1.94320212514813e-05,
      "loss": 0.6173,
      "step": 894
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8360185465033739,
      "learning_rate": 1.9430398975834333e-05,
      "loss": 0.6352,
      "step": 895
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6536131487246389,
      "learning_rate": 1.9428774454610845e-05,
      "loss": 0.5991,
      "step": 896
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9661416898458469,
      "learning_rate": 1.9427147688197662e-05,
      "loss": 0.6284,
      "step": 897
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7369949002173071,
      "learning_rate": 1.942551867698216e-05,
      "loss": 0.5992,
      "step": 898
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7116548647355779,
      "learning_rate": 1.942388742135224e-05,
      "loss": 0.6033,
      "step": 899
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6912251760203055,
      "learning_rate": 1.942225392169633e-05,
      "loss": 0.5959,
      "step": 900
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5812729043295686,
      "learning_rate": 1.9420618178403412e-05,
      "loss": 0.5996,
      "step": 901
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6675822410965759,
      "learning_rate": 1.941898019186298e-05,
      "loss": 0.6,
      "step": 902
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6934745220720596,
      "learning_rate": 1.9417339962465084e-05,
      "loss": 0.6054,
      "step": 903
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.835238642348984,
      "learning_rate": 1.9415697490600287e-05,
      "loss": 0.5892,
      "step": 904
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7376904574185248,
      "learning_rate": 1.9414052776659705e-05,
      "loss": 0.6089,
      "step": 905
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9071807708276016,
      "learning_rate": 1.9412405821034973e-05,
      "loss": 0.6093,
      "step": 906
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7035537676885433,
      "learning_rate": 1.9410756624118267e-05,
      "loss": 0.6185,
      "step": 907
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6551793145808442,
      "learning_rate": 1.9409105186302292e-05,
      "loss": 0.6231,
      "step": 908
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7600935100615208,
      "learning_rate": 1.9407451507980298e-05,
      "loss": 0.6293,
      "step": 909
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7549862421847495,
      "learning_rate": 1.940579558954606e-05,
      "loss": 0.6159,
      "step": 910
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7743615366223672,
      "learning_rate": 1.940413743139388e-05,
      "loss": 0.6261,
      "step": 911
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8941597391025313,
      "learning_rate": 1.9402477033918607e-05,
      "loss": 0.6406,
      "step": 912
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7033285969744142,
      "learning_rate": 1.9400814397515612e-05,
      "loss": 0.6192,
      "step": 913
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6981366556949055,
      "learning_rate": 1.9399149522580805e-05,
      "loss": 0.5903,
      "step": 914
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7240372369609608,
      "learning_rate": 1.9397482409510627e-05,
      "loss": 0.6375,
      "step": 915
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.802047929190546,
      "learning_rate": 1.9395813058702057e-05,
      "loss": 0.6206,
      "step": 916
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7844062034270127,
      "learning_rate": 1.93941414705526e-05,
      "loss": 0.6366,
      "step": 917
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6525003885089735,
      "learning_rate": 1.9392467645460293e-05,
      "loss": 0.6259,
      "step": 918
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7003843369764982,
      "learning_rate": 1.9390791583823713e-05,
      "loss": 0.6224,
      "step": 919
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7168920331311668,
      "learning_rate": 1.9389113286041965e-05,
      "loss": 0.626,
      "step": 920
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7302892270887388,
      "learning_rate": 1.9387432752514686e-05,
      "loss": 0.6523,
      "step": 921
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.72804738398997,
      "learning_rate": 1.938574998364205e-05,
      "loss": 0.6006,
      "step": 922
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7267532911365068,
      "learning_rate": 1.9384064979824753e-05,
      "loss": 0.6008,
      "step": 923
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7080954973998344,
      "learning_rate": 1.9382377741464032e-05,
      "loss": 0.5872,
      "step": 924
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7271921974718173,
      "learning_rate": 1.938068826896166e-05,
      "loss": 0.6052,
      "step": 925
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7421162013955066,
      "learning_rate": 1.937899656271993e-05,
      "loss": 0.6098,
      "step": 926
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7332850011948303,
      "learning_rate": 1.9377302623141672e-05,
      "loss": 0.6145,
      "step": 927
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.652190024628748,
      "learning_rate": 1.9375606450630253e-05,
      "loss": 0.5974,
      "step": 928
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7895818727818744,
      "learning_rate": 1.9373908045589566e-05,
      "loss": 0.6116,
      "step": 929
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7478618343005538,
      "learning_rate": 1.9372207408424034e-05,
      "loss": 0.6063,
      "step": 930
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2485154119780861,
      "learning_rate": 1.937050453953862e-05,
      "loss": 0.5893,
      "step": 931
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7640763571883183,
      "learning_rate": 1.936879943933881e-05,
      "loss": 0.618,
      "step": 932
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.166923420714208,
      "learning_rate": 1.936709210823062e-05,
      "loss": 0.6377,
      "step": 933
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7849884079721925,
      "learning_rate": 1.9365382546620607e-05,
      "loss": 0.6057,
      "step": 934
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.827994814421874,
      "learning_rate": 1.9363670754915855e-05,
      "loss": 0.5866,
      "step": 935
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7085509895022793,
      "learning_rate": 1.936195673352397e-05,
      "loss": 0.619,
      "step": 936
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8074771179001425,
      "learning_rate": 1.9360240482853104e-05,
      "loss": 0.6179,
      "step": 937
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7978798055564547,
      "learning_rate": 1.9358522003311927e-05,
      "loss": 0.615,
      "step": 938
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7351160759576867,
      "learning_rate": 1.935680129530965e-05,
      "loss": 0.6049,
      "step": 939
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.705000172454822,
      "learning_rate": 1.935507835925601e-05,
      "loss": 0.608,
      "step": 940
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8499207687913607,
      "learning_rate": 1.9353353195561274e-05,
      "loss": 0.6426,
      "step": 941
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7244049875285213,
      "learning_rate": 1.9351625804636232e-05,
      "loss": 0.6164,
      "step": 942
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7768796409217661,
      "learning_rate": 1.934989618689222e-05,
      "loss": 0.6052,
      "step": 943
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7192804756444787,
      "learning_rate": 1.9348164342741095e-05,
      "loss": 0.5877,
      "step": 944
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6587968993969502,
      "learning_rate": 1.9346430272595244e-05,
      "loss": 0.6204,
      "step": 945
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7239927534792663,
      "learning_rate": 1.934469397686759e-05,
      "loss": 0.6114,
      "step": 946
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7271270171123475,
      "learning_rate": 1.9342955455971576e-05,
      "loss": 0.6164,
      "step": 947
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6574210030584338,
      "learning_rate": 1.9341214710321178e-05,
      "loss": 0.601,
      "step": 948
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8217603869980132,
      "learning_rate": 1.9339471740330917e-05,
      "loss": 0.6423,
      "step": 949
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6979685426235882,
      "learning_rate": 1.9337726546415815e-05,
      "loss": 0.6178,
      "step": 950
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7124525675647895,
      "learning_rate": 1.933597912899145e-05,
      "loss": 0.6482,
      "step": 951
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7076021104445245,
      "learning_rate": 1.9334229488473917e-05,
      "loss": 0.6378,
      "step": 952
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.791165453810713,
      "learning_rate": 1.933247762527984e-05,
      "loss": 0.6055,
      "step": 953
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6663348805277864,
      "learning_rate": 1.9330723539826373e-05,
      "loss": 0.6026,
      "step": 954
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6062450029277702,
      "learning_rate": 1.932896723253121e-05,
      "loss": 0.6093,
      "step": 955
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6902765249492427,
      "learning_rate": 1.9327208703812553e-05,
      "loss": 0.6206,
      "step": 956
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6283592045090637,
      "learning_rate": 1.9325447954089148e-05,
      "loss": 0.6008,
      "step": 957
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6487008445690206,
      "learning_rate": 1.9323684983780273e-05,
      "loss": 0.6058,
      "step": 958
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9126954927209444,
      "learning_rate": 1.9321919793305723e-05,
      "loss": 0.6284,
      "step": 959
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4178615831753079,
      "learning_rate": 1.9320152383085826e-05,
      "loss": 0.5827,
      "step": 960
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7798206067938044,
      "learning_rate": 1.931838275354144e-05,
      "loss": 0.6245,
      "step": 961
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7149354004197548,
      "learning_rate": 1.9316610905093957e-05,
      "loss": 0.6238,
      "step": 962
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7443064183592594,
      "learning_rate": 1.931483683816528e-05,
      "loss": 0.6311,
      "step": 963
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7078285869677127,
      "learning_rate": 1.9313060553177865e-05,
      "loss": 0.6052,
      "step": 964
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8549179488648055,
      "learning_rate": 1.931128205055467e-05,
      "loss": 0.6299,
      "step": 965
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7774540320182158,
      "learning_rate": 1.9309501330719205e-05,
      "loss": 0.5798,
      "step": 966
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7883288747921433,
      "learning_rate": 1.9307718394095493e-05,
      "loss": 0.6168,
      "step": 967
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.946469557843595,
      "learning_rate": 1.9305933241108086e-05,
      "loss": 0.6291,
      "step": 968
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7360555870755733,
      "learning_rate": 1.9304145872182064e-05,
      "loss": 0.607,
      "step": 969
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7756771450303076,
      "learning_rate": 1.9302356287743048e-05,
      "loss": 0.6275,
      "step": 970
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6766311764793317,
      "learning_rate": 1.9300564488217164e-05,
      "loss": 0.6343,
      "step": 971
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7188994648467893,
      "learning_rate": 1.9298770474031086e-05,
      "loss": 0.6009,
      "step": 972
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6750529366025588,
      "learning_rate": 1.9296974245612e-05,
      "loss": 0.6196,
      "step": 973
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7561915947912583,
      "learning_rate": 1.9295175803387633e-05,
      "loss": 0.6297,
      "step": 974
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8290928907507275,
      "learning_rate": 1.9293375147786225e-05,
      "loss": 0.6131,
      "step": 975
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8310142102557582,
      "learning_rate": 1.929157227923655e-05,
      "loss": 0.622,
      "step": 976
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7122039636690828,
      "learning_rate": 1.9289767198167918e-05,
      "loss": 0.616,
      "step": 977
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6540433320595809,
      "learning_rate": 1.9287959905010144e-05,
      "loss": 0.602,
      "step": 978
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6377770355253077,
      "learning_rate": 1.9286150400193593e-05,
      "loss": 0.6021,
      "step": 979
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7274592478789627,
      "learning_rate": 1.928433868414914e-05,
      "loss": 0.6399,
      "step": 980
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6632696715465773,
      "learning_rate": 1.9282524757308197e-05,
      "loss": 0.6259,
      "step": 981
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7006995675515623,
      "learning_rate": 1.9280708620102695e-05,
      "loss": 0.6091,
      "step": 982
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6223113046033273,
      "learning_rate": 1.9278890272965097e-05,
      "loss": 0.5993,
      "step": 983
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6476660392011026,
      "learning_rate": 1.9277069716328385e-05,
      "loss": 0.5733,
      "step": 984
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8308980098081431,
      "learning_rate": 1.9275246950626077e-05,
      "loss": 0.6296,
      "step": 985
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7569920194874641,
      "learning_rate": 1.927342197629221e-05,
      "loss": 0.6085,
      "step": 986
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6859663741926855,
      "learning_rate": 1.927159479376135e-05,
      "loss": 0.5989,
      "step": 987
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7606458671769926,
      "learning_rate": 1.9269765403468583e-05,
      "loss": 0.6398,
      "step": 988
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8037588079981138,
      "learning_rate": 1.9267933805849534e-05,
      "loss": 0.6327,
      "step": 989
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7587194607515481,
      "learning_rate": 1.9266100001340337e-05,
      "loss": 0.5953,
      "step": 990
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7598820837433717,
      "learning_rate": 1.926426399037766e-05,
      "loss": 0.6154,
      "step": 991
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6592325896029957,
      "learning_rate": 1.92624257733987e-05,
      "loss": 0.6248,
      "step": 992
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8192231163376772,
      "learning_rate": 1.9260585350841174e-05,
      "loss": 0.6283,
      "step": 993
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7460749803810885,
      "learning_rate": 1.9258742723143324e-05,
      "loss": 0.6189,
      "step": 994
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8366894863362077,
      "learning_rate": 1.925689789074392e-05,
      "loss": 0.6084,
      "step": 995
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6938941639658621,
      "learning_rate": 1.9255050854082255e-05,
      "loss": 0.6012,
      "step": 996
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7547979397820499,
      "learning_rate": 1.9253201613598145e-05,
      "loss": 0.633,
      "step": 997
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7298418081313954,
      "learning_rate": 1.9251350169731935e-05,
      "loss": 0.6162,
      "step": 998
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6981337625885464,
      "learning_rate": 1.9249496522924492e-05,
      "loss": 0.5836,
      "step": 999
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7624033676080971,
      "learning_rate": 1.9247640673617213e-05,
      "loss": 0.6183,
      "step": 1000
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7229982698001403,
      "learning_rate": 1.9245782622252008e-05,
      "loss": 0.6288,
      "step": 1001
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7175014046844768,
      "learning_rate": 1.924392236927132e-05,
      "loss": 0.5985,
      "step": 1002
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7026816679028524,
      "learning_rate": 1.9242059915118117e-05,
      "loss": 0.612,
      "step": 1003
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7733408172025907,
      "learning_rate": 1.9240195260235883e-05,
      "loss": 0.6029,
      "step": 1004
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6984812315493724,
      "learning_rate": 1.923832840506864e-05,
      "loss": 0.6195,
      "step": 1005
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8586741975990974,
      "learning_rate": 1.9236459350060918e-05,
      "loss": 0.6177,
      "step": 1006
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7423673250743222,
      "learning_rate": 1.9234588095657783e-05,
      "loss": 0.6078,
      "step": 1007
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7109604745207248,
      "learning_rate": 1.923271464230482e-05,
      "loss": 0.604,
      "step": 1008
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6835666140615029,
      "learning_rate": 1.9230838990448134e-05,
      "loss": 0.6181,
      "step": 1009
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6951177843450684,
      "learning_rate": 1.922896114053436e-05,
      "loss": 0.6053,
      "step": 1010
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6469885755478186,
      "learning_rate": 1.9227081093010647e-05,
      "loss": 0.5979,
      "step": 1011
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7971667640078239,
      "learning_rate": 1.9225198848324687e-05,
      "loss": 0.6358,
      "step": 1012
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6958133690253047,
      "learning_rate": 1.9223314406924675e-05,
      "loss": 0.6211,
      "step": 1013
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6620385583401297,
      "learning_rate": 1.9221427769259333e-05,
      "loss": 0.5868,
      "step": 1014
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6608545099680336,
      "learning_rate": 1.9219538935777912e-05,
      "loss": 0.6021,
      "step": 1015
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.694258471765309,
      "learning_rate": 1.9217647906930183e-05,
      "loss": 0.624,
      "step": 1016
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8840442169450834,
      "learning_rate": 1.9215754683166442e-05,
      "loss": 0.5999,
      "step": 1017
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7144925179558652,
      "learning_rate": 1.9213859264937503e-05,
      "loss": 0.621,
      "step": 1018
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6581582908269362,
      "learning_rate": 1.9211961652694704e-05,
      "loss": 0.6033,
      "step": 1019
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7705525998395379,
      "learning_rate": 1.9210061846889908e-05,
      "loss": 0.6082,
      "step": 1020
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.753136313060494,
      "learning_rate": 1.92081598479755e-05,
      "loss": 0.6447,
      "step": 1021
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8751663317906946,
      "learning_rate": 1.9206255656404384e-05,
      "loss": 0.6421,
      "step": 1022
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7038718816176662,
      "learning_rate": 1.9204349272629988e-05,
      "loss": 0.5923,
      "step": 1023
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6545727669579694,
      "learning_rate": 1.9202440697106263e-05,
      "loss": 0.5896,
      "step": 1024
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7143263817098455,
      "learning_rate": 1.920052993028768e-05,
      "loss": 0.6135,
      "step": 1025
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6403203327267423,
      "learning_rate": 1.919861697262923e-05,
      "loss": 0.6073,
      "step": 1026
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6548554167191524,
      "learning_rate": 1.919670182458644e-05,
      "loss": 0.62,
      "step": 1027
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7150563685072396,
      "learning_rate": 1.9194784486615333e-05,
      "loss": 0.6306,
      "step": 1028
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6895753842624802,
      "learning_rate": 1.9192864959172475e-05,
      "loss": 0.5979,
      "step": 1029
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.79396892193246,
      "learning_rate": 1.9190943242714947e-05,
      "loss": 0.6289,
      "step": 1030
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6993731358151569,
      "learning_rate": 1.9189019337700344e-05,
      "loss": 0.5833,
      "step": 1031
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.698168471791629,
      "learning_rate": 1.9187093244586793e-05,
      "loss": 0.5774,
      "step": 1032
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6505110401440101,
      "learning_rate": 1.9185164963832938e-05,
      "loss": 0.5999,
      "step": 1033
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7203011112453799,
      "learning_rate": 1.918323449589794e-05,
      "loss": 0.5906,
      "step": 1034
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7193031773697043,
      "learning_rate": 1.9181301841241486e-05,
      "loss": 0.6124,
      "step": 1035
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6885227504737675,
      "learning_rate": 1.917936700032378e-05,
      "loss": 0.5973,
      "step": 1036
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6917572654775688,
      "learning_rate": 1.917742997360555e-05,
      "loss": 0.6162,
      "step": 1037
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7344510700280423,
      "learning_rate": 1.9175490761548047e-05,
      "loss": 0.6343,
      "step": 1038
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7432492209921788,
      "learning_rate": 1.917354936461303e-05,
      "loss": 0.5947,
      "step": 1039
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6621133559084311,
      "learning_rate": 1.917160578326279e-05,
      "loss": 0.6194,
      "step": 1040
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8293858890575486,
      "learning_rate": 1.9169660017960135e-05,
      "loss": 0.6076,
      "step": 1041
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6585007548179668,
      "learning_rate": 1.916771206916839e-05,
      "loss": 0.6053,
      "step": 1042
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7535990323751358,
      "learning_rate": 1.9165761937351412e-05,
      "loss": 0.5958,
      "step": 1043
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6821407238302626,
      "learning_rate": 1.9163809622973555e-05,
      "loss": 0.5925,
      "step": 1044
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.743219681504195,
      "learning_rate": 1.9161855126499716e-05,
      "loss": 0.6,
      "step": 1045
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6757805472976454,
      "learning_rate": 1.91598984483953e-05,
      "loss": 0.5762,
      "step": 1046
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8893447406909694,
      "learning_rate": 1.9157939589126226e-05,
      "loss": 0.6332,
      "step": 1047
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7019683043246642,
      "learning_rate": 1.9155978549158952e-05,
      "loss": 0.6061,
      "step": 1048
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6990684797286649,
      "learning_rate": 1.915401532896043e-05,
      "loss": 0.5834,
      "step": 1049
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7621031095661669,
      "learning_rate": 1.9152049928998157e-05,
      "loss": 0.6257,
      "step": 1050
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6967397401599443,
      "learning_rate": 1.9150082349740123e-05,
      "loss": 0.6114,
      "step": 1051
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7250104867765016,
      "learning_rate": 1.9148112591654858e-05,
      "loss": 0.6536,
      "step": 1052
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7982002931880168,
      "learning_rate": 1.9146140655211405e-05,
      "loss": 0.6372,
      "step": 1053
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6943740172375892,
      "learning_rate": 1.9144166540879315e-05,
      "loss": 0.6126,
      "step": 1054
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7779563961534407,
      "learning_rate": 1.9142190249128677e-05,
      "loss": 0.6075,
      "step": 1055
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7173274497495413,
      "learning_rate": 1.9140211780430076e-05,
      "loss": 0.6003,
      "step": 1056
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6500575644289253,
      "learning_rate": 1.9138231135254635e-05,
      "loss": 0.6023,
      "step": 1057
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7853678577067646,
      "learning_rate": 1.9136248314073985e-05,
      "loss": 0.5957,
      "step": 1058
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7874571726655841,
      "learning_rate": 1.9134263317360277e-05,
      "loss": 0.6113,
      "step": 1059
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1818315561142991,
      "learning_rate": 1.913227614558618e-05,
      "loss": 0.6232,
      "step": 1060
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7463433918002932,
      "learning_rate": 1.9130286799224887e-05,
      "loss": 0.6384,
      "step": 1061
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6918498248228361,
      "learning_rate": 1.9128295278750094e-05,
      "loss": 0.6118,
      "step": 1062
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8333064297150972,
      "learning_rate": 1.9126301584636034e-05,
      "loss": 0.6077,
      "step": 1063
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6763206134464432,
      "learning_rate": 1.9124305717357437e-05,
      "loss": 0.5964,
      "step": 1064
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6644019835329056,
      "learning_rate": 1.912230767738957e-05,
      "loss": 0.5997,
      "step": 1065
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7494642291362389,
      "learning_rate": 1.91203074652082e-05,
      "loss": 0.6434,
      "step": 1066
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7504805065834519,
      "learning_rate": 1.9118305081289626e-05,
      "loss": 0.6185,
      "step": 1067
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7072756993038933,
      "learning_rate": 1.911630052611066e-05,
      "loss": 0.626,
      "step": 1068
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.800187788420381,
      "learning_rate": 1.9114293800148622e-05,
      "loss": 0.6025,
      "step": 1069
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7128398302579053,
      "learning_rate": 1.911228490388136e-05,
      "loss": 0.595,
      "step": 1070
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6817101146504935,
      "learning_rate": 1.911027383778723e-05,
      "loss": 0.6349,
      "step": 1071
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8148578678328573,
      "learning_rate": 1.9108260602345114e-05,
      "loss": 0.6262,
      "step": 1072
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0728800163066807,
      "learning_rate": 1.9106245198034402e-05,
      "loss": 0.6071,
      "step": 1073
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6720216455538829,
      "learning_rate": 1.910422762533501e-05,
      "loss": 0.5934,
      "step": 1074
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6912444659095367,
      "learning_rate": 1.910220788472736e-05,
      "loss": 0.6136,
      "step": 1075
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6420494852047184,
      "learning_rate": 1.910018597669239e-05,
      "loss": 0.593,
      "step": 1076
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8699845718945016,
      "learning_rate": 1.909816190171157e-05,
      "loss": 0.6541,
      "step": 1077
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6512456275943187,
      "learning_rate": 1.909613566026687e-05,
      "loss": 0.6106,
      "step": 1078
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7291916151773172,
      "learning_rate": 1.9094107252840778e-05,
      "loss": 0.6022,
      "step": 1079
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6713062784242807,
      "learning_rate": 1.9092076679916302e-05,
      "loss": 0.608,
      "step": 1080
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6398237368232861,
      "learning_rate": 1.909004394197696e-05,
      "loss": 0.6016,
      "step": 1081
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3815168098656156,
      "learning_rate": 1.90880090395068e-05,
      "loss": 0.6071,
      "step": 1082
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6420840584261446,
      "learning_rate": 1.9085971972990366e-05,
      "loss": 0.6059,
      "step": 1083
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6849902982920437,
      "learning_rate": 1.9083932742912733e-05,
      "loss": 0.6154,
      "step": 1084
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.655026176612022,
      "learning_rate": 1.908189134975948e-05,
      "loss": 0.6054,
      "step": 1085
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7158431329366663,
      "learning_rate": 1.907984779401671e-05,
      "loss": 0.6207,
      "step": 1086
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.642921520991938,
      "learning_rate": 1.907780207617103e-05,
      "loss": 0.5961,
      "step": 1087
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6771344975699938,
      "learning_rate": 1.9075754196709574e-05,
      "loss": 0.5953,
      "step": 1088
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7228839169829289,
      "learning_rate": 1.907370415611998e-05,
      "loss": 0.6118,
      "step": 1089
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6831029680402461,
      "learning_rate": 1.907165195489041e-05,
      "loss": 0.5952,
      "step": 1090
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7404912403178175,
      "learning_rate": 1.9069597593509538e-05,
      "loss": 0.6149,
      "step": 1091
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7393730670099675,
      "learning_rate": 1.906754107246655e-05,
      "loss": 0.6092,
      "step": 1092
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.697426656723135,
      "learning_rate": 1.9065482392251142e-05,
      "loss": 0.6109,
      "step": 1093
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7909024462163136,
      "learning_rate": 1.9063421553353535e-05,
      "loss": 0.6344,
      "step": 1094
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7304639882078083,
      "learning_rate": 1.9061358556264455e-05,
      "loss": 0.6212,
      "step": 1095
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7086994059547599,
      "learning_rate": 1.905929340147514e-05,
      "loss": 0.6206,
      "step": 1096
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7587841559688621,
      "learning_rate": 1.9057226089477358e-05,
      "loss": 0.6127,
      "step": 1097
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6894867506972107,
      "learning_rate": 1.9055156620763372e-05,
      "loss": 0.6121,
      "step": 1098
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7954388726507495,
      "learning_rate": 1.905308499582597e-05,
      "loss": 0.6346,
      "step": 1099
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6292364270115334,
      "learning_rate": 1.9051011215158445e-05,
      "loss": 0.5966,
      "step": 1100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6970005374865067,
      "learning_rate": 1.904893527925461e-05,
      "loss": 0.6138,
      "step": 1101
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6755189649363342,
      "learning_rate": 1.904685718860879e-05,
      "loss": 0.6038,
      "step": 1102
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6662843481690619,
      "learning_rate": 1.904477694371582e-05,
      "loss": 0.5906,
      "step": 1103
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.760439402593433,
      "learning_rate": 1.9042694545071055e-05,
      "loss": 0.6339,
      "step": 1104
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7167727410375813,
      "learning_rate": 1.9040609993170352e-05,
      "loss": 0.6009,
      "step": 1105
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5344042877997075,
      "learning_rate": 1.9038523288510088e-05,
      "loss": 0.5786,
      "step": 1106
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6887007773917607,
      "learning_rate": 1.9036434431587155e-05,
      "loss": 0.5846,
      "step": 1107
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7177774114582823,
      "learning_rate": 1.9034343422898952e-05,
      "loss": 0.5904,
      "step": 1108
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7834081551006415,
      "learning_rate": 1.9032250262943388e-05,
      "loss": 0.6276,
      "step": 1109
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7474027640406299,
      "learning_rate": 1.9030154952218895e-05,
      "loss": 0.6222,
      "step": 1110
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7112360865725054,
      "learning_rate": 1.9028057491224405e-05,
      "loss": 0.613,
      "step": 1111
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6825304570105598,
      "learning_rate": 1.902595788045937e-05,
      "loss": 0.6056,
      "step": 1112
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.669218673730651,
      "learning_rate": 1.9023856120423754e-05,
      "loss": 0.5908,
      "step": 1113
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.728781341473331,
      "learning_rate": 1.9021752211618026e-05,
      "loss": 0.6052,
      "step": 1114
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6631364718347144,
      "learning_rate": 1.9019646154543173e-05,
      "loss": 0.5906,
      "step": 1115
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6628709153269217,
      "learning_rate": 1.9017537949700693e-05,
      "loss": 0.622,
      "step": 1116
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7418474435110829,
      "learning_rate": 1.901542759759259e-05,
      "loss": 0.6245,
      "step": 1117
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7929234845869509,
      "learning_rate": 1.901331509872139e-05,
      "loss": 0.624,
      "step": 1118
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6596166910131636,
      "learning_rate": 1.9011200453590116e-05,
      "loss": 0.608,
      "step": 1119
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5980720813223348,
      "learning_rate": 1.900908366270231e-05,
      "loss": 0.6129,
      "step": 1120
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7387811449513924,
      "learning_rate": 1.900696472656203e-05,
      "loss": 0.6386,
      "step": 1121
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5724375550796177,
      "learning_rate": 1.9004843645673835e-05,
      "loss": 0.5886,
      "step": 1122
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7836876661643577,
      "learning_rate": 1.9002720420542803e-05,
      "loss": 0.6232,
      "step": 1123
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6754086768733204,
      "learning_rate": 1.9000595051674518e-05,
      "loss": 0.5962,
      "step": 1124
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6480013717184133,
      "learning_rate": 1.899846753957507e-05,
      "loss": 0.5988,
      "step": 1125
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9407573652766873,
      "learning_rate": 1.8996337884751064e-05,
      "loss": 0.6622,
      "step": 1126
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7077840393554568,
      "learning_rate": 1.8994206087709623e-05,
      "loss": 0.6209,
      "step": 1127
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6497066611366842,
      "learning_rate": 1.8992072148958368e-05,
      "loss": 0.6052,
      "step": 1128
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7617906318793758,
      "learning_rate": 1.8989936069005437e-05,
      "loss": 0.6246,
      "step": 1129
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6872349506110091,
      "learning_rate": 1.8987797848359472e-05,
      "loss": 0.6019,
      "step": 1130
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.678690296195298,
      "learning_rate": 1.8985657487529633e-05,
      "loss": 0.6147,
      "step": 1131
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7302829029570754,
      "learning_rate": 1.8983514987025583e-05,
      "loss": 0.6177,
      "step": 1132
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2157709066413114,
      "learning_rate": 1.8981370347357494e-05,
      "loss": 0.6283,
      "step": 1133
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.653970430100776,
      "learning_rate": 1.8979223569036055e-05,
      "loss": 0.6073,
      "step": 1134
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6996153039212468,
      "learning_rate": 1.8977074652572452e-05,
      "loss": 0.5997,
      "step": 1135
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6248268335494997,
      "learning_rate": 1.8974923598478393e-05,
      "loss": 0.616,
      "step": 1136
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7121465176834578,
      "learning_rate": 1.897277040726609e-05,
      "loss": 0.6401,
      "step": 1137
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6664548601637247,
      "learning_rate": 1.8970615079448254e-05,
      "loss": 0.5935,
      "step": 1138
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6885494748816947,
      "learning_rate": 1.8968457615538127e-05,
      "loss": 0.612,
      "step": 1139
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7014905840710198,
      "learning_rate": 1.8966298016049438e-05,
      "loss": 0.5959,
      "step": 1140
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6583741310448121,
      "learning_rate": 1.8964136281496433e-05,
      "loss": 0.6061,
      "step": 1141
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.74771599145121,
      "learning_rate": 1.8961972412393873e-05,
      "loss": 0.6209,
      "step": 1142
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6724490905215105,
      "learning_rate": 1.8959806409257014e-05,
      "loss": 0.5997,
      "step": 1143
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.689977991386939,
      "learning_rate": 1.895763827260163e-05,
      "loss": 0.5991,
      "step": 1144
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7688206535099406,
      "learning_rate": 1.8955468002943996e-05,
      "loss": 0.6067,
      "step": 1145
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7465508348181252,
      "learning_rate": 1.8953295600800906e-05,
      "loss": 0.6438,
      "step": 1146
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7144660589809594,
      "learning_rate": 1.8951121066689655e-05,
      "loss": 0.6166,
      "step": 1147
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6821923738068959,
      "learning_rate": 1.8948944401128035e-05,
      "loss": 0.6141,
      "step": 1148
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6798243622024949,
      "learning_rate": 1.8946765604634368e-05,
      "loss": 0.598,
      "step": 1149
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6762295400125014,
      "learning_rate": 1.894458467772746e-05,
      "loss": 0.6105,
      "step": 1150
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6323599813663727,
      "learning_rate": 1.8942401620926643e-05,
      "loss": 0.6102,
      "step": 1151
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.735944830363536,
      "learning_rate": 1.894021643475175e-05,
      "loss": 0.6006,
      "step": 1152
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7849402923644142,
      "learning_rate": 1.8938029119723113e-05,
      "loss": 0.6264,
      "step": 1153
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6508516193057435,
      "learning_rate": 1.8935839676361584e-05,
      "loss": 0.623,
      "step": 1154
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5953109368423845,
      "learning_rate": 1.893364810518851e-05,
      "loss": 0.584,
      "step": 1155
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7241317256321534,
      "learning_rate": 1.8931454406725757e-05,
      "loss": 0.6129,
      "step": 1156
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7738321677063196,
      "learning_rate": 1.8929258581495688e-05,
      "loss": 0.6138,
      "step": 1157
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.740174611352315,
      "learning_rate": 1.892706063002117e-05,
      "loss": 0.6084,
      "step": 1158
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6640459117950981,
      "learning_rate": 1.892486055282559e-05,
      "loss": 0.6179,
      "step": 1159
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6983428235514813,
      "learning_rate": 1.8922658350432827e-05,
      "loss": 0.593,
      "step": 1160
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7862488815475364,
      "learning_rate": 1.892045402336727e-05,
      "loss": 0.608,
      "step": 1161
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.783104072146204,
      "learning_rate": 1.8918247572153822e-05,
      "loss": 0.599,
      "step": 1162
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7540980557629522,
      "learning_rate": 1.8916038997317887e-05,
      "loss": 0.5853,
      "step": 1163
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7036595720670173,
      "learning_rate": 1.8913828299385362e-05,
      "loss": 0.6212,
      "step": 1164
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6262914801317764,
      "learning_rate": 1.8911615478882672e-05,
      "loss": 0.5979,
      "step": 1165
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7609632316384588,
      "learning_rate": 1.8909400536336728e-05,
      "loss": 0.6312,
      "step": 1166
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6749529991561528,
      "learning_rate": 1.890718347227496e-05,
      "loss": 0.6019,
      "step": 1167
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.757659687438456,
      "learning_rate": 1.8904964287225292e-05,
      "loss": 0.6208,
      "step": 1168
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6779168067839313,
      "learning_rate": 1.8902742981716166e-05,
      "loss": 0.6151,
      "step": 1169
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6285611224412462,
      "learning_rate": 1.890051955627652e-05,
      "loss": 0.5845,
      "step": 1170
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6978130891469554,
      "learning_rate": 1.889829401143579e-05,
      "loss": 0.6003,
      "step": 1171
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7037994104554686,
      "learning_rate": 1.889606634772394e-05,
      "loss": 0.618,
      "step": 1172
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.757307736113978,
      "learning_rate": 1.8893836565671408e-05,
      "loss": 0.6321,
      "step": 1173
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7129799666987509,
      "learning_rate": 1.8891604665809162e-05,
      "loss": 0.6216,
      "step": 1174
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7516344064587895,
      "learning_rate": 1.888937064866866e-05,
      "loss": 0.6477,
      "step": 1175
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.626312090841905,
      "learning_rate": 1.8887134514781872e-05,
      "loss": 0.6181,
      "step": 1176
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.647531970837858,
      "learning_rate": 1.8884896264681264e-05,
      "loss": 0.6046,
      "step": 1177
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6718429374816889,
      "learning_rate": 1.8882655898899812e-05,
      "loss": 0.6148,
      "step": 1178
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7965907973994819,
      "learning_rate": 1.8880413417970998e-05,
      "loss": 0.5996,
      "step": 1179
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9349930203970829,
      "learning_rate": 1.88781688224288e-05,
      "loss": 0.6477,
      "step": 1180
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6492622131233361,
      "learning_rate": 1.8875922112807706e-05,
      "loss": 0.6133,
      "step": 1181
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7956391802159468,
      "learning_rate": 1.88736732896427e-05,
      "loss": 0.633,
      "step": 1182
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7167718275226442,
      "learning_rate": 1.887142235346928e-05,
      "loss": 0.6252,
      "step": 1183
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0304570756126907,
      "learning_rate": 1.8869169304823438e-05,
      "loss": 0.6215,
      "step": 1184
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6508341604166205,
      "learning_rate": 1.8866914144241673e-05,
      "loss": 0.5971,
      "step": 1185
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.628981371071229,
      "learning_rate": 1.8864656872260985e-05,
      "loss": 0.6008,
      "step": 1186
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.633597438313175,
      "learning_rate": 1.8862397489418885e-05,
      "loss": 0.6292,
      "step": 1187
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.748834303622585,
      "learning_rate": 1.8860135996253368e-05,
      "loss": 0.6191,
      "step": 1188
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6679105788360218,
      "learning_rate": 1.8857872393302955e-05,
      "loss": 0.6099,
      "step": 1189
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6551033969723922,
      "learning_rate": 1.8855606681106646e-05,
      "loss": 0.6009,
      "step": 1190
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.790696630435795,
      "learning_rate": 1.8853338860203962e-05,
      "loss": 0.6476,
      "step": 1191
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6664792835664871,
      "learning_rate": 1.885106893113492e-05,
      "loss": 0.623,
      "step": 1192
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7193688370707544,
      "learning_rate": 1.884879689444003e-05,
      "loss": 0.6002,
      "step": 1193
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7846159068677512,
      "learning_rate": 1.884652275066032e-05,
      "loss": 0.5972,
      "step": 1194
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6083147385742068,
      "learning_rate": 1.8844246500337308e-05,
      "loss": 0.5959,
      "step": 1195
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6380843404306752,
      "learning_rate": 1.884196814401302e-05,
      "loss": 0.593,
      "step": 1196
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6775753530834959,
      "learning_rate": 1.8839687682229978e-05,
      "loss": 0.5932,
      "step": 1197
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6641883608071887,
      "learning_rate": 1.8837405115531205e-05,
      "loss": 0.6143,
      "step": 1198
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.67707700338518,
      "learning_rate": 1.883512044446023e-05,
      "loss": 0.6013,
      "step": 1199
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6906503134060769,
      "learning_rate": 1.8832833669561087e-05,
      "loss": 0.6174,
      "step": 1200
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6905376126504097,
      "learning_rate": 1.8830544791378298e-05,
      "loss": 0.6182,
      "step": 1201
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6691379371007519,
      "learning_rate": 1.8828253810456898e-05,
      "loss": 0.5816,
      "step": 1202
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6934543958456515,
      "learning_rate": 1.8825960727342414e-05,
      "loss": 0.6298,
      "step": 1203
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6703842238780334,
      "learning_rate": 1.8823665542580878e-05,
      "loss": 0.607,
      "step": 1204
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6209923644637304,
      "learning_rate": 1.8821368256718825e-05,
      "loss": 0.5997,
      "step": 1205
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6643050762247594,
      "learning_rate": 1.8819068870303287e-05,
      "loss": 0.5781,
      "step": 1206
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7010058687481772,
      "learning_rate": 1.881676738388179e-05,
      "loss": 0.6126,
      "step": 1207
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7254187959058769,
      "learning_rate": 1.881446379800237e-05,
      "loss": 0.592,
      "step": 1208
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8280311580275442,
      "learning_rate": 1.8812158113213564e-05,
      "loss": 0.6188,
      "step": 1209
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6572601547909908,
      "learning_rate": 1.88098503300644e-05,
      "loss": 0.5859,
      "step": 1210
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7470047892699333,
      "learning_rate": 1.880754044910441e-05,
      "loss": 0.5854,
      "step": 1211
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7433877746490177,
      "learning_rate": 1.8805228470883624e-05,
      "loss": 0.6126,
      "step": 1212
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7366815895788541,
      "learning_rate": 1.880291439595257e-05,
      "loss": 0.6077,
      "step": 1213
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.781311773277207,
      "learning_rate": 1.8800598224862286e-05,
      "loss": 0.6411,
      "step": 1214
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9416594611693238,
      "learning_rate": 1.8798279958164295e-05,
      "loss": 0.6392,
      "step": 1215
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6226921959393235,
      "learning_rate": 1.879595959641063e-05,
      "loss": 0.6006,
      "step": 1216
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.734371062875331,
      "learning_rate": 1.8793637140153812e-05,
      "loss": 0.6092,
      "step": 1217
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6265934121119363,
      "learning_rate": 1.8791312589946867e-05,
      "loss": 0.5753,
      "step": 1218
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.739933890470168,
      "learning_rate": 1.8788985946343327e-05,
      "loss": 0.631,
      "step": 1219
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6651510118343045,
      "learning_rate": 1.8786657209897207e-05,
      "loss": 0.6245,
      "step": 1220
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.699860025003005,
      "learning_rate": 1.8784326381163038e-05,
      "loss": 0.6393,
      "step": 1221
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7448662853086321,
      "learning_rate": 1.8781993460695824e-05,
      "loss": 0.6005,
      "step": 1222
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6419334156069902,
      "learning_rate": 1.8779658449051092e-05,
      "loss": 0.622,
      "step": 1223
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7766034497723208,
      "learning_rate": 1.8777321346784858e-05,
      "loss": 0.6185,
      "step": 1224
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6787863035821172,
      "learning_rate": 1.8774982154453632e-05,
      "loss": 0.6228,
      "step": 1225
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7977150059394935,
      "learning_rate": 1.877264087261443e-05,
      "loss": 0.619,
      "step": 1226
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6186982789475034,
      "learning_rate": 1.877029750182475e-05,
      "loss": 0.5862,
      "step": 1227
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7846290982893951,
      "learning_rate": 1.8767952042642613e-05,
      "loss": 0.6072,
      "step": 1228
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6784818380446809,
      "learning_rate": 1.876560449562651e-05,
      "loss": 0.6078,
      "step": 1229
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7190027314630455,
      "learning_rate": 1.8763254861335445e-05,
      "loss": 0.6421,
      "step": 1230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7172572088291781,
      "learning_rate": 1.8760903140328917e-05,
      "loss": 0.611,
      "step": 1231
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6481929553212784,
      "learning_rate": 1.875854933316692e-05,
      "loss": 0.5952,
      "step": 1232
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6957893552292598,
      "learning_rate": 1.8756193440409945e-05,
      "loss": 0.6051,
      "step": 1233
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6843199848871396,
      "learning_rate": 1.8753835462618976e-05,
      "loss": 0.6036,
      "step": 1234
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6729631080873538,
      "learning_rate": 1.87514754003555e-05,
      "loss": 0.5989,
      "step": 1235
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6937722187689618,
      "learning_rate": 1.8749113254181498e-05,
      "loss": 0.6302,
      "step": 1236
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6496506192291986,
      "learning_rate": 1.8746749024659445e-05,
      "loss": 0.6045,
      "step": 1237
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6821880173677086,
      "learning_rate": 1.8744382712352317e-05,
      "loss": 0.5841,
      "step": 1238
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8325669893729474,
      "learning_rate": 1.8742014317823583e-05,
      "loss": 0.6122,
      "step": 1239
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6775817378364123,
      "learning_rate": 1.8739643841637202e-05,
      "loss": 0.6048,
      "step": 1240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7005624618506732,
      "learning_rate": 1.873727128435764e-05,
      "loss": 0.6202,
      "step": 1241
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.646809548963158,
      "learning_rate": 1.873489664654985e-05,
      "loss": 0.6039,
      "step": 1242
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6534467257733058,
      "learning_rate": 1.873251992877928e-05,
      "loss": 0.5955,
      "step": 1243
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6218843859986373,
      "learning_rate": 1.8730141131611882e-05,
      "loss": 0.6332,
      "step": 1244
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6764841527194144,
      "learning_rate": 1.8727760255614097e-05,
      "loss": 0.6328,
      "step": 1245
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.65147064307679,
      "learning_rate": 1.872537730135286e-05,
      "loss": 0.6194,
      "step": 1246
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6629079700097793,
      "learning_rate": 1.8722992269395605e-05,
      "loss": 0.617,
      "step": 1247
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6320562636957038,
      "learning_rate": 1.8720605160310257e-05,
      "loss": 0.5951,
      "step": 1248
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0840614144930716,
      "learning_rate": 1.8718215974665235e-05,
      "loss": 0.6217,
      "step": 1249
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7613982674017729,
      "learning_rate": 1.8715824713029455e-05,
      "loss": 0.5902,
      "step": 1250
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6413666715749927,
      "learning_rate": 1.871343137597233e-05,
      "loss": 0.6032,
      "step": 1251
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6776398440877589,
      "learning_rate": 1.871103596406376e-05,
      "loss": 0.6329,
      "step": 1252
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6670358110642369,
      "learning_rate": 1.8708638477874145e-05,
      "loss": 0.5848,
      "step": 1253
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6616605479459893,
      "learning_rate": 1.8706238917974377e-05,
      "loss": 0.6154,
      "step": 1254
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7053092402354052,
      "learning_rate": 1.870383728493584e-05,
      "loss": 0.6008,
      "step": 1255
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6289172794422315,
      "learning_rate": 1.8701433579330414e-05,
      "loss": 0.6008,
      "step": 1256
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7395598363904446,
      "learning_rate": 1.8699027801730473e-05,
      "loss": 0.5871,
      "step": 1257
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7190368313758222,
      "learning_rate": 1.8696619952708885e-05,
      "loss": 0.609,
      "step": 1258
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5885669554042297,
      "learning_rate": 1.8694210032839005e-05,
      "loss": 0.605,
      "step": 1259
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6684946747451337,
      "learning_rate": 1.869179804269469e-05,
      "loss": 0.6096,
      "step": 1260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6823383925079861,
      "learning_rate": 1.8689383982850284e-05,
      "loss": 0.5932,
      "step": 1261
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6699977131549762,
      "learning_rate": 1.868696785388062e-05,
      "loss": 0.5998,
      "step": 1262
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7937523658241524,
      "learning_rate": 1.868454965636104e-05,
      "loss": 0.6377,
      "step": 1263
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6439563423813502,
      "learning_rate": 1.868212939086736e-05,
      "loss": 0.5853,
      "step": 1264
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6866838899936522,
      "learning_rate": 1.8679707057975894e-05,
      "loss": 0.5839,
      "step": 1265
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6899491190030382,
      "learning_rate": 1.867728265826346e-05,
      "loss": 0.6058,
      "step": 1266
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6950104285215248,
      "learning_rate": 1.8674856192307354e-05,
      "loss": 0.6143,
      "step": 1267
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7260061091964034,
      "learning_rate": 1.8672427660685365e-05,
      "loss": 0.5985,
      "step": 1268
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6245025921169199,
      "learning_rate": 1.8669997063975783e-05,
      "loss": 0.5827,
      "step": 1269
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7444614170550568,
      "learning_rate": 1.866756440275738e-05,
      "loss": 0.6215,
      "step": 1270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6908986143415199,
      "learning_rate": 1.8665129677609425e-05,
      "loss": 0.619,
      "step": 1271
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.672389082978755,
      "learning_rate": 1.8662692889111675e-05,
      "loss": 0.5882,
      "step": 1272
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6781752557207543,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.6244,
      "step": 1273
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.611840852726918,
      "learning_rate": 1.86578131243883e-05,
      "loss": 0.6001,
      "step": 1274
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6005031778860889,
      "learning_rate": 1.8655370149324647e-05,
      "loss": 0.5919,
      "step": 1275
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6985255416683156,
      "learning_rate": 1.8652925113235146e-05,
      "loss": 0.5905,
      "step": 1276
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6341385589765429,
      "learning_rate": 1.865047801670202e-05,
      "loss": 0.6146,
      "step": 1277
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7937355479809106,
      "learning_rate": 1.864802886030797e-05,
      "loss": 0.6317,
      "step": 1278
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5563881808053549,
      "learning_rate": 1.8645577644636193e-05,
      "loss": 0.5888,
      "step": 1279
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7436169555134403,
      "learning_rate": 1.8643124370270373e-05,
      "loss": 0.609,
      "step": 1280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7016076678107337,
      "learning_rate": 1.864066903779469e-05,
      "loss": 0.604,
      "step": 1281
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7087046298680165,
      "learning_rate": 1.8638211647793807e-05,
      "loss": 0.5847,
      "step": 1282
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6220950170659441,
      "learning_rate": 1.8635752200852878e-05,
      "loss": 0.6001,
      "step": 1283
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6839954771337048,
      "learning_rate": 1.8633290697557557e-05,
      "loss": 0.6163,
      "step": 1284
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7136462708046283,
      "learning_rate": 1.8630827138493975e-05,
      "loss": 0.6302,
      "step": 1285
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6854124690000145,
      "learning_rate": 1.8628361524248755e-05,
      "loss": 0.608,
      "step": 1286
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6417589019568359,
      "learning_rate": 1.8625893855409014e-05,
      "loss": 0.5929,
      "step": 1287
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6495994962437398,
      "learning_rate": 1.8623424132562358e-05,
      "loss": 0.6179,
      "step": 1288
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6401692158714323,
      "learning_rate": 1.8620952356296876e-05,
      "loss": 0.6054,
      "step": 1289
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.837689345421963,
      "learning_rate": 1.8618478527201154e-05,
      "loss": 0.5897,
      "step": 1290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7091224596062282,
      "learning_rate": 1.861600264586426e-05,
      "loss": 0.6405,
      "step": 1291
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7188137082843952,
      "learning_rate": 1.8613524712875755e-05,
      "loss": 0.5974,
      "step": 1292
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6583420997474698,
      "learning_rate": 1.8611044728825683e-05,
      "loss": 0.6102,
      "step": 1293
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.762995373161564,
      "learning_rate": 1.8608562694304592e-05,
      "loss": 0.617,
      "step": 1294
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.629620325384154,
      "learning_rate": 1.860607860990349e-05,
      "loss": 0.6091,
      "step": 1295
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6851207729177878,
      "learning_rate": 1.86035924762139e-05,
      "loss": 0.6072,
      "step": 1296
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7397728774676737,
      "learning_rate": 1.8601104293827823e-05,
      "loss": 0.6396,
      "step": 1297
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7481025027891457,
      "learning_rate": 1.8598614063337743e-05,
      "loss": 0.6199,
      "step": 1298
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.609958777845809,
      "learning_rate": 1.859612178533664e-05,
      "loss": 0.5821,
      "step": 1299
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6631949448094269,
      "learning_rate": 1.8593627460417977e-05,
      "loss": 0.578,
      "step": 1300
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.66590508797875,
      "learning_rate": 1.8591131089175704e-05,
      "loss": 0.626,
      "step": 1301
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6932519533065356,
      "learning_rate": 1.8588632672204264e-05,
      "loss": 0.6096,
      "step": 1302
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7538483036671589,
      "learning_rate": 1.8586132210098573e-05,
      "loss": 0.621,
      "step": 1303
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.692339565700301,
      "learning_rate": 1.8583629703454055e-05,
      "loss": 0.624,
      "step": 1304
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6699870118327373,
      "learning_rate": 1.85811251528666e-05,
      "loss": 0.6339,
      "step": 1305
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7439436315536981,
      "learning_rate": 1.85786185589326e-05,
      "loss": 0.5941,
      "step": 1306
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8939299830972885,
      "learning_rate": 1.8576109922248932e-05,
      "loss": 0.5845,
      "step": 1307
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6932556062593529,
      "learning_rate": 1.8573599243412944e-05,
      "loss": 0.6221,
      "step": 1308
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.620688322652189,
      "learning_rate": 1.8571086523022486e-05,
      "loss": 0.5959,
      "step": 1309
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6748136900310663,
      "learning_rate": 1.8568571761675893e-05,
      "loss": 0.6183,
      "step": 1310
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.647585683952202,
      "learning_rate": 1.8566054959971983e-05,
      "loss": 0.589,
      "step": 1311
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6154048179640372,
      "learning_rate": 1.856353611851005e-05,
      "loss": 0.6087,
      "step": 1312
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6879400516700018,
      "learning_rate": 1.8561015237889896e-05,
      "loss": 0.6344,
      "step": 1313
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.621074328990284,
      "learning_rate": 1.8558492318711786e-05,
      "loss": 0.5982,
      "step": 1314
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6646972662043361,
      "learning_rate": 1.8555967361576484e-05,
      "loss": 0.605,
      "step": 1315
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7004746716502712,
      "learning_rate": 1.8553440367085238e-05,
      "loss": 0.6062,
      "step": 1316
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7674931192315084,
      "learning_rate": 1.8550911335839774e-05,
      "loss": 0.6255,
      "step": 1317
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6448901833397968,
      "learning_rate": 1.854838026844231e-05,
      "loss": 0.6101,
      "step": 1318
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7157552799498506,
      "learning_rate": 1.8545847165495547e-05,
      "loss": 0.6183,
      "step": 1319
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7453839695576239,
      "learning_rate": 1.8543312027602668e-05,
      "loss": 0.6263,
      "step": 1320
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6912767521959561,
      "learning_rate": 1.8540774855367344e-05,
      "loss": 0.5928,
      "step": 1321
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9802039655323872,
      "learning_rate": 1.8538235649393727e-05,
      "loss": 0.6298,
      "step": 1322
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7101768435453327,
      "learning_rate": 1.853569441028646e-05,
      "loss": 0.6103,
      "step": 1323
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6608722386991565,
      "learning_rate": 1.8533151138650657e-05,
      "loss": 0.601,
      "step": 1324
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.683258517682998,
      "learning_rate": 1.8530605835091936e-05,
      "loss": 0.6348,
      "step": 1325
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0832439526830258,
      "learning_rate": 1.8528058500216383e-05,
      "loss": 0.6543,
      "step": 1326
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6719163422781431,
      "learning_rate": 1.8525509134630566e-05,
      "loss": 0.5926,
      "step": 1327
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6598967036085842,
      "learning_rate": 1.852295773894155e-05,
      "loss": 0.6023,
      "step": 1328
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6571679691944511,
      "learning_rate": 1.852040431375687e-05,
      "loss": 0.6126,
      "step": 1329
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7033990627425559,
      "learning_rate": 1.851784885968456e-05,
      "loss": 0.6005,
      "step": 1330
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5594946758688845,
      "learning_rate": 1.8515291377333114e-05,
      "loss": 0.5997,
      "step": 1331
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6572170844403138,
      "learning_rate": 1.8512731867311534e-05,
      "loss": 0.5941,
      "step": 1332
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7425063567374167,
      "learning_rate": 1.851017033022929e-05,
      "loss": 0.6108,
      "step": 1333
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.769746075976278,
      "learning_rate": 1.850760676669633e-05,
      "loss": 0.5985,
      "step": 1334
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6314107635965408,
      "learning_rate": 1.85050411773231e-05,
      "loss": 0.5976,
      "step": 1335
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7472165557622943,
      "learning_rate": 1.8502473562720525e-05,
      "loss": 0.6294,
      "step": 1336
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9448569369501024,
      "learning_rate": 1.8499903923500002e-05,
      "loss": 0.634,
      "step": 1337
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6505544483157228,
      "learning_rate": 1.8497332260273415e-05,
      "loss": 0.6311,
      "step": 1338
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7678032688573174,
      "learning_rate": 1.8494758573653134e-05,
      "loss": 0.6065,
      "step": 1339
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7571525903381319,
      "learning_rate": 1.849218286425201e-05,
      "loss": 0.6416,
      "step": 1340
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5990920639311724,
      "learning_rate": 1.8489605132683365e-05,
      "loss": 0.5865,
      "step": 1341
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7156713903519215,
      "learning_rate": 1.848702537956102e-05,
      "loss": 0.6096,
      "step": 1342
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8244507395395435,
      "learning_rate": 1.8484443605499266e-05,
      "loss": 0.6336,
      "step": 1343
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7129649636476977,
      "learning_rate": 1.848185981111288e-05,
      "loss": 0.6277,
      "step": 1344
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6150630282944628,
      "learning_rate": 1.8479273997017112e-05,
      "loss": 0.6028,
      "step": 1345
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7114940614080478,
      "learning_rate": 1.84766861638277e-05,
      "loss": 0.6497,
      "step": 1346
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7206003460152216,
      "learning_rate": 1.8474096312160866e-05,
      "loss": 0.62,
      "step": 1347
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.823264629854062,
      "learning_rate": 1.8471504442633304e-05,
      "loss": 0.6747,
      "step": 1348
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6698466721224119,
      "learning_rate": 1.8468910555862196e-05,
      "loss": 0.6051,
      "step": 1349
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.743519924900364,
      "learning_rate": 1.84663146524652e-05,
      "loss": 0.6188,
      "step": 1350
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.671227899298752,
      "learning_rate": 1.8463716733060454e-05,
      "loss": 0.6156,
      "step": 1351
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6652401860263372,
      "learning_rate": 1.846111679826658e-05,
      "loss": 0.6037,
      "step": 1352
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6606559149803369,
      "learning_rate": 1.845851484870267e-05,
      "loss": 0.5991,
      "step": 1353
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7397604109539303,
      "learning_rate": 1.8455910884988312e-05,
      "loss": 0.6266,
      "step": 1354
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7791463268843197,
      "learning_rate": 1.8453304907743562e-05,
      "loss": 0.6287,
      "step": 1355
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7121063523855554,
      "learning_rate": 1.8450696917588954e-05,
      "loss": 0.6052,
      "step": 1356
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.720672157231405,
      "learning_rate": 1.8448086915145508e-05,
      "loss": 0.6147,
      "step": 1357
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7416606372996478,
      "learning_rate": 1.844547490103472e-05,
      "loss": 0.588,
      "step": 1358
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6137346541981364,
      "learning_rate": 1.844286087587857e-05,
      "loss": 0.5849,
      "step": 1359
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6619140973683831,
      "learning_rate": 1.8440244840299507e-05,
      "loss": 0.5955,
      "step": 1360
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.693812437575388,
      "learning_rate": 1.8437626794920464e-05,
      "loss": 0.6003,
      "step": 1361
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7074812874286074,
      "learning_rate": 1.8435006740364857e-05,
      "loss": 0.6169,
      "step": 1362
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8482117607998818,
      "learning_rate": 1.843238467725657e-05,
      "loss": 0.6151,
      "step": 1363
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7036560333733294,
      "learning_rate": 1.8429760606219974e-05,
      "loss": 0.602,
      "step": 1364
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6779939873773297,
      "learning_rate": 1.8427134527879923e-05,
      "loss": 0.5933,
      "step": 1365
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.728371308104298,
      "learning_rate": 1.8424506442861733e-05,
      "loss": 0.6249,
      "step": 1366
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8552058382768243,
      "learning_rate": 1.8421876351791206e-05,
      "loss": 0.6102,
      "step": 1367
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6400275362370907,
      "learning_rate": 1.8419244255294628e-05,
      "loss": 0.6307,
      "step": 1368
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6670257316892859,
      "learning_rate": 1.8416610153998748e-05,
      "loss": 0.6026,
      "step": 1369
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7437451674854967,
      "learning_rate": 1.8413974048530813e-05,
      "loss": 0.6221,
      "step": 1370
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7603222384184396,
      "learning_rate": 1.8411335939518523e-05,
      "loss": 0.6259,
      "step": 1371
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7076891949121623,
      "learning_rate": 1.8408695827590073e-05,
      "loss": 0.6254,
      "step": 1372
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7076821062555306,
      "learning_rate": 1.840605371337413e-05,
      "loss": 0.5975,
      "step": 1373
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7192916728737218,
      "learning_rate": 1.8403409597499835e-05,
      "loss": 0.6012,
      "step": 1374
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6079168696681151,
      "learning_rate": 1.840076348059681e-05,
      "loss": 0.5809,
      "step": 1375
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6570987008965246,
      "learning_rate": 1.8398115363295152e-05,
      "loss": 0.6036,
      "step": 1376
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7415338909943806,
      "learning_rate": 1.8395465246225425e-05,
      "loss": 0.5911,
      "step": 1377
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6280659288715535,
      "learning_rate": 1.8392813130018687e-05,
      "loss": 0.6087,
      "step": 1378
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.806311034409592,
      "learning_rate": 1.839015901530646e-05,
      "loss": 0.612,
      "step": 1379
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7196400039552057,
      "learning_rate": 1.8387502902720744e-05,
      "loss": 0.6102,
      "step": 1380
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.708825368872833,
      "learning_rate": 1.8384844792894014e-05,
      "loss": 0.608,
      "step": 1381
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7416269893998962,
      "learning_rate": 1.8382184686459225e-05,
      "loss": 0.6123,
      "step": 1382
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6585064436285797,
      "learning_rate": 1.8379522584049803e-05,
      "loss": 0.5701,
      "step": 1383
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6469859375284992,
      "learning_rate": 1.837685848629965e-05,
      "loss": 0.6171,
      "step": 1384
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6949442788513889,
      "learning_rate": 1.8374192393843143e-05,
      "loss": 0.5967,
      "step": 1385
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6542662047918175,
      "learning_rate": 1.8371524307315135e-05,
      "loss": 0.5939,
      "step": 1386
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5779422073734575,
      "learning_rate": 1.8368854227350955e-05,
      "loss": 0.5996,
      "step": 1387
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6657134599704209,
      "learning_rate": 1.8366182154586408e-05,
      "loss": 0.6074,
      "step": 1388
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7575887339266892,
      "learning_rate": 1.8363508089657763e-05,
      "loss": 0.6136,
      "step": 1389
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6875575931566412,
      "learning_rate": 1.8360832033201777e-05,
      "loss": 0.6109,
      "step": 1390
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7300024114146781,
      "learning_rate": 1.8358153985855675e-05,
      "loss": 0.5968,
      "step": 1391
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6635740973858345,
      "learning_rate": 1.8355473948257156e-05,
      "loss": 0.6006,
      "step": 1392
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7019879288800099,
      "learning_rate": 1.835279192104439e-05,
      "loss": 0.5893,
      "step": 1393
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7291348805929987,
      "learning_rate": 1.8350107904856026e-05,
      "loss": 0.6289,
      "step": 1394
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7682569685370039,
      "learning_rate": 1.834742190033119e-05,
      "loss": 0.6248,
      "step": 1395
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6487487205197013,
      "learning_rate": 1.834473390810947e-05,
      "loss": 0.6275,
      "step": 1396
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6733024861719364,
      "learning_rate": 1.8342043928830932e-05,
      "loss": 0.6035,
      "step": 1397
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6448336678304265,
      "learning_rate": 1.833935196313612e-05,
      "loss": 0.5704,
      "step": 1398
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7344428685717422,
      "learning_rate": 1.8336658011666055e-05,
      "loss": 0.6247,
      "step": 1399
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7501724430715377,
      "learning_rate": 1.833396207506221e-05,
      "loss": 0.6223,
      "step": 1400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7098121746180656,
      "learning_rate": 1.833126415396655e-05,
      "loss": 0.609,
      "step": 1401
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6997361099017182,
      "learning_rate": 1.8328564249021514e-05,
      "loss": 0.6215,
      "step": 1402
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6273379198996384,
      "learning_rate": 1.8325862360869994e-05,
      "loss": 0.6153,
      "step": 1403
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7986512982446448,
      "learning_rate": 1.8323158490155374e-05,
      "loss": 0.6729,
      "step": 1404
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6925630739847997,
      "learning_rate": 1.83204526375215e-05,
      "loss": 0.6087,
      "step": 1405
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6580449806724031,
      "learning_rate": 1.8317744803612693e-05,
      "loss": 0.6103,
      "step": 1406
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6427837434696525,
      "learning_rate": 1.831503498907375e-05,
      "loss": 0.5715,
      "step": 1407
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6685940052171028,
      "learning_rate": 1.8312323194549923e-05,
      "loss": 0.5912,
      "step": 1408
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.628834127445035,
      "learning_rate": 1.8309609420686956e-05,
      "loss": 0.588,
      "step": 1409
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6394856047099526,
      "learning_rate": 1.8306893668131058e-05,
      "loss": 0.6074,
      "step": 1410
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7473246883832265,
      "learning_rate": 1.8304175937528902e-05,
      "loss": 0.6204,
      "step": 1411
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6940342891817015,
      "learning_rate": 1.8301456229527637e-05,
      "loss": 0.6281,
      "step": 1412
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7162934646060891,
      "learning_rate": 1.8298734544774882e-05,
      "loss": 0.6025,
      "step": 1413
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.765291735550678,
      "learning_rate": 1.829601088391873e-05,
      "loss": 0.6306,
      "step": 1414
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7514587463274455,
      "learning_rate": 1.8293285247607743e-05,
      "loss": 0.6138,
      "step": 1415
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.752703928094448,
      "learning_rate": 1.829055763649095e-05,
      "loss": 0.6426,
      "step": 1416
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6572397542544877,
      "learning_rate": 1.8287828051217854e-05,
      "loss": 0.5847,
      "step": 1417
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6654028515062426,
      "learning_rate": 1.8285096492438424e-05,
      "loss": 0.6118,
      "step": 1418
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7316270184760694,
      "learning_rate": 1.82823629608031e-05,
      "loss": 0.6188,
      "step": 1419
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8799736180032971,
      "learning_rate": 1.8279627456962804e-05,
      "loss": 0.6506,
      "step": 1420
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6424608830424703,
      "learning_rate": 1.827688998156891e-05,
      "loss": 0.6183,
      "step": 1421
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7758821134718052,
      "learning_rate": 1.8274150535273264e-05,
      "loss": 0.6279,
      "step": 1422
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7389471857209832,
      "learning_rate": 1.8271409118728192e-05,
      "loss": 0.608,
      "step": 1423
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6446073920420484,
      "learning_rate": 1.826866573258648e-05,
      "loss": 0.5863,
      "step": 1424
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6583713479497387,
      "learning_rate": 1.826592037750139e-05,
      "loss": 0.5969,
      "step": 1425
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6523620901892393,
      "learning_rate": 1.8263173054126643e-05,
      "loss": 0.6075,
      "step": 1426
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7561852741945351,
      "learning_rate": 1.826042376311644e-05,
      "loss": 0.6382,
      "step": 1427
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7253448834258918,
      "learning_rate": 1.8257672505125445e-05,
      "loss": 0.6139,
      "step": 1428
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5826173386378104,
      "learning_rate": 1.8254919280808782e-05,
      "loss": 0.5826,
      "step": 1429
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7556307305212318,
      "learning_rate": 1.8252164090822064e-05,
      "loss": 0.6175,
      "step": 1430
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6698997550705017,
      "learning_rate": 1.824940693582135e-05,
      "loss": 0.5754,
      "step": 1431
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7634538137874116,
      "learning_rate": 1.824664781646318e-05,
      "loss": 0.6335,
      "step": 1432
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8255138945263221,
      "learning_rate": 1.8243886733404563e-05,
      "loss": 0.6413,
      "step": 1433
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6719641983847138,
      "learning_rate": 1.824112368730296e-05,
      "loss": 0.6236,
      "step": 1434
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6479593619707441,
      "learning_rate": 1.8238358678816324e-05,
      "loss": 0.6001,
      "step": 1435
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.704005739362322,
      "learning_rate": 1.823559170860305e-05,
      "loss": 0.6059,
      "step": 1436
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6694476091993573,
      "learning_rate": 1.823282277732202e-05,
      "loss": 0.6088,
      "step": 1437
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6096768262720234,
      "learning_rate": 1.823005188563257e-05,
      "loss": 0.6085,
      "step": 1438
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5924304700397942,
      "learning_rate": 1.822727903419451e-05,
      "loss": 0.605,
      "step": 1439
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7296037616806325,
      "learning_rate": 1.8224504223668112e-05,
      "loss": 0.6133,
      "step": 1440
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6399834497097241,
      "learning_rate": 1.822172745471412e-05,
      "loss": 0.6156,
      "step": 1441
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7034970282958536,
      "learning_rate": 1.8218948727993736e-05,
      "loss": 0.5988,
      "step": 1442
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6389690164544449,
      "learning_rate": 1.821616804416864e-05,
      "loss": 0.607,
      "step": 1443
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6355616973926238,
      "learning_rate": 1.8213385403900966e-05,
      "loss": 0.5657,
      "step": 1444
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7693668330294495,
      "learning_rate": 1.8210600807853325e-05,
      "loss": 0.619,
      "step": 1445
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6634689930150585,
      "learning_rate": 1.820781425668878e-05,
      "loss": 0.6292,
      "step": 1446
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5803326882578396,
      "learning_rate": 1.8205025751070878e-05,
      "loss": 0.5856,
      "step": 1447
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6648963779702213,
      "learning_rate": 1.820223529166361e-05,
      "loss": 0.6383,
      "step": 1448
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6491000312028189,
      "learning_rate": 1.819944287913145e-05,
      "loss": 0.6032,
      "step": 1449
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6260646229032659,
      "learning_rate": 1.8196648514139325e-05,
      "loss": 0.5961,
      "step": 1450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5786173991744632,
      "learning_rate": 1.8193852197352636e-05,
      "loss": 0.5982,
      "step": 1451
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7196657891064646,
      "learning_rate": 1.8191053929437243e-05,
      "loss": 0.6235,
      "step": 1452
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7145303616085551,
      "learning_rate": 1.8188253711059475e-05,
      "loss": 0.6016,
      "step": 1453
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8501057757927852,
      "learning_rate": 1.8185451542886123e-05,
      "loss": 0.6242,
      "step": 1454
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7164636649712791,
      "learning_rate": 1.818264742558444e-05,
      "loss": 0.6056,
      "step": 1455
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8079830116289449,
      "learning_rate": 1.8179841359822144e-05,
      "loss": 0.6073,
      "step": 1456
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.700980356084714,
      "learning_rate": 1.8177033346267424e-05,
      "loss": 0.6091,
      "step": 1457
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6245240187181126,
      "learning_rate": 1.817422338558892e-05,
      "loss": 0.6098,
      "step": 1458
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7681744612738487,
      "learning_rate": 1.8171411478455746e-05,
      "loss": 0.6112,
      "step": 1459
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7291457634414602,
      "learning_rate": 1.816859762553748e-05,
      "loss": 0.5986,
      "step": 1460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7163585885185013,
      "learning_rate": 1.8165781827504155e-05,
      "loss": 0.6109,
      "step": 1461
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.662558694818876,
      "learning_rate": 1.8162964085026272e-05,
      "loss": 0.5881,
      "step": 1462
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7270804748554932,
      "learning_rate": 1.8160144398774797e-05,
      "loss": 0.6249,
      "step": 1463
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6940547857543679,
      "learning_rate": 1.8157322769421155e-05,
      "loss": 0.61,
      "step": 1464
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.608830755058314,
      "learning_rate": 1.8154499197637236e-05,
      "loss": 0.5764,
      "step": 1465
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6865510093428993,
      "learning_rate": 1.8151673684095393e-05,
      "loss": 0.5933,
      "step": 1466
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6605733277436211,
      "learning_rate": 1.8148846229468436e-05,
      "loss": 0.5807,
      "step": 1467
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6122216841984044,
      "learning_rate": 1.814601683442965e-05,
      "loss": 0.608,
      "step": 1468
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6563814543699601,
      "learning_rate": 1.8143185499652757e-05,
      "loss": 0.5887,
      "step": 1469
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7749347903110269,
      "learning_rate": 1.8140352225811976e-05,
      "loss": 0.6284,
      "step": 1470
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.625845160717316,
      "learning_rate": 1.8137517013581963e-05,
      "loss": 0.6116,
      "step": 1471
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6261979828188207,
      "learning_rate": 1.813467986363784e-05,
      "loss": 0.5851,
      "step": 1472
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6832901340929168,
      "learning_rate": 1.8131840776655186e-05,
      "loss": 0.6274,
      "step": 1473
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6155017837645719,
      "learning_rate": 1.8128999753310062e-05,
      "loss": 0.6068,
      "step": 1474
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7319082540661226,
      "learning_rate": 1.8126156794278962e-05,
      "loss": 0.6015,
      "step": 1475
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7177187549517242,
      "learning_rate": 1.812331190023886e-05,
      "loss": 0.5994,
      "step": 1476
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.841668291531027,
      "learning_rate": 1.8120465071867186e-05,
      "loss": 0.6593,
      "step": 1477
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.624823778695531,
      "learning_rate": 1.811761630984183e-05,
      "loss": 0.6008,
      "step": 1478
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6173893487577957,
      "learning_rate": 1.8114765614841137e-05,
      "loss": 0.588,
      "step": 1479
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.714156505067405,
      "learning_rate": 1.8111912987543924e-05,
      "loss": 0.5907,
      "step": 1480
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7015681749574461,
      "learning_rate": 1.8109058428629457e-05,
      "loss": 0.5761,
      "step": 1481
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6163368245250909,
      "learning_rate": 1.810620193877747e-05,
      "loss": 0.6041,
      "step": 1482
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7211095578750794,
      "learning_rate": 1.810334351866815e-05,
      "loss": 0.5969,
      "step": 1483
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6531846740090711,
      "learning_rate": 1.8100483168982152e-05,
      "loss": 0.5993,
      "step": 1484
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5462564217345839,
      "learning_rate": 1.8097620890400583e-05,
      "loss": 0.5902,
      "step": 1485
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7825269043066544,
      "learning_rate": 1.809475668360501e-05,
      "loss": 0.6391,
      "step": 1486
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7498224791612242,
      "learning_rate": 1.809189054927746e-05,
      "loss": 0.6037,
      "step": 1487
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6742705217776627,
      "learning_rate": 1.808902248810042e-05,
      "loss": 0.6201,
      "step": 1488
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7201509830200779,
      "learning_rate": 1.808615250075684e-05,
      "loss": 0.6083,
      "step": 1489
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9593820322463581,
      "learning_rate": 1.8083280587930124e-05,
      "loss": 0.6139,
      "step": 1490
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6653898265950247,
      "learning_rate": 1.8080406750304133e-05,
      "loss": 0.5985,
      "step": 1491
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6333732309276222,
      "learning_rate": 1.8077530988563184e-05,
      "loss": 0.5977,
      "step": 1492
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7068290490972479,
      "learning_rate": 1.8074653303392063e-05,
      "loss": 0.612,
      "step": 1493
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6655911922665514,
      "learning_rate": 1.8071773695476003e-05,
      "loss": 0.6087,
      "step": 1494
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7037232182177893,
      "learning_rate": 1.8068892165500704e-05,
      "loss": 0.6046,
      "step": 1495
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8332481803465059,
      "learning_rate": 1.8066008714152317e-05,
      "loss": 0.612,
      "step": 1496
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.97851199142385,
      "learning_rate": 1.806312334211745e-05,
      "loss": 0.584,
      "step": 1497
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0906653592805649,
      "learning_rate": 1.8060236050083173e-05,
      "loss": 0.604,
      "step": 1498
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6750254396346869,
      "learning_rate": 1.805734683873701e-05,
      "loss": 0.6044,
      "step": 1499
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7548251350384798,
      "learning_rate": 1.8054455708766946e-05,
      "loss": 0.6009,
      "step": 1500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8066628111971161,
      "learning_rate": 1.8051562660861415e-05,
      "loss": 0.6154,
      "step": 1501
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6736627724845086,
      "learning_rate": 1.8048667695709317e-05,
      "loss": 0.6091,
      "step": 1502
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7740036871724276,
      "learning_rate": 1.8045770813999998e-05,
      "loss": 0.5908,
      "step": 1503
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.680261199903679,
      "learning_rate": 1.8042872016423274e-05,
      "loss": 0.6013,
      "step": 1504
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6396884918857213,
      "learning_rate": 1.8039971303669407e-05,
      "loss": 0.6343,
      "step": 1505
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.616987353136675,
      "learning_rate": 1.8037068676429114e-05,
      "loss": 0.5943,
      "step": 1506
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7532150405826352,
      "learning_rate": 1.803416413539358e-05,
      "loss": 0.6276,
      "step": 1507
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.771640055624305,
      "learning_rate": 1.803125768125443e-05,
      "loss": 0.6285,
      "step": 1508
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5790149375136145,
      "learning_rate": 1.8028349314703754e-05,
      "loss": 0.5848,
      "step": 1509
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5842107641865937,
      "learning_rate": 1.8025439036434094e-05,
      "loss": 0.5863,
      "step": 1510
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7109911826110628,
      "learning_rate": 1.8022526847138454e-05,
      "loss": 0.5903,
      "step": 1511
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8362043397795474,
      "learning_rate": 1.801961274751028e-05,
      "loss": 0.6208,
      "step": 1512
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7099315953309663,
      "learning_rate": 1.8016696738243483e-05,
      "loss": 0.6211,
      "step": 1513
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9464040347520712,
      "learning_rate": 1.801377882003243e-05,
      "loss": 0.5855,
      "step": 1514
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.787071963393706,
      "learning_rate": 1.8010858993571937e-05,
      "loss": 0.612,
      "step": 1515
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6902309953818847,
      "learning_rate": 1.8007937259557274e-05,
      "loss": 0.6087,
      "step": 1516
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6535834593500894,
      "learning_rate": 1.8005013618684168e-05,
      "loss": 0.6228,
      "step": 1517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7034898638756387,
      "learning_rate": 1.80020880716488e-05,
      "loss": 0.6057,
      "step": 1518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6296100446897315,
      "learning_rate": 1.7999160619147806e-05,
      "loss": 0.5951,
      "step": 1519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.692440406451605,
      "learning_rate": 1.799623126187827e-05,
      "loss": 0.6003,
      "step": 1520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6222812074369692,
      "learning_rate": 1.799330000053774e-05,
      "loss": 0.5995,
      "step": 1521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6829017155190036,
      "learning_rate": 1.79903668358242e-05,
      "loss": 0.6165,
      "step": 1522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6746693589023088,
      "learning_rate": 1.798743176843611e-05,
      "loss": 0.5776,
      "step": 1523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7927548992249484,
      "learning_rate": 1.798449479907237e-05,
      "loss": 0.6312,
      "step": 1524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6668006044078469,
      "learning_rate": 1.7981555928432327e-05,
      "loss": 0.6314,
      "step": 1525
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7442894247675637,
      "learning_rate": 1.7978615157215792e-05,
      "loss": 0.6054,
      "step": 1526
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8210941241125133,
      "learning_rate": 1.7975672486123025e-05,
      "loss": 0.6109,
      "step": 1527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7327439447479919,
      "learning_rate": 1.7972727915854735e-05,
      "loss": 0.6358,
      "step": 1528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6567656936567128,
      "learning_rate": 1.796978144711209e-05,
      "loss": 0.6119,
      "step": 1529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.622343211881444,
      "learning_rate": 1.7966833080596705e-05,
      "loss": 0.5732,
      "step": 1530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7262208526609747,
      "learning_rate": 1.7963882817010647e-05,
      "loss": 0.6204,
      "step": 1531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7276165026948189,
      "learning_rate": 1.796093065705644e-05,
      "loss": 0.6085,
      "step": 1532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7017678400516972,
      "learning_rate": 1.795797660143705e-05,
      "loss": 0.6238,
      "step": 1533
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6576084758753449,
      "learning_rate": 1.79550206508559e-05,
      "loss": 0.6313,
      "step": 1534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6520695198310542,
      "learning_rate": 1.7952062806016867e-05,
      "loss": 0.6146,
      "step": 1535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7971406617003554,
      "learning_rate": 1.7949103067624277e-05,
      "loss": 0.6121,
      "step": 1536
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6331819849144062,
      "learning_rate": 1.7946141436382904e-05,
      "loss": 0.6084,
      "step": 1537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7345520801458834,
      "learning_rate": 1.7943177912997974e-05,
      "loss": 0.6159,
      "step": 1538
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6735543389182002,
      "learning_rate": 1.7940212498175168e-05,
      "loss": 0.5924,
      "step": 1539
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6713318787512268,
      "learning_rate": 1.7937245192620606e-05,
      "loss": 0.5955,
      "step": 1540
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.709768580897599,
      "learning_rate": 1.7934275997040873e-05,
      "loss": 0.5918,
      "step": 1541
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7377770582719555,
      "learning_rate": 1.7931304912142998e-05,
      "loss": 0.6093,
      "step": 1542
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.745132911065759,
      "learning_rate": 1.7928331938634452e-05,
      "loss": 0.6135,
      "step": 1543
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6636442930973059,
      "learning_rate": 1.792535707722317e-05,
      "loss": 0.6196,
      "step": 1544
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.678822667004372,
      "learning_rate": 1.792238032861752e-05,
      "loss": 0.6231,
      "step": 1545
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6887928761150991,
      "learning_rate": 1.7919401693526338e-05,
      "loss": 0.6152,
      "step": 1546
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7171621835617604,
      "learning_rate": 1.7916421172658892e-05,
      "loss": 0.6018,
      "step": 1547
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6610611896033312,
      "learning_rate": 1.7913438766724914e-05,
      "loss": 0.6059,
      "step": 1548
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6791978521750266,
      "learning_rate": 1.7910454476434574e-05,
      "loss": 0.5908,
      "step": 1549
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6583857776054884,
      "learning_rate": 1.7907468302498495e-05,
      "loss": 0.5804,
      "step": 1550
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.649205759953637,
      "learning_rate": 1.7904480245627743e-05,
      "loss": 0.5864,
      "step": 1551
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7102182997785454,
      "learning_rate": 1.7901490306533845e-05,
      "loss": 0.6304,
      "step": 1552
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6850204848873105,
      "learning_rate": 1.7898498485928764e-05,
      "loss": 0.6209,
      "step": 1553
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7157618616317725,
      "learning_rate": 1.7895504784524917e-05,
      "loss": 0.5995,
      "step": 1554
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7045348020534744,
      "learning_rate": 1.7892509203035166e-05,
      "loss": 0.6049,
      "step": 1555
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5609216767570527,
      "learning_rate": 1.7889511742172822e-05,
      "loss": 0.6111,
      "step": 1556
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6974812723793149,
      "learning_rate": 1.7886512402651645e-05,
      "loss": 0.6078,
      "step": 1557
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6055367888142633,
      "learning_rate": 1.7883511185185843e-05,
      "loss": 0.5891,
      "step": 1558
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6846047767618952,
      "learning_rate": 1.7880508090490062e-05,
      "loss": 0.6311,
      "step": 1559
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6635442883323622,
      "learning_rate": 1.7877503119279408e-05,
      "loss": 0.5941,
      "step": 1560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7487780329391045,
      "learning_rate": 1.7874496272269426e-05,
      "loss": 0.6345,
      "step": 1561
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6405424324516873,
      "learning_rate": 1.787148755017611e-05,
      "loss": 0.6233,
      "step": 1562
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5385903714246424,
      "learning_rate": 1.78684769537159e-05,
      "loss": 0.5827,
      "step": 1563
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6186053134980374,
      "learning_rate": 1.7865464483605684e-05,
      "loss": 0.5897,
      "step": 1564
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6799870069217395,
      "learning_rate": 1.786245014056279e-05,
      "loss": 0.5902,
      "step": 1565
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6762890807601928,
      "learning_rate": 1.7859433925305e-05,
      "loss": 0.594,
      "step": 1566
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7321030343357319,
      "learning_rate": 1.785641583855054e-05,
      "loss": 0.6275,
      "step": 1567
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6873771352360233,
      "learning_rate": 1.7853395881018075e-05,
      "loss": 0.6086,
      "step": 1568
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.612658292844178,
      "learning_rate": 1.7850374053426725e-05,
      "loss": 0.579,
      "step": 1569
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.687550724447703,
      "learning_rate": 1.7847350356496047e-05,
      "loss": 0.6242,
      "step": 1570
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6417235985417862,
      "learning_rate": 1.784432479094605e-05,
      "loss": 0.5896,
      "step": 1571
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6216009991159868,
      "learning_rate": 1.7841297357497184e-05,
      "loss": 0.5783,
      "step": 1572
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6191670417634182,
      "learning_rate": 1.7838268056870345e-05,
      "loss": 0.5849,
      "step": 1573
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6607469357167972,
      "learning_rate": 1.7835236889786868e-05,
      "loss": 0.607,
      "step": 1574
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6335156292041666,
      "learning_rate": 1.7832203856968545e-05,
      "loss": 0.6168,
      "step": 1575
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8192439371568366,
      "learning_rate": 1.7829168959137606e-05,
      "loss": 0.6098,
      "step": 1576
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7973182555911664,
      "learning_rate": 1.7826132197016715e-05,
      "loss": 0.6117,
      "step": 1577
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7030101811479327,
      "learning_rate": 1.7823093571328998e-05,
      "loss": 0.596,
      "step": 1578
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6875156494129674,
      "learning_rate": 1.782005308279801e-05,
      "loss": 0.594,
      "step": 1579
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7796457396436464,
      "learning_rate": 1.7817010732147758e-05,
      "loss": 0.6413,
      "step": 1580
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8682030350634626,
      "learning_rate": 1.781396652010269e-05,
      "loss": 0.6127,
      "step": 1581
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7719788101277362,
      "learning_rate": 1.7810920447387696e-05,
      "loss": 0.6238,
      "step": 1582
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7837033417805459,
      "learning_rate": 1.7807872514728105e-05,
      "loss": 0.5915,
      "step": 1583
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7026204676150098,
      "learning_rate": 1.7804822722849704e-05,
      "loss": 0.6135,
      "step": 1584
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7703128633955472,
      "learning_rate": 1.7801771072478705e-05,
      "loss": 0.5921,
      "step": 1585
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7767660189640087,
      "learning_rate": 1.7798717564341775e-05,
      "loss": 0.546,
      "step": 1586
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6772164974890791,
      "learning_rate": 1.7795662199166017e-05,
      "loss": 0.5782,
      "step": 1587
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7164433441462114,
      "learning_rate": 1.7792604977678974e-05,
      "loss": 0.6038,
      "step": 1588
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.775816401608737,
      "learning_rate": 1.778954590060864e-05,
      "loss": 0.5894,
      "step": 1589
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6648952758120988,
      "learning_rate": 1.7786484968683442e-05,
      "loss": 0.617,
      "step": 1590
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7402192312157746,
      "learning_rate": 1.7783422182632257e-05,
      "loss": 0.6247,
      "step": 1591
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6841046264626502,
      "learning_rate": 1.7780357543184396e-05,
      "loss": 0.5717,
      "step": 1592
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7035949815616438,
      "learning_rate": 1.7777291051069614e-05,
      "loss": 0.6359,
      "step": 1593
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6361019025947726,
      "learning_rate": 1.7774222707018106e-05,
      "loss": 0.6028,
      "step": 1594
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7845644615190445,
      "learning_rate": 1.777115251176051e-05,
      "loss": 0.6234,
      "step": 1595
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.709414797123088,
      "learning_rate": 1.7768080466027904e-05,
      "loss": 0.6084,
      "step": 1596
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7173112173811834,
      "learning_rate": 1.776500657055181e-05,
      "loss": 0.6149,
      "step": 1597
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7097603440714834,
      "learning_rate": 1.776193082606418e-05,
      "loss": 0.6201,
      "step": 1598
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7257842243829465,
      "learning_rate": 1.7758853233297422e-05,
      "loss": 0.6062,
      "step": 1599
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.747489262209451,
      "learning_rate": 1.7755773792984374e-05,
      "loss": 0.6007,
      "step": 1600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6529884265460801,
      "learning_rate": 1.775269250585831e-05,
      "loss": 0.595,
      "step": 1601
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6879702602266908,
      "learning_rate": 1.7749609372652953e-05,
      "loss": 0.6075,
      "step": 1602
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7197992145165341,
      "learning_rate": 1.774652439410246e-05,
      "loss": 0.6136,
      "step": 1603
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6707933400775657,
      "learning_rate": 1.7743437570941435e-05,
      "loss": 0.6068,
      "step": 1604
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.664618719918338,
      "learning_rate": 1.7740348903904913e-05,
      "loss": 0.5984,
      "step": 1605
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6949363751937103,
      "learning_rate": 1.7737258393728363e-05,
      "loss": 0.5797,
      "step": 1606
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6776047689837895,
      "learning_rate": 1.7734166041147713e-05,
      "loss": 0.6002,
      "step": 1607
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7161557112225574,
      "learning_rate": 1.7731071846899307e-05,
      "loss": 0.5953,
      "step": 1608
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7438764439343474,
      "learning_rate": 1.7727975811719942e-05,
      "loss": 0.613,
      "step": 1609
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7148853192488868,
      "learning_rate": 1.772487793634685e-05,
      "loss": 0.6377,
      "step": 1610
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7481519758610665,
      "learning_rate": 1.77217782215177e-05,
      "loss": 0.5954,
      "step": 1611
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6976156327518414,
      "learning_rate": 1.7718676667970597e-05,
      "loss": 0.6168,
      "step": 1612
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7492218086172716,
      "learning_rate": 1.7715573276444086e-05,
      "loss": 0.6438,
      "step": 1613
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7331758943507667,
      "learning_rate": 1.771246804767716e-05,
      "loss": 0.6134,
      "step": 1614
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6041210066081472,
      "learning_rate": 1.770936098240922e-05,
      "loss": 0.6018,
      "step": 1615
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6922065831755951,
      "learning_rate": 1.7706252081380143e-05,
      "loss": 0.6339,
      "step": 1616
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6607832539016674,
      "learning_rate": 1.7703141345330212e-05,
      "loss": 0.5934,
      "step": 1617
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6036990387174894,
      "learning_rate": 1.770002877500016e-05,
      "loss": 0.5873,
      "step": 1618
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7222326267951289,
      "learning_rate": 1.7696914371131164e-05,
      "loss": 0.6137,
      "step": 1619
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6023102059132677,
      "learning_rate": 1.7693798134464818e-05,
      "loss": 0.5664,
      "step": 1620
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.667361465389337,
      "learning_rate": 1.769068006574317e-05,
      "loss": 0.5674,
      "step": 1621
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8847133246290058,
      "learning_rate": 1.7687560165708695e-05,
      "loss": 0.599,
      "step": 1622
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6014526596705811,
      "learning_rate": 1.7684438435104308e-05,
      "loss": 0.5956,
      "step": 1623
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5962345315036977,
      "learning_rate": 1.768131487467336e-05,
      "loss": 0.5812,
      "step": 1624
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5984991660471262,
      "learning_rate": 1.7678189485159635e-05,
      "loss": 0.6021,
      "step": 1625
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7451576360590897,
      "learning_rate": 1.7675062267307358e-05,
      "loss": 0.6133,
      "step": 1626
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7243407771762991,
      "learning_rate": 1.7671933221861178e-05,
      "loss": 0.5988,
      "step": 1627
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.647975825655551,
      "learning_rate": 1.766880234956619e-05,
      "loss": 0.5921,
      "step": 1628
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6489397838686444,
      "learning_rate": 1.766566965116792e-05,
      "loss": 0.5948,
      "step": 1629
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6558263925580294,
      "learning_rate": 1.7662535127412334e-05,
      "loss": 0.5876,
      "step": 1630
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5934360320547403,
      "learning_rate": 1.7659398779045824e-05,
      "loss": 0.5882,
      "step": 1631
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6498515113686311,
      "learning_rate": 1.765626060681522e-05,
      "loss": 0.6267,
      "step": 1632
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7027376725753904,
      "learning_rate": 1.7653120611467786e-05,
      "loss": 0.6021,
      "step": 1633
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.612869021978046,
      "learning_rate": 1.7649978793751226e-05,
      "loss": 0.572,
      "step": 1634
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6815901203048336,
      "learning_rate": 1.764683515441367e-05,
      "loss": 0.6099,
      "step": 1635
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6788237262237463,
      "learning_rate": 1.7643689694203682e-05,
      "loss": 0.5957,
      "step": 1636
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6649451140834398,
      "learning_rate": 1.764054241387027e-05,
      "loss": 0.6052,
      "step": 1637
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8434404124594787,
      "learning_rate": 1.7637393314162856e-05,
      "loss": 0.6398,
      "step": 1638
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6778694038503871,
      "learning_rate": 1.7634242395831316e-05,
      "loss": 0.5736,
      "step": 1639
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7243653286793399,
      "learning_rate": 1.763108965962595e-05,
      "loss": 0.6094,
      "step": 1640
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6679443583021469,
      "learning_rate": 1.762793510629749e-05,
      "loss": 0.6149,
      "step": 1641
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6916869272574876,
      "learning_rate": 1.7624778736597094e-05,
      "loss": 0.6103,
      "step": 1642
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6775321536026792,
      "learning_rate": 1.7621620551276366e-05,
      "loss": 0.6043,
      "step": 1643
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6320751165183953,
      "learning_rate": 1.7618460551087338e-05,
      "loss": 0.5994,
      "step": 1644
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6674215779965829,
      "learning_rate": 1.761529873678247e-05,
      "loss": 0.6205,
      "step": 1645
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6963069090020945,
      "learning_rate": 1.761213510911466e-05,
      "loss": 0.619,
      "step": 1646
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5941094200031333,
      "learning_rate": 1.760896966883723e-05,
      "loss": 0.6,
      "step": 1647
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6337300845114947,
      "learning_rate": 1.7605802416703946e-05,
      "loss": 0.5793,
      "step": 1648
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6724689472723541,
      "learning_rate": 1.7602633353468987e-05,
      "loss": 0.5854,
      "step": 1649
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6155061730181258,
      "learning_rate": 1.7599462479886976e-05,
      "loss": 0.6098,
      "step": 1650
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6282678854181399,
      "learning_rate": 1.759628979671297e-05,
      "loss": 0.6242,
      "step": 1651
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6515491299552011,
      "learning_rate": 1.7593115304702448e-05,
      "loss": 0.6242,
      "step": 1652
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6420555626921008,
      "learning_rate": 1.7589939004611318e-05,
      "loss": 0.592,
      "step": 1653
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6280113503768013,
      "learning_rate": 1.7586760897195935e-05,
      "loss": 0.6169,
      "step": 1654
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6123947216547505,
      "learning_rate": 1.7583580983213068e-05,
      "loss": 0.5896,
      "step": 1655
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6478435611793082,
      "learning_rate": 1.758039926341992e-05,
      "loss": 0.5976,
      "step": 1656
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.572411801318021,
      "learning_rate": 1.757721573857413e-05,
      "loss": 0.5917,
      "step": 1657
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6405970222060057,
      "learning_rate": 1.7574030409433755e-05,
      "loss": 0.6223,
      "step": 1658
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7384678787500129,
      "learning_rate": 1.757084327675729e-05,
      "loss": 0.6234,
      "step": 1659
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7161211479401558,
      "learning_rate": 1.7567654341303668e-05,
      "loss": 0.6129,
      "step": 1660
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6025757335768955,
      "learning_rate": 1.7564463603832234e-05,
      "loss": 0.6116,
      "step": 1661
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7108565911786756,
      "learning_rate": 1.756127106510277e-05,
      "loss": 0.6211,
      "step": 1662
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7829599689519348,
      "learning_rate": 1.7558076725875487e-05,
      "loss": 0.6418,
      "step": 1663
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.606923108445432,
      "learning_rate": 1.7554880586911025e-05,
      "loss": 0.5954,
      "step": 1664
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6523428216180104,
      "learning_rate": 1.755168264897045e-05,
      "loss": 0.6148,
      "step": 1665
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.713520885911629,
      "learning_rate": 1.7548482912815265e-05,
      "loss": 0.6001,
      "step": 1666
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6137128074974048,
      "learning_rate": 1.7545281379207385e-05,
      "loss": 0.602,
      "step": 1667
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6950771508154617,
      "learning_rate": 1.7542078048909172e-05,
      "loss": 0.6093,
      "step": 1668
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6915701898740584,
      "learning_rate": 1.75388729226834e-05,
      "loss": 0.6409,
      "step": 1669
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7226510047353882,
      "learning_rate": 1.753566600129328e-05,
      "loss": 0.5941,
      "step": 1670
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7140665071404,
      "learning_rate": 1.7532457285502444e-05,
      "loss": 0.6107,
      "step": 1671
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5809211707908102,
      "learning_rate": 1.7529246776074958e-05,
      "loss": 0.5963,
      "step": 1672
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6152798657726072,
      "learning_rate": 1.752603447377531e-05,
      "loss": 0.6128,
      "step": 1673
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6942511738784232,
      "learning_rate": 1.7522820379368417e-05,
      "loss": 0.6177,
      "step": 1674
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7911139203572608,
      "learning_rate": 1.7519604493619622e-05,
      "loss": 0.6297,
      "step": 1675
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7105374242830454,
      "learning_rate": 1.75163868172947e-05,
      "loss": 0.6284,
      "step": 1676
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6445325140989326,
      "learning_rate": 1.7513167351159846e-05,
      "loss": 0.612,
      "step": 1677
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.755364966081996,
      "learning_rate": 1.7509946095981676e-05,
      "loss": 0.5909,
      "step": 1678
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6835756162327685,
      "learning_rate": 1.7506723052527243e-05,
      "loss": 0.609,
      "step": 1679
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.70417090465472,
      "learning_rate": 1.7503498221564026e-05,
      "loss": 0.5981,
      "step": 1680
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6412859792912269,
      "learning_rate": 1.750027160385992e-05,
      "loss": 0.5845,
      "step": 1681
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6950228409334109,
      "learning_rate": 1.7497043200183247e-05,
      "loss": 0.6056,
      "step": 1682
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7139562376333646,
      "learning_rate": 1.7493813011302764e-05,
      "loss": 0.5807,
      "step": 1683
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6320568595134136,
      "learning_rate": 1.7490581037987645e-05,
      "loss": 0.6041,
      "step": 1684
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6828168238263848,
      "learning_rate": 1.7487347281007493e-05,
      "loss": 0.6008,
      "step": 1685
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7727726868510874,
      "learning_rate": 1.7484111741132332e-05,
      "loss": 0.5827,
      "step": 1686
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7200629519458676,
      "learning_rate": 1.7480874419132613e-05,
      "loss": 0.5979,
      "step": 1687
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7242200165129086,
      "learning_rate": 1.7477635315779205e-05,
      "loss": 0.6225,
      "step": 1688
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7184515466081791,
      "learning_rate": 1.7474394431843414e-05,
      "loss": 0.5898,
      "step": 1689
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.729448266622724,
      "learning_rate": 1.747115176809696e-05,
      "loss": 0.6022,
      "step": 1690
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7750066313990859,
      "learning_rate": 1.7467907325311987e-05,
      "loss": 0.6363,
      "step": 1691
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7350120113673692,
      "learning_rate": 1.746466110426107e-05,
      "loss": 0.6082,
      "step": 1692
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5993017902034864,
      "learning_rate": 1.74614131057172e-05,
      "loss": 0.5767,
      "step": 1693
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7101017060220045,
      "learning_rate": 1.7458163330453794e-05,
      "loss": 0.5979,
      "step": 1694
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.660241993808396,
      "learning_rate": 1.745491177924469e-05,
      "loss": 0.5888,
      "step": 1695
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6896325656765486,
      "learning_rate": 1.7451658452864152e-05,
      "loss": 0.6284,
      "step": 1696
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7065616982270682,
      "learning_rate": 1.7448403352086868e-05,
      "loss": 0.6144,
      "step": 1697
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6474599589615406,
      "learning_rate": 1.7445146477687943e-05,
      "loss": 0.6021,
      "step": 1698
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8179478136952858,
      "learning_rate": 1.7441887830442913e-05,
      "loss": 0.6143,
      "step": 1699
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6534971855028678,
      "learning_rate": 1.743862741112772e-05,
      "loss": 0.6126,
      "step": 1700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6926649502957952,
      "learning_rate": 1.7435365220518746e-05,
      "loss": 0.6059,
      "step": 1701
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7342930302885706,
      "learning_rate": 1.7432101259392786e-05,
      "loss": 0.6002,
      "step": 1702
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6449349536180872,
      "learning_rate": 1.742883552852706e-05,
      "loss": 0.6023,
      "step": 1703
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5757588606311878,
      "learning_rate": 1.74255680286992e-05,
      "loss": 0.5926,
      "step": 1704
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7165119138592311,
      "learning_rate": 1.7422298760687275e-05,
      "loss": 0.5845,
      "step": 1705
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6191650422033094,
      "learning_rate": 1.7419027725269763e-05,
      "loss": 0.5937,
      "step": 1706
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6992002730786461,
      "learning_rate": 1.7415754923225564e-05,
      "loss": 0.6177,
      "step": 1707
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7466067019934234,
      "learning_rate": 1.7412480355334006e-05,
      "loss": 0.6209,
      "step": 1708
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7610173985387936,
      "learning_rate": 1.7409204022374826e-05,
      "loss": 0.6046,
      "step": 1709
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6938988180173797,
      "learning_rate": 1.7405925925128193e-05,
      "loss": 0.5965,
      "step": 1710
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6903681905400755,
      "learning_rate": 1.740264606437469e-05,
      "loss": 0.63,
      "step": 1711
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7322616051963219,
      "learning_rate": 1.7399364440895323e-05,
      "loss": 0.636,
      "step": 1712
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6976605483376723,
      "learning_rate": 1.739608105547151e-05,
      "loss": 0.5636,
      "step": 1713
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6990863154051565,
      "learning_rate": 1.73927959088851e-05,
      "loss": 0.5938,
      "step": 1714
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.731199025473645,
      "learning_rate": 1.738950900191835e-05,
      "loss": 0.6171,
      "step": 1715
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6535621325738198,
      "learning_rate": 1.7386220335353945e-05,
      "loss": 0.5933,
      "step": 1716
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7868594628120035,
      "learning_rate": 1.7382929909974988e-05,
      "loss": 0.6251,
      "step": 1717
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.665326206461159,
      "learning_rate": 1.7379637726564994e-05,
      "loss": 0.5915,
      "step": 1718
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6800392509474047,
      "learning_rate": 1.7376343785907905e-05,
      "loss": 0.6046,
      "step": 1719
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6714222761261258,
      "learning_rate": 1.7373048088788076e-05,
      "loss": 0.5826,
      "step": 1720
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.745527338217347,
      "learning_rate": 1.7369750635990278e-05,
      "loss": 0.6173,
      "step": 1721
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7263806543223533,
      "learning_rate": 1.7366451428299713e-05,
      "loss": 0.6272,
      "step": 1722
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6274448473505879,
      "learning_rate": 1.736315046650198e-05,
      "loss": 0.6165,
      "step": 1723
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7216992320984983,
      "learning_rate": 1.7359847751383115e-05,
      "loss": 0.6066,
      "step": 1724
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6625786950071283,
      "learning_rate": 1.735654328372957e-05,
      "loss": 0.6043,
      "step": 1725
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6607907688489582,
      "learning_rate": 1.735323706432819e-05,
      "loss": 0.5939,
      "step": 1726
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5992559900825009,
      "learning_rate": 1.7349929093966275e-05,
      "loss": 0.5747,
      "step": 1727
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6347746907400887,
      "learning_rate": 1.7346619373431513e-05,
      "loss": 0.6427,
      "step": 1728
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6380433297358536,
      "learning_rate": 1.734330790351202e-05,
      "loss": 0.6011,
      "step": 1729
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6606739880304799,
      "learning_rate": 1.733999468499632e-05,
      "loss": 0.6092,
      "step": 1730
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6801774104662649,
      "learning_rate": 1.733667971867337e-05,
      "loss": 0.6021,
      "step": 1731
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6919028559229786,
      "learning_rate": 1.733336300533253e-05,
      "loss": 0.607,
      "step": 1732
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6691266234986999,
      "learning_rate": 1.7330044545763574e-05,
      "loss": 0.6009,
      "step": 1733
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.587975016053544,
      "learning_rate": 1.7326724340756706e-05,
      "loss": 0.5917,
      "step": 1734
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7740431578433303,
      "learning_rate": 1.732340239110253e-05,
      "loss": 0.6158,
      "step": 1735
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9223374424642119,
      "learning_rate": 1.7320078697592077e-05,
      "loss": 0.6403,
      "step": 1736
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7313738361683453,
      "learning_rate": 1.7316753261016782e-05,
      "loss": 0.6314,
      "step": 1737
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7028104199145103,
      "learning_rate": 1.731342608216851e-05,
      "loss": 0.5972,
      "step": 1738
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6713895805135152,
      "learning_rate": 1.7310097161839526e-05,
      "loss": 0.5815,
      "step": 1739
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6746073742794296,
      "learning_rate": 1.7306766500822516e-05,
      "loss": 0.5755,
      "step": 1740
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6724055770350796,
      "learning_rate": 1.730343409991058e-05,
      "loss": 0.6295,
      "step": 1741
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7237136931403726,
      "learning_rate": 1.730009995989724e-05,
      "loss": 0.618,
      "step": 1742
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6172470178447877,
      "learning_rate": 1.7296764081576417e-05,
      "loss": 0.5932,
      "step": 1743
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6952749721581812,
      "learning_rate": 1.729342646574246e-05,
      "loss": 0.5974,
      "step": 1744
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6814126767007608,
      "learning_rate": 1.7290087113190117e-05,
      "loss": 0.5915,
      "step": 1745
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6662788334173622,
      "learning_rate": 1.7286746024714566e-05,
      "loss": 0.6224,
      "step": 1746
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9369380898851956,
      "learning_rate": 1.7283403201111384e-05,
      "loss": 0.6641,
      "step": 1747
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6045327449042293,
      "learning_rate": 1.728005864317658e-05,
      "loss": 0.6059,
      "step": 1748
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7821487682165729,
      "learning_rate": 1.7276712351706548e-05,
      "loss": 0.6281,
      "step": 1749
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6771028985673575,
      "learning_rate": 1.7273364327498118e-05,
      "loss": 0.604,
      "step": 1750
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7132313792056149,
      "learning_rate": 1.7270014571348528e-05,
      "loss": 0.6037,
      "step": 1751
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6937241888718266,
      "learning_rate": 1.726666308405542e-05,
      "loss": 0.5893,
      "step": 1752
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6961635463525235,
      "learning_rate": 1.726330986641686e-05,
      "loss": 0.6019,
      "step": 1753
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.665654139428949,
      "learning_rate": 1.725995491923131e-05,
      "loss": 0.6268,
      "step": 1754
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6481929340279893,
      "learning_rate": 1.7256598243297662e-05,
      "loss": 0.5886,
      "step": 1755
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.695832908188197,
      "learning_rate": 1.7253239839415207e-05,
      "loss": 0.5991,
      "step": 1756
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6783422448889033,
      "learning_rate": 1.7249879708383654e-05,
      "loss": 0.6055,
      "step": 1757
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6540299080595523,
      "learning_rate": 1.724651785100312e-05,
      "loss": 0.5703,
      "step": 1758
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5949007278191176,
      "learning_rate": 1.7243154268074132e-05,
      "loss": 0.5924,
      "step": 1759
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6298036798703515,
      "learning_rate": 1.7239788960397636e-05,
      "loss": 0.593,
      "step": 1760
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6402107606071932,
      "learning_rate": 1.7236421928774975e-05,
      "loss": 0.6015,
      "step": 1761
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.661295174472384,
      "learning_rate": 1.7233053174007914e-05,
      "loss": 0.6074,
      "step": 1762
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6972206925611064,
      "learning_rate": 1.7229682696898623e-05,
      "loss": 0.6181,
      "step": 1763
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.681271117476565,
      "learning_rate": 1.7226310498249688e-05,
      "loss": 0.5894,
      "step": 1764
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.64979086430654,
      "learning_rate": 1.7222936578864094e-05,
      "loss": 0.6069,
      "step": 1765
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7403303368863553,
      "learning_rate": 1.7219560939545246e-05,
      "loss": 0.6031,
      "step": 1766
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5991932824308888,
      "learning_rate": 1.7216183581096955e-05,
      "loss": 0.6136,
      "step": 1767
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7174697333467484,
      "learning_rate": 1.7212804504323437e-05,
      "loss": 0.6077,
      "step": 1768
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6747544725706832,
      "learning_rate": 1.720942371002933e-05,
      "loss": 0.5828,
      "step": 1769
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6514027994904169,
      "learning_rate": 1.7206041199019664e-05,
      "loss": 0.5977,
      "step": 1770
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6799686232146536,
      "learning_rate": 1.7202656972099886e-05,
      "loss": 0.6371,
      "step": 1771
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5876902404249339,
      "learning_rate": 1.719927103007586e-05,
      "loss": 0.5931,
      "step": 1772
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6402821941269039,
      "learning_rate": 1.719588337375384e-05,
      "loss": 0.5658,
      "step": 1773
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7477045757723367,
      "learning_rate": 1.7192494003940504e-05,
      "loss": 0.5585,
      "step": 1774
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.621395917755218,
      "learning_rate": 1.718910292144293e-05,
      "loss": 0.6304,
      "step": 1775
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5997991653596108,
      "learning_rate": 1.7185710127068614e-05,
      "loss": 0.5895,
      "step": 1776
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6326619075127855,
      "learning_rate": 1.718231562162544e-05,
      "loss": 0.6408,
      "step": 1777
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6852211115569367,
      "learning_rate": 1.7178919405921717e-05,
      "loss": 0.6034,
      "step": 1778
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5931172914058555,
      "learning_rate": 1.7175521480766152e-05,
      "loss": 0.5845,
      "step": 1779
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7356814846574025,
      "learning_rate": 1.717212184696787e-05,
      "loss": 0.5958,
      "step": 1780
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6420850620192456,
      "learning_rate": 1.7168720505336385e-05,
      "loss": 0.5983,
      "step": 1781
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7022376770402581,
      "learning_rate": 1.7165317456681642e-05,
      "loss": 0.6181,
      "step": 1782
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.631221619601243,
      "learning_rate": 1.716191270181396e-05,
      "loss": 0.6141,
      "step": 1783
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5484564341913422,
      "learning_rate": 1.71585062415441e-05,
      "loss": 0.5801,
      "step": 1784
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6849922598860502,
      "learning_rate": 1.7155098076683203e-05,
      "loss": 0.5945,
      "step": 1785
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6057362755970046,
      "learning_rate": 1.7151688208042826e-05,
      "loss": 0.6024,
      "step": 1786
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6756778937699621,
      "learning_rate": 1.7148276636434933e-05,
      "loss": 0.5993,
      "step": 1787
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7227174495450678,
      "learning_rate": 1.714486336267189e-05,
      "loss": 0.6045,
      "step": 1788
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6549747750257309,
      "learning_rate": 1.7141448387566467e-05,
      "loss": 0.5972,
      "step": 1789
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6558049119727064,
      "learning_rate": 1.7138031711931842e-05,
      "loss": 0.5833,
      "step": 1790
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.684627726838437,
      "learning_rate": 1.7134613336581602e-05,
      "loss": 0.5967,
      "step": 1791
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6762886241569018,
      "learning_rate": 1.7131193262329726e-05,
      "loss": 0.5877,
      "step": 1792
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7822531402614845,
      "learning_rate": 1.7127771489990616e-05,
      "loss": 0.6157,
      "step": 1793
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7096293106591207,
      "learning_rate": 1.7124348020379056e-05,
      "loss": 0.6137,
      "step": 1794
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6353663419278949,
      "learning_rate": 1.712092285431026e-05,
      "loss": 0.6046,
      "step": 1795
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6895160151034134,
      "learning_rate": 1.7117495992599814e-05,
      "loss": 0.6047,
      "step": 1796
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7163017091206453,
      "learning_rate": 1.7114067436063745e-05,
      "loss": 0.6065,
      "step": 1797
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6174227638423814,
      "learning_rate": 1.7110637185518453e-05,
      "loss": 0.6385,
      "step": 1798
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7592062533474795,
      "learning_rate": 1.710720524178076e-05,
      "loss": 0.6254,
      "step": 1799
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6318528051692076,
      "learning_rate": 1.7103771605667874e-05,
      "loss": 0.6028,
      "step": 1800
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.705909822844947,
      "learning_rate": 1.7100336277997424e-05,
      "loss": 0.6194,
      "step": 1801
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6708979260649942,
      "learning_rate": 1.7096899259587432e-05,
      "loss": 0.5871,
      "step": 1802
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6088290846124488,
      "learning_rate": 1.7093460551256325e-05,
      "loss": 0.5999,
      "step": 1803
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6866695847996201,
      "learning_rate": 1.7090020153822934e-05,
      "loss": 0.5821,
      "step": 1804
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6554604482727218,
      "learning_rate": 1.7086578068106484e-05,
      "loss": 0.5907,
      "step": 1805
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7206409467173562,
      "learning_rate": 1.708313429492661e-05,
      "loss": 0.6149,
      "step": 1806
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6334171267518309,
      "learning_rate": 1.707968883510335e-05,
      "loss": 0.6256,
      "step": 1807
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6093683449125109,
      "learning_rate": 1.7076241689457134e-05,
      "loss": 0.5931,
      "step": 1808
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6334597979746514,
      "learning_rate": 1.707279285880881e-05,
      "loss": 0.5969,
      "step": 1809
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7641263634081724,
      "learning_rate": 1.706934234397961e-05,
      "loss": 0.6135,
      "step": 1810
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6234236657799048,
      "learning_rate": 1.7065890145791177e-05,
      "loss": 0.6083,
      "step": 1811
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6553989802945993,
      "learning_rate": 1.706243626506555e-05,
      "loss": 0.606,
      "step": 1812
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6398773303923584,
      "learning_rate": 1.7058980702625172e-05,
      "loss": 0.5711,
      "step": 1813
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7274720812451468,
      "learning_rate": 1.7055523459292888e-05,
      "loss": 0.5917,
      "step": 1814
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6683831546589611,
      "learning_rate": 1.7052064535891935e-05,
      "loss": 0.6042,
      "step": 1815
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6822397251007756,
      "learning_rate": 1.7048603933245956e-05,
      "loss": 0.6094,
      "step": 1816
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6661056039436193,
      "learning_rate": 1.7045141652178997e-05,
      "loss": 0.6065,
      "step": 1817
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7231419219727517,
      "learning_rate": 1.7041677693515497e-05,
      "loss": 0.5983,
      "step": 1818
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5882226496720692,
      "learning_rate": 1.70382120580803e-05,
      "loss": 0.5683,
      "step": 1819
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6662495639504237,
      "learning_rate": 1.7034744746698644e-05,
      "loss": 0.6285,
      "step": 1820
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6992334119948705,
      "learning_rate": 1.7031275760196172e-05,
      "loss": 0.6277,
      "step": 1821
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8346168913013347,
      "learning_rate": 1.702780509939892e-05,
      "loss": 0.6468,
      "step": 1822
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6493512237602754,
      "learning_rate": 1.7024332765133325e-05,
      "loss": 0.6103,
      "step": 1823
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6667480815071375,
      "learning_rate": 1.702085875822623e-05,
      "loss": 0.5843,
      "step": 1824
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6392832306340201,
      "learning_rate": 1.7017383079504858e-05,
      "loss": 0.6111,
      "step": 1825
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.647147193197308,
      "learning_rate": 1.7013905729796845e-05,
      "loss": 0.5917,
      "step": 1826
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6787752257269924,
      "learning_rate": 1.701042670993023e-05,
      "loss": 0.6127,
      "step": 1827
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6019997511353872,
      "learning_rate": 1.7006946020733426e-05,
      "loss": 0.6003,
      "step": 1828
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6751467816381357,
      "learning_rate": 1.700346366303527e-05,
      "loss": 0.5885,
      "step": 1829
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6442827642687858,
      "learning_rate": 1.6999979637664982e-05,
      "loss": 0.5869,
      "step": 1830
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7395673208952322,
      "learning_rate": 1.699649394545218e-05,
      "loss": 0.5962,
      "step": 1831
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7199398186444771,
      "learning_rate": 1.6993006587226876e-05,
      "loss": 0.6246,
      "step": 1832
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6895325033663152,
      "learning_rate": 1.6989517563819496e-05,
      "loss": 0.5971,
      "step": 1833
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6465999786045518,
      "learning_rate": 1.698602687606084e-05,
      "loss": 0.5846,
      "step": 1834
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.677643602209896,
      "learning_rate": 1.6982534524782117e-05,
      "loss": 0.6018,
      "step": 1835
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8053108302510901,
      "learning_rate": 1.697904051081493e-05,
      "loss": 0.6187,
      "step": 1836
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7932622892590487,
      "learning_rate": 1.6975544834991273e-05,
      "loss": 0.6077,
      "step": 1837
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7143843421229633,
      "learning_rate": 1.6972047498143544e-05,
      "loss": 0.602,
      "step": 1838
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6739516028896694,
      "learning_rate": 1.6968548501104532e-05,
      "loss": 0.6175,
      "step": 1839
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6092003809076021,
      "learning_rate": 1.6965047844707428e-05,
      "loss": 0.5733,
      "step": 1840
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7729755861234835,
      "learning_rate": 1.69615455297858e-05,
      "loss": 0.6165,
      "step": 1841
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6802494459721224,
      "learning_rate": 1.695804155717363e-05,
      "loss": 0.5869,
      "step": 1842
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6137278147372114,
      "learning_rate": 1.6954535927705288e-05,
      "loss": 0.5944,
      "step": 1843
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7464968801599263,
      "learning_rate": 1.6951028642215533e-05,
      "loss": 0.6143,
      "step": 1844
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7987870549259584,
      "learning_rate": 1.694751970153953e-05,
      "loss": 0.5852,
      "step": 1845
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6178723801389933,
      "learning_rate": 1.6944009106512828e-05,
      "loss": 0.583,
      "step": 1846
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6374871087993462,
      "learning_rate": 1.6940496857971375e-05,
      "loss": 0.6005,
      "step": 1847
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7519120900354006,
      "learning_rate": 1.693698295675151e-05,
      "loss": 0.6057,
      "step": 1848
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7312546043107612,
      "learning_rate": 1.693346740368997e-05,
      "loss": 0.5996,
      "step": 1849
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8487574772274836,
      "learning_rate": 1.6929950199623875e-05,
      "loss": 0.5978,
      "step": 1850
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7738503413464253,
      "learning_rate": 1.692643134539075e-05,
      "loss": 0.5946,
      "step": 1851
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7566051704686855,
      "learning_rate": 1.6922910841828513e-05,
      "loss": 0.6094,
      "step": 1852
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7417919750326488,
      "learning_rate": 1.6919388689775463e-05,
      "loss": 0.6464,
      "step": 1853
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7669636399932046,
      "learning_rate": 1.69158648900703e-05,
      "loss": 0.6075,
      "step": 1854
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7012379310671543,
      "learning_rate": 1.691233944355212e-05,
      "loss": 0.6117,
      "step": 1855
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7960343868525623,
      "learning_rate": 1.6908812351060397e-05,
      "loss": 0.606,
      "step": 1856
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7620796501983392,
      "learning_rate": 1.6905283613435012e-05,
      "loss": 0.6014,
      "step": 1857
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6790991129670275,
      "learning_rate": 1.690175323151623e-05,
      "loss": 0.5951,
      "step": 1858
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6291369025188999,
      "learning_rate": 1.689822120614471e-05,
      "loss": 0.5924,
      "step": 1859
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6679778942970458,
      "learning_rate": 1.6894687538161503e-05,
      "loss": 0.5812,
      "step": 1860
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.700398122265365,
      "learning_rate": 1.6891152228408046e-05,
      "loss": 0.5646,
      "step": 1861
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7972817583073993,
      "learning_rate": 1.6887615277726174e-05,
      "loss": 0.628,
      "step": 1862
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6438658340018977,
      "learning_rate": 1.6884076686958107e-05,
      "loss": 0.595,
      "step": 1863
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6720102342837712,
      "learning_rate": 1.688053645694646e-05,
      "loss": 0.6104,
      "step": 1864
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6270918474165273,
      "learning_rate": 1.6876994588534234e-05,
      "loss": 0.5803,
      "step": 1865
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6102503756248223,
      "learning_rate": 1.6873451082564828e-05,
      "loss": 0.5921,
      "step": 1866
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6299518158291247,
      "learning_rate": 1.6869905939882015e-05,
      "loss": 0.5899,
      "step": 1867
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6445721940310828,
      "learning_rate": 1.686635916132998e-05,
      "loss": 0.5964,
      "step": 1868
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6773589544346452,
      "learning_rate": 1.6862810747753274e-05,
      "loss": 0.5962,
      "step": 1869
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6672250883414577,
      "learning_rate": 1.685926069999686e-05,
      "loss": 0.6116,
      "step": 1870
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.607475420405834,
      "learning_rate": 1.6855709018906073e-05,
      "loss": 0.5789,
      "step": 1871
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5402679452837859,
      "learning_rate": 1.6852155705326644e-05,
      "loss": 0.5851,
      "step": 1872
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7079696388620926,
      "learning_rate": 1.684860076010469e-05,
      "loss": 0.6174,
      "step": 1873
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7568861394428937,
      "learning_rate": 1.684504418408672e-05,
      "loss": 0.6249,
      "step": 1874
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5909050072530909,
      "learning_rate": 1.684148597811963e-05,
      "loss": 0.6002,
      "step": 1875
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.696625796877673,
      "learning_rate": 1.6837926143050707e-05,
      "loss": 0.6193,
      "step": 1876
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6235422287709886,
      "learning_rate": 1.6834364679727614e-05,
      "loss": 0.614,
      "step": 1877
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5762653815626427,
      "learning_rate": 1.683080158899842e-05,
      "loss": 0.5832,
      "step": 1878
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7511988995685137,
      "learning_rate": 1.6827236871711566e-05,
      "loss": 0.593,
      "step": 1879
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6877874175529135,
      "learning_rate": 1.6823670528715886e-05,
      "loss": 0.6,
      "step": 1880
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6685679699587276,
      "learning_rate": 1.6820102560860607e-05,
      "loss": 0.5963,
      "step": 1881
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7238721923727348,
      "learning_rate": 1.681653296899533e-05,
      "loss": 0.6041,
      "step": 1882
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6747302243573299,
      "learning_rate": 1.6812961753970054e-05,
      "loss": 0.6072,
      "step": 1883
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5892177044287131,
      "learning_rate": 1.6809388916635158e-05,
      "loss": 0.5833,
      "step": 1884
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6813052287990913,
      "learning_rate": 1.6805814457841416e-05,
      "loss": 0.6157,
      "step": 1885
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7275432565761268,
      "learning_rate": 1.6802238378439976e-05,
      "loss": 0.6023,
      "step": 1886
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.627113078135882,
      "learning_rate": 1.6798660679282374e-05,
      "loss": 0.5933,
      "step": 1887
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.674258718477368,
      "learning_rate": 1.6795081361220547e-05,
      "loss": 0.6121,
      "step": 1888
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9223221838541948,
      "learning_rate": 1.6791500425106795e-05,
      "loss": 0.6765,
      "step": 1889
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3694925943138283,
      "learning_rate": 1.6787917871793823e-05,
      "loss": 0.6328,
      "step": 1890
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7735305766398504,
      "learning_rate": 1.678433370213471e-05,
      "loss": 0.5901,
      "step": 1891
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6517974345477973,
      "learning_rate": 1.6780747916982912e-05,
      "loss": 0.6159,
      "step": 1892
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6719830967501966,
      "learning_rate": 1.6777160517192297e-05,
      "loss": 0.6006,
      "step": 1893
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5741649327997186,
      "learning_rate": 1.677357150361709e-05,
      "loss": 0.5839,
      "step": 1894
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7508374285818511,
      "learning_rate": 1.6769980877111906e-05,
      "loss": 0.6234,
      "step": 1895
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6783296054764437,
      "learning_rate": 1.6766388638531762e-05,
      "loss": 0.592,
      "step": 1896
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6745987730098442,
      "learning_rate": 1.6762794788732037e-05,
      "loss": 0.5874,
      "step": 1897
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6510322190686046,
      "learning_rate": 1.6759199328568506e-05,
      "loss": 0.5827,
      "step": 1898
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6603933653745835,
      "learning_rate": 1.6755602258897323e-05,
      "loss": 0.6083,
      "step": 1899
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6171914629035893,
      "learning_rate": 1.675200358057502e-05,
      "loss": 0.5949,
      "step": 1900
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.761137184971375,
      "learning_rate": 1.6748403294458528e-05,
      "loss": 0.6287,
      "step": 1901
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6531828410842538,
      "learning_rate": 1.6744801401405138e-05,
      "loss": 0.5891,
      "step": 1902
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6143225713116828,
      "learning_rate": 1.6741197902272553e-05,
      "loss": 0.5804,
      "step": 1903
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7059888787742089,
      "learning_rate": 1.673759279791883e-05,
      "loss": 0.6138,
      "step": 1904
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6852686931349037,
      "learning_rate": 1.6733986089202427e-05,
      "loss": 0.5883,
      "step": 1905
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6088790046053734,
      "learning_rate": 1.6730377776982173e-05,
      "loss": 0.6164,
      "step": 1906
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6846002165458434,
      "learning_rate": 1.6726767862117283e-05,
      "loss": 0.5847,
      "step": 1907
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6434962702772383,
      "learning_rate": 1.6723156345467354e-05,
      "loss": 0.605,
      "step": 1908
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6717778181949585,
      "learning_rate": 1.6719543227892367e-05,
      "loss": 0.6081,
      "step": 1909
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5893660986993524,
      "learning_rate": 1.6715928510252678e-05,
      "loss": 0.6117,
      "step": 1910
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6871490353641498,
      "learning_rate": 1.6712312193409032e-05,
      "loss": 0.5753,
      "step": 1911
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.678975349367144,
      "learning_rate": 1.6708694278222542e-05,
      "loss": 0.5836,
      "step": 1912
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6405909557682835,
      "learning_rate": 1.670507476555472e-05,
      "loss": 0.611,
      "step": 1913
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6065616502904173,
      "learning_rate": 1.6701453656267437e-05,
      "loss": 0.5845,
      "step": 1914
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6485947068600428,
      "learning_rate": 1.6697830951222963e-05,
      "loss": 0.5867,
      "step": 1915
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7157318069219143,
      "learning_rate": 1.6694206651283938e-05,
      "loss": 0.6241,
      "step": 1916
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6086748667417556,
      "learning_rate": 1.669058075731339e-05,
      "loss": 0.5935,
      "step": 1917
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6405430677640077,
      "learning_rate": 1.668695327017471e-05,
      "loss": 0.6322,
      "step": 1918
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5843701937546871,
      "learning_rate": 1.6683324190731686e-05,
      "loss": 0.6172,
      "step": 1919
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6812768071114182,
      "learning_rate": 1.667969351984848e-05,
      "loss": 0.5933,
      "step": 1920
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7094404973370815,
      "learning_rate": 1.667606125838962e-05,
      "loss": 0.6168,
      "step": 1921
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6351753696899005,
      "learning_rate": 1.6672427407220038e-05,
      "loss": 0.5972,
      "step": 1922
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6129053365344901,
      "learning_rate": 1.6668791967205026e-05,
      "loss": 0.5932,
      "step": 1923
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6780679537002031,
      "learning_rate": 1.666515493921025e-05,
      "loss": 0.5839,
      "step": 1924
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6983469757065446,
      "learning_rate": 1.6661516324101775e-05,
      "loss": 0.6027,
      "step": 1925
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6099839996180596,
      "learning_rate": 1.6657876122746027e-05,
      "loss": 0.6281,
      "step": 1926
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6454132636833424,
      "learning_rate": 1.6654234336009812e-05,
      "loss": 0.6053,
      "step": 1927
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6381485129261868,
      "learning_rate": 1.665059096476032e-05,
      "loss": 0.5656,
      "step": 1928
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7446970542131645,
      "learning_rate": 1.664694600986511e-05,
      "loss": 0.5984,
      "step": 1929
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6638590775881609,
      "learning_rate": 1.664329947219213e-05,
      "loss": 0.5754,
      "step": 1930
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6244366224253219,
      "learning_rate": 1.663965135260969e-05,
      "loss": 0.5956,
      "step": 1931
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6931085808676314,
      "learning_rate": 1.6636001651986485e-05,
      "loss": 0.595,
      "step": 1932
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6159731671214101,
      "learning_rate": 1.663235037119159e-05,
      "loss": 0.6099,
      "step": 1933
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.687907203252858,
      "learning_rate": 1.6628697511094446e-05,
      "loss": 0.5913,
      "step": 1934
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.659893970911617,
      "learning_rate": 1.662504307256488e-05,
      "loss": 0.6,
      "step": 1935
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7226084677028546,
      "learning_rate": 1.6621387056473094e-05,
      "loss": 0.5823,
      "step": 1936
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6828895259831614,
      "learning_rate": 1.6617729463689655e-05,
      "loss": 0.6099,
      "step": 1937
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6747085204418715,
      "learning_rate": 1.661407029508552e-05,
      "loss": 0.6129,
      "step": 1938
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7167750799478424,
      "learning_rate": 1.6610409551532006e-05,
      "loss": 0.5864,
      "step": 1939
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6691242375516591,
      "learning_rate": 1.6606747233900816e-05,
      "loss": 0.6042,
      "step": 1940
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6439896163725728,
      "learning_rate": 1.6603083343064028e-05,
      "loss": 0.5999,
      "step": 1941
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6650016506288791,
      "learning_rate": 1.659941787989409e-05,
      "loss": 0.5969,
      "step": 1942
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7513367085005029,
      "learning_rate": 1.6595750845263825e-05,
      "loss": 0.6026,
      "step": 1943
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6260501952278811,
      "learning_rate": 1.659208224004643e-05,
      "loss": 0.5925,
      "step": 1944
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6983721603715678,
      "learning_rate": 1.6588412065115483e-05,
      "loss": 0.5996,
      "step": 1945
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6851105367974296,
      "learning_rate": 1.6584740321344927e-05,
      "loss": 0.5722,
      "step": 1946
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6512843676666041,
      "learning_rate": 1.6581067009609073e-05,
      "loss": 0.6011,
      "step": 1947
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7712314202987408,
      "learning_rate": 1.6577392130782625e-05,
      "loss": 0.6315,
      "step": 1948
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.689450035180266,
      "learning_rate": 1.6573715685740647e-05,
      "loss": 0.6162,
      "step": 1949
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6802073204081116,
      "learning_rate": 1.6570037675358573e-05,
      "loss": 0.6061,
      "step": 1950
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6247778704519112,
      "learning_rate": 1.6566358100512222e-05,
      "loss": 0.6094,
      "step": 1951
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6650855650055889,
      "learning_rate": 1.656267696207777e-05,
      "loss": 0.5973,
      "step": 1952
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6920208610742103,
      "learning_rate": 1.655899426093178e-05,
      "loss": 0.6134,
      "step": 1953
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7060606460399308,
      "learning_rate": 1.6555309997951178e-05,
      "loss": 0.6105,
      "step": 1954
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.648527039733465,
      "learning_rate": 1.6551624174013264e-05,
      "loss": 0.586,
      "step": 1955
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6415030606013791,
      "learning_rate": 1.654793678999571e-05,
      "loss": 0.6189,
      "step": 1956
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6542044150981694,
      "learning_rate": 1.6544247846776562e-05,
      "loss": 0.5993,
      "step": 1957
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6999884217173349,
      "learning_rate": 1.6540557345234235e-05,
      "loss": 0.6159,
      "step": 1958
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6791242392739011,
      "learning_rate": 1.6536865286247512e-05,
      "loss": 0.5945,
      "step": 1959
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.663898342889356,
      "learning_rate": 1.6533171670695555e-05,
      "loss": 0.6033,
      "step": 1960
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6695096725217191,
      "learning_rate": 1.6529476499457886e-05,
      "loss": 0.6044,
      "step": 1961
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.691843374347992,
      "learning_rate": 1.6525779773414406e-05,
      "loss": 0.6559,
      "step": 1962
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6190085520259632,
      "learning_rate": 1.6522081493445386e-05,
      "loss": 0.5623,
      "step": 1963
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6733570457250068,
      "learning_rate": 1.6518381660431456e-05,
      "loss": 0.5869,
      "step": 1964
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7233414745531647,
      "learning_rate": 1.6514680275253634e-05,
      "loss": 0.6082,
      "step": 1965
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6920231987343965,
      "learning_rate": 1.6510977338793293e-05,
      "loss": 0.5972,
      "step": 1966
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6472779799222145,
      "learning_rate": 1.650727285193218e-05,
      "loss": 0.5908,
      "step": 1967
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6324637468025757,
      "learning_rate": 1.6503566815552408e-05,
      "loss": 0.5991,
      "step": 1968
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7190435988192431,
      "learning_rate": 1.6499859230536468e-05,
      "loss": 0.6146,
      "step": 1969
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.664235420190212,
      "learning_rate": 1.6496150097767212e-05,
      "loss": 0.6123,
      "step": 1970
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8145196504033442,
      "learning_rate": 1.6492439418127865e-05,
      "loss": 0.6269,
      "step": 1971
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6633995167941812,
      "learning_rate": 1.648872719250201e-05,
      "loss": 0.6131,
      "step": 1972
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7780920194687235,
      "learning_rate": 1.6485013421773616e-05,
      "loss": 0.622,
      "step": 1973
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7206308816658217,
      "learning_rate": 1.6481298106827005e-05,
      "loss": 0.6048,
      "step": 1974
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7135882554631527,
      "learning_rate": 1.647758124854687e-05,
      "loss": 0.592,
      "step": 1975
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7850136535267687,
      "learning_rate": 1.647386284781828e-05,
      "loss": 0.604,
      "step": 1976
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6831462484339123,
      "learning_rate": 1.6470142905526652e-05,
      "loss": 0.6026,
      "step": 1977
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6844412578092819,
      "learning_rate": 1.6466421422557798e-05,
      "loss": 0.5633,
      "step": 1978
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6675576986343367,
      "learning_rate": 1.6462698399797873e-05,
      "loss": 0.6201,
      "step": 1979
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7009429357341727,
      "learning_rate": 1.6458973838133405e-05,
      "loss": 0.6217,
      "step": 1980
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7642188750135603,
      "learning_rate": 1.6455247738451296e-05,
      "loss": 0.6238,
      "step": 1981
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7342290353383314,
      "learning_rate": 1.6451520101638805e-05,
      "loss": 0.5882,
      "step": 1982
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6364660593937621,
      "learning_rate": 1.644779092858356e-05,
      "loss": 0.5998,
      "step": 1983
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6736111463978225,
      "learning_rate": 1.644406022017356e-05,
      "loss": 0.6011,
      "step": 1984
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471213698215001,
      "learning_rate": 1.6440327977297165e-05,
      "loss": 0.603,
      "step": 1985
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6207743536880022,
      "learning_rate": 1.6436594200843097e-05,
      "loss": 0.592,
      "step": 1986
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7124780993604093,
      "learning_rate": 1.6432858891700445e-05,
      "loss": 0.5924,
      "step": 1987
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6098637911056654,
      "learning_rate": 1.642912205075867e-05,
      "loss": 0.6013,
      "step": 1988
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6161609937629993,
      "learning_rate": 1.6425383678907594e-05,
      "loss": 0.5948,
      "step": 1989
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7585677422209909,
      "learning_rate": 1.6421643777037397e-05,
      "loss": 0.6013,
      "step": 1990
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6589519741023366,
      "learning_rate": 1.641790234603863e-05,
      "loss": 0.6099,
      "step": 1991
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.695316573718371,
      "learning_rate": 1.6414159386802204e-05,
      "loss": 0.5886,
      "step": 1992
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7071139603707186,
      "learning_rate": 1.64104149002194e-05,
      "loss": 0.6021,
      "step": 1993
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5689002487034992,
      "learning_rate": 1.640666888718186e-05,
      "loss": 0.571,
      "step": 1994
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6663677605928552,
      "learning_rate": 1.6402921348581586e-05,
      "loss": 0.6126,
      "step": 1995
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6287785676274485,
      "learning_rate": 1.6399172285310943e-05,
      "loss": 0.5932,
      "step": 1996
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6971086294533984,
      "learning_rate": 1.6395421698262665e-05,
      "loss": 0.6282,
      "step": 1997
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6466367567395099,
      "learning_rate": 1.639166958832985e-05,
      "loss": 0.5967,
      "step": 1998
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.770228102332453,
      "learning_rate": 1.638791595640595e-05,
      "loss": 0.6257,
      "step": 1999
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6634350694944328,
      "learning_rate": 1.6384160803384782e-05,
      "loss": 0.5802,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6599734309304826,
      "learning_rate": 1.6380404130160528e-05,
      "loss": 0.6068,
      "step": 2001
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7228431651966175,
      "learning_rate": 1.6376645937627734e-05,
      "loss": 0.6075,
      "step": 2002
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5993648319023211,
      "learning_rate": 1.6372886226681303e-05,
      "loss": 0.5997,
      "step": 2003
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1134542905020277,
      "learning_rate": 1.63691249982165e-05,
      "loss": 0.6671,
      "step": 2004
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6338648449362693,
      "learning_rate": 1.6365362253128956e-05,
      "loss": 0.5809,
      "step": 2005
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7508594704741691,
      "learning_rate": 1.6361597992314657e-05,
      "loss": 0.5967,
      "step": 2006
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6705905839835794,
      "learning_rate": 1.6357832216669956e-05,
      "loss": 0.5857,
      "step": 2007
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6313990806298996,
      "learning_rate": 1.6354064927091555e-05,
      "loss": 0.6171,
      "step": 2008
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7015446315176228,
      "learning_rate": 1.6350296124476534e-05,
      "loss": 0.5974,
      "step": 2009
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6475975614804205,
      "learning_rate": 1.634652580972232e-05,
      "loss": 0.6134,
      "step": 2010
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6496710305531369,
      "learning_rate": 1.6342753983726704e-05,
      "loss": 0.5797,
      "step": 2011
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6777114112414183,
      "learning_rate": 1.6338980647387843e-05,
      "loss": 0.5804,
      "step": 2012
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6650724054032437,
      "learning_rate": 1.6335205801604242e-05,
      "loss": 0.6049,
      "step": 2013
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.71880411502911,
      "learning_rate": 1.633142944727477e-05,
      "loss": 0.6179,
      "step": 2014
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471857039618738,
      "learning_rate": 1.6327651585298657e-05,
      "loss": 0.609,
      "step": 2015
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.59676365771531,
      "learning_rate": 1.6323872216575498e-05,
      "loss": 0.5874,
      "step": 2016
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7525020621938365,
      "learning_rate": 1.6320091342005238e-05,
      "loss": 0.6213,
      "step": 2017
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7503705713711802,
      "learning_rate": 1.6316308962488173e-05,
      "loss": 0.6128,
      "step": 2018
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7076625030353092,
      "learning_rate": 1.631252507892498e-05,
      "loss": 0.618,
      "step": 2019
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6515831637345803,
      "learning_rate": 1.6308739692216675e-05,
      "loss": 0.5739,
      "step": 2020
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6875274888241037,
      "learning_rate": 1.6304952803264643e-05,
      "loss": 0.5944,
      "step": 2021
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6606335544009875,
      "learning_rate": 1.6301164412970612e-05,
      "loss": 0.5922,
      "step": 2022
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6642609608018171,
      "learning_rate": 1.629737452223669e-05,
      "loss": 0.6263,
      "step": 2023
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7599326215706652,
      "learning_rate": 1.6293583131965317e-05,
      "loss": 0.6541,
      "step": 2024
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7702042445494943,
      "learning_rate": 1.6289790243059313e-05,
      "loss": 0.6212,
      "step": 2025
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7566088646949041,
      "learning_rate": 1.6285995856421843e-05,
      "loss": 0.6237,
      "step": 2026
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6515132735721905,
      "learning_rate": 1.6282199972956425e-05,
      "loss": 0.599,
      "step": 2027
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6651524952805004,
      "learning_rate": 1.6278402593566946e-05,
      "loss": 0.566,
      "step": 2028
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7417340705222062,
      "learning_rate": 1.6274603719157633e-05,
      "loss": 0.6085,
      "step": 2029
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6944628712853237,
      "learning_rate": 1.6270803350633085e-05,
      "loss": 0.5832,
      "step": 2030
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8043337703296971,
      "learning_rate": 1.626700148889825e-05,
      "loss": 0.6268,
      "step": 2031
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7445349046704591,
      "learning_rate": 1.6263198134858428e-05,
      "loss": 0.593,
      "step": 2032
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6920706000927842,
      "learning_rate": 1.625939328941928e-05,
      "loss": 0.6198,
      "step": 2033
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6857046768335133,
      "learning_rate": 1.6255586953486817e-05,
      "loss": 0.6212,
      "step": 2034
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6622777317195089,
      "learning_rate": 1.6251779127967412e-05,
      "loss": 0.6215,
      "step": 2035
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6792796748085671,
      "learning_rate": 1.6247969813767784e-05,
      "loss": 0.5967,
      "step": 2036
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6660894815187176,
      "learning_rate": 1.6244159011795012e-05,
      "loss": 0.5701,
      "step": 2037
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5770924291259197,
      "learning_rate": 1.624034672295653e-05,
      "loss": 0.5823,
      "step": 2038
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7234456410764145,
      "learning_rate": 1.6236532948160123e-05,
      "loss": 0.6262,
      "step": 2039
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7072820892665567,
      "learning_rate": 1.6232717688313933e-05,
      "loss": 0.6015,
      "step": 2040
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6287576457105444,
      "learning_rate": 1.622890094432645e-05,
      "loss": 0.6097,
      "step": 2041
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7115475368480542,
      "learning_rate": 1.6225082717106525e-05,
      "loss": 0.6186,
      "step": 2042
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5985166045829932,
      "learning_rate": 1.6221263007563352e-05,
      "loss": 0.6198,
      "step": 2043
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6396536756155031,
      "learning_rate": 1.6217441816606494e-05,
      "loss": 0.5892,
      "step": 2044
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6469935087340453,
      "learning_rate": 1.621361914514585e-05,
      "loss": 0.6062,
      "step": 2045
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6445496189458209,
      "learning_rate": 1.6209794994091676e-05,
      "loss": 0.6109,
      "step": 2046
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6879298198541857,
      "learning_rate": 1.620596936435459e-05,
      "loss": 0.5995,
      "step": 2047
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7708503335348119,
      "learning_rate": 1.6202142256845553e-05,
      "loss": 0.6258,
      "step": 2048
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.666017054461228,
      "learning_rate": 1.6198313672475875e-05,
      "loss": 0.5791,
      "step": 2049
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7501041688824491,
      "learning_rate": 1.6194483612157232e-05,
      "loss": 0.6365,
      "step": 2050
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7022389619883334,
      "learning_rate": 1.6190652076801635e-05,
      "loss": 0.5757,
      "step": 2051
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6155965962979512,
      "learning_rate": 1.618681906732145e-05,
      "loss": 0.5997,
      "step": 2052
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6454824736115756,
      "learning_rate": 1.6182984584629405e-05,
      "loss": 0.567,
      "step": 2053
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6056229175071359,
      "learning_rate": 1.6179148629638567e-05,
      "loss": 0.5846,
      "step": 2054
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7433612036549987,
      "learning_rate": 1.617531120326236e-05,
      "loss": 0.6282,
      "step": 2055
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7579906726699168,
      "learning_rate": 1.6171472306414554e-05,
      "loss": 0.6298,
      "step": 2056
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6215023367002198,
      "learning_rate": 1.6167631940009273e-05,
      "loss": 0.5727,
      "step": 2057
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6662512873693995,
      "learning_rate": 1.6163790104960984e-05,
      "loss": 0.6017,
      "step": 2058
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7339123975476414,
      "learning_rate": 1.6159946802184518e-05,
      "loss": 0.6072,
      "step": 2059
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7628444796361602,
      "learning_rate": 1.6156102032595032e-05,
      "loss": 0.6047,
      "step": 2060
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7309281164985622,
      "learning_rate": 1.6152255797108064e-05,
      "loss": 0.6471,
      "step": 2061
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6382577258196153,
      "learning_rate": 1.614840809663947e-05,
      "loss": 0.5828,
      "step": 2062
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6632877259394762,
      "learning_rate": 1.6144558932105473e-05,
      "loss": 0.5949,
      "step": 2063
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6464761610095942,
      "learning_rate": 1.6140708304422647e-05,
      "loss": 0.616,
      "step": 2064
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6730224241744551,
      "learning_rate": 1.61368562145079e-05,
      "loss": 0.5921,
      "step": 2065
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6958777213609278,
      "learning_rate": 1.613300266327849e-05,
      "loss": 0.5847,
      "step": 2066
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6523583849728118,
      "learning_rate": 1.612914765165204e-05,
      "loss": 0.5986,
      "step": 2067
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6107423575370826,
      "learning_rate": 1.6125291180546508e-05,
      "loss": 0.5925,
      "step": 2068
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6900241579676372,
      "learning_rate": 1.6121433250880193e-05,
      "loss": 0.5864,
      "step": 2069
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6168098842499028,
      "learning_rate": 1.6117573863571756e-05,
      "loss": 0.5883,
      "step": 2070
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6324814240331675,
      "learning_rate": 1.6113713019540196e-05,
      "loss": 0.5988,
      "step": 2071
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7048917579201693,
      "learning_rate": 1.6109850719704864e-05,
      "loss": 0.5936,
      "step": 2072
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7140485063634407,
      "learning_rate": 1.6105986964985453e-05,
      "loss": 0.6025,
      "step": 2073
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6267070430043907,
      "learning_rate": 1.6102121756302e-05,
      "loss": 0.6171,
      "step": 2074
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6748396703585599,
      "learning_rate": 1.60982550945749e-05,
      "loss": 0.6109,
      "step": 2075
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.622440408450489,
      "learning_rate": 1.609438698072488e-05,
      "loss": 0.6174,
      "step": 2076
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6162445655625806,
      "learning_rate": 1.6090517415673024e-05,
      "loss": 0.5942,
      "step": 2077
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6337022969938448,
      "learning_rate": 1.6086646400340756e-05,
      "loss": 0.6105,
      "step": 2078
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6104888805295675,
      "learning_rate": 1.6082773935649843e-05,
      "loss": 0.5925,
      "step": 2079
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6272733311541707,
      "learning_rate": 1.607890002252241e-05,
      "loss": 0.6067,
      "step": 2080
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6284988618720627,
      "learning_rate": 1.6075024661880903e-05,
      "loss": 0.5955,
      "step": 2081
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6279742250098971,
      "learning_rate": 1.6071147854648137e-05,
      "loss": 0.5988,
      "step": 2082
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.737869246516799,
      "learning_rate": 1.606726960174726e-05,
      "loss": 0.6151,
      "step": 2083
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7132140166897173,
      "learning_rate": 1.6063389904101764e-05,
      "loss": 0.5949,
      "step": 2084
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5981409367578778,
      "learning_rate": 1.6059508762635482e-05,
      "loss": 0.5946,
      "step": 2085
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6670210093423784,
      "learning_rate": 1.6055626178272606e-05,
      "loss": 0.6047,
      "step": 2086
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6240197145362942,
      "learning_rate": 1.6051742151937655e-05,
      "loss": 0.5958,
      "step": 2087
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6798610098480536,
      "learning_rate": 1.6047856684555493e-05,
      "loss": 0.6214,
      "step": 2088
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6569115475506464,
      "learning_rate": 1.6043969777051342e-05,
      "loss": 0.6156,
      "step": 2089
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7165102631463022,
      "learning_rate": 1.604008143035075e-05,
      "loss": 0.6231,
      "step": 2090
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6339496503998806,
      "learning_rate": 1.6036191645379613e-05,
      "loss": 0.5983,
      "step": 2091
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6461902574557974,
      "learning_rate": 1.6032300423064174e-05,
      "loss": 0.62,
      "step": 2092
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7940128145268187,
      "learning_rate": 1.6028407764331015e-05,
      "loss": 0.6221,
      "step": 2093
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6387787715837138,
      "learning_rate": 1.602451367010706e-05,
      "loss": 0.5976,
      "step": 2094
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.714293623289142,
      "learning_rate": 1.602061814131957e-05,
      "loss": 0.6061,
      "step": 2095
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7032046899039923,
      "learning_rate": 1.601672117889616e-05,
      "loss": 0.5954,
      "step": 2096
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6784678640583861,
      "learning_rate": 1.6012822783764774e-05,
      "loss": 0.6004,
      "step": 2097
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7612244401602836,
      "learning_rate": 1.60089229568537e-05,
      "loss": 0.632,
      "step": 2098
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6348524809139874,
      "learning_rate": 1.6005021699091576e-05,
      "loss": 0.5975,
      "step": 2099
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.627961182434351,
      "learning_rate": 1.6001119011407366e-05,
      "loss": 0.5998,
      "step": 2100
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6765795946983133,
      "learning_rate": 1.5997214894730387e-05,
      "loss": 0.5945,
      "step": 2101
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8430605999967865,
      "learning_rate": 1.5993309349990293e-05,
      "loss": 0.628,
      "step": 2102
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7806536659549623,
      "learning_rate": 1.5989402378117068e-05,
      "loss": 0.5996,
      "step": 2103
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6694954456333314,
      "learning_rate": 1.598549398004105e-05,
      "loss": 0.617,
      "step": 2104
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6592194157491142,
      "learning_rate": 1.5981584156692912e-05,
      "loss": 0.5959,
      "step": 2105
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.608208823314461,
      "learning_rate": 1.5977672909003664e-05,
      "loss": 0.589,
      "step": 2106
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.83932309475919,
      "learning_rate": 1.5973760237904653e-05,
      "loss": 0.5982,
      "step": 2107
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6876990601718534,
      "learning_rate": 1.5969846144327574e-05,
      "loss": 0.6055,
      "step": 2108
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6897953804408434,
      "learning_rate": 1.596593062920445e-05,
      "loss": 0.5901,
      "step": 2109
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6231717324037529,
      "learning_rate": 1.5962013693467652e-05,
      "loss": 0.6101,
      "step": 2110
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7814238305816216,
      "learning_rate": 1.5958095338049882e-05,
      "loss": 0.6022,
      "step": 2111
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6555843675759586,
      "learning_rate": 1.5954175563884187e-05,
      "loss": 0.6022,
      "step": 2112
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6665360825140374,
      "learning_rate": 1.595025437190394e-05,
      "loss": 0.5723,
      "step": 2113
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6977913875540762,
      "learning_rate": 1.594633176304287e-05,
      "loss": 0.5728,
      "step": 2114
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.638044434684015,
      "learning_rate": 1.594240773823502e-05,
      "loss": 0.5898,
      "step": 2115
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7620917116227771,
      "learning_rate": 1.5938482298414794e-05,
      "loss": 0.6168,
      "step": 2116
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6628338420078815,
      "learning_rate": 1.5934555444516916e-05,
      "loss": 0.5775,
      "step": 2117
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6835018036558909,
      "learning_rate": 1.5930627177476452e-05,
      "loss": 0.6024,
      "step": 2118
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6741543858855553,
      "learning_rate": 1.5926697498228808e-05,
      "loss": 0.615,
      "step": 2119
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6733187260829191,
      "learning_rate": 1.5922766407709724e-05,
      "loss": 0.5773,
      "step": 2120
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6558820707136274,
      "learning_rate": 1.5918833906855274e-05,
      "loss": 0.5761,
      "step": 2121
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6124842884396654,
      "learning_rate": 1.5914899996601865e-05,
      "loss": 0.5616,
      "step": 2122
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.691478804828533,
      "learning_rate": 1.591096467788625e-05,
      "loss": 0.6159,
      "step": 2123
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7555257813817181,
      "learning_rate": 1.590702795164551e-05,
      "loss": 0.5871,
      "step": 2124
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6465794986383748,
      "learning_rate": 1.5903089818817067e-05,
      "loss": 0.6109,
      "step": 2125
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7452328067304338,
      "learning_rate": 1.589915028033866e-05,
      "loss": 0.6013,
      "step": 2126
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6526146380651029,
      "learning_rate": 1.5895209337148388e-05,
      "loss": 0.5939,
      "step": 2127
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6640222555926911,
      "learning_rate": 1.5891266990184665e-05,
      "loss": 0.5872,
      "step": 2128
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6043857422759942,
      "learning_rate": 1.5887323240386252e-05,
      "loss": 0.5918,
      "step": 2129
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5445929581988869,
      "learning_rate": 1.5883378088692238e-05,
      "loss": 0.5832,
      "step": 2130
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6274669452749879,
      "learning_rate": 1.5879431536042047e-05,
      "loss": 0.5952,
      "step": 2131
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6911861401337472,
      "learning_rate": 1.5875483583375434e-05,
      "loss": 0.5915,
      "step": 2132
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6794973023147408,
      "learning_rate": 1.5871534231632488e-05,
      "loss": 0.6091,
      "step": 2133
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6620508257881272,
      "learning_rate": 1.5867583481753638e-05,
      "loss": 0.5975,
      "step": 2134
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6929795066949057,
      "learning_rate": 1.5863631334679638e-05,
      "loss": 0.5942,
      "step": 2135
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5875538039646863,
      "learning_rate": 1.5859677791351577e-05,
      "loss": 0.6162,
      "step": 2136
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6279313443953549,
      "learning_rate": 1.5855722852710878e-05,
      "loss": 0.5983,
      "step": 2137
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6705850153223561,
      "learning_rate": 1.5851766519699294e-05,
      "loss": 0.6141,
      "step": 2138
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6456591246805101,
      "learning_rate": 1.584780879325891e-05,
      "loss": 0.5842,
      "step": 2139
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6393741131836355,
      "learning_rate": 1.584384967433215e-05,
      "loss": 0.5993,
      "step": 2140
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6355113319626255,
      "learning_rate": 1.5839889163861756e-05,
      "loss": 0.5842,
      "step": 2141
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7288643162842169,
      "learning_rate": 1.5835927262790812e-05,
      "loss": 0.6086,
      "step": 2142
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6671896194293089,
      "learning_rate": 1.5831963972062734e-05,
      "loss": 0.6076,
      "step": 2143
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.729222752111582,
      "learning_rate": 1.5827999292621263e-05,
      "loss": 0.5953,
      "step": 2144
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5897788946276816,
      "learning_rate": 1.5824033225410463e-05,
      "loss": 0.595,
      "step": 2145
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6627039196875185,
      "learning_rate": 1.582006577137475e-05,
      "loss": 0.6114,
      "step": 2146
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6667469388525606,
      "learning_rate": 1.5816096931458854e-05,
      "loss": 0.6011,
      "step": 2147
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6246382276619443,
      "learning_rate": 1.5812126706607846e-05,
      "loss": 0.5827,
      "step": 2148
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6915018093559678,
      "learning_rate": 1.5808155097767107e-05,
      "loss": 0.5731,
      "step": 2149
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6480994357377479,
      "learning_rate": 1.580418210588237e-05,
      "loss": 0.6053,
      "step": 2150
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6451375491327838,
      "learning_rate": 1.5800207731899683e-05,
      "loss": 0.6049,
      "step": 2151
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6174537737524287,
      "learning_rate": 1.579623197676543e-05,
      "loss": 0.602,
      "step": 2152
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6393899683767474,
      "learning_rate": 1.579225484142633e-05,
      "loss": 0.6065,
      "step": 2153
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6740951026166947,
      "learning_rate": 1.5788276326829408e-05,
      "loss": 0.6208,
      "step": 2154
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7071966043894831,
      "learning_rate": 1.578429643392204e-05,
      "loss": 0.6387,
      "step": 2155
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6130820927419297,
      "learning_rate": 1.5780315163651922e-05,
      "loss": 0.5827,
      "step": 2156
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5391799107974966,
      "learning_rate": 1.577633251696708e-05,
      "loss": 0.5659,
      "step": 2157
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6129912934262173,
      "learning_rate": 1.5772348494815864e-05,
      "loss": 0.6144,
      "step": 2158
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6670883526740916,
      "learning_rate": 1.576836309814695e-05,
      "loss": 0.6185,
      "step": 2159
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5926753639727231,
      "learning_rate": 1.5764376327909353e-05,
      "loss": 0.5939,
      "step": 2160
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6916451621048172,
      "learning_rate": 1.57603881850524e-05,
      "loss": 0.5963,
      "step": 2161
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6441810979787805,
      "learning_rate": 1.5756398670525753e-05,
      "loss": 0.5891,
      "step": 2162
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6325293785381719,
      "learning_rate": 1.57524077852794e-05,
      "loss": 0.6053,
      "step": 2163
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6159571640136973,
      "learning_rate": 1.5748415530263656e-05,
      "loss": 0.5931,
      "step": 2164
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5976929036215347,
      "learning_rate": 1.5744421906429163e-05,
      "loss": 0.5731,
      "step": 2165
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6525774397719374,
      "learning_rate": 1.5740426914726882e-05,
      "loss": 0.6006,
      "step": 2166
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5985579840589627,
      "learning_rate": 1.5736430556108104e-05,
      "loss": 0.5663,
      "step": 2167
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7208420049818208,
      "learning_rate": 1.5732432831524448e-05,
      "loss": 0.6216,
      "step": 2168
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6701991741889645,
      "learning_rate": 1.572843374192786e-05,
      "loss": 0.5894,
      "step": 2169
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6471633467045068,
      "learning_rate": 1.5724433288270606e-05,
      "loss": 0.6063,
      "step": 2170
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7114510670975042,
      "learning_rate": 1.572043147150528e-05,
      "loss": 0.6158,
      "step": 2171
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.648078077141092,
      "learning_rate": 1.5716428292584788e-05,
      "loss": 0.6147,
      "step": 2172
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6346027766862848,
      "learning_rate": 1.571242375246238e-05,
      "loss": 0.6066,
      "step": 2173
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7294793160208628,
      "learning_rate": 1.5708417852091625e-05,
      "loss": 0.5943,
      "step": 2174
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6186457647267415,
      "learning_rate": 1.5704410592426406e-05,
      "loss": 0.6233,
      "step": 2175
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6761819096169498,
      "learning_rate": 1.5700401974420935e-05,
      "loss": 0.612,
      "step": 2176
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.67100183403496,
      "learning_rate": 1.5696391999029753e-05,
      "loss": 0.6087,
      "step": 2177
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.606451987786465,
      "learning_rate": 1.5692380667207715e-05,
      "loss": 0.5928,
      "step": 2178
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5786852835366489,
      "learning_rate": 1.5688367979910004e-05,
      "loss": 0.5861,
      "step": 2179
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5965696945464232,
      "learning_rate": 1.568435393809213e-05,
      "loss": 0.5929,
      "step": 2180
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6130121776284763,
      "learning_rate": 1.5680338542709917e-05,
      "loss": 0.5987,
      "step": 2181
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6713869320713862,
      "learning_rate": 1.5676321794719517e-05,
      "loss": 0.5864,
      "step": 2182
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.001313005517303,
      "learning_rate": 1.56723036950774e-05,
      "loss": 0.5965,
      "step": 2183
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6235411762063624,
      "learning_rate": 1.566828424474036e-05,
      "loss": 0.6082,
      "step": 2184
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6519596463306125,
      "learning_rate": 1.5664263444665518e-05,
      "loss": 0.6067,
      "step": 2185
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7448620801586588,
      "learning_rate": 1.5660241295810307e-05,
      "loss": 0.6188,
      "step": 2186
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6558862747653865,
      "learning_rate": 1.565621779913248e-05,
      "loss": 0.5686,
      "step": 2187
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.765582538705813,
      "learning_rate": 1.5652192955590127e-05,
      "loss": 0.614,
      "step": 2188
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6909009420076274,
      "learning_rate": 1.564816676614164e-05,
      "loss": 0.5971,
      "step": 2189
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6843900118286678,
      "learning_rate": 1.5644139231745743e-05,
      "loss": 0.6062,
      "step": 2190
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7119450008182545,
      "learning_rate": 1.5640110353361478e-05,
      "loss": 0.5879,
      "step": 2191
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6653908948573644,
      "learning_rate": 1.5636080131948204e-05,
      "loss": 0.6126,
      "step": 2192
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6834203444771193,
      "learning_rate": 1.56320485684656e-05,
      "loss": 0.6073,
      "step": 2193
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7538794073356009,
      "learning_rate": 1.5628015663873664e-05,
      "loss": 0.5939,
      "step": 2194
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.595182829192296,
      "learning_rate": 1.5623981419132722e-05,
      "loss": 0.6129,
      "step": 2195
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7519267423969099,
      "learning_rate": 1.5619945835203413e-05,
      "loss": 0.6236,
      "step": 2196
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6690204629534116,
      "learning_rate": 1.5615908913046686e-05,
      "loss": 0.621,
      "step": 2197
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6203036034648306,
      "learning_rate": 1.5611870653623826e-05,
      "loss": 0.5708,
      "step": 2198
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6423563008776022,
      "learning_rate": 1.560783105789642e-05,
      "loss": 0.5785,
      "step": 2199
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6427344696436611,
      "learning_rate": 1.560379012682639e-05,
      "loss": 0.6106,
      "step": 2200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7238561405475982,
      "learning_rate": 1.5599747861375957e-05,
      "loss": 0.6018,
      "step": 2201
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7708958549136212,
      "learning_rate": 1.5595704262507672e-05,
      "loss": 0.6023,
      "step": 2202
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6671202392776475,
      "learning_rate": 1.5591659331184407e-05,
      "loss": 0.6184,
      "step": 2203
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6648019480862885,
      "learning_rate": 1.5587613068369337e-05,
      "loss": 0.5896,
      "step": 2204
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6319859324849557,
      "learning_rate": 1.558356547502597e-05,
      "loss": 0.5776,
      "step": 2205
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7123213488242025,
      "learning_rate": 1.557951655211812e-05,
      "loss": 0.6179,
      "step": 2206
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6278815049204538,
      "learning_rate": 1.5575466300609917e-05,
      "loss": 0.6137,
      "step": 2207
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6261374149181158,
      "learning_rate": 1.5571414721465818e-05,
      "loss": 0.5833,
      "step": 2208
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7111761307989594,
      "learning_rate": 1.5567361815650586e-05,
      "loss": 0.6096,
      "step": 2209
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6127100646924926,
      "learning_rate": 1.5563307584129302e-05,
      "loss": 0.6076,
      "step": 2210
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6434102356898731,
      "learning_rate": 1.5559252027867368e-05,
      "loss": 0.5993,
      "step": 2211
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6854541783584319,
      "learning_rate": 1.5555195147830488e-05,
      "loss": 0.6214,
      "step": 2212
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6570934795936713,
      "learning_rate": 1.55511369449847e-05,
      "loss": 0.597,
      "step": 2213
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6201845846564606,
      "learning_rate": 1.5547077420296345e-05,
      "loss": 0.6037,
      "step": 2214
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7592473825333094,
      "learning_rate": 1.554301657473208e-05,
      "loss": 0.6135,
      "step": 2215
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6306935019111757,
      "learning_rate": 1.553895440925888e-05,
      "loss": 0.5963,
      "step": 2216
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.686822136971876,
      "learning_rate": 1.5534890924844025e-05,
      "loss": 0.6103,
      "step": 2217
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6541193530085067,
      "learning_rate": 1.5530826122455128e-05,
      "loss": 0.5962,
      "step": 2218
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.664392961664467,
      "learning_rate": 1.5526760003060095e-05,
      "loss": 0.62,
      "step": 2219
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6195216301805262,
      "learning_rate": 1.5522692567627157e-05,
      "loss": 0.5878,
      "step": 2220
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6093267759432759,
      "learning_rate": 1.5518623817124858e-05,
      "loss": 0.5783,
      "step": 2221
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6157159829310918,
      "learning_rate": 1.551455375252205e-05,
      "loss": 0.6085,
      "step": 2222
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6430803455738368,
      "learning_rate": 1.5510482374787902e-05,
      "loss": 0.585,
      "step": 2223
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6995243649742997,
      "learning_rate": 1.5506409684891897e-05,
      "loss": 0.6269,
      "step": 2224
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6527398972490864,
      "learning_rate": 1.5502335683803828e-05,
      "loss": 0.6061,
      "step": 2225
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6700784837088646,
      "learning_rate": 1.5498260372493795e-05,
      "loss": 0.6042,
      "step": 2226
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.667322339657429,
      "learning_rate": 1.5494183751932227e-05,
      "loss": 0.6065,
      "step": 2227
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.008878288622194,
      "learning_rate": 1.549010582308984e-05,
      "loss": 0.6585,
      "step": 2228
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6931816377868014,
      "learning_rate": 1.548602658693768e-05,
      "loss": 0.5744,
      "step": 2229
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.572742141595676,
      "learning_rate": 1.54819460444471e-05,
      "loss": 0.5847,
      "step": 2230
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6234312709513854,
      "learning_rate": 1.5477864196589762e-05,
      "loss": 0.5865,
      "step": 2231
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6829566913790239,
      "learning_rate": 1.547378104433764e-05,
      "loss": 0.6299,
      "step": 2232
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.700483314116681,
      "learning_rate": 1.546969658866302e-05,
      "loss": 0.6912,
      "step": 2233
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.708761986222273,
      "learning_rate": 1.5465610830538496e-05,
      "loss": 0.5962,
      "step": 2234
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6781261801991474,
      "learning_rate": 1.546152377093697e-05,
      "loss": 0.5982,
      "step": 2235
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6832881283451332,
      "learning_rate": 1.5457435410831662e-05,
      "loss": 0.6361,
      "step": 2236
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6742349317974999,
      "learning_rate": 1.5453345751196095e-05,
      "loss": 0.6048,
      "step": 2237
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6975079813177535,
      "learning_rate": 1.54492547930041e-05,
      "loss": 0.5819,
      "step": 2238
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6615298156185224,
      "learning_rate": 1.5445162537229825e-05,
      "loss": 0.6016,
      "step": 2239
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6499065597714421,
      "learning_rate": 1.5441068984847718e-05,
      "loss": 0.6034,
      "step": 2240
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.695361483928165,
      "learning_rate": 1.5436974136832542e-05,
      "loss": 0.5835,
      "step": 2241
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.475540030478772,
      "learning_rate": 1.5432877994159365e-05,
      "loss": 0.641,
      "step": 2242
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6805222645014024,
      "learning_rate": 1.542878055780357e-05,
      "loss": 0.6027,
      "step": 2243
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6589397110183698,
      "learning_rate": 1.5424681828740837e-05,
      "loss": 0.5897,
      "step": 2244
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6713317856127986,
      "learning_rate": 1.5420581807947158e-05,
      "loss": 0.5917,
      "step": 2245
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7528869040317719,
      "learning_rate": 1.5416480496398843e-05,
      "loss": 0.5517,
      "step": 2246
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6306773933948566,
      "learning_rate": 1.541237789507249e-05,
      "loss": 0.5984,
      "step": 2247
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0892886213731443,
      "learning_rate": 1.5408274004945024e-05,
      "loss": 0.6057,
      "step": 2248
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8804993495451449,
      "learning_rate": 1.5404168826993665e-05,
      "loss": 0.6512,
      "step": 2249
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.368018929050576,
      "learning_rate": 1.540006236219594e-05,
      "loss": 0.6183,
      "step": 2250
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6271563414282109,
      "learning_rate": 1.5395954611529684e-05,
      "loss": 0.6082,
      "step": 2251
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8322585536818564,
      "learning_rate": 1.5391845575973046e-05,
      "loss": 0.5977,
      "step": 2252
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.323911757826253,
      "learning_rate": 1.5387735256504464e-05,
      "loss": 0.6761,
      "step": 2253
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6381181673937265,
      "learning_rate": 1.5383623654102697e-05,
      "loss": 0.6179,
      "step": 2254
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6571636221440611,
      "learning_rate": 1.5379510769746803e-05,
      "loss": 0.5943,
      "step": 2255
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6355712449217025,
      "learning_rate": 1.537539660441615e-05,
      "loss": 0.635,
      "step": 2256
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.724053505574232,
      "learning_rate": 1.53712811590904e-05,
      "loss": 0.6103,
      "step": 2257
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.681888174236165,
      "learning_rate": 1.5367164434749536e-05,
      "loss": 0.5823,
      "step": 2258
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6889585782917843,
      "learning_rate": 1.5363046432373824e-05,
      "loss": 0.6196,
      "step": 2259
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6071497206632643,
      "learning_rate": 1.535892715294386e-05,
      "loss": 0.5988,
      "step": 2260
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.234550308026726,
      "learning_rate": 1.535480659744053e-05,
      "loss": 0.6285,
      "step": 2261
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7744949533377783,
      "learning_rate": 1.5350684766845017e-05,
      "loss": 0.6112,
      "step": 2262
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6882179090851055,
      "learning_rate": 1.5346561662138817e-05,
      "loss": 0.6056,
      "step": 2263
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6481970724336378,
      "learning_rate": 1.5342437284303738e-05,
      "loss": 0.5985,
      "step": 2264
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6590554281055637,
      "learning_rate": 1.5338311634321868e-05,
      "loss": 0.6171,
      "step": 2265
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.953371561968166,
      "learning_rate": 1.5334184713175618e-05,
      "loss": 0.6358,
      "step": 2266
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7387470593282393,
      "learning_rate": 1.5330056521847695e-05,
      "loss": 0.6288,
      "step": 2267
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7168735861516415,
      "learning_rate": 1.532592706132111e-05,
      "loss": 0.6265,
      "step": 2268
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6663419160902614,
      "learning_rate": 1.5321796332579167e-05,
      "loss": 0.5798,
      "step": 2269
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6218265934961675,
      "learning_rate": 1.5317664336605488e-05,
      "loss": 0.5648,
      "step": 2270
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7024232258239037,
      "learning_rate": 1.5313531074383984e-05,
      "loss": 0.6053,
      "step": 2271
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.161154256216162,
      "learning_rate": 1.530939654689887e-05,
      "loss": 0.6404,
      "step": 2272
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6221818867204687,
      "learning_rate": 1.5305260755134668e-05,
      "loss": 0.5955,
      "step": 2273
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1466075088459777,
      "learning_rate": 1.530112370007619e-05,
      "loss": 0.6536,
      "step": 2274
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7478098843101622,
      "learning_rate": 1.5296985382708573e-05,
      "loss": 0.5903,
      "step": 2275
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.906640245323243,
      "learning_rate": 1.5292845804017216e-05,
      "loss": 0.676,
      "step": 2276
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6732029648430545,
      "learning_rate": 1.528870496498785e-05,
      "loss": 0.6257,
      "step": 2277
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.693568528230171,
      "learning_rate": 1.52845628666065e-05,
      "loss": 0.5817,
      "step": 2278
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6665368699121943,
      "learning_rate": 1.5280419509859477e-05,
      "loss": 0.6325,
      "step": 2279
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6164698928899979,
      "learning_rate": 1.527627489573341e-05,
      "loss": 0.5952,
      "step": 2280
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8871203303562296,
      "learning_rate": 1.5272129025215217e-05,
      "loss": 0.6008,
      "step": 2281
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6720254989447976,
      "learning_rate": 1.526798189929211e-05,
      "loss": 0.5943,
      "step": 2282
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6275095696365394,
      "learning_rate": 1.5263833518951616e-05,
      "loss": 0.5811,
      "step": 2283
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6749476328147515,
      "learning_rate": 1.525968388518155e-05,
      "loss": 0.6106,
      "step": 2284
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6511118901356789,
      "learning_rate": 1.5255532998970023e-05,
      "loss": 0.5955,
      "step": 2285
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9044151939550623,
      "learning_rate": 1.5251380861305452e-05,
      "loss": 0.634,
      "step": 2286
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6577274850218251,
      "learning_rate": 1.5247227473176547e-05,
      "loss": 0.5935,
      "step": 2287
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6426959883880929,
      "learning_rate": 1.5243072835572319e-05,
      "loss": 0.597,
      "step": 2288
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7764568719805157,
      "learning_rate": 1.523891694948207e-05,
      "loss": 0.6396,
      "step": 2289
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7368081178348983,
      "learning_rate": 1.523475981589541e-05,
      "loss": 0.587,
      "step": 2290
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7365955041523506,
      "learning_rate": 1.5230601435802235e-05,
      "loss": 0.6067,
      "step": 2291
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7783349927164628,
      "learning_rate": 1.5226441810192744e-05,
      "loss": 0.6326,
      "step": 2292
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6585096130576116,
      "learning_rate": 1.5222280940057436e-05,
      "loss": 0.6076,
      "step": 2293
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6702802668444032,
      "learning_rate": 1.5218118826387099e-05,
      "loss": 0.6024,
      "step": 2294
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6921464292744164,
      "learning_rate": 1.5213955470172814e-05,
      "loss": 0.6241,
      "step": 2295
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6057022104378911,
      "learning_rate": 1.5209790872405972e-05,
      "loss": 0.5881,
      "step": 2296
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6572205986838595,
      "learning_rate": 1.5205625034078244e-05,
      "loss": 0.5811,
      "step": 2297
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6964964936300572,
      "learning_rate": 1.5201457956181612e-05,
      "loss": 0.5859,
      "step": 2298
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6636351957527278,
      "learning_rate": 1.5197289639708341e-05,
      "loss": 0.6029,
      "step": 2299
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6576852988358431,
      "learning_rate": 1.5193120085650996e-05,
      "loss": 0.6024,
      "step": 2300
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7552106782527116,
      "learning_rate": 1.5188949295002431e-05,
      "loss": 0.6338,
      "step": 2301
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.251687226957976,
      "learning_rate": 1.5184777268755803e-05,
      "loss": 0.64,
      "step": 2302
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6729079058730125,
      "learning_rate": 1.518060400790456e-05,
      "loss": 0.6136,
      "step": 2303
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6603117220333655,
      "learning_rate": 1.5176429513442443e-05,
      "loss": 0.5806,
      "step": 2304
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6629332242401234,
      "learning_rate": 1.5172253786363484e-05,
      "loss": 0.6125,
      "step": 2305
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6516053536983649,
      "learning_rate": 1.5168076827662014e-05,
      "loss": 0.6062,
      "step": 2306
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6338877465906607,
      "learning_rate": 1.5163898638332652e-05,
      "loss": 0.5877,
      "step": 2307
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6172196670177551,
      "learning_rate": 1.515971921937032e-05,
      "loss": 0.5924,
      "step": 2308
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6078594241464895,
      "learning_rate": 1.515553857177022e-05,
      "loss": 0.5867,
      "step": 2309
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.63292180029032,
      "learning_rate": 1.5151356696527852e-05,
      "loss": 0.5767,
      "step": 2310
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6427909923359005,
      "learning_rate": 1.5147173594639007e-05,
      "loss": 0.5911,
      "step": 2311
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.035886174570441,
      "learning_rate": 1.5142989267099775e-05,
      "loss": 0.6294,
      "step": 2312
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6461741371784953,
      "learning_rate": 1.5138803714906529e-05,
      "loss": 0.5841,
      "step": 2313
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6288408520297523,
      "learning_rate": 1.513461693905594e-05,
      "loss": 0.5987,
      "step": 2314
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6694833396642741,
      "learning_rate": 1.5130428940544963e-05,
      "loss": 0.609,
      "step": 2315
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5732749708016537,
      "learning_rate": 1.5126239720370852e-05,
      "loss": 0.5881,
      "step": 2316
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6865002010894448,
      "learning_rate": 1.5122049279531143e-05,
      "loss": 0.5884,
      "step": 2317
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7180079255555121,
      "learning_rate": 1.5117857619023677e-05,
      "loss": 0.5995,
      "step": 2318
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6100528421504219,
      "learning_rate": 1.5113664739846571e-05,
      "loss": 0.5895,
      "step": 2319
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5776351491721247,
      "learning_rate": 1.5109470642998241e-05,
      "loss": 0.5787,
      "step": 2320
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7032180189542605,
      "learning_rate": 1.5105275329477384e-05,
      "loss": 0.6227,
      "step": 2321
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6017633714364087,
      "learning_rate": 1.5101078800282998e-05,
      "loss": 0.5896,
      "step": 2322
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6269993800910806,
      "learning_rate": 1.5096881056414364e-05,
      "loss": 0.5684,
      "step": 2323
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6449646476458186,
      "learning_rate": 1.5092682098871052e-05,
      "loss": 0.6292,
      "step": 2324
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.658177755979956,
      "learning_rate": 1.5088481928652921e-05,
      "loss": 0.5953,
      "step": 2325
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6464647513176688,
      "learning_rate": 1.5084280546760122e-05,
      "loss": 0.6147,
      "step": 2326
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7472006241127037,
      "learning_rate": 1.5080077954193093e-05,
      "loss": 0.6503,
      "step": 2327
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6002511877674827,
      "learning_rate": 1.5075874151952557e-05,
      "loss": 0.5852,
      "step": 2328
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6518138423341598,
      "learning_rate": 1.507166914103953e-05,
      "loss": 0.6091,
      "step": 2329
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6865004561147199,
      "learning_rate": 1.5067462922455314e-05,
      "loss": 0.6078,
      "step": 2330
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6692764299556285,
      "learning_rate": 1.5063255497201497e-05,
      "loss": 0.6096,
      "step": 2331
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6298018343445762,
      "learning_rate": 1.5059046866279957e-05,
      "loss": 0.5928,
      "step": 2332
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.576350538003907,
      "learning_rate": 1.5054837030692855e-05,
      "loss": 0.5621,
      "step": 2333
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.633171149919194,
      "learning_rate": 1.5050625991442642e-05,
      "loss": 0.5839,
      "step": 2334
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6540413582352169,
      "learning_rate": 1.5046413749532058e-05,
      "loss": 0.5973,
      "step": 2335
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5670688425530768,
      "learning_rate": 1.5042200305964123e-05,
      "loss": 0.6214,
      "step": 2336
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7630483106466941,
      "learning_rate": 1.503798566174215e-05,
      "loss": 0.6125,
      "step": 2337
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7249694238787334,
      "learning_rate": 1.503376981786973e-05,
      "loss": 0.6141,
      "step": 2338
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7204193284376541,
      "learning_rate": 1.5029552775350746e-05,
      "loss": 0.6108,
      "step": 2339
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8240677033871723,
      "learning_rate": 1.5025334535189368e-05,
      "loss": 0.6071,
      "step": 2340
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5812785894563697,
      "learning_rate": 1.5021115098390046e-05,
      "loss": 0.5964,
      "step": 2341
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6349452797164659,
      "learning_rate": 1.5016894465957514e-05,
      "loss": 0.591,
      "step": 2342
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7503055071901437,
      "learning_rate": 1.5012672638896797e-05,
      "loss": 0.6046,
      "step": 2343
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6596300562610632,
      "learning_rate": 1.5008449618213196e-05,
      "loss": 0.5989,
      "step": 2344
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8249123616773596,
      "learning_rate": 1.500422540491231e-05,
      "loss": 0.6127,
      "step": 2345
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6847865242727448,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.621,
      "step": 2346
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7320489943332819,
      "learning_rate": 1.4995773404482436e-05,
      "loss": 0.5994,
      "step": 2347
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5948632303007972,
      "learning_rate": 1.4991545619366055e-05,
      "loss": 0.5908,
      "step": 2348
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.780818256949603,
      "learning_rate": 1.4987316645657581e-05,
      "loss": 0.6684,
      "step": 2349
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6169677267969315,
      "learning_rate": 1.4983086484364023e-05,
      "loss": 0.594,
      "step": 2350
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6760750677342737,
      "learning_rate": 1.4978855136492666e-05,
      "loss": 0.5987,
      "step": 2351
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6464420482759855,
      "learning_rate": 1.4974622603051093e-05,
      "loss": 0.6346,
      "step": 2352
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6642356437791392,
      "learning_rate": 1.497038888504715e-05,
      "loss": 0.5997,
      "step": 2353
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7431190721963632,
      "learning_rate": 1.4966153983488978e-05,
      "loss": 0.6091,
      "step": 2354
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2009511497798877,
      "learning_rate": 1.4961917899384999e-05,
      "loss": 0.6093,
      "step": 2355
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.630067738824808,
      "learning_rate": 1.495768063374391e-05,
      "loss": 0.6014,
      "step": 2356
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5870176752500378,
      "learning_rate": 1.4953442187574694e-05,
      "loss": 0.6001,
      "step": 2357
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6526695382230676,
      "learning_rate": 1.4949202561886615e-05,
      "loss": 0.596,
      "step": 2358
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5838902698419781,
      "learning_rate": 1.4944961757689216e-05,
      "loss": 0.5791,
      "step": 2359
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5543705354368677,
      "learning_rate": 1.4940719775992326e-05,
      "loss": 0.5643,
      "step": 2360
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5903638090037195,
      "learning_rate": 1.4936476617806041e-05,
      "loss": 0.5919,
      "step": 2361
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6856202274543421,
      "learning_rate": 1.4932232284140755e-05,
      "loss": 0.6195,
      "step": 2362
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6599075077061134,
      "learning_rate": 1.4927986776007129e-05,
      "loss": 0.5945,
      "step": 2363
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6487446425745201,
      "learning_rate": 1.492374009441611e-05,
      "loss": 0.5836,
      "step": 2364
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6574163294312556,
      "learning_rate": 1.4919492240378923e-05,
      "loss": 0.593,
      "step": 2365
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.145460809276835,
      "learning_rate": 1.4915243214907067e-05,
      "loss": 0.6064,
      "step": 2366
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7636996191024492,
      "learning_rate": 1.4910993019012328e-05,
      "loss": 0.6272,
      "step": 2367
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5866855008231171,
      "learning_rate": 1.4906741653706767e-05,
      "loss": 0.6002,
      "step": 2368
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7007921317902166,
      "learning_rate": 1.4902489120002722e-05,
      "loss": 0.5956,
      "step": 2369
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.648179786327304,
      "learning_rate": 1.4898235418912812e-05,
      "loss": 0.5734,
      "step": 2370
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6023461803831578,
      "learning_rate": 1.4893980551449932e-05,
      "loss": 0.5658,
      "step": 2371
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6178609150805209,
      "learning_rate": 1.4889724518627255e-05,
      "loss": 0.5861,
      "step": 2372
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6788789182313159,
      "learning_rate": 1.4885467321458234e-05,
      "loss": 0.578,
      "step": 2373
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7072707640932642,
      "learning_rate": 1.4881208960956596e-05,
      "loss": 0.6335,
      "step": 2374
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6719943738584503,
      "learning_rate": 1.4876949438136348e-05,
      "loss": 0.5917,
      "step": 2375
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.659730102671151,
      "learning_rate": 1.4872688754011768e-05,
      "loss": 0.6155,
      "step": 2376
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0456745871120636,
      "learning_rate": 1.4868426909597417e-05,
      "loss": 0.6237,
      "step": 2377
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5449358358570793,
      "learning_rate": 1.4864163905908133e-05,
      "loss": 0.6759,
      "step": 2378
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6242652076248209,
      "learning_rate": 1.4859899743959023e-05,
      "loss": 0.5993,
      "step": 2379
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5846482742816171,
      "learning_rate": 1.4855634424765473e-05,
      "loss": 0.5887,
      "step": 2380
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6921013782283254,
      "learning_rate": 1.485136794934315e-05,
      "loss": 0.6163,
      "step": 2381
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6016849560279548,
      "learning_rate": 1.4847100318707983e-05,
      "loss": 0.5989,
      "step": 2382
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5980585250954605,
      "learning_rate": 1.4842831533876196e-05,
      "loss": 0.594,
      "step": 2383
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6156299586618095,
      "learning_rate": 1.4838561595864269e-05,
      "loss": 0.5828,
      "step": 2384
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6356819945318439,
      "learning_rate": 1.4834290505688967e-05,
      "loss": 0.6135,
      "step": 2385
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6477979367538816,
      "learning_rate": 1.4830018264367322e-05,
      "loss": 0.5465,
      "step": 2386
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5842533296481344,
      "learning_rate": 1.482574487291665e-05,
      "loss": 0.5903,
      "step": 2387
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1289001174496436,
      "learning_rate": 1.4821470332354536e-05,
      "loss": 0.6156,
      "step": 2388
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.692841464697432,
      "learning_rate": 1.481719464369883e-05,
      "loss": 0.6035,
      "step": 2389
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.618745458225919,
      "learning_rate": 1.4812917807967672e-05,
      "loss": 0.573,
      "step": 2390
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.647883547401936,
      "learning_rate": 1.4808639826179464e-05,
      "loss": 0.6047,
      "step": 2391
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6819533380115873,
      "learning_rate": 1.4804360699352884e-05,
      "loss": 0.5791,
      "step": 2392
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6028900237531621,
      "learning_rate": 1.4800080428506883e-05,
      "loss": 0.6037,
      "step": 2393
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6537369544080288,
      "learning_rate": 1.4795799014660677e-05,
      "loss": 0.5901,
      "step": 2394
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6957004221958176,
      "learning_rate": 1.4791516458833771e-05,
      "loss": 0.6067,
      "step": 2395
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6542277362117673,
      "learning_rate": 1.478723276204592e-05,
      "loss": 0.578,
      "step": 2396
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5996670242844544,
      "learning_rate": 1.4782947925317173e-05,
      "loss": 0.5917,
      "step": 2397
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.703643420764037,
      "learning_rate": 1.4778661949667836e-05,
      "loss": 0.6031,
      "step": 2398
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6559474374680428,
      "learning_rate": 1.477437483611849e-05,
      "loss": 0.5925,
      "step": 2399
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6177811515055196,
      "learning_rate": 1.4770086585689982e-05,
      "loss": 0.5984,
      "step": 2400
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6814004256472886,
      "learning_rate": 1.4765797199403445e-05,
      "loss": 0.6077,
      "step": 2401
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7024412957049457,
      "learning_rate": 1.4761506678280264e-05,
      "loss": 0.615,
      "step": 2402
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6760659544690024,
      "learning_rate": 1.4757215023342105e-05,
      "loss": 0.5688,
      "step": 2403
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8636125592634742,
      "learning_rate": 1.47529222356109e-05,
      "loss": 0.6082,
      "step": 2404
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6055354977025341,
      "learning_rate": 1.4748628316108855e-05,
      "loss": 0.5952,
      "step": 2405
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6519702619961736,
      "learning_rate": 1.4744333265858442e-05,
      "loss": 0.5862,
      "step": 2406
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7194913113523713,
      "learning_rate": 1.4740037085882401e-05,
      "loss": 0.623,
      "step": 2407
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6612323715046722,
      "learning_rate": 1.4735739777203746e-05,
      "loss": 0.5889,
      "step": 2408
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6536485180735251,
      "learning_rate": 1.473144134084575e-05,
      "loss": 0.6121,
      "step": 2409
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6464629785460925,
      "learning_rate": 1.4727141777831969e-05,
      "loss": 0.612,
      "step": 2410
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6028828445195892,
      "learning_rate": 1.4722841089186214e-05,
      "loss": 0.5966,
      "step": 2411
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.795544660040685,
      "learning_rate": 1.4718539275932575e-05,
      "loss": 0.6799,
      "step": 2412
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6974443209120342,
      "learning_rate": 1.4714236339095397e-05,
      "loss": 0.5902,
      "step": 2413
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6452503400093992,
      "learning_rate": 1.4709932279699309e-05,
      "loss": 0.5968,
      "step": 2414
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.68632380954961,
      "learning_rate": 1.4705627098769186e-05,
      "loss": 0.6089,
      "step": 2415
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6868658716404404,
      "learning_rate": 1.4701320797330196e-05,
      "loss": 0.6078,
      "step": 2416
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.752941741700579,
      "learning_rate": 1.469701337640775e-05,
      "loss": 0.6506,
      "step": 2417
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5913203901737134,
      "learning_rate": 1.4692704837027538e-05,
      "loss": 0.5866,
      "step": 2418
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6785288448457716,
      "learning_rate": 1.4688395180215515e-05,
      "loss": 0.5932,
      "step": 2419
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.665875760307257,
      "learning_rate": 1.4684084406997903e-05,
      "loss": 0.5877,
      "step": 2420
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6207115702739142,
      "learning_rate": 1.4679772518401185e-05,
      "loss": 0.594,
      "step": 2421
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6701487572606895,
      "learning_rate": 1.4675459515452113e-05,
      "loss": 0.5952,
      "step": 2422
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5989286938917648,
      "learning_rate": 1.4671145399177702e-05,
      "loss": 0.6152,
      "step": 2423
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6145798138702994,
      "learning_rate": 1.4666830170605238e-05,
      "loss": 0.6115,
      "step": 2424
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6766643728000219,
      "learning_rate": 1.4662513830762262e-05,
      "loss": 0.5828,
      "step": 2425
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6053242578336688,
      "learning_rate": 1.465819638067659e-05,
      "loss": 0.5971,
      "step": 2426
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6637106767052112,
      "learning_rate": 1.4653877821376301e-05,
      "loss": 0.6095,
      "step": 2427
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6491885858779456,
      "learning_rate": 1.4649558153889726e-05,
      "loss": 0.6409,
      "step": 2428
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6796396382572777,
      "learning_rate": 1.4645237379245476e-05,
      "loss": 0.5768,
      "step": 2429
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5888749025264773,
      "learning_rate": 1.4640915498472415e-05,
      "loss": 0.607,
      "step": 2430
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.66980601444993,
      "learning_rate": 1.4636592512599674e-05,
      "loss": 0.5765,
      "step": 2431
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6151083818075014,
      "learning_rate": 1.4632268422656645e-05,
      "loss": 0.566,
      "step": 2432
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6384129590997393,
      "learning_rate": 1.4627943229672992e-05,
      "loss": 0.6002,
      "step": 2433
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7158191433320324,
      "learning_rate": 1.4623616934678627e-05,
      "loss": 0.6127,
      "step": 2434
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7083014763010729,
      "learning_rate": 1.4619289538703736e-05,
      "loss": 0.5932,
      "step": 2435
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8053193219166362,
      "learning_rate": 1.4614961042778762e-05,
      "loss": 0.5981,
      "step": 2436
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6842994353464504,
      "learning_rate": 1.461063144793441e-05,
      "loss": 0.6009,
      "step": 2437
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6543977338640904,
      "learning_rate": 1.4606300755201646e-05,
      "loss": 0.6098,
      "step": 2438
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6950244351835788,
      "learning_rate": 1.4601968965611704e-05,
      "loss": 0.583,
      "step": 2439
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.735063649371256,
      "learning_rate": 1.4597636080196074e-05,
      "loss": 0.6377,
      "step": 2440
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6780925015724154,
      "learning_rate": 1.4593302099986502e-05,
      "loss": 0.5947,
      "step": 2441
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6140716252417918,
      "learning_rate": 1.4588967026015004e-05,
      "loss": 0.572,
      "step": 2442
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6394279923762076,
      "learning_rate": 1.458463085931385e-05,
      "loss": 0.5967,
      "step": 2443
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7020551851113525,
      "learning_rate": 1.4580293600915579e-05,
      "loss": 0.6122,
      "step": 2444
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6694328433471742,
      "learning_rate": 1.4575955251852973e-05,
      "loss": 0.5938,
      "step": 2445
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7586270924672183,
      "learning_rate": 1.457161581315909e-05,
      "loss": 0.6125,
      "step": 2446
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6194402282979431,
      "learning_rate": 1.4567275285867243e-05,
      "loss": 0.5782,
      "step": 2447
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7215334452791189,
      "learning_rate": 1.4562933671011001e-05,
      "loss": 0.6213,
      "step": 2448
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6013672472451295,
      "learning_rate": 1.4558590969624198e-05,
      "loss": 0.6066,
      "step": 2449
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7064908625576828,
      "learning_rate": 1.4554247182740915e-05,
      "loss": 0.5867,
      "step": 2450
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6091070495836003,
      "learning_rate": 1.4549902311395503e-05,
      "loss": 0.5924,
      "step": 2451
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7237965014991403,
      "learning_rate": 1.4545556356622568e-05,
      "loss": 0.6011,
      "step": 2452
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7443428488584564,
      "learning_rate": 1.4541209319456972e-05,
      "loss": 0.6086,
      "step": 2453
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6253024597897687,
      "learning_rate": 1.4536861200933838e-05,
      "loss": 0.5972,
      "step": 2454
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.642501734880737,
      "learning_rate": 1.4532512002088544e-05,
      "loss": 0.5893,
      "step": 2455
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6892294830371651,
      "learning_rate": 1.4528161723956725e-05,
      "loss": 0.5924,
      "step": 2456
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6161863407298275,
      "learning_rate": 1.4523810367574271e-05,
      "loss": 0.5951,
      "step": 2457
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6251700850307159,
      "learning_rate": 1.4519457933977337e-05,
      "loss": 0.5936,
      "step": 2458
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6289617153610868,
      "learning_rate": 1.4515104424202325e-05,
      "loss": 0.5796,
      "step": 2459
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6125669452742845,
      "learning_rate": 1.45107498392859e-05,
      "loss": 0.5688,
      "step": 2460
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.779249225232535,
      "learning_rate": 1.4506394180264978e-05,
      "loss": 0.63,
      "step": 2461
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6333253990158062,
      "learning_rate": 1.4502037448176734e-05,
      "loss": 0.5937,
      "step": 2462
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8029221647429328,
      "learning_rate": 1.44976796440586e-05,
      "loss": 0.6091,
      "step": 2463
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.581669259093701,
      "learning_rate": 1.4493320768948258e-05,
      "loss": 0.5926,
      "step": 2464
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6088862093222326,
      "learning_rate": 1.4488960823883647e-05,
      "loss": 0.5911,
      "step": 2465
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6812026381117233,
      "learning_rate": 1.4484599809902964e-05,
      "loss": 0.5608,
      "step": 2466
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6017027992034696,
      "learning_rate": 1.448023772804466e-05,
      "loss": 0.6066,
      "step": 2467
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6426267142924121,
      "learning_rate": 1.4475874579347436e-05,
      "loss": 0.607,
      "step": 2468
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6733468011511209,
      "learning_rate": 1.4471510364850248e-05,
      "loss": 0.5968,
      "step": 2469
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6559937677917598,
      "learning_rate": 1.4467145085592309e-05,
      "loss": 0.5948,
      "step": 2470
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6049944498768087,
      "learning_rate": 1.4462778742613082e-05,
      "loss": 0.585,
      "step": 2471
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6259126851696085,
      "learning_rate": 1.4458411336952292e-05,
      "loss": 0.5945,
      "step": 2472
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6670543595383552,
      "learning_rate": 1.4454042869649902e-05,
      "loss": 0.5853,
      "step": 2473
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6243034618261296,
      "learning_rate": 1.4449673341746142e-05,
      "loss": 0.5995,
      "step": 2474
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6821993296232126,
      "learning_rate": 1.4445302754281483e-05,
      "loss": 0.5799,
      "step": 2475
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.613142923194752,
      "learning_rate": 1.4440931108296658e-05,
      "loss": 0.6016,
      "step": 2476
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.707595089041513,
      "learning_rate": 1.443655840483265e-05,
      "loss": 0.6041,
      "step": 2477
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6925848088127065,
      "learning_rate": 1.4432184644930685e-05,
      "loss": 0.6352,
      "step": 2478
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5365702369871098,
      "learning_rate": 1.4427809829632252e-05,
      "loss": 0.5944,
      "step": 2479
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6170820478565715,
      "learning_rate": 1.4423433959979085e-05,
      "loss": 0.6092,
      "step": 2480
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5970521730105459,
      "learning_rate": 1.4419057037013171e-05,
      "loss": 0.5789,
      "step": 2481
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6821999834419481,
      "learning_rate": 1.4414679061776746e-05,
      "loss": 0.5943,
      "step": 2482
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6414033208609287,
      "learning_rate": 1.4410300035312304e-05,
      "loss": 0.5913,
      "step": 2483
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.684800252331019,
      "learning_rate": 1.4405919958662575e-05,
      "loss": 0.5958,
      "step": 2484
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7707885158673116,
      "learning_rate": 1.4401538832870555e-05,
      "loss": 0.5911,
      "step": 2485
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6988765804508791,
      "learning_rate": 1.439715665897948e-05,
      "loss": 0.5911,
      "step": 2486
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6472233327522062,
      "learning_rate": 1.4392773438032833e-05,
      "loss": 0.589,
      "step": 2487
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7146061242500901,
      "learning_rate": 1.4388389171074356e-05,
      "loss": 0.5963,
      "step": 2488
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6361763449177703,
      "learning_rate": 1.4384003859148035e-05,
      "loss": 0.6138,
      "step": 2489
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6347503728034447,
      "learning_rate": 1.43796175032981e-05,
      "loss": 0.5996,
      "step": 2490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.646675487005103,
      "learning_rate": 1.4375230104569044e-05,
      "loss": 0.5933,
      "step": 2491
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6411747421222441,
      "learning_rate": 1.4370841664005592e-05,
      "loss": 0.5639,
      "step": 2492
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6717475525405543,
      "learning_rate": 1.4366452182652729e-05,
      "loss": 0.6229,
      "step": 2493
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6476871118460547,
      "learning_rate": 1.4362061661555675e-05,
      "loss": 0.6093,
      "step": 2494
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.515262845155613,
      "learning_rate": 1.4357670101759914e-05,
      "loss": 0.5469,
      "step": 2495
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6112477774938772,
      "learning_rate": 1.4353277504311164e-05,
      "loss": 0.5866,
      "step": 2496
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7033443311289705,
      "learning_rate": 1.4348883870255397e-05,
      "loss": 0.6515,
      "step": 2497
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.088032716906332,
      "learning_rate": 1.4344489200638827e-05,
      "loss": 0.6063,
      "step": 2498
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.636717391000228,
      "learning_rate": 1.4340093496507921e-05,
      "loss": 0.6046,
      "step": 2499
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.657921169509806,
      "learning_rate": 1.4335696758909386e-05,
      "loss": 0.6165,
      "step": 2500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6312280174443011,
      "learning_rate": 1.433129898889018e-05,
      "loss": 0.6124,
      "step": 2501
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5891913782452455,
      "learning_rate": 1.4326900187497504e-05,
      "loss": 0.5847,
      "step": 2502
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5776044183212418,
      "learning_rate": 1.4322500355778804e-05,
      "loss": 0.5935,
      "step": 2503
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6254423162457259,
      "learning_rate": 1.431809949478177e-05,
      "loss": 0.6221,
      "step": 2504
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5554039271665531,
      "learning_rate": 1.4313697605554349e-05,
      "loss": 0.6159,
      "step": 2505
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5904484160928205,
      "learning_rate": 1.4309294689144712e-05,
      "loss": 0.6015,
      "step": 2506
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1350006653260556,
      "learning_rate": 1.4304890746601294e-05,
      "loss": 0.5888,
      "step": 2507
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.725039539633756,
      "learning_rate": 1.4300485778972761e-05,
      "loss": 0.6221,
      "step": 2508
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5956338123018016,
      "learning_rate": 1.4296079787308035e-05,
      "loss": 0.5764,
      "step": 2509
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6381703740575206,
      "learning_rate": 1.429167277265627e-05,
      "loss": 0.6264,
      "step": 2510
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.598600174734266,
      "learning_rate": 1.428726473606687e-05,
      "loss": 0.5841,
      "step": 2511
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.659286760927626,
      "learning_rate": 1.4282855678589482e-05,
      "loss": 0.5914,
      "step": 2512
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6187954277092009,
      "learning_rate": 1.4278445601273998e-05,
      "loss": 0.5854,
      "step": 2513
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7178209837184808,
      "learning_rate": 1.4274034505170543e-05,
      "loss": 0.5828,
      "step": 2514
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7226589788544184,
      "learning_rate": 1.4269622391329501e-05,
      "loss": 0.6143,
      "step": 2515
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7104081495207951,
      "learning_rate": 1.4265209260801483e-05,
      "loss": 0.6067,
      "step": 2516
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6324649976116535,
      "learning_rate": 1.426079511463735e-05,
      "loss": 0.6108,
      "step": 2517
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6517621862637044,
      "learning_rate": 1.4256379953888202e-05,
      "loss": 0.6016,
      "step": 2518
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6632697976822143,
      "learning_rate": 1.4251963779605383e-05,
      "loss": 0.5907,
      "step": 2519
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6463325253792058,
      "learning_rate": 1.424754659284048e-05,
      "loss": 0.6099,
      "step": 2520
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6739304577927229,
      "learning_rate": 1.424312839464531e-05,
      "loss": 0.5913,
      "step": 2521
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.72963366507478,
      "learning_rate": 1.4238709186071948e-05,
      "loss": 0.5672,
      "step": 2522
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6132545457230766,
      "learning_rate": 1.4234288968172696e-05,
      "loss": 0.5991,
      "step": 2523
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6588441592424061,
      "learning_rate": 1.4229867742000103e-05,
      "loss": 0.5963,
      "step": 2524
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6495526533040382,
      "learning_rate": 1.4225445508606951e-05,
      "loss": 0.5967,
      "step": 2525
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7237709832097475,
      "learning_rate": 1.4221022269046273e-05,
      "loss": 0.5941,
      "step": 2526
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8110353055123701,
      "learning_rate": 1.4216598024371332e-05,
      "loss": 0.637,
      "step": 2527
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.793294933269033,
      "learning_rate": 1.4212172775635633e-05,
      "loss": 0.6102,
      "step": 2528
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6317401700965717,
      "learning_rate": 1.4207746523892928e-05,
      "loss": 0.5881,
      "step": 2529
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6568313308526806,
      "learning_rate": 1.4203319270197188e-05,
      "loss": 0.5865,
      "step": 2530
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6652494900217635,
      "learning_rate": 1.4198891015602648e-05,
      "loss": 0.5687,
      "step": 2531
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8694466385887616,
      "learning_rate": 1.419446176116376e-05,
      "loss": 0.6083,
      "step": 2532
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6583671411229818,
      "learning_rate": 1.4190031507935227e-05,
      "loss": 0.582,
      "step": 2533
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7335687804841802,
      "learning_rate": 1.4185600256971987e-05,
      "loss": 0.5957,
      "step": 2534
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6533196555922189,
      "learning_rate": 1.4181168009329209e-05,
      "loss": 0.5853,
      "step": 2535
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.579632360250677,
      "learning_rate": 1.4176734766062307e-05,
      "loss": 0.5895,
      "step": 2536
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.678985617852527,
      "learning_rate": 1.417230052822693e-05,
      "loss": 0.5615,
      "step": 2537
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6559081807059398,
      "learning_rate": 1.4167865296878964e-05,
      "loss": 0.5743,
      "step": 2538
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6423330634176194,
      "learning_rate": 1.416342907307453e-05,
      "loss": 0.604,
      "step": 2539
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5848182086158334,
      "learning_rate": 1.4158991857869987e-05,
      "loss": 0.5946,
      "step": 2540
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6856089627930141,
      "learning_rate": 1.4154553652321928e-05,
      "loss": 0.5984,
      "step": 2541
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5621241063458609,
      "learning_rate": 1.4150114457487183e-05,
      "loss": 0.5835,
      "step": 2542
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6559825105564661,
      "learning_rate": 1.414567427442282e-05,
      "loss": 0.6078,
      "step": 2543
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7029224878643059,
      "learning_rate": 1.4141233104186136e-05,
      "loss": 0.6014,
      "step": 2544
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7713035496717259,
      "learning_rate": 1.4136790947834673e-05,
      "loss": 0.6273,
      "step": 2545
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.710055708348854,
      "learning_rate": 1.4132347806426197e-05,
      "loss": 0.6045,
      "step": 2546
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0088247751393653,
      "learning_rate": 1.4127903681018715e-05,
      "loss": 0.6482,
      "step": 2547
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6925350246589473,
      "learning_rate": 1.4123458572670468e-05,
      "loss": 0.5955,
      "step": 2548
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6427645961080157,
      "learning_rate": 1.4119012482439929e-05,
      "loss": 0.5716,
      "step": 2549
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8250609553135185,
      "learning_rate": 1.4114565411385803e-05,
      "loss": 0.5823,
      "step": 2550
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8101469761622205,
      "learning_rate": 1.4110117360567037e-05,
      "loss": 0.6144,
      "step": 2551
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6853671498330038,
      "learning_rate": 1.41056683310428e-05,
      "loss": 0.6098,
      "step": 2552
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6060985781888628,
      "learning_rate": 1.4101218323872506e-05,
      "loss": 0.5927,
      "step": 2553
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6935959763944137,
      "learning_rate": 1.4096767340115787e-05,
      "loss": 0.6007,
      "step": 2554
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6431943652823551,
      "learning_rate": 1.4092315380832522e-05,
      "loss": 0.6335,
      "step": 2555
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6185574337127892,
      "learning_rate": 1.4087862447082814e-05,
      "loss": 0.5907,
      "step": 2556
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6850849683936957,
      "learning_rate": 1.4083408539927002e-05,
      "loss": 0.6086,
      "step": 2557
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6735455278092267,
      "learning_rate": 1.4078953660425651e-05,
      "loss": 0.6039,
      "step": 2558
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6727328104013184,
      "learning_rate": 1.4074497809639569e-05,
      "loss": 0.5996,
      "step": 2559
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6917535542346642,
      "learning_rate": 1.4070040988629778e-05,
      "loss": 0.6104,
      "step": 2560
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6315526386246582,
      "learning_rate": 1.406558319845755e-05,
      "loss": 0.621,
      "step": 2561
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5558559827204064,
      "learning_rate": 1.4061124440184375e-05,
      "loss": 0.5744,
      "step": 2562
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6341175963390567,
      "learning_rate": 1.4056664714871974e-05,
      "loss": 0.5741,
      "step": 2563
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6424072534681715,
      "learning_rate": 1.4052204023582305e-05,
      "loss": 0.5855,
      "step": 2564
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6720106134136462,
      "learning_rate": 1.4047742367377555e-05,
      "loss": 0.5868,
      "step": 2565
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6968423719144783,
      "learning_rate": 1.4043279747320133e-05,
      "loss": 0.6167,
      "step": 2566
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6652203765879963,
      "learning_rate": 1.4038816164472686e-05,
      "loss": 0.5966,
      "step": 2567
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6400799465615923,
      "learning_rate": 1.4034351619898088e-05,
      "loss": 0.6252,
      "step": 2568
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6279358731921791,
      "learning_rate": 1.4029886114659434e-05,
      "loss": 0.6294,
      "step": 2569
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.593759059535663,
      "learning_rate": 1.4025419649820065e-05,
      "loss": 0.6023,
      "step": 2570
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6394324939358667,
      "learning_rate": 1.4020952226443534e-05,
      "loss": 0.6066,
      "step": 2571
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6999632041902091,
      "learning_rate": 1.4016483845593629e-05,
      "loss": 0.6024,
      "step": 2572
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6122023288049173,
      "learning_rate": 1.4012014508334366e-05,
      "loss": 0.5838,
      "step": 2573
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6582499640324546,
      "learning_rate": 1.4007544215729991e-05,
      "loss": 0.5963,
      "step": 2574
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7136037951691465,
      "learning_rate": 1.4003072968844969e-05,
      "loss": 0.5983,
      "step": 2575
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.564759048676059,
      "learning_rate": 1.3998600768744006e-05,
      "loss": 0.5833,
      "step": 2576
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6019258263614602,
      "learning_rate": 1.399412761649202e-05,
      "loss": 0.5956,
      "step": 2577
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6019066519229791,
      "learning_rate": 1.3989653513154165e-05,
      "loss": 0.5791,
      "step": 2578
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6101731823305009,
      "learning_rate": 1.3985178459795819e-05,
      "loss": 0.5679,
      "step": 2579
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5255801740398696,
      "learning_rate": 1.3980702457482588e-05,
      "loss": 0.5752,
      "step": 2580
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5726169043222405,
      "learning_rate": 1.39762255072803e-05,
      "loss": 0.5712,
      "step": 2581
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6593857862676724,
      "learning_rate": 1.3971747610255015e-05,
      "loss": 0.5652,
      "step": 2582
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6309006210872158,
      "learning_rate": 1.3967268767473008e-05,
      "loss": 0.6077,
      "step": 2583
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5986774889781592,
      "learning_rate": 1.3962788980000793e-05,
      "loss": 0.5918,
      "step": 2584
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6047589057113255,
      "learning_rate": 1.3958308248905097e-05,
      "loss": 0.597,
      "step": 2585
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6164744192133613,
      "learning_rate": 1.3953826575252878e-05,
      "loss": 0.5783,
      "step": 2586
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.684935916133258,
      "learning_rate": 1.3949343960111315e-05,
      "loss": 0.6376,
      "step": 2587
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.682122671398694,
      "learning_rate": 1.3944860404547816e-05,
      "loss": 0.5927,
      "step": 2588
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7541099152457099,
      "learning_rate": 1.3940375909630006e-05,
      "loss": 0.6412,
      "step": 2589
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6481360937503471,
      "learning_rate": 1.3935890476425744e-05,
      "loss": 0.5999,
      "step": 2590
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7445280739798534,
      "learning_rate": 1.3931404106003101e-05,
      "loss": 0.606,
      "step": 2591
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7158985194456997,
      "learning_rate": 1.3926916799430376e-05,
      "loss": 0.5967,
      "step": 2592
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6573731506669958,
      "learning_rate": 1.3922428557776094e-05,
      "loss": 0.5836,
      "step": 2593
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.655988715869466,
      "learning_rate": 1.3917939382108998e-05,
      "loss": 0.5907,
      "step": 2594
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6739391276149579,
      "learning_rate": 1.3913449273498057e-05,
      "loss": 0.6121,
      "step": 2595
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6476372463680402,
      "learning_rate": 1.3908958233012457e-05,
      "loss": 0.6035,
      "step": 2596
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6929096156342818,
      "learning_rate": 1.3904466261721613e-05,
      "loss": 0.5974,
      "step": 2597
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1124879599151443,
      "learning_rate": 1.3899973360695154e-05,
      "loss": 0.6317,
      "step": 2598
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6344406569980513,
      "learning_rate": 1.3895479531002942e-05,
      "loss": 0.564,
      "step": 2599
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.613928123298479,
      "learning_rate": 1.3890984773715042e-05,
      "loss": 0.5889,
      "step": 2600
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5945710999877611,
      "learning_rate": 1.3886489089901756e-05,
      "loss": 0.5554,
      "step": 2601
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5996207356268803,
      "learning_rate": 1.3881992480633601e-05,
      "loss": 0.5826,
      "step": 2602
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6371930222469616,
      "learning_rate": 1.3877494946981313e-05,
      "loss": 0.5942,
      "step": 2603
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5931281560186665,
      "learning_rate": 1.3872996490015853e-05,
      "loss": 0.5461,
      "step": 2604
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7089117542480033,
      "learning_rate": 1.3868497110808394e-05,
      "loss": 0.6241,
      "step": 2605
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6953592833594501,
      "learning_rate": 1.3863996810430333e-05,
      "loss": 0.6089,
      "step": 2606
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6478416252966349,
      "learning_rate": 1.3859495589953289e-05,
      "loss": 0.5906,
      "step": 2607
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7573308128811483,
      "learning_rate": 1.3854993450449095e-05,
      "loss": 0.643,
      "step": 2608
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6458111300519656,
      "learning_rate": 1.3850490392989808e-05,
      "loss": 0.554,
      "step": 2609
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6128414921383571,
      "learning_rate": 1.3845986418647697e-05,
      "loss": 0.5944,
      "step": 2610
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6952363055693056,
      "learning_rate": 1.3841481528495255e-05,
      "loss": 0.593,
      "step": 2611
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6816390031636316,
      "learning_rate": 1.383697572360519e-05,
      "loss": 0.5918,
      "step": 2612
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6248506683156622,
      "learning_rate": 1.3832469005050433e-05,
      "loss": 0.5741,
      "step": 2613
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6506398007321821,
      "learning_rate": 1.3827961373904126e-05,
      "loss": 0.5732,
      "step": 2614
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6462464166170265,
      "learning_rate": 1.382345283123963e-05,
      "loss": 0.5878,
      "step": 2615
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6325098346362149,
      "learning_rate": 1.3818943378130522e-05,
      "loss": 0.5874,
      "step": 2616
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5884257797015272,
      "learning_rate": 1.3814433015650605e-05,
      "loss": 0.6039,
      "step": 2617
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.629445520265009,
      "learning_rate": 1.3809921744873885e-05,
      "loss": 0.6029,
      "step": 2618
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6569951412379768,
      "learning_rate": 1.380540956687459e-05,
      "loss": 0.5933,
      "step": 2619
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6532789531476376,
      "learning_rate": 1.380089648272717e-05,
      "loss": 0.588,
      "step": 2620
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6237541462851641,
      "learning_rate": 1.3796382493506277e-05,
      "loss": 0.5816,
      "step": 2621
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6555871216583027,
      "learning_rate": 1.3791867600286795e-05,
      "loss": 0.6097,
      "step": 2622
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6198726234158799,
      "learning_rate": 1.3787351804143812e-05,
      "loss": 0.6021,
      "step": 2623
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6187525203903775,
      "learning_rate": 1.3782835106152634e-05,
      "loss": 0.5774,
      "step": 2624
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7348282933703318,
      "learning_rate": 1.3778317507388778e-05,
      "loss": 0.6127,
      "step": 2625
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7112551297600634,
      "learning_rate": 1.377379900892799e-05,
      "loss": 0.6049,
      "step": 2626
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6256664899784542,
      "learning_rate": 1.3769279611846209e-05,
      "loss": 0.605,
      "step": 2627
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6437651973235785,
      "learning_rate": 1.3764759317219607e-05,
      "loss": 0.6086,
      "step": 2628
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.742632480709837,
      "learning_rate": 1.3760238126124552e-05,
      "loss": 0.611,
      "step": 2629
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7211552321533506,
      "learning_rate": 1.375571603963764e-05,
      "loss": 0.5655,
      "step": 2630
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6194403295822389,
      "learning_rate": 1.3751193058835675e-05,
      "loss": 0.6231,
      "step": 2631
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6577921489477874,
      "learning_rate": 1.3746669184795677e-05,
      "loss": 0.5956,
      "step": 2632
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6269316655633069,
      "learning_rate": 1.3742144418594869e-05,
      "loss": 0.6053,
      "step": 2633
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6284556554228546,
      "learning_rate": 1.3737618761310695e-05,
      "loss": 0.5985,
      "step": 2634
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8888130054533591,
      "learning_rate": 1.3733092214020813e-05,
      "loss": 0.6502,
      "step": 2635
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6239969122722532,
      "learning_rate": 1.3728564777803089e-05,
      "loss": 0.5903,
      "step": 2636
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6957331658121856,
      "learning_rate": 1.3724036453735593e-05,
      "loss": 0.5805,
      "step": 2637
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6550537997453668,
      "learning_rate": 1.3719507242896625e-05,
      "loss": 0.6006,
      "step": 2638
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7005325058666046,
      "learning_rate": 1.3714977146364676e-05,
      "loss": 0.6119,
      "step": 2639
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7447683271018557,
      "learning_rate": 1.3710446165218465e-05,
      "loss": 0.5945,
      "step": 2640
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7386852464728645,
      "learning_rate": 1.370591430053691e-05,
      "loss": 0.6037,
      "step": 2641
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7099495780534905,
      "learning_rate": 1.3701381553399147e-05,
      "loss": 0.6352,
      "step": 2642
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.669750114031773,
      "learning_rate": 1.369684792488451e-05,
      "loss": 0.6023,
      "step": 2643
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6343767316526393,
      "learning_rate": 1.369231341607256e-05,
      "loss": 0.5703,
      "step": 2644
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.635607190891683,
      "learning_rate": 1.3687778028043055e-05,
      "loss": 0.6082,
      "step": 2645
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6265462192320949,
      "learning_rate": 1.368324176187597e-05,
      "loss": 0.5865,
      "step": 2646
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6403578128891015,
      "learning_rate": 1.3678704618651481e-05,
      "loss": 0.5974,
      "step": 2647
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7039566939109346,
      "learning_rate": 1.3674166599449978e-05,
      "loss": 0.6233,
      "step": 2648
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5971529935153149,
      "learning_rate": 1.3669627705352063e-05,
      "loss": 0.5866,
      "step": 2649
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7266986254347653,
      "learning_rate": 1.3665087937438537e-05,
      "loss": 0.5805,
      "step": 2650
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7055471137202884,
      "learning_rate": 1.3660547296790418e-05,
      "loss": 0.5953,
      "step": 2651
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.60187078100539,
      "learning_rate": 1.3656005784488925e-05,
      "loss": 0.6049,
      "step": 2652
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6505810099876885,
      "learning_rate": 1.365146340161549e-05,
      "loss": 0.5924,
      "step": 2653
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7882492979278427,
      "learning_rate": 1.364692014925175e-05,
      "loss": 0.6064,
      "step": 2654
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6401342127021544,
      "learning_rate": 1.364237602847955e-05,
      "loss": 0.5973,
      "step": 2655
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6806083313387239,
      "learning_rate": 1.3637831040380937e-05,
      "loss": 0.5928,
      "step": 2656
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6475784553140269,
      "learning_rate": 1.3633285186038172e-05,
      "loss": 0.5939,
      "step": 2657
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6745573896111974,
      "learning_rate": 1.3628738466533716e-05,
      "loss": 0.6355,
      "step": 2658
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5634157680431997,
      "learning_rate": 1.3624190882950241e-05,
      "loss": 0.5994,
      "step": 2659
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6313546281652004,
      "learning_rate": 1.361964243637062e-05,
      "loss": 0.5896,
      "step": 2660
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6112281151358624,
      "learning_rate": 1.3615093127877938e-05,
      "loss": 0.5999,
      "step": 2661
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6801382366037984,
      "learning_rate": 1.3610542958555475e-05,
      "loss": 0.6094,
      "step": 2662
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7032390862423592,
      "learning_rate": 1.3605991929486728e-05,
      "loss": 0.6089,
      "step": 2663
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6366684057776563,
      "learning_rate": 1.3601440041755386e-05,
      "loss": 0.5787,
      "step": 2664
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5756005389664344,
      "learning_rate": 1.359688729644536e-05,
      "loss": 0.6037,
      "step": 2665
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5882718015620684,
      "learning_rate": 1.3592333694640743e-05,
      "loss": 0.5932,
      "step": 2666
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.597080462827227,
      "learning_rate": 1.358777923742585e-05,
      "loss": 0.586,
      "step": 2667
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6094343197979925,
      "learning_rate": 1.3583223925885191e-05,
      "loss": 0.5994,
      "step": 2668
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6296927494643544,
      "learning_rate": 1.3578667761103483e-05,
      "loss": 0.6014,
      "step": 2669
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6000749922045655,
      "learning_rate": 1.3574110744165644e-05,
      "loss": 0.5843,
      "step": 2670
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6051295269691391,
      "learning_rate": 1.3569552876156798e-05,
      "loss": 0.5812,
      "step": 2671
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6785228198764014,
      "learning_rate": 1.3564994158162261e-05,
      "loss": 0.5892,
      "step": 2672
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5917023709674424,
      "learning_rate": 1.356043459126757e-05,
      "loss": 0.5909,
      "step": 2673
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6453615266408496,
      "learning_rate": 1.3555874176558445e-05,
      "loss": 0.5889,
      "step": 2674
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6976438687831291,
      "learning_rate": 1.3551312915120826e-05,
      "loss": 0.6257,
      "step": 2675
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7534416088160648,
      "learning_rate": 1.3546750808040841e-05,
      "loss": 0.5791,
      "step": 2676
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6794194208360202,
      "learning_rate": 1.3542187856404823e-05,
      "loss": 0.6029,
      "step": 2677
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6058565967138131,
      "learning_rate": 1.3537624061299303e-05,
      "loss": 0.6079,
      "step": 2678
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7098269697452598,
      "learning_rate": 1.3533059423811026e-05,
      "loss": 0.6127,
      "step": 2679
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7087441755051843,
      "learning_rate": 1.352849394502692e-05,
      "loss": 0.5667,
      "step": 2680
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6661213630053333,
      "learning_rate": 1.3523927626034126e-05,
      "loss": 0.6008,
      "step": 2681
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7384506386945201,
      "learning_rate": 1.3519360467919977e-05,
      "loss": 0.6286,
      "step": 2682
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6593404080188271,
      "learning_rate": 1.3514792471772013e-05,
      "loss": 0.5824,
      "step": 2683
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7331073058469236,
      "learning_rate": 1.3510223638677973e-05,
      "loss": 0.594,
      "step": 2684
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6734863854433195,
      "learning_rate": 1.3505653969725785e-05,
      "loss": 0.5838,
      "step": 2685
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6780518939560027,
      "learning_rate": 1.3501083466003589e-05,
      "loss": 0.5982,
      "step": 2686
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6428158350906255,
      "learning_rate": 1.3496512128599713e-05,
      "loss": 0.5788,
      "step": 2687
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6333996690476603,
      "learning_rate": 1.3491939958602691e-05,
      "loss": 0.5892,
      "step": 2688
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.717021031486782,
      "learning_rate": 1.3487366957101258e-05,
      "loss": 0.594,
      "step": 2689
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6233095620269624,
      "learning_rate": 1.3482793125184332e-05,
      "loss": 0.587,
      "step": 2690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7562034492106153,
      "learning_rate": 1.3478218463941047e-05,
      "loss": 0.6025,
      "step": 2691
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6599526686700827,
      "learning_rate": 1.3473642974460724e-05,
      "loss": 0.5821,
      "step": 2692
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6325087848622067,
      "learning_rate": 1.3469066657832882e-05,
      "loss": 0.5744,
      "step": 2693
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7137958001161847,
      "learning_rate": 1.3464489515147239e-05,
      "loss": 0.5692,
      "step": 2694
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6686657872944741,
      "learning_rate": 1.3459911547493704e-05,
      "loss": 0.6081,
      "step": 2695
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7130266933284446,
      "learning_rate": 1.3455332755962398e-05,
      "loss": 0.5968,
      "step": 2696
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6883338656142735,
      "learning_rate": 1.345075314164362e-05,
      "loss": 0.6042,
      "step": 2697
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6719875380287863,
      "learning_rate": 1.3446172705627878e-05,
      "loss": 0.6002,
      "step": 2698
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5901906878328588,
      "learning_rate": 1.3441591449005862e-05,
      "loss": 0.5784,
      "step": 2699
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9907799853922614,
      "learning_rate": 1.3437009372868476e-05,
      "loss": 0.5785,
      "step": 2700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7305151614627236,
      "learning_rate": 1.3432426478306796e-05,
      "loss": 0.606,
      "step": 2701
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.614875338562151,
      "learning_rate": 1.3427842766412118e-05,
      "loss": 0.5706,
      "step": 2702
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6002112443978267,
      "learning_rate": 1.3423258238275915e-05,
      "loss": 0.6007,
      "step": 2703
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.626673299536179,
      "learning_rate": 1.341867289498986e-05,
      "loss": 0.6035,
      "step": 2704
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6182136959377279,
      "learning_rate": 1.3414086737645819e-05,
      "loss": 0.5946,
      "step": 2705
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7366709958497745,
      "learning_rate": 1.3409499767335854e-05,
      "loss": 0.605,
      "step": 2706
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5911594762104765,
      "learning_rate": 1.3404911985152216e-05,
      "loss": 0.5689,
      "step": 2707
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6813268968900199,
      "learning_rate": 1.3400323392187358e-05,
      "loss": 0.5998,
      "step": 2708
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6244239968178222,
      "learning_rate": 1.3395733989533916e-05,
      "loss": 0.5875,
      "step": 2709
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8248480718723116,
      "learning_rate": 1.3391143778284725e-05,
      "loss": 0.6275,
      "step": 2710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7240436552384649,
      "learning_rate": 1.3386552759532814e-05,
      "loss": 0.608,
      "step": 2711
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6468756759991242,
      "learning_rate": 1.3381960934371394e-05,
      "loss": 0.6083,
      "step": 2712
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6626127417298835,
      "learning_rate": 1.3377368303893882e-05,
      "loss": 0.5747,
      "step": 2713
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5800521868099373,
      "learning_rate": 1.3372774869193876e-05,
      "loss": 0.5912,
      "step": 2714
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5944183597923131,
      "learning_rate": 1.3368180631365171e-05,
      "loss": 0.6043,
      "step": 2715
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6299888494039801,
      "learning_rate": 1.3363585591501751e-05,
      "loss": 0.5894,
      "step": 2716
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6013563193384497,
      "learning_rate": 1.3358989750697797e-05,
      "loss": 0.5882,
      "step": 2717
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6386096955743051,
      "learning_rate": 1.3354393110047665e-05,
      "loss": 0.5819,
      "step": 2718
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6661244945254595,
      "learning_rate": 1.334979567064592e-05,
      "loss": 0.6079,
      "step": 2719
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.626311111943774,
      "learning_rate": 1.3345197433587306e-05,
      "loss": 0.6013,
      "step": 2720
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6631694295529584,
      "learning_rate": 1.3340598399966762e-05,
      "loss": 0.6153,
      "step": 2721
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.654464288731563,
      "learning_rate": 1.3335998570879414e-05,
      "loss": 0.5765,
      "step": 2722
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6336628948984468,
      "learning_rate": 1.3331397947420578e-05,
      "loss": 0.581,
      "step": 2723
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6460638258917362,
      "learning_rate": 1.3326796530685757e-05,
      "loss": 0.6029,
      "step": 2724
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6666393285215256,
      "learning_rate": 1.3322194321770647e-05,
      "loss": 0.5936,
      "step": 2725
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7339252145017334,
      "learning_rate": 1.3317591321771135e-05,
      "loss": 0.619,
      "step": 2726
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6968017004409147,
      "learning_rate": 1.3312987531783285e-05,
      "loss": 0.6344,
      "step": 2727
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.621686986652866,
      "learning_rate": 1.3308382952903358e-05,
      "loss": 0.5609,
      "step": 2728
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6348466969845685,
      "learning_rate": 1.3303777586227806e-05,
      "loss": 0.6072,
      "step": 2729
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6857742808086961,
      "learning_rate": 1.3299171432853259e-05,
      "loss": 0.5949,
      "step": 2730
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6030902539754488,
      "learning_rate": 1.329456449387654e-05,
      "loss": 0.601,
      "step": 2731
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5921138224194001,
      "learning_rate": 1.3289956770394661e-05,
      "loss": 0.6099,
      "step": 2732
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6130814344643953,
      "learning_rate": 1.3285348263504814e-05,
      "loss": 0.6097,
      "step": 2733
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6073856850802263,
      "learning_rate": 1.3280738974304383e-05,
      "loss": 0.5865,
      "step": 2734
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6683630807249338,
      "learning_rate": 1.327612890389094e-05,
      "loss": 0.614,
      "step": 2735
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6146030370791976,
      "learning_rate": 1.3271518053362233e-05,
      "loss": 0.5791,
      "step": 2736
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6329963014474204,
      "learning_rate": 1.3266906423816206e-05,
      "loss": 0.5958,
      "step": 2737
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6711760200523282,
      "learning_rate": 1.3262294016350987e-05,
      "loss": 0.5826,
      "step": 2738
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6756360107566766,
      "learning_rate": 1.3257680832064884e-05,
      "loss": 0.5926,
      "step": 2739
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7078200810032563,
      "learning_rate": 1.3253066872056402e-05,
      "loss": 0.5811,
      "step": 2740
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6291037419277445,
      "learning_rate": 1.3248452137424208e-05,
      "loss": 0.5647,
      "step": 2741
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5685630539267076,
      "learning_rate": 1.3243836629267177e-05,
      "loss": 0.5737,
      "step": 2742
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.611087664335368,
      "learning_rate": 1.3239220348684353e-05,
      "loss": 0.5583,
      "step": 2743
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.614215685319096,
      "learning_rate": 1.3234603296774978e-05,
      "loss": 0.5744,
      "step": 2744
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.622298390311003,
      "learning_rate": 1.322998547463846e-05,
      "loss": 0.5891,
      "step": 2745
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6864751228163479,
      "learning_rate": 1.3225366883374409e-05,
      "loss": 0.6144,
      "step": 2746
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.626227930430021,
      "learning_rate": 1.3220747524082595e-05,
      "loss": 0.5735,
      "step": 2747
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6754804676675155,
      "learning_rate": 1.3216127397863001e-05,
      "loss": 0.6101,
      "step": 2748
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7307356218292166,
      "learning_rate": 1.3211506505815764e-05,
      "loss": 0.5756,
      "step": 2749
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7411398146526191,
      "learning_rate": 1.3206884849041224e-05,
      "loss": 0.6444,
      "step": 2750
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6080098546801869,
      "learning_rate": 1.3202262428639886e-05,
      "loss": 0.608,
      "step": 2751
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5933111920386361,
      "learning_rate": 1.3197639245712454e-05,
      "loss": 0.606,
      "step": 2752
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6134287658394467,
      "learning_rate": 1.31930153013598e-05,
      "loss": 0.5808,
      "step": 2753
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5840858828719185,
      "learning_rate": 1.3188390596682985e-05,
      "loss": 0.5812,
      "step": 2754
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6545875317504112,
      "learning_rate": 1.318376513278325e-05,
      "loss": 0.5978,
      "step": 2755
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6616465176204834,
      "learning_rate": 1.3179138910762013e-05,
      "loss": 0.5869,
      "step": 2756
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7310132940957121,
      "learning_rate": 1.317451193172087e-05,
      "loss": 0.5936,
      "step": 2757
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6222532921219173,
      "learning_rate": 1.316988419676161e-05,
      "loss": 0.6072,
      "step": 2758
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6547542154590263,
      "learning_rate": 1.3165255706986193e-05,
      "loss": 0.5875,
      "step": 2759
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6567567886011019,
      "learning_rate": 1.3160626463496756e-05,
      "loss": 0.5775,
      "step": 2760
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6727984501765049,
      "learning_rate": 1.3155996467395623e-05,
      "loss": 0.5957,
      "step": 2761
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6189706955134987,
      "learning_rate": 1.315136571978529e-05,
      "loss": 0.5895,
      "step": 2762
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5909325632887681,
      "learning_rate": 1.3146734221768439e-05,
      "loss": 0.5727,
      "step": 2763
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6520620498711789,
      "learning_rate": 1.3142101974447923e-05,
      "loss": 0.5847,
      "step": 2764
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6391339526516485,
      "learning_rate": 1.3137468978926784e-05,
      "loss": 0.6022,
      "step": 2765
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6605563609794908,
      "learning_rate": 1.3132835236308228e-05,
      "loss": 0.6029,
      "step": 2766
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6330411041452324,
      "learning_rate": 1.3128200747695651e-05,
      "loss": 0.5938,
      "step": 2767
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6390279569719252,
      "learning_rate": 1.3123565514192625e-05,
      "loss": 0.6127,
      "step": 2768
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6485370230502427,
      "learning_rate": 1.3118929536902894e-05,
      "loss": 0.5892,
      "step": 2769
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5776470412413589,
      "learning_rate": 1.3114292816930378e-05,
      "loss": 0.5709,
      "step": 2770
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6884144137764797,
      "learning_rate": 1.3109655355379183e-05,
      "loss": 0.5905,
      "step": 2771
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6379851044693474,
      "learning_rate": 1.3105017153353583e-05,
      "loss": 0.6267,
      "step": 2772
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6267396217710562,
      "learning_rate": 1.3100378211958036e-05,
      "loss": 0.6022,
      "step": 2773
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6611828763436216,
      "learning_rate": 1.3095738532297164e-05,
      "loss": 0.5742,
      "step": 2774
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5994624142766619,
      "learning_rate": 1.309109811547578e-05,
      "loss": 0.6078,
      "step": 2775
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6384348389041644,
      "learning_rate": 1.3086456962598859e-05,
      "loss": 0.5642,
      "step": 2776
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5961229324314073,
      "learning_rate": 1.3081815074771562e-05,
      "loss": 0.5917,
      "step": 2777
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6748208057877048,
      "learning_rate": 1.3077172453099219e-05,
      "loss": 0.5914,
      "step": 2778
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5907525000264252,
      "learning_rate": 1.3072529098687334e-05,
      "loss": 0.576,
      "step": 2779
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0978183696276078,
      "learning_rate": 1.3067885012641589e-05,
      "loss": 0.6257,
      "step": 2780
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7053653201093003,
      "learning_rate": 1.3063240196067837e-05,
      "loss": 0.5972,
      "step": 2781
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7557038926405619,
      "learning_rate": 1.3058594650072106e-05,
      "loss": 0.6232,
      "step": 2782
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5871237897045098,
      "learning_rate": 1.3053948375760604e-05,
      "loss": 0.5738,
      "step": 2783
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7057694475798123,
      "learning_rate": 1.3049301374239702e-05,
      "loss": 0.6369,
      "step": 2784
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6899513001808715,
      "learning_rate": 1.3044653646615948e-05,
      "loss": 0.6067,
      "step": 2785
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.612196350409052,
      "learning_rate": 1.3040005193996065e-05,
      "loss": 0.5961,
      "step": 2786
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6583499576962192,
      "learning_rate": 1.3035356017486951e-05,
      "loss": 0.6028,
      "step": 2787
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6636010171325253,
      "learning_rate": 1.3030706118195669e-05,
      "loss": 0.6322,
      "step": 2788
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6365575526040282,
      "learning_rate": 1.3026055497229457e-05,
      "loss": 0.5872,
      "step": 2789
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6121989413654054,
      "learning_rate": 1.3021404155695728e-05,
      "loss": 0.597,
      "step": 2790
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6101523029838366,
      "learning_rate": 1.3016752094702064e-05,
      "loss": 0.5474,
      "step": 2791
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7900480338062238,
      "learning_rate": 1.3012099315356222e-05,
      "loss": 0.5838,
      "step": 2792
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6970893413421955,
      "learning_rate": 1.3007445818766116e-05,
      "loss": 0.609,
      "step": 2793
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6786348680698943,
      "learning_rate": 1.3002791606039853e-05,
      "loss": 0.5955,
      "step": 2794
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5970745900912974,
      "learning_rate": 1.2998136678285694e-05,
      "loss": 0.6094,
      "step": 2795
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7007708037347704,
      "learning_rate": 1.2993481036612074e-05,
      "loss": 0.592,
      "step": 2796
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6563853156345577,
      "learning_rate": 1.2988824682127605e-05,
      "loss": 0.6116,
      "step": 2797
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7233008218044049,
      "learning_rate": 1.2984167615941056e-05,
      "loss": 0.5999,
      "step": 2798
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.702520861655505,
      "learning_rate": 1.2979509839161377e-05,
      "loss": 0.6128,
      "step": 2799
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5488357265270434,
      "learning_rate": 1.2974851352897681e-05,
      "loss": 0.5644,
      "step": 2800
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6541414491043824,
      "learning_rate": 1.2970192158259251e-05,
      "loss": 0.5804,
      "step": 2801
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.632417680768741,
      "learning_rate": 1.2965532256355547e-05,
      "loss": 0.6065,
      "step": 2802
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6900182351360823,
      "learning_rate": 1.2960871648296176e-05,
      "loss": 0.6028,
      "step": 2803
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6999311395866472,
      "learning_rate": 1.2956210335190934e-05,
      "loss": 0.602,
      "step": 2804
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7060627610823739,
      "learning_rate": 1.295154831814978e-05,
      "loss": 0.5908,
      "step": 2805
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6839672512528076,
      "learning_rate": 1.2946885598282839e-05,
      "loss": 0.6016,
      "step": 2806
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8613753654251725,
      "learning_rate": 1.2942222176700397e-05,
      "loss": 0.6482,
      "step": 2807
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6422176610508766,
      "learning_rate": 1.2937558054512916e-05,
      "loss": 0.5834,
      "step": 2808
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7871352520700124,
      "learning_rate": 1.2932893232831021e-05,
      "loss": 0.6171,
      "step": 2809
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6524303185474716,
      "learning_rate": 1.2928227712765504e-05,
      "loss": 0.5763,
      "step": 2810
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5616091983115071,
      "learning_rate": 1.2923561495427327e-05,
      "loss": 0.589,
      "step": 2811
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6403121795186777,
      "learning_rate": 1.291889458192761e-05,
      "loss": 0.6078,
      "step": 2812
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6704184635954055,
      "learning_rate": 1.2914226973377646e-05,
      "loss": 0.6201,
      "step": 2813
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6461223491624738,
      "learning_rate": 1.2909558670888891e-05,
      "loss": 0.5741,
      "step": 2814
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7281080430806406,
      "learning_rate": 1.2904889675572964e-05,
      "loss": 0.6252,
      "step": 2815
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6552096771516412,
      "learning_rate": 1.2900219988541652e-05,
      "loss": 0.5928,
      "step": 2816
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6542156689293125,
      "learning_rate": 1.2895549610906909e-05,
      "loss": 0.601,
      "step": 2817
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8713263646022694,
      "learning_rate": 1.2890878543780843e-05,
      "loss": 0.6336,
      "step": 2818
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6480444116446601,
      "learning_rate": 1.2886206788275739e-05,
      "loss": 0.5889,
      "step": 2819
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.699336845755618,
      "learning_rate": 1.2881534345504041e-05,
      "loss": 0.6076,
      "step": 2820
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6570585148161123,
      "learning_rate": 1.2876861216578354e-05,
      "loss": 0.6017,
      "step": 2821
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.664322978560969,
      "learning_rate": 1.2872187402611446e-05,
      "loss": 0.6257,
      "step": 2822
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.185498339308399,
      "learning_rate": 1.2867512904716255e-05,
      "loss": 0.6286,
      "step": 2823
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6944150110508286,
      "learning_rate": 1.2862837724005872e-05,
      "loss": 0.5875,
      "step": 2824
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6101600675766988,
      "learning_rate": 1.2858161861593566e-05,
      "loss": 0.6151,
      "step": 2825
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6902294789848636,
      "learning_rate": 1.2853485318592744e-05,
      "loss": 0.5896,
      "step": 2826
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6554151956699968,
      "learning_rate": 1.2848808096117003e-05,
      "loss": 0.6097,
      "step": 2827
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5858437450810416,
      "learning_rate": 1.2844130195280076e-05,
      "loss": 0.5833,
      "step": 2828
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6004337384206234,
      "learning_rate": 1.2839451617195879e-05,
      "loss": 0.57,
      "step": 2829
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6361760341974048,
      "learning_rate": 1.2834772362978476e-05,
      "loss": 0.6132,
      "step": 2830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6682283175761966,
      "learning_rate": 1.2830092433742098e-05,
      "loss": 0.5909,
      "step": 2831
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6545948200080682,
      "learning_rate": 1.282541183060113e-05,
      "loss": 0.5767,
      "step": 2832
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7333602971765459,
      "learning_rate": 1.2820730554670128e-05,
      "loss": 0.6141,
      "step": 2833
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.617722662275663,
      "learning_rate": 1.28160486070638e-05,
      "loss": 0.6006,
      "step": 2834
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6440380080447496,
      "learning_rate": 1.2811365988897015e-05,
      "loss": 0.6019,
      "step": 2835
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6764456705853319,
      "learning_rate": 1.2806682701284803e-05,
      "loss": 0.575,
      "step": 2836
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6977989360828047,
      "learning_rate": 1.2801998745342354e-05,
      "loss": 0.5789,
      "step": 2837
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6641252551285782,
      "learning_rate": 1.2797314122185018e-05,
      "loss": 0.5908,
      "step": 2838
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7566582532047437,
      "learning_rate": 1.2792628832928302e-05,
      "loss": 0.6162,
      "step": 2839
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6162315614042702,
      "learning_rate": 1.2787942878687871e-05,
      "loss": 0.594,
      "step": 2840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7068615183379933,
      "learning_rate": 1.278325626057955e-05,
      "loss": 0.5982,
      "step": 2841
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6617048316922381,
      "learning_rate": 1.277856897971932e-05,
      "loss": 0.5735,
      "step": 2842
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6479742762036615,
      "learning_rate": 1.2773881037223321e-05,
      "loss": 0.5733,
      "step": 2843
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6721064221561188,
      "learning_rate": 1.2769192434207853e-05,
      "loss": 0.6169,
      "step": 2844
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6336117290003388,
      "learning_rate": 1.2764503171789369e-05,
      "loss": 0.6043,
      "step": 2845
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6358378957656099,
      "learning_rate": 1.2759813251084486e-05,
      "loss": 0.5865,
      "step": 2846
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6109804155962737,
      "learning_rate": 1.2755122673209963e-05,
      "loss": 0.6309,
      "step": 2847
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5798191869278048,
      "learning_rate": 1.2750431439282736e-05,
      "loss": 0.5999,
      "step": 2848
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.620469581388412,
      "learning_rate": 1.2745739550419882e-05,
      "loss": 0.6269,
      "step": 2849
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7150626102208415,
      "learning_rate": 1.2741047007738638e-05,
      "loss": 0.6097,
      "step": 2850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6654898984178338,
      "learning_rate": 1.2736353812356396e-05,
      "loss": 0.5942,
      "step": 2851
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6010833790339339,
      "learning_rate": 1.2731659965390707e-05,
      "loss": 0.5653,
      "step": 2852
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6011117696497597,
      "learning_rate": 1.2726965467959275e-05,
      "loss": 0.5605,
      "step": 2853
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6078424271703496,
      "learning_rate": 1.2722270321179958e-05,
      "loss": 0.5981,
      "step": 2854
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6147426061628354,
      "learning_rate": 1.2717574526170769e-05,
      "loss": 0.5738,
      "step": 2855
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8074547543561329,
      "learning_rate": 1.2712878084049873e-05,
      "loss": 0.613,
      "step": 2856
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6641823442793262,
      "learning_rate": 1.2708180995935595e-05,
      "loss": 0.5892,
      "step": 2857
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7190058671177609,
      "learning_rate": 1.2703483262946415e-05,
      "loss": 0.6053,
      "step": 2858
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6676689378482258,
      "learning_rate": 1.2698784886200953e-05,
      "loss": 0.5985,
      "step": 2859
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5817217135772756,
      "learning_rate": 1.2694085866817995e-05,
      "loss": 0.5861,
      "step": 2860
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7028658360856588,
      "learning_rate": 1.2689386205916477e-05,
      "loss": 0.5869,
      "step": 2861
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6525848408755479,
      "learning_rate": 1.2684685904615488e-05,
      "loss": 0.6019,
      "step": 2862
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.647187932345636,
      "learning_rate": 1.2679984964034269e-05,
      "loss": 0.6226,
      "step": 2863
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6253436318679861,
      "learning_rate": 1.2675283385292212e-05,
      "loss": 0.5935,
      "step": 2864
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.556640546033559,
      "learning_rate": 1.2670581169508857e-05,
      "loss": 0.5888,
      "step": 2865
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.584714897235693,
      "learning_rate": 1.266587831780391e-05,
      "loss": 0.5963,
      "step": 2866
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6325190561137486,
      "learning_rate": 1.2661174831297212e-05,
      "loss": 0.5972,
      "step": 2867
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5949838032210698,
      "learning_rate": 1.2656470711108763e-05,
      "loss": 0.5849,
      "step": 2868
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6645100529523673,
      "learning_rate": 1.2651765958358717e-05,
      "loss": 0.6026,
      "step": 2869
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6431752976061872,
      "learning_rate": 1.2647060574167374e-05,
      "loss": 0.5831,
      "step": 2870
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6580277657472198,
      "learning_rate": 1.2642354559655177e-05,
      "loss": 0.59,
      "step": 2871
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6942682274495378,
      "learning_rate": 1.263764791594274e-05,
      "loss": 0.6057,
      "step": 2872
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.654083067558896,
      "learning_rate": 1.2632940644150803e-05,
      "loss": 0.5879,
      "step": 2873
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6588396962782563,
      "learning_rate": 1.2628232745400269e-05,
      "loss": 0.5678,
      "step": 2874
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6293284247277653,
      "learning_rate": 1.2623524220812193e-05,
      "loss": 0.5884,
      "step": 2875
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5965635967450346,
      "learning_rate": 1.2618815071507768e-05,
      "loss": 0.5554,
      "step": 2876
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.635192532308121,
      "learning_rate": 1.2614105298608347e-05,
      "loss": 0.5885,
      "step": 2877
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6907080303433992,
      "learning_rate": 1.260939490323542e-05,
      "loss": 0.6213,
      "step": 2878
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5896932520513485,
      "learning_rate": 1.2604683886510635e-05,
      "loss": 0.589,
      "step": 2879
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6148797497572487,
      "learning_rate": 1.2599972249555782e-05,
      "loss": 0.5899,
      "step": 2880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6133192716039879,
      "learning_rate": 1.2595259993492808e-05,
      "loss": 0.5887,
      "step": 2881
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6653847007662218,
      "learning_rate": 1.259054711944379e-05,
      "loss": 0.5813,
      "step": 2882
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6934819765548345,
      "learning_rate": 1.2585833628530967e-05,
      "loss": 0.6128,
      "step": 2883
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7312901298227287,
      "learning_rate": 1.2581119521876724e-05,
      "loss": 0.6229,
      "step": 2884
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7531501122372045,
      "learning_rate": 1.2576404800603583e-05,
      "loss": 0.6167,
      "step": 2885
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6003432965848102,
      "learning_rate": 1.2571689465834223e-05,
      "loss": 0.5549,
      "step": 2886
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6271816347734306,
      "learning_rate": 1.2566973518691463e-05,
      "loss": 0.5815,
      "step": 2887
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6826702835061326,
      "learning_rate": 1.2562256960298267e-05,
      "loss": 0.605,
      "step": 2888
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.883976022787751,
      "learning_rate": 1.2557539791777749e-05,
      "loss": 0.6621,
      "step": 2889
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6193978295407891,
      "learning_rate": 1.2552822014253165e-05,
      "loss": 0.5798,
      "step": 2890
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6867548502865433,
      "learning_rate": 1.2548103628847923e-05,
      "loss": 0.6018,
      "step": 2891
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5976370608723429,
      "learning_rate": 1.2543384636685561e-05,
      "loss": 0.5984,
      "step": 2892
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6418206176852634,
      "learning_rate": 1.2538665038889775e-05,
      "loss": 0.5876,
      "step": 2893
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6063964139213707,
      "learning_rate": 1.2533944836584397e-05,
      "loss": 0.5918,
      "step": 2894
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5850749375130525,
      "learning_rate": 1.2529224030893415e-05,
      "loss": 0.582,
      "step": 2895
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6039094382650253,
      "learning_rate": 1.2524502622940944e-05,
      "loss": 0.6033,
      "step": 2896
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6485924503623914,
      "learning_rate": 1.2519780613851254e-05,
      "loss": 0.5744,
      "step": 2897
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6345432651325358,
      "learning_rate": 1.2515058004748753e-05,
      "loss": 0.6039,
      "step": 2898
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6345352907491619,
      "learning_rate": 1.2510334796757997e-05,
      "loss": 0.5846,
      "step": 2899
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6200117089656405,
      "learning_rate": 1.2505610991003678e-05,
      "loss": 0.5929,
      "step": 2900
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.679784135371385,
      "learning_rate": 1.250088658861063e-05,
      "loss": 0.6239,
      "step": 2901
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6126940807133633,
      "learning_rate": 1.2496161590703844e-05,
      "loss": 0.5779,
      "step": 2902
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6402735895855088,
      "learning_rate": 1.249143599840843e-05,
      "loss": 0.551,
      "step": 2903
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7807826078920705,
      "learning_rate": 1.2486709812849659e-05,
      "loss": 0.5722,
      "step": 2904
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6146053888920715,
      "learning_rate": 1.248198303515293e-05,
      "loss": 0.5583,
      "step": 2905
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6082142767219552,
      "learning_rate": 1.2477255666443793e-05,
      "loss": 0.591,
      "step": 2906
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5453342328286045,
      "learning_rate": 1.2472527707847926e-05,
      "loss": 0.5777,
      "step": 2907
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6828922013471965,
      "learning_rate": 1.2467799160491165e-05,
      "loss": 0.5758,
      "step": 2908
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6188811358611422,
      "learning_rate": 1.246307002549947e-05,
      "loss": 0.5873,
      "step": 2909
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6957728634671274,
      "learning_rate": 1.2458340303998954e-05,
      "loss": 0.5862,
      "step": 2910
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6518409778433197,
      "learning_rate": 1.2453609997115856e-05,
      "loss": 0.5991,
      "step": 2911
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7210668557533474,
      "learning_rate": 1.2448879105976567e-05,
      "loss": 0.614,
      "step": 2912
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.64738895485026,
      "learning_rate": 1.2444147631707606e-05,
      "loss": 0.5601,
      "step": 2913
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7453334341783014,
      "learning_rate": 1.2439415575435647e-05,
      "loss": 0.5726,
      "step": 2914
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6212137782222545,
      "learning_rate": 1.2434682938287478e-05,
      "loss": 0.5806,
      "step": 2915
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6340932141395459,
      "learning_rate": 1.2429949721390053e-05,
      "loss": 0.608,
      "step": 2916
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6388566261109111,
      "learning_rate": 1.2425215925870439e-05,
      "loss": 0.5956,
      "step": 2917
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5960472798510533,
      "learning_rate": 1.2420481552855862e-05,
      "loss": 0.5495,
      "step": 2918
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6145640784218825,
      "learning_rate": 1.2415746603473673e-05,
      "loss": 0.581,
      "step": 2919
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6473211145419321,
      "learning_rate": 1.2411011078851361e-05,
      "loss": 0.6017,
      "step": 2920
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6304987963452632,
      "learning_rate": 1.2406274980116552e-05,
      "loss": 0.6162,
      "step": 2921
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6416946284953098,
      "learning_rate": 1.2401538308397019e-05,
      "loss": 0.5942,
      "step": 2922
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5448189435272807,
      "learning_rate": 1.2396801064820654e-05,
      "loss": 0.593,
      "step": 2923
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6344613525661139,
      "learning_rate": 1.23920632505155e-05,
      "loss": 0.5853,
      "step": 2924
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6089921223826487,
      "learning_rate": 1.2387324866609732e-05,
      "loss": 0.5982,
      "step": 2925
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6859768430233338,
      "learning_rate": 1.238258591423165e-05,
      "loss": 0.6183,
      "step": 2926
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6526715789313926,
      "learning_rate": 1.237784639450971e-05,
      "loss": 0.6081,
      "step": 2927
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6112195347915172,
      "learning_rate": 1.2373106308572483e-05,
      "loss": 0.5854,
      "step": 2928
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6229956937731403,
      "learning_rate": 1.2368365657548686e-05,
      "loss": 0.5865,
      "step": 2929
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6159722794177116,
      "learning_rate": 1.2363624442567167e-05,
      "loss": 0.5658,
      "step": 2930
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6497214812272757,
      "learning_rate": 1.235888266475691e-05,
      "loss": 0.6036,
      "step": 2931
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6745521346887,
      "learning_rate": 1.2354140325247033e-05,
      "loss": 0.5721,
      "step": 2932
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6724128194645553,
      "learning_rate": 1.2349397425166788e-05,
      "loss": 0.6269,
      "step": 2933
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.618469621042292,
      "learning_rate": 1.2344653965645553e-05,
      "loss": 0.5995,
      "step": 2934
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.582182119841888,
      "learning_rate": 1.2339909947812851e-05,
      "loss": 0.5988,
      "step": 2935
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5570270916527433,
      "learning_rate": 1.233516537279833e-05,
      "loss": 0.5808,
      "step": 2936
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.638032283059082,
      "learning_rate": 1.2330420241731778e-05,
      "loss": 0.5846,
      "step": 2937
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6429854152906733,
      "learning_rate": 1.2325674555743106e-05,
      "loss": 0.5921,
      "step": 2938
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7177257220577946,
      "learning_rate": 1.2320928315962362e-05,
      "loss": 0.5838,
      "step": 2939
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7410935254514553,
      "learning_rate": 1.2316181523519725e-05,
      "loss": 0.6132,
      "step": 2940
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5966936031418065,
      "learning_rate": 1.231143417954551e-05,
      "loss": 0.6086,
      "step": 2941
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6672640810875103,
      "learning_rate": 1.2306686285170156e-05,
      "loss": 0.5824,
      "step": 2942
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7135535788455951,
      "learning_rate": 1.230193784152424e-05,
      "loss": 0.5837,
      "step": 2943
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6148420306553957,
      "learning_rate": 1.2297188849738462e-05,
      "loss": 0.5913,
      "step": 2944
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6987802780724421,
      "learning_rate": 1.2292439310943658e-05,
      "loss": 0.6045,
      "step": 2945
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6425944693145421,
      "learning_rate": 1.2287689226270794e-05,
      "loss": 0.5933,
      "step": 2946
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6004968150666596,
      "learning_rate": 1.2282938596850968e-05,
      "loss": 0.5777,
      "step": 2947
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7009925679441407,
      "learning_rate": 1.2278187423815402e-05,
      "loss": 0.6058,
      "step": 2948
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6311898400132977,
      "learning_rate": 1.2273435708295451e-05,
      "loss": 0.5727,
      "step": 2949
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6394841898238813,
      "learning_rate": 1.2268683451422596e-05,
      "loss": 0.5775,
      "step": 2950
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1203527015625168,
      "learning_rate": 1.2263930654328452e-05,
      "loss": 0.6019,
      "step": 2951
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6593269436799455,
      "learning_rate": 1.2259177318144762e-05,
      "loss": 0.6079,
      "step": 2952
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7266514862117212,
      "learning_rate": 1.2254423444003387e-05,
      "loss": 0.5923,
      "step": 2953
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5796779657203601,
      "learning_rate": 1.2249669033036336e-05,
      "loss": 0.5711,
      "step": 2954
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6443234535787599,
      "learning_rate": 1.2244914086375726e-05,
      "loss": 0.6011,
      "step": 2955
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6119946454525058,
      "learning_rate": 1.2240158605153814e-05,
      "loss": 0.6144,
      "step": 2956
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5865239602715347,
      "learning_rate": 1.223540259050298e-05,
      "loss": 0.5642,
      "step": 2957
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6704395592696327,
      "learning_rate": 1.2230646043555729e-05,
      "loss": 0.5941,
      "step": 2958
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6208655949800149,
      "learning_rate": 1.2225888965444694e-05,
      "loss": 0.5805,
      "step": 2959
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6207507743850033,
      "learning_rate": 1.2221131357302643e-05,
      "loss": 0.5903,
      "step": 2960
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6800226862770069,
      "learning_rate": 1.2216373220262453e-05,
      "loss": 0.5654,
      "step": 2961
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5845299531596593,
      "learning_rate": 1.221161455545714e-05,
      "loss": 0.5866,
      "step": 2962
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6406564056554386,
      "learning_rate": 1.2206855364019845e-05,
      "loss": 0.594,
      "step": 2963
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6169254113780113,
      "learning_rate": 1.220209564708383e-05,
      "loss": 0.5951,
      "step": 2964
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5465004148978793,
      "learning_rate": 1.219733540578248e-05,
      "loss": 0.577,
      "step": 2965
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6716714725783118,
      "learning_rate": 1.2192574641249318e-05,
      "loss": 0.5655,
      "step": 2966
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5700993749197709,
      "learning_rate": 1.2187813354617973e-05,
      "loss": 0.5792,
      "step": 2967
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6844168164605169,
      "learning_rate": 1.2183051547022212e-05,
      "loss": 0.6052,
      "step": 2968
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6699116250782566,
      "learning_rate": 1.2178289219595917e-05,
      "loss": 0.5765,
      "step": 2969
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5929700690498657,
      "learning_rate": 1.2173526373473105e-05,
      "loss": 0.567,
      "step": 2970
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6756830007308517,
      "learning_rate": 1.216876300978791e-05,
      "loss": 0.5806,
      "step": 2971
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6809347285107139,
      "learning_rate": 1.2163999129674583e-05,
      "loss": 0.6182,
      "step": 2972
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5499577283442488,
      "learning_rate": 1.2159234734267504e-05,
      "loss": 0.5936,
      "step": 2973
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6709967084452046,
      "learning_rate": 1.2154469824701185e-05,
      "loss": 0.6067,
      "step": 2974
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6015312737608575,
      "learning_rate": 1.2149704402110243e-05,
      "loss": 0.6216,
      "step": 2975
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6776538055573677,
      "learning_rate": 1.2144938467629426e-05,
      "loss": 0.5966,
      "step": 2976
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5873305217509949,
      "learning_rate": 1.2140172022393608e-05,
      "loss": 0.5882,
      "step": 2977
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6350537385433701,
      "learning_rate": 1.2135405067537778e-05,
      "loss": 0.6013,
      "step": 2978
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6008980378496281,
      "learning_rate": 1.2130637604197043e-05,
      "loss": 0.5857,
      "step": 2979
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6141280124238995,
      "learning_rate": 1.2125869633506643e-05,
      "loss": 0.6008,
      "step": 2980
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6134572539034259,
      "learning_rate": 1.2121101156601932e-05,
      "loss": 0.5725,
      "step": 2981
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6610317647092222,
      "learning_rate": 1.2116332174618378e-05,
      "loss": 0.594,
      "step": 2982
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5928826878074431,
      "learning_rate": 1.2111562688691587e-05,
      "loss": 0.5828,
      "step": 2983
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7896928433181061,
      "learning_rate": 1.2106792699957264e-05,
      "loss": 0.6018,
      "step": 2984
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5888185008557024,
      "learning_rate": 1.2102022209551249e-05,
      "loss": 0.593,
      "step": 2985
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6251367804626301,
      "learning_rate": 1.2097251218609494e-05,
      "loss": 0.5891,
      "step": 2986
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5996458406197904,
      "learning_rate": 1.2092479728268073e-05,
      "loss": 0.5828,
      "step": 2987
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6352427607451615,
      "learning_rate": 1.2087707739663177e-05,
      "loss": 0.5988,
      "step": 2988
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5660714850433015,
      "learning_rate": 1.2082935253931121e-05,
      "loss": 0.6132,
      "step": 2989
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5783652337163259,
      "learning_rate": 1.2078162272208332e-05,
      "loss": 0.5768,
      "step": 2990
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7013069103205473,
      "learning_rate": 1.2073388795631355e-05,
      "loss": 0.5965,
      "step": 2991
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.55904991602892,
      "learning_rate": 1.2068614825336856e-05,
      "loss": 0.5895,
      "step": 2992
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6477764160440307,
      "learning_rate": 1.2063840362461621e-05,
      "loss": 0.5913,
      "step": 2993
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6269260096623375,
      "learning_rate": 1.205906540814255e-05,
      "loss": 0.5711,
      "step": 2994
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7579651467828253,
      "learning_rate": 1.2054289963516656e-05,
      "loss": 0.5827,
      "step": 2995
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6495155088303837,
      "learning_rate": 1.2049514029721077e-05,
      "loss": 0.596,
      "step": 2996
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5776556827460032,
      "learning_rate": 1.204473760789306e-05,
      "loss": 0.5883,
      "step": 2997
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6243512936703092,
      "learning_rate": 1.2039960699169972e-05,
      "loss": 0.5943,
      "step": 2998
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5458105243551886,
      "learning_rate": 1.2035183304689303e-05,
      "loss": 0.571,
      "step": 2999
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6239332633177068,
      "learning_rate": 1.2030405425588638e-05,
      "loss": 0.5823,
      "step": 3000
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6296579201639557,
      "learning_rate": 1.2025627063005703e-05,
      "loss": 0.5746,
      "step": 3001
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.625270226107122,
      "learning_rate": 1.2020848218078315e-05,
      "loss": 0.5822,
      "step": 3002
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5918436826570685,
      "learning_rate": 1.201606889194443e-05,
      "loss": 0.5907,
      "step": 3003
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6494394769958656,
      "learning_rate": 1.2011289085742099e-05,
      "loss": 0.6177,
      "step": 3004
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8570348283066984,
      "learning_rate": 1.2006508800609495e-05,
      "loss": 0.5855,
      "step": 3005
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.584040158743361,
      "learning_rate": 1.2001728037684903e-05,
      "loss": 0.5868,
      "step": 3006
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6417324731111731,
      "learning_rate": 1.1996946798106728e-05,
      "loss": 0.5853,
      "step": 3007
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.631201574276749,
      "learning_rate": 1.199216508301348e-05,
      "loss": 0.5791,
      "step": 3008
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6101233592490096,
      "learning_rate": 1.1987382893543786e-05,
      "loss": 0.6042,
      "step": 3009
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5894970335330688,
      "learning_rate": 1.198260023083639e-05,
      "loss": 0.5841,
      "step": 3010
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4763081670990779,
      "learning_rate": 1.1977817096030138e-05,
      "loss": 0.6064,
      "step": 3011
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6694415494366353,
      "learning_rate": 1.1973033490264e-05,
      "loss": 0.6262,
      "step": 3012
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5885363446408192,
      "learning_rate": 1.1968249414677055e-05,
      "loss": 0.6009,
      "step": 3013
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6858975051189713,
      "learning_rate": 1.196346487040849e-05,
      "loss": 0.6086,
      "step": 3014
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5910317220327039,
      "learning_rate": 1.1958679858597599e-05,
      "loss": 0.5401,
      "step": 3015
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6789738492278697,
      "learning_rate": 1.1953894380383805e-05,
      "loss": 0.5859,
      "step": 3016
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.612271044438309,
      "learning_rate": 1.1949108436906625e-05,
      "loss": 0.582,
      "step": 3017
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6339131810841488,
      "learning_rate": 1.1944322029305697e-05,
      "loss": 0.5801,
      "step": 3018
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6879419340056145,
      "learning_rate": 1.193953515872076e-05,
      "loss": 0.5789,
      "step": 3019
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5950827481760903,
      "learning_rate": 1.193474782629167e-05,
      "loss": 0.595,
      "step": 3020
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6175428459996379,
      "learning_rate": 1.1929960033158392e-05,
      "loss": 0.6064,
      "step": 3021
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5792465456409237,
      "learning_rate": 1.1925171780461004e-05,
      "loss": 0.577,
      "step": 3022
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6793462509741789,
      "learning_rate": 1.1920383069339684e-05,
      "loss": 0.5829,
      "step": 3023
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6245956560668406,
      "learning_rate": 1.191559390093473e-05,
      "loss": 0.6018,
      "step": 3024
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6225263332985561,
      "learning_rate": 1.1910804276386541e-05,
      "loss": 0.5742,
      "step": 3025
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8032213244764893,
      "learning_rate": 1.190601419683563e-05,
      "loss": 0.5925,
      "step": 3026
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6429473570194856,
      "learning_rate": 1.1901223663422611e-05,
      "loss": 0.5819,
      "step": 3027
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6390513076722435,
      "learning_rate": 1.1896432677288215e-05,
      "loss": 0.5958,
      "step": 3028
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5945418782150862,
      "learning_rate": 1.1891641239573273e-05,
      "loss": 0.589,
      "step": 3029
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6264317385572279,
      "learning_rate": 1.1886849351418732e-05,
      "loss": 0.584,
      "step": 3030
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6137721664000398,
      "learning_rate": 1.1882057013965637e-05,
      "loss": 0.6011,
      "step": 3031
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6593278972003308,
      "learning_rate": 1.187726422835515e-05,
      "loss": 0.5855,
      "step": 3032
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6203856366568737,
      "learning_rate": 1.1872470995728529e-05,
      "loss": 0.5635,
      "step": 3033
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6990567349676515,
      "learning_rate": 1.1867677317227144e-05,
      "loss": 0.6256,
      "step": 3034
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6374066327145121,
      "learning_rate": 1.1862883193992471e-05,
      "loss": 0.5724,
      "step": 3035
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5745816821707768,
      "learning_rate": 1.1858088627166096e-05,
      "loss": 0.5922,
      "step": 3036
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6834429160895605,
      "learning_rate": 1.1853293617889701e-05,
      "loss": 0.618,
      "step": 3037
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6061541391986749,
      "learning_rate": 1.1848498167305078e-05,
      "loss": 0.5933,
      "step": 3038
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5746883916962078,
      "learning_rate": 1.1843702276554131e-05,
      "loss": 0.5671,
      "step": 3039
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6738741862209696,
      "learning_rate": 1.1838905946778856e-05,
      "loss": 0.5732,
      "step": 3040
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6688021535179505,
      "learning_rate": 1.1834109179121367e-05,
      "loss": 0.5908,
      "step": 3041
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5838999151143166,
      "learning_rate": 1.1829311974723868e-05,
      "loss": 0.5955,
      "step": 3042
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7152895207055072,
      "learning_rate": 1.1824514334728678e-05,
      "loss": 0.5749,
      "step": 3043
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.718767218416572,
      "learning_rate": 1.1819716260278215e-05,
      "loss": 0.6255,
      "step": 3044
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5696569516940512,
      "learning_rate": 1.1814917752515005e-05,
      "loss": 0.5952,
      "step": 3045
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7488433452515051,
      "learning_rate": 1.1810118812581671e-05,
      "loss": 0.6162,
      "step": 3046
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7013429802833988,
      "learning_rate": 1.180531944162094e-05,
      "loss": 0.5885,
      "step": 3047
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6578780660557414,
      "learning_rate": 1.1800519640775642e-05,
      "loss": 0.6042,
      "step": 3048
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5782340714050399,
      "learning_rate": 1.1795719411188717e-05,
      "loss": 0.5616,
      "step": 3049
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5885415250580095,
      "learning_rate": 1.1790918754003195e-05,
      "loss": 0.5916,
      "step": 3050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6401212003016632,
      "learning_rate": 1.178611767036222e-05,
      "loss": 0.6064,
      "step": 3051
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7192317502353542,
      "learning_rate": 1.1781316161409024e-05,
      "loss": 0.6015,
      "step": 3052
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.63603339501862,
      "learning_rate": 1.1776514228286951e-05,
      "loss": 0.6074,
      "step": 3053
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6724789674599334,
      "learning_rate": 1.177171187213944e-05,
      "loss": 0.6009,
      "step": 3054
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6671409025695557,
      "learning_rate": 1.1766909094110036e-05,
      "loss": 0.5717,
      "step": 3055
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6357168750479433,
      "learning_rate": 1.176210589534238e-05,
      "loss": 0.585,
      "step": 3056
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6795523391486133,
      "learning_rate": 1.1757302276980213e-05,
      "loss": 0.593,
      "step": 3057
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5865908348703853,
      "learning_rate": 1.175249824016738e-05,
      "loss": 0.5838,
      "step": 3058
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6605769799939855,
      "learning_rate": 1.1747693786047827e-05,
      "loss": 0.6001,
      "step": 3059
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6734686582245314,
      "learning_rate": 1.174288891576559e-05,
      "loss": 0.5933,
      "step": 3060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.938740368254021,
      "learning_rate": 1.1738083630464807e-05,
      "loss": 0.6426,
      "step": 3061
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5916357023246503,
      "learning_rate": 1.1733277931289726e-05,
      "loss": 0.5877,
      "step": 3062
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5505918920608474,
      "learning_rate": 1.1728471819384679e-05,
      "loss": 0.5828,
      "step": 3063
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6839689929984891,
      "learning_rate": 1.1723665295894104e-05,
      "loss": 0.581,
      "step": 3064
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6567366076491723,
      "learning_rate": 1.171885836196254e-05,
      "loss": 0.6216,
      "step": 3065
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6298802337568211,
      "learning_rate": 1.1714051018734612e-05,
      "loss": 0.5688,
      "step": 3066
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6092247258694448,
      "learning_rate": 1.170924326735505e-05,
      "loss": 0.5736,
      "step": 3067
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5932064699303442,
      "learning_rate": 1.1704435108968688e-05,
      "loss": 0.5809,
      "step": 3068
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6155259352669759,
      "learning_rate": 1.1699626544720446e-05,
      "loss": 0.5986,
      "step": 3069
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6439217124422393,
      "learning_rate": 1.1694817575755342e-05,
      "loss": 0.6042,
      "step": 3070
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6361269185472741,
      "learning_rate": 1.1690008203218493e-05,
      "loss": 0.575,
      "step": 3071
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6599785989880699,
      "learning_rate": 1.1685198428255114e-05,
      "loss": 0.6029,
      "step": 3072
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.656001800621621,
      "learning_rate": 1.1680388252010511e-05,
      "loss": 0.5879,
      "step": 3073
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6684936196719119,
      "learning_rate": 1.1675577675630093e-05,
      "loss": 0.5844,
      "step": 3074
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6157244408919245,
      "learning_rate": 1.1670766700259355e-05,
      "loss": 0.595,
      "step": 3075
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6071185633917827,
      "learning_rate": 1.1665955327043897e-05,
      "loss": 0.5557,
      "step": 3076
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6743687094624382,
      "learning_rate": 1.1661143557129397e-05,
      "loss": 0.6087,
      "step": 3077
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6893488929822356,
      "learning_rate": 1.1656331391661652e-05,
      "loss": 0.6064,
      "step": 3078
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6137161500939102,
      "learning_rate": 1.1651518831786533e-05,
      "loss": 0.5637,
      "step": 3079
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6421378390881723,
      "learning_rate": 1.1646705878650012e-05,
      "loss": 0.5831,
      "step": 3080
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6152558766952143,
      "learning_rate": 1.1641892533398156e-05,
      "loss": 0.5835,
      "step": 3081
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6341770234721512,
      "learning_rate": 1.1637078797177122e-05,
      "loss": 0.5558,
      "step": 3082
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6988502171295082,
      "learning_rate": 1.1632264671133163e-05,
      "loss": 0.5702,
      "step": 3083
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6143544307689863,
      "learning_rate": 1.1627450156412628e-05,
      "loss": 0.5654,
      "step": 3084
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6607437419309796,
      "learning_rate": 1.1622635254161945e-05,
      "loss": 0.58,
      "step": 3085
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6357446415334488,
      "learning_rate": 1.161781996552765e-05,
      "loss": 0.6191,
      "step": 3086
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6172796452812391,
      "learning_rate": 1.1613004291656362e-05,
      "loss": 0.5713,
      "step": 3087
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6338317279427671,
      "learning_rate": 1.1608188233694797e-05,
      "loss": 0.5928,
      "step": 3088
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6443505908590823,
      "learning_rate": 1.1603371792789759e-05,
      "loss": 0.5867,
      "step": 3089
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6384316887358927,
      "learning_rate": 1.1598554970088142e-05,
      "loss": 0.5952,
      "step": 3090
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6097373148673344,
      "learning_rate": 1.1593737766736936e-05,
      "loss": 0.5759,
      "step": 3091
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6462323961347678,
      "learning_rate": 1.1588920183883216e-05,
      "loss": 0.599,
      "step": 3092
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6226842979662855,
      "learning_rate": 1.1584102222674152e-05,
      "loss": 0.6041,
      "step": 3093
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6222767961938076,
      "learning_rate": 1.1579283884256997e-05,
      "loss": 0.5933,
      "step": 3094
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.614136618450228,
      "learning_rate": 1.1574465169779106e-05,
      "loss": 0.6031,
      "step": 3095
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6240058552050145,
      "learning_rate": 1.156964608038791e-05,
      "loss": 0.5919,
      "step": 3096
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6079877046012065,
      "learning_rate": 1.1564826617230943e-05,
      "loss": 0.5816,
      "step": 3097
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5799269308909927,
      "learning_rate": 1.1560006781455813e-05,
      "loss": 0.583,
      "step": 3098
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6242898113900891,
      "learning_rate": 1.1555186574210229e-05,
      "loss": 0.6136,
      "step": 3099
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6285413617778196,
      "learning_rate": 1.155036599664198e-05,
      "loss": 0.5968,
      "step": 3100
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6044211107473394,
      "learning_rate": 1.154554504989895e-05,
      "loss": 0.5834,
      "step": 3101
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6741750394284695,
      "learning_rate": 1.154072373512911e-05,
      "loss": 0.5972,
      "step": 3102
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7463660654782285,
      "learning_rate": 1.153590205348051e-05,
      "loss": 0.5965,
      "step": 3103
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.529042353627978,
      "learning_rate": 1.1531080006101298e-05,
      "loss": 0.5859,
      "step": 3104
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5665969135450936,
      "learning_rate": 1.1526257594139706e-05,
      "loss": 0.5749,
      "step": 3105
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6015458960692204,
      "learning_rate": 1.152143481874405e-05,
      "loss": 0.5633,
      "step": 3106
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6663170132736146,
      "learning_rate": 1.1516611681062738e-05,
      "loss": 0.619,
      "step": 3107
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5992823335211755,
      "learning_rate": 1.1511788182244252e-05,
      "loss": 0.5867,
      "step": 3108
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7140301997503615,
      "learning_rate": 1.1506964323437178e-05,
      "loss": 0.5767,
      "step": 3109
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.602815782042564,
      "learning_rate": 1.1502140105790172e-05,
      "loss": 0.5976,
      "step": 3110
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5897462811726688,
      "learning_rate": 1.1497315530451985e-05,
      "loss": 0.5916,
      "step": 3111
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6207458534175396,
      "learning_rate": 1.1492490598571451e-05,
      "loss": 0.5539,
      "step": 3112
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5958029393390093,
      "learning_rate": 1.1487665311297484e-05,
      "loss": 0.5773,
      "step": 3113
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6386506634907909,
      "learning_rate": 1.1482839669779087e-05,
      "loss": 0.6097,
      "step": 3114
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6019757521858486,
      "learning_rate": 1.147801367516535e-05,
      "loss": 0.5655,
      "step": 3115
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7946398332141029,
      "learning_rate": 1.1473187328605444e-05,
      "loss": 0.598,
      "step": 3116
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6215204123228419,
      "learning_rate": 1.1468360631248618e-05,
      "loss": 0.5736,
      "step": 3117
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6687303999847889,
      "learning_rate": 1.1463533584244218e-05,
      "loss": 0.6073,
      "step": 3118
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5856190786811987,
      "learning_rate": 1.1458706188741657e-05,
      "loss": 0.6024,
      "step": 3119
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6946936504114326,
      "learning_rate": 1.145387844589045e-05,
      "loss": 0.6047,
      "step": 3120
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7142487006301276,
      "learning_rate": 1.1449050356840175e-05,
      "loss": 0.6027,
      "step": 3121
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5862764954235161,
      "learning_rate": 1.1444221922740507e-05,
      "loss": 0.6061,
      "step": 3122
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5877397730450278,
      "learning_rate": 1.1439393144741192e-05,
      "loss": 0.5782,
      "step": 3123
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7027107089389333,
      "learning_rate": 1.143456402399207e-05,
      "loss": 0.5949,
      "step": 3124
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6837978631919357,
      "learning_rate": 1.1429734561643053e-05,
      "loss": 0.5635,
      "step": 3125
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6323735036607574,
      "learning_rate": 1.1424904758844141e-05,
      "loss": 0.5894,
      "step": 3126
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6297414953950771,
      "learning_rate": 1.1420074616745407e-05,
      "loss": 0.5788,
      "step": 3127
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6534967618077547,
      "learning_rate": 1.1415244136497012e-05,
      "loss": 0.5758,
      "step": 3128
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7033184981765737,
      "learning_rate": 1.1410413319249193e-05,
      "loss": 0.6051,
      "step": 3129
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6471604200086062,
      "learning_rate": 1.1405582166152276e-05,
      "loss": 0.5932,
      "step": 3130
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6467218779750767,
      "learning_rate": 1.1400750678356652e-05,
      "loss": 0.5788,
      "step": 3131
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6722661975397803,
      "learning_rate": 1.1395918857012803e-05,
      "loss": 0.5964,
      "step": 3132
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6027155815515892,
      "learning_rate": 1.1391086703271285e-05,
      "loss": 0.5947,
      "step": 3133
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6928770865592608,
      "learning_rate": 1.1386254218282744e-05,
      "loss": 0.5698,
      "step": 3134
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7163032580462892,
      "learning_rate": 1.1381421403197888e-05,
      "loss": 0.6027,
      "step": 3135
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6558162006687028,
      "learning_rate": 1.1376588259167516e-05,
      "loss": 0.5883,
      "step": 3136
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6878747959339265,
      "learning_rate": 1.1371754787342497e-05,
      "loss": 0.5871,
      "step": 3137
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7206465854570181,
      "learning_rate": 1.1366920988873787e-05,
      "loss": 0.6091,
      "step": 3138
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6751718850396344,
      "learning_rate": 1.1362086864912411e-05,
      "loss": 0.5792,
      "step": 3139
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7691280674803991,
      "learning_rate": 1.1357252416609482e-05,
      "loss": 0.6205,
      "step": 3140
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6287577680431223,
      "learning_rate": 1.1352417645116178e-05,
      "loss": 0.5994,
      "step": 3141
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7042393414472421,
      "learning_rate": 1.1347582551583764e-05,
      "loss": 0.6015,
      "step": 3142
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6237375516870188,
      "learning_rate": 1.1342747137163571e-05,
      "loss": 0.59,
      "step": 3143
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6521416868047027,
      "learning_rate": 1.1337911403007023e-05,
      "loss": 0.5791,
      "step": 3144
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6260945259014872,
      "learning_rate": 1.1333075350265601e-05,
      "loss": 0.5955,
      "step": 3145
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7154362070927281,
      "learning_rate": 1.1328238980090876e-05,
      "loss": 0.59,
      "step": 3146
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6204079824476125,
      "learning_rate": 1.1323402293634487e-05,
      "loss": 0.5957,
      "step": 3147
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6329780244543864,
      "learning_rate": 1.131856529204815e-05,
      "loss": 0.5904,
      "step": 3148
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6259707645673888,
      "learning_rate": 1.1313727976483666e-05,
      "loss": 0.5921,
      "step": 3149
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6501965418336639,
      "learning_rate": 1.130889034809289e-05,
      "loss": 0.5774,
      "step": 3150
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6505207944269582,
      "learning_rate": 1.1304052408027766e-05,
      "loss": 0.5934,
      "step": 3151
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6216516744693562,
      "learning_rate": 1.1299214157440313e-05,
      "loss": 0.5729,
      "step": 3152
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6713256424667953,
      "learning_rate": 1.129437559748262e-05,
      "loss": 0.5741,
      "step": 3153
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6324742284823326,
      "learning_rate": 1.1289536729306844e-05,
      "loss": 0.5626,
      "step": 3154
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6089276411436282,
      "learning_rate": 1.128469755406523e-05,
      "loss": 0.5752,
      "step": 3155
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.677587623748924,
      "learning_rate": 1.127985807291008e-05,
      "loss": 0.6034,
      "step": 3156
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.663776280520753,
      "learning_rate": 1.1275018286993782e-05,
      "loss": 0.6,
      "step": 3157
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.570483539999573,
      "learning_rate": 1.1270178197468788e-05,
      "loss": 0.563,
      "step": 3158
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5603007040881339,
      "learning_rate": 1.1265337805487628e-05,
      "loss": 0.5704,
      "step": 3159
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6081131339502253,
      "learning_rate": 1.1260497112202895e-05,
      "loss": 0.5889,
      "step": 3160
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6527442621100358,
      "learning_rate": 1.1255656118767266e-05,
      "loss": 0.6112,
      "step": 3161
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.579486636997354,
      "learning_rate": 1.125081482633348e-05,
      "loss": 0.5684,
      "step": 3162
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5706721764362468,
      "learning_rate": 1.1245973236054355e-05,
      "loss": 0.5804,
      "step": 3163
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6662549036576194,
      "learning_rate": 1.124113134908277e-05,
      "loss": 0.5807,
      "step": 3164
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6287274839230863,
      "learning_rate": 1.1236289166571683e-05,
      "loss": 0.5805,
      "step": 3165
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7139001096021598,
      "learning_rate": 1.1231446689674119e-05,
      "loss": 0.6498,
      "step": 3166
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5847548390196584,
      "learning_rate": 1.1226603919543176e-05,
      "loss": 0.5842,
      "step": 3167
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6268558255076215,
      "learning_rate": 1.1221760857332016e-05,
      "loss": 0.5826,
      "step": 3168
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6669460640336452,
      "learning_rate": 1.1216917504193877e-05,
      "loss": 0.5956,
      "step": 3169
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5871263558230565,
      "learning_rate": 1.121207386128206e-05,
      "loss": 0.5965,
      "step": 3170
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6419649409100842,
      "learning_rate": 1.1207229929749944e-05,
      "loss": 0.5996,
      "step": 3171
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5979227302763175,
      "learning_rate": 1.1202385710750962e-05,
      "loss": 0.5902,
      "step": 3172
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.396086860980004,
      "learning_rate": 1.1197541205438634e-05,
      "loss": 0.6112,
      "step": 3173
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5962045310468512,
      "learning_rate": 1.1192696414966533e-05,
      "loss": 0.5938,
      "step": 3174
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6389565381513886,
      "learning_rate": 1.1187851340488308e-05,
      "loss": 0.5721,
      "step": 3175
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6421655673895311,
      "learning_rate": 1.1183005983157672e-05,
      "loss": 0.5672,
      "step": 3176
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7758710409553295,
      "learning_rate": 1.1178160344128409e-05,
      "loss": 0.6162,
      "step": 3177
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6763094187310271,
      "learning_rate": 1.1173314424554365e-05,
      "loss": 0.6159,
      "step": 3178
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6341799963561482,
      "learning_rate": 1.1168468225589452e-05,
      "loss": 0.5871,
      "step": 3179
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.682902821330859,
      "learning_rate": 1.1163621748387663e-05,
      "loss": 0.5823,
      "step": 3180
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6215634613365396,
      "learning_rate": 1.1158774994103035e-05,
      "loss": 0.5776,
      "step": 3181
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6429740358250616,
      "learning_rate": 1.115392796388969e-05,
      "loss": 0.573,
      "step": 3182
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7231254916943314,
      "learning_rate": 1.11490806589018e-05,
      "loss": 0.5901,
      "step": 3183
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6268008978761795,
      "learning_rate": 1.1144233080293619e-05,
      "loss": 0.5685,
      "step": 3184
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6309141267025985,
      "learning_rate": 1.1139385229219451e-05,
      "loss": 0.6063,
      "step": 3185
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.673596543155491,
      "learning_rate": 1.1134537106833673e-05,
      "loss": 0.6198,
      "step": 3186
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6903386910793061,
      "learning_rate": 1.112968871429073e-05,
      "loss": 0.5895,
      "step": 3187
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5783660724861575,
      "learning_rate": 1.1124840052745121e-05,
      "loss": 0.5801,
      "step": 3188
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6372415116355908,
      "learning_rate": 1.1119991123351413e-05,
      "loss": 0.5927,
      "step": 3189
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5994597110573312,
      "learning_rate": 1.1115141927264244e-05,
      "loss": 0.569,
      "step": 3190
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6977311289484817,
      "learning_rate": 1.1110292465638301e-05,
      "loss": 0.6027,
      "step": 3191
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.659348083924995,
      "learning_rate": 1.1105442739628357e-05,
      "loss": 0.6029,
      "step": 3192
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6295475163603829,
      "learning_rate": 1.1100592750389219e-05,
      "loss": 0.5914,
      "step": 3193
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5703488821831817,
      "learning_rate": 1.109574249907578e-05,
      "loss": 0.5794,
      "step": 3194
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6544098201040771,
      "learning_rate": 1.1090891986842982e-05,
      "loss": 0.5978,
      "step": 3195
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5451902226224858,
      "learning_rate": 1.108604121484584e-05,
      "loss": 0.5636,
      "step": 3196
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6681769302996253,
      "learning_rate": 1.1081190184239418e-05,
      "loss": 0.618,
      "step": 3197
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.55546721735155,
      "learning_rate": 1.1076338896178858e-05,
      "loss": 0.5853,
      "step": 3198
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6088048663970226,
      "learning_rate": 1.1071487351819341e-05,
      "loss": 0.5854,
      "step": 3199
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5794220285374725,
      "learning_rate": 1.1066635552316133e-05,
      "loss": 0.583,
      "step": 3200
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5618043995414983,
      "learning_rate": 1.1061783498824545e-05,
      "loss": 0.5687,
      "step": 3201
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6640155201990188,
      "learning_rate": 1.1056931192499954e-05,
      "loss": 0.603,
      "step": 3202
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6442823124295779,
      "learning_rate": 1.1052078634497795e-05,
      "loss": 0.6123,
      "step": 3203
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6881228808078084,
      "learning_rate": 1.1047225825973565e-05,
      "loss": 0.5873,
      "step": 3204
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6369352635657938,
      "learning_rate": 1.104237276808282e-05,
      "loss": 0.5985,
      "step": 3205
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.639529821460624,
      "learning_rate": 1.1037519461981176e-05,
      "loss": 0.5808,
      "step": 3206
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5983194224337366,
      "learning_rate": 1.1032665908824307e-05,
      "loss": 0.5964,
      "step": 3207
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6777113556706627,
      "learning_rate": 1.1027812109767944e-05,
      "loss": 0.6,
      "step": 3208
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5904800670211903,
      "learning_rate": 1.1022958065967879e-05,
      "loss": 0.5883,
      "step": 3209
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6574954951221262,
      "learning_rate": 1.1018103778579966e-05,
      "loss": 0.6023,
      "step": 3210
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6531256788287961,
      "learning_rate": 1.1013249248760111e-05,
      "loss": 0.6033,
      "step": 3211
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6604163729668753,
      "learning_rate": 1.1008394477664278e-05,
      "loss": 0.593,
      "step": 3212
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6564417692415082,
      "learning_rate": 1.1003539466448492e-05,
      "loss": 0.5778,
      "step": 3213
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6211678886870303,
      "learning_rate": 1.0998684216268831e-05,
      "loss": 0.5674,
      "step": 3214
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6087112840913976,
      "learning_rate": 1.099382872828144e-05,
      "loss": 0.5751,
      "step": 3215
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6280443748861884,
      "learning_rate": 1.09889730036425e-05,
      "loss": 0.5786,
      "step": 3216
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6063221674461767,
      "learning_rate": 1.0984117043508271e-05,
      "loss": 0.5852,
      "step": 3217
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.586769990325059,
      "learning_rate": 1.0979260849035054e-05,
      "loss": 0.5955,
      "step": 3218
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6484238792810143,
      "learning_rate": 1.0974404421379217e-05,
      "loss": 0.5849,
      "step": 3219
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6036141444784151,
      "learning_rate": 1.0969547761697174e-05,
      "loss": 0.5852,
      "step": 3220
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6360427549966806,
      "learning_rate": 1.0964690871145398e-05,
      "loss": 0.6035,
      "step": 3221
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6765035191212322,
      "learning_rate": 1.0959833750880415e-05,
      "loss": 0.5909,
      "step": 3222
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5921608443685759,
      "learning_rate": 1.0954976402058812e-05,
      "loss": 0.5743,
      "step": 3223
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6362234893572994,
      "learning_rate": 1.0950118825837226e-05,
      "loss": 0.5644,
      "step": 3224
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6310316347381144,
      "learning_rate": 1.094526102337234e-05,
      "loss": 0.583,
      "step": 3225
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6152248477487875,
      "learning_rate": 1.0940402995820908e-05,
      "loss": 0.5893,
      "step": 3226
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6288623096279056,
      "learning_rate": 1.0935544744339727e-05,
      "loss": 0.578,
      "step": 3227
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6431655068600521,
      "learning_rate": 1.0930686270085646e-05,
      "loss": 0.6118,
      "step": 3228
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6060490659163955,
      "learning_rate": 1.0925827574215573e-05,
      "loss": 0.5656,
      "step": 3229
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6886577759529474,
      "learning_rate": 1.0920968657886463e-05,
      "loss": 0.5589,
      "step": 3230
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6258016832817868,
      "learning_rate": 1.0916109522255326e-05,
      "loss": 0.5893,
      "step": 3231
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6858829562622687,
      "learning_rate": 1.091125016847923e-05,
      "loss": 0.5925,
      "step": 3232
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6146474940911149,
      "learning_rate": 1.0906390597715281e-05,
      "loss": 0.5888,
      "step": 3233
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6099227426384924,
      "learning_rate": 1.0901530811120655e-05,
      "loss": 0.5629,
      "step": 3234
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.662189875432919,
      "learning_rate": 1.0896670809852558e-05,
      "loss": 0.5793,
      "step": 3235
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5978585517260554,
      "learning_rate": 1.0891810595068271e-05,
      "loss": 0.5913,
      "step": 3236
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6186695635908208,
      "learning_rate": 1.0886950167925101e-05,
      "loss": 0.59,
      "step": 3237
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9107667866044631,
      "learning_rate": 1.0882089529580428e-05,
      "loss": 0.5983,
      "step": 3238
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6392505014659099,
      "learning_rate": 1.0877228681191667e-05,
      "loss": 0.5942,
      "step": 3239
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6053605983911458,
      "learning_rate": 1.0872367623916291e-05,
      "loss": 0.5792,
      "step": 3240
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6089732524915779,
      "learning_rate": 1.0867506358911818e-05,
      "loss": 0.5831,
      "step": 3241
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6027861555681102,
      "learning_rate": 1.0862644887335819e-05,
      "loss": 0.5836,
      "step": 3242
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6354851637014294,
      "learning_rate": 1.0857783210345914e-05,
      "loss": 0.5892,
      "step": 3243
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6083808810501127,
      "learning_rate": 1.0852921329099766e-05,
      "loss": 0.5778,
      "step": 3244
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5812819908806137,
      "learning_rate": 1.0848059244755093e-05,
      "loss": 0.5873,
      "step": 3245
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6538580922987096,
      "learning_rate": 1.0843196958469665e-05,
      "loss": 0.5695,
      "step": 3246
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.690604941005413,
      "learning_rate": 1.0838334471401285e-05,
      "loss": 0.6152,
      "step": 3247
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6148089997267089,
      "learning_rate": 1.0833471784707825e-05,
      "loss": 0.584,
      "step": 3248
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.655736441433198,
      "learning_rate": 1.0828608899547181e-05,
      "loss": 0.5745,
      "step": 3249
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6961801812919118,
      "learning_rate": 1.0823745817077318e-05,
      "loss": 0.5572,
      "step": 3250
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6300632050122518,
      "learning_rate": 1.0818882538456232e-05,
      "loss": 0.6029,
      "step": 3251
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6243358529000196,
      "learning_rate": 1.0814019064841979e-05,
      "loss": 0.5724,
      "step": 3252
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7151882311562908,
      "learning_rate": 1.0809155397392648e-05,
      "loss": 0.6193,
      "step": 3253
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6892665362628402,
      "learning_rate": 1.0804291537266381e-05,
      "loss": 0.596,
      "step": 3254
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.624544293674445,
      "learning_rate": 1.079942748562137e-05,
      "loss": 0.5755,
      "step": 3255
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.607809009302885,
      "learning_rate": 1.0794563243615843e-05,
      "loss": 0.5567,
      "step": 3256
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6191973201220076,
      "learning_rate": 1.0789698812408082e-05,
      "loss": 0.5533,
      "step": 3257
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5945048010735878,
      "learning_rate": 1.0784834193156408e-05,
      "loss": 0.5706,
      "step": 3258
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6803014893484866,
      "learning_rate": 1.0779969387019193e-05,
      "loss": 0.5988,
      "step": 3259
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5246865559848268,
      "learning_rate": 1.0775104395154845e-05,
      "loss": 0.5447,
      "step": 3260
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5930125522526913,
      "learning_rate": 1.0770239218721822e-05,
      "loss": 0.5709,
      "step": 3261
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.649719376737111,
      "learning_rate": 1.0765373858878632e-05,
      "loss": 0.5769,
      "step": 3262
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6549821684801727,
      "learning_rate": 1.0760508316783809e-05,
      "loss": 0.6104,
      "step": 3263
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6138606694005204,
      "learning_rate": 1.0755642593595945e-05,
      "loss": 0.5922,
      "step": 3264
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7059816675428097,
      "learning_rate": 1.0750776690473674e-05,
      "loss": 0.6297,
      "step": 3265
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6624599200446826,
      "learning_rate": 1.0745910608575667e-05,
      "loss": 0.5957,
      "step": 3266
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6229615284739678,
      "learning_rate": 1.0741044349060646e-05,
      "loss": 0.5703,
      "step": 3267
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6139123701813587,
      "learning_rate": 1.0736177913087358e-05,
      "loss": 0.5988,
      "step": 3268
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5978350112433329,
      "learning_rate": 1.0731311301814616e-05,
      "loss": 0.5885,
      "step": 3269
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6477847693807918,
      "learning_rate": 1.0726444516401253e-05,
      "loss": 0.5899,
      "step": 3270
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6093570038905263,
      "learning_rate": 1.0721577558006164e-05,
      "loss": 0.6045,
      "step": 3271
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6275198414255001,
      "learning_rate": 1.0716710427788263e-05,
      "loss": 0.59,
      "step": 3272
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6265532399900561,
      "learning_rate": 1.0711843126906522e-05,
      "loss": 0.5842,
      "step": 3273
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5493812560610531,
      "learning_rate": 1.0706975656519946e-05,
      "loss": 0.5603,
      "step": 3274
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6567540731337133,
      "learning_rate": 1.0702108017787587e-05,
      "loss": 0.5865,
      "step": 3275
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.633337948948849,
      "learning_rate": 1.0697240211868527e-05,
      "loss": 0.5677,
      "step": 3276
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5911862873326197,
      "learning_rate": 1.0692372239921894e-05,
      "loss": 0.5512,
      "step": 3277
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6693765923747725,
      "learning_rate": 1.0687504103106854e-05,
      "loss": 0.5923,
      "step": 3278
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6003960835046847,
      "learning_rate": 1.0682635802582616e-05,
      "loss": 0.5765,
      "step": 3279
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6632291801561575,
      "learning_rate": 1.0677767339508418e-05,
      "loss": 0.6015,
      "step": 3280
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6972362995678417,
      "learning_rate": 1.067289871504355e-05,
      "loss": 0.5857,
      "step": 3281
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6082671997699317,
      "learning_rate": 1.0668029930347336e-05,
      "loss": 0.6055,
      "step": 3282
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.759411212590372,
      "learning_rate": 1.0663160986579129e-05,
      "loss": 0.639,
      "step": 3283
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6042862614274812,
      "learning_rate": 1.0658291884898333e-05,
      "loss": 0.557,
      "step": 3284
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6458655314879459,
      "learning_rate": 1.065342262646438e-05,
      "loss": 0.5767,
      "step": 3285
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.594198225028437,
      "learning_rate": 1.0648553212436746e-05,
      "loss": 0.5826,
      "step": 3286
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6094159206778349,
      "learning_rate": 1.0643683643974934e-05,
      "loss": 0.5854,
      "step": 3287
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6799400058723186,
      "learning_rate": 1.0638813922238498e-05,
      "loss": 0.5751,
      "step": 3288
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6628303947102959,
      "learning_rate": 1.0633944048387017e-05,
      "loss": 0.5981,
      "step": 3289
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6822415381665454,
      "learning_rate": 1.062907402358012e-05,
      "loss": 0.5823,
      "step": 3290
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6828648042849819,
      "learning_rate": 1.0624203848977446e-05,
      "loss": 0.5801,
      "step": 3291
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7146735066689485,
      "learning_rate": 1.06193335257387e-05,
      "loss": 0.567,
      "step": 3292
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6229154324777045,
      "learning_rate": 1.0614463055023601e-05,
      "loss": 0.5812,
      "step": 3293
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5872923091452813,
      "learning_rate": 1.0609592437991913e-05,
      "loss": 0.5892,
      "step": 3294
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6881905086224053,
      "learning_rate": 1.0604721675803436e-05,
      "loss": 0.5804,
      "step": 3295
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6324181987821196,
      "learning_rate": 1.0599850769617996e-05,
      "loss": 0.616,
      "step": 3296
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7011120928961999,
      "learning_rate": 1.0594979720595457e-05,
      "loss": 0.6245,
      "step": 3297
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6351465622870203,
      "learning_rate": 1.0590108529895723e-05,
      "loss": 0.5624,
      "step": 3298
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6333868112903797,
      "learning_rate": 1.0585237198678724e-05,
      "loss": 0.5731,
      "step": 3299
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5843347628254649,
      "learning_rate": 1.0580365728104433e-05,
      "loss": 0.583,
      "step": 3300
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5849138324973119,
      "learning_rate": 1.0575494119332838e-05,
      "loss": 0.5733,
      "step": 3301
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7219009539325167,
      "learning_rate": 1.0570622373523983e-05,
      "loss": 0.5957,
      "step": 3302
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.598326374953446,
      "learning_rate": 1.0565750491837925e-05,
      "loss": 0.5792,
      "step": 3303
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7475505801588537,
      "learning_rate": 1.0560878475434766e-05,
      "loss": 0.636,
      "step": 3304
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.683628374066146,
      "learning_rate": 1.0556006325474634e-05,
      "loss": 0.6033,
      "step": 3305
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6422883701299418,
      "learning_rate": 1.0551134043117693e-05,
      "loss": 0.5866,
      "step": 3306
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6240585010117703,
      "learning_rate": 1.054626162952413e-05,
      "loss": 0.5917,
      "step": 3307
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6908807658230026,
      "learning_rate": 1.0541389085854177e-05,
      "loss": 0.6097,
      "step": 3308
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.651163883289212,
      "learning_rate": 1.0536516413268085e-05,
      "loss": 0.5746,
      "step": 3309
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6971330808219914,
      "learning_rate": 1.0531643612926136e-05,
      "loss": 0.5982,
      "step": 3310
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6393834184889007,
      "learning_rate": 1.0526770685988657e-05,
      "loss": 0.6251,
      "step": 3311
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6828056502157729,
      "learning_rate": 1.0521897633615983e-05,
      "loss": 0.6131,
      "step": 3312
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6401451970942696,
      "learning_rate": 1.0517024456968498e-05,
      "loss": 0.6042,
      "step": 3313
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.739476091256883,
      "learning_rate": 1.0512151157206606e-05,
      "loss": 0.5921,
      "step": 3314
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6925912776718998,
      "learning_rate": 1.0507277735490743e-05,
      "loss": 0.5747,
      "step": 3315
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6068251826367145,
      "learning_rate": 1.050240419298137e-05,
      "loss": 0.5897,
      "step": 3316
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.643244947258669,
      "learning_rate": 1.0497530530838985e-05,
      "loss": 0.588,
      "step": 3317
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6237722945586831,
      "learning_rate": 1.0492656750224103e-05,
      "loss": 0.5747,
      "step": 3318
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.649867745018638,
      "learning_rate": 1.048778285229728e-05,
      "loss": 0.5679,
      "step": 3319
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6990929880630962,
      "learning_rate": 1.0482908838219088e-05,
      "loss": 0.5771,
      "step": 3320
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5819051090454697,
      "learning_rate": 1.047803470915014e-05,
      "loss": 0.5918,
      "step": 3321
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6296589846689244,
      "learning_rate": 1.047316046625106e-05,
      "loss": 0.5986,
      "step": 3322
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6916005833914498,
      "learning_rate": 1.0468286110682518e-05,
      "loss": 0.5546,
      "step": 3323
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6394612466574594,
      "learning_rate": 1.0463411643605189e-05,
      "loss": 0.5951,
      "step": 3324
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6702172361838298,
      "learning_rate": 1.0458537066179795e-05,
      "loss": 0.5854,
      "step": 3325
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6058853899345304,
      "learning_rate": 1.0453662379567069e-05,
      "loss": 0.5581,
      "step": 3326
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6013141741978772,
      "learning_rate": 1.0448787584927782e-05,
      "loss": 0.5775,
      "step": 3327
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.583318220432075,
      "learning_rate": 1.0443912683422726e-05,
      "loss": 0.5907,
      "step": 3328
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7419465764090043,
      "learning_rate": 1.0439037676212713e-05,
      "loss": 0.6132,
      "step": 3329
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6718732780565642,
      "learning_rate": 1.0434162564458585e-05,
      "loss": 0.598,
      "step": 3330
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5853137443249707,
      "learning_rate": 1.0429287349321214e-05,
      "loss": 0.6063,
      "step": 3331
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6088893900188477,
      "learning_rate": 1.0424412031961485e-05,
      "loss": 0.5812,
      "step": 3332
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6454372738028734,
      "learning_rate": 1.0419536613540317e-05,
      "loss": 0.5699,
      "step": 3333
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6193453546817901,
      "learning_rate": 1.041466109521865e-05,
      "loss": 0.5815,
      "step": 3334
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7012347830668733,
      "learning_rate": 1.0409785478157448e-05,
      "loss": 0.5967,
      "step": 3335
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6962645900389337,
      "learning_rate": 1.0404909763517695e-05,
      "loss": 0.6241,
      "step": 3336
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7378633012168955,
      "learning_rate": 1.0400033952460405e-05,
      "loss": 0.6179,
      "step": 3337
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6542805621375509,
      "learning_rate": 1.0395158046146608e-05,
      "loss": 0.6026,
      "step": 3338
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6277555154840665,
      "learning_rate": 1.039028204573736e-05,
      "loss": 0.5981,
      "step": 3339
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6302550251603217,
      "learning_rate": 1.0385405952393743e-05,
      "loss": 0.5967,
      "step": 3340
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.600986938697162,
      "learning_rate": 1.0380529767276856e-05,
      "loss": 0.5751,
      "step": 3341
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.610095686157058,
      "learning_rate": 1.0375653491547822e-05,
      "loss": 0.5944,
      "step": 3342
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6033334676487158,
      "learning_rate": 1.037077712636778e-05,
      "loss": 0.6129,
      "step": 3343
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6586453552849909,
      "learning_rate": 1.03659006728979e-05,
      "loss": 0.5978,
      "step": 3344
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6537489295595567,
      "learning_rate": 1.0361024132299364e-05,
      "loss": 0.5745,
      "step": 3345
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5943377981357091,
      "learning_rate": 1.0356147505733385e-05,
      "loss": 0.5932,
      "step": 3346
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.755811840780994,
      "learning_rate": 1.0351270794361188e-05,
      "loss": 0.5793,
      "step": 3347
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5877907485525709,
      "learning_rate": 1.034639399934402e-05,
      "loss": 0.5594,
      "step": 3348
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6082394686737358,
      "learning_rate": 1.0341517121843147e-05,
      "loss": 0.5413,
      "step": 3349
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6792546970956842,
      "learning_rate": 1.0336640163019857e-05,
      "loss": 0.6124,
      "step": 3350
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6235752627285405,
      "learning_rate": 1.0331763124035461e-05,
      "loss": 0.5852,
      "step": 3351
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.541377359591189,
      "learning_rate": 1.032688600605128e-05,
      "loss": 0.6045,
      "step": 3352
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6617242060270349,
      "learning_rate": 1.0322008810228657e-05,
      "loss": 0.5903,
      "step": 3353
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6126777343252078,
      "learning_rate": 1.031713153772896e-05,
      "loss": 0.5879,
      "step": 3354
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7042633840480479,
      "learning_rate": 1.0312254189713564e-05,
      "loss": 0.6122,
      "step": 3355
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6051695351762216,
      "learning_rate": 1.0307376767343877e-05,
      "loss": 0.5694,
      "step": 3356
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6034352643843616,
      "learning_rate": 1.0302499271781306e-05,
      "loss": 0.596,
      "step": 3357
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6402365782744825,
      "learning_rate": 1.0297621704187292e-05,
      "loss": 0.5812,
      "step": 3358
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6738384236727949,
      "learning_rate": 1.0292744065723286e-05,
      "loss": 0.5775,
      "step": 3359
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6526366555736173,
      "learning_rate": 1.0287866357550753e-05,
      "loss": 0.5891,
      "step": 3360
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.676116190108646,
      "learning_rate": 1.0282988580831183e-05,
      "loss": 0.5709,
      "step": 3361
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5865786019165299,
      "learning_rate": 1.0278110736726074e-05,
      "loss": 0.5959,
      "step": 3362
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6450637377874876,
      "learning_rate": 1.0273232826396942e-05,
      "loss": 0.5784,
      "step": 3363
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6216108265728992,
      "learning_rate": 1.0268354851005322e-05,
      "loss": 0.5624,
      "step": 3364
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5535125069848275,
      "learning_rate": 1.0263476811712765e-05,
      "loss": 0.5822,
      "step": 3365
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6734933234066817,
      "learning_rate": 1.0258598709680829e-05,
      "loss": 0.5781,
      "step": 3366
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.625578523687992,
      "learning_rate": 1.02537205460711e-05,
      "loss": 0.5751,
      "step": 3367
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6939930959251451,
      "learning_rate": 1.0248842322045165e-05,
      "loss": 0.5756,
      "step": 3368
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.658982793862038,
      "learning_rate": 1.0243964038764636e-05,
      "loss": 0.5831,
      "step": 3369
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5915144527254327,
      "learning_rate": 1.0239085697391134e-05,
      "loss": 0.5568,
      "step": 3370
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5908096972368595,
      "learning_rate": 1.0234207299086294e-05,
      "loss": 0.6049,
      "step": 3371
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5926567165614316,
      "learning_rate": 1.0229328845011763e-05,
      "loss": 0.5958,
      "step": 3372
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5791737433774528,
      "learning_rate": 1.0224450336329207e-05,
      "loss": 0.5629,
      "step": 3373
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6906417884359627,
      "learning_rate": 1.0219571774200303e-05,
      "loss": 0.6012,
      "step": 3374
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6351722476966519,
      "learning_rate": 1.021469315978674e-05,
      "loss": 0.5658,
      "step": 3375
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7209701580440103,
      "learning_rate": 1.020981449425021e-05,
      "loss": 0.6164,
      "step": 3376
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6781174418381085,
      "learning_rate": 1.0204935778752434e-05,
      "loss": 0.5818,
      "step": 3377
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7260646356912973,
      "learning_rate": 1.0200057014455134e-05,
      "loss": 0.5767,
      "step": 3378
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6059816207386498,
      "learning_rate": 1.0195178202520048e-05,
      "loss": 0.5663,
      "step": 3379
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6232353990286668,
      "learning_rate": 1.0190299344108924e-05,
      "loss": 0.5863,
      "step": 3380
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6599576878933521,
      "learning_rate": 1.0185420440383522e-05,
      "loss": 0.6152,
      "step": 3381
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5905679638837784,
      "learning_rate": 1.0180541492505605e-05,
      "loss": 0.5544,
      "step": 3382
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6316871285936227,
      "learning_rate": 1.017566250163696e-05,
      "loss": 0.5784,
      "step": 3383
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6673689433072684,
      "learning_rate": 1.0170783468939379e-05,
      "loss": 0.5877,
      "step": 3384
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6371793885353882,
      "learning_rate": 1.0165904395574655e-05,
      "loss": 0.5985,
      "step": 3385
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6417836385237053,
      "learning_rate": 1.01610252827046e-05,
      "loss": 0.6016,
      "step": 3386
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6254184731044458,
      "learning_rate": 1.0156146131491038e-05,
      "loss": 0.5991,
      "step": 3387
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6183926866658042,
      "learning_rate": 1.0151266943095792e-05,
      "loss": 0.5795,
      "step": 3388
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6059068861518669,
      "learning_rate": 1.0146387718680706e-05,
      "loss": 0.6189,
      "step": 3389
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5668969612755133,
      "learning_rate": 1.0141508459407622e-05,
      "loss": 0.585,
      "step": 3390
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0864064808608191,
      "learning_rate": 1.0136629166438396e-05,
      "loss": 0.6326,
      "step": 3391
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6144573766984985,
      "learning_rate": 1.0131749840934885e-05,
      "loss": 0.5765,
      "step": 3392
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7012056739167986,
      "learning_rate": 1.0126870484058966e-05,
      "loss": 0.5855,
      "step": 3393
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6682738900562579,
      "learning_rate": 1.0121991096972514e-05,
      "loss": 0.5683,
      "step": 3394
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3160962425824023,
      "learning_rate": 1.0117111680837412e-05,
      "loss": 0.6216,
      "step": 3395
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5979028101594189,
      "learning_rate": 1.0112232236815555e-05,
      "loss": 0.5654,
      "step": 3396
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6099245723406055,
      "learning_rate": 1.0107352766068838e-05,
      "loss": 0.5932,
      "step": 3397
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6385139155700987,
      "learning_rate": 1.0102473269759172e-05,
      "loss": 0.5974,
      "step": 3398
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6646915188807243,
      "learning_rate": 1.009759374904846e-05,
      "loss": 0.5943,
      "step": 3399
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8353680028951077,
      "learning_rate": 1.0092714205098622e-05,
      "loss": 0.5891,
      "step": 3400
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6941624834839227,
      "learning_rate": 1.0087834639071578e-05,
      "loss": 0.5996,
      "step": 3401
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.58038055786597,
      "learning_rate": 1.0082955052129259e-05,
      "loss": 0.5812,
      "step": 3402
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5893633279572974,
      "learning_rate": 1.0078075445433598e-05,
      "loss": 0.5538,
      "step": 3403
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6280379293404368,
      "learning_rate": 1.0073195820146529e-05,
      "loss": 0.5865,
      "step": 3404
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7182138763346245,
      "learning_rate": 1.006831617742999e-05,
      "loss": 0.6114,
      "step": 3405
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6208538684863142,
      "learning_rate": 1.0063436518445936e-05,
      "loss": 0.5851,
      "step": 3406
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6315727739583149,
      "learning_rate": 1.0058556844356306e-05,
      "loss": 0.5714,
      "step": 3407
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6075555830176662,
      "learning_rate": 1.0053677156323067e-05,
      "loss": 0.5797,
      "step": 3408
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.613476601849385,
      "learning_rate": 1.004879745550816e-05,
      "loss": 0.5654,
      "step": 3409
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5658719035111854,
      "learning_rate": 1.0043917743073555e-05,
      "loss": 0.5631,
      "step": 3410
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6886268880086922,
      "learning_rate": 1.003903802018121e-05,
      "loss": 0.575,
      "step": 3411
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.598847918481728,
      "learning_rate": 1.0034158287993092e-05,
      "loss": 0.5897,
      "step": 3412
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6751966294286138,
      "learning_rate": 1.0029278547671162e-05,
      "loss": 0.6321,
      "step": 3413
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6511995293795664,
      "learning_rate": 1.0024398800377398e-05,
      "loss": 0.5953,
      "step": 3414
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5890491668315402,
      "learning_rate": 1.0019519047273762e-05,
      "loss": 0.5909,
      "step": 3415
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.622104524815821,
      "learning_rate": 1.0014639289522232e-05,
      "loss": 0.5691,
      "step": 3416
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5419857929340229,
      "learning_rate": 1.0009759528284781e-05,
      "loss": 0.5845,
      "step": 3417
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6701133368344336,
      "learning_rate": 1.0004879764723377e-05,
      "loss": 0.6112,
      "step": 3418
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6199126291817003,
      "learning_rate": 1e-05,
      "loss": 0.575,
      "step": 3419
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7156135306025839,
      "learning_rate": 9.995120235276625e-06,
      "loss": 0.606,
      "step": 3420
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6222538618608157,
      "learning_rate": 9.990240471715222e-06,
      "loss": 0.5693,
      "step": 3421
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6410521899012647,
      "learning_rate": 9.98536071047777e-06,
      "loss": 0.6003,
      "step": 3422
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6381875933404813,
      "learning_rate": 9.980480952726238e-06,
      "loss": 0.5951,
      "step": 3423
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6640539969187678,
      "learning_rate": 9.975601199622605e-06,
      "loss": 0.6004,
      "step": 3424
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6189238670862454,
      "learning_rate": 9.970721452328841e-06,
      "loss": 0.5775,
      "step": 3425
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6643149037769039,
      "learning_rate": 9.965841712006911e-06,
      "loss": 0.5722,
      "step": 3426
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6554256200125288,
      "learning_rate": 9.960961979818796e-06,
      "loss": 0.5747,
      "step": 3427
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6658808240506616,
      "learning_rate": 9.956082256926448e-06,
      "loss": 0.5976,
      "step": 3428
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5911390134355263,
      "learning_rate": 9.951202544491843e-06,
      "loss": 0.545,
      "step": 3429
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7242070333091817,
      "learning_rate": 9.946322843676938e-06,
      "loss": 0.5803,
      "step": 3430
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5990314430842245,
      "learning_rate": 9.941443155643694e-06,
      "loss": 0.5988,
      "step": 3431
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6637877869726261,
      "learning_rate": 9.936563481554068e-06,
      "loss": 0.6133,
      "step": 3432
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.656031041872133,
      "learning_rate": 9.93168382257001e-06,
      "loss": 0.5895,
      "step": 3433
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6092931402122154,
      "learning_rate": 9.926804179853476e-06,
      "loss": 0.5753,
      "step": 3434
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.67045915725406,
      "learning_rate": 9.921924554566407e-06,
      "loss": 0.5658,
      "step": 3435
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6324519502300984,
      "learning_rate": 9.917044947870741e-06,
      "loss": 0.5904,
      "step": 3436
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8904110836707659,
      "learning_rate": 9.912165360928425e-06,
      "loss": 0.5921,
      "step": 3437
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6197716690430978,
      "learning_rate": 9.907285794901383e-06,
      "loss": 0.5856,
      "step": 3438
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6363546634532492,
      "learning_rate": 9.902406250951544e-06,
      "loss": 0.5856,
      "step": 3439
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7378833739263382,
      "learning_rate": 9.897526730240833e-06,
      "loss": 0.6206,
      "step": 3440
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6724301672349084,
      "learning_rate": 9.892647233931162e-06,
      "loss": 0.6073,
      "step": 3441
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.660594251761273,
      "learning_rate": 9.887767763184448e-06,
      "loss": 0.5889,
      "step": 3442
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6604356038112362,
      "learning_rate": 9.882888319162591e-06,
      "loss": 0.5791,
      "step": 3443
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6796849054890733,
      "learning_rate": 9.878008903027488e-06,
      "loss": 0.6081,
      "step": 3444
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7239056210292056,
      "learning_rate": 9.873129515941036e-06,
      "loss": 0.5725,
      "step": 3445
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6253975248215847,
      "learning_rate": 9.868250159065115e-06,
      "loss": 0.5955,
      "step": 3446
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6673263843375452,
      "learning_rate": 9.863370833561607e-06,
      "loss": 0.5766,
      "step": 3447
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6363424606376323,
      "learning_rate": 9.858491540592383e-06,
      "loss": 0.5771,
      "step": 3448
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6425654902017892,
      "learning_rate": 9.853612281319296e-06,
      "loss": 0.5793,
      "step": 3449
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5708971646023027,
      "learning_rate": 9.84873305690421e-06,
      "loss": 0.5657,
      "step": 3450
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8940440629005075,
      "learning_rate": 9.843853868508965e-06,
      "loss": 0.612,
      "step": 3451
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6311230269564229,
      "learning_rate": 9.838974717295403e-06,
      "loss": 0.573,
      "step": 3452
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7179581458153169,
      "learning_rate": 9.834095604425351e-06,
      "loss": 0.5918,
      "step": 3453
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6065616725391667,
      "learning_rate": 9.829216531060626e-06,
      "loss": 0.5735,
      "step": 3454
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5917276761938792,
      "learning_rate": 9.824337498363042e-06,
      "loss": 0.588,
      "step": 3455
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5776824021203077,
      "learning_rate": 9.819458507494395e-06,
      "loss": 0.5732,
      "step": 3456
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6538541758682856,
      "learning_rate": 9.81457955961648e-06,
      "loss": 0.5879,
      "step": 3457
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6142251632511062,
      "learning_rate": 9.80970065589108e-06,
      "loss": 0.5881,
      "step": 3458
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6237423510597284,
      "learning_rate": 9.804821797479952e-06,
      "loss": 0.561,
      "step": 3459
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6055797429283671,
      "learning_rate": 9.79994298554487e-06,
      "loss": 0.576,
      "step": 3460
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6556688850777487,
      "learning_rate": 9.795064221247566e-06,
      "loss": 0.5861,
      "step": 3461
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.665209795416182,
      "learning_rate": 9.790185505749794e-06,
      "loss": 0.5417,
      "step": 3462
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6719433748402317,
      "learning_rate": 9.785306840213267e-06,
      "loss": 0.5726,
      "step": 3463
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6953716021769079,
      "learning_rate": 9.780428225799699e-06,
      "loss": 0.5994,
      "step": 3464
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6170432190588949,
      "learning_rate": 9.775549663670794e-06,
      "loss": 0.5759,
      "step": 3465
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6914222983252234,
      "learning_rate": 9.77067115498824e-06,
      "loss": 0.5882,
      "step": 3466
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6233456256836687,
      "learning_rate": 9.76579270091371e-06,
      "loss": 0.5641,
      "step": 3467
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.583721100888317,
      "learning_rate": 9.76091430260887e-06,
      "loss": 0.5673,
      "step": 3468
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5846063971871843,
      "learning_rate": 9.756035961235366e-06,
      "loss": 0.5998,
      "step": 3469
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6891221721976141,
      "learning_rate": 9.751157677954839e-06,
      "loss": 0.589,
      "step": 3470
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8245303124131376,
      "learning_rate": 9.746279453928905e-06,
      "loss": 0.5986,
      "step": 3471
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.673638448018894,
      "learning_rate": 9.741401290319173e-06,
      "loss": 0.5973,
      "step": 3472
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7264414452086536,
      "learning_rate": 9.73652318828724e-06,
      "loss": 0.616,
      "step": 3473
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6172374183740054,
      "learning_rate": 9.731645148994681e-06,
      "loss": 0.5945,
      "step": 3474
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6293182676813036,
      "learning_rate": 9.726767173603063e-06,
      "loss": 0.5704,
      "step": 3475
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6291580280063381,
      "learning_rate": 9.721889263273931e-06,
      "loss": 0.5758,
      "step": 3476
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6449169912989114,
      "learning_rate": 9.71701141916882e-06,
      "loss": 0.5854,
      "step": 3477
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6196177316815482,
      "learning_rate": 9.712133642449249e-06,
      "loss": 0.5798,
      "step": 3478
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6790134556210577,
      "learning_rate": 9.707255934276715e-06,
      "loss": 0.5885,
      "step": 3479
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5933863911760882,
      "learning_rate": 9.70237829581271e-06,
      "loss": 0.5751,
      "step": 3480
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6412938167892943,
      "learning_rate": 9.697500728218699e-06,
      "loss": 0.5749,
      "step": 3481
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5730843053472634,
      "learning_rate": 9.692623232656126e-06,
      "loss": 0.5555,
      "step": 3482
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6809286920078135,
      "learning_rate": 9.687745810286439e-06,
      "loss": 0.5815,
      "step": 3483
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6572463433367484,
      "learning_rate": 9.682868462271042e-06,
      "loss": 0.5833,
      "step": 3484
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6055685345948852,
      "learning_rate": 9.677991189771347e-06,
      "loss": 0.5755,
      "step": 3485
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6197429194821185,
      "learning_rate": 9.673113993948726e-06,
      "loss": 0.5834,
      "step": 3486
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.564873124367959,
      "learning_rate": 9.668236875964542e-06,
      "loss": 0.5705,
      "step": 3487
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6886706203544677,
      "learning_rate": 9.663359836980144e-06,
      "loss": 0.6043,
      "step": 3488
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.672132777490386,
      "learning_rate": 9.658482878156854e-06,
      "loss": 0.572,
      "step": 3489
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5980969926729509,
      "learning_rate": 9.653606000655983e-06,
      "loss": 0.5994,
      "step": 3490
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6692182697132895,
      "learning_rate": 9.648729205638816e-06,
      "loss": 0.5797,
      "step": 3491
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6090988775907328,
      "learning_rate": 9.643852494266615e-06,
      "loss": 0.571,
      "step": 3492
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6796829995220274,
      "learning_rate": 9.638975867700638e-06,
      "loss": 0.5955,
      "step": 3493
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6506235977439351,
      "learning_rate": 9.634099327102102e-06,
      "loss": 0.595,
      "step": 3494
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6844943107732402,
      "learning_rate": 9.629222873632224e-06,
      "loss": 0.5983,
      "step": 3495
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6524244593210382,
      "learning_rate": 9.624346508452185e-06,
      "loss": 0.6096,
      "step": 3496
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6203851322730122,
      "learning_rate": 9.619470232723145e-06,
      "loss": 0.5532,
      "step": 3497
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6518216862265881,
      "learning_rate": 9.61459404760626e-06,
      "loss": 0.6044,
      "step": 3498
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6077295760903934,
      "learning_rate": 9.609717954262643e-06,
      "loss": 0.5916,
      "step": 3499
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6284201793448476,
      "learning_rate": 9.604841953853396e-06,
      "loss": 0.5788,
      "step": 3500
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5226807383788704,
      "learning_rate": 9.599966047539599e-06,
      "loss": 0.5611,
      "step": 3501
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7008685334011112,
      "learning_rate": 9.595090236482307e-06,
      "loss": 0.6036,
      "step": 3502
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6447621314794924,
      "learning_rate": 9.590214521842555e-06,
      "loss": 0.5801,
      "step": 3503
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5721369745792395,
      "learning_rate": 9.585338904781355e-06,
      "loss": 0.563,
      "step": 3504
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6355108291065888,
      "learning_rate": 9.580463386459686e-06,
      "loss": 0.5981,
      "step": 3505
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6479866839805072,
      "learning_rate": 9.57558796803852e-06,
      "loss": 0.5823,
      "step": 3506
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5949260325627296,
      "learning_rate": 9.57071265067879e-06,
      "loss": 0.5885,
      "step": 3507
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5491704131635982,
      "learning_rate": 9.565837435541418e-06,
      "loss": 0.5659,
      "step": 3508
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.595822873501851,
      "learning_rate": 9.560962323787292e-06,
      "loss": 0.5928,
      "step": 3509
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.599395441761663,
      "learning_rate": 9.556087316577278e-06,
      "loss": 0.5589,
      "step": 3510
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6051617661301335,
      "learning_rate": 9.55121241507222e-06,
      "loss": 0.5618,
      "step": 3511
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6911210711580587,
      "learning_rate": 9.546337620432931e-06,
      "loss": 0.6149,
      "step": 3512
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6629843482138579,
      "learning_rate": 9.541462933820208e-06,
      "loss": 0.5977,
      "step": 3513
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6489364324462014,
      "learning_rate": 9.536588356394816e-06,
      "loss": 0.5616,
      "step": 3514
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6166427286791143,
      "learning_rate": 9.531713889317487e-06,
      "loss": 0.6019,
      "step": 3515
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6368659712353262,
      "learning_rate": 9.526839533748943e-06,
      "loss": 0.5799,
      "step": 3516
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6238738096962393,
      "learning_rate": 9.521965290849861e-06,
      "loss": 0.5893,
      "step": 3517
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6035397566661654,
      "learning_rate": 9.517091161780914e-06,
      "loss": 0.5787,
      "step": 3518
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6410713086337395,
      "learning_rate": 9.512217147702725e-06,
      "loss": 0.5734,
      "step": 3519
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6051106025227397,
      "learning_rate": 9.507343249775898e-06,
      "loss": 0.577,
      "step": 3520
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.65476554718514,
      "learning_rate": 9.50246946916102e-06,
      "loss": 0.5606,
      "step": 3521
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6288040153125418,
      "learning_rate": 9.49759580701863e-06,
      "loss": 0.5674,
      "step": 3522
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6016962834049757,
      "learning_rate": 9.492722264509258e-06,
      "loss": 0.5968,
      "step": 3523
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7268967119381324,
      "learning_rate": 9.487848842793395e-06,
      "loss": 0.6149,
      "step": 3524
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6831923390125306,
      "learning_rate": 9.4829755430315e-06,
      "loss": 0.6029,
      "step": 3525
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6206046695673305,
      "learning_rate": 9.478102366384019e-06,
      "loss": 0.5738,
      "step": 3526
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5798499557773941,
      "learning_rate": 9.473229314011345e-06,
      "loss": 0.5752,
      "step": 3527
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5827516938038547,
      "learning_rate": 9.468356387073866e-06,
      "loss": 0.5871,
      "step": 3528
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5788836983692397,
      "learning_rate": 9.46348358673192e-06,
      "loss": 0.6004,
      "step": 3529
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6106581688607021,
      "learning_rate": 9.458610914145826e-06,
      "loss": 0.5779,
      "step": 3530
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6294349498804307,
      "learning_rate": 9.453738370475871e-06,
      "loss": 0.5999,
      "step": 3531
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.637333821485295,
      "learning_rate": 9.448865956882312e-06,
      "loss": 0.591,
      "step": 3532
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6611661385732897,
      "learning_rate": 9.443993674525367e-06,
      "loss": 0.6067,
      "step": 3533
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6600911766536743,
      "learning_rate": 9.439121524565237e-06,
      "loss": 0.5983,
      "step": 3534
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6181154921142025,
      "learning_rate": 9.434249508162076e-06,
      "loss": 0.6066,
      "step": 3535
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6622666827989003,
      "learning_rate": 9.42937762647602e-06,
      "loss": 0.588,
      "step": 3536
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6905700230778133,
      "learning_rate": 9.424505880667167e-06,
      "loss": 0.6,
      "step": 3537
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6328091536716434,
      "learning_rate": 9.41963427189557e-06,
      "loss": 0.5843,
      "step": 3538
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6677286621073831,
      "learning_rate": 9.414762801321277e-06,
      "loss": 0.6108,
      "step": 3539
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.712297415423857,
      "learning_rate": 9.409891470104277e-06,
      "loss": 0.6238,
      "step": 3540
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.629943256500636,
      "learning_rate": 9.405020279404545e-06,
      "loss": 0.5835,
      "step": 3541
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.625604147986321,
      "learning_rate": 9.40014923038201e-06,
      "loss": 0.5726,
      "step": 3542
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5654082000882764,
      "learning_rate": 9.395278324196568e-06,
      "loss": 0.5596,
      "step": 3543
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6376193302432946,
      "learning_rate": 9.390407562008088e-06,
      "loss": 0.5898,
      "step": 3544
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.679409453655597,
      "learning_rate": 9.3855369449764e-06,
      "loss": 0.5995,
      "step": 3545
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6958473095901632,
      "learning_rate": 9.380666474261303e-06,
      "loss": 0.5752,
      "step": 3546
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6622035964389933,
      "learning_rate": 9.375796151022558e-06,
      "loss": 0.5842,
      "step": 3547
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6068054375305509,
      "learning_rate": 9.370925976419885e-06,
      "loss": 0.5699,
      "step": 3548
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6182051969131892,
      "learning_rate": 9.366055951612985e-06,
      "loss": 0.5716,
      "step": 3549
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6662660350167716,
      "learning_rate": 9.361186077761503e-06,
      "loss": 0.589,
      "step": 3550
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.581760270798829,
      "learning_rate": 9.356316356025069e-06,
      "loss": 0.5876,
      "step": 3551
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6274553624326062,
      "learning_rate": 9.351446787563261e-06,
      "loss": 0.574,
      "step": 3552
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.790413557394591,
      "learning_rate": 9.346577373535623e-06,
      "loss": 0.6209,
      "step": 3553
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6352841875896149,
      "learning_rate": 9.34170811510167e-06,
      "loss": 0.6071,
      "step": 3554
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6229954962398814,
      "learning_rate": 9.336839013420871e-06,
      "loss": 0.5493,
      "step": 3555
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5961863177778497,
      "learning_rate": 9.331970069652665e-06,
      "loss": 0.5878,
      "step": 3556
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6798851802448636,
      "learning_rate": 9.327101284956451e-06,
      "loss": 0.5776,
      "step": 3557
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6400362249663696,
      "learning_rate": 9.322232660491582e-06,
      "loss": 0.5918,
      "step": 3558
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6344760264477046,
      "learning_rate": 9.31736419741739e-06,
      "loss": 0.569,
      "step": 3559
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6997905012920468,
      "learning_rate": 9.312495896893153e-06,
      "loss": 0.6001,
      "step": 3560
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6192898133920374,
      "learning_rate": 9.307627760078108e-06,
      "loss": 0.5862,
      "step": 3561
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5868941592278616,
      "learning_rate": 9.302759788131478e-06,
      "loss": 0.59,
      "step": 3562
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5914574621808749,
      "learning_rate": 9.297891982212416e-06,
      "loss": 0.574,
      "step": 3563
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6097934042365977,
      "learning_rate": 9.293024343480056e-06,
      "loss": 0.6036,
      "step": 3564
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6123445827065801,
      "learning_rate": 9.288156873093481e-06,
      "loss": 0.5683,
      "step": 3565
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6238379928696552,
      "learning_rate": 9.28328957221174e-06,
      "loss": 0.5822,
      "step": 3566
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5734619341801148,
      "learning_rate": 9.278422441993841e-06,
      "loss": 0.596,
      "step": 3567
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7002097578472358,
      "learning_rate": 9.273555483598747e-06,
      "loss": 0.5901,
      "step": 3568
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.653746567966922,
      "learning_rate": 9.268688698185388e-06,
      "loss": 0.5879,
      "step": 3569
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6303731785343545,
      "learning_rate": 9.263822086912646e-06,
      "loss": 0.5866,
      "step": 3570
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6758185559062635,
      "learning_rate": 9.25895565093936e-06,
      "loss": 0.575,
      "step": 3571
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5917165443998761,
      "learning_rate": 9.254089391424335e-06,
      "loss": 0.5957,
      "step": 3572
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5898461965184904,
      "learning_rate": 9.249223309526324e-06,
      "loss": 0.5519,
      "step": 3573
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5904202011708269,
      "learning_rate": 9.244357406404057e-06,
      "loss": 0.589,
      "step": 3574
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5731075335557331,
      "learning_rate": 9.239491683216195e-06,
      "loss": 0.5649,
      "step": 3575
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6426245641783513,
      "learning_rate": 9.234626141121372e-06,
      "loss": 0.5792,
      "step": 3576
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5828649528923986,
      "learning_rate": 9.22976078127818e-06,
      "loss": 0.5615,
      "step": 3577
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5909715525524435,
      "learning_rate": 9.224895604845157e-06,
      "loss": 0.5779,
      "step": 3578
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5762903595447544,
      "learning_rate": 9.22003061298081e-06,
      "loss": 0.5398,
      "step": 3579
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.687681006046133,
      "learning_rate": 9.215165806843594e-06,
      "loss": 0.5741,
      "step": 3580
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.684325162469507,
      "learning_rate": 9.21030118759192e-06,
      "loss": 0.5918,
      "step": 3581
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5973992888402869,
      "learning_rate": 9.205436756384159e-06,
      "loss": 0.5845,
      "step": 3582
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6055858126564325,
      "learning_rate": 9.200572514378631e-06,
      "loss": 0.5973,
      "step": 3583
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6581527239450523,
      "learning_rate": 9.195708462733622e-06,
      "loss": 0.5901,
      "step": 3584
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6424616742903639,
      "learning_rate": 9.190844602607357e-06,
      "loss": 0.5741,
      "step": 3585
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6835900978390228,
      "learning_rate": 9.185980935158025e-06,
      "loss": 0.5936,
      "step": 3586
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6393619992380906,
      "learning_rate": 9.181117461543771e-06,
      "loss": 0.5896,
      "step": 3587
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6324929228368676,
      "learning_rate": 9.176254182922685e-06,
      "loss": 0.5873,
      "step": 3588
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6277851160905646,
      "learning_rate": 9.17139110045282e-06,
      "loss": 0.6012,
      "step": 3589
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6142788032931262,
      "learning_rate": 9.166528215292181e-06,
      "loss": 0.589,
      "step": 3590
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7100756244356393,
      "learning_rate": 9.161665528598717e-06,
      "loss": 0.6102,
      "step": 3591
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6432076709658403,
      "learning_rate": 9.15680304153034e-06,
      "loss": 0.5519,
      "step": 3592
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6141015613686776,
      "learning_rate": 9.151940755244912e-06,
      "loss": 0.5643,
      "step": 3593
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5936722122039058,
      "learning_rate": 9.147078670900238e-06,
      "loss": 0.5809,
      "step": 3594
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6564523693287254,
      "learning_rate": 9.142216789654093e-06,
      "loss": 0.573,
      "step": 3595
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7221491531470441,
      "learning_rate": 9.137355112664181e-06,
      "loss": 0.5882,
      "step": 3596
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6485389658947287,
      "learning_rate": 9.132493641088184e-06,
      "loss": 0.5704,
      "step": 3597
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5936746482829292,
      "learning_rate": 9.127632376083712e-06,
      "loss": 0.577,
      "step": 3598
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6517804531299307,
      "learning_rate": 9.122771318808335e-06,
      "loss": 0.5893,
      "step": 3599
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7354706917923255,
      "learning_rate": 9.117910470419575e-06,
      "loss": 0.5951,
      "step": 3600
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6393110348995402,
      "learning_rate": 9.113049832074899e-06,
      "loss": 0.5536,
      "step": 3601
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5125045876405372,
      "learning_rate": 9.108189404931732e-06,
      "loss": 0.5668,
      "step": 3602
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6375831039656291,
      "learning_rate": 9.103329190147444e-06,
      "loss": 0.5988,
      "step": 3603
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6294638128611082,
      "learning_rate": 9.098469188879348e-06,
      "loss": 0.5733,
      "step": 3604
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6547848724986052,
      "learning_rate": 9.093609402284722e-06,
      "loss": 0.5799,
      "step": 3605
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6972618162932585,
      "learning_rate": 9.088749831520771e-06,
      "loss": 0.6064,
      "step": 3606
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6801389212709676,
      "learning_rate": 9.083890477744675e-06,
      "loss": 0.6154,
      "step": 3607
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6062950998536771,
      "learning_rate": 9.079031342113542e-06,
      "loss": 0.5717,
      "step": 3608
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6457173112353943,
      "learning_rate": 9.07417242578443e-06,
      "loss": 0.6066,
      "step": 3609
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6195486424813679,
      "learning_rate": 9.069313729914356e-06,
      "loss": 0.5566,
      "step": 3610
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5915859456130305,
      "learning_rate": 9.064455255660274e-06,
      "loss": 0.5661,
      "step": 3611
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6342258541810519,
      "learning_rate": 9.059597004179093e-06,
      "loss": 0.5721,
      "step": 3612
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6596357087672491,
      "learning_rate": 9.054738976627662e-06,
      "loss": 0.5708,
      "step": 3613
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6335938771687679,
      "learning_rate": 9.049881174162779e-06,
      "loss": 0.5786,
      "step": 3614
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7101194243334825,
      "learning_rate": 9.04502359794119e-06,
      "loss": 0.6037,
      "step": 3615
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.617894259727918,
      "learning_rate": 9.040166249119583e-06,
      "loss": 0.5823,
      "step": 3616
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5995434951787232,
      "learning_rate": 9.035309128854605e-06,
      "loss": 0.5755,
      "step": 3617
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6726037478030663,
      "learning_rate": 9.030452238302831e-06,
      "loss": 0.6018,
      "step": 3618
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5754639643670797,
      "learning_rate": 9.025595578620783e-06,
      "loss": 0.5577,
      "step": 3619
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6210622236899008,
      "learning_rate": 9.020739150964947e-06,
      "loss": 0.6038,
      "step": 3620
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6536789975645003,
      "learning_rate": 9.015882956491732e-06,
      "loss": 0.575,
      "step": 3621
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.63137146786354,
      "learning_rate": 9.011026996357504e-06,
      "loss": 0.5919,
      "step": 3622
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6227844093255384,
      "learning_rate": 9.006171271718567e-06,
      "loss": 0.596,
      "step": 3623
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.560329679564074,
      "learning_rate": 9.001315783731169e-06,
      "loss": 0.5998,
      "step": 3624
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6304660688004808,
      "learning_rate": 8.996460533551512e-06,
      "loss": 0.5726,
      "step": 3625
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.682348622280806,
      "learning_rate": 8.991605522335727e-06,
      "loss": 0.5719,
      "step": 3626
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6441496752148902,
      "learning_rate": 8.98675075123989e-06,
      "loss": 0.6318,
      "step": 3627
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.912335766713085,
      "learning_rate": 8.981896221420039e-06,
      "loss": 0.5684,
      "step": 3628
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6652477366563128,
      "learning_rate": 8.97704193403212e-06,
      "loss": 0.5763,
      "step": 3629
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6419557136267321,
      "learning_rate": 8.972187890232061e-06,
      "loss": 0.5998,
      "step": 3630
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6334842480291578,
      "learning_rate": 8.967334091175698e-06,
      "loss": 0.5646,
      "step": 3631
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5956927114910707,
      "learning_rate": 8.962480538018828e-06,
      "loss": 0.6022,
      "step": 3632
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5935985297557291,
      "learning_rate": 8.957627231917183e-06,
      "loss": 0.5765,
      "step": 3633
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6141477853166998,
      "learning_rate": 8.952774174026437e-06,
      "loss": 0.5796,
      "step": 3634
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5623824995512839,
      "learning_rate": 8.947921365502208e-06,
      "loss": 0.5444,
      "step": 3635
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7082964120799913,
      "learning_rate": 8.94306880750005e-06,
      "loss": 0.5866,
      "step": 3636
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5749850217738294,
      "learning_rate": 8.938216501175457e-06,
      "loss": 0.5968,
      "step": 3637
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6320125847903993,
      "learning_rate": 8.933364447683868e-06,
      "loss": 0.5669,
      "step": 3638
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6726298255445491,
      "learning_rate": 8.928512648180659e-06,
      "loss": 0.5821,
      "step": 3639
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6216107775697395,
      "learning_rate": 8.923661103821146e-06,
      "loss": 0.5949,
      "step": 3640
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6592998328182962,
      "learning_rate": 8.918809815760585e-06,
      "loss": 0.5792,
      "step": 3641
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6488747384530087,
      "learning_rate": 8.913958785154165e-06,
      "loss": 0.6073,
      "step": 3642
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.631931856716016,
      "learning_rate": 8.909108013157021e-06,
      "loss": 0.6139,
      "step": 3643
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6496646964289029,
      "learning_rate": 8.904257500924224e-06,
      "loss": 0.6102,
      "step": 3644
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7147563856939133,
      "learning_rate": 8.899407249610785e-06,
      "loss": 0.6087,
      "step": 3645
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7684500592890321,
      "learning_rate": 8.894557260371648e-06,
      "loss": 0.5827,
      "step": 3646
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.664107292124564,
      "learning_rate": 8.889707534361699e-06,
      "loss": 0.5886,
      "step": 3647
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6416119086542871,
      "learning_rate": 8.88485807273576e-06,
      "loss": 0.584,
      "step": 3648
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6358417742936826,
      "learning_rate": 8.880008876648588e-06,
      "loss": 0.5816,
      "step": 3649
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6480135713313382,
      "learning_rate": 8.875159947254882e-06,
      "loss": 0.5945,
      "step": 3650
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5923562279650835,
      "learning_rate": 8.870311285709274e-06,
      "loss": 0.6049,
      "step": 3651
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7058294373434935,
      "learning_rate": 8.865462893166325e-06,
      "loss": 0.5869,
      "step": 3652
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6383275611384467,
      "learning_rate": 8.860614770780552e-06,
      "loss": 0.5769,
      "step": 3653
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5580418774839024,
      "learning_rate": 8.855766919706386e-06,
      "loss": 0.5631,
      "step": 3654
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6688935597417797,
      "learning_rate": 8.850919341098202e-06,
      "loss": 0.5818,
      "step": 3655
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6124001595463925,
      "learning_rate": 8.846072036110316e-06,
      "loss": 0.5739,
      "step": 3656
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5972464865665833,
      "learning_rate": 8.841225005896967e-06,
      "loss": 0.5973,
      "step": 3657
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6101307336141002,
      "learning_rate": 8.83637825161234e-06,
      "loss": 0.5733,
      "step": 3658
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5919756788801056,
      "learning_rate": 8.83153177441055e-06,
      "loss": 0.5709,
      "step": 3659
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6397241295946606,
      "learning_rate": 8.826685575445638e-06,
      "loss": 0.5783,
      "step": 3660
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7193068055291707,
      "learning_rate": 8.821839655871593e-06,
      "loss": 0.6206,
      "step": 3661
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6534594063794702,
      "learning_rate": 8.81699401684233e-06,
      "loss": 0.605,
      "step": 3662
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.612446609241036,
      "learning_rate": 8.812148659511695e-06,
      "loss": 0.5785,
      "step": 3663
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6070410489298179,
      "learning_rate": 8.80730358503347e-06,
      "loss": 0.5757,
      "step": 3664
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.634984066490653,
      "learning_rate": 8.80245879456137e-06,
      "loss": 0.5914,
      "step": 3665
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6054389065746065,
      "learning_rate": 8.797614289249041e-06,
      "loss": 0.5685,
      "step": 3666
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5648290432608108,
      "learning_rate": 8.792770070250061e-06,
      "loss": 0.5728,
      "step": 3667
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5984988806252426,
      "learning_rate": 8.787926138717942e-06,
      "loss": 0.5593,
      "step": 3668
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7198786162202782,
      "learning_rate": 8.783082495806128e-06,
      "loss": 0.5878,
      "step": 3669
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7016109979957087,
      "learning_rate": 8.778239142667985e-06,
      "loss": 0.5707,
      "step": 3670
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5899964710105217,
      "learning_rate": 8.773396080456828e-06,
      "loss": 0.5851,
      "step": 3671
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6291502891008419,
      "learning_rate": 8.768553310325883e-06,
      "loss": 0.5851,
      "step": 3672
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6575354779011473,
      "learning_rate": 8.763710833428319e-06,
      "loss": 0.6173,
      "step": 3673
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6147013189900814,
      "learning_rate": 8.758868650917236e-06,
      "loss": 0.6041,
      "step": 3674
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.647579902587223,
      "learning_rate": 8.754026763945648e-06,
      "loss": 0.5825,
      "step": 3675
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6140787347873581,
      "learning_rate": 8.749185173666523e-06,
      "loss": 0.5755,
      "step": 3676
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7986648608216732,
      "learning_rate": 8.744343881232736e-06,
      "loss": 0.6158,
      "step": 3677
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6429918458791456,
      "learning_rate": 8.739502887797108e-06,
      "loss": 0.6197,
      "step": 3678
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5550438701333668,
      "learning_rate": 8.734662194512377e-06,
      "loss": 0.5494,
      "step": 3679
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7315176607873679,
      "learning_rate": 8.729821802531213e-06,
      "loss": 0.6028,
      "step": 3680
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7319344966777492,
      "learning_rate": 8.72498171300622e-06,
      "loss": 0.615,
      "step": 3681
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6784179484783991,
      "learning_rate": 8.720141927089921e-06,
      "loss": 0.5918,
      "step": 3682
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6391059128301152,
      "learning_rate": 8.715302445934773e-06,
      "loss": 0.5644,
      "step": 3683
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6158511004013395,
      "learning_rate": 8.710463270693159e-06,
      "loss": 0.5747,
      "step": 3684
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6574074677339155,
      "learning_rate": 8.705624402517382e-06,
      "loss": 0.5974,
      "step": 3685
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7725307760231451,
      "learning_rate": 8.70078584255969e-06,
      "loss": 0.6027,
      "step": 3686
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6813875566407345,
      "learning_rate": 8.695947591972238e-06,
      "loss": 0.5821,
      "step": 3687
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6184925533470573,
      "learning_rate": 8.691109651907114e-06,
      "loss": 0.5843,
      "step": 3688
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7141178349384589,
      "learning_rate": 8.686272023516339e-06,
      "loss": 0.5873,
      "step": 3689
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7455171726124447,
      "learning_rate": 8.68143470795185e-06,
      "loss": 0.5813,
      "step": 3690
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6594384150528669,
      "learning_rate": 8.676597706365516e-06,
      "loss": 0.5604,
      "step": 3691
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6178391438297803,
      "learning_rate": 8.671761019909129e-06,
      "loss": 0.5652,
      "step": 3692
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6385842044956653,
      "learning_rate": 8.6669246497344e-06,
      "loss": 0.588,
      "step": 3693
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7072971094550436,
      "learning_rate": 8.662088596992982e-06,
      "loss": 0.5776,
      "step": 3694
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6453245973922734,
      "learning_rate": 8.65725286283643e-06,
      "loss": 0.5501,
      "step": 3695
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6268180063579408,
      "learning_rate": 8.652417448416239e-06,
      "loss": 0.5766,
      "step": 3696
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.666582337306442,
      "learning_rate": 8.647582354883827e-06,
      "loss": 0.5903,
      "step": 3697
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.758741068872693,
      "learning_rate": 8.642747583390522e-06,
      "loss": 0.5831,
      "step": 3698
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6271586209069864,
      "learning_rate": 8.637913135087592e-06,
      "loss": 0.5774,
      "step": 3699
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5819639063586642,
      "learning_rate": 8.633079011126216e-06,
      "loss": 0.5671,
      "step": 3700
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6839563917686743,
      "learning_rate": 8.628245212657506e-06,
      "loss": 0.59,
      "step": 3701
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6694948026256242,
      "learning_rate": 8.62341174083249e-06,
      "loss": 0.6091,
      "step": 3702
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5932635854605798,
      "learning_rate": 8.618578596802114e-06,
      "loss": 0.5795,
      "step": 3703
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6474484443821179,
      "learning_rate": 8.61374578171726e-06,
      "loss": 0.5958,
      "step": 3704
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6592299224556459,
      "learning_rate": 8.608913296728713e-06,
      "loss": 0.5567,
      "step": 3705
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9113943701088421,
      "learning_rate": 8.604081142987199e-06,
      "loss": 0.5752,
      "step": 3706
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6708650945800332,
      "learning_rate": 8.599249321643353e-06,
      "loss": 0.5919,
      "step": 3707
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7377888128879668,
      "learning_rate": 8.594417833847728e-06,
      "loss": 0.5989,
      "step": 3708
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.619468944915587,
      "learning_rate": 8.58958668075081e-06,
      "loss": 0.5955,
      "step": 3709
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8388667518052881,
      "learning_rate": 8.584755863502988e-06,
      "loss": 0.6461,
      "step": 3710
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6350053772909169,
      "learning_rate": 8.579925383254598e-06,
      "loss": 0.5831,
      "step": 3711
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6426393412226945,
      "learning_rate": 8.575095241155864e-06,
      "loss": 0.5895,
      "step": 3712
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6462028591339563,
      "learning_rate": 8.570265438356949e-06,
      "loss": 0.5726,
      "step": 3713
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6230838472187346,
      "learning_rate": 8.565435976007932e-06,
      "loss": 0.5728,
      "step": 3714
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5949274258515123,
      "learning_rate": 8.560606855258808e-06,
      "loss": 0.587,
      "step": 3715
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6635509564849554,
      "learning_rate": 8.555778077259496e-06,
      "loss": 0.5747,
      "step": 3716
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6873651759082553,
      "learning_rate": 8.550949643159829e-06,
      "loss": 0.5824,
      "step": 3717
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6829712903915968,
      "learning_rate": 8.546121554109552e-06,
      "loss": 0.6108,
      "step": 3718
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6455012332137978,
      "learning_rate": 8.541293811258345e-06,
      "loss": 0.5794,
      "step": 3719
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6860589960039561,
      "learning_rate": 8.536466415755787e-06,
      "loss": 0.5791,
      "step": 3720
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6037345553188098,
      "learning_rate": 8.531639368751384e-06,
      "loss": 0.584,
      "step": 3721
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6421446985885838,
      "learning_rate": 8.52681267139456e-06,
      "loss": 0.581,
      "step": 3722
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6181659652025395,
      "learning_rate": 8.521986324834653e-06,
      "loss": 0.5874,
      "step": 3723
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6279366410088356,
      "learning_rate": 8.517160330220915e-06,
      "loss": 0.588,
      "step": 3724
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.670062196685405,
      "learning_rate": 8.512334688702521e-06,
      "loss": 0.6251,
      "step": 3725
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9391085021382437,
      "learning_rate": 8.507509401428554e-06,
      "loss": 0.6071,
      "step": 3726
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6700796479871785,
      "learning_rate": 8.502684469548017e-06,
      "loss": 0.5756,
      "step": 3727
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6287283189828867,
      "learning_rate": 8.49785989420983e-06,
      "loss": 0.5883,
      "step": 3728
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6514822367986026,
      "learning_rate": 8.493035676562826e-06,
      "loss": 0.5783,
      "step": 3729
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6374203560975037,
      "learning_rate": 8.488211817755753e-06,
      "loss": 0.5911,
      "step": 3730
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6089498177555974,
      "learning_rate": 8.483388318937264e-06,
      "loss": 0.5731,
      "step": 3731
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6967472261404132,
      "learning_rate": 8.478565181255953e-06,
      "loss": 0.6058,
      "step": 3732
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5889942370548241,
      "learning_rate": 8.473742405860294e-06,
      "loss": 0.5958,
      "step": 3733
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5983189217258296,
      "learning_rate": 8.468919993898704e-06,
      "loss": 0.5856,
      "step": 3734
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6324246441400496,
      "learning_rate": 8.464097946519494e-06,
      "loss": 0.581,
      "step": 3735
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5999284407068777,
      "learning_rate": 8.459276264870895e-06,
      "loss": 0.5803,
      "step": 3736
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.67314048021972,
      "learning_rate": 8.454454950101052e-06,
      "loss": 0.5963,
      "step": 3737
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.621317232161455,
      "learning_rate": 8.449634003358022e-06,
      "loss": 0.617,
      "step": 3738
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6823826008842583,
      "learning_rate": 8.444813425789776e-06,
      "loss": 0.5752,
      "step": 3739
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6119651357839501,
      "learning_rate": 8.439993218544192e-06,
      "loss": 0.5868,
      "step": 3740
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7127536797955354,
      "learning_rate": 8.435173382769059e-06,
      "loss": 0.572,
      "step": 3741
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5637331544481734,
      "learning_rate": 8.430353919612091e-06,
      "loss": 0.5719,
      "step": 3742
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6136912396004248,
      "learning_rate": 8.425534830220894e-06,
      "loss": 0.5833,
      "step": 3743
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6874555664267753,
      "learning_rate": 8.420716115743004e-06,
      "loss": 0.5994,
      "step": 3744
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7117421619517547,
      "learning_rate": 8.415897777325853e-06,
      "loss": 0.5754,
      "step": 3745
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6452885911003376,
      "learning_rate": 8.411079816116786e-06,
      "loss": 0.5902,
      "step": 3746
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6109126942574463,
      "learning_rate": 8.406262233263067e-06,
      "loss": 0.5817,
      "step": 3747
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.676829932609987,
      "learning_rate": 8.401445029911861e-06,
      "loss": 0.5981,
      "step": 3748
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6263814147667683,
      "learning_rate": 8.396628207210243e-06,
      "loss": 0.5653,
      "step": 3749
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6627122531575761,
      "learning_rate": 8.391811766305205e-06,
      "loss": 0.5851,
      "step": 3750
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6204171257736762,
      "learning_rate": 8.38699570834364e-06,
      "loss": 0.5819,
      "step": 3751
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6910035654922231,
      "learning_rate": 8.382180034472353e-06,
      "loss": 0.6218,
      "step": 3752
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6302766841554887,
      "learning_rate": 8.377364745838062e-06,
      "loss": 0.5864,
      "step": 3753
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6110542945923875,
      "learning_rate": 8.372549843587377e-06,
      "loss": 0.5931,
      "step": 3754
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6595571111039364,
      "learning_rate": 8.36773532886684e-06,
      "loss": 0.5952,
      "step": 3755
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5439071641703244,
      "learning_rate": 8.36292120282288e-06,
      "loss": 0.5882,
      "step": 3756
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6226011155450006,
      "learning_rate": 8.358107466601848e-06,
      "loss": 0.5288,
      "step": 3757
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6170634191177243,
      "learning_rate": 8.353294121349993e-06,
      "loss": 0.6185,
      "step": 3758
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6001408162001888,
      "learning_rate": 8.348481168213468e-06,
      "loss": 0.5472,
      "step": 3759
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5932522479486794,
      "learning_rate": 8.343668608338351e-06,
      "loss": 0.5584,
      "step": 3760
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6432059302317751,
      "learning_rate": 8.338856442870603e-06,
      "loss": 0.6038,
      "step": 3761
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6183295541768983,
      "learning_rate": 8.334044672956108e-06,
      "loss": 0.5852,
      "step": 3762
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6604864862711846,
      "learning_rate": 8.329233299740649e-06,
      "loss": 0.5684,
      "step": 3763
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6100179844236524,
      "learning_rate": 8.324422324369908e-06,
      "loss": 0.5745,
      "step": 3764
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.591253732761286,
      "learning_rate": 8.319611747989492e-06,
      "loss": 0.5757,
      "step": 3765
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7314349279596094,
      "learning_rate": 8.314801571744888e-06,
      "loss": 0.5928,
      "step": 3766
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5605537260463648,
      "learning_rate": 8.309991796781512e-06,
      "loss": 0.5705,
      "step": 3767
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6438176628599344,
      "learning_rate": 8.305182424244665e-06,
      "loss": 0.6215,
      "step": 3768
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6872102100082791,
      "learning_rate": 8.300373455279557e-06,
      "loss": 0.6052,
      "step": 3769
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6582808533839225,
      "learning_rate": 8.295564891031314e-06,
      "loss": 0.5954,
      "step": 3770
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6648612007017505,
      "learning_rate": 8.29075673264495e-06,
      "loss": 0.5831,
      "step": 3771
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6590578781531415,
      "learning_rate": 8.285948981265392e-06,
      "loss": 0.5887,
      "step": 3772
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6016647329558675,
      "learning_rate": 8.281141638037464e-06,
      "loss": 0.5894,
      "step": 3773
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6416733166245977,
      "learning_rate": 8.276334704105896e-06,
      "loss": 0.5869,
      "step": 3774
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5676582223187437,
      "learning_rate": 8.271528180615322e-06,
      "loss": 0.6047,
      "step": 3775
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6295383793154725,
      "learning_rate": 8.266722068710276e-06,
      "loss": 0.5985,
      "step": 3776
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6243877970686827,
      "learning_rate": 8.261916369535196e-06,
      "loss": 0.6112,
      "step": 3777
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6107238069207156,
      "learning_rate": 8.257111084234417e-06,
      "loss": 0.594,
      "step": 3778
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.61729264614804,
      "learning_rate": 8.252306213952178e-06,
      "loss": 0.572,
      "step": 3779
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6636937205487005,
      "learning_rate": 8.247501759832621e-06,
      "loss": 0.5751,
      "step": 3780
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6829852313884219,
      "learning_rate": 8.24269772301979e-06,
      "loss": 0.601,
      "step": 3781
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6488529150329063,
      "learning_rate": 8.237894104657624e-06,
      "loss": 0.6023,
      "step": 3782
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6294555666997281,
      "learning_rate": 8.233090905889969e-06,
      "loss": 0.5646,
      "step": 3783
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.570410520302905,
      "learning_rate": 8.228288127860561e-06,
      "loss": 0.5778,
      "step": 3784
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5955495896242229,
      "learning_rate": 8.223485771713052e-06,
      "loss": 0.5621,
      "step": 3785
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.713938436813007,
      "learning_rate": 8.218683838590981e-06,
      "loss": 0.5821,
      "step": 3786
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6677399458682457,
      "learning_rate": 8.213882329637781e-06,
      "loss": 0.5644,
      "step": 3787
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6381618555376362,
      "learning_rate": 8.209081245996807e-06,
      "loss": 0.5925,
      "step": 3788
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6135647668439506,
      "learning_rate": 8.204280588811283e-06,
      "loss": 0.5769,
      "step": 3789
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6827176138958752,
      "learning_rate": 8.199480359224359e-06,
      "loss": 0.5912,
      "step": 3790
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5576336989185864,
      "learning_rate": 8.194680558379066e-06,
      "loss": 0.5936,
      "step": 3791
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5980977787868949,
      "learning_rate": 8.189881187418332e-06,
      "loss": 0.5915,
      "step": 3792
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6432876478994057,
      "learning_rate": 8.185082247484999e-06,
      "loss": 0.6037,
      "step": 3793
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5896244913464119,
      "learning_rate": 8.180283739721785e-06,
      "loss": 0.5826,
      "step": 3794
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6281718583206453,
      "learning_rate": 8.175485665271324e-06,
      "loss": 0.572,
      "step": 3795
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6094059646739965,
      "learning_rate": 8.170688025276134e-06,
      "loss": 0.5818,
      "step": 3796
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.587835439681876,
      "learning_rate": 8.165890820878635e-06,
      "loss": 0.5767,
      "step": 3797
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6182088506628937,
      "learning_rate": 8.161094053221148e-06,
      "loss": 0.5904,
      "step": 3798
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7166832223532028,
      "learning_rate": 8.15629772344587e-06,
      "loss": 0.613,
      "step": 3799
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6721938856386374,
      "learning_rate": 8.151501832694923e-06,
      "loss": 0.5819,
      "step": 3800
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5965247355337259,
      "learning_rate": 8.146706382110304e-06,
      "loss": 0.5829,
      "step": 3801
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6014999426363676,
      "learning_rate": 8.141911372833909e-06,
      "loss": 0.5682,
      "step": 3802
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7458580443561929,
      "learning_rate": 8.137116806007532e-06,
      "loss": 0.5935,
      "step": 3803
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5868136995686194,
      "learning_rate": 8.132322682772859e-06,
      "loss": 0.561,
      "step": 3804
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5713021473113968,
      "learning_rate": 8.127529004271474e-06,
      "loss": 0.5912,
      "step": 3805
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6447463750050499,
      "learning_rate": 8.122735771644853e-06,
      "loss": 0.6132,
      "step": 3806
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6278351446795338,
      "learning_rate": 8.117942986034365e-06,
      "loss": 0.5823,
      "step": 3807
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6373372740152361,
      "learning_rate": 8.11315064858127e-06,
      "loss": 0.5922,
      "step": 3808
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6367265114804485,
      "learning_rate": 8.108358760426727e-06,
      "loss": 0.5674,
      "step": 3809
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.59306066354714,
      "learning_rate": 8.103567322711787e-06,
      "loss": 0.5956,
      "step": 3810
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7168760779458118,
      "learning_rate": 8.098776336577394e-06,
      "loss": 0.5919,
      "step": 3811
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6296975036956648,
      "learning_rate": 8.093985803164372e-06,
      "loss": 0.602,
      "step": 3812
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6641882888562213,
      "learning_rate": 8.089195723613462e-06,
      "loss": 0.5573,
      "step": 3813
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5786949537808074,
      "learning_rate": 8.084406099065273e-06,
      "loss": 0.5825,
      "step": 3814
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.654435998747993,
      "learning_rate": 8.079616930660317e-06,
      "loss": 0.5841,
      "step": 3815
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6518910639534604,
      "learning_rate": 8.074828219539001e-06,
      "loss": 0.6048,
      "step": 3816
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6872466286881628,
      "learning_rate": 8.07003996684161e-06,
      "loss": 0.6004,
      "step": 3817
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6462254356973444,
      "learning_rate": 8.065252173708334e-06,
      "loss": 0.5976,
      "step": 3818
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5807505143155958,
      "learning_rate": 8.060464841279247e-06,
      "loss": 0.575,
      "step": 3819
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6077662348561012,
      "learning_rate": 8.055677970694306e-06,
      "loss": 0.5915,
      "step": 3820
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6030131634825098,
      "learning_rate": 8.050891563093378e-06,
      "loss": 0.5551,
      "step": 3821
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6211201472807328,
      "learning_rate": 8.046105619616195e-06,
      "loss": 0.5564,
      "step": 3822
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5671718665946797,
      "learning_rate": 8.041320141402403e-06,
      "loss": 0.5685,
      "step": 3823
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6026521016240577,
      "learning_rate": 8.036535129591517e-06,
      "loss": 0.5881,
      "step": 3824
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6622334745107595,
      "learning_rate": 8.031750585322948e-06,
      "loss": 0.5805,
      "step": 3825
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6920092352013156,
      "learning_rate": 8.026966509736001e-06,
      "loss": 0.5972,
      "step": 3826
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5511804598171768,
      "learning_rate": 8.022182903969863e-06,
      "loss": 0.5715,
      "step": 3827
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6201692048170034,
      "learning_rate": 8.017399769163614e-06,
      "loss": 0.5746,
      "step": 3828
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5999269654662471,
      "learning_rate": 8.012617106456217e-06,
      "loss": 0.5761,
      "step": 3829
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5840796930461397,
      "learning_rate": 8.007834916986523e-06,
      "loss": 0.5782,
      "step": 3830
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6745487832178498,
      "learning_rate": 8.003053201893275e-06,
      "loss": 0.5828,
      "step": 3831
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6439937471572423,
      "learning_rate": 7.998271962315099e-06,
      "loss": 0.571,
      "step": 3832
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5950057422671994,
      "learning_rate": 7.993491199390508e-06,
      "loss": 0.5516,
      "step": 3833
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5920724016120251,
      "learning_rate": 7.988710914257906e-06,
      "loss": 0.5795,
      "step": 3834
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.645681272365996,
      "learning_rate": 7.983931108055574e-06,
      "loss": 0.5574,
      "step": 3835
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6124724700797554,
      "learning_rate": 7.979151781921686e-06,
      "loss": 0.5791,
      "step": 3836
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7559167191902372,
      "learning_rate": 7.974372936994302e-06,
      "loss": 0.5555,
      "step": 3837
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6817312174900184,
      "learning_rate": 7.969594574411364e-06,
      "loss": 0.585,
      "step": 3838
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6020589356580623,
      "learning_rate": 7.964816695310702e-06,
      "loss": 0.5867,
      "step": 3839
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7040003530333727,
      "learning_rate": 7.960039300830028e-06,
      "loss": 0.612,
      "step": 3840
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6519260985421906,
      "learning_rate": 7.955262392106943e-06,
      "loss": 0.6041,
      "step": 3841
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6461729170685644,
      "learning_rate": 7.95048597027893e-06,
      "loss": 0.6054,
      "step": 3842
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6164930721926113,
      "learning_rate": 7.945710036483345e-06,
      "loss": 0.5984,
      "step": 3843
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6916182031619544,
      "learning_rate": 7.940934591857455e-06,
      "loss": 0.584,
      "step": 3844
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6302428375192867,
      "learning_rate": 7.936159637538379e-06,
      "loss": 0.5892,
      "step": 3845
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6479439435296989,
      "learning_rate": 7.931385174663146e-06,
      "loss": 0.5768,
      "step": 3846
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5975209829843598,
      "learning_rate": 7.92661120436865e-06,
      "loss": 0.5853,
      "step": 3847
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5830121193590843,
      "learning_rate": 7.921837727791673e-06,
      "loss": 0.5915,
      "step": 3848
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5891087567990743,
      "learning_rate": 7.917064746068882e-06,
      "loss": 0.5669,
      "step": 3849
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6147511631320051,
      "learning_rate": 7.912292260336823e-06,
      "loss": 0.5752,
      "step": 3850
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5638508492649064,
      "learning_rate": 7.90752027173193e-06,
      "loss": 0.5769,
      "step": 3851
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5676000821272018,
      "learning_rate": 7.902748781390509e-06,
      "loss": 0.5818,
      "step": 3852
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6615559219123522,
      "learning_rate": 7.897977790448753e-06,
      "loss": 0.593,
      "step": 3853
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6466636406696142,
      "learning_rate": 7.89320730004274e-06,
      "loss": 0.5817,
      "step": 3854
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.570084931126133,
      "learning_rate": 7.888437311308415e-06,
      "loss": 0.5971,
      "step": 3855
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5705258015259805,
      "learning_rate": 7.883667825381623e-06,
      "loss": 0.5561,
      "step": 3856
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6675374257441391,
      "learning_rate": 7.878898843398073e-06,
      "loss": 0.619,
      "step": 3857
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7420881803973364,
      "learning_rate": 7.874130366493359e-06,
      "loss": 0.6064,
      "step": 3858
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7103542030161568,
      "learning_rate": 7.869362395802959e-06,
      "loss": 0.5961,
      "step": 3859
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5693143388895799,
      "learning_rate": 7.864594932462227e-06,
      "loss": 0.5933,
      "step": 3860
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6718137675927958,
      "learning_rate": 7.859827977606394e-06,
      "loss": 0.5633,
      "step": 3861
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.603147633477335,
      "learning_rate": 7.855061532370575e-06,
      "loss": 0.5862,
      "step": 3862
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5719328916196443,
      "learning_rate": 7.85029559788976e-06,
      "loss": 0.5601,
      "step": 3863
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7683366444684611,
      "learning_rate": 7.845530175298818e-06,
      "loss": 0.6299,
      "step": 3864
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6580496411002555,
      "learning_rate": 7.840765265732495e-06,
      "loss": 0.5894,
      "step": 3865
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.609094723874891,
      "learning_rate": 7.83600087032542e-06,
      "loss": 0.573,
      "step": 3866
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6429411570050613,
      "learning_rate": 7.831236990212097e-06,
      "loss": 0.5841,
      "step": 3867
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6205271095153541,
      "learning_rate": 7.826473626526895e-06,
      "loss": 0.5899,
      "step": 3868
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6852157358582778,
      "learning_rate": 7.821710780404086e-06,
      "loss": 0.5978,
      "step": 3869
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.582287513108944,
      "learning_rate": 7.816948452977792e-06,
      "loss": 0.5871,
      "step": 3870
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6372261610644809,
      "learning_rate": 7.81218664538203e-06,
      "loss": 0.5564,
      "step": 3871
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5866694951151218,
      "learning_rate": 7.807425358750687e-06,
      "loss": 0.5799,
      "step": 3872
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6403635870790662,
      "learning_rate": 7.80266459421752e-06,
      "loss": 0.6089,
      "step": 3873
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6956951735366639,
      "learning_rate": 7.797904352916174e-06,
      "loss": 0.5996,
      "step": 3874
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.654337170576096,
      "learning_rate": 7.79314463598016e-06,
      "loss": 0.5546,
      "step": 3875
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5748169412165235,
      "learning_rate": 7.78838544454286e-06,
      "loss": 0.5615,
      "step": 3876
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6224240157852681,
      "learning_rate": 7.783626779737552e-06,
      "loss": 0.5996,
      "step": 3877
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5652961212693312,
      "learning_rate": 7.778868642697359e-06,
      "loss": 0.5718,
      "step": 3878
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.668109445149047,
      "learning_rate": 7.774111034555308e-06,
      "loss": 0.5841,
      "step": 3879
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6272867189107145,
      "learning_rate": 7.769353956444275e-06,
      "loss": 0.5778,
      "step": 3880
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5991387753955366,
      "learning_rate": 7.764597409497023e-06,
      "loss": 0.5824,
      "step": 3881
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6169396603549058,
      "learning_rate": 7.759841394846187e-06,
      "loss": 0.5951,
      "step": 3882
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6895810056703027,
      "learning_rate": 7.755085913624274e-06,
      "loss": 0.5431,
      "step": 3883
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6321325343642314,
      "learning_rate": 7.750330966963666e-06,
      "loss": 0.556,
      "step": 3884
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5719301872546042,
      "learning_rate": 7.745576555996615e-06,
      "loss": 0.5758,
      "step": 3885
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5952964563707136,
      "learning_rate": 7.740822681855242e-06,
      "loss": 0.5646,
      "step": 3886
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6128790827156083,
      "learning_rate": 7.736069345671551e-06,
      "loss": 0.5788,
      "step": 3887
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6785314335583275,
      "learning_rate": 7.731316548577406e-06,
      "loss": 0.5757,
      "step": 3888
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6569069235343822,
      "learning_rate": 7.726564291704552e-06,
      "loss": 0.5635,
      "step": 3889
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6287405027777693,
      "learning_rate": 7.721812576184603e-06,
      "loss": 0.6051,
      "step": 3890
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6086164604765318,
      "learning_rate": 7.717061403149034e-06,
      "loss": 0.566,
      "step": 3891
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7005081549151572,
      "learning_rate": 7.712310773729208e-06,
      "loss": 0.6067,
      "step": 3892
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6030019577370451,
      "learning_rate": 7.707560689056343e-06,
      "loss": 0.5751,
      "step": 3893
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6098517331635461,
      "learning_rate": 7.702811150261543e-06,
      "loss": 0.5613,
      "step": 3894
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6493553543547256,
      "learning_rate": 7.698062158475766e-06,
      "loss": 0.5987,
      "step": 3895
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6223534041665237,
      "learning_rate": 7.693313714829846e-06,
      "loss": 0.5668,
      "step": 3896
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6371407292658582,
      "learning_rate": 7.688565820454492e-06,
      "loss": 0.5862,
      "step": 3897
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5942040347223029,
      "learning_rate": 7.683818476480275e-06,
      "loss": 0.584,
      "step": 3898
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6878456985871562,
      "learning_rate": 7.67907168403764e-06,
      "loss": 0.6124,
      "step": 3899
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6550605512972102,
      "learning_rate": 7.674325444256899e-06,
      "loss": 0.6157,
      "step": 3900
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6387513636788913,
      "learning_rate": 7.669579758268222e-06,
      "loss": 0.6028,
      "step": 3901
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6379774006273894,
      "learning_rate": 7.664834627201671e-06,
      "loss": 0.5802,
      "step": 3902
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5998626056771449,
      "learning_rate": 7.660090052187149e-06,
      "loss": 0.5771,
      "step": 3903
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6995264486552721,
      "learning_rate": 7.655346034354449e-06,
      "loss": 0.5926,
      "step": 3904
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5802654361476869,
      "learning_rate": 7.650602574833217e-06,
      "loss": 0.572,
      "step": 3905
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6159638134153432,
      "learning_rate": 7.645859674752968e-06,
      "loss": 0.5802,
      "step": 3906
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6644107494330949,
      "learning_rate": 7.641117335243091e-06,
      "loss": 0.5874,
      "step": 3907
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5888222046882786,
      "learning_rate": 7.636375557432835e-06,
      "loss": 0.5763,
      "step": 3908
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6520035671079388,
      "learning_rate": 7.631634342451315e-06,
      "loss": 0.5859,
      "step": 3909
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5843435648270475,
      "learning_rate": 7.62689369142752e-06,
      "loss": 0.5686,
      "step": 3910
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5862579765537527,
      "learning_rate": 7.622153605490292e-06,
      "loss": 0.5609,
      "step": 3911
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6229824351369655,
      "learning_rate": 7.617414085768352e-06,
      "loss": 0.595,
      "step": 3912
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6230196568533692,
      "learning_rate": 7.612675133390275e-06,
      "loss": 0.5631,
      "step": 3913
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6490319082572127,
      "learning_rate": 7.607936749484503e-06,
      "loss": 0.5652,
      "step": 3914
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6035133601101587,
      "learning_rate": 7.603198935179349e-06,
      "loss": 0.5855,
      "step": 3915
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6489380611688261,
      "learning_rate": 7.598461691602984e-06,
      "loss": 0.5602,
      "step": 3916
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6003892417202256,
      "learning_rate": 7.593725019883449e-06,
      "loss": 0.5691,
      "step": 3917
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.57066662795649,
      "learning_rate": 7.588988921148644e-06,
      "loss": 0.5717,
      "step": 3918
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6177556399928141,
      "learning_rate": 7.58425339652633e-06,
      "loss": 0.5766,
      "step": 3919
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6137373073533755,
      "learning_rate": 7.579518447144139e-06,
      "loss": 0.5832,
      "step": 3920
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6212167041947173,
      "learning_rate": 7.57478407412956e-06,
      "loss": 0.6038,
      "step": 3921
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6321051112577365,
      "learning_rate": 7.570050278609951e-06,
      "loss": 0.6303,
      "step": 3922
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5963282144361179,
      "learning_rate": 7.565317061712525e-06,
      "loss": 0.5854,
      "step": 3923
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6246985449813233,
      "learning_rate": 7.560584424564356e-06,
      "loss": 0.6122,
      "step": 3924
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7134976468368948,
      "learning_rate": 7.555852368292396e-06,
      "loss": 0.5877,
      "step": 3925
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6610857056035545,
      "learning_rate": 7.551120894023433e-06,
      "loss": 0.5955,
      "step": 3926
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6390251260064839,
      "learning_rate": 7.546390002884147e-06,
      "loss": 0.5535,
      "step": 3927
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.669517480871212,
      "learning_rate": 7.54165969600105e-06,
      "loss": 0.5836,
      "step": 3928
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5987062893152436,
      "learning_rate": 7.53692997450053e-06,
      "loss": 0.5705,
      "step": 3929
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7038965486712053,
      "learning_rate": 7.532200839508838e-06,
      "loss": 0.5737,
      "step": 3930
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5723688520117074,
      "learning_rate": 7.527472292152074e-06,
      "loss": 0.579,
      "step": 3931
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6098737073330212,
      "learning_rate": 7.522744333556211e-06,
      "loss": 0.5694,
      "step": 3932
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6460352321112816,
      "learning_rate": 7.5180169648470744e-06,
      "loss": 0.5575,
      "step": 3933
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5860829508848729,
      "learning_rate": 7.513290187150343e-06,
      "loss": 0.5856,
      "step": 3934
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6600190874052676,
      "learning_rate": 7.508564001591573e-06,
      "loss": 0.6266,
      "step": 3935
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5845294728197783,
      "learning_rate": 7.503838409296162e-06,
      "loss": 0.5501,
      "step": 3936
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6688678032222872,
      "learning_rate": 7.499113411389371e-06,
      "loss": 0.5843,
      "step": 3937
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5914020075480483,
      "learning_rate": 7.494389008996328e-06,
      "loss": 0.5788,
      "step": 3938
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5605140104512465,
      "learning_rate": 7.489665203242007e-06,
      "loss": 0.5866,
      "step": 3939
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.645820148299109,
      "learning_rate": 7.4849419952512495e-06,
      "loss": 0.5957,
      "step": 3940
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6088713730521849,
      "learning_rate": 7.480219386148751e-06,
      "loss": 0.551,
      "step": 3941
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5714554483919789,
      "learning_rate": 7.475497377059058e-06,
      "loss": 0.584,
      "step": 3942
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6661972177798132,
      "learning_rate": 7.470775969106587e-06,
      "loss": 0.587,
      "step": 3943
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6175364325198348,
      "learning_rate": 7.466055163415602e-06,
      "loss": 0.5903,
      "step": 3944
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.728089428532739,
      "learning_rate": 7.461334961110227e-06,
      "loss": 0.6184,
      "step": 3945
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6374808001107573,
      "learning_rate": 7.456615363314445e-06,
      "loss": 0.6033,
      "step": 3946
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6157469929668986,
      "learning_rate": 7.45189637115208e-06,
      "loss": 0.5682,
      "step": 3947
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.62078304842222,
      "learning_rate": 7.447177985746837e-06,
      "loss": 0.5865,
      "step": 3948
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5879420504540634,
      "learning_rate": 7.442460208222253e-06,
      "loss": 0.5761,
      "step": 3949
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6525983172641716,
      "learning_rate": 7.437743039701737e-06,
      "loss": 0.5687,
      "step": 3950
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6325563208268927,
      "learning_rate": 7.433026481308543e-06,
      "loss": 0.5764,
      "step": 3951
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.509882949413642,
      "learning_rate": 7.428310534165779e-06,
      "loss": 0.5659,
      "step": 3952
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6693808928333139,
      "learning_rate": 7.4235951993964195e-06,
      "loss": 0.604,
      "step": 3953
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5526047658503344,
      "learning_rate": 7.418880478123278e-06,
      "loss": 0.5652,
      "step": 3954
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.635540151139767,
      "learning_rate": 7.414166371469034e-06,
      "loss": 0.5958,
      "step": 3955
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5770888442552126,
      "learning_rate": 7.409452880556215e-06,
      "loss": 0.574,
      "step": 3956
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6588923310635465,
      "learning_rate": 7.404740006507194e-06,
      "loss": 0.5686,
      "step": 3957
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5699189452625469,
      "learning_rate": 7.40002775044422e-06,
      "loss": 0.5907,
      "step": 3958
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5713842613414672,
      "learning_rate": 7.3953161134893644e-06,
      "loss": 0.5432,
      "step": 3959
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6348944947902022,
      "learning_rate": 7.390605096764583e-06,
      "loss": 0.5933,
      "step": 3960
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6397625640406706,
      "learning_rate": 7.385894701391658e-06,
      "loss": 0.5592,
      "step": 3961
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5773994290950221,
      "learning_rate": 7.381184928492233e-06,
      "loss": 0.5789,
      "step": 3962
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.650044778167239,
      "learning_rate": 7.376475779187811e-06,
      "loss": 0.5832,
      "step": 3963
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.618353986125613,
      "learning_rate": 7.371767254599731e-06,
      "loss": 0.5823,
      "step": 3964
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5468824050656436,
      "learning_rate": 7.3670593558492e-06,
      "loss": 0.5909,
      "step": 3965
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7070864577937593,
      "learning_rate": 7.362352084057265e-06,
      "loss": 0.6039,
      "step": 3966
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5994037240171944,
      "learning_rate": 7.357645440344823e-06,
      "loss": 0.5806,
      "step": 3967
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6705763553948412,
      "learning_rate": 7.3529394258326305e-06,
      "loss": 0.6207,
      "step": 3968
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5887073042504971,
      "learning_rate": 7.3482340416412865e-06,
      "loss": 0.5799,
      "step": 3969
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7268702954082301,
      "learning_rate": 7.343529288891239e-06,
      "loss": 0.5866,
      "step": 3970
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5600016195213661,
      "learning_rate": 7.338825168702792e-06,
      "loss": 0.5794,
      "step": 3971
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6456803777155153,
      "learning_rate": 7.3341216821960935e-06,
      "loss": 0.5588,
      "step": 3972
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5664042847887434,
      "learning_rate": 7.329418830491145e-06,
      "loss": 0.5916,
      "step": 3973
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6921944707039305,
      "learning_rate": 7.324716614707794e-06,
      "loss": 0.6067,
      "step": 3974
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.627520462863688,
      "learning_rate": 7.320015035965734e-06,
      "loss": 0.5756,
      "step": 3975
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.644057603060079,
      "learning_rate": 7.315314095384515e-06,
      "loss": 0.5819,
      "step": 3976
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.583789188981561,
      "learning_rate": 7.310613794083524e-06,
      "loss": 0.581,
      "step": 3977
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6351487975834307,
      "learning_rate": 7.305914133182008e-06,
      "loss": 0.587,
      "step": 3978
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6746256999020347,
      "learning_rate": 7.301215113799054e-06,
      "loss": 0.5714,
      "step": 3979
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9394542922036817,
      "learning_rate": 7.296516737053587e-06,
      "loss": 0.6237,
      "step": 3980
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6211000842843705,
      "learning_rate": 7.291819004064407e-06,
      "loss": 0.5845,
      "step": 3981
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6750906993817274,
      "learning_rate": 7.287121915950126e-06,
      "loss": 0.5879,
      "step": 3982
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6009930232960378,
      "learning_rate": 7.282425473829235e-06,
      "loss": 0.5422,
      "step": 3983
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6276484308998088,
      "learning_rate": 7.2777296788200466e-06,
      "loss": 0.5881,
      "step": 3984
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5922267650118582,
      "learning_rate": 7.273034532040727e-06,
      "loss": 0.5837,
      "step": 3985
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6605509505796787,
      "learning_rate": 7.268340034609296e-06,
      "loss": 0.5776,
      "step": 3986
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6056379938246617,
      "learning_rate": 7.263646187643605e-06,
      "loss": 0.5694,
      "step": 3987
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.828283562648704,
      "learning_rate": 7.258952992261366e-06,
      "loss": 0.601,
      "step": 3988
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6383708046647574,
      "learning_rate": 7.254260449580122e-06,
      "loss": 0.5584,
      "step": 3989
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5920714103398301,
      "learning_rate": 7.249568560717264e-06,
      "loss": 0.5956,
      "step": 3990
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5935906917213934,
      "learning_rate": 7.244877326790039e-06,
      "loss": 0.5718,
      "step": 3991
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6983044207453284,
      "learning_rate": 7.240186748915517e-06,
      "loss": 0.5869,
      "step": 3992
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6189357270513877,
      "learning_rate": 7.235496828210633e-06,
      "loss": 0.5816,
      "step": 3993
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5920810698629729,
      "learning_rate": 7.230807565792151e-06,
      "loss": 0.5484,
      "step": 3994
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6285809477929963,
      "learning_rate": 7.226118962776683e-06,
      "loss": 0.5554,
      "step": 3995
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6057700949368632,
      "learning_rate": 7.221431020280685e-06,
      "loss": 0.5664,
      "step": 3996
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5646666405870414,
      "learning_rate": 7.216743739420453e-06,
      "loss": 0.586,
      "step": 3997
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6089581473339097,
      "learning_rate": 7.212057121312133e-06,
      "loss": 0.5725,
      "step": 3998
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5723100733282868,
      "learning_rate": 7.2073711670717e-06,
      "loss": 0.5876,
      "step": 3999
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6065241646081824,
      "learning_rate": 7.202685877814982e-06,
      "loss": 0.5896,
      "step": 4000
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6130976348167237,
      "learning_rate": 7.1980012546576474e-06,
      "loss": 0.5978,
      "step": 4001
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6511111059132856,
      "learning_rate": 7.193317298715201e-06,
      "loss": 0.6047,
      "step": 4002
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5820209372139049,
      "learning_rate": 7.188634011102987e-06,
      "loss": 0.5528,
      "step": 4003
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6388098312048793,
      "learning_rate": 7.183951392936205e-06,
      "loss": 0.5729,
      "step": 4004
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6658225281419454,
      "learning_rate": 7.179269445329874e-06,
      "loss": 0.5575,
      "step": 4005
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6104561033033974,
      "learning_rate": 7.1745881693988725e-06,
      "loss": 0.5876,
      "step": 4006
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5792232734760108,
      "learning_rate": 7.169907566257907e-06,
      "loss": 0.5726,
      "step": 4007
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.614421638119897,
      "learning_rate": 7.1652276370215254e-06,
      "loss": 0.5412,
      "step": 4008
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6313938033511802,
      "learning_rate": 7.160548382804125e-06,
      "loss": 0.5793,
      "step": 4009
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6166972491156804,
      "learning_rate": 7.155869804719925e-06,
      "loss": 0.5842,
      "step": 4010
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6497971461911898,
      "learning_rate": 7.1511919038830016e-06,
      "loss": 0.5716,
      "step": 4011
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6350237975062664,
      "learning_rate": 7.1465146814072605e-06,
      "loss": 0.5624,
      "step": 4012
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5865312817246339,
      "learning_rate": 7.141838138406439e-06,
      "loss": 0.5587,
      "step": 4013
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6355718146011456,
      "learning_rate": 7.137162275994131e-06,
      "loss": 0.5712,
      "step": 4014
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5562810330616826,
      "learning_rate": 7.132487095283747e-06,
      "loss": 0.5852,
      "step": 4015
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5832516659051322,
      "learning_rate": 7.127812597388557e-06,
      "loss": 0.5653,
      "step": 4016
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6441679901737033,
      "learning_rate": 7.1231387834216515e-06,
      "loss": 0.5871,
      "step": 4017
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6710228405598927,
      "learning_rate": 7.118465654495962e-06,
      "loss": 0.598,
      "step": 4018
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6534451530637967,
      "learning_rate": 7.113793211724263e-06,
      "loss": 0.605,
      "step": 4019
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6673990569394233,
      "learning_rate": 7.109121456219159e-06,
      "loss": 0.6086,
      "step": 4020
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6214038708316642,
      "learning_rate": 7.104450389093095e-06,
      "loss": 0.5489,
      "step": 4021
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6966345534147025,
      "learning_rate": 7.0997800114583505e-06,
      "loss": 0.6059,
      "step": 4022
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6180550859264514,
      "learning_rate": 7.0951103244270366e-06,
      "loss": 0.5752,
      "step": 4023
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6132555427909984,
      "learning_rate": 7.090441329111111e-06,
      "loss": 0.585,
      "step": 4024
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6030904861057237,
      "learning_rate": 7.085773026622353e-06,
      "loss": 0.5744,
      "step": 4025
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6080292043591302,
      "learning_rate": 7.081105418072392e-06,
      "loss": 0.5568,
      "step": 4026
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6244936650100242,
      "learning_rate": 7.076438504572677e-06,
      "loss": 0.5786,
      "step": 4027
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6261678649837082,
      "learning_rate": 7.071772287234497e-06,
      "loss": 0.5817,
      "step": 4028
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7183795771373733,
      "learning_rate": 7.067106767168983e-06,
      "loss": 0.6201,
      "step": 4029
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6641769259603765,
      "learning_rate": 7.062441945487088e-06,
      "loss": 0.5831,
      "step": 4030
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5961007608444461,
      "learning_rate": 7.057777823299607e-06,
      "loss": 0.5923,
      "step": 4031
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6135405907127552,
      "learning_rate": 7.0531144017171645e-06,
      "loss": 0.5735,
      "step": 4032
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6682645833810501,
      "learning_rate": 7.04845168185022e-06,
      "loss": 0.5581,
      "step": 4033
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6493423972521397,
      "learning_rate": 7.043789664809066e-06,
      "loss": 0.574,
      "step": 4034
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5859106798793807,
      "learning_rate": 7.03912835170383e-06,
      "loss": 0.5819,
      "step": 4035
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5903401537365597,
      "learning_rate": 7.034467743644457e-06,
      "loss": 0.5574,
      "step": 4036
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6140793310765635,
      "learning_rate": 7.0298078417407515e-06,
      "loss": 0.6052,
      "step": 4037
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6435402940625101,
      "learning_rate": 7.025148647102319e-06,
      "loss": 0.5648,
      "step": 4038
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5946891899143749,
      "learning_rate": 7.020490160838627e-06,
      "loss": 0.5665,
      "step": 4039
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6499118686407519,
      "learning_rate": 7.015832384058948e-06,
      "loss": 0.5827,
      "step": 4040
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6498429289478108,
      "learning_rate": 7.011175317872398e-06,
      "loss": 0.5778,
      "step": 4041
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.615269144748512,
      "learning_rate": 7.006518963387928e-06,
      "loss": 0.5711,
      "step": 4042
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5861900115708337,
      "learning_rate": 7.001863321714308e-06,
      "loss": 0.5952,
      "step": 4043
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5630371955290596,
      "learning_rate": 6.99720839396015e-06,
      "loss": 0.5441,
      "step": 4044
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5666708330621734,
      "learning_rate": 6.992554181233886e-06,
      "loss": 0.544,
      "step": 4045
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6379026467250031,
      "learning_rate": 6.987900684643782e-06,
      "loss": 0.5624,
      "step": 4046
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6202121781262382,
      "learning_rate": 6.98324790529794e-06,
      "loss": 0.5718,
      "step": 4047
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6382075685953189,
      "learning_rate": 6.978595844304272e-06,
      "loss": 0.5917,
      "step": 4048
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6483683826220277,
      "learning_rate": 6.973944502770547e-06,
      "loss": 0.577,
      "step": 4049
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6275877556745876,
      "learning_rate": 6.969293881804335e-06,
      "loss": 0.568,
      "step": 4050
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6152570497107422,
      "learning_rate": 6.964643982513052e-06,
      "loss": 0.565,
      "step": 4051
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5626079391848111,
      "learning_rate": 6.9599948060039364e-06,
      "loss": 0.5787,
      "step": 4052
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6449610097035647,
      "learning_rate": 6.955346353384054e-06,
      "loss": 0.5791,
      "step": 4053
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6172805332441914,
      "learning_rate": 6.950698625760302e-06,
      "loss": 0.5618,
      "step": 4054
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.630993866893239,
      "learning_rate": 6.9460516242394e-06,
      "loss": 0.5841,
      "step": 4055
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6495543821376322,
      "learning_rate": 6.941405349927894e-06,
      "loss": 0.5758,
      "step": 4056
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6329003502392853,
      "learning_rate": 6.936759803932167e-06,
      "loss": 0.5681,
      "step": 4057
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6080362970611574,
      "learning_rate": 6.932114987358414e-06,
      "loss": 0.5673,
      "step": 4058
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6418870756685607,
      "learning_rate": 6.927470901312668e-06,
      "loss": 0.574,
      "step": 4059
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6264961719468167,
      "learning_rate": 6.9228275469007854e-06,
      "loss": 0.5747,
      "step": 4060
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6282419068053996,
      "learning_rate": 6.918184925228438e-06,
      "loss": 0.5917,
      "step": 4061
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5733034609148536,
      "learning_rate": 6.913543037401144e-06,
      "loss": 0.5718,
      "step": 4062
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6354279095729671,
      "learning_rate": 6.908901884524225e-06,
      "loss": 0.5779,
      "step": 4063
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.6905705664945074,
      "learning_rate": 6.904261467702838e-06,
      "loss": 0.6319,
      "step": 4064
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.62180127592362,
      "learning_rate": 6.8996217880419694e-06,
      "loss": 0.5605,
      "step": 4065
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5632649146971046,
      "learning_rate": 6.894982846646419e-06,
      "loss": 0.5568,
      "step": 4066
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6192778138438205,
      "learning_rate": 6.890344644620819e-06,
      "loss": 0.5642,
      "step": 4067
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6353718931751433,
      "learning_rate": 6.885707183069627e-06,
      "loss": 0.5654,
      "step": 4068
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5560530642543577,
      "learning_rate": 6.881070463097108e-06,
      "loss": 0.5608,
      "step": 4069
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6222588219816889,
      "learning_rate": 6.8764344858073795e-06,
      "loss": 0.5855,
      "step": 4070
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7741384401316311,
      "learning_rate": 6.871799252304348e-06,
      "loss": 0.5979,
      "step": 4071
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6128206628166601,
      "learning_rate": 6.867164763691775e-06,
      "loss": 0.5912,
      "step": 4072
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6422410074283826,
      "learning_rate": 6.862531021073222e-06,
      "loss": 0.5693,
      "step": 4073
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6188366175984995,
      "learning_rate": 6.857898025552079e-06,
      "loss": 0.5926,
      "step": 4074
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6408485736011817,
      "learning_rate": 6.8532657782315655e-06,
      "loss": 0.5443,
      "step": 4075
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6143015037074813,
      "learning_rate": 6.8486342802147125e-06,
      "loss": 0.5996,
      "step": 4076
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.586846850895715,
      "learning_rate": 6.844003532604381e-06,
      "loss": 0.5837,
      "step": 4077
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6278704330645949,
      "learning_rate": 6.839373536503247e-06,
      "loss": 0.5908,
      "step": 4078
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6487219216532836,
      "learning_rate": 6.834744293013809e-06,
      "loss": 0.5858,
      "step": 4079
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6442225344681554,
      "learning_rate": 6.8301158032383905e-06,
      "loss": 0.5871,
      "step": 4080
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6347708201606905,
      "learning_rate": 6.82548806827913e-06,
      "loss": 0.5529,
      "step": 4081
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7884421338773271,
      "learning_rate": 6.820861089237991e-06,
      "loss": 0.5752,
      "step": 4082
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.646683025964278,
      "learning_rate": 6.816234867216755e-06,
      "loss": 0.5928,
      "step": 4083
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6704462683361881,
      "learning_rate": 6.811609403317017e-06,
      "loss": 0.5886,
      "step": 4084
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6408387163587916,
      "learning_rate": 6.806984698640202e-06,
      "loss": 0.5897,
      "step": 4085
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.665575879470817,
      "learning_rate": 6.802360754287548e-06,
      "loss": 0.5636,
      "step": 4086
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6305963796244536,
      "learning_rate": 6.797737571360115e-06,
      "loss": 0.5942,
      "step": 4087
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.603864213974122,
      "learning_rate": 6.7931151509587815e-06,
      "loss": 0.5679,
      "step": 4088
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6811136956953441,
      "learning_rate": 6.788493494184237e-06,
      "loss": 0.6097,
      "step": 4089
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6126549134885437,
      "learning_rate": 6.783872602137002e-06,
      "loss": 0.578,
      "step": 4090
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6680122589455656,
      "learning_rate": 6.779252475917408e-06,
      "loss": 0.5872,
      "step": 4091
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5466148250988206,
      "learning_rate": 6.774633116625595e-06,
      "loss": 0.5786,
      "step": 4092
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6097070713425822,
      "learning_rate": 6.770014525361544e-06,
      "loss": 0.5772,
      "step": 4093
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5928722734654113,
      "learning_rate": 6.765396703225024e-06,
      "loss": 0.5549,
      "step": 4094
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.541113381846613,
      "learning_rate": 6.760779651315649e-06,
      "loss": 0.5844,
      "step": 4095
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6631591070383956,
      "learning_rate": 6.756163370732828e-06,
      "loss": 0.6123,
      "step": 4096
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6506381684542611,
      "learning_rate": 6.751547862575796e-06,
      "loss": 0.5879,
      "step": 4097
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.801380112070296,
      "learning_rate": 6.746933127943604e-06,
      "loss": 0.5493,
      "step": 4098
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5956663214200167,
      "learning_rate": 6.7423191679351155e-06,
      "loss": 0.5639,
      "step": 4099
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5907162040894053,
      "learning_rate": 6.7377059836490145e-06,
      "loss": 0.577,
      "step": 4100
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6121078469164315,
      "learning_rate": 6.7330935761837955e-06,
      "loss": 0.5494,
      "step": 4101
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5990572862132371,
      "learning_rate": 6.7284819466377705e-06,
      "loss": 0.5758,
      "step": 4102
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6681030961924517,
      "learning_rate": 6.723871096109065e-06,
      "loss": 0.5945,
      "step": 4103
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6329013842556088,
      "learning_rate": 6.719261025695618e-06,
      "loss": 0.5698,
      "step": 4104
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6140491885293425,
      "learning_rate": 6.71465173649519e-06,
      "loss": 0.5795,
      "step": 4105
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.693131850407786,
      "learning_rate": 6.710043229605343e-06,
      "loss": 0.5763,
      "step": 4106
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6043875800304864,
      "learning_rate": 6.705435506123461e-06,
      "loss": 0.5553,
      "step": 4107
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6332317414750538,
      "learning_rate": 6.7008285671467445e-06,
      "loss": 0.578,
      "step": 4108
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6181022205283278,
      "learning_rate": 6.696222413772195e-06,
      "loss": 0.5576,
      "step": 4109
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5571664375935066,
      "learning_rate": 6.691617047096643e-06,
      "loss": 0.5671,
      "step": 4110
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5906650122649051,
      "learning_rate": 6.687012468216719e-06,
      "loss": 0.5698,
      "step": 4111
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9417066845239648,
      "learning_rate": 6.682408678228869e-06,
      "loss": 0.5916,
      "step": 4112
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6198081116459834,
      "learning_rate": 6.677805678229354e-06,
      "loss": 0.5623,
      "step": 4113
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6070514886872327,
      "learning_rate": 6.673203469314245e-06,
      "loss": 0.5807,
      "step": 4114
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6999143877044393,
      "learning_rate": 6.668602052579425e-06,
      "loss": 0.5911,
      "step": 4115
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6511515712163249,
      "learning_rate": 6.6640014291205904e-06,
      "loss": 0.5787,
      "step": 4116
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6316942516356926,
      "learning_rate": 6.659401600033238e-06,
      "loss": 0.5764,
      "step": 4117
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6089178468373215,
      "learning_rate": 6.654802566412697e-06,
      "loss": 0.5801,
      "step": 4118
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6765990391136474,
      "learning_rate": 6.6502043293540795e-06,
      "loss": 0.6182,
      "step": 4119
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6115932526534513,
      "learning_rate": 6.645606889952337e-06,
      "loss": 0.5988,
      "step": 4120
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5659590380110584,
      "learning_rate": 6.6410102493022085e-06,
      "loss": 0.5763,
      "step": 4121
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6265663319618372,
      "learning_rate": 6.636414408498249e-06,
      "loss": 0.6,
      "step": 4122
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5671797736529789,
      "learning_rate": 6.631819368634831e-06,
      "loss": 0.5524,
      "step": 4123
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6149697154450671,
      "learning_rate": 6.627225130806127e-06,
      "loss": 0.5647,
      "step": 4124
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.580574311322042,
      "learning_rate": 6.62263169610612e-06,
      "loss": 0.5986,
      "step": 4125
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5779595980491128,
      "learning_rate": 6.61803906562861e-06,
      "loss": 0.5779,
      "step": 4126
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6530716316019038,
      "learning_rate": 6.613447240467188e-06,
      "loss": 0.5936,
      "step": 4127
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7420233124553606,
      "learning_rate": 6.608856221715278e-06,
      "loss": 0.5879,
      "step": 4128
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6179758990095214,
      "learning_rate": 6.604266010466087e-06,
      "loss": 0.5804,
      "step": 4129
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6823123730600077,
      "learning_rate": 6.599676607812645e-06,
      "loss": 0.6003,
      "step": 4130
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6184041468142231,
      "learning_rate": 6.595088014847786e-06,
      "loss": 0.5602,
      "step": 4131
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6312900413938324,
      "learning_rate": 6.590500232664149e-06,
      "loss": 0.5898,
      "step": 4132
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2793356698777207,
      "learning_rate": 6.585913262354184e-06,
      "loss": 0.5771,
      "step": 4133
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.673069581811607,
      "learning_rate": 6.581327105010144e-06,
      "loss": 0.8248,
      "step": 4134
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6262065469844061,
      "learning_rate": 6.5767417617240855e-06,
      "loss": 0.5633,
      "step": 4135
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5723217219231762,
      "learning_rate": 6.572157233587884e-06,
      "loss": 0.5629,
      "step": 4136
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6304001325779831,
      "learning_rate": 6.567573521693204e-06,
      "loss": 0.5846,
      "step": 4137
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7092869483921308,
      "learning_rate": 6.562990627131528e-06,
      "loss": 0.5731,
      "step": 4138
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7765604519509387,
      "learning_rate": 6.558408550994142e-06,
      "loss": 0.574,
      "step": 4139
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6345338408039172,
      "learning_rate": 6.5538272943721235e-06,
      "loss": 0.5787,
      "step": 4140
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6729545832637657,
      "learning_rate": 6.549246858356382e-06,
      "loss": 0.5825,
      "step": 4141
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6283172982603852,
      "learning_rate": 6.5446672440376036e-06,
      "loss": 0.5553,
      "step": 4142
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7989447751168213,
      "learning_rate": 6.5400884525062965e-06,
      "loss": 0.6333,
      "step": 4143
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.613428380189208,
      "learning_rate": 6.535510484852767e-06,
      "loss": 0.5579,
      "step": 4144
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6327150479482821,
      "learning_rate": 6.530933342167122e-06,
      "loss": 0.5563,
      "step": 4145
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6596824796382748,
      "learning_rate": 6.52635702553928e-06,
      "loss": 0.5807,
      "step": 4146
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.671347560960846,
      "learning_rate": 6.521781536058954e-06,
      "loss": 0.5998,
      "step": 4147
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6734620928803513,
      "learning_rate": 6.517206874815669e-06,
      "loss": 0.5998,
      "step": 4148
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6326686109520508,
      "learning_rate": 6.512633042898748e-06,
      "loss": 0.5839,
      "step": 4149
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6536291968011225,
      "learning_rate": 6.508060041397308e-06,
      "loss": 0.5957,
      "step": 4150
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6022552063261929,
      "learning_rate": 6.503487871400291e-06,
      "loss": 0.5985,
      "step": 4151
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6072050700593212,
      "learning_rate": 6.4989165339964134e-06,
      "loss": 0.5763,
      "step": 4152
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6173731868378253,
      "learning_rate": 6.494346030274218e-06,
      "loss": 0.5911,
      "step": 4153
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6946595799723855,
      "learning_rate": 6.489776361322031e-06,
      "loss": 0.5905,
      "step": 4154
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6303692884876879,
      "learning_rate": 6.485207528227986e-06,
      "loss": 0.5847,
      "step": 4155
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6822311884270404,
      "learning_rate": 6.480639532080024e-06,
      "loss": 0.5914,
      "step": 4156
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5970740641811793,
      "learning_rate": 6.476072373965878e-06,
      "loss": 0.5822,
      "step": 4157
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6810548171321016,
      "learning_rate": 6.471506054973082e-06,
      "loss": 0.586,
      "step": 4158
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.657721153566595,
      "learning_rate": 6.466940576188978e-06,
      "loss": 0.5872,
      "step": 4159
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5840869435201954,
      "learning_rate": 6.462375938700697e-06,
      "loss": 0.5577,
      "step": 4160
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6309733004380956,
      "learning_rate": 6.457812143595181e-06,
      "loss": 0.5546,
      "step": 4161
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6091947025260626,
      "learning_rate": 6.453249191959164e-06,
      "loss": 0.5785,
      "step": 4162
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5843376821855875,
      "learning_rate": 6.448687084879175e-06,
      "loss": 0.5673,
      "step": 4163
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.729907513052622,
      "learning_rate": 6.444125823441556e-06,
      "loss": 0.5637,
      "step": 4164
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5272920505288355,
      "learning_rate": 6.439565408732434e-06,
      "loss": 0.5685,
      "step": 4165
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6014976712306674,
      "learning_rate": 6.4350058418377415e-06,
      "loss": 0.5642,
      "step": 4166
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6876899445371008,
      "learning_rate": 6.430447123843209e-06,
      "loss": 0.5463,
      "step": 4167
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5615541081627863,
      "learning_rate": 6.425889255834357e-06,
      "loss": 0.5993,
      "step": 4168
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5506430841154839,
      "learning_rate": 6.421332238896519e-06,
      "loss": 0.5482,
      "step": 4169
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5893925181848663,
      "learning_rate": 6.41677607411481e-06,
      "loss": 0.574,
      "step": 4170
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5922581910342908,
      "learning_rate": 6.412220762574152e-06,
      "loss": 0.5828,
      "step": 4171
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6824276263076088,
      "learning_rate": 6.407666305359261e-06,
      "loss": 0.6052,
      "step": 4172
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6834340899767409,
      "learning_rate": 6.403112703554643e-06,
      "loss": 0.5784,
      "step": 4173
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6580485254126499,
      "learning_rate": 6.398559958244615e-06,
      "loss": 0.617,
      "step": 4174
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6105858480192543,
      "learning_rate": 6.394008070513273e-06,
      "loss": 0.5828,
      "step": 4175
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6230944412355643,
      "learning_rate": 6.389457041444528e-06,
      "loss": 0.5479,
      "step": 4176
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0058699152351558,
      "learning_rate": 6.384906872122067e-06,
      "loss": 0.6653,
      "step": 4177
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6036196100090985,
      "learning_rate": 6.380357563629381e-06,
      "loss": 0.5709,
      "step": 4178
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5976744621714272,
      "learning_rate": 6.3758091170497614e-06,
      "loss": 0.5764,
      "step": 4179
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6190202838463034,
      "learning_rate": 6.3712615334662844e-06,
      "loss": 0.5705,
      "step": 4180
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6373111997179569,
      "learning_rate": 6.3667148139618294e-06,
      "loss": 0.5749,
      "step": 4181
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.606429147330575,
      "learning_rate": 6.362168959619066e-06,
      "loss": 0.5698,
      "step": 4182
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8620973198159823,
      "learning_rate": 6.357623971520452e-06,
      "loss": 0.5751,
      "step": 4183
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6357510509198793,
      "learning_rate": 6.353079850748254e-06,
      "loss": 0.5789,
      "step": 4184
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6089605793489422,
      "learning_rate": 6.348536598384513e-06,
      "loss": 0.5648,
      "step": 4185
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5451577623145814,
      "learning_rate": 6.343994215511078e-06,
      "loss": 0.5692,
      "step": 4186
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.620352879334212,
      "learning_rate": 6.3394527032095865e-06,
      "loss": 0.574,
      "step": 4187
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6426616107575542,
      "learning_rate": 6.334912062561466e-06,
      "loss": 0.5675,
      "step": 4188
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.594745352911965,
      "learning_rate": 6.330372294647942e-06,
      "loss": 0.5627,
      "step": 4189
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.667072888503124,
      "learning_rate": 6.325833400550026e-06,
      "loss": 0.5966,
      "step": 4190
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.656970192920483,
      "learning_rate": 6.321295381348522e-06,
      "loss": 0.5777,
      "step": 4191
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7224180157610645,
      "learning_rate": 6.316758238124033e-06,
      "loss": 0.5765,
      "step": 4192
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5752069766987418,
      "learning_rate": 6.3122219719569444e-06,
      "loss": 0.591,
      "step": 4193
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5754773817015263,
      "learning_rate": 6.307686583927442e-06,
      "loss": 0.5826,
      "step": 4194
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5800286174347591,
      "learning_rate": 6.303152075115494e-06,
      "loss": 0.5703,
      "step": 4195
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6794975483147263,
      "learning_rate": 6.298618446600856e-06,
      "loss": 0.5821,
      "step": 4196
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6956372861801923,
      "learning_rate": 6.294085699463093e-06,
      "loss": 0.5597,
      "step": 4197
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7588352017128936,
      "learning_rate": 6.289553834781537e-06,
      "loss": 0.6112,
      "step": 4198
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7582640115928612,
      "learning_rate": 6.285022853635325e-06,
      "loss": 0.5821,
      "step": 4199
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6236026381178462,
      "learning_rate": 6.280492757103379e-06,
      "loss": 0.5828,
      "step": 4200
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5832486672593242,
      "learning_rate": 6.2759635462644075e-06,
      "loss": 0.5825,
      "step": 4201
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6383499607864986,
      "learning_rate": 6.2714352221969155e-06,
      "loss": 0.5578,
      "step": 4202
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6395059807980966,
      "learning_rate": 6.266907785979187e-06,
      "loss": 0.5819,
      "step": 4203
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.666790041056347,
      "learning_rate": 6.2623812386893055e-06,
      "loss": 0.6061,
      "step": 4204
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5707474471395009,
      "learning_rate": 6.257855581405137e-06,
      "loss": 0.5876,
      "step": 4205
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6356012837569861,
      "learning_rate": 6.253330815204326e-06,
      "loss": 0.5818,
      "step": 4206
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5650026413298428,
      "learning_rate": 6.248806941164327e-06,
      "loss": 0.5704,
      "step": 4207
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6242752365619841,
      "learning_rate": 6.244283960362361e-06,
      "loss": 0.5989,
      "step": 4208
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6018491627505227,
      "learning_rate": 6.239761873875452e-06,
      "loss": 0.5784,
      "step": 4209
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6408476714535555,
      "learning_rate": 6.235240682780399e-06,
      "loss": 0.5825,
      "step": 4210
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5873548941625363,
      "learning_rate": 6.2307203881537926e-06,
      "loss": 0.5596,
      "step": 4211
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6012091846319502,
      "learning_rate": 6.2262009910720136e-06,
      "loss": 0.5767,
      "step": 4212
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5912165630423508,
      "learning_rate": 6.22168249261122e-06,
      "loss": 0.5669,
      "step": 4213
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5821314571794001,
      "learning_rate": 6.217164893847369e-06,
      "loss": 0.5778,
      "step": 4214
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6800437080642419,
      "learning_rate": 6.212648195856191e-06,
      "loss": 0.5903,
      "step": 4215
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5815650278530576,
      "learning_rate": 6.208132399713207e-06,
      "loss": 0.6018,
      "step": 4216
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5978042477038524,
      "learning_rate": 6.2036175064937245e-06,
      "loss": 0.5696,
      "step": 4217
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5920230604316864,
      "learning_rate": 6.199103517272837e-06,
      "loss": 0.5803,
      "step": 4218
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6252783211258425,
      "learning_rate": 6.194590433125414e-06,
      "loss": 0.6148,
      "step": 4219
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6013445896520504,
      "learning_rate": 6.1900782551261205e-06,
      "loss": 0.5588,
      "step": 4220
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6294100391008659,
      "learning_rate": 6.185566984349398e-06,
      "loss": 0.5974,
      "step": 4221
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5916306508098785,
      "learning_rate": 6.1810566218694786e-06,
      "loss": 0.552,
      "step": 4222
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6619104637835295,
      "learning_rate": 6.176547168760374e-06,
      "loss": 0.6063,
      "step": 4223
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6461601029620436,
      "learning_rate": 6.172038626095876e-06,
      "loss": 0.6072,
      "step": 4224
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7054656357378596,
      "learning_rate": 6.167530994949569e-06,
      "loss": 0.6016,
      "step": 4225
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5942691941030658,
      "learning_rate": 6.163024276394809e-06,
      "loss": 0.5833,
      "step": 4226
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6192555355776951,
      "learning_rate": 6.158518471504746e-06,
      "loss": 0.5958,
      "step": 4227
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7434525807599235,
      "learning_rate": 6.154013581352308e-06,
      "loss": 0.5857,
      "step": 4228
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5902778982814704,
      "learning_rate": 6.149509607010195e-06,
      "loss": 0.5732,
      "step": 4229
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5882082725134485,
      "learning_rate": 6.145006549550909e-06,
      "loss": 0.546,
      "step": 4230
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.623648012560824,
      "learning_rate": 6.140504410046712e-06,
      "loss": 0.5773,
      "step": 4231
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3764346968932561,
      "learning_rate": 6.13600318956967e-06,
      "loss": 0.6001,
      "step": 4232
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6165636682951797,
      "learning_rate": 6.131502889191611e-06,
      "loss": 0.5634,
      "step": 4233
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5973014002760412,
      "learning_rate": 6.12700350998415e-06,
      "loss": 0.5627,
      "step": 4234
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5603248953497331,
      "learning_rate": 6.122505053018688e-06,
      "loss": 0.5677,
      "step": 4235
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6749472519178001,
      "learning_rate": 6.1180075193664e-06,
      "loss": 0.5966,
      "step": 4236
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6519330838381859,
      "learning_rate": 6.113510910098246e-06,
      "loss": 0.5605,
      "step": 4237
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6298127233092945,
      "learning_rate": 6.109015226284961e-06,
      "loss": 0.587,
      "step": 4238
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5876851026879484,
      "learning_rate": 6.104520468997062e-06,
      "loss": 0.5615,
      "step": 4239
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6631476655319659,
      "learning_rate": 6.100026639304848e-06,
      "loss": 0.5876,
      "step": 4240
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5497647197883261,
      "learning_rate": 6.095533738278387e-06,
      "loss": 0.5841,
      "step": 4241
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5932178595079592,
      "learning_rate": 6.091041766987546e-06,
      "loss": 0.5717,
      "step": 4242
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7078673502271867,
      "learning_rate": 6.086550726501947e-06,
      "loss": 0.5738,
      "step": 4243
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.612430956941572,
      "learning_rate": 6.082060617891004e-06,
      "loss": 0.5686,
      "step": 4244
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5968218340258854,
      "learning_rate": 6.0775714422239085e-06,
      "loss": 0.5904,
      "step": 4245
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6080252483211376,
      "learning_rate": 6.073083200569625e-06,
      "loss": 0.5951,
      "step": 4246
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6008609389312585,
      "learning_rate": 6.068595893996903e-06,
      "loss": 0.5505,
      "step": 4247
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6044934247839927,
      "learning_rate": 6.0641095235742596e-06,
      "loss": 0.5734,
      "step": 4248
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.53069351647301,
      "learning_rate": 6.059624090369994e-06,
      "loss": 0.5854,
      "step": 4249
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5793221325717381,
      "learning_rate": 6.0551395954521864e-06,
      "loss": 0.5489,
      "step": 4250
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.64867718329413,
      "learning_rate": 6.050656039888689e-06,
      "loss": 0.5626,
      "step": 4251
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5895707959804565,
      "learning_rate": 6.0461734247471245e-06,
      "loss": 0.5563,
      "step": 4252
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6617797362886796,
      "learning_rate": 6.041691751094908e-06,
      "loss": 0.6093,
      "step": 4253
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5309179106602449,
      "learning_rate": 6.037211019999207e-06,
      "loss": 0.5847,
      "step": 4254
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6484913842943943,
      "learning_rate": 6.032731232526994e-06,
      "loss": 0.5925,
      "step": 4255
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5539370564381453,
      "learning_rate": 6.0282523897449895e-06,
      "loss": 0.5755,
      "step": 4256
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6558692225690412,
      "learning_rate": 6.023774492719702e-06,
      "loss": 0.584,
      "step": 4257
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6241978889517902,
      "learning_rate": 6.019297542517415e-06,
      "loss": 0.5434,
      "step": 4258
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6403257998332333,
      "learning_rate": 6.014821540204182e-06,
      "loss": 0.5665,
      "step": 4259
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6162296946377644,
      "learning_rate": 6.010346486845837e-06,
      "loss": 0.5674,
      "step": 4260
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.621414067884576,
      "learning_rate": 6.005872383507986e-06,
      "loss": 0.5613,
      "step": 4261
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6323399108740716,
      "learning_rate": 6.0013992312559955e-06,
      "loss": 0.598,
      "step": 4262
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.651600490946828,
      "learning_rate": 5.996927031155033e-06,
      "loss": 0.5628,
      "step": 4263
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7079476116782633,
      "learning_rate": 5.992455784270011e-06,
      "loss": 0.5956,
      "step": 4264
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5753274735132877,
      "learning_rate": 5.987985491665637e-06,
      "loss": 0.5694,
      "step": 4265
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6207229712388463,
      "learning_rate": 5.9835161544063746e-06,
      "loss": 0.5745,
      "step": 4266
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8009723923572675,
      "learning_rate": 5.979047773556469e-06,
      "loss": 0.5987,
      "step": 4267
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7605666900004916,
      "learning_rate": 5.974580350179938e-06,
      "loss": 0.5911,
      "step": 4268
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6641739013806596,
      "learning_rate": 5.970113885340566e-06,
      "loss": 0.5801,
      "step": 4269
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5734508842674214,
      "learning_rate": 5.965648380101916e-06,
      "loss": 0.5722,
      "step": 4270
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6008083783865612,
      "learning_rate": 5.961183835527317e-06,
      "loss": 0.5696,
      "step": 4271
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.675503945086961,
      "learning_rate": 5.956720252679869e-06,
      "loss": 0.5673,
      "step": 4272
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6419224154635005,
      "learning_rate": 5.952257632622448e-06,
      "loss": 0.5829,
      "step": 4273
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6053351677239636,
      "learning_rate": 5.947795976417694e-06,
      "loss": 0.5992,
      "step": 4274
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6856951842375871,
      "learning_rate": 5.943335285128027e-06,
      "loss": 0.5797,
      "step": 4275
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5809973627538684,
      "learning_rate": 5.938875559815631e-06,
      "loss": 0.5788,
      "step": 4276
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6589056322943072,
      "learning_rate": 5.934416801542453e-06,
      "loss": 0.5925,
      "step": 4277
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6283574967610425,
      "learning_rate": 5.929959011370226e-06,
      "loss": 0.5619,
      "step": 4278
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6048136949902778,
      "learning_rate": 5.925502190360437e-06,
      "loss": 0.5663,
      "step": 4279
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5557915894947436,
      "learning_rate": 5.92104633957435e-06,
      "loss": 0.5854,
      "step": 4280
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5876026302708013,
      "learning_rate": 5.9165914600730026e-06,
      "loss": 0.5731,
      "step": 4281
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5966293786374078,
      "learning_rate": 5.912137552917188e-06,
      "loss": 0.5896,
      "step": 4282
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6287676787353856,
      "learning_rate": 5.907684619167481e-06,
      "loss": 0.5626,
      "step": 4283
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6736098166505884,
      "learning_rate": 5.903232659884218e-06,
      "loss": 0.5721,
      "step": 4284
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6340828244266502,
      "learning_rate": 5.898781676127497e-06,
      "loss": 0.5875,
      "step": 4285
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6901663097480641,
      "learning_rate": 5.894331668957203e-06,
      "loss": 0.554,
      "step": 4286
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6263020887040318,
      "learning_rate": 5.889882639432965e-06,
      "loss": 0.5768,
      "step": 4287
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6376992862667291,
      "learning_rate": 5.8854345886142e-06,
      "loss": 0.5595,
      "step": 4288
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6674534345301325,
      "learning_rate": 5.880987517560075e-06,
      "loss": 0.5706,
      "step": 4289
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6510360328754435,
      "learning_rate": 5.876541427329534e-06,
      "loss": 0.5811,
      "step": 4290
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7149609814349266,
      "learning_rate": 5.872096318981288e-06,
      "loss": 0.5836,
      "step": 4291
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5661984701688645,
      "learning_rate": 5.867652193573805e-06,
      "loss": 0.5455,
      "step": 4292
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6204798890774035,
      "learning_rate": 5.863209052165331e-06,
      "loss": 0.5865,
      "step": 4293
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6531891704970622,
      "learning_rate": 5.858766895813865e-06,
      "loss": 0.6054,
      "step": 4294
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6835195513111316,
      "learning_rate": 5.854325725577183e-06,
      "loss": 0.5557,
      "step": 4295
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6505906009294369,
      "learning_rate": 5.849885542512819e-06,
      "loss": 0.573,
      "step": 4296
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6507666023109974,
      "learning_rate": 5.845446347678073e-06,
      "loss": 0.5831,
      "step": 4297
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6263939959187865,
      "learning_rate": 5.841008142130016e-06,
      "loss": 0.5838,
      "step": 4298
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6308139250959613,
      "learning_rate": 5.836570926925473e-06,
      "loss": 0.5713,
      "step": 4299
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6174891967430988,
      "learning_rate": 5.8321347031210374e-06,
      "loss": 0.554,
      "step": 4300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5638597665913717,
      "learning_rate": 5.82769947177307e-06,
      "loss": 0.5528,
      "step": 4301
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6946164765933296,
      "learning_rate": 5.823265233937692e-06,
      "loss": 0.6093,
      "step": 4302
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6814667326279733,
      "learning_rate": 5.818831990670794e-06,
      "loss": 0.5557,
      "step": 4303
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6748848064824073,
      "learning_rate": 5.814399743028018e-06,
      "loss": 0.5756,
      "step": 4304
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7875958836990125,
      "learning_rate": 5.809968492064775e-06,
      "loss": 0.5982,
      "step": 4305
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6064528849425075,
      "learning_rate": 5.8055382388362415e-06,
      "loss": 0.5599,
      "step": 4306
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6816652063283669,
      "learning_rate": 5.801108984397355e-06,
      "loss": 0.5778,
      "step": 4307
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6091223884779309,
      "learning_rate": 5.796680729802815e-06,
      "loss": 0.5642,
      "step": 4308
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6908833528797491,
      "learning_rate": 5.792253476107077e-06,
      "loss": 0.5683,
      "step": 4309
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.639567667934252,
      "learning_rate": 5.787827224364366e-06,
      "loss": 0.5765,
      "step": 4310
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7099138038624936,
      "learning_rate": 5.783401975628672e-06,
      "loss": 0.5807,
      "step": 4311
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.612961197209185,
      "learning_rate": 5.778977730953734e-06,
      "loss": 0.5727,
      "step": 4312
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6669152728530543,
      "learning_rate": 5.7745544913930495e-06,
      "loss": 0.5948,
      "step": 4313
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6061472467033744,
      "learning_rate": 5.770132257999902e-06,
      "loss": 0.5664,
      "step": 4314
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6754723257511849,
      "learning_rate": 5.765711031827306e-06,
      "loss": 0.5659,
      "step": 4315
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.584880028545052,
      "learning_rate": 5.761290813928056e-06,
      "loss": 0.5903,
      "step": 4316
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6183370497458098,
      "learning_rate": 5.75687160535469e-06,
      "loss": 0.5803,
      "step": 4317
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6510911700296205,
      "learning_rate": 5.752453407159521e-06,
      "loss": 0.5493,
      "step": 4318
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6103080945286183,
      "learning_rate": 5.7480362203946195e-06,
      "loss": 0.5602,
      "step": 4319
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6305780093531017,
      "learning_rate": 5.7436200461117995e-06,
      "loss": 0.5621,
      "step": 4320
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7102188621605886,
      "learning_rate": 5.739204885362653e-06,
      "loss": 0.6075,
      "step": 4321
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6341448518424224,
      "learning_rate": 5.734790739198522e-06,
      "loss": 0.5547,
      "step": 4322
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7120270416072471,
      "learning_rate": 5.730377608670502e-06,
      "loss": 0.5678,
      "step": 4323
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.184693135431425,
      "learning_rate": 5.725965494829457e-06,
      "loss": 0.6005,
      "step": 4324
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.608648950124236,
      "learning_rate": 5.721554398726003e-06,
      "loss": 0.5522,
      "step": 4325
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7849999238785578,
      "learning_rate": 5.7171443214105205e-06,
      "loss": 0.5769,
      "step": 4326
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6678329770695874,
      "learning_rate": 5.712735263933136e-06,
      "loss": 0.5739,
      "step": 4327
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6369507921988197,
      "learning_rate": 5.708327227343735e-06,
      "loss": 0.5976,
      "step": 4328
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6237809312399482,
      "learning_rate": 5.703920212691969e-06,
      "loss": 0.565,
      "step": 4329
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6606721993749712,
      "learning_rate": 5.699514221027241e-06,
      "loss": 0.5617,
      "step": 4330
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7037041395777299,
      "learning_rate": 5.6951092533987114e-06,
      "loss": 0.6129,
      "step": 4331
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5969739406440493,
      "learning_rate": 5.690705310855291e-06,
      "loss": 0.5805,
      "step": 4332
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6908406177616088,
      "learning_rate": 5.686302394445654e-06,
      "loss": 0.5665,
      "step": 4333
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5502100842027592,
      "learning_rate": 5.6819005052182315e-06,
      "loss": 0.5561,
      "step": 4334
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7156192639858983,
      "learning_rate": 5.6774996442211995e-06,
      "loss": 0.5851,
      "step": 4335
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6962352597530171,
      "learning_rate": 5.673099812502499e-06,
      "loss": 0.5668,
      "step": 4336
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6477541664098461,
      "learning_rate": 5.668701011109823e-06,
      "loss": 0.5484,
      "step": 4337
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5911151774030308,
      "learning_rate": 5.664303241090615e-06,
      "loss": 0.57,
      "step": 4338
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6699109750199348,
      "learning_rate": 5.659906503492083e-06,
      "loss": 0.5701,
      "step": 4339
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6783361180350862,
      "learning_rate": 5.655510799361172e-06,
      "loss": 0.5784,
      "step": 4340
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7857393199077529,
      "learning_rate": 5.651116129744605e-06,
      "loss": 0.5796,
      "step": 4341
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5429821379159638,
      "learning_rate": 5.6467224956888415e-06,
      "loss": 0.5774,
      "step": 4342
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.694513031234128,
      "learning_rate": 5.642329898240089e-06,
      "loss": 0.6007,
      "step": 4343
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.655826652294444,
      "learning_rate": 5.637938338444325e-06,
      "loss": 0.595,
      "step": 4344
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6005717969667481,
      "learning_rate": 5.633547817347277e-06,
      "loss": 0.5643,
      "step": 4345
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6644793428098794,
      "learning_rate": 5.6291583359944095e-06,
      "loss": 0.5657,
      "step": 4346
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5856681590493444,
      "learning_rate": 5.6247698954309616e-06,
      "loss": 0.5759,
      "step": 4347
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6542624042493758,
      "learning_rate": 5.620382496701897e-06,
      "loss": 0.5962,
      "step": 4348
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6338891150471423,
      "learning_rate": 5.6159961408519695e-06,
      "loss": 0.5757,
      "step": 4349
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6077901757130985,
      "learning_rate": 5.61161082892565e-06,
      "loss": 0.5708,
      "step": 4350
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5417532471771076,
      "learning_rate": 5.607226561967171e-06,
      "loss": 0.671,
      "step": 4351
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6799506261378305,
      "learning_rate": 5.602843341020525e-06,
      "loss": 0.5793,
      "step": 4352
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.638627399871396,
      "learning_rate": 5.598461167129445e-06,
      "loss": 0.5542,
      "step": 4353
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6851104074288451,
      "learning_rate": 5.594080041337426e-06,
      "loss": 0.5861,
      "step": 4354
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6102204396217369,
      "learning_rate": 5.589699964687698e-06,
      "loss": 0.5559,
      "step": 4355
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6634727340236586,
      "learning_rate": 5.585320938223253e-06,
      "loss": 0.5916,
      "step": 4356
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6255730583195074,
      "learning_rate": 5.580942962986833e-06,
      "loss": 0.5623,
      "step": 4357
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.604212650389939,
      "learning_rate": 5.576566040020917e-06,
      "loss": 0.5587,
      "step": 4358
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5969931098486032,
      "learning_rate": 5.57219017036775e-06,
      "loss": 0.5657,
      "step": 4359
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5976078277452168,
      "learning_rate": 5.567815355069319e-06,
      "loss": 0.6059,
      "step": 4360
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6085861926765798,
      "learning_rate": 5.5634415951673536e-06,
      "loss": 0.5778,
      "step": 4361
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6534823868962406,
      "learning_rate": 5.5590688917033454e-06,
      "loss": 0.5853,
      "step": 4362
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6005220560680405,
      "learning_rate": 5.554697245718519e-06,
      "loss": 0.5822,
      "step": 4363
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6260829214613144,
      "learning_rate": 5.550326658253861e-06,
      "loss": 0.5518,
      "step": 4364
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6414692124639836,
      "learning_rate": 5.545957130350102e-06,
      "loss": 0.5706,
      "step": 4365
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5803086770605344,
      "learning_rate": 5.541588663047711e-06,
      "loss": 0.5779,
      "step": 4366
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6130790949375039,
      "learning_rate": 5.5372212573869175e-06,
      "loss": 0.573,
      "step": 4367
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6579637590268183,
      "learning_rate": 5.532854914407693e-06,
      "loss": 0.5592,
      "step": 4368
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5889917057238047,
      "learning_rate": 5.5284896351497566e-06,
      "loss": 0.5625,
      "step": 4369
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6129041780239711,
      "learning_rate": 5.524125420652571e-06,
      "loss": 0.5717,
      "step": 4370
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6323658540644849,
      "learning_rate": 5.51976227195534e-06,
      "loss": 0.5997,
      "step": 4371
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.64542722752174,
      "learning_rate": 5.515400190097038e-06,
      "loss": 0.5756,
      "step": 4372
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6229251898474316,
      "learning_rate": 5.511039176116357e-06,
      "loss": 0.5896,
      "step": 4373
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.560808767899697,
      "learning_rate": 5.506679231051747e-06,
      "loss": 0.5887,
      "step": 4374
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5952100918162784,
      "learning_rate": 5.502320355941404e-06,
      "loss": 0.571,
      "step": 4375
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.640790499700804,
      "learning_rate": 5.497962551823266e-06,
      "loss": 0.5802,
      "step": 4376
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.538476760402279,
      "learning_rate": 5.493605819735026e-06,
      "loss": 0.5486,
      "step": 4377
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6226641582455901,
      "learning_rate": 5.4892501607141036e-06,
      "loss": 0.5671,
      "step": 4378
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6369737059037823,
      "learning_rate": 5.4848955757976775e-06,
      "loss": 0.5847,
      "step": 4379
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5737897258674434,
      "learning_rate": 5.480542066022667e-06,
      "loss": 0.5768,
      "step": 4380
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.63542688871931,
      "learning_rate": 5.476189632425732e-06,
      "loss": 0.5761,
      "step": 4381
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.587266642330005,
      "learning_rate": 5.471838276043278e-06,
      "loss": 0.564,
      "step": 4382
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5752003833446413,
      "learning_rate": 5.46748799791146e-06,
      "loss": 0.577,
      "step": 4383
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5554057637845452,
      "learning_rate": 5.4631387990661635e-06,
      "loss": 0.5489,
      "step": 4384
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6603033374729118,
      "learning_rate": 5.458790680543031e-06,
      "loss": 0.5787,
      "step": 4385
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6150111322329599,
      "learning_rate": 5.454443643377435e-06,
      "loss": 0.5567,
      "step": 4386
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6231059196394173,
      "learning_rate": 5.450097688604498e-06,
      "loss": 0.5758,
      "step": 4387
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5738722859587357,
      "learning_rate": 5.44575281725909e-06,
      "loss": 0.5522,
      "step": 4388
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6694287065195859,
      "learning_rate": 5.441409030375806e-06,
      "loss": 0.5968,
      "step": 4389
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6804003585113683,
      "learning_rate": 5.437066328988999e-06,
      "loss": 0.5811,
      "step": 4390
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5639456318617001,
      "learning_rate": 5.432724714132756e-06,
      "loss": 0.5566,
      "step": 4391
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5759096889672248,
      "learning_rate": 5.428384186840912e-06,
      "loss": 0.5756,
      "step": 4392
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6327743298365242,
      "learning_rate": 5.424044748147032e-06,
      "loss": 0.5767,
      "step": 4393
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6631144949903681,
      "learning_rate": 5.419706399084424e-06,
      "loss": 0.5678,
      "step": 4394
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6522025876978667,
      "learning_rate": 5.415369140686151e-06,
      "loss": 0.6119,
      "step": 4395
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6947095962360144,
      "learning_rate": 5.411032973984997e-06,
      "loss": 0.5822,
      "step": 4396
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6859932196921613,
      "learning_rate": 5.406697900013502e-06,
      "loss": 0.6092,
      "step": 4397
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6101858440001523,
      "learning_rate": 5.40236391980393e-06,
      "loss": 0.5533,
      "step": 4398
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6513482925295485,
      "learning_rate": 5.3980310343882955e-06,
      "loss": 0.6053,
      "step": 4399
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6524806248953091,
      "learning_rate": 5.393699244798357e-06,
      "loss": 0.5869,
      "step": 4400
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6551904748722196,
      "learning_rate": 5.389368552065595e-06,
      "loss": 0.5803,
      "step": 4401
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6374691230415036,
      "learning_rate": 5.385038957221241e-06,
      "loss": 0.5552,
      "step": 4402
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5876530850252384,
      "learning_rate": 5.380710461296268e-06,
      "loss": 0.5766,
      "step": 4403
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6134442481584222,
      "learning_rate": 5.376383065321376e-06,
      "loss": 0.5768,
      "step": 4404
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6198496823476646,
      "learning_rate": 5.3720567703270135e-06,
      "loss": 0.5775,
      "step": 4405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7181756361913089,
      "learning_rate": 5.367731577343357e-06,
      "loss": 0.6169,
      "step": 4406
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.582628302881711,
      "learning_rate": 5.36340748740033e-06,
      "loss": 0.5894,
      "step": 4407
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7444746253648747,
      "learning_rate": 5.35908450152759e-06,
      "loss": 0.5847,
      "step": 4408
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5675051343444081,
      "learning_rate": 5.354762620754528e-06,
      "loss": 0.5605,
      "step": 4409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6473399130496875,
      "learning_rate": 5.350441846110274e-06,
      "loss": 0.5768,
      "step": 4410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6642189245276304,
      "learning_rate": 5.346122178623705e-06,
      "loss": 0.6203,
      "step": 4411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6489014950078786,
      "learning_rate": 5.3418036193234115e-06,
      "loss": 0.5985,
      "step": 4412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7278632549862386,
      "learning_rate": 5.337486169237739e-06,
      "loss": 0.6079,
      "step": 4413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6469995512692753,
      "learning_rate": 5.3331698293947645e-06,
      "loss": 0.5713,
      "step": 4414
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6127504363434527,
      "learning_rate": 5.328854600822302e-06,
      "loss": 0.5878,
      "step": 4415
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6172554599772125,
      "learning_rate": 5.324540484547894e-06,
      "loss": 0.5505,
      "step": 4416
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7358626753081658,
      "learning_rate": 5.320227481598816e-06,
      "loss": 0.5855,
      "step": 4417
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5871358576177598,
      "learning_rate": 5.3159155930021e-06,
      "loss": 0.5642,
      "step": 4418
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5844372459409641,
      "learning_rate": 5.3116048197844845e-06,
      "loss": 0.5821,
      "step": 4419
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6340860634407393,
      "learning_rate": 5.307295162972466e-06,
      "loss": 0.5668,
      "step": 4420
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6222285015492923,
      "learning_rate": 5.302986623592253e-06,
      "loss": 0.5872,
      "step": 4421
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6001374530984681,
      "learning_rate": 5.298679202669806e-06,
      "loss": 0.5589,
      "step": 4422
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5983228865593799,
      "learning_rate": 5.294372901230815e-06,
      "loss": 0.5735,
      "step": 4423
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6227846540117286,
      "learning_rate": 5.290067720300695e-06,
      "loss": 0.5559,
      "step": 4424
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6543535283749303,
      "learning_rate": 5.2857636609046026e-06,
      "loss": 0.6012,
      "step": 4425
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7009358000221099,
      "learning_rate": 5.2814607240674285e-06,
      "loss": 0.5923,
      "step": 4426
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.592116859538832,
      "learning_rate": 5.277158910813786e-06,
      "loss": 0.6061,
      "step": 4427
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6093018696286875,
      "learning_rate": 5.2728582221680355e-06,
      "loss": 0.5644,
      "step": 4428
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7115428824985682,
      "learning_rate": 5.268558659154248e-06,
      "loss": 0.604,
      "step": 4429
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6564730525732356,
      "learning_rate": 5.264260222796258e-06,
      "loss": 0.5907,
      "step": 4430
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6294357342219901,
      "learning_rate": 5.259962914117603e-06,
      "loss": 0.5904,
      "step": 4431
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.603628177564068,
      "learning_rate": 5.255666734141561e-06,
      "loss": 0.5629,
      "step": 4432
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7081991043110814,
      "learning_rate": 5.2513716838911465e-06,
      "loss": 0.5878,
      "step": 4433
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6266331068755672,
      "learning_rate": 5.247077764389099e-06,
      "loss": 0.5507,
      "step": 4434
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6355667128573279,
      "learning_rate": 5.242784976657899e-06,
      "loss": 0.6009,
      "step": 4435
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6482942486478281,
      "learning_rate": 5.238493321719739e-06,
      "loss": 0.5571,
      "step": 4436
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6110262072862388,
      "learning_rate": 5.2342028005965575e-06,
      "loss": 0.5682,
      "step": 4437
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6477868553780773,
      "learning_rate": 5.229913414310019e-06,
      "loss": 0.5775,
      "step": 4438
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6878062967882882,
      "learning_rate": 5.225625163881518e-06,
      "loss": 0.5619,
      "step": 4439
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6459658962425797,
      "learning_rate": 5.22133805033217e-06,
      "loss": 0.5496,
      "step": 4440
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6645294604400309,
      "learning_rate": 5.217052074682829e-06,
      "loss": 0.5801,
      "step": 4441
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5857042510412659,
      "learning_rate": 5.212767237954081e-06,
      "loss": 0.5602,
      "step": 4442
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6006328586177473,
      "learning_rate": 5.208483541166236e-06,
      "loss": 0.5872,
      "step": 4443
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.655524742889709,
      "learning_rate": 5.2042009853393245e-06,
      "loss": 0.5931,
      "step": 4444
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7161665216528849,
      "learning_rate": 5.1999195714931205e-06,
      "loss": 0.6043,
      "step": 4445
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5579284430138475,
      "learning_rate": 5.195639300647119e-06,
      "loss": 0.5967,
      "step": 4446
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6778834737114503,
      "learning_rate": 5.191360173820536e-06,
      "loss": 0.5556,
      "step": 4447
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5563928464024738,
      "learning_rate": 5.1870821920323275e-06,
      "loss": 0.5744,
      "step": 4448
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7743492044358881,
      "learning_rate": 5.182805356301173e-06,
      "loss": 0.6023,
      "step": 4449
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6491658091150085,
      "learning_rate": 5.1785296676454685e-06,
      "loss": 0.5876,
      "step": 4450
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6634419578615546,
      "learning_rate": 5.174255127083354e-06,
      "loss": 0.5942,
      "step": 4451
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5921701851870578,
      "learning_rate": 5.169981735632677e-06,
      "loss": 0.5847,
      "step": 4452
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6371954168971778,
      "learning_rate": 5.165709494311037e-06,
      "loss": 0.5461,
      "step": 4453
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6816101149580495,
      "learning_rate": 5.1614384041357356e-06,
      "loss": 0.5841,
      "step": 4454
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5798830421452587,
      "learning_rate": 5.1571684661238075e-06,
      "loss": 0.5577,
      "step": 4455
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6195979895845417,
      "learning_rate": 5.1528996812920166e-06,
      "loss": 0.5823,
      "step": 4456
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6345130339843671,
      "learning_rate": 5.148632050656852e-06,
      "loss": 0.5689,
      "step": 4457
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6067511644286576,
      "learning_rate": 5.144365575234529e-06,
      "loss": 0.5777,
      "step": 4458
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5988854539420542,
      "learning_rate": 5.140100256040979e-06,
      "loss": 0.5763,
      "step": 4459
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6876375882632212,
      "learning_rate": 5.135836094091867e-06,
      "loss": 0.5964,
      "step": 4460
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6066781244505529,
      "learning_rate": 5.131573090402584e-06,
      "loss": 0.5933,
      "step": 4461
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6151787551930545,
      "learning_rate": 5.127311245988233e-06,
      "loss": 0.5697,
      "step": 4462
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5766381841485517,
      "learning_rate": 5.1230505618636575e-06,
      "loss": 0.5333,
      "step": 4463
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6436503265167904,
      "learning_rate": 5.118791039043407e-06,
      "loss": 0.5881,
      "step": 4464
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5726714540525146,
      "learning_rate": 5.114532678541768e-06,
      "loss": 0.5875,
      "step": 4465
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6964652661770602,
      "learning_rate": 5.110275481372748e-06,
      "loss": 0.5792,
      "step": 4466
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6956626399811792,
      "learning_rate": 5.106019448550073e-06,
      "loss": 0.5931,
      "step": 4467
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7398909744122575,
      "learning_rate": 5.10176458108719e-06,
      "loss": 0.6284,
      "step": 4468
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7378897985227231,
      "learning_rate": 5.097510879997283e-06,
      "loss": 0.6036,
      "step": 4469
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6007536043490743,
      "learning_rate": 5.093258346293237e-06,
      "loss": 0.5796,
      "step": 4470
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6230284072161915,
      "learning_rate": 5.089006980987674e-06,
      "loss": 0.5698,
      "step": 4471
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6250380115031017,
      "learning_rate": 5.084756785092937e-06,
      "loss": 0.5794,
      "step": 4472
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6072933859720552,
      "learning_rate": 5.080507759621081e-06,
      "loss": 0.5882,
      "step": 4473
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6550333310192595,
      "learning_rate": 5.076259905583894e-06,
      "loss": 0.6272,
      "step": 4474
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6559592626717875,
      "learning_rate": 5.07201322399287e-06,
      "loss": 0.5881,
      "step": 4475
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7038963096723803,
      "learning_rate": 5.067767715859248e-06,
      "loss": 0.5781,
      "step": 4476
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6043646582964688,
      "learning_rate": 5.063523382193963e-06,
      "loss": 0.5974,
      "step": 4477
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.555964027176949,
      "learning_rate": 5.05928022400768e-06,
      "loss": 0.5522,
      "step": 4478
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5705626194986504,
      "learning_rate": 5.055038242310786e-06,
      "loss": 0.5773,
      "step": 4479
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.637534014673388,
      "learning_rate": 5.050797438113387e-06,
      "loss": 0.5812,
      "step": 4480
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6243305332475112,
      "learning_rate": 5.0465578124253104e-06,
      "loss": 0.582,
      "step": 4481
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6234808255232378,
      "learning_rate": 5.042319366256096e-06,
      "loss": 0.5673,
      "step": 4482
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6271634735228154,
      "learning_rate": 5.038082100615003e-06,
      "loss": 0.5594,
      "step": 4483
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6024755639338867,
      "learning_rate": 5.0338460165110235e-06,
      "loss": 0.5767,
      "step": 4484
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6475425354958582,
      "learning_rate": 5.029611114952852e-06,
      "loss": 0.5889,
      "step": 4485
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5810169105991811,
      "learning_rate": 5.025377396948914e-06,
      "loss": 0.5627,
      "step": 4486
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6199626377878353,
      "learning_rate": 5.021144863507337e-06,
      "loss": 0.5874,
      "step": 4487
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7303114526243251,
      "learning_rate": 5.016913515635981e-06,
      "loss": 0.6022,
      "step": 4488
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6267522347406206,
      "learning_rate": 5.012683354342424e-06,
      "loss": 0.572,
      "step": 4489
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6638638048203074,
      "learning_rate": 5.008454380633948e-06,
      "loss": 0.5509,
      "step": 4490
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5241707913555439,
      "learning_rate": 5.004226595517565e-06,
      "loss": 0.5841,
      "step": 4491
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5935196696394327,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.5596,
      "step": 4492
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6028378571981096,
      "learning_rate": 4.995774595087695e-06,
      "loss": 0.5582,
      "step": 4493
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6864364495984538,
      "learning_rate": 4.991550381786804e-06,
      "loss": 0.612,
      "step": 4494
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5561089514308977,
      "learning_rate": 4.9873273611032035e-06,
      "loss": 0.5687,
      "step": 4495
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6831624526356164,
      "learning_rate": 4.983105534042489e-06,
      "loss": 0.5839,
      "step": 4496
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6271567429483923,
      "learning_rate": 4.9788849016099595e-06,
      "loss": 0.579,
      "step": 4497
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6007846503456582,
      "learning_rate": 4.974665464810635e-06,
      "loss": 0.5757,
      "step": 4498
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7149855375717011,
      "learning_rate": 4.970447224649255e-06,
      "loss": 0.598,
      "step": 4499
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7389756116295924,
      "learning_rate": 4.966230182130275e-06,
      "loss": 0.6173,
      "step": 4500
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6065685967587185,
      "learning_rate": 4.962014338257856e-06,
      "loss": 0.5691,
      "step": 4501
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6144569495035302,
      "learning_rate": 4.95779969403588e-06,
      "loss": 0.5685,
      "step": 4502
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.633969807381721,
      "learning_rate": 4.9535862504679435e-06,
      "loss": 0.5914,
      "step": 4503
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6615430431913929,
      "learning_rate": 4.949374008557361e-06,
      "loss": 0.5825,
      "step": 4504
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7367075587134612,
      "learning_rate": 4.945162969307152e-06,
      "loss": 0.5878,
      "step": 4505
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.631880733084774,
      "learning_rate": 4.940953133720045e-06,
      "loss": 0.5955,
      "step": 4506
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6544257604284821,
      "learning_rate": 4.936744502798506e-06,
      "loss": 0.5982,
      "step": 4507
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6319931823809107,
      "learning_rate": 4.9325370775446866e-06,
      "loss": 0.564,
      "step": 4508
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6822972139529907,
      "learning_rate": 4.9283308589604725e-06,
      "loss": 0.6119,
      "step": 4509
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.549512099320622,
      "learning_rate": 4.9241258480474454e-06,
      "loss": 0.5432,
      "step": 4510
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6574087704549659,
      "learning_rate": 4.9199220458069085e-06,
      "loss": 0.5743,
      "step": 4511
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6438057832705145,
      "learning_rate": 4.915719453239882e-06,
      "loss": 0.5702,
      "step": 4512
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5986134140097166,
      "learning_rate": 4.911518071347081e-06,
      "loss": 0.5818,
      "step": 4513
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5889915294032723,
      "learning_rate": 4.907317901128951e-06,
      "loss": 0.5699,
      "step": 4514
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5894816835649026,
      "learning_rate": 4.90311894358564e-06,
      "loss": 0.556,
      "step": 4515
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6038470869616696,
      "learning_rate": 4.898921199717004e-06,
      "loss": 0.563,
      "step": 4516
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7847655890356454,
      "learning_rate": 4.894724670522617e-06,
      "loss": 0.5969,
      "step": 4517
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6133967669014498,
      "learning_rate": 4.89052935700176e-06,
      "loss": 0.5729,
      "step": 4518
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6940368946836005,
      "learning_rate": 4.886335260153431e-06,
      "loss": 0.6052,
      "step": 4519
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.608666554295606,
      "learning_rate": 4.882142380976327e-06,
      "loss": 0.5797,
      "step": 4520
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6498261201608675,
      "learning_rate": 4.8779507204688595e-06,
      "loss": 0.5668,
      "step": 4521
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6352706684283035,
      "learning_rate": 4.873760279629152e-06,
      "loss": 0.5664,
      "step": 4522
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6655421970422792,
      "learning_rate": 4.86957105945504e-06,
      "loss": 0.6059,
      "step": 4523
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6185354290169118,
      "learning_rate": 4.865383060944065e-06,
      "loss": 0.5867,
      "step": 4524
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6082031718843028,
      "learning_rate": 4.861196285093473e-06,
      "loss": 0.5797,
      "step": 4525
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.646232618970827,
      "learning_rate": 4.857010732900225e-06,
      "loss": 0.5831,
      "step": 4526
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.546747966627919,
      "learning_rate": 4.852826405360996e-06,
      "loss": 0.5761,
      "step": 4527
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6547894394796044,
      "learning_rate": 4.848643303472151e-06,
      "loss": 0.5739,
      "step": 4528
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6463685004410359,
      "learning_rate": 4.844461428229782e-06,
      "loss": 0.5666,
      "step": 4529
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5949484246755965,
      "learning_rate": 4.840280780629683e-06,
      "loss": 0.5753,
      "step": 4530
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5430826682954278,
      "learning_rate": 4.836101361667348e-06,
      "loss": 0.5477,
      "step": 4531
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6326322183339809,
      "learning_rate": 4.831923172337991e-06,
      "loss": 0.5725,
      "step": 4532
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6301961317644168,
      "learning_rate": 4.827746213636519e-06,
      "loss": 0.5836,
      "step": 4533
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5932219059016035,
      "learning_rate": 4.823570486557561e-06,
      "loss": 0.5803,
      "step": 4534
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5855484543541768,
      "learning_rate": 4.8193959920954435e-06,
      "loss": 0.5484,
      "step": 4535
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6932241838288722,
      "learning_rate": 4.815222731244199e-06,
      "loss": 0.6167,
      "step": 4536
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6613293718988869,
      "learning_rate": 4.8110507049975705e-06,
      "loss": 0.5829,
      "step": 4537
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.652860383505928,
      "learning_rate": 4.80687991434901e-06,
      "loss": 0.5755,
      "step": 4538
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6175013139700856,
      "learning_rate": 4.8027103602916615e-06,
      "loss": 0.5802,
      "step": 4539
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5937184176530421,
      "learning_rate": 4.798542043818391e-06,
      "loss": 0.5901,
      "step": 4540
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5897538301615829,
      "learning_rate": 4.794374965921754e-06,
      "loss": 0.566,
      "step": 4541
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6707311376873556,
      "learning_rate": 4.790209127594032e-06,
      "loss": 0.5863,
      "step": 4542
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6561930834868891,
      "learning_rate": 4.786044529827191e-06,
      "loss": 0.5841,
      "step": 4543
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6279566008497701,
      "learning_rate": 4.781881173612906e-06,
      "loss": 0.5519,
      "step": 4544
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6612041732987344,
      "learning_rate": 4.777719059942566e-06,
      "loss": 0.5724,
      "step": 4545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6529998523246828,
      "learning_rate": 4.773558189807255e-06,
      "loss": 0.5922,
      "step": 4546
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6448372197956386,
      "learning_rate": 4.769398564197768e-06,
      "loss": 0.5796,
      "step": 4547
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6038370657167977,
      "learning_rate": 4.765240184104592e-06,
      "loss": 0.581,
      "step": 4548
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6324210675249027,
      "learning_rate": 4.7610830505179295e-06,
      "loss": 0.5504,
      "step": 4549
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6559509681186431,
      "learning_rate": 4.756927164427685e-06,
      "loss": 0.5898,
      "step": 4550
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5860163764958537,
      "learning_rate": 4.752772526823453e-06,
      "loss": 0.5697,
      "step": 4551
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6504070815958456,
      "learning_rate": 4.748619138694548e-06,
      "loss": 0.5544,
      "step": 4552
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6576400624372657,
      "learning_rate": 4.74446700102998e-06,
      "loss": 0.5867,
      "step": 4553
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6032770188628608,
      "learning_rate": 4.7403161148184515e-06,
      "loss": 0.5782,
      "step": 4554
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6446002786481246,
      "learning_rate": 4.736166481048388e-06,
      "loss": 0.5821,
      "step": 4555
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6392844284452383,
      "learning_rate": 4.732018100707892e-06,
      "loss": 0.5506,
      "step": 4556
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6196391921465212,
      "learning_rate": 4.727870974784787e-06,
      "loss": 0.5655,
      "step": 4557
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6305839809645051,
      "learning_rate": 4.723725104266594e-06,
      "loss": 0.5772,
      "step": 4558
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5943393789798964,
      "learning_rate": 4.719580490140525e-06,
      "loss": 0.5621,
      "step": 4559
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6451717815048396,
      "learning_rate": 4.715437133393503e-06,
      "loss": 0.5347,
      "step": 4560
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6772699397032628,
      "learning_rate": 4.711295035012153e-06,
      "loss": 0.5878,
      "step": 4561
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6663484003790805,
      "learning_rate": 4.707154195982788e-06,
      "loss": 0.5519,
      "step": 4562
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6245277120366622,
      "learning_rate": 4.703014617291436e-06,
      "loss": 0.5583,
      "step": 4563
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6487352577010665,
      "learning_rate": 4.698876299923807e-06,
      "loss": 0.5553,
      "step": 4564
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6368467672637151,
      "learning_rate": 4.694739244865335e-06,
      "loss": 0.5866,
      "step": 4565
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5548583658141479,
      "learning_rate": 4.690603453101134e-06,
      "loss": 0.5695,
      "step": 4566
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6395853704297735,
      "learning_rate": 4.686468925616021e-06,
      "loss": 0.5544,
      "step": 4567
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6175512882002981,
      "learning_rate": 4.6823356633945136e-06,
      "loss": 0.5621,
      "step": 4568
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5916718990395402,
      "learning_rate": 4.678203667420832e-06,
      "loss": 0.5666,
      "step": 4569
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5960881028402212,
      "learning_rate": 4.674072938678894e-06,
      "loss": 0.5862,
      "step": 4570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7058576688085443,
      "learning_rate": 4.669943478152305e-06,
      "loss": 0.5997,
      "step": 4571
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6634504137638306,
      "learning_rate": 4.665815286824381e-06,
      "loss": 0.5848,
      "step": 4572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.647749644135465,
      "learning_rate": 4.661688365678135e-06,
      "loss": 0.5897,
      "step": 4573
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7238204758887614,
      "learning_rate": 4.657562715696266e-06,
      "loss": 0.5658,
      "step": 4574
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5754145396854531,
      "learning_rate": 4.653438337861182e-06,
      "loss": 0.5709,
      "step": 4575
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7471050432149048,
      "learning_rate": 4.649315233154988e-06,
      "loss": 0.6269,
      "step": 4576
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6279510540013163,
      "learning_rate": 4.645193402559473e-06,
      "loss": 0.5533,
      "step": 4577
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.649033613366684,
      "learning_rate": 4.641072847056142e-06,
      "loss": 0.5742,
      "step": 4578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6689760094199162,
      "learning_rate": 4.636953567626176e-06,
      "loss": 0.5916,
      "step": 4579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7175486195330271,
      "learning_rate": 4.6328355652504686e-06,
      "loss": 0.6086,
      "step": 4580
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6526263366943833,
      "learning_rate": 4.628718840909604e-06,
      "loss": 0.5799,
      "step": 4581
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6543553443130213,
      "learning_rate": 4.624603395583854e-06,
      "loss": 0.5817,
      "step": 4582
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.612144183801067,
      "learning_rate": 4.620489230253198e-06,
      "loss": 0.5749,
      "step": 4583
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6150430705156223,
      "learning_rate": 4.616376345897303e-06,
      "loss": 0.588,
      "step": 4584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6756737676739911,
      "learning_rate": 4.612264743495539e-06,
      "loss": 0.5791,
      "step": 4585
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6114507806169428,
      "learning_rate": 4.60815442402696e-06,
      "loss": 0.5698,
      "step": 4586
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6430171307903717,
      "learning_rate": 4.604045388470314e-06,
      "loss": 0.5876,
      "step": 4587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6082097270460749,
      "learning_rate": 4.599937637804063e-06,
      "loss": 0.5506,
      "step": 4588
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6131214130446788,
      "learning_rate": 4.595831173006335e-06,
      "loss": 0.5602,
      "step": 4589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6108307415001649,
      "learning_rate": 4.5917259950549775e-06,
      "loss": 0.5727,
      "step": 4590
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6260736089168263,
      "learning_rate": 4.587622104927511e-06,
      "loss": 0.566,
      "step": 4591
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6997730311831755,
      "learning_rate": 4.583519503601159e-06,
      "loss": 0.5614,
      "step": 4592
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6350324925274267,
      "learning_rate": 4.579418192052844e-06,
      "loss": 0.5737,
      "step": 4593
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6268123820141094,
      "learning_rate": 4.5753181712591675e-06,
      "loss": 0.5726,
      "step": 4594
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6731357064608511,
      "learning_rate": 4.571219442196433e-06,
      "loss": 0.5676,
      "step": 4595
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6277692600105557,
      "learning_rate": 4.567122005840639e-06,
      "loss": 0.5878,
      "step": 4596
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6688389455959668,
      "learning_rate": 4.563025863167461e-06,
      "loss": 0.5598,
      "step": 4597
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6338170074357194,
      "learning_rate": 4.558931015152288e-06,
      "loss": 0.5864,
      "step": 4598
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6424233934831348,
      "learning_rate": 4.55483746277018e-06,
      "loss": 0.5471,
      "step": 4599
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.658417526700379,
      "learning_rate": 4.550745206995901e-06,
      "loss": 0.5701,
      "step": 4600
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.619527330066766,
      "learning_rate": 4.54665424880391e-06,
      "loss": 0.5892,
      "step": 4601
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6362995339075603,
      "learning_rate": 4.54256458916834e-06,
      "loss": 0.5737,
      "step": 4602
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5834566883055193,
      "learning_rate": 4.53847622906303e-06,
      "loss": 0.5683,
      "step": 4603
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5692029884654553,
      "learning_rate": 4.5343891694615094e-06,
      "loss": 0.5583,
      "step": 4604
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6339776531713363,
      "learning_rate": 4.530303411336983e-06,
      "loss": 0.5864,
      "step": 4605
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6455552773600487,
      "learning_rate": 4.526218955662361e-06,
      "loss": 0.5878,
      "step": 4606
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7613714917051614,
      "learning_rate": 4.522135803410238e-06,
      "loss": 0.6059,
      "step": 4607
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6357132501385705,
      "learning_rate": 4.518053955552903e-06,
      "loss": 0.5742,
      "step": 4608
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5770444127809703,
      "learning_rate": 4.513973413062326e-06,
      "loss": 0.5733,
      "step": 4609
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8655370610337187,
      "learning_rate": 4.509894176910161e-06,
      "loss": 0.6201,
      "step": 4610
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6232392267029774,
      "learning_rate": 4.505816248067778e-06,
      "loss": 0.5917,
      "step": 4611
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7853653159403667,
      "learning_rate": 4.501739627506203e-06,
      "loss": 0.5569,
      "step": 4612
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5921725908234721,
      "learning_rate": 4.497664316196175e-06,
      "loss": 0.5665,
      "step": 4613
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5643868408485125,
      "learning_rate": 4.493590315108103e-06,
      "loss": 0.5585,
      "step": 4614
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6311139732201959,
      "learning_rate": 4.489517625212096e-06,
      "loss": 0.567,
      "step": 4615
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6186861196399319,
      "learning_rate": 4.485446247477953e-06,
      "loss": 0.5937,
      "step": 4616
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5866730362855263,
      "learning_rate": 4.4813761828751445e-06,
      "loss": 0.576,
      "step": 4617
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6178976978314047,
      "learning_rate": 4.477307432372844e-06,
      "loss": 0.5764,
      "step": 4618
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6077356890196103,
      "learning_rate": 4.473239996939909e-06,
      "loss": 0.5439,
      "step": 4619
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.59801600107108,
      "learning_rate": 4.4691738775448755e-06,
      "loss": 0.5435,
      "step": 4620
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6391738683940354,
      "learning_rate": 4.4651090751559775e-06,
      "loss": 0.5846,
      "step": 4621
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5932595665061452,
      "learning_rate": 4.461045590741122e-06,
      "loss": 0.5519,
      "step": 4622
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9416438911677819,
      "learning_rate": 4.4569834252679235e-06,
      "loss": 0.5982,
      "step": 4623
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6126214029089195,
      "learning_rate": 4.45292257970366e-06,
      "loss": 0.5657,
      "step": 4624
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5619387973046667,
      "learning_rate": 4.4488630550153036e-06,
      "loss": 0.5769,
      "step": 4625
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.612755540325664,
      "learning_rate": 4.444804852169515e-06,
      "loss": 0.5921,
      "step": 4626
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5731455870927824,
      "learning_rate": 4.44074797213264e-06,
      "loss": 0.5726,
      "step": 4627
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.575836989774451,
      "learning_rate": 4.436692415870701e-06,
      "loss": 0.56,
      "step": 4628
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6119907079533512,
      "learning_rate": 4.432638184349416e-06,
      "loss": 0.5882,
      "step": 4629
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.571580735484904,
      "learning_rate": 4.428585278534181e-06,
      "loss": 0.5568,
      "step": 4630
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6559916371442853,
      "learning_rate": 4.424533699390083e-06,
      "loss": 0.585,
      "step": 4631
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5422526343618528,
      "learning_rate": 4.420483447881885e-06,
      "loss": 0.5602,
      "step": 4632
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6194465810440327,
      "learning_rate": 4.416434524974033e-06,
      "loss": 0.5602,
      "step": 4633
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6295501292522002,
      "learning_rate": 4.412386931630663e-06,
      "loss": 0.5883,
      "step": 4634
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5977839342726785,
      "learning_rate": 4.408340668815595e-06,
      "loss": 0.5602,
      "step": 4635
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6194096805398105,
      "learning_rate": 4.4042957374923316e-06,
      "loss": 0.5755,
      "step": 4636
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5927225645596278,
      "learning_rate": 4.400252138624047e-06,
      "loss": 0.554,
      "step": 4637
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5844603625021769,
      "learning_rate": 4.396209873173614e-06,
      "loss": 0.5418,
      "step": 4638
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5578997679213533,
      "learning_rate": 4.392168942103583e-06,
      "loss": 0.5734,
      "step": 4639
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6614239933028766,
      "learning_rate": 4.388129346376177e-06,
      "loss": 0.5609,
      "step": 4640
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6040547726015645,
      "learning_rate": 4.384091086953314e-06,
      "loss": 0.5528,
      "step": 4641
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6558265706586872,
      "learning_rate": 4.380054164796591e-06,
      "loss": 0.5675,
      "step": 4642
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6426183679416164,
      "learning_rate": 4.376018580867278e-06,
      "loss": 0.5581,
      "step": 4643
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6457890822599868,
      "learning_rate": 4.371984336126338e-06,
      "loss": 0.5702,
      "step": 4644
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6172667762211093,
      "learning_rate": 4.367951431534401e-06,
      "loss": 0.5727,
      "step": 4645
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6146280527075497,
      "learning_rate": 4.363919868051799e-06,
      "loss": 0.5699,
      "step": 4646
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6090069302743752,
      "learning_rate": 4.359889646638527e-06,
      "loss": 0.5561,
      "step": 4647
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6059173155819849,
      "learning_rate": 4.355860768254259e-06,
      "loss": 0.5629,
      "step": 4648
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6037825510985251,
      "learning_rate": 4.351833233858361e-06,
      "loss": 0.5977,
      "step": 4649
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7022225236617712,
      "learning_rate": 4.347807044409874e-06,
      "loss": 0.6079,
      "step": 4650
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5890989013239464,
      "learning_rate": 4.343782200867522e-06,
      "loss": 0.5431,
      "step": 4651
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5858041084721203,
      "learning_rate": 4.339758704189699e-06,
      "loss": 0.5618,
      "step": 4652
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6772568619805849,
      "learning_rate": 4.3357365553344836e-06,
      "loss": 0.5573,
      "step": 4653
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6364977234202509,
      "learning_rate": 4.331715755259642e-06,
      "loss": 0.5784,
      "step": 4654
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6357414868584739,
      "learning_rate": 4.327696304922606e-06,
      "loss": 0.5624,
      "step": 4655
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6079416140895443,
      "learning_rate": 4.3236782052804884e-06,
      "loss": 0.5333,
      "step": 4656
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6438470812271272,
      "learning_rate": 4.3196614572900865e-06,
      "loss": 0.5864,
      "step": 4657
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6098360514486684,
      "learning_rate": 4.315646061907872e-06,
      "loss": 0.5451,
      "step": 4658
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7213760791603121,
      "learning_rate": 4.31163202009e-06,
      "loss": 0.583,
      "step": 4659
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6003755184609004,
      "learning_rate": 4.307619332792289e-06,
      "loss": 0.5826,
      "step": 4660
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6011586833890445,
      "learning_rate": 4.303608000970251e-06,
      "loss": 0.566,
      "step": 4661
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5972178144583947,
      "learning_rate": 4.299598025579069e-06,
      "loss": 0.5848,
      "step": 4662
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6083976321422595,
      "learning_rate": 4.295589407573598e-06,
      "loss": 0.5676,
      "step": 4663
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6534148812590205,
      "learning_rate": 4.291582147908376e-06,
      "loss": 0.5969,
      "step": 4664
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6098709129358829,
      "learning_rate": 4.2875762475376204e-06,
      "loss": 0.5877,
      "step": 4665
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6373314776642164,
      "learning_rate": 4.283571707415214e-06,
      "loss": 0.5713,
      "step": 4666
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6507416559846642,
      "learning_rate": 4.279568528494727e-06,
      "loss": 0.5742,
      "step": 4667
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5738225074533779,
      "learning_rate": 4.275566711729392e-06,
      "loss": 0.5408,
      "step": 4668
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7353665895003872,
      "learning_rate": 4.271566258072139e-06,
      "loss": 0.6107,
      "step": 4669
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.658703779379185,
      "learning_rate": 4.267567168475554e-06,
      "loss": 0.5558,
      "step": 4670
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7175871069315144,
      "learning_rate": 4.2635694438919e-06,
      "loss": 0.6037,
      "step": 4671
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5689890262374196,
      "learning_rate": 4.259573085273122e-06,
      "loss": 0.5802,
      "step": 4672
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6600093979744396,
      "learning_rate": 4.2555780935708405e-06,
      "loss": 0.5795,
      "step": 4673
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6134094618088834,
      "learning_rate": 4.2515844697363465e-06,
      "loss": 0.5804,
      "step": 4674
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.626345792946055,
      "learning_rate": 4.247592214720605e-06,
      "loss": 0.557,
      "step": 4675
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6306381118861238,
      "learning_rate": 4.243601329474248e-06,
      "loss": 0.5589,
      "step": 4676
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5670088252120847,
      "learning_rate": 4.239611814947605e-06,
      "loss": 0.5556,
      "step": 4677
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.644587079910625,
      "learning_rate": 4.23562367209065e-06,
      "loss": 0.5835,
      "step": 4678
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5924796348850546,
      "learning_rate": 4.231636901853053e-06,
      "loss": 0.6079,
      "step": 4679
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6383210760451575,
      "learning_rate": 4.227651505184141e-06,
      "loss": 0.5463,
      "step": 4680
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.59313801815422,
      "learning_rate": 4.223667483032921e-06,
      "loss": 0.5604,
      "step": 4681
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6051877424593581,
      "learning_rate": 4.2196848363480815e-06,
      "loss": 0.5541,
      "step": 4682
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6183269040408707,
      "learning_rate": 4.215703566077962e-06,
      "loss": 0.5681,
      "step": 4683
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.185402974095776,
      "learning_rate": 4.211723673170595e-06,
      "loss": 0.6226,
      "step": 4684
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6101905311863363,
      "learning_rate": 4.207745158573677e-06,
      "loss": 0.5859,
      "step": 4685
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6473406979531492,
      "learning_rate": 4.20376802323457e-06,
      "loss": 0.5897,
      "step": 4686
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7243751509868331,
      "learning_rate": 4.199792268100318e-06,
      "loss": 0.5993,
      "step": 4687
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6505885114500449,
      "learning_rate": 4.1958178941176355e-06,
      "loss": 0.5786,
      "step": 4688
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6695525049531456,
      "learning_rate": 4.191844902232896e-06,
      "loss": 0.5841,
      "step": 4689
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5790027137181964,
      "learning_rate": 4.187873293392161e-06,
      "loss": 0.5551,
      "step": 4690
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6369169655922539,
      "learning_rate": 4.183903068541146e-06,
      "loss": 0.5616,
      "step": 4691
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5707912282999587,
      "learning_rate": 4.17993422862525e-06,
      "loss": 0.5674,
      "step": 4692
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6656023546932154,
      "learning_rate": 4.1759667745895395e-06,
      "loss": 0.5827,
      "step": 4693
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5920551264718906,
      "learning_rate": 4.1720007073787425e-06,
      "loss": 0.5656,
      "step": 4694
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5902492707176794,
      "learning_rate": 4.168036027937267e-06,
      "loss": 0.5798,
      "step": 4695
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6115281485442007,
      "learning_rate": 4.164072737209186e-06,
      "loss": 0.5616,
      "step": 4696
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6091759089661132,
      "learning_rate": 4.160110836138246e-06,
      "loss": 0.5896,
      "step": 4697
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.602917223407156,
      "learning_rate": 4.156150325667856e-06,
      "loss": 0.59,
      "step": 4698
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6007293779765897,
      "learning_rate": 4.15219120674109e-06,
      "loss": 0.5722,
      "step": 4699
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5908707609038529,
      "learning_rate": 4.148233480300709e-06,
      "loss": 0.5566,
      "step": 4700
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6432383105951237,
      "learning_rate": 4.144277147289125e-06,
      "loss": 0.5533,
      "step": 4701
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7349196550825818,
      "learning_rate": 4.140322208648428e-06,
      "loss": 0.6357,
      "step": 4702
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.587499929720271,
      "learning_rate": 4.136368665320366e-06,
      "loss": 0.5663,
      "step": 4703
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5833532031608747,
      "learning_rate": 4.132416518246365e-06,
      "loss": 0.5463,
      "step": 4704
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5826416486064488,
      "learning_rate": 4.1284657683675164e-06,
      "loss": 0.576,
      "step": 4705
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6053236078485797,
      "learning_rate": 4.124516416624571e-06,
      "loss": 0.5664,
      "step": 4706
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6700755057076427,
      "learning_rate": 4.120568463957956e-06,
      "loss": 0.5698,
      "step": 4707
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6066094996499679,
      "learning_rate": 4.116621911307765e-06,
      "loss": 0.5504,
      "step": 4708
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5662613523572188,
      "learning_rate": 4.11267675961375e-06,
      "loss": 0.5695,
      "step": 4709
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5751010742562032,
      "learning_rate": 4.108733009815335e-06,
      "loss": 0.5781,
      "step": 4710
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6644122720906511,
      "learning_rate": 4.104790662851613e-06,
      "loss": 0.5577,
      "step": 4711
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.614576298550255,
      "learning_rate": 4.1008497196613415e-06,
      "loss": 0.5849,
      "step": 4712
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6473035465503426,
      "learning_rate": 4.0969101811829406e-06,
      "loss": 0.5486,
      "step": 4713
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5433896605720661,
      "learning_rate": 4.092972048354491e-06,
      "loss": 0.5823,
      "step": 4714
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5840606019152873,
      "learning_rate": 4.089035322113749e-06,
      "loss": 0.557,
      "step": 4715
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7052910408677534,
      "learning_rate": 4.085100003398138e-06,
      "loss": 0.6008,
      "step": 4716
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6298340990016208,
      "learning_rate": 4.0811660931447305e-06,
      "loss": 0.5854,
      "step": 4717
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6403121375413304,
      "learning_rate": 4.077233592290279e-06,
      "loss": 0.5467,
      "step": 4718
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6362660864582792,
      "learning_rate": 4.073302501771192e-06,
      "loss": 0.5684,
      "step": 4719
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6686533863185768,
      "learning_rate": 4.069372822523552e-06,
      "loss": 0.5719,
      "step": 4720
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.671615696975606,
      "learning_rate": 4.065444555483091e-06,
      "loss": 0.5886,
      "step": 4721
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6278050634175764,
      "learning_rate": 4.061517701585208e-06,
      "loss": 0.5837,
      "step": 4722
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7380232863825429,
      "learning_rate": 4.057592261764982e-06,
      "loss": 0.5949,
      "step": 4723
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5789172284907852,
      "learning_rate": 4.053668236957135e-06,
      "loss": 0.581,
      "step": 4724
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5791560716340115,
      "learning_rate": 4.049745628096062e-06,
      "loss": 0.559,
      "step": 4725
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6229463695944782,
      "learning_rate": 4.045824436115816e-06,
      "loss": 0.5607,
      "step": 4726
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6295970572545052,
      "learning_rate": 4.041904661950118e-06,
      "loss": 0.5887,
      "step": 4727
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6794674106771974,
      "learning_rate": 4.037986306532351e-06,
      "loss": 0.563,
      "step": 4728
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6601136161316831,
      "learning_rate": 4.034069370795552e-06,
      "loss": 0.582,
      "step": 4729
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6517327172009642,
      "learning_rate": 4.030153855672428e-06,
      "loss": 0.5905,
      "step": 4730
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7385459120759347,
      "learning_rate": 4.026239762095351e-06,
      "loss": 0.5969,
      "step": 4731
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.66887140902903,
      "learning_rate": 4.0223270909963405e-06,
      "loss": 0.5783,
      "step": 4732
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6941334825297242,
      "learning_rate": 4.018415843307094e-06,
      "loss": 0.6025,
      "step": 4733
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5820249471041833,
      "learning_rate": 4.01450601995895e-06,
      "loss": 0.5512,
      "step": 4734
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6595553847414313,
      "learning_rate": 4.010597621882936e-06,
      "loss": 0.5827,
      "step": 4735
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6133667164881617,
      "learning_rate": 4.0066906500097145e-06,
      "loss": 0.5529,
      "step": 4736
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6251587941044348,
      "learning_rate": 4.002785105269616e-06,
      "loss": 0.5691,
      "step": 4737
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5923546680690608,
      "learning_rate": 3.998880988592635e-06,
      "loss": 0.5916,
      "step": 4738
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0953627508642432,
      "learning_rate": 3.994978300908425e-06,
      "loss": 0.6205,
      "step": 4739
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6177098432435544,
      "learning_rate": 3.991077043146302e-06,
      "loss": 0.5663,
      "step": 4740
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6344891277977248,
      "learning_rate": 3.987177216235229e-06,
      "loss": 0.5771,
      "step": 4741
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6306168013782731,
      "learning_rate": 3.98327882110384e-06,
      "loss": 0.5812,
      "step": 4742
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5629630364230869,
      "learning_rate": 3.979381858680431e-06,
      "loss": 0.5649,
      "step": 4743
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6891713664948772,
      "learning_rate": 3.9754863298929414e-06,
      "loss": 0.5692,
      "step": 4744
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6573455381963512,
      "learning_rate": 3.971592235668984e-06,
      "loss": 0.5793,
      "step": 4745
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7339013138802403,
      "learning_rate": 3.967699576935828e-06,
      "loss": 0.5701,
      "step": 4746
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6066774921147631,
      "learning_rate": 3.963808354620388e-06,
      "loss": 0.5368,
      "step": 4747
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5413564326266515,
      "learning_rate": 3.959918569649255e-06,
      "loss": 0.5556,
      "step": 4748
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5689175561414759,
      "learning_rate": 3.9560302229486604e-06,
      "loss": 0.5413,
      "step": 4749
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6188567304264009,
      "learning_rate": 3.952143315444506e-06,
      "loss": 0.5773,
      "step": 4750
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6807730163506522,
      "learning_rate": 3.948257848062351e-06,
      "loss": 0.5766,
      "step": 4751
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6624202701554652,
      "learning_rate": 3.944373821727398e-06,
      "loss": 0.5956,
      "step": 4752
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6418323432208665,
      "learning_rate": 3.940491237364519e-06,
      "loss": 0.5997,
      "step": 4753
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6135146562468128,
      "learning_rate": 3.9366100958982425e-06,
      "loss": 0.5875,
      "step": 4754
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6184144784106657,
      "learning_rate": 3.9327303982527445e-06,
      "loss": 0.5872,
      "step": 4755
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7093658458969571,
      "learning_rate": 3.928852145351868e-06,
      "loss": 0.5723,
      "step": 4756
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6118736977619681,
      "learning_rate": 3.924975338119097e-06,
      "loss": 0.5491,
      "step": 4757
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6313506978498433,
      "learning_rate": 3.921099977477595e-06,
      "loss": 0.5601,
      "step": 4758
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6454194240215158,
      "learning_rate": 3.917226064350159e-06,
      "loss": 0.5906,
      "step": 4759
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5618495784257236,
      "learning_rate": 3.913353599659248e-06,
      "loss": 0.541,
      "step": 4760
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6025163434307085,
      "learning_rate": 3.9094825843269775e-06,
      "loss": 0.5756,
      "step": 4761
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.651244532719023,
      "learning_rate": 3.905613019275122e-06,
      "loss": 0.5729,
      "step": 4762
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6176166423709457,
      "learning_rate": 3.9017449054251055e-06,
      "loss": 0.582,
      "step": 4763
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6616436058627608,
      "learning_rate": 3.897878243698004e-06,
      "loss": 0.5805,
      "step": 4764
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6003162810842915,
      "learning_rate": 3.8940130350145515e-06,
      "loss": 0.569,
      "step": 4765
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6734189165146053,
      "learning_rate": 3.8901492802951404e-06,
      "loss": 0.5676,
      "step": 4766
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6396653675936403,
      "learning_rate": 3.886286980459806e-06,
      "loss": 0.5938,
      "step": 4767
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6280792282559482,
      "learning_rate": 3.882426136428245e-06,
      "loss": 0.5642,
      "step": 4768
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6519263096477462,
      "learning_rate": 3.878566749119809e-06,
      "loss": 0.5992,
      "step": 4769
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5721741689816355,
      "learning_rate": 3.874708819453496e-06,
      "loss": 0.5635,
      "step": 4770
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6554588090794337,
      "learning_rate": 3.870852348347963e-06,
      "loss": 0.5819,
      "step": 4771
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5868771027702377,
      "learning_rate": 3.866997336721511e-06,
      "loss": 0.5403,
      "step": 4772
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6015559258098505,
      "learning_rate": 3.8631437854921035e-06,
      "loss": 0.5575,
      "step": 4773
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5847333356457607,
      "learning_rate": 3.859291695577358e-06,
      "loss": 0.5796,
      "step": 4774
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6444082968975396,
      "learning_rate": 3.8554410678945265e-06,
      "loss": 0.5674,
      "step": 4775
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5915726249956188,
      "learning_rate": 3.851591903360531e-06,
      "loss": 0.574,
      "step": 4776
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5913669874029593,
      "learning_rate": 3.847744202891938e-06,
      "loss": 0.5903,
      "step": 4777
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5433202161407228,
      "learning_rate": 3.843897967404968e-06,
      "loss": 0.5604,
      "step": 4778
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6010153123142411,
      "learning_rate": 3.84005319781549e-06,
      "loss": 0.5814,
      "step": 4779
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8272033724777454,
      "learning_rate": 3.836209895039016e-06,
      "loss": 0.6091,
      "step": 4780
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8276090176965448,
      "learning_rate": 3.832368059990732e-06,
      "loss": 0.6288,
      "step": 4781
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6507189949208821,
      "learning_rate": 3.828527693585451e-06,
      "loss": 0.6171,
      "step": 4782
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6563303799079564,
      "learning_rate": 3.824688796737643e-06,
      "loss": 0.55,
      "step": 4783
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5825416293290532,
      "learning_rate": 3.820851370361435e-06,
      "loss": 0.568,
      "step": 4784
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6116220946784333,
      "learning_rate": 3.817015415370596e-06,
      "loss": 0.567,
      "step": 4785
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6238431401852425,
      "learning_rate": 3.813180932678553e-06,
      "loss": 0.5687,
      "step": 4786
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5626110894474766,
      "learning_rate": 3.80934792319837e-06,
      "loss": 0.5547,
      "step": 4787
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6617262169111965,
      "learning_rate": 3.8055163878427703e-06,
      "loss": 0.5534,
      "step": 4788
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6468909577945657,
      "learning_rate": 3.801686327524126e-06,
      "loss": 0.556,
      "step": 4789
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6546206975944056,
      "learning_rate": 3.7978577431544495e-06,
      "loss": 0.578,
      "step": 4790
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5993617938857834,
      "learning_rate": 3.7940306356454136e-06,
      "loss": 0.5568,
      "step": 4791
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.710086384505655,
      "learning_rate": 3.7902050059083264e-06,
      "loss": 0.5559,
      "step": 4792
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6086285807666401,
      "learning_rate": 3.7863808548541536e-06,
      "loss": 0.5795,
      "step": 4793
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6359851151813504,
      "learning_rate": 3.782558183393511e-06,
      "loss": 0.5599,
      "step": 4794
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6723246131862316,
      "learning_rate": 3.778736992436649e-06,
      "loss": 0.5805,
      "step": 4795
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5510574144549232,
      "learning_rate": 3.7749172828934777e-06,
      "loss": 0.5533,
      "step": 4796
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6410588883971925,
      "learning_rate": 3.771099055673553e-06,
      "loss": 0.5857,
      "step": 4797
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6484975292738744,
      "learning_rate": 3.767282311686069e-06,
      "loss": 0.5987,
      "step": 4798
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6053183484759092,
      "learning_rate": 3.7634670518398764e-06,
      "loss": 0.5519,
      "step": 4799
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5913576515843129,
      "learning_rate": 3.759653277043469e-06,
      "loss": 0.5667,
      "step": 4800
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6445779040332643,
      "learning_rate": 3.7558409882049897e-06,
      "loss": 0.5853,
      "step": 4801
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6297061618102241,
      "learning_rate": 3.7520301862322207e-06,
      "loss": 0.553,
      "step": 4802
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6668137355879361,
      "learning_rate": 3.7482208720325886e-06,
      "loss": 0.5589,
      "step": 4803
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5950909096651876,
      "learning_rate": 3.7444130465131836e-06,
      "loss": 0.5625,
      "step": 4804
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7205231301317352,
      "learning_rate": 3.740606710580721e-06,
      "loss": 0.5661,
      "step": 4805
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.632429686797222,
      "learning_rate": 3.7368018651415738e-06,
      "loss": 0.5717,
      "step": 4806
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6476550256546226,
      "learning_rate": 3.732998511101752e-06,
      "loss": 0.5698,
      "step": 4807
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6123850095981735,
      "learning_rate": 3.729196649366914e-06,
      "loss": 0.5692,
      "step": 4808
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6650812079248767,
      "learning_rate": 3.725396280842369e-06,
      "loss": 0.5832,
      "step": 4809
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6292395412111638,
      "learning_rate": 3.721597406433062e-06,
      "loss": 0.585,
      "step": 4810
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6713335613617184,
      "learning_rate": 3.7178000270435765e-06,
      "loss": 0.585,
      "step": 4811
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5716483407928636,
      "learning_rate": 3.7140041435781616e-06,
      "loss": 0.554,
      "step": 4812
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6543988614105887,
      "learning_rate": 3.7102097569406893e-06,
      "loss": 0.5854,
      "step": 4813
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6867147054825022,
      "learning_rate": 3.706416868034687e-06,
      "loss": 0.5706,
      "step": 4814
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6065874867539446,
      "learning_rate": 3.7026254777633164e-06,
      "loss": 0.5663,
      "step": 4815
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6723731242219281,
      "learning_rate": 3.698835587029389e-06,
      "loss": 0.5603,
      "step": 4816
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7399585307373505,
      "learning_rate": 3.6950471967353642e-06,
      "loss": 0.5765,
      "step": 4817
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6735043968045528,
      "learning_rate": 3.6912603077833274e-06,
      "loss": 0.5633,
      "step": 4818
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6475955364086996,
      "learning_rate": 3.6874749210750207e-06,
      "loss": 0.5816,
      "step": 4819
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5944364602563974,
      "learning_rate": 3.6836910375118294e-06,
      "loss": 0.5737,
      "step": 4820
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5649206176813368,
      "learning_rate": 3.6799086579947674e-06,
      "loss": 0.5563,
      "step": 4821
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5279839150892907,
      "learning_rate": 3.6761277834245023e-06,
      "loss": 0.5669,
      "step": 4822
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5774431627738701,
      "learning_rate": 3.672348414701341e-06,
      "loss": 0.5665,
      "step": 4823
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6655095677278376,
      "learning_rate": 3.6685705527252337e-06,
      "loss": 0.5829,
      "step": 4824
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6483900003989171,
      "learning_rate": 3.6647941983957647e-06,
      "loss": 0.5788,
      "step": 4825
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6148422528158296,
      "learning_rate": 3.661019352612162e-06,
      "loss": 0.5682,
      "step": 4826
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6252042975628321,
      "learning_rate": 3.657246016273297e-06,
      "loss": 0.5664,
      "step": 4827
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6059197475531111,
      "learning_rate": 3.6534741902776816e-06,
      "loss": 0.5723,
      "step": 4828
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6167647693377706,
      "learning_rate": 3.6497038755234703e-06,
      "loss": 0.5791,
      "step": 4829
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5913899998752349,
      "learning_rate": 3.6459350729084473e-06,
      "loss": 0.566,
      "step": 4830
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7052439925579611,
      "learning_rate": 3.6421677833300474e-06,
      "loss": 0.573,
      "step": 4831
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6310824161324354,
      "learning_rate": 3.6384020076853453e-06,
      "loss": 0.5894,
      "step": 4832
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6809692922163272,
      "learning_rate": 3.634637746871045e-06,
      "loss": 0.596,
      "step": 4833
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6081201476406284,
      "learning_rate": 3.6308750017834984e-06,
      "loss": 0.5677,
      "step": 4834
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6344754654919621,
      "learning_rate": 3.6271137733186977e-06,
      "loss": 0.5892,
      "step": 4835
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6038818424088188,
      "learning_rate": 3.6233540623722662e-06,
      "loss": 0.5801,
      "step": 4836
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6268065735268639,
      "learning_rate": 3.619595869839474e-06,
      "loss": 0.5636,
      "step": 4837
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6138195670569694,
      "learning_rate": 3.6158391966152175e-06,
      "loss": 0.5513,
      "step": 4838
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7291872319222982,
      "learning_rate": 3.6120840435940517e-06,
      "loss": 0.614,
      "step": 4839
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6264389123274916,
      "learning_rate": 3.6083304116701535e-06,
      "loss": 0.5633,
      "step": 4840
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6897152482900546,
      "learning_rate": 3.604578301737336e-06,
      "loss": 0.5608,
      "step": 4841
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6720324782639503,
      "learning_rate": 3.600827714689059e-06,
      "loss": 0.5836,
      "step": 4842
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6386390305301177,
      "learning_rate": 3.59707865141842e-06,
      "loss": 0.5574,
      "step": 4843
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6382198245427035,
      "learning_rate": 3.593331112818144e-06,
      "loss": 0.5581,
      "step": 4844
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6139949825236831,
      "learning_rate": 3.5895850997806016e-06,
      "loss": 0.5876,
      "step": 4845
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6108298423822393,
      "learning_rate": 3.5858406131977965e-06,
      "loss": 0.5737,
      "step": 4846
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.707977017899423,
      "learning_rate": 3.5820976539613738e-06,
      "loss": 0.5915,
      "step": 4847
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6671911735397763,
      "learning_rate": 3.5783562229626078e-06,
      "loss": 0.5878,
      "step": 4848
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6163256360245746,
      "learning_rate": 3.574616321092409e-06,
      "loss": 0.5917,
      "step": 4849
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7091289507692232,
      "learning_rate": 3.5708779492413294e-06,
      "loss": 0.606,
      "step": 4850
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6195013881878318,
      "learning_rate": 3.567141108299554e-06,
      "loss": 0.5824,
      "step": 4851
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6162855319407782,
      "learning_rate": 3.5634057991569082e-06,
      "loss": 0.5483,
      "step": 4852
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6317603745691336,
      "learning_rate": 3.5596720227028382e-06,
      "loss": 0.5755,
      "step": 4853
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6082610896981095,
      "learning_rate": 3.5559397798264404e-06,
      "loss": 0.5676,
      "step": 4854
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6794619432367333,
      "learning_rate": 3.5522090714164425e-06,
      "loss": 0.576,
      "step": 4855
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6139098627190821,
      "learning_rate": 3.548479898361199e-06,
      "loss": 0.5414,
      "step": 4856
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6513325548650157,
      "learning_rate": 3.544752261548706e-06,
      "loss": 0.6071,
      "step": 4857
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6232360317408926,
      "learning_rate": 3.5410261618665985e-06,
      "loss": 0.5588,
      "step": 4858
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6505660397495355,
      "learning_rate": 3.537301600202131e-06,
      "loss": 0.5672,
      "step": 4859
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6199537138453127,
      "learning_rate": 3.533578577442206e-06,
      "loss": 0.5719,
      "step": 4860
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6061791386224543,
      "learning_rate": 3.5298570944733447e-06,
      "loss": 0.5838,
      "step": 4861
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6605273786471234,
      "learning_rate": 3.5261371521817247e-06,
      "loss": 0.5733,
      "step": 4862
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6392554912415068,
      "learning_rate": 3.522418751453133e-06,
      "loss": 0.5286,
      "step": 4863
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5747044878978673,
      "learning_rate": 3.5187018931729987e-06,
      "loss": 0.5835,
      "step": 4864
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5732054121837796,
      "learning_rate": 3.514986578226386e-06,
      "loss": 0.5641,
      "step": 4865
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5816450530080808,
      "learning_rate": 3.5112728074979896e-06,
      "loss": 0.587,
      "step": 4866
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6233515886237524,
      "learning_rate": 3.507560581872139e-06,
      "loss": 0.5597,
      "step": 4867
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7159381187118645,
      "learning_rate": 3.503849902232792e-06,
      "loss": 0.5958,
      "step": 4868
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6389459274316664,
      "learning_rate": 3.5001407694635326e-06,
      "loss": 0.579,
      "step": 4869
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5919630761353694,
      "learning_rate": 3.4964331844475953e-06,
      "loss": 0.535,
      "step": 4870
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5686737242109995,
      "learning_rate": 3.492727148067824e-06,
      "loss": 0.5746,
      "step": 4871
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6403157982615575,
      "learning_rate": 3.4890226612067124e-06,
      "loss": 0.5581,
      "step": 4872
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6783670189590135,
      "learning_rate": 3.485319724746369e-06,
      "loss": 0.5655,
      "step": 4873
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.588470549960733,
      "learning_rate": 3.4816183395685433e-06,
      "loss": 0.5538,
      "step": 4874
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6885776221669804,
      "learning_rate": 3.4779185065546183e-06,
      "loss": 0.5838,
      "step": 4875
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6491916072829085,
      "learning_rate": 3.474220226585595e-06,
      "loss": 0.5775,
      "step": 4876
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6019915973196324,
      "learning_rate": 3.4705235005421144e-06,
      "loss": 0.5465,
      "step": 4877
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6068723597813861,
      "learning_rate": 3.4668283293044492e-06,
      "loss": 0.553,
      "step": 4878
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6705999244493251,
      "learning_rate": 3.4631347137524896e-06,
      "loss": 0.5913,
      "step": 4879
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6651310482015131,
      "learning_rate": 3.4594426547657667e-06,
      "loss": 0.6054,
      "step": 4880
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6737207345631887,
      "learning_rate": 3.4557521532234405e-06,
      "loss": 0.5683,
      "step": 4881
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6431851608302372,
      "learning_rate": 3.4520632100042916e-06,
      "loss": 0.5799,
      "step": 4882
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.651833992213339,
      "learning_rate": 3.448375825986742e-06,
      "loss": 0.5771,
      "step": 4883
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6840281270859886,
      "learning_rate": 3.444690002048826e-06,
      "loss": 0.5705,
      "step": 4884
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6358576434153252,
      "learning_rate": 3.441005739068223e-06,
      "loss": 0.5821,
      "step": 4885
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6334151618934964,
      "learning_rate": 3.4373230379222344e-06,
      "loss": 0.5698,
      "step": 4886
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6383764181868641,
      "learning_rate": 3.433641899487783e-06,
      "loss": 0.558,
      "step": 4887
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.672039996833442,
      "learning_rate": 3.4299623246414283e-06,
      "loss": 0.5712,
      "step": 4888
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5857663344591096,
      "learning_rate": 3.4262843142593536e-06,
      "loss": 0.574,
      "step": 4889
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6404908165165167,
      "learning_rate": 3.422607869217377e-06,
      "loss": 0.5734,
      "step": 4890
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7346522782950287,
      "learning_rate": 3.4189329903909307e-06,
      "loss": 0.5722,
      "step": 4891
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5513104982417493,
      "learning_rate": 3.4152596786550764e-06,
      "loss": 0.5514,
      "step": 4892
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6009078215753282,
      "learning_rate": 3.4115879348845194e-06,
      "loss": 0.5643,
      "step": 4893
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6226666126544305,
      "learning_rate": 3.4079177599535695e-06,
      "loss": 0.5523,
      "step": 4894
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6779605553799631,
      "learning_rate": 3.404249154736179e-06,
      "loss": 0.5818,
      "step": 4895
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5682407990938658,
      "learning_rate": 3.400582120105913e-06,
      "loss": 0.5673,
      "step": 4896
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6344644989362542,
      "learning_rate": 3.3969166569359734e-06,
      "loss": 0.5899,
      "step": 4897
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6585532321687773,
      "learning_rate": 3.3932527660991877e-06,
      "loss": 0.5674,
      "step": 4898
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6151550194595674,
      "learning_rate": 3.3895904484679986e-06,
      "loss": 0.5498,
      "step": 4899
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5837807158749906,
      "learning_rate": 3.3859297049144833e-06,
      "loss": 0.5761,
      "step": 4900
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6464358585650078,
      "learning_rate": 3.382270536310347e-06,
      "loss": 0.5832,
      "step": 4901
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7131654447751339,
      "learning_rate": 3.3786129435269076e-06,
      "loss": 0.6026,
      "step": 4902
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6577843132479008,
      "learning_rate": 3.374956927435118e-06,
      "loss": 0.5675,
      "step": 4903
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6604734836765688,
      "learning_rate": 3.3713024889055557e-06,
      "loss": 0.5798,
      "step": 4904
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7292060807768278,
      "learning_rate": 3.3676496288084127e-06,
      "loss": 0.584,
      "step": 4905
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.638622160269619,
      "learning_rate": 3.3639983480135197e-06,
      "loss": 0.5637,
      "step": 4906
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5983235001694759,
      "learning_rate": 3.3603486473903147e-06,
      "loss": 0.5363,
      "step": 4907
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5907648286762036,
      "learning_rate": 3.3567005278078736e-06,
      "loss": 0.5474,
      "step": 4908
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6088332578663829,
      "learning_rate": 3.3530539901348937e-06,
      "loss": 0.5831,
      "step": 4909
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5845459889918345,
      "learning_rate": 3.349409035239685e-06,
      "loss": 0.5472,
      "step": 4910
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6180248525698573,
      "learning_rate": 3.3457656639901903e-06,
      "loss": 0.5799,
      "step": 4911
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.589858291438929,
      "learning_rate": 3.3421238772539755e-06,
      "loss": 0.5659,
      "step": 4912
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6747874480221078,
      "learning_rate": 3.3384836758982277e-06,
      "loss": 0.5719,
      "step": 4913
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6203677031757725,
      "learning_rate": 3.334845060789753e-06,
      "loss": 0.5738,
      "step": 4914
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6327610936445632,
      "learning_rate": 3.331208032794977e-06,
      "loss": 0.6096,
      "step": 4915
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6076713109655137,
      "learning_rate": 3.327572592779963e-06,
      "loss": 0.5667,
      "step": 4916
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5631192642169841,
      "learning_rate": 3.3239387416103786e-06,
      "loss": 0.5404,
      "step": 4917
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6876523136158139,
      "learning_rate": 3.320306480151526e-06,
      "loss": 0.5892,
      "step": 4918
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5752516665125551,
      "learning_rate": 3.316675809268316e-06,
      "loss": 0.5818,
      "step": 4919
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6224351718969509,
      "learning_rate": 3.313046729825291e-06,
      "loss": 0.5698,
      "step": 4920
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6345004496002764,
      "learning_rate": 3.3094192426866144e-06,
      "loss": 0.5549,
      "step": 4921
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6472131303565551,
      "learning_rate": 3.305793348716062e-06,
      "loss": 0.5668,
      "step": 4922
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5957331125848221,
      "learning_rate": 3.3021690487770374e-06,
      "loss": 0.5524,
      "step": 4923
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6538635360693924,
      "learning_rate": 3.298546343732567e-06,
      "loss": 0.5587,
      "step": 4924
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6191638950941867,
      "learning_rate": 3.2949252344452855e-06,
      "loss": 0.5839,
      "step": 4925
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.556493836440054,
      "learning_rate": 3.2913057217774636e-06,
      "loss": 0.5331,
      "step": 4926
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5791952521802086,
      "learning_rate": 3.2876878065909714e-06,
      "loss": 0.561,
      "step": 4927
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7210625560790375,
      "learning_rate": 3.2840714897473247e-06,
      "loss": 0.6118,
      "step": 4928
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5928311926760486,
      "learning_rate": 3.2804567721076385e-06,
      "loss": 0.5431,
      "step": 4929
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6161168841272254,
      "learning_rate": 3.276843654532649e-06,
      "loss": 0.5724,
      "step": 4930
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5880912396282438,
      "learning_rate": 3.2732321378827204e-06,
      "loss": 0.582,
      "step": 4931
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5926905658699098,
      "learning_rate": 3.2696222230178286e-06,
      "loss": 0.5603,
      "step": 4932
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7004116369772495,
      "learning_rate": 3.2660139107975764e-06,
      "loss": 0.5933,
      "step": 4933
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.702767705198501,
      "learning_rate": 3.2624072020811703e-06,
      "loss": 0.5921,
      "step": 4934
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6143455846434712,
      "learning_rate": 3.258802097727447e-06,
      "loss": 0.5689,
      "step": 4935
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5985944625929754,
      "learning_rate": 3.255198598594862e-06,
      "loss": 0.5458,
      "step": 4936
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6421266898016232,
      "learning_rate": 3.25159670554148e-06,
      "loss": 0.5725,
      "step": 4937
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.584887508873097,
      "learning_rate": 3.2479964194249813e-06,
      "loss": 0.5649,
      "step": 4938
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6532526377650075,
      "learning_rate": 3.244397741102683e-06,
      "loss": 0.5804,
      "step": 4939
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5882588825832014,
      "learning_rate": 3.2408006714314967e-06,
      "loss": 0.5568,
      "step": 4940
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6346159435025871,
      "learning_rate": 3.2372052112679666e-06,
      "loss": 0.5627,
      "step": 4941
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5706211886254408,
      "learning_rate": 3.2336113614682405e-06,
      "loss": 0.5819,
      "step": 4942
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5840237576797104,
      "learning_rate": 3.230019122888094e-06,
      "loss": 0.5623,
      "step": 4943
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7574421538238267,
      "learning_rate": 3.2264284963829175e-06,
      "loss": 0.5961,
      "step": 4944
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5787875887120599,
      "learning_rate": 3.222839482807707e-06,
      "loss": 0.5543,
      "step": 4945
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6281429896992078,
      "learning_rate": 3.2192520830170882e-06,
      "loss": 0.5815,
      "step": 4946
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5517104133475504,
      "learning_rate": 3.2156662978652975e-06,
      "loss": 0.5391,
      "step": 4947
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6045723195098518,
      "learning_rate": 3.2120821282061798e-06,
      "loss": 0.5801,
      "step": 4948
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5559502874289899,
      "learning_rate": 3.208499574893208e-06,
      "loss": 0.5603,
      "step": 4949
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5896942986753538,
      "learning_rate": 3.2049186387794538e-06,
      "loss": 0.5745,
      "step": 4950
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6469405035185517,
      "learning_rate": 3.2013393207176267e-06,
      "loss": 0.5787,
      "step": 4951
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5706714362240053,
      "learning_rate": 3.1977616215600304e-06,
      "loss": 0.5617,
      "step": 4952
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6636625805559315,
      "learning_rate": 3.1941855421585876e-06,
      "loss": 0.5848,
      "step": 4953
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6839579476027613,
      "learning_rate": 3.1906110833648417e-06,
      "loss": 0.6043,
      "step": 4954
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.621784229452243,
      "learning_rate": 3.1870382460299466e-06,
      "loss": 0.5591,
      "step": 4955
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.583926987145605,
      "learning_rate": 3.1834670310046735e-06,
      "loss": 0.5625,
      "step": 4956
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5854267740894892,
      "learning_rate": 3.1798974391393953e-06,
      "loss": 0.577,
      "step": 4957
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6001235842542424,
      "learning_rate": 3.176329471284113e-06,
      "loss": 0.536,
      "step": 4958
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6194048029770127,
      "learning_rate": 3.1727631282884363e-06,
      "loss": 0.5685,
      "step": 4959
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6711976168615612,
      "learning_rate": 3.1691984110015818e-06,
      "loss": 0.5514,
      "step": 4960
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6059684296402201,
      "learning_rate": 3.1656353202723876e-06,
      "loss": 0.5531,
      "step": 4961
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6234294576852881,
      "learning_rate": 3.162073856949296e-06,
      "loss": 0.564,
      "step": 4962
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5856275168385675,
      "learning_rate": 3.15851402188037e-06,
      "loss": 0.5802,
      "step": 4963
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2714728637874462,
      "learning_rate": 3.154955815913283e-06,
      "loss": 0.5952,
      "step": 4964
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6112321879517744,
      "learning_rate": 3.151399239895313e-06,
      "loss": 0.5676,
      "step": 4965
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6353182936760294,
      "learning_rate": 3.147844294673359e-06,
      "loss": 0.5864,
      "step": 4966
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5910606480960338,
      "learning_rate": 3.1442909810939316e-06,
      "loss": 0.5548,
      "step": 4967
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6197019127449298,
      "learning_rate": 3.1407393000031426e-06,
      "loss": 0.5812,
      "step": 4968
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6275171246627468,
      "learning_rate": 3.137189252246726e-06,
      "loss": 0.5378,
      "step": 4969
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.606758990180046,
      "learning_rate": 3.1336408386700256e-06,
      "loss": 0.5541,
      "step": 4970
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6162346253935688,
      "learning_rate": 3.130094060117986e-06,
      "loss": 0.5531,
      "step": 4971
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6710694124730192,
      "learning_rate": 3.126548917435179e-06,
      "loss": 0.571,
      "step": 4972
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.64082836484089,
      "learning_rate": 3.123005411465766e-06,
      "loss": 0.561,
      "step": 4973
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6554857275980609,
      "learning_rate": 3.1194635430535426e-06,
      "loss": 0.57,
      "step": 4974
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7329231559420089,
      "learning_rate": 3.1159233130418975e-06,
      "loss": 0.5716,
      "step": 4975
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5992060637366844,
      "learning_rate": 3.11238472227383e-06,
      "loss": 0.5516,
      "step": 4976
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6523296251182931,
      "learning_rate": 3.108847771591956e-06,
      "loss": 0.5532,
      "step": 4977
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6517750641888723,
      "learning_rate": 3.105312461838499e-06,
      "loss": 0.5706,
      "step": 4978
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5725324187609794,
      "learning_rate": 3.1017787938552925e-06,
      "loss": 0.5629,
      "step": 4979
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5446128826703162,
      "learning_rate": 3.0982467684837724e-06,
      "loss": 0.5605,
      "step": 4980
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6504710806925246,
      "learning_rate": 3.0947163865649897e-06,
      "loss": 0.5789,
      "step": 4981
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5995844956954457,
      "learning_rate": 3.0911876489396063e-06,
      "loss": 0.5465,
      "step": 4982
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7043147843536065,
      "learning_rate": 3.0876605564478832e-06,
      "loss": 0.5547,
      "step": 4983
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5937616704210618,
      "learning_rate": 3.0841351099297025e-06,
      "loss": 0.5738,
      "step": 4984
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7008550991009201,
      "learning_rate": 3.0806113102245395e-06,
      "loss": 0.5715,
      "step": 4985
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6418530073893597,
      "learning_rate": 3.0770891581714877e-06,
      "loss": 0.5555,
      "step": 4986
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7388376925379818,
      "learning_rate": 3.0735686546092514e-06,
      "loss": 0.6039,
      "step": 4987
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6466771035062036,
      "learning_rate": 3.070049800376127e-06,
      "loss": 0.5585,
      "step": 4988
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6488497500990696,
      "learning_rate": 3.0665325963100334e-06,
      "loss": 0.586,
      "step": 4989
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6547335339718485,
      "learning_rate": 3.063017043248493e-06,
      "loss": 0.5725,
      "step": 4990
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5854884961010371,
      "learning_rate": 3.059503142028627e-06,
      "loss": 0.5319,
      "step": 4991
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5967499850981272,
      "learning_rate": 3.055990893487173e-06,
      "loss": 0.5777,
      "step": 4992
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6088687104748702,
      "learning_rate": 3.0524802984604694e-06,
      "loss": 0.5645,
      "step": 4993
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6210531024151024,
      "learning_rate": 3.0489713577844683e-06,
      "loss": 0.5485,
      "step": 4994
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.622708317570684,
      "learning_rate": 3.045464072294717e-06,
      "loss": 0.5897,
      "step": 4995
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.622415794900689,
      "learning_rate": 3.0419584428263692e-06,
      "loss": 0.57,
      "step": 4996
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7038590684912528,
      "learning_rate": 3.038454470214203e-06,
      "loss": 0.6,
      "step": 4997
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6646887948723207,
      "learning_rate": 3.0349521552925774e-06,
      "loss": 0.5662,
      "step": 4998
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5575546569931839,
      "learning_rate": 3.031451498895468e-06,
      "loss": 0.5441,
      "step": 4999
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6898703444580082,
      "learning_rate": 3.027952501856457e-06,
      "loss": 0.5405,
      "step": 5000
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7070872762605988,
      "learning_rate": 3.0244551650087286e-06,
      "loss": 0.6023,
      "step": 5001
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.613408676540853,
      "learning_rate": 3.0209594891850757e-06,
      "loss": 0.5878,
      "step": 5002
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6246481704788291,
      "learning_rate": 3.01746547521789e-06,
      "loss": 0.5711,
      "step": 5003
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5876732940528254,
      "learning_rate": 3.0139731239391625e-06,
      "loss": 0.6,
      "step": 5004
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.625518208532328,
      "learning_rate": 3.0104824361805074e-06,
      "loss": 0.5558,
      "step": 5005
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.58745566019133,
      "learning_rate": 3.006993412773124e-06,
      "loss": 0.5451,
      "step": 5006
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5338508612845311,
      "learning_rate": 3.003506054547827e-06,
      "loss": 0.5525,
      "step": 5007
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6576288864341766,
      "learning_rate": 3.0000203623350223e-06,
      "loss": 0.5593,
      "step": 5008
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5537225234249081,
      "learning_rate": 2.9965363369647317e-06,
      "loss": 0.568,
      "step": 5009
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7403165458062899,
      "learning_rate": 2.9930539792665767e-06,
      "loss": 0.5678,
      "step": 5010
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6291082246814355,
      "learning_rate": 2.989573290069776e-06,
      "loss": 0.5779,
      "step": 5011
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6345200345155974,
      "learning_rate": 2.986094270203156e-06,
      "loss": 0.5607,
      "step": 5012
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.610932667880943,
      "learning_rate": 2.982616920495147e-06,
      "loss": 0.5698,
      "step": 5013
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6411441869828944,
      "learning_rate": 2.979141241773775e-06,
      "loss": 0.5815,
      "step": 5014
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5926304845917117,
      "learning_rate": 2.975667234866675e-06,
      "loss": 0.5736,
      "step": 5015
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6272324632681673,
      "learning_rate": 2.9721949006010807e-06,
      "loss": 0.5673,
      "step": 5016
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6203022360015855,
      "learning_rate": 2.968724239803831e-06,
      "loss": 0.5794,
      "step": 5017
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6131457339913292,
      "learning_rate": 2.96525525330136e-06,
      "loss": 0.5634,
      "step": 5018
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6115323629083947,
      "learning_rate": 2.9617879419197037e-06,
      "loss": 0.556,
      "step": 5019
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6041149306922772,
      "learning_rate": 2.9583223064845057e-06,
      "loss": 0.5358,
      "step": 5020
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5532134294960177,
      "learning_rate": 2.9548583478210045e-06,
      "loss": 0.5732,
      "step": 5021
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7993832707135256,
      "learning_rate": 2.9513960667540475e-06,
      "loss": 0.6072,
      "step": 5022
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7216735514450869,
      "learning_rate": 2.947935464108069e-06,
      "loss": 0.5785,
      "step": 5023
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6190733494000892,
      "learning_rate": 2.9444765407071153e-06,
      "loss": 0.559,
      "step": 5024
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6242272700544949,
      "learning_rate": 2.9410192973748298e-06,
      "loss": 0.575,
      "step": 5025
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6268138254686035,
      "learning_rate": 2.937563734934451e-06,
      "loss": 0.5777,
      "step": 5026
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7451106845552987,
      "learning_rate": 2.9341098542088232e-06,
      "loss": 0.5845,
      "step": 5027
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6053313942906361,
      "learning_rate": 2.9306576560203926e-06,
      "loss": 0.5504,
      "step": 5028
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6532781008003382,
      "learning_rate": 2.927207141191192e-06,
      "loss": 0.57,
      "step": 5029
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6157640033859043,
      "learning_rate": 2.923758310542868e-06,
      "loss": 0.5707,
      "step": 5030
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6342877944165707,
      "learning_rate": 2.920311164896655e-06,
      "loss": 0.5677,
      "step": 5031
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6640109018546116,
      "learning_rate": 2.916865705073393e-06,
      "loss": 0.5543,
      "step": 5032
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.571209266537621,
      "learning_rate": 2.913421931893523e-06,
      "loss": 0.5735,
      "step": 5033
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6514784267894156,
      "learning_rate": 2.909979846177071e-06,
      "loss": 0.5787,
      "step": 5034
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6392239432295667,
      "learning_rate": 2.906539448743676e-06,
      "loss": 0.5695,
      "step": 5035
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6422297655145336,
      "learning_rate": 2.903100740412571e-06,
      "loss": 0.5701,
      "step": 5036
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6571701356523566,
      "learning_rate": 2.8996637220025782e-06,
      "loss": 0.5787,
      "step": 5037
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6181785169221887,
      "learning_rate": 2.8962283943321277e-06,
      "loss": 0.5592,
      "step": 5038
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6089322038096938,
      "learning_rate": 2.892794758219243e-06,
      "loss": 0.5532,
      "step": 5039
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7291678913358174,
      "learning_rate": 2.889362814481549e-06,
      "loss": 0.5887,
      "step": 5040
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6440800934845093,
      "learning_rate": 2.885932563936259e-06,
      "loss": 0.5611,
      "step": 5041
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.56394264486635,
      "learning_rate": 2.8825040074001877e-06,
      "loss": 0.5862,
      "step": 5042
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6536557473294352,
      "learning_rate": 2.879077145689746e-06,
      "loss": 0.5364,
      "step": 5043
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6022254475098291,
      "learning_rate": 2.875651979620945e-06,
      "loss": 0.5602,
      "step": 5044
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.631083112123194,
      "learning_rate": 2.8722285100093894e-06,
      "loss": 0.5838,
      "step": 5045
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5994636394303378,
      "learning_rate": 2.8688067376702743e-06,
      "loss": 0.573,
      "step": 5046
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6209218456254798,
      "learning_rate": 2.8653866634184e-06,
      "loss": 0.5349,
      "step": 5047
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5810071781408356,
      "learning_rate": 2.86196828806816e-06,
      "loss": 0.577,
      "step": 5048
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6110401983165621,
      "learning_rate": 2.8585516124335355e-06,
      "loss": 0.5327,
      "step": 5049
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6313468967021685,
      "learning_rate": 2.8551366373281107e-06,
      "loss": 0.578,
      "step": 5050
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5952619751156194,
      "learning_rate": 2.8517233635650687e-06,
      "loss": 0.5734,
      "step": 5051
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6692941970184018,
      "learning_rate": 2.8483117919571748e-06,
      "loss": 0.5759,
      "step": 5052
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6452917536260042,
      "learning_rate": 2.844901923316801e-06,
      "loss": 0.5893,
      "step": 5053
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6318382618633904,
      "learning_rate": 2.8414937584559e-06,
      "loss": 0.5643,
      "step": 5054
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6274529592813481,
      "learning_rate": 2.8380872981860396e-06,
      "loss": 0.5632,
      "step": 5055
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5687554883841626,
      "learning_rate": 2.8346825433183654e-06,
      "loss": 0.5743,
      "step": 5056
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6579761153961656,
      "learning_rate": 2.831279494663616e-06,
      "loss": 0.5794,
      "step": 5057
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6711417288874857,
      "learning_rate": 2.827878153032133e-06,
      "loss": 0.5529,
      "step": 5058
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6769463818106668,
      "learning_rate": 2.8244785192338488e-06,
      "loss": 0.5783,
      "step": 5059
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6410417761573304,
      "learning_rate": 2.821080594078288e-06,
      "loss": 0.5999,
      "step": 5060
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6288147139733656,
      "learning_rate": 2.8176843783745665e-06,
      "loss": 0.5773,
      "step": 5061
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6142755505147466,
      "learning_rate": 2.8142898729313885e-06,
      "loss": 0.5579,
      "step": 5062
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6025265548759413,
      "learning_rate": 2.81089707855707e-06,
      "loss": 0.574,
      "step": 5063
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6200985654437094,
      "learning_rate": 2.8075059960594998e-06,
      "loss": 0.5733,
      "step": 5064
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6533460383924645,
      "learning_rate": 2.804116626246164e-06,
      "loss": 0.5862,
      "step": 5065
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6585238495257004,
      "learning_rate": 2.8007289699241435e-06,
      "loss": 0.5496,
      "step": 5066
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6424958516334657,
      "learning_rate": 2.7973430279001146e-06,
      "loss": 0.5684,
      "step": 5067
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6293126807806013,
      "learning_rate": 2.793958800980341e-06,
      "loss": 0.5625,
      "step": 5068
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6840321375452753,
      "learning_rate": 2.7905762899706734e-06,
      "loss": 0.5983,
      "step": 5069
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6760426484254771,
      "learning_rate": 2.7871954956765625e-06,
      "loss": 0.5694,
      "step": 5070
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6790878830304499,
      "learning_rate": 2.7838164189030493e-06,
      "loss": 0.5755,
      "step": 5071
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6054458533331707,
      "learning_rate": 2.780439060454756e-06,
      "loss": 0.5568,
      "step": 5072
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.641733824444045,
      "learning_rate": 2.777063421135907e-06,
      "loss": 0.5864,
      "step": 5073
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7222344591827594,
      "learning_rate": 2.7736895017503163e-06,
      "loss": 0.5532,
      "step": 5074
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.641613600953448,
      "learning_rate": 2.7703173031013773e-06,
      "loss": 0.5913,
      "step": 5075
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5951450952342761,
      "learning_rate": 2.766946825992091e-06,
      "loss": 0.5602,
      "step": 5076
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6717745325569264,
      "learning_rate": 2.763578071225028e-06,
      "loss": 0.5728,
      "step": 5077
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6396493782985471,
      "learning_rate": 2.7602110396023672e-06,
      "loss": 0.567,
      "step": 5078
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6167793964635964,
      "learning_rate": 2.7568457319258714e-06,
      "loss": 0.5714,
      "step": 5079
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6120264637112417,
      "learning_rate": 2.7534821489968833e-06,
      "loss": 0.5764,
      "step": 5080
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6000877891789935,
      "learning_rate": 2.7501202916163484e-06,
      "loss": 0.5573,
      "step": 5081
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7628420143426332,
      "learning_rate": 2.7467601605847937e-06,
      "loss": 0.5925,
      "step": 5082
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6623309331255449,
      "learning_rate": 2.743401756702341e-06,
      "loss": 0.5918,
      "step": 5083
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6314403857479683,
      "learning_rate": 2.740045080768694e-06,
      "loss": 0.5469,
      "step": 5084
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6349216441809163,
      "learning_rate": 2.736690133583143e-06,
      "loss": 0.5865,
      "step": 5085
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5890146574670724,
      "learning_rate": 2.733336915944581e-06,
      "loss": 0.5557,
      "step": 5086
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6065459366773877,
      "learning_rate": 2.7299854286514727e-06,
      "loss": 0.5634,
      "step": 5087
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6297444630235585,
      "learning_rate": 2.726635672501884e-06,
      "loss": 0.5656,
      "step": 5088
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6110477618466105,
      "learning_rate": 2.7232876482934545e-06,
      "loss": 0.5539,
      "step": 5089
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5690393980108365,
      "learning_rate": 2.7199413568234245e-06,
      "loss": 0.5491,
      "step": 5090
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6409393871002829,
      "learning_rate": 2.7165967988886168e-06,
      "loss": 0.572,
      "step": 5091
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6398063775213312,
      "learning_rate": 2.713253975285437e-06,
      "loss": 0.5651,
      "step": 5092
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6204986184681892,
      "learning_rate": 2.7099128868098847e-06,
      "loss": 0.5733,
      "step": 5093
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6030009989073799,
      "learning_rate": 2.7065735342575463e-06,
      "loss": 0.5639,
      "step": 5094
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6087334982749361,
      "learning_rate": 2.7032359184235845e-06,
      "loss": 0.585,
      "step": 5095
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6678431997097196,
      "learning_rate": 2.6999000401027607e-06,
      "loss": 0.5495,
      "step": 5096
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6275504015936142,
      "learning_rate": 2.6965659000894205e-06,
      "loss": 0.5713,
      "step": 5097
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.657022883037171,
      "learning_rate": 2.693233499177487e-06,
      "loss": 0.5694,
      "step": 5098
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6192337464023825,
      "learning_rate": 2.6899028381604786e-06,
      "loss": 0.5553,
      "step": 5099
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.640613677999764,
      "learning_rate": 2.6865739178314933e-06,
      "loss": 0.5612,
      "step": 5100
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6290866447423838,
      "learning_rate": 2.683246738983217e-06,
      "loss": 0.5538,
      "step": 5101
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6489646311427917,
      "learning_rate": 2.6799213024079263e-06,
      "loss": 0.569,
      "step": 5102
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6748417664565021,
      "learning_rate": 2.67659760889747e-06,
      "loss": 0.5723,
      "step": 5103
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6280545038394882,
      "learning_rate": 2.673275659243294e-06,
      "loss": 0.5908,
      "step": 5104
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6054782214311393,
      "learning_rate": 2.6699554542364237e-06,
      "loss": 0.5603,
      "step": 5105
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6493783945330517,
      "learning_rate": 2.666636994667473e-06,
      "loss": 0.563,
      "step": 5106
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5823429885642489,
      "learning_rate": 2.6633202813266334e-06,
      "loss": 0.5645,
      "step": 5107
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.621822606663938,
      "learning_rate": 2.6600053150036798e-06,
      "loss": 0.5648,
      "step": 5108
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7755776136798271,
      "learning_rate": 2.656692096487985e-06,
      "loss": 0.5879,
      "step": 5109
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6686720772446859,
      "learning_rate": 2.6533806265684892e-06,
      "loss": 0.565,
      "step": 5110
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6232761274276266,
      "learning_rate": 2.650070906033727e-06,
      "loss": 0.5843,
      "step": 5111
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.665206007558697,
      "learning_rate": 2.6467629356718095e-06,
      "loss": 0.5769,
      "step": 5112
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6471819062023617,
      "learning_rate": 2.6434567162704338e-06,
      "loss": 0.554,
      "step": 5113
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6664306315415273,
      "learning_rate": 2.640152248616886e-06,
      "loss": 0.5902,
      "step": 5114
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6128857117579586,
      "learning_rate": 2.6368495334980214e-06,
      "loss": 0.5674,
      "step": 5115
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6291168848752493,
      "learning_rate": 2.633548571700291e-06,
      "loss": 0.5759,
      "step": 5116
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6439926465709623,
      "learning_rate": 2.630249364009725e-06,
      "loss": 0.5928,
      "step": 5117
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5942370178754464,
      "learning_rate": 2.626951911211928e-06,
      "loss": 0.5651,
      "step": 5118
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6165600015428186,
      "learning_rate": 2.6236562140920998e-06,
      "loss": 0.586,
      "step": 5119
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6813200021971207,
      "learning_rate": 2.6203622734350054e-06,
      "loss": 0.5548,
      "step": 5120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6095596424602392,
      "learning_rate": 2.6170700900250146e-06,
      "loss": 0.5522,
      "step": 5121
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6216445574089852,
      "learning_rate": 2.6137796646460576e-06,
      "loss": 0.5933,
      "step": 5122
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6455358767699105,
      "learning_rate": 2.6104909980816527e-06,
      "loss": 0.5939,
      "step": 5123
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5874563426404067,
      "learning_rate": 2.6072040911149033e-06,
      "loss": 0.5652,
      "step": 5124
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6225972432740148,
      "learning_rate": 2.603918944528494e-06,
      "loss": 0.552,
      "step": 5125
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6351095748311809,
      "learning_rate": 2.6006355591046806e-06,
      "loss": 0.5856,
      "step": 5126
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6745954622765635,
      "learning_rate": 2.597353935625311e-06,
      "loss": 0.575,
      "step": 5127
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6010164910260317,
      "learning_rate": 2.5940740748718074e-06,
      "loss": 0.5545,
      "step": 5128
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7220779307170213,
      "learning_rate": 2.5907959776251768e-06,
      "loss": 0.5734,
      "step": 5129
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6116288939654889,
      "learning_rate": 2.587519644666001e-06,
      "loss": 0.5722,
      "step": 5130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6005899030025991,
      "learning_rate": 2.5842450767744378e-06,
      "loss": 0.5622,
      "step": 5131
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6493139461844002,
      "learning_rate": 2.5809722747302414e-06,
      "loss": 0.5702,
      "step": 5132
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6176910404085217,
      "learning_rate": 2.5777012393127265e-06,
      "loss": 0.5778,
      "step": 5133
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5760734869178201,
      "learning_rate": 2.5744319713008025e-06,
      "loss": 0.5758,
      "step": 5134
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6449154479819317,
      "learning_rate": 2.5711644714729443e-06,
      "loss": 0.5688,
      "step": 5135
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6507587654658197,
      "learning_rate": 2.567898740607215e-06,
      "loss": 0.591,
      "step": 5136
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6935222246493755,
      "learning_rate": 2.5646347794812566e-06,
      "loss": 0.5486,
      "step": 5137
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6050256826729787,
      "learning_rate": 2.561372588872283e-06,
      "loss": 0.546,
      "step": 5138
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.613619827419747,
      "learning_rate": 2.558112169557091e-06,
      "loss": 0.5941,
      "step": 5139
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6771363336997124,
      "learning_rate": 2.5548535223120584e-06,
      "loss": 0.5714,
      "step": 5140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6192257735865359,
      "learning_rate": 2.5515966479131325e-06,
      "loss": 0.5637,
      "step": 5141
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6068966920418793,
      "learning_rate": 2.548341547135851e-06,
      "loss": 0.5673,
      "step": 5142
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6605788094478464,
      "learning_rate": 2.5450882207553095e-06,
      "loss": 0.5458,
      "step": 5143
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6143242102134049,
      "learning_rate": 2.5418366695462095e-06,
      "loss": 0.5544,
      "step": 5144
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6390039572823275,
      "learning_rate": 2.5385868942828042e-06,
      "loss": 0.5529,
      "step": 5145
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6184419134202961,
      "learning_rate": 2.5353388957389323e-06,
      "loss": 0.5575,
      "step": 5146
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6973819686328276,
      "learning_rate": 2.5320926746880137e-06,
      "loss": 0.597,
      "step": 5147
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6579950310978705,
      "learning_rate": 2.5288482319030417e-06,
      "loss": 0.5718,
      "step": 5148
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6166738119546463,
      "learning_rate": 2.5256055681565883e-06,
      "loss": 0.5676,
      "step": 5149
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6581691794157271,
      "learning_rate": 2.522364684220796e-06,
      "loss": 0.5595,
      "step": 5150
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.645765434913503,
      "learning_rate": 2.51912558086739e-06,
      "loss": 0.556,
      "step": 5151
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.610925856610954,
      "learning_rate": 2.5158882588676704e-06,
      "loss": 0.5615,
      "step": 5152
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6659135005884267,
      "learning_rate": 2.512652718992508e-06,
      "loss": 0.5911,
      "step": 5153
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6878116876149998,
      "learning_rate": 2.5094189620123565e-06,
      "loss": 0.5676,
      "step": 5154
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6120914823880343,
      "learning_rate": 2.5061869886972378e-06,
      "loss": 0.5656,
      "step": 5155
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6637555413649496,
      "learning_rate": 2.5029567998167546e-06,
      "loss": 0.5844,
      "step": 5156
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6165960422465305,
      "learning_rate": 2.4997283961400866e-06,
      "loss": 0.5609,
      "step": 5157
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5875147548545354,
      "learning_rate": 2.496501778435977e-06,
      "loss": 0.5623,
      "step": 5158
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.702413274495182,
      "learning_rate": 2.493276947472756e-06,
      "loss": 0.5593,
      "step": 5159
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5423677994492987,
      "learning_rate": 2.490053904018327e-06,
      "loss": 0.5395,
      "step": 5160
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7203662382265472,
      "learning_rate": 2.4868326488401575e-06,
      "loss": 0.5852,
      "step": 5161
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6637951145600877,
      "learning_rate": 2.483613182705299e-06,
      "loss": 0.5515,
      "step": 5162
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5885796149310557,
      "learning_rate": 2.4803955063803775e-06,
      "loss": 0.576,
      "step": 5163
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5883746717298303,
      "learning_rate": 2.4771796206315846e-06,
      "loss": 0.5769,
      "step": 5164
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6583921172503684,
      "learning_rate": 2.4739655262246954e-06,
      "loss": 0.5798,
      "step": 5165
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.654326957261001,
      "learning_rate": 2.4707532239250442e-06,
      "loss": 0.5785,
      "step": 5166
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5956685608517748,
      "learning_rate": 2.4675427144975593e-06,
      "loss": 0.5692,
      "step": 5167
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5777425914536971,
      "learning_rate": 2.464333998706726e-06,
      "loss": 0.5827,
      "step": 5168
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6034285718679375,
      "learning_rate": 2.461127077316603e-06,
      "loss": 0.5601,
      "step": 5169
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6198081206079534,
      "learning_rate": 2.4579219510908294e-06,
      "loss": 0.5693,
      "step": 5170
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6637678080948349,
      "learning_rate": 2.4547186207926134e-06,
      "loss": 0.5701,
      "step": 5171
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6209083930563348,
      "learning_rate": 2.4515170871847383e-06,
      "loss": 0.5764,
      "step": 5172
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5751141778066206,
      "learning_rate": 2.44831735102955e-06,
      "loss": 0.5818,
      "step": 5173
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6833512743783504,
      "learning_rate": 2.4451194130889754e-06,
      "loss": 0.5789,
      "step": 5174
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6254015732277669,
      "learning_rate": 2.4419232741245158e-06,
      "loss": 0.6109,
      "step": 5175
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6425771127919786,
      "learning_rate": 2.438728934897232e-06,
      "loss": 0.5849,
      "step": 5176
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6778192123772527,
      "learning_rate": 2.4355363961677702e-06,
      "loss": 0.5867,
      "step": 5177
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.650093598633319,
      "learning_rate": 2.4323456586963333e-06,
      "loss": 0.5731,
      "step": 5178
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6543129822891146,
      "learning_rate": 2.4291567232427095e-06,
      "loss": 0.5791,
      "step": 5179
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6943743099451367,
      "learning_rate": 2.4259695905662506e-06,
      "loss": 0.5488,
      "step": 5180
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6592778949544047,
      "learning_rate": 2.4227842614258758e-06,
      "loss": 0.5676,
      "step": 5181
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6610143103945308,
      "learning_rate": 2.419600736580081e-06,
      "loss": 0.5826,
      "step": 5182
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6318969743471221,
      "learning_rate": 2.416419016786936e-06,
      "loss": 0.5597,
      "step": 5183
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6024777324629612,
      "learning_rate": 2.413239102804067e-06,
      "loss": 0.5547,
      "step": 5184
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6419542584222193,
      "learning_rate": 2.410060995388681e-06,
      "loss": 0.5692,
      "step": 5185
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6751409203022639,
      "learning_rate": 2.406884695297558e-06,
      "loss": 0.5808,
      "step": 5186
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6424895031274612,
      "learning_rate": 2.403710203287033e-06,
      "loss": 0.5609,
      "step": 5187
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6773644690058108,
      "learning_rate": 2.4005375201130275e-06,
      "loss": 0.5919,
      "step": 5188
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7147042846332795,
      "learning_rate": 2.3973666465310143e-06,
      "loss": 0.5537,
      "step": 5189
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6067862709559934,
      "learning_rate": 2.3941975832960563e-06,
      "loss": 0.5692,
      "step": 5190
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6982273782915284,
      "learning_rate": 2.39103033116277e-06,
      "loss": 0.5989,
      "step": 5191
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6224166234747347,
      "learning_rate": 2.387864890885341e-06,
      "loss": 0.572,
      "step": 5192
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6617252315583243,
      "learning_rate": 2.3847012632175293e-06,
      "loss": 0.5874,
      "step": 5193
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.570316000253902,
      "learning_rate": 2.381539448912662e-06,
      "loss": 0.5586,
      "step": 5194
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5969477584289196,
      "learning_rate": 2.3783794487236367e-06,
      "loss": 0.5558,
      "step": 5195
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6401285142514777,
      "learning_rate": 2.3752212634029127e-06,
      "loss": 0.5597,
      "step": 5196
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6221377286941322,
      "learning_rate": 2.3720648937025157e-06,
      "loss": 0.5608,
      "step": 5197
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6155292794615738,
      "learning_rate": 2.368910340374054e-06,
      "loss": 0.5626,
      "step": 5198
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.662074913809814,
      "learning_rate": 2.365757604168686e-06,
      "loss": 0.5362,
      "step": 5199
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.653934733808823,
      "learning_rate": 2.3626066858371477e-06,
      "loss": 0.5659,
      "step": 5200
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6046128803049122,
      "learning_rate": 2.3594575861297355e-06,
      "loss": 0.559,
      "step": 5201
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5723292156818445,
      "learning_rate": 2.3563103057963188e-06,
      "loss": 0.6077,
      "step": 5202
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5970230569954039,
      "learning_rate": 2.3531648455863345e-06,
      "loss": 0.5559,
      "step": 5203
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.721113606933481,
      "learning_rate": 2.350021206248777e-06,
      "loss": 0.5895,
      "step": 5204
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6225039282373156,
      "learning_rate": 2.346879388532215e-06,
      "loss": 0.559,
      "step": 5205
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6186025933100087,
      "learning_rate": 2.3437393931847842e-06,
      "loss": 0.5466,
      "step": 5206
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5891164912400145,
      "learning_rate": 2.3406012209541795e-06,
      "loss": 0.5552,
      "step": 5207
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6029757072672648,
      "learning_rate": 2.337464872587667e-06,
      "loss": 0.5907,
      "step": 5208
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6396776818088699,
      "learning_rate": 2.334330348832079e-06,
      "loss": 0.5602,
      "step": 5209
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5948094497415457,
      "learning_rate": 2.331197650433813e-06,
      "loss": 0.5593,
      "step": 5210
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.627294615772193,
      "learning_rate": 2.3280667781388276e-06,
      "loss": 0.5738,
      "step": 5211
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6093629509210505,
      "learning_rate": 2.324937732692647e-06,
      "loss": 0.5759,
      "step": 5212
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6418286377822544,
      "learning_rate": 2.3218105148403657e-06,
      "loss": 0.5831,
      "step": 5213
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.706515655224374,
      "learning_rate": 2.3186851253266397e-06,
      "loss": 0.5961,
      "step": 5214
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6370951341467634,
      "learning_rate": 2.3155615648956944e-06,
      "loss": 0.5604,
      "step": 5215
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6931151456494519,
      "learning_rate": 2.312439834291307e-06,
      "loss": 0.5563,
      "step": 5216
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5998119471740535,
      "learning_rate": 2.3093199342568316e-06,
      "loss": 0.5552,
      "step": 5217
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6567467277270337,
      "learning_rate": 2.306201865535186e-06,
      "loss": 0.5568,
      "step": 5218
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6156172826351259,
      "learning_rate": 2.303085628868843e-06,
      "loss": 0.5494,
      "step": 5219
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6572415137742513,
      "learning_rate": 2.2999712249998396e-06,
      "loss": 0.5687,
      "step": 5220
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5989137499669018,
      "learning_rate": 2.2968586546697914e-06,
      "loss": 0.5588,
      "step": 5221
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5948465364817642,
      "learning_rate": 2.293747918619861e-06,
      "loss": 0.5456,
      "step": 5222
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6601198689811283,
      "learning_rate": 2.2906390175907823e-06,
      "loss": 0.5631,
      "step": 5223
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5919161936253085,
      "learning_rate": 2.2875319523228466e-06,
      "loss": 0.5671,
      "step": 5224
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6516639305826245,
      "learning_rate": 2.284426723555914e-06,
      "loss": 0.5722,
      "step": 5225
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7059651802344822,
      "learning_rate": 2.281323332029407e-06,
      "loss": 0.6005,
      "step": 5226
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6725361012474752,
      "learning_rate": 2.2782217784823036e-06,
      "loss": 0.5715,
      "step": 5227
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6399769391893672,
      "learning_rate": 2.2751220636531523e-06,
      "loss": 0.5842,
      "step": 5228
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6101147876309417,
      "learning_rate": 2.272024188280062e-06,
      "loss": 0.5849,
      "step": 5229
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6022710805836101,
      "learning_rate": 2.268928153100697e-06,
      "loss": 0.5617,
      "step": 5230
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5857014530289211,
      "learning_rate": 2.2658339588522906e-06,
      "loss": 0.5754,
      "step": 5231
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5918605089273534,
      "learning_rate": 2.2627416062716366e-06,
      "loss": 0.5682,
      "step": 5232
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6978837536851127,
      "learning_rate": 2.259651096095091e-06,
      "loss": 0.5962,
      "step": 5233
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6715440436272758,
      "learning_rate": 2.2565624290585674e-06,
      "loss": 0.5846,
      "step": 5234
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6342965369740857,
      "learning_rate": 2.2534756058975403e-06,
      "loss": 0.5736,
      "step": 5235
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.636783677196929,
      "learning_rate": 2.250390627347049e-06,
      "loss": 0.5915,
      "step": 5236
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6759432504237523,
      "learning_rate": 2.2473074941416916e-06,
      "loss": 0.5521,
      "step": 5237
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6598067814342886,
      "learning_rate": 2.2442262070156294e-06,
      "loss": 0.5822,
      "step": 5238
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6315402699297894,
      "learning_rate": 2.2411467667025787e-06,
      "loss": 0.5603,
      "step": 5239
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6654194747138561,
      "learning_rate": 2.2380691739358186e-06,
      "loss": 0.5369,
      "step": 5240
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6619593141653894,
      "learning_rate": 2.2349934294481943e-06,
      "loss": 0.5745,
      "step": 5241
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8419441723578382,
      "learning_rate": 2.231919533972098e-06,
      "loss": 0.5666,
      "step": 5242
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.584291123901812,
      "learning_rate": 2.228847488239492e-06,
      "loss": 0.5478,
      "step": 5243
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6836226665562435,
      "learning_rate": 2.2257772929818977e-06,
      "loss": 0.5807,
      "step": 5244
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.645442131972635,
      "learning_rate": 2.222708948930389e-06,
      "loss": 0.5705,
      "step": 5245
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6373844931345733,
      "learning_rate": 2.2196424568156073e-06,
      "loss": 0.5651,
      "step": 5246
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5411196632679949,
      "learning_rate": 2.216577817367741e-06,
      "loss": 0.5451,
      "step": 5247
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6080547297621614,
      "learning_rate": 2.2135150313165566e-06,
      "loss": 0.5464,
      "step": 5248
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6493425840646643,
      "learning_rate": 2.2104540993913613e-06,
      "loss": 0.5905,
      "step": 5249
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6260426210961831,
      "learning_rate": 2.2073950223210274e-06,
      "loss": 0.5677,
      "step": 5250
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6119876616339411,
      "learning_rate": 2.2043378008339845e-06,
      "loss": 0.5543,
      "step": 5251
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6400419944285015,
      "learning_rate": 2.2012824356582275e-06,
      "loss": 0.5862,
      "step": 5252
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5670680036272657,
      "learning_rate": 2.1982289275212954e-06,
      "loss": 0.5717,
      "step": 5253
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6359374046344111,
      "learning_rate": 2.1951772771503e-06,
      "loss": 0.554,
      "step": 5254
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5982522358252532,
      "learning_rate": 2.1921274852718944e-06,
      "loss": 0.5666,
      "step": 5255
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.630465796074457,
      "learning_rate": 2.1890795526123086e-06,
      "loss": 0.569,
      "step": 5256
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5943557194178636,
      "learning_rate": 2.186033479897316e-06,
      "loss": 0.5378,
      "step": 5257
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6751143616510604,
      "learning_rate": 2.182989267852246e-06,
      "loss": 0.589,
      "step": 5258
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5821642088262767,
      "learning_rate": 2.1799469172019914e-06,
      "loss": 0.5464,
      "step": 5259
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.664321193730273,
      "learning_rate": 2.176906428671003e-06,
      "loss": 0.5863,
      "step": 5260
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.597918690496956,
      "learning_rate": 2.173867802983286e-06,
      "loss": 0.5771,
      "step": 5261
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.625596620453047,
      "learning_rate": 2.170831040862397e-06,
      "loss": 0.5744,
      "step": 5262
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5983758466650558,
      "learning_rate": 2.167796143031453e-06,
      "loss": 0.5653,
      "step": 5263
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6355140250668305,
      "learning_rate": 2.1647631102131328e-06,
      "loss": 0.5567,
      "step": 5264
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6979589770419351,
      "learning_rate": 2.161731943129658e-06,
      "loss": 0.5859,
      "step": 5265
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5708697071976139,
      "learning_rate": 2.1587026425028168e-06,
      "loss": 0.5675,
      "step": 5266
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6246925774552706,
      "learning_rate": 2.1556752090539523e-06,
      "loss": 0.5843,
      "step": 5267
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6931676239738338,
      "learning_rate": 2.1526496435039547e-06,
      "loss": 0.596,
      "step": 5268
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6811012579275987,
      "learning_rate": 2.1496259465732783e-06,
      "loss": 0.5795,
      "step": 5269
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7036063018378688,
      "learning_rate": 2.1466041189819266e-06,
      "loss": 0.5872,
      "step": 5270
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5373808276418679,
      "learning_rate": 2.1435841614494625e-06,
      "loss": 0.5826,
      "step": 5271
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5850177575854363,
      "learning_rate": 2.140566074695002e-06,
      "loss": 0.5744,
      "step": 5272
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7015222837978123,
      "learning_rate": 2.1375498594372113e-06,
      "loss": 0.5938,
      "step": 5273
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6055593391431378,
      "learning_rate": 2.1345355163943173e-06,
      "loss": 0.5554,
      "step": 5274
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.584346145952104,
      "learning_rate": 2.1315230462840985e-06,
      "loss": 0.5732,
      "step": 5275
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5964064072699917,
      "learning_rate": 2.1285124498238905e-06,
      "loss": 0.577,
      "step": 5276
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6173337632708447,
      "learning_rate": 2.125503727730577e-06,
      "loss": 0.5665,
      "step": 5277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.666823522624521,
      "learning_rate": 2.1224968807205914e-06,
      "loss": 0.5948,
      "step": 5278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6128130708711318,
      "learning_rate": 2.1194919095099396e-06,
      "loss": 0.5746,
      "step": 5279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6294947708556557,
      "learning_rate": 2.116488814814163e-06,
      "loss": 0.5824,
      "step": 5280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6996680820474886,
      "learning_rate": 2.113487597348357e-06,
      "loss": 0.595,
      "step": 5281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5820875004408823,
      "learning_rate": 2.110488257827179e-06,
      "loss": 0.5647,
      "step": 5282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6559368288627706,
      "learning_rate": 2.107490796964835e-06,
      "loss": 0.5331,
      "step": 5283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6193035786279598,
      "learning_rate": 2.1044952154750864e-06,
      "loss": 0.557,
      "step": 5284
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6187476195243474,
      "learning_rate": 2.101501514071238e-06,
      "loss": 0.568,
      "step": 5285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6071227234461656,
      "learning_rate": 2.0985096934661563e-06,
      "loss": 0.5648,
      "step": 5286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6241341859879185,
      "learning_rate": 2.0955197543722595e-06,
      "loss": 0.5741,
      "step": 5287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6504137756770589,
      "learning_rate": 2.0925316975015087e-06,
      "loss": 0.5794,
      "step": 5288
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7018622929530205,
      "learning_rate": 2.0895455235654306e-06,
      "loss": 0.5952,
      "step": 5289
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6826395027780683,
      "learning_rate": 2.0865612332750883e-06,
      "loss": 0.5749,
      "step": 5290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6476253457399653,
      "learning_rate": 2.0835788273411084e-06,
      "loss": 0.5835,
      "step": 5291
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.623445275560438,
      "learning_rate": 2.0805983064736668e-06,
      "loss": 0.5778,
      "step": 5292
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6568656635215769,
      "learning_rate": 2.0776196713824825e-06,
      "loss": 0.5731,
      "step": 5293
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6088628249764487,
      "learning_rate": 2.074642922776834e-06,
      "loss": 0.5918,
      "step": 5294
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6148355991135404,
      "learning_rate": 2.0716680613655515e-06,
      "loss": 0.5643,
      "step": 5295
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6677267997720419,
      "learning_rate": 2.0686950878570058e-06,
      "loss": 0.5521,
      "step": 5296
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6208089174188709,
      "learning_rate": 2.0657240029591276e-06,
      "loss": 0.5759,
      "step": 5297
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6263790223767632,
      "learning_rate": 2.0627548073793933e-06,
      "loss": 0.555,
      "step": 5298
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.675055469335233,
      "learning_rate": 2.059787501824836e-06,
      "loss": 0.5534,
      "step": 5299
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.653362976260901,
      "learning_rate": 2.0568220870020296e-06,
      "loss": 0.5481,
      "step": 5300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6884852668131206,
      "learning_rate": 2.053858563617096e-06,
      "loss": 0.5649,
      "step": 5301
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5999993600256596,
      "learning_rate": 2.0508969323757243e-06,
      "loss": 0.558,
      "step": 5302
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5904834656247445,
      "learning_rate": 2.0479371939831325e-06,
      "loss": 0.5681,
      "step": 5303
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.600521769912323,
      "learning_rate": 2.0449793491441026e-06,
      "loss": 0.5695,
      "step": 5304
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6232935853324721,
      "learning_rate": 2.0420233985629534e-06,
      "loss": 0.5773,
      "step": 5305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6389540001623611,
      "learning_rate": 2.0390693429435626e-06,
      "loss": 0.5581,
      "step": 5306
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.61658089595662,
      "learning_rate": 2.0361171829893545e-06,
      "loss": 0.5607,
      "step": 5307
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6036980585927457,
      "learning_rate": 2.0331669194032968e-06,
      "loss": 0.5492,
      "step": 5308
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.632639900059736,
      "learning_rate": 2.0302185528879104e-06,
      "loss": 0.5596,
      "step": 5309
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.644568353948571,
      "learning_rate": 2.0272720841452674e-06,
      "loss": 0.5548,
      "step": 5310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6310238750057914,
      "learning_rate": 2.024327513876979e-06,
      "loss": 0.562,
      "step": 5311
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6162502697139798,
      "learning_rate": 2.0213848427842133e-06,
      "loss": 0.5685,
      "step": 5312
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5939293101157334,
      "learning_rate": 2.0184440715676767e-06,
      "loss": 0.559,
      "step": 5313
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6450459518715435,
      "learning_rate": 2.015505200927633e-06,
      "loss": 0.6123,
      "step": 5314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7389657624502607,
      "learning_rate": 2.0125682315638915e-06,
      "loss": 0.6313,
      "step": 5315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6281767856587255,
      "learning_rate": 2.0096331641758006e-06,
      "loss": 0.5752,
      "step": 5316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6643111774690097,
      "learning_rate": 2.006699999462264e-06,
      "loss": 0.5582,
      "step": 5317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.588788665389874,
      "learning_rate": 2.003768738121732e-06,
      "loss": 0.5735,
      "step": 5318
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6287819856093568,
      "learning_rate": 2.0008393808521966e-06,
      "loss": 0.5782,
      "step": 5319
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6007183525605082,
      "learning_rate": 1.9979119283512006e-06,
      "loss": 0.552,
      "step": 5320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.618763475628038,
      "learning_rate": 1.994986381315832e-06,
      "loss": 0.5796,
      "step": 5321
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.682162168423708,
      "learning_rate": 1.9920627404427275e-06,
      "loss": 0.576,
      "step": 5322
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5934383263495643,
      "learning_rate": 1.989141006428066e-06,
      "loss": 0.5683,
      "step": 5323
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6234939206269339,
      "learning_rate": 1.986221179967568e-06,
      "loss": 0.5563,
      "step": 5324
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6593729929034188,
      "learning_rate": 1.9833032617565173e-06,
      "loss": 0.5759,
      "step": 5325
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6541635321727437,
      "learning_rate": 1.9803872524897215e-06,
      "loss": 0.5386,
      "step": 5326
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6532239950622069,
      "learning_rate": 1.97747315286155e-06,
      "loss": 0.6008,
      "step": 5327
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6349944999502638,
      "learning_rate": 1.974560963565907e-06,
      "loss": 0.5659,
      "step": 5328
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6347381385215431,
      "learning_rate": 1.971650685296247e-06,
      "loss": 0.5601,
      "step": 5329
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6498321447171813,
      "learning_rate": 1.9687423187455735e-06,
      "loss": 0.553,
      "step": 5330
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5687972561257534,
      "learning_rate": 1.965835864606422e-06,
      "loss": 0.5569,
      "step": 5331
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6273062691255347,
      "learning_rate": 1.9629313235708846e-06,
      "loss": 0.5728,
      "step": 5332
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6020761031189071,
      "learning_rate": 1.960028696330596e-06,
      "loss": 0.5536,
      "step": 5333
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.676381342840967,
      "learning_rate": 1.9571279835767276e-06,
      "loss": 0.5756,
      "step": 5334
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6046934516764654,
      "learning_rate": 1.954229186000005e-06,
      "loss": 0.5736,
      "step": 5335
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6254746915009493,
      "learning_rate": 1.951332304290685e-06,
      "loss": 0.5573,
      "step": 5336
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.667696276422499,
      "learning_rate": 1.948437339138588e-06,
      "loss": 0.5833,
      "step": 5337
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6717591804909068,
      "learning_rate": 1.945544291233059e-06,
      "loss": 0.5637,
      "step": 5338
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5592395851358318,
      "learning_rate": 1.9426531612629917e-06,
      "loss": 0.5463,
      "step": 5339
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6261985281431238,
      "learning_rate": 1.9397639499168285e-06,
      "loss": 0.5834,
      "step": 5340
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5704080857181653,
      "learning_rate": 1.9368766578825503e-06,
      "loss": 0.5469,
      "step": 5341
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5906656738833314,
      "learning_rate": 1.933991285847686e-06,
      "loss": 0.5689,
      "step": 5342
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6121864317354061,
      "learning_rate": 1.931107834499296e-06,
      "loss": 0.5505,
      "step": 5343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6434072155492723,
      "learning_rate": 1.928226304523996e-06,
      "loss": 0.581,
      "step": 5344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6159663936795867,
      "learning_rate": 1.9253466966079393e-06,
      "loss": 0.5984,
      "step": 5345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6059354657347172,
      "learning_rate": 1.9224690114368205e-06,
      "loss": 0.5733,
      "step": 5346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6211038680613817,
      "learning_rate": 1.9195932496958738e-06,
      "loss": 0.5403,
      "step": 5347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6153184174844218,
      "learning_rate": 1.9167194120698797e-06,
      "loss": 0.5682,
      "step": 5348
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6681092455183884,
      "learning_rate": 1.913847499243161e-06,
      "loss": 0.5574,
      "step": 5349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6884594021410612,
      "learning_rate": 1.9109775118995823e-06,
      "loss": 0.5643,
      "step": 5350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6256195832186512,
      "learning_rate": 1.908109450722544e-06,
      "loss": 0.5783,
      "step": 5351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6078455043795232,
      "learning_rate": 1.9052433163949935e-06,
      "loss": 0.5506,
      "step": 5352
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.599251629128412,
      "learning_rate": 1.9023791095994214e-06,
      "loss": 0.5363,
      "step": 5353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6605216041573295,
      "learning_rate": 1.899516831017849e-06,
      "loss": 0.5736,
      "step": 5354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6956736265605443,
      "learning_rate": 1.8966564813318489e-06,
      "loss": 0.5838,
      "step": 5355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6180178888112428,
      "learning_rate": 1.8937980612225315e-06,
      "loss": 0.5616,
      "step": 5356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6465117119942058,
      "learning_rate": 1.8909415713705438e-06,
      "loss": 0.5593,
      "step": 5357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6312173535442566,
      "learning_rate": 1.888087012456079e-06,
      "loss": 0.5917,
      "step": 5358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6757941000898688,
      "learning_rate": 1.8852343851588617e-06,
      "loss": 0.5508,
      "step": 5359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7035636736156249,
      "learning_rate": 1.8823836901581727e-06,
      "loss": 0.5671,
      "step": 5360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6414203769196276,
      "learning_rate": 1.8795349281328167e-06,
      "loss": 0.5677,
      "step": 5361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7147442640653794,
      "learning_rate": 1.8766880997611424e-06,
      "loss": 0.5886,
      "step": 5362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.597857238869112,
      "learning_rate": 1.8738432057210398e-06,
      "loss": 0.5546,
      "step": 5363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6313797021946249,
      "learning_rate": 1.8710002466899414e-06,
      "loss": 0.5802,
      "step": 5364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6948689097294001,
      "learning_rate": 1.8681592233448142e-06,
      "loss": 0.5794,
      "step": 5365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6136265603170543,
      "learning_rate": 1.8653201363621643e-06,
      "loss": 0.5381,
      "step": 5366
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6512066153610366,
      "learning_rate": 1.8624829864180384e-06,
      "loss": 0.5693,
      "step": 5367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6065570690560417,
      "learning_rate": 1.8596477741880248e-06,
      "loss": 0.5901,
      "step": 5368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6935642856894103,
      "learning_rate": 1.8568145003472427e-06,
      "loss": 0.5918,
      "step": 5369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5953712737117854,
      "learning_rate": 1.8539831655703577e-06,
      "loss": 0.5623,
      "step": 5370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6623530241083843,
      "learning_rate": 1.8511537705315674e-06,
      "loss": 0.5819,
      "step": 5371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5767593141028314,
      "learning_rate": 1.8483263159046104e-06,
      "loss": 0.5458,
      "step": 5372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6412457873409061,
      "learning_rate": 1.8455008023627674e-06,
      "loss": 0.579,
      "step": 5373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6566559350975023,
      "learning_rate": 1.8426772305788476e-06,
      "loss": 0.5542,
      "step": 5374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6380650705713609,
      "learning_rate": 1.8398556012252044e-06,
      "loss": 0.5766,
      "step": 5375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.584526397385808,
      "learning_rate": 1.8370359149737304e-06,
      "loss": 0.5431,
      "step": 5376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6096289023525182,
      "learning_rate": 1.8342181724958474e-06,
      "loss": 0.5696,
      "step": 5377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6527855799559551,
      "learning_rate": 1.831402374462521e-06,
      "loss": 0.577,
      "step": 5378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.611166580073851,
      "learning_rate": 1.8285885215442556e-06,
      "loss": 0.5895,
      "step": 5379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.577970588884454,
      "learning_rate": 1.8257766144110823e-06,
      "loss": 0.591,
      "step": 5380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5660871824990316,
      "learning_rate": 1.8229666537325819e-06,
      "loss": 0.5682,
      "step": 5381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.648951668574401,
      "learning_rate": 1.820158640177856e-06,
      "loss": 0.5503,
      "step": 5382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6357604655271597,
      "learning_rate": 1.817352574415563e-06,
      "loss": 0.5523,
      "step": 5383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6569802454279678,
      "learning_rate": 1.8145484571138805e-06,
      "loss": 0.6004,
      "step": 5384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6285148668924647,
      "learning_rate": 1.811746288940527e-06,
      "loss": 0.5579,
      "step": 5385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6186210105343578,
      "learning_rate": 1.808946070562758e-06,
      "loss": 0.5683,
      "step": 5386
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5848993714471069,
      "learning_rate": 1.8061478026473656e-06,
      "loss": 0.5677,
      "step": 5387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5839297226406498,
      "learning_rate": 1.8033514858606783e-06,
      "loss": 0.5692,
      "step": 5388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7004095673336423,
      "learning_rate": 1.8005571208685567e-06,
      "loss": 0.581,
      "step": 5389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5880902371649783,
      "learning_rate": 1.7977647083363914e-06,
      "loss": 0.5578,
      "step": 5390
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6404403183767634,
      "learning_rate": 1.7949742489291256e-06,
      "loss": 0.547,
      "step": 5391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.634572466492501,
      "learning_rate": 1.7921857433112188e-06,
      "loss": 0.5678,
      "step": 5392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6384205546113194,
      "learning_rate": 1.789399192146678e-06,
      "loss": 0.5835,
      "step": 5393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.616820844907087,
      "learning_rate": 1.7866145960990334e-06,
      "loss": 0.561,
      "step": 5394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7191183836847598,
      "learning_rate": 1.7838319558313598e-06,
      "loss": 0.5633,
      "step": 5395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6504179590064708,
      "learning_rate": 1.7810512720062655e-06,
      "loss": 0.558,
      "step": 5396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6023547919512776,
      "learning_rate": 1.778272545285883e-06,
      "loss": 0.5401,
      "step": 5397
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6823108765424187,
      "learning_rate": 1.7754957763318892e-06,
      "loss": 0.5846,
      "step": 5398
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.622614337040353,
      "learning_rate": 1.7727209658054933e-06,
      "loss": 0.5718,
      "step": 5399
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5701004780801088,
      "learning_rate": 1.7699481143674323e-06,
      "loss": 0.5432,
      "step": 5400
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6221460804237794,
      "learning_rate": 1.7671772226779814e-06,
      "loss": 0.5747,
      "step": 5401
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6925637746204081,
      "learning_rate": 1.7644082913969496e-06,
      "loss": 0.5803,
      "step": 5402
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6363872499849856,
      "learning_rate": 1.7616413211836792e-06,
      "loss": 0.567,
      "step": 5403
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.740084902556674,
      "learning_rate": 1.758876312697041e-06,
      "loss": 0.5884,
      "step": 5404
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6192032666937723,
      "learning_rate": 1.7561132665954418e-06,
      "loss": 0.544,
      "step": 5405
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6091371401311104,
      "learning_rate": 1.7533521835368206e-06,
      "loss": 0.5626,
      "step": 5406
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5932304710095446,
      "learning_rate": 1.750593064178654e-06,
      "loss": 0.5786,
      "step": 5407
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5731618273344997,
      "learning_rate": 1.7478359091779395e-06,
      "loss": 0.5904,
      "step": 5408
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6267766540304772,
      "learning_rate": 1.7450807191912188e-06,
      "loss": 0.5344,
      "step": 5409
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.588780845003133,
      "learning_rate": 1.7423274948745584e-06,
      "loss": 0.5674,
      "step": 5410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5873873030098137,
      "learning_rate": 1.7395762368835623e-06,
      "loss": 0.5506,
      "step": 5411
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5960171579453178,
      "learning_rate": 1.7368269458733611e-06,
      "loss": 0.544,
      "step": 5412
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7543689583047133,
      "learning_rate": 1.7340796224986123e-06,
      "loss": 0.5806,
      "step": 5413
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6488866429953218,
      "learning_rate": 1.7313342674135225e-06,
      "loss": 0.5651,
      "step": 5414
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6858867966910478,
      "learning_rate": 1.728590881271811e-06,
      "loss": 0.5746,
      "step": 5415
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6526612765705662,
      "learning_rate": 1.725849464726741e-06,
      "loss": 0.5641,
      "step": 5416
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6426399948799865,
      "learning_rate": 1.7231100184310955e-06,
      "loss": 0.5688,
      "step": 5417
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6643258150032847,
      "learning_rate": 1.7203725430371976e-06,
      "loss": 0.5763,
      "step": 5418
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6536542767008094,
      "learning_rate": 1.7176370391968999e-06,
      "loss": 0.582,
      "step": 5419
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.602150364230825,
      "learning_rate": 1.7149035075615795e-06,
      "loss": 0.5585,
      "step": 5420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6221902828959798,
      "learning_rate": 1.7121719487821498e-06,
      "loss": 0.5794,
      "step": 5421
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6242551252335333,
      "learning_rate": 1.709442363509054e-06,
      "loss": 0.549,
      "step": 5422
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6026619305953936,
      "learning_rate": 1.7067147523922589e-06,
      "loss": 0.5783,
      "step": 5423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6193590922844516,
      "learning_rate": 1.7039891160812705e-06,
      "loss": 0.5776,
      "step": 5424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6579557451536201,
      "learning_rate": 1.7012654552251183e-06,
      "loss": 0.5767,
      "step": 5425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6287507836654863,
      "learning_rate": 1.6985437704723673e-06,
      "loss": 0.5903,
      "step": 5426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6012407563729977,
      "learning_rate": 1.6958240624711032e-06,
      "loss": 0.5675,
      "step": 5427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6441022344633849,
      "learning_rate": 1.6931063318689456e-06,
      "loss": 0.5844,
      "step": 5428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6230139811883408,
      "learning_rate": 1.690390579313045e-06,
      "loss": 0.5717,
      "step": 5429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.592818091771867,
      "learning_rate": 1.6876768054500781e-06,
      "loss": 0.582,
      "step": 5430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6202457122800873,
      "learning_rate": 1.684965010926256e-06,
      "loss": 0.5527,
      "step": 5431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6137661097967072,
      "learning_rate": 1.682255196387308e-06,
      "loss": 0.5763,
      "step": 5432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5844439839389689,
      "learning_rate": 1.6795473624784998e-06,
      "loss": 0.5503,
      "step": 5433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6383496150457573,
      "learning_rate": 1.6768415098446277e-06,
      "loss": 0.5552,
      "step": 5434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6404108406662443,
      "learning_rate": 1.674137639130009e-06,
      "loss": 0.5529,
      "step": 5435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6372605422384705,
      "learning_rate": 1.6714357509784873e-06,
      "loss": 0.564,
      "step": 5436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6984281194659531,
      "learning_rate": 1.6687358460334491e-06,
      "loss": 0.5621,
      "step": 5437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6447337894883783,
      "learning_rate": 1.6660379249377912e-06,
      "loss": 0.5592,
      "step": 5438
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6367375466945714,
      "learning_rate": 1.6633419883339496e-06,
      "loss": 0.5746,
      "step": 5439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.660106734547602,
      "learning_rate": 1.66064803686388e-06,
      "loss": 0.5822,
      "step": 5440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6095171630896534,
      "learning_rate": 1.6579560711690702e-06,
      "loss": 0.5907,
      "step": 5441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6677143422472959,
      "learning_rate": 1.6552660918905361e-06,
      "loss": 0.5591,
      "step": 5442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.658312882956476,
      "learning_rate": 1.6525780996688146e-06,
      "loss": 0.5686,
      "step": 5443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.685311123109449,
      "learning_rate": 1.6498920951439756e-06,
      "loss": 0.5601,
      "step": 5444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.63123269350227,
      "learning_rate": 1.6472080789556145e-06,
      "loss": 0.5489,
      "step": 5445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6504010370156236,
      "learning_rate": 1.6445260517428486e-06,
      "loss": 0.5619,
      "step": 5446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6680216483700053,
      "learning_rate": 1.6418460141443303e-06,
      "loss": 0.5544,
      "step": 5447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6350238597463168,
      "learning_rate": 1.6391679667982241e-06,
      "loss": 0.5656,
      "step": 5448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6839270731124479,
      "learning_rate": 1.6364919103422394e-06,
      "loss": 0.5726,
      "step": 5449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5758181149160336,
      "learning_rate": 1.6338178454135977e-06,
      "loss": 0.5786,
      "step": 5450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6110166298584634,
      "learning_rate": 1.6311457726490466e-06,
      "loss": 0.5565,
      "step": 5451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6301852240121807,
      "learning_rate": 1.6284756926848667e-06,
      "loss": 0.565,
      "step": 5452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6713391241542473,
      "learning_rate": 1.6258076061568585e-06,
      "loss": 0.5775,
      "step": 5453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5909101957306033,
      "learning_rate": 1.6231415137003536e-06,
      "loss": 0.5605,
      "step": 5454
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6387783851436956,
      "learning_rate": 1.6204774159501991e-06,
      "loss": 0.5542,
      "step": 5455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6401560389654561,
      "learning_rate": 1.6178153135407748e-06,
      "loss": 0.5573,
      "step": 5456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6239022807612169,
      "learning_rate": 1.6151552071059874e-06,
      "loss": 0.5461,
      "step": 5457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6355662067767461,
      "learning_rate": 1.6124970972792574e-06,
      "loss": 0.5919,
      "step": 5458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5788798414646011,
      "learning_rate": 1.6098409846935402e-06,
      "loss": 0.5871,
      "step": 5459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5803994507887936,
      "learning_rate": 1.6071868699813143e-06,
      "loss": 0.5567,
      "step": 5460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6362825025993337,
      "learning_rate": 1.6045347537745759e-06,
      "loss": 0.5659,
      "step": 5461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6458943767807107,
      "learning_rate": 1.6018846367048534e-06,
      "loss": 0.5652,
      "step": 5462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6152203863101294,
      "learning_rate": 1.5992365194031922e-06,
      "loss": 0.5769,
      "step": 5463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6172092543562167,
      "learning_rate": 1.5965904025001656e-06,
      "loss": 0.5664,
      "step": 5464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.62016321012933,
      "learning_rate": 1.5939462866258725e-06,
      "loss": 0.5679,
      "step": 5465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6041351641587767,
      "learning_rate": 1.5913041724099288e-06,
      "loss": 0.5616,
      "step": 5466
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5748408480738791,
      "learning_rate": 1.5886640604814796e-06,
      "loss": 0.5426,
      "step": 5467
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5922017907616337,
      "learning_rate": 1.5860259514691933e-06,
      "loss": 0.5664,
      "step": 5468
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5903657715352725,
      "learning_rate": 1.5833898460012531e-06,
      "loss": 0.5644,
      "step": 5469
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7851756319268838,
      "learning_rate": 1.5807557447053779e-06,
      "loss": 0.567,
      "step": 5470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6214675812245356,
      "learning_rate": 1.5781236482087947e-06,
      "loss": 0.5804,
      "step": 5471
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6095743164987044,
      "learning_rate": 1.5754935571382712e-06,
      "loss": 0.5827,
      "step": 5472
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6357713173534177,
      "learning_rate": 1.5728654721200808e-06,
      "loss": 0.5741,
      "step": 5473
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6537044433434285,
      "learning_rate": 1.5702393937800264e-06,
      "loss": 0.585,
      "step": 5474
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6345247560765257,
      "learning_rate": 1.567615322743432e-06,
      "loss": 0.5919,
      "step": 5475
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5925571799892349,
      "learning_rate": 1.5649932596351469e-06,
      "loss": 0.5522,
      "step": 5476
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6363959123758535,
      "learning_rate": 1.5623732050795393e-06,
      "loss": 0.5744,
      "step": 5477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6289828533183462,
      "learning_rate": 1.5597551597004968e-06,
      "loss": 0.5678,
      "step": 5478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6363497906841623,
      "learning_rate": 1.557139124121433e-06,
      "loss": 0.5827,
      "step": 5479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6428323906402743,
      "learning_rate": 1.5545250989652816e-06,
      "loss": 0.5783,
      "step": 5480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6663938399967876,
      "learning_rate": 1.551913084854494e-06,
      "loss": 0.5686,
      "step": 5481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6459156651409191,
      "learning_rate": 1.54930308241105e-06,
      "loss": 0.5668,
      "step": 5482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5840808297096107,
      "learning_rate": 1.5466950922564428e-06,
      "loss": 0.5544,
      "step": 5483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6636267998817817,
      "learning_rate": 1.5440891150116888e-06,
      "loss": 0.5803,
      "step": 5484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6443545117160318,
      "learning_rate": 1.5414851512973317e-06,
      "loss": 0.5482,
      "step": 5485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6039818600558239,
      "learning_rate": 1.5388832017334244e-06,
      "loss": 0.5749,
      "step": 5486
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6316021106385575,
      "learning_rate": 1.5362832669395466e-06,
      "loss": 0.5973,
      "step": 5487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6025920998919899,
      "learning_rate": 1.5336853475348023e-06,
      "loss": 0.5489,
      "step": 5488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5946857552800131,
      "learning_rate": 1.5310894441378043e-06,
      "loss": 0.5571,
      "step": 5489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6162194266519734,
      "learning_rate": 1.528495557366696e-06,
      "loss": 0.5657,
      "step": 5490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6266626460648828,
      "learning_rate": 1.5259036878391342e-06,
      "loss": 0.5722,
      "step": 5491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6002866340599065,
      "learning_rate": 1.523313836172301e-06,
      "loss": 0.5476,
      "step": 5492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6087721000996348,
      "learning_rate": 1.520726002982893e-06,
      "loss": 0.5559,
      "step": 5493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6662540450619996,
      "learning_rate": 1.5181401888871218e-06,
      "loss": 0.5745,
      "step": 5494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.576859644883379,
      "learning_rate": 1.5155563945007345e-06,
      "loss": 0.5654,
      "step": 5495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6388504659958791,
      "learning_rate": 1.5129746204389795e-06,
      "loss": 0.5953,
      "step": 5496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.600395973065311,
      "learning_rate": 1.5103948673166369e-06,
      "loss": 0.5345,
      "step": 5497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6210949273494621,
      "learning_rate": 1.5078171357479942e-06,
      "loss": 0.5633,
      "step": 5498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6297106135350603,
      "learning_rate": 1.5052414263468663e-06,
      "loss": 0.5664,
      "step": 5499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6439944181358521,
      "learning_rate": 1.5026677397265876e-06,
      "loss": 0.5503,
      "step": 5500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5932173146276812,
      "learning_rate": 1.5000960765000006e-06,
      "loss": 0.5599,
      "step": 5501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6924690670628683,
      "learning_rate": 1.497526437279475e-06,
      "loss": 0.5834,
      "step": 5502
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.620145531815181,
      "learning_rate": 1.4949588226768996e-06,
      "loss": 0.5485,
      "step": 5503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6283802455738061,
      "learning_rate": 1.4923932333036717e-06,
      "loss": 0.5629,
      "step": 5504
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5943228044264248,
      "learning_rate": 1.489829669770717e-06,
      "loss": 0.5479,
      "step": 5505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6428435814049704,
      "learning_rate": 1.4872681326884696e-06,
      "loss": 0.5568,
      "step": 5506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.703345170404814,
      "learning_rate": 1.4847086226668871e-06,
      "loss": 0.5886,
      "step": 5507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5616496850227489,
      "learning_rate": 1.4821511403154465e-06,
      "loss": 0.5422,
      "step": 5508
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6096707622659531,
      "learning_rate": 1.4795956862431315e-06,
      "loss": 0.5677,
      "step": 5509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6252349390655307,
      "learning_rate": 1.4770422610584534e-06,
      "loss": 0.5627,
      "step": 5510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6166123151542855,
      "learning_rate": 1.474490865369438e-06,
      "loss": 0.5679,
      "step": 5511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6290164305030489,
      "learning_rate": 1.4719414997836223e-06,
      "loss": 0.5579,
      "step": 5512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6163823188910741,
      "learning_rate": 1.4693941649080656e-06,
      "loss": 0.5601,
      "step": 5513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6205741866007781,
      "learning_rate": 1.4668488613493426e-06,
      "loss": 0.5766,
      "step": 5514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.551167920360871,
      "learning_rate": 1.464305589713545e-06,
      "loss": 0.5555,
      "step": 5515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6468067051912508,
      "learning_rate": 1.4617643506062774e-06,
      "loss": 0.5471,
      "step": 5516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6106041423658095,
      "learning_rate": 1.459225144632659e-06,
      "loss": 0.5843,
      "step": 5517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6119711182841419,
      "learning_rate": 1.456687972397336e-06,
      "loss": 0.5706,
      "step": 5518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5807185829417546,
      "learning_rate": 1.4541528345044553e-06,
      "loss": 0.5445,
      "step": 5519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6848111311773944,
      "learning_rate": 1.451619731557693e-06,
      "loss": 0.5591,
      "step": 5520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6281667613086812,
      "learning_rate": 1.4490886641602275e-06,
      "loss": 0.5708,
      "step": 5521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6521404426353327,
      "learning_rate": 1.4465596329147635e-06,
      "loss": 0.5637,
      "step": 5522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5846576211104146,
      "learning_rate": 1.444032638423517e-06,
      "loss": 0.5843,
      "step": 5523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6132942481872773,
      "learning_rate": 1.4415076812882156e-06,
      "loss": 0.5806,
      "step": 5524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6185127637691737,
      "learning_rate": 1.4389847621101061e-06,
      "loss": 0.5645,
      "step": 5525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5969392137946592,
      "learning_rate": 1.4364638814899513e-06,
      "loss": 0.569,
      "step": 5526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6509740718150363,
      "learning_rate": 1.4339450400280208e-06,
      "loss": 0.5411,
      "step": 5527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6719050829276946,
      "learning_rate": 1.4314282383241097e-06,
      "loss": 0.5474,
      "step": 5528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6311302465594356,
      "learning_rate": 1.4289134769775147e-06,
      "loss": 0.558,
      "step": 5529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5966079670451214,
      "learning_rate": 1.4264007565870586e-06,
      "loss": 0.5405,
      "step": 5530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5806526431196868,
      "learning_rate": 1.4238900777510734e-06,
      "loss": 0.564,
      "step": 5531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6822940623562641,
      "learning_rate": 1.4213814410673999e-06,
      "loss": 0.5705,
      "step": 5532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6755294697244177,
      "learning_rate": 1.4188748471334003e-06,
      "loss": 0.5598,
      "step": 5533
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6318748400741562,
      "learning_rate": 1.416370296545949e-06,
      "loss": 0.5492,
      "step": 5534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.652528855549961,
      "learning_rate": 1.4138677899014276e-06,
      "loss": 0.5816,
      "step": 5535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6485393063389655,
      "learning_rate": 1.4113673277957395e-06,
      "loss": 0.5668,
      "step": 5536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6462311309119105,
      "learning_rate": 1.4088689108242959e-06,
      "loss": 0.5913,
      "step": 5537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6452705397134414,
      "learning_rate": 1.406372539582025e-06,
      "loss": 0.5578,
      "step": 5538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5608946110438167,
      "learning_rate": 1.4038782146633634e-06,
      "loss": 0.5264,
      "step": 5539
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6230367556837717,
      "learning_rate": 1.4013859366622595e-06,
      "loss": 0.5594,
      "step": 5540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5913109577707173,
      "learning_rate": 1.3988957061721797e-06,
      "loss": 0.5779,
      "step": 5541
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6135689696319503,
      "learning_rate": 1.3964075237861007e-06,
      "loss": 0.5578,
      "step": 5542
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.719158480057417,
      "learning_rate": 1.3939213900965133e-06,
      "loss": 0.5939,
      "step": 5543
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6863270815953888,
      "learning_rate": 1.3914373056954122e-06,
      "loss": 0.5715,
      "step": 5544
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6097896267447782,
      "learning_rate": 1.3889552711743147e-06,
      "loss": 0.5639,
      "step": 5545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6296384825060306,
      "learning_rate": 1.386475287124247e-06,
      "loss": 0.5558,
      "step": 5546
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6054960781495676,
      "learning_rate": 1.383997354135741e-06,
      "loss": 0.5609,
      "step": 5547
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5766899064247477,
      "learning_rate": 1.381521472798847e-06,
      "loss": 0.5744,
      "step": 5548
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.605833840927251,
      "learning_rate": 1.3790476437031252e-06,
      "loss": 0.5348,
      "step": 5549
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.684232255954573,
      "learning_rate": 1.376575867437644e-06,
      "loss": 0.5721,
      "step": 5550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6160475148462224,
      "learning_rate": 1.3741061445909886e-06,
      "loss": 0.5587,
      "step": 5551
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6050377174218753,
      "learning_rate": 1.3716384757512458e-06,
      "loss": 0.5456,
      "step": 5552
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6387076462775834,
      "learning_rate": 1.3691728615060284e-06,
      "loss": 0.5699,
      "step": 5553
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5218833368180053,
      "learning_rate": 1.366709302442446e-06,
      "loss": 0.5611,
      "step": 5554
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6034328747249339,
      "learning_rate": 1.3642477991471225e-06,
      "loss": 0.5838,
      "step": 5555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.679318206139652,
      "learning_rate": 1.361788352206196e-06,
      "loss": 0.5901,
      "step": 5556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5487231387652045,
      "learning_rate": 1.3593309622053118e-06,
      "loss": 0.5665,
      "step": 5557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5788967417095706,
      "learning_rate": 1.3568756297296292e-06,
      "loss": 0.547,
      "step": 5558
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6180690845465511,
      "learning_rate": 1.3544223553638091e-06,
      "loss": 0.5604,
      "step": 5559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5914523812883261,
      "learning_rate": 1.351971139692031e-06,
      "loss": 0.5666,
      "step": 5560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5474104032950987,
      "learning_rate": 1.3495219832979821e-06,
      "loss": 0.5436,
      "step": 5561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6861385540338986,
      "learning_rate": 1.3470748867648576e-06,
      "loss": 0.5646,
      "step": 5562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6299828605161341,
      "learning_rate": 1.3446298506753585e-06,
      "loss": 0.5786,
      "step": 5563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6324216787765521,
      "learning_rate": 1.3421868756117028e-06,
      "loss": 0.5755,
      "step": 5564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6561275637472103,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.5695,
      "step": 5565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5737981286462575,
      "learning_rate": 1.3373071108883263e-06,
      "loss": 0.5582,
      "step": 5566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.576117215345895,
      "learning_rate": 1.334870322390579e-06,
      "loss": 0.5675,
      "step": 5567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5962542510916236,
      "learning_rate": 1.3324355972426228e-06,
      "loss": 0.5726,
      "step": 5568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6391916259638765,
      "learning_rate": 1.3300029360242218e-06,
      "loss": 0.5651,
      "step": 5569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.589178404817528,
      "learning_rate": 1.3275723393146367e-06,
      "loss": 0.5509,
      "step": 5570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6453352994502568,
      "learning_rate": 1.325143807692648e-06,
      "loss": 0.5706,
      "step": 5571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6714892070361029,
      "learning_rate": 1.3227173417365413e-06,
      "loss": 0.5912,
      "step": 5572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5756233152154661,
      "learning_rate": 1.320292942024105e-06,
      "loss": 0.553,
      "step": 5573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6605824047859921,
      "learning_rate": 1.3178706091326455e-06,
      "loss": 0.5663,
      "step": 5574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6040078229440524,
      "learning_rate": 1.315450343638962e-06,
      "loss": 0.5353,
      "step": 5575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6418029933259713,
      "learning_rate": 1.3130321461193807e-06,
      "loss": 0.5646,
      "step": 5576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6337128061185311,
      "learning_rate": 1.3106160171497217e-06,
      "loss": 0.5629,
      "step": 5577
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.698686861556483,
      "learning_rate": 1.308201957305314e-06,
      "loss": 0.5865,
      "step": 5578
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5724309925022285,
      "learning_rate": 1.3057899671609974e-06,
      "loss": 0.5556,
      "step": 5579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6475882057643619,
      "learning_rate": 1.3033800472911174e-06,
      "loss": 0.5869,
      "step": 5580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5811887282862747,
      "learning_rate": 1.300972198269529e-06,
      "loss": 0.555,
      "step": 5581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6641922277504947,
      "learning_rate": 1.2985664206695902e-06,
      "loss": 0.5523,
      "step": 5582
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6703254472257169,
      "learning_rate": 1.296162715064162e-06,
      "loss": 0.5639,
      "step": 5583
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6255623129582402,
      "learning_rate": 1.2937610820256275e-06,
      "loss": 0.5673,
      "step": 5584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6561569703000806,
      "learning_rate": 1.2913615221258579e-06,
      "loss": 0.5585,
      "step": 5585
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6826152180338398,
      "learning_rate": 1.2889640359362432e-06,
      "loss": 0.5503,
      "step": 5586
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5517104186184045,
      "learning_rate": 1.2865686240276732e-06,
      "loss": 0.5556,
      "step": 5587
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5974691301092983,
      "learning_rate": 1.2841752869705459e-06,
      "loss": 0.5684,
      "step": 5588
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6706963086973307,
      "learning_rate": 1.2817840253347679e-06,
      "loss": 0.5729,
      "step": 5589
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6498619622929783,
      "learning_rate": 1.279394839689746e-06,
      "loss": 0.532,
      "step": 5590
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6105630286285397,
      "learning_rate": 1.277007730604396e-06,
      "loss": 0.5861,
      "step": 5591
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6205479110148233,
      "learning_rate": 1.274622698647141e-06,
      "loss": 0.5725,
      "step": 5592
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5966402236868459,
      "learning_rate": 1.2722397443859036e-06,
      "loss": 0.5677,
      "step": 5593
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5998597611497956,
      "learning_rate": 1.2698588683881185e-06,
      "loss": 0.5506,
      "step": 5594
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6044774069534479,
      "learning_rate": 1.2674800712207226e-06,
      "loss": 0.5689,
      "step": 5595
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6565183199429272,
      "learning_rate": 1.2651033534501543e-06,
      "loss": 0.5619,
      "step": 5596
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9958017040158006,
      "learning_rate": 1.262728715642365e-06,
      "loss": 0.5862,
      "step": 5597
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6111755230407836,
      "learning_rate": 1.260356158362801e-06,
      "loss": 0.5924,
      "step": 5598
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6475599922624558,
      "learning_rate": 1.2579856821764202e-06,
      "loss": 0.5522,
      "step": 5599
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.678518533323929,
      "learning_rate": 1.2556172876476846e-06,
      "loss": 0.5735,
      "step": 5600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6587612597633205,
      "learning_rate": 1.2532509753405554e-06,
      "loss": 0.5769,
      "step": 5601
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6814436696750141,
      "learning_rate": 1.2508867458185037e-06,
      "loss": 0.5845,
      "step": 5602
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6036787806612784,
      "learning_rate": 1.2485245996445006e-06,
      "loss": 0.5568,
      "step": 5603
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5975741399870128,
      "learning_rate": 1.2461645373810272e-06,
      "loss": 0.5873,
      "step": 5604
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7130057871054347,
      "learning_rate": 1.2438065595900605e-06,
      "loss": 0.6097,
      "step": 5605
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.623274359347084,
      "learning_rate": 1.2414506668330805e-06,
      "loss": 0.5427,
      "step": 5606
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6807524829129833,
      "learning_rate": 1.239096859671084e-06,
      "loss": 0.5959,
      "step": 5607
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6192497252610639,
      "learning_rate": 1.2367451386645558e-06,
      "loss": 0.5679,
      "step": 5608
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6526265881243973,
      "learning_rate": 1.2343955043734924e-06,
      "loss": 0.6037,
      "step": 5609
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.655553057913997,
      "learning_rate": 1.2320479573573895e-06,
      "loss": 0.5458,
      "step": 5610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6034227184798745,
      "learning_rate": 1.2297024981752482e-06,
      "loss": 0.5576,
      "step": 5611
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5807035035455281,
      "learning_rate": 1.2273591273855744e-06,
      "loss": 0.5527,
      "step": 5612
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.587927943569368,
      "learning_rate": 1.225017845546369e-06,
      "loss": 0.5632,
      "step": 5613
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6413968932572067,
      "learning_rate": 1.2226786532151435e-06,
      "loss": 0.5653,
      "step": 5614
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6535385391982129,
      "learning_rate": 1.2203415509489102e-06,
      "loss": 0.564,
      "step": 5615
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.650107196665328,
      "learning_rate": 1.2180065393041796e-06,
      "loss": 0.5629,
      "step": 5616
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6762790234468732,
      "learning_rate": 1.2156736188369667e-06,
      "loss": 0.5826,
      "step": 5617
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6649059977780543,
      "learning_rate": 1.2133427901027916e-06,
      "loss": 0.5659,
      "step": 5618
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6263952988623949,
      "learning_rate": 1.211014053656674e-06,
      "loss": 0.5619,
      "step": 5619
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6383139171826311,
      "learning_rate": 1.2086874100531342e-06,
      "loss": 0.5577,
      "step": 5620
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6687514022646305,
      "learning_rate": 1.206362859846192e-06,
      "loss": 0.5571,
      "step": 5621
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6639340387772328,
      "learning_rate": 1.2040404035893737e-06,
      "loss": 0.5903,
      "step": 5622
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6490309980840434,
      "learning_rate": 1.2017200418357077e-06,
      "loss": 0.5806,
      "step": 5623
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.534247435771868,
      "learning_rate": 1.1994017751377175e-06,
      "loss": 0.5448,
      "step": 5624
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6219116131017847,
      "learning_rate": 1.1970856040474311e-06,
      "loss": 0.5762,
      "step": 5625
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6420302400813507,
      "learning_rate": 1.1947715291163797e-06,
      "loss": 0.5784,
      "step": 5626
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.61131910438501,
      "learning_rate": 1.1924595508955949e-06,
      "loss": 0.5509,
      "step": 5627
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6472852257190013,
      "learning_rate": 1.1901496699356041e-06,
      "loss": 0.5632,
      "step": 5628
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7031703499821015,
      "learning_rate": 1.187841886786436e-06,
      "loss": 0.5996,
      "step": 5629
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6638964386822112,
      "learning_rate": 1.1855362019976302e-06,
      "loss": 0.579,
      "step": 5630
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6307371371144364,
      "learning_rate": 1.1832326161182118e-06,
      "loss": 0.5492,
      "step": 5631
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5986985011869935,
      "learning_rate": 1.1809311296967184e-06,
      "loss": 0.563,
      "step": 5632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6284251099813007,
      "learning_rate": 1.1786317432811767e-06,
      "loss": 0.5661,
      "step": 5633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6713040076388473,
      "learning_rate": 1.1763344574191227e-06,
      "loss": 0.5797,
      "step": 5634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6426989016411977,
      "learning_rate": 1.17403927265759e-06,
      "loss": 0.5616,
      "step": 5635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6239199321125886,
      "learning_rate": 1.171746189543106e-06,
      "loss": 0.5949,
      "step": 5636
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6004577823868661,
      "learning_rate": 1.1694552086217037e-06,
      "loss": 0.5645,
      "step": 5637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5973469986835784,
      "learning_rate": 1.1671663304389169e-06,
      "loss": 0.535,
      "step": 5638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6281560312209848,
      "learning_rate": 1.1648795555397719e-06,
      "loss": 0.5528,
      "step": 5639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5919684698137684,
      "learning_rate": 1.1625948844688007e-06,
      "loss": 0.5321,
      "step": 5640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5885290267207351,
      "learning_rate": 1.160312317770026e-06,
      "loss": 0.5482,
      "step": 5641
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6171483433359172,
      "learning_rate": 1.158031855986983e-06,
      "loss": 0.5666,
      "step": 5642
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6193960880449251,
      "learning_rate": 1.155753499662694e-06,
      "loss": 0.5623,
      "step": 5643
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6279518193369725,
      "learning_rate": 1.1534772493396818e-06,
      "loss": 0.5789,
      "step": 5644
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5865833903878411,
      "learning_rate": 1.1512031055599703e-06,
      "loss": 0.5653,
      "step": 5645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6138126634954277,
      "learning_rate": 1.1489310688650834e-06,
      "loss": 0.5639,
      "step": 5646
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6384424664306669,
      "learning_rate": 1.1466611397960404e-06,
      "loss": 0.556,
      "step": 5647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6569098707847658,
      "learning_rate": 1.1443933188933554e-06,
      "loss": 0.5853,
      "step": 5648
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5485877443523639,
      "learning_rate": 1.1421276066970478e-06,
      "loss": 0.5608,
      "step": 5649
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5745048028350213,
      "learning_rate": 1.1398640037466325e-06,
      "loss": 0.5547,
      "step": 5650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6171433082247935,
      "learning_rate": 1.1376025105811172e-06,
      "loss": 0.5443,
      "step": 5651
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6151661870719836,
      "learning_rate": 1.1353431277390125e-06,
      "loss": 0.5611,
      "step": 5652
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5516451876423385,
      "learning_rate": 1.1330858557583279e-06,
      "loss": 0.5433,
      "step": 5653
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6476057392756945,
      "learning_rate": 1.1308306951765635e-06,
      "loss": 0.5637,
      "step": 5654
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5920243156729522,
      "learning_rate": 1.128577646530723e-06,
      "loss": 0.5529,
      "step": 5655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5992698271254083,
      "learning_rate": 1.1263267103573016e-06,
      "loss": 0.5353,
      "step": 5656
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.57807905439476,
      "learning_rate": 1.1240778871922976e-06,
      "loss": 0.5507,
      "step": 5657
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6229879860421623,
      "learning_rate": 1.1218311775712031e-06,
      "loss": 0.5848,
      "step": 5658
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6287991323831155,
      "learning_rate": 1.1195865820290052e-06,
      "loss": 0.5858,
      "step": 5659
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6244417079944827,
      "learning_rate": 1.1173441011001895e-06,
      "loss": 0.5718,
      "step": 5660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6627731755905817,
      "learning_rate": 1.11510373531874e-06,
      "loss": 0.5564,
      "step": 5661
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6615476387440857,
      "learning_rate": 1.112865485218132e-06,
      "loss": 0.5701,
      "step": 5662
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6394831815876743,
      "learning_rate": 1.1106293513313437e-06,
      "loss": 0.5738,
      "step": 5663
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6678735984797651,
      "learning_rate": 1.1083953341908393e-06,
      "loss": 0.5721,
      "step": 5664
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7294715735950664,
      "learning_rate": 1.1061634343285944e-06,
      "loss": 0.6,
      "step": 5665
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8266455276924443,
      "learning_rate": 1.1039336522760659e-06,
      "loss": 0.5891,
      "step": 5666
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5449110826165351,
      "learning_rate": 1.10170598856421e-06,
      "loss": 0.5404,
      "step": 5667
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6852735621688899,
      "learning_rate": 1.099480443723483e-06,
      "loss": 0.5624,
      "step": 5668
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6460453382951052,
      "learning_rate": 1.0972570182838327e-06,
      "loss": 0.5733,
      "step": 5669
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6233241802426832,
      "learning_rate": 1.095035712774708e-06,
      "loss": 0.5642,
      "step": 5670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5840118425012077,
      "learning_rate": 1.0928165277250425e-06,
      "loss": 0.571,
      "step": 5671
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6427321876028976,
      "learning_rate": 1.0905994636632732e-06,
      "loss": 0.592,
      "step": 5672
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6024191247789298,
      "learning_rate": 1.0883845211173315e-06,
      "loss": 0.5615,
      "step": 5673
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.58604971885849,
      "learning_rate": 1.0861717006146388e-06,
      "loss": 0.5587,
      "step": 5674
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6404301671633988,
      "learning_rate": 1.0839610026821179e-06,
      "loss": 0.5594,
      "step": 5675
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6400316137622091,
      "learning_rate": 1.0817524278461777e-06,
      "loss": 0.5821,
      "step": 5676
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6459395447019797,
      "learning_rate": 1.07954597663273e-06,
      "loss": 0.5448,
      "step": 5677
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.586833191133254,
      "learning_rate": 1.0773416495671773e-06,
      "loss": 0.5732,
      "step": 5678
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6433013274726225,
      "learning_rate": 1.0751394471744138e-06,
      "loss": 0.5672,
      "step": 5679
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6322822885194321,
      "learning_rate": 1.0729393699788303e-06,
      "loss": 0.5462,
      "step": 5680
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6367826428925966,
      "learning_rate": 1.0707414185043163e-06,
      "loss": 0.5681,
      "step": 5681
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5647177630124289,
      "learning_rate": 1.068545593274245e-06,
      "loss": 0.5604,
      "step": 5682
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6087766614516823,
      "learning_rate": 1.0663518948114892e-06,
      "loss": 0.5769,
      "step": 5683
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6088294809452011,
      "learning_rate": 1.064160323638418e-06,
      "loss": 0.5512,
      "step": 5684
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5894927695914957,
      "learning_rate": 1.0619708802768891e-06,
      "loss": 0.5605,
      "step": 5685
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.623319630811217,
      "learning_rate": 1.0597835652482545e-06,
      "loss": 0.5724,
      "step": 5686
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6586902295760652,
      "learning_rate": 1.0575983790733569e-06,
      "loss": 0.5584,
      "step": 5687
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6893312727637745,
      "learning_rate": 1.0554153222725417e-06,
      "loss": 0.5631,
      "step": 5688
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6239711938155726,
      "learning_rate": 1.0532343953656377e-06,
      "loss": 0.5832,
      "step": 5689
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6190941390942957,
      "learning_rate": 1.0510555988719663e-06,
      "loss": 0.5668,
      "step": 5690
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5960450029959312,
      "learning_rate": 1.0488789333103488e-06,
      "loss": 0.5461,
      "step": 5691
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6336770306732148,
      "learning_rate": 1.0467043991990932e-06,
      "loss": 0.6061,
      "step": 5692
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6364322746151064,
      "learning_rate": 1.0445319970560042e-06,
      "loss": 0.5546,
      "step": 5693
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6415519446458355,
      "learning_rate": 1.0423617273983733e-06,
      "loss": 0.5391,
      "step": 5694
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6180655509274008,
      "learning_rate": 1.0401935907429883e-06,
      "loss": 0.5599,
      "step": 5695
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6047284805220455,
      "learning_rate": 1.0380275876061307e-06,
      "loss": 0.5497,
      "step": 5696
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6370198962575218,
      "learning_rate": 1.035863718503568e-06,
      "loss": 0.5724,
      "step": 5697
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6707013379748626,
      "learning_rate": 1.0337019839505669e-06,
      "loss": 0.5699,
      "step": 5698
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5844308869630567,
      "learning_rate": 1.0315423844618767e-06,
      "loss": 0.5447,
      "step": 5699
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6585171053932398,
      "learning_rate": 1.029384920551747e-06,
      "loss": 0.5374,
      "step": 5700
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5410546911545374,
      "learning_rate": 1.027229592733916e-06,
      "loss": 0.5332,
      "step": 5701
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6611333065854856,
      "learning_rate": 1.0250764015216097e-06,
      "loss": 0.5433,
      "step": 5702
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6950360434268384,
      "learning_rate": 1.0229253474275502e-06,
      "loss": 0.5523,
      "step": 5703
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6986341304990653,
      "learning_rate": 1.02077643096395e-06,
      "loss": 0.5432,
      "step": 5704
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6175985096616463,
      "learning_rate": 1.0186296526425076e-06,
      "loss": 0.5524,
      "step": 5705
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7108868401251779,
      "learning_rate": 1.0164850129744186e-06,
      "loss": 0.5656,
      "step": 5706
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6854884690232206,
      "learning_rate": 1.0143425124703665e-06,
      "loss": 0.5624,
      "step": 5707
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6002868776019208,
      "learning_rate": 1.012202151640528e-06,
      "loss": 0.556,
      "step": 5708
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6136738063870494,
      "learning_rate": 1.0100639309945659e-06,
      "loss": 0.5557,
      "step": 5709
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7689523892612219,
      "learning_rate": 1.0079278510416313e-06,
      "loss": 0.5714,
      "step": 5710
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6676528772223393,
      "learning_rate": 1.005793912290378e-06,
      "loss": 0.5885,
      "step": 5711
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6579639148876085,
      "learning_rate": 1.0036621152489357e-06,
      "loss": 0.5778,
      "step": 5712
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6496186133392244,
      "learning_rate": 1.0015324604249343e-06,
      "loss": 0.589,
      "step": 5713
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.65305363291709,
      "learning_rate": 9.994049483254865e-07,
      "loss": 0.597,
      "step": 5714
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6198006400218625,
      "learning_rate": 9.972795794571976e-07,
      "loss": 0.545,
      "step": 5715
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6592587600403158,
      "learning_rate": 9.951563543261656e-07,
      "loss": 0.5635,
      "step": 5716
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6244084385149953,
      "learning_rate": 9.930352734379734e-07,
      "loss": 0.5635,
      "step": 5717
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6188964014974753,
      "learning_rate": 9.909163372976903e-07,
      "loss": 0.5437,
      "step": 5718
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6007066073017998,
      "learning_rate": 9.887995464098888e-07,
      "loss": 0.5642,
      "step": 5719
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5830464974721424,
      "learning_rate": 9.86684901278614e-07,
      "loss": 0.5563,
      "step": 5720
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6944886197889149,
      "learning_rate": 9.845724024074122e-07,
      "loss": 0.5689,
      "step": 5721
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6012229661178727,
      "learning_rate": 9.824620502993098e-07,
      "loss": 0.5499,
      "step": 5722
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6216329473377932,
      "learning_rate": 9.803538454568285e-07,
      "loss": 0.5769,
      "step": 5723
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6837635990236801,
      "learning_rate": 9.782477883819775e-07,
      "loss": 0.59,
      "step": 5724
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.681037803358642,
      "learning_rate": 9.761438795762491e-07,
      "loss": 0.5808,
      "step": 5725
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6403819856531344,
      "learning_rate": 9.740421195406314e-07,
      "loss": 0.537,
      "step": 5726
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.578329497188657,
      "learning_rate": 9.719425087755984e-07,
      "loss": 0.5657,
      "step": 5727
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5803934058989836,
      "learning_rate": 9.698450477811095e-07,
      "loss": 0.5591,
      "step": 5728
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7066738972726413,
      "learning_rate": 9.677497370566135e-07,
      "loss": 0.552,
      "step": 5729
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6301792550380438,
      "learning_rate": 9.656565771010507e-07,
      "loss": 0.5608,
      "step": 5730
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6546582174270882,
      "learning_rate": 9.635655684128475e-07,
      "loss": 0.5595,
      "step": 5731
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6647546441218154,
      "learning_rate": 9.61476711489915e-07,
      "loss": 0.5698,
      "step": 5732
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6206117475106162,
      "learning_rate": 9.593900068296512e-07,
      "loss": 0.5498,
      "step": 5733
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6067976193227218,
      "learning_rate": 9.573054549289485e-07,
      "loss": 0.5602,
      "step": 5734
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6884550110029412,
      "learning_rate": 9.552230562841802e-07,
      "loss": 0.5727,
      "step": 5735
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6040375926812288,
      "learning_rate": 9.531428113912134e-07,
      "loss": 0.5688,
      "step": 5736
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7055306607440656,
      "learning_rate": 9.510647207453927e-07,
      "loss": 0.5697,
      "step": 5737
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6251880639205337,
      "learning_rate": 9.489887848415569e-07,
      "loss": 0.5542,
      "step": 5738
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6461461176310945,
      "learning_rate": 9.469150041740338e-07,
      "loss": 0.5717,
      "step": 5739
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.676489451247401,
      "learning_rate": 9.448433792366296e-07,
      "loss": 0.5687,
      "step": 5740
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7043459716237557,
      "learning_rate": 9.42773910522643e-07,
      "loss": 0.5767,
      "step": 5741
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.667839102885621,
      "learning_rate": 9.40706598524861e-07,
      "loss": 0.5681,
      "step": 5742
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7174710901124429,
      "learning_rate": 9.386414437355495e-07,
      "loss": 0.5575,
      "step": 5743
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7215692704136684,
      "learning_rate": 9.365784466464689e-07,
      "loss": 0.5708,
      "step": 5744
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5905827279234684,
      "learning_rate": 9.345176077488583e-07,
      "loss": 0.5655,
      "step": 5745
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6602143900889303,
      "learning_rate": 9.324589275334517e-07,
      "loss": 0.6099,
      "step": 5746
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5734408813433248,
      "learning_rate": 9.304024064904626e-07,
      "loss": 0.5581,
      "step": 5747
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6106154395465144,
      "learning_rate": 9.283480451095894e-07,
      "loss": 0.5508,
      "step": 5748
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5536402496032428,
      "learning_rate": 9.262958438800207e-07,
      "loss": 0.5716,
      "step": 5749
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6945206966297174,
      "learning_rate": 9.242458032904311e-07,
      "loss": 0.5331,
      "step": 5750
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6941520778664024,
      "learning_rate": 9.221979238289736e-07,
      "loss": 0.5747,
      "step": 5751
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7029107271183288,
      "learning_rate": 9.201522059832935e-07,
      "loss": 0.5838,
      "step": 5752
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6476083349710973,
      "learning_rate": 9.181086502405201e-07,
      "loss": 0.5725,
      "step": 5753
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7323134274364502,
      "learning_rate": 9.160672570872687e-07,
      "loss": 0.5602,
      "step": 5754
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6058325195497022,
      "learning_rate": 9.140280270096358e-07,
      "loss": 0.563,
      "step": 5755
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5811456632656877,
      "learning_rate": 9.119909604932031e-07,
      "loss": 0.5574,
      "step": 5756
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6296273211087151,
      "learning_rate": 9.099560580230405e-07,
      "loss": 0.5992,
      "step": 5757
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6313319695015991,
      "learning_rate": 9.079233200837023e-07,
      "loss": 0.5492,
      "step": 5758
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6586013784867923,
      "learning_rate": 9.058927471592272e-07,
      "loss": 0.5685,
      "step": 5759
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5905949694491365,
      "learning_rate": 9.038643397331337e-07,
      "loss": 0.5585,
      "step": 5760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6405422377609133,
      "learning_rate": 9.018380982884311e-07,
      "loss": 0.5733,
      "step": 5761
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6361964108383867,
      "learning_rate": 8.998140233076103e-07,
      "loss": 0.5479,
      "step": 5762
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5483744244536749,
      "learning_rate": 8.977921152726432e-07,
      "loss": 0.5701,
      "step": 5763
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.605451762728419,
      "learning_rate": 8.957723746649916e-07,
      "loss": 0.5747,
      "step": 5764
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9648912171619192,
      "learning_rate": 8.937548019655984e-07,
      "loss": 0.5793,
      "step": 5765
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6212242423948735,
      "learning_rate": 8.917393976548882e-07,
      "loss": 0.5739,
      "step": 5766
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6598903025842897,
      "learning_rate": 8.897261622127728e-07,
      "loss": 0.5715,
      "step": 5767
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6430158145197202,
      "learning_rate": 8.87715096118642e-07,
      "loss": 0.5604,
      "step": 5768
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7028742321998184,
      "learning_rate": 8.857061998513794e-07,
      "loss": 0.6032,
      "step": 5769
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6498631189507282,
      "learning_rate": 8.836994738893434e-07,
      "loss": 0.5608,
      "step": 5770
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7065015075672392,
      "learning_rate": 8.816949187103741e-07,
      "loss": 0.5664,
      "step": 5771
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.568252199615863,
      "learning_rate": 8.796925347918006e-07,
      "loss": 0.5596,
      "step": 5772
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5999222821972433,
      "learning_rate": 8.776923226104328e-07,
      "loss": 0.5404,
      "step": 5773
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6182363942520853,
      "learning_rate": 8.756942826425652e-07,
      "loss": 0.5584,
      "step": 5774
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7096147969187653,
      "learning_rate": 8.736984153639716e-07,
      "loss": 0.563,
      "step": 5775
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5809335278629923,
      "learning_rate": 8.717047212499052e-07,
      "loss": 0.559,
      "step": 5776
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5547469658566676,
      "learning_rate": 8.697132007751163e-07,
      "loss": 0.5516,
      "step": 5777
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6241203758357738,
      "learning_rate": 8.677238544138201e-07,
      "loss": 0.5609,
      "step": 5778
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6416886817857231,
      "learning_rate": 8.657366826397251e-07,
      "loss": 0.5531,
      "step": 5779
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6792374899594569,
      "learning_rate": 8.637516859260176e-07,
      "loss": 0.5928,
      "step": 5780
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6756116133693122,
      "learning_rate": 8.61768864745367e-07,
      "loss": 0.5947,
      "step": 5781
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5714435490871749,
      "learning_rate": 8.59788219569927e-07,
      "loss": 0.5548,
      "step": 5782
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.66880473964041,
      "learning_rate": 8.578097508713279e-07,
      "loss": 0.5616,
      "step": 5783
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6559829603824449,
      "learning_rate": 8.558334591206852e-07,
      "loss": 0.5665,
      "step": 5784
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6300402381313942,
      "learning_rate": 8.53859344788599e-07,
      "loss": 0.5614,
      "step": 5785
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6709497884500135,
      "learning_rate": 8.518874083451434e-07,
      "loss": 0.5581,
      "step": 5786
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6397013512546293,
      "learning_rate": 8.499176502598783e-07,
      "loss": 0.5807,
      "step": 5787
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6699291179989549,
      "learning_rate": 8.479500710018484e-07,
      "loss": 0.5763,
      "step": 5788
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5497593106026654,
      "learning_rate": 8.459846710395714e-07,
      "loss": 0.5687,
      "step": 5789
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6331644171876505,
      "learning_rate": 8.440214508410527e-07,
      "loss": 0.5423,
      "step": 5790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6461880151673924,
      "learning_rate": 8.420604108737751e-07,
      "loss": 0.5638,
      "step": 5791
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5405985891058153,
      "learning_rate": 8.401015516047039e-07,
      "loss": 0.5415,
      "step": 5792
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6095554150231984,
      "learning_rate": 8.381448735002861e-07,
      "loss": 0.562,
      "step": 5793
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6255738294082971,
      "learning_rate": 8.361903770264457e-07,
      "loss": 0.5571,
      "step": 5794
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6211953583798966,
      "learning_rate": 8.342380626485902e-07,
      "loss": 0.5482,
      "step": 5795
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.598487273116422,
      "learning_rate": 8.322879308316078e-07,
      "loss": 0.5548,
      "step": 5796
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6500106933665514,
      "learning_rate": 8.303399820398672e-07,
      "loss": 0.5759,
      "step": 5797
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5996356282736349,
      "learning_rate": 8.283942167372128e-07,
      "loss": 0.5776,
      "step": 5798
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5826798401931941,
      "learning_rate": 8.264506353869717e-07,
      "loss": 0.5541,
      "step": 5799
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6777397846829961,
      "learning_rate": 8.24509238451956e-07,
      "loss": 0.5704,
      "step": 5800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6465390639473159,
      "learning_rate": 8.225700263944481e-07,
      "loss": 0.5812,
      "step": 5801
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6658779342209908,
      "learning_rate": 8.206329996762208e-07,
      "loss": 0.5724,
      "step": 5802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8330409050010061,
      "learning_rate": 8.186981587585152e-07,
      "loss": 0.5441,
      "step": 5803
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5925829650968275,
      "learning_rate": 8.167655041020606e-07,
      "loss": 0.5681,
      "step": 5804
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6447636860025934,
      "learning_rate": 8.148350361670643e-07,
      "loss": 0.5681,
      "step": 5805
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6067027081503862,
      "learning_rate": 8.129067554132075e-07,
      "loss": 0.5761,
      "step": 5806
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6695362413045725,
      "learning_rate": 8.109806622996563e-07,
      "loss": 0.5489,
      "step": 5807
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5749321449622535,
      "learning_rate": 8.090567572850561e-07,
      "loss": 0.5635,
      "step": 5808
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6628131487804725,
      "learning_rate": 8.071350408275258e-07,
      "loss": 0.5666,
      "step": 5809
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7568810111975013,
      "learning_rate": 8.052155133846695e-07,
      "loss": 0.5382,
      "step": 5810
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.615289956528749,
      "learning_rate": 8.032981754135638e-07,
      "loss": 0.5581,
      "step": 5811
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6207617729914912,
      "learning_rate": 8.01383027370769e-07,
      "loss": 0.5508,
      "step": 5812
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6567569955701847,
      "learning_rate": 7.994700697123247e-07,
      "loss": 0.5731,
      "step": 5813
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6222408749434861,
      "learning_rate": 7.975593028937412e-07,
      "loss": 0.549,
      "step": 5814
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5992961542824257,
      "learning_rate": 7.956507273700154e-07,
      "loss": 0.547,
      "step": 5815
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.596636038734861,
      "learning_rate": 7.937443435956205e-07,
      "loss": 0.5357,
      "step": 5816
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6964998621384252,
      "learning_rate": 7.918401520245033e-07,
      "loss": 0.5666,
      "step": 5817
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5736568607595747,
      "learning_rate": 7.899381531100936e-07,
      "loss": 0.5816,
      "step": 5818
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6671131295126457,
      "learning_rate": 7.880383473052966e-07,
      "loss": 0.5634,
      "step": 5819
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6490646974973376,
      "learning_rate": 7.861407350624994e-07,
      "loss": 0.572,
      "step": 5820
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6415842875142371,
      "learning_rate": 7.842453168335607e-07,
      "loss": 0.5897,
      "step": 5821
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6031914748912153,
      "learning_rate": 7.82352093069817e-07,
      "loss": 0.5462,
      "step": 5822
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6903424909346659,
      "learning_rate": 7.8046106422209e-07,
      "loss": 0.5864,
      "step": 5823
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6356844509374928,
      "learning_rate": 7.785722307406685e-07,
      "loss": 0.5696,
      "step": 5824
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5918056151355708,
      "learning_rate": 7.766855930753281e-07,
      "loss": 0.5555,
      "step": 5825
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6306739484306971,
      "learning_rate": 7.74801151675314e-07,
      "loss": 0.5525,
      "step": 5826
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6090849250635636,
      "learning_rate": 7.729189069893506e-07,
      "loss": 0.5716,
      "step": 5827
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6670578170638045,
      "learning_rate": 7.710388594656449e-07,
      "loss": 0.5837,
      "step": 5828
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.66534311035265,
      "learning_rate": 7.691610095518686e-07,
      "loss": 0.5502,
      "step": 5829
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6321233780903426,
      "learning_rate": 7.672853576951822e-07,
      "loss": 0.5642,
      "step": 5830
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6299215416286862,
      "learning_rate": 7.654119043422192e-07,
      "loss": 0.5615,
      "step": 5831
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6152986821803937,
      "learning_rate": 7.635406499390829e-07,
      "loss": 0.5695,
      "step": 5832
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6152259207474653,
      "learning_rate": 7.616715949313636e-07,
      "loss": 0.5379,
      "step": 5833
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6613898617018487,
      "learning_rate": 7.598047397641162e-07,
      "loss": 0.5734,
      "step": 5834
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5744223913382174,
      "learning_rate": 7.579400848818864e-07,
      "loss": 0.5349,
      "step": 5835
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6635579523306571,
      "learning_rate": 7.560776307286843e-07,
      "loss": 0.5694,
      "step": 5836
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5876562540117716,
      "learning_rate": 7.542173777479966e-07,
      "loss": 0.554,
      "step": 5837
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5768252490569179,
      "learning_rate": 7.5235932638279e-07,
      "loss": 0.5616,
      "step": 5838
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6230860269552428,
      "learning_rate": 7.505034770755082e-07,
      "loss": 0.5723,
      "step": 5839
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6267685350282644,
      "learning_rate": 7.486498302680679e-07,
      "loss": 0.5718,
      "step": 5840
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6411860631761472,
      "learning_rate": 7.467983864018568e-07,
      "loss": 0.5726,
      "step": 5841
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6535906582213611,
      "learning_rate": 7.449491459177471e-07,
      "loss": 0.5913,
      "step": 5842
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7057974824333392,
      "learning_rate": 7.431021092560819e-07,
      "loss": 0.5929,
      "step": 5843
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.745450478344984,
      "learning_rate": 7.412572768566783e-07,
      "loss": 0.5825,
      "step": 5844
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6300255175297687,
      "learning_rate": 7.394146491588261e-07,
      "loss": 0.5478,
      "step": 5845
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6050460066580907,
      "learning_rate": 7.375742266013009e-07,
      "loss": 0.5404,
      "step": 5846
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6260079857451624,
      "learning_rate": 7.357360096223409e-07,
      "loss": 0.5621,
      "step": 5847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6508891914380133,
      "learning_rate": 7.338999986596673e-07,
      "loss": 0.5723,
      "step": 5848
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5958147244611939,
      "learning_rate": 7.320661941504703e-07,
      "loss": 0.5843,
      "step": 5849
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6538385818948159,
      "learning_rate": 7.302345965314173e-07,
      "loss": 0.5918,
      "step": 5850
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6469953779740865,
      "learning_rate": 7.284052062386538e-07,
      "loss": 0.5391,
      "step": 5851
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6003382778615418,
      "learning_rate": 7.265780237077924e-07,
      "loss": 0.5654,
      "step": 5852
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.622777311510329,
      "learning_rate": 7.247530493739252e-07,
      "loss": 0.5658,
      "step": 5853
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6105165631823456,
      "learning_rate": 7.229302836716179e-07,
      "loss": 0.567,
      "step": 5854
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.618824247653524,
      "learning_rate": 7.211097270349065e-07,
      "loss": 0.5664,
      "step": 5855
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5819475011915084,
      "learning_rate": 7.192913798973089e-07,
      "loss": 0.5465,
      "step": 5856
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6787502776032842,
      "learning_rate": 7.174752426918041e-07,
      "loss": 0.5558,
      "step": 5857
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6171280951502537,
      "learning_rate": 7.156613158508618e-07,
      "loss": 0.5698,
      "step": 5858
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6103633741825153,
      "learning_rate": 7.138495998064099e-07,
      "loss": 0.5591,
      "step": 5859
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.738189172078162,
      "learning_rate": 7.120400949898576e-07,
      "loss": 0.5617,
      "step": 5860
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6698384629446569,
      "learning_rate": 7.102328018320859e-07,
      "loss": 0.5572,
      "step": 5861
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.630569827144095,
      "learning_rate": 7.084277207634494e-07,
      "loss": 0.5566,
      "step": 5862
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6154109017855667,
      "learning_rate": 7.066248522137786e-07,
      "loss": 0.5592,
      "step": 5863
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5826097009099569,
      "learning_rate": 7.048241966123703e-07,
      "loss": 0.5642,
      "step": 5864
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.600029781900714,
      "learning_rate": 7.030257543879992e-07,
      "loss": 0.5776,
      "step": 5865
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6773417797463112,
      "learning_rate": 7.012295259689161e-07,
      "loss": 0.5498,
      "step": 5866
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.742987400953712,
      "learning_rate": 6.994355117828366e-07,
      "loss": 0.5693,
      "step": 5867
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6818933352135397,
      "learning_rate": 6.976437122569557e-07,
      "loss": 0.5689,
      "step": 5868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.629189249745075,
      "learning_rate": 6.958541278179365e-07,
      "loss": 0.5591,
      "step": 5869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6108616027978169,
      "learning_rate": 6.94066758891917e-07,
      "loss": 0.5398,
      "step": 5870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6289607711351594,
      "learning_rate": 6.922816059045112e-07,
      "loss": 0.5391,
      "step": 5871
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.616840944597574,
      "learning_rate": 6.904986692807958e-07,
      "loss": 0.5689,
      "step": 5872
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5917999869477919,
      "learning_rate": 6.887179494453289e-07,
      "loss": 0.5439,
      "step": 5873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6753832168438675,
      "learning_rate": 6.86939446822138e-07,
      "loss": 0.5574,
      "step": 5874
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6861127091052931,
      "learning_rate": 6.851631618347198e-07,
      "loss": 0.5804,
      "step": 5875
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5848366962467494,
      "learning_rate": 6.833890949060462e-07,
      "loss": 0.5585,
      "step": 5876
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.597636642046953,
      "learning_rate": 6.816172464585614e-07,
      "loss": 0.5514,
      "step": 5877
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6256172974352764,
      "learning_rate": 6.798476169141766e-07,
      "loss": 0.5659,
      "step": 5878
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6567064331413895,
      "learning_rate": 6.780802066942816e-07,
      "loss": 0.5687,
      "step": 5879
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5791500730453193,
      "learning_rate": 6.763150162197274e-07,
      "loss": 0.5613,
      "step": 5880
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6325547973515608,
      "learning_rate": 6.745520459108523e-07,
      "loss": 0.5694,
      "step": 5881
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6814573781630953,
      "learning_rate": 6.727912961874505e-07,
      "loss": 0.6047,
      "step": 5882
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6232449635239774,
      "learning_rate": 6.710327674687944e-07,
      "loss": 0.561,
      "step": 5883
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6041767118650793,
      "learning_rate": 6.692764601736268e-07,
      "loss": 0.5559,
      "step": 5884
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6653381220820074,
      "learning_rate": 6.67522374720162e-07,
      "loss": 0.5717,
      "step": 5885
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6025046841639842,
      "learning_rate": 6.657705115260859e-07,
      "loss": 0.5763,
      "step": 5886
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1826239987854161,
      "learning_rate": 6.640208710085517e-07,
      "loss": 0.5525,
      "step": 5887
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6877036743743561,
      "learning_rate": 6.622734535841868e-07,
      "loss": 0.5725,
      "step": 5888
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5516600304672328,
      "learning_rate": 6.605282596690888e-07,
      "loss": 0.5678,
      "step": 5889
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6744895926059957,
      "learning_rate": 6.587852896788227e-07,
      "loss": 0.5657,
      "step": 5890
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6472409367417125,
      "learning_rate": 6.57044544028429e-07,
      "loss": 0.5717,
      "step": 5891
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6551058866217289,
      "learning_rate": 6.553060231324137e-07,
      "loss": 0.5575,
      "step": 5892
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.62715011404072,
      "learning_rate": 6.53569727404757e-07,
      "loss": 0.576,
      "step": 5893
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6059397059965617,
      "learning_rate": 6.518356572589079e-07,
      "loss": 0.5549,
      "step": 5894
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.646766042460902,
      "learning_rate": 6.50103813107782e-07,
      "loss": 0.5562,
      "step": 5895
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7650035007925183,
      "learning_rate": 6.48374195363769e-07,
      "loss": 0.5915,
      "step": 5896
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6359527646940606,
      "learning_rate": 6.466468044387308e-07,
      "loss": 0.5805,
      "step": 5897
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.596838552470831,
      "learning_rate": 6.449216407439906e-07,
      "loss": 0.5724,
      "step": 5898
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6439124914092829,
      "learning_rate": 6.431987046903487e-07,
      "loss": 0.5673,
      "step": 5899
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6005980221817196,
      "learning_rate": 6.414779966880714e-07,
      "loss": 0.5525,
      "step": 5900
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6765085352842591,
      "learning_rate": 6.397595171468984e-07,
      "loss": 0.5755,
      "step": 5901
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6707250095151333,
      "learning_rate": 6.380432664760327e-07,
      "loss": 0.5526,
      "step": 5902
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7012601518991474,
      "learning_rate": 6.363292450841485e-07,
      "loss": 0.5948,
      "step": 5903
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6243009855764597,
      "learning_rate": 6.346174533793948e-07,
      "loss": 0.5662,
      "step": 5904
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5740720855386763,
      "learning_rate": 6.329078917693832e-07,
      "loss": 0.5551,
      "step": 5905
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6614629241805857,
      "learning_rate": 6.312005606611949e-07,
      "loss": 0.5741,
      "step": 5906
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.58560755449029,
      "learning_rate": 6.294954604613824e-07,
      "loss": 0.5472,
      "step": 5907
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6854071291614878,
      "learning_rate": 6.277925915759664e-07,
      "loss": 0.5734,
      "step": 5908
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.608056983370541,
      "learning_rate": 6.260919544104371e-07,
      "loss": 0.5636,
      "step": 5909
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.74397099904607,
      "learning_rate": 6.243935493697495e-07,
      "loss": 0.5917,
      "step": 5910
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5932596697056289,
      "learning_rate": 6.226973768583278e-07,
      "loss": 0.5631,
      "step": 5911
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6321562308492733,
      "learning_rate": 6.210034372800733e-07,
      "loss": 0.5545,
      "step": 5912
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.615556735755657,
      "learning_rate": 6.193117310383412e-07,
      "loss": 0.5538,
      "step": 5913
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5815712598565139,
      "learning_rate": 6.176222585359681e-07,
      "loss": 0.556,
      "step": 5914
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6042441645397406,
      "learning_rate": 6.15935020175249e-07,
      "loss": 0.5655,
      "step": 5915
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6276337475980469,
      "learning_rate": 6.142500163579534e-07,
      "loss": 0.5653,
      "step": 5916
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.623130598400157,
      "learning_rate": 6.12567247485315e-07,
      "loss": 0.5543,
      "step": 5917
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5806806630963087,
      "learning_rate": 6.108867139580365e-07,
      "loss": 0.5741,
      "step": 5918
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6396600338228078,
      "learning_rate": 6.09208416176288e-07,
      "loss": 0.5656,
      "step": 5919
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5731533243270515,
      "learning_rate": 6.075323545397093e-07,
      "loss": 0.5695,
      "step": 5920
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6598942520926525,
      "learning_rate": 6.058585294474028e-07,
      "loss": 0.58,
      "step": 5921
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5982951734651193,
      "learning_rate": 6.04186941297944e-07,
      "loss": 0.5811,
      "step": 5922
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6254593923420224,
      "learning_rate": 6.02517590489372e-07,
      "loss": 0.5605,
      "step": 5923
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6202062089784476,
      "learning_rate": 6.00850477419197e-07,
      "loss": 0.5414,
      "step": 5924
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6364403720242158,
      "learning_rate": 5.991856024843923e-07,
      "loss": 0.5221,
      "step": 5925
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6845420651962707,
      "learning_rate": 5.975229660813964e-07,
      "loss": 0.6025,
      "step": 5926
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6637835027801413,
      "learning_rate": 5.958625686061215e-07,
      "loss": 0.5737,
      "step": 5927
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5915096529925457,
      "learning_rate": 5.942044104539412e-07,
      "loss": 0.5746,
      "step": 5928
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6192117849906789,
      "learning_rate": 5.925484920197022e-07,
      "loss": 0.5535,
      "step": 5929
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6654005115955961,
      "learning_rate": 5.908948136977078e-07,
      "loss": 0.5411,
      "step": 5930
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6279676343385545,
      "learning_rate": 5.892433758817362e-07,
      "loss": 0.5677,
      "step": 5931
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6892633352367072,
      "learning_rate": 5.875941789650319e-07,
      "loss": 0.5645,
      "step": 5932
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5889565697256114,
      "learning_rate": 5.859472233402985e-07,
      "loss": 0.5686,
      "step": 5933
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.637966402645585,
      "learning_rate": 5.843025093997134e-07,
      "loss": 0.5629,
      "step": 5934
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6527038053456194,
      "learning_rate": 5.826600375349201e-07,
      "loss": 0.5638,
      "step": 5935
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6308420843024483,
      "learning_rate": 5.8101980813702e-07,
      "loss": 0.5345,
      "step": 5936
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6522536504815242,
      "learning_rate": 5.793818215965918e-07,
      "loss": 0.5746,
      "step": 5937
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5720708467194022,
      "learning_rate": 5.777460783036714e-07,
      "loss": 0.5554,
      "step": 5938
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6803948316492731,
      "learning_rate": 5.76112578647764e-07,
      "loss": 0.58,
      "step": 5939
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.612053962858455,
      "learning_rate": 5.744813230178415e-07,
      "loss": 0.5615,
      "step": 5940
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6177579039986753,
      "learning_rate": 5.728523118023388e-07,
      "loss": 0.5572,
      "step": 5941
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6143070409516856,
      "learning_rate": 5.71225545389158e-07,
      "loss": 0.5642,
      "step": 5942
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6134418709729115,
      "learning_rate": 5.696010241656691e-07,
      "loss": 0.5682,
      "step": 5943
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6329347320606192,
      "learning_rate": 5.679787485187005e-07,
      "loss": 0.562,
      "step": 5944
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6294137947101334,
      "learning_rate": 5.663587188345521e-07,
      "loss": 0.5482,
      "step": 5945
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6331017269686485,
      "learning_rate": 5.647409354989874e-07,
      "loss": 0.5552,
      "step": 5946
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5793493190906029,
      "learning_rate": 5.631253988972362e-07,
      "loss": 0.5502,
      "step": 5947
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6566512174074521,
      "learning_rate": 5.615121094139897e-07,
      "loss": 0.5435,
      "step": 5948
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5798065921771262,
      "learning_rate": 5.599010674334049e-07,
      "loss": 0.5402,
      "step": 5949
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6547110002427718,
      "learning_rate": 5.582922733391071e-07,
      "loss": 0.5561,
      "step": 5950
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6427863776425489,
      "learning_rate": 5.566857275141824e-07,
      "loss": 0.5912,
      "step": 5951
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6440768228470264,
      "learning_rate": 5.550814303411856e-07,
      "loss": 0.5718,
      "step": 5952
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.647870669270731,
      "learning_rate": 5.534793822021311e-07,
      "loss": 0.558,
      "step": 5953
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6000327293997925,
      "learning_rate": 5.518795834785018e-07,
      "loss": 0.5529,
      "step": 5954
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6338003900952502,
      "learning_rate": 5.502820345512438e-07,
      "loss": 0.586,
      "step": 5955
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6177413219246088,
      "learning_rate": 5.486867358007642e-07,
      "loss": 0.5561,
      "step": 5956
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5850855209747946,
      "learning_rate": 5.4709368760694e-07,
      "loss": 0.5671,
      "step": 5957
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5777922552312327,
      "learning_rate": 5.455028903491099e-07,
      "loss": 0.5709,
      "step": 5958
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6788131454112817,
      "learning_rate": 5.439143444060746e-07,
      "loss": 0.5833,
      "step": 5959
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6162214503814999,
      "learning_rate": 5.423280501561013e-07,
      "loss": 0.5779,
      "step": 5960
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6290894377654198,
      "learning_rate": 5.407440079769188e-07,
      "loss": 0.5538,
      "step": 5961
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6664819096445186,
      "learning_rate": 5.391622182457212e-07,
      "loss": 0.5751,
      "step": 5962
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5732087250849585,
      "learning_rate": 5.375826813391682e-07,
      "loss": 0.5725,
      "step": 5963
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6491754776930874,
      "learning_rate": 5.360053976333779e-07,
      "loss": 0.556,
      "step": 5964
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6557270144261068,
      "learning_rate": 5.344303675039353e-07,
      "loss": 0.5448,
      "step": 5965
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6455274807668472,
      "learning_rate": 5.328575913258893e-07,
      "loss": 0.5887,
      "step": 5966
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6142538593612178,
      "learning_rate": 5.312870694737516e-07,
      "loss": 0.5574,
      "step": 5967
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6662975532554798,
      "learning_rate": 5.297188023214961e-07,
      "loss": 0.5368,
      "step": 5968
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.633089021295785,
      "learning_rate": 5.281527902425565e-07,
      "loss": 0.5368,
      "step": 5969
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6165208864509769,
      "learning_rate": 5.265890336098389e-07,
      "loss": 0.5596,
      "step": 5970
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6041004750575961,
      "learning_rate": 5.250275327957033e-07,
      "loss": 0.5724,
      "step": 5971
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6398818689235158,
      "learning_rate": 5.234682881719766e-07,
      "loss": 0.5458,
      "step": 5972
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5847193572639499,
      "learning_rate": 5.219113001099474e-07,
      "loss": 0.5604,
      "step": 5973
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5971377505297237,
      "learning_rate": 5.20356568980368e-07,
      "loss": 0.5411,
      "step": 5974
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6170854090922476,
      "learning_rate": 5.188040951534534e-07,
      "loss": 0.5421,
      "step": 5975
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6273574900619718,
      "learning_rate": 5.172538789988779e-07,
      "loss": 0.577,
      "step": 5976
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6465398837743946,
      "learning_rate": 5.157059208857817e-07,
      "loss": 0.5784,
      "step": 5977
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6792096627130513,
      "learning_rate": 5.14160221182769e-07,
      "loss": 0.5591,
      "step": 5978
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5924522609604332,
      "learning_rate": 5.126167802578985e-07,
      "loss": 0.5818,
      "step": 5979
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6426932514892219,
      "learning_rate": 5.110755984786997e-07,
      "loss": 0.5626,
      "step": 5980
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6106286409001912,
      "learning_rate": 5.095366762121601e-07,
      "loss": 0.5601,
      "step": 5981
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6672347200134011,
      "learning_rate": 5.080000138247265e-07,
      "loss": 0.577,
      "step": 5982
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5818976231789096,
      "learning_rate": 5.064656116823141e-07,
      "loss": 0.5452,
      "step": 5983
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6727891499854904,
      "learning_rate": 5.049334701502939e-07,
      "loss": 0.5665,
      "step": 5984
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5947201584621304,
      "learning_rate": 5.03403589593503e-07,
      "loss": 0.5552,
      "step": 5985
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6517646275111243,
      "learning_rate": 5.018759703762377e-07,
      "loss": 0.5635,
      "step": 5986
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6369898857072159,
      "learning_rate": 5.003506128622537e-07,
      "loss": 0.5721,
      "step": 5987
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6113039086667983,
      "learning_rate": 4.988275174147739e-07,
      "loss": 0.5874,
      "step": 5988
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6303568360881779,
      "learning_rate": 4.97306684396478e-07,
      "loss": 0.5638,
      "step": 5989
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6761227515535676,
      "learning_rate": 4.957881141695098e-07,
      "loss": 0.5547,
      "step": 5990
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7021514935533941,
      "learning_rate": 4.942718070954721e-07,
      "loss": 0.5486,
      "step": 5991
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5554428645962342,
      "learning_rate": 4.927577635354253e-07,
      "loss": 0.5406,
      "step": 5992
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6459073394002215,
      "learning_rate": 4.912459838499029e-07,
      "loss": 0.5606,
      "step": 5993
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6876376884234456,
      "learning_rate": 4.897364683988859e-07,
      "loss": 0.5907,
      "step": 5994
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5659024077322587,
      "learning_rate": 4.882292175418246e-07,
      "loss": 0.5777,
      "step": 5995
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6625165918029462,
      "learning_rate": 4.867242316376241e-07,
      "loss": 0.5535,
      "step": 5996
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5852714081986724,
      "learning_rate": 4.852215110446556e-07,
      "loss": 0.5307,
      "step": 5997
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7132919322609079,
      "learning_rate": 4.837210561207484e-07,
      "loss": 0.6019,
      "step": 5998
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5968722332978612,
      "learning_rate": 4.82222867223191e-07,
      "loss": 0.5557,
      "step": 5999
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.692378606456316,
      "learning_rate": 4.807269447087348e-07,
      "loss": 0.564,
      "step": 6000
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5864131009404661,
      "learning_rate": 4.792332889335915e-07,
      "loss": 0.5541,
      "step": 6001
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6431542135118415,
      "learning_rate": 4.777419002534289e-07,
      "loss": 0.5409,
      "step": 6002
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6325404595858037,
      "learning_rate": 4.762527790233817e-07,
      "loss": 0.565,
      "step": 6003
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5978511107858264,
      "learning_rate": 4.7476592559803744e-07,
      "loss": 0.5517,
      "step": 6004
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6099502209786996,
      "learning_rate": 4.7328134033144955e-07,
      "loss": 0.5929,
      "step": 6005
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6509240705724234,
      "learning_rate": 4.717990235771297e-07,
      "loss": 0.5733,
      "step": 6006
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6660559147387477,
      "learning_rate": 4.7031897568804553e-07,
      "loss": 0.5593,
      "step": 6007
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6416514991803511,
      "learning_rate": 4.6884119701662954e-07,
      "loss": 0.5701,
      "step": 6008
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.784201651466995,
      "learning_rate": 4.6736568791477367e-07,
      "loss": 0.6192,
      "step": 6009
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6274507726058116,
      "learning_rate": 4.6589244873382454e-07,
      "loss": 0.5793,
      "step": 6010
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6047728770725415,
      "learning_rate": 4.6442147982459276e-07,
      "loss": 0.5658,
      "step": 6011
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5844106404803688,
      "learning_rate": 4.6295278153734693e-07,
      "loss": 0.5571,
      "step": 6012
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6576881865770189,
      "learning_rate": 4.6148635422181733e-07,
      "loss": 0.5637,
      "step": 6013
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5919521379329862,
      "learning_rate": 4.6002219822718794e-07,
      "loss": 0.5542,
      "step": 6014
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6868123046410411,
      "learning_rate": 4.585603139021044e-07,
      "loss": 0.5645,
      "step": 6015
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6218303137617116,
      "learning_rate": 4.5710070159467604e-07,
      "loss": 0.5535,
      "step": 6016
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.564729483163916,
      "learning_rate": 4.556433616524636e-07,
      "loss": 0.5807,
      "step": 6017
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5880756608234832,
      "learning_rate": 4.54188294422494e-07,
      "loss": 0.5373,
      "step": 6018
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6953069535173934,
      "learning_rate": 4.527355002512457e-07,
      "loss": 0.5671,
      "step": 6019
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6043295272095836,
      "learning_rate": 4.512849794846608e-07,
      "loss": 0.5746,
      "step": 6020
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5984757403522838,
      "learning_rate": 4.498367324681407e-07,
      "loss": 0.551,
      "step": 6021
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.625342382290063,
      "learning_rate": 4.48390759546542e-07,
      "loss": 0.5645,
      "step": 6022
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6615610831488415,
      "learning_rate": 4.4694706106418016e-07,
      "loss": 0.5524,
      "step": 6023
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6388675812413914,
      "learning_rate": 4.455056373648325e-07,
      "loss": 0.5616,
      "step": 6024
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7655857541841007,
      "learning_rate": 4.440664887917301e-07,
      "loss": 0.5544,
      "step": 6025
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6435291907055317,
      "learning_rate": 4.426296156875676e-07,
      "loss": 0.5574,
      "step": 6026
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5693353772309623,
      "learning_rate": 4.411950183944902e-07,
      "loss": 0.5532,
      "step": 6027
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7048911262421,
      "learning_rate": 4.3976269725411026e-07,
      "loss": 0.5663,
      "step": 6028
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6195101281923298,
      "learning_rate": 4.3833265260749157e-07,
      "loss": 0.5454,
      "step": 6029
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.597832673990884,
      "learning_rate": 4.3690488479515735e-07,
      "loss": 0.5677,
      "step": 6030
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6192401095475639,
      "learning_rate": 4.3547939415708893e-07,
      "loss": 0.5905,
      "step": 6031
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6668390805946695,
      "learning_rate": 4.3405618103272816e-07,
      "loss": 0.5559,
      "step": 6032
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5928995499614075,
      "learning_rate": 4.326352457609695e-07,
      "loss": 0.5447,
      "step": 6033
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6240846125565387,
      "learning_rate": 4.312165886801678e-07,
      "loss": 0.554,
      "step": 6034
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.667691913072858,
      "learning_rate": 4.298002101281362e-07,
      "loss": 0.5938,
      "step": 6035
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6782943752831371,
      "learning_rate": 4.2838611044214496e-07,
      "loss": 0.5803,
      "step": 6036
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6481421030469308,
      "learning_rate": 4.269742899589191e-07,
      "loss": 0.5562,
      "step": 6037
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6305875315470949,
      "learning_rate": 4.2556474901464195e-07,
      "loss": 0.5795,
      "step": 6038
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6280801378384254,
      "learning_rate": 4.241574879449595e-07,
      "loss": 0.5715,
      "step": 6039
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.660932398574651,
      "learning_rate": 4.2275250708496475e-07,
      "loss": 0.5502,
      "step": 6040
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6223215203196261,
      "learning_rate": 4.2134980676921787e-07,
      "loss": 0.5602,
      "step": 6041
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6239996865246256,
      "learning_rate": 4.199493873317273e-07,
      "loss": 0.5819,
      "step": 6042
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6224700092907504,
      "learning_rate": 4.18551249105964e-07,
      "loss": 0.5585,
      "step": 6043
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6167722846862461,
      "learning_rate": 4.1715539242485613e-07,
      "loss": 0.5324,
      "step": 6044
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5898317080449677,
      "learning_rate": 4.157618176207834e-07,
      "loss": 0.5546,
      "step": 6045
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6714172750497487,
      "learning_rate": 4.1437052502558693e-07,
      "loss": 0.5979,
      "step": 6046
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6011489964752297,
      "learning_rate": 4.1298151497056404e-07,
      "loss": 0.5601,
      "step": 6047
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6816801598223506,
      "learning_rate": 4.1159478778646347e-07,
      "loss": 0.594,
      "step": 6048
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6614819719867496,
      "learning_rate": 4.102103438034988e-07,
      "loss": 0.5728,
      "step": 6049
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6173560853412534,
      "learning_rate": 4.088281833513297e-07,
      "loss": 0.5625,
      "step": 6050
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6426759970179521,
      "learning_rate": 4.07448306759084e-07,
      "loss": 0.5693,
      "step": 6051
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6219076633418195,
      "learning_rate": 4.0607071435533554e-07,
      "loss": 0.5726,
      "step": 6052
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6036657332380749,
      "learning_rate": 4.0469540646811856e-07,
      "loss": 0.5715,
      "step": 6053
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.649419741692739,
      "learning_rate": 4.0332238342492223e-07,
      "loss": 0.5616,
      "step": 6054
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6492597533547066,
      "learning_rate": 4.019516455526928e-07,
      "loss": 0.5739,
      "step": 6055
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6386743555441282,
      "learning_rate": 4.0058319317783366e-07,
      "loss": 0.5666,
      "step": 6056
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6857862827462233,
      "learning_rate": 3.9921702662619966e-07,
      "loss": 0.5797,
      "step": 6057
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6055545676990367,
      "learning_rate": 3.97853146223105e-07,
      "loss": 0.5646,
      "step": 6058
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5845378689938413,
      "learning_rate": 3.964915522933188e-07,
      "loss": 0.5645,
      "step": 6059
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6182014846728426,
      "learning_rate": 3.9513224516106507e-07,
      "loss": 0.5707,
      "step": 6060
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5685474899960898,
      "learning_rate": 3.937752251500204e-07,
      "loss": 0.5583,
      "step": 6061
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6945058949031052,
      "learning_rate": 3.9242049258332306e-07,
      "loss": 0.556,
      "step": 6062
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.588452669845098,
      "learning_rate": 3.910680477835627e-07,
      "loss": 0.559,
      "step": 6063
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6219300679843293,
      "learning_rate": 3.897178910727861e-07,
      "loss": 0.5593,
      "step": 6064
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6321928165570488,
      "learning_rate": 3.883700227724907e-07,
      "loss": 0.5646,
      "step": 6065
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5946853935076837,
      "learning_rate": 3.870244432036352e-07,
      "loss": 0.5766,
      "step": 6066
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5963876399278774,
      "learning_rate": 3.8568115268663e-07,
      "loss": 0.5431,
      "step": 6067
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7242779013813054,
      "learning_rate": 3.843401515413392e-07,
      "loss": 0.596,
      "step": 6068
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5881716649923076,
      "learning_rate": 3.8300144008708516e-07,
      "loss": 0.538,
      "step": 6069
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6856835110135591,
      "learning_rate": 3.8166501864264404e-07,
      "loss": 0.5731,
      "step": 6070
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5863510129859817,
      "learning_rate": 3.8033088752624347e-07,
      "loss": 0.5578,
      "step": 6071
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5661101320561338,
      "learning_rate": 3.7899904705556936e-07,
      "loss": 0.534,
      "step": 6072
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7340797292531602,
      "learning_rate": 3.7766949754775905e-07,
      "loss": 0.6105,
      "step": 6073
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6618860520860176,
      "learning_rate": 3.763422393194105e-07,
      "loss": 0.5457,
      "step": 6074
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7339544847559629,
      "learning_rate": 3.7501727268656974e-07,
      "loss": 0.6033,
      "step": 6075
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5794975906368092,
      "learning_rate": 3.736945979647377e-07,
      "loss": 0.5578,
      "step": 6076
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5951583405022529,
      "learning_rate": 3.723742154688714e-07,
      "loss": 0.5511,
      "step": 6077
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5927556713840468,
      "learning_rate": 3.7105612551338377e-07,
      "loss": 0.602,
      "step": 6078
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5763633568019499,
      "learning_rate": 3.6974032841213923e-07,
      "loss": 0.5462,
      "step": 6079
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6523527279227347,
      "learning_rate": 3.68426824478455e-07,
      "loss": 0.5729,
      "step": 6080
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6139778302120502,
      "learning_rate": 3.671156140251053e-07,
      "loss": 0.5578,
      "step": 6081
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7030925393737769,
      "learning_rate": 3.658066973643171e-07,
      "loss": 0.6024,
      "step": 6082
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5900270825053089,
      "learning_rate": 3.645000748077709e-07,
      "loss": 0.562,
      "step": 6083
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6986526031803489,
      "learning_rate": 3.631957466666003e-07,
      "loss": 0.6121,
      "step": 6084
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6362555309051862,
      "learning_rate": 3.6189371325139444e-07,
      "loss": 0.5738,
      "step": 6085
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6664741345862839,
      "learning_rate": 3.6059397487219315e-07,
      "loss": 0.5787,
      "step": 6086
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.577599193298096,
      "learning_rate": 3.5929653183849444e-07,
      "loss": 0.5816,
      "step": 6087
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6365759945076644,
      "learning_rate": 3.5800138445924336e-07,
      "loss": 0.5702,
      "step": 6088
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6308238709055012,
      "learning_rate": 3.5670853304284324e-07,
      "loss": 0.5685,
      "step": 6089
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6492916450021303,
      "learning_rate": 3.5541797789715115e-07,
      "loss": 0.5538,
      "step": 6090
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6540255814542403,
      "learning_rate": 3.5412971932947236e-07,
      "loss": 0.5814,
      "step": 6091
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6218138382516091,
      "learning_rate": 3.528437576465693e-07,
      "loss": 0.5557,
      "step": 6092
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6224221351119812,
      "learning_rate": 3.5156009315465813e-07,
      "loss": 0.6022,
      "step": 6093
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6056434413140784,
      "learning_rate": 3.5027872615940426e-07,
      "loss": 0.5381,
      "step": 6094
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5568252218662235,
      "learning_rate": 3.489996569659293e-07,
      "loss": 0.5665,
      "step": 6095
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6110137358859178,
      "learning_rate": 3.477228858788051e-07,
      "loss": 0.5451,
      "step": 6096
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6392956605902856,
      "learning_rate": 3.464484132020607e-07,
      "loss": 0.5706,
      "step": 6097
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5807005630260521,
      "learning_rate": 3.451762392391733e-07,
      "loss": 0.5473,
      "step": 6098
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6150214193926506,
      "learning_rate": 3.439063642930729e-07,
      "loss": 0.545,
      "step": 6099
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5993291982345842,
      "learning_rate": 3.426387886661442e-07,
      "loss": 0.5498,
      "step": 6100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6156438576975234,
      "learning_rate": 3.413735126602247e-07,
      "loss": 0.5759,
      "step": 6101
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6298578519250823,
      "learning_rate": 3.401105365766033e-07,
      "loss": 0.5759,
      "step": 6102
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.603691440067513,
      "learning_rate": 3.388498607160207e-07,
      "loss": 0.5624,
      "step": 6103
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6860260779478075,
      "learning_rate": 3.375914853786677e-07,
      "loss": 0.5794,
      "step": 6104
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5912296237957305,
      "learning_rate": 3.3633541086419477e-07,
      "loss": 0.585,
      "step": 6105
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6936197132676248,
      "learning_rate": 3.35081637471697e-07,
      "loss": 0.5701,
      "step": 6106
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6082633705982303,
      "learning_rate": 3.338301654997245e-07,
      "loss": 0.5788,
      "step": 6107
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6384972232222889,
      "learning_rate": 3.3258099524627884e-07,
      "loss": 0.5594,
      "step": 6108
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6215053575692648,
      "learning_rate": 3.313341270088144e-07,
      "loss": 0.5224,
      "step": 6109
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.623030580403707,
      "learning_rate": 3.3008956108423696e-07,
      "loss": 0.5702,
      "step": 6110
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6134185419741451,
      "learning_rate": 3.2884729776890276e-07,
      "loss": 0.5837,
      "step": 6111
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.628010650970135,
      "learning_rate": 3.276073373586208e-07,
      "loss": 0.5752,
      "step": 6112
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6121799819267449,
      "learning_rate": 3.263696801486538e-07,
      "loss": 0.5518,
      "step": 6113
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6082032978791241,
      "learning_rate": 3.2513432643371144e-07,
      "loss": 0.5775,
      "step": 6114
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6470098785068275,
      "learning_rate": 3.2390127650795857e-07,
      "loss": 0.556,
      "step": 6115
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6513734482128778,
      "learning_rate": 3.226705306650113e-07,
      "loss": 0.5916,
      "step": 6116
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6335319328383059,
      "learning_rate": 3.214420891979353e-07,
      "loss": 0.5552,
      "step": 6117
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6063347309271515,
      "learning_rate": 3.2021595239924874e-07,
      "loss": 0.5812,
      "step": 6118
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.862395299316082,
      "learning_rate": 3.1899212056091923e-07,
      "loss": 0.5366,
      "step": 6119
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6405697751098725,
      "learning_rate": 3.1777059397436693e-07,
      "loss": 0.5685,
      "step": 6120
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6344796465470173,
      "learning_rate": 3.165513729304648e-07,
      "loss": 0.559,
      "step": 6121
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6350949618778199,
      "learning_rate": 3.15334457719535e-07,
      "loss": 0.575,
      "step": 6122
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.610212552296033,
      "learning_rate": 3.141198486313479e-07,
      "loss": 0.5383,
      "step": 6123
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6148605939572029,
      "learning_rate": 3.129075459551301e-07,
      "loss": 0.558,
      "step": 6124
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6229379654241395,
      "learning_rate": 3.1169754997955715e-07,
      "loss": 0.5408,
      "step": 6125
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6232769542181517,
      "learning_rate": 3.1048986099275204e-07,
      "loss": 0.5554,
      "step": 6126
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6781061879166492,
      "learning_rate": 3.092844792822902e-07,
      "loss": 0.5532,
      "step": 6127
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6480508948901857,
      "learning_rate": 3.0808140513520213e-07,
      "loss": 0.5753,
      "step": 6128
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6373592398620562,
      "learning_rate": 3.06880638837963e-07,
      "loss": 0.5527,
      "step": 6129
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6159308845142402,
      "learning_rate": 3.056821806765009e-07,
      "loss": 0.579,
      "step": 6130
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6459361788283123,
      "learning_rate": 3.04486030936193e-07,
      "loss": 0.5648,
      "step": 6131
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6001863205641447,
      "learning_rate": 3.032921899018681e-07,
      "loss": 0.5627,
      "step": 6132
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6987197726074137,
      "learning_rate": 3.021006578578067e-07,
      "loss": 0.5974,
      "step": 6133
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6018314574675754,
      "learning_rate": 3.009114350877351e-07,
      "loss": 0.59,
      "step": 6134
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6002584724801424,
      "learning_rate": 2.997245218748335e-07,
      "loss": 0.5781,
      "step": 6135
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6019749161627915,
      "learning_rate": 2.985399185017324e-07,
      "loss": 0.5737,
      "step": 6136
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.694249930460923,
      "learning_rate": 2.9735762525050727e-07,
      "loss": 0.5793,
      "step": 6137
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.59647217918814,
      "learning_rate": 2.9617764240269076e-07,
      "loss": 0.5588,
      "step": 6138
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5379812851238284,
      "learning_rate": 2.94999970239257e-07,
      "loss": 0.5612,
      "step": 6139
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5839618681100163,
      "learning_rate": 2.938246090406405e-07,
      "loss": 0.5366,
      "step": 6140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6275492388250422,
      "learning_rate": 2.9265155908671516e-07,
      "loss": 0.571,
      "step": 6141
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6583309348507442,
      "learning_rate": 2.914808206568098e-07,
      "loss": 0.5529,
      "step": 6142
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6851103002124933,
      "learning_rate": 2.903123940297015e-07,
      "loss": 0.5922,
      "step": 6143
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6418345747155152,
      "learning_rate": 2.891462794836186e-07,
      "loss": 0.559,
      "step": 6144
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5350670141155172,
      "learning_rate": 2.879824772962381e-07,
      "loss": 0.5819,
      "step": 6145
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6139036962450489,
      "learning_rate": 2.8682098774468257e-07,
      "loss": 0.5512,
      "step": 6146
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.675412779299054,
      "learning_rate": 2.856618111055298e-07,
      "loss": 0.5744,
      "step": 6147
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6525192492249596,
      "learning_rate": 2.845049476548045e-07,
      "loss": 0.5861,
      "step": 6148
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7110303757279431,
      "learning_rate": 2.8335039766797745e-07,
      "loss": 0.5691,
      "step": 6149
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6457822094081704,
      "learning_rate": 2.8219816141997315e-07,
      "loss": 0.5474,
      "step": 6150
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6934234048957526,
      "learning_rate": 2.810482391851643e-07,
      "loss": 0.6152,
      "step": 6151
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6648150137077892,
      "learning_rate": 2.799006312373698e-07,
      "loss": 0.5467,
      "step": 6152
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6415333117570596,
      "learning_rate": 2.787553378498609e-07,
      "loss": 0.573,
      "step": 6153
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6579209798406831,
      "learning_rate": 2.776123592953539e-07,
      "loss": 0.5807,
      "step": 6154
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5731358943479724,
      "learning_rate": 2.7647169584601677e-07,
      "loss": 0.5532,
      "step": 6155
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7189677031110981,
      "learning_rate": 2.7533334777346764e-07,
      "loss": 0.5676,
      "step": 6156
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5897019359784802,
      "learning_rate": 2.741973153487687e-07,
      "loss": 0.5385,
      "step": 6157
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6788019728334652,
      "learning_rate": 2.730635988424335e-07,
      "loss": 0.6025,
      "step": 6158
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6240992866237427,
      "learning_rate": 2.71932198524425e-07,
      "loss": 0.5475,
      "step": 6159
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6488067892399508,
      "learning_rate": 2.708031146641521e-07,
      "loss": 0.5341,
      "step": 6160
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.655277607084754,
      "learning_rate": 2.6967634753047424e-07,
      "loss": 0.5685,
      "step": 6161
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6032409144830617,
      "learning_rate": 2.6855189739169673e-07,
      "loss": 0.5708,
      "step": 6162
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6198495907802128,
      "learning_rate": 2.674297645155788e-07,
      "loss": 0.5972,
      "step": 6163
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5811158999466862,
      "learning_rate": 2.6630994916932107e-07,
      "loss": 0.5892,
      "step": 6164
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6012425340123563,
      "learning_rate": 2.6519245161957364e-07,
      "loss": 0.5499,
      "step": 6165
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5443451446973046,
      "learning_rate": 2.640772721324392e-07,
      "loss": 0.5752,
      "step": 6166
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6502071582289858,
      "learning_rate": 2.62964410973463e-07,
      "loss": 0.5667,
      "step": 6167
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6489415820724161,
      "learning_rate": 2.618538684076444e-07,
      "loss": 0.5653,
      "step": 6168
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6424592259852037,
      "learning_rate": 2.6074564469942277e-07,
      "loss": 0.5747,
      "step": 6169
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6653004197196529,
      "learning_rate": 2.5963974011269156e-07,
      "loss": 0.5652,
      "step": 6170
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6260708832330857,
      "learning_rate": 2.585361549107901e-07,
      "loss": 0.5708,
      "step": 6171
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5887062298815533,
      "learning_rate": 2.5743488935650483e-07,
      "loss": 0.5602,
      "step": 6172
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6160963467920882,
      "learning_rate": 2.563359437120694e-07,
      "loss": 0.5436,
      "step": 6173
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6698477519485784,
      "learning_rate": 2.552393182391677e-07,
      "loss": 0.5903,
      "step": 6174
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6670241593321516,
      "learning_rate": 2.541450131989287e-07,
      "loss": 0.5651,
      "step": 6175
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6229618631939087,
      "learning_rate": 2.530530288519284e-07,
      "loss": 0.5748,
      "step": 6176
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6125237656622797,
      "learning_rate": 2.519633654581921e-07,
      "loss": 0.5713,
      "step": 6177
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6537883424845718,
      "learning_rate": 2.5087602327719117e-07,
      "loss": 0.5525,
      "step": 6178
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5430692804469428,
      "learning_rate": 2.497910025678463e-07,
      "loss": 0.554,
      "step": 6179
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6969420188965462,
      "learning_rate": 2.487083035885218e-07,
      "loss": 0.5383,
      "step": 6180
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6462882571374368,
      "learning_rate": 2.476279265970316e-07,
      "loss": 0.5649,
      "step": 6181
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6449241324454414,
      "learning_rate": 2.465498718506354e-07,
      "loss": 0.5397,
      "step": 6182
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6492550075413082,
      "learning_rate": 2.4547413960604336e-07,
      "loss": 0.5731,
      "step": 6183
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5979824903200327,
      "learning_rate": 2.4440073011940845e-07,
      "loss": 0.554,
      "step": 6184
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6473365939563508,
      "learning_rate": 2.433296436463306e-07,
      "loss": 0.5337,
      "step": 6185
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6738227639831645,
      "learning_rate": 2.4226088044186026e-07,
      "loss": 0.5911,
      "step": 6186
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7281302703881575,
      "learning_rate": 2.411944407604927e-07,
      "loss": 0.5863,
      "step": 6187
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6146439766578257,
      "learning_rate": 2.4013032485616595e-07,
      "loss": 0.5686,
      "step": 6188
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.599763910836122,
      "learning_rate": 2.390685329822728e-07,
      "loss": 0.5787,
      "step": 6189
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5897362070227877,
      "learning_rate": 2.3800906539164558e-07,
      "loss": 0.5184,
      "step": 6190
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6549429874958466,
      "learning_rate": 2.3695192233656682e-07,
      "loss": 0.5802,
      "step": 6191
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6361396200642202,
      "learning_rate": 2.3589710406876408e-07,
      "loss": 0.5689,
      "step": 6192
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7539421172935411,
      "learning_rate": 2.3484461083941312e-07,
      "loss": 0.5849,
      "step": 6193
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6104779198847732,
      "learning_rate": 2.3379444289913344e-07,
      "loss": 0.5516,
      "step": 6194
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5774152953006423,
      "learning_rate": 2.327466004979917e-07,
      "loss": 0.5661,
      "step": 6195
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.510534147754674,
      "learning_rate": 2.3170108388550384e-07,
      "loss": 0.5496,
      "step": 6196
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6470612965873249,
      "learning_rate": 2.3065789331062515e-07,
      "loss": 0.5797,
      "step": 6197
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6916054822583659,
      "learning_rate": 2.2961702902176476e-07,
      "loss": 0.5694,
      "step": 6198
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5703147712070974,
      "learning_rate": 2.2857849126677433e-07,
      "loss": 0.5635,
      "step": 6199
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6290993315373514,
      "learning_rate": 2.2754228029294833e-07,
      "loss": 0.5606,
      "step": 6200
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6189288950433929,
      "learning_rate": 2.265083963470327e-07,
      "loss": 0.5602,
      "step": 6201
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6494615775918525,
      "learning_rate": 2.2547683967521827e-07,
      "loss": 0.5798,
      "step": 6202
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8877125874326905,
      "learning_rate": 2.2444761052313857e-07,
      "loss": 0.6333,
      "step": 6203
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5884073254155852,
      "learning_rate": 2.2342070913587423e-07,
      "loss": 0.5891,
      "step": 6204
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5924548857200247,
      "learning_rate": 2.2239613575795294e-07,
      "loss": 0.5522,
      "step": 6205
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6655899635774224,
      "learning_rate": 2.2137389063334846e-07,
      "loss": 0.5579,
      "step": 6206
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6429397328394254,
      "learning_rate": 2.2035397400547832e-07,
      "loss": 0.5651,
      "step": 6207
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6255375759061009,
      "learning_rate": 2.1933638611720265e-07,
      "loss": 0.5634,
      "step": 6208
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6144707521054289,
      "learning_rate": 2.1832112721083542e-07,
      "loss": 0.5596,
      "step": 6209
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6365497725633665,
      "learning_rate": 2.1730819752812772e-07,
      "loss": 0.5994,
      "step": 6210
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6208031875261446,
      "learning_rate": 2.162975973102821e-07,
      "loss": 0.5588,
      "step": 6211
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6442556013547744,
      "learning_rate": 2.1528932679794168e-07,
      "loss": 0.5599,
      "step": 6212
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5774389062694179,
      "learning_rate": 2.142833862311977e-07,
      "loss": 0.5549,
      "step": 6213
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5929589649529352,
      "learning_rate": 2.1327977584958637e-07,
      "loss": 0.5292,
      "step": 6214
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6017984716018032,
      "learning_rate": 2.1227849589208648e-07,
      "loss": 0.5744,
      "step": 6215
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.589982349132712,
      "learning_rate": 2.1127954659712513e-07,
      "loss": 0.5429,
      "step": 6216
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6416166149902358,
      "learning_rate": 2.1028292820257535e-07,
      "loss": 0.5587,
      "step": 6217
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6141138540748011,
      "learning_rate": 2.0928864094574842e-07,
      "loss": 0.5484,
      "step": 6218
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6272990940850068,
      "learning_rate": 2.082966850634094e-07,
      "loss": 0.5629,
      "step": 6219
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.587394894243686,
      "learning_rate": 2.0730706079176156e-07,
      "loss": 0.5537,
      "step": 6220
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6270773433647854,
      "learning_rate": 2.0631976836645418e-07,
      "loss": 0.5516,
      "step": 6221
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6138264876866611,
      "learning_rate": 2.0533480802258587e-07,
      "loss": 0.5718,
      "step": 6222
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.579815498584087,
      "learning_rate": 2.043521799946935e-07,
      "loss": 0.5589,
      "step": 6223
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7049550475867038,
      "learning_rate": 2.0337188451676205e-07,
      "loss": 0.5669,
      "step": 6224
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6824920885806883,
      "learning_rate": 2.023939218222215e-07,
      "loss": 0.5796,
      "step": 6225
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6564137408105388,
      "learning_rate": 2.0141829214394447e-07,
      "loss": 0.5729,
      "step": 6226
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6557647156209588,
      "learning_rate": 2.0044499571424846e-07,
      "loss": 0.5763,
      "step": 6227
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5785363680953101,
      "learning_rate": 1.9947403276489808e-07,
      "loss": 0.5366,
      "step": 6228
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6399536050375553,
      "learning_rate": 1.9850540352709836e-07,
      "loss": 0.5636,
      "step": 6229
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6498300312744487,
      "learning_rate": 1.9753910823150035e-07,
      "loss": 0.5735,
      "step": 6230
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.610221160010349,
      "learning_rate": 1.965751471081978e-07,
      "loss": 0.5504,
      "step": 6231
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5766400557472771,
      "learning_rate": 1.9561352038673264e-07,
      "loss": 0.5698,
      "step": 6232
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5979699577869045,
      "learning_rate": 1.9465422829608838e-07,
      "loss": 0.5592,
      "step": 6233
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6497220615373525,
      "learning_rate": 1.9369727106469116e-07,
      "loss": 0.5804,
      "step": 6234
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6816356576946452,
      "learning_rate": 1.9274264892041428e-07,
      "loss": 0.5582,
      "step": 6235
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6247048155799078,
      "learning_rate": 1.9179036209057034e-07,
      "loss": 0.5619,
      "step": 6236
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.61846812266756,
      "learning_rate": 1.908404108019235e-07,
      "loss": 0.5495,
      "step": 6237
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6421006958259033,
      "learning_rate": 1.8989279528067283e-07,
      "loss": 0.5763,
      "step": 6238
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6454042733701901,
      "learning_rate": 1.8894751575246783e-07,
      "loss": 0.5745,
      "step": 6239
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6676226338201187,
      "learning_rate": 1.8800457244239957e-07,
      "loss": 0.5674,
      "step": 6240
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5760542853245436,
      "learning_rate": 1.8706396557500172e-07,
      "loss": 0.5508,
      "step": 6241
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6090625440696675,
      "learning_rate": 1.8612569537425295e-07,
      "loss": 0.5621,
      "step": 6242
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7295203737139405,
      "learning_rate": 1.8518976206357452e-07,
      "loss": 0.5726,
      "step": 6243
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6014014737910555,
      "learning_rate": 1.842561658658337e-07,
      "loss": 0.5636,
      "step": 6244
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6562180449408184,
      "learning_rate": 1.8332490700333828e-07,
      "loss": 0.5648,
      "step": 6245
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6631904432950995,
      "learning_rate": 1.8239598569783856e-07,
      "loss": 0.5431,
      "step": 6246
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6801259358601232,
      "learning_rate": 1.8146940217053322e-07,
      "loss": 0.564,
      "step": 6247
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5656841319588665,
      "learning_rate": 1.805451566420613e-07,
      "loss": 0.5489,
      "step": 6248
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6710489576676839,
      "learning_rate": 1.796232493325023e-07,
      "loss": 0.5596,
      "step": 6249
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6503007358996429,
      "learning_rate": 1.7870368046138508e-07,
      "loss": 0.5786,
      "step": 6250
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6950320770413514,
      "learning_rate": 1.777864502476756e-07,
      "loss": 0.5765,
      "step": 6251
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.625390220057247,
      "learning_rate": 1.76871558909788e-07,
      "loss": 0.5795,
      "step": 6252
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6363885171251061,
      "learning_rate": 1.7595900666557585e-07,
      "loss": 0.567,
      "step": 6253
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.573644390393102,
      "learning_rate": 1.7504879373233752e-07,
      "loss": 0.538,
      "step": 6254
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6433402971581067,
      "learning_rate": 1.7414092032681294e-07,
      "loss": 0.5721,
      "step": 6255
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7191429609690105,
      "learning_rate": 1.732353866651859e-07,
      "loss": 0.5587,
      "step": 6256
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6078738826504203,
      "learning_rate": 1.72332192963085e-07,
      "loss": 0.5656,
      "step": 6257
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6842896261974585,
      "learning_rate": 1.714313394355782e-07,
      "loss": 0.5534,
      "step": 6258
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7048344636156201,
      "learning_rate": 1.7053282629717726e-07,
      "loss": 0.5797,
      "step": 6259
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5941570060096918,
      "learning_rate": 1.696366537618388e-07,
      "loss": 0.5647,
      "step": 6260
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6583281846237785,
      "learning_rate": 1.6874282204295765e-07,
      "loss": 0.5968,
      "step": 6261
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6103317357057139,
      "learning_rate": 1.6785133135337584e-07,
      "loss": 0.5488,
      "step": 6262
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6092698325081863,
      "learning_rate": 1.6696218190537682e-07,
      "loss": 0.5817,
      "step": 6263
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6568754608966988,
      "learning_rate": 1.6607537391068463e-07,
      "loss": 0.5743,
      "step": 6264
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6119158992916258,
      "learning_rate": 1.6519090758046696e-07,
      "loss": 0.5644,
      "step": 6265
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6298381388672775,
      "learning_rate": 1.6430878312533205e-07,
      "loss": 0.5438,
      "step": 6266
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5854093657565679,
      "learning_rate": 1.634290007553363e-07,
      "loss": 0.5687,
      "step": 6267
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5707200153485201,
      "learning_rate": 1.6255156067997325e-07,
      "loss": 0.5576,
      "step": 6268
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5625216489760382,
      "learning_rate": 1.6167646310817686e-07,
      "loss": 0.5517,
      "step": 6269
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6223357333258496,
      "learning_rate": 1.6080370824832824e-07,
      "loss": 0.5735,
      "step": 6270
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6399327861483546,
      "learning_rate": 1.5993329630825005e-07,
      "loss": 0.5471,
      "step": 6271
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5937482973528104,
      "learning_rate": 1.5906522749520426e-07,
      "loss": 0.5626,
      "step": 6272
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.566104719136141,
      "learning_rate": 1.5819950201589552e-07,
      "loss": 0.5645,
      "step": 6273
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6438706837976403,
      "learning_rate": 1.5733612007647226e-07,
      "loss": 0.5608,
      "step": 6274
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6473435059528634,
      "learning_rate": 1.564750818825256e-07,
      "loss": 0.5855,
      "step": 6275
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6273981307026909,
      "learning_rate": 1.5561638763908372e-07,
      "loss": 0.5732,
      "step": 6276
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6949585586780903,
      "learning_rate": 1.5476003755062197e-07,
      "loss": 0.5767,
      "step": 6277
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6085208449686842,
      "learning_rate": 1.539060318210539e-07,
      "loss": 0.5512,
      "step": 6278
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6135086409850292,
      "learning_rate": 1.5305437065373797e-07,
      "loss": 0.5292,
      "step": 6279
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.653752751780739,
      "learning_rate": 1.522050542514708e-07,
      "loss": 0.5726,
      "step": 6280
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6183429380527908,
      "learning_rate": 1.51358082816494e-07,
      "loss": 0.5469,
      "step": 6281
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6524483235439561,
      "learning_rate": 1.5051345655048733e-07,
      "loss": 0.5771,
      "step": 6282
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6104423871893347,
      "learning_rate": 1.4967117565457657e-07,
      "loss": 0.5645,
      "step": 6283
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.618676203997302,
      "learning_rate": 1.488312403293257e-07,
      "loss": 0.5577,
      "step": 6284
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6646928470344102,
      "learning_rate": 1.4799365077473926e-07,
      "loss": 0.5718,
      "step": 6285
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5959838073091963,
      "learning_rate": 1.4715840719026763e-07,
      "loss": 0.5434,
      "step": 6286
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6122318646750682,
      "learning_rate": 1.4632550977479843e-07,
      "loss": 0.5654,
      "step": 6287
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9489824698105461,
      "learning_rate": 1.4549495872666186e-07,
      "loss": 0.5725,
      "step": 6288
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5944410394961112,
      "learning_rate": 1.4466675424362863e-07,
      "loss": 0.5666,
      "step": 6289
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.658811169889194,
      "learning_rate": 1.4384089652291544e-07,
      "loss": 0.5419,
      "step": 6290
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6186411806237321,
      "learning_rate": 1.4301738576117386e-07,
      "loss": 0.5536,
      "step": 6291
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6455510386920353,
      "learning_rate": 1.4219622215449814e-07,
      "loss": 0.5754,
      "step": 6292
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5492261003356887,
      "learning_rate": 1.4137740589842519e-07,
      "loss": 0.5463,
      "step": 6293
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5724423484092074,
      "learning_rate": 1.4056093718793352e-07,
      "loss": 0.5784,
      "step": 6294
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5860342948390549,
      "learning_rate": 1.3974681621744202e-07,
      "loss": 0.5476,
      "step": 6295
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6688704438382412,
      "learning_rate": 1.3893504318080897e-07,
      "loss": 0.5749,
      "step": 6296
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6158156982871954,
      "learning_rate": 1.3812561827133307e-07,
      "loss": 0.5762,
      "step": 6297
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6136859916403187,
      "learning_rate": 1.3731854168175796e-07,
      "loss": 0.5269,
      "step": 6298
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5786259449380019,
      "learning_rate": 1.365138136042643e-07,
      "loss": 0.5581,
      "step": 6299
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6067346342545287,
      "learning_rate": 1.357114342304755e-07,
      "loss": 0.572,
      "step": 6300
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6927162396590648,
      "learning_rate": 1.349114037514543e-07,
      "loss": 0.5686,
      "step": 6301
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6740027621199238,
      "learning_rate": 1.3411372235770493e-07,
      "loss": 0.5239,
      "step": 6302
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5603382669465446,
      "learning_rate": 1.3331839023917216e-07,
      "loss": 0.5681,
      "step": 6303
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6110820327692954,
      "learning_rate": 1.3252540758524113e-07,
      "loss": 0.5817,
      "step": 6304
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5842676086723627,
      "learning_rate": 1.317347745847386e-07,
      "loss": 0.557,
      "step": 6305
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5862824997922874,
      "learning_rate": 1.3094649142593064e-07,
      "loss": 0.5651,
      "step": 6306
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6224367453682987,
      "learning_rate": 1.3016055829652263e-07,
      "loss": 0.5596,
      "step": 6307
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6169628286130808,
      "learning_rate": 1.2937697538366378e-07,
      "loss": 0.5561,
      "step": 6308
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6891188525883815,
      "learning_rate": 1.2859574287393928e-07,
      "loss": 0.5889,
      "step": 6309
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5697900095693244,
      "learning_rate": 1.278168609533803e-07,
      "loss": 0.5552,
      "step": 6310
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6692514755927683,
      "learning_rate": 1.2704032980745296e-07,
      "loss": 0.5837,
      "step": 6311
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6071135617316166,
      "learning_rate": 1.2626614962106598e-07,
      "loss": 0.5806,
      "step": 6312
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6137838289022854,
      "learning_rate": 1.2549432057856746e-07,
      "loss": 0.5725,
      "step": 6313
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.62509719094514,
      "learning_rate": 1.247248428637471e-07,
      "loss": 0.5691,
      "step": 6314
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6118562222013085,
      "learning_rate": 1.239577166598327e-07,
      "loss": 0.5538,
      "step": 6315
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6419369502287721,
      "learning_rate": 1.2319294214949373e-07,
      "loss": 0.5821,
      "step": 6316
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6312579986303589,
      "learning_rate": 1.2243051951483898e-07,
      "loss": 0.5599,
      "step": 6317
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6145098643527873,
      "learning_rate": 1.2167044893741876e-07,
      "loss": 0.5541,
      "step": 6318
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6341390606336731,
      "learning_rate": 1.209127305982205e-07,
      "loss": 0.5567,
      "step": 6319
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.571347758525733,
      "learning_rate": 1.2015736467767102e-07,
      "loss": 0.5476,
      "step": 6320
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5929802911750771,
      "learning_rate": 1.194043513556431e-07,
      "loss": 0.5445,
      "step": 6321
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6335134033574591,
      "learning_rate": 1.1865369081144107e-07,
      "loss": 0.5786,
      "step": 6322
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6174172825083679,
      "learning_rate": 1.1790538322381528e-07,
      "loss": 0.5608,
      "step": 6323
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.631685819365676,
      "learning_rate": 1.171594287709532e-07,
      "loss": 0.5685,
      "step": 6324
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6620451794919252,
      "learning_rate": 1.1641582763048165e-07,
      "loss": 0.5642,
      "step": 6325
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6562007035950577,
      "learning_rate": 1.1567457997946896e-07,
      "loss": 0.5836,
      "step": 6326
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6115397759855254,
      "learning_rate": 1.1493568599441951e-07,
      "loss": 0.5554,
      "step": 6327
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5909413388479148,
      "learning_rate": 1.1419914585128145e-07,
      "loss": 0.5531,
      "step": 6328
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5785048870755922,
      "learning_rate": 1.1346495972544003e-07,
      "loss": 0.5369,
      "step": 6329
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.613331749033247,
      "learning_rate": 1.1273312779172096e-07,
      "loss": 0.6006,
      "step": 6330
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.654262144506682,
      "learning_rate": 1.1200365022438819e-07,
      "loss": 0.5457,
      "step": 6331
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6229230408721643,
      "learning_rate": 1.1127652719714388e-07,
      "loss": 0.5598,
      "step": 6332
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6508014918050511,
      "learning_rate": 1.1055175888313507e-07,
      "loss": 0.5855,
      "step": 6333
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5950379434199085,
      "learning_rate": 1.0982934545494262e-07,
      "loss": 0.5586,
      "step": 6334
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7280759895579574,
      "learning_rate": 1.0910928708458779e-07,
      "loss": 0.6023,
      "step": 6335
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7672917805002042,
      "learning_rate": 1.0839158394353122e-07,
      "loss": 0.5677,
      "step": 6336
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6216306489732993,
      "learning_rate": 1.0767623620267509e-07,
      "loss": 0.5509,
      "step": 6337
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7294297417591498,
      "learning_rate": 1.0696324403235759e-07,
      "loss": 0.5866,
      "step": 6338
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6771401193253139,
      "learning_rate": 1.0625260760235623e-07,
      "loss": 0.608,
      "step": 6339
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6786748640055517,
      "learning_rate": 1.055443270818901e-07,
      "loss": 0.5604,
      "step": 6340
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5704180104232831,
      "learning_rate": 1.0483840263961543e-07,
      "loss": 0.5582,
      "step": 6341
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6360268825943007,
      "learning_rate": 1.041348344436277e-07,
      "loss": 0.5845,
      "step": 6342
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6492267231368706,
      "learning_rate": 1.0343362266145963e-07,
      "loss": 0.5535,
      "step": 6343
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6265432166476423,
      "learning_rate": 1.0273476746008649e-07,
      "loss": 0.5453,
      "step": 6344
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7126528075897332,
      "learning_rate": 1.0203826900591962e-07,
      "loss": 0.5914,
      "step": 6345
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5953309968264467,
      "learning_rate": 1.0134412746481082e-07,
      "loss": 0.5748,
      "step": 6346
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5859114285009317,
      "learning_rate": 1.0065234300204895e-07,
      "loss": 0.5531,
      "step": 6347
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5734434918627724,
      "learning_rate": 9.996291578236228e-08,
      "loss": 0.5626,
      "step": 6348
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6038680758258902,
      "learning_rate": 9.927584596991835e-08,
      "loss": 0.5765,
      "step": 6349
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.612605667814685,
      "learning_rate": 9.859113372832297e-08,
      "loss": 0.5431,
      "step": 6350
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6210931546828173,
      "learning_rate": 9.790877922062015e-08,
      "loss": 0.5732,
      "step": 6351
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6084022391056093,
      "learning_rate": 9.722878260929325e-08,
      "loss": 0.5728,
      "step": 6352
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.593561769866803,
      "learning_rate": 9.655114405626387e-08,
      "loss": 0.5618,
      "step": 6353
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6235713898408585,
      "learning_rate": 9.587586372289182e-08,
      "loss": 0.5611,
      "step": 6354
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6049108842621558,
      "learning_rate": 9.520294176997514e-08,
      "loss": 0.5777,
      "step": 6355
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6793072320560274,
      "learning_rate": 9.453237835775119e-08,
      "loss": 0.574,
      "step": 6356
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5728084525643748,
      "learning_rate": 9.386417364589562e-08,
      "loss": 0.5571,
      "step": 6357
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6148386086108282,
      "learning_rate": 9.319832779352223e-08,
      "loss": 0.5827,
      "step": 6358
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5971494680118105,
      "learning_rate": 9.253484095918087e-08,
      "loss": 0.5469,
      "step": 6359
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5863823694871222,
      "learning_rate": 9.187371330086404e-08,
      "loss": 0.563,
      "step": 6360
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6322578907145052,
      "learning_rate": 9.121494497599915e-08,
      "loss": 0.5514,
      "step": 6361
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6469812136776109,
      "learning_rate": 9.055853614145294e-08,
      "loss": 0.5521,
      "step": 6362
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6059652836818963,
      "learning_rate": 8.990448695352927e-08,
      "loss": 0.5657,
      "step": 6363
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5867612893706623,
      "learning_rate": 8.925279756797357e-08,
      "loss": 0.5576,
      "step": 6364
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6623764329574018,
      "learning_rate": 8.86034681399639e-08,
      "loss": 0.572,
      "step": 6365
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6436613808976017,
      "learning_rate": 8.795649882412105e-08,
      "loss": 0.5783,
      "step": 6366
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5985578058578714,
      "learning_rate": 8.731188977450178e-08,
      "loss": 0.5537,
      "step": 6367
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6337008855168134,
      "learning_rate": 8.666964114459997e-08,
      "loss": 0.5786,
      "step": 6368
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6415184651558188,
      "learning_rate": 8.602975308734996e-08,
      "loss": 0.5753,
      "step": 6369
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6719754116811515,
      "learning_rate": 8.539222575512096e-08,
      "loss": 0.5502,
      "step": 6370
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6065159355705522,
      "learning_rate": 8.475705929972377e-08,
      "loss": 0.586,
      "step": 6371
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6126863760233003,
      "learning_rate": 8.412425387240408e-08,
      "loss": 0.5451,
      "step": 6372
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6081999263293006,
      "learning_rate": 8.349380962384468e-08,
      "loss": 0.5504,
      "step": 6373
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6324497392055608,
      "learning_rate": 8.286572670416993e-08,
      "loss": 0.5423,
      "step": 6374
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.602560182821042,
      "learning_rate": 8.224000526293796e-08,
      "loss": 0.5569,
      "step": 6375
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5953520281451772,
      "learning_rate": 8.161664544914627e-08,
      "loss": 0.5606,
      "step": 6376
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6279298570761448,
      "learning_rate": 8.099564741123167e-08,
      "loss": 0.5632,
      "step": 6377
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6070597151555512,
      "learning_rate": 8.037701129706366e-08,
      "loss": 0.5799,
      "step": 6378
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6049036428684542,
      "learning_rate": 7.97607372539566e-08,
      "loss": 0.5677,
      "step": 6379
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.695807587685676,
      "learning_rate": 7.914682542865537e-08,
      "loss": 0.5734,
      "step": 6380
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.613841307937507,
      "learning_rate": 7.853527596734634e-08,
      "loss": 0.5742,
      "step": 6381
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6768247191774991,
      "learning_rate": 7.792608901565191e-08,
      "loss": 0.5532,
      "step": 6382
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6178335180984968,
      "learning_rate": 7.73192647186316e-08,
      "loss": 0.5573,
      "step": 6383
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6534831452017744,
      "learning_rate": 7.67148032207854e-08,
      "loss": 0.5719,
      "step": 6384
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.593356997664956,
      "learning_rate": 7.611270466604592e-08,
      "loss": 0.5752,
      "step": 6385
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.760486170735005,
      "learning_rate": 7.551296919778738e-08,
      "loss": 0.6112,
      "step": 6386
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.604116346277985,
      "learning_rate": 7.491559695881779e-08,
      "loss": 0.5595,
      "step": 6387
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6580239375299676,
      "learning_rate": 7.432058809138442e-08,
      "loss": 0.5466,
      "step": 6388
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6389490272004543,
      "learning_rate": 7.372794273717288e-08,
      "loss": 0.5531,
      "step": 6389
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7031183407868526,
      "learning_rate": 7.313766103730246e-08,
      "loss": 0.5766,
      "step": 6390
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.651055963046969,
      "learning_rate": 7.254974313233298e-08,
      "loss": 0.5723,
      "step": 6391
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5879851996533052,
      "learning_rate": 7.196418916226022e-08,
      "loss": 0.5684,
      "step": 6392
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5643732529873443,
      "learning_rate": 7.138099926651487e-08,
      "loss": 0.5314,
      "step": 6393
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6098777065191062,
      "learning_rate": 7.080017358396917e-08,
      "loss": 0.5534,
      "step": 6394
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5899225504730073,
      "learning_rate": 7.022171225292918e-08,
      "loss": 0.5563,
      "step": 6395
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6755901941604264,
      "learning_rate": 6.964561541113801e-08,
      "loss": 0.5982,
      "step": 6396
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.592904356379865,
      "learning_rate": 6.907188319577706e-08,
      "loss": 0.5748,
      "step": 6397
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6659332031709211,
      "learning_rate": 6.850051574346373e-08,
      "loss": 0.5716,
      "step": 6398
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6341800072320176,
      "learning_rate": 6.79315131902536e-08,
      "loss": 0.5786,
      "step": 6399
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6564560992497543,
      "learning_rate": 6.736487567163719e-08,
      "loss": 0.5791,
      "step": 6400
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6528187302748303,
      "learning_rate": 6.680060332254212e-08,
      "loss": 0.5573,
      "step": 6401
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6252759657122914,
      "learning_rate": 6.623869627733537e-08,
      "loss": 0.5628,
      "step": 6402
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5836114944188386,
      "learning_rate": 6.567915466981767e-08,
      "loss": 0.5618,
      "step": 6403
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6111106131449321,
      "learning_rate": 6.5121978633228e-08,
      "loss": 0.5585,
      "step": 6404
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5708276218975182,
      "learning_rate": 6.456716830024135e-08,
      "loss": 0.5499,
      "step": 6405
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6359161043692093,
      "learning_rate": 6.401472380297091e-08,
      "loss": 0.5863,
      "step": 6406
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6048582998357203,
      "learning_rate": 6.34646452729637e-08,
      "loss": 0.5291,
      "step": 6407
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6334758994018823,
      "learning_rate": 6.291693284120492e-08,
      "loss": 0.5656,
      "step": 6408
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6876546452906458,
      "learning_rate": 6.237158663811805e-08,
      "loss": 0.551,
      "step": 6409
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6047781547515981,
      "learning_rate": 6.18286067935614e-08,
      "loss": 0.533,
      "step": 6410
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6003848516703713,
      "learning_rate": 6.128799343682712e-08,
      "loss": 0.5496,
      "step": 6411
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6307075498733091,
      "learning_rate": 6.074974669665112e-08,
      "loss": 0.579,
      "step": 6412
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6344483049024646,
      "learning_rate": 6.021386670119755e-08,
      "loss": 0.5553,
      "step": 6413
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.584049412266616,
      "learning_rate": 5.9680353578071e-08,
      "loss": 0.5645,
      "step": 6414
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5882364881523097,
      "learning_rate": 5.9149207454315404e-08,
      "loss": 0.5608,
      "step": 6415
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5653185273884632,
      "learning_rate": 5.862042845640403e-08,
      "loss": 0.5385,
      "step": 6416
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6326007491590924,
      "learning_rate": 5.809401671025283e-08,
      "loss": 0.5696,
      "step": 6417
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6794948873948332,
      "learning_rate": 5.7569972341210424e-08,
      "loss": 0.5787,
      "step": 6418
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6376274360401408,
      "learning_rate": 5.704829547406143e-08,
      "loss": 0.5466,
      "step": 6419
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6504784363150949,
      "learning_rate": 5.652898623303094e-08,
      "loss": 0.5763,
      "step": 6420
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6370514660344744,
      "learning_rate": 5.601204474177668e-08,
      "loss": 0.5783,
      "step": 6421
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6561736760997064,
      "learning_rate": 5.5497471123391325e-08,
      "loss": 0.5726,
      "step": 6422
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6675121136284937,
      "learning_rate": 5.498526550040906e-08,
      "loss": 0.5672,
      "step": 6423
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5623547836257189,
      "learning_rate": 5.447542799479233e-08,
      "loss": 0.5549,
      "step": 6424
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6783958462946509,
      "learning_rate": 5.396795872794846e-08,
      "loss": 0.5591,
      "step": 6425
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.671520415419185,
      "learning_rate": 5.346285782071414e-08,
      "loss": 0.5535,
      "step": 6426
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5652086737090345,
      "learning_rate": 5.296012539336537e-08,
      "loss": 0.5471,
      "step": 6427
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.629346560674772,
      "learning_rate": 5.2459761565613054e-08,
      "loss": 0.5822,
      "step": 6428
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5892815926551586,
      "learning_rate": 5.196176645660411e-08,
      "loss": 0.5578,
      "step": 6429
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6003307879769696,
      "learning_rate": 5.146614018492369e-08,
      "loss": 0.556,
      "step": 6430
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6168220523325448,
      "learning_rate": 5.097288286858737e-08,
      "loss": 0.5566,
      "step": 6431
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6284790191371614,
      "learning_rate": 5.048199462505232e-08,
      "loss": 0.5305,
      "step": 6432
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6732523692021166,
      "learning_rate": 4.9993475571209485e-08,
      "loss": 0.5777,
      "step": 6433
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5703950680616939,
      "learning_rate": 4.950732582338358e-08,
      "loss": 0.5396,
      "step": 6434
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6193917539873006,
      "learning_rate": 4.902354549733979e-08,
      "loss": 0.5652,
      "step": 6435
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6089474984714803,
      "learning_rate": 4.8542134708274844e-08,
      "loss": 0.5392,
      "step": 6436
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6120704339934504,
      "learning_rate": 4.8063093570822615e-08,
      "loss": 0.5767,
      "step": 6437
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.645448394117943,
      "learning_rate": 4.7586422199054075e-08,
      "loss": 0.5633,
      "step": 6438
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6595181696121003,
      "learning_rate": 4.711212070647286e-08,
      "loss": 0.5617,
      "step": 6439
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7101692259202028,
      "learning_rate": 4.664018920602198e-08,
      "loss": 0.545,
      "step": 6440
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5660502599644609,
      "learning_rate": 4.6170627810077086e-08,
      "loss": 0.5624,
      "step": 6441
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5843558535664827,
      "learning_rate": 4.570343663045096e-08,
      "loss": 0.5398,
      "step": 6442
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6292761902913976,
      "learning_rate": 4.5238615778392394e-08,
      "loss": 0.5671,
      "step": 6443
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6283910202180191,
      "learning_rate": 4.477616536458396e-08,
      "loss": 0.5687,
      "step": 6444
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6577868480079334,
      "learning_rate": 4.431608549914535e-08,
      "loss": 0.5367,
      "step": 6445
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5976073607805957,
      "learning_rate": 4.385837629163114e-08,
      "loss": 0.5571,
      "step": 6446
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.599330880174151,
      "learning_rate": 4.3403037851031945e-08,
      "loss": 0.5495,
      "step": 6447
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5963365101069611,
      "learning_rate": 4.295007028577214e-08,
      "loss": 0.5764,
      "step": 6448
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7843253379097824,
      "learning_rate": 4.249947370371432e-08,
      "loss": 0.5746,
      "step": 6449
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6122865827846522,
      "learning_rate": 4.2051248212154897e-08,
      "loss": 0.5597,
      "step": 6450
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6321976747463154,
      "learning_rate": 4.1605393917824034e-08,
      "loss": 0.5647,
      "step": 6451
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6522390881382375,
      "learning_rate": 4.116191092689126e-08,
      "loss": 0.5801,
      "step": 6452
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6475228382940452,
      "learning_rate": 4.072079934495765e-08,
      "loss": 0.5741,
      "step": 6453
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.660387836821809,
      "learning_rate": 4.0282059277062524e-08,
      "loss": 0.59,
      "step": 6454
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6411591313823338,
      "learning_rate": 3.9845690827677866e-08,
      "loss": 0.5877,
      "step": 6455
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5351195674391999,
      "learning_rate": 3.941169410071277e-08,
      "loss": 0.5588,
      "step": 6456
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5873463210023414,
      "learning_rate": 3.898006919951125e-08,
      "loss": 0.5584,
      "step": 6457
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6498275287012032,
      "learning_rate": 3.855081622685219e-08,
      "loss": 0.5758,
      "step": 6458
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6057180502717355,
      "learning_rate": 3.812393528494829e-08,
      "loss": 0.5505,
      "step": 6459
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.643706305104524,
      "learning_rate": 3.769942647545044e-08,
      "loss": 0.5874,
      "step": 6460
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7250594807175842,
      "learning_rate": 3.727728989944446e-08,
      "loss": 0.6002,
      "step": 6461
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6363602591012452,
      "learning_rate": 3.68575256574466e-08,
      "loss": 0.5475,
      "step": 6462
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6779895217082723,
      "learning_rate": 3.6440133849413586e-08,
      "loss": 0.58,
      "step": 6463
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6080984372487644,
      "learning_rate": 3.602511457473479e-08,
      "loss": 0.5641,
      "step": 6464
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6827397354691077,
      "learning_rate": 3.561246793223561e-08,
      "loss": 0.5606,
      "step": 6465
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5995939701102898,
      "learning_rate": 3.5202194020175215e-08,
      "loss": 0.5672,
      "step": 6466
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7177600882961843,
      "learning_rate": 3.479429293624881e-08,
      "loss": 0.6026,
      "step": 6467
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.734192087727876,
      "learning_rate": 3.438876477758646e-08,
      "loss": 0.6043,
      "step": 6468
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5422148024429644,
      "learning_rate": 3.398560964075204e-08,
      "loss": 0.5411,
      "step": 6469
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6417758031754253,
      "learning_rate": 3.358482762174542e-08,
      "loss": 0.55,
      "step": 6470
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5796740047516412,
      "learning_rate": 3.318641881600249e-08,
      "loss": 0.5849,
      "step": 6471
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6295797058886149,
      "learning_rate": 3.279038331839068e-08,
      "loss": 0.5521,
      "step": 6472
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5228201221962301,
      "learning_rate": 3.2396721223216795e-08,
      "loss": 0.5627,
      "step": 6473
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6621500029862115,
      "learning_rate": 3.200543262421807e-08,
      "loss": 0.565,
      "step": 6474
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5792844689402363,
      "learning_rate": 3.161651761456996e-08,
      "loss": 0.5518,
      "step": 6475
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5780878651046308,
      "learning_rate": 3.1229976286880624e-08,
      "loss": 0.5669,
      "step": 6476
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6407284812812579,
      "learning_rate": 3.08458087331942e-08,
      "loss": 0.5409,
      "step": 6477
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6073246211633264,
      "learning_rate": 3.046401504498753e-08,
      "loss": 0.5564,
      "step": 6478
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5686337987684441,
      "learning_rate": 3.008459531317564e-08,
      "loss": 0.5501,
      "step": 6479
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5993117055551586,
      "learning_rate": 2.970754962810518e-08,
      "loss": 0.5806,
      "step": 6480
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6183387753410514,
      "learning_rate": 2.933287807955987e-08,
      "loss": 0.5572,
      "step": 6481
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6170349618134474,
      "learning_rate": 2.8960580756756118e-08,
      "loss": 0.5618,
      "step": 6482
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6789778315156786,
      "learning_rate": 2.859065774834413e-08,
      "loss": 0.5875,
      "step": 6483
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6691565043111097,
      "learning_rate": 2.8223109142413442e-08,
      "loss": 0.5648,
      "step": 6484
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6613592526206179,
      "learning_rate": 2.7857935026482928e-08,
      "loss": 0.6043,
      "step": 6485
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5943934239135318,
      "learning_rate": 2.7495135487509706e-08,
      "loss": 0.5508,
      "step": 6486
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6340198243404533,
      "learning_rate": 2.713471061188244e-08,
      "loss": 0.5437,
      "step": 6487
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5912698651071288,
      "learning_rate": 2.6776660485426932e-08,
      "loss": 0.568,
      "step": 6488
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5920505908937648,
      "learning_rate": 2.642098519340275e-08,
      "loss": 0.551,
      "step": 6489
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6533132137625406,
      "learning_rate": 2.606768482050215e-08,
      "loss": 0.5538,
      "step": 6490
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5815199795267596,
      "learning_rate": 2.57167594508545e-08,
      "loss": 0.5563,
      "step": 6491
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6477705674900327,
      "learning_rate": 2.5368209168021852e-08,
      "loss": 0.5879,
      "step": 6492
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6112974471311692,
      "learning_rate": 2.5022034055003363e-08,
      "loss": 0.5481,
      "step": 6493
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5996498701015902,
      "learning_rate": 2.4678234194227546e-08,
      "loss": 0.5607,
      "step": 6494
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5694556803996196,
      "learning_rate": 2.4336809667563355e-08,
      "loss": 0.5581,
      "step": 6495
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6064945695385562,
      "learning_rate": 2.399776055630909e-08,
      "loss": 0.5483,
      "step": 6496
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6387486906186259,
      "learning_rate": 2.3661086941200174e-08,
      "loss": 0.5627,
      "step": 6497
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6757085140786692,
      "learning_rate": 2.3326788902404695e-08,
      "loss": 0.5593,
      "step": 6498
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6316498730600013,
      "learning_rate": 2.299486651952898e-08,
      "loss": 0.5716,
      "step": 6499
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5705955707217412,
      "learning_rate": 2.2665319871607584e-08,
      "loss": 0.5604,
      "step": 6500
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5759255814749408,
      "learning_rate": 2.2338149037113287e-08,
      "loss": 0.5515,
      "step": 6501
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7941661480229412,
      "learning_rate": 2.2013354093953775e-08,
      "loss": 0.5686,
      "step": 6502
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6297388354796303,
      "learning_rate": 2.1690935119468294e-08,
      "loss": 0.5466,
      "step": 6503
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.628334214469636,
      "learning_rate": 2.137089219043098e-08,
      "loss": 0.5611,
      "step": 6504
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6376501282994149,
      "learning_rate": 2.1053225383051988e-08,
      "loss": 0.5777,
      "step": 6505
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6015258155700278,
      "learning_rate": 2.0737934772975254e-08,
      "loss": 0.5408,
      "step": 6506
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6548726846441378,
      "learning_rate": 2.0425020435276278e-08,
      "loss": 0.5849,
      "step": 6507
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5988171864352942,
      "learning_rate": 2.0114482444466565e-08,
      "loss": 0.563,
      "step": 6508
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.603010939993788,
      "learning_rate": 1.9806320874492523e-08,
      "loss": 0.5843,
      "step": 6509
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6128385552746018,
      "learning_rate": 1.9500535798734342e-08,
      "loss": 0.5788,
      "step": 6510
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5977671953840574,
      "learning_rate": 1.9197127290006003e-08,
      "loss": 0.5561,
      "step": 6511
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6035431928413375,
      "learning_rate": 1.889609542055415e-08,
      "loss": 0.5374,
      "step": 6512
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6703276876457988,
      "learning_rate": 1.859744026206145e-08,
      "loss": 0.5801,
      "step": 6513
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6581868959482469,
      "learning_rate": 1.8301161885644347e-08,
      "loss": 0.578,
      "step": 6514
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6124598946317599,
      "learning_rate": 1.8007260361851963e-08,
      "loss": 0.5885,
      "step": 6515
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6399209774720596,
      "learning_rate": 1.7715735760669428e-08,
      "loss": 0.5831,
      "step": 6516
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6340573986295871,
      "learning_rate": 1.7426588151514546e-08,
      "loss": 0.5895,
      "step": 6517
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6344116675725231,
      "learning_rate": 1.7139817603240015e-08,
      "loss": 0.5609,
      "step": 6518
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6287615320346815,
      "learning_rate": 1.685542418413122e-08,
      "loss": 0.5642,
      "step": 6519
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6584613523676088,
      "learning_rate": 1.6573407961907317e-08,
      "loss": 0.5879,
      "step": 6520
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6278061678431875,
      "learning_rate": 1.6293769003724592e-08,
      "loss": 0.5724,
      "step": 6521
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5926023992775143,
      "learning_rate": 1.6016507376169776e-08,
      "loss": 0.5753,
      "step": 6522
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6181109312049737,
      "learning_rate": 1.5741623145263396e-08,
      "loss": 0.5616,
      "step": 6523
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6528593525357226,
      "learning_rate": 1.54691163764642e-08,
      "loss": 0.563,
      "step": 6524
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6645220847255392,
      "learning_rate": 1.519898713465806e-08,
      "loss": 0.5728,
      "step": 6525
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6554608630269512,
      "learning_rate": 1.4931235484172412e-08,
      "loss": 0.5473,
      "step": 6526
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6499064765736792,
      "learning_rate": 1.4665861488761813e-08,
      "loss": 0.5679,
      "step": 6527
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5360649648157378,
      "learning_rate": 1.4402865211617934e-08,
      "loss": 0.5466,
      "step": 6528
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6333038436453611,
      "learning_rate": 1.4142246715366238e-08,
      "loss": 0.5809,
      "step": 6529
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6279112223651215,
      "learning_rate": 1.388400606206486e-08,
      "loss": 0.566,
      "step": 6530
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.588355551583381,
      "learning_rate": 1.3628143313206833e-08,
      "loss": 0.5592,
      "step": 6531
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6221640629558394,
      "learning_rate": 1.3374658529717866e-08,
      "loss": 0.5495,
      "step": 6532
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7014072074209231,
      "learning_rate": 1.3123551771958564e-08,
      "loss": 0.5777,
      "step": 6533
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.598551750992874,
      "learning_rate": 1.287482309972332e-08,
      "loss": 0.5686,
      "step": 6534
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5899251859514298,
      "learning_rate": 1.2628472572239204e-08,
      "loss": 0.5573,
      "step": 6535
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6131790059500649,
      "learning_rate": 1.2384500248165954e-08,
      "loss": 0.5593,
      "step": 6536
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5765573038670757,
      "learning_rate": 1.2142906185600433e-08,
      "loss": 0.547,
      "step": 6537
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7470230177724838,
      "learning_rate": 1.190369044206996e-08,
      "loss": 0.5886,
      "step": 6538
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6358288653465857,
      "learning_rate": 1.1666853074538965e-08,
      "loss": 0.5231,
      "step": 6539
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6457455337240113,
      "learning_rate": 1.1432394139401226e-08,
      "loss": 0.6001,
      "step": 6540
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6081420615534961,
      "learning_rate": 1.120031369248653e-08,
      "loss": 0.569,
      "step": 6541
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7997015634511262,
      "learning_rate": 1.097061178905956e-08,
      "loss": 0.5948,
      "step": 6542
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6153880864160831,
      "learning_rate": 1.074328848381545e-08,
      "loss": 0.5643,
      "step": 6543
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5429109371126793,
      "learning_rate": 1.0518343830885346e-08,
      "loss": 0.5411,
      "step": 6544
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6560962099567372,
      "learning_rate": 1.029577788383418e-08,
      "loss": 0.577,
      "step": 6545
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6070099872990045,
      "learning_rate": 1.0075590695658444e-08,
      "loss": 0.5708,
      "step": 6546
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6185394951201979,
      "learning_rate": 9.857782318790643e-09,
      "loss": 0.5821,
      "step": 6547
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6153349753985956,
      "learning_rate": 9.642352805093735e-09,
      "loss": 0.5663,
      "step": 6548
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6007583339919934,
      "learning_rate": 9.429302205866686e-09,
      "loss": 0.5632,
      "step": 6549
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5595325156207198,
      "learning_rate": 9.218630571842247e-09,
      "loss": 0.5381,
      "step": 6550
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6626389894659596,
      "learning_rate": 9.010337953185843e-09,
      "loss": 0.5698,
      "step": 6551
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6578710007816689,
      "learning_rate": 8.80442439949447e-09,
      "loss": 0.5835,
      "step": 6552
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6626612291809267,
      "learning_rate": 8.600889959801128e-09,
      "loss": 0.5746,
      "step": 6553
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6117395293290318,
      "learning_rate": 8.399734682573712e-09,
      "loss": 0.5609,
      "step": 6554
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6914873668092402,
      "learning_rate": 8.200958615708354e-09,
      "loss": 0.5954,
      "step": 6555
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6893728561177143,
      "learning_rate": 8.004561806540523e-09,
      "loss": 0.5827,
      "step": 6556
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6197622178074332,
      "learning_rate": 7.810544301835032e-09,
      "loss": 0.5415,
      "step": 6557
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6205431319925464,
      "learning_rate": 7.618906147791594e-09,
      "loss": 0.5728,
      "step": 6558
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6346581878445655,
      "learning_rate": 7.429647390043704e-09,
      "loss": 0.5659,
      "step": 6559
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.670727403202772,
      "learning_rate": 7.2427680736575355e-09,
      "loss": 0.5589,
      "step": 6560
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6086484695249447,
      "learning_rate": 7.058268243133048e-09,
      "loss": 0.562,
      "step": 6561
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5711950786784271,
      "learning_rate": 6.876147942403988e-09,
      "loss": 0.5568,
      "step": 6562
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6234828967347271,
      "learning_rate": 6.696407214835665e-09,
      "loss": 0.5521,
      "step": 6563
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6264872768710233,
      "learning_rate": 6.5190461032305085e-09,
      "loss": 0.5546,
      "step": 6564
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6561923083526894,
      "learning_rate": 6.344064649819182e-09,
      "loss": 0.571,
      "step": 6565
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6280812641599335,
      "learning_rate": 6.171462896270575e-09,
      "loss": 0.5586,
      "step": 6566
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6919085391767646,
      "learning_rate": 6.001240883684034e-09,
      "loss": 0.5753,
      "step": 6567
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6368199693409674,
      "learning_rate": 5.833398652593802e-09,
      "loss": 0.5597,
      "step": 6568
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5670020788583297,
      "learning_rate": 5.667936242964578e-09,
      "loss": 0.5526,
      "step": 6569
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6517725637086152,
      "learning_rate": 5.504853694198176e-09,
      "loss": 0.5645,
      "step": 6570
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6513605715628056,
      "learning_rate": 5.344151045126866e-09,
      "loss": 0.5616,
      "step": 6571
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5675789946425044,
      "learning_rate": 5.185828334018928e-09,
      "loss": 0.5495,
      "step": 6572
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6369293669048143,
      "learning_rate": 5.029885598573092e-09,
      "loss": 0.5918,
      "step": 6573
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6095306719187098,
      "learning_rate": 4.87632287592299e-09,
      "loss": 0.5721,
      "step": 6574
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5558231717043298,
      "learning_rate": 4.725140202634926e-09,
      "loss": 0.5352,
      "step": 6575
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6401363844220932,
      "learning_rate": 4.576337614708992e-09,
      "loss": 0.5836,
      "step": 6576
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7725591227657623,
      "learning_rate": 4.4299151475779565e-09,
      "loss": 0.5793,
      "step": 6577
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6406942787072162,
      "learning_rate": 4.285872836108374e-09,
      "loss": 0.5833,
      "step": 6578
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7691880660149498,
      "learning_rate": 4.144210714599472e-09,
      "loss": 0.5894,
      "step": 6579
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5995672047253197,
      "learning_rate": 4.00492881678427e-09,
      "loss": 0.579,
      "step": 6580
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6607463144068729,
      "learning_rate": 3.868027175827349e-09,
      "loss": 0.5799,
      "step": 6581
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6469609790160139,
      "learning_rate": 3.733505824330408e-09,
      "loss": 0.5747,
      "step": 6582
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5856975603259224,
      "learning_rate": 3.6013647943233808e-09,
      "loss": 0.5586,
      "step": 6583
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6184050699599231,
      "learning_rate": 3.4716041172733195e-09,
      "loss": 0.5432,
      "step": 6584
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5977707527926425,
      "learning_rate": 3.3442238240788403e-09,
      "loss": 0.5814,
      "step": 6585
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6304002827844055,
      "learning_rate": 3.219223945071237e-09,
      "loss": 0.5627,
      "step": 6586
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5777329892187025,
      "learning_rate": 3.0966045100155885e-09,
      "loss": 0.5418,
      "step": 6587
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6435857385184411,
      "learning_rate": 2.9763655481118705e-09,
      "loss": 0.5422,
      "step": 6588
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6220550768519639,
      "learning_rate": 2.858507087988294e-09,
      "loss": 0.5387,
      "step": 6589
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6910423430570529,
      "learning_rate": 2.743029157712407e-09,
      "loss": 0.575,
      "step": 6590
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6528554529744419,
      "learning_rate": 2.6299317847811034e-09,
      "loss": 0.5465,
      "step": 6591
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6111300632410296,
      "learning_rate": 2.5192149961250633e-09,
      "loss": 0.5648,
      "step": 6592
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6231756561679894,
      "learning_rate": 2.4108788181076428e-09,
      "loss": 0.5565,
      "step": 6593
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.652511369650803,
      "learning_rate": 2.3049232765259834e-09,
      "loss": 0.5851,
      "step": 6594
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6378814941011552,
      "learning_rate": 2.201348396612124e-09,
      "loss": 0.5713,
      "step": 6595
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6487464960085633,
      "learning_rate": 2.100154203027449e-09,
      "loss": 0.5738,
      "step": 6596
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6758962880539175,
      "learning_rate": 2.0013407198693492e-09,
      "loss": 0.5769,
      "step": 6597
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.656375722166194,
      "learning_rate": 1.9049079706667804e-09,
      "loss": 0.5571,
      "step": 6598
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.659828479580366,
      "learning_rate": 1.8108559783824863e-09,
      "loss": 0.5711,
      "step": 6599
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6121484593158358,
      "learning_rate": 1.719184765414106e-09,
      "loss": 0.5421,
      "step": 6600
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6631140586889157,
      "learning_rate": 1.6298943535875133e-09,
      "loss": 0.5776,
      "step": 6601
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5884125453939113,
      "learning_rate": 1.5429847641668106e-09,
      "loss": 0.5267,
      "step": 6602
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6488688920148517,
      "learning_rate": 1.4584560178465546e-09,
      "loss": 0.5765,
      "step": 6603
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6453552325521485,
      "learning_rate": 1.3763081347539785e-09,
      "loss": 0.5898,
      "step": 6604
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7219348600768241,
      "learning_rate": 1.2965411344501023e-09,
      "loss": 0.6061,
      "step": 6605
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.621848184441379,
      "learning_rate": 1.2191550359308413e-09,
      "loss": 0.5483,
      "step": 6606
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6435587061906904,
      "learning_rate": 1.1441498576225673e-09,
      "loss": 0.5689,
      "step": 6607
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.680204142455661,
      "learning_rate": 1.071525617384328e-09,
      "loss": 0.5686,
      "step": 6608
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7822008526632316,
      "learning_rate": 1.0012823325111776e-09,
      "loss": 0.5937,
      "step": 6609
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6345850868274766,
      "learning_rate": 9.334200197286258e-10,
      "loss": 0.5788,
      "step": 6610
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6058858537499411,
      "learning_rate": 8.679386951970792e-10,
      "loss": 0.5621,
      "step": 6611
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6308316130905844,
      "learning_rate": 8.048383745085097e-10,
      "loss": 0.5444,
      "step": 6612
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6218985084231007,
      "learning_rate": 7.441190726875658e-10,
      "loss": 0.5668,
      "step": 6613
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6066278601433752,
      "learning_rate": 6.857808041937919e-10,
      "loss": 0.5361,
      "step": 6614
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5789862007925324,
      "learning_rate": 6.298235829182986e-10,
      "loss": 0.5697,
      "step": 6615
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6365567233668313,
      "learning_rate": 5.762474221859826e-10,
      "loss": 0.5867,
      "step": 6616
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5633746401106743,
      "learning_rate": 5.250523347544167e-10,
      "loss": 0.5316,
      "step": 6617
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.628355359856209,
      "learning_rate": 4.762383328138498e-10,
      "loss": 0.5601,
      "step": 6618
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6122460852623793,
      "learning_rate": 4.298054279883168e-10,
      "loss": 0.5685,
      "step": 6619
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5907977186731707,
      "learning_rate": 3.857536313345289e-10,
      "loss": 0.5385,
      "step": 6620
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5948656831859032,
      "learning_rate": 3.4408295334187324e-10,
      "loss": 0.5539,
      "step": 6621
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5905812181038466,
      "learning_rate": 3.047934039335232e-10,
      "loss": 0.5641,
      "step": 6622
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6120674856383039,
      "learning_rate": 2.6788499246421794e-10,
      "loss": 0.5571,
      "step": 6623
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6291524271108334,
      "learning_rate": 2.333577277235932e-10,
      "loss": 0.5896,
      "step": 6624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6031615438199496,
      "learning_rate": 2.012116179328505e-10,
      "loss": 0.5758,
      "step": 6625
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5688059902579582,
      "learning_rate": 1.714466707469775e-10,
      "loss": 0.5482,
      "step": 6626
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6285740424710093,
      "learning_rate": 1.4406289325252788e-10,
      "loss": 0.5697,
      "step": 6627
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6287508057963594,
      "learning_rate": 1.190602919709516e-10,
      "loss": 0.5468,
      "step": 6628
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6067530603267308,
      "learning_rate": 9.643887285637476e-11,
      "loss": 0.5707,
      "step": 6629
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6621483845356202,
      "learning_rate": 7.619864129559951e-11,
      "loss": 0.5713,
      "step": 6630
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6218007882908773,
      "learning_rate": 5.833960210588352e-11,
      "loss": 0.5837,
      "step": 6631
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5991594628397772,
      "learning_rate": 4.286175954271166e-11,
      "loss": 0.5588,
      "step": 6632
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6241881797714341,
      "learning_rate": 2.976511729091414e-11,
      "loss": 0.5506,
      "step": 6633
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5497178699484356,
      "learning_rate": 1.9049678467997213e-11,
      "loss": 0.5539,
      "step": 6634
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7000850043906851,
      "learning_rate": 1.0715445626363618e-11,
      "loss": 0.534,
      "step": 6635
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6411799017222726,
      "learning_rate": 4.762420751092123e-12,
      "loss": 0.5601,
      "step": 6636
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6429039377347664,
      "learning_rate": 1.1906052588273044e-12,
      "loss": 0.5652,
      "step": 6637
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5835997914727289,
      "learning_rate": 0.0,
      "loss": 0.5641,
      "step": 6638
    },
    {
      "epoch": 1.0,
      "step": 6638,
      "total_flos": 0.0,
      "train_loss": 0.11632079654660847,
      "train_runtime": 10674.7287,
      "train_samples_per_second": 159.273,
      "train_steps_per_second": 0.622
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6638,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}