{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.20144203657455892,
  "eval_steps": 500,
  "global_step": 5001,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1379.3720991798398,
      "learning_rate": 2.6845637583892618e-08,
      "loss": 15.4268,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.8534206219106,
      "learning_rate": 5.3691275167785235e-08,
      "loss": 1.5295,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 1367.2156892086268,
      "learning_rate": 8.053691275167787e-08,
      "loss": 14.914,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 1356.0751967393617,
      "learning_rate": 1.0738255033557047e-07,
      "loss": 15.3003,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 1332.7461428880738,
      "learning_rate": 1.342281879194631e-07,
      "loss": 15.3399,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 1327.7087244291388,
      "learning_rate": 1.6107382550335574e-07,
      "loss": 15.6574,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 1357.2838160327776,
      "learning_rate": 1.8791946308724833e-07,
      "loss": 15.3619,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 1460.9880894767196,
      "learning_rate": 2.1476510067114094e-07,
      "loss": 14.8984,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.685025255387016,
      "learning_rate": 2.4161073825503355e-07,
      "loss": 1.6682,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 1433.944513190723,
      "learning_rate": 2.684563758389262e-07,
      "loss": 15.0011,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1302.7623411450743,
      "learning_rate": 2.9530201342281884e-07,
      "loss": 15.6583,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1345.8042090611784,
      "learning_rate": 3.221476510067115e-07,
      "loss": 15.4581,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1208.1489105112164,
      "learning_rate": 3.48993288590604e-07,
      "loss": 14.4632,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1122.3733969914717,
      "learning_rate": 3.7583892617449665e-07,
      "loss": 13.5493,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1036.9703513692707,
      "learning_rate": 4.026845637583893e-07,
      "loss": 13.0069,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 989.8774194879742,
      "learning_rate": 4.295302013422819e-07,
      "loss": 13.0841,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 1070.3499952623663,
      "learning_rate": 4.563758389261745e-07,
      "loss": 13.8049,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 1022.0857382159567,
      "learning_rate": 4.832214765100671e-07,
      "loss": 13.1312,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 922.0295847794329,
      "learning_rate": 5.100671140939598e-07,
      "loss": 12.2413,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 626.8057262987901,
      "learning_rate": 5.369127516778524e-07,
      "loss": 10.251,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 574.7762030056041,
      "learning_rate": 5.63758389261745e-07,
      "loss": 10.5845,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 621.5321839899642,
      "learning_rate": 5.906040268456377e-07,
      "loss": 10.1517,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 627.1730841493847,
      "learning_rate": 6.174496644295303e-07,
      "loss": 10.0554,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 510.36730831743716,
      "learning_rate": 6.44295302013423e-07,
      "loss": 10.2728,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 519.3995865128837,
      "learning_rate": 6.711409395973155e-07,
      "loss": 9.6979,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 457.2100248409554,
      "learning_rate": 6.97986577181208e-07,
      "loss": 8.404,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 333.7980179207978,
      "learning_rate": 7.248322147651007e-07,
      "loss": 7.0196,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 230.65095048133824,
      "learning_rate": 7.516778523489933e-07,
      "loss": 7.0535,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 202.45785726173838,
      "learning_rate": 7.78523489932886e-07,
      "loss": 7.3524,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 141.1784162971984,
      "learning_rate": 8.053691275167786e-07,
      "loss": 5.9181,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 160.36072995219004,
      "learning_rate": 8.322147651006713e-07,
      "loss": 6.7424,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 153.3983638890247,
      "learning_rate": 8.590604026845638e-07,
      "loss": 5.3766,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 160.0499921485999,
      "learning_rate": 8.859060402684565e-07,
      "loss": 5.067,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 144.8789271164616,
      "learning_rate": 9.12751677852349e-07,
      "loss": 5.6964,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 142.97309552434834,
      "learning_rate": 9.395973154362417e-07,
      "loss": 5.2783,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 120.28867130605691,
      "learning_rate": 9.664429530201342e-07,
      "loss": 5.8628,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 117.39289909868887,
      "learning_rate": 9.93288590604027e-07,
      "loss": 6.0932,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 109.1450812867474,
      "learning_rate": 1.0201342281879196e-06,
      "loss": 5.2595,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 119.76168539749345,
      "learning_rate": 1.0469798657718122e-06,
      "loss": 5.8302,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 138.6191466863621,
      "learning_rate": 1.0738255033557048e-06,
      "loss": 4.4134,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 104.40519142078718,
      "learning_rate": 1.1006711409395974e-06,
      "loss": 5.6586,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 70.673119264975,
      "learning_rate": 1.12751677852349e-06,
      "loss": 4.8559,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 88.61579685891138,
      "learning_rate": 1.1543624161073828e-06,
      "loss": 4.5503,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 55.93590312064801,
      "learning_rate": 1.1812080536912753e-06,
      "loss": 4.311,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 118.27309796259414,
      "learning_rate": 1.208053691275168e-06,
      "loss": 3.9019,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 78.85209279862124,
      "learning_rate": 1.2348993288590605e-06,
      "loss": 4.1337,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 98.12405486611367,
      "learning_rate": 1.2617449664429531e-06,
      "loss": 4.5948,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 63.959240832161754,
      "learning_rate": 1.288590604026846e-06,
      "loss": 3.5753,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 56.58840985353587,
      "learning_rate": 1.3154362416107383e-06,
      "loss": 3.6236,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 55.24870913585886,
      "learning_rate": 1.342281879194631e-06,
      "loss": 3.4079,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.87746896035972,
      "learning_rate": 1.3691275167785237e-06,
      "loss": 3.6738,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 51.97487694440137,
      "learning_rate": 1.395973154362416e-06,
      "loss": 3.8814,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 42.98929196517629,
      "learning_rate": 1.4228187919463088e-06,
      "loss": 3.3119,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 41.24350579976683,
      "learning_rate": 1.4496644295302014e-06,
      "loss": 3.0046,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 45.75625157306433,
      "learning_rate": 1.4765100671140942e-06,
      "loss": 3.3642,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 45.14172474761506,
      "learning_rate": 1.5033557046979866e-06,
      "loss": 3.3027,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 53.02674522833074,
      "learning_rate": 1.5302013422818792e-06,
      "loss": 3.4546,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.83432483620663,
      "learning_rate": 1.557046979865772e-06,
      "loss": 3.2131,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.0919898696877,
      "learning_rate": 1.5838926174496646e-06,
      "loss": 3.2642,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 42.87961228631677,
      "learning_rate": 1.6107382550335572e-06,
      "loss": 3.0634,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 42.971138864100254,
      "learning_rate": 1.6375838926174498e-06,
      "loss": 3.3706,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 48.06803919950474,
      "learning_rate": 1.6644295302013426e-06,
      "loss": 3.4155,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.48043635353635,
      "learning_rate": 1.6912751677852351e-06,
      "loss": 2.839,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.969839843504406,
      "learning_rate": 1.7181208053691275e-06,
      "loss": 2.7469,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.65704331081703,
      "learning_rate": 1.7449664429530203e-06,
      "loss": 2.6614,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.509605083858826,
      "learning_rate": 1.771812080536913e-06,
      "loss": 2.5327,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.797477758110855,
      "learning_rate": 1.7986577181208057e-06,
      "loss": 2.5899,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.33846675375741,
      "learning_rate": 1.825503355704698e-06,
      "loss": 2.9863,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.92341785904472,
      "learning_rate": 1.8523489932885907e-06,
      "loss": 2.7372,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.679791412739323,
      "learning_rate": 1.8791946308724835e-06,
      "loss": 3.014,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.13369507776243,
      "learning_rate": 1.906040268456376e-06,
      "loss": 2.602,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.02156210830726,
      "learning_rate": 1.9328859060402684e-06,
      "loss": 3.0853,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.70853474511914,
      "learning_rate": 1.959731543624161e-06,
      "loss": 2.1585,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.27251874476091,
      "learning_rate": 1.986577181208054e-06,
      "loss": 2.4983,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.32702895458006,
      "learning_rate": 2.013422818791946e-06,
      "loss": 2.4734,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.51841515770848,
      "learning_rate": 2.0402684563758392e-06,
      "loss": 2.162,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.064990372140535,
      "learning_rate": 2.067114093959732e-06,
      "loss": 2.3108,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.064437504570606,
      "learning_rate": 2.0939597315436244e-06,
      "loss": 2.9155,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.536350362321539,
      "learning_rate": 2.120805369127517e-06,
      "loss": 1.4456,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.978376246878465,
      "learning_rate": 2.1476510067114096e-06,
      "loss": 2.6396,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.562256155967653,
      "learning_rate": 2.174496644295302e-06,
      "loss": 2.215,
      "step": 81
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.77336038919066,
      "learning_rate": 2.2013422818791947e-06,
      "loss": 2.0772,
      "step": 82
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.049999307203677,
      "learning_rate": 2.2281879194630873e-06,
      "loss": 2.3279,
      "step": 83
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.24508052793857,
      "learning_rate": 2.25503355704698e-06,
      "loss": 1.4648,
      "step": 84
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.313229785568076,
      "learning_rate": 2.2818791946308725e-06,
      "loss": 2.7208,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.129347964745936,
      "learning_rate": 2.3087248322147655e-06,
      "loss": 1.9661,
      "step": 86
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.958667114185133,
      "learning_rate": 2.3355704697986577e-06,
      "loss": 2.2202,
      "step": 87
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.74421340201436,
      "learning_rate": 2.3624161073825507e-06,
      "loss": 2.124,
      "step": 88
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.638125408192659,
      "learning_rate": 2.3892617449664433e-06,
      "loss": 1.4054,
      "step": 89
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.547612050266068,
      "learning_rate": 2.416107382550336e-06,
      "loss": 1.8086,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.728919128173345,
      "learning_rate": 2.4429530201342285e-06,
      "loss": 2.1664,
      "step": 91
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.818225129062817,
      "learning_rate": 2.469798657718121e-06,
      "loss": 2.0583,
      "step": 92
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.987351354977292,
      "learning_rate": 2.4966442953020136e-06,
      "loss": 2.2603,
      "step": 93
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.296590866071373,
      "learning_rate": 2.5234899328859062e-06,
      "loss": 2.1282,
      "step": 94
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.214574496499722,
      "learning_rate": 2.5503355704697992e-06,
      "loss": 2.0341,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.256988084631885,
      "learning_rate": 2.577181208053692e-06,
      "loss": 2.1784,
      "step": 96
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.750386975252132,
      "learning_rate": 2.604026845637584e-06,
      "loss": 1.7499,
      "step": 97
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.10847467657068,
      "learning_rate": 2.6308724832214766e-06,
      "loss": 1.9537,
      "step": 98
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.306771416543913,
      "learning_rate": 2.657718120805369e-06,
      "loss": 2.0663,
      "step": 99
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.84553552279815,
      "learning_rate": 2.684563758389262e-06,
      "loss": 2.1094,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.510941228250555,
      "learning_rate": 2.7114093959731548e-06,
      "loss": 2.0442,
      "step": 101
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.797589490096449,
      "learning_rate": 2.7382550335570473e-06,
      "loss": 1.7963,
      "step": 102
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.723040840925055,
      "learning_rate": 2.76510067114094e-06,
      "loss": 2.0927,
      "step": 103
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.56979983597794,
      "learning_rate": 2.791946308724832e-06,
      "loss": 1.9796,
      "step": 104
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.38876629404607,
      "learning_rate": 2.8187919463087247e-06,
      "loss": 1.9783,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.199703074114808,
      "learning_rate": 2.8456375838926177e-06,
      "loss": 1.9211,
      "step": 106
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.789568611069523,
      "learning_rate": 2.8724832214765103e-06,
      "loss": 1.767,
      "step": 107
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.58163221726849,
      "learning_rate": 2.899328859060403e-06,
      "loss": 1.9911,
      "step": 108
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.91176329218805,
      "learning_rate": 2.9261744966442955e-06,
      "loss": 1.8932,
      "step": 109
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.956470177399684,
      "learning_rate": 2.9530201342281885e-06,
      "loss": 1.8611,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.733270877123367,
      "learning_rate": 2.979865771812081e-06,
      "loss": 2.0041,
      "step": 111
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.18002295197271,
      "learning_rate": 3.0067114093959732e-06,
      "loss": 1.5112,
      "step": 112
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.70936859670564,
      "learning_rate": 3.033557046979866e-06,
      "loss": 2.4545,
      "step": 113
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.926470122370823,
      "learning_rate": 3.0604026845637584e-06,
      "loss": 1.7448,
      "step": 114
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.943350791797702,
      "learning_rate": 3.0872483221476514e-06,
      "loss": 1.9604,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.197824441019495,
      "learning_rate": 3.114093959731544e-06,
      "loss": 2.0546,
      "step": 116
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.3138879828343,
      "learning_rate": 3.1409395973154366e-06,
      "loss": 1.2785,
      "step": 117
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.35395249592559,
      "learning_rate": 3.167785234899329e-06,
      "loss": 2.7419,
      "step": 118
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.69111880960645,
      "learning_rate": 3.194630872483222e-06,
      "loss": 1.6434,
      "step": 119
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.982455849431444,
      "learning_rate": 3.2214765100671143e-06,
      "loss": 2.1976,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.518962918485057,
      "learning_rate": 3.248322147651007e-06,
      "loss": 1.5123,
      "step": 121
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.064316885133437,
      "learning_rate": 3.2751677852348995e-06,
      "loss": 2.206,
      "step": 122
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.196567443999271,
      "learning_rate": 3.302013422818792e-06,
      "loss": 1.4922,
      "step": 123
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.813290931368686,
      "learning_rate": 3.328859060402685e-06,
      "loss": 1.1576,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.643274264416528,
      "learning_rate": 3.3557046979865777e-06,
      "loss": 1.9169,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.764007126181044,
      "learning_rate": 3.3825503355704703e-06,
      "loss": 1.8815,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.096268086959679,
      "learning_rate": 3.4093959731543625e-06,
      "loss": 1.567,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.250938395234062,
      "learning_rate": 3.436241610738255e-06,
      "loss": 1.668,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.573891263834847,
      "learning_rate": 3.4630872483221476e-06,
      "loss": 1.6717,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.141761954303966,
      "learning_rate": 3.4899328859060407e-06,
      "loss": 1.7085,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.860055788677856,
      "learning_rate": 3.5167785234899332e-06,
      "loss": 1.8231,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.50483829259475,
      "learning_rate": 3.543624161073826e-06,
      "loss": 2.04,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.340768519036331,
      "learning_rate": 3.5704697986577184e-06,
      "loss": 1.4616,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.286266651787656,
      "learning_rate": 3.5973154362416114e-06,
      "loss": 1.4944,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.223707110673606,
      "learning_rate": 3.6241610738255036e-06,
      "loss": 1.6233,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.48345026987458,
      "learning_rate": 3.651006711409396e-06,
      "loss": 1.9901,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.26552789758307,
      "learning_rate": 3.6778523489932888e-06,
      "loss": 2.0787,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.72525228652948,
      "learning_rate": 3.7046979865771814e-06,
      "loss": 2.0334,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.780049120791178,
      "learning_rate": 3.7315436241610744e-06,
      "loss": 1.6787,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.546074380485582,
      "learning_rate": 3.758389261744967e-06,
      "loss": 1.8702,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.709097755903235,
      "learning_rate": 3.7852348993288595e-06,
      "loss": 1.2595,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.063162323488752,
      "learning_rate": 3.812080536912752e-06,
      "loss": 1.784,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.957625227865835,
      "learning_rate": 3.838926174496644e-06,
      "loss": 1.8175,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.085079054839767,
      "learning_rate": 3.865771812080537e-06,
      "loss": 1.6911,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.31556562796878,
      "learning_rate": 3.8926174496644295e-06,
      "loss": 1.3445,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.540520671108675,
      "learning_rate": 3.919463087248322e-06,
      "loss": 1.5777,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.576257909876029,
      "learning_rate": 3.9463087248322155e-06,
      "loss": 1.5895,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.052530982896844,
      "learning_rate": 3.973154362416108e-06,
      "loss": 1.6578,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.71377622166453,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.8639,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.918372148849897,
      "learning_rate": 4.026845637583892e-06,
      "loss": 1.4887,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.821435242090589,
      "learning_rate": 4.053691275167785e-06,
      "loss": 1.8509,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.446613557133626,
      "learning_rate": 4.0805369127516784e-06,
      "loss": 1.643,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.579093564838542,
      "learning_rate": 4.107382550335571e-06,
      "loss": 1.339,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.299064293707799,
      "learning_rate": 4.134228187919464e-06,
      "loss": 1.8641,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.139099152761217,
      "learning_rate": 4.161073825503356e-06,
      "loss": 1.6654,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.686682477529438,
      "learning_rate": 4.187919463087249e-06,
      "loss": 1.412,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.51065934826115,
      "learning_rate": 4.214765100671141e-06,
      "loss": 2.0085,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.265061302846355,
      "learning_rate": 4.241610738255034e-06,
      "loss": 2.324,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.807505565522677,
      "learning_rate": 4.2684563758389265e-06,
      "loss": 1.7117,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.153231537480185,
      "learning_rate": 4.295302013422819e-06,
      "loss": 1.9677,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.489437743942664,
      "learning_rate": 4.322147651006712e-06,
      "loss": 1.9075,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.627439994995036,
      "learning_rate": 4.348993288590604e-06,
      "loss": 1.7908,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.314315042757203,
      "learning_rate": 4.375838926174497e-06,
      "loss": 1.628,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.894482362082506,
      "learning_rate": 4.4026845637583895e-06,
      "loss": 1.6236,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.345393611291524,
      "learning_rate": 4.429530201342283e-06,
      "loss": 2.0266,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.603244940585046,
      "learning_rate": 4.456375838926175e-06,
      "loss": 1.5495,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.345492404743105,
      "learning_rate": 4.483221476510067e-06,
      "loss": 1.4911,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.440021844240855,
      "learning_rate": 4.51006711409396e-06,
      "loss": 1.8804,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.531648439709466,
      "learning_rate": 4.536912751677852e-06,
      "loss": 1.697,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.08337602148859,
      "learning_rate": 4.563758389261745e-06,
      "loss": 1.5035,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.535791614438475,
      "learning_rate": 4.5906040268456384e-06,
      "loss": 1.8912,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.881529290309787,
      "learning_rate": 4.617449664429531e-06,
      "loss": 1.6131,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.17970250569803,
      "learning_rate": 4.644295302013423e-06,
      "loss": 1.7546,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.143087308647942,
      "learning_rate": 4.671140939597315e-06,
      "loss": 1.5703,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.28331408382497,
      "learning_rate": 4.697986577181208e-06,
      "loss": 2.0012,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.884699065314573,
      "learning_rate": 4.724832214765101e-06,
      "loss": 1.5294,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.722896782080753,
      "learning_rate": 4.751677852348994e-06,
      "loss": 1.5955,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.347335472195118,
      "learning_rate": 4.7785234899328866e-06,
      "loss": 1.6365,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.851492785664655,
      "learning_rate": 4.805369127516779e-06,
      "loss": 1.9486,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.971523057766166,
      "learning_rate": 4.832214765100672e-06,
      "loss": 1.8916,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.29731024397963,
      "learning_rate": 4.859060402684564e-06,
      "loss": 1.5239,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.239374412559837,
      "learning_rate": 4.885906040268457e-06,
      "loss": 1.6987,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.579998991834582,
      "learning_rate": 4.9127516778523495e-06,
      "loss": 1.7507,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.827398285369991,
      "learning_rate": 4.939597315436242e-06,
      "loss": 1.8709,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.180745238261542,
      "learning_rate": 4.966442953020135e-06,
      "loss": 1.4853,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.77188804594341,
      "learning_rate": 4.993288590604027e-06,
      "loss": 1.2317,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.764297684473298,
      "learning_rate": 5.02013422818792e-06,
      "loss": 1.529,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.410177769159763,
      "learning_rate": 5.0469798657718124e-06,
      "loss": 1.2431,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.296675687538578,
      "learning_rate": 5.073825503355705e-06,
      "loss": 1.4392,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.151768475375025,
      "learning_rate": 5.1006711409395985e-06,
      "loss": 1.7412,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.601085935647076,
      "learning_rate": 5.12751677852349e-06,
      "loss": 2.1132,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.31842403632272,
      "learning_rate": 5.154362416107384e-06,
      "loss": 1.5016,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.061021935372647,
      "learning_rate": 5.181208053691275e-06,
      "loss": 1.8035,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.943096562508227,
      "learning_rate": 5.208053691275168e-06,
      "loss": 1.6704,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.91609745320432,
      "learning_rate": 5.234899328859061e-06,
      "loss": 1.4274,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.043075066679481,
      "learning_rate": 5.261744966442953e-06,
      "loss": 1.5762,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.17379490776072,
      "learning_rate": 5.2885906040268466e-06,
      "loss": 1.6552,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.652944866848367,
      "learning_rate": 5.315436241610738e-06,
      "loss": 1.516,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.061365814480085,
      "learning_rate": 5.342281879194632e-06,
      "loss": 1.6705,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.024462008727843,
      "learning_rate": 5.369127516778524e-06,
      "loss": 1.7595,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.35316912645751,
      "learning_rate": 5.395973154362416e-06,
      "loss": 1.4622,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.825025297260247,
      "learning_rate": 5.4228187919463095e-06,
      "loss": 2.0792,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.765515690477317,
      "learning_rate": 5.449664429530201e-06,
      "loss": 1.6338,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.8581687685068,
      "learning_rate": 5.476510067114095e-06,
      "loss": 1.7885,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.189091030248958,
      "learning_rate": 5.503355704697987e-06,
      "loss": 1.828,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.667264981245788,
      "learning_rate": 5.53020134228188e-06,
      "loss": 1.3538,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.86597105681116,
      "learning_rate": 5.5570469798657725e-06,
      "loss": 1.6074,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.725591120764053,
      "learning_rate": 5.583892617449664e-06,
      "loss": 1.4364,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.989212120333164,
      "learning_rate": 5.610738255033558e-06,
      "loss": 1.4904,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.446222742017182,
      "learning_rate": 5.637583892617449e-06,
      "loss": 1.4446,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8196189133302445,
      "learning_rate": 5.664429530201343e-06,
      "loss": 1.155,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.57547711753834,
      "learning_rate": 5.691275167785235e-06,
      "loss": 1.764,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.485919062651456,
      "learning_rate": 5.718120805369128e-06,
      "loss": 1.3615,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.24774036826691,
      "learning_rate": 5.7449664429530206e-06,
      "loss": 1.3537,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.484321234211297,
      "learning_rate": 5.771812080536914e-06,
      "loss": 1.6412,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.507940120152089,
      "learning_rate": 5.798657718120806e-06,
      "loss": 1.3493,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.754718070726947,
      "learning_rate": 5.825503355704698e-06,
      "loss": 1.8523,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.401522533753987,
      "learning_rate": 5.852348993288591e-06,
      "loss": 1.6014,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.353450906498342,
      "learning_rate": 5.8791946308724835e-06,
      "loss": 1.5356,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.596158219614574,
      "learning_rate": 5.906040268456377e-06,
      "loss": 1.5215,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.028007652005108,
      "learning_rate": 5.932885906040269e-06,
      "loss": 1.5457,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.60145639881299,
      "learning_rate": 5.959731543624162e-06,
      "loss": 1.5872,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.957821931272354,
      "learning_rate": 5.986577181208054e-06,
      "loss": 1.4347,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.133189588283791,
      "learning_rate": 6.0134228187919464e-06,
      "loss": 1.7833,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.41756147762758,
      "learning_rate": 6.04026845637584e-06,
      "loss": 1.8138,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.610841931715774,
      "learning_rate": 6.067114093959732e-06,
      "loss": 1.8726,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.495916536168222,
      "learning_rate": 6.093959731543625e-06,
      "loss": 1.7612,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.156497571691345,
      "learning_rate": 6.120805369127517e-06,
      "loss": 1.51,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.983284316136041,
      "learning_rate": 6.14765100671141e-06,
      "loss": 1.4645,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.12661480251222,
      "learning_rate": 6.174496644295303e-06,
      "loss": 1.658,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.552454706231153,
      "learning_rate": 6.2013422818791946e-06,
      "loss": 1.2123,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.123196858785338,
      "learning_rate": 6.228187919463088e-06,
      "loss": 1.7071,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.580620826008001,
      "learning_rate": 6.25503355704698e-06,
      "loss": 1.3564,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.263579162132157,
      "learning_rate": 6.281879194630873e-06,
      "loss": 1.6229,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.598528740649439,
      "learning_rate": 6.308724832214766e-06,
      "loss": 2.0826,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.671772121388846,
      "learning_rate": 6.335570469798658e-06,
      "loss": 1.9092,
      "step": 236
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.134116326125955,
      "learning_rate": 6.362416107382551e-06,
      "loss": 1.1773,
      "step": 237
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.00806840581287,
      "learning_rate": 6.389261744966444e-06,
      "loss": 1.2386,
      "step": 238
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.6349220086916,
      "learning_rate": 6.416107382550336e-06,
      "loss": 1.2972,
      "step": 239
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.870003388516592,
      "learning_rate": 6.442953020134229e-06,
      "loss": 1.4886,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.703293154317208,
      "learning_rate": 6.469798657718121e-06,
      "loss": 1.5812,
      "step": 241
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.477762977854425,
      "learning_rate": 6.496644295302014e-06,
      "loss": 2.0074,
      "step": 242
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.364091339313735,
      "learning_rate": 6.523489932885907e-06,
      "loss": 1.5767,
      "step": 243
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.061778043450656,
      "learning_rate": 6.550335570469799e-06,
      "loss": 1.1556,
      "step": 244
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.235527266761915,
      "learning_rate": 6.5771812080536925e-06,
      "loss": 1.5016,
      "step": 245
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.76044715570968,
      "learning_rate": 6.604026845637584e-06,
      "loss": 1.6558,
      "step": 246
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.211734022710534,
      "learning_rate": 6.630872483221477e-06,
      "loss": 2.0527,
      "step": 247
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.4877471188776,
      "learning_rate": 6.65771812080537e-06,
      "loss": 1.7114,
      "step": 248
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.952922265439943,
      "learning_rate": 6.684563758389262e-06,
      "loss": 1.4065,
      "step": 249
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.752231909907449,
      "learning_rate": 6.711409395973155e-06,
      "loss": 1.7585,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.964973395053712,
      "learning_rate": 6.738255033557047e-06,
      "loss": 1.758,
      "step": 251
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.964169821449593,
      "learning_rate": 6.765100671140941e-06,
      "loss": 1.1376,
      "step": 252
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.68719837794281,
      "learning_rate": 6.791946308724832e-06,
      "loss": 1.282,
      "step": 253
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.346014460083014,
      "learning_rate": 6.818791946308725e-06,
      "loss": 1.4305,
      "step": 254
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.393880235366709,
      "learning_rate": 6.845637583892618e-06,
      "loss": 1.4726,
      "step": 255
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.574973700346327,
      "learning_rate": 6.87248322147651e-06,
      "loss": 1.5764,
      "step": 256
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.5889622707587,
      "learning_rate": 6.8993288590604035e-06,
      "loss": 1.5886,
      "step": 257
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.053009784693288,
      "learning_rate": 6.926174496644295e-06,
      "loss": 1.3649,
      "step": 258
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.569822734540786,
      "learning_rate": 6.953020134228189e-06,
      "loss": 1.7326,
      "step": 259
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.014095218785567,
      "learning_rate": 6.979865771812081e-06,
      "loss": 1.3926,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.626409654072866,
      "learning_rate": 7.006711409395974e-06,
      "loss": 1.909,
      "step": 261
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.946460199204028,
      "learning_rate": 7.0335570469798665e-06,
      "loss": 1.5285,
      "step": 262
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.367394362597063,
      "learning_rate": 7.060402684563758e-06,
      "loss": 1.6252,
      "step": 263
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.128127598162862,
      "learning_rate": 7.087248322147652e-06,
      "loss": 1.7979,
      "step": 264
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.493707412532908,
      "learning_rate": 7.114093959731544e-06,
      "loss": 1.4242,
      "step": 265
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.54427329173879,
      "learning_rate": 7.140939597315437e-06,
      "loss": 1.9175,
      "step": 266
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.302456341955967,
      "learning_rate": 7.167785234899329e-06,
      "loss": 1.354,
      "step": 267
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.965065298588401,
      "learning_rate": 7.194630872483223e-06,
      "loss": 1.8103,
      "step": 268
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.773296535998654,
      "learning_rate": 7.221476510067115e-06,
      "loss": 1.2827,
      "step": 269
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.965233599793144,
      "learning_rate": 7.248322147651007e-06,
      "loss": 1.6364,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.18406341014796,
      "learning_rate": 7.2751677852349e-06,
      "loss": 1.2708,
      "step": 271
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.635674408248562,
      "learning_rate": 7.302013422818792e-06,
      "loss": 1.4547,
      "step": 272
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.059144232433749,
      "learning_rate": 7.328859060402686e-06,
      "loss": 1.4559,
      "step": 273
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.207189326824452,
      "learning_rate": 7.3557046979865775e-06,
      "loss": 1.9118,
      "step": 274
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.658580091519708,
      "learning_rate": 7.382550335570471e-06,
      "loss": 1.8017,
      "step": 275
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.913271685246535,
      "learning_rate": 7.409395973154363e-06,
      "loss": 1.5528,
      "step": 276
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.383906560364299,
      "learning_rate": 7.436241610738255e-06,
      "loss": 1.3903,
      "step": 277
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.784505758233623,
      "learning_rate": 7.463087248322149e-06,
      "loss": 1.5208,
      "step": 278
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.79469433278323,
      "learning_rate": 7.4899328859060405e-06,
      "loss": 1.6199,
      "step": 279
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.56381120904719,
      "learning_rate": 7.516778523489934e-06,
      "loss": 1.3765,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.094434540039053,
      "learning_rate": 7.543624161073826e-06,
      "loss": 1.4641,
      "step": 281
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.34365961384772,
      "learning_rate": 7.570469798657719e-06,
      "loss": 2.1032,
      "step": 282
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.035124596658779,
      "learning_rate": 7.597315436241612e-06,
      "loss": 1.5777,
      "step": 283
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.997683545546973,
      "learning_rate": 7.624161073825504e-06,
      "loss": 1.6408,
      "step": 284
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.065155618120626,
      "learning_rate": 7.651006711409396e-06,
      "loss": 1.4344,
      "step": 285
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.361079722623494,
      "learning_rate": 7.677852348993289e-06,
      "loss": 1.4586,
      "step": 286
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.695438794503787,
      "learning_rate": 7.704697986577183e-06,
      "loss": 1.6616,
      "step": 287
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.299179387335887,
      "learning_rate": 7.731543624161074e-06,
      "loss": 1.785,
      "step": 288
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.568308646025505,
      "learning_rate": 7.758389261744968e-06,
      "loss": 1.4455,
      "step": 289
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.732738386479873,
      "learning_rate": 7.785234899328859e-06,
      "loss": 1.5233,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.833384094395637,
      "learning_rate": 7.812080536912753e-06,
      "loss": 1.4055,
      "step": 291
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.54529072920182,
      "learning_rate": 7.838926174496644e-06,
      "loss": 1.1067,
      "step": 292
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.48750810634312,
      "learning_rate": 7.865771812080537e-06,
      "loss": 2.0899,
      "step": 293
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.906559405220005,
      "learning_rate": 7.892617449664431e-06,
      "loss": 1.6334,
      "step": 294
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.699846895401677,
      "learning_rate": 7.919463087248322e-06,
      "loss": 1.5975,
      "step": 295
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.39621817297778,
      "learning_rate": 7.946308724832216e-06,
      "loss": 1.5089,
      "step": 296
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.789277280291532,
      "learning_rate": 7.973154362416107e-06,
      "loss": 2.038,
      "step": 297
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.295142628075283,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2353,
      "step": 298
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.807328434768754,
      "learning_rate": 8.026845637583894e-06,
      "loss": 1.5409,
      "step": 299
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.473396773857809,
      "learning_rate": 8.053691275167785e-06,
      "loss": 1.6243,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.904258622721057,
      "learning_rate": 8.080536912751679e-06,
      "loss": 2.04,
      "step": 301
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.673097454710115,
      "learning_rate": 8.10738255033557e-06,
      "loss": 1.2377,
      "step": 302
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.044854709390615,
      "learning_rate": 8.134228187919464e-06,
      "loss": 1.7277,
      "step": 303
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.900336713035701,
      "learning_rate": 8.161073825503357e-06,
      "loss": 1.5998,
      "step": 304
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.174609108609152,
      "learning_rate": 8.18791946308725e-06,
      "loss": 1.7433,
      "step": 305
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.844639630043693,
      "learning_rate": 8.214765100671142e-06,
      "loss": 1.6152,
      "step": 306
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.139723228716852,
      "learning_rate": 8.241610738255035e-06,
      "loss": 1.6333,
      "step": 307
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.727332261052586,
      "learning_rate": 8.268456375838927e-06,
      "loss": 1.5606,
      "step": 308
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.663628255979177,
      "learning_rate": 8.29530201342282e-06,
      "loss": 1.7879,
      "step": 309
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.29019286777234,
      "learning_rate": 8.322147651006712e-06,
      "loss": 1.5749,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.775092407800736,
      "learning_rate": 8.348993288590605e-06,
      "loss": 1.5944,
      "step": 311
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.67006642374828,
      "learning_rate": 8.375838926174498e-06,
      "loss": 1.6951,
      "step": 312
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.313097629781833,
      "learning_rate": 8.40268456375839e-06,
      "loss": 1.119,
      "step": 313
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.740240080737678,
      "learning_rate": 8.429530201342283e-06,
      "loss": 1.5149,
      "step": 314
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.340689028123274,
      "learning_rate": 8.456375838926175e-06,
      "loss": 1.9048,
      "step": 315
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.273432871141575,
      "learning_rate": 8.483221476510068e-06,
      "loss": 1.6202,
      "step": 316
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.644259856017797,
      "learning_rate": 8.51006711409396e-06,
      "loss": 1.6766,
      "step": 317
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.657875111729533,
      "learning_rate": 8.536912751677853e-06,
      "loss": 1.9949,
      "step": 318
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.246742026987623,
      "learning_rate": 8.563758389261746e-06,
      "loss": 1.1047,
      "step": 319
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.18864242956882,
      "learning_rate": 8.590604026845638e-06,
      "loss": 1.2312,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.090116088685697,
      "learning_rate": 8.617449664429531e-06,
      "loss": 1.5047,
      "step": 321
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.564187038931198,
      "learning_rate": 8.644295302013423e-06,
      "loss": 1.4331,
      "step": 322
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.496333819942201,
      "learning_rate": 8.671140939597316e-06,
      "loss": 1.5035,
      "step": 323
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.272267597231734,
      "learning_rate": 8.697986577181209e-06,
      "loss": 1.2826,
      "step": 324
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.576186066897877,
      "learning_rate": 8.724832214765101e-06,
      "loss": 1.7093,
      "step": 325
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.864505869737435,
      "learning_rate": 8.751677852348994e-06,
      "loss": 1.3858,
      "step": 326
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.182285214380252,
      "learning_rate": 8.778523489932886e-06,
      "loss": 1.2842,
      "step": 327
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.14634103815586,
      "learning_rate": 8.805369127516779e-06,
      "loss": 1.1746,
      "step": 328
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.688655254608909,
      "learning_rate": 8.832214765100672e-06,
      "loss": 1.5103,
      "step": 329
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.554938803093232,
      "learning_rate": 8.859060402684566e-06,
      "loss": 1.2835,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.255985803647404,
      "learning_rate": 8.885906040268457e-06,
      "loss": 1.9845,
      "step": 331
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.181282883855086,
      "learning_rate": 8.91275167785235e-06,
      "loss": 1.661,
      "step": 332
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.36346818139929,
      "learning_rate": 8.939597315436242e-06,
      "loss": 1.5244,
      "step": 333
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.388853705491451,
      "learning_rate": 8.966442953020134e-06,
      "loss": 1.4875,
      "step": 334
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.737028575065583,
      "learning_rate": 8.993288590604027e-06,
      "loss": 1.4506,
      "step": 335
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.432560448611623,
      "learning_rate": 9.02013422818792e-06,
      "loss": 1.6521,
      "step": 336
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.427772696312779,
      "learning_rate": 9.046979865771814e-06,
      "loss": 1.3052,
      "step": 337
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.344212539496946,
      "learning_rate": 9.073825503355705e-06,
      "loss": 1.4408,
      "step": 338
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.725747313321792,
      "learning_rate": 9.100671140939597e-06,
      "loss": 1.8887,
      "step": 339
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.311809411600889,
      "learning_rate": 9.12751677852349e-06,
      "loss": 1.3505,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.558711767623944,
      "learning_rate": 9.154362416107383e-06,
      "loss": 1.3347,
      "step": 341
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.543979641674689,
      "learning_rate": 9.181208053691277e-06,
      "loss": 1.8786,
      "step": 342
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.767906823118505,
      "learning_rate": 9.208053691275168e-06,
      "loss": 1.6337,
      "step": 343
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.349390558250093,
      "learning_rate": 9.234899328859062e-06,
      "loss": 1.2518,
      "step": 344
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.295244084871817,
      "learning_rate": 9.261744966442953e-06,
      "loss": 1.646,
      "step": 345
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.067964926192612,
      "learning_rate": 9.288590604026846e-06,
      "loss": 1.5712,
      "step": 346
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.988413240516904,
      "learning_rate": 9.31543624161074e-06,
      "loss": 1.8023,
      "step": 347
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.725082292402449,
      "learning_rate": 9.34228187919463e-06,
      "loss": 1.3904,
      "step": 348
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.538385697186461,
      "learning_rate": 9.369127516778525e-06,
      "loss": 1.2176,
      "step": 349
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.954462854500267,
      "learning_rate": 9.395973154362416e-06,
      "loss": 1.3867,
      "step": 350
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.69562591117132,
      "learning_rate": 9.42281879194631e-06,
      "loss": 1.2014,
      "step": 351
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.907438289910699,
      "learning_rate": 9.449664429530203e-06,
      "loss": 1.3065,
      "step": 352
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.331833282275065,
      "learning_rate": 9.476510067114095e-06,
      "loss": 1.8804,
      "step": 353
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.335076987176812,
      "learning_rate": 9.503355704697988e-06,
      "loss": 1.4837,
      "step": 354
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.267759303994664,
      "learning_rate": 9.530201342281879e-06,
      "loss": 1.7094,
      "step": 355
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.952949430161986,
      "learning_rate": 9.557046979865773e-06,
      "loss": 1.4585,
      "step": 356
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.981061721370573,
      "learning_rate": 9.583892617449666e-06,
      "loss": 1.6428,
      "step": 357
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.750352290003713,
      "learning_rate": 9.610738255033558e-06,
      "loss": 1.4157,
      "step": 358
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.537089893461811,
      "learning_rate": 9.637583892617451e-06,
      "loss": 1.5694,
      "step": 359
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.721233187634588,
      "learning_rate": 9.664429530201343e-06,
      "loss": 1.428,
      "step": 360
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.894043355202989,
      "learning_rate": 9.691275167785236e-06,
      "loss": 1.2399,
      "step": 361
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.89419547831076,
      "learning_rate": 9.718120805369129e-06,
      "loss": 1.4593,
      "step": 362
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.444037192684798,
      "learning_rate": 9.744966442953021e-06,
      "loss": 1.84,
      "step": 363
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.6825192853110265,
      "learning_rate": 9.771812080536914e-06,
      "loss": 1.291,
      "step": 364
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.913625663843959,
      "learning_rate": 9.798657718120806e-06,
      "loss": 1.4084,
      "step": 365
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.955604322140983,
      "learning_rate": 9.825503355704699e-06,
      "loss": 1.7061,
      "step": 366
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.62089959227162,
      "learning_rate": 9.852348993288592e-06,
      "loss": 1.7951,
      "step": 367
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.29940556199203,
      "learning_rate": 9.879194630872484e-06,
      "loss": 1.644,
      "step": 368
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.888406606746397,
      "learning_rate": 9.906040268456377e-06,
      "loss": 1.4007,
      "step": 369
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.324435974424938,
      "learning_rate": 9.93288590604027e-06,
      "loss": 1.5139,
      "step": 370
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.092469886188237,
      "learning_rate": 9.959731543624162e-06,
      "loss": 1.658,
      "step": 371
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.798239784673768,
      "learning_rate": 9.986577181208055e-06,
      "loss": 1.5002,
      "step": 372
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.174780869324398,
      "learning_rate": 1.0013422818791947e-05,
      "loss": 1.9242,
      "step": 373
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.21923255015902,
      "learning_rate": 1.004026845637584e-05,
      "loss": 1.3241,
      "step": 374
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.46030760835296,
      "learning_rate": 1.0067114093959734e-05,
      "loss": 1.1467,
      "step": 375
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.104507369826752,
      "learning_rate": 1.0093959731543625e-05,
      "loss": 1.4861,
      "step": 376
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.82452396622584,
      "learning_rate": 1.0120805369127517e-05,
      "loss": 1.2164,
      "step": 377
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.237099080996998,
      "learning_rate": 1.014765100671141e-05,
      "loss": 1.7492,
      "step": 378
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.650935437381616,
      "learning_rate": 1.0174496644295303e-05,
      "loss": 1.1814,
      "step": 379
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.035852084864297,
      "learning_rate": 1.0201342281879197e-05,
      "loss": 1.6595,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.41862935575625,
      "learning_rate": 1.0228187919463088e-05,
      "loss": 1.4773,
      "step": 381
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.095710098628684,
      "learning_rate": 1.025503355704698e-05,
      "loss": 1.7986,
      "step": 382
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.909356812828454,
      "learning_rate": 1.0281879194630873e-05,
      "loss": 1.4372,
      "step": 383
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.819748827331054,
      "learning_rate": 1.0308724832214767e-05,
      "loss": 1.5856,
      "step": 384
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.327594488710123,
      "learning_rate": 1.033557046979866e-05,
      "loss": 1.826,
      "step": 385
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.13503962517979,
      "learning_rate": 1.036241610738255e-05,
      "loss": 1.6619,
      "step": 386
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.377050703898767,
      "learning_rate": 1.0389261744966443e-05,
      "loss": 1.2555,
      "step": 387
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.072628247604198,
      "learning_rate": 1.0416107382550336e-05,
      "loss": 1.5963,
      "step": 388
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.297840978482256,
      "learning_rate": 1.044295302013423e-05,
      "loss": 1.3466,
      "step": 389
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.414736317088886,
      "learning_rate": 1.0469798657718123e-05,
      "loss": 1.5266,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.46327313730743,
      "learning_rate": 1.0496644295302014e-05,
      "loss": 1.3284,
      "step": 391
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.251013017465008,
      "learning_rate": 1.0523489932885906e-05,
      "loss": 1.2931,
      "step": 392
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.578294615207088,
      "learning_rate": 1.0550335570469799e-05,
      "loss": 1.6163,
      "step": 393
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.911961559019799,
      "learning_rate": 1.0577181208053693e-05,
      "loss": 1.3804,
      "step": 394
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.316430892253736,
      "learning_rate": 1.0604026845637586e-05,
      "loss": 1.3293,
      "step": 395
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.383118553593106,
      "learning_rate": 1.0630872483221477e-05,
      "loss": 1.3164,
      "step": 396
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.844864708845973,
      "learning_rate": 1.065771812080537e-05,
      "loss": 1.4031,
      "step": 397
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.950385801454448,
      "learning_rate": 1.0684563758389264e-05,
      "loss": 1.4896,
      "step": 398
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.393193707835785,
      "learning_rate": 1.0711409395973156e-05,
      "loss": 1.3456,
      "step": 399
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.909450442572261,
      "learning_rate": 1.0738255033557049e-05,
      "loss": 1.4749,
      "step": 400
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.082460375776544,
      "learning_rate": 1.076510067114094e-05,
      "loss": 1.284,
      "step": 401
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.675805571191248,
      "learning_rate": 1.0791946308724832e-05,
      "loss": 1.5959,
      "step": 402
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.12779406435023,
      "learning_rate": 1.0818791946308726e-05,
      "loss": 1.8493,
      "step": 403
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.335440232238135,
      "learning_rate": 1.0845637583892619e-05,
      "loss": 1.6222,
      "step": 404
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.573394308412109,
      "learning_rate": 1.0872483221476512e-05,
      "loss": 1.4336,
      "step": 405
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.977355478954481,
      "learning_rate": 1.0899328859060403e-05,
      "loss": 1.5827,
      "step": 406
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.696029088146323,
      "learning_rate": 1.0926174496644297e-05,
      "loss": 1.924,
      "step": 407
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.201859611500188,
      "learning_rate": 1.095302013422819e-05,
      "loss": 1.5042,
      "step": 408
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.00011203211513,
      "learning_rate": 1.0979865771812082e-05,
      "loss": 1.3995,
      "step": 409
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.551486110298208,
      "learning_rate": 1.1006711409395975e-05,
      "loss": 1.6962,
      "step": 410
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.675182744412622,
      "learning_rate": 1.1033557046979865e-05,
      "loss": 1.0448,
      "step": 411
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.097196057725439,
      "learning_rate": 1.106040268456376e-05,
      "loss": 1.4764,
      "step": 412
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.32493336522974,
      "learning_rate": 1.1087248322147652e-05,
      "loss": 1.6457,
      "step": 413
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.253346597613184,
      "learning_rate": 1.1114093959731545e-05,
      "loss": 1.2916,
      "step": 414
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.323028106691497,
      "learning_rate": 1.1140939597315436e-05,
      "loss": 1.3274,
      "step": 415
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.09821426153491,
      "learning_rate": 1.1167785234899328e-05,
      "loss": 1.6463,
      "step": 416
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.277224133702717,
      "learning_rate": 1.1194630872483223e-05,
      "loss": 1.4417,
      "step": 417
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.220819649715269,
      "learning_rate": 1.1221476510067115e-05,
      "loss": 1.8499,
      "step": 418
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.50967380209388,
      "learning_rate": 1.1248322147651008e-05,
      "loss": 1.6063,
      "step": 419
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.494909302554124,
      "learning_rate": 1.1275167785234899e-05,
      "loss": 1.1973,
      "step": 420
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.268122149772186,
      "learning_rate": 1.1302013422818795e-05,
      "loss": 1.5379,
      "step": 421
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.067738434152155,
      "learning_rate": 1.1328859060402686e-05,
      "loss": 1.4953,
      "step": 422
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.427611759674664,
      "learning_rate": 1.1355704697986578e-05,
      "loss": 1.5466,
      "step": 423
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.31902039398064,
      "learning_rate": 1.138255033557047e-05,
      "loss": 1.4364,
      "step": 424
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.99627372078485,
      "learning_rate": 1.1409395973154362e-05,
      "loss": 1.5627,
      "step": 425
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.604063946254929,
      "learning_rate": 1.1436241610738256e-05,
      "loss": 1.8446,
      "step": 426
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.569096113951334,
      "learning_rate": 1.1463087248322149e-05,
      "loss": 1.3505,
      "step": 427
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.485028254891345,
      "learning_rate": 1.1489932885906041e-05,
      "loss": 1.2645,
      "step": 428
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.653880927228366,
      "learning_rate": 1.1516778523489934e-05,
      "loss": 1.3512,
      "step": 429
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.573778927063621,
      "learning_rate": 1.1543624161073828e-05,
      "loss": 1.4285,
      "step": 430
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.414640141107622,
      "learning_rate": 1.1570469798657719e-05,
      "loss": 1.4266,
      "step": 431
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.229242957248958,
      "learning_rate": 1.1597315436241611e-05,
      "loss": 1.6794,
      "step": 432
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.421603714600863,
      "learning_rate": 1.1624161073825504e-05,
      "loss": 1.4813,
      "step": 433
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.865355602236088,
      "learning_rate": 1.1651006711409397e-05,
      "loss": 1.4764,
      "step": 434
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.035964443677242,
      "learning_rate": 1.1677852348993291e-05,
      "loss": 2.1341,
      "step": 435
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.2320562576311,
      "learning_rate": 1.1704697986577182e-05,
      "loss": 1.6651,
      "step": 436
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.252894253733398,
      "learning_rate": 1.1731543624161074e-05,
      "loss": 1.5525,
      "step": 437
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.839349451519443,
      "learning_rate": 1.1758389261744967e-05,
      "loss": 1.3654,
      "step": 438
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.079801548707691,
      "learning_rate": 1.178523489932886e-05,
      "loss": 1.1164,
      "step": 439
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.385378902213176,
      "learning_rate": 1.1812080536912754e-05,
      "loss": 1.5868,
      "step": 440
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.999054327947057,
      "learning_rate": 1.1838926174496645e-05,
      "loss": 1.5177,
      "step": 441
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.47426552667939,
      "learning_rate": 1.1865771812080537e-05,
      "loss": 1.3416,
      "step": 442
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.670943549709241,
      "learning_rate": 1.189261744966443e-05,
      "loss": 1.479,
      "step": 443
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.154212116148663,
      "learning_rate": 1.1919463087248324e-05,
      "loss": 1.3498,
      "step": 444
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.532682033293778,
      "learning_rate": 1.1946308724832217e-05,
      "loss": 1.3744,
      "step": 445
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.34991911053489,
      "learning_rate": 1.1973154362416108e-05,
      "loss": 1.4076,
      "step": 446
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.438665318262275,
      "learning_rate": 1.2e-05,
      "loss": 1.4545,
      "step": 447
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.316898371045484,
      "learning_rate": 1.2026845637583893e-05,
      "loss": 1.58,
      "step": 448
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.145984928059145,
      "learning_rate": 1.2053691275167787e-05,
      "loss": 1.6299,
      "step": 449
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.17763415555075,
      "learning_rate": 1.208053691275168e-05,
      "loss": 1.6198,
      "step": 450
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.8501911393393575,
      "learning_rate": 1.210738255033557e-05,
      "loss": 1.3471,
      "step": 451
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.710169940848424,
      "learning_rate": 1.2134228187919463e-05,
      "loss": 1.6027,
      "step": 452
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.480272569957146,
      "learning_rate": 1.2161073825503358e-05,
      "loss": 1.6442,
      "step": 453
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.061662406799057,
      "learning_rate": 1.218791946308725e-05,
      "loss": 1.1247,
      "step": 454
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.796630545625053,
      "learning_rate": 1.2214765100671143e-05,
      "loss": 1.6916,
      "step": 455
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.916702945209782,
      "learning_rate": 1.2241610738255034e-05,
      "loss": 1.4768,
      "step": 456
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.301497312682491,
      "learning_rate": 1.2268456375838926e-05,
      "loss": 1.5865,
      "step": 457
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.757332208179617,
      "learning_rate": 1.229530201342282e-05,
      "loss": 1.1263,
      "step": 458
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.172429477316673,
      "learning_rate": 1.2322147651006713e-05,
      "loss": 1.5916,
      "step": 459
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.764459373362694,
      "learning_rate": 1.2348993288590606e-05,
      "loss": 1.3644,
      "step": 460
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.793327597742426,
      "learning_rate": 1.2375838926174497e-05,
      "loss": 1.6128,
      "step": 461
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.153653293119383,
      "learning_rate": 1.2402684563758389e-05,
      "loss": 1.3732,
      "step": 462
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.982019540661202,
      "learning_rate": 1.2429530201342283e-05,
      "loss": 1.4621,
      "step": 463
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.745604520467719,
      "learning_rate": 1.2456375838926176e-05,
      "loss": 1.5017,
      "step": 464
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.907962121888232,
      "learning_rate": 1.2483221476510069e-05,
      "loss": 1.4954,
      "step": 465
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.025972186983962,
      "learning_rate": 1.251006711409396e-05,
      "loss": 1.3027,
      "step": 466
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.148161832078976,
      "learning_rate": 1.2536912751677854e-05,
      "loss": 1.1679,
      "step": 467
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.191485177006818,
      "learning_rate": 1.2563758389261746e-05,
      "loss": 1.3467,
      "step": 468
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.38838657551247,
      "learning_rate": 1.2590604026845639e-05,
      "loss": 1.3592,
      "step": 469
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.428320566074404,
      "learning_rate": 1.2617449664429532e-05,
      "loss": 1.9306,
      "step": 470
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.670737252850074,
      "learning_rate": 1.2644295302013422e-05,
      "loss": 1.4233,
      "step": 471
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.375400769249934,
      "learning_rate": 1.2671140939597317e-05,
      "loss": 1.3632,
      "step": 472
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.113950100576592,
      "learning_rate": 1.269798657718121e-05,
      "loss": 0.9677,
      "step": 473
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.101881978912405,
      "learning_rate": 1.2724832214765102e-05,
      "loss": 1.3013,
      "step": 474
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.566489460814012,
      "learning_rate": 1.2751677852348994e-05,
      "loss": 1.5764,
      "step": 475
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.81306177129485,
      "learning_rate": 1.2778523489932889e-05,
      "loss": 1.5302,
      "step": 476
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.103011929112583,
      "learning_rate": 1.280536912751678e-05,
      "loss": 1.5293,
      "step": 477
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.034707274716586,
      "learning_rate": 1.2832214765100672e-05,
      "loss": 1.605,
      "step": 478
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.78190053653468,
      "learning_rate": 1.2859060402684565e-05,
      "loss": 1.8285,
      "step": 479
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.832945997807883,
      "learning_rate": 1.2885906040268457e-05,
      "loss": 1.2634,
      "step": 480
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.28774557452969,
      "learning_rate": 1.2912751677852352e-05,
      "loss": 1.4308,
      "step": 481
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.778102054842449,
      "learning_rate": 1.2939597315436243e-05,
      "loss": 1.4375,
      "step": 482
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.007807198159542,
      "learning_rate": 1.2966442953020135e-05,
      "loss": 1.5594,
      "step": 483
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.555946738014196,
      "learning_rate": 1.2993288590604028e-05,
      "loss": 1.6496,
      "step": 484
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.648784491062047,
      "learning_rate": 1.302013422818792e-05,
      "loss": 1.7619,
      "step": 485
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.805502468961682,
      "learning_rate": 1.3046979865771815e-05,
      "loss": 1.2919,
      "step": 486
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.248174788555783,
      "learning_rate": 1.3073825503355706e-05,
      "loss": 1.1833,
      "step": 487
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.810163711229256,
      "learning_rate": 1.3100671140939598e-05,
      "loss": 1.5023,
      "step": 488
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.037078629041334,
      "learning_rate": 1.312751677852349e-05,
      "loss": 1.6849,
      "step": 489
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.665302782234119,
      "learning_rate": 1.3154362416107385e-05,
      "loss": 1.1481,
      "step": 490
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.062356207185401,
      "learning_rate": 1.3181208053691278e-05,
      "loss": 1.2876,
      "step": 491
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.486829501735757,
      "learning_rate": 1.3208053691275168e-05,
      "loss": 1.2023,
      "step": 492
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.019840304216956,
      "learning_rate": 1.3234899328859061e-05,
      "loss": 1.5196,
      "step": 493
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.336089736601796,
      "learning_rate": 1.3261744966442954e-05,
      "loss": 1.5339,
      "step": 494
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.415309981188626,
      "learning_rate": 1.3288590604026848e-05,
      "loss": 1.3937,
      "step": 495
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.283136231413104,
      "learning_rate": 1.331543624161074e-05,
      "loss": 1.5499,
      "step": 496
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.072416553588251,
      "learning_rate": 1.3342281879194631e-05,
      "loss": 1.681,
      "step": 497
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.856148105266359,
      "learning_rate": 1.3369127516778524e-05,
      "loss": 1.2094,
      "step": 498
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.16530480403354,
      "learning_rate": 1.3395973154362418e-05,
      "loss": 1.4894,
      "step": 499
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.695566334632854,
      "learning_rate": 1.342281879194631e-05,
      "loss": 1.1902,
      "step": 500
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.94896777394868,
      "learning_rate": 1.3449664429530202e-05,
      "loss": 1.582,
      "step": 501
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.416645425890673,
      "learning_rate": 1.3476510067114094e-05,
      "loss": 1.7216,
      "step": 502
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.111735282585837,
      "learning_rate": 1.3503355704697987e-05,
      "loss": 1.5494,
      "step": 503
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.0788022868739,
      "learning_rate": 1.3530201342281881e-05,
      "loss": 1.395,
      "step": 504
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.67167001885283,
      "learning_rate": 1.3557046979865774e-05,
      "loss": 1.3778,
      "step": 505
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.408622900237423,
      "learning_rate": 1.3583892617449665e-05,
      "loss": 1.4904,
      "step": 506
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.723404671606452,
      "learning_rate": 1.3610738255033557e-05,
      "loss": 1.4276,
      "step": 507
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.538371607976068,
      "learning_rate": 1.363758389261745e-05,
      "loss": 1.254,
      "step": 508
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.152355195574536,
      "learning_rate": 1.3664429530201344e-05,
      "loss": 1.3815,
      "step": 509
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.075484270640901,
      "learning_rate": 1.3691275167785237e-05,
      "loss": 1.5075,
      "step": 510
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.929160859051723,
      "learning_rate": 1.3718120805369128e-05,
      "loss": 1.5102,
      "step": 511
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.521422411126925,
      "learning_rate": 1.374496644295302e-05,
      "loss": 1.3788,
      "step": 512
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.662761368199288,
      "learning_rate": 1.3771812080536914e-05,
      "loss": 1.1429,
      "step": 513
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.568898730051162,
      "learning_rate": 1.3798657718120807e-05,
      "loss": 1.4793,
      "step": 514
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.31628783591004,
      "learning_rate": 1.38255033557047e-05,
      "loss": 1.4851,
      "step": 515
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.932900830669347,
      "learning_rate": 1.385234899328859e-05,
      "loss": 1.5712,
      "step": 516
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.249113168548576,
      "learning_rate": 1.3879194630872483e-05,
      "loss": 1.3454,
      "step": 517
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.881798747617315,
      "learning_rate": 1.3906040268456377e-05,
      "loss": 1.8441,
      "step": 518
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.155168464491197,
      "learning_rate": 1.393288590604027e-05,
      "loss": 1.5913,
      "step": 519
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.453412654675237,
      "learning_rate": 1.3959731543624163e-05,
      "loss": 1.5557,
      "step": 520
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.041093106879847,
      "learning_rate": 1.3986577181208053e-05,
      "loss": 1.156,
      "step": 521
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.137265776600472,
      "learning_rate": 1.4013422818791948e-05,
      "loss": 1.1797,
      "step": 522
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.870416818990308,
      "learning_rate": 1.404026845637584e-05,
      "loss": 1.3991,
      "step": 523
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.5708998645491,
      "learning_rate": 1.4067114093959733e-05,
      "loss": 1.4046,
      "step": 524
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.745510809295142,
      "learning_rate": 1.4093959731543626e-05,
      "loss": 1.3977,
      "step": 525
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.943868135824137,
      "learning_rate": 1.4120805369127516e-05,
      "loss": 1.3955,
      "step": 526
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.996875242097958,
      "learning_rate": 1.414765100671141e-05,
      "loss": 1.2694,
      "step": 527
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.496375775379063,
      "learning_rate": 1.4174496644295303e-05,
      "loss": 1.4434,
      "step": 528
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.317487581743314,
      "learning_rate": 1.4201342281879196e-05,
      "loss": 1.3041,
      "step": 529
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.2485209867741345,
      "learning_rate": 1.4228187919463088e-05,
      "loss": 1.4963,
      "step": 530
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.560976821678934,
      "learning_rate": 1.425503355704698e-05,
      "loss": 1.5704,
      "step": 531
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.811254700919813,
      "learning_rate": 1.4281879194630874e-05,
      "loss": 1.5331,
      "step": 532
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.465413006052787,
      "learning_rate": 1.4308724832214766e-05,
      "loss": 1.5635,
      "step": 533
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.101774379901071,
      "learning_rate": 1.4335570469798659e-05,
      "loss": 1.1154,
      "step": 534
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.159371358502538,
      "learning_rate": 1.4362416107382551e-05,
      "loss": 1.4458,
      "step": 535
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.174260961502805,
      "learning_rate": 1.4389261744966446e-05,
      "loss": 1.5627,
      "step": 536
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.545421669624485,
      "learning_rate": 1.4416107382550337e-05,
      "loss": 1.2197,
      "step": 537
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.708844544090837,
      "learning_rate": 1.444295302013423e-05,
      "loss": 1.5157,
      "step": 538
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.17160886194052,
      "learning_rate": 1.4469798657718122e-05,
      "loss": 1.3182,
      "step": 539
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.419679652994198,
      "learning_rate": 1.4496644295302014e-05,
      "loss": 1.3181,
      "step": 540
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.125527777852835,
      "learning_rate": 1.4523489932885909e-05,
      "loss": 1.3603,
      "step": 541
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.6874506097130375,
      "learning_rate": 1.45503355704698e-05,
      "loss": 1.0992,
      "step": 542
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.270491378056562,
      "learning_rate": 1.4577181208053692e-05,
      "loss": 1.2715,
      "step": 543
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.583169553596723,
      "learning_rate": 1.4604026845637585e-05,
      "loss": 1.2255,
      "step": 544
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.406054115096355,
      "learning_rate": 1.4630872483221479e-05,
      "loss": 1.2699,
      "step": 545
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.353522345368765,
      "learning_rate": 1.4657718120805372e-05,
      "loss": 1.4357,
      "step": 546
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.511655148061715,
      "learning_rate": 1.4684563758389262e-05,
      "loss": 1.3119,
      "step": 547
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.851821039613545,
      "learning_rate": 1.4711409395973155e-05,
      "loss": 1.4612,
      "step": 548
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.47303434090239,
      "learning_rate": 1.4738255033557048e-05,
      "loss": 1.3566,
      "step": 549
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.510422595093781,
      "learning_rate": 1.4765100671140942e-05,
      "loss": 1.5438,
      "step": 550
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.739921214927813,
      "learning_rate": 1.4791946308724835e-05,
      "loss": 1.8695,
      "step": 551
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.119992350323915,
      "learning_rate": 1.4818791946308725e-05,
      "loss": 1.1185,
      "step": 552
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.520701360288286,
      "learning_rate": 1.4845637583892618e-05,
      "loss": 1.6391,
      "step": 553
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.383802738815257,
      "learning_rate": 1.487248322147651e-05,
      "loss": 1.328,
      "step": 554
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.62589206342016,
      "learning_rate": 1.4899328859060405e-05,
      "loss": 1.2322,
      "step": 555
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.30216351796237,
      "learning_rate": 1.4926174496644297e-05,
      "loss": 1.2199,
      "step": 556
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.228721412554105,
      "learning_rate": 1.4953020134228188e-05,
      "loss": 1.2169,
      "step": 557
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.742477711805231,
      "learning_rate": 1.4979865771812081e-05,
      "loss": 1.4394,
      "step": 558
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.937906435587678,
      "learning_rate": 1.5006711409395975e-05,
      "loss": 1.2919,
      "step": 559
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.4501171918562985,
      "learning_rate": 1.5033557046979868e-05,
      "loss": 1.424,
      "step": 560
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.316736783331523,
      "learning_rate": 1.506040268456376e-05,
      "loss": 1.565,
      "step": 561
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.749884676260847,
      "learning_rate": 1.5087248322147651e-05,
      "loss": 1.2053,
      "step": 562
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.81823142785112,
      "learning_rate": 1.5114093959731544e-05,
      "loss": 1.5824,
      "step": 563
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.7225795080504875,
      "learning_rate": 1.5140939597315438e-05,
      "loss": 1.3721,
      "step": 564
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3485080045617135,
      "learning_rate": 1.516778523489933e-05,
      "loss": 1.0952,
      "step": 565
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.37887808090431,
      "learning_rate": 1.5194630872483223e-05,
      "loss": 1.3754,
      "step": 566
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.87861084672472,
      "learning_rate": 1.5221476510067114e-05,
      "loss": 1.7421,
      "step": 567
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.778248564677485,
      "learning_rate": 1.5248322147651009e-05,
      "loss": 1.711,
      "step": 568
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.62918022335809,
      "learning_rate": 1.5275167785234903e-05,
      "loss": 1.4417,
      "step": 569
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.834517594911215,
      "learning_rate": 1.5302013422818792e-05,
      "loss": 1.8575,
      "step": 570
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.17769990533976,
      "learning_rate": 1.5328859060402685e-05,
      "loss": 1.5623,
      "step": 571
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.211237958354836,
      "learning_rate": 1.5355704697986577e-05,
      "loss": 1.7896,
      "step": 572
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.3952703994595534,
      "learning_rate": 1.5382550335570473e-05,
      "loss": 1.4155,
      "step": 573
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.583809823056281,
      "learning_rate": 1.5409395973154366e-05,
      "loss": 1.236,
      "step": 574
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.116588935418603,
      "learning_rate": 1.5436241610738255e-05,
      "loss": 1.5395,
      "step": 575
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.834567717938464,
      "learning_rate": 1.5463087248322148e-05,
      "loss": 1.6032,
      "step": 576
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.313015440352208,
      "learning_rate": 1.548993288590604e-05,
      "loss": 1.531,
      "step": 577
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.761453775322154,
      "learning_rate": 1.5516778523489936e-05,
      "loss": 1.4034,
      "step": 578
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.733988907897115,
      "learning_rate": 1.554362416107383e-05,
      "loss": 1.4758,
      "step": 579
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.723471651705433,
      "learning_rate": 1.5570469798657718e-05,
      "loss": 1.6519,
      "step": 580
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.683024073717439,
      "learning_rate": 1.559731543624161e-05,
      "loss": 1.1023,
      "step": 581
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.771382530687474,
      "learning_rate": 1.5624161073825506e-05,
      "loss": 1.2879,
      "step": 582
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.333162236828377,
      "learning_rate": 1.56510067114094e-05,
      "loss": 1.8738,
      "step": 583
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.726603314237087,
      "learning_rate": 1.5677852348993288e-05,
      "loss": 1.2181,
      "step": 584
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.742760599928726,
      "learning_rate": 1.570469798657718e-05,
      "loss": 1.501,
      "step": 585
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.966656136793154,
      "learning_rate": 1.5731543624161073e-05,
      "loss": 1.5028,
      "step": 586
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.18356366854356,
      "learning_rate": 1.575838926174497e-05,
      "loss": 1.3861,
      "step": 587
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.947862619777618,
      "learning_rate": 1.5785234899328862e-05,
      "loss": 1.1892,
      "step": 588
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4826065850478445,
      "learning_rate": 1.581208053691275e-05,
      "loss": 1.1479,
      "step": 589
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.320773479811276,
      "learning_rate": 1.5838926174496644e-05,
      "loss": 1.3655,
      "step": 590
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.914291018541771,
      "learning_rate": 1.586577181208054e-05,
      "loss": 1.3441,
      "step": 591
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.516654803971699,
      "learning_rate": 1.5892617449664432e-05,
      "loss": 1.8111,
      "step": 592
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.51706100074725,
      "learning_rate": 1.5919463087248325e-05,
      "loss": 1.6239,
      "step": 593
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.671907382657079,
      "learning_rate": 1.5946308724832214e-05,
      "loss": 1.3289,
      "step": 594
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.026837225200687,
      "learning_rate": 1.5973154362416107e-05,
      "loss": 1.3409,
      "step": 595
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.869668467611291,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.3098,
      "step": 596
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.282826546264316,
      "learning_rate": 1.6026845637583895e-05,
      "loss": 1.4488,
      "step": 597
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.690974793437086,
      "learning_rate": 1.6053691275167788e-05,
      "loss": 1.0872,
      "step": 598
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.356392804768612,
      "learning_rate": 1.6080536912751677e-05,
      "loss": 1.3607,
      "step": 599
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.048842024036013,
      "learning_rate": 1.610738255033557e-05,
      "loss": 1.4449,
      "step": 600
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.155717759001321,
      "learning_rate": 1.6134228187919466e-05,
      "loss": 1.5859,
      "step": 601
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.858674187094271,
      "learning_rate": 1.6161073825503358e-05,
      "loss": 1.1479,
      "step": 602
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.558384009973683,
      "learning_rate": 1.618791946308725e-05,
      "loss": 1.653,
      "step": 603
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.775744137326498,
      "learning_rate": 1.621476510067114e-05,
      "loss": 1.4949,
      "step": 604
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.464495909492927,
      "learning_rate": 1.6241610738255036e-05,
      "loss": 1.505,
      "step": 605
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.282144098657279,
      "learning_rate": 1.626845637583893e-05,
      "loss": 1.4929,
      "step": 606
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.774937894051398,
      "learning_rate": 1.629530201342282e-05,
      "loss": 1.0399,
      "step": 607
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.515020436246944,
      "learning_rate": 1.6322147651006714e-05,
      "loss": 1.9131,
      "step": 608
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.02323023790672,
      "learning_rate": 1.6348993288590603e-05,
      "loss": 1.6139,
      "step": 609
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.622876690045713,
      "learning_rate": 1.63758389261745e-05,
      "loss": 1.0577,
      "step": 610
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.776784253430833,
      "learning_rate": 1.640268456375839e-05,
      "loss": 1.9954,
      "step": 611
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.695960454965206,
      "learning_rate": 1.6429530201342284e-05,
      "loss": 1.1116,
      "step": 612
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.016577281751768,
      "learning_rate": 1.6456375838926177e-05,
      "loss": 1.5672,
      "step": 613
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.28260412750175,
      "learning_rate": 1.648322147651007e-05,
      "loss": 1.5242,
      "step": 614
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.842548267981796,
      "learning_rate": 1.6510067114093962e-05,
      "loss": 1.4001,
      "step": 615
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.797958727681141,
      "learning_rate": 1.6536912751677854e-05,
      "loss": 1.5679,
      "step": 616
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.134279512769903,
      "learning_rate": 1.6563758389261747e-05,
      "loss": 1.4982,
      "step": 617
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.93637686416352,
      "learning_rate": 1.659060402684564e-05,
      "loss": 1.6387,
      "step": 618
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.911455523137747,
      "learning_rate": 1.6617449664429532e-05,
      "loss": 1.3803,
      "step": 619
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.111460667724544,
      "learning_rate": 1.6644295302013425e-05,
      "loss": 1.2912,
      "step": 620
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.84826865811465,
      "learning_rate": 1.6671140939597317e-05,
      "loss": 1.3408,
      "step": 621
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.811612554386186,
      "learning_rate": 1.669798657718121e-05,
      "loss": 1.4216,
      "step": 622
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.600228955930593,
      "learning_rate": 1.6724832214765103e-05,
      "loss": 1.6831,
      "step": 623
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.237916989274682,
      "learning_rate": 1.6751677852348995e-05,
      "loss": 1.5187,
      "step": 624
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.690984881873106,
      "learning_rate": 1.6778523489932888e-05,
      "loss": 1.2515,
      "step": 625
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.629821009030652,
      "learning_rate": 1.680536912751678e-05,
      "loss": 1.7682,
      "step": 626
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.768552078380317,
      "learning_rate": 1.6832214765100673e-05,
      "loss": 1.6374,
      "step": 627
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.826202406673499,
      "learning_rate": 1.6859060402684565e-05,
      "loss": 1.5675,
      "step": 628
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.343875400904523,
      "learning_rate": 1.6885906040268458e-05,
      "loss": 1.9749,
      "step": 629
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.442940900110244,
      "learning_rate": 1.691275167785235e-05,
      "loss": 1.3851,
      "step": 630
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.49809278340363,
      "learning_rate": 1.6939597315436243e-05,
      "loss": 1.5163,
      "step": 631
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.87558836941394,
      "learning_rate": 1.6966442953020136e-05,
      "loss": 1.4327,
      "step": 632
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.879065893643617,
      "learning_rate": 1.699328859060403e-05,
      "loss": 1.6641,
      "step": 633
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.970638435658616,
      "learning_rate": 1.702013422818792e-05,
      "loss": 1.3324,
      "step": 634
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.150071994776928,
      "learning_rate": 1.7046979865771814e-05,
      "loss": 1.2787,
      "step": 635
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.721376703536926,
      "learning_rate": 1.7073825503355706e-05,
      "loss": 1.6006,
      "step": 636
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.798452481275602,
      "learning_rate": 1.71006711409396e-05,
      "loss": 1.7199,
      "step": 637
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.493434676423813,
      "learning_rate": 1.712751677852349e-05,
      "loss": 1.5992,
      "step": 638
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.584695602309373,
      "learning_rate": 1.7154362416107384e-05,
      "loss": 1.5779,
      "step": 639
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.211231575961541,
      "learning_rate": 1.7181208053691277e-05,
      "loss": 1.2577,
      "step": 640
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.607595585767653,
      "learning_rate": 1.720805369127517e-05,
      "loss": 1.7324,
      "step": 641
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.043581327014544,
      "learning_rate": 1.7234899328859062e-05,
      "loss": 1.587,
      "step": 642
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.914377064064201,
      "learning_rate": 1.7261744966442954e-05,
      "loss": 1.7005,
      "step": 643
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.383295887873215,
      "learning_rate": 1.7288590604026847e-05,
      "loss": 1.5572,
      "step": 644
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.998904184716423,
      "learning_rate": 1.731543624161074e-05,
      "loss": 1.2511,
      "step": 645
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.115506829839945,
      "learning_rate": 1.7342281879194632e-05,
      "loss": 1.3593,
      "step": 646
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.054047838124271,
      "learning_rate": 1.7369127516778525e-05,
      "loss": 1.3131,
      "step": 647
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.129573932882154,
      "learning_rate": 1.7395973154362417e-05,
      "loss": 1.3129,
      "step": 648
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.761645123310167,
      "learning_rate": 1.742281879194631e-05,
      "loss": 1.3701,
      "step": 649
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.9777522086385035,
      "learning_rate": 1.7449664429530202e-05,
      "loss": 1.4474,
      "step": 650
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.003757630729373,
      "learning_rate": 1.7476510067114095e-05,
      "loss": 1.1496,
      "step": 651
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.243792899357986,
      "learning_rate": 1.7503355704697988e-05,
      "loss": 1.8197,
      "step": 652
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.751956755261069,
      "learning_rate": 1.753020134228188e-05,
      "loss": 1.5824,
      "step": 653
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.096351528690645,
      "learning_rate": 1.7557046979865773e-05,
      "loss": 1.497,
      "step": 654
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.694022663612402,
      "learning_rate": 1.7583892617449665e-05,
      "loss": 1.3844,
      "step": 655
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.58025552246381,
      "learning_rate": 1.7610738255033558e-05,
      "loss": 1.5863,
      "step": 656
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.547482488927036,
      "learning_rate": 1.763758389261745e-05,
      "loss": 1.0899,
      "step": 657
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.416158381154374,
      "learning_rate": 1.7664429530201343e-05,
      "loss": 1.5087,
      "step": 658
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.294839563754406,
      "learning_rate": 1.7691275167785236e-05,
      "loss": 1.8399,
      "step": 659
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.630333333437656,
      "learning_rate": 1.771812080536913e-05,
      "loss": 1.2181,
      "step": 660
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.113713147515277,
      "learning_rate": 1.774496644295302e-05,
      "loss": 1.4961,
      "step": 661
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.248928916187554,
      "learning_rate": 1.7771812080536913e-05,
      "loss": 1.506,
      "step": 662
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.843347056037487,
      "learning_rate": 1.7798657718120806e-05,
      "loss": 1.7997,
      "step": 663
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.595666452267448,
      "learning_rate": 1.78255033557047e-05,
      "loss": 1.4877,
      "step": 664
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.573118819718279,
      "learning_rate": 1.7852348993288595e-05,
      "loss": 1.4087,
      "step": 665
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.990766588736186,
      "learning_rate": 1.7879194630872484e-05,
      "loss": 1.8962,
      "step": 666
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.799215913727351,
      "learning_rate": 1.7906040268456376e-05,
      "loss": 1.3199,
      "step": 667
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.683206986329194,
      "learning_rate": 1.793288590604027e-05,
      "loss": 1.0974,
      "step": 668
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.351257159317212,
      "learning_rate": 1.795973154362416e-05,
      "loss": 1.3103,
      "step": 669
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.950992592804711,
      "learning_rate": 1.7986577181208054e-05,
      "loss": 1.4563,
      "step": 670
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.550490056202355,
      "learning_rate": 1.8013422818791947e-05,
      "loss": 1.6629,
      "step": 671
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.37748526166874,
      "learning_rate": 1.804026845637584e-05,
      "loss": 1.4842,
      "step": 672
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.969724880593675,
      "learning_rate": 1.8067114093959732e-05,
      "loss": 1.1297,
      "step": 673
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.320598290114177,
      "learning_rate": 1.8093959731543628e-05,
      "loss": 1.4965,
      "step": 674
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.66158812672837,
      "learning_rate": 1.8120805369127517e-05,
      "loss": 1.5017,
      "step": 675
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.024446627271338,
      "learning_rate": 1.814765100671141e-05,
      "loss": 1.7178,
      "step": 676
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.722779569375733,
      "learning_rate": 1.8174496644295302e-05,
      "loss": 1.4876,
      "step": 677
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.997168282674599,
      "learning_rate": 1.8201342281879195e-05,
      "loss": 1.22,
      "step": 678
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.814857150025032,
      "learning_rate": 1.822818791946309e-05,
      "loss": 1.3637,
      "step": 679
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.720562058525064,
      "learning_rate": 1.825503355704698e-05,
      "loss": 1.4558,
      "step": 680
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.095251042606805,
      "learning_rate": 1.8281879194630873e-05,
      "loss": 1.6135,
      "step": 681
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.379690528812599,
      "learning_rate": 1.8308724832214765e-05,
      "loss": 1.4294,
      "step": 682
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.237698216152336,
      "learning_rate": 1.833557046979866e-05,
      "loss": 1.4633,
      "step": 683
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.920617156667952,
      "learning_rate": 1.8362416107382554e-05,
      "loss": 1.3385,
      "step": 684
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.978482030153378,
      "learning_rate": 1.8389261744966443e-05,
      "loss": 1.4339,
      "step": 685
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.675762067153812,
      "learning_rate": 1.8416107382550336e-05,
      "loss": 1.7475,
      "step": 686
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.91537531960125,
      "learning_rate": 1.8442953020134228e-05,
      "loss": 1.3903,
      "step": 687
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.39191489746502,
      "learning_rate": 1.8469798657718124e-05,
      "loss": 1.1273,
      "step": 688
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.698657895502848,
      "learning_rate": 1.8496644295302017e-05,
      "loss": 1.2752,
      "step": 689
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.040502632693768,
      "learning_rate": 1.8523489932885906e-05,
      "loss": 1.3141,
      "step": 690
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.41594770225767,
      "learning_rate": 1.85503355704698e-05,
      "loss": 1.8202,
      "step": 691
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.059940052491621,
      "learning_rate": 1.857718120805369e-05,
      "loss": 1.6422,
      "step": 692
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.998902675284476,
      "learning_rate": 1.8604026845637587e-05,
      "loss": 1.6059,
      "step": 693
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.060663849685163,
      "learning_rate": 1.863087248322148e-05,
      "loss": 1.3515,
      "step": 694
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.44681127573044,
      "learning_rate": 1.865771812080537e-05,
      "loss": 1.4895,
      "step": 695
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.924586200696046,
      "learning_rate": 1.868456375838926e-05,
      "loss": 1.4124,
      "step": 696
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.753865952420195,
      "learning_rate": 1.8711409395973157e-05,
      "loss": 1.2813,
      "step": 697
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.585563320327854,
      "learning_rate": 1.873825503355705e-05,
      "loss": 1.2427,
      "step": 698
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.28238335349942,
      "learning_rate": 1.8765100671140943e-05,
      "loss": 1.6496,
      "step": 699
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.320841708865357,
      "learning_rate": 1.8791946308724832e-05,
      "loss": 1.4567,
      "step": 700
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.720938619177465,
      "learning_rate": 1.8818791946308724e-05,
      "loss": 1.4006,
      "step": 701
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.087994039308818,
      "learning_rate": 1.884563758389262e-05,
      "loss": 1.738,
      "step": 702
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.348231111030693,
      "learning_rate": 1.8872483221476513e-05,
      "loss": 1.1448,
      "step": 703
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.934053643986842,
      "learning_rate": 1.8899328859060406e-05,
      "loss": 1.4155,
      "step": 704
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.804127198362616,
      "learning_rate": 1.8926174496644295e-05,
      "loss": 1.2542,
      "step": 705
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.454901448724996,
      "learning_rate": 1.895302013422819e-05,
      "loss": 1.5187,
      "step": 706
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.667501023111416,
      "learning_rate": 1.8979865771812083e-05,
      "loss": 1.3869,
      "step": 707
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.685829475079491,
      "learning_rate": 1.9006711409395976e-05,
      "loss": 1.4483,
      "step": 708
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.851068920924944,
      "learning_rate": 1.903355704697987e-05,
      "loss": 1.5229,
      "step": 709
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.360581657943092,
      "learning_rate": 1.9060402684563758e-05,
      "loss": 1.5678,
      "step": 710
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.838471833579849,
      "learning_rate": 1.9087248322147654e-05,
      "loss": 1.5814,
      "step": 711
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.538814897314536,
      "learning_rate": 1.9114093959731546e-05,
      "loss": 1.5919,
      "step": 712
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.973865796091992,
      "learning_rate": 1.914093959731544e-05,
      "loss": 1.5036,
      "step": 713
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.942099494799981,
      "learning_rate": 1.916778523489933e-05,
      "loss": 1.346,
      "step": 714
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.928561960545991,
      "learning_rate": 1.919463087248322e-05,
      "loss": 1.5113,
      "step": 715
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.327535615218974,
      "learning_rate": 1.9221476510067117e-05,
      "loss": 1.4813,
      "step": 716
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.111254074950109,
      "learning_rate": 1.924832214765101e-05,
      "loss": 1.3902,
      "step": 717
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.795375983098658,
      "learning_rate": 1.9275167785234902e-05,
      "loss": 1.4768,
      "step": 718
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.106205038701756,
      "learning_rate": 1.9302013422818794e-05,
      "loss": 1.3735,
      "step": 719
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.282268008053936,
      "learning_rate": 1.9328859060402687e-05,
      "loss": 0.9056,
      "step": 720
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.643635341717253,
      "learning_rate": 1.935570469798658e-05,
      "loss": 1.9102,
      "step": 721
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.40585713760294,
      "learning_rate": 1.9382550335570472e-05,
      "loss": 1.1561,
      "step": 722
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.501838932148951,
      "learning_rate": 1.9409395973154365e-05,
      "loss": 1.1236,
      "step": 723
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.98236565771981,
      "learning_rate": 1.9436241610738257e-05,
      "loss": 1.8247,
      "step": 724
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.727741036215868,
      "learning_rate": 1.946308724832215e-05,
      "loss": 1.2078,
      "step": 725
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.416768621462328,
      "learning_rate": 1.9489932885906042e-05,
      "loss": 1.7006,
      "step": 726
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.513142860504843,
      "learning_rate": 1.9516778523489935e-05,
      "loss": 1.1217,
      "step": 727
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.072295401301412,
      "learning_rate": 1.9543624161073828e-05,
      "loss": 1.38,
      "step": 728
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.6748219663576736,
      "learning_rate": 1.957046979865772e-05,
      "loss": 1.7309,
      "step": 729
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.254250623148174,
      "learning_rate": 1.9597315436241613e-05,
      "loss": 1.972,
      "step": 730
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.830264236163746,
      "learning_rate": 1.9624161073825505e-05,
      "loss": 1.1723,
      "step": 731
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.223272899053619,
      "learning_rate": 1.9651006711409398e-05,
      "loss": 1.6147,
      "step": 732
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.85084381573008,
      "learning_rate": 1.967785234899329e-05,
      "loss": 1.3413,
      "step": 733
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.742970632841864,
      "learning_rate": 1.9704697986577183e-05,
      "loss": 1.3835,
      "step": 734
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.360610460935072,
      "learning_rate": 1.9731543624161076e-05,
      "loss": 1.08,
      "step": 735
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.285160081855017,
      "learning_rate": 1.975838926174497e-05,
      "loss": 1.6392,
      "step": 736
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.1117153040527645,
      "learning_rate": 1.978523489932886e-05,
      "loss": 1.2834,
      "step": 737
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.785895560013108,
      "learning_rate": 1.9812080536912754e-05,
      "loss": 1.1575,
      "step": 738
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.703110716339846,
      "learning_rate": 1.9838926174496646e-05,
      "loss": 1.7975,
      "step": 739
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.237705248850261,
      "learning_rate": 1.986577181208054e-05,
      "loss": 1.291,
      "step": 740
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.116558227935354,
      "learning_rate": 1.989261744966443e-05,
      "loss": 2.0239,
      "step": 741
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.90484734944185,
      "learning_rate": 1.9919463087248324e-05,
      "loss": 1.3373,
      "step": 742
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.5733463525343225,
      "learning_rate": 1.9946308724832216e-05,
      "loss": 1.3538,
      "step": 743
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.116657752010826,
      "learning_rate": 1.997315436241611e-05,
      "loss": 1.5555,
      "step": 744
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.469860543505684,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 745
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.706829985744727,
      "learning_rate": 1.9999999914901735e-05,
      "loss": 1.2363,
      "step": 746
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.206991344480622,
      "learning_rate": 1.9999999659606937e-05,
      "loss": 1.599,
      "step": 747
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.706840586256803,
      "learning_rate": 1.999999923411561e-05,
      "loss": 1.4984,
      "step": 748
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.264141889429116,
      "learning_rate": 1.9999998638427763e-05,
      "loss": 1.716,
      "step": 749
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.274320343622609,
      "learning_rate": 1.9999997872543408e-05,
      "loss": 1.7843,
      "step": 750
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.8960701568138205,
      "learning_rate": 1.9999996936462554e-05,
      "loss": 1.7174,
      "step": 751
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.598929548023946,
      "learning_rate": 1.9999995830185223e-05,
      "loss": 1.0959,
      "step": 752
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.780972960595795,
      "learning_rate": 1.999999455371143e-05,
      "loss": 1.1483,
      "step": 753
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.184131294296247,
      "learning_rate": 1.9999993107041193e-05,
      "loss": 1.7082,
      "step": 754
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.781349946752146,
      "learning_rate": 1.9999991490174538e-05,
      "loss": 2.0316,
      "step": 755
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.640662260274956,
      "learning_rate": 1.99999897031115e-05,
      "loss": 1.3356,
      "step": 756
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.88159666557774,
      "learning_rate": 1.99999877458521e-05,
      "loss": 1.4823,
      "step": 757
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.44327637549819,
      "learning_rate": 1.9999985618396377e-05,
      "loss": 1.3769,
      "step": 758
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.395754990201366,
      "learning_rate": 1.999998332074437e-05,
      "loss": 1.1389,
      "step": 759
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.37935042952115,
      "learning_rate": 1.9999980852896106e-05,
      "loss": 1.3471,
      "step": 760
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.228485876678234,
      "learning_rate": 1.999997821485164e-05,
      "loss": 1.042,
      "step": 761
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.7488458530356,
      "learning_rate": 1.999997540661101e-05,
      "loss": 1.3817,
      "step": 762
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.917897366076083,
      "learning_rate": 1.9999972428174262e-05,
      "loss": 1.6128,
      "step": 763
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.550449975188234,
      "learning_rate": 1.9999969279541456e-05,
      "loss": 1.3963,
      "step": 764
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.13585240811636,
      "learning_rate": 1.9999965960712635e-05,
      "loss": 1.2469,
      "step": 765
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.7245083934014644,
      "learning_rate": 1.9999962471687865e-05,
      "loss": 1.3719,
      "step": 766
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.296826926203583,
      "learning_rate": 1.9999958812467196e-05,
      "loss": 2.0527,
      "step": 767
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.024472436586068,
      "learning_rate": 1.9999954983050696e-05,
      "loss": 1.4095,
      "step": 768
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.260654994613297,
      "learning_rate": 1.9999950983438433e-05,
      "loss": 1.6362,
      "step": 769
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.186866729649955,
      "learning_rate": 1.9999946813630468e-05,
      "loss": 1.6968,
      "step": 770
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.813063620379354,
      "learning_rate": 1.999994247362687e-05,
      "loss": 1.4802,
      "step": 771
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.981096229430154,
      "learning_rate": 1.9999937963427728e-05,
      "loss": 1.4514,
      "step": 772
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.03754199655554,
      "learning_rate": 1.99999332830331e-05,
      "loss": 1.3815,
      "step": 773
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.023180925830445,
      "learning_rate": 1.9999928432443077e-05,
      "loss": 1.498,
      "step": 774
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.131724684558895,
      "learning_rate": 1.9999923411657742e-05,
      "loss": 1.4913,
      "step": 775
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.818180068795018,
      "learning_rate": 1.9999918220677176e-05,
      "loss": 1.4362,
      "step": 776
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.546406079156196,
      "learning_rate": 1.9999912859501466e-05,
      "loss": 1.3896,
      "step": 777
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.86169432171516,
      "learning_rate": 1.9999907328130708e-05,
      "loss": 1.2364,
      "step": 778
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.166797940058492,
      "learning_rate": 1.9999901626564993e-05,
      "loss": 1.8125,
      "step": 779
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.681415371586207,
      "learning_rate": 1.9999895754804423e-05,
      "loss": 1.6726,
      "step": 780
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.785825979286638,
      "learning_rate": 1.999988971284909e-05,
      "loss": 1.4207,
      "step": 781
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.096810940912092,
      "learning_rate": 1.9999883500699103e-05,
      "loss": 1.1938,
      "step": 782
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.439366157570156,
      "learning_rate": 1.9999877118354564e-05,
      "loss": 1.5035,
      "step": 783
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.785683165772193,
      "learning_rate": 1.9999870565815582e-05,
      "loss": 1.619,
      "step": 784
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.644861384471609,
      "learning_rate": 1.9999863843082274e-05,
      "loss": 1.4064,
      "step": 785
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.9769501926234625,
      "learning_rate": 1.9999856950154747e-05,
      "loss": 1.5519,
      "step": 786
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.067900778530898,
      "learning_rate": 1.999984988703312e-05,
      "loss": 1.6625,
      "step": 787
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.061841807549257,
      "learning_rate": 1.999984265371752e-05,
      "loss": 1.585,
      "step": 788
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.023966626054792,
      "learning_rate": 1.999983525020806e-05,
      "loss": 1.619,
      "step": 789
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.374809501106074,
      "learning_rate": 1.9999827676504874e-05,
      "loss": 1.3745,
      "step": 790
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.212699114977627,
      "learning_rate": 1.999981993260809e-05,
      "loss": 1.4163,
      "step": 791
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.955192400510422,
      "learning_rate": 1.999981201851783e-05,
      "loss": 1.4073,
      "step": 792
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.131343692873796,
      "learning_rate": 1.999980393423424e-05,
      "loss": 1.1097,
      "step": 793
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.389692371614512,
      "learning_rate": 1.9999795679757456e-05,
      "loss": 1.4518,
      "step": 794
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.418405174133042,
      "learning_rate": 1.9999787255087613e-05,
      "loss": 1.1745,
      "step": 795
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.112785581933728,
      "learning_rate": 1.9999778660224863e-05,
      "loss": 1.5497,
      "step": 796
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.979924218349447,
      "learning_rate": 1.999976989516934e-05,
      "loss": 1.6604,
      "step": 797
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.744524953387931,
      "learning_rate": 1.9999760959921204e-05,
      "loss": 1.1642,
      "step": 798
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.497287409860748,
      "learning_rate": 1.9999751854480604e-05,
      "loss": 1.7451,
      "step": 799
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.37670110198183,
      "learning_rate": 1.9999742578847692e-05,
      "loss": 1.1948,
      "step": 800
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.60652184238769,
      "learning_rate": 1.999973313302263e-05,
      "loss": 1.633,
      "step": 801
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.566660149471107,
      "learning_rate": 1.9999723517005574e-05,
      "loss": 1.1491,
      "step": 802
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.582237190888199,
      "learning_rate": 1.9999713730796693e-05,
      "loss": 1.8714,
      "step": 803
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.649713840485242,
      "learning_rate": 1.999970377439615e-05,
      "loss": 1.8329,
      "step": 804
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.302606868619392,
      "learning_rate": 1.9999693647804115e-05,
      "loss": 1.332,
      "step": 805
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.3297002103004,
      "learning_rate": 1.9999683351020765e-05,
      "loss": 1.3406,
      "step": 806
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.387169241072521,
      "learning_rate": 1.9999672884046265e-05,
      "loss": 1.2219,
      "step": 807
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.21946052432099,
      "learning_rate": 1.9999662246880804e-05,
      "loss": 1.125,
      "step": 808
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.686452505019419,
      "learning_rate": 1.9999651439524553e-05,
      "loss": 1.5402,
      "step": 809
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.943442565837149,
      "learning_rate": 1.9999640461977706e-05,
      "loss": 1.5689,
      "step": 810
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.555902913489458,
      "learning_rate": 1.999962931424044e-05,
      "loss": 1.4912,
      "step": 811
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.916068438554028,
      "learning_rate": 1.999961799631296e-05,
      "loss": 1.4335,
      "step": 812
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.13841769076635,
      "learning_rate": 1.999960650819544e-05,
      "loss": 1.306,
      "step": 813
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.509678119174156,
      "learning_rate": 1.9999594849888083e-05,
      "loss": 1.2639,
      "step": 814
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.671413232654003,
      "learning_rate": 1.9999583021391093e-05,
      "loss": 1.533,
      "step": 815
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.170979308483551,
      "learning_rate": 1.9999571022704665e-05,
      "loss": 1.3236,
      "step": 816
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.965625377361226,
      "learning_rate": 1.9999558853829003e-05,
      "loss": 1.3139,
      "step": 817
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.3054427065930865,
      "learning_rate": 1.999954651476432e-05,
      "loss": 1.2987,
      "step": 818
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.710995486022545,
      "learning_rate": 1.999953400551082e-05,
      "loss": 1.4309,
      "step": 819
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.881028038869501,
      "learning_rate": 1.999952132606872e-05,
      "loss": 1.4738,
      "step": 820
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.814300233674139,
      "learning_rate": 1.999950847643823e-05,
      "loss": 1.2071,
      "step": 821
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.62747515332661,
      "learning_rate": 1.9999495456619572e-05,
      "loss": 1.3801,
      "step": 822
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.505578199197759,
      "learning_rate": 1.999948226661297e-05,
      "loss": 1.3126,
      "step": 823
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.957752844563435,
      "learning_rate": 1.9999468906418647e-05,
      "loss": 1.3338,
      "step": 824
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.933677100266538,
      "learning_rate": 1.999945537603683e-05,
      "loss": 1.2617,
      "step": 825
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.0164568690635996,
      "learning_rate": 1.9999441675467748e-05,
      "loss": 1.3407,
      "step": 826
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.020236962220256,
      "learning_rate": 1.9999427804711638e-05,
      "loss": 1.7017,
      "step": 827
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.146532602788197,
      "learning_rate": 1.999941376376873e-05,
      "loss": 1.5422,
      "step": 828
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.600247752067845,
      "learning_rate": 1.9999399552639268e-05,
      "loss": 1.449,
      "step": 829
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.858498831341781,
      "learning_rate": 1.9999385171323492e-05,
      "loss": 1.5386,
      "step": 830
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.460704209311796,
      "learning_rate": 1.9999370619821646e-05,
      "loss": 1.3679,
      "step": 831
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.63946032846754,
      "learning_rate": 1.999935589813398e-05,
      "loss": 1.795,
      "step": 832
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.027083400066376,
      "learning_rate": 1.999934100626074e-05,
      "loss": 1.3929,
      "step": 833
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.039673303531014,
      "learning_rate": 1.9999325944202187e-05,
      "loss": 1.3605,
      "step": 834
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.1670429422022215,
      "learning_rate": 1.999931071195857e-05,
      "loss": 1.2207,
      "step": 835
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.710611359121355,
      "learning_rate": 1.9999295309530155e-05,
      "loss": 1.3509,
      "step": 836
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.9800599831532715,
      "learning_rate": 1.99992797369172e-05,
      "loss": 1.4956,
      "step": 837
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.076902527027667,
      "learning_rate": 1.999926399411997e-05,
      "loss": 1.4288,
      "step": 838
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.945709014067419,
      "learning_rate": 1.999924808113873e-05,
      "loss": 1.5127,
      "step": 839
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.946234903174766,
      "learning_rate": 1.9999231997973756e-05,
      "loss": 1.2649,
      "step": 840
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.261989801534517,
      "learning_rate": 1.999921574462532e-05,
      "loss": 1.3994,
      "step": 841
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.364302618827185,
      "learning_rate": 1.99991993210937e-05,
      "loss": 1.4293,
      "step": 842
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.348881751103375,
      "learning_rate": 1.9999182727379173e-05,
      "loss": 1.8878,
      "step": 843
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.822638419472803,
      "learning_rate": 1.9999165963482025e-05,
      "loss": 0.9755,
      "step": 844
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.823924191768002,
      "learning_rate": 1.9999149029402533e-05,
      "loss": 1.1731,
      "step": 845
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.441260653035092,
      "learning_rate": 1.9999131925140994e-05,
      "loss": 1.2607,
      "step": 846
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.137848826561035,
      "learning_rate": 1.99991146506977e-05,
      "loss": 1.3407,
      "step": 847
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.348558348268355,
      "learning_rate": 1.9999097206072936e-05,
      "loss": 1.55,
      "step": 848
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.90711494647997,
      "learning_rate": 1.9999079591267008e-05,
      "loss": 1.5706,
      "step": 849
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.531440156458352,
      "learning_rate": 1.999906180628021e-05,
      "loss": 1.1552,
      "step": 850
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.423840931270778,
      "learning_rate": 1.9999043851112848e-05,
      "loss": 1.0926,
      "step": 851
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.470800628277083,
      "learning_rate": 1.9999025725765223e-05,
      "loss": 1.6047,
      "step": 852
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.481258890278603,
      "learning_rate": 1.9999007430237648e-05,
      "loss": 1.2276,
      "step": 853
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.599166988714669,
      "learning_rate": 1.9998988964530434e-05,
      "loss": 1.2592,
      "step": 854
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.998311713897696,
      "learning_rate": 1.9998970328643895e-05,
      "loss": 1.4086,
      "step": 855
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.81383571468493,
      "learning_rate": 1.9998951522578344e-05,
      "loss": 1.2964,
      "step": 856
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.357954055525788,
      "learning_rate": 1.999893254633411e-05,
      "loss": 1.4832,
      "step": 857
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.593746315651964,
      "learning_rate": 1.9998913399911507e-05,
      "loss": 1.1391,
      "step": 858
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.1076663544153185,
      "learning_rate": 1.9998894083310864e-05,
      "loss": 1.2948,
      "step": 859
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.99522056733465,
      "learning_rate": 1.9998874596532512e-05,
      "loss": 1.1518,
      "step": 860
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.14299161592401,
      "learning_rate": 1.999885493957678e-05,
      "loss": 1.2017,
      "step": 861
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.656668942223748,
      "learning_rate": 1.9998835112444005e-05,
      "loss": 1.2726,
      "step": 862
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.4357093681983795,
      "learning_rate": 1.999881511513452e-05,
      "loss": 1.3603,
      "step": 863
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.527095207488873,
      "learning_rate": 1.9998794947648667e-05,
      "loss": 1.6349,
      "step": 864
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.529976740925223,
      "learning_rate": 1.9998774609986794e-05,
      "loss": 1.2691,
      "step": 865
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.486917256158089,
      "learning_rate": 1.9998754102149243e-05,
      "loss": 1.4847,
      "step": 866
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.3840159519485535,
      "learning_rate": 1.9998733424136362e-05,
      "loss": 1.2487,
      "step": 867
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.786659163729986,
      "learning_rate": 1.9998712575948508e-05,
      "loss": 1.4958,
      "step": 868
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9899254222054235,
      "learning_rate": 1.999869155758603e-05,
      "loss": 1.0337,
      "step": 869
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.466241359445439,
      "learning_rate": 1.9998670369049287e-05,
      "loss": 1.4944,
      "step": 870
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.281120623943987,
      "learning_rate": 1.999864901033864e-05,
      "loss": 1.4187,
      "step": 871
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.743635079779636,
      "learning_rate": 1.9998627481454455e-05,
      "loss": 1.0909,
      "step": 872
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.8700526621491536,
      "learning_rate": 1.9998605782397095e-05,
      "loss": 1.5574,
      "step": 873
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.178490473313165,
      "learning_rate": 1.9998583913166937e-05,
      "loss": 1.2867,
      "step": 874
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.707439956768373,
      "learning_rate": 1.999856187376434e-05,
      "loss": 1.6709,
      "step": 875
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.869414675758467,
      "learning_rate": 1.999853966418969e-05,
      "loss": 1.2027,
      "step": 876
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.862271494855543,
      "learning_rate": 1.999851728444336e-05,
      "loss": 1.3763,
      "step": 877
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.92081644880729,
      "learning_rate": 1.999849473452573e-05,
      "loss": 1.5602,
      "step": 878
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.068815683800614,
      "learning_rate": 1.999847201443719e-05,
      "loss": 1.4446,
      "step": 879
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.570821762072489,
      "learning_rate": 1.9998449124178122e-05,
      "loss": 1.3241,
      "step": 880
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.490764591285981,
      "learning_rate": 1.9998426063748915e-05,
      "loss": 1.3843,
      "step": 881
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.792883202515735,
      "learning_rate": 1.9998402833149962e-05,
      "loss": 1.0749,
      "step": 882
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.772212550643707,
      "learning_rate": 1.999837943238166e-05,
      "loss": 1.2109,
      "step": 883
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.523506950103476,
      "learning_rate": 1.9998355861444408e-05,
      "loss": 1.747,
      "step": 884
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.616832527487637,
      "learning_rate": 1.9998332120338603e-05,
      "loss": 1.5266,
      "step": 885
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.840466841293714,
      "learning_rate": 1.999830820906465e-05,
      "loss": 1.4726,
      "step": 886
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.769396106789772,
      "learning_rate": 1.9998284127622963e-05,
      "loss": 1.5841,
      "step": 887
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.458809827355406,
      "learning_rate": 1.999825987601394e-05,
      "loss": 1.4778,
      "step": 888
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.839335232516739,
      "learning_rate": 1.9998235454238006e-05,
      "loss": 1.1632,
      "step": 889
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.924268410262047,
      "learning_rate": 1.9998210862295568e-05,
      "loss": 1.0121,
      "step": 890
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.132790765966691,
      "learning_rate": 1.999818610018705e-05,
      "loss": 1.2826,
      "step": 891
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.638807362039483,
      "learning_rate": 1.9998161167912865e-05,
      "loss": 1.8299,
      "step": 892
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.348266224987619,
      "learning_rate": 1.9998136065473447e-05,
      "loss": 1.5679,
      "step": 893
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.1697973469581155,
      "learning_rate": 1.999811079286922e-05,
      "loss": 1.3121,
      "step": 894
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.675506559745794,
      "learning_rate": 1.9998085350100614e-05,
      "loss": 1.2425,
      "step": 895
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.009689644410746,
      "learning_rate": 1.999805973716806e-05,
      "loss": 1.0485,
      "step": 896
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.241446875853482,
      "learning_rate": 1.9998033954071997e-05,
      "loss": 1.5707,
      "step": 897
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.80428414964604,
      "learning_rate": 1.999800800081286e-05,
      "loss": 1.2991,
      "step": 898
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.61366199279083,
      "learning_rate": 1.999798187739109e-05,
      "loss": 1.7387,
      "step": 899
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.313297502305878,
      "learning_rate": 1.999795558380714e-05,
      "loss": 1.3132,
      "step": 900
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.3202478094818995,
      "learning_rate": 1.999792912006145e-05,
      "loss": 1.7162,
      "step": 901
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.37181925081806,
      "learning_rate": 1.9997902486154475e-05,
      "loss": 1.5806,
      "step": 902
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.338108099523437,
      "learning_rate": 1.9997875682086666e-05,
      "loss": 1.3067,
      "step": 903
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.758093689236459,
      "learning_rate": 1.9997848707858476e-05,
      "loss": 1.1247,
      "step": 904
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.069832365251186,
      "learning_rate": 1.9997821563470366e-05,
      "loss": 1.5612,
      "step": 905
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.920955496689187,
      "learning_rate": 1.99977942489228e-05,
      "loss": 1.7987,
      "step": 906
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.071893410173617,
      "learning_rate": 1.9997766764216243e-05,
      "loss": 1.2922,
      "step": 907
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.7541456085918705,
      "learning_rate": 1.9997739109351164e-05,
      "loss": 1.4874,
      "step": 908
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.223701762165308,
      "learning_rate": 1.999771128432803e-05,
      "loss": 1.5645,
      "step": 909
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.091512140312028,
      "learning_rate": 1.9997683289147314e-05,
      "loss": 1.0261,
      "step": 910
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.646671091088304,
      "learning_rate": 1.9997655123809495e-05,
      "loss": 1.6216,
      "step": 911
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.17266627377912,
      "learning_rate": 1.9997626788315056e-05,
      "loss": 1.3076,
      "step": 912
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.879290927576934,
      "learning_rate": 1.9997598282664467e-05,
      "loss": 1.4595,
      "step": 913
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.775935161279197,
      "learning_rate": 1.9997569606858227e-05,
      "loss": 1.6569,
      "step": 914
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.834018863311575,
      "learning_rate": 1.9997540760896817e-05,
      "loss": 1.0737,
      "step": 915
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.574795828722781,
      "learning_rate": 1.9997511744780732e-05,
      "loss": 1.5492,
      "step": 916
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.93404331579417,
      "learning_rate": 1.999748255851046e-05,
      "loss": 1.6876,
      "step": 917
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.144081585405643,
      "learning_rate": 1.99974532020865e-05,
      "loss": 1.3208,
      "step": 918
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.365463924681675,
      "learning_rate": 1.9997423675509354e-05,
      "loss": 1.6095,
      "step": 919
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.84847043991765,
      "learning_rate": 1.9997393978779524e-05,
      "loss": 1.878,
      "step": 920
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.625038671080887,
      "learning_rate": 1.9997364111897512e-05,
      "loss": 1.3495,
      "step": 921
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.459081527348825,
      "learning_rate": 1.999733407486383e-05,
      "loss": 1.7466,
      "step": 922
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.798782472513242,
      "learning_rate": 1.9997303867678986e-05,
      "loss": 1.2825,
      "step": 923
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.811219471879745,
      "learning_rate": 1.99972734903435e-05,
      "loss": 1.197,
      "step": 924
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.449292848412114,
      "learning_rate": 1.9997242942857883e-05,
      "loss": 1.5957,
      "step": 925
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.077292673562203,
      "learning_rate": 1.9997212225222657e-05,
      "loss": 1.6007,
      "step": 926
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.920999201052989,
      "learning_rate": 1.9997181337438346e-05,
      "loss": 1.7541,
      "step": 927
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.803679177788871,
      "learning_rate": 1.9997150279505476e-05,
      "loss": 1.2353,
      "step": 928
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.885007517961068,
      "learning_rate": 1.999711905142457e-05,
      "loss": 1.3159,
      "step": 929
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.795986035123478,
      "learning_rate": 1.9997087653196166e-05,
      "loss": 1.486,
      "step": 930
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.694050399489354,
      "learning_rate": 1.9997056084820797e-05,
      "loss": 1.3817,
      "step": 931
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.540591863416799,
      "learning_rate": 1.9997024346298996e-05,
      "loss": 1.4426,
      "step": 932
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.114919773107593,
      "learning_rate": 1.999699243763131e-05,
      "loss": 1.4093,
      "step": 933
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.317358439773628,
      "learning_rate": 1.9996960358818277e-05,
      "loss": 1.1704,
      "step": 934
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.624932923310686,
      "learning_rate": 1.9996928109860442e-05,
      "loss": 1.5912,
      "step": 935
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.52554577414443,
      "learning_rate": 1.9996895690758358e-05,
      "loss": 1.48,
      "step": 936
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.018647240373812,
      "learning_rate": 1.9996863101512576e-05,
      "loss": 1.2244,
      "step": 937
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.9780436074401395,
      "learning_rate": 1.9996830342123648e-05,
      "loss": 1.6141,
      "step": 938
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.816056822306007,
      "learning_rate": 1.9996797412592137e-05,
      "loss": 1.1869,
      "step": 939
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.184321348035454,
      "learning_rate": 1.99967643129186e-05,
      "loss": 1.4055,
      "step": 940
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.261313600473951,
      "learning_rate": 1.999673104310359e-05,
      "loss": 1.6616,
      "step": 941
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.1815032758984145,
      "learning_rate": 1.9996697603147696e-05,
      "loss": 1.3349,
      "step": 942
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.980776032792113,
      "learning_rate": 1.999666399305147e-05,
      "loss": 1.5637,
      "step": 943
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.415747393477452,
      "learning_rate": 1.9996630212815486e-05,
      "loss": 1.2148,
      "step": 944
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.98050192302215,
      "learning_rate": 1.999659626244032e-05,
      "loss": 1.3023,
      "step": 945
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.472954369019742,
      "learning_rate": 1.9996562141926557e-05,
      "loss": 1.5546,
      "step": 946
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.559545023053875,
      "learning_rate": 1.999652785127477e-05,
      "loss": 1.6487,
      "step": 947
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.144467829269692,
      "learning_rate": 1.9996493390485545e-05,
      "loss": 1.4299,
      "step": 948
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.671945924005451,
      "learning_rate": 1.9996458759559468e-05,
      "loss": 1.4013,
      "step": 949
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.3728415567973,
      "learning_rate": 1.999642395849713e-05,
      "loss": 1.8359,
      "step": 950
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.200381167291844,
      "learning_rate": 1.9996388987299117e-05,
      "loss": 1.6116,
      "step": 951
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.141567012674161,
      "learning_rate": 1.9996353845966033e-05,
      "loss": 1.3249,
      "step": 952
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.616278529581532,
      "learning_rate": 1.999631853449847e-05,
      "loss": 1.4081,
      "step": 953
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.010885292179237,
      "learning_rate": 1.9996283052897035e-05,
      "loss": 1.2937,
      "step": 954
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.593531996326042,
      "learning_rate": 1.9996247401162324e-05,
      "loss": 1.3638,
      "step": 955
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.240717284488935,
      "learning_rate": 1.9996211579294947e-05,
      "loss": 1.4056,
      "step": 956
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.393327986303736,
      "learning_rate": 1.999617558729552e-05,
      "loss": 1.4116,
      "step": 957
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.011379452200711,
      "learning_rate": 1.9996139425164644e-05,
      "loss": 1.2791,
      "step": 958
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.075644223576186,
      "learning_rate": 1.9996103092902944e-05,
      "loss": 1.5012,
      "step": 959
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.966514585664333,
      "learning_rate": 1.9996066590511032e-05,
      "loss": 1.3111,
      "step": 960
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.011836781072017,
      "learning_rate": 1.9996029917989537e-05,
      "loss": 1.68,
      "step": 961
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.747085036174793,
      "learning_rate": 1.999599307533907e-05,
      "loss": 1.396,
      "step": 962
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.875664391433856,
      "learning_rate": 1.9995956062560274e-05,
      "loss": 1.5381,
      "step": 963
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.612536592671347,
      "learning_rate": 1.999591887965377e-05,
      "loss": 1.3646,
      "step": 964
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.0717609666381005,
      "learning_rate": 1.999588152662019e-05,
      "loss": 1.2591,
      "step": 965
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.2311529315973,
      "learning_rate": 1.9995844003460173e-05,
      "loss": 1.6544,
      "step": 966
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.683708922212679,
      "learning_rate": 1.9995806310174357e-05,
      "loss": 1.1797,
      "step": 967
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.892594769306384,
      "learning_rate": 1.9995768446763383e-05,
      "loss": 1.0544,
      "step": 968
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.680636210293175,
      "learning_rate": 1.9995730413227894e-05,
      "loss": 1.6938,
      "step": 969
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.679661199812549,
      "learning_rate": 1.999569220956854e-05,
      "loss": 1.7427,
      "step": 970
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.513474271290572,
      "learning_rate": 1.999565383578597e-05,
      "loss": 1.3672,
      "step": 971
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.74898764281988,
      "learning_rate": 1.9995615291880838e-05,
      "loss": 1.4479,
      "step": 972
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.706665898881583,
      "learning_rate": 1.9995576577853794e-05,
      "loss": 1.7845,
      "step": 973
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.444991628194857,
      "learning_rate": 1.9995537693705506e-05,
      "loss": 1.5322,
      "step": 974
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.829607735225315,
      "learning_rate": 1.9995498639436634e-05,
      "loss": 1.5932,
      "step": 975
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.42211434977875,
      "learning_rate": 1.9995459415047835e-05,
      "loss": 1.3356,
      "step": 976
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.837787731391683,
      "learning_rate": 1.9995420020539783e-05,
      "loss": 1.684,
      "step": 977
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.901813918701222,
      "learning_rate": 1.9995380455913153e-05,
      "loss": 1.5276,
      "step": 978
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.79887242832733,
      "learning_rate": 1.999534072116861e-05,
      "loss": 1.5357,
      "step": 979
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.534717527296219,
      "learning_rate": 1.9995300816306832e-05,
      "loss": 1.5181,
      "step": 980
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.040834051294794,
      "learning_rate": 1.9995260741328496e-05,
      "loss": 1.5596,
      "step": 981
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.180512668344116,
      "learning_rate": 1.9995220496234292e-05,
      "loss": 1.3328,
      "step": 982
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.772007216600503,
      "learning_rate": 1.99951800810249e-05,
      "loss": 1.5205,
      "step": 983
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.6853981053446905,
      "learning_rate": 1.999513949570101e-05,
      "loss": 1.3577,
      "step": 984
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.22475066772042,
      "learning_rate": 1.9995098740263308e-05,
      "loss": 1.3033,
      "step": 985
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.833301425126583,
      "learning_rate": 1.9995057814712488e-05,
      "loss": 1.5611,
      "step": 986
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.434479476544725,
      "learning_rate": 1.9995016719049255e-05,
      "loss": 1.6259,
      "step": 987
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.871294747787312,
      "learning_rate": 1.99949754532743e-05,
      "loss": 1.4339,
      "step": 988
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.140778691577871,
      "learning_rate": 1.9994934017388326e-05,
      "loss": 1.5787,
      "step": 989
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.505653973831968,
      "learning_rate": 1.9994892411392043e-05,
      "loss": 1.2681,
      "step": 990
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.700502500654452,
      "learning_rate": 1.9994850635286153e-05,
      "loss": 1.2548,
      "step": 991
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.448845610114987,
      "learning_rate": 1.9994808689071374e-05,
      "loss": 1.475,
      "step": 992
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.137376055457139,
      "learning_rate": 1.9994766572748413e-05,
      "loss": 1.5991,
      "step": 993
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.697492373898864,
      "learning_rate": 1.9994724286317996e-05,
      "loss": 1.5039,
      "step": 994
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.322005883310929,
      "learning_rate": 1.999468182978083e-05,
      "loss": 1.1898,
      "step": 995
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.125618165303836,
      "learning_rate": 1.9994639203137647e-05,
      "loss": 1.2526,
      "step": 996
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.727283980793183,
      "learning_rate": 1.999459640638917e-05,
      "loss": 1.3563,
      "step": 997
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.516814097533212,
      "learning_rate": 1.9994553439536125e-05,
      "loss": 1.651,
      "step": 998
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.174027831910559,
      "learning_rate": 1.9994510302579248e-05,
      "loss": 1.445,
      "step": 999
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.600799823324763,
      "learning_rate": 1.999446699551927e-05,
      "loss": 1.242,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.054816642540922,
      "learning_rate": 1.9994423518356926e-05,
      "loss": 1.3757,
      "step": 1001
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.19061321706073,
      "learning_rate": 1.999437987109296e-05,
      "loss": 1.5227,
      "step": 1002
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.280727576755848,
      "learning_rate": 1.9994336053728118e-05,
      "loss": 1.2921,
      "step": 1003
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.419717502732885,
      "learning_rate": 1.9994292066263138e-05,
      "loss": 1.3911,
      "step": 1004
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.801942725902244,
      "learning_rate": 1.999424790869877e-05,
      "loss": 1.8957,
      "step": 1005
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.873363949094151,
      "learning_rate": 1.999420358103577e-05,
      "loss": 1.7875,
      "step": 1006
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.1484383855677125,
      "learning_rate": 1.9994159083274886e-05,
      "loss": 1.3423,
      "step": 1007
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.896922711767914,
      "learning_rate": 1.9994114415416884e-05,
      "loss": 1.0747,
      "step": 1008
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.178020421771567,
      "learning_rate": 1.9994069577462517e-05,
      "loss": 1.3024,
      "step": 1009
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.426796083788023,
      "learning_rate": 1.999402456941255e-05,
      "loss": 1.3702,
      "step": 1010
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.196536156645012,
      "learning_rate": 1.9993979391267752e-05,
      "loss": 1.6199,
      "step": 1011
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.268881628158054,
      "learning_rate": 1.9993934043028886e-05,
      "loss": 1.5233,
      "step": 1012
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.732571140657865,
      "learning_rate": 1.999388852469673e-05,
      "loss": 1.3941,
      "step": 1013
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.521183511351005,
      "learning_rate": 1.9993842836272054e-05,
      "loss": 1.6458,
      "step": 1014
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.114005870907644,
      "learning_rate": 1.999379697775564e-05,
      "loss": 1.376,
      "step": 1015
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.712721168899753,
      "learning_rate": 1.9993750949148266e-05,
      "loss": 1.4283,
      "step": 1016
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.536587072851685,
      "learning_rate": 1.999370475045072e-05,
      "loss": 1.4373,
      "step": 1017
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.59251515402904,
      "learning_rate": 1.9993658381663773e-05,
      "loss": 1.6832,
      "step": 1018
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.311404974494539,
      "learning_rate": 1.9993611842788233e-05,
      "loss": 1.5442,
      "step": 1019
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.654078000564674,
      "learning_rate": 1.9993565133824882e-05,
      "loss": 1.3342,
      "step": 1020
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.867182038706527,
      "learning_rate": 1.9993518254774517e-05,
      "loss": 1.3485,
      "step": 1021
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.481603750153433,
      "learning_rate": 1.9993471205637934e-05,
      "loss": 1.1429,
      "step": 1022
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.691290119472744,
      "learning_rate": 1.999342398641594e-05,
      "loss": 1.3617,
      "step": 1023
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.788521076069781,
      "learning_rate": 1.9993376597109332e-05,
      "loss": 1.3175,
      "step": 1024
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.858053878268503,
      "learning_rate": 1.9993329037718916e-05,
      "loss": 1.2998,
      "step": 1025
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.147726962331111,
      "learning_rate": 1.9993281308245507e-05,
      "loss": 1.8507,
      "step": 1026
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.046559033580077,
      "learning_rate": 1.9993233408689915e-05,
      "loss": 1.5495,
      "step": 1027
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.497128773485002,
      "learning_rate": 1.9993185339052956e-05,
      "loss": 1.3018,
      "step": 1028
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.00542676787873,
      "learning_rate": 1.9993137099335445e-05,
      "loss": 1.8672,
      "step": 1029
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.014732185832456,
      "learning_rate": 1.9993088689538206e-05,
      "loss": 1.4667,
      "step": 1030
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.444986002334135,
      "learning_rate": 1.999304010966206e-05,
      "loss": 1.1086,
      "step": 1031
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.662606660008082,
      "learning_rate": 1.9992991359707837e-05,
      "loss": 1.4395,
      "step": 1032
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.489408846017827,
      "learning_rate": 1.999294243967637e-05,
      "loss": 1.1524,
      "step": 1033
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.203708282250208,
      "learning_rate": 1.9992893349568482e-05,
      "loss": 1.4725,
      "step": 1034
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.135229336920883,
      "learning_rate": 1.999284408938501e-05,
      "loss": 1.6593,
      "step": 1035
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.035658867456814,
      "learning_rate": 1.9992794659126806e-05,
      "loss": 1.4783,
      "step": 1036
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.54452779435401,
      "learning_rate": 1.9992745058794694e-05,
      "loss": 1.5312,
      "step": 1037
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.7457926274126985,
      "learning_rate": 1.9992695288389527e-05,
      "loss": 1.3954,
      "step": 1038
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.725314644713983,
      "learning_rate": 1.999264534791215e-05,
      "loss": 1.2731,
      "step": 1039
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.177025012280493,
      "learning_rate": 1.9992595237363415e-05,
      "loss": 1.6093,
      "step": 1040
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.587725964785506,
      "learning_rate": 1.9992544956744174e-05,
      "loss": 1.3668,
      "step": 1041
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.827510771105642,
      "learning_rate": 1.999249450605528e-05,
      "loss": 1.41,
      "step": 1042
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.951484119297973,
      "learning_rate": 1.9992443885297592e-05,
      "loss": 1.7087,
      "step": 1043
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.675934905959847,
      "learning_rate": 1.9992393094471976e-05,
      "loss": 1.3769,
      "step": 1044
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.617488507898361,
      "learning_rate": 1.9992342133579295e-05,
      "loss": 1.3218,
      "step": 1045
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.424149798609915,
      "learning_rate": 1.9992291002620412e-05,
      "loss": 1.384,
      "step": 1046
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.125964742977812,
      "learning_rate": 1.9992239701596202e-05,
      "loss": 1.6464,
      "step": 1047
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.711751849837585,
      "learning_rate": 1.999218823050754e-05,
      "loss": 1.3026,
      "step": 1048
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.7981425952959835,
      "learning_rate": 1.9992136589355292e-05,
      "loss": 1.4123,
      "step": 1049
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.174696792080061,
      "learning_rate": 1.9992084778140347e-05,
      "loss": 1.5308,
      "step": 1050
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.15037456468554,
      "learning_rate": 1.9992032796863584e-05,
      "loss": 1.4467,
      "step": 1051
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.769767446998471,
      "learning_rate": 1.9991980645525882e-05,
      "loss": 1.3248,
      "step": 1052
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.989813356447352,
      "learning_rate": 1.999192832412814e-05,
      "loss": 1.4501,
      "step": 1053
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.08183533405334,
      "learning_rate": 1.999187583267124e-05,
      "loss": 1.3976,
      "step": 1054
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.202110512643497,
      "learning_rate": 1.9991823171156072e-05,
      "loss": 1.5613,
      "step": 1055
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.552436474798843,
      "learning_rate": 1.9991770339583544e-05,
      "loss": 1.5689,
      "step": 1056
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.75306964654212,
      "learning_rate": 1.9991717337954545e-05,
      "loss": 1.3197,
      "step": 1057
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.270917445460211,
      "learning_rate": 1.999166416626998e-05,
      "loss": 1.3265,
      "step": 1058
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.846781645425409,
      "learning_rate": 1.999161082453076e-05,
      "loss": 1.3841,
      "step": 1059
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.422739618053509,
      "learning_rate": 1.9991557312737787e-05,
      "loss": 1.2695,
      "step": 1060
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.790371697998435,
      "learning_rate": 1.9991503630891967e-05,
      "loss": 1.1158,
      "step": 1061
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.221746567787163,
      "learning_rate": 1.9991449778994223e-05,
      "loss": 1.6151,
      "step": 1062
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.592692375084283,
      "learning_rate": 1.9991395757045467e-05,
      "loss": 1.0144,
      "step": 1063
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.434188342257871,
      "learning_rate": 1.999134156504662e-05,
      "loss": 1.5019,
      "step": 1064
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.8651413437217945,
      "learning_rate": 1.9991287202998602e-05,
      "loss": 1.454,
      "step": 1065
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.232482694551734,
      "learning_rate": 1.9991232670902343e-05,
      "loss": 1.5478,
      "step": 1066
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.632265934540515,
      "learning_rate": 1.9991177968758764e-05,
      "loss": 1.5036,
      "step": 1067
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.04449328756143,
      "learning_rate": 1.9991123096568802e-05,
      "loss": 1.2138,
      "step": 1068
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.275985218567797,
      "learning_rate": 1.9991068054333387e-05,
      "loss": 1.7905,
      "step": 1069
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.266771284315821,
      "learning_rate": 1.999101284205346e-05,
      "loss": 1.4896,
      "step": 1070
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.00377099296883,
      "learning_rate": 1.9990957459729956e-05,
      "loss": 1.1116,
      "step": 1071
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.550448812268904,
      "learning_rate": 1.999090190736382e-05,
      "loss": 1.6304,
      "step": 1072
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.439208143688307,
      "learning_rate": 1.9990846184955998e-05,
      "loss": 1.5372,
      "step": 1073
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.904235022439844,
      "learning_rate": 1.9990790292507434e-05,
      "loss": 1.6194,
      "step": 1074
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.177638332439812,
      "learning_rate": 1.9990734230019086e-05,
      "loss": 1.2,
      "step": 1075
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.96615675756491,
      "learning_rate": 1.9990677997491906e-05,
      "loss": 1.9356,
      "step": 1076
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.300314222220389,
      "learning_rate": 1.999062159492685e-05,
      "loss": 1.3247,
      "step": 1077
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.3795764096347645,
      "learning_rate": 1.9990565022324877e-05,
      "loss": 1.3849,
      "step": 1078
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.5033674087850635,
      "learning_rate": 1.9990508279686952e-05,
      "loss": 1.5566,
      "step": 1079
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.516357839503339,
      "learning_rate": 1.9990451367014033e-05,
      "loss": 1.4577,
      "step": 1080
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.895091733776823,
      "learning_rate": 1.9990394284307105e-05,
      "loss": 1.3428,
      "step": 1081
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.982195042920531,
      "learning_rate": 1.9990337031567124e-05,
      "loss": 1.0707,
      "step": 1082
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.703179984539122,
      "learning_rate": 1.999027960879507e-05,
      "loss": 1.4839,
      "step": 1083
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.091692019999666,
      "learning_rate": 1.9990222015991925e-05,
      "loss": 1.1921,
      "step": 1084
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.073775033096179,
      "learning_rate": 1.999016425315866e-05,
      "loss": 1.1458,
      "step": 1085
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.893720118063953,
      "learning_rate": 1.9990106320296266e-05,
      "loss": 1.2642,
      "step": 1086
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.829825252599104,
      "learning_rate": 1.9990048217405725e-05,
      "loss": 1.366,
      "step": 1087
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.689235875990117,
      "learning_rate": 1.9989989944488028e-05,
      "loss": 1.2178,
      "step": 1088
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.37470020278995,
      "learning_rate": 1.998993150154416e-05,
      "loss": 1.2546,
      "step": 1089
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.497384595560334,
      "learning_rate": 1.998987288857513e-05,
      "loss": 1.493,
      "step": 1090
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.027493470938268,
      "learning_rate": 1.9989814105581923e-05,
      "loss": 1.2675,
      "step": 1091
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.421792511325926,
      "learning_rate": 1.998975515256554e-05,
      "loss": 1.4063,
      "step": 1092
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.436662927099091,
      "learning_rate": 1.998969602952699e-05,
      "loss": 1.4665,
      "step": 1093
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.0810915679503,
      "learning_rate": 1.9989636736467278e-05,
      "loss": 1.3567,
      "step": 1094
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.641916997220099,
      "learning_rate": 1.9989577273387412e-05,
      "loss": 1.251,
      "step": 1095
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.373019408079179,
      "learning_rate": 1.9989517640288402e-05,
      "loss": 1.3061,
      "step": 1096
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.162205263710801,
      "learning_rate": 1.9989457837171266e-05,
      "loss": 1.783,
      "step": 1097
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.3618504008006935,
      "learning_rate": 1.998939786403702e-05,
      "loss": 1.2785,
      "step": 1098
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.9690183771913095,
      "learning_rate": 1.9989337720886685e-05,
      "loss": 1.3401,
      "step": 1099
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.571866725799089,
      "learning_rate": 1.9989277407721287e-05,
      "loss": 1.2627,
      "step": 1100
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.376809660231449,
      "learning_rate": 1.998921692454185e-05,
      "loss": 1.3289,
      "step": 1101
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.199059706561627,
      "learning_rate": 1.9989156271349405e-05,
      "loss": 1.1648,
      "step": 1102
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.121096665549203,
      "learning_rate": 1.998909544814498e-05,
      "loss": 1.2336,
      "step": 1103
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.943995396406024,
      "learning_rate": 1.9989034454929616e-05,
      "loss": 1.4747,
      "step": 1104
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.553381325124945,
      "learning_rate": 1.998897329170435e-05,
      "loss": 1.4922,
      "step": 1105
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.764048438199058,
      "learning_rate": 1.998891195847022e-05,
      "loss": 1.335,
      "step": 1106
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.041997050681997,
      "learning_rate": 1.9988850455228273e-05,
      "loss": 1.4317,
      "step": 1107
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.960319512562217,
      "learning_rate": 1.9988788781979554e-05,
      "loss": 1.327,
      "step": 1108
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.731941091844931,
      "learning_rate": 1.9988726938725113e-05,
      "loss": 1.4945,
      "step": 1109
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.900597433402245,
      "learning_rate": 1.9988664925465998e-05,
      "loss": 1.5907,
      "step": 1110
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.306397185153227,
      "learning_rate": 1.9988602742203275e-05,
      "loss": 1.2107,
      "step": 1111
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.40629404572251,
      "learning_rate": 1.9988540388937993e-05,
      "loss": 1.0309,
      "step": 1112
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.374400902175227,
      "learning_rate": 1.9988477865671217e-05,
      "loss": 1.613,
      "step": 1113
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.523030994512139,
      "learning_rate": 1.998841517240401e-05,
      "loss": 1.4283,
      "step": 1114
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.830620363542188,
      "learning_rate": 1.998835230913744e-05,
      "loss": 1.655,
      "step": 1115
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.353142651649498,
      "learning_rate": 1.9988289275872576e-05,
      "loss": 1.3767,
      "step": 1116
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.21912231695581,
      "learning_rate": 1.9988226072610494e-05,
      "loss": 1.3033,
      "step": 1117
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.814991805506658,
      "learning_rate": 1.9988162699352264e-05,
      "loss": 1.3728,
      "step": 1118
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.288018848422014,
      "learning_rate": 1.9988099156098966e-05,
      "loss": 1.1978,
      "step": 1119
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.420668415998671,
      "learning_rate": 1.9988035442851686e-05,
      "loss": 1.6142,
      "step": 1120
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.451789256923396,
      "learning_rate": 1.9987971559611507e-05,
      "loss": 1.3732,
      "step": 1121
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.408693781857968,
      "learning_rate": 1.9987907506379513e-05,
      "loss": 1.4327,
      "step": 1122
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.32806376782246,
      "learning_rate": 1.9987843283156795e-05,
      "loss": 1.4698,
      "step": 1123
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.8335292889247725,
      "learning_rate": 1.9987778889944448e-05,
      "loss": 1.4701,
      "step": 1124
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.6765760616675385,
      "learning_rate": 1.9987714326743566e-05,
      "loss": 1.0584,
      "step": 1125
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.871944992225571,
      "learning_rate": 1.998764959355525e-05,
      "loss": 1.3631,
      "step": 1126
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.758705382655558,
      "learning_rate": 1.99875846903806e-05,
      "loss": 1.5984,
      "step": 1127
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.337426195414603,
      "learning_rate": 1.998751961722072e-05,
      "loss": 1.5327,
      "step": 1128
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.045436515490048,
      "learning_rate": 1.998745437407672e-05,
      "loss": 1.6327,
      "step": 1129
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.146993945227617,
      "learning_rate": 1.998738896094971e-05,
      "loss": 1.5954,
      "step": 1130
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.01110302002289,
      "learning_rate": 1.9987323377840802e-05,
      "loss": 1.7811,
      "step": 1131
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.410782251349829,
      "learning_rate": 1.9987257624751112e-05,
      "loss": 1.1588,
      "step": 1132
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.963648512349303,
      "learning_rate": 1.9987191701681757e-05,
      "loss": 1.1468,
      "step": 1133
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.707446085421224,
      "learning_rate": 1.9987125608633863e-05,
      "loss": 1.5631,
      "step": 1134
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.65735247981638,
      "learning_rate": 1.9987059345608557e-05,
      "loss": 1.4013,
      "step": 1135
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.726827874090467,
      "learning_rate": 1.998699291260696e-05,
      "loss": 1.7462,
      "step": 1136
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.7340754261406,
      "learning_rate": 1.9986926309630208e-05,
      "loss": 1.4953,
      "step": 1137
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.017154576768546,
      "learning_rate": 1.998685953667943e-05,
      "loss": 1.2793,
      "step": 1138
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.6742403901590786,
      "learning_rate": 1.998679259375577e-05,
      "loss": 1.1419,
      "step": 1139
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.576496007981129,
      "learning_rate": 1.9986725480860355e-05,
      "loss": 1.2548,
      "step": 1140
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.145869110067948,
      "learning_rate": 1.998665819799434e-05,
      "loss": 1.4126,
      "step": 1141
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.513297955882855,
      "learning_rate": 1.9986590745158864e-05,
      "loss": 1.8085,
      "step": 1142
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.178169495109167,
      "learning_rate": 1.9986523122355073e-05,
      "loss": 1.3133,
      "step": 1143
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.17626904446269,
      "learning_rate": 1.998645532958412e-05,
      "loss": 1.3063,
      "step": 1144
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.556993280300207,
      "learning_rate": 1.9986387366847164e-05,
      "loss": 1.0484,
      "step": 1145
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.659703484377311,
      "learning_rate": 1.9986319234145356e-05,
      "loss": 1.4591,
      "step": 1146
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.240316368792554,
      "learning_rate": 1.9986250931479857e-05,
      "loss": 1.3616,
      "step": 1147
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.995479266482959,
      "learning_rate": 1.9986182458851824e-05,
      "loss": 1.4465,
      "step": 1148
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.754078073630623,
      "learning_rate": 1.9986113816262433e-05,
      "loss": 1.3476,
      "step": 1149
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.213156906259343,
      "learning_rate": 1.9986045003712845e-05,
      "loss": 1.3594,
      "step": 1150
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.44331677429967,
      "learning_rate": 1.9985976021204236e-05,
      "loss": 1.5596,
      "step": 1151
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.722628412584643,
      "learning_rate": 1.998590686873777e-05,
      "loss": 1.2554,
      "step": 1152
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.201822462865456,
      "learning_rate": 1.9985837546314637e-05,
      "loss": 1.6201,
      "step": 1153
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.1481882654429505,
      "learning_rate": 1.9985768053936012e-05,
      "loss": 1.3285,
      "step": 1154
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.697784955203706,
      "learning_rate": 1.9985698391603076e-05,
      "loss": 1.2146,
      "step": 1155
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.575946803991559,
      "learning_rate": 1.998562855931701e-05,
      "loss": 1.4095,
      "step": 1156
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.6831928515601575,
      "learning_rate": 1.9985558557079013e-05,
      "loss": 1.5531,
      "step": 1157
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.055767479507656,
      "learning_rate": 1.9985488384890268e-05,
      "loss": 1.153,
      "step": 1158
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.647366893362251,
      "learning_rate": 1.9985418042751975e-05,
      "loss": 1.4351,
      "step": 1159
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.379921029593668,
      "learning_rate": 1.9985347530665327e-05,
      "loss": 1.281,
      "step": 1160
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.122118194239811,
      "learning_rate": 1.9985276848631526e-05,
      "loss": 1.5779,
      "step": 1161
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.080564586428036,
      "learning_rate": 1.9985205996651776e-05,
      "loss": 1.4505,
      "step": 1162
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.3925020052979225,
      "learning_rate": 1.998513497472728e-05,
      "loss": 1.3472,
      "step": 1163
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.336455626270903,
      "learning_rate": 1.998506378285925e-05,
      "loss": 1.3863,
      "step": 1164
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.466951551893215,
      "learning_rate": 1.9984992421048898e-05,
      "loss": 1.1468,
      "step": 1165
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.858953394755017,
      "learning_rate": 1.9984920889297432e-05,
      "loss": 1.4421,
      "step": 1166
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.400100271260996,
      "learning_rate": 1.9984849187606078e-05,
      "loss": 1.3454,
      "step": 1167
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.120267437658351,
      "learning_rate": 1.998477731597605e-05,
      "loss": 1.3566,
      "step": 1168
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.066949897519249,
      "learning_rate": 1.9984705274408574e-05,
      "loss": 1.282,
      "step": 1169
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.666664242902167,
      "learning_rate": 1.9984633062904873e-05,
      "loss": 1.7577,
      "step": 1170
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.282790140133076,
      "learning_rate": 1.9984560681466184e-05,
      "loss": 1.3367,
      "step": 1171
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.451778489057455,
      "learning_rate": 1.9984488130093732e-05,
      "loss": 1.231,
      "step": 1172
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.793557985622377,
      "learning_rate": 1.998441540878875e-05,
      "loss": 1.2448,
      "step": 1173
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.816243707827969,
      "learning_rate": 1.9984342517552485e-05,
      "loss": 1.3335,
      "step": 1174
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.6671012390821,
      "learning_rate": 1.998426945638617e-05,
      "loss": 1.6317,
      "step": 1175
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.344876307572289,
      "learning_rate": 1.9984196225291052e-05,
      "loss": 1.0844,
      "step": 1176
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4696141731361325,
      "learning_rate": 1.9984122824268372e-05,
      "loss": 0.9681,
      "step": 1177
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.009083209285846,
      "learning_rate": 1.9984049253319383e-05,
      "loss": 1.4656,
      "step": 1178
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.534199764603587,
      "learning_rate": 1.9983975512445337e-05,
      "loss": 1.6483,
      "step": 1179
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.50637888443161,
      "learning_rate": 1.998390160164749e-05,
      "loss": 1.6886,
      "step": 1180
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.642440148319819,
      "learning_rate": 1.99838275209271e-05,
      "loss": 1.5254,
      "step": 1181
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.678685568390385,
      "learning_rate": 1.9983753270285423e-05,
      "loss": 1.7299,
      "step": 1182
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.50557903745635,
      "learning_rate": 1.998367884972373e-05,
      "loss": 1.7553,
      "step": 1183
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.985319747550658,
      "learning_rate": 1.9983604259243284e-05,
      "loss": 1.482,
      "step": 1184
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.555131735123524,
      "learning_rate": 1.9983529498845355e-05,
      "loss": 1.3364,
      "step": 1185
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.143094192730966,
      "learning_rate": 1.9983454568531215e-05,
      "loss": 1.5031,
      "step": 1186
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.310516474342615,
      "learning_rate": 1.9983379468302136e-05,
      "loss": 1.4171,
      "step": 1187
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.0725500170010855,
      "learning_rate": 1.9983304198159404e-05,
      "loss": 0.9257,
      "step": 1188
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.083362159453924,
      "learning_rate": 1.9983228758104295e-05,
      "loss": 1.1782,
      "step": 1189
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.463812605050002,
      "learning_rate": 1.9983153148138094e-05,
      "loss": 1.6367,
      "step": 1190
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.618756562321986,
      "learning_rate": 1.9983077368262084e-05,
      "loss": 1.6135,
      "step": 1191
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.545773539425761,
      "learning_rate": 1.998300141847756e-05,
      "loss": 1.2085,
      "step": 1192
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.125736597001142,
      "learning_rate": 1.9982925298785813e-05,
      "loss": 1.6518,
      "step": 1193
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.8792477957971245,
      "learning_rate": 1.9982849009188138e-05,
      "loss": 1.3272,
      "step": 1194
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.952758240196858,
      "learning_rate": 1.9982772549685834e-05,
      "loss": 1.6702,
      "step": 1195
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.558048025667784,
      "learning_rate": 1.9982695920280203e-05,
      "loss": 1.3478,
      "step": 1196
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.130211683446369,
      "learning_rate": 1.9982619120972547e-05,
      "loss": 1.4685,
      "step": 1197
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.94752275980667,
      "learning_rate": 1.9982542151764176e-05,
      "loss": 1.5225,
      "step": 1198
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.051667879884348,
      "learning_rate": 1.9982465012656395e-05,
      "loss": 1.1539,
      "step": 1199
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.111269567684271,
      "learning_rate": 1.998238770365052e-05,
      "loss": 1.412,
      "step": 1200
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.781262453022093,
      "learning_rate": 1.998231022474787e-05,
      "loss": 1.2496,
      "step": 1201
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.797413020456786,
      "learning_rate": 1.9982232575949764e-05,
      "loss": 1.2036,
      "step": 1202
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.015556562711532,
      "learning_rate": 1.9982154757257515e-05,
      "loss": 1.4245,
      "step": 1203
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.866238147136575,
      "learning_rate": 1.9982076768672453e-05,
      "loss": 1.2116,
      "step": 1204
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.561758853761197,
      "learning_rate": 1.998199861019591e-05,
      "loss": 1.4311,
      "step": 1205
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.174754019457683,
      "learning_rate": 1.9981920281829207e-05,
      "loss": 1.3279,
      "step": 1206
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.619812424309057,
      "learning_rate": 1.998184178357368e-05,
      "loss": 1.4269,
      "step": 1207
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.394987705318756,
      "learning_rate": 1.9981763115430666e-05,
      "loss": 1.6773,
      "step": 1208
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.426016299014888,
      "learning_rate": 1.9981684277401506e-05,
      "loss": 1.3274,
      "step": 1209
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.294887037371584,
      "learning_rate": 1.9981605269487543e-05,
      "loss": 1.5305,
      "step": 1210
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.796464300055534,
      "learning_rate": 1.9981526091690117e-05,
      "loss": 1.5138,
      "step": 1211
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.893280444083912,
      "learning_rate": 1.9981446744010577e-05,
      "loss": 1.3877,
      "step": 1212
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.212464041628949,
      "learning_rate": 1.9981367226450272e-05,
      "loss": 1.2192,
      "step": 1213
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.352197947992044,
      "learning_rate": 1.998128753901056e-05,
      "loss": 1.2298,
      "step": 1214
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.4543117523989,
      "learning_rate": 1.998120768169279e-05,
      "loss": 1.3555,
      "step": 1215
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.132273085524825,
      "learning_rate": 1.9981127654498335e-05,
      "loss": 1.2322,
      "step": 1216
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.388557301176299,
      "learning_rate": 1.9981047457428538e-05,
      "loss": 1.1142,
      "step": 1217
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.177780432484276,
      "learning_rate": 1.9980967090484776e-05,
      "loss": 1.4194,
      "step": 1218
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.676748628908222,
      "learning_rate": 1.9980886553668418e-05,
      "loss": 1.3767,
      "step": 1219
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.278656816329185,
      "learning_rate": 1.9980805846980828e-05,
      "loss": 1.4169,
      "step": 1220
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.608800775152978,
      "learning_rate": 1.9980724970423383e-05,
      "loss": 1.2306,
      "step": 1221
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.035396988791753,
      "learning_rate": 1.9980643923997462e-05,
      "loss": 1.4411,
      "step": 1222
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.000639820694462,
      "learning_rate": 1.9980562707704436e-05,
      "loss": 1.3169,
      "step": 1223
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.81233099634652,
      "learning_rate": 1.9980481321545695e-05,
      "loss": 1.3637,
      "step": 1224
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.842308611076228,
      "learning_rate": 1.998039976552262e-05,
      "loss": 1.5162,
      "step": 1225
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.58499874189754,
      "learning_rate": 1.9980318039636607e-05,
      "loss": 1.4434,
      "step": 1226
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.367101475806799,
      "learning_rate": 1.9980236143889038e-05,
      "loss": 1.6265,
      "step": 1227
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.98801771822008,
      "learning_rate": 1.998015407828131e-05,
      "loss": 1.5983,
      "step": 1228
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.009980584536178,
      "learning_rate": 1.998007184281482e-05,
      "loss": 1.378,
      "step": 1229
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.495230931097211,
      "learning_rate": 1.9979989437490967e-05,
      "loss": 1.3763,
      "step": 1230
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.519108385378533,
      "learning_rate": 1.9979906862311153e-05,
      "loss": 1.0846,
      "step": 1231
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.997626650944431,
      "learning_rate": 1.9979824117276784e-05,
      "loss": 1.436,
      "step": 1232
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.327859077093838,
      "learning_rate": 1.9979741202389266e-05,
      "loss": 1.4388,
      "step": 1233
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.721771545577175,
      "learning_rate": 1.997965811765002e-05,
      "loss": 1.3019,
      "step": 1234
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.48673392795667,
      "learning_rate": 1.9979574863060444e-05,
      "loss": 1.3877,
      "step": 1235
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.276455521815421,
      "learning_rate": 1.9979491438621967e-05,
      "loss": 1.4666,
      "step": 1236
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.403137209355509,
      "learning_rate": 1.9979407844336008e-05,
      "loss": 1.228,
      "step": 1237
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.021694275945022,
      "learning_rate": 1.9979324080203987e-05,
      "loss": 1.3729,
      "step": 1238
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.491883420784804,
      "learning_rate": 1.9979240146227326e-05,
      "loss": 1.2864,
      "step": 1239
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.742516425050354,
      "learning_rate": 1.997915604240746e-05,
      "loss": 1.1955,
      "step": 1240
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.438750471792213,
      "learning_rate": 1.9979071768745816e-05,
      "loss": 1.2317,
      "step": 1241
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.840317232397019,
      "learning_rate": 1.9978987325243834e-05,
      "loss": 1.0969,
      "step": 1242
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.9528970834182,
      "learning_rate": 1.9978902711902944e-05,
      "loss": 1.2987,
      "step": 1243
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.521246901237202,
      "learning_rate": 1.9978817928724588e-05,
      "loss": 1.2004,
      "step": 1244
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.865414721252148,
      "learning_rate": 1.997873297571021e-05,
      "loss": 1.2301,
      "step": 1245
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.49196308520183,
      "learning_rate": 1.997864785286126e-05,
      "loss": 1.2564,
      "step": 1246
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.505189403364644,
      "learning_rate": 1.9978562560179183e-05,
      "loss": 1.4784,
      "step": 1247
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.182909692563706,
      "learning_rate": 1.997847709766543e-05,
      "loss": 1.5122,
      "step": 1248
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.702263858571601,
      "learning_rate": 1.9978391465321453e-05,
      "loss": 1.1709,
      "step": 1249
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.204931571719213,
      "learning_rate": 1.9978305663148717e-05,
      "loss": 1.4563,
      "step": 1250
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.532946374106809,
      "learning_rate": 1.9978219691148676e-05,
      "loss": 1.5769,
      "step": 1251
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.299408141140899,
      "learning_rate": 1.9978133549322792e-05,
      "loss": 1.6734,
      "step": 1252
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.048391828637755,
      "learning_rate": 1.997804723767254e-05,
      "loss": 1.1673,
      "step": 1253
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.308771093030084,
      "learning_rate": 1.997796075619938e-05,
      "loss": 1.281,
      "step": 1254
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.135544222229945,
      "learning_rate": 1.997787410490479e-05,
      "loss": 1.1979,
      "step": 1255
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.109479217772318,
      "learning_rate": 1.997778728379024e-05,
      "loss": 1.5992,
      "step": 1256
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.9237348275568715,
      "learning_rate": 1.9977700292857207e-05,
      "loss": 1.2393,
      "step": 1257
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.993804642586544,
      "learning_rate": 1.9977613132107177e-05,
      "loss": 1.5255,
      "step": 1258
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.798595554244061,
      "learning_rate": 1.9977525801541628e-05,
      "loss": 1.4897,
      "step": 1259
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.104167810309012,
      "learning_rate": 1.997743830116205e-05,
      "loss": 1.1029,
      "step": 1260
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.4435354330140395,
      "learning_rate": 1.9977350630969933e-05,
      "loss": 1.477,
      "step": 1261
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.924202902590135,
      "learning_rate": 1.9977262790966768e-05,
      "loss": 1.1487,
      "step": 1262
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.594166592787112,
      "learning_rate": 1.9977174781154046e-05,
      "loss": 0.9261,
      "step": 1263
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.289278572250316,
      "learning_rate": 1.9977086601533272e-05,
      "loss": 1.7257,
      "step": 1264
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.781225272463888,
      "learning_rate": 1.997699825210594e-05,
      "loss": 1.4252,
      "step": 1265
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.984428742447479,
      "learning_rate": 1.9976909732873556e-05,
      "loss": 1.602,
      "step": 1266
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.021206146700822,
      "learning_rate": 1.997682104383763e-05,
      "loss": 1.4888,
      "step": 1267
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.280136120764609,
      "learning_rate": 1.9976732184999665e-05,
      "loss": 1.6424,
      "step": 1268
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.251768817059288,
      "learning_rate": 1.997664315636118e-05,
      "loss": 1.451,
      "step": 1269
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.174717110567478,
      "learning_rate": 1.9976553957923688e-05,
      "loss": 1.0553,
      "step": 1270
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.438107690991323,
      "learning_rate": 1.99764645896887e-05,
      "loss": 1.4878,
      "step": 1271
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.662200244215648,
      "learning_rate": 1.9976375051657753e-05,
      "loss": 1.4619,
      "step": 1272
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.548822899378578,
      "learning_rate": 1.9976285343832353e-05,
      "loss": 1.1558,
      "step": 1273
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.636682695192623,
      "learning_rate": 1.9976195466214042e-05,
      "loss": 1.323,
      "step": 1274
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.130434143017432,
      "learning_rate": 1.9976105418804338e-05,
      "loss": 1.3274,
      "step": 1275
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.545043591919084,
      "learning_rate": 1.997601520160478e-05,
      "loss": 1.4263,
      "step": 1276
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.412584827796017,
      "learning_rate": 1.99759248146169e-05,
      "loss": 1.6126,
      "step": 1277
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.42135890707537,
      "learning_rate": 1.9975834257842242e-05,
      "loss": 1.3288,
      "step": 1278
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.13431366627439,
      "learning_rate": 1.997574353128234e-05,
      "loss": 1.3081,
      "step": 1279
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.001090135687505,
      "learning_rate": 1.997565263493874e-05,
      "loss": 1.4558,
      "step": 1280
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.635698819584958,
      "learning_rate": 1.9975561568812994e-05,
      "loss": 1.2734,
      "step": 1281
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.9314357646433615,
      "learning_rate": 1.997547033290665e-05,
      "loss": 1.4843,
      "step": 1282
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.480072415970632,
      "learning_rate": 1.9975378927221257e-05,
      "loss": 1.5015,
      "step": 1283
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.505497398699267,
      "learning_rate": 1.9975287351758373e-05,
      "loss": 1.3908,
      "step": 1284
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.902488915356631,
      "learning_rate": 1.9975195606519556e-05,
      "loss": 1.1291,
      "step": 1285
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.941932615270398,
      "learning_rate": 1.9975103691506368e-05,
      "loss": 1.3984,
      "step": 1286
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.47877951575868,
      "learning_rate": 1.9975011606720374e-05,
      "loss": 0.9977,
      "step": 1287
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.566348842636724,
      "learning_rate": 1.997491935216314e-05,
      "loss": 1.5744,
      "step": 1288
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.223989082166951,
      "learning_rate": 1.9974826927836236e-05,
      "loss": 1.3923,
      "step": 1289
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.339019579066466,
      "learning_rate": 1.997473433374124e-05,
      "loss": 1.5108,
      "step": 1290
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.319203663600513,
      "learning_rate": 1.997464156987972e-05,
      "loss": 1.5018,
      "step": 1291
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.526390276771469,
      "learning_rate": 1.997454863625326e-05,
      "loss": 1.2867,
      "step": 1292
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.668274949695046,
      "learning_rate": 1.997445553286344e-05,
      "loss": 1.3103,
      "step": 1293
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.672380716918383,
      "learning_rate": 1.9974362259711844e-05,
      "loss": 1.5279,
      "step": 1294
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.990287659986135,
      "learning_rate": 1.997426881680006e-05,
      "loss": 1.4137,
      "step": 1295
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.861426070700963,
      "learning_rate": 1.997417520412968e-05,
      "loss": 1.8529,
      "step": 1296
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.619709724346227,
      "learning_rate": 1.9974081421702296e-05,
      "loss": 1.342,
      "step": 1297
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.281780397698333,
      "learning_rate": 1.9973987469519504e-05,
      "loss": 1.4139,
      "step": 1298
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.309080637932666,
      "learning_rate": 1.9973893347582904e-05,
      "loss": 1.6374,
      "step": 1299
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.04046855663974,
      "learning_rate": 1.9973799055894094e-05,
      "loss": 1.603,
      "step": 1300
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.500141862449772,
      "learning_rate": 1.997370459445468e-05,
      "loss": 1.5292,
      "step": 1301
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.922333711416567,
      "learning_rate": 1.9973609963266277e-05,
      "loss": 1.2449,
      "step": 1302
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.276098913412875,
      "learning_rate": 1.997351516233049e-05,
      "loss": 1.5701,
      "step": 1303
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.414216014922543,
      "learning_rate": 1.9973420191648928e-05,
      "loss": 1.3047,
      "step": 1304
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.840338640641262,
      "learning_rate": 1.9973325051223215e-05,
      "loss": 1.7686,
      "step": 1305
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.549427471224725,
      "learning_rate": 1.997322974105497e-05,
      "loss": 1.4307,
      "step": 1306
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.877854618742849,
      "learning_rate": 1.9973134261145807e-05,
      "loss": 1.4297,
      "step": 1307
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.835052890888657,
      "learning_rate": 1.997303861149736e-05,
      "loss": 1.2405,
      "step": 1308
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.870056983620416,
      "learning_rate": 1.997294279211125e-05,
      "loss": 1.2845,
      "step": 1309
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.136738218515481,
      "learning_rate": 1.9972846802989113e-05,
      "loss": 1.3764,
      "step": 1310
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.5352214864921745,
      "learning_rate": 1.997275064413258e-05,
      "loss": 1.3409,
      "step": 1311
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.149582103843652,
      "learning_rate": 1.9972654315543293e-05,
      "loss": 1.3812,
      "step": 1312
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.398368386300991,
      "learning_rate": 1.9972557817222885e-05,
      "loss": 1.2974,
      "step": 1313
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.248684577690094,
      "learning_rate": 1.9972461149172997e-05,
      "loss": 1.373,
      "step": 1314
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.048471522523522,
      "learning_rate": 1.997236431139528e-05,
      "loss": 1.5065,
      "step": 1315
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.842490955727602,
      "learning_rate": 1.9972267303891378e-05,
      "loss": 1.4348,
      "step": 1316
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.367193806942393,
      "learning_rate": 1.9972170126662945e-05,
      "loss": 1.3992,
      "step": 1317
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.339883110487712,
      "learning_rate": 1.9972072779711634e-05,
      "loss": 1.6242,
      "step": 1318
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.185983676218573,
      "learning_rate": 1.9971975263039102e-05,
      "loss": 1.0995,
      "step": 1319
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.205812040527204,
      "learning_rate": 1.9971877576647005e-05,
      "loss": 1.2978,
      "step": 1320
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.702314679345192,
      "learning_rate": 1.997177972053701e-05,
      "loss": 0.9797,
      "step": 1321
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.587133445717258,
      "learning_rate": 1.9971681694710784e-05,
      "loss": 1.3709,
      "step": 1322
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.408010095518702,
      "learning_rate": 1.9971583499169988e-05,
      "loss": 1.325,
      "step": 1323
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.668658527773487,
      "learning_rate": 1.99714851339163e-05,
      "loss": 1.2033,
      "step": 1324
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.909971208772203,
      "learning_rate": 1.9971386598951395e-05,
      "loss": 1.2628,
      "step": 1325
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.061806241487598,
      "learning_rate": 1.9971287894276946e-05,
      "loss": 1.2654,
      "step": 1326
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.0032750031536075,
      "learning_rate": 1.997118901989463e-05,
      "loss": 1.2757,
      "step": 1327
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.1985029655935335,
      "learning_rate": 1.9971089975806135e-05,
      "loss": 1.4569,
      "step": 1328
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.657985684216344,
      "learning_rate": 1.997099076201315e-05,
      "loss": 1.4145,
      "step": 1329
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.226220000720181,
      "learning_rate": 1.9970891378517355e-05,
      "loss": 1.2174,
      "step": 1330
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.750864239524953,
      "learning_rate": 1.9970791825320444e-05,
      "loss": 1.4603,
      "step": 1331
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.958585449582653,
      "learning_rate": 1.9970692102424115e-05,
      "loss": 1.4582,
      "step": 1332
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.35505735223703,
      "learning_rate": 1.9970592209830064e-05,
      "loss": 1.333,
      "step": 1333
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.78835772535074,
      "learning_rate": 1.997049214753999e-05,
      "loss": 1.6088,
      "step": 1334
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.159407594791994,
      "learning_rate": 1.9970391915555595e-05,
      "loss": 1.4315,
      "step": 1335
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.905671630994518,
      "learning_rate": 1.997029151387859e-05,
      "loss": 1.3165,
      "step": 1336
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.631150416929365,
      "learning_rate": 1.9970190942510674e-05,
      "loss": 1.4673,
      "step": 1337
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.200698391171232,
      "learning_rate": 1.997009020145357e-05,
      "loss": 1.2484,
      "step": 1338
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.392567442450567,
      "learning_rate": 1.9969989290708982e-05,
      "loss": 1.5052,
      "step": 1339
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.794361035108619,
      "learning_rate": 1.9969888210278637e-05,
      "loss": 1.3799,
      "step": 1340
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.465148174357662,
      "learning_rate": 1.996978696016425e-05,
      "loss": 1.5378,
      "step": 1341
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.045859333313788,
      "learning_rate": 1.9969685540367544e-05,
      "loss": 1.6475,
      "step": 1342
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.5918414474142955,
      "learning_rate": 1.9969583950890245e-05,
      "loss": 1.4184,
      "step": 1343
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.825143488116609,
      "learning_rate": 1.9969482191734086e-05,
      "loss": 1.6977,
      "step": 1344
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.674388114273383,
      "learning_rate": 1.9969380262900796e-05,
      "loss": 1.3704,
      "step": 1345
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.616565869653101,
      "learning_rate": 1.9969278164392116e-05,
      "loss": 1.3853,
      "step": 1346
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.431165634106452,
      "learning_rate": 1.996917589620977e-05,
      "loss": 1.5915,
      "step": 1347
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.948670467241131,
      "learning_rate": 1.996907345835551e-05,
      "loss": 1.4315,
      "step": 1348
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.871450895343091,
      "learning_rate": 1.996897085083107e-05,
      "loss": 1.2682,
      "step": 1349
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.434795215898948,
      "learning_rate": 1.996886807363821e-05,
      "loss": 1.634,
      "step": 1350
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.974471061751191,
      "learning_rate": 1.9968765126778666e-05,
      "loss": 1.3612,
      "step": 1351
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.598605382044358,
      "learning_rate": 1.9968662010254198e-05,
      "loss": 1.7867,
      "step": 1352
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.991346003621144,
      "learning_rate": 1.996855872406656e-05,
      "loss": 1.3096,
      "step": 1353
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.135931785313358,
      "learning_rate": 1.9968455268217506e-05,
      "loss": 1.1562,
      "step": 1354
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.163616841890807,
      "learning_rate": 1.99683516427088e-05,
      "loss": 1.6955,
      "step": 1355
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.504699794197886,
      "learning_rate": 1.9968247847542205e-05,
      "loss": 1.562,
      "step": 1356
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.174707742827942,
      "learning_rate": 1.9968143882719486e-05,
      "loss": 1.1909,
      "step": 1357
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.756016632837737,
      "learning_rate": 1.996803974824242e-05,
      "loss": 1.3687,
      "step": 1358
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.659431660539806,
      "learning_rate": 1.9967935444112766e-05,
      "loss": 1.573,
      "step": 1359
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.397334151266401,
      "learning_rate": 1.9967830970332305e-05,
      "loss": 1.2908,
      "step": 1360
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.019052399944366,
      "learning_rate": 1.9967726326902825e-05,
      "loss": 1.395,
      "step": 1361
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.179139620525485,
      "learning_rate": 1.9967621513826092e-05,
      "loss": 1.2932,
      "step": 1362
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.62573276640963,
      "learning_rate": 1.99675165311039e-05,
      "loss": 1.6737,
      "step": 1363
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.473040022097914,
      "learning_rate": 1.996741137873803e-05,
      "loss": 1.2636,
      "step": 1364
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.422617529299315,
      "learning_rate": 1.9967306056730276e-05,
      "loss": 1.1477,
      "step": 1365
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.69153725119968,
      "learning_rate": 1.9967200565082426e-05,
      "loss": 1.3223,
      "step": 1366
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.260358551013038,
      "learning_rate": 1.9967094903796282e-05,
      "loss": 1.744,
      "step": 1367
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.053172632007911,
      "learning_rate": 1.9966989072873636e-05,
      "loss": 1.2347,
      "step": 1368
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.628946146524508,
      "learning_rate": 1.996688307231629e-05,
      "loss": 1.1935,
      "step": 1369
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.7071394218458,
      "learning_rate": 1.996677690212605e-05,
      "loss": 1.677,
      "step": 1370
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.011482228103249,
      "learning_rate": 1.996667056230472e-05,
      "loss": 1.0503,
      "step": 1371
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.294646943173595,
      "learning_rate": 1.996656405285412e-05,
      "loss": 1.6745,
      "step": 1372
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.359235718467013,
      "learning_rate": 1.9966457373776048e-05,
      "loss": 1.5218,
      "step": 1373
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.083855847892963,
      "learning_rate": 1.9966350525072328e-05,
      "loss": 1.6229,
      "step": 1374
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.903834660424176,
      "learning_rate": 1.996624350674478e-05,
      "loss": 1.4559,
      "step": 1375
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.982463557818338,
      "learning_rate": 1.9966136318795216e-05,
      "loss": 1.4578,
      "step": 1376
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.133956956738988,
      "learning_rate": 1.9966028961225475e-05,
      "loss": 1.2484,
      "step": 1377
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.871959477089958,
      "learning_rate": 1.996592143403737e-05,
      "loss": 1.4797,
      "step": 1378
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.681588409948699,
      "learning_rate": 1.996581373723274e-05,
      "loss": 1.3188,
      "step": 1379
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.342168302014338,
      "learning_rate": 1.9965705870813416e-05,
      "loss": 1.6737,
      "step": 1380
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.337915034132228,
      "learning_rate": 1.9965597834781235e-05,
      "loss": 1.7259,
      "step": 1381
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.060582252549427,
      "learning_rate": 1.9965489629138032e-05,
      "loss": 1.3318,
      "step": 1382
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.8255501345718335,
      "learning_rate": 1.9965381253885646e-05,
      "loss": 1.8348,
      "step": 1383
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.884381742112118,
      "learning_rate": 1.996527270902593e-05,
      "loss": 1.3294,
      "step": 1384
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.829997124670875,
      "learning_rate": 1.996516399456073e-05,
      "loss": 1.4629,
      "step": 1385
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.116597884062246,
      "learning_rate": 1.996505511049189e-05,
      "loss": 1.71,
      "step": 1386
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.555178881607057,
      "learning_rate": 1.9964946056821266e-05,
      "loss": 1.4158,
      "step": 1387
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.90115804167898,
      "learning_rate": 1.9964836833550715e-05,
      "loss": 1.4301,
      "step": 1388
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.291682362882667,
      "learning_rate": 1.9964727440682097e-05,
      "loss": 1.4802,
      "step": 1389
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.021321365084674,
      "learning_rate": 1.9964617878217273e-05,
      "loss": 1.554,
      "step": 1390
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.831753044292781,
      "learning_rate": 1.9964508146158107e-05,
      "loss": 1.3284,
      "step": 1391
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.687964436965135,
      "learning_rate": 1.9964398244506468e-05,
      "loss": 1.1057,
      "step": 1392
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.730245361736973,
      "learning_rate": 1.996428817326422e-05,
      "loss": 1.2512,
      "step": 1393
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.591611226009329,
      "learning_rate": 1.996417793243325e-05,
      "loss": 1.2635,
      "step": 1394
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.445458182302627,
      "learning_rate": 1.996406752201542e-05,
      "loss": 1.3182,
      "step": 1395
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.519498646951889,
      "learning_rate": 1.9963956942012615e-05,
      "loss": 1.3275,
      "step": 1396
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.438288964606097,
      "learning_rate": 1.996384619242672e-05,
      "loss": 1.4527,
      "step": 1397
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.506660684958048,
      "learning_rate": 1.9963735273259618e-05,
      "loss": 1.509,
      "step": 1398
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.578134223133227,
      "learning_rate": 1.9963624184513194e-05,
      "loss": 1.2091,
      "step": 1399
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.557938957712244,
      "learning_rate": 1.9963512926189337e-05,
      "loss": 1.4336,
      "step": 1400
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.309800897864221,
      "learning_rate": 1.9963401498289952e-05,
      "loss": 1.6587,
      "step": 1401
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.275750734392879,
      "learning_rate": 1.9963289900816923e-05,
      "loss": 1.5638,
      "step": 1402
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.980202899940684,
      "learning_rate": 1.9963178133772153e-05,
      "loss": 1.5954,
      "step": 1403
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.670835507005246,
      "learning_rate": 1.996306619715755e-05,
      "loss": 1.5459,
      "step": 1404
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.65282837425003,
      "learning_rate": 1.9962954090975007e-05,
      "loss": 1.3876,
      "step": 1405
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.645863043399876,
      "learning_rate": 1.9962841815226447e-05,
      "loss": 1.2799,
      "step": 1406
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.542246425417072,
      "learning_rate": 1.996272936991377e-05,
      "loss": 1.339,
      "step": 1407
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.551154463507817,
      "learning_rate": 1.9962616755038892e-05,
      "loss": 1.6935,
      "step": 1408
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.7856825501857925,
      "learning_rate": 1.9962503970603733e-05,
      "loss": 1.3675,
      "step": 1409
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.344699352998903,
      "learning_rate": 1.9962391016610207e-05,
      "loss": 1.2029,
      "step": 1410
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.49984258139571,
      "learning_rate": 1.9962277893060244e-05,
      "loss": 1.1598,
      "step": 1411
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.621348140223294,
      "learning_rate": 1.9962164599955762e-05,
      "loss": 1.5974,
      "step": 1412
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.378576489746932,
      "learning_rate": 1.9962051137298693e-05,
      "loss": 1.4864,
      "step": 1413
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.38889274891855,
      "learning_rate": 1.9961937505090968e-05,
      "loss": 1.3161,
      "step": 1414
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.317106323828197,
      "learning_rate": 1.9961823703334518e-05,
      "loss": 1.341,
      "step": 1415
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.524252607842972,
      "learning_rate": 1.9961709732031284e-05,
      "loss": 1.4369,
      "step": 1416
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.055702442798704,
      "learning_rate": 1.9961595591183205e-05,
      "loss": 1.4266,
      "step": 1417
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.063598853684944,
      "learning_rate": 1.996148128079222e-05,
      "loss": 1.1427,
      "step": 1418
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.235377435692664,
      "learning_rate": 1.996136680086028e-05,
      "loss": 1.3123,
      "step": 1419
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.859673502151679,
      "learning_rate": 1.9961252151389328e-05,
      "loss": 1.3697,
      "step": 1420
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.995510368566718,
      "learning_rate": 1.996113733238132e-05,
      "loss": 1.7422,
      "step": 1421
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.28704253852011,
      "learning_rate": 1.9961022343838206e-05,
      "loss": 1.7036,
      "step": 1422
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.700495649750436,
      "learning_rate": 1.9960907185761945e-05,
      "loss": 1.2192,
      "step": 1423
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.635391920762192,
      "learning_rate": 1.99607918581545e-05,
      "loss": 1.1413,
      "step": 1424
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.56365265542115,
      "learning_rate": 1.9960676361017827e-05,
      "loss": 1.2294,
      "step": 1425
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.721291352004608,
      "learning_rate": 1.99605606943539e-05,
      "loss": 1.613,
      "step": 1426
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.745725039514284,
      "learning_rate": 1.9960444858164678e-05,
      "loss": 1.342,
      "step": 1427
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.25555042936773,
      "learning_rate": 1.996032885245214e-05,
      "loss": 1.8052,
      "step": 1428
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.873998256905931,
      "learning_rate": 1.996021267721826e-05,
      "loss": 1.4539,
      "step": 1429
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.182174424982235,
      "learning_rate": 1.996009633246501e-05,
      "loss": 1.3021,
      "step": 1430
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.290054311175747,
      "learning_rate": 1.9959979818194373e-05,
      "loss": 1.471,
      "step": 1431
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.760008867452599,
      "learning_rate": 1.9959863134408337e-05,
      "loss": 1.0637,
      "step": 1432
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.90735698477957,
      "learning_rate": 1.995974628110888e-05,
      "loss": 1.2686,
      "step": 1433
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.19552721123613,
      "learning_rate": 1.9959629258297995e-05,
      "loss": 1.1687,
      "step": 1434
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.108050288538617,
      "learning_rate": 1.9959512065977673e-05,
      "loss": 1.7228,
      "step": 1435
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.316459531445684,
      "learning_rate": 1.9959394704149908e-05,
      "loss": 1.2216,
      "step": 1436
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.487244895961331,
      "learning_rate": 1.9959277172816698e-05,
      "loss": 1.1444,
      "step": 1437
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.258707138502286,
      "learning_rate": 1.9959159471980045e-05,
      "loss": 1.4857,
      "step": 1438
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.170354732182427,
      "learning_rate": 1.9959041601641948e-05,
      "loss": 1.1608,
      "step": 1439
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.102090795510799,
      "learning_rate": 1.9958923561804417e-05,
      "loss": 1.1645,
      "step": 1440
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.737535906696977,
      "learning_rate": 1.995880535246946e-05,
      "loss": 1.398,
      "step": 1441
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.852537355125488,
      "learning_rate": 1.9958686973639088e-05,
      "loss": 1.291,
      "step": 1442
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.623558687956683,
      "learning_rate": 1.9958568425315316e-05,
      "loss": 1.3199,
      "step": 1443
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.62514592417538,
      "learning_rate": 1.995844970750016e-05,
      "loss": 1.4647,
      "step": 1444
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.127517128104072,
      "learning_rate": 1.9958330820195644e-05,
      "loss": 1.522,
      "step": 1445
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.460841400722731,
      "learning_rate": 1.995821176340379e-05,
      "loss": 1.4628,
      "step": 1446
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.417868220660701,
      "learning_rate": 1.9958092537126627e-05,
      "loss": 1.3646,
      "step": 1447
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.733327752591579,
      "learning_rate": 1.9957973141366177e-05,
      "loss": 1.51,
      "step": 1448
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.582031803694507,
      "learning_rate": 1.9957853576124478e-05,
      "loss": 1.1981,
      "step": 1449
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.040515310500903,
      "learning_rate": 1.9957733841403564e-05,
      "loss": 1.2464,
      "step": 1450
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.593812483583821,
      "learning_rate": 1.995761393720547e-05,
      "loss": 1.2816,
      "step": 1451
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.642966983467225,
      "learning_rate": 1.9957493863532242e-05,
      "loss": 1.4182,
      "step": 1452
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.6762412941854175,
      "learning_rate": 1.9957373620385917e-05,
      "loss": 1.8309,
      "step": 1453
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.29165738375946,
      "learning_rate": 1.9957253207768547e-05,
      "loss": 1.3729,
      "step": 1454
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.453883533880601,
      "learning_rate": 1.9957132625682177e-05,
      "loss": 1.1561,
      "step": 1455
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.00676070441879,
      "learning_rate": 1.9957011874128866e-05,
      "loss": 1.0691,
      "step": 1456
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.06257821384189,
      "learning_rate": 1.9956890953110662e-05,
      "loss": 1.3211,
      "step": 1457
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.494155595739766,
      "learning_rate": 1.995676986262963e-05,
      "loss": 1.7024,
      "step": 1458
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.036051458018038,
      "learning_rate": 1.9956648602687824e-05,
      "loss": 1.4724,
      "step": 1459
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.421741318285281,
      "learning_rate": 1.995652717328731e-05,
      "loss": 1.0659,
      "step": 1460
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.445206413407089,
      "learning_rate": 1.9956405574430157e-05,
      "loss": 1.1788,
      "step": 1461
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.42401125335881,
      "learning_rate": 1.995628380611843e-05,
      "loss": 1.1567,
      "step": 1462
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.909183887961802,
      "learning_rate": 1.9956161868354207e-05,
      "loss": 1.4171,
      "step": 1463
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.142008319180877,
      "learning_rate": 1.9956039761139558e-05,
      "loss": 1.457,
      "step": 1464
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.939658985538146,
      "learning_rate": 1.9955917484476567e-05,
      "loss": 1.4544,
      "step": 1465
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.971951006394367,
      "learning_rate": 1.995579503836731e-05,
      "loss": 1.1686,
      "step": 1466
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.633004642298342,
      "learning_rate": 1.9955672422813873e-05,
      "loss": 1.6463,
      "step": 1467
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.265404154294693,
      "learning_rate": 1.9955549637818343e-05,
      "loss": 1.3659,
      "step": 1468
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.122616632776786,
      "learning_rate": 1.995542668338281e-05,
      "loss": 1.528,
      "step": 1469
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.270860678395396,
      "learning_rate": 1.9955303559509365e-05,
      "loss": 1.2669,
      "step": 1470
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.907035937604945,
      "learning_rate": 1.9955180266200107e-05,
      "loss": 1.79,
      "step": 1471
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.6045138801594385,
      "learning_rate": 1.995505680345713e-05,
      "loss": 1.5054,
      "step": 1472
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.840056759815627,
      "learning_rate": 1.9954933171282542e-05,
      "loss": 1.3857,
      "step": 1473
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.215959167176934,
      "learning_rate": 1.9954809369678435e-05,
      "loss": 1.5313,
      "step": 1474
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.712740973719321,
      "learning_rate": 1.995468539864693e-05,
      "loss": 1.6483,
      "step": 1475
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.632704324433016,
      "learning_rate": 1.995456125819013e-05,
      "loss": 1.3361,
      "step": 1476
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.991152607595547,
      "learning_rate": 1.9954436948310144e-05,
      "loss": 1.3354,
      "step": 1477
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.956666104002904,
      "learning_rate": 1.9954312469009095e-05,
      "loss": 1.3973,
      "step": 1478
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.663911308799256,
      "learning_rate": 1.9954187820289102e-05,
      "loss": 1.3989,
      "step": 1479
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.08768547949442,
      "learning_rate": 1.9954063002152278e-05,
      "loss": 1.2128,
      "step": 1480
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.084507535424869,
      "learning_rate": 1.9953938014600757e-05,
      "loss": 1.3111,
      "step": 1481
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.248581230638397,
      "learning_rate": 1.995381285763666e-05,
      "loss": 1.2533,
      "step": 1482
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.741567905340336,
      "learning_rate": 1.9953687531262117e-05,
      "loss": 1.5419,
      "step": 1483
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.001529962007812,
      "learning_rate": 1.9953562035479267e-05,
      "loss": 1.2979,
      "step": 1484
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.511616847772875,
      "learning_rate": 1.995343637029024e-05,
      "loss": 1.118,
      "step": 1485
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.581325701859086,
      "learning_rate": 1.9953310535697173e-05,
      "loss": 1.5552,
      "step": 1486
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.724582240646722,
      "learning_rate": 1.9953184531702215e-05,
      "loss": 1.5721,
      "step": 1487
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.372612877513773,
      "learning_rate": 1.9953058358307512e-05,
      "loss": 1.5329,
      "step": 1488
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.311344892737381,
      "learning_rate": 1.9952932015515198e-05,
      "loss": 1.8343,
      "step": 1489
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.0780310089049,
      "learning_rate": 1.995280550332744e-05,
      "loss": 1.5196,
      "step": 1490
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.184315064769623,
      "learning_rate": 1.995267882174637e-05,
      "loss": 1.5368,
      "step": 1491
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.09806800037185,
      "learning_rate": 1.9952551970774166e-05,
      "loss": 1.2071,
      "step": 1492
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.924723751243145,
      "learning_rate": 1.9952424950412978e-05,
      "loss": 1.3975,
      "step": 1493
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.413953116505157,
      "learning_rate": 1.9952297760664966e-05,
      "loss": 1.6823,
      "step": 1494
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.802940647087575,
      "learning_rate": 1.9952170401532296e-05,
      "loss": 1.3648,
      "step": 1495
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7516086853913877,
      "learning_rate": 1.9952042873017136e-05,
      "loss": 0.9918,
      "step": 1496
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.035630361418338,
      "learning_rate": 1.995191517512166e-05,
      "loss": 1.607,
      "step": 1497
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.201077764268946,
      "learning_rate": 1.995178730784803e-05,
      "loss": 1.1725,
      "step": 1498
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.985591555531078,
      "learning_rate": 1.9951659271198437e-05,
      "loss": 2.1272,
      "step": 1499
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.513750348852558,
      "learning_rate": 1.9951531065175054e-05,
      "loss": 1.2252,
      "step": 1500
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.683200390746712,
      "learning_rate": 1.995140268978006e-05,
      "loss": 1.6882,
      "step": 1501
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.848679770613399,
      "learning_rate": 1.995127414501564e-05,
      "loss": 1.556,
      "step": 1502
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.3941788132556345,
      "learning_rate": 1.9951145430883986e-05,
      "loss": 1.3069,
      "step": 1503
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.194172717057551,
      "learning_rate": 1.9951016547387286e-05,
      "loss": 1.3932,
      "step": 1504
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.597246146880204,
      "learning_rate": 1.9950887494527735e-05,
      "loss": 1.4012,
      "step": 1505
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.054413820458573,
      "learning_rate": 1.995075827230753e-05,
      "loss": 1.4402,
      "step": 1506
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.086833181709755,
      "learning_rate": 1.9950628880728868e-05,
      "loss": 1.12,
      "step": 1507
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.806341080716645,
      "learning_rate": 1.995049931979395e-05,
      "loss": 1.2329,
      "step": 1508
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.392345445619857,
      "learning_rate": 1.9950369589504987e-05,
      "loss": 1.4104,
      "step": 1509
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.098024377502341,
      "learning_rate": 1.9950239689864182e-05,
      "loss": 1.3549,
      "step": 1510
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.226256123252309,
      "learning_rate": 1.9950109620873746e-05,
      "loss": 1.0022,
      "step": 1511
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.6360826043349155,
      "learning_rate": 1.9949979382535894e-05,
      "loss": 1.4581,
      "step": 1512
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.544144963732201,
      "learning_rate": 1.9949848974852843e-05,
      "loss": 1.6059,
      "step": 1513
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.165575590524718,
      "learning_rate": 1.994971839782681e-05,
      "loss": 1.4665,
      "step": 1514
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.272202055463505,
      "learning_rate": 1.994958765146002e-05,
      "loss": 1.2935,
      "step": 1515
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.56313624952119,
      "learning_rate": 1.9949456735754697e-05,
      "loss": 1.1577,
      "step": 1516
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.539039786243627,
      "learning_rate": 1.9949325650713072e-05,
      "loss": 1.5931,
      "step": 1517
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.685723790866573,
      "learning_rate": 1.9949194396337374e-05,
      "loss": 1.1646,
      "step": 1518
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.223923164237922,
      "learning_rate": 1.9949062972629835e-05,
      "loss": 0.9914,
      "step": 1519
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.023196817327144,
      "learning_rate": 1.9948931379592692e-05,
      "loss": 1.2849,
      "step": 1520
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.479767445041039,
      "learning_rate": 1.994879961722819e-05,
      "loss": 1.3771,
      "step": 1521
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.865406509365259,
      "learning_rate": 1.9948667685538564e-05,
      "loss": 1.2343,
      "step": 1522
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.870246739719876,
      "learning_rate": 1.9948535584526066e-05,
      "loss": 1.1168,
      "step": 1523
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.141149797676787,
      "learning_rate": 1.9948403314192938e-05,
      "loss": 1.295,
      "step": 1524
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.335691686724497,
      "learning_rate": 1.9948270874541438e-05,
      "loss": 1.3613,
      "step": 1525
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.545276913963258,
      "learning_rate": 1.9948138265573815e-05,
      "loss": 1.4848,
      "step": 1526
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.638456702770252,
      "learning_rate": 1.994800548729233e-05,
      "loss": 1.0175,
      "step": 1527
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.351059645817972,
      "learning_rate": 1.9947872539699236e-05,
      "loss": 1.0924,
      "step": 1528
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.658952670205018,
      "learning_rate": 1.9947739422796803e-05,
      "loss": 1.2638,
      "step": 1529
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.824508558053438,
      "learning_rate": 1.9947606136587293e-05,
      "loss": 1.6483,
      "step": 1530
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.760923289785385,
      "learning_rate": 1.9947472681072974e-05,
      "loss": 1.4872,
      "step": 1531
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.143965236547297,
      "learning_rate": 1.994733905625612e-05,
      "loss": 1.4459,
      "step": 1532
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.4480628440869205,
      "learning_rate": 1.9947205262139006e-05,
      "loss": 1.5218,
      "step": 1533
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.631708833704664,
      "learning_rate": 1.9947071298723904e-05,
      "loss": 1.5779,
      "step": 1534
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.485497750679615,
      "learning_rate": 1.99469371660131e-05,
      "loss": 1.6444,
      "step": 1535
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.722309497031518,
      "learning_rate": 1.994680286400887e-05,
      "loss": 1.4287,
      "step": 1536
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.992546940333161,
      "learning_rate": 1.9946668392713503e-05,
      "loss": 1.6253,
      "step": 1537
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.342102541772421,
      "learning_rate": 1.9946533752129292e-05,
      "loss": 1.2453,
      "step": 1538
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.654906082102196,
      "learning_rate": 1.9946398942258523e-05,
      "loss": 1.3879,
      "step": 1539
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.208205537170746,
      "learning_rate": 1.9946263963103493e-05,
      "loss": 1.2924,
      "step": 1540
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.28143628400916,
      "learning_rate": 1.9946128814666496e-05,
      "loss": 1.3187,
      "step": 1541
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.0501705117234135,
      "learning_rate": 1.994599349694984e-05,
      "loss": 1.4295,
      "step": 1542
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.079207926087998,
      "learning_rate": 1.994585800995582e-05,
      "loss": 1.0657,
      "step": 1543
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.529469507098824,
      "learning_rate": 1.9945722353686743e-05,
      "loss": 1.4487,
      "step": 1544
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.35757365334837,
      "learning_rate": 1.9945586528144923e-05,
      "loss": 1.4333,
      "step": 1545
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.571609072814372,
      "learning_rate": 1.9945450533332663e-05,
      "loss": 1.9596,
      "step": 1546
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.032395663945293,
      "learning_rate": 1.994531436925229e-05,
      "loss": 1.717,
      "step": 1547
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.7972952803423725,
      "learning_rate": 1.994517803590611e-05,
      "loss": 1.2925,
      "step": 1548
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.08034214608452,
      "learning_rate": 1.9945041533296448e-05,
      "loss": 1.3213,
      "step": 1549
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.539644529234815,
      "learning_rate": 1.9944904861425626e-05,
      "loss": 1.4502,
      "step": 1550
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.31531444417317,
      "learning_rate": 1.994476802029597e-05,
      "loss": 1.7819,
      "step": 1551
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.921529556124726,
      "learning_rate": 1.9944631009909815e-05,
      "loss": 1.7807,
      "step": 1552
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.5716019322330785,
      "learning_rate": 1.9944493830269487e-05,
      "loss": 1.6417,
      "step": 1553
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.519358117715136,
      "learning_rate": 1.994435648137732e-05,
      "loss": 1.9101,
      "step": 1554
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.532529571551585,
      "learning_rate": 1.9944218963235653e-05,
      "loss": 1.4121,
      "step": 1555
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.907671414538133,
      "learning_rate": 1.9944081275846826e-05,
      "loss": 1.112,
      "step": 1556
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.32721150332738,
      "learning_rate": 1.9943943419213182e-05,
      "loss": 1.1694,
      "step": 1557
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.11909419991208,
      "learning_rate": 1.9943805393337073e-05,
      "loss": 1.45,
      "step": 1558
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.286928875595552,
      "learning_rate": 1.994366719822084e-05,
      "loss": 1.2329,
      "step": 1559
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7927439174877033,
      "learning_rate": 1.994352883386684e-05,
      "loss": 1.0477,
      "step": 1560
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.958128889538223,
      "learning_rate": 1.9943390300277427e-05,
      "loss": 1.603,
      "step": 1561
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.170189289590508,
      "learning_rate": 1.994325159745496e-05,
      "loss": 1.4423,
      "step": 1562
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.50071470960183,
      "learning_rate": 1.9943112725401793e-05,
      "loss": 1.1979,
      "step": 1563
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.67036573703531,
      "learning_rate": 1.9942973684120297e-05,
      "loss": 1.3554,
      "step": 1564
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.899842506755757,
      "learning_rate": 1.9942834473612836e-05,
      "loss": 1.2885,
      "step": 1565
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.670129116540577,
      "learning_rate": 1.9942695093881782e-05,
      "loss": 1.4537,
      "step": 1566
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.7312256194945554,
      "learning_rate": 1.9942555544929497e-05,
      "loss": 1.263,
      "step": 1567
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.185744260633248,
      "learning_rate": 1.994241582675837e-05,
      "loss": 1.2302,
      "step": 1568
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.098596941884958,
      "learning_rate": 1.994227593937077e-05,
      "loss": 1.3897,
      "step": 1569
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.522767975133329,
      "learning_rate": 1.9942135882769083e-05,
      "loss": 1.5482,
      "step": 1570
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.714566595500402,
      "learning_rate": 1.9941995656955687e-05,
      "loss": 1.1318,
      "step": 1571
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.2981293350866645,
      "learning_rate": 1.9941855261932975e-05,
      "loss": 1.2576,
      "step": 1572
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.2250672496196415,
      "learning_rate": 1.9941714697703333e-05,
      "loss": 1.0657,
      "step": 1573
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.094744842113075,
      "learning_rate": 1.994157396426915e-05,
      "loss": 1.2815,
      "step": 1574
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.525858474415809,
      "learning_rate": 1.9941433061632827e-05,
      "loss": 1.3282,
      "step": 1575
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.461860155676494,
      "learning_rate": 1.9941291989796756e-05,
      "loss": 1.2583,
      "step": 1576
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.7787951119602505,
      "learning_rate": 1.9941150748763344e-05,
      "loss": 1.022,
      "step": 1577
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.401451954512998,
      "learning_rate": 1.9941009338534995e-05,
      "loss": 1.8156,
      "step": 1578
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.58227379164404,
      "learning_rate": 1.9940867759114113e-05,
      "loss": 1.3461,
      "step": 1579
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.690634179228754,
      "learning_rate": 1.9940726010503105e-05,
      "loss": 1.4187,
      "step": 1580
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.77280965565021,
      "learning_rate": 1.994058409270439e-05,
      "loss": 1.3978,
      "step": 1581
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.083777859803659,
      "learning_rate": 1.994044200572038e-05,
      "loss": 1.2461,
      "step": 1582
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.856843418443755,
      "learning_rate": 1.9940299749553492e-05,
      "loss": 1.458,
      "step": 1583
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.31283912339637,
      "learning_rate": 1.994015732420615e-05,
      "loss": 1.4488,
      "step": 1584
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.057728244559561,
      "learning_rate": 1.9940014729680775e-05,
      "loss": 1.2361,
      "step": 1585
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.595304766012098,
      "learning_rate": 1.9939871965979794e-05,
      "loss": 1.5832,
      "step": 1586
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.60111861956935,
      "learning_rate": 1.993972903310564e-05,
      "loss": 1.1715,
      "step": 1587
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.989689784232833,
      "learning_rate": 1.9939585931060744e-05,
      "loss": 1.1792,
      "step": 1588
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.17453886949581,
      "learning_rate": 1.993944265984754e-05,
      "loss": 1.3627,
      "step": 1589
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.409664314217006,
      "learning_rate": 1.9939299219468467e-05,
      "loss": 1.5373,
      "step": 1590
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.075693106741561,
      "learning_rate": 1.993915560992597e-05,
      "loss": 1.1777,
      "step": 1591
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.192403901172698,
      "learning_rate": 1.993901183122249e-05,
      "loss": 1.5035,
      "step": 1592
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.640393897120675,
      "learning_rate": 1.9938867883360472e-05,
      "loss": 1.3254,
      "step": 1593
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.809090601586043,
      "learning_rate": 1.993872376634237e-05,
      "loss": 1.3879,
      "step": 1594
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.355182527982758,
      "learning_rate": 1.9938579480170633e-05,
      "loss": 1.1732,
      "step": 1595
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.322432509837842,
      "learning_rate": 1.9938435024847723e-05,
      "loss": 1.075,
      "step": 1596
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.981143220828466,
      "learning_rate": 1.9938290400376094e-05,
      "loss": 1.3783,
      "step": 1597
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.619628262979774,
      "learning_rate": 1.9938145606758202e-05,
      "loss": 1.3298,
      "step": 1598
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.623960361259372,
      "learning_rate": 1.9938000643996522e-05,
      "loss": 1.5042,
      "step": 1599
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.5739225559456145,
      "learning_rate": 1.9937855512093516e-05,
      "loss": 1.3372,
      "step": 1600
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.266145277377041,
      "learning_rate": 1.9937710211051653e-05,
      "loss": 1.509,
      "step": 1601
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.469312510968996,
      "learning_rate": 1.993756474087341e-05,
      "loss": 1.4383,
      "step": 1602
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.978471542836006,
      "learning_rate": 1.9937419101561258e-05,
      "loss": 1.2334,
      "step": 1603
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.392335320682871,
      "learning_rate": 1.9937273293117675e-05,
      "loss": 1.4389,
      "step": 1604
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.158813927089674,
      "learning_rate": 1.9937127315545153e-05,
      "loss": 1.3663,
      "step": 1605
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.378488620897281,
      "learning_rate": 1.9936981168846166e-05,
      "loss": 1.5565,
      "step": 1606
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.62754883387644,
      "learning_rate": 1.99368348530232e-05,
      "loss": 1.4329,
      "step": 1607
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.943410745982573,
      "learning_rate": 1.9936688368078756e-05,
      "loss": 1.3482,
      "step": 1608
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.451168088621057,
      "learning_rate": 1.9936541714015317e-05,
      "loss": 1.5532,
      "step": 1609
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.559605427719264,
      "learning_rate": 1.9936394890835387e-05,
      "loss": 1.4402,
      "step": 1610
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.142675489715711,
      "learning_rate": 1.9936247898541457e-05,
      "loss": 1.3562,
      "step": 1611
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.390666476163092,
      "learning_rate": 1.9936100737136034e-05,
      "loss": 1.048,
      "step": 1612
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.527392428232301,
      "learning_rate": 1.993595340662162e-05,
      "loss": 1.2397,
      "step": 1613
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.8514051690548365,
      "learning_rate": 1.993580590700073e-05,
      "loss": 1.1742,
      "step": 1614
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.478998822309698,
      "learning_rate": 1.9935658238275858e-05,
      "loss": 1.1014,
      "step": 1615
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.608661681964139,
      "learning_rate": 1.9935510400449534e-05,
      "loss": 1.092,
      "step": 1616
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.089407265384593,
      "learning_rate": 1.9935362393524262e-05,
      "loss": 1.7726,
      "step": 1617
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.774298791242347,
      "learning_rate": 1.993521421750257e-05,
      "loss": 1.571,
      "step": 1618
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.781802512135609,
      "learning_rate": 1.9935065872386977e-05,
      "loss": 1.6905,
      "step": 1619
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.0197893745602835,
      "learning_rate": 1.993491735818001e-05,
      "loss": 1.3041,
      "step": 1620
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.544392804378859,
      "learning_rate": 1.9934768674884187e-05,
      "loss": 1.5142,
      "step": 1621
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.805665932824067,
      "learning_rate": 1.993461982250205e-05,
      "loss": 1.322,
      "step": 1622
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.805394971521824,
      "learning_rate": 1.993447080103613e-05,
      "loss": 1.3419,
      "step": 1623
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.865843585738184,
      "learning_rate": 1.9934321610488955e-05,
      "loss": 1.1941,
      "step": 1624
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.677701994350402,
      "learning_rate": 1.9934172250863074e-05,
      "loss": 1.6682,
      "step": 1625
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.656563521687858,
      "learning_rate": 1.9934022722161025e-05,
      "loss": 1.1174,
      "step": 1626
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.074739077062604,
      "learning_rate": 1.9933873024385352e-05,
      "loss": 1.2706,
      "step": 1627
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.116254423287817,
      "learning_rate": 1.9933723157538606e-05,
      "loss": 1.5758,
      "step": 1628
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.43568384602253,
      "learning_rate": 1.993357312162333e-05,
      "loss": 1.2362,
      "step": 1629
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.290182240848821,
      "learning_rate": 1.9933422916642093e-05,
      "loss": 1.3019,
      "step": 1630
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.120901478111954,
      "learning_rate": 1.9933272542597432e-05,
      "loss": 1.5725,
      "step": 1631
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.111223430836484,
      "learning_rate": 1.9933121999491923e-05,
      "loss": 1.4675,
      "step": 1632
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.071313114492433,
      "learning_rate": 1.993297128732812e-05,
      "loss": 1.4817,
      "step": 1633
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.710873816747146,
      "learning_rate": 1.9932820406108586e-05,
      "loss": 1.4678,
      "step": 1634
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.327068150483543,
      "learning_rate": 1.9932669355835893e-05,
      "loss": 1.1287,
      "step": 1635
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.856110823537345,
      "learning_rate": 1.9932518136512612e-05,
      "loss": 1.2326,
      "step": 1636
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.5763091821901805,
      "learning_rate": 1.9932366748141318e-05,
      "loss": 1.3405,
      "step": 1637
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.49273271358971,
      "learning_rate": 1.9932215190724582e-05,
      "loss": 1.15,
      "step": 1638
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.398065629815027,
      "learning_rate": 1.993206346426499e-05,
      "loss": 1.0625,
      "step": 1639
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.750787524045484,
      "learning_rate": 1.993191156876512e-05,
      "loss": 1.7256,
      "step": 1640
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.730581934470894,
      "learning_rate": 1.9931759504227555e-05,
      "loss": 1.5309,
      "step": 1641
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.2088961963128195,
      "learning_rate": 1.993160727065489e-05,
      "loss": 1.2774,
      "step": 1642
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.651327937181227,
      "learning_rate": 1.9931454868049713e-05,
      "loss": 1.4965,
      "step": 1643
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.92699279504504,
      "learning_rate": 1.9931302296414615e-05,
      "loss": 1.3109,
      "step": 1644
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.309450043311207,
      "learning_rate": 1.9931149555752195e-05,
      "loss": 1.5808,
      "step": 1645
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.749626345156699,
      "learning_rate": 1.9930996646065055e-05,
      "loss": 1.6116,
      "step": 1646
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.112384855649576,
      "learning_rate": 1.9930843567355794e-05,
      "loss": 1.4236,
      "step": 1647
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.568626866352402,
      "learning_rate": 1.9930690319627017e-05,
      "loss": 1.3355,
      "step": 1648
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.644718595756402,
      "learning_rate": 1.9930536902881333e-05,
      "loss": 1.4223,
      "step": 1649
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.894857796854881,
      "learning_rate": 1.9930383317121355e-05,
      "loss": 1.3005,
      "step": 1650
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.006740734699825,
      "learning_rate": 1.9930229562349697e-05,
      "loss": 1.2257,
      "step": 1651
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.4882022628806,
      "learning_rate": 1.9930075638568973e-05,
      "loss": 1.7219,
      "step": 1652
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.938144432325534,
      "learning_rate": 1.9929921545781805e-05,
      "loss": 1.6193,
      "step": 1653
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.37878731753769,
      "learning_rate": 1.9929767283990814e-05,
      "loss": 1.468,
      "step": 1654
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.712880185992707,
      "learning_rate": 1.9929612853198628e-05,
      "loss": 1.4272,
      "step": 1655
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.082371310995654,
      "learning_rate": 1.992945825340787e-05,
      "loss": 1.185,
      "step": 1656
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.985156578753303,
      "learning_rate": 1.9929303484621177e-05,
      "loss": 1.3921,
      "step": 1657
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.033665445543282,
      "learning_rate": 1.992914854684118e-05,
      "loss": 1.5983,
      "step": 1658
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.326442660226737,
      "learning_rate": 1.9928993440070522e-05,
      "loss": 1.4395,
      "step": 1659
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.870200774828051,
      "learning_rate": 1.9928838164311834e-05,
      "loss": 1.0925,
      "step": 1660
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.072785921063068,
      "learning_rate": 1.9928682719567762e-05,
      "loss": 1.2887,
      "step": 1661
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.888497958751173,
      "learning_rate": 1.9928527105840952e-05,
      "loss": 1.3087,
      "step": 1662
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.534549608752247,
      "learning_rate": 1.9928371323134055e-05,
      "loss": 1.2264,
      "step": 1663
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.7763821422269785,
      "learning_rate": 1.992821537144972e-05,
      "loss": 1.1615,
      "step": 1664
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.073106206221608,
      "learning_rate": 1.99280592507906e-05,
      "loss": 1.293,
      "step": 1665
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.074976451536681,
      "learning_rate": 1.9927902961159353e-05,
      "loss": 1.1778,
      "step": 1666
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.909053159561319,
      "learning_rate": 1.992774650255864e-05,
      "loss": 1.4633,
      "step": 1667
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.385949656089326,
      "learning_rate": 1.9927589874991125e-05,
      "loss": 1.2306,
      "step": 1668
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.715482874351792,
      "learning_rate": 1.992743307845947e-05,
      "loss": 1.1861,
      "step": 1669
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.778867765226718,
      "learning_rate": 1.9927276112966347e-05,
      "loss": 1.3787,
      "step": 1670
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.648479773841119,
      "learning_rate": 1.9927118978514426e-05,
      "loss": 1.6182,
      "step": 1671
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.15540401833208,
      "learning_rate": 1.992696167510638e-05,
      "loss": 1.4082,
      "step": 1672
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.254012974044487,
      "learning_rate": 1.992680420274489e-05,
      "loss": 1.3287,
      "step": 1673
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.518865110024242,
      "learning_rate": 1.9926646561432634e-05,
      "loss": 1.653,
      "step": 1674
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.923994973352675,
      "learning_rate": 1.992648875117229e-05,
      "loss": 1.662,
      "step": 1675
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.2048350053733365,
      "learning_rate": 1.9926330771966554e-05,
      "loss": 1.2386,
      "step": 1676
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.699161398714065,
      "learning_rate": 1.9926172623818108e-05,
      "loss": 1.2519,
      "step": 1677
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.34755248933042,
      "learning_rate": 1.9926014306729645e-05,
      "loss": 1.316,
      "step": 1678
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.152965599653538,
      "learning_rate": 1.9925855820703862e-05,
      "loss": 1.3969,
      "step": 1679
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.69933018387624,
      "learning_rate": 1.9925697165743453e-05,
      "loss": 1.2576,
      "step": 1680
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.031168273676834,
      "learning_rate": 1.992553834185112e-05,
      "loss": 1.1726,
      "step": 1681
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.496433759523101,
      "learning_rate": 1.992537934902956e-05,
      "loss": 1.1888,
      "step": 1682
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.729325085418545,
      "learning_rate": 1.992522018728149e-05,
      "loss": 1.1776,
      "step": 1683
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.8770115341714195,
      "learning_rate": 1.9925060856609614e-05,
      "loss": 1.6919,
      "step": 1684
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.0878552238194095,
      "learning_rate": 1.992490135701664e-05,
      "loss": 1.4694,
      "step": 1685
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.478208461704142,
      "learning_rate": 1.9924741688505284e-05,
      "loss": 1.2665,
      "step": 1686
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.24583709703786,
      "learning_rate": 1.9924581851078272e-05,
      "loss": 1.5426,
      "step": 1687
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.190483850669683,
      "learning_rate": 1.992442184473831e-05,
      "loss": 1.4105,
      "step": 1688
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.248388133122466,
      "learning_rate": 1.992426166948813e-05,
      "loss": 1.2239,
      "step": 1689
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.0589390011053395,
      "learning_rate": 1.992410132533046e-05,
      "loss": 1.6185,
      "step": 1690
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.145311335723086,
      "learning_rate": 1.992394081226803e-05,
      "loss": 1.5274,
      "step": 1691
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.45199735989477,
      "learning_rate": 1.992378013030356e-05,
      "loss": 1.2775,
      "step": 1692
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.10188760423342,
      "learning_rate": 1.99236192794398e-05,
      "loss": 1.0359,
      "step": 1693
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.680400597548671,
      "learning_rate": 1.9923458259679473e-05,
      "loss": 1.3229,
      "step": 1694
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.346560758139622,
      "learning_rate": 1.992329707102533e-05,
      "loss": 1.3244,
      "step": 1695
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.556713394630111,
      "learning_rate": 1.9923135713480112e-05,
      "loss": 1.1206,
      "step": 1696
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.9654618664773045,
      "learning_rate": 1.9922974187046563e-05,
      "loss": 1.3522,
      "step": 1697
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.133248190517277,
      "learning_rate": 1.992281249172743e-05,
      "loss": 1.2864,
      "step": 1698
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.377822359872365,
      "learning_rate": 1.9922650627525475e-05,
      "loss": 1.2607,
      "step": 1699
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.768660810924517,
      "learning_rate": 1.9922488594443445e-05,
      "loss": 1.5015,
      "step": 1700
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.853470733131885,
      "learning_rate": 1.99223263924841e-05,
      "loss": 1.21,
      "step": 1701
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.520803231728888,
      "learning_rate": 1.9922164021650195e-05,
      "loss": 1.3648,
      "step": 1702
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.822571381701583,
      "learning_rate": 1.99220014819445e-05,
      "loss": 1.3823,
      "step": 1703
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.917994216538084,
      "learning_rate": 1.9921838773369782e-05,
      "loss": 1.43,
      "step": 1704
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.8088448831396216,
      "learning_rate": 1.9921675895928806e-05,
      "loss": 1.4698,
      "step": 1705
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.946736114359994,
      "learning_rate": 1.9921512849624345e-05,
      "loss": 1.5381,
      "step": 1706
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.79832887747123,
      "learning_rate": 1.9921349634459176e-05,
      "loss": 1.4969,
      "step": 1707
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.304559368916484,
      "learning_rate": 1.9921186250436075e-05,
      "loss": 1.1857,
      "step": 1708
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.3264970553984,
      "learning_rate": 1.9921022697557824e-05,
      "loss": 1.3402,
      "step": 1709
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.149684063867502,
      "learning_rate": 1.9920858975827207e-05,
      "loss": 1.4257,
      "step": 1710
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.451057532094731,
      "learning_rate": 1.9920695085247012e-05,
      "loss": 1.3762,
      "step": 1711
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.3843049881880995,
      "learning_rate": 1.992053102582002e-05,
      "loss": 1.5508,
      "step": 1712
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.117902170348703,
      "learning_rate": 1.9920366797549032e-05,
      "loss": 1.2479,
      "step": 1713
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.319116056589735,
      "learning_rate": 1.9920202400436843e-05,
      "loss": 1.5845,
      "step": 1714
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.00822832168836,
      "learning_rate": 1.9920037834486245e-05,
      "loss": 1.5245,
      "step": 1715
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.2934451740367665,
      "learning_rate": 1.9919873099700045e-05,
      "loss": 1.4578,
      "step": 1716
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.159304496800559,
      "learning_rate": 1.991970819608104e-05,
      "loss": 1.4281,
      "step": 1717
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.952747536882591,
      "learning_rate": 1.9919543123632044e-05,
      "loss": 1.1817,
      "step": 1718
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.211977119387033,
      "learning_rate": 1.9919377882355863e-05,
      "loss": 1.44,
      "step": 1719
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.331147977211939,
      "learning_rate": 1.9919212472255305e-05,
      "loss": 1.2883,
      "step": 1720
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.3346054916498185,
      "learning_rate": 1.9919046893333194e-05,
      "loss": 1.3507,
      "step": 1721
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.731406890546832,
      "learning_rate": 1.991888114559234e-05,
      "loss": 1.6637,
      "step": 1722
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.134700365445143,
      "learning_rate": 1.991871522903557e-05,
      "loss": 1.7547,
      "step": 1723
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.136288929620964,
      "learning_rate": 1.9918549143665705e-05,
      "loss": 1.1381,
      "step": 1724
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.603047677756543,
      "learning_rate": 1.9918382889485578e-05,
      "loss": 1.6167,
      "step": 1725
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.557569322985294,
      "learning_rate": 1.9918216466498004e-05,
      "loss": 2.2483,
      "step": 1726
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.109442732408645,
      "learning_rate": 1.991804987470583e-05,
      "loss": 1.4275,
      "step": 1727
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.3797191129216815,
      "learning_rate": 1.9917883114111884e-05,
      "loss": 1.4498,
      "step": 1728
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.506771305809261,
      "learning_rate": 1.9917716184719e-05,
      "loss": 1.6498,
      "step": 1729
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.310880322621589,
      "learning_rate": 1.9917549086530034e-05,
      "loss": 1.2824,
      "step": 1730
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.811035140124221,
      "learning_rate": 1.9917381819547814e-05,
      "loss": 1.0696,
      "step": 1731
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.608228013349219,
      "learning_rate": 1.99172143837752e-05,
      "loss": 1.2104,
      "step": 1732
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.027231981468973,
      "learning_rate": 1.991704677921503e-05,
      "loss": 1.4918,
      "step": 1733
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.0489565763917765,
      "learning_rate": 1.9916879005870164e-05,
      "loss": 1.3998,
      "step": 1734
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.42963262852247,
      "learning_rate": 1.9916711063743456e-05,
      "loss": 1.4929,
      "step": 1735
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.017362157096668,
      "learning_rate": 1.9916542952837764e-05,
      "loss": 1.4392,
      "step": 1736
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.045204604129488,
      "learning_rate": 1.9916374673155945e-05,
      "loss": 1.1419,
      "step": 1737
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.145834818494214,
      "learning_rate": 1.991620622470087e-05,
      "loss": 1.1655,
      "step": 1738
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.081079699341444,
      "learning_rate": 1.9916037607475402e-05,
      "loss": 1.2701,
      "step": 1739
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.226159187826429,
      "learning_rate": 1.991586882148241e-05,
      "loss": 1.2725,
      "step": 1740
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.349359066104512,
      "learning_rate": 1.9915699866724773e-05,
      "loss": 1.4241,
      "step": 1741
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.926721703308994,
      "learning_rate": 1.9915530743205358e-05,
      "loss": 1.59,
      "step": 1742
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.999344282514739,
      "learning_rate": 1.991536145092705e-05,
      "loss": 1.8143,
      "step": 1743
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.818289030500545,
      "learning_rate": 1.9915191989892725e-05,
      "loss": 1.2075,
      "step": 1744
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.601293365287694,
      "learning_rate": 1.991502236010527e-05,
      "loss": 1.4127,
      "step": 1745
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.298343420921267,
      "learning_rate": 1.9914852561567574e-05,
      "loss": 1.0145,
      "step": 1746
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.474593316675642,
      "learning_rate": 1.9914682594282526e-05,
      "loss": 1.388,
      "step": 1747
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.195612080501038,
      "learning_rate": 1.9914512458253014e-05,
      "loss": 1.2526,
      "step": 1748
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.324070732912341,
      "learning_rate": 1.9914342153481935e-05,
      "loss": 1.222,
      "step": 1749
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.895319279343103,
      "learning_rate": 1.9914171679972196e-05,
      "loss": 1.424,
      "step": 1750
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.353797203841676,
      "learning_rate": 1.991400103772669e-05,
      "loss": 1.3574,
      "step": 1751
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.042433336959572,
      "learning_rate": 1.9913830226748322e-05,
      "loss": 1.3862,
      "step": 1752
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.998261005360554,
      "learning_rate": 1.9913659247039998e-05,
      "loss": 1.4126,
      "step": 1753
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.898870494013171,
      "learning_rate": 1.9913488098604636e-05,
      "loss": 1.3144,
      "step": 1754
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.833619034022559,
      "learning_rate": 1.991331678144514e-05,
      "loss": 1.4492,
      "step": 1755
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.803320290471051,
      "learning_rate": 1.991314529556443e-05,
      "loss": 1.3331,
      "step": 1756
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.191269472544466,
      "learning_rate": 1.9912973640965423e-05,
      "loss": 1.3654,
      "step": 1757
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.89152675124716,
      "learning_rate": 1.991280181765104e-05,
      "loss": 1.3809,
      "step": 1758
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.386224116385923,
      "learning_rate": 1.9912629825624212e-05,
      "loss": 1.2998,
      "step": 1759
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.277462624411385,
      "learning_rate": 1.9912457664887855e-05,
      "loss": 1.0777,
      "step": 1760
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.639883888556958,
      "learning_rate": 1.991228533544491e-05,
      "loss": 1.3858,
      "step": 1761
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.71621459688264,
      "learning_rate": 1.9912112837298304e-05,
      "loss": 1.1989,
      "step": 1762
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.556813046672799,
      "learning_rate": 1.9911940170450972e-05,
      "loss": 1.5478,
      "step": 1763
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.382872366027941,
      "learning_rate": 1.9911767334905858e-05,
      "loss": 1.5322,
      "step": 1764
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.784716661821816,
      "learning_rate": 1.9911594330665895e-05,
      "loss": 1.7565,
      "step": 1765
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.628468038670719,
      "learning_rate": 1.9911421157734036e-05,
      "loss": 1.5098,
      "step": 1766
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.177711320719025,
      "learning_rate": 1.9911247816113224e-05,
      "loss": 1.1364,
      "step": 1767
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.561688252484943,
      "learning_rate": 1.991107430580641e-05,
      "loss": 1.2444,
      "step": 1768
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.4568799363013,
      "learning_rate": 1.991090062681655e-05,
      "loss": 1.4806,
      "step": 1769
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.394164282685899,
      "learning_rate": 1.9910726779146597e-05,
      "loss": 1.3055,
      "step": 1770
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.07188287178954,
      "learning_rate": 1.9910552762799506e-05,
      "loss": 1.2697,
      "step": 1771
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.277541721371888,
      "learning_rate": 1.9910378577778248e-05,
      "loss": 1.6219,
      "step": 1772
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.432089175637452,
      "learning_rate": 1.9910204224085777e-05,
      "loss": 1.474,
      "step": 1773
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.569606106332312,
      "learning_rate": 1.991002970172507e-05,
      "loss": 1.3646,
      "step": 1774
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.596953282022571,
      "learning_rate": 1.990985501069909e-05,
      "loss": 1.0274,
      "step": 1775
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.357675756044595,
      "learning_rate": 1.9909680151010815e-05,
      "loss": 1.3306,
      "step": 1776
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.377739488685105,
      "learning_rate": 1.9909505122663222e-05,
      "loss": 1.3393,
      "step": 1777
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.6138783849353775,
      "learning_rate": 1.990932992565928e-05,
      "loss": 1.2284,
      "step": 1778
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.60313478546979,
      "learning_rate": 1.9909154560001986e-05,
      "loss": 1.5439,
      "step": 1779
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.972049318222225,
      "learning_rate": 1.9908979025694312e-05,
      "loss": 1.6239,
      "step": 1780
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.858085993130201,
      "learning_rate": 1.990880332273925e-05,
      "loss": 1.1449,
      "step": 1781
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.423170453073888,
      "learning_rate": 1.9908627451139795e-05,
      "loss": 1.2426,
      "step": 1782
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.9883928300979745,
      "learning_rate": 1.9908451410898936e-05,
      "loss": 1.5397,
      "step": 1783
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.542155496212223,
      "learning_rate": 1.9908275202019664e-05,
      "loss": 1.4943,
      "step": 1784
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.452378534502823,
      "learning_rate": 1.9908098824504986e-05,
      "loss": 1.2322,
      "step": 1785
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.2635571017663505,
      "learning_rate": 1.99079222783579e-05,
      "loss": 1.3524,
      "step": 1786
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.703191484416512,
      "learning_rate": 1.9907745563581414e-05,
      "loss": 1.1532,
      "step": 1787
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.309888596478867,
      "learning_rate": 1.990756868017853e-05,
      "loss": 1.4145,
      "step": 1788
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.425418491896762,
      "learning_rate": 1.9907391628152265e-05,
      "loss": 1.6318,
      "step": 1789
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.748837398982095,
      "learning_rate": 1.9907214407505626e-05,
      "loss": 1.5433,
      "step": 1790
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.610148343171284,
      "learning_rate": 1.9907037018241636e-05,
      "loss": 1.4925,
      "step": 1791
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.886287135748048,
      "learning_rate": 1.9906859460363307e-05,
      "loss": 1.3725,
      "step": 1792
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.934101973483032,
      "learning_rate": 1.9906681733873667e-05,
      "loss": 1.2098,
      "step": 1793
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.821121171717437,
      "learning_rate": 1.990650383877574e-05,
      "loss": 1.5023,
      "step": 1794
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.074021152154163,
      "learning_rate": 1.990632577507255e-05,
      "loss": 1.4374,
      "step": 1795
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.933658849380015,
      "learning_rate": 1.990614754276713e-05,
      "loss": 1.9002,
      "step": 1796
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.871279041465596,
      "learning_rate": 1.9905969141862512e-05,
      "loss": 1.5813,
      "step": 1797
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.941290167208677,
      "learning_rate": 1.9905790572361736e-05,
      "loss": 1.3539,
      "step": 1798
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.316097887667811,
      "learning_rate": 1.9905611834267834e-05,
      "loss": 1.2964,
      "step": 1799
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.522166752742381,
      "learning_rate": 1.9905432927583858e-05,
      "loss": 1.7959,
      "step": 1800
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.78242110236598,
      "learning_rate": 1.9905253852312845e-05,
      "loss": 1.5231,
      "step": 1801
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.157257795621279,
      "learning_rate": 1.9905074608457843e-05,
      "loss": 1.0522,
      "step": 1802
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.289875723341624,
      "learning_rate": 1.990489519602191e-05,
      "loss": 1.2414,
      "step": 1803
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.894820781939416,
      "learning_rate": 1.9904715615008092e-05,
      "loss": 1.6934,
      "step": 1804
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.429420225885721,
      "learning_rate": 1.990453586541945e-05,
      "loss": 1.2645,
      "step": 1805
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.206983720684911,
      "learning_rate": 1.990435594725904e-05,
      "loss": 1.1669,
      "step": 1806
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.429944245912889,
      "learning_rate": 1.9904175860529922e-05,
      "loss": 1.0718,
      "step": 1807
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.344498414761276,
      "learning_rate": 1.9903995605235172e-05,
      "loss": 1.657,
      "step": 1808
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.1099072782204615,
      "learning_rate": 1.9903815181377847e-05,
      "loss": 1.2116,
      "step": 1809
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.612526679436819,
      "learning_rate": 1.990363458896102e-05,
      "loss": 1.504,
      "step": 1810
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.185900296286562,
      "learning_rate": 1.9903453827987765e-05,
      "loss": 1.3022,
      "step": 1811
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.95702676462048,
      "learning_rate": 1.990327289846116e-05,
      "loss": 1.2666,
      "step": 1812
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.220178227341689,
      "learning_rate": 1.9903091800384283e-05,
      "loss": 1.2833,
      "step": 1813
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.082063465426698,
      "learning_rate": 1.9902910533760217e-05,
      "loss": 1.4117,
      "step": 1814
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.884736141590249,
      "learning_rate": 1.990272909859205e-05,
      "loss": 1.2908,
      "step": 1815
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.5416372703163175,
      "learning_rate": 1.9902547494882865e-05,
      "loss": 1.7995,
      "step": 1816
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.925866921431485,
      "learning_rate": 1.9902365722635753e-05,
      "loss": 1.6132,
      "step": 1817
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.922915269005915,
      "learning_rate": 1.990218378185381e-05,
      "loss": 1.4446,
      "step": 1818
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.795380954368281,
      "learning_rate": 1.990200167254013e-05,
      "loss": 1.3096,
      "step": 1819
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.195386670293692,
      "learning_rate": 1.990181939469782e-05,
      "loss": 1.3163,
      "step": 1820
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.912284694575628,
      "learning_rate": 1.9901636948329974e-05,
      "loss": 1.4034,
      "step": 1821
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.32285080006908,
      "learning_rate": 1.99014543334397e-05,
      "loss": 1.4367,
      "step": 1822
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.598316673353837,
      "learning_rate": 1.9901271550030107e-05,
      "loss": 1.5952,
      "step": 1823
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.514889830843357,
      "learning_rate": 1.9901088598104305e-05,
      "loss": 1.3828,
      "step": 1824
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.739268656764798,
      "learning_rate": 1.99009054776654e-05,
      "loss": 1.2074,
      "step": 1825
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.741307080442253,
      "learning_rate": 1.9900722188716526e-05,
      "loss": 1.7741,
      "step": 1826
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.838050987195043,
      "learning_rate": 1.9900538731260786e-05,
      "loss": 1.873,
      "step": 1827
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.035000673892894,
      "learning_rate": 1.9900355105301314e-05,
      "loss": 1.4176,
      "step": 1828
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.0587549412363995,
      "learning_rate": 1.990017131084123e-05,
      "loss": 1.513,
      "step": 1829
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.687590853577961,
      "learning_rate": 1.9899987347883662e-05,
      "loss": 1.2701,
      "step": 1830
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.001211163053075,
      "learning_rate": 1.989980321643174e-05,
      "loss": 1.5076,
      "step": 1831
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.544513499111193,
      "learning_rate": 1.98996189164886e-05,
      "loss": 1.4251,
      "step": 1832
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.843430567631214,
      "learning_rate": 1.989943444805738e-05,
      "loss": 1.1513,
      "step": 1833
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.899542014326216,
      "learning_rate": 1.9899249811141215e-05,
      "loss": 1.2903,
      "step": 1834
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.877236704650846,
      "learning_rate": 1.989906500574325e-05,
      "loss": 1.3307,
      "step": 1835
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.51169605999953,
      "learning_rate": 1.9898880031866632e-05,
      "loss": 1.4188,
      "step": 1836
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.844452648018087,
      "learning_rate": 1.9898694889514506e-05,
      "loss": 1.647,
      "step": 1837
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.61160811105431,
      "learning_rate": 1.9898509578690027e-05,
      "loss": 1.2722,
      "step": 1838
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.356485262920222,
      "learning_rate": 1.9898324099396346e-05,
      "loss": 1.5551,
      "step": 1839
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.711797400099966,
      "learning_rate": 1.9898138451636618e-05,
      "loss": 1.2139,
      "step": 1840
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.250110614898936,
      "learning_rate": 1.9897952635414007e-05,
      "loss": 1.3928,
      "step": 1841
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.960504154851947,
      "learning_rate": 1.9897766650731677e-05,
      "loss": 1.422,
      "step": 1842
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.551607915883396,
      "learning_rate": 1.9897580497592784e-05,
      "loss": 1.4695,
      "step": 1843
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.377224571925624,
      "learning_rate": 1.9897394176000508e-05,
      "loss": 1.1154,
      "step": 1844
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.60280508667392,
      "learning_rate": 1.989720768595801e-05,
      "loss": 1.2947,
      "step": 1845
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.756358955303432,
      "learning_rate": 1.989702102746847e-05,
      "loss": 1.2054,
      "step": 1846
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.571809357760466,
      "learning_rate": 1.9896834200535062e-05,
      "loss": 1.4072,
      "step": 1847
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.238190866052183,
      "learning_rate": 1.989664720516097e-05,
      "loss": 1.249,
      "step": 1848
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.007015988014855,
      "learning_rate": 1.989646004134937e-05,
      "loss": 1.3853,
      "step": 1849
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.08241283981454,
      "learning_rate": 1.9896272709103455e-05,
      "loss": 1.8438,
      "step": 1850
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.4160827198627715,
      "learning_rate": 1.9896085208426405e-05,
      "loss": 1.1178,
      "step": 1851
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.167960950431204,
      "learning_rate": 1.9895897539321418e-05,
      "loss": 1.382,
      "step": 1852
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.653280038247735,
      "learning_rate": 1.9895709701791687e-05,
      "loss": 1.4701,
      "step": 1853
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.878093247756022,
      "learning_rate": 1.9895521695840406e-05,
      "loss": 1.1745,
      "step": 1854
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.541696865083197,
      "learning_rate": 1.9895333521470775e-05,
      "loss": 1.1126,
      "step": 1855
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.723480103350438,
      "learning_rate": 1.9895145178686e-05,
      "loss": 1.5545,
      "step": 1856
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.660580261372116,
      "learning_rate": 1.9894956667489286e-05,
      "loss": 1.3655,
      "step": 1857
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.37325426878096,
      "learning_rate": 1.9894767987883835e-05,
      "loss": 1.2893,
      "step": 1858
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.905116105498393,
      "learning_rate": 1.9894579139872868e-05,
      "loss": 1.3496,
      "step": 1859
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9599799363560337,
      "learning_rate": 1.9894390123459592e-05,
      "loss": 1.2177,
      "step": 1860
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.216751001268892,
      "learning_rate": 1.9894200938647224e-05,
      "loss": 1.2694,
      "step": 1861
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.087187324380208,
      "learning_rate": 1.989401158543899e-05,
      "loss": 0.9905,
      "step": 1862
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.405171557016595,
      "learning_rate": 1.9893822063838108e-05,
      "loss": 1.4704,
      "step": 1863
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.728880050427066,
      "learning_rate": 1.9893632373847803e-05,
      "loss": 1.5873,
      "step": 1864
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.6554711554294,
      "learning_rate": 1.989344251547131e-05,
      "loss": 1.5349,
      "step": 1865
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.216398772182142,
      "learning_rate": 1.989325248871185e-05,
      "loss": 1.0868,
      "step": 1866
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.735331624295228,
      "learning_rate": 1.989306229357266e-05,
      "loss": 1.2725,
      "step": 1867
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.472735416115238,
      "learning_rate": 1.9892871930056987e-05,
      "loss": 1.3095,
      "step": 1868
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.414520252337363,
      "learning_rate": 1.9892681398168062e-05,
      "loss": 1.0833,
      "step": 1869
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.034319195959007,
      "learning_rate": 1.9892490697909127e-05,
      "loss": 1.3502,
      "step": 1870
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.290917496925557,
      "learning_rate": 1.989229982928343e-05,
      "loss": 1.3774,
      "step": 1871
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.5135485726364815,
      "learning_rate": 1.989210879229422e-05,
      "loss": 1.1419,
      "step": 1872
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.521960658831057,
      "learning_rate": 1.989191758694475e-05,
      "loss": 1.8081,
      "step": 1873
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.330767326989305,
      "learning_rate": 1.9891726213238267e-05,
      "loss": 1.246,
      "step": 1874
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.909940074987961,
      "learning_rate": 1.9891534671178035e-05,
      "loss": 1.5355,
      "step": 1875
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.391192753244623,
      "learning_rate": 1.9891342960767312e-05,
      "loss": 1.2977,
      "step": 1876
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.809855256836489,
      "learning_rate": 1.9891151082009364e-05,
      "loss": 1.4352,
      "step": 1877
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.645338893387367,
      "learning_rate": 1.9890959034907452e-05,
      "loss": 1.3508,
      "step": 1878
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.472118473117855,
      "learning_rate": 1.9890766819464847e-05,
      "loss": 1.5358,
      "step": 1879
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.074032989461991,
      "learning_rate": 1.9890574435684818e-05,
      "loss": 1.5075,
      "step": 1880
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.992202967631186,
      "learning_rate": 1.989038188357064e-05,
      "loss": 1.4892,
      "step": 1881
    },
    {
      "epoch": 0.08,
      "grad_norm": 32.29535162413436,
      "learning_rate": 1.9890189163125594e-05,
      "loss": 1.2364,
      "step": 1882
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.303141467011429,
      "learning_rate": 1.9889996274352955e-05,
      "loss": 1.4995,
      "step": 1883
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.961460165241721,
      "learning_rate": 1.988980321725601e-05,
      "loss": 1.2742,
      "step": 1884
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.903422782355507,
      "learning_rate": 1.9889609991838042e-05,
      "loss": 1.7074,
      "step": 1885
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.291231553119069,
      "learning_rate": 1.988941659810234e-05,
      "loss": 1.2152,
      "step": 1886
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.835324326467648,
      "learning_rate": 1.9889223036052198e-05,
      "loss": 1.5583,
      "step": 1887
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.607203745269564,
      "learning_rate": 1.988902930569091e-05,
      "loss": 1.1908,
      "step": 1888
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.733304406953474,
      "learning_rate": 1.9888835407021765e-05,
      "loss": 1.5966,
      "step": 1889
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.715246070931457,
      "learning_rate": 1.9888641340048075e-05,
      "loss": 1.5123,
      "step": 1890
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.3958815925201735,
      "learning_rate": 1.9888447104773133e-05,
      "loss": 1.277,
      "step": 1891
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.214163076957846,
      "learning_rate": 1.9888252701200253e-05,
      "loss": 1.5097,
      "step": 1892
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.043811599076709,
      "learning_rate": 1.988805812933274e-05,
      "loss": 1.3704,
      "step": 1893
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.016743686393313,
      "learning_rate": 1.9887863389173908e-05,
      "loss": 1.3935,
      "step": 1894
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.426743542356977,
      "learning_rate": 1.9887668480727066e-05,
      "loss": 1.3407,
      "step": 1895
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.797615012238835,
      "learning_rate": 1.9887473403995534e-05,
      "loss": 1.2813,
      "step": 1896
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.0018845640442,
      "learning_rate": 1.988727815898263e-05,
      "loss": 1.6259,
      "step": 1897
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.484024172122658,
      "learning_rate": 1.9887082745691685e-05,
      "loss": 1.3847,
      "step": 1898
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.581495327619253,
      "learning_rate": 1.988688716412602e-05,
      "loss": 1.2899,
      "step": 1899
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.148368174049432,
      "learning_rate": 1.988669141428896e-05,
      "loss": 1.492,
      "step": 1900
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.058422598656176,
      "learning_rate": 1.988649549618384e-05,
      "loss": 1.2799,
      "step": 1901
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.099246718032856,
      "learning_rate": 1.988629940981399e-05,
      "loss": 1.3575,
      "step": 1902
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.75550331744537,
      "learning_rate": 1.988610315518276e-05,
      "loss": 1.4692,
      "step": 1903
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.6383837643469485,
      "learning_rate": 1.9885906732293473e-05,
      "loss": 1.5472,
      "step": 1904
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.307233565069014,
      "learning_rate": 1.9885710141149488e-05,
      "loss": 1.2707,
      "step": 1905
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.250488980604428,
      "learning_rate": 1.988551338175414e-05,
      "loss": 1.3601,
      "step": 1906
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.94454103123784,
      "learning_rate": 1.988531645411078e-05,
      "loss": 1.5703,
      "step": 1907
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.47266529671148,
      "learning_rate": 1.9885119358222765e-05,
      "loss": 1.2022,
      "step": 1908
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.304969827303509,
      "learning_rate": 1.9884922094093445e-05,
      "loss": 1.9959,
      "step": 1909
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.195758014392338,
      "learning_rate": 1.9884724661726176e-05,
      "loss": 1.5371,
      "step": 1910
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.536575092917892,
      "learning_rate": 1.988452706112432e-05,
      "loss": 1.3836,
      "step": 1911
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.915181813887974,
      "learning_rate": 1.988432929229124e-05,
      "loss": 1.4546,
      "step": 1912
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.081173340000653,
      "learning_rate": 1.9884131355230303e-05,
      "loss": 1.3403,
      "step": 1913
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.793320254187598,
      "learning_rate": 1.988393324994488e-05,
      "loss": 1.444,
      "step": 1914
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.313101418786349,
      "learning_rate": 1.9883734976438335e-05,
      "loss": 1.2467,
      "step": 1915
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.947451746894204,
      "learning_rate": 1.988353653471405e-05,
      "loss": 1.1859,
      "step": 1916
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.045131783586373,
      "learning_rate": 1.9883337924775397e-05,
      "loss": 1.4838,
      "step": 1917
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.623865516138536,
      "learning_rate": 1.9883139146625763e-05,
      "loss": 1.472,
      "step": 1918
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.848043277361003,
      "learning_rate": 1.9882940200268525e-05,
      "loss": 1.2968,
      "step": 1919
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.028172657779085,
      "learning_rate": 1.988274108570707e-05,
      "loss": 1.4765,
      "step": 1920
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.305263445746768,
      "learning_rate": 1.988254180294479e-05,
      "loss": 1.4632,
      "step": 1921
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.448802281639567,
      "learning_rate": 1.9882342351985072e-05,
      "loss": 1.4273,
      "step": 1922
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.101088181749497,
      "learning_rate": 1.9882142732831316e-05,
      "loss": 1.4528,
      "step": 1923
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.727522198138094,
      "learning_rate": 1.9881942945486916e-05,
      "loss": 1.3927,
      "step": 1924
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.65456179362216,
      "learning_rate": 1.988174298995527e-05,
      "loss": 1.4535,
      "step": 1925
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.244434902691974,
      "learning_rate": 1.9881542866239786e-05,
      "loss": 1.1475,
      "step": 1926
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.339281823390445,
      "learning_rate": 1.988134257434387e-05,
      "loss": 1.0999,
      "step": 1927
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.4584276036091195,
      "learning_rate": 1.9881142114270928e-05,
      "loss": 1.2863,
      "step": 1928
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.344711810640193,
      "learning_rate": 1.9880941486024375e-05,
      "loss": 1.3002,
      "step": 1929
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.351686169973188,
      "learning_rate": 1.988074068960762e-05,
      "loss": 1.3365,
      "step": 1930
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.769722034786174,
      "learning_rate": 1.9880539725024082e-05,
      "loss": 1.5159,
      "step": 1931
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.795344799719702,
      "learning_rate": 1.9880338592277188e-05,
      "loss": 1.4249,
      "step": 1932
    },
    {
      "epoch": 0.08,
      "grad_norm": 49.66664152571337,
      "learning_rate": 1.9880137291370356e-05,
      "loss": 1.5723,
      "step": 1933
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.549454416769481,
      "learning_rate": 1.987993582230701e-05,
      "loss": 1.1613,
      "step": 1934
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.131330768663847,
      "learning_rate": 1.9879734185090585e-05,
      "loss": 1.3872,
      "step": 1935
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.411637757687284,
      "learning_rate": 1.9879532379724507e-05,
      "loss": 1.9966,
      "step": 1936
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.771896070647335,
      "learning_rate": 1.9879330406212214e-05,
      "loss": 1.5774,
      "step": 1937
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.3398532727425145,
      "learning_rate": 1.9879128264557142e-05,
      "loss": 1.2974,
      "step": 1938
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.52315372868552,
      "learning_rate": 1.987892595476273e-05,
      "loss": 1.4551,
      "step": 1939
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.4556568476893545,
      "learning_rate": 1.9878723476832427e-05,
      "loss": 1.4021,
      "step": 1940
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.155423123645429,
      "learning_rate": 1.9878520830769675e-05,
      "loss": 1.6988,
      "step": 1941
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.36117962714811,
      "learning_rate": 1.9878318016577917e-05,
      "loss": 1.2595,
      "step": 1942
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.08492112780158,
      "learning_rate": 1.9878115034260616e-05,
      "loss": 1.5837,
      "step": 1943
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.869138684675926,
      "learning_rate": 1.9877911883821223e-05,
      "loss": 1.5631,
      "step": 1944
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.0845420574577,
      "learning_rate": 1.987770856526319e-05,
      "loss": 1.3692,
      "step": 1945
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.225301032281044,
      "learning_rate": 1.9877505078589988e-05,
      "loss": 1.2714,
      "step": 1946
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.046137332716192,
      "learning_rate": 1.9877301423805068e-05,
      "loss": 1.3963,
      "step": 1947
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.325081274730213,
      "learning_rate": 1.9877097600911903e-05,
      "loss": 1.4354,
      "step": 1948
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.458713001584281,
      "learning_rate": 1.9876893609913965e-05,
      "loss": 1.0767,
      "step": 1949
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.7971848401223545,
      "learning_rate": 1.9876689450814718e-05,
      "loss": 1.3194,
      "step": 1950
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.667725455264923,
      "learning_rate": 1.9876485123617643e-05,
      "loss": 2.0666,
      "step": 1951
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.004045602884313,
      "learning_rate": 1.9876280628326215e-05,
      "loss": 1.3416,
      "step": 1952
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.164829069345282,
      "learning_rate": 1.987607596494391e-05,
      "loss": 1.437,
      "step": 1953
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.952126644912351,
      "learning_rate": 1.9875871133474223e-05,
      "loss": 1.5062,
      "step": 1954
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.2870983332966714,
      "learning_rate": 1.987566613392063e-05,
      "loss": 1.294,
      "step": 1955
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.100405620816359,
      "learning_rate": 1.987546096628662e-05,
      "loss": 1.7477,
      "step": 1956
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.342764950449752,
      "learning_rate": 1.9875255630575693e-05,
      "loss": 1.4715,
      "step": 1957
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.137647740131131,
      "learning_rate": 1.9875050126791334e-05,
      "loss": 1.2572,
      "step": 1958
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.747413580892788,
      "learning_rate": 1.987484445493705e-05,
      "loss": 1.5153,
      "step": 1959
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.827275100563909,
      "learning_rate": 1.9874638615016337e-05,
      "loss": 1.2175,
      "step": 1960
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.819189551363388,
      "learning_rate": 1.9874432607032696e-05,
      "loss": 1.5515,
      "step": 1961
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.391896935253675,
      "learning_rate": 1.9874226430989637e-05,
      "loss": 1.6351,
      "step": 1962
    },
    {
      "epoch": 0.08,
      "grad_norm": 50.76994407845396,
      "learning_rate": 1.9874020086890664e-05,
      "loss": 1.8266,
      "step": 1963
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.729947535579848,
      "learning_rate": 1.9873813574739293e-05,
      "loss": 1.32,
      "step": 1964
    },
    {
      "epoch": 0.08,
      "grad_norm": 71.17585192663728,
      "learning_rate": 1.9873606894539043e-05,
      "loss": 2.0312,
      "step": 1965
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.982800640683587,
      "learning_rate": 1.9873400046293422e-05,
      "loss": 1.2879,
      "step": 1966
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.2020414650697235,
      "learning_rate": 1.987319303000595e-05,
      "loss": 1.4783,
      "step": 1967
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.610199168823236,
      "learning_rate": 1.9872985845680164e-05,
      "loss": 1.484,
      "step": 1968
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.680698284824642,
      "learning_rate": 1.987277849331958e-05,
      "loss": 1.7848,
      "step": 1969
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.356293134043534,
      "learning_rate": 1.987257097292773e-05,
      "loss": 1.4553,
      "step": 1970
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.543439321803046,
      "learning_rate": 1.9872363284508143e-05,
      "loss": 1.5538,
      "step": 1971
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.886208274946951,
      "learning_rate": 1.9872155428064356e-05,
      "loss": 1.2102,
      "step": 1972
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.030215927281574,
      "learning_rate": 1.9871947403599904e-05,
      "loss": 1.4892,
      "step": 1973
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.427589184249705,
      "learning_rate": 1.9871739211118332e-05,
      "loss": 1.3802,
      "step": 1974
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.9571556013096,
      "learning_rate": 1.987153085062318e-05,
      "loss": 2.2489,
      "step": 1975
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.13438091681017,
      "learning_rate": 1.9871322322117998e-05,
      "loss": 0.9751,
      "step": 1976
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.340738806307808,
      "learning_rate": 1.987111362560633e-05,
      "loss": 1.4222,
      "step": 1977
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.066015523174409,
      "learning_rate": 1.9870904761091728e-05,
      "loss": 1.4344,
      "step": 1978
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.376372344209014,
      "learning_rate": 1.987069572857775e-05,
      "loss": 1.2253,
      "step": 1979
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.567556856907022,
      "learning_rate": 1.9870486528067957e-05,
      "loss": 1.4916,
      "step": 1980
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.352045204240113,
      "learning_rate": 1.9870277159565902e-05,
      "loss": 1.4423,
      "step": 1981
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.758888762307567,
      "learning_rate": 1.987006762307515e-05,
      "loss": 1.7159,
      "step": 1982
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.705161845761703,
      "learning_rate": 1.9869857918599273e-05,
      "loss": 1.4703,
      "step": 1983
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.080581947700933,
      "learning_rate": 1.986964804614183e-05,
      "loss": 1.3423,
      "step": 1984
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.801902283726717,
      "learning_rate": 1.9869438005706404e-05,
      "loss": 1.5858,
      "step": 1985
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.194431286339335,
      "learning_rate": 1.9869227797296565e-05,
      "loss": 1.6454,
      "step": 1986
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.033776752601409,
      "learning_rate": 1.9869017420915888e-05,
      "loss": 1.5596,
      "step": 1987
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.258548302769505,
      "learning_rate": 1.9868806876567954e-05,
      "loss": 1.4228,
      "step": 1988
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.175625414139162,
      "learning_rate": 1.9868596164256352e-05,
      "loss": 1.729,
      "step": 1989
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.4132349723929,
      "learning_rate": 1.9868385283984665e-05,
      "loss": 1.4251,
      "step": 1990
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.928297429089372,
      "learning_rate": 1.986817423575648e-05,
      "loss": 1.6279,
      "step": 1991
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.732851828082064,
      "learning_rate": 1.986796301957539e-05,
      "loss": 1.5941,
      "step": 1992
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.577160730653409,
      "learning_rate": 1.986775163544499e-05,
      "loss": 1.7442,
      "step": 1993
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.3395855298044035,
      "learning_rate": 1.986754008336888e-05,
      "loss": 1.1593,
      "step": 1994
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.9369983081322015,
      "learning_rate": 1.9867328363350658e-05,
      "loss": 1.2943,
      "step": 1995
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.393487589835019,
      "learning_rate": 1.986711647539393e-05,
      "loss": 1.6736,
      "step": 1996
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.560275518553596,
      "learning_rate": 1.98669044195023e-05,
      "loss": 1.3487,
      "step": 1997
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.126566641487436,
      "learning_rate": 1.9866692195679375e-05,
      "loss": 1.6894,
      "step": 1998
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.119523502501558,
      "learning_rate": 1.986647980392877e-05,
      "loss": 1.355,
      "step": 1999
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.682670636332944,
      "learning_rate": 1.9866267244254103e-05,
      "loss": 1.1045,
      "step": 2000
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.478065499779025,
      "learning_rate": 1.9866054516658985e-05,
      "loss": 1.3317,
      "step": 2001
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.898600535774286,
      "learning_rate": 1.986584162114704e-05,
      "loss": 1.4485,
      "step": 2002
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.062876729158292,
      "learning_rate": 1.986562855772189e-05,
      "loss": 1.358,
      "step": 2003
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.273172695020129,
      "learning_rate": 1.986541532638716e-05,
      "loss": 1.2718,
      "step": 2004
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.613439102606318,
      "learning_rate": 1.9865201927146486e-05,
      "loss": 1.4793,
      "step": 2005
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.441628410394232,
      "learning_rate": 1.986498836000349e-05,
      "loss": 0.9922,
      "step": 2006
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.202889276326259,
      "learning_rate": 1.986477462496182e-05,
      "loss": 1.355,
      "step": 2007
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.3635050064863385,
      "learning_rate": 1.9864560722025098e-05,
      "loss": 1.5329,
      "step": 2008
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.41000838214148,
      "learning_rate": 1.9864346651196976e-05,
      "loss": 1.6468,
      "step": 2009
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.692878247565852,
      "learning_rate": 1.9864132412481094e-05,
      "loss": 1.0813,
      "step": 2010
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.738002300884941,
      "learning_rate": 1.98639180058811e-05,
      "loss": 1.0475,
      "step": 2011
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.0385714517372096,
      "learning_rate": 1.986370343140064e-05,
      "loss": 1.407,
      "step": 2012
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.823605859600802,
      "learning_rate": 1.9863488689043363e-05,
      "loss": 1.8481,
      "step": 2013
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.329536391505625,
      "learning_rate": 1.9863273778812934e-05,
      "loss": 1.7969,
      "step": 2014
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.236965320588755,
      "learning_rate": 1.9863058700713e-05,
      "loss": 1.3936,
      "step": 2015
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.457369005091717,
      "learning_rate": 1.986284345474723e-05,
      "loss": 1.8438,
      "step": 2016
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.718536519614692,
      "learning_rate": 1.986262804091928e-05,
      "loss": 1.0857,
      "step": 2017
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.324054694663023,
      "learning_rate": 1.986241245923282e-05,
      "loss": 1.4625,
      "step": 2018
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.984901507079154,
      "learning_rate": 1.9862196709691522e-05,
      "loss": 1.1212,
      "step": 2019
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.7095705009407975,
      "learning_rate": 1.9861980792299056e-05,
      "loss": 1.2673,
      "step": 2020
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.938045672856821,
      "learning_rate": 1.9861764707059094e-05,
      "loss": 1.3506,
      "step": 2021
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.212348386769186,
      "learning_rate": 1.9861548453975315e-05,
      "loss": 1.355,
      "step": 2022
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.95386858016036,
      "learning_rate": 1.9861332033051402e-05,
      "loss": 1.4413,
      "step": 2023
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.3363356268296815,
      "learning_rate": 1.9861115444291034e-05,
      "loss": 1.6278,
      "step": 2024
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.391962594516691,
      "learning_rate": 1.9860898687697904e-05,
      "loss": 1.3893,
      "step": 2025
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.361945359189987,
      "learning_rate": 1.9860681763275693e-05,
      "loss": 1.3278,
      "step": 2026
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.691938593058895,
      "learning_rate": 1.98604646710281e-05,
      "loss": 1.7167,
      "step": 2027
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.936805249320885,
      "learning_rate": 1.9860247410958812e-05,
      "loss": 1.3407,
      "step": 2028
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.343126756429472,
      "learning_rate": 1.9860029983071537e-05,
      "loss": 1.8915,
      "step": 2029
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.7400645903005385,
      "learning_rate": 1.9859812387369965e-05,
      "loss": 1.4172,
      "step": 2030
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.734630976959931,
      "learning_rate": 1.9859594623857806e-05,
      "loss": 1.3161,
      "step": 2031
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.2376251754373575,
      "learning_rate": 1.985937669253877e-05,
      "loss": 1.5185,
      "step": 2032
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.3593288744757475,
      "learning_rate": 1.9859158593416554e-05,
      "loss": 1.119,
      "step": 2033
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.0016955696470555,
      "learning_rate": 1.9858940326494877e-05,
      "loss": 1.4015,
      "step": 2034
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.093204002074062,
      "learning_rate": 1.9858721891777457e-05,
      "loss": 1.4231,
      "step": 2035
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.546781331476035,
      "learning_rate": 1.9858503289268007e-05,
      "loss": 1.0932,
      "step": 2036
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.104858795656092,
      "learning_rate": 1.985828451897025e-05,
      "loss": 1.8787,
      "step": 2037
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.74825242973996,
      "learning_rate": 1.9858065580887907e-05,
      "loss": 1.5114,
      "step": 2038
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.28404120686469,
      "learning_rate": 1.9857846475024704e-05,
      "loss": 1.4589,
      "step": 2039
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.865812391855606,
      "learning_rate": 1.9857627201384372e-05,
      "loss": 1.391,
      "step": 2040
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.422128495422681,
      "learning_rate": 1.9857407759970645e-05,
      "loss": 1.3031,
      "step": 2041
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.740899762848786,
      "learning_rate": 1.9857188150787254e-05,
      "loss": 1.5544,
      "step": 2042
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.40047595690418,
      "learning_rate": 1.985696837383794e-05,
      "loss": 1.4017,
      "step": 2043
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.454280308096974,
      "learning_rate": 1.9856748429126438e-05,
      "loss": 1.0161,
      "step": 2044
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.502385374574426,
      "learning_rate": 1.98565283166565e-05,
      "loss": 1.1975,
      "step": 2045
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.404091452828083,
      "learning_rate": 1.985630803643186e-05,
      "loss": 1.6209,
      "step": 2046
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.595585731363609,
      "learning_rate": 1.9856087588456278e-05,
      "loss": 1.3848,
      "step": 2047
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.248203717006554,
      "learning_rate": 1.9855866972733505e-05,
      "loss": 1.3173,
      "step": 2048
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.753662816339862,
      "learning_rate": 1.985564618926729e-05,
      "loss": 1.1622,
      "step": 2049
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.480586541317035,
      "learning_rate": 1.9855425238061393e-05,
      "loss": 1.5633,
      "step": 2050
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.341995987455431,
      "learning_rate": 1.9855204119119572e-05,
      "loss": 1.356,
      "step": 2051
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.475278231967259,
      "learning_rate": 1.98549828324456e-05,
      "loss": 1.177,
      "step": 2052
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.932490616127661,
      "learning_rate": 1.9854761378043234e-05,
      "loss": 1.2633,
      "step": 2053
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.451900146705143,
      "learning_rate": 1.9854539755916245e-05,
      "loss": 1.4707,
      "step": 2054
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.21180800605403,
      "learning_rate": 1.985431796606841e-05,
      "loss": 1.3845,
      "step": 2055
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.555010389264111,
      "learning_rate": 1.9854096008503495e-05,
      "loss": 1.3051,
      "step": 2056
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.766848976093007,
      "learning_rate": 1.9853873883225282e-05,
      "loss": 1.6173,
      "step": 2057
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.161336498618304,
      "learning_rate": 1.9853651590237554e-05,
      "loss": 1.7824,
      "step": 2058
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.043646570582553,
      "learning_rate": 1.9853429129544092e-05,
      "loss": 1.3753,
      "step": 2059
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.335500231804678,
      "learning_rate": 1.985320650114868e-05,
      "loss": 1.248,
      "step": 2060
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.486632047340316,
      "learning_rate": 1.9852983705055112e-05,
      "loss": 1.1288,
      "step": 2061
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.015814769319924,
      "learning_rate": 1.9852760741267177e-05,
      "loss": 1.395,
      "step": 2062
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.471297949698899,
      "learning_rate": 1.985253760978867e-05,
      "loss": 1.593,
      "step": 2063
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.615892953807744,
      "learning_rate": 1.9852314310623385e-05,
      "loss": 1.3446,
      "step": 2064
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.6764861541814495,
      "learning_rate": 1.985209084377513e-05,
      "loss": 1.224,
      "step": 2065
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.984682132826466,
      "learning_rate": 1.9851867209247704e-05,
      "loss": 1.2029,
      "step": 2066
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.286963968874016,
      "learning_rate": 1.9851643407044914e-05,
      "loss": 1.6169,
      "step": 2067
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.521578102390017,
      "learning_rate": 1.9851419437170566e-05,
      "loss": 1.2492,
      "step": 2068
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.7339445328126,
      "learning_rate": 1.9851195299628476e-05,
      "loss": 1.3129,
      "step": 2069
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.420731971776329,
      "learning_rate": 1.985097099442246e-05,
      "loss": 1.297,
      "step": 2070
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.925840543569701,
      "learning_rate": 1.9850746521556328e-05,
      "loss": 1.4849,
      "step": 2071
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.524618399040382,
      "learning_rate": 1.9850521881033907e-05,
      "loss": 1.6939,
      "step": 2072
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.213465774504431,
      "learning_rate": 1.985029707285902e-05,
      "loss": 1.224,
      "step": 2073
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.8137052616530065,
      "learning_rate": 1.9850072097035493e-05,
      "loss": 1.4034,
      "step": 2074
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.439276558837327,
      "learning_rate": 1.984984695356715e-05,
      "loss": 1.2804,
      "step": 2075
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.23039211308543,
      "learning_rate": 1.984962164245783e-05,
      "loss": 1.1844,
      "step": 2076
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.997671619595233,
      "learning_rate": 1.9849396163711363e-05,
      "loss": 1.1771,
      "step": 2077
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.899571590860373,
      "learning_rate": 1.9849170517331587e-05,
      "loss": 0.9749,
      "step": 2078
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.4484078862909255,
      "learning_rate": 1.9848944703322345e-05,
      "loss": 1.4435,
      "step": 2079
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.553159947970482,
      "learning_rate": 1.9848718721687478e-05,
      "loss": 1.4433,
      "step": 2080
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.02600219154172,
      "learning_rate": 1.9848492572430832e-05,
      "loss": 1.2229,
      "step": 2081
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.659733282924702,
      "learning_rate": 1.9848266255556255e-05,
      "loss": 1.3973,
      "step": 2082
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.49003489161571,
      "learning_rate": 1.9848039771067604e-05,
      "loss": 1.2507,
      "step": 2083
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.072160922255867,
      "learning_rate": 1.9847813118968727e-05,
      "loss": 1.5853,
      "step": 2084
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.184045547019883,
      "learning_rate": 1.9847586299263487e-05,
      "loss": 1.2774,
      "step": 2085
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.218850798424014,
      "learning_rate": 1.9847359311955743e-05,
      "loss": 1.443,
      "step": 2086
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.081527867587629,
      "learning_rate": 1.9847132157049357e-05,
      "loss": 1.2054,
      "step": 2087
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.058633030220518,
      "learning_rate": 1.9846904834548196e-05,
      "loss": 1.159,
      "step": 2088
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.169304717094344,
      "learning_rate": 1.9846677344456127e-05,
      "loss": 1.6185,
      "step": 2089
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.8593668527065965,
      "learning_rate": 1.9846449686777025e-05,
      "loss": 1.284,
      "step": 2090
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.323128931275353,
      "learning_rate": 1.984622186151476e-05,
      "loss": 1.3999,
      "step": 2091
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.559913890501203,
      "learning_rate": 1.9845993868673216e-05,
      "loss": 1.742,
      "step": 2092
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.8157740668673625,
      "learning_rate": 1.9845765708256265e-05,
      "loss": 1.2514,
      "step": 2093
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.3539166770356195,
      "learning_rate": 1.98455373802678e-05,
      "loss": 1.2092,
      "step": 2094
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.199856207051168,
      "learning_rate": 1.98453088847117e-05,
      "loss": 1.2876,
      "step": 2095
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.521129345934831,
      "learning_rate": 1.984508022159186e-05,
      "loss": 1.3859,
      "step": 2096
    },
    {
      "epoch": 0.08,
      "grad_norm": 58.11827962530425,
      "learning_rate": 1.9844851390912163e-05,
      "loss": 1.9576,
      "step": 2097
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.766235199335897,
      "learning_rate": 1.984462239267651e-05,
      "loss": 1.2589,
      "step": 2098
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.718447267821798,
      "learning_rate": 1.9844393226888793e-05,
      "loss": 1.5563,
      "step": 2099
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.998866019231098,
      "learning_rate": 1.9844163893552923e-05,
      "loss": 1.5821,
      "step": 2100
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.6937655745645,
      "learning_rate": 1.984393439267279e-05,
      "loss": 1.4352,
      "step": 2101
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.05595963983662,
      "learning_rate": 1.9843704724252308e-05,
      "loss": 1.4137,
      "step": 2102
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.119930206321781,
      "learning_rate": 1.9843474888295388e-05,
      "loss": 1.185,
      "step": 2103
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.126472705493255,
      "learning_rate": 1.9843244884805935e-05,
      "loss": 1.3151,
      "step": 2104
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.848231348397727,
      "learning_rate": 1.9843014713787866e-05,
      "loss": 1.7134,
      "step": 2105
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.107833459485515,
      "learning_rate": 1.98427843752451e-05,
      "loss": 1.3304,
      "step": 2106
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.27156983478511,
      "learning_rate": 1.9842553869181555e-05,
      "loss": 1.6161,
      "step": 2107
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.346621323021084,
      "learning_rate": 1.984232319560116e-05,
      "loss": 1.5305,
      "step": 2108
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.06230000514128,
      "learning_rate": 1.9842092354507835e-05,
      "loss": 1.4227,
      "step": 2109
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.405568751567361,
      "learning_rate": 1.9841861345905505e-05,
      "loss": 1.2207,
      "step": 2110
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.997328427118234,
      "learning_rate": 1.984163016979811e-05,
      "loss": 2.2131,
      "step": 2111
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.96372065153358,
      "learning_rate": 1.9841398826189585e-05,
      "loss": 1.2718,
      "step": 2112
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.167961223986975,
      "learning_rate": 1.9841167315083858e-05,
      "loss": 1.3156,
      "step": 2113
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.449055423860171,
      "learning_rate": 1.9840935636484877e-05,
      "loss": 1.1934,
      "step": 2114
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.254981984664196,
      "learning_rate": 1.9840703790396583e-05,
      "loss": 1.537,
      "step": 2115
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.784466640951036,
      "learning_rate": 1.9840471776822927e-05,
      "loss": 1.6054,
      "step": 2116
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.197839595501872,
      "learning_rate": 1.9840239595767847e-05,
      "loss": 1.2749,
      "step": 2117
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.418675478498772,
      "learning_rate": 1.9840007247235303e-05,
      "loss": 1.4637,
      "step": 2118
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.807871732267039,
      "learning_rate": 1.9839774731229247e-05,
      "loss": 1.2672,
      "step": 2119
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.69749765223561,
      "learning_rate": 1.9839542047753634e-05,
      "loss": 1.6809,
      "step": 2120
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.431462192437819,
      "learning_rate": 1.9839309196812428e-05,
      "loss": 1.0529,
      "step": 2121
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.588562292992778,
      "learning_rate": 1.9839076178409593e-05,
      "loss": 1.4438,
      "step": 2122
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.89138412003577,
      "learning_rate": 1.983884299254909e-05,
      "loss": 1.3326,
      "step": 2123
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.851090815440423,
      "learning_rate": 1.9838609639234892e-05,
      "loss": 1.1389,
      "step": 2124
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.115450581213647,
      "learning_rate": 1.9838376118470965e-05,
      "loss": 1.3287,
      "step": 2125
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.598472923197704,
      "learning_rate": 1.9838142430261293e-05,
      "loss": 1.2751,
      "step": 2126
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.309393209448436,
      "learning_rate": 1.9837908574609842e-05,
      "loss": 1.7558,
      "step": 2127
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.218686822109227,
      "learning_rate": 1.98376745515206e-05,
      "loss": 1.2729,
      "step": 2128
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.943066851311584,
      "learning_rate": 1.983744036099755e-05,
      "loss": 1.4527,
      "step": 2129
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.215902907933666,
      "learning_rate": 1.9837206003044674e-05,
      "loss": 1.3402,
      "step": 2130
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.973109497639802,
      "learning_rate": 1.9836971477665962e-05,
      "loss": 1.5662,
      "step": 2131
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.21149359569075,
      "learning_rate": 1.9836736784865406e-05,
      "loss": 1.5492,
      "step": 2132
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.018203124705373,
      "learning_rate": 1.9836501924646998e-05,
      "loss": 1.3717,
      "step": 2133
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.737189938770085,
      "learning_rate": 1.983626689701474e-05,
      "loss": 1.1464,
      "step": 2134
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.515106223177204,
      "learning_rate": 1.983603170197263e-05,
      "loss": 1.5919,
      "step": 2135
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.715792116718323,
      "learning_rate": 1.983579633952467e-05,
      "loss": 1.7391,
      "step": 2136
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.351822784282908,
      "learning_rate": 1.983556080967486e-05,
      "loss": 1.2781,
      "step": 2137
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.121850344172062,
      "learning_rate": 1.9835325112427223e-05,
      "loss": 1.3982,
      "step": 2138
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.5617134171531655,
      "learning_rate": 1.9835089247785763e-05,
      "loss": 1.473,
      "step": 2139
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.812970555955127,
      "learning_rate": 1.9834853215754488e-05,
      "loss": 1.5728,
      "step": 2140
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.853917566749489,
      "learning_rate": 1.9834617016337424e-05,
      "loss": 1.2893,
      "step": 2141
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.8519349787536,
      "learning_rate": 1.9834380649538586e-05,
      "loss": 1.1881,
      "step": 2142
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.536763702885692,
      "learning_rate": 1.9834144115362e-05,
      "loss": 1.5859,
      "step": 2143
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.405226438623203,
      "learning_rate": 1.983390741381169e-05,
      "loss": 1.5649,
      "step": 2144
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.679808971907038,
      "learning_rate": 1.9833670544891685e-05,
      "loss": 1.3384,
      "step": 2145
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.450287416090553,
      "learning_rate": 1.9833433508606018e-05,
      "loss": 1.529,
      "step": 2146
    },
    {
      "epoch": 0.09,
      "grad_norm": 24.249720411959476,
      "learning_rate": 1.983319630495872e-05,
      "loss": 1.3843,
      "step": 2147
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.818429244666359,
      "learning_rate": 1.983295893395383e-05,
      "loss": 1.4132,
      "step": 2148
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.392360692370362,
      "learning_rate": 1.983272139559539e-05,
      "loss": 1.5431,
      "step": 2149
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.029588964620906,
      "learning_rate": 1.9832483689887438e-05,
      "loss": 1.2658,
      "step": 2150
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.551146151310274,
      "learning_rate": 1.9832245816834023e-05,
      "loss": 1.4433,
      "step": 2151
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.922382842549953,
      "learning_rate": 1.9832007776439194e-05,
      "loss": 1.4921,
      "step": 2152
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.601642708939318,
      "learning_rate": 1.9831769568707e-05,
      "loss": 1.4252,
      "step": 2153
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.613467230248213,
      "learning_rate": 1.98315311936415e-05,
      "loss": 1.7025,
      "step": 2154
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.044813847208802,
      "learning_rate": 1.9831292651246742e-05,
      "loss": 1.4549,
      "step": 2155
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.990937121081952,
      "learning_rate": 1.9831053941526796e-05,
      "loss": 1.5714,
      "step": 2156
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.797955484443314,
      "learning_rate": 1.983081506448572e-05,
      "loss": 1.3849,
      "step": 2157
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.243140945057242,
      "learning_rate": 1.9830576020127577e-05,
      "loss": 1.1203,
      "step": 2158
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.194466445992653,
      "learning_rate": 1.983033680845644e-05,
      "loss": 1.2226,
      "step": 2159
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.5177857425702586,
      "learning_rate": 1.9830097429476377e-05,
      "loss": 1.3112,
      "step": 2160
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.037636701157427,
      "learning_rate": 1.9829857883191466e-05,
      "loss": 1.655,
      "step": 2161
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.897443426397933,
      "learning_rate": 1.982961816960578e-05,
      "loss": 1.4438,
      "step": 2162
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.874977978404665,
      "learning_rate": 1.98293782887234e-05,
      "loss": 1.255,
      "step": 2163
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.646579244033633,
      "learning_rate": 1.9829138240548413e-05,
      "loss": 1.4475,
      "step": 2164
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.681079010827928,
      "learning_rate": 1.9828898025084898e-05,
      "loss": 1.3579,
      "step": 2165
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.078049597857948,
      "learning_rate": 1.9828657642336947e-05,
      "loss": 1.6121,
      "step": 2166
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.734300445580301,
      "learning_rate": 1.9828417092308646e-05,
      "loss": 1.2257,
      "step": 2167
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.315772418202527,
      "learning_rate": 1.9828176375004097e-05,
      "loss": 1.2124,
      "step": 2168
    },
    {
      "epoch": 0.09,
      "grad_norm": 33.484953359548484,
      "learning_rate": 1.9827935490427393e-05,
      "loss": 1.6639,
      "step": 2169
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.441566891901978,
      "learning_rate": 1.9827694438582635e-05,
      "loss": 1.2254,
      "step": 2170
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.779562889206817,
      "learning_rate": 1.9827453219473925e-05,
      "loss": 1.286,
      "step": 2171
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.422568988701434,
      "learning_rate": 1.982721183310537e-05,
      "loss": 1.4057,
      "step": 2172
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.170934217107806,
      "learning_rate": 1.9826970279481072e-05,
      "loss": 1.5012,
      "step": 2173
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.619348316438499,
      "learning_rate": 1.982672855860515e-05,
      "loss": 1.4628,
      "step": 2174
    },
    {
      "epoch": 0.09,
      "grad_norm": 119.68899419043944,
      "learning_rate": 1.9826486670481712e-05,
      "loss": 3.5284,
      "step": 2175
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.33062875343638,
      "learning_rate": 1.982624461511488e-05,
      "loss": 1.6982,
      "step": 2176
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.931613305513558,
      "learning_rate": 1.982600239250877e-05,
      "loss": 1.3873,
      "step": 2177
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.827554319094738,
      "learning_rate": 1.9825760002667506e-05,
      "loss": 1.3944,
      "step": 2178
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.107013922183744,
      "learning_rate": 1.9825517445595212e-05,
      "loss": 1.5091,
      "step": 2179
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.081900530013561,
      "learning_rate": 1.982527472129602e-05,
      "loss": 1.2978,
      "step": 2180
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.860203157097322,
      "learning_rate": 1.9825031829774057e-05,
      "loss": 1.5169,
      "step": 2181
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.914802620545139,
      "learning_rate": 1.9824788771033462e-05,
      "loss": 1.548,
      "step": 2182
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.590894387333212,
      "learning_rate": 1.9824545545078366e-05,
      "loss": 0.9895,
      "step": 2183
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.032637597648499,
      "learning_rate": 1.9824302151912907e-05,
      "loss": 1.6021,
      "step": 2184
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.177045173185003,
      "learning_rate": 1.9824058591541234e-05,
      "loss": 1.6869,
      "step": 2185
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.843339249264051,
      "learning_rate": 1.982381486396749e-05,
      "loss": 1.4369,
      "step": 2186
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.37092532836726,
      "learning_rate": 1.9823570969195822e-05,
      "loss": 1.5016,
      "step": 2187
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.132022144009358,
      "learning_rate": 1.9823326907230384e-05,
      "loss": 1.5888,
      "step": 2188
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.716857741974544,
      "learning_rate": 1.9823082678075323e-05,
      "loss": 1.3036,
      "step": 2189
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.557280712698354,
      "learning_rate": 1.9822838281734807e-05,
      "loss": 1.9741,
      "step": 2190
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.67467083225736,
      "learning_rate": 1.9822593718212985e-05,
      "loss": 1.6161,
      "step": 2191
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.74868170786922,
      "learning_rate": 1.9822348987514025e-05,
      "loss": 1.1185,
      "step": 2192
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.65723469052163,
      "learning_rate": 1.982210408964209e-05,
      "loss": 1.4581,
      "step": 2193
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.242940490403177,
      "learning_rate": 1.9821859024601345e-05,
      "loss": 1.0819,
      "step": 2194
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.27109662109477,
      "learning_rate": 1.9821613792395967e-05,
      "loss": 0.9394,
      "step": 2195
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.463391549800129,
      "learning_rate": 1.9821368393030127e-05,
      "loss": 1.3728,
      "step": 2196
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.570914939074665,
      "learning_rate": 1.9821122826508e-05,
      "loss": 1.2208,
      "step": 2197
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.714525603016291,
      "learning_rate": 1.9820877092833773e-05,
      "loss": 1.5355,
      "step": 2198
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.842768182807974,
      "learning_rate": 1.9820631192011618e-05,
      "loss": 1.9009,
      "step": 2199
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.6021668328893774,
      "learning_rate": 1.9820385124045725e-05,
      "loss": 1.1475,
      "step": 2200
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.926957060186036,
      "learning_rate": 1.9820138888940286e-05,
      "loss": 1.5232,
      "step": 2201
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.12948475778242,
      "learning_rate": 1.9819892486699485e-05,
      "loss": 1.6132,
      "step": 2202
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.4955331526551845,
      "learning_rate": 1.9819645917327516e-05,
      "loss": 1.4578,
      "step": 2203
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.037800495883764,
      "learning_rate": 1.981939918082858e-05,
      "loss": 1.6176,
      "step": 2204
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.833791577414573,
      "learning_rate": 1.9819152277206874e-05,
      "loss": 1.2693,
      "step": 2205
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.519088830593718,
      "learning_rate": 1.98189052064666e-05,
      "loss": 1.2969,
      "step": 2206
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.776263505534228,
      "learning_rate": 1.9818657968611963e-05,
      "loss": 1.4722,
      "step": 2207
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.907905931489199,
      "learning_rate": 1.9818410563647172e-05,
      "loss": 1.2125,
      "step": 2208
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.011724837052742,
      "learning_rate": 1.981816299157644e-05,
      "loss": 1.0828,
      "step": 2209
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.75490711624516,
      "learning_rate": 1.9817915252403973e-05,
      "loss": 1.326,
      "step": 2210
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.85572245274576,
      "learning_rate": 1.9817667346133995e-05,
      "loss": 1.2886,
      "step": 2211
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.358118676115781,
      "learning_rate": 1.9817419272770718e-05,
      "loss": 1.3831,
      "step": 2212
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.574388800189298,
      "learning_rate": 1.981717103231837e-05,
      "loss": 1.2479,
      "step": 2213
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.930142040638645,
      "learning_rate": 1.981692262478118e-05,
      "loss": 1.1691,
      "step": 2214
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.293104411838765,
      "learning_rate": 1.9816674050163366e-05,
      "loss": 1.6866,
      "step": 2215
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.979976522826998,
      "learning_rate": 1.9816425308469163e-05,
      "loss": 1.9124,
      "step": 2216
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.775595961718293,
      "learning_rate": 1.9816176399702806e-05,
      "loss": 1.2914,
      "step": 2217
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.821757472794938,
      "learning_rate": 1.9815927323868528e-05,
      "loss": 1.6291,
      "step": 2218
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.464951897816095,
      "learning_rate": 1.9815678080970574e-05,
      "loss": 1.3126,
      "step": 2219
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.623158993669226,
      "learning_rate": 1.981542867101318e-05,
      "loss": 1.5193,
      "step": 2220
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.52193403309273,
      "learning_rate": 1.9815179094000588e-05,
      "loss": 1.8671,
      "step": 2221
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.713833295812381,
      "learning_rate": 1.9814929349937054e-05,
      "loss": 1.0225,
      "step": 2222
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.216328518432011,
      "learning_rate": 1.9814679438826826e-05,
      "loss": 1.3291,
      "step": 2223
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.930286917116389,
      "learning_rate": 1.9814429360674158e-05,
      "loss": 1.5232,
      "step": 2224
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.6446798731208645,
      "learning_rate": 1.9814179115483303e-05,
      "loss": 1.3697,
      "step": 2225
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.398420866200483,
      "learning_rate": 1.981392870325852e-05,
      "loss": 1.0964,
      "step": 2226
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.675559635444843,
      "learning_rate": 1.9813678124004078e-05,
      "loss": 1.2761,
      "step": 2227
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.286088278388259,
      "learning_rate": 1.981342737772423e-05,
      "loss": 1.4092,
      "step": 2228
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.329069935826217,
      "learning_rate": 1.9813176464423255e-05,
      "loss": 1.6465,
      "step": 2229
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.122298097986522,
      "learning_rate": 1.9812925384105416e-05,
      "loss": 0.9894,
      "step": 2230
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.799462363435874,
      "learning_rate": 1.981267413677499e-05,
      "loss": 1.3434,
      "step": 2231
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.972480107966845,
      "learning_rate": 1.9812422722436254e-05,
      "loss": 1.443,
      "step": 2232
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.650724217988361,
      "learning_rate": 1.9812171141093484e-05,
      "loss": 1.0992,
      "step": 2233
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.935915311958544,
      "learning_rate": 1.981191939275096e-05,
      "loss": 1.6155,
      "step": 2234
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.700876084149738,
      "learning_rate": 1.9811667477412972e-05,
      "loss": 1.1604,
      "step": 2235
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.896066610940687,
      "learning_rate": 1.9811415395083803e-05,
      "loss": 1.3248,
      "step": 2236
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.050652963231763,
      "learning_rate": 1.981116314576775e-05,
      "loss": 1.3466,
      "step": 2237
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.051142127930135,
      "learning_rate": 1.98109107294691e-05,
      "loss": 1.5764,
      "step": 2238
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.22467871666585,
      "learning_rate": 1.981065814619215e-05,
      "loss": 1.4366,
      "step": 2239
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.3612733858837895,
      "learning_rate": 1.98104053959412e-05,
      "loss": 1.1638,
      "step": 2240
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.677878483592012,
      "learning_rate": 1.981015247872055e-05,
      "loss": 1.6954,
      "step": 2241
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.735076921680322,
      "learning_rate": 1.9809899394534505e-05,
      "loss": 1.9427,
      "step": 2242
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.64146162873456,
      "learning_rate": 1.9809646143387374e-05,
      "loss": 1.5667,
      "step": 2243
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.252219092040173,
      "learning_rate": 1.9809392725283466e-05,
      "loss": 1.3843,
      "step": 2244
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.382678712283265,
      "learning_rate": 1.9809139140227097e-05,
      "loss": 1.4533,
      "step": 2245
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.704497503856482,
      "learning_rate": 1.9808885388222578e-05,
      "loss": 1.2716,
      "step": 2246
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.643919518517507,
      "learning_rate": 1.9808631469274234e-05,
      "loss": 1.2191,
      "step": 2247
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.104766447960723,
      "learning_rate": 1.980837738338638e-05,
      "loss": 1.3034,
      "step": 2248
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.264482665874045,
      "learning_rate": 1.9808123130563343e-05,
      "loss": 1.4667,
      "step": 2249
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.262810167578607,
      "learning_rate": 1.980786871080945e-05,
      "loss": 1.3823,
      "step": 2250
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.229917165581827,
      "learning_rate": 1.9807614124129034e-05,
      "loss": 1.2753,
      "step": 2251
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.8002524729154,
      "learning_rate": 1.9807359370526425e-05,
      "loss": 1.4502,
      "step": 2252
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.514224622960106,
      "learning_rate": 1.9807104450005962e-05,
      "loss": 1.2071,
      "step": 2253
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.117451877344915,
      "learning_rate": 1.9806849362571978e-05,
      "loss": 1.7524,
      "step": 2254
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.446803291189573,
      "learning_rate": 1.980659410822882e-05,
      "loss": 1.4324,
      "step": 2255
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.204477024225661,
      "learning_rate": 1.9806338686980825e-05,
      "loss": 1.2728,
      "step": 2256
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.513502680165082,
      "learning_rate": 1.980608309883235e-05,
      "loss": 1.1821,
      "step": 2257
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.864596488954305,
      "learning_rate": 1.980582734378774e-05,
      "loss": 1.174,
      "step": 2258
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.57540194428217,
      "learning_rate": 1.9805571421851345e-05,
      "loss": 1.3875,
      "step": 2259
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.871270072300933,
      "learning_rate": 1.980531533302753e-05,
      "loss": 1.6042,
      "step": 2260
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.153764303858159,
      "learning_rate": 1.9805059077320643e-05,
      "loss": 1.2587,
      "step": 2261
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.634854873773975,
      "learning_rate": 1.9804802654735048e-05,
      "loss": 1.2827,
      "step": 2262
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.436011980111569,
      "learning_rate": 1.9804546065275116e-05,
      "loss": 1.4932,
      "step": 2263
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.1529211812132525,
      "learning_rate": 1.9804289308945207e-05,
      "loss": 1.5369,
      "step": 2264
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.5276050598796,
      "learning_rate": 1.9804032385749693e-05,
      "loss": 1.3374,
      "step": 2265
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.955919974843701,
      "learning_rate": 1.9803775295692944e-05,
      "loss": 1.4445,
      "step": 2266
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.735865948029496,
      "learning_rate": 1.980351803877934e-05,
      "loss": 1.6915,
      "step": 2267
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.091024703482934,
      "learning_rate": 1.9803260615013257e-05,
      "loss": 1.6724,
      "step": 2268
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.299801763832618,
      "learning_rate": 1.980300302439908e-05,
      "loss": 1.2007,
      "step": 2269
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.086150301725009,
      "learning_rate": 1.980274526694119e-05,
      "loss": 1.398,
      "step": 2270
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.154678537577464,
      "learning_rate": 1.980248734264397e-05,
      "loss": 1.513,
      "step": 2271
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.670652538745526,
      "learning_rate": 1.9802229251511815e-05,
      "loss": 1.3006,
      "step": 2272
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.738828673167921,
      "learning_rate": 1.980197099354912e-05,
      "loss": 1.2935,
      "step": 2273
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.670523290173077,
      "learning_rate": 1.9801712568760273e-05,
      "loss": 1.2846,
      "step": 2274
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.814570880174142,
      "learning_rate": 1.9801453977149674e-05,
      "loss": 1.2825,
      "step": 2275
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.053366255786789,
      "learning_rate": 1.9801195218721734e-05,
      "loss": 1.3144,
      "step": 2276
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.113979128022825,
      "learning_rate": 1.9800936293480847e-05,
      "loss": 1.6025,
      "step": 2277
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.273020099675095,
      "learning_rate": 1.9800677201431417e-05,
      "loss": 1.3955,
      "step": 2278
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.9208848384884165,
      "learning_rate": 1.9800417942577865e-05,
      "loss": 1.5187,
      "step": 2279
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.336788340396834,
      "learning_rate": 1.9800158516924593e-05,
      "loss": 1.3084,
      "step": 2280
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.722623191784169,
      "learning_rate": 1.9799898924476022e-05,
      "loss": 1.2517,
      "step": 2281
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.837008197305857,
      "learning_rate": 1.9799639165236568e-05,
      "loss": 1.3607,
      "step": 2282
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.47358606709877,
      "learning_rate": 1.9799379239210654e-05,
      "loss": 1.1804,
      "step": 2283
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.388501971117802,
      "learning_rate": 1.9799119146402703e-05,
      "loss": 1.3098,
      "step": 2284
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.549232759883415,
      "learning_rate": 1.979885888681714e-05,
      "loss": 1.4922,
      "step": 2285
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.969302243087315,
      "learning_rate": 1.9798598460458394e-05,
      "loss": 1.2463,
      "step": 2286
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.292745131968066,
      "learning_rate": 1.9798337867330907e-05,
      "loss": 1.1767,
      "step": 2287
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.23170360050032,
      "learning_rate": 1.97980771074391e-05,
      "loss": 1.4472,
      "step": 2288
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.623997762440592,
      "learning_rate": 1.9797816180787418e-05,
      "loss": 1.3854,
      "step": 2289
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.792177052243116,
      "learning_rate": 1.97975550873803e-05,
      "loss": 1.3113,
      "step": 2290
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.089261522258525,
      "learning_rate": 1.9797293827222194e-05,
      "loss": 1.1542,
      "step": 2291
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.787551284696548,
      "learning_rate": 1.9797032400317543e-05,
      "loss": 1.1647,
      "step": 2292
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.6498494989094725,
      "learning_rate": 1.9796770806670795e-05,
      "loss": 1.7065,
      "step": 2293
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.225555189168992,
      "learning_rate": 1.979650904628641e-05,
      "loss": 1.2545,
      "step": 2294
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.359320306566103,
      "learning_rate": 1.9796247119168832e-05,
      "loss": 1.3106,
      "step": 2295
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.01055905853141,
      "learning_rate": 1.9795985025322525e-05,
      "loss": 1.7033,
      "step": 2296
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.891219777461577,
      "learning_rate": 1.979572276475195e-05,
      "loss": 1.2933,
      "step": 2297
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.589534035004933,
      "learning_rate": 1.979546033746157e-05,
      "loss": 1.3456,
      "step": 2298
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.828405956998997,
      "learning_rate": 1.979519774345585e-05,
      "loss": 1.3914,
      "step": 2299
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.407218418186931,
      "learning_rate": 1.979493498273926e-05,
      "loss": 1.3972,
      "step": 2300
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.300819827888258,
      "learning_rate": 1.979467205531627e-05,
      "loss": 1.5014,
      "step": 2301
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.369858996830136,
      "learning_rate": 1.9794408961191362e-05,
      "loss": 1.2413,
      "step": 2302
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.470620070124043,
      "learning_rate": 1.9794145700369007e-05,
      "loss": 1.2812,
      "step": 2303
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.05400563217698,
      "learning_rate": 1.9793882272853688e-05,
      "loss": 1.7169,
      "step": 2304
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.148956365044722,
      "learning_rate": 1.9793618678649888e-05,
      "loss": 1.2752,
      "step": 2305
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.055338612325986,
      "learning_rate": 1.9793354917762096e-05,
      "loss": 1.3632,
      "step": 2306
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.951062331253587,
      "learning_rate": 1.9793090990194795e-05,
      "loss": 1.3419,
      "step": 2307
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.204507590274996,
      "learning_rate": 1.979282689595248e-05,
      "loss": 1.2839,
      "step": 2308
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.807214933211405,
      "learning_rate": 1.979256263503965e-05,
      "loss": 1.5369,
      "step": 2309
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.020449480062286,
      "learning_rate": 1.9792298207460793e-05,
      "loss": 1.4177,
      "step": 2310
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.625559408723252,
      "learning_rate": 1.9792033613220422e-05,
      "loss": 1.0572,
      "step": 2311
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.277135341343127,
      "learning_rate": 1.979176885232303e-05,
      "loss": 1.5635,
      "step": 2312
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.258081832944338,
      "learning_rate": 1.9791503924773126e-05,
      "loss": 1.2927,
      "step": 2313
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.211124612288353,
      "learning_rate": 1.979123883057522e-05,
      "loss": 1.1576,
      "step": 2314
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.810563655285875,
      "learning_rate": 1.9790973569733825e-05,
      "loss": 1.5758,
      "step": 2315
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.235994249562045,
      "learning_rate": 1.9790708142253456e-05,
      "loss": 1.261,
      "step": 2316
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.769641259717488,
      "learning_rate": 1.9790442548138626e-05,
      "loss": 1.4045,
      "step": 2317
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.338530163532364,
      "learning_rate": 1.979017678739386e-05,
      "loss": 1.4369,
      "step": 2318
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.023206439533541,
      "learning_rate": 1.9789910860023677e-05,
      "loss": 1.3618,
      "step": 2319
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.176496351653952,
      "learning_rate": 1.9789644766032604e-05,
      "loss": 1.4055,
      "step": 2320
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.370978828060599,
      "learning_rate": 1.9789378505425173e-05,
      "loss": 1.4803,
      "step": 2321
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.49617763684483,
      "learning_rate": 1.9789112078205918e-05,
      "loss": 1.4739,
      "step": 2322
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.563416070876334,
      "learning_rate": 1.978884548437936e-05,
      "loss": 1.1277,
      "step": 2323
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.5005460012442375,
      "learning_rate": 1.9788578723950053e-05,
      "loss": 1.4434,
      "step": 2324
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.290455558746135,
      "learning_rate": 1.9788311796922527e-05,
      "loss": 1.582,
      "step": 2325
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.288105712047095,
      "learning_rate": 1.9788044703301327e-05,
      "loss": 1.86,
      "step": 2326
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.746960049002564,
      "learning_rate": 1.9787777443091e-05,
      "loss": 1.4346,
      "step": 2327
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.935723944753187,
      "learning_rate": 1.9787510016296095e-05,
      "loss": 1.0585,
      "step": 2328
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.803863764278566,
      "learning_rate": 1.9787242422921163e-05,
      "loss": 1.4575,
      "step": 2329
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.392008227940538,
      "learning_rate": 1.9786974662970758e-05,
      "loss": 1.3128,
      "step": 2330
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.654788540628237,
      "learning_rate": 1.9786706736449436e-05,
      "loss": 1.2252,
      "step": 2331
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.962868483444048,
      "learning_rate": 1.978643864336176e-05,
      "loss": 1.5635,
      "step": 2332
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.890644268893915,
      "learning_rate": 1.978617038371229e-05,
      "loss": 1.6848,
      "step": 2333
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.978357958792903,
      "learning_rate": 1.9785901957505592e-05,
      "loss": 1.4512,
      "step": 2334
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.157856887797164,
      "learning_rate": 1.9785633364746234e-05,
      "loss": 1.4105,
      "step": 2335
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.818300383704187,
      "learning_rate": 1.978536460543879e-05,
      "loss": 1.3506,
      "step": 2336
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.0187521832239,
      "learning_rate": 1.9785095679587836e-05,
      "loss": 1.5442,
      "step": 2337
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.336400475364567,
      "learning_rate": 1.9784826587197942e-05,
      "loss": 1.607,
      "step": 2338
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.46176835177031,
      "learning_rate": 1.9784557328273696e-05,
      "loss": 1.5016,
      "step": 2339
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.436280238490885,
      "learning_rate": 1.9784287902819673e-05,
      "loss": 1.1873,
      "step": 2340
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.395440181414065,
      "learning_rate": 1.978401831084046e-05,
      "loss": 1.455,
      "step": 2341
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.49295549620925,
      "learning_rate": 1.978374855234065e-05,
      "loss": 1.0875,
      "step": 2342
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.553754161335859,
      "learning_rate": 1.9783478627324833e-05,
      "loss": 1.7287,
      "step": 2343
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.091099174184596,
      "learning_rate": 1.9783208535797602e-05,
      "loss": 1.3867,
      "step": 2344
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.816272472701812,
      "learning_rate": 1.9782938277763547e-05,
      "loss": 1.3163,
      "step": 2345
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.325533281728228,
      "learning_rate": 1.9782667853227278e-05,
      "loss": 1.2925,
      "step": 2346
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.774503574457326,
      "learning_rate": 1.9782397262193398e-05,
      "loss": 1.4665,
      "step": 2347
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.239155409430976,
      "learning_rate": 1.9782126504666502e-05,
      "loss": 1.5979,
      "step": 2348
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.308282359683663,
      "learning_rate": 1.9781855580651207e-05,
      "loss": 1.519,
      "step": 2349
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.782115078365031,
      "learning_rate": 1.978158449015212e-05,
      "loss": 1.6723,
      "step": 2350
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.286537351194205,
      "learning_rate": 1.9781313233173858e-05,
      "loss": 1.4596,
      "step": 2351
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.737455978053662,
      "learning_rate": 1.9781041809721032e-05,
      "loss": 1.5336,
      "step": 2352
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.909025841860506,
      "learning_rate": 1.978077021979827e-05,
      "loss": 1.6065,
      "step": 2353
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.041378923459224,
      "learning_rate": 1.9780498463410187e-05,
      "loss": 1.7557,
      "step": 2354
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.697575369707412,
      "learning_rate": 1.9780226540561413e-05,
      "loss": 1.286,
      "step": 2355
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.645396053787499,
      "learning_rate": 1.977995445125657e-05,
      "loss": 1.5277,
      "step": 2356
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.139478509198994,
      "learning_rate": 1.9779682195500297e-05,
      "loss": 1.404,
      "step": 2357
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.608353730408532,
      "learning_rate": 1.977940977329722e-05,
      "loss": 1.3109,
      "step": 2358
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.446476810984435,
      "learning_rate": 1.9779137184651983e-05,
      "loss": 1.252,
      "step": 2359
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.681720462200904,
      "learning_rate": 1.9778864429569218e-05,
      "loss": 1.5071,
      "step": 2360
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.113596827536437,
      "learning_rate": 1.9778591508053573e-05,
      "loss": 1.4697,
      "step": 2361
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.365578117426721,
      "learning_rate": 1.977831842010969e-05,
      "loss": 1.2852,
      "step": 2362
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.872350184857425,
      "learning_rate": 1.9778045165742216e-05,
      "loss": 1.1824,
      "step": 2363
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.115627604945798,
      "learning_rate": 1.9777771744955804e-05,
      "loss": 1.3251,
      "step": 2364
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.197142668936221,
      "learning_rate": 1.9777498157755105e-05,
      "loss": 1.5379,
      "step": 2365
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.069140332784883,
      "learning_rate": 1.977722440414478e-05,
      "loss": 1.6133,
      "step": 2366
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.282741907563678,
      "learning_rate": 1.9776950484129484e-05,
      "loss": 1.3122,
      "step": 2367
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.579124253701632,
      "learning_rate": 1.9776676397713882e-05,
      "loss": 1.5344,
      "step": 2368
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.234074244359075,
      "learning_rate": 1.9776402144902636e-05,
      "loss": 1.5122,
      "step": 2369
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.104892642666924,
      "learning_rate": 1.9776127725700413e-05,
      "loss": 1.1049,
      "step": 2370
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.101666703954131,
      "learning_rate": 1.9775853140111887e-05,
      "loss": 1.4982,
      "step": 2371
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.818277688615716,
      "learning_rate": 1.977557838814173e-05,
      "loss": 1.2647,
      "step": 2372
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.542974421187077,
      "learning_rate": 1.977530346979462e-05,
      "loss": 1.4095,
      "step": 2373
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.439748550519008,
      "learning_rate": 1.977502838507523e-05,
      "loss": 0.8275,
      "step": 2374
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.025433194786921,
      "learning_rate": 1.9774753133988246e-05,
      "loss": 1.1935,
      "step": 2375
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.323258265745918,
      "learning_rate": 1.9774477716538352e-05,
      "loss": 1.3953,
      "step": 2376
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.234839979396902,
      "learning_rate": 1.977420213273024e-05,
      "loss": 1.6676,
      "step": 2377
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.1756150533742895,
      "learning_rate": 1.9773926382568592e-05,
      "loss": 1.4695,
      "step": 2378
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.6299358078533155,
      "learning_rate": 1.9773650466058107e-05,
      "loss": 1.1667,
      "step": 2379
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.709647800653659,
      "learning_rate": 1.977337438320348e-05,
      "loss": 1.1521,
      "step": 2380
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.352227388811451,
      "learning_rate": 1.977309813400941e-05,
      "loss": 0.992,
      "step": 2381
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.0209300111154835,
      "learning_rate": 1.9772821718480596e-05,
      "loss": 1.1522,
      "step": 2382
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.970260572492284,
      "learning_rate": 1.9772545136621746e-05,
      "loss": 1.5545,
      "step": 2383
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.2576045814755465,
      "learning_rate": 1.9772268388437562e-05,
      "loss": 1.2986,
      "step": 2384
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.753774365377731,
      "learning_rate": 1.9771991473932764e-05,
      "loss": 1.5024,
      "step": 2385
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.879106583684735,
      "learning_rate": 1.9771714393112057e-05,
      "loss": 1.3666,
      "step": 2386
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.816612076620179,
      "learning_rate": 1.9771437145980155e-05,
      "loss": 1.3493,
      "step": 2387
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.275079342119255,
      "learning_rate": 1.9771159732541783e-05,
      "loss": 1.6081,
      "step": 2388
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.240677902769774,
      "learning_rate": 1.9770882152801662e-05,
      "loss": 1.2584,
      "step": 2389
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.3073863214931025,
      "learning_rate": 1.977060440676451e-05,
      "loss": 1.149,
      "step": 2390
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.8025112277602275,
      "learning_rate": 1.977032649443506e-05,
      "loss": 1.2676,
      "step": 2391
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.286296544568502,
      "learning_rate": 1.9770048415818042e-05,
      "loss": 1.4232,
      "step": 2392
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.863040857611469,
      "learning_rate": 1.9769770170918187e-05,
      "loss": 0.8656,
      "step": 2393
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.539837149411529,
      "learning_rate": 1.976949175974023e-05,
      "loss": 1.8224,
      "step": 2394
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.589612391885002,
      "learning_rate": 1.976921318228891e-05,
      "loss": 1.4666,
      "step": 2395
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.289036891601582,
      "learning_rate": 1.9768934438568967e-05,
      "loss": 1.2597,
      "step": 2396
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.100475058053136,
      "learning_rate": 1.9768655528585147e-05,
      "loss": 1.3084,
      "step": 2397
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.727862760370415,
      "learning_rate": 1.9768376452342197e-05,
      "loss": 1.332,
      "step": 2398
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.633192839215411,
      "learning_rate": 1.9768097209844867e-05,
      "loss": 0.99,
      "step": 2399
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.30743139076143,
      "learning_rate": 1.976781780109791e-05,
      "loss": 1.2059,
      "step": 2400
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.947547214362693,
      "learning_rate": 1.9767538226106078e-05,
      "loss": 1.3394,
      "step": 2401
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.573479943471469,
      "learning_rate": 1.9767258484874136e-05,
      "loss": 1.7008,
      "step": 2402
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.664828573057713,
      "learning_rate": 1.9766978577406834e-05,
      "loss": 1.3304,
      "step": 2403
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.6189670903477635,
      "learning_rate": 1.9766698503708945e-05,
      "loss": 1.33,
      "step": 2404
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.96458374837972,
      "learning_rate": 1.9766418263785237e-05,
      "loss": 1.2183,
      "step": 2405
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.487236312312979,
      "learning_rate": 1.976613785764047e-05,
      "loss": 1.469,
      "step": 2406
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.4274288304752005,
      "learning_rate": 1.9765857285279432e-05,
      "loss": 1.4245,
      "step": 2407
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.20778941634519,
      "learning_rate": 1.976557654670688e-05,
      "loss": 1.2476,
      "step": 2408
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.328902456172008,
      "learning_rate": 1.9765295641927604e-05,
      "loss": 1.3873,
      "step": 2409
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.993270302209906,
      "learning_rate": 1.9765014570946383e-05,
      "loss": 1.5852,
      "step": 2410
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.513559657351155,
      "learning_rate": 1.9764733333767998e-05,
      "loss": 1.4948,
      "step": 2411
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.111248953835878,
      "learning_rate": 1.9764451930397235e-05,
      "loss": 1.4332,
      "step": 2412
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.1986550466949675,
      "learning_rate": 1.9764170360838886e-05,
      "loss": 1.2678,
      "step": 2413
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.315287515718362,
      "learning_rate": 1.976388862509774e-05,
      "loss": 1.0943,
      "step": 2414
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.4549355052132205,
      "learning_rate": 1.9763606723178602e-05,
      "loss": 1.084,
      "step": 2415
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.4637995556816215,
      "learning_rate": 1.976332465508626e-05,
      "loss": 1.6363,
      "step": 2416
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.07483323233534,
      "learning_rate": 1.9763042420825513e-05,
      "loss": 1.3429,
      "step": 2417
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.212156582204302,
      "learning_rate": 1.9762760020401172e-05,
      "loss": 1.3462,
      "step": 2418
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.6254916782672195,
      "learning_rate": 1.976247745381804e-05,
      "loss": 1.2216,
      "step": 2419
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.3778045481600985,
      "learning_rate": 1.9762194721080926e-05,
      "loss": 1.1359,
      "step": 2420
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.8702363290839275,
      "learning_rate": 1.9761911822194643e-05,
      "loss": 1.6536,
      "step": 2421
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.773962552878389,
      "learning_rate": 1.9761628757164002e-05,
      "loss": 1.2052,
      "step": 2422
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.963056452181607,
      "learning_rate": 1.9761345525993826e-05,
      "loss": 1.7302,
      "step": 2423
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.416849015630392,
      "learning_rate": 1.9761062128688932e-05,
      "loss": 1.3611,
      "step": 2424
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.564680309555223,
      "learning_rate": 1.9760778565254148e-05,
      "loss": 1.1967,
      "step": 2425
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.846261665463614,
      "learning_rate": 1.976049483569429e-05,
      "loss": 1.4635,
      "step": 2426
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.712065082087612,
      "learning_rate": 1.97602109400142e-05,
      "loss": 1.421,
      "step": 2427
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.089016323328259,
      "learning_rate": 1.9759926878218703e-05,
      "loss": 1.1574,
      "step": 2428
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.347513902152747,
      "learning_rate": 1.9759642650312633e-05,
      "loss": 1.3844,
      "step": 2429
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.073420672555265,
      "learning_rate": 1.9759358256300827e-05,
      "loss": 1.1729,
      "step": 2430
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.995792486720135,
      "learning_rate": 1.9759073696188127e-05,
      "loss": 1.1315,
      "step": 2431
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.287106373752574,
      "learning_rate": 1.975878896997938e-05,
      "loss": 1.3228,
      "step": 2432
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.138740153735842,
      "learning_rate": 1.9758504077679425e-05,
      "loss": 1.4879,
      "step": 2433
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.174036650312244,
      "learning_rate": 1.9758219019293115e-05,
      "loss": 1.3063,
      "step": 2434
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.668052127880092,
      "learning_rate": 1.97579337948253e-05,
      "loss": 1.2467,
      "step": 2435
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.883660742892354,
      "learning_rate": 1.975764840428083e-05,
      "loss": 1.2907,
      "step": 2436
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.487756411917772,
      "learning_rate": 1.9757362847664574e-05,
      "loss": 1.1775,
      "step": 2437
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.711475171237999,
      "learning_rate": 1.9757077124981383e-05,
      "loss": 1.0746,
      "step": 2438
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.333239556863978,
      "learning_rate": 1.975679123623612e-05,
      "loss": 1.5081,
      "step": 2439
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.560943980113215,
      "learning_rate": 1.9756505181433654e-05,
      "loss": 1.3991,
      "step": 2440
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.063242720032983,
      "learning_rate": 1.9756218960578854e-05,
      "loss": 1.4994,
      "step": 2441
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.82615857484061,
      "learning_rate": 1.9755932573676587e-05,
      "loss": 1.6439,
      "step": 2442
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.051744772190796,
      "learning_rate": 1.9755646020731728e-05,
      "loss": 1.2911,
      "step": 2443
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.685018516542887,
      "learning_rate": 1.975535930174916e-05,
      "loss": 1.4109,
      "step": 2444
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.2151942726645,
      "learning_rate": 1.9755072416733756e-05,
      "loss": 1.2421,
      "step": 2445
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.207488791982005,
      "learning_rate": 1.9754785365690403e-05,
      "loss": 1.3384,
      "step": 2446
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.190489932409248,
      "learning_rate": 1.9754498148623985e-05,
      "loss": 1.4638,
      "step": 2447
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.696121120359858,
      "learning_rate": 1.975421076553939e-05,
      "loss": 1.4681,
      "step": 2448
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.070757299860934,
      "learning_rate": 1.975392321644151e-05,
      "loss": 1.2239,
      "step": 2449
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.100002574189488,
      "learning_rate": 1.9753635501335236e-05,
      "loss": 1.3039,
      "step": 2450
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.445605511542439,
      "learning_rate": 1.975334762022547e-05,
      "loss": 1.4239,
      "step": 2451
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.984800751996431,
      "learning_rate": 1.9753059573117105e-05,
      "loss": 1.1344,
      "step": 2452
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.445415734833298,
      "learning_rate": 1.9752771360015053e-05,
      "loss": 1.1388,
      "step": 2453
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.536766513256212,
      "learning_rate": 1.975248298092421e-05,
      "loss": 1.5078,
      "step": 2454
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.94465643624607,
      "learning_rate": 1.975219443584949e-05,
      "loss": 1.371,
      "step": 2455
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.57138242400799,
      "learning_rate": 1.97519057247958e-05,
      "loss": 1.5209,
      "step": 2456
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.6006928040398885,
      "learning_rate": 1.9751616847768053e-05,
      "loss": 1.2998,
      "step": 2457
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.47192036814895,
      "learning_rate": 1.9751327804771172e-05,
      "loss": 1.3711,
      "step": 2458
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.643635758346635,
      "learning_rate": 1.9751038595810068e-05,
      "loss": 1.4132,
      "step": 2459
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.883925815532739,
      "learning_rate": 1.9750749220889673e-05,
      "loss": 1.4994,
      "step": 2460
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.755740301849055,
      "learning_rate": 1.9750459680014904e-05,
      "loss": 1.3762,
      "step": 2461
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.89682322350582,
      "learning_rate": 1.975016997319069e-05,
      "loss": 1.7078,
      "step": 2462
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.575981286381875,
      "learning_rate": 1.9749880100421967e-05,
      "loss": 1.2477,
      "step": 2463
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.184497377969023,
      "learning_rate": 1.9749590061713663e-05,
      "loss": 1.3656,
      "step": 2464
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.958324399610432,
      "learning_rate": 1.9749299857070716e-05,
      "loss": 1.3767,
      "step": 2465
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.12023312740827,
      "learning_rate": 1.9749009486498064e-05,
      "loss": 1.5188,
      "step": 2466
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.497908280092942,
      "learning_rate": 1.9748718950000653e-05,
      "loss": 1.3222,
      "step": 2467
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.2370476186823005,
      "learning_rate": 1.9748428247583423e-05,
      "loss": 1.2115,
      "step": 2468
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.965137462496832,
      "learning_rate": 1.9748137379251324e-05,
      "loss": 1.2071,
      "step": 2469
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.514050449460222,
      "learning_rate": 1.9747846345009306e-05,
      "loss": 1.1752,
      "step": 2470
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.341545435811534,
      "learning_rate": 1.9747555144862327e-05,
      "loss": 1.3015,
      "step": 2471
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.423157995644816,
      "learning_rate": 1.9747263778815335e-05,
      "loss": 1.5654,
      "step": 2472
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.409162197740796,
      "learning_rate": 1.974697224687329e-05,
      "loss": 1.0982,
      "step": 2473
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.586100238607348,
      "learning_rate": 1.9746680549041163e-05,
      "loss": 1.5776,
      "step": 2474
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.848624016694678,
      "learning_rate": 1.9746388685323907e-05,
      "loss": 1.3578,
      "step": 2475
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.300857202920487,
      "learning_rate": 1.9746096655726497e-05,
      "loss": 1.3031,
      "step": 2476
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.480346173294212,
      "learning_rate": 1.9745804460253903e-05,
      "loss": 0.8772,
      "step": 2477
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.832706186773727,
      "learning_rate": 1.9745512098911092e-05,
      "loss": 1.3348,
      "step": 2478
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.009632440307114,
      "learning_rate": 1.9745219571703042e-05,
      "loss": 1.7949,
      "step": 2479
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.954167734482992,
      "learning_rate": 1.974492687863474e-05,
      "loss": 1.1958,
      "step": 2480
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.935506098639694,
      "learning_rate": 1.9744634019711155e-05,
      "loss": 1.3224,
      "step": 2481
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.407714859873503,
      "learning_rate": 1.974434099493728e-05,
      "loss": 1.5656,
      "step": 2482
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.636815000135334,
      "learning_rate": 1.9744047804318097e-05,
      "loss": 1.6051,
      "step": 2483
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.4316843546847715,
      "learning_rate": 1.97437544478586e-05,
      "loss": 1.062,
      "step": 2484
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.835939241298115,
      "learning_rate": 1.9743460925563783e-05,
      "loss": 1.1627,
      "step": 2485
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.537300126249334,
      "learning_rate": 1.9743167237438637e-05,
      "loss": 1.597,
      "step": 2486
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.64606556085158,
      "learning_rate": 1.9742873383488163e-05,
      "loss": 1.4732,
      "step": 2487
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.4696880934698004,
      "learning_rate": 1.974257936371736e-05,
      "loss": 1.4101,
      "step": 2488
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.734206369883296,
      "learning_rate": 1.9742285178131233e-05,
      "loss": 1.5091,
      "step": 2489
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.646663051702214,
      "learning_rate": 1.9741990826734793e-05,
      "loss": 1.8179,
      "step": 2490
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.949321939249056,
      "learning_rate": 1.9741696309533042e-05,
      "loss": 1.4236,
      "step": 2491
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.615022293077521,
      "learning_rate": 1.9741401626531003e-05,
      "loss": 1.0857,
      "step": 2492
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.190171276770775,
      "learning_rate": 1.974110677773368e-05,
      "loss": 1.3074,
      "step": 2493
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.019877268896954,
      "learning_rate": 1.97408117631461e-05,
      "loss": 1.3625,
      "step": 2494
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.6182056571985575,
      "learning_rate": 1.9740516582773283e-05,
      "loss": 1.1987,
      "step": 2495
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.416483147531896,
      "learning_rate": 1.9740221236620245e-05,
      "loss": 1.2295,
      "step": 2496
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.839186228160043,
      "learning_rate": 1.973992572469202e-05,
      "loss": 1.1997,
      "step": 2497
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.138526379447447,
      "learning_rate": 1.973963004699364e-05,
      "loss": 1.1367,
      "step": 2498
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.71341067783331,
      "learning_rate": 1.973933420353013e-05,
      "loss": 1.5714,
      "step": 2499
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.293538561554527,
      "learning_rate": 1.9739038194306528e-05,
      "loss": 1.297,
      "step": 2500
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.291853478249205,
      "learning_rate": 1.9738742019327875e-05,
      "loss": 1.089,
      "step": 2501
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.070496979564334,
      "learning_rate": 1.9738445678599206e-05,
      "loss": 1.543,
      "step": 2502
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.454494272585979,
      "learning_rate": 1.9738149172125566e-05,
      "loss": 1.1625,
      "step": 2503
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.455262123675351,
      "learning_rate": 1.9737852499912006e-05,
      "loss": 1.4755,
      "step": 2504
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.840162638767774,
      "learning_rate": 1.9737555661963573e-05,
      "loss": 1.6069,
      "step": 2505
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.780915941394309,
      "learning_rate": 1.973725865828532e-05,
      "loss": 1.4268,
      "step": 2506
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.320601516683957,
      "learning_rate": 1.97369614888823e-05,
      "loss": 1.2379,
      "step": 2507
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.720291432432725,
      "learning_rate": 1.9736664153759565e-05,
      "loss": 1.0744,
      "step": 2508
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.987449368240055,
      "learning_rate": 1.9736366652922188e-05,
      "loss": 1.6249,
      "step": 2509
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.532215216540594,
      "learning_rate": 1.9736068986375222e-05,
      "loss": 1.5233,
      "step": 2510
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.069008746314104,
      "learning_rate": 1.9735771154123742e-05,
      "loss": 1.213,
      "step": 2511
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.824199356299683,
      "learning_rate": 1.9735473156172808e-05,
      "loss": 1.5735,
      "step": 2512
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.082075189932723,
      "learning_rate": 1.9735174992527498e-05,
      "loss": 1.0434,
      "step": 2513
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.923635342204987,
      "learning_rate": 1.9734876663192883e-05,
      "loss": 1.3727,
      "step": 2514
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.04703745672627,
      "learning_rate": 1.9734578168174044e-05,
      "loss": 1.3505,
      "step": 2515
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.779894182461942,
      "learning_rate": 1.9734279507476057e-05,
      "loss": 1.0725,
      "step": 2516
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.658138648044491,
      "learning_rate": 1.973398068110401e-05,
      "loss": 1.3463,
      "step": 2517
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.839489411405488,
      "learning_rate": 1.973368168906299e-05,
      "loss": 1.3811,
      "step": 2518
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.118705737640947,
      "learning_rate": 1.9733382531358073e-05,
      "loss": 1.1019,
      "step": 2519
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.842196335714272,
      "learning_rate": 1.9733083207994367e-05,
      "loss": 1.126,
      "step": 2520
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.795827233205269,
      "learning_rate": 1.9732783718976957e-05,
      "loss": 1.2167,
      "step": 2521
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.34291033975161,
      "learning_rate": 1.973248406431094e-05,
      "loss": 1.3082,
      "step": 2522
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.485082348198171,
      "learning_rate": 1.973218424400142e-05,
      "loss": 1.5184,
      "step": 2523
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.201681497338545,
      "learning_rate": 1.97318842580535e-05,
      "loss": 1.4952,
      "step": 2524
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.786421578703849,
      "learning_rate": 1.973158410647228e-05,
      "loss": 1.5083,
      "step": 2525
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.170509306928906,
      "learning_rate": 1.9731283789262877e-05,
      "loss": 1.3967,
      "step": 2526
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.485243579930072,
      "learning_rate": 1.9730983306430396e-05,
      "loss": 1.3816,
      "step": 2527
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.132264137214185,
      "learning_rate": 1.973068265797995e-05,
      "loss": 1.3127,
      "step": 2528
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.977391876260524,
      "learning_rate": 1.973038184391666e-05,
      "loss": 1.5879,
      "step": 2529
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.652429282391596,
      "learning_rate": 1.9730080864245647e-05,
      "loss": 1.53,
      "step": 2530
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.30106402723896,
      "learning_rate": 1.972977971897203e-05,
      "loss": 1.515,
      "step": 2531
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.057634385169399,
      "learning_rate": 1.9729478408100933e-05,
      "loss": 1.4551,
      "step": 2532
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.2130591298821765,
      "learning_rate": 1.9729176931637487e-05,
      "loss": 1.1922,
      "step": 2533
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.811616563202031,
      "learning_rate": 1.9728875289586824e-05,
      "loss": 1.3936,
      "step": 2534
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.0430246481648595,
      "learning_rate": 1.9728573481954075e-05,
      "loss": 1.3444,
      "step": 2535
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.093425293607571,
      "learning_rate": 1.972827150874438e-05,
      "loss": 1.3472,
      "step": 2536
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.73977166428045,
      "learning_rate": 1.9727969369962877e-05,
      "loss": 1.2742,
      "step": 2537
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.402948572001233,
      "learning_rate": 1.9727667065614703e-05,
      "loss": 1.3686,
      "step": 2538
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.522379616520328,
      "learning_rate": 1.9727364595705012e-05,
      "loss": 1.4152,
      "step": 2539
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.45297160480035,
      "learning_rate": 1.9727061960238948e-05,
      "loss": 1.0922,
      "step": 2540
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.116096096203538,
      "learning_rate": 1.9726759159221662e-05,
      "loss": 1.0823,
      "step": 2541
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.123595476813397,
      "learning_rate": 1.972645619265831e-05,
      "loss": 1.4807,
      "step": 2542
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.5752367803110765,
      "learning_rate": 1.972615306055404e-05,
      "loss": 1.4202,
      "step": 2543
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.228200499914623,
      "learning_rate": 1.9725849762914015e-05,
      "loss": 1.1249,
      "step": 2544
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.2736127969757005,
      "learning_rate": 1.9725546299743407e-05,
      "loss": 1.3513,
      "step": 2545
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.962823614733744,
      "learning_rate": 1.9725242671047365e-05,
      "loss": 1.312,
      "step": 2546
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.480483980456306,
      "learning_rate": 1.9724938876831067e-05,
      "loss": 1.1961,
      "step": 2547
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.577530021198025,
      "learning_rate": 1.9724634917099683e-05,
      "loss": 1.3993,
      "step": 2548
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.618886043389187,
      "learning_rate": 1.9724330791858382e-05,
      "loss": 1.443,
      "step": 2549
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.479859292745246,
      "learning_rate": 1.9724026501112343e-05,
      "loss": 1.2056,
      "step": 2550
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.97982456503693,
      "learning_rate": 1.9723722044866745e-05,
      "loss": 1.3498,
      "step": 2551
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.940270348244879,
      "learning_rate": 1.972341742312677e-05,
      "loss": 1.4322,
      "step": 2552
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.4979506781185945,
      "learning_rate": 1.9723112635897595e-05,
      "loss": 1.7645,
      "step": 2553
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.4608654751304675,
      "learning_rate": 1.9722807683184418e-05,
      "loss": 1.4636,
      "step": 2554
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.577685968851478,
      "learning_rate": 1.9722502564992428e-05,
      "loss": 1.37,
      "step": 2555
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.396471204200455,
      "learning_rate": 1.972219728132681e-05,
      "loss": 1.245,
      "step": 2556
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.261552051716292,
      "learning_rate": 1.9721891832192764e-05,
      "loss": 1.4803,
      "step": 2557
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.844550835074468,
      "learning_rate": 1.9721586217595496e-05,
      "loss": 1.4101,
      "step": 2558
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.111566223673272,
      "learning_rate": 1.9721280437540197e-05,
      "loss": 1.3702,
      "step": 2559
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.964504024698751,
      "learning_rate": 1.9720974492032073e-05,
      "loss": 1.4542,
      "step": 2560
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.0427961928055804,
      "learning_rate": 1.9720668381076337e-05,
      "loss": 1.2805,
      "step": 2561
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.890183302965444,
      "learning_rate": 1.9720362104678193e-05,
      "loss": 1.4862,
      "step": 2562
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.760601390188951,
      "learning_rate": 1.9720055662842856e-05,
      "loss": 1.3909,
      "step": 2563
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.670609260341819,
      "learning_rate": 1.9719749055575544e-05,
      "loss": 1.1174,
      "step": 2564
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.10808418327401,
      "learning_rate": 1.971944228288147e-05,
      "loss": 1.3224,
      "step": 2565
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.581019467013325,
      "learning_rate": 1.971913534476586e-05,
      "loss": 1.6409,
      "step": 2566
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.147327137615482,
      "learning_rate": 1.9718828241233935e-05,
      "loss": 1.4296,
      "step": 2567
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.925467265580668,
      "learning_rate": 1.971852097229092e-05,
      "loss": 1.304,
      "step": 2568
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.947998284566251,
      "learning_rate": 1.971821353794205e-05,
      "loss": 1.3643,
      "step": 2569
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.210875232762943,
      "learning_rate": 1.9717905938192558e-05,
      "loss": 1.5467,
      "step": 2570
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.0957404310164,
      "learning_rate": 1.9717598173047673e-05,
      "loss": 1.2647,
      "step": 2571
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.397809068048734,
      "learning_rate": 1.9717290242512632e-05,
      "loss": 1.3148,
      "step": 2572
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.6121276094373314,
      "learning_rate": 1.9716982146592684e-05,
      "loss": 1.133,
      "step": 2573
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.90533337923404,
      "learning_rate": 1.9716673885293066e-05,
      "loss": 1.4559,
      "step": 2574
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.522578140924586,
      "learning_rate": 1.971636545861903e-05,
      "loss": 1.3052,
      "step": 2575
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.666716266027498,
      "learning_rate": 1.9716056866575822e-05,
      "loss": 1.2982,
      "step": 2576
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.718605776142715,
      "learning_rate": 1.9715748109168696e-05,
      "loss": 0.9182,
      "step": 2577
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.48792606287642,
      "learning_rate": 1.9715439186402903e-05,
      "loss": 1.4046,
      "step": 2578
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.205638488319808,
      "learning_rate": 1.9715130098283702e-05,
      "loss": 1.0977,
      "step": 2579
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.626440399506317,
      "learning_rate": 1.9714820844816358e-05,
      "loss": 1.2659,
      "step": 2580
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.111753445647643,
      "learning_rate": 1.971451142600613e-05,
      "loss": 1.3977,
      "step": 2581
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.7160376626386595,
      "learning_rate": 1.971420184185829e-05,
      "loss": 1.5231,
      "step": 2582
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.384911200035775,
      "learning_rate": 1.9713892092378093e-05,
      "loss": 1.3073,
      "step": 2583
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.371804409585002,
      "learning_rate": 1.9713582177570826e-05,
      "loss": 1.2322,
      "step": 2584
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.232292813034817,
      "learning_rate": 1.9713272097441755e-05,
      "loss": 1.2428,
      "step": 2585
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.609874843452016,
      "learning_rate": 1.9712961851996163e-05,
      "loss": 1.2985,
      "step": 2586
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.647828417674238,
      "learning_rate": 1.9712651441239328e-05,
      "loss": 1.0808,
      "step": 2587
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.546445954421866,
      "learning_rate": 1.9712340865176532e-05,
      "loss": 1.3253,
      "step": 2588
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.299834495007353,
      "learning_rate": 1.9712030123813058e-05,
      "loss": 1.2245,
      "step": 2589
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.84215651674454,
      "learning_rate": 1.9711719217154204e-05,
      "loss": 1.669,
      "step": 2590
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.8009027873066845,
      "learning_rate": 1.9711408145205253e-05,
      "loss": 1.1867,
      "step": 2591
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.361569737684312,
      "learning_rate": 1.97110969079715e-05,
      "loss": 1.469,
      "step": 2592
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.482124003750835,
      "learning_rate": 1.971078550545825e-05,
      "loss": 1.2488,
      "step": 2593
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.617585624826382,
      "learning_rate": 1.9710473937670792e-05,
      "loss": 1.287,
      "step": 2594
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.562722403308764,
      "learning_rate": 1.9710162204614435e-05,
      "loss": 1.4877,
      "step": 2595
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.639918939120277,
      "learning_rate": 1.9709850306294485e-05,
      "loss": 1.5055,
      "step": 2596
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.378697982189352,
      "learning_rate": 1.970953824271625e-05,
      "loss": 1.1987,
      "step": 2597
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.917312702005351,
      "learning_rate": 1.970922601388504e-05,
      "loss": 1.1321,
      "step": 2598
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.872735214192032,
      "learning_rate": 1.970891361980617e-05,
      "loss": 1.267,
      "step": 2599
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.983803978661779,
      "learning_rate": 1.970860106048495e-05,
      "loss": 1.1915,
      "step": 2600
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.536393733639486,
      "learning_rate": 1.9708288335926714e-05,
      "loss": 1.3072,
      "step": 2601
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.8302395304708945,
      "learning_rate": 1.970797544613677e-05,
      "loss": 1.3533,
      "step": 2602
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.772681751432056,
      "learning_rate": 1.9707662391120456e-05,
      "loss": 1.3297,
      "step": 2603
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.003462753384896,
      "learning_rate": 1.9707349170883087e-05,
      "loss": 1.4141,
      "step": 2604
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.975087336141571,
      "learning_rate": 1.9707035785430005e-05,
      "loss": 1.2704,
      "step": 2605
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.857051644240189,
      "learning_rate": 1.9706722234766537e-05,
      "loss": 1.3668,
      "step": 2606
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.753991782344862,
      "learning_rate": 1.9706408518898025e-05,
      "loss": 1.456,
      "step": 2607
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.571156882821755,
      "learning_rate": 1.9706094637829797e-05,
      "loss": 1.7173,
      "step": 2608
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.185155369222365,
      "learning_rate": 1.9705780591567213e-05,
      "loss": 1.5724,
      "step": 2609
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.278671248295658,
      "learning_rate": 1.9705466380115603e-05,
      "loss": 1.2371,
      "step": 2610
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.582594741907626,
      "learning_rate": 1.970515200348032e-05,
      "loss": 1.3111,
      "step": 2611
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.290958544940048,
      "learning_rate": 1.9704837461666715e-05,
      "loss": 1.2586,
      "step": 2612
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.007766331586256,
      "learning_rate": 1.9704522754680143e-05,
      "loss": 1.3112,
      "step": 2613
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.71701695075993,
      "learning_rate": 1.9704207882525955e-05,
      "loss": 1.6345,
      "step": 2614
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.445357235939104,
      "learning_rate": 1.9703892845209513e-05,
      "loss": 1.3071,
      "step": 2615
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.129241481667356,
      "learning_rate": 1.9703577642736182e-05,
      "loss": 1.2857,
      "step": 2616
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.794384650257385,
      "learning_rate": 1.970326227511132e-05,
      "loss": 1.3616,
      "step": 2617
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.893052752147094,
      "learning_rate": 1.97029467423403e-05,
      "loss": 1.2112,
      "step": 2618
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.814211109622403,
      "learning_rate": 1.9702631044428492e-05,
      "loss": 1.221,
      "step": 2619
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.972677239655188,
      "learning_rate": 1.9702315181381264e-05,
      "loss": 1.2337,
      "step": 2620
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.196352352861117,
      "learning_rate": 1.9701999153203997e-05,
      "loss": 1.4647,
      "step": 2621
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.905569600372065,
      "learning_rate": 1.970168295990207e-05,
      "loss": 1.4314,
      "step": 2622
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.843204010319253,
      "learning_rate": 1.970136660148086e-05,
      "loss": 1.3417,
      "step": 2623
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.093669898283489,
      "learning_rate": 1.9701050077945754e-05,
      "loss": 0.9207,
      "step": 2624
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.371296593125958,
      "learning_rate": 1.970073338930214e-05,
      "loss": 1.364,
      "step": 2625
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.456231878503099,
      "learning_rate": 1.9700416535555404e-05,
      "loss": 1.3838,
      "step": 2626
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.367962455969566,
      "learning_rate": 1.9700099516710944e-05,
      "loss": 1.6727,
      "step": 2627
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.092825874018594,
      "learning_rate": 1.9699782332774154e-05,
      "loss": 1.3588,
      "step": 2628
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.648597273201228,
      "learning_rate": 1.969946498375043e-05,
      "loss": 1.0985,
      "step": 2629
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.244211271625733,
      "learning_rate": 1.969914746964517e-05,
      "loss": 1.2563,
      "step": 2630
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.5734040374852025,
      "learning_rate": 1.9698829790463792e-05,
      "loss": 1.5872,
      "step": 2631
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.9132671764684295,
      "learning_rate": 1.969851194621169e-05,
      "loss": 1.3093,
      "step": 2632
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.869013547994214,
      "learning_rate": 1.969819393689427e-05,
      "loss": 1.3391,
      "step": 2633
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.935787370378075,
      "learning_rate": 1.969787576251696e-05,
      "loss": 1.3522,
      "step": 2634
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.693761333803899,
      "learning_rate": 1.9697557423085163e-05,
      "loss": 1.2907,
      "step": 2635
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.946515616282997,
      "learning_rate": 1.9697238918604303e-05,
      "loss": 1.4278,
      "step": 2636
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.508275403223399,
      "learning_rate": 1.9696920249079795e-05,
      "loss": 1.3119,
      "step": 2637
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.6256961738392,
      "learning_rate": 1.969660141451707e-05,
      "loss": 1.6451,
      "step": 2638
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.446599124142236,
      "learning_rate": 1.969628241492155e-05,
      "loss": 1.2343,
      "step": 2639
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.452133046659204,
      "learning_rate": 1.969596325029866e-05,
      "loss": 1.4679,
      "step": 2640
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.592834168939742,
      "learning_rate": 1.9695643920653845e-05,
      "loss": 1.2738,
      "step": 2641
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.432758786585461,
      "learning_rate": 1.9695324425992527e-05,
      "loss": 1.2182,
      "step": 2642
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.188941705415736,
      "learning_rate": 1.9695004766320148e-05,
      "loss": 1.1953,
      "step": 2643
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.4436755639824215,
      "learning_rate": 1.969468494164215e-05,
      "loss": 1.0823,
      "step": 2644
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.490861642998686,
      "learning_rate": 1.9694364951963972e-05,
      "loss": 1.3339,
      "step": 2645
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.848845453207603,
      "learning_rate": 1.9694044797291068e-05,
      "loss": 1.2896,
      "step": 2646
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.194653474198149,
      "learning_rate": 1.9693724477628882e-05,
      "loss": 1.3313,
      "step": 2647
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.72612689262227,
      "learning_rate": 1.9693403992982865e-05,
      "loss": 1.3308,
      "step": 2648
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.21344519617811,
      "learning_rate": 1.9693083343358473e-05,
      "loss": 1.4584,
      "step": 2649
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.86341747952875,
      "learning_rate": 1.969276252876116e-05,
      "loss": 1.2772,
      "step": 2650
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.786564555091909,
      "learning_rate": 1.9692441549196393e-05,
      "loss": 1.2751,
      "step": 2651
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.8928547299243546,
      "learning_rate": 1.969212040466963e-05,
      "loss": 1.3377,
      "step": 2652
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.149200571263584,
      "learning_rate": 1.9691799095186338e-05,
      "loss": 1.5266,
      "step": 2653
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.625527245206051,
      "learning_rate": 1.9691477620751985e-05,
      "loss": 1.9414,
      "step": 2654
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.649545192142785,
      "learning_rate": 1.9691155981372043e-05,
      "loss": 1.0669,
      "step": 2655
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.100652777122418,
      "learning_rate": 1.9690834177051984e-05,
      "loss": 1.4298,
      "step": 2656
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.42682685603318,
      "learning_rate": 1.9690512207797287e-05,
      "loss": 1.4025,
      "step": 2657
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.485111633744882,
      "learning_rate": 1.9690190073613434e-05,
      "loss": 1.3335,
      "step": 2658
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.344349622565846,
      "learning_rate": 1.9689867774505906e-05,
      "loss": 1.1668,
      "step": 2659
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.503404580393387,
      "learning_rate": 1.9689545310480186e-05,
      "loss": 1.4361,
      "step": 2660
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.697256452857929,
      "learning_rate": 1.9689222681541766e-05,
      "loss": 1.7503,
      "step": 2661
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.745235298324686,
      "learning_rate": 1.9688899887696134e-05,
      "loss": 1.2887,
      "step": 2662
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.710038701970412,
      "learning_rate": 1.9688576928948783e-05,
      "loss": 1.3471,
      "step": 2663
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.107554587208237,
      "learning_rate": 1.968825380530521e-05,
      "loss": 1.4368,
      "step": 2664
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.757835486457908,
      "learning_rate": 1.9687930516770922e-05,
      "loss": 1.2799,
      "step": 2665
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.116851055686256,
      "learning_rate": 1.9687607063351414e-05,
      "loss": 1.7312,
      "step": 2666
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.5675377980561445,
      "learning_rate": 1.9687283445052192e-05,
      "loss": 1.1898,
      "step": 2667
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.5652486917275397,
      "learning_rate": 1.9686959661878763e-05,
      "loss": 1.1707,
      "step": 2668
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.608071758920661,
      "learning_rate": 1.9686635713836638e-05,
      "loss": 1.5562,
      "step": 2669
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.698834640255228,
      "learning_rate": 1.9686311600931334e-05,
      "loss": 1.3651,
      "step": 2670
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.793840444509706,
      "learning_rate": 1.968598732316836e-05,
      "loss": 1.3487,
      "step": 2671
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.77514727182542,
      "learning_rate": 1.9685662880553242e-05,
      "loss": 1.3989,
      "step": 2672
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.203341460102947,
      "learning_rate": 1.96853382730915e-05,
      "loss": 1.4155,
      "step": 2673
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.21709180764204,
      "learning_rate": 1.968501350078866e-05,
      "loss": 1.0167,
      "step": 2674
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.801862324736657,
      "learning_rate": 1.9684688563650247e-05,
      "loss": 1.3805,
      "step": 2675
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.03908982357433,
      "learning_rate": 1.9684363461681792e-05,
      "loss": 1.4309,
      "step": 2676
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.681365024019717,
      "learning_rate": 1.9684038194888827e-05,
      "loss": 1.1512,
      "step": 2677
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.669043516411221,
      "learning_rate": 1.968371276327689e-05,
      "loss": 1.2958,
      "step": 2678
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.542748346643057,
      "learning_rate": 1.968338716685152e-05,
      "loss": 1.4848,
      "step": 2679
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.623009384238525,
      "learning_rate": 1.968306140561826e-05,
      "loss": 1.5436,
      "step": 2680
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.3282366296029755,
      "learning_rate": 1.9682735479582648e-05,
      "loss": 1.3925,
      "step": 2681
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.990307853502858,
      "learning_rate": 1.9682409388750234e-05,
      "loss": 1.074,
      "step": 2682
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.819775468128765,
      "learning_rate": 1.9682083133126572e-05,
      "loss": 1.293,
      "step": 2683
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.763354357104793,
      "learning_rate": 1.968175671271721e-05,
      "loss": 1.4267,
      "step": 2684
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.975424420129589,
      "learning_rate": 1.9681430127527705e-05,
      "loss": 1.5483,
      "step": 2685
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.555192594316251,
      "learning_rate": 1.9681103377563617e-05,
      "loss": 1.1957,
      "step": 2686
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.180674169508494,
      "learning_rate": 1.9680776462830506e-05,
      "loss": 1.5546,
      "step": 2687
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.31878764622501,
      "learning_rate": 1.968044938333393e-05,
      "loss": 1.2565,
      "step": 2688
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.533048670545851,
      "learning_rate": 1.968012213907947e-05,
      "loss": 1.0506,
      "step": 2689
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.735540246500043,
      "learning_rate": 1.9679794730072684e-05,
      "loss": 1.0938,
      "step": 2690
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.024936087787572,
      "learning_rate": 1.9679467156319148e-05,
      "loss": 1.6703,
      "step": 2691
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.045798465348672,
      "learning_rate": 1.9679139417824435e-05,
      "loss": 1.1684,
      "step": 2692
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.579634980579847,
      "learning_rate": 1.9678811514594124e-05,
      "loss": 1.3536,
      "step": 2693
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.745255431764085,
      "learning_rate": 1.9678483446633797e-05,
      "loss": 1.1109,
      "step": 2694
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.542947265265834,
      "learning_rate": 1.967815521394904e-05,
      "loss": 1.4376,
      "step": 2695
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.164446908759423,
      "learning_rate": 1.9677826816545434e-05,
      "loss": 1.1871,
      "step": 2696
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.82784491661167,
      "learning_rate": 1.9677498254428572e-05,
      "loss": 1.6281,
      "step": 2697
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.353131912383838,
      "learning_rate": 1.9677169527604047e-05,
      "loss": 1.3564,
      "step": 2698
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.755077056494895,
      "learning_rate": 1.9676840636077447e-05,
      "loss": 1.3584,
      "step": 2699
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.424367356868997,
      "learning_rate": 1.9676511579854375e-05,
      "loss": 1.0231,
      "step": 2700
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.08466756546122,
      "learning_rate": 1.9676182358940434e-05,
      "loss": 1.6196,
      "step": 2701
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.343602729307134,
      "learning_rate": 1.967585297334122e-05,
      "loss": 1.4593,
      "step": 2702
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.058906287184488,
      "learning_rate": 1.9675523423062347e-05,
      "loss": 1.4502,
      "step": 2703
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.768534045984661,
      "learning_rate": 1.9675193708109415e-05,
      "loss": 1.4222,
      "step": 2704
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.325415876580892,
      "learning_rate": 1.967486382848804e-05,
      "loss": 1.4382,
      "step": 2705
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.489690987699771,
      "learning_rate": 1.967453378420384e-05,
      "loss": 1.2465,
      "step": 2706
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.098141216297506,
      "learning_rate": 1.967420357526243e-05,
      "loss": 1.2797,
      "step": 2707
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.95364769501616,
      "learning_rate": 1.9673873201669428e-05,
      "loss": 1.37,
      "step": 2708
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.117781522531783,
      "learning_rate": 1.9673542663430457e-05,
      "loss": 1.3079,
      "step": 2709
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.485972373108887,
      "learning_rate": 1.967321196055114e-05,
      "loss": 1.1891,
      "step": 2710
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.53221064131135,
      "learning_rate": 1.9672881093037114e-05,
      "loss": 1.5583,
      "step": 2711
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.310811151339031,
      "learning_rate": 1.9672550060894003e-05,
      "loss": 1.2058,
      "step": 2712
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.422410736448932,
      "learning_rate": 1.9672218864127445e-05,
      "loss": 1.3707,
      "step": 2713
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.176430183471184,
      "learning_rate": 1.9671887502743075e-05,
      "loss": 1.3765,
      "step": 2714
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.590180176362584,
      "learning_rate": 1.9671555976746536e-05,
      "loss": 1.4214,
      "step": 2715
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.0084746626061385,
      "learning_rate": 1.9671224286143462e-05,
      "loss": 1.335,
      "step": 2716
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.98658812168835,
      "learning_rate": 1.9670892430939504e-05,
      "loss": 1.4309,
      "step": 2717
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.108684537305699,
      "learning_rate": 1.967056041114031e-05,
      "loss": 1.1104,
      "step": 2718
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.905700178962875,
      "learning_rate": 1.967022822675153e-05,
      "loss": 1.2274,
      "step": 2719
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.893630922726997,
      "learning_rate": 1.966989587777882e-05,
      "loss": 1.3914,
      "step": 2720
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.305385370644963,
      "learning_rate": 1.9669563364227834e-05,
      "loss": 1.5114,
      "step": 2721
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.793130231758179,
      "learning_rate": 1.9669230686104233e-05,
      "loss": 1.1672,
      "step": 2722
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.47311372819021,
      "learning_rate": 1.9668897843413676e-05,
      "loss": 1.3378,
      "step": 2723
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.264044963137334,
      "learning_rate": 1.966856483616183e-05,
      "loss": 1.3463,
      "step": 2724
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.010400668513547,
      "learning_rate": 1.9668231664354365e-05,
      "loss": 0.8856,
      "step": 2725
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.8070727574670595,
      "learning_rate": 1.9667898327996947e-05,
      "loss": 1.592,
      "step": 2726
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.185424625114049,
      "learning_rate": 1.9667564827095248e-05,
      "loss": 1.4691,
      "step": 2727
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.790047347138205,
      "learning_rate": 1.9667231161654952e-05,
      "loss": 1.187,
      "step": 2728
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.035057492653119,
      "learning_rate": 1.9666897331681732e-05,
      "loss": 1.4052,
      "step": 2729
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.931391081861229,
      "learning_rate": 1.966656333718127e-05,
      "loss": 1.3659,
      "step": 2730
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.169570998747281,
      "learning_rate": 1.966622917815925e-05,
      "loss": 1.3168,
      "step": 2731
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.877566618953312,
      "learning_rate": 1.966589485462136e-05,
      "loss": 1.1932,
      "step": 2732
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.3657600811029935,
      "learning_rate": 1.9665560366573292e-05,
      "loss": 1.2801,
      "step": 2733
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.071324244342275,
      "learning_rate": 1.966522571402074e-05,
      "loss": 1.4703,
      "step": 2734
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.515860057439115,
      "learning_rate": 1.9664890896969397e-05,
      "loss": 1.1866,
      "step": 2735
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.984958897150914,
      "learning_rate": 1.9664555915424957e-05,
      "loss": 1.3443,
      "step": 2736
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.334276366194055,
      "learning_rate": 1.9664220769393127e-05,
      "loss": 1.5344,
      "step": 2737
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.557632553846091,
      "learning_rate": 1.9663885458879614e-05,
      "loss": 1.4431,
      "step": 2738
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.754010481727069,
      "learning_rate": 1.9663549983890117e-05,
      "loss": 1.286,
      "step": 2739
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.657492117621428,
      "learning_rate": 1.966321434443035e-05,
      "loss": 1.0604,
      "step": 2740
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.484174994932781,
      "learning_rate": 1.9662878540506026e-05,
      "loss": 1.5294,
      "step": 2741
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.462947855182102,
      "learning_rate": 1.966254257212286e-05,
      "loss": 1.5214,
      "step": 2742
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.748969895592326,
      "learning_rate": 1.9662206439286566e-05,
      "loss": 1.3349,
      "step": 2743
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.846406109520501,
      "learning_rate": 1.966187014200287e-05,
      "loss": 1.2443,
      "step": 2744
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.42855112727872,
      "learning_rate": 1.9661533680277495e-05,
      "loss": 1.2886,
      "step": 2745
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.489372436870935,
      "learning_rate": 1.9661197054116165e-05,
      "loss": 1.2051,
      "step": 2746
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.097993313979912,
      "learning_rate": 1.9660860263524612e-05,
      "loss": 1.344,
      "step": 2747
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.503484134715703,
      "learning_rate": 1.9660523308508565e-05,
      "loss": 1.552,
      "step": 2748
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.951035554240658,
      "learning_rate": 1.9660186189073763e-05,
      "loss": 1.6213,
      "step": 2749
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.213123144938951,
      "learning_rate": 1.965984890522594e-05,
      "loss": 1.4438,
      "step": 2750
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.347535284783067,
      "learning_rate": 1.9659511456970833e-05,
      "loss": 1.4343,
      "step": 2751
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.027988667183151,
      "learning_rate": 1.9659173844314194e-05,
      "loss": 1.7236,
      "step": 2752
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.989486344905474,
      "learning_rate": 1.9658836067261766e-05,
      "loss": 0.9638,
      "step": 2753
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.083279259266226,
      "learning_rate": 1.9658498125819293e-05,
      "loss": 1.3267,
      "step": 2754
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.775133600965877,
      "learning_rate": 1.9658160019992534e-05,
      "loss": 1.2873,
      "step": 2755
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.320594295275376,
      "learning_rate": 1.9657821749787237e-05,
      "loss": 1.2924,
      "step": 2756
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.572308131025181,
      "learning_rate": 1.9657483315209162e-05,
      "loss": 1.5642,
      "step": 2757
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.083294500059953,
      "learning_rate": 1.9657144716264073e-05,
      "loss": 1.5336,
      "step": 2758
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.28399374268687,
      "learning_rate": 1.9656805952957726e-05,
      "loss": 1.4155,
      "step": 2759
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.029630607930598,
      "learning_rate": 1.965646702529589e-05,
      "loss": 1.3834,
      "step": 2760
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.006328454238826,
      "learning_rate": 1.9656127933284334e-05,
      "loss": 1.6977,
      "step": 2761
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.858457426903794,
      "learning_rate": 1.9655788676928826e-05,
      "loss": 1.465,
      "step": 2762
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.285633985799881,
      "learning_rate": 1.9655449256235143e-05,
      "loss": 1.2721,
      "step": 2763
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.890785900426548,
      "learning_rate": 1.9655109671209062e-05,
      "loss": 1.0141,
      "step": 2764
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.104977486636603,
      "learning_rate": 1.965476992185636e-05,
      "loss": 1.3042,
      "step": 2765
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.987633889899133,
      "learning_rate": 1.9654430008182824e-05,
      "loss": 1.1932,
      "step": 2766
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.894923459308026,
      "learning_rate": 1.9654089930194234e-05,
      "loss": 1.3016,
      "step": 2767
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.821217418502508,
      "learning_rate": 1.965374968789638e-05,
      "loss": 1.4806,
      "step": 2768
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.702403826431151,
      "learning_rate": 1.9653409281295053e-05,
      "loss": 1.5811,
      "step": 2769
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.519920633596432,
      "learning_rate": 1.965306871039605e-05,
      "loss": 1.1832,
      "step": 2770
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.426337340831826,
      "learning_rate": 1.9652727975205163e-05,
      "loss": 1.5694,
      "step": 2771
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.456171605167301,
      "learning_rate": 1.965238707572819e-05,
      "loss": 1.2235,
      "step": 2772
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.6982703207900745,
      "learning_rate": 1.9652046011970937e-05,
      "loss": 1.2918,
      "step": 2773
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.294801423525001,
      "learning_rate": 1.9651704783939206e-05,
      "loss": 1.5028,
      "step": 2774
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.842042936543816,
      "learning_rate": 1.9651363391638804e-05,
      "loss": 1.0369,
      "step": 2775
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.80397930166349,
      "learning_rate": 1.9651021835075545e-05,
      "loss": 1.7982,
      "step": 2776
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.889976636647789,
      "learning_rate": 1.9650680114255244e-05,
      "loss": 1.1349,
      "step": 2777
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.26751993008654,
      "learning_rate": 1.9650338229183708e-05,
      "loss": 1.2002,
      "step": 2778
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.749358005633514,
      "learning_rate": 1.964999617986676e-05,
      "loss": 1.1048,
      "step": 2779
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.528745578930723,
      "learning_rate": 1.9649653966310227e-05,
      "loss": 1.3761,
      "step": 2780
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.225904791585423,
      "learning_rate": 1.9649311588519927e-05,
      "loss": 1.3215,
      "step": 2781
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4567084198471654,
      "learning_rate": 1.9648969046501688e-05,
      "loss": 1.1459,
      "step": 2782
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.729173500930345,
      "learning_rate": 1.9648626340261346e-05,
      "loss": 1.3909,
      "step": 2783
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.795136701138339,
      "learning_rate": 1.9648283469804723e-05,
      "loss": 1.3115,
      "step": 2784
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.2922720186001655,
      "learning_rate": 1.964794043513766e-05,
      "loss": 1.2719,
      "step": 2785
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.541398203476618,
      "learning_rate": 1.9647597236266e-05,
      "loss": 1.3593,
      "step": 2786
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.796024424037487,
      "learning_rate": 1.9647253873195578e-05,
      "loss": 1.3737,
      "step": 2787
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.717613274294006,
      "learning_rate": 1.9646910345932237e-05,
      "loss": 1.4179,
      "step": 2788
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.789135693081646,
      "learning_rate": 1.9646566654481833e-05,
      "loss": 1.0573,
      "step": 2789
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.8996782589804075,
      "learning_rate": 1.96462227988502e-05,
      "loss": 1.6745,
      "step": 2790
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.386067592439238,
      "learning_rate": 1.9645878779043203e-05,
      "loss": 1.5231,
      "step": 2791
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.52714251171493,
      "learning_rate": 1.9645534595066697e-05,
      "loss": 1.1773,
      "step": 2792
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.974401519825416,
      "learning_rate": 1.9645190246926532e-05,
      "loss": 1.4538,
      "step": 2793
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.425727593177253,
      "learning_rate": 1.9644845734628573e-05,
      "loss": 1.3373,
      "step": 2794
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.457637090283118,
      "learning_rate": 1.9644501058178686e-05,
      "loss": 1.1194,
      "step": 2795
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.261110202674557,
      "learning_rate": 1.964415621758273e-05,
      "loss": 1.2325,
      "step": 2796
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.727085281587915,
      "learning_rate": 1.9643811212846583e-05,
      "loss": 1.4017,
      "step": 2797
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.669661160780835,
      "learning_rate": 1.964346604397611e-05,
      "loss": 1.1224,
      "step": 2798
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.603343268310057,
      "learning_rate": 1.964312071097719e-05,
      "loss": 1.2152,
      "step": 2799
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.584081278612616,
      "learning_rate": 1.9642775213855696e-05,
      "loss": 1.4744,
      "step": 2800
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.861484195069706,
      "learning_rate": 1.964242955261751e-05,
      "loss": 1.0113,
      "step": 2801
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.140797000024259,
      "learning_rate": 1.9642083727268522e-05,
      "loss": 0.9178,
      "step": 2802
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.953665568602358,
      "learning_rate": 1.9641737737814608e-05,
      "loss": 1.3652,
      "step": 2803
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.727611758723014,
      "learning_rate": 1.964139158426166e-05,
      "loss": 1.2361,
      "step": 2804
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.022872001862229,
      "learning_rate": 1.964104526661557e-05,
      "loss": 1.3895,
      "step": 2805
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.760873124369996,
      "learning_rate": 1.9640698784882234e-05,
      "loss": 1.3244,
      "step": 2806
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.445761211332446,
      "learning_rate": 1.9640352139067544e-05,
      "loss": 1.4343,
      "step": 2807
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.959768970519632,
      "learning_rate": 1.9640005329177405e-05,
      "loss": 1.3001,
      "step": 2808
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.077627419149165,
      "learning_rate": 1.9639658355217716e-05,
      "loss": 1.4133,
      "step": 2809
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.489544276868499,
      "learning_rate": 1.9639311217194383e-05,
      "loss": 0.9325,
      "step": 2810
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.954737840251701,
      "learning_rate": 1.9638963915113317e-05,
      "loss": 1.1183,
      "step": 2811
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.510906637869646,
      "learning_rate": 1.9638616448980428e-05,
      "loss": 1.276,
      "step": 2812
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.620640085005474,
      "learning_rate": 1.9638268818801624e-05,
      "loss": 1.4719,
      "step": 2813
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.964273454987224,
      "learning_rate": 1.963792102458283e-05,
      "loss": 1.2706,
      "step": 2814
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.453076955862479,
      "learning_rate": 1.963757306632996e-05,
      "loss": 1.2797,
      "step": 2815
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.65631590131971,
      "learning_rate": 1.9637224944048938e-05,
      "loss": 1.2578,
      "step": 2816
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.949128000237121,
      "learning_rate": 1.9636876657745687e-05,
      "loss": 1.2016,
      "step": 2817
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.2880866058449465,
      "learning_rate": 1.963652820742614e-05,
      "loss": 1.219,
      "step": 2818
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.884923106688556,
      "learning_rate": 1.963617959309622e-05,
      "loss": 1.1964,
      "step": 2819
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.005678290210574,
      "learning_rate": 1.9635830814761864e-05,
      "loss": 1.2213,
      "step": 2820
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.737778248751051,
      "learning_rate": 1.963548187242901e-05,
      "loss": 1.7709,
      "step": 2821
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.557764276695044,
      "learning_rate": 1.9635132766103596e-05,
      "loss": 1.587,
      "step": 2822
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.990558305498386,
      "learning_rate": 1.963478349579156e-05,
      "loss": 1.196,
      "step": 2823
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.728168765473515,
      "learning_rate": 1.963443406149885e-05,
      "loss": 1.4557,
      "step": 2824
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.435177731692842,
      "learning_rate": 1.9634084463231414e-05,
      "loss": 1.5095,
      "step": 2825
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.19998539063743,
      "learning_rate": 1.96337347009952e-05,
      "loss": 1.2355,
      "step": 2826
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6980594723944615,
      "learning_rate": 1.963338477479616e-05,
      "loss": 1.0948,
      "step": 2827
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.665371280262307,
      "learning_rate": 1.9633034684640248e-05,
      "loss": 1.4163,
      "step": 2828
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.8403540953147495,
      "learning_rate": 1.9632684430533432e-05,
      "loss": 1.4926,
      "step": 2829
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.121091710661659,
      "learning_rate": 1.9632334012481662e-05,
      "loss": 1.576,
      "step": 2830
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.82688596177203,
      "learning_rate": 1.9631983430490907e-05,
      "loss": 1.3069,
      "step": 2831
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.257572257648184,
      "learning_rate": 1.9631632684567133e-05,
      "loss": 1.3457,
      "step": 2832
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.523104392929014,
      "learning_rate": 1.963128177471631e-05,
      "loss": 1.7196,
      "step": 2833
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.997030793879166,
      "learning_rate": 1.9630930700944413e-05,
      "loss": 1.5087,
      "step": 2834
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.302666638407603,
      "learning_rate": 1.9630579463257414e-05,
      "loss": 1.2299,
      "step": 2835
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.920095918709255,
      "learning_rate": 1.963022806166129e-05,
      "loss": 1.2553,
      "step": 2836
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.177563167127155,
      "learning_rate": 1.962987649616202e-05,
      "loss": 1.3203,
      "step": 2837
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.246860194724832,
      "learning_rate": 1.9629524766765593e-05,
      "loss": 1.4603,
      "step": 2838
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.7638675826899615,
      "learning_rate": 1.9629172873477995e-05,
      "loss": 1.2964,
      "step": 2839
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.418423443124514,
      "learning_rate": 1.9628820816305213e-05,
      "loss": 1.3539,
      "step": 2840
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.089433770382736,
      "learning_rate": 1.9628468595253233e-05,
      "loss": 1.5603,
      "step": 2841
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.918410746020326,
      "learning_rate": 1.9628116210328062e-05,
      "loss": 1.5613,
      "step": 2842
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.617093920342838,
      "learning_rate": 1.9627763661535687e-05,
      "loss": 1.7142,
      "step": 2843
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.836870930690821,
      "learning_rate": 1.9627410948882115e-05,
      "loss": 1.1748,
      "step": 2844
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.740540623794242,
      "learning_rate": 1.9627058072373347e-05,
      "loss": 1.3688,
      "step": 2845
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.78262268344502,
      "learning_rate": 1.9626705032015386e-05,
      "loss": 0.9966,
      "step": 2846
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5933736281000765,
      "learning_rate": 1.9626351827814242e-05,
      "loss": 1.1126,
      "step": 2847
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.504756426907731,
      "learning_rate": 1.962599845977593e-05,
      "loss": 1.1528,
      "step": 2848
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6605424783366542,
      "learning_rate": 1.9625644927906458e-05,
      "loss": 1.2022,
      "step": 2849
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.127310216186128,
      "learning_rate": 1.962529123221185e-05,
      "loss": 1.1252,
      "step": 2850
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.429840219216996,
      "learning_rate": 1.962493737269812e-05,
      "loss": 1.3465,
      "step": 2851
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.8463620064832575,
      "learning_rate": 1.962458334937129e-05,
      "loss": 1.2957,
      "step": 2852
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.386598681357348,
      "learning_rate": 1.9624229162237394e-05,
      "loss": 1.4593,
      "step": 2853
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.809225249662573,
      "learning_rate": 1.9623874811302454e-05,
      "loss": 1.3638,
      "step": 2854
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.045619293079117,
      "learning_rate": 1.9623520296572497e-05,
      "loss": 1.4809,
      "step": 2855
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.776247598722774,
      "learning_rate": 1.962316561805356e-05,
      "loss": 1.0748,
      "step": 2856
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.177123213498283,
      "learning_rate": 1.9622810775751688e-05,
      "loss": 1.4757,
      "step": 2857
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.21474537755649,
      "learning_rate": 1.9622455769672904e-05,
      "loss": 1.3392,
      "step": 2858
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.931093733180505,
      "learning_rate": 1.962210059982326e-05,
      "loss": 1.1643,
      "step": 2859
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.401431797382821,
      "learning_rate": 1.9621745266208803e-05,
      "loss": 1.0452,
      "step": 2860
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.876646164511465,
      "learning_rate": 1.962138976883558e-05,
      "loss": 1.1907,
      "step": 2861
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.6906775656598,
      "learning_rate": 1.9621034107709635e-05,
      "loss": 1.6123,
      "step": 2862
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.690848291450209,
      "learning_rate": 1.9620678282837023e-05,
      "loss": 1.1326,
      "step": 2863
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.6236062903143145,
      "learning_rate": 1.9620322294223805e-05,
      "loss": 1.2682,
      "step": 2864
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.315662759418646,
      "learning_rate": 1.9619966141876035e-05,
      "loss": 1.5175,
      "step": 2865
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.142642505500878,
      "learning_rate": 1.961960982579978e-05,
      "loss": 1.2871,
      "step": 2866
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.436541144391051,
      "learning_rate": 1.9619253346001096e-05,
      "loss": 1.2387,
      "step": 2867
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.105759302399907,
      "learning_rate": 1.9618896702486055e-05,
      "loss": 1.4664,
      "step": 2868
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.351610552585991,
      "learning_rate": 1.961853989526073e-05,
      "loss": 1.3793,
      "step": 2869
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.6163534310487275,
      "learning_rate": 1.9618182924331187e-05,
      "loss": 1.2818,
      "step": 2870
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.672578767145308,
      "learning_rate": 1.961782578970351e-05,
      "loss": 1.6266,
      "step": 2871
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.668347899213302,
      "learning_rate": 1.9617468491383768e-05,
      "loss": 1.4731,
      "step": 2872
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.42370501469077,
      "learning_rate": 1.961711102937805e-05,
      "loss": 1.4226,
      "step": 2873
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.414895713083783,
      "learning_rate": 1.9616753403692433e-05,
      "loss": 1.5084,
      "step": 2874
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.1980634292722065,
      "learning_rate": 1.961639561433301e-05,
      "loss": 1.4234,
      "step": 2875
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.571689709055699,
      "learning_rate": 1.9616037661305867e-05,
      "loss": 1.0961,
      "step": 2876
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.132075438621875,
      "learning_rate": 1.9615679544617094e-05,
      "loss": 1.1206,
      "step": 2877
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.3487393555883775,
      "learning_rate": 1.9615321264272795e-05,
      "loss": 1.2646,
      "step": 2878
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.324496386006089,
      "learning_rate": 1.9614962820279056e-05,
      "loss": 1.4227,
      "step": 2879
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.659609525772356,
      "learning_rate": 1.9614604212641985e-05,
      "loss": 1.298,
      "step": 2880
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.695250021756731,
      "learning_rate": 1.9614245441367682e-05,
      "loss": 1.4738,
      "step": 2881
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.6184096888809165,
      "learning_rate": 1.9613886506462253e-05,
      "loss": 1.1892,
      "step": 2882
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.837886466306968,
      "learning_rate": 1.9613527407931815e-05,
      "loss": 1.2968,
      "step": 2883
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.498531798787343,
      "learning_rate": 1.9613168145782468e-05,
      "loss": 1.1254,
      "step": 2884
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.498342773474677,
      "learning_rate": 1.9612808720020335e-05,
      "loss": 1.2909,
      "step": 2885
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.71947128340825,
      "learning_rate": 1.9612449130651528e-05,
      "loss": 1.3005,
      "step": 2886
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.550068955528291,
      "learning_rate": 1.961208937768217e-05,
      "loss": 1.3433,
      "step": 2887
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.117216278986792,
      "learning_rate": 1.9611729461118384e-05,
      "loss": 1.3155,
      "step": 2888
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.401941916760231,
      "learning_rate": 1.9611369380966297e-05,
      "loss": 1.2425,
      "step": 2889
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.26270200597905,
      "learning_rate": 1.961100913723203e-05,
      "loss": 1.8019,
      "step": 2890
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.76039505527485,
      "learning_rate": 1.961064872992172e-05,
      "loss": 1.5588,
      "step": 2891
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.29796211946602,
      "learning_rate": 1.9610288159041503e-05,
      "loss": 1.3669,
      "step": 2892
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.848291883336264,
      "learning_rate": 1.9609927424597515e-05,
      "loss": 1.6341,
      "step": 2893
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.2425399622976,
      "learning_rate": 1.960956652659589e-05,
      "loss": 1.4863,
      "step": 2894
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.127132957625464,
      "learning_rate": 1.9609205465042778e-05,
      "loss": 1.3095,
      "step": 2895
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.169493650840664,
      "learning_rate": 1.9608844239944317e-05,
      "loss": 1.5671,
      "step": 2896
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.693782543691986,
      "learning_rate": 1.960848285130666e-05,
      "loss": 1.3775,
      "step": 2897
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.329634978392567,
      "learning_rate": 1.9608121299135954e-05,
      "loss": 1.3524,
      "step": 2898
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.996100142190463,
      "learning_rate": 1.9607759583438357e-05,
      "loss": 1.2188,
      "step": 2899
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.38763252106279,
      "learning_rate": 1.960739770422002e-05,
      "loss": 1.2908,
      "step": 2900
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.0256171393472675,
      "learning_rate": 1.9607035661487107e-05,
      "loss": 1.3418,
      "step": 2901
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.807626817356524,
      "learning_rate": 1.9606673455245778e-05,
      "loss": 1.2549,
      "step": 2902
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.736051956146063,
      "learning_rate": 1.9606311085502198e-05,
      "loss": 1.3496,
      "step": 2903
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.5476951421386325,
      "learning_rate": 1.960594855226253e-05,
      "loss": 1.4083,
      "step": 2904
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.510502432609987,
      "learning_rate": 1.9605585855532953e-05,
      "loss": 1.1249,
      "step": 2905
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.55168109731261,
      "learning_rate": 1.960522299531963e-05,
      "loss": 1.2036,
      "step": 2906
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.544873416622423,
      "learning_rate": 1.9604859971628743e-05,
      "loss": 1.4241,
      "step": 2907
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.030756357373765,
      "learning_rate": 1.960449678446647e-05,
      "loss": 1.3429,
      "step": 2908
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.788832050066066,
      "learning_rate": 1.960413343383899e-05,
      "loss": 1.1346,
      "step": 2909
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.777799628065474,
      "learning_rate": 1.9603769919752493e-05,
      "loss": 1.1554,
      "step": 2910
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.969208262810722,
      "learning_rate": 1.9603406242213156e-05,
      "loss": 1.3351,
      "step": 2911
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.935278337666744,
      "learning_rate": 1.9603042401227178e-05,
      "loss": 1.4676,
      "step": 2912
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.891660418861789,
      "learning_rate": 1.9602678396800744e-05,
      "loss": 1.1546,
      "step": 2913
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.804670352643215,
      "learning_rate": 1.9602314228940055e-05,
      "loss": 1.2661,
      "step": 2914
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.365024282921082,
      "learning_rate": 1.9601949897651308e-05,
      "loss": 1.3712,
      "step": 2915
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.467231080517495,
      "learning_rate": 1.96015854029407e-05,
      "loss": 1.2419,
      "step": 2916
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.606533187938361,
      "learning_rate": 1.960122074481444e-05,
      "loss": 1.4522,
      "step": 2917
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.894385989418491,
      "learning_rate": 1.960085592327873e-05,
      "loss": 1.1086,
      "step": 2918
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.733907321614068,
      "learning_rate": 1.9600490938339778e-05,
      "loss": 1.2835,
      "step": 2919
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.664715080744172,
      "learning_rate": 1.9600125790003802e-05,
      "loss": 1.1556,
      "step": 2920
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.328742808705013,
      "learning_rate": 1.9599760478277014e-05,
      "loss": 1.5612,
      "step": 2921
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.004330361434522,
      "learning_rate": 1.959939500316563e-05,
      "loss": 1.2427,
      "step": 2922
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.083135901483623,
      "learning_rate": 1.959902936467587e-05,
      "loss": 1.5178,
      "step": 2923
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.745019161552831,
      "learning_rate": 1.9598663562813957e-05,
      "loss": 1.1685,
      "step": 2924
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.892650304796013,
      "learning_rate": 1.959829759758612e-05,
      "loss": 1.3835,
      "step": 2925
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.372157940105152,
      "learning_rate": 1.959793146899858e-05,
      "loss": 1.3857,
      "step": 2926
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.591501721097316,
      "learning_rate": 1.9597565177057582e-05,
      "loss": 1.2305,
      "step": 2927
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.615674859692715,
      "learning_rate": 1.9597198721769345e-05,
      "loss": 1.1188,
      "step": 2928
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.89080026928698,
      "learning_rate": 1.9596832103140118e-05,
      "loss": 1.1787,
      "step": 2929
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.58841971043338,
      "learning_rate": 1.9596465321176136e-05,
      "loss": 1.4848,
      "step": 2930
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.148700622030725,
      "learning_rate": 1.9596098375883635e-05,
      "loss": 1.188,
      "step": 2931
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.153032019131572,
      "learning_rate": 1.9595731267268874e-05,
      "loss": 1.5297,
      "step": 2932
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.863641315739823,
      "learning_rate": 1.959536399533809e-05,
      "loss": 1.3637,
      "step": 2933
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.283600782479322,
      "learning_rate": 1.9594996560097538e-05,
      "loss": 1.2388,
      "step": 2934
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.352890084338497,
      "learning_rate": 1.959462896155347e-05,
      "loss": 1.5276,
      "step": 2935
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.811615391789324,
      "learning_rate": 1.9594261199712147e-05,
      "loss": 1.4688,
      "step": 2936
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.578410856272615,
      "learning_rate": 1.959389327457982e-05,
      "loss": 1.0948,
      "step": 2937
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.313321012670502,
      "learning_rate": 1.9593525186162758e-05,
      "loss": 1.3573,
      "step": 2938
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.589468440784611,
      "learning_rate": 1.9593156934467222e-05,
      "loss": 1.3421,
      "step": 2939
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.910560080529263,
      "learning_rate": 1.959278851949948e-05,
      "loss": 1.2332,
      "step": 2940
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.876533685395539,
      "learning_rate": 1.9592419941265804e-05,
      "loss": 1.0218,
      "step": 2941
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.39355662596893,
      "learning_rate": 1.9592051199772467e-05,
      "loss": 1.6571,
      "step": 2942
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.026297884341463,
      "learning_rate": 1.959168229502574e-05,
      "loss": 1.2139,
      "step": 2943
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.812318755952964,
      "learning_rate": 1.9591313227031913e-05,
      "loss": 1.3714,
      "step": 2944
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.210162094672603,
      "learning_rate": 1.9590943995797252e-05,
      "loss": 1.5276,
      "step": 2945
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.187543343787898,
      "learning_rate": 1.9590574601328056e-05,
      "loss": 1.3987,
      "step": 2946
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.271347237694114,
      "learning_rate": 1.9590205043630604e-05,
      "loss": 1.2376,
      "step": 2947
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.898819521750856,
      "learning_rate": 1.9589835322711183e-05,
      "loss": 1.5076,
      "step": 2948
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.733115833031504,
      "learning_rate": 1.9589465438576093e-05,
      "loss": 1.08,
      "step": 2949
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.760782388702946,
      "learning_rate": 1.9589095391231625e-05,
      "loss": 1.4793,
      "step": 2950
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.416027700542925,
      "learning_rate": 1.9588725180684077e-05,
      "loss": 1.3132,
      "step": 2951
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5557127270216236,
      "learning_rate": 1.9588354806939752e-05,
      "loss": 1.1222,
      "step": 2952
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.04421466213256,
      "learning_rate": 1.958798427000495e-05,
      "loss": 1.0641,
      "step": 2953
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.094901227752529,
      "learning_rate": 1.9587613569885984e-05,
      "loss": 1.3609,
      "step": 2954
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.322958656263504,
      "learning_rate": 1.9587242706589155e-05,
      "loss": 1.4906,
      "step": 2955
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.343438138792018,
      "learning_rate": 1.958687168012078e-05,
      "loss": 1.0237,
      "step": 2956
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.626874726987884,
      "learning_rate": 1.9586500490487174e-05,
      "loss": 1.7888,
      "step": 2957
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.044770377191504,
      "learning_rate": 1.9586129137694653e-05,
      "loss": 1.6433,
      "step": 2958
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.902634948759568,
      "learning_rate": 1.9585757621749536e-05,
      "loss": 1.315,
      "step": 2959
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.059915875074615,
      "learning_rate": 1.958538594265815e-05,
      "loss": 1.0832,
      "step": 2960
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.640305931146878,
      "learning_rate": 1.9585014100426813e-05,
      "loss": 1.3399,
      "step": 2961
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.540717001933553,
      "learning_rate": 1.958464209506186e-05,
      "loss": 1.3087,
      "step": 2962
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.386719805919371,
      "learning_rate": 1.9584269926569626e-05,
      "loss": 1.3857,
      "step": 2963
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.231246768925852,
      "learning_rate": 1.958389759495644e-05,
      "loss": 1.3003,
      "step": 2964
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4351415414924475,
      "learning_rate": 1.9583525100228638e-05,
      "loss": 1.0146,
      "step": 2965
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.524253826156421,
      "learning_rate": 1.958315244239256e-05,
      "loss": 1.2228,
      "step": 2966
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.365544611681355,
      "learning_rate": 1.958277962145455e-05,
      "loss": 1.4578,
      "step": 2967
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.237198047052827,
      "learning_rate": 1.9582406637420954e-05,
      "loss": 1.1376,
      "step": 2968
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.6663691464538575,
      "learning_rate": 1.958203349029812e-05,
      "loss": 1.2138,
      "step": 2969
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.130814467633388,
      "learning_rate": 1.9581660180092394e-05,
      "loss": 1.293,
      "step": 2970
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4531606997283935,
      "learning_rate": 1.958128670681014e-05,
      "loss": 0.9759,
      "step": 2971
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.8434341914874315,
      "learning_rate": 1.9580913070457705e-05,
      "loss": 1.5946,
      "step": 2972
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.871851083499828,
      "learning_rate": 1.958053927104145e-05,
      "loss": 1.1846,
      "step": 2973
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.3896781752599425,
      "learning_rate": 1.9580165308567737e-05,
      "loss": 1.3184,
      "step": 2974
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.09958137414239,
      "learning_rate": 1.9579791183042933e-05,
      "loss": 1.2932,
      "step": 2975
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.693450824601202,
      "learning_rate": 1.9579416894473407e-05,
      "loss": 1.2035,
      "step": 2976
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.571185821447305,
      "learning_rate": 1.957904244286552e-05,
      "loss": 1.119,
      "step": 2977
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.232555225273163,
      "learning_rate": 1.9578667828225657e-05,
      "loss": 1.5448,
      "step": 2978
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.642243778391934,
      "learning_rate": 1.9578293050560185e-05,
      "loss": 1.3442,
      "step": 2979
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.875077957848286,
      "learning_rate": 1.957791810987549e-05,
      "loss": 1.1479,
      "step": 2980
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.327984929629464,
      "learning_rate": 1.9577543006177948e-05,
      "loss": 1.5649,
      "step": 2981
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.2493053310253535,
      "learning_rate": 1.9577167739473945e-05,
      "loss": 1.0351,
      "step": 2982
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.690070603375901,
      "learning_rate": 1.9576792309769863e-05,
      "loss": 1.5636,
      "step": 2983
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.472415118386781,
      "learning_rate": 1.95764167170721e-05,
      "loss": 1.2785,
      "step": 2984
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.7858990556448,
      "learning_rate": 1.9576040961387043e-05,
      "loss": 1.194,
      "step": 2985
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.788544842937453,
      "learning_rate": 1.957566504272109e-05,
      "loss": 1.0681,
      "step": 2986
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.593798199226103,
      "learning_rate": 1.9575288961080637e-05,
      "loss": 1.1911,
      "step": 2987
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.671293214817249,
      "learning_rate": 1.9574912716472086e-05,
      "loss": 1.2671,
      "step": 2988
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.417457574391445,
      "learning_rate": 1.957453630890184e-05,
      "loss": 1.5844,
      "step": 2989
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.007017462945115,
      "learning_rate": 1.9574159738376304e-05,
      "loss": 1.3579,
      "step": 2990
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.57823210297244,
      "learning_rate": 1.957378300490189e-05,
      "loss": 0.9591,
      "step": 2991
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.3944908180625015,
      "learning_rate": 1.9573406108485006e-05,
      "loss": 1.4853,
      "step": 2992
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.631522651851654,
      "learning_rate": 1.957302904913207e-05,
      "loss": 1.1462,
      "step": 2993
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.088771004797125,
      "learning_rate": 1.95726518268495e-05,
      "loss": 1.5986,
      "step": 2994
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.3358824000871286,
      "learning_rate": 1.9572274441643715e-05,
      "loss": 1.3787,
      "step": 2995
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.8932885477376535,
      "learning_rate": 1.9571896893521135e-05,
      "loss": 1.6467,
      "step": 2996
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.234227830629614,
      "learning_rate": 1.957151918248819e-05,
      "loss": 1.1829,
      "step": 2997
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.908963945306555,
      "learning_rate": 1.9571141308551306e-05,
      "loss": 1.5723,
      "step": 2998
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.791431990796311,
      "learning_rate": 1.957076327171692e-05,
      "loss": 1.1533,
      "step": 2999
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.0185539576857785,
      "learning_rate": 1.9570385071991455e-05,
      "loss": 1.3065,
      "step": 3000
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.530716919064747,
      "learning_rate": 1.9570006709381352e-05,
      "loss": 1.6475,
      "step": 3001
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.266439417205829,
      "learning_rate": 1.9569628183893058e-05,
      "loss": 1.5153,
      "step": 3002
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.4523597666476675,
      "learning_rate": 1.956924949553301e-05,
      "loss": 1.119,
      "step": 3003
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.670794335594339,
      "learning_rate": 1.956887064430765e-05,
      "loss": 1.4032,
      "step": 3004
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.752957253056093,
      "learning_rate": 1.956849163022343e-05,
      "loss": 1.1935,
      "step": 3005
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.085954711966279,
      "learning_rate": 1.95681124532868e-05,
      "loss": 0.9464,
      "step": 3006
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.384637233585121,
      "learning_rate": 1.9567733113504213e-05,
      "loss": 1.1936,
      "step": 3007
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.646012852902901,
      "learning_rate": 1.9567353610882126e-05,
      "loss": 1.0686,
      "step": 3008
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.194704576514314,
      "learning_rate": 1.9566973945427e-05,
      "loss": 1.7048,
      "step": 3009
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.976242487833728,
      "learning_rate": 1.956659411714529e-05,
      "loss": 1.4069,
      "step": 3010
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.577012794802915,
      "learning_rate": 1.9566214126043467e-05,
      "loss": 1.8297,
      "step": 3011
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.446067739588253,
      "learning_rate": 1.9565833972127996e-05,
      "loss": 1.0778,
      "step": 3012
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.741439398314465,
      "learning_rate": 1.9565453655405342e-05,
      "loss": 1.2241,
      "step": 3013
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.3475571467680005,
      "learning_rate": 1.9565073175881986e-05,
      "loss": 1.2833,
      "step": 3014
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.955427454994196,
      "learning_rate": 1.9564692533564404e-05,
      "loss": 1.2292,
      "step": 3015
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.64942812661909,
      "learning_rate": 1.956431172845907e-05,
      "loss": 1.3799,
      "step": 3016
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.163095006138523,
      "learning_rate": 1.9563930760572462e-05,
      "loss": 1.5989,
      "step": 3017
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.540869041804654,
      "learning_rate": 1.956354962991107e-05,
      "loss": 1.4673,
      "step": 3018
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.010788689807262,
      "learning_rate": 1.9563168336481384e-05,
      "loss": 1.3297,
      "step": 3019
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.876265498028198,
      "learning_rate": 1.9562786880289883e-05,
      "loss": 1.4176,
      "step": 3020
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.575270548171447,
      "learning_rate": 1.9562405261343066e-05,
      "loss": 1.514,
      "step": 3021
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.536256304518745,
      "learning_rate": 1.956202347964743e-05,
      "loss": 1.2446,
      "step": 3022
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.428698878852939,
      "learning_rate": 1.956164153520946e-05,
      "loss": 1.416,
      "step": 3023
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.041139765511129,
      "learning_rate": 1.9561259428035675e-05,
      "loss": 1.4111,
      "step": 3024
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.461622632040706,
      "learning_rate": 1.9560877158132565e-05,
      "loss": 1.03,
      "step": 3025
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.186863638675823,
      "learning_rate": 1.9560494725506644e-05,
      "loss": 1.4132,
      "step": 3026
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.82533876966083,
      "learning_rate": 1.9560112130164412e-05,
      "loss": 1.4313,
      "step": 3027
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.10960376436543,
      "learning_rate": 1.955972937211239e-05,
      "loss": 1.3614,
      "step": 3028
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.41487829928482,
      "learning_rate": 1.9559346451357085e-05,
      "loss": 1.5139,
      "step": 3029
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.3565909930691324,
      "learning_rate": 1.955896336790502e-05,
      "loss": 1.1268,
      "step": 3030
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.005849207104483,
      "learning_rate": 1.9558580121762712e-05,
      "loss": 1.108,
      "step": 3031
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.296166926016051,
      "learning_rate": 1.9558196712936682e-05,
      "loss": 1.3326,
      "step": 3032
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.674354796676517,
      "learning_rate": 1.9557813141433462e-05,
      "loss": 1.2763,
      "step": 3033
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.994637448520102,
      "learning_rate": 1.9557429407259573e-05,
      "loss": 1.7612,
      "step": 3034
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.122383604544614,
      "learning_rate": 1.955704551042155e-05,
      "loss": 1.3702,
      "step": 3035
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.4407046974262006,
      "learning_rate": 1.9556661450925923e-05,
      "loss": 1.727,
      "step": 3036
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.7153045348909615,
      "learning_rate": 1.9556277228779237e-05,
      "loss": 1.3907,
      "step": 3037
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.944929222233087,
      "learning_rate": 1.955589284398802e-05,
      "loss": 1.1955,
      "step": 3038
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.311154124956176,
      "learning_rate": 1.955550829655882e-05,
      "loss": 1.4433,
      "step": 3039
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.471146382141527,
      "learning_rate": 1.9555123586498187e-05,
      "loss": 1.4908,
      "step": 3040
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.865120433965442,
      "learning_rate": 1.9554738713812657e-05,
      "loss": 1.0397,
      "step": 3041
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.352302617216924,
      "learning_rate": 1.955435367850879e-05,
      "loss": 1.4875,
      "step": 3042
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.67850082155871,
      "learning_rate": 1.9553968480593133e-05,
      "loss": 1.4699,
      "step": 3043
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.225597444846306,
      "learning_rate": 1.955358312007225e-05,
      "loss": 1.2078,
      "step": 3044
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.447491290475621,
      "learning_rate": 1.955319759695269e-05,
      "loss": 1.3521,
      "step": 3045
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.273068525614477,
      "learning_rate": 1.955281191124102e-05,
      "loss": 1.3991,
      "step": 3046
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.892519769589509,
      "learning_rate": 1.95524260629438e-05,
      "loss": 1.3323,
      "step": 3047
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.6377192759838755,
      "learning_rate": 1.9552040052067604e-05,
      "loss": 1.4178,
      "step": 3048
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.453432864443718,
      "learning_rate": 1.9551653878618998e-05,
      "loss": 1.5771,
      "step": 3049
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.555419331136935,
      "learning_rate": 1.955126754260455e-05,
      "loss": 1.063,
      "step": 3050
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.391439764954109,
      "learning_rate": 1.9550881044030844e-05,
      "loss": 1.2256,
      "step": 3051
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.717759878989044,
      "learning_rate": 1.955049438290445e-05,
      "loss": 1.2581,
      "step": 3052
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.567993637139406,
      "learning_rate": 1.955010755923196e-05,
      "loss": 1.6009,
      "step": 3053
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.191616474726164,
      "learning_rate": 1.9549720573019945e-05,
      "loss": 1.343,
      "step": 3054
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.851952047077054,
      "learning_rate": 1.9549333424274998e-05,
      "loss": 1.4578,
      "step": 3055
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.236155484130217,
      "learning_rate": 1.9548946113003703e-05,
      "loss": 0.8406,
      "step": 3056
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.605833266726817,
      "learning_rate": 1.954855863921266e-05,
      "loss": 1.7349,
      "step": 3057
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.341293251682161,
      "learning_rate": 1.9548171002908458e-05,
      "loss": 1.2011,
      "step": 3058
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.621043118896074,
      "learning_rate": 1.9547783204097694e-05,
      "loss": 1.0242,
      "step": 3059
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.987528890735899,
      "learning_rate": 1.9547395242786973e-05,
      "loss": 1.2363,
      "step": 3060
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.856406915139954,
      "learning_rate": 1.954700711898289e-05,
      "loss": 1.3745,
      "step": 3061
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.038433718282434,
      "learning_rate": 1.9546618832692062e-05,
      "loss": 1.6385,
      "step": 3062
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.881645646146623,
      "learning_rate": 1.954623038392109e-05,
      "loss": 1.4833,
      "step": 3063
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.525638146966041,
      "learning_rate": 1.954584177267658e-05,
      "loss": 1.1032,
      "step": 3064
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.575045041497217,
      "learning_rate": 1.954545299896516e-05,
      "loss": 1.178,
      "step": 3065
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.807885969302533,
      "learning_rate": 1.9545064062793434e-05,
      "loss": 1.3116,
      "step": 3066
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.393033748070535,
      "learning_rate": 1.954467496416803e-05,
      "loss": 1.3919,
      "step": 3067
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.093672654520014,
      "learning_rate": 1.9544285703095565e-05,
      "loss": 1.4815,
      "step": 3068
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.779054658204869,
      "learning_rate": 1.954389627958267e-05,
      "loss": 1.4736,
      "step": 3069
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.359269409318602,
      "learning_rate": 1.9543506693635965e-05,
      "loss": 1.1374,
      "step": 3070
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.058726471822921,
      "learning_rate": 1.9543116945262087e-05,
      "loss": 1.1055,
      "step": 3071
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.465982245119378,
      "learning_rate": 1.9542727034467667e-05,
      "loss": 1.0212,
      "step": 3072
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.707224513120302,
      "learning_rate": 1.9542336961259338e-05,
      "loss": 1.5051,
      "step": 3073
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.97382912673923,
      "learning_rate": 1.9541946725643747e-05,
      "loss": 1.3319,
      "step": 3074
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.948759052204712,
      "learning_rate": 1.9541556327627525e-05,
      "loss": 1.0344,
      "step": 3075
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.407563038946613,
      "learning_rate": 1.9541165767217327e-05,
      "loss": 1.2038,
      "step": 3076
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.394440742720365,
      "learning_rate": 1.9540775044419793e-05,
      "loss": 1.2591,
      "step": 3077
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.8547254876401995,
      "learning_rate": 1.9540384159241577e-05,
      "loss": 1.1788,
      "step": 3078
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.782167733510908,
      "learning_rate": 1.953999311168933e-05,
      "loss": 1.3366,
      "step": 3079
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.721799313264839,
      "learning_rate": 1.9539601901769708e-05,
      "loss": 1.1487,
      "step": 3080
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.662441523037463,
      "learning_rate": 1.9539210529489367e-05,
      "loss": 1.2138,
      "step": 3081
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.754428460318666,
      "learning_rate": 1.953881899485497e-05,
      "loss": 1.4099,
      "step": 3082
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.3788485642736,
      "learning_rate": 1.9538427297873183e-05,
      "loss": 1.4588,
      "step": 3083
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.655618133586845,
      "learning_rate": 1.9538035438550665e-05,
      "loss": 1.1919,
      "step": 3084
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.641410299590693,
      "learning_rate": 1.9537643416894097e-05,
      "loss": 1.2265,
      "step": 3085
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.844856901940121,
      "learning_rate": 1.9537251232910143e-05,
      "loss": 1.4873,
      "step": 3086
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.396194844009791,
      "learning_rate": 1.9536858886605477e-05,
      "loss": 1.0394,
      "step": 3087
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.511212779095034,
      "learning_rate": 1.953646637798678e-05,
      "loss": 1.4258,
      "step": 3088
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.168810476323793,
      "learning_rate": 1.9536073707060733e-05,
      "loss": 1.5882,
      "step": 3089
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.322385518885021,
      "learning_rate": 1.9535680873834017e-05,
      "loss": 1.681,
      "step": 3090
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.533059218061694,
      "learning_rate": 1.9535287878313315e-05,
      "loss": 1.3936,
      "step": 3091
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.758483617366037,
      "learning_rate": 1.9534894720505326e-05,
      "loss": 1.0865,
      "step": 3092
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.109722057695535,
      "learning_rate": 1.9534501400416727e-05,
      "loss": 1.2768,
      "step": 3093
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.087739345191718,
      "learning_rate": 1.9534107918054223e-05,
      "loss": 1.4132,
      "step": 3094
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.968139949887867,
      "learning_rate": 1.9533714273424506e-05,
      "loss": 1.4629,
      "step": 3095
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.454359057454755,
      "learning_rate": 1.953332046653428e-05,
      "loss": 1.352,
      "step": 3096
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.432247309200816,
      "learning_rate": 1.9532926497390244e-05,
      "loss": 1.2346,
      "step": 3097
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.710200808342539,
      "learning_rate": 1.9532532365999102e-05,
      "loss": 1.3329,
      "step": 3098
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.751932441236711,
      "learning_rate": 1.9532138072367567e-05,
      "loss": 1.2681,
      "step": 3099
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.448049972556608,
      "learning_rate": 1.9531743616502345e-05,
      "loss": 1.6228,
      "step": 3100
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.280045510637844,
      "learning_rate": 1.953134899841015e-05,
      "loss": 1.6846,
      "step": 3101
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.6073916841500395,
      "learning_rate": 1.9530954218097703e-05,
      "loss": 1.2591,
      "step": 3102
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7326692307282205,
      "learning_rate": 1.9530559275571714e-05,
      "loss": 1.1636,
      "step": 3103
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.246469503453881,
      "learning_rate": 1.9530164170838915e-05,
      "loss": 1.2267,
      "step": 3104
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.696156586714013,
      "learning_rate": 1.9529768903906022e-05,
      "loss": 1.1701,
      "step": 3105
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.759558738368552,
      "learning_rate": 1.952937347477977e-05,
      "loss": 1.2415,
      "step": 3106
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.183835929593885,
      "learning_rate": 1.9528977883466883e-05,
      "loss": 1.3643,
      "step": 3107
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.390243423847495,
      "learning_rate": 1.9528582129974098e-05,
      "loss": 1.3506,
      "step": 3108
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.33701614284104,
      "learning_rate": 1.9528186214308144e-05,
      "loss": 1.152,
      "step": 3109
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.417570859709311,
      "learning_rate": 1.9527790136475767e-05,
      "loss": 1.5009,
      "step": 3110
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.139641464746353,
      "learning_rate": 1.9527393896483703e-05,
      "loss": 1.264,
      "step": 3111
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.61139589048605,
      "learning_rate": 1.95269974943387e-05,
      "loss": 1.4734,
      "step": 3112
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.23824752848678,
      "learning_rate": 1.9526600930047503e-05,
      "loss": 0.9956,
      "step": 3113
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.807227945536994,
      "learning_rate": 1.952620420361686e-05,
      "loss": 1.3152,
      "step": 3114
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.106900305471056,
      "learning_rate": 1.9525807315053523e-05,
      "loss": 1.2571,
      "step": 3115
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.388020990542762,
      "learning_rate": 1.952541026436425e-05,
      "loss": 1.3577,
      "step": 3116
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.213904734356749,
      "learning_rate": 1.9525013051555794e-05,
      "loss": 1.222,
      "step": 3117
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.489748683967728,
      "learning_rate": 1.9524615676634923e-05,
      "loss": 1.2105,
      "step": 3118
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.302755564249802,
      "learning_rate": 1.952421813960839e-05,
      "loss": 1.1385,
      "step": 3119
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.1986965276448,
      "learning_rate": 1.952382044048297e-05,
      "loss": 1.6333,
      "step": 3120
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.746355989314025,
      "learning_rate": 1.9523422579265426e-05,
      "loss": 1.1581,
      "step": 3121
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.056781487093378,
      "learning_rate": 1.952302455596253e-05,
      "loss": 1.4657,
      "step": 3122
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.094425361183786,
      "learning_rate": 1.9522626370581058e-05,
      "loss": 1.2324,
      "step": 3123
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.347728731077913,
      "learning_rate": 1.9522228023127788e-05,
      "loss": 1.2746,
      "step": 3124
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.4641902310283115,
      "learning_rate": 1.95218295136095e-05,
      "loss": 1.2687,
      "step": 3125
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.461242472567538,
      "learning_rate": 1.952143084203297e-05,
      "loss": 1.6961,
      "step": 3126
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.147368545218562,
      "learning_rate": 1.9521032008404992e-05,
      "loss": 1.008,
      "step": 3127
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.293863915920318,
      "learning_rate": 1.952063301273235e-05,
      "loss": 1.3189,
      "step": 3128
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.07060437331083,
      "learning_rate": 1.952023385502183e-05,
      "loss": 1.6048,
      "step": 3129
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.447779137593787,
      "learning_rate": 1.9519834535280237e-05,
      "loss": 1.0686,
      "step": 3130
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.293321291177359,
      "learning_rate": 1.9519435053514356e-05,
      "loss": 1.4016,
      "step": 3131
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.160184887594016,
      "learning_rate": 1.9519035409730992e-05,
      "loss": 1.3989,
      "step": 3132
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.029311753194242,
      "learning_rate": 1.9518635603936943e-05,
      "loss": 1.3441,
      "step": 3133
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.200473158804808,
      "learning_rate": 1.951823563613902e-05,
      "loss": 1.2821,
      "step": 3134
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.589873426352561,
      "learning_rate": 1.9517835506344025e-05,
      "loss": 1.5579,
      "step": 3135
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.674041607977303,
      "learning_rate": 1.9517435214558768e-05,
      "loss": 1.1926,
      "step": 3136
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.073125887395572,
      "learning_rate": 1.9517034760790064e-05,
      "loss": 1.1299,
      "step": 3137
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.596435977833834,
      "learning_rate": 1.9516634145044728e-05,
      "loss": 1.3055,
      "step": 3138
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.233413033341089,
      "learning_rate": 1.9516233367329578e-05,
      "loss": 1.303,
      "step": 3139
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.823768848641338,
      "learning_rate": 1.9515832427651436e-05,
      "loss": 1.3427,
      "step": 3140
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.870714170790582,
      "learning_rate": 1.9515431326017124e-05,
      "loss": 1.4737,
      "step": 3141
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.839561530875974,
      "learning_rate": 1.951503006243347e-05,
      "loss": 1.1626,
      "step": 3142
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.634065462722403,
      "learning_rate": 1.9514628636907303e-05,
      "loss": 1.1391,
      "step": 3143
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.0998195375773525,
      "learning_rate": 1.9514227049445454e-05,
      "loss": 1.8265,
      "step": 3144
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.525217785291463,
      "learning_rate": 1.951382530005476e-05,
      "loss": 1.4533,
      "step": 3145
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.779742753850481,
      "learning_rate": 1.9513423388742063e-05,
      "loss": 1.0843,
      "step": 3146
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.911795214129019,
      "learning_rate": 1.9513021315514188e-05,
      "loss": 1.8046,
      "step": 3147
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.535562190103548,
      "learning_rate": 1.9512619080377994e-05,
      "loss": 1.1025,
      "step": 3148
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.438452427184618,
      "learning_rate": 1.951221668334032e-05,
      "loss": 1.4489,
      "step": 3149
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.826261881253335,
      "learning_rate": 1.951181412440802e-05,
      "loss": 1.3695,
      "step": 3150
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.626958487669006,
      "learning_rate": 1.9511411403587937e-05,
      "loss": 1.1484,
      "step": 3151
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.474240700334718,
      "learning_rate": 1.9511008520886928e-05,
      "loss": 1.6691,
      "step": 3152
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.9715911382301385,
      "learning_rate": 1.9510605476311854e-05,
      "loss": 1.129,
      "step": 3153
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.288235779622477,
      "learning_rate": 1.9510202269869573e-05,
      "loss": 1.7531,
      "step": 3154
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.347282279406751,
      "learning_rate": 1.9509798901566943e-05,
      "loss": 1.352,
      "step": 3155
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.290169032711555,
      "learning_rate": 1.950939537141083e-05,
      "loss": 0.9036,
      "step": 3156
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.119428081977788,
      "learning_rate": 1.9508991679408113e-05,
      "loss": 1.1192,
      "step": 3157
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.832228268612211,
      "learning_rate": 1.950858782556565e-05,
      "loss": 1.2977,
      "step": 3158
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.2643133361941175,
      "learning_rate": 1.950818380989032e-05,
      "loss": 1.2638,
      "step": 3159
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.394257114093725,
      "learning_rate": 1.9507779632388997e-05,
      "loss": 1.41,
      "step": 3160
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.351244974840778,
      "learning_rate": 1.950737529306856e-05,
      "loss": 1.479,
      "step": 3161
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.2873959834810025,
      "learning_rate": 1.950697079193589e-05,
      "loss": 1.3747,
      "step": 3162
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.5091983518889664,
      "learning_rate": 1.9506566128997877e-05,
      "loss": 1.4221,
      "step": 3163
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.114140882061283,
      "learning_rate": 1.9506161304261404e-05,
      "loss": 1.334,
      "step": 3164
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.278889627160714,
      "learning_rate": 1.9505756317733356e-05,
      "loss": 1.2872,
      "step": 3165
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.590371198434953,
      "learning_rate": 1.9505351169420637e-05,
      "loss": 1.4958,
      "step": 3166
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.280514788857349,
      "learning_rate": 1.9504945859330134e-05,
      "loss": 1.2974,
      "step": 3167
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.755227227450106,
      "learning_rate": 1.950454038746875e-05,
      "loss": 1.1727,
      "step": 3168
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.470652611003924,
      "learning_rate": 1.950413475384338e-05,
      "loss": 1.2225,
      "step": 3169
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.980494039062044,
      "learning_rate": 1.9503728958460932e-05,
      "loss": 1.4493,
      "step": 3170
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.497028765723917,
      "learning_rate": 1.9503323001328313e-05,
      "loss": 1.4733,
      "step": 3171
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.438736381098128,
      "learning_rate": 1.9502916882452433e-05,
      "loss": 1.4552,
      "step": 3172
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.014814383555338,
      "learning_rate": 1.9502510601840198e-05,
      "loss": 1.5364,
      "step": 3173
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.11365221813519,
      "learning_rate": 1.9502104159498532e-05,
      "loss": 1.4995,
      "step": 3174
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.448499025644143,
      "learning_rate": 1.9501697555434342e-05,
      "loss": 1.4112,
      "step": 3175
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.556659212265499,
      "learning_rate": 1.9501290789654558e-05,
      "loss": 1.4222,
      "step": 3176
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.044358101204118,
      "learning_rate": 1.9500883862166097e-05,
      "loss": 1.36,
      "step": 3177
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.152249167669299,
      "learning_rate": 1.9500476772975886e-05,
      "loss": 1.455,
      "step": 3178
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.174292309946439,
      "learning_rate": 1.9500069522090853e-05,
      "loss": 1.4961,
      "step": 3179
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.419515867315315,
      "learning_rate": 1.9499662109517933e-05,
      "loss": 1.5951,
      "step": 3180
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.848041107085185,
      "learning_rate": 1.9499254535264055e-05,
      "loss": 1.1112,
      "step": 3181
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.811772059718595,
      "learning_rate": 1.9498846799336162e-05,
      "loss": 1.2166,
      "step": 3182
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.164532820496349,
      "learning_rate": 1.9498438901741186e-05,
      "loss": 1.3728,
      "step": 3183
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.750138822888145,
      "learning_rate": 1.9498030842486072e-05,
      "loss": 1.3979,
      "step": 3184
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.22778534124478,
      "learning_rate": 1.949762262157777e-05,
      "loss": 1.1116,
      "step": 3185
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.297112288792508,
      "learning_rate": 1.949721423902322e-05,
      "loss": 1.2778,
      "step": 3186
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.692693435108846,
      "learning_rate": 1.9496805694829376e-05,
      "loss": 1.4646,
      "step": 3187
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.757566110643624,
      "learning_rate": 1.9496396989003195e-05,
      "loss": 1.7272,
      "step": 3188
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.120495598479416,
      "learning_rate": 1.9495988121551628e-05,
      "loss": 1.417,
      "step": 3189
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.268010284859559,
      "learning_rate": 1.9495579092481635e-05,
      "loss": 1.1349,
      "step": 3190
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.966486651588832,
      "learning_rate": 1.9495169901800178e-05,
      "loss": 1.1294,
      "step": 3191
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.198491498076987,
      "learning_rate": 1.949476054951422e-05,
      "loss": 1.6101,
      "step": 3192
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.974929077504478,
      "learning_rate": 1.949435103563073e-05,
      "loss": 1.2375,
      "step": 3193
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.851274058449346,
      "learning_rate": 1.9493941360156676e-05,
      "loss": 0.9031,
      "step": 3194
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.085427871696805,
      "learning_rate": 1.949353152309903e-05,
      "loss": 1.3395,
      "step": 3195
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.270046507465551,
      "learning_rate": 1.949312152446477e-05,
      "loss": 1.4827,
      "step": 3196
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.79421732536088,
      "learning_rate": 1.949271136426088e-05,
      "loss": 1.2993,
      "step": 3197
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.268235799217651,
      "learning_rate": 1.9492301042494324e-05,
      "loss": 1.2778,
      "step": 3198
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.914332370010181,
      "learning_rate": 1.9491890559172098e-05,
      "loss": 1.4354,
      "step": 3199
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.3276898572050975,
      "learning_rate": 1.9491479914301184e-05,
      "loss": 1.3066,
      "step": 3200
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.109008374133051,
      "learning_rate": 1.9491069107888578e-05,
      "loss": 1.2693,
      "step": 3201
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.047042053394044,
      "learning_rate": 1.9490658139941262e-05,
      "loss": 1.5876,
      "step": 3202
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.365138874510876,
      "learning_rate": 1.9490247010466234e-05,
      "loss": 1.2326,
      "step": 3203
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.532562104234933,
      "learning_rate": 1.9489835719470494e-05,
      "loss": 1.1141,
      "step": 3204
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.788554536756972,
      "learning_rate": 1.9489424266961037e-05,
      "loss": 1.1927,
      "step": 3205
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.611499604994943,
      "learning_rate": 1.9489012652944874e-05,
      "loss": 1.5178,
      "step": 3206
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.438750031904858,
      "learning_rate": 1.9488600877429e-05,
      "loss": 1.1693,
      "step": 3207
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.2529604867609,
      "learning_rate": 1.9488188940420432e-05,
      "loss": 1.3445,
      "step": 3208
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.315975839745974,
      "learning_rate": 1.9487776841926176e-05,
      "loss": 1.436,
      "step": 3209
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.012619624850473,
      "learning_rate": 1.948736458195325e-05,
      "loss": 1.2594,
      "step": 3210
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.890717702991227,
      "learning_rate": 1.9486952160508665e-05,
      "loss": 1.5454,
      "step": 3211
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.129146337493982,
      "learning_rate": 1.9486539577599445e-05,
      "loss": 1.2022,
      "step": 3212
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.456061686636017,
      "learning_rate": 1.948612683323261e-05,
      "loss": 1.6361,
      "step": 3213
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.835789233262676,
      "learning_rate": 1.9485713927415186e-05,
      "loss": 1.1138,
      "step": 3214
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.136426438011736,
      "learning_rate": 1.9485300860154194e-05,
      "loss": 1.3093,
      "step": 3215
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.327201137961394,
      "learning_rate": 1.9484887631456674e-05,
      "loss": 1.5607,
      "step": 3216
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.837548562590902,
      "learning_rate": 1.9484474241329656e-05,
      "loss": 1.318,
      "step": 3217
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.352309975911152,
      "learning_rate": 1.9484060689780172e-05,
      "loss": 1.466,
      "step": 3218
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.735711007284067,
      "learning_rate": 1.9483646976815264e-05,
      "loss": 1.0882,
      "step": 3219
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.256733010334577,
      "learning_rate": 1.9483233102441974e-05,
      "loss": 1.5666,
      "step": 3220
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.028540282110542,
      "learning_rate": 1.9482819066667342e-05,
      "loss": 1.4483,
      "step": 3221
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.504197065860445,
      "learning_rate": 1.948240486949842e-05,
      "loss": 1.4249,
      "step": 3222
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.548819065160275,
      "learning_rate": 1.948199051094225e-05,
      "loss": 1.3833,
      "step": 3223
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.2136892827810515,
      "learning_rate": 1.948157599100589e-05,
      "loss": 1.2258,
      "step": 3224
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.4880970593812055,
      "learning_rate": 1.9481161309696393e-05,
      "loss": 1.2484,
      "step": 3225
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.671116439321732,
      "learning_rate": 1.9480746467020825e-05,
      "loss": 1.4631,
      "step": 3226
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.296735583885339,
      "learning_rate": 1.948033146298623e-05,
      "loss": 1.1563,
      "step": 3227
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.016455086786798,
      "learning_rate": 1.9479916297599683e-05,
      "loss": 1.2724,
      "step": 3228
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.741368219442508,
      "learning_rate": 1.947950097086825e-05,
      "loss": 1.3168,
      "step": 3229
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.942241250653897,
      "learning_rate": 1.9479085482798995e-05,
      "loss": 1.2419,
      "step": 3230
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.233880497656015,
      "learning_rate": 1.947866983339899e-05,
      "loss": 1.0405,
      "step": 3231
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.686138213576816,
      "learning_rate": 1.9478254022675313e-05,
      "loss": 1.3642,
      "step": 3232
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.198563422292533,
      "learning_rate": 1.9477838050635032e-05,
      "loss": 1.4815,
      "step": 3233
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.8195837294042,
      "learning_rate": 1.947742191728524e-05,
      "loss": 1.2152,
      "step": 3234
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.919356606267648,
      "learning_rate": 1.9477005622633012e-05,
      "loss": 1.2265,
      "step": 3235
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.538992435059917,
      "learning_rate": 1.947658916668543e-05,
      "loss": 1.2607,
      "step": 3236
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.242175342325214,
      "learning_rate": 1.9476172549449593e-05,
      "loss": 1.2295,
      "step": 3237
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.36514967763002,
      "learning_rate": 1.9475755770932578e-05,
      "loss": 1.3858,
      "step": 3238
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.423527561574612,
      "learning_rate": 1.9475338831141485e-05,
      "loss": 1.3123,
      "step": 3239
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.638508019465149,
      "learning_rate": 1.9474921730083413e-05,
      "loss": 1.504,
      "step": 3240
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.943571819457716,
      "learning_rate": 1.9474504467765457e-05,
      "loss": 1.2554,
      "step": 3241
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.840141511832041,
      "learning_rate": 1.947408704419472e-05,
      "loss": 1.2764,
      "step": 3242
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.842227644229233,
      "learning_rate": 1.9473669459378305e-05,
      "loss": 1.1892,
      "step": 3243
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.231742095563332,
      "learning_rate": 1.9473251713323323e-05,
      "loss": 1.3548,
      "step": 3244
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.965579620840349,
      "learning_rate": 1.9472833806036876e-05,
      "loss": 1.2348,
      "step": 3245
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.291797988219052,
      "learning_rate": 1.9472415737526086e-05,
      "loss": 1.1578,
      "step": 3246
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.05889135336164,
      "learning_rate": 1.9471997507798062e-05,
      "loss": 1.6284,
      "step": 3247
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.679425184609992,
      "learning_rate": 1.9471579116859926e-05,
      "loss": 1.2238,
      "step": 3248
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.108669539692443,
      "learning_rate": 1.9471160564718795e-05,
      "loss": 0.8654,
      "step": 3249
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.76392612498795,
      "learning_rate": 1.9470741851381796e-05,
      "loss": 1.3839,
      "step": 3250
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.253218852365156,
      "learning_rate": 1.9470322976856053e-05,
      "loss": 1.1897,
      "step": 3251
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.079527753969774,
      "learning_rate": 1.94699039411487e-05,
      "loss": 1.2545,
      "step": 3252
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.663969590548592,
      "learning_rate": 1.946948474426686e-05,
      "loss": 1.4431,
      "step": 3253
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.484699512101987,
      "learning_rate": 1.9469065386217673e-05,
      "loss": 1.0928,
      "step": 3254
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.717445961901015,
      "learning_rate": 1.9468645867008276e-05,
      "loss": 1.3254,
      "step": 3255
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.071149164123177,
      "learning_rate": 1.9468226186645813e-05,
      "loss": 1.0902,
      "step": 3256
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.218806457313732,
      "learning_rate": 1.946780634513742e-05,
      "loss": 1.6235,
      "step": 3257
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.333969292158284,
      "learning_rate": 1.9467386342490245e-05,
      "loss": 1.1253,
      "step": 3258
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.559886561860392,
      "learning_rate": 1.9466966178711438e-05,
      "loss": 1.2546,
      "step": 3259
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.69814946934972,
      "learning_rate": 1.9466545853808147e-05,
      "loss": 1.5185,
      "step": 3260
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.238771490499749,
      "learning_rate": 1.9466125367787528e-05,
      "loss": 1.408,
      "step": 3261
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.628580859231505,
      "learning_rate": 1.9465704720656736e-05,
      "loss": 1.6864,
      "step": 3262
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.751367572304234,
      "learning_rate": 1.9465283912422934e-05,
      "loss": 1.5183,
      "step": 3263
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.538599781838982,
      "learning_rate": 1.946486294309328e-05,
      "loss": 1.5341,
      "step": 3264
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.820825510610087,
      "learning_rate": 1.946444181267494e-05,
      "loss": 1.2936,
      "step": 3265
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.928673362332088,
      "learning_rate": 1.9464020521175083e-05,
      "loss": 1.0408,
      "step": 3266
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.6252149959206506,
      "learning_rate": 1.9463599068600875e-05,
      "loss": 1.2583,
      "step": 3267
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.412580713080168,
      "learning_rate": 1.9463177454959493e-05,
      "loss": 1.3512,
      "step": 3268
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.653030957375395,
      "learning_rate": 1.9462755680258113e-05,
      "loss": 1.1338,
      "step": 3269
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.447617988641233,
      "learning_rate": 1.946233374450391e-05,
      "loss": 1.5323,
      "step": 3270
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.282261222145705,
      "learning_rate": 1.9461911647704066e-05,
      "loss": 1.4106,
      "step": 3271
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.960131756637283,
      "learning_rate": 1.946148938986577e-05,
      "loss": 1.3172,
      "step": 3272
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.674550950269288,
      "learning_rate": 1.94610669709962e-05,
      "loss": 1.4782,
      "step": 3273
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.506198869122454,
      "learning_rate": 1.946064439110255e-05,
      "loss": 1.4238,
      "step": 3274
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.886270450800704,
      "learning_rate": 1.9460221650192016e-05,
      "loss": 1.1621,
      "step": 3275
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.342368421347022,
      "learning_rate": 1.945979874827179e-05,
      "loss": 1.4895,
      "step": 3276
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.974476170875982,
      "learning_rate": 1.9459375685349066e-05,
      "loss": 1.3778,
      "step": 3277
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.723824414064462,
      "learning_rate": 1.9458952461431047e-05,
      "loss": 1.2274,
      "step": 3278
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.9309333494473275,
      "learning_rate": 1.945852907652494e-05,
      "loss": 1.3131,
      "step": 3279
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.174028374815243,
      "learning_rate": 1.9458105530637945e-05,
      "loss": 1.4126,
      "step": 3280
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.42526151683423,
      "learning_rate": 1.945768182377727e-05,
      "loss": 1.391,
      "step": 3281
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.7815508600708405,
      "learning_rate": 1.9457257955950136e-05,
      "loss": 1.1637,
      "step": 3282
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.235192022263862,
      "learning_rate": 1.9456833927163745e-05,
      "loss": 1.3215,
      "step": 3283
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.860765433504844,
      "learning_rate": 1.9456409737425322e-05,
      "loss": 1.5687,
      "step": 3284
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.50325663611714,
      "learning_rate": 1.945598538674208e-05,
      "loss": 1.2808,
      "step": 3285
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.042301135948131,
      "learning_rate": 1.9455560875121245e-05,
      "loss": 1.0842,
      "step": 3286
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.23677810978767,
      "learning_rate": 1.9455136202570045e-05,
      "loss": 1.6341,
      "step": 3287
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.433045526404774,
      "learning_rate": 1.9454711369095703e-05,
      "loss": 1.3169,
      "step": 3288
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.775568158855489,
      "learning_rate": 1.9454286374705454e-05,
      "loss": 1.237,
      "step": 3289
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.689083871178388,
      "learning_rate": 1.9453861219406524e-05,
      "loss": 0.9635,
      "step": 3290
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.372991689863902,
      "learning_rate": 1.9453435903206158e-05,
      "loss": 1.5761,
      "step": 3291
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.63275167794525,
      "learning_rate": 1.945301042611159e-05,
      "loss": 1.4305,
      "step": 3292
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.930546492153795,
      "learning_rate": 1.9452584788130056e-05,
      "loss": 1.2588,
      "step": 3293
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.366031774389702,
      "learning_rate": 1.9452158989268812e-05,
      "loss": 1.2981,
      "step": 3294
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.620615471082825,
      "learning_rate": 1.94517330295351e-05,
      "loss": 1.5391,
      "step": 3295
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.274268881662023,
      "learning_rate": 1.9451306908936166e-05,
      "loss": 1.2932,
      "step": 3296
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.281238755349078,
      "learning_rate": 1.9450880627479262e-05,
      "loss": 1.383,
      "step": 3297
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.673910053830081,
      "learning_rate": 1.945045418517165e-05,
      "loss": 1.2052,
      "step": 3298
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.618628733128624,
      "learning_rate": 1.9450027582020582e-05,
      "loss": 1.4226,
      "step": 3299
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.645187814093119,
      "learning_rate": 1.9449600818033322e-05,
      "loss": 1.5259,
      "step": 3300
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.099596961234936,
      "learning_rate": 1.9449173893217134e-05,
      "loss": 1.3621,
      "step": 3301
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.921301316880527,
      "learning_rate": 1.944874680757928e-05,
      "loss": 1.4232,
      "step": 3302
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.654607927474763,
      "learning_rate": 1.9448319561127032e-05,
      "loss": 1.3699,
      "step": 3303
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.439714898858324,
      "learning_rate": 1.944789215386766e-05,
      "loss": 1.2159,
      "step": 3304
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.72204302312801,
      "learning_rate": 1.9447464585808438e-05,
      "loss": 1.0172,
      "step": 3305
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.818956633735157,
      "learning_rate": 1.9447036856956643e-05,
      "loss": 1.3676,
      "step": 3306
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.262279286287546,
      "learning_rate": 1.944660896731956e-05,
      "loss": 1.5104,
      "step": 3307
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.573475144694797,
      "learning_rate": 1.9446180916904467e-05,
      "loss": 1.2324,
      "step": 3308
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.285351516155106,
      "learning_rate": 1.9445752705718644e-05,
      "loss": 1.347,
      "step": 3309
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.240872637600388,
      "learning_rate": 1.944532433376939e-05,
      "loss": 0.9876,
      "step": 3310
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.328598112214239,
      "learning_rate": 1.944489580106399e-05,
      "loss": 1.2482,
      "step": 3311
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.876037093380104,
      "learning_rate": 1.944446710760974e-05,
      "loss": 1.0243,
      "step": 3312
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.68922648656827,
      "learning_rate": 1.9444038253413927e-05,
      "loss": 1.2498,
      "step": 3313
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.386042769752434,
      "learning_rate": 1.944360923848386e-05,
      "loss": 1.1413,
      "step": 3314
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.659303162081216,
      "learning_rate": 1.9443180062826842e-05,
      "loss": 1.0009,
      "step": 3315
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.670410536949369,
      "learning_rate": 1.9442750726450166e-05,
      "loss": 1.2824,
      "step": 3316
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.413963606263734,
      "learning_rate": 1.9442321229361153e-05,
      "loss": 1.4245,
      "step": 3317
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.585528788406191,
      "learning_rate": 1.9441891571567102e-05,
      "loss": 0.8945,
      "step": 3318
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.685591388886318,
      "learning_rate": 1.9441461753075333e-05,
      "loss": 1.3506,
      "step": 3319
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.145652990902651,
      "learning_rate": 1.9441031773893154e-05,
      "loss": 1.0681,
      "step": 3320
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.352750874389817,
      "learning_rate": 1.9440601634027892e-05,
      "loss": 1.2163,
      "step": 3321
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.849806879905485,
      "learning_rate": 1.944017133348686e-05,
      "loss": 1.2149,
      "step": 3322
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.628944191212095,
      "learning_rate": 1.9439740872277384e-05,
      "loss": 1.4039,
      "step": 3323
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.410014045362807,
      "learning_rate": 1.9439310250406794e-05,
      "loss": 1.288,
      "step": 3324
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.8274339350327224,
      "learning_rate": 1.9438879467882413e-05,
      "loss": 1.3408,
      "step": 3325
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.580427472801265,
      "learning_rate": 1.9438448524711576e-05,
      "loss": 1.1246,
      "step": 3326
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.773255568507889,
      "learning_rate": 1.943801742090162e-05,
      "loss": 1.6044,
      "step": 3327
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.264925055453032,
      "learning_rate": 1.943758615645988e-05,
      "loss": 1.0845,
      "step": 3328
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.1285509559694455,
      "learning_rate": 1.9437154731393694e-05,
      "loss": 1.1623,
      "step": 3329
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.889338185632538,
      "learning_rate": 1.94367231457104e-05,
      "loss": 1.4098,
      "step": 3330
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.47445467463117,
      "learning_rate": 1.943629139941736e-05,
      "loss": 1.1482,
      "step": 3331
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.552692855026888,
      "learning_rate": 1.9435859492521906e-05,
      "loss": 1.3552,
      "step": 3332
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.4598459268808694,
      "learning_rate": 1.9435427425031393e-05,
      "loss": 1.1868,
      "step": 3333
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.120259090094493,
      "learning_rate": 1.943499519695318e-05,
      "loss": 1.4921,
      "step": 3334
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.792501300172822,
      "learning_rate": 1.9434562808294622e-05,
      "loss": 1.061,
      "step": 3335
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.598181886120825,
      "learning_rate": 1.943413025906307e-05,
      "loss": 1.0647,
      "step": 3336
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.230043662129069,
      "learning_rate": 1.9433697549265896e-05,
      "loss": 1.0181,
      "step": 3337
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.856936026810484,
      "learning_rate": 1.943326467891046e-05,
      "loss": 1.6702,
      "step": 3338
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.03830172688048,
      "learning_rate": 1.9432831648004128e-05,
      "loss": 1.4255,
      "step": 3339
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.63577456919776,
      "learning_rate": 1.943239845655427e-05,
      "loss": 1.2956,
      "step": 3340
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.774999816236365,
      "learning_rate": 1.943196510456826e-05,
      "loss": 1.1748,
      "step": 3341
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.843903043950551,
      "learning_rate": 1.943153159205348e-05,
      "loss": 1.1477,
      "step": 3342
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.234248514806017,
      "learning_rate": 1.94310979190173e-05,
      "loss": 1.5955,
      "step": 3343
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.338884870818518,
      "learning_rate": 1.94306640854671e-05,
      "loss": 0.9635,
      "step": 3344
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.23531744425188,
      "learning_rate": 1.9430230091410265e-05,
      "loss": 1.223,
      "step": 3345
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.563420286292663,
      "learning_rate": 1.942979593685419e-05,
      "loss": 1.5077,
      "step": 3346
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.431782743218271,
      "learning_rate": 1.942936162180625e-05,
      "loss": 1.1606,
      "step": 3347
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.290189750667179,
      "learning_rate": 1.942892714627385e-05,
      "loss": 1.2706,
      "step": 3348
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.4201539775121725,
      "learning_rate": 1.9428492510264374e-05,
      "loss": 1.2016,
      "step": 3349
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.988778742524293,
      "learning_rate": 1.9428057713785226e-05,
      "loss": 1.2799,
      "step": 3350
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.041848148760502,
      "learning_rate": 1.9427622756843803e-05,
      "loss": 1.3313,
      "step": 3351
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.683812301600324,
      "learning_rate": 1.9427187639447514e-05,
      "loss": 1.3284,
      "step": 3352
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.037413054633752,
      "learning_rate": 1.942675236160376e-05,
      "loss": 1.3698,
      "step": 3353
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.495613753991952,
      "learning_rate": 1.942631692331994e-05,
      "loss": 0.9804,
      "step": 3354
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.543844139044633,
      "learning_rate": 1.942588132460348e-05,
      "loss": 1.4795,
      "step": 3355
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.469374523114687,
      "learning_rate": 1.942544556546179e-05,
      "loss": 1.206,
      "step": 3356
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.031339884769263,
      "learning_rate": 1.942500964590228e-05,
      "loss": 1.3744,
      "step": 3357
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.632459982394275,
      "learning_rate": 1.9424573565932375e-05,
      "loss": 1.3592,
      "step": 3358
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.152457338704897,
      "learning_rate": 1.9424137325559496e-05,
      "loss": 1.3744,
      "step": 3359
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.817381467132538,
      "learning_rate": 1.9423700924791065e-05,
      "loss": 1.1671,
      "step": 3360
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.856140417085936,
      "learning_rate": 1.9423264363634513e-05,
      "loss": 1.2425,
      "step": 3361
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.307620813916284,
      "learning_rate": 1.9422827642097266e-05,
      "loss": 1.4451,
      "step": 3362
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.016422738751547,
      "learning_rate": 1.9422390760186762e-05,
      "loss": 1.5191,
      "step": 3363
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.259095271282117,
      "learning_rate": 1.9421953717910434e-05,
      "loss": 1.3417,
      "step": 3364
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.196925698008163,
      "learning_rate": 1.9421516515275716e-05,
      "loss": 1.1784,
      "step": 3365
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.072968875541047,
      "learning_rate": 1.9421079152290058e-05,
      "loss": 1.3248,
      "step": 3366
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.10942870247672,
      "learning_rate": 1.9420641628960897e-05,
      "loss": 1.2596,
      "step": 3367
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.577252628633583,
      "learning_rate": 1.942020394529568e-05,
      "loss": 1.2609,
      "step": 3368
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.93289491121317,
      "learning_rate": 1.941976610130186e-05,
      "loss": 1.1995,
      "step": 3369
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.244429828615214,
      "learning_rate": 1.9419328096986887e-05,
      "loss": 1.3102,
      "step": 3370
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.1170779650590354,
      "learning_rate": 1.9418889932358213e-05,
      "loss": 1.3383,
      "step": 3371
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.9837898830308545,
      "learning_rate": 1.94184516074233e-05,
      "loss": 1.2413,
      "step": 3372
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.802853367765761,
      "learning_rate": 1.9418013122189603e-05,
      "loss": 1.4723,
      "step": 3373
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.985108001041933,
      "learning_rate": 1.941757447666459e-05,
      "loss": 1.4141,
      "step": 3374
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.824270072485116,
      "learning_rate": 1.9417135670855725e-05,
      "loss": 1.2629,
      "step": 3375
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.991615788394417,
      "learning_rate": 1.9416696704770476e-05,
      "loss": 1.1989,
      "step": 3376
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.155859940372491,
      "learning_rate": 1.941625757841631e-05,
      "loss": 1.2238,
      "step": 3377
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.224106117372199,
      "learning_rate": 1.9415818291800706e-05,
      "loss": 1.4948,
      "step": 3378
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.589497109648999,
      "learning_rate": 1.9415378844931144e-05,
      "loss": 1.5896,
      "step": 3379
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.170033320692142,
      "learning_rate": 1.941493923781509e-05,
      "loss": 1.0294,
      "step": 3380
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.051430615475649,
      "learning_rate": 1.9414499470460038e-05,
      "loss": 1.3544,
      "step": 3381
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.496059669335801,
      "learning_rate": 1.941405954287347e-05,
      "loss": 1.2456,
      "step": 3382
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.16012270156543,
      "learning_rate": 1.9413619455062873e-05,
      "loss": 1.316,
      "step": 3383
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.670541146981436,
      "learning_rate": 1.941317920703574e-05,
      "loss": 1.1069,
      "step": 3384
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.712483386603127,
      "learning_rate": 1.941273879879955e-05,
      "loss": 0.9328,
      "step": 3385
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.9802664301070703,
      "learning_rate": 1.9412298230361816e-05,
      "loss": 1.0938,
      "step": 3386
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.463753723647389,
      "learning_rate": 1.9411857501730028e-05,
      "loss": 1.3868,
      "step": 3387
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.511544210395661,
      "learning_rate": 1.9411416612911688e-05,
      "loss": 1.3139,
      "step": 3388
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.611414218736167,
      "learning_rate": 1.9410975563914298e-05,
      "loss": 1.4016,
      "step": 3389
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.964293718582946,
      "learning_rate": 1.9410534354745367e-05,
      "loss": 1.2763,
      "step": 3390
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.362813387979358,
      "learning_rate": 1.9410092985412405e-05,
      "loss": 1.3941,
      "step": 3391
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.376554532814177,
      "learning_rate": 1.9409651455922925e-05,
      "loss": 1.5512,
      "step": 3392
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.406336684032062,
      "learning_rate": 1.9409209766284438e-05,
      "loss": 1.1554,
      "step": 3393
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.743753070043522,
      "learning_rate": 1.940876791650446e-05,
      "loss": 1.6023,
      "step": 3394
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.383232165187367,
      "learning_rate": 1.9408325906590517e-05,
      "loss": 1.4496,
      "step": 3395
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.906339305535815,
      "learning_rate": 1.9407883736550128e-05,
      "loss": 1.2932,
      "step": 3396
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.025327306896144,
      "learning_rate": 1.9407441406390818e-05,
      "loss": 1.2405,
      "step": 3397
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.769775916082093,
      "learning_rate": 1.940699891612012e-05,
      "loss": 1.4101,
      "step": 3398
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.052871484635951,
      "learning_rate": 1.940655626574556e-05,
      "loss": 1.2884,
      "step": 3399
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.414912173761975,
      "learning_rate": 1.9406113455274672e-05,
      "loss": 1.4323,
      "step": 3400
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.83116374807513,
      "learning_rate": 1.9405670484714995e-05,
      "loss": 1.1161,
      "step": 3401
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.821701934315629,
      "learning_rate": 1.940522735407407e-05,
      "loss": 1.4214,
      "step": 3402
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.0175844553651725,
      "learning_rate": 1.940478406335943e-05,
      "loss": 1.5578,
      "step": 3403
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.48628861303603,
      "learning_rate": 1.9404340612578628e-05,
      "loss": 1.5568,
      "step": 3404
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.446606043686788,
      "learning_rate": 1.940389700173921e-05,
      "loss": 1.4454,
      "step": 3405
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.819481838956032,
      "learning_rate": 1.9403453230848726e-05,
      "loss": 1.4707,
      "step": 3406
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.7041510195737315,
      "learning_rate": 1.9403009299914727e-05,
      "loss": 1.2329,
      "step": 3407
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.361500083965648,
      "learning_rate": 1.940256520894477e-05,
      "loss": 1.3406,
      "step": 3408
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1733373194692565,
      "learning_rate": 1.9402120957946413e-05,
      "loss": 1.5265,
      "step": 3409
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.465194621615663,
      "learning_rate": 1.9401676546927218e-05,
      "loss": 1.0876,
      "step": 3410
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.603399231957344,
      "learning_rate": 1.9401231975894744e-05,
      "loss": 1.4524,
      "step": 3411
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.824476685524571,
      "learning_rate": 1.9400787244856564e-05,
      "loss": 1.715,
      "step": 3412
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.589583707872581,
      "learning_rate": 1.9400342353820244e-05,
      "loss": 1.3078,
      "step": 3413
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.374150355455869,
      "learning_rate": 1.9399897302793356e-05,
      "loss": 1.4007,
      "step": 3414
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.3259788621077595,
      "learning_rate": 1.9399452091783475e-05,
      "loss": 1.7372,
      "step": 3415
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.046465292883325,
      "learning_rate": 1.939900672079818e-05,
      "loss": 1.3025,
      "step": 3416
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.202166410265881,
      "learning_rate": 1.9398561189845045e-05,
      "loss": 1.3197,
      "step": 3417
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0489215579169695,
      "learning_rate": 1.9398115498931657e-05,
      "loss": 1.099,
      "step": 3418
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.529183044772131,
      "learning_rate": 1.9397669648065603e-05,
      "loss": 1.4582,
      "step": 3419
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.70313803446688,
      "learning_rate": 1.939722363725447e-05,
      "loss": 1.2427,
      "step": 3420
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.891997811350824,
      "learning_rate": 1.9396777466505846e-05,
      "loss": 1.4855,
      "step": 3421
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.553798407386221,
      "learning_rate": 1.939633113582733e-05,
      "loss": 1.6285,
      "step": 3422
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.431800125437125,
      "learning_rate": 1.9395884645226515e-05,
      "loss": 1.7717,
      "step": 3423
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.758094550645932,
      "learning_rate": 1.9395437994711002e-05,
      "loss": 1.36,
      "step": 3424
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.106706396827898,
      "learning_rate": 1.939499118428839e-05,
      "loss": 1.4688,
      "step": 3425
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.354318890115715,
      "learning_rate": 1.9394544213966288e-05,
      "loss": 1.3066,
      "step": 3426
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.88545503745137,
      "learning_rate": 1.9394097083752297e-05,
      "loss": 1.3422,
      "step": 3427
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.164132069473866,
      "learning_rate": 1.939364979365403e-05,
      "loss": 1.2147,
      "step": 3428
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.572123127985721,
      "learning_rate": 1.93932023436791e-05,
      "loss": 1.3116,
      "step": 3429
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.96948694150734,
      "learning_rate": 1.9392754733835128e-05,
      "loss": 1.1239,
      "step": 3430
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.96746578011833,
      "learning_rate": 1.939230696412972e-05,
      "loss": 1.2308,
      "step": 3431
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.886858589409234,
      "learning_rate": 1.939185903457051e-05,
      "loss": 1.5237,
      "step": 3432
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.14458108874245,
      "learning_rate": 1.939141094516511e-05,
      "loss": 1.4112,
      "step": 3433
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.563895841498034,
      "learning_rate": 1.9390962695921152e-05,
      "loss": 1.2075,
      "step": 3434
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.688612434360364,
      "learning_rate": 1.9390514286846266e-05,
      "loss": 1.2809,
      "step": 3435
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.455404088915596,
      "learning_rate": 1.9390065717948084e-05,
      "loss": 1.124,
      "step": 3436
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.377092463913481,
      "learning_rate": 1.9389616989234236e-05,
      "loss": 1.2928,
      "step": 3437
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.3292860420178,
      "learning_rate": 1.9389168100712363e-05,
      "loss": 1.4396,
      "step": 3438
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.865940936365773,
      "learning_rate": 1.9388719052390104e-05,
      "loss": 1.3292,
      "step": 3439
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.844952206312766,
      "learning_rate": 1.9388269844275103e-05,
      "loss": 1.4415,
      "step": 3440
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.39040531383545,
      "learning_rate": 1.9387820476375005e-05,
      "loss": 1.4864,
      "step": 3441
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.2925562792158996,
      "learning_rate": 1.938737094869745e-05,
      "loss": 1.1135,
      "step": 3442
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.245415351294139,
      "learning_rate": 1.9386921261250104e-05,
      "loss": 1.0453,
      "step": 3443
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.707016036573508,
      "learning_rate": 1.938647141404061e-05,
      "loss": 1.6276,
      "step": 3444
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.607878055502544,
      "learning_rate": 1.9386021407076626e-05,
      "loss": 1.31,
      "step": 3445
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.433957679482305,
      "learning_rate": 1.9385571240365813e-05,
      "loss": 1.4047,
      "step": 3446
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.558277596922672,
      "learning_rate": 1.938512091391583e-05,
      "loss": 1.5027,
      "step": 3447
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.543241024622766,
      "learning_rate": 1.938467042773434e-05,
      "loss": 1.4807,
      "step": 3448
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.155714350358287,
      "learning_rate": 1.938421978182902e-05,
      "loss": 1.1894,
      "step": 3449
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.291657158779086,
      "learning_rate": 1.9383768976207526e-05,
      "loss": 1.3246,
      "step": 3450
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.304015317762432,
      "learning_rate": 1.938331801087754e-05,
      "loss": 1.2248,
      "step": 3451
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.762949836475672,
      "learning_rate": 1.9382866885846735e-05,
      "loss": 1.3467,
      "step": 3452
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.608611672444151,
      "learning_rate": 1.9382415601122786e-05,
      "loss": 1.1973,
      "step": 3453
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.543379816820437,
      "learning_rate": 1.938196415671338e-05,
      "loss": 1.3761,
      "step": 3454
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.354132069285957,
      "learning_rate": 1.9381512552626192e-05,
      "loss": 1.3247,
      "step": 3455
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.903248979810366,
      "learning_rate": 1.9381060788868916e-05,
      "loss": 1.2662,
      "step": 3456
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.389771314102732,
      "learning_rate": 1.938060886544924e-05,
      "loss": 1.2107,
      "step": 3457
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.015642641356586,
      "learning_rate": 1.938015678237485e-05,
      "loss": 1.35,
      "step": 3458
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.581190872631078,
      "learning_rate": 1.9379704539653443e-05,
      "loss": 1.1677,
      "step": 3459
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.1322004279129,
      "learning_rate": 1.9379252137292716e-05,
      "loss": 1.3736,
      "step": 3460
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1417265371124214,
      "learning_rate": 1.9378799575300372e-05,
      "loss": 1.1147,
      "step": 3461
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.478441920222657,
      "learning_rate": 1.937834685368411e-05,
      "loss": 1.6412,
      "step": 3462
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.861013409056604,
      "learning_rate": 1.9377893972451636e-05,
      "loss": 1.3654,
      "step": 3463
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.762769559743608,
      "learning_rate": 1.9377440931610655e-05,
      "loss": 1.3497,
      "step": 3464
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.806424827527615,
      "learning_rate": 1.937698773116888e-05,
      "loss": 1.1595,
      "step": 3465
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.163343008450323,
      "learning_rate": 1.937653437113403e-05,
      "loss": 1.2692,
      "step": 3466
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.523775991468714,
      "learning_rate": 1.9376080851513814e-05,
      "loss": 1.158,
      "step": 3467
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.241072416361618,
      "learning_rate": 1.937562717231595e-05,
      "loss": 1.2557,
      "step": 3468
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.948099192001277,
      "learning_rate": 1.9375173333548164e-05,
      "loss": 1.6766,
      "step": 3469
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.716205195254176,
      "learning_rate": 1.9374719335218176e-05,
      "loss": 1.3781,
      "step": 3470
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.894590509924929,
      "learning_rate": 1.9374265177333716e-05,
      "loss": 1.1931,
      "step": 3471
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.358373304822158,
      "learning_rate": 1.9373810859902513e-05,
      "loss": 1.074,
      "step": 3472
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.872093054416607,
      "learning_rate": 1.93733563829323e-05,
      "loss": 1.3643,
      "step": 3473
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.124914847642334,
      "learning_rate": 1.937290174643081e-05,
      "loss": 1.3216,
      "step": 3474
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.24917466785008,
      "learning_rate": 1.937244695040578e-05,
      "loss": 1.1541,
      "step": 3475
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.574337029204829,
      "learning_rate": 1.9371991994864956e-05,
      "loss": 1.3603,
      "step": 3476
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.5933238798800415,
      "learning_rate": 1.9371536879816074e-05,
      "loss": 1.4235,
      "step": 3477
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.718265799348814,
      "learning_rate": 1.9371081605266884e-05,
      "loss": 1.4624,
      "step": 3478
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.294645468376381,
      "learning_rate": 1.9370626171225134e-05,
      "loss": 1.5719,
      "step": 3479
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.993040484422911,
      "learning_rate": 1.9370170577698575e-05,
      "loss": 1.7173,
      "step": 3480
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.181489864014435,
      "learning_rate": 1.9369714824694964e-05,
      "loss": 1.093,
      "step": 3481
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.965665987668762,
      "learning_rate": 1.9369258912222052e-05,
      "loss": 1.1253,
      "step": 3482
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.12412523605497,
      "learning_rate": 1.9368802840287603e-05,
      "loss": 1.4009,
      "step": 3483
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.742624776225516,
      "learning_rate": 1.9368346608899376e-05,
      "loss": 1.2876,
      "step": 3484
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.992417496795538,
      "learning_rate": 1.9367890218065136e-05,
      "loss": 1.3265,
      "step": 3485
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.37341153192314,
      "learning_rate": 1.9367433667792658e-05,
      "loss": 1.2273,
      "step": 3486
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.34479826162872,
      "learning_rate": 1.9366976958089704e-05,
      "loss": 1.3512,
      "step": 3487
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.382871018399976,
      "learning_rate": 1.9366520088964046e-05,
      "loss": 1.2288,
      "step": 3488
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.462706475481963,
      "learning_rate": 1.9366063060423468e-05,
      "loss": 1.6517,
      "step": 3489
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1258245338650354,
      "learning_rate": 1.936560587247574e-05,
      "loss": 1.3657,
      "step": 3490
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.75781319110719,
      "learning_rate": 1.9365148525128653e-05,
      "loss": 1.0261,
      "step": 3491
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.461145532762875,
      "learning_rate": 1.936469101838998e-05,
      "loss": 1.1841,
      "step": 3492
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.179499326549803,
      "learning_rate": 1.936423335226751e-05,
      "loss": 1.2588,
      "step": 3493
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.5727749207936075,
      "learning_rate": 1.936377552676904e-05,
      "loss": 1.22,
      "step": 3494
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.196754930823824,
      "learning_rate": 1.9363317541902353e-05,
      "loss": 1.0986,
      "step": 3495
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.556516018727316,
      "learning_rate": 1.9362859397675252e-05,
      "loss": 1.4878,
      "step": 3496
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.658884717243499,
      "learning_rate": 1.9362401094095527e-05,
      "loss": 1.323,
      "step": 3497
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.3152435849005615,
      "learning_rate": 1.936194263117098e-05,
      "loss": 1.1911,
      "step": 3498
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.4156630325329385,
      "learning_rate": 1.9361484008909418e-05,
      "loss": 1.1041,
      "step": 3499
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.7797491483193,
      "learning_rate": 1.936102522731864e-05,
      "loss": 1.11,
      "step": 3500
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.231591885572993,
      "learning_rate": 1.936056628640646e-05,
      "loss": 1.2652,
      "step": 3501
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.874898910780507,
      "learning_rate": 1.936010718618069e-05,
      "loss": 1.2977,
      "step": 3502
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.910529618788939,
      "learning_rate": 1.9359647926649137e-05,
      "loss": 1.2985,
      "step": 3503
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.127822701745876,
      "learning_rate": 1.9359188507819623e-05,
      "loss": 1.5123,
      "step": 3504
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.637024614745705,
      "learning_rate": 1.9358728929699966e-05,
      "loss": 1.1047,
      "step": 3505
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.359722410178401,
      "learning_rate": 1.9358269192297986e-05,
      "loss": 1.5618,
      "step": 3506
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.750770619213121,
      "learning_rate": 1.935780929562151e-05,
      "loss": 1.4019,
      "step": 3507
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.58412025929698,
      "learning_rate": 1.935734923967836e-05,
      "loss": 1.7258,
      "step": 3508
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.547147867446062,
      "learning_rate": 1.935688902447637e-05,
      "loss": 1.4428,
      "step": 3509
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.778229935074341,
      "learning_rate": 1.9356428650023377e-05,
      "loss": 1.0937,
      "step": 3510
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.60732683791496,
      "learning_rate": 1.935596811632721e-05,
      "loss": 1.3196,
      "step": 3511
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.700535746591945,
      "learning_rate": 1.935550742339571e-05,
      "loss": 1.3589,
      "step": 3512
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.131380948001451,
      "learning_rate": 1.9355046571236715e-05,
      "loss": 1.2749,
      "step": 3513
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.621406242803609,
      "learning_rate": 1.9354585559858073e-05,
      "loss": 1.3357,
      "step": 3514
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.661460594129713,
      "learning_rate": 1.9354124389267625e-05,
      "loss": 1.2304,
      "step": 3515
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.801585180083996,
      "learning_rate": 1.9353663059473225e-05,
      "loss": 1.3014,
      "step": 3516
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.983408719264548,
      "learning_rate": 1.9353201570482723e-05,
      "loss": 1.7416,
      "step": 3517
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.272217801127376,
      "learning_rate": 1.9352739922303973e-05,
      "loss": 1.1886,
      "step": 3518
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.731418781738641,
      "learning_rate": 1.935227811494483e-05,
      "loss": 1.3161,
      "step": 3519
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.4041217862465585,
      "learning_rate": 1.935181614841316e-05,
      "loss": 1.4277,
      "step": 3520
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.212395122218816,
      "learning_rate": 1.9351354022716817e-05,
      "loss": 1.3881,
      "step": 3521
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.30944805217136,
      "learning_rate": 1.9350891737863672e-05,
      "loss": 1.2659,
      "step": 3522
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.397898307814744,
      "learning_rate": 1.935042929386159e-05,
      "loss": 1.1778,
      "step": 3523
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.146021414355779,
      "learning_rate": 1.9349966690718448e-05,
      "loss": 1.499,
      "step": 3524
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.6445677890048405,
      "learning_rate": 1.9349503928442107e-05,
      "loss": 1.4357,
      "step": 3525
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.113995305391045,
      "learning_rate": 1.9349041007040456e-05,
      "loss": 1.138,
      "step": 3526
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.279183692764449,
      "learning_rate": 1.9348577926521365e-05,
      "loss": 1.2013,
      "step": 3527
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.70110630698537,
      "learning_rate": 1.9348114686892722e-05,
      "loss": 1.2864,
      "step": 3528
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.450872096008785,
      "learning_rate": 1.9347651288162405e-05,
      "loss": 1.5178,
      "step": 3529
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.342921691811906,
      "learning_rate": 1.9347187730338307e-05,
      "loss": 1.1821,
      "step": 3530
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.670805418502913,
      "learning_rate": 1.934672401342831e-05,
      "loss": 1.6262,
      "step": 3531
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.476008241308268,
      "learning_rate": 1.9346260137440313e-05,
      "loss": 1.4287,
      "step": 3532
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.818720125648486,
      "learning_rate": 1.9345796102382207e-05,
      "loss": 1.3461,
      "step": 3533
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.046703201104345,
      "learning_rate": 1.9345331908261893e-05,
      "loss": 1.3369,
      "step": 3534
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.615088791054546,
      "learning_rate": 1.934486755508727e-05,
      "loss": 1.492,
      "step": 3535
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.301105297708707,
      "learning_rate": 1.9344403042866242e-05,
      "loss": 1.2015,
      "step": 3536
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.463109159601858,
      "learning_rate": 1.9343938371606714e-05,
      "loss": 1.3505,
      "step": 3537
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.186095209401906,
      "learning_rate": 1.934347354131659e-05,
      "loss": 0.9972,
      "step": 3538
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.51826927075693,
      "learning_rate": 1.934300855200379e-05,
      "loss": 1.2761,
      "step": 3539
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.595616800886077,
      "learning_rate": 1.9342543403676222e-05,
      "loss": 1.2862,
      "step": 3540
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.0756309610333625,
      "learning_rate": 1.9342078096341804e-05,
      "loss": 1.157,
      "step": 3541
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.905058340147622,
      "learning_rate": 1.9341612630008457e-05,
      "loss": 1.0459,
      "step": 3542
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.046927815146896,
      "learning_rate": 1.9341147004684098e-05,
      "loss": 1.4215,
      "step": 3543
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.396685621252678,
      "learning_rate": 1.934068122037666e-05,
      "loss": 1.5611,
      "step": 3544
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.876620266252898,
      "learning_rate": 1.934021527709406e-05,
      "loss": 1.2384,
      "step": 3545
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.736525717937083,
      "learning_rate": 1.933974917484424e-05,
      "loss": 1.2211,
      "step": 3546
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.517526527577385,
      "learning_rate": 1.9339282913635125e-05,
      "loss": 1.2161,
      "step": 3547
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.755477376298946,
      "learning_rate": 1.933881649347465e-05,
      "loss": 1.4884,
      "step": 3548
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0292536306221365,
      "learning_rate": 1.9338349914370764e-05,
      "loss": 1.1779,
      "step": 3549
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.75411921483523,
      "learning_rate": 1.9337883176331392e-05,
      "loss": 1.2395,
      "step": 3550
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.676796576915265,
      "learning_rate": 1.9337416279364486e-05,
      "loss": 1.4636,
      "step": 3551
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.424827810644559,
      "learning_rate": 1.9336949223477993e-05,
      "loss": 1.5046,
      "step": 3552
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.84517265145236,
      "learning_rate": 1.9336482008679863e-05,
      "loss": 1.4599,
      "step": 3553
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.5351893978123945,
      "learning_rate": 1.9336014634978046e-05,
      "loss": 1.2653,
      "step": 3554
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.351882071932221,
      "learning_rate": 1.9335547102380493e-05,
      "loss": 1.4666,
      "step": 3555
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.23298100591561,
      "learning_rate": 1.933507941089517e-05,
      "loss": 1.3103,
      "step": 3556
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.755909101930564,
      "learning_rate": 1.933461156053003e-05,
      "loss": 1.4024,
      "step": 3557
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.131644604105007,
      "learning_rate": 1.9334143551293036e-05,
      "loss": 1.4308,
      "step": 3558
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.73181816984935,
      "learning_rate": 1.9333675383192155e-05,
      "loss": 1.3873,
      "step": 3559
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.127282418627163,
      "learning_rate": 1.9333207056235353e-05,
      "loss": 1.2409,
      "step": 3560
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.385636718875025,
      "learning_rate": 1.933273857043061e-05,
      "loss": 1.1726,
      "step": 3561
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.897729141243967,
      "learning_rate": 1.9332269925785886e-05,
      "loss": 1.2604,
      "step": 3562
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.544996930364611,
      "learning_rate": 1.9331801122309165e-05,
      "loss": 1.419,
      "step": 3563
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.316018173358168,
      "learning_rate": 1.933133216000842e-05,
      "loss": 1.3462,
      "step": 3564
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.588162260642491,
      "learning_rate": 1.933086303889164e-05,
      "loss": 1.231,
      "step": 3565
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.447751074374453,
      "learning_rate": 1.9330393758966808e-05,
      "loss": 1.186,
      "step": 3566
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2497593448365647,
      "learning_rate": 1.9329924320241905e-05,
      "loss": 1.1019,
      "step": 3567
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.070092271897849,
      "learning_rate": 1.932945472272493e-05,
      "loss": 1.3323,
      "step": 3568
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.954542640271877,
      "learning_rate": 1.932898496642386e-05,
      "loss": 1.3721,
      "step": 3569
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.487115156244947,
      "learning_rate": 1.932851505134671e-05,
      "loss": 1.3476,
      "step": 3570
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.843672548931465,
      "learning_rate": 1.9328044977501465e-05,
      "loss": 1.3099,
      "step": 3571
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.081740237232703,
      "learning_rate": 1.9327574744896125e-05,
      "loss": 1.0944,
      "step": 3572
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.368916255204438,
      "learning_rate": 1.9327104353538702e-05,
      "loss": 1.452,
      "step": 3573
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.662836395515465,
      "learning_rate": 1.9326633803437197e-05,
      "loss": 1.1854,
      "step": 3574
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.242383917318552,
      "learning_rate": 1.932616309459961e-05,
      "loss": 1.4741,
      "step": 3575
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.690959504717035,
      "learning_rate": 1.9325692227033966e-05,
      "loss": 1.4331,
      "step": 3576
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.080305621749548,
      "learning_rate": 1.9325221200748276e-05,
      "loss": 1.4988,
      "step": 3577
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.598070086629607,
      "learning_rate": 1.9324750015750552e-05,
      "loss": 1.412,
      "step": 3578
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.034487462878759,
      "learning_rate": 1.9324278672048812e-05,
      "loss": 1.5655,
      "step": 3579
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.728796048200182,
      "learning_rate": 1.9323807169651086e-05,
      "loss": 1.1206,
      "step": 3580
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.316272963710362,
      "learning_rate": 1.932333550856539e-05,
      "loss": 1.3065,
      "step": 3581
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.933249928534773,
      "learning_rate": 1.932286368879976e-05,
      "loss": 1.468,
      "step": 3582
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.402785082308136,
      "learning_rate": 1.9322391710362218e-05,
      "loss": 1.4435,
      "step": 3583
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.816580075304339,
      "learning_rate": 1.9321919573260803e-05,
      "loss": 1.3101,
      "step": 3584
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.5722148320596165,
      "learning_rate": 1.9321447277503548e-05,
      "loss": 1.3343,
      "step": 3585
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.944792166548335,
      "learning_rate": 1.9320974823098496e-05,
      "loss": 1.133,
      "step": 3586
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.5003254459739015,
      "learning_rate": 1.932050221005368e-05,
      "loss": 1.0491,
      "step": 3587
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.075978641757509,
      "learning_rate": 1.932002943837715e-05,
      "loss": 1.5595,
      "step": 3588
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.399880362003879,
      "learning_rate": 1.931955650807695e-05,
      "loss": 1.3211,
      "step": 3589
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.10246550786571,
      "learning_rate": 1.9319083419161125e-05,
      "loss": 1.1488,
      "step": 3590
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.288271731388031,
      "learning_rate": 1.9318610171637734e-05,
      "loss": 1.4547,
      "step": 3591
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.923431648573905,
      "learning_rate": 1.931813676551483e-05,
      "loss": 1.3313,
      "step": 3592
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.768131559573413,
      "learning_rate": 1.9317663200800468e-05,
      "loss": 1.2237,
      "step": 3593
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.676869049064393,
      "learning_rate": 1.9317189477502703e-05,
      "loss": 1.3421,
      "step": 3594
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.913947831849508,
      "learning_rate": 1.9316715595629613e-05,
      "loss": 1.486,
      "step": 3595
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.120575406662452,
      "learning_rate": 1.9316241555189245e-05,
      "loss": 1.3117,
      "step": 3596
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.852399156032039,
      "learning_rate": 1.931576735618968e-05,
      "loss": 1.3234,
      "step": 3597
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.797687182123218,
      "learning_rate": 1.9315292998638983e-05,
      "loss": 1.2737,
      "step": 3598
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.987113542103044,
      "learning_rate": 1.931481848254523e-05,
      "loss": 1.3531,
      "step": 3599
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.824767118872224,
      "learning_rate": 1.9314343807916495e-05,
      "loss": 1.3492,
      "step": 3600
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.596404174813881,
      "learning_rate": 1.9313868974760854e-05,
      "loss": 1.3098,
      "step": 3601
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.472603541163481,
      "learning_rate": 1.9313393983086396e-05,
      "loss": 1.4864,
      "step": 3602
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.130093005860818,
      "learning_rate": 1.93129188329012e-05,
      "loss": 1.2642,
      "step": 3603
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.39032586262664,
      "learning_rate": 1.9312443524213356e-05,
      "loss": 1.2612,
      "step": 3604
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.698058669876713,
      "learning_rate": 1.931196805703095e-05,
      "loss": 1.3444,
      "step": 3605
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.715010709611209,
      "learning_rate": 1.9311492431362077e-05,
      "loss": 1.1977,
      "step": 3606
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.020022402447842,
      "learning_rate": 1.9311016647214828e-05,
      "loss": 1.2581,
      "step": 3607
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.5807724501818425,
      "learning_rate": 1.9310540704597307e-05,
      "loss": 1.209,
      "step": 3608
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.668415692756213,
      "learning_rate": 1.931006460351761e-05,
      "loss": 1.3596,
      "step": 3609
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.108981395578428,
      "learning_rate": 1.9309588343983844e-05,
      "loss": 0.9325,
      "step": 3610
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.131992702399826,
      "learning_rate": 1.9309111926004107e-05,
      "loss": 1.27,
      "step": 3611
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.042034294276209,
      "learning_rate": 1.9308635349586512e-05,
      "loss": 1.2546,
      "step": 3612
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.312828712542513,
      "learning_rate": 1.9308158614739177e-05,
      "loss": 1.295,
      "step": 3613
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.504771280480764,
      "learning_rate": 1.9307681721470202e-05,
      "loss": 1.3963,
      "step": 3614
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.050241099585464,
      "learning_rate": 1.9307204669787716e-05,
      "loss": 1.0411,
      "step": 3615
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.523750735929576,
      "learning_rate": 1.930672745969983e-05,
      "loss": 1.1821,
      "step": 3616
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.872808469155872,
      "learning_rate": 1.9306250091214672e-05,
      "loss": 1.5609,
      "step": 3617
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.717052631414772,
      "learning_rate": 1.930577256434036e-05,
      "loss": 1.6624,
      "step": 3618
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.0116159017622985,
      "learning_rate": 1.930529487908503e-05,
      "loss": 1.2977,
      "step": 3619
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.139016423607786,
      "learning_rate": 1.9304817035456804e-05,
      "loss": 1.3727,
      "step": 3620
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.065429177253061,
      "learning_rate": 1.9304339033463816e-05,
      "loss": 1.009,
      "step": 3621
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.579654315103523,
      "learning_rate": 1.9303860873114207e-05,
      "loss": 1.2083,
      "step": 3622
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.322308122966746,
      "learning_rate": 1.930338255441611e-05,
      "loss": 1.0228,
      "step": 3623
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.492213567756576,
      "learning_rate": 1.9302904077377668e-05,
      "loss": 1.2802,
      "step": 3624
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.906902712563606,
      "learning_rate": 1.9302425442007025e-05,
      "loss": 1.1329,
      "step": 3625
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.667151005763552,
      "learning_rate": 1.930194664831232e-05,
      "loss": 1.2075,
      "step": 3626
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.431314148854089,
      "learning_rate": 1.9301467696301714e-05,
      "loss": 1.3872,
      "step": 3627
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.2163390874354585,
      "learning_rate": 1.9300988585983354e-05,
      "loss": 1.2297,
      "step": 3628
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.907424457661549,
      "learning_rate": 1.9300509317365387e-05,
      "loss": 1.2644,
      "step": 3629
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.525480119631577,
      "learning_rate": 1.930002989045598e-05,
      "loss": 1.2747,
      "step": 3630
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.510632692476909,
      "learning_rate": 1.9299550305263285e-05,
      "loss": 1.2827,
      "step": 3631
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.043368806106176,
      "learning_rate": 1.929907056179547e-05,
      "loss": 1.3859,
      "step": 3632
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.859155643966201,
      "learning_rate": 1.9298590660060698e-05,
      "loss": 0.9335,
      "step": 3633
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.672651385402915,
      "learning_rate": 1.9298110600067134e-05,
      "loss": 1.2317,
      "step": 3634
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.490775705702057,
      "learning_rate": 1.9297630381822956e-05,
      "loss": 1.526,
      "step": 3635
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.323688208103202,
      "learning_rate": 1.9297150005336327e-05,
      "loss": 1.1783,
      "step": 3636
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.851350916321758,
      "learning_rate": 1.9296669470615432e-05,
      "loss": 1.0798,
      "step": 3637
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.977923548771917,
      "learning_rate": 1.9296188777668444e-05,
      "loss": 1.3619,
      "step": 3638
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.804242503767528,
      "learning_rate": 1.9295707926503545e-05,
      "loss": 1.2245,
      "step": 3639
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.668574121363205,
      "learning_rate": 1.929522691712892e-05,
      "loss": 1.366,
      "step": 3640
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.984019835954693,
      "learning_rate": 1.929474574955276e-05,
      "loss": 0.9846,
      "step": 3641
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.274181238217217,
      "learning_rate": 1.9294264423783246e-05,
      "loss": 1.1889,
      "step": 3642
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.728230889564322,
      "learning_rate": 1.929378293982857e-05,
      "loss": 1.3088,
      "step": 3643
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.970012673731528,
      "learning_rate": 1.9293301297696936e-05,
      "loss": 1.222,
      "step": 3644
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.583279825897293,
      "learning_rate": 1.9292819497396532e-05,
      "loss": 1.3263,
      "step": 3645
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.57970882903596,
      "learning_rate": 1.9292337538935568e-05,
      "loss": 1.3843,
      "step": 3646
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.899798365587663,
      "learning_rate": 1.9291855422322233e-05,
      "loss": 1.261,
      "step": 3647
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.2634117257346515,
      "learning_rate": 1.9291373147564743e-05,
      "loss": 1.4518,
      "step": 3648
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.393498341951171,
      "learning_rate": 1.9290890714671304e-05,
      "loss": 1.128,
      "step": 3649
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.716729883723744,
      "learning_rate": 1.9290408123650127e-05,
      "loss": 0.804,
      "step": 3650
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.517149975457266,
      "learning_rate": 1.9289925374509422e-05,
      "loss": 1.1292,
      "step": 3651
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.567846451666844,
      "learning_rate": 1.9289442467257408e-05,
      "loss": 1.3339,
      "step": 3652
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.984416937200025,
      "learning_rate": 1.9288959401902306e-05,
      "loss": 1.0181,
      "step": 3653
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.895444387772017,
      "learning_rate": 1.9288476178452333e-05,
      "loss": 1.3377,
      "step": 3654
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.582332886043637,
      "learning_rate": 1.9287992796915716e-05,
      "loss": 1.337,
      "step": 3655
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.772375719885593,
      "learning_rate": 1.9287509257300684e-05,
      "loss": 1.3948,
      "step": 3656
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.39170452683877,
      "learning_rate": 1.9287025559615463e-05,
      "loss": 1.0504,
      "step": 3657
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.346128561062332,
      "learning_rate": 1.9286541703868287e-05,
      "loss": 1.4128,
      "step": 3658
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.711095436137454,
      "learning_rate": 1.928605769006739e-05,
      "loss": 1.3212,
      "step": 3659
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.849773416393885,
      "learning_rate": 1.9285573518221008e-05,
      "loss": 1.0052,
      "step": 3660
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.750127909569548,
      "learning_rate": 1.9285089188337385e-05,
      "loss": 1.4988,
      "step": 3661
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.486853898468992,
      "learning_rate": 1.9284604700424764e-05,
      "loss": 1.1074,
      "step": 3662
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.283655832184039,
      "learning_rate": 1.9284120054491393e-05,
      "loss": 1.3574,
      "step": 3663
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.893076616364612,
      "learning_rate": 1.9283635250545514e-05,
      "loss": 1.3644,
      "step": 3664
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.812246406475565,
      "learning_rate": 1.928315028859538e-05,
      "loss": 1.4042,
      "step": 3665
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.352552300886805,
      "learning_rate": 1.928266516864925e-05,
      "loss": 1.3715,
      "step": 3666
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.31070807562294,
      "learning_rate": 1.9282179890715377e-05,
      "loss": 1.4893,
      "step": 3667
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.452256551316395,
      "learning_rate": 1.9281694454802017e-05,
      "loss": 1.244,
      "step": 3668
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.17309529174972,
      "learning_rate": 1.9281208860917436e-05,
      "loss": 1.4553,
      "step": 3669
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.735994118998276,
      "learning_rate": 1.9280723109069898e-05,
      "loss": 1.2584,
      "step": 3670
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.648520444258801,
      "learning_rate": 1.9280237199267672e-05,
      "loss": 1.3496,
      "step": 3671
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.56478061535713,
      "learning_rate": 1.9279751131519027e-05,
      "loss": 1.0981,
      "step": 3672
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.2071165548835205,
      "learning_rate": 1.927926490583223e-05,
      "loss": 1.3553,
      "step": 3673
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.079019620365373,
      "learning_rate": 1.9278778522215567e-05,
      "loss": 1.2601,
      "step": 3674
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.122498246668452,
      "learning_rate": 1.9278291980677308e-05,
      "loss": 1.35,
      "step": 3675
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.790148936939341,
      "learning_rate": 1.9277805281225734e-05,
      "loss": 1.3816,
      "step": 3676
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.27965247935062,
      "learning_rate": 1.9277318423869133e-05,
      "loss": 1.2692,
      "step": 3677
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.817572267617018,
      "learning_rate": 1.9276831408615788e-05,
      "loss": 1.0147,
      "step": 3678
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.875090241137434,
      "learning_rate": 1.9276344235473988e-05,
      "loss": 1.0579,
      "step": 3679
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.402732151671582,
      "learning_rate": 1.9275856904452026e-05,
      "loss": 1.1325,
      "step": 3680
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.693677607393764,
      "learning_rate": 1.9275369415558196e-05,
      "loss": 1.3578,
      "step": 3681
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.986115359918962,
      "learning_rate": 1.9274881768800793e-05,
      "loss": 1.3884,
      "step": 3682
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.665038641187422,
      "learning_rate": 1.9274393964188118e-05,
      "loss": 0.9523,
      "step": 3683
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7636628425190217,
      "learning_rate": 1.9273906001728474e-05,
      "loss": 1.1186,
      "step": 3684
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.026186379803525,
      "learning_rate": 1.9273417881430163e-05,
      "loss": 1.1833,
      "step": 3685
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.033531731303629,
      "learning_rate": 1.9272929603301496e-05,
      "loss": 1.0802,
      "step": 3686
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.328588956025378,
      "learning_rate": 1.927244116735078e-05,
      "loss": 1.165,
      "step": 3687
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.018169671147362,
      "learning_rate": 1.927195257358633e-05,
      "loss": 1.4584,
      "step": 3688
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.956268215344535,
      "learning_rate": 1.9271463822016465e-05,
      "loss": 1.339,
      "step": 3689
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.075293777667218,
      "learning_rate": 1.9270974912649498e-05,
      "loss": 1.18,
      "step": 3690
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.452895779661805,
      "learning_rate": 1.9270485845493753e-05,
      "loss": 1.2107,
      "step": 3691
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.398338202271196,
      "learning_rate": 1.9269996620557553e-05,
      "loss": 1.3696,
      "step": 3692
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.368699449515494,
      "learning_rate": 1.9269507237849225e-05,
      "loss": 1.0349,
      "step": 3693
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.008223011938114,
      "learning_rate": 1.9269017697377094e-05,
      "loss": 1.3826,
      "step": 3694
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.213925188757125,
      "learning_rate": 1.9268527999149498e-05,
      "loss": 1.525,
      "step": 3695
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.000479089852423,
      "learning_rate": 1.9268038143174768e-05,
      "loss": 1.5339,
      "step": 3696
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.936056037864134,
      "learning_rate": 1.926754812946124e-05,
      "loss": 1.2142,
      "step": 3697
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.883685056970128,
      "learning_rate": 1.9267057958017258e-05,
      "loss": 1.2474,
      "step": 3698
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.103146379228568,
      "learning_rate": 1.926656762885116e-05,
      "loss": 1.1131,
      "step": 3699
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.432743471184644,
      "learning_rate": 1.9266077141971296e-05,
      "loss": 1.2827,
      "step": 3700
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.9905415171978555,
      "learning_rate": 1.9265586497386013e-05,
      "loss": 1.2263,
      "step": 3701
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.97507455174489,
      "learning_rate": 1.926509569510366e-05,
      "loss": 1.3248,
      "step": 3702
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.9220059987667675,
      "learning_rate": 1.9264604735132586e-05,
      "loss": 1.4234,
      "step": 3703
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.754350554195363,
      "learning_rate": 1.9264113617481153e-05,
      "loss": 1.1399,
      "step": 3704
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.178642875162765,
      "learning_rate": 1.926362234215772e-05,
      "loss": 1.6125,
      "step": 3705
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.6604005382348515,
      "learning_rate": 1.926313090917064e-05,
      "loss": 1.3311,
      "step": 3706
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.070124717070623,
      "learning_rate": 1.9262639318528292e-05,
      "loss": 1.2972,
      "step": 3707
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.731064718254311,
      "learning_rate": 1.9262147570239028e-05,
      "loss": 1.2734,
      "step": 3708
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.429318271011033,
      "learning_rate": 1.9261655664311226e-05,
      "loss": 1.2221,
      "step": 3709
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.302304958869578,
      "learning_rate": 1.9261163600753256e-05,
      "loss": 1.4537,
      "step": 3710
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.533648539056662,
      "learning_rate": 1.926067137957349e-05,
      "loss": 1.2812,
      "step": 3711
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.048886952960785,
      "learning_rate": 1.926017900078031e-05,
      "loss": 1.5396,
      "step": 3712
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.265197260912028,
      "learning_rate": 1.9259686464382093e-05,
      "loss": 1.4943,
      "step": 3713
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.8310756397495345,
      "learning_rate": 1.925919377038722e-05,
      "loss": 1.5629,
      "step": 3714
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.251915872759406,
      "learning_rate": 1.9258700918804084e-05,
      "loss": 1.0858,
      "step": 3715
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.005725986569699,
      "learning_rate": 1.9258207909641066e-05,
      "loss": 1.3501,
      "step": 3716
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.557941995569214,
      "learning_rate": 1.925771474290656e-05,
      "loss": 1.3268,
      "step": 3717
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.138057897210028,
      "learning_rate": 1.9257221418608954e-05,
      "loss": 1.2849,
      "step": 3718
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.488108963349086,
      "learning_rate": 1.9256727936756652e-05,
      "loss": 1.5539,
      "step": 3719
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.817781541738744,
      "learning_rate": 1.925623429735805e-05,
      "loss": 1.088,
      "step": 3720
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.245826881472131,
      "learning_rate": 1.9255740500421548e-05,
      "loss": 1.4417,
      "step": 3721
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.724485877740637,
      "learning_rate": 1.9255246545955554e-05,
      "loss": 1.4726,
      "step": 3722
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.767269610676141,
      "learning_rate": 1.925475243396847e-05,
      "loss": 1.2265,
      "step": 3723
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.67284850788565,
      "learning_rate": 1.9254258164468707e-05,
      "loss": 1.2075,
      "step": 3724
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.996101563161321,
      "learning_rate": 1.9253763737464683e-05,
      "loss": 1.2114,
      "step": 3725
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.241654358941049,
      "learning_rate": 1.9253269152964805e-05,
      "loss": 1.2924,
      "step": 3726
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.0739761135009545,
      "learning_rate": 1.9252774410977494e-05,
      "loss": 1.19,
      "step": 3727
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.627544700876754,
      "learning_rate": 1.925227951151117e-05,
      "loss": 1.114,
      "step": 3728
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.6427805283444705,
      "learning_rate": 1.9251784454574262e-05,
      "loss": 1.144,
      "step": 3729
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.9774769806845205,
      "learning_rate": 1.9251289240175182e-05,
      "loss": 1.3762,
      "step": 3730
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.578277478599333,
      "learning_rate": 1.925079386832237e-05,
      "loss": 1.46,
      "step": 3731
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.804664393037738,
      "learning_rate": 1.9250298339024255e-05,
      "loss": 1.3968,
      "step": 3732
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.208521764488242,
      "learning_rate": 1.9249802652289267e-05,
      "loss": 1.3355,
      "step": 3733
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.182292652665391,
      "learning_rate": 1.9249306808125846e-05,
      "loss": 1.3366,
      "step": 3734
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.743112719592296,
      "learning_rate": 1.924881080654243e-05,
      "loss": 1.4569,
      "step": 3735
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.450416471870751,
      "learning_rate": 1.9248314647547457e-05,
      "loss": 1.5415,
      "step": 3736
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.440774616437502,
      "learning_rate": 1.924781833114938e-05,
      "loss": 1.4191,
      "step": 3737
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.810002646614609,
      "learning_rate": 1.9247321857356634e-05,
      "loss": 1.1696,
      "step": 3738
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.852139207604734,
      "learning_rate": 1.924682522617768e-05,
      "loss": 1.3536,
      "step": 3739
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.032221007590808,
      "learning_rate": 1.924632843762097e-05,
      "loss": 1.5368,
      "step": 3740
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.698590505776723,
      "learning_rate": 1.9245831491694955e-05,
      "loss": 1.3329,
      "step": 3741
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.576343666943949,
      "learning_rate": 1.924533438840809e-05,
      "loss": 1.2937,
      "step": 3742
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.188691956455258,
      "learning_rate": 1.924483712776884e-05,
      "loss": 1.3262,
      "step": 3743
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.551430712996993,
      "learning_rate": 1.9244339709785666e-05,
      "loss": 1.1791,
      "step": 3744
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.21766330659627,
      "learning_rate": 1.9243842134467035e-05,
      "loss": 1.378,
      "step": 3745
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.1124838385395535,
      "learning_rate": 1.9243344401821415e-05,
      "loss": 1.3612,
      "step": 3746
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.686074514033127,
      "learning_rate": 1.9242846511857277e-05,
      "loss": 1.2782,
      "step": 3747
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.26241414437536,
      "learning_rate": 1.92423484645831e-05,
      "loss": 1.3612,
      "step": 3748
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.519000917290419,
      "learning_rate": 1.9241850260007354e-05,
      "loss": 1.3866,
      "step": 3749
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.375136986811197,
      "learning_rate": 1.9241351898138522e-05,
      "loss": 1.5753,
      "step": 3750
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.863259096332787,
      "learning_rate": 1.9240853378985084e-05,
      "loss": 1.1668,
      "step": 3751
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.786533408740255,
      "learning_rate": 1.9240354702555524e-05,
      "loss": 1.3227,
      "step": 3752
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.820147376534494,
      "learning_rate": 1.923985586885833e-05,
      "loss": 1.1194,
      "step": 3753
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.744365054160161,
      "learning_rate": 1.9239356877901993e-05,
      "loss": 1.4594,
      "step": 3754
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.506276529928117,
      "learning_rate": 1.9238857729695007e-05,
      "loss": 1.3011,
      "step": 3755
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.317328298623736,
      "learning_rate": 1.923835842424586e-05,
      "loss": 1.196,
      "step": 3756
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.359263538222685,
      "learning_rate": 1.9237858961563063e-05,
      "loss": 1.5914,
      "step": 3757
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.362459258567222,
      "learning_rate": 1.9237359341655108e-05,
      "loss": 1.171,
      "step": 3758
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.270209784498621,
      "learning_rate": 1.923685956453049e-05,
      "loss": 1.2476,
      "step": 3759
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.891320776726994,
      "learning_rate": 1.9236359630197733e-05,
      "loss": 1.5037,
      "step": 3760
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.194767654159692,
      "learning_rate": 1.923585953866534e-05,
      "loss": 1.6077,
      "step": 3761
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.279986118357161,
      "learning_rate": 1.9235359289941814e-05,
      "loss": 1.4265,
      "step": 3762
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.878088776021628,
      "learning_rate": 1.9234858884035676e-05,
      "loss": 1.1897,
      "step": 3763
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.550540300259057,
      "learning_rate": 1.923435832095544e-05,
      "loss": 1.084,
      "step": 3764
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.752809656776331,
      "learning_rate": 1.9233857600709628e-05,
      "loss": 1.1469,
      "step": 3765
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.669034517079199,
      "learning_rate": 1.923335672330676e-05,
      "loss": 1.5183,
      "step": 3766
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.145025531869675,
      "learning_rate": 1.9232855688755365e-05,
      "loss": 1.1538,
      "step": 3767
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.423097826020777,
      "learning_rate": 1.923235449706396e-05,
      "loss": 1.3414,
      "step": 3768
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.9319014618644275,
      "learning_rate": 1.9231853148241092e-05,
      "loss": 1.0655,
      "step": 3769
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.919737016052818,
      "learning_rate": 1.9231351642295278e-05,
      "loss": 1.3515,
      "step": 3770
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.729249224771176,
      "learning_rate": 1.923084997923506e-05,
      "loss": 1.2049,
      "step": 3771
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.034395189026535,
      "learning_rate": 1.923034815906898e-05,
      "loss": 1.4821,
      "step": 3772
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.974134307215962,
      "learning_rate": 1.9229846181805565e-05,
      "loss": 1.2205,
      "step": 3773
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.192448997433581,
      "learning_rate": 1.9229344047453374e-05,
      "loss": 1.5085,
      "step": 3774
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.798058468258428,
      "learning_rate": 1.9228841756020948e-05,
      "loss": 1.376,
      "step": 3775
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.387259076644892,
      "learning_rate": 1.922833930751683e-05,
      "loss": 1.3605,
      "step": 3776
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.330209575658079,
      "learning_rate": 1.9227836701949577e-05,
      "loss": 1.3744,
      "step": 3777
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.979954302492915,
      "learning_rate": 1.9227333939327746e-05,
      "loss": 1.2736,
      "step": 3778
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.645129768018499,
      "learning_rate": 1.9226831019659887e-05,
      "loss": 1.2831,
      "step": 3779
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.319076667147134,
      "learning_rate": 1.9226327942954566e-05,
      "loss": 1.1212,
      "step": 3780
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.615559871893295,
      "learning_rate": 1.922582470922034e-05,
      "loss": 1.4999,
      "step": 3781
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.9468578876842,
      "learning_rate": 1.922532131846578e-05,
      "loss": 1.308,
      "step": 3782
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.688015286109089,
      "learning_rate": 1.9224817770699445e-05,
      "loss": 1.2843,
      "step": 3783
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.641621215101204,
      "learning_rate": 1.922431406592991e-05,
      "loss": 0.9755,
      "step": 3784
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.894752119471582,
      "learning_rate": 1.922381020416575e-05,
      "loss": 1.2113,
      "step": 3785
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.097364666716114,
      "learning_rate": 1.922330618541554e-05,
      "loss": 1.2074,
      "step": 3786
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.647640820032474,
      "learning_rate": 1.922280200968785e-05,
      "loss": 1.2878,
      "step": 3787
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.3250466723271215,
      "learning_rate": 1.922229767699127e-05,
      "loss": 1.2344,
      "step": 3788
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.123384543670834,
      "learning_rate": 1.9221793187334378e-05,
      "loss": 1.3558,
      "step": 3789
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.512702869317206,
      "learning_rate": 1.9221288540725768e-05,
      "loss": 1.4625,
      "step": 3790
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.1838517742396,
      "learning_rate": 1.922078373717402e-05,
      "loss": 1.1259,
      "step": 3791
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.995466628782504,
      "learning_rate": 1.9220278776687735e-05,
      "loss": 1.5003,
      "step": 3792
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.151449390600595,
      "learning_rate": 1.9219773659275495e-05,
      "loss": 1.4624,
      "step": 3793
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.1720808761756,
      "learning_rate": 1.9219268384945907e-05,
      "loss": 1.3972,
      "step": 3794
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.465595626878422,
      "learning_rate": 1.9218762953707568e-05,
      "loss": 1.1815,
      "step": 3795
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.158897803475003,
      "learning_rate": 1.9218257365569077e-05,
      "loss": 1.1699,
      "step": 3796
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.230357586269678,
      "learning_rate": 1.9217751620539044e-05,
      "loss": 1.8189,
      "step": 3797
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.730949594137556,
      "learning_rate": 1.9217245718626076e-05,
      "loss": 0.9981,
      "step": 3798
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.667613083579999,
      "learning_rate": 1.9216739659838776e-05,
      "loss": 1.353,
      "step": 3799
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.340806530774783,
      "learning_rate": 1.9216233444185762e-05,
      "loss": 1.3024,
      "step": 3800
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.163995474749633,
      "learning_rate": 1.9215727071675654e-05,
      "loss": 1.5443,
      "step": 3801
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.665633643031014,
      "learning_rate": 1.9215220542317066e-05,
      "loss": 1.4613,
      "step": 3802
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.173796924601962,
      "learning_rate": 1.9214713856118615e-05,
      "loss": 1.3302,
      "step": 3803
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.605443035063351,
      "learning_rate": 1.9214207013088935e-05,
      "loss": 1.5524,
      "step": 3804
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.658690125701912,
      "learning_rate": 1.9213700013236644e-05,
      "loss": 1.2231,
      "step": 3805
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.993704515848969,
      "learning_rate": 1.9213192856570368e-05,
      "loss": 1.2258,
      "step": 3806
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.760948090715163,
      "learning_rate": 1.921268554309875e-05,
      "loss": 1.5158,
      "step": 3807
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.067065161736737,
      "learning_rate": 1.9212178072830416e-05,
      "loss": 1.4492,
      "step": 3808
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.472669688632329,
      "learning_rate": 1.9211670445774e-05,
      "loss": 1.1818,
      "step": 3809
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.714497295994995,
      "learning_rate": 1.9211162661938155e-05,
      "loss": 1.3543,
      "step": 3810
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.577241083585308,
      "learning_rate": 1.921065472133151e-05,
      "loss": 1.3984,
      "step": 3811
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.700557794574102,
      "learning_rate": 1.9210146623962718e-05,
      "loss": 1.3243,
      "step": 3812
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.80797467472245,
      "learning_rate": 1.920963836984042e-05,
      "loss": 1.3319,
      "step": 3813
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.749698605675931,
      "learning_rate": 1.920912995897327e-05,
      "loss": 1.3987,
      "step": 3814
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.138633510085447,
      "learning_rate": 1.9208621391369922e-05,
      "loss": 1.0934,
      "step": 3815
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.469142077486504,
      "learning_rate": 1.920811266703903e-05,
      "loss": 1.3873,
      "step": 3816
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.022262116622826,
      "learning_rate": 1.920760378598925e-05,
      "loss": 1.1021,
      "step": 3817
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.20752288032026,
      "learning_rate": 1.9207094748229252e-05,
      "loss": 0.9833,
      "step": 3818
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.135915586436894,
      "learning_rate": 1.920658555376769e-05,
      "loss": 1.1488,
      "step": 3819
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.535549625733738,
      "learning_rate": 1.9206076202613232e-05,
      "loss": 1.1257,
      "step": 3820
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.640631376982391,
      "learning_rate": 1.920556669477455e-05,
      "loss": 1.0853,
      "step": 3821
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.323387246951501,
      "learning_rate": 1.9205057030260316e-05,
      "loss": 1.5059,
      "step": 3822
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.690342472497452,
      "learning_rate": 1.92045472090792e-05,
      "loss": 1.3229,
      "step": 3823
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.130675119792466,
      "learning_rate": 1.9204037231239882e-05,
      "loss": 1.5292,
      "step": 3824
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.19741804077045,
      "learning_rate": 1.9203527096751045e-05,
      "loss": 1.7438,
      "step": 3825
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.157199695102383,
      "learning_rate": 1.9203016805621364e-05,
      "loss": 1.6524,
      "step": 3826
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.404130745570322,
      "learning_rate": 1.920250635785953e-05,
      "loss": 1.551,
      "step": 3827
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.738529910094059,
      "learning_rate": 1.9201995753474224e-05,
      "loss": 1.3145,
      "step": 3828
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.772856275114295,
      "learning_rate": 1.920148499247414e-05,
      "loss": 1.3349,
      "step": 3829
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.074887256067584,
      "learning_rate": 1.920097407486798e-05,
      "loss": 1.4214,
      "step": 3830
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.697891003247016,
      "learning_rate": 1.9200463000664422e-05,
      "loss": 1.3174,
      "step": 3831
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.128807077075823,
      "learning_rate": 1.9199951769872174e-05,
      "loss": 1.2457,
      "step": 3832
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.792345818206707,
      "learning_rate": 1.919944038249994e-05,
      "loss": 1.6631,
      "step": 3833
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.974756661409931,
      "learning_rate": 1.919892883855642e-05,
      "loss": 1.0845,
      "step": 3834
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.436150015078166,
      "learning_rate": 1.919841713805032e-05,
      "loss": 1.3107,
      "step": 3835
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.372083409615469,
      "learning_rate": 1.9197905280990348e-05,
      "loss": 1.1329,
      "step": 3836
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.8916240096357244,
      "learning_rate": 1.919739326738522e-05,
      "loss": 1.1311,
      "step": 3837
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.788597186741629,
      "learning_rate": 1.9196881097243642e-05,
      "loss": 1.1001,
      "step": 3838
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.21082485911273,
      "learning_rate": 1.919636877057434e-05,
      "loss": 1.7171,
      "step": 3839
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.436244014812157,
      "learning_rate": 1.9195856287386027e-05,
      "loss": 1.3476,
      "step": 3840
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.0814037512354915,
      "learning_rate": 1.919534364768743e-05,
      "loss": 1.1749,
      "step": 3841
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.1856133308859365,
      "learning_rate": 1.9194830851487273e-05,
      "loss": 1.182,
      "step": 3842
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.65762750758182,
      "learning_rate": 1.919431789879428e-05,
      "loss": 1.1558,
      "step": 3843
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.007061909464161,
      "learning_rate": 1.9193804789617188e-05,
      "loss": 1.5005,
      "step": 3844
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.5035270428692735,
      "learning_rate": 1.9193291523964722e-05,
      "loss": 1.4588,
      "step": 3845
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.086662395379156,
      "learning_rate": 1.9192778101845622e-05,
      "loss": 1.3039,
      "step": 3846
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.910266420009931,
      "learning_rate": 1.9192264523268624e-05,
      "loss": 1.1328,
      "step": 3847
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.698133319571874,
      "learning_rate": 1.9191750788242472e-05,
      "loss": 1.2088,
      "step": 3848
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.33086963615655,
      "learning_rate": 1.919123689677591e-05,
      "loss": 1.7101,
      "step": 3849
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.356822169575082,
      "learning_rate": 1.9190722848877683e-05,
      "loss": 1.3061,
      "step": 3850
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.249389388081088,
      "learning_rate": 1.9190208644556538e-05,
      "loss": 1.3839,
      "step": 3851
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.897460340172067,
      "learning_rate": 1.918969428382123e-05,
      "loss": 1.5199,
      "step": 3852
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.98671194544257,
      "learning_rate": 1.9189179766680507e-05,
      "loss": 1.1599,
      "step": 3853
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.850408992493337,
      "learning_rate": 1.9188665093143133e-05,
      "loss": 1.3345,
      "step": 3854
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.391009796388568,
      "learning_rate": 1.9188150263217864e-05,
      "loss": 1.5095,
      "step": 3855
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.838441234242977,
      "learning_rate": 1.9187635276913463e-05,
      "loss": 1.4144,
      "step": 3856
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.034984575417448,
      "learning_rate": 1.9187120134238696e-05,
      "loss": 1.2125,
      "step": 3857
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.822944790439971,
      "learning_rate": 1.9186604835202324e-05,
      "loss": 1.0103,
      "step": 3858
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.889684936936214,
      "learning_rate": 1.918608937981313e-05,
      "loss": 1.3846,
      "step": 3859
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.21692384744174,
      "learning_rate": 1.9185573768079876e-05,
      "loss": 1.2952,
      "step": 3860
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.809230616007657,
      "learning_rate": 1.9185058000011342e-05,
      "loss": 1.401,
      "step": 3861
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.576857951911166,
      "learning_rate": 1.9184542075616302e-05,
      "loss": 1.2086,
      "step": 3862
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.734310243437038,
      "learning_rate": 1.9184025994903544e-05,
      "loss": 1.4372,
      "step": 3863
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.837163982255326,
      "learning_rate": 1.9183509757881843e-05,
      "loss": 1.3568,
      "step": 3864
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.12725320200854,
      "learning_rate": 1.9182993364559996e-05,
      "loss": 1.6745,
      "step": 3865
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.633506535689906,
      "learning_rate": 1.918247681494678e-05,
      "loss": 0.9801,
      "step": 3866
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.700271228741421,
      "learning_rate": 1.9181960109050993e-05,
      "loss": 1.3088,
      "step": 3867
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.464754223041804,
      "learning_rate": 1.918144324688143e-05,
      "loss": 1.2367,
      "step": 3868
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.913514429796562,
      "learning_rate": 1.9180926228446886e-05,
      "loss": 1.1111,
      "step": 3869
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.537275438626429,
      "learning_rate": 1.9180409053756158e-05,
      "loss": 1.4125,
      "step": 3870
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.738478179416889,
      "learning_rate": 1.9179891722818054e-05,
      "loss": 1.6166,
      "step": 3871
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.827985215295418,
      "learning_rate": 1.9179374235641375e-05,
      "loss": 1.3614,
      "step": 3872
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.622787491929182,
      "learning_rate": 1.9178856592234927e-05,
      "loss": 1.6012,
      "step": 3873
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.242189412616971,
      "learning_rate": 1.9178338792607522e-05,
      "loss": 1.1982,
      "step": 3874
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.4039197343953225,
      "learning_rate": 1.9177820836767974e-05,
      "loss": 1.4711,
      "step": 3875
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.390833561885452,
      "learning_rate": 1.9177302724725095e-05,
      "loss": 1.4224,
      "step": 3876
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.728029590502367,
      "learning_rate": 1.917678445648771e-05,
      "loss": 1.135,
      "step": 3877
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.0386771633276926,
      "learning_rate": 1.917626603206463e-05,
      "loss": 1.267,
      "step": 3878
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.722515838757173,
      "learning_rate": 1.9175747451464683e-05,
      "loss": 1.6525,
      "step": 3879
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.161512264660469,
      "learning_rate": 1.91752287146967e-05,
      "loss": 1.4847,
      "step": 3880
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.181849465314917,
      "learning_rate": 1.9174709821769504e-05,
      "loss": 1.3885,
      "step": 3881
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.502587012621865,
      "learning_rate": 1.9174190772691924e-05,
      "loss": 1.3475,
      "step": 3882
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.5929713414010696,
      "learning_rate": 1.9173671567472803e-05,
      "loss": 1.1267,
      "step": 3883
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.532333076128531,
      "learning_rate": 1.9173152206120968e-05,
      "loss": 1.4223,
      "step": 3884
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.403691751019014,
      "learning_rate": 1.9172632688645265e-05,
      "loss": 1.3165,
      "step": 3885
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.477158275100342,
      "learning_rate": 1.917211301505453e-05,
      "loss": 1.1999,
      "step": 3886
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.533871107985907,
      "learning_rate": 1.9171593185357614e-05,
      "loss": 1.1917,
      "step": 3887
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.247619270820096,
      "learning_rate": 1.9171073199563364e-05,
      "loss": 1.4068,
      "step": 3888
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.5892145983715045,
      "learning_rate": 1.9170553057680627e-05,
      "loss": 1.2391,
      "step": 3889
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.587957018158503,
      "learning_rate": 1.9170032759718252e-05,
      "loss": 1.2336,
      "step": 3890
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.026958200540149,
      "learning_rate": 1.9169512305685103e-05,
      "loss": 1.2288,
      "step": 3891
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.988735901117943,
      "learning_rate": 1.916899169559003e-05,
      "loss": 1.3029,
      "step": 3892
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.716828089818758,
      "learning_rate": 1.91684709294419e-05,
      "loss": 1.3467,
      "step": 3893
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.202088366155682,
      "learning_rate": 1.9167950007249572e-05,
      "loss": 1.3521,
      "step": 3894
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.609252817072948,
      "learning_rate": 1.9167428929021913e-05,
      "loss": 1.541,
      "step": 3895
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.057070051912977,
      "learning_rate": 1.916690769476779e-05,
      "loss": 1.4667,
      "step": 3896
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.507965385433437,
      "learning_rate": 1.916638630449608e-05,
      "loss": 1.5024,
      "step": 3897
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.215247349603151,
      "learning_rate": 1.916586475821565e-05,
      "loss": 1.3783,
      "step": 3898
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.033434260138074,
      "learning_rate": 1.916534305593538e-05,
      "loss": 1.4885,
      "step": 3899
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.304444790246893,
      "learning_rate": 1.916482119766415e-05,
      "loss": 1.1599,
      "step": 3900
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.072540441390557,
      "learning_rate": 1.9164299183410842e-05,
      "loss": 1.2316,
      "step": 3901
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.536055589870417,
      "learning_rate": 1.9163777013184334e-05,
      "loss": 1.3214,
      "step": 3902
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.061469530045128,
      "learning_rate": 1.9163254686993523e-05,
      "loss": 1.4698,
      "step": 3903
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.9952480850266365,
      "learning_rate": 1.916273220484729e-05,
      "loss": 1.1851,
      "step": 3904
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.005845166514685,
      "learning_rate": 1.9162209566754538e-05,
      "loss": 1.3878,
      "step": 3905
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.524911528632204,
      "learning_rate": 1.916168677272415e-05,
      "loss": 1.2975,
      "step": 3906
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.594218512839721,
      "learning_rate": 1.916116382276503e-05,
      "loss": 1.3037,
      "step": 3907
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.496392316228006,
      "learning_rate": 1.9160640716886078e-05,
      "loss": 1.2733,
      "step": 3908
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.016168457297015,
      "learning_rate": 1.9160117455096196e-05,
      "loss": 1.138,
      "step": 3909
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.395068242671233,
      "learning_rate": 1.9159594037404293e-05,
      "loss": 1.3595,
      "step": 3910
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.010449209608262,
      "learning_rate": 1.9159070463819272e-05,
      "loss": 1.2519,
      "step": 3911
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.6229324037390045,
      "learning_rate": 1.9158546734350052e-05,
      "loss": 1.1025,
      "step": 3912
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.6434831585598495,
      "learning_rate": 1.9158022849005537e-05,
      "loss": 1.4361,
      "step": 3913
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.24781432772015,
      "learning_rate": 1.915749880779465e-05,
      "loss": 1.4862,
      "step": 3914
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.355221530516276,
      "learning_rate": 1.9156974610726305e-05,
      "loss": 1.5043,
      "step": 3915
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.013331507711504,
      "learning_rate": 1.9156450257809435e-05,
      "loss": 1.3182,
      "step": 3916
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.915691744731051,
      "learning_rate": 1.915592574905295e-05,
      "loss": 1.3181,
      "step": 3917
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.356850204145977,
      "learning_rate": 1.9155401084465782e-05,
      "loss": 1.0183,
      "step": 3918
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.089190928528394,
      "learning_rate": 1.9154876264056863e-05,
      "loss": 1.2056,
      "step": 3919
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.897977324407091,
      "learning_rate": 1.915435128783512e-05,
      "loss": 1.5153,
      "step": 3920
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.6166049406825795,
      "learning_rate": 1.91538261558095e-05,
      "loss": 1.425,
      "step": 3921
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.75718428060117,
      "learning_rate": 1.9153300867988927e-05,
      "loss": 1.0657,
      "step": 3922
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.7666433423794246,
      "learning_rate": 1.9152775424382347e-05,
      "loss": 0.9287,
      "step": 3923
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.5582159572859675,
      "learning_rate": 1.9152249824998703e-05,
      "loss": 1.6585,
      "step": 3924
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.582317105012463,
      "learning_rate": 1.9151724069846937e-05,
      "loss": 1.1968,
      "step": 3925
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.916294923686202,
      "learning_rate": 1.9151198158936007e-05,
      "loss": 1.1931,
      "step": 3926
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.957959479994896,
      "learning_rate": 1.915067209227485e-05,
      "loss": 1.2911,
      "step": 3927
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.510601133539533,
      "learning_rate": 1.915014586987243e-05,
      "loss": 1.1864,
      "step": 3928
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.077513241994799,
      "learning_rate": 1.91496194917377e-05,
      "loss": 1.5012,
      "step": 3929
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.338665777749406,
      "learning_rate": 1.9149092957879617e-05,
      "loss": 1.0849,
      "step": 3930
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.979967275998465,
      "learning_rate": 1.9148566268307144e-05,
      "loss": 1.7117,
      "step": 3931
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.402761380258971,
      "learning_rate": 1.9148039423029244e-05,
      "loss": 1.3435,
      "step": 3932
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.397586508024166,
      "learning_rate": 1.9147512422054883e-05,
      "loss": 1.2432,
      "step": 3933
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.651701699204425,
      "learning_rate": 1.9146985265393034e-05,
      "loss": 1.2454,
      "step": 3934
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.218726204895244,
      "learning_rate": 1.914645795305267e-05,
      "loss": 1.1791,
      "step": 3935
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.255495130330186,
      "learning_rate": 1.914593048504276e-05,
      "loss": 1.2026,
      "step": 3936
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.677419869572245,
      "learning_rate": 1.914540286137228e-05,
      "loss": 1.2492,
      "step": 3937
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.672548882527528,
      "learning_rate": 1.9144875082050217e-05,
      "loss": 1.4637,
      "step": 3938
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.828162437648522,
      "learning_rate": 1.914434714708555e-05,
      "loss": 1.224,
      "step": 3939
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.025841275952991,
      "learning_rate": 1.914381905648727e-05,
      "loss": 1.1636,
      "step": 3940
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.0095438998974835,
      "learning_rate": 1.914329081026435e-05,
      "loss": 1.1343,
      "step": 3941
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.408204680472971,
      "learning_rate": 1.9142762408425797e-05,
      "loss": 1.3587,
      "step": 3942
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.882122571608768,
      "learning_rate": 1.9142233850980597e-05,
      "loss": 1.3185,
      "step": 3943
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.389104872503573,
      "learning_rate": 1.9141705137937744e-05,
      "loss": 1.2037,
      "step": 3944
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.811735139937446,
      "learning_rate": 1.914117626930624e-05,
      "loss": 1.3879,
      "step": 3945
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.451260668935955,
      "learning_rate": 1.9140647245095085e-05,
      "loss": 1.3158,
      "step": 3946
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.433150912453171,
      "learning_rate": 1.9140118065313283e-05,
      "loss": 1.3322,
      "step": 3947
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.096560605172103,
      "learning_rate": 1.9139588729969842e-05,
      "loss": 1.4495,
      "step": 3948
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.26722098626651,
      "learning_rate": 1.9139059239073768e-05,
      "loss": 1.3397,
      "step": 3949
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.389180427294422,
      "learning_rate": 1.913852959263407e-05,
      "loss": 1.402,
      "step": 3950
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.60419850154709,
      "learning_rate": 1.9137999790659773e-05,
      "loss": 1.434,
      "step": 3951
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.10981211237569,
      "learning_rate": 1.9137469833159885e-05,
      "loss": 1.3826,
      "step": 3952
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.031626752744284,
      "learning_rate": 1.9136939720143427e-05,
      "loss": 1.2693,
      "step": 3953
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.289170127508096,
      "learning_rate": 1.9136409451619424e-05,
      "loss": 1.0715,
      "step": 3954
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.743561585095742,
      "learning_rate": 1.91358790275969e-05,
      "loss": 1.3814,
      "step": 3955
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.860988125599921,
      "learning_rate": 1.9135348448084882e-05,
      "loss": 1.3517,
      "step": 3956
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.080799217369657,
      "learning_rate": 1.9134817713092398e-05,
      "loss": 1.2271,
      "step": 3957
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.566551905934027,
      "learning_rate": 1.9134286822628487e-05,
      "loss": 1.3495,
      "step": 3958
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.695460027798251,
      "learning_rate": 1.913375577670218e-05,
      "loss": 1.2411,
      "step": 3959
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.51972245641186,
      "learning_rate": 1.9133224575322512e-05,
      "loss": 1.4759,
      "step": 3960
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.697427461179988,
      "learning_rate": 1.9132693218498533e-05,
      "loss": 1.1836,
      "step": 3961
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.240553494703757,
      "learning_rate": 1.9132161706239283e-05,
      "loss": 1.2489,
      "step": 3962
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.0458551102618,
      "learning_rate": 1.9131630038553803e-05,
      "loss": 1.3072,
      "step": 3963
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.276745761285537,
      "learning_rate": 1.9131098215451146e-05,
      "loss": 1.401,
      "step": 3964
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.068671066868721,
      "learning_rate": 1.9130566236940363e-05,
      "loss": 1.3303,
      "step": 3965
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.753395219577663,
      "learning_rate": 1.913003410303051e-05,
      "loss": 1.347,
      "step": 3966
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.993137010254468,
      "learning_rate": 1.912950181373064e-05,
      "loss": 1.2497,
      "step": 3967
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.633084078479786,
      "learning_rate": 1.9128969369049813e-05,
      "loss": 1.3824,
      "step": 3968
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.235285590479765,
      "learning_rate": 1.9128436768997096e-05,
      "loss": 1.2654,
      "step": 3969
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.33870591397474,
      "learning_rate": 1.9127904013581548e-05,
      "loss": 1.0856,
      "step": 3970
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.910078706990259,
      "learning_rate": 1.912737110281224e-05,
      "loss": 1.2468,
      "step": 3971
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.687189801457925,
      "learning_rate": 1.9126838036698236e-05,
      "loss": 1.4292,
      "step": 3972
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.2162623685819955,
      "learning_rate": 1.9126304815248616e-05,
      "loss": 1.5462,
      "step": 3973
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.938850309744121,
      "learning_rate": 1.9125771438472453e-05,
      "loss": 1.2516,
      "step": 3974
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.351161935380967,
      "learning_rate": 1.912523790637882e-05,
      "loss": 1.3084,
      "step": 3975
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.519061067936587,
      "learning_rate": 1.91247042189768e-05,
      "loss": 1.0883,
      "step": 3976
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.858608754899617,
      "learning_rate": 1.9124170376275485e-05,
      "loss": 1.3124,
      "step": 3977
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.839305567671578,
      "learning_rate": 1.9123636378283948e-05,
      "loss": 1.4764,
      "step": 3978
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.880174468505113,
      "learning_rate": 1.9123102225011286e-05,
      "loss": 1.2123,
      "step": 3979
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.3924268671000775,
      "learning_rate": 1.9122567916466584e-05,
      "loss": 1.3654,
      "step": 3980
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.797929460736927,
      "learning_rate": 1.9122033452658938e-05,
      "loss": 1.2953,
      "step": 3981
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.462425674597181,
      "learning_rate": 1.912149883359745e-05,
      "loss": 1.3913,
      "step": 3982
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.7755814204741265,
      "learning_rate": 1.912096405929121e-05,
      "loss": 1.2286,
      "step": 3983
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.042746204710508,
      "learning_rate": 1.9120429129749323e-05,
      "loss": 1.351,
      "step": 3984
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.5600375864339,
      "learning_rate": 1.9119894044980895e-05,
      "loss": 1.2232,
      "step": 3985
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.954628936058114,
      "learning_rate": 1.9119358804995035e-05,
      "loss": 1.2885,
      "step": 3986
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.789663616420473,
      "learning_rate": 1.911882340980085e-05,
      "loss": 1.2644,
      "step": 3987
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.624606870472025,
      "learning_rate": 1.911828785940745e-05,
      "loss": 1.0498,
      "step": 3988
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.136777154300308,
      "learning_rate": 1.911775215382395e-05,
      "loss": 1.4429,
      "step": 3989
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.789210439437904,
      "learning_rate": 1.9117216293059472e-05,
      "loss": 1.0914,
      "step": 3990
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.820218220206617,
      "learning_rate": 1.911668027712313e-05,
      "loss": 1.1221,
      "step": 3991
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.692253755503386,
      "learning_rate": 1.911614410602405e-05,
      "loss": 1.3284,
      "step": 3992
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.939064276741031,
      "learning_rate": 1.9115607779771365e-05,
      "loss": 1.3062,
      "step": 3993
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.565265641101854,
      "learning_rate": 1.911507129837419e-05,
      "loss": 1.3111,
      "step": 3994
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.9710799451178005,
      "learning_rate": 1.9114534661841663e-05,
      "loss": 1.1121,
      "step": 3995
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.541410297033705,
      "learning_rate": 1.9113997870182915e-05,
      "loss": 1.1924,
      "step": 3996
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.484909188411632,
      "learning_rate": 1.9113460923407086e-05,
      "loss": 1.2646,
      "step": 3997
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.1294841701586735,
      "learning_rate": 1.9112923821523307e-05,
      "loss": 1.155,
      "step": 3998
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.060400873576505,
      "learning_rate": 1.911238656454073e-05,
      "loss": 1.0395,
      "step": 3999
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.981002062322405,
      "learning_rate": 1.9111849152468488e-05,
      "loss": 1.2215,
      "step": 4000
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.351345809523174,
      "learning_rate": 1.9111311585315733e-05,
      "loss": 1.6578,
      "step": 4001
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.2902344118359,
      "learning_rate": 1.9110773863091614e-05,
      "loss": 1.0908,
      "step": 4002
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.510523565078586,
      "learning_rate": 1.911023598580528e-05,
      "loss": 1.6234,
      "step": 4003
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.94672909636757,
      "learning_rate": 1.9109697953465893e-05,
      "loss": 1.2226,
      "step": 4004
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.13362832513854,
      "learning_rate": 1.9109159766082603e-05,
      "loss": 1.5722,
      "step": 4005
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.538853039712078,
      "learning_rate": 1.9108621423664568e-05,
      "loss": 1.5411,
      "step": 4006
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.465210824671681,
      "learning_rate": 1.9108082926220958e-05,
      "loss": 0.9195,
      "step": 4007
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.356385050736317,
      "learning_rate": 1.9107544273760937e-05,
      "loss": 1.2664,
      "step": 4008
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.916384097075544,
      "learning_rate": 1.9107005466293666e-05,
      "loss": 1.0919,
      "step": 4009
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.507429889435454,
      "learning_rate": 1.9106466503828317e-05,
      "loss": 1.0175,
      "step": 4010
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.342193800888606,
      "learning_rate": 1.910592738637407e-05,
      "loss": 1.0172,
      "step": 4011
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.4553139610356585,
      "learning_rate": 1.9105388113940093e-05,
      "loss": 1.4864,
      "step": 4012
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.208767929592601,
      "learning_rate": 1.9104848686535567e-05,
      "loss": 1.4763,
      "step": 4013
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.041356404532198,
      "learning_rate": 1.9104309104169674e-05,
      "loss": 1.2814,
      "step": 4014
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.531589273096803,
      "learning_rate": 1.9103769366851596e-05,
      "loss": 1.1889,
      "step": 4015
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.602721974375811,
      "learning_rate": 1.910322947459052e-05,
      "loss": 0.8348,
      "step": 4016
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.676311704820899,
      "learning_rate": 1.9102689427395633e-05,
      "loss": 1.5055,
      "step": 4017
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.442611143275396,
      "learning_rate": 1.910214922527613e-05,
      "loss": 1.4062,
      "step": 4018
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.998029811766504,
      "learning_rate": 1.91016088682412e-05,
      "loss": 1.2014,
      "step": 4019
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.638442250038815,
      "learning_rate": 1.9101068356300044e-05,
      "loss": 1.2616,
      "step": 4020
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.4308125467135975,
      "learning_rate": 1.910052768946186e-05,
      "loss": 1.3831,
      "step": 4021
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.700519019361235,
      "learning_rate": 1.909998686773585e-05,
      "loss": 1.4969,
      "step": 4022
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.849079879604387,
      "learning_rate": 1.909944589113122e-05,
      "loss": 1.4318,
      "step": 4023
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.108428068547263,
      "learning_rate": 1.909890475965717e-05,
      "loss": 1.3843,
      "step": 4024
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.595965653962027,
      "learning_rate": 1.9098363473322918e-05,
      "loss": 1.2431,
      "step": 4025
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.028703473267938,
      "learning_rate": 1.9097822032137676e-05,
      "loss": 1.1114,
      "step": 4026
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.4354098365999,
      "learning_rate": 1.9097280436110654e-05,
      "loss": 1.2692,
      "step": 4027
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.537722732343423,
      "learning_rate": 1.9096738685251078e-05,
      "loss": 1.3151,
      "step": 4028
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.999544876461522,
      "learning_rate": 1.909619677956816e-05,
      "loss": 1.4168,
      "step": 4029
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.478221563824725,
      "learning_rate": 1.9095654719071123e-05,
      "loss": 1.3935,
      "step": 4030
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.911313510294487,
      "learning_rate": 1.90951125037692e-05,
      "loss": 1.2622,
      "step": 4031
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.996130345867584,
      "learning_rate": 1.9094570133671614e-05,
      "loss": 1.0175,
      "step": 4032
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.479017377905313,
      "learning_rate": 1.9094027608787598e-05,
      "loss": 1.0729,
      "step": 4033
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.261298019559854,
      "learning_rate": 1.9093484929126383e-05,
      "loss": 1.147,
      "step": 4034
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.752432555672791,
      "learning_rate": 1.9092942094697213e-05,
      "loss": 1.3411,
      "step": 4035
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.300407204066554,
      "learning_rate": 1.9092399105509315e-05,
      "loss": 1.3357,
      "step": 4036
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.925035058879272,
      "learning_rate": 1.909185596157194e-05,
      "loss": 1.2917,
      "step": 4037
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5230609460436884,
      "learning_rate": 1.9091312662894325e-05,
      "loss": 1.0933,
      "step": 4038
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.702579278503415,
      "learning_rate": 1.909076920948572e-05,
      "loss": 1.3678,
      "step": 4039
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.982367437833771,
      "learning_rate": 1.909022560135538e-05,
      "loss": 1.1192,
      "step": 4040
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.729083469654225,
      "learning_rate": 1.9089681838512547e-05,
      "loss": 1.0381,
      "step": 4041
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.513821335628773,
      "learning_rate": 1.908913792096648e-05,
      "loss": 1.1477,
      "step": 4042
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.773388251653469,
      "learning_rate": 1.908859384872644e-05,
      "loss": 1.1699,
      "step": 4043
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.651860651937749,
      "learning_rate": 1.9088049621801684e-05,
      "loss": 1.0998,
      "step": 4044
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4349738027490617,
      "learning_rate": 1.9087505240201468e-05,
      "loss": 1.255,
      "step": 4045
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.104153851918781,
      "learning_rate": 1.908696070393507e-05,
      "loss": 1.1829,
      "step": 4046
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.635607179539315,
      "learning_rate": 1.9086416013011752e-05,
      "loss": 1.0639,
      "step": 4047
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.561638287270044,
      "learning_rate": 1.9085871167440775e-05,
      "loss": 1.3473,
      "step": 4048
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.543053350369894,
      "learning_rate": 1.908532616723143e-05,
      "loss": 1.3694,
      "step": 4049
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.7217636391201685,
      "learning_rate": 1.9084781012392975e-05,
      "loss": 1.3103,
      "step": 4050
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.852843198092934,
      "learning_rate": 1.9084235702934705e-05,
      "loss": 1.4944,
      "step": 4051
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.541005270850799,
      "learning_rate": 1.9083690238865887e-05,
      "loss": 1.3616,
      "step": 4052
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.197655348001722,
      "learning_rate": 1.9083144620195814e-05,
      "loss": 1.2814,
      "step": 4053
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.229290844987736,
      "learning_rate": 1.9082598846933767e-05,
      "loss": 1.5011,
      "step": 4054
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.242491562680106,
      "learning_rate": 1.9082052919089034e-05,
      "loss": 1.2355,
      "step": 4055
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.826152213947616,
      "learning_rate": 1.9081506836670912e-05,
      "loss": 1.5492,
      "step": 4056
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.967482843484629,
      "learning_rate": 1.908096059968869e-05,
      "loss": 1.2665,
      "step": 4057
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.929183704047125,
      "learning_rate": 1.9080414208151668e-05,
      "loss": 1.1782,
      "step": 4058
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.210670718484714,
      "learning_rate": 1.9079867662069144e-05,
      "loss": 1.2923,
      "step": 4059
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.475009653548973,
      "learning_rate": 1.907932096145042e-05,
      "loss": 1.2918,
      "step": 4060
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.032073573249129,
      "learning_rate": 1.90787741063048e-05,
      "loss": 1.1418,
      "step": 4061
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.925142894109444,
      "learning_rate": 1.9078227096641594e-05,
      "loss": 1.2424,
      "step": 4062
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.242719694152994,
      "learning_rate": 1.9077679932470113e-05,
      "loss": 1.2966,
      "step": 4063
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.246429605300468,
      "learning_rate": 1.9077132613799662e-05,
      "loss": 1.1688,
      "step": 4064
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.196638174202283,
      "learning_rate": 1.907658514063956e-05,
      "loss": 1.4808,
      "step": 4065
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.277122466137503,
      "learning_rate": 1.907603751299913e-05,
      "loss": 1.2063,
      "step": 4066
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.27510211772358,
      "learning_rate": 1.9075489730887685e-05,
      "loss": 1.1375,
      "step": 4067
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.115132460349061,
      "learning_rate": 1.907494179431455e-05,
      "loss": 1.1361,
      "step": 4068
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.622161903685695,
      "learning_rate": 1.9074393703289056e-05,
      "loss": 1.7304,
      "step": 4069
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.297744031137932,
      "learning_rate": 1.9073845457820522e-05,
      "loss": 1.0489,
      "step": 4070
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.848921161460434,
      "learning_rate": 1.907329705791829e-05,
      "loss": 1.1773,
      "step": 4071
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.609906984655678,
      "learning_rate": 1.907274850359168e-05,
      "loss": 1.6249,
      "step": 4072
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.632729223568407,
      "learning_rate": 1.9072199794850043e-05,
      "loss": 1.4483,
      "step": 4073
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.551924924177208,
      "learning_rate": 1.907165093170271e-05,
      "loss": 1.3681,
      "step": 4074
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.169654819379074,
      "learning_rate": 1.907110191415902e-05,
      "loss": 1.3127,
      "step": 4075
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.456541155587372,
      "learning_rate": 1.907055274222832e-05,
      "loss": 1.4727,
      "step": 4076
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.92732094597193,
      "learning_rate": 1.907000341591996e-05,
      "loss": 1.2995,
      "step": 4077
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.701744017646784,
      "learning_rate": 1.9069453935243286e-05,
      "loss": 1.2846,
      "step": 4078
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.537891497220597,
      "learning_rate": 1.9068904300207647e-05,
      "loss": 1.4928,
      "step": 4079
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.388034364632162,
      "learning_rate": 1.9068354510822402e-05,
      "loss": 1.2935,
      "step": 4080
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.881248442673398,
      "learning_rate": 1.906780456709691e-05,
      "loss": 1.1108,
      "step": 4081
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.691952758896949,
      "learning_rate": 1.9067254469040527e-05,
      "loss": 1.1865,
      "step": 4082
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.884892010250429,
      "learning_rate": 1.906670421666261e-05,
      "loss": 1.2785,
      "step": 4083
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.8612112472085025,
      "learning_rate": 1.9066153809972537e-05,
      "loss": 1.2009,
      "step": 4084
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.574998306287548,
      "learning_rate": 1.906560324897967e-05,
      "loss": 1.2508,
      "step": 4085
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.876479189937202,
      "learning_rate": 1.9065052533693376e-05,
      "loss": 1.2509,
      "step": 4086
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.962294925941931,
      "learning_rate": 1.9064501664123033e-05,
      "loss": 1.4141,
      "step": 4087
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.760471564740774,
      "learning_rate": 1.9063950640278014e-05,
      "loss": 1.3964,
      "step": 4088
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.203148923464315,
      "learning_rate": 1.9063399462167697e-05,
      "loss": 1.4512,
      "step": 4089
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.299949418464628,
      "learning_rate": 1.9062848129801458e-05,
      "loss": 1.1381,
      "step": 4090
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.119950786517242,
      "learning_rate": 1.9062296643188693e-05,
      "loss": 1.29,
      "step": 4091
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.908816551484177,
      "learning_rate": 1.9061745002338777e-05,
      "loss": 1.4608,
      "step": 4092
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.489005019678511,
      "learning_rate": 1.9061193207261102e-05,
      "loss": 1.1705,
      "step": 4093
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.973546667733247,
      "learning_rate": 1.9060641257965064e-05,
      "loss": 1.5166,
      "step": 4094
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.901476009621438,
      "learning_rate": 1.906008915446005e-05,
      "loss": 1.3733,
      "step": 4095
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.315880935154338,
      "learning_rate": 1.905953689675546e-05,
      "loss": 0.9727,
      "step": 4096
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.865692163863106,
      "learning_rate": 1.9058984484860695e-05,
      "loss": 1.2484,
      "step": 4097
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.393028979758367,
      "learning_rate": 1.905843191878515e-05,
      "loss": 1.1992,
      "step": 4098
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.726953082386982,
      "learning_rate": 1.9057879198538236e-05,
      "loss": 1.0842,
      "step": 4099
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.109123650366585,
      "learning_rate": 1.905732632412936e-05,
      "loss": 1.2144,
      "step": 4100
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.586129751350574,
      "learning_rate": 1.905677329556793e-05,
      "loss": 1.2536,
      "step": 4101
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.870874733055597,
      "learning_rate": 1.9056220112863356e-05,
      "loss": 1.3207,
      "step": 4102
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.667085183209596,
      "learning_rate": 1.905566677602506e-05,
      "loss": 1.3671,
      "step": 4103
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.198058525992584,
      "learning_rate": 1.905511328506245e-05,
      "loss": 1.0882,
      "step": 4104
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.889736921681723,
      "learning_rate": 1.9054559639984952e-05,
      "loss": 1.4496,
      "step": 4105
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.173461754686899,
      "learning_rate": 1.905400584080199e-05,
      "loss": 0.8316,
      "step": 4106
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.782626931693055,
      "learning_rate": 1.9053451887522988e-05,
      "loss": 1.1679,
      "step": 4107
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.24056730017825,
      "learning_rate": 1.9052897780157372e-05,
      "loss": 1.6221,
      "step": 4108
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.105805126529109,
      "learning_rate": 1.9052343518714572e-05,
      "loss": 1.1856,
      "step": 4109
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.200808170312423,
      "learning_rate": 1.9051789103204025e-05,
      "loss": 1.0288,
      "step": 4110
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.310740425012478,
      "learning_rate": 1.905123453363517e-05,
      "loss": 1.4674,
      "step": 4111
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.401006686130842,
      "learning_rate": 1.9050679810017438e-05,
      "loss": 1.2391,
      "step": 4112
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.4757669435233325,
      "learning_rate": 1.9050124932360273e-05,
      "loss": 1.4394,
      "step": 4113
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.851951259336822,
      "learning_rate": 1.9049569900673118e-05,
      "loss": 1.1741,
      "step": 4114
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.542666383593923,
      "learning_rate": 1.9049014714965422e-05,
      "loss": 1.069,
      "step": 4115
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.420199798812453,
      "learning_rate": 1.9048459375246634e-05,
      "loss": 1.3526,
      "step": 4116
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.554377579626602,
      "learning_rate": 1.90479038815262e-05,
      "loss": 1.1253,
      "step": 4117
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.889652536431984,
      "learning_rate": 1.9047348233813583e-05,
      "loss": 1.129,
      "step": 4118
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.502665099466654,
      "learning_rate": 1.9046792432118236e-05,
      "loss": 1.3682,
      "step": 4119
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.620223849385193,
      "learning_rate": 1.9046236476449618e-05,
      "loss": 1.4377,
      "step": 4120
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.402728737983948,
      "learning_rate": 1.904568036681719e-05,
      "loss": 1.4661,
      "step": 4121
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.046263441294367,
      "learning_rate": 1.9045124103230417e-05,
      "loss": 1.0424,
      "step": 4122
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.646090579412489,
      "learning_rate": 1.904456768569877e-05,
      "loss": 1.2779,
      "step": 4123
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.707571479826666,
      "learning_rate": 1.9044011114231713e-05,
      "loss": 1.4202,
      "step": 4124
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.0289642625186595,
      "learning_rate": 1.9043454388838724e-05,
      "loss": 1.1951,
      "step": 4125
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.976845069796216,
      "learning_rate": 1.904289750952928e-05,
      "loss": 1.2837,
      "step": 4126
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.767008015357354,
      "learning_rate": 1.9042340476312852e-05,
      "loss": 1.2335,
      "step": 4127
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.452213153648672,
      "learning_rate": 1.9041783289198923e-05,
      "loss": 1.1895,
      "step": 4128
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.14924579669854,
      "learning_rate": 1.9041225948196977e-05,
      "loss": 1.019,
      "step": 4129
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.2480955391546855,
      "learning_rate": 1.9040668453316503e-05,
      "loss": 1.3784,
      "step": 4130
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.460208038320138,
      "learning_rate": 1.9040110804566986e-05,
      "loss": 1.1634,
      "step": 4131
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.332927046039073,
      "learning_rate": 1.9039553001957915e-05,
      "loss": 1.1034,
      "step": 4132
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.196843591382779,
      "learning_rate": 1.9038995045498786e-05,
      "loss": 1.5509,
      "step": 4133
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.087264325863853,
      "learning_rate": 1.9038436935199095e-05,
      "loss": 1.2561,
      "step": 4134
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.144802973655612,
      "learning_rate": 1.9037878671068343e-05,
      "loss": 1.331,
      "step": 4135
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.6076507022294555,
      "learning_rate": 1.9037320253116026e-05,
      "loss": 1.2651,
      "step": 4136
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.738789407328971,
      "learning_rate": 1.9036761681351653e-05,
      "loss": 1.1613,
      "step": 4137
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.157789660414331,
      "learning_rate": 1.9036202955784732e-05,
      "loss": 1.2435,
      "step": 4138
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.660465149977502,
      "learning_rate": 1.9035644076424766e-05,
      "loss": 1.377,
      "step": 4139
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.2396666040428705,
      "learning_rate": 1.9035085043281272e-05,
      "loss": 1.1554,
      "step": 4140
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.69115708202659,
      "learning_rate": 1.903452585636376e-05,
      "loss": 1.2674,
      "step": 4141
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.371007482887036,
      "learning_rate": 1.9033966515681754e-05,
      "loss": 1.4455,
      "step": 4142
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.514316492059914,
      "learning_rate": 1.9033407021244768e-05,
      "loss": 1.3736,
      "step": 4143
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.2104188642885525,
      "learning_rate": 1.9032847373062323e-05,
      "loss": 1.3408,
      "step": 4144
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.111703791044323,
      "learning_rate": 1.9032287571143954e-05,
      "loss": 1.4128,
      "step": 4145
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.246914915422701,
      "learning_rate": 1.9031727615499174e-05,
      "loss": 1.3446,
      "step": 4146
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.48137919799662,
      "learning_rate": 1.903116750613753e-05,
      "loss": 1.1962,
      "step": 4147
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.510867209155573,
      "learning_rate": 1.9030607243068538e-05,
      "loss": 1.2643,
      "step": 4148
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.118363623608781,
      "learning_rate": 1.9030046826301746e-05,
      "loss": 1.3494,
      "step": 4149
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.267618543365224,
      "learning_rate": 1.9029486255846688e-05,
      "loss": 1.6005,
      "step": 4150
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.891015552381707,
      "learning_rate": 1.9028925531712903e-05,
      "loss": 1.3349,
      "step": 4151
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.724143422510819,
      "learning_rate": 1.9028364653909937e-05,
      "loss": 1.1487,
      "step": 4152
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.52367496957695,
      "learning_rate": 1.902780362244733e-05,
      "loss": 1.3595,
      "step": 4153
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.588870029794739,
      "learning_rate": 1.902724243733464e-05,
      "loss": 1.3281,
      "step": 4154
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.39501779418519,
      "learning_rate": 1.902668109858141e-05,
      "loss": 1.311,
      "step": 4155
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.898455721056232,
      "learning_rate": 1.90261196061972e-05,
      "loss": 1.3004,
      "step": 4156
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.357064018067873,
      "learning_rate": 1.902555796019156e-05,
      "loss": 1.4259,
      "step": 4157
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.492053233673439,
      "learning_rate": 1.902499616057406e-05,
      "loss": 1.2429,
      "step": 4158
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.22538259298864,
      "learning_rate": 1.9024434207354248e-05,
      "loss": 1.1377,
      "step": 4159
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.7177887618253695,
      "learning_rate": 1.9023872100541692e-05,
      "loss": 1.4894,
      "step": 4160
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.83359452997663,
      "learning_rate": 1.9023309840145966e-05,
      "loss": 1.1451,
      "step": 4161
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.390113412692883,
      "learning_rate": 1.9022747426176633e-05,
      "loss": 0.9082,
      "step": 4162
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.22015351576059,
      "learning_rate": 1.902218485864327e-05,
      "loss": 1.3903,
      "step": 4163
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.791844895773235,
      "learning_rate": 1.9021622137555445e-05,
      "loss": 1.3022,
      "step": 4164
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.757263776438203,
      "learning_rate": 1.9021059262922738e-05,
      "loss": 1.4691,
      "step": 4165
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.61741147073761,
      "learning_rate": 1.9020496234754732e-05,
      "loss": 1.2111,
      "step": 4166
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.718783430484432,
      "learning_rate": 1.9019933053061008e-05,
      "loss": 1.2667,
      "step": 4167
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.276472599988779,
      "learning_rate": 1.9019369717851148e-05,
      "loss": 1.1607,
      "step": 4168
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.700138567126512,
      "learning_rate": 1.9018806229134746e-05,
      "loss": 1.2683,
      "step": 4169
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.97917893400863,
      "learning_rate": 1.9018242586921386e-05,
      "loss": 1.1441,
      "step": 4170
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.674491098824714,
      "learning_rate": 1.9017678791220665e-05,
      "loss": 1.405,
      "step": 4171
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.811535977505021,
      "learning_rate": 1.9017114842042174e-05,
      "loss": 1.2406,
      "step": 4172
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.506173308628981,
      "learning_rate": 1.9016550739395517e-05,
      "loss": 1.3702,
      "step": 4173
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.773617095044542,
      "learning_rate": 1.9015986483290295e-05,
      "loss": 1.4584,
      "step": 4174
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.963510153289954,
      "learning_rate": 1.9015422073736106e-05,
      "loss": 1.253,
      "step": 4175
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.223371264488661,
      "learning_rate": 1.901485751074256e-05,
      "loss": 1.3695,
      "step": 4176
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.125777335669692,
      "learning_rate": 1.901429279431926e-05,
      "loss": 1.2943,
      "step": 4177
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.303999755082373,
      "learning_rate": 1.9013727924475826e-05,
      "loss": 1.1347,
      "step": 4178
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.968884518911134,
      "learning_rate": 1.9013162901221865e-05,
      "loss": 1.2601,
      "step": 4179
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.5777634966437795,
      "learning_rate": 1.9012597724567003e-05,
      "loss": 1.291,
      "step": 4180
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.410349299046547,
      "learning_rate": 1.9012032394520843e-05,
      "loss": 1.2015,
      "step": 4181
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.577495444980912,
      "learning_rate": 1.9011466911093022e-05,
      "loss": 1.1122,
      "step": 4182
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.409329341758364,
      "learning_rate": 1.9010901274293155e-05,
      "loss": 1.2332,
      "step": 4183
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.5128201706210715,
      "learning_rate": 1.9010335484130875e-05,
      "loss": 0.8123,
      "step": 4184
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.433498204616782,
      "learning_rate": 1.9009769540615805e-05,
      "loss": 1.1675,
      "step": 4185
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.581343348196284,
      "learning_rate": 1.9009203443757585e-05,
      "loss": 1.1597,
      "step": 4186
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.008334863854525,
      "learning_rate": 1.9008637193565844e-05,
      "loss": 1.268,
      "step": 4187
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.666244154036399,
      "learning_rate": 1.9008070790050222e-05,
      "loss": 1.3138,
      "step": 4188
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.513396578522295,
      "learning_rate": 1.9007504233220356e-05,
      "loss": 1.5344,
      "step": 4189
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.689918154723225,
      "learning_rate": 1.900693752308589e-05,
      "loss": 1.1962,
      "step": 4190
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.730261335277553,
      "learning_rate": 1.9006370659656475e-05,
      "loss": 0.9682,
      "step": 4191
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.780503003484608,
      "learning_rate": 1.9005803642941753e-05,
      "loss": 1.3375,
      "step": 4192
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.374370435815554,
      "learning_rate": 1.9005236472951374e-05,
      "loss": 1.2969,
      "step": 4193
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.197691962394294,
      "learning_rate": 1.900466914969499e-05,
      "loss": 1.3945,
      "step": 4194
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.963987768302976,
      "learning_rate": 1.900410167318226e-05,
      "loss": 1.1416,
      "step": 4195
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.003354891023903,
      "learning_rate": 1.900353404342284e-05,
      "loss": 1.2174,
      "step": 4196
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.304299760505899,
      "learning_rate": 1.9002966260426396e-05,
      "loss": 1.3525,
      "step": 4197
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.729549876712869,
      "learning_rate": 1.9002398324202583e-05,
      "loss": 1.2705,
      "step": 4198
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.768477788732904,
      "learning_rate": 1.9001830234761074e-05,
      "loss": 1.2811,
      "step": 4199
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.340185716854088,
      "learning_rate": 1.9001261992111535e-05,
      "loss": 1.0455,
      "step": 4200
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.127369634443601,
      "learning_rate": 1.9000693596263636e-05,
      "loss": 1.2094,
      "step": 4201
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.838092044002746,
      "learning_rate": 1.9000125047227053e-05,
      "loss": 1.1596,
      "step": 4202
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.931177384866107,
      "learning_rate": 1.8999556345011464e-05,
      "loss": 1.2924,
      "step": 4203
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.225745293601149,
      "learning_rate": 1.8998987489626545e-05,
      "loss": 1.1341,
      "step": 4204
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.735379037971675,
      "learning_rate": 1.8998418481081975e-05,
      "loss": 1.153,
      "step": 4205
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.758671824270253,
      "learning_rate": 1.8997849319387448e-05,
      "loss": 1.3788,
      "step": 4206
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.925834453560451,
      "learning_rate": 1.899728000455264e-05,
      "loss": 1.042,
      "step": 4207
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.390627915233522,
      "learning_rate": 1.899671053658725e-05,
      "loss": 1.1889,
      "step": 4208
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.5423754870824,
      "learning_rate": 1.8996140915500964e-05,
      "loss": 0.9527,
      "step": 4209
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.117016306735069,
      "learning_rate": 1.899557114130348e-05,
      "loss": 1.2741,
      "step": 4210
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.208528470147233,
      "learning_rate": 1.8995001214004487e-05,
      "loss": 1.2389,
      "step": 4211
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.4466878052555225,
      "learning_rate": 1.8994431133613695e-05,
      "loss": 0.9779,
      "step": 4212
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.537736222068103,
      "learning_rate": 1.8993860900140807e-05,
      "loss": 1.3798,
      "step": 4213
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.16920820506357,
      "learning_rate": 1.899329051359552e-05,
      "loss": 1.2395,
      "step": 4214
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.372975613187559,
      "learning_rate": 1.899271997398755e-05,
      "loss": 1.2014,
      "step": 4215
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.656028995646227,
      "learning_rate": 1.8992149281326603e-05,
      "loss": 1.3257,
      "step": 4216
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.5945922563721915,
      "learning_rate": 1.899157843562239e-05,
      "loss": 1.3995,
      "step": 4217
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.903911745808679,
      "learning_rate": 1.8991007436884633e-05,
      "loss": 1.4616,
      "step": 4218
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.415036367593793,
      "learning_rate": 1.8990436285123042e-05,
      "loss": 1.2587,
      "step": 4219
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.2093674135774135,
      "learning_rate": 1.898986498034735e-05,
      "loss": 1.3395,
      "step": 4220
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.123614662395101,
      "learning_rate": 1.8989293522567265e-05,
      "loss": 1.3079,
      "step": 4221
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.206658805395513,
      "learning_rate": 1.8988721911792523e-05,
      "loss": 1.3163,
      "step": 4222
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.888924777133074,
      "learning_rate": 1.8988150148032854e-05,
      "loss": 1.1933,
      "step": 4223
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.965790676796993,
      "learning_rate": 1.898757823129798e-05,
      "loss": 1.1544,
      "step": 4224
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.155195982004498,
      "learning_rate": 1.8987006161597643e-05,
      "loss": 1.1177,
      "step": 4225
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.097808316632964,
      "learning_rate": 1.8986433938941577e-05,
      "loss": 1.5003,
      "step": 4226
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.7449644979935615,
      "learning_rate": 1.898586156333952e-05,
      "loss": 1.2691,
      "step": 4227
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.103774421891713,
      "learning_rate": 1.8985289034801212e-05,
      "loss": 1.243,
      "step": 4228
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.498461143290428,
      "learning_rate": 1.8984716353336404e-05,
      "loss": 1.3659,
      "step": 4229
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.294445868735889,
      "learning_rate": 1.8984143518954838e-05,
      "loss": 1.3825,
      "step": 4230
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.757614800385568,
      "learning_rate": 1.898357053166626e-05,
      "loss": 1.4647,
      "step": 4231
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.556306499899737,
      "learning_rate": 1.898299739148043e-05,
      "loss": 0.8725,
      "step": 4232
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.793241043531141,
      "learning_rate": 1.89824240984071e-05,
      "loss": 1.1133,
      "step": 4233
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.018227606322503,
      "learning_rate": 1.8981850652456024e-05,
      "loss": 1.0434,
      "step": 4234
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.361342587091893,
      "learning_rate": 1.8981277053636963e-05,
      "loss": 1.4224,
      "step": 4235
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.147018316009382,
      "learning_rate": 1.898070330195968e-05,
      "loss": 1.5018,
      "step": 4236
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.098059028257036,
      "learning_rate": 1.898012939743394e-05,
      "loss": 1.6081,
      "step": 4237
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.150083170476777,
      "learning_rate": 1.8979555340069512e-05,
      "loss": 1.152,
      "step": 4238
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.995017937740366,
      "learning_rate": 1.8978981129876165e-05,
      "loss": 1.307,
      "step": 4239
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.7810164915567235,
      "learning_rate": 1.897840676686367e-05,
      "loss": 1.1702,
      "step": 4240
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.083915294583586,
      "learning_rate": 1.897783225104181e-05,
      "loss": 0.9839,
      "step": 4241
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.6308581010287515,
      "learning_rate": 1.8977257582420354e-05,
      "loss": 1.4667,
      "step": 4242
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.216692789242337,
      "learning_rate": 1.8976682761009086e-05,
      "loss": 0.992,
      "step": 4243
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.344948975694708,
      "learning_rate": 1.897610778681779e-05,
      "loss": 1.454,
      "step": 4244
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.594603256448575,
      "learning_rate": 1.897553265985625e-05,
      "loss": 1.2748,
      "step": 4245
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.175387913595413,
      "learning_rate": 1.8974957380134258e-05,
      "loss": 1.1561,
      "step": 4246
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.154112426744577,
      "learning_rate": 1.8974381947661604e-05,
      "loss": 1.0737,
      "step": 4247
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.022794834860591,
      "learning_rate": 1.8973806362448076e-05,
      "loss": 1.1573,
      "step": 4248
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.425517336601293,
      "learning_rate": 1.897323062450348e-05,
      "loss": 1.2308,
      "step": 4249
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.99113089443778,
      "learning_rate": 1.8972654733837605e-05,
      "loss": 1.2826,
      "step": 4250
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.732592895401215,
      "learning_rate": 1.897207869046026e-05,
      "loss": 1.1966,
      "step": 4251
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.7832659368942547,
      "learning_rate": 1.8971502494381244e-05,
      "loss": 1.0807,
      "step": 4252
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.661746696794355,
      "learning_rate": 1.8970926145610367e-05,
      "loss": 1.1946,
      "step": 4253
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.005951955564505,
      "learning_rate": 1.8970349644157438e-05,
      "loss": 1.195,
      "step": 4254
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.1558635044994485,
      "learning_rate": 1.8969772990032267e-05,
      "loss": 1.4774,
      "step": 4255
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.09245462302887,
      "learning_rate": 1.896919618324467e-05,
      "loss": 1.4298,
      "step": 4256
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.397221070844202,
      "learning_rate": 1.8968619223804464e-05,
      "loss": 1.0146,
      "step": 4257
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.000394952102271,
      "learning_rate": 1.8968042111721464e-05,
      "loss": 1.5147,
      "step": 4258
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.628687354439735,
      "learning_rate": 1.89674648470055e-05,
      "loss": 1.2967,
      "step": 4259
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.47750713636404,
      "learning_rate": 1.896688742966639e-05,
      "loss": 1.3162,
      "step": 4260
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.393961090797919,
      "learning_rate": 1.8966309859713972e-05,
      "loss": 1.2164,
      "step": 4261
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.278199991008887,
      "learning_rate": 1.896573213715806e-05,
      "loss": 1.14,
      "step": 4262
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.150660068598242,
      "learning_rate": 1.8965154262008504e-05,
      "loss": 1.5391,
      "step": 4263
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.995955815178639,
      "learning_rate": 1.8964576234275123e-05,
      "loss": 1.4985,
      "step": 4264
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.339509295519993,
      "learning_rate": 1.8963998053967767e-05,
      "loss": 1.2305,
      "step": 4265
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.222936851594538,
      "learning_rate": 1.896341972109627e-05,
      "loss": 1.3357,
      "step": 4266
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.773137705096902,
      "learning_rate": 1.8962841235670475e-05,
      "loss": 1.3041,
      "step": 4267
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.740417396603743,
      "learning_rate": 1.8962262597700233e-05,
      "loss": 1.1314,
      "step": 4268
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.373949504056772,
      "learning_rate": 1.8961683807195387e-05,
      "loss": 1.1384,
      "step": 4269
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.7772118498666343,
      "learning_rate": 1.896110486416579e-05,
      "loss": 1.0885,
      "step": 4270
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.665475800644459,
      "learning_rate": 1.8960525768621296e-05,
      "loss": 1.3285,
      "step": 4271
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.549486977308858,
      "learning_rate": 1.8959946520571755e-05,
      "loss": 1.4301,
      "step": 4272
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.838483660110163,
      "learning_rate": 1.8959367120027035e-05,
      "loss": 1.356,
      "step": 4273
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.684825200050149,
      "learning_rate": 1.895878756699699e-05,
      "loss": 1.1585,
      "step": 4274
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.263663518684972,
      "learning_rate": 1.8958207861491488e-05,
      "loss": 1.4996,
      "step": 4275
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.551195626193033,
      "learning_rate": 1.8957628003520394e-05,
      "loss": 1.2472,
      "step": 4276
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.697831817455648,
      "learning_rate": 1.8957047993093578e-05,
      "loss": 1.2591,
      "step": 4277
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.077108015498099,
      "learning_rate": 1.895646783022091e-05,
      "loss": 1.1692,
      "step": 4278
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.794474729322474,
      "learning_rate": 1.895588751491226e-05,
      "loss": 1.0932,
      "step": 4279
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.106190316816176,
      "learning_rate": 1.8955307047177513e-05,
      "loss": 1.1443,
      "step": 4280
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.73926457438282,
      "learning_rate": 1.8954726427026545e-05,
      "loss": 1.1674,
      "step": 4281
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.990018960399158,
      "learning_rate": 1.8954145654469238e-05,
      "loss": 1.4399,
      "step": 4282
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.793995788615614,
      "learning_rate": 1.8953564729515477e-05,
      "loss": 1.0895,
      "step": 4283
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.321907130397331,
      "learning_rate": 1.8952983652175145e-05,
      "loss": 1.1901,
      "step": 4284
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.096628903280536,
      "learning_rate": 1.895240242245814e-05,
      "loss": 1.4192,
      "step": 4285
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.436370493843423,
      "learning_rate": 1.8951821040374345e-05,
      "loss": 1.1864,
      "step": 4286
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.876592943922851,
      "learning_rate": 1.8951239505933663e-05,
      "loss": 1.3472,
      "step": 4287
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.36632982465595,
      "learning_rate": 1.8950657819145987e-05,
      "loss": 1.3469,
      "step": 4288
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.814468973651315,
      "learning_rate": 1.8950075980021215e-05,
      "loss": 1.2157,
      "step": 4289
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.715411853835615,
      "learning_rate": 1.8949493988569255e-05,
      "loss": 1.034,
      "step": 4290
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.947809600546948,
      "learning_rate": 1.894891184480001e-05,
      "loss": 1.3961,
      "step": 4291
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.709186018797437,
      "learning_rate": 1.894832954872339e-05,
      "loss": 1.2778,
      "step": 4292
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.162888226498755,
      "learning_rate": 1.8947747100349298e-05,
      "loss": 1.2603,
      "step": 4293
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.629955481142523,
      "learning_rate": 1.8947164499687655e-05,
      "loss": 1.2178,
      "step": 4294
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0233820489206,
      "learning_rate": 1.8946581746748374e-05,
      "loss": 1.3454,
      "step": 4295
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.927142539383322,
      "learning_rate": 1.8945998841541374e-05,
      "loss": 1.361,
      "step": 4296
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.168033191393253,
      "learning_rate": 1.8945415784076574e-05,
      "loss": 1.3914,
      "step": 4297
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.437293337136122,
      "learning_rate": 1.89448325743639e-05,
      "loss": 0.9796,
      "step": 4298
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.824913662082511,
      "learning_rate": 1.8944249212413276e-05,
      "loss": 1.4317,
      "step": 4299
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.028606151290398,
      "learning_rate": 1.894366569823463e-05,
      "loss": 1.336,
      "step": 4300
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.328452649476066,
      "learning_rate": 1.8943082031837897e-05,
      "loss": 1.2715,
      "step": 4301
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.794809047626491,
      "learning_rate": 1.8942498213233004e-05,
      "loss": 1.4298,
      "step": 4302
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.135302713747916,
      "learning_rate": 1.8941914242429897e-05,
      "loss": 1.3603,
      "step": 4303
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.365593516392254,
      "learning_rate": 1.8941330119438506e-05,
      "loss": 1.3723,
      "step": 4304
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.132494157221942,
      "learning_rate": 1.894074584426878e-05,
      "loss": 1.2061,
      "step": 4305
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.733069821896607,
      "learning_rate": 1.8940161416930656e-05,
      "loss": 1.3013,
      "step": 4306
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.9387316338394776,
      "learning_rate": 1.8939576837434084e-05,
      "loss": 1.241,
      "step": 4307
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.355142967288342,
      "learning_rate": 1.8938992105789012e-05,
      "loss": 1.153,
      "step": 4308
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.696270706624069,
      "learning_rate": 1.8938407222005398e-05,
      "loss": 1.4796,
      "step": 4309
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.613084490145604,
      "learning_rate": 1.893782218609319e-05,
      "loss": 1.1218,
      "step": 4310
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.41515307852449,
      "learning_rate": 1.8937236998062344e-05,
      "loss": 1.0516,
      "step": 4311
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.027167532917759,
      "learning_rate": 1.8936651657922827e-05,
      "loss": 1.2802,
      "step": 4312
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.397578294957767,
      "learning_rate": 1.8936066165684598e-05,
      "loss": 1.1003,
      "step": 4313
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.389898195355006,
      "learning_rate": 1.8935480521357616e-05,
      "loss": 1.3039,
      "step": 4314
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.776881115905419,
      "learning_rate": 1.8934894724951855e-05,
      "loss": 1.4138,
      "step": 4315
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.070534612381212,
      "learning_rate": 1.8934308776477283e-05,
      "loss": 1.2252,
      "step": 4316
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.3668608436410175,
      "learning_rate": 1.8933722675943874e-05,
      "loss": 1.2269,
      "step": 4317
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.138648330707551,
      "learning_rate": 1.8933136423361604e-05,
      "loss": 1.2831,
      "step": 4318
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.907119050083833,
      "learning_rate": 1.8932550018740447e-05,
      "loss": 0.9935,
      "step": 4319
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.649321260624856,
      "learning_rate": 1.8931963462090384e-05,
      "loss": 1.1005,
      "step": 4320
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.5461662683998645,
      "learning_rate": 1.89313767534214e-05,
      "loss": 1.358,
      "step": 4321
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.41675452869121,
      "learning_rate": 1.8930789892743485e-05,
      "loss": 1.2606,
      "step": 4322
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.461829900212319,
      "learning_rate": 1.8930202880066618e-05,
      "loss": 0.9678,
      "step": 4323
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.547301724679606,
      "learning_rate": 1.8929615715400792e-05,
      "loss": 1.3516,
      "step": 4324
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.193958606566885,
      "learning_rate": 1.8929028398756008e-05,
      "loss": 1.3388,
      "step": 4325
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.450831187389043,
      "learning_rate": 1.892844093014225e-05,
      "loss": 1.4321,
      "step": 4326
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.4640168248911465,
      "learning_rate": 1.8927853309569526e-05,
      "loss": 1.3913,
      "step": 4327
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.565248543857628,
      "learning_rate": 1.8927265537047832e-05,
      "loss": 1.3392,
      "step": 4328
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.852347757732991,
      "learning_rate": 1.8926677612587177e-05,
      "loss": 1.2285,
      "step": 4329
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.23178338552767,
      "learning_rate": 1.8926089536197563e-05,
      "loss": 1.226,
      "step": 4330
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.697973490525278,
      "learning_rate": 1.8925501307889e-05,
      "loss": 1.349,
      "step": 4331
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.859960368416996,
      "learning_rate": 1.8924912927671495e-05,
      "loss": 1.2498,
      "step": 4332
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.595200223930764,
      "learning_rate": 1.8924324395555066e-05,
      "loss": 1.2582,
      "step": 4333
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.144888384523135,
      "learning_rate": 1.8923735711549735e-05,
      "loss": 1.491,
      "step": 4334
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.4057908836843165,
      "learning_rate": 1.892314687566551e-05,
      "loss": 1.2136,
      "step": 4335
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0447616586800015,
      "learning_rate": 1.8922557887912424e-05,
      "loss": 1.509,
      "step": 4336
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.943509989891014,
      "learning_rate": 1.8921968748300495e-05,
      "loss": 1.5511,
      "step": 4337
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.694518776077535,
      "learning_rate": 1.8921379456839747e-05,
      "loss": 1.1658,
      "step": 4338
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.993000186310493,
      "learning_rate": 1.8920790013540218e-05,
      "loss": 1.3424,
      "step": 4339
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.766819295212575,
      "learning_rate": 1.8920200418411933e-05,
      "loss": 1.0428,
      "step": 4340
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.082405476770038,
      "learning_rate": 1.8919610671464933e-05,
      "loss": 1.0525,
      "step": 4341
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0136655816599385,
      "learning_rate": 1.891902077270925e-05,
      "loss": 1.3672,
      "step": 4342
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.545127992661719,
      "learning_rate": 1.8918430722154924e-05,
      "loss": 1.3403,
      "step": 4343
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.9411959338882334,
      "learning_rate": 1.8917840519812e-05,
      "loss": 1.2174,
      "step": 4344
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.034176306693609,
      "learning_rate": 1.8917250165690523e-05,
      "loss": 1.5069,
      "step": 4345
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.706656355508024,
      "learning_rate": 1.8916659659800535e-05,
      "loss": 1.0406,
      "step": 4346
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.377725795847853,
      "learning_rate": 1.8916069002152096e-05,
      "loss": 1.4078,
      "step": 4347
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.295479034387314,
      "learning_rate": 1.891547819275525e-05,
      "loss": 1.2969,
      "step": 4348
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.218547221210901,
      "learning_rate": 1.891488723162006e-05,
      "loss": 1.2081,
      "step": 4349
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.36522186300687,
      "learning_rate": 1.891429611875658e-05,
      "loss": 1.4318,
      "step": 4350
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.436941180292118,
      "learning_rate": 1.8913704854174867e-05,
      "loss": 1.2503,
      "step": 4351
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.136405922579668,
      "learning_rate": 1.8913113437884987e-05,
      "loss": 1.2039,
      "step": 4352
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.8755568653850085,
      "learning_rate": 1.891252186989701e-05,
      "loss": 1.2686,
      "step": 4353
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.400171292512557,
      "learning_rate": 1.8911930150221e-05,
      "loss": 1.3839,
      "step": 4354
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.855400962344863,
      "learning_rate": 1.8911338278867028e-05,
      "loss": 1.2567,
      "step": 4355
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.982819755432895,
      "learning_rate": 1.8910746255845168e-05,
      "loss": 1.2508,
      "step": 4356
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.231276469993783,
      "learning_rate": 1.8910154081165494e-05,
      "loss": 1.5624,
      "step": 4357
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8790626975499602,
      "learning_rate": 1.890956175483809e-05,
      "loss": 1.2836,
      "step": 4358
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.931027291510924,
      "learning_rate": 1.890896927687303e-05,
      "loss": 1.6742,
      "step": 4359
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.6187023880244125,
      "learning_rate": 1.8908376647280404e-05,
      "loss": 1.0636,
      "step": 4360
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.705412210480778,
      "learning_rate": 1.8907783866070296e-05,
      "loss": 1.2547,
      "step": 4361
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.181676317917846,
      "learning_rate": 1.8907190933252794e-05,
      "loss": 1.1798,
      "step": 4362
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.81482616461269,
      "learning_rate": 1.890659784883799e-05,
      "loss": 1.3226,
      "step": 4363
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.953661017174649,
      "learning_rate": 1.8906004612835977e-05,
      "loss": 1.2867,
      "step": 4364
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.144618242040318,
      "learning_rate": 1.8905411225256856e-05,
      "loss": 1.3516,
      "step": 4365
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.022810711668597,
      "learning_rate": 1.890481768611072e-05,
      "loss": 1.205,
      "step": 4366
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.071178090962952,
      "learning_rate": 1.890422399540768e-05,
      "loss": 1.3281,
      "step": 4367
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.751429598600528,
      "learning_rate": 1.890363015315783e-05,
      "loss": 1.3129,
      "step": 4368
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.597953196327177,
      "learning_rate": 1.8903036159371282e-05,
      "loss": 1.2595,
      "step": 4369
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.617677540853075,
      "learning_rate": 1.8902442014058144e-05,
      "loss": 1.0626,
      "step": 4370
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.074914047144023,
      "learning_rate": 1.890184771722853e-05,
      "loss": 1.2469,
      "step": 4371
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.881060101272912,
      "learning_rate": 1.8901253268892557e-05,
      "loss": 1.2362,
      "step": 4372
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.467233175883716,
      "learning_rate": 1.8900658669060336e-05,
      "loss": 1.4732,
      "step": 4373
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.745584029051529,
      "learning_rate": 1.890006391774199e-05,
      "loss": 1.2648,
      "step": 4374
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.582305153260924,
      "learning_rate": 1.8899469014947644e-05,
      "loss": 1.3636,
      "step": 4375
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.689762419806045,
      "learning_rate": 1.889887396068742e-05,
      "loss": 1.4946,
      "step": 4376
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.256219532959471,
      "learning_rate": 1.8898278754971445e-05,
      "loss": 1.3302,
      "step": 4377
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.493968695076601,
      "learning_rate": 1.8897683397809854e-05,
      "loss": 1.2045,
      "step": 4378
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.336877559634579,
      "learning_rate": 1.8897087889212772e-05,
      "loss": 1.3283,
      "step": 4379
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.570182029669496,
      "learning_rate": 1.889649222919034e-05,
      "loss": 1.3018,
      "step": 4380
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.045856246825819,
      "learning_rate": 1.8895896417752694e-05,
      "loss": 1.5388,
      "step": 4381
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.911413200736854,
      "learning_rate": 1.8895300454909976e-05,
      "loss": 1.2405,
      "step": 4382
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.085391023456443,
      "learning_rate": 1.889470434067233e-05,
      "loss": 1.3358,
      "step": 4383
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.417037606037555,
      "learning_rate": 1.88941080750499e-05,
      "loss": 1.3578,
      "step": 4384
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.712752240324182,
      "learning_rate": 1.889351165805283e-05,
      "loss": 0.9522,
      "step": 4385
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.154277962239984,
      "learning_rate": 1.8892915089691276e-05,
      "loss": 1.1169,
      "step": 4386
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.375079693148147,
      "learning_rate": 1.889231836997539e-05,
      "loss": 1.2072,
      "step": 4387
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.085273106848625,
      "learning_rate": 1.889172149891533e-05,
      "loss": 1.5563,
      "step": 4388
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.324312061107912,
      "learning_rate": 1.8891124476521253e-05,
      "loss": 1.1974,
      "step": 4389
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.394357947145928,
      "learning_rate": 1.889052730280332e-05,
      "loss": 1.2813,
      "step": 4390
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.690621147828054,
      "learning_rate": 1.888992997777169e-05,
      "loss": 1.0738,
      "step": 4391
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.034193477990593,
      "learning_rate": 1.888933250143654e-05,
      "loss": 1.289,
      "step": 4392
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.930537294931636,
      "learning_rate": 1.888873487380803e-05,
      "loss": 1.3259,
      "step": 4393
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.787568030138284,
      "learning_rate": 1.888813709489633e-05,
      "loss": 1.4088,
      "step": 4394
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.801297225045891,
      "learning_rate": 1.888753916471162e-05,
      "loss": 1.5406,
      "step": 4395
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.150690603786375,
      "learning_rate": 1.888694108326408e-05,
      "loss": 0.9393,
      "step": 4396
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.127693256107507,
      "learning_rate": 1.888634285056388e-05,
      "loss": 1.2916,
      "step": 4397
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.639459294737565,
      "learning_rate": 1.8885744466621206e-05,
      "loss": 1.338,
      "step": 4398
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.808000470405496,
      "learning_rate": 1.888514593144624e-05,
      "loss": 1.0546,
      "step": 4399
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.008501925098046,
      "learning_rate": 1.888454724504917e-05,
      "loss": 1.1262,
      "step": 4400
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.146473411509136,
      "learning_rate": 1.888394840744019e-05,
      "loss": 1.1292,
      "step": 4401
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.8389111086878,
      "learning_rate": 1.8883349418629487e-05,
      "loss": 1.7098,
      "step": 4402
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.79196616933261,
      "learning_rate": 1.8882750278627254e-05,
      "loss": 1.3485,
      "step": 4403
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.733437592596033,
      "learning_rate": 1.888215098744369e-05,
      "loss": 1.6601,
      "step": 4404
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.989369394058566,
      "learning_rate": 1.8881551545089e-05,
      "loss": 1.1521,
      "step": 4405
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.549303542577163,
      "learning_rate": 1.8880951951573378e-05,
      "loss": 1.3346,
      "step": 4406
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.332827649702478,
      "learning_rate": 1.8880352206907037e-05,
      "loss": 1.4142,
      "step": 4407
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.1733283381223085,
      "learning_rate": 1.8879752311100176e-05,
      "loss": 1.3729,
      "step": 4408
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.750456079645053,
      "learning_rate": 1.887915226416301e-05,
      "loss": 1.2957,
      "step": 4409
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.614484725518983,
      "learning_rate": 1.8878552066105752e-05,
      "loss": 1.2285,
      "step": 4410
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.194949779984246,
      "learning_rate": 1.8877951716938618e-05,
      "loss": 1.0329,
      "step": 4411
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6544631414064614,
      "learning_rate": 1.8877351216671817e-05,
      "loss": 1.3231,
      "step": 4412
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.4397153094455915,
      "learning_rate": 1.887675056531558e-05,
      "loss": 1.2889,
      "step": 4413
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.399622064545052,
      "learning_rate": 1.8876149762880123e-05,
      "loss": 0.9711,
      "step": 4414
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.007096995843496,
      "learning_rate": 1.887554880937568e-05,
      "loss": 1.2303,
      "step": 4415
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.859219769918822,
      "learning_rate": 1.887494770481247e-05,
      "loss": 1.2725,
      "step": 4416
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.3695845494951095,
      "learning_rate": 1.8874346449200727e-05,
      "loss": 1.0699,
      "step": 4417
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.642858324588418,
      "learning_rate": 1.8873745042550685e-05,
      "loss": 1.3867,
      "step": 4418
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.288058024253876,
      "learning_rate": 1.8873143484872577e-05,
      "loss": 1.3379,
      "step": 4419
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.345322234192439,
      "learning_rate": 1.8872541776176644e-05,
      "loss": 1.2178,
      "step": 4420
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.664321215138521,
      "learning_rate": 1.8871939916473126e-05,
      "loss": 1.2527,
      "step": 4421
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.94631377783516,
      "learning_rate": 1.8871337905772264e-05,
      "loss": 1.3158,
      "step": 4422
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.996948982237108,
      "learning_rate": 1.8870735744084313e-05,
      "loss": 1.4475,
      "step": 4423
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.251995088309851,
      "learning_rate": 1.8870133431419506e-05,
      "loss": 1.2211,
      "step": 4424
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.400090132349033,
      "learning_rate": 1.886953096778811e-05,
      "loss": 1.1054,
      "step": 4425
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.661036230029953,
      "learning_rate": 1.8868928353200367e-05,
      "loss": 1.0564,
      "step": 4426
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.217034556981774,
      "learning_rate": 1.886832558766654e-05,
      "loss": 0.9428,
      "step": 4427
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.796610734406143,
      "learning_rate": 1.886772267119689e-05,
      "loss": 1.1896,
      "step": 4428
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.995910908699163,
      "learning_rate": 1.8867119603801673e-05,
      "loss": 1.5249,
      "step": 4429
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.477827035439924,
      "learning_rate": 1.8866516385491152e-05,
      "loss": 1.4581,
      "step": 4430
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.389102548868236,
      "learning_rate": 1.8865913016275596e-05,
      "loss": 1.1133,
      "step": 4431
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.5281028767873845,
      "learning_rate": 1.8865309496165275e-05,
      "loss": 1.1693,
      "step": 4432
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.215969664441569,
      "learning_rate": 1.886470582517046e-05,
      "loss": 0.9721,
      "step": 4433
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.052442293107154,
      "learning_rate": 1.8864102003301426e-05,
      "loss": 1.2633,
      "step": 4434
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.382132576376022,
      "learning_rate": 1.8863498030568448e-05,
      "loss": 1.3172,
      "step": 4435
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.269198782687323,
      "learning_rate": 1.8862893906981805e-05,
      "loss": 1.4431,
      "step": 4436
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.033302687913594,
      "learning_rate": 1.886228963255178e-05,
      "loss": 1.0094,
      "step": 4437
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.0596317545262295,
      "learning_rate": 1.886168520728866e-05,
      "loss": 1.3735,
      "step": 4438
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.914637482848103,
      "learning_rate": 1.886108063120273e-05,
      "loss": 1.3126,
      "step": 4439
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.4171174897263095,
      "learning_rate": 1.886047590430428e-05,
      "loss": 1.2406,
      "step": 4440
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.591025540082884,
      "learning_rate": 1.88598710266036e-05,
      "loss": 1.4657,
      "step": 4441
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.631936261743646,
      "learning_rate": 1.8859265998110986e-05,
      "loss": 1.5003,
      "step": 4442
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.117394961107376,
      "learning_rate": 1.885866081883674e-05,
      "loss": 1.315,
      "step": 4443
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.73695124662593,
      "learning_rate": 1.8858055488791155e-05,
      "loss": 1.2914,
      "step": 4444
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.653257776719424,
      "learning_rate": 1.885745000798454e-05,
      "loss": 1.0961,
      "step": 4445
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.534803301492843,
      "learning_rate": 1.8856844376427194e-05,
      "loss": 1.1918,
      "step": 4446
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.629857229163721,
      "learning_rate": 1.8856238594129426e-05,
      "loss": 1.1383,
      "step": 4447
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.502217825642043,
      "learning_rate": 1.885563266110155e-05,
      "loss": 1.49,
      "step": 4448
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.161735519050004,
      "learning_rate": 1.8855026577353872e-05,
      "loss": 1.2604,
      "step": 4449
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.396950571496014,
      "learning_rate": 1.8854420342896716e-05,
      "loss": 1.3279,
      "step": 4450
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.100674797603801,
      "learning_rate": 1.8853813957740392e-05,
      "loss": 1.2933,
      "step": 4451
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.902191212169402,
      "learning_rate": 1.885320742189523e-05,
      "loss": 1.3177,
      "step": 4452
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.666627676107658,
      "learning_rate": 1.885260073537154e-05,
      "loss": 1.2332,
      "step": 4453
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.560875947090635,
      "learning_rate": 1.8851993898179655e-05,
      "loss": 1.2338,
      "step": 4454
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.24279625717281,
      "learning_rate": 1.885138691032991e-05,
      "loss": 1.4482,
      "step": 4455
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.561727693309757,
      "learning_rate": 1.8850779771832622e-05,
      "loss": 1.2817,
      "step": 4456
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.979381781168331,
      "learning_rate": 1.885017248269813e-05,
      "loss": 1.6861,
      "step": 4457
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.503194501713442,
      "learning_rate": 1.8849565042936775e-05,
      "loss": 1.2245,
      "step": 4458
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.531002107179251,
      "learning_rate": 1.884895745255889e-05,
      "loss": 1.187,
      "step": 4459
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.703881548803801,
      "learning_rate": 1.8848349711574813e-05,
      "loss": 1.1069,
      "step": 4460
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.250100002370529,
      "learning_rate": 1.8847741819994895e-05,
      "loss": 1.4136,
      "step": 4461
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.670462990005993,
      "learning_rate": 1.884713377782948e-05,
      "loss": 1.4004,
      "step": 4462
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.875641512910392,
      "learning_rate": 1.884652558508891e-05,
      "loss": 1.1917,
      "step": 4463
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.200519119639992,
      "learning_rate": 1.8845917241783547e-05,
      "loss": 1.2745,
      "step": 4464
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.73768108435634,
      "learning_rate": 1.8845308747923735e-05,
      "loss": 0.9994,
      "step": 4465
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.924002435505012,
      "learning_rate": 1.8844700103519836e-05,
      "loss": 1.5162,
      "step": 4466
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.2661329633140275,
      "learning_rate": 1.884409130858221e-05,
      "loss": 1.2029,
      "step": 4467
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.079354909110589,
      "learning_rate": 1.8843482363121213e-05,
      "loss": 1.4302,
      "step": 4468
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.220705856247863,
      "learning_rate": 1.8842873267147214e-05,
      "loss": 1.3525,
      "step": 4469
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.444105243229465,
      "learning_rate": 1.8842264020670578e-05,
      "loss": 1.2266,
      "step": 4470
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.7769081668323645,
      "learning_rate": 1.8841654623701673e-05,
      "loss": 1.2906,
      "step": 4471
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.053724802602392,
      "learning_rate": 1.8841045076250873e-05,
      "loss": 1.3564,
      "step": 4472
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6293095884343485,
      "learning_rate": 1.884043537832855e-05,
      "loss": 1.2075,
      "step": 4473
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.198617731542617,
      "learning_rate": 1.8839825529945076e-05,
      "loss": 1.652,
      "step": 4474
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.913149500051442,
      "learning_rate": 1.8839215531110842e-05,
      "loss": 1.3059,
      "step": 4475
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.262451722333823,
      "learning_rate": 1.8838605381836225e-05,
      "loss": 1.4842,
      "step": 4476
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.197614891059246,
      "learning_rate": 1.883799508213161e-05,
      "loss": 1.28,
      "step": 4477
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.084304776563417,
      "learning_rate": 1.8837384632007376e-05,
      "loss": 1.0778,
      "step": 4478
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.047925893665262,
      "learning_rate": 1.8836774031473923e-05,
      "loss": 1.2157,
      "step": 4479
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.888862402044357,
      "learning_rate": 1.883616328054164e-05,
      "loss": 1.3635,
      "step": 4480
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.256445800711791,
      "learning_rate": 1.883555237922092e-05,
      "loss": 1.3279,
      "step": 4481
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.813584618718623,
      "learning_rate": 1.8834941327522162e-05,
      "loss": 1.2737,
      "step": 4482
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.584289698864137,
      "learning_rate": 1.883433012545577e-05,
      "loss": 1.2498,
      "step": 4483
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.414363332104274,
      "learning_rate": 1.8833718773032136e-05,
      "loss": 1.1305,
      "step": 4484
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.0934537586672315,
      "learning_rate": 1.883310727026167e-05,
      "loss": 1.4291,
      "step": 4485
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.885027733210101,
      "learning_rate": 1.8832495617154784e-05,
      "loss": 1.2216,
      "step": 4486
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.977923895060429,
      "learning_rate": 1.8831883813721888e-05,
      "loss": 1.2047,
      "step": 4487
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.341002217793443,
      "learning_rate": 1.8831271859973386e-05,
      "loss": 1.4237,
      "step": 4488
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.299189659586374,
      "learning_rate": 1.88306597559197e-05,
      "loss": 1.4654,
      "step": 4489
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.985043440378593,
      "learning_rate": 1.8830047501571247e-05,
      "loss": 1.3946,
      "step": 4490
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.950037599983838,
      "learning_rate": 1.8829435096938447e-05,
      "loss": 1.4906,
      "step": 4491
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.250181536800797,
      "learning_rate": 1.8828822542031723e-05,
      "loss": 1.2741,
      "step": 4492
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.291412877206207,
      "learning_rate": 1.8828209836861496e-05,
      "loss": 1.5185,
      "step": 4493
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.834209368600863,
      "learning_rate": 1.8827596981438202e-05,
      "loss": 1.35,
      "step": 4494
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.145165080295693,
      "learning_rate": 1.882698397577227e-05,
      "loss": 1.0564,
      "step": 4495
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.323295911055921,
      "learning_rate": 1.8826370819874128e-05,
      "loss": 0.9857,
      "step": 4496
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.08787784288154,
      "learning_rate": 1.8825757513754215e-05,
      "loss": 1.5257,
      "step": 4497
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.2022767310068145,
      "learning_rate": 1.882514405742297e-05,
      "loss": 0.9796,
      "step": 4498
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.443515107702634,
      "learning_rate": 1.882453045089083e-05,
      "loss": 1.318,
      "step": 4499
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.319359484340461,
      "learning_rate": 1.8823916694168242e-05,
      "loss": 1.1088,
      "step": 4500
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.40264256631867,
      "learning_rate": 1.882330278726565e-05,
      "loss": 1.5257,
      "step": 4501
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.078610948200963,
      "learning_rate": 1.8822688730193506e-05,
      "loss": 1.4138,
      "step": 4502
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.329282975402542,
      "learning_rate": 1.8822074522962258e-05,
      "loss": 1.1031,
      "step": 4503
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.121290600714152,
      "learning_rate": 1.8821460165582358e-05,
      "loss": 1.2793,
      "step": 4504
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.63084566307773,
      "learning_rate": 1.8820845658064265e-05,
      "loss": 1.4674,
      "step": 4505
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.54951902816772,
      "learning_rate": 1.8820231000418442e-05,
      "loss": 1.2521,
      "step": 4506
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.486140188615365,
      "learning_rate": 1.8819616192655342e-05,
      "loss": 1.1938,
      "step": 4507
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.652051170633008,
      "learning_rate": 1.881900123478543e-05,
      "loss": 1.3378,
      "step": 4508
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.392778309910515,
      "learning_rate": 1.8818386126819173e-05,
      "loss": 1.239,
      "step": 4509
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.444398620839957,
      "learning_rate": 1.8817770868767046e-05,
      "loss": 1.4517,
      "step": 4510
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.79088154847275,
      "learning_rate": 1.8817155460639512e-05,
      "loss": 1.2352,
      "step": 4511
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.183225135734056,
      "learning_rate": 1.881653990244705e-05,
      "loss": 1.0793,
      "step": 4512
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.101924002078076,
      "learning_rate": 1.8815924194200138e-05,
      "loss": 1.08,
      "step": 4513
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.164645497525183,
      "learning_rate": 1.881530833590925e-05,
      "loss": 1.1646,
      "step": 4514
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.8367478485278,
      "learning_rate": 1.881469232758487e-05,
      "loss": 1.4225,
      "step": 4515
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.336827465241128,
      "learning_rate": 1.8814076169237485e-05,
      "loss": 1.2651,
      "step": 4516
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.18071839971492,
      "learning_rate": 1.8813459860877575e-05,
      "loss": 1.356,
      "step": 4517
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.830981049738741,
      "learning_rate": 1.8812843402515638e-05,
      "loss": 1.3964,
      "step": 4518
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.456459304913187,
      "learning_rate": 1.881222679416216e-05,
      "loss": 1.1463,
      "step": 4519
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.229189759046091,
      "learning_rate": 1.8811610035827636e-05,
      "loss": 1.2574,
      "step": 4520
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.781623820813202,
      "learning_rate": 1.8810993127522566e-05,
      "loss": 1.4418,
      "step": 4521
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.15274658246238,
      "learning_rate": 1.8810376069257445e-05,
      "loss": 1.5474,
      "step": 4522
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.396109744878242,
      "learning_rate": 1.880975886104278e-05,
      "loss": 1.0957,
      "step": 4523
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.244068499328444,
      "learning_rate": 1.8809141502889073e-05,
      "loss": 1.4457,
      "step": 4524
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.125810355867455,
      "learning_rate": 1.880852399480683e-05,
      "loss": 1.2754,
      "step": 4525
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.157993597198173,
      "learning_rate": 1.8807906336806564e-05,
      "loss": 1.3595,
      "step": 4526
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8368633356268362,
      "learning_rate": 1.8807288528898786e-05,
      "loss": 1.2655,
      "step": 4527
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.318300551986323,
      "learning_rate": 1.880667057109401e-05,
      "loss": 1.1486,
      "step": 4528
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.092027253668999,
      "learning_rate": 1.8806052463402755e-05,
      "loss": 1.1191,
      "step": 4529
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.022168399803412,
      "learning_rate": 1.8805434205835536e-05,
      "loss": 1.1899,
      "step": 4530
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.558288559325908,
      "learning_rate": 1.8804815798402886e-05,
      "loss": 1.4704,
      "step": 4531
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.490299621511517,
      "learning_rate": 1.880419724111532e-05,
      "loss": 0.9065,
      "step": 4532
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1961573242747745,
      "learning_rate": 1.8803578533983367e-05,
      "loss": 1.0668,
      "step": 4533
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.157320858413727,
      "learning_rate": 1.8802959677017562e-05,
      "loss": 0.8709,
      "step": 4534
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.830504418412042,
      "learning_rate": 1.8802340670228437e-05,
      "loss": 1.2743,
      "step": 4535
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.858982838307921,
      "learning_rate": 1.8801721513626523e-05,
      "loss": 1.2311,
      "step": 4536
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.406588774617404,
      "learning_rate": 1.880110220722236e-05,
      "loss": 1.319,
      "step": 4537
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.00581185794077,
      "learning_rate": 1.880048275102649e-05,
      "loss": 1.1021,
      "step": 4538
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.997605535728802,
      "learning_rate": 1.8799863145049454e-05,
      "loss": 1.4435,
      "step": 4539
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.891767821457742,
      "learning_rate": 1.8799243389301796e-05,
      "loss": 1.2838,
      "step": 4540
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.750310468618979,
      "learning_rate": 1.879862348379407e-05,
      "loss": 1.0947,
      "step": 4541
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4129831900261065,
      "learning_rate": 1.879800342853682e-05,
      "loss": 1.1225,
      "step": 4542
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.4702797657156195,
      "learning_rate": 1.87973832235406e-05,
      "loss": 1.2621,
      "step": 4543
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.623334244018804,
      "learning_rate": 1.8796762868815974e-05,
      "loss": 1.2064,
      "step": 4544
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.718570690838693,
      "learning_rate": 1.879614236437349e-05,
      "loss": 1.0247,
      "step": 4545
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.09042954782156,
      "learning_rate": 1.879552171022371e-05,
      "loss": 1.3439,
      "step": 4546
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.225182128387661,
      "learning_rate": 1.8794900906377202e-05,
      "loss": 1.391,
      "step": 4547
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.404872875788206,
      "learning_rate": 1.879427995284453e-05,
      "loss": 1.2436,
      "step": 4548
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.30569607372079,
      "learning_rate": 1.879365884963626e-05,
      "loss": 1.4946,
      "step": 4549
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.272256502943383,
      "learning_rate": 1.8793037596762967e-05,
      "loss": 0.8016,
      "step": 4550
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.10225632387139,
      "learning_rate": 1.8792416194235223e-05,
      "loss": 1.438,
      "step": 4551
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.320523792105241,
      "learning_rate": 1.8791794642063604e-05,
      "loss": 0.9384,
      "step": 4552
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.731900497466726,
      "learning_rate": 1.8791172940258685e-05,
      "loss": 1.263,
      "step": 4553
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.586534174244404,
      "learning_rate": 1.8790551088831054e-05,
      "loss": 1.3709,
      "step": 4554
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.212164205088169,
      "learning_rate": 1.878992908779129e-05,
      "loss": 1.3088,
      "step": 4555
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.881734470457721,
      "learning_rate": 1.8789306937149975e-05,
      "loss": 1.3407,
      "step": 4556
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.445264846656285,
      "learning_rate": 1.8788684636917712e-05,
      "loss": 1.2234,
      "step": 4557
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.121657642443756,
      "learning_rate": 1.8788062187105075e-05,
      "loss": 1.2993,
      "step": 4558
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.077963657685995,
      "learning_rate": 1.8787439587722673e-05,
      "loss": 1.2244,
      "step": 4559
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.505667260562164,
      "learning_rate": 1.8786816838781093e-05,
      "loss": 1.1706,
      "step": 4560
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.266002524489676,
      "learning_rate": 1.8786193940290937e-05,
      "loss": 1.0087,
      "step": 4561
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.4241248487741744,
      "learning_rate": 1.8785570892262806e-05,
      "loss": 1.0572,
      "step": 4562
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.2249844914321715,
      "learning_rate": 1.87849476947073e-05,
      "loss": 0.9661,
      "step": 4563
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.664364482199374,
      "learning_rate": 1.8784324347635035e-05,
      "loss": 1.3471,
      "step": 4564
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.388554141565434,
      "learning_rate": 1.8783700851056616e-05,
      "loss": 1.3395,
      "step": 4565
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.280304828072684,
      "learning_rate": 1.878307720498265e-05,
      "loss": 1.1092,
      "step": 4566
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.205254374877958,
      "learning_rate": 1.8782453409423756e-05,
      "loss": 1.322,
      "step": 4567
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.517086561841382,
      "learning_rate": 1.878182946439055e-05,
      "loss": 1.4162,
      "step": 4568
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.494534003979354,
      "learning_rate": 1.8781205369893653e-05,
      "loss": 1.4238,
      "step": 4569
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.547568806078674,
      "learning_rate": 1.8780581125943685e-05,
      "loss": 1.4186,
      "step": 4570
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.531575482961443,
      "learning_rate": 1.8779956732551267e-05,
      "loss": 0.9405,
      "step": 4571
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.936868487851518,
      "learning_rate": 1.877933218972703e-05,
      "loss": 1.22,
      "step": 4572
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.547228826696501,
      "learning_rate": 1.8778707497481602e-05,
      "loss": 1.4736,
      "step": 4573
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.908756792158426,
      "learning_rate": 1.877808265582562e-05,
      "loss": 1.1622,
      "step": 4574
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.672945452763215,
      "learning_rate": 1.8777457664769712e-05,
      "loss": 1.1965,
      "step": 4575
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.833688159377586,
      "learning_rate": 1.8776832524324518e-05,
      "loss": 1.3671,
      "step": 4576
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.9739441911370035,
      "learning_rate": 1.8776207234500676e-05,
      "loss": 1.1654,
      "step": 4577
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.127439467696708,
      "learning_rate": 1.8775581795308828e-05,
      "loss": 1.2655,
      "step": 4578
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.388913525756902,
      "learning_rate": 1.8774956206759626e-05,
      "loss": 1.163,
      "step": 4579
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6864056583035705,
      "learning_rate": 1.8774330468863707e-05,
      "loss": 1.279,
      "step": 4580
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6234500912637495,
      "learning_rate": 1.8773704581631725e-05,
      "loss": 1.1634,
      "step": 4581
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.410155215759155,
      "learning_rate": 1.8773078545074336e-05,
      "loss": 1.0441,
      "step": 4582
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.8050926723433145,
      "learning_rate": 1.877245235920219e-05,
      "loss": 1.3381,
      "step": 4583
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.124954067591887,
      "learning_rate": 1.8771826024025944e-05,
      "loss": 1.4467,
      "step": 4584
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.757296912547764,
      "learning_rate": 1.8771199539556262e-05,
      "loss": 1.3221,
      "step": 4585
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.23109552952346,
      "learning_rate": 1.8770572905803806e-05,
      "loss": 1.2868,
      "step": 4586
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.892327266522782,
      "learning_rate": 1.8769946122779235e-05,
      "loss": 1.4066,
      "step": 4587
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.6814255807880185,
      "learning_rate": 1.8769319190493226e-05,
      "loss": 1.4081,
      "step": 4588
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.463828364007628,
      "learning_rate": 1.8768692108956443e-05,
      "loss": 1.4741,
      "step": 4589
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.196970472837772,
      "learning_rate": 1.876806487817956e-05,
      "loss": 1.404,
      "step": 4590
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.358945799541403,
      "learning_rate": 1.8767437498173257e-05,
      "loss": 1.3551,
      "step": 4591
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.865053196910596,
      "learning_rate": 1.87668099689482e-05,
      "loss": 1.1321,
      "step": 4592
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.6755994415718805,
      "learning_rate": 1.8766182290515083e-05,
      "loss": 1.0784,
      "step": 4593
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.017823036368646,
      "learning_rate": 1.8765554462884576e-05,
      "loss": 1.0638,
      "step": 4594
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.138934463331966,
      "learning_rate": 1.8764926486067375e-05,
      "loss": 1.3486,
      "step": 4595
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.020186588290887,
      "learning_rate": 1.8764298360074167e-05,
      "loss": 1.0099,
      "step": 4596
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.1393466002442025,
      "learning_rate": 1.876367008491563e-05,
      "loss": 1.1537,
      "step": 4597
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.121708751944723,
      "learning_rate": 1.8763041660602478e-05,
      "loss": 1.231,
      "step": 4598
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.147660527714543,
      "learning_rate": 1.876241308714539e-05,
      "loss": 1.2973,
      "step": 4599
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.659432205867825,
      "learning_rate": 1.8761784364555066e-05,
      "loss": 1.0015,
      "step": 4600
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.859273407726427,
      "learning_rate": 1.8761155492842213e-05,
      "loss": 1.4491,
      "step": 4601
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.485255724409126,
      "learning_rate": 1.8760526472017528e-05,
      "loss": 1.1516,
      "step": 4602
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.162433020104971,
      "learning_rate": 1.875989730209172e-05,
      "loss": 1.3195,
      "step": 4603
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.129361773647051,
      "learning_rate": 1.8759267983075503e-05,
      "loss": 1.1531,
      "step": 4604
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.167393974714415,
      "learning_rate": 1.8758638514979573e-05,
      "loss": 1.346,
      "step": 4605
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.831129756704401,
      "learning_rate": 1.8758008897814657e-05,
      "loss": 1.3465,
      "step": 4606
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.780306454822088,
      "learning_rate": 1.8757379131591468e-05,
      "loss": 1.1159,
      "step": 4607
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.695916249574915,
      "learning_rate": 1.875674921632072e-05,
      "loss": 1.0702,
      "step": 4608
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.491460442180391,
      "learning_rate": 1.8756119152013134e-05,
      "loss": 1.1601,
      "step": 4609
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.9081847515599994,
      "learning_rate": 1.8755488938679433e-05,
      "loss": 1.187,
      "step": 4610
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.6560089705652965,
      "learning_rate": 1.875485857633035e-05,
      "loss": 1.0686,
      "step": 4611
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.424595659308349,
      "learning_rate": 1.875422806497661e-05,
      "loss": 1.391,
      "step": 4612
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.618950523358616,
      "learning_rate": 1.875359740462894e-05,
      "loss": 1.3955,
      "step": 4613
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.3840893587452685,
      "learning_rate": 1.8752966595298078e-05,
      "loss": 1.1914,
      "step": 4614
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.219649099067193,
      "learning_rate": 1.8752335636994762e-05,
      "loss": 1.2548,
      "step": 4615
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.6137060618783945,
      "learning_rate": 1.8751704529729726e-05,
      "loss": 1.5777,
      "step": 4616
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.449309697393977,
      "learning_rate": 1.8751073273513715e-05,
      "loss": 1.5551,
      "step": 4617
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.20166536894782,
      "learning_rate": 1.8750441868357464e-05,
      "loss": 1.3862,
      "step": 4618
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.082021441184097,
      "learning_rate": 1.8749810314271733e-05,
      "loss": 1.1982,
      "step": 4619
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.177859040929787,
      "learning_rate": 1.874917861126726e-05,
      "loss": 0.864,
      "step": 4620
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.521633611874776,
      "learning_rate": 1.8748546759354803e-05,
      "loss": 1.349,
      "step": 4621
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.414736798599506,
      "learning_rate": 1.874791475854511e-05,
      "loss": 1.5094,
      "step": 4622
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.396283329276596,
      "learning_rate": 1.8747282608848946e-05,
      "loss": 1.2148,
      "step": 4623
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.604300847576402,
      "learning_rate": 1.874665031027706e-05,
      "loss": 1.2823,
      "step": 4624
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.7904999295724355,
      "learning_rate": 1.8746017862840218e-05,
      "loss": 1.1947,
      "step": 4625
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.49728619137556,
      "learning_rate": 1.8745385266549184e-05,
      "loss": 1.3574,
      "step": 4626
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.852817692011626,
      "learning_rate": 1.8744752521414725e-05,
      "loss": 1.0126,
      "step": 4627
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.348966239069032,
      "learning_rate": 1.8744119627447612e-05,
      "loss": 1.3497,
      "step": 4628
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.2150873360854,
      "learning_rate": 1.8743486584658613e-05,
      "loss": 1.0525,
      "step": 4629
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.092850605431305,
      "learning_rate": 1.8742853393058505e-05,
      "loss": 1.3406,
      "step": 4630
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.096191779352404,
      "learning_rate": 1.8742220052658063e-05,
      "loss": 1.3553,
      "step": 4631
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.9512716695438055,
      "learning_rate": 1.8741586563468064e-05,
      "loss": 1.2562,
      "step": 4632
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.169173002548652,
      "learning_rate": 1.8740952925499295e-05,
      "loss": 1.1415,
      "step": 4633
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.41358987106939,
      "learning_rate": 1.8740319138762536e-05,
      "loss": 1.1705,
      "step": 4634
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.78606065719576,
      "learning_rate": 1.8739685203268573e-05,
      "loss": 1.4717,
      "step": 4635
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.469488269044575,
      "learning_rate": 1.87390511190282e-05,
      "loss": 1.1829,
      "step": 4636
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.577792405729955,
      "learning_rate": 1.8738416886052208e-05,
      "loss": 1.165,
      "step": 4637
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.252371663529462,
      "learning_rate": 1.873778250435139e-05,
      "loss": 1.1163,
      "step": 4638
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.298618564353992,
      "learning_rate": 1.873714797393654e-05,
      "loss": 1.3149,
      "step": 4639
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.281427872699856,
      "learning_rate": 1.8736513294818465e-05,
      "loss": 1.3689,
      "step": 4640
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.013751609211,
      "learning_rate": 1.873587846700796e-05,
      "loss": 1.1238,
      "step": 4641
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.1759704400751065,
      "learning_rate": 1.8735243490515832e-05,
      "loss": 1.2527,
      "step": 4642
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.37545654314104,
      "learning_rate": 1.873460836535289e-05,
      "loss": 1.2973,
      "step": 4643
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.911569698591918,
      "learning_rate": 1.8733973091529936e-05,
      "loss": 1.16,
      "step": 4644
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.52670946247441,
      "learning_rate": 1.8733337669057796e-05,
      "loss": 1.1358,
      "step": 4645
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.810140535345715,
      "learning_rate": 1.873270209794727e-05,
      "loss": 1.3117,
      "step": 4646
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.22342388165926,
      "learning_rate": 1.8732066378209186e-05,
      "loss": 1.5277,
      "step": 4647
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.573172407288157,
      "learning_rate": 1.873143050985436e-05,
      "loss": 1.0776,
      "step": 4648
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.626195102224377,
      "learning_rate": 1.8730794492893612e-05,
      "loss": 1.2823,
      "step": 4649
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.050868454277971,
      "learning_rate": 1.8730158327337767e-05,
      "loss": 1.1642,
      "step": 4650
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.583836315220715,
      "learning_rate": 1.8729522013197655e-05,
      "loss": 1.2055,
      "step": 4651
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.876269783971336,
      "learning_rate": 1.8728885550484105e-05,
      "loss": 1.2426,
      "step": 4652
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.839442632105295,
      "learning_rate": 1.872824893920795e-05,
      "loss": 1.2379,
      "step": 4653
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.869462641483248,
      "learning_rate": 1.8727612179380022e-05,
      "loss": 1.4272,
      "step": 4654
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.142222768693301,
      "learning_rate": 1.8726975271011163e-05,
      "loss": 1.3829,
      "step": 4655
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.477050488453169,
      "learning_rate": 1.8726338214112207e-05,
      "loss": 0.9291,
      "step": 4656
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.760471681900068,
      "learning_rate": 1.8725701008694002e-05,
      "loss": 1.2014,
      "step": 4657
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.23165387265446,
      "learning_rate": 1.872506365476739e-05,
      "loss": 1.3719,
      "step": 4658
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.69488356473868,
      "learning_rate": 1.8724426152343224e-05,
      "loss": 1.3387,
      "step": 4659
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.2598398998205225,
      "learning_rate": 1.8723788501432347e-05,
      "loss": 1.4809,
      "step": 4660
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.510469888872794,
      "learning_rate": 1.872315070204561e-05,
      "loss": 1.2569,
      "step": 4661
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.290037189522641,
      "learning_rate": 1.8722512754193872e-05,
      "loss": 1.3445,
      "step": 4662
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.722829418661295,
      "learning_rate": 1.8721874657887996e-05,
      "loss": 1.2164,
      "step": 4663
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.242417218377923,
      "learning_rate": 1.8721236413138837e-05,
      "loss": 1.3205,
      "step": 4664
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.03173622775544,
      "learning_rate": 1.8720598019957254e-05,
      "loss": 0.998,
      "step": 4665
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.1653729022379125,
      "learning_rate": 1.871995947835412e-05,
      "loss": 1.2475,
      "step": 4666
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.8132025519639505,
      "learning_rate": 1.8719320788340296e-05,
      "loss": 1.3997,
      "step": 4667
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.92939768599124,
      "learning_rate": 1.8718681949926656e-05,
      "loss": 1.0802,
      "step": 4668
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.52746340990976,
      "learning_rate": 1.8718042963124072e-05,
      "loss": 1.0331,
      "step": 4669
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.688522826851687,
      "learning_rate": 1.8717403827943416e-05,
      "loss": 1.3063,
      "step": 4670
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.223806733064901,
      "learning_rate": 1.871676454439557e-05,
      "loss": 1.1858,
      "step": 4671
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.469480515964721,
      "learning_rate": 1.871612511249142e-05,
      "loss": 1.1416,
      "step": 4672
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.680627864934252,
      "learning_rate": 1.8715485532241836e-05,
      "loss": 1.1724,
      "step": 4673
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.013016402554069,
      "learning_rate": 1.8714845803657714e-05,
      "loss": 1.2225,
      "step": 4674
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.495111014401297,
      "learning_rate": 1.8714205926749933e-05,
      "loss": 0.8749,
      "step": 4675
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.126560349128588,
      "learning_rate": 1.871356590152939e-05,
      "loss": 1.2205,
      "step": 4676
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.173737337865287,
      "learning_rate": 1.8712925728006976e-05,
      "loss": 1.2147,
      "step": 4677
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.973901495091993,
      "learning_rate": 1.8712285406193585e-05,
      "loss": 1.2448,
      "step": 4678
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.260395959744758,
      "learning_rate": 1.871164493610012e-05,
      "loss": 1.4158,
      "step": 4679
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.281084294582049,
      "learning_rate": 1.871100431773748e-05,
      "loss": 1.2511,
      "step": 4680
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.145637835699388,
      "learning_rate": 1.8710363551116565e-05,
      "loss": 1.3473,
      "step": 4681
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.264344050579813,
      "learning_rate": 1.8709722636248278e-05,
      "loss": 1.1152,
      "step": 4682
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.750085556906818,
      "learning_rate": 1.8709081573143537e-05,
      "loss": 0.9851,
      "step": 4683
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.074255591068009,
      "learning_rate": 1.8708440361813242e-05,
      "loss": 1.1979,
      "step": 4684
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.601059231004394,
      "learning_rate": 1.8707799002268315e-05,
      "loss": 1.2024,
      "step": 4685
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.548259819117604,
      "learning_rate": 1.8707157494519666e-05,
      "loss": 1.1473,
      "step": 4686
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.329652485437876,
      "learning_rate": 1.8706515838578216e-05,
      "loss": 1.0674,
      "step": 4687
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.190452401253011,
      "learning_rate": 1.8705874034454883e-05,
      "loss": 1.1267,
      "step": 4688
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.027125936288616,
      "learning_rate": 1.8705232082160593e-05,
      "loss": 1.3743,
      "step": 4689
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.522286997811931,
      "learning_rate": 1.8704589981706273e-05,
      "loss": 1.3945,
      "step": 4690
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.531727844668713,
      "learning_rate": 1.8703947733102846e-05,
      "loss": 1.2108,
      "step": 4691
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.358274275813516,
      "learning_rate": 1.8703305336361247e-05,
      "loss": 1.1512,
      "step": 4692
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.219741413152358,
      "learning_rate": 1.8702662791492413e-05,
      "loss": 1.2593,
      "step": 4693
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.649957748419576,
      "learning_rate": 1.8702020098507268e-05,
      "loss": 1.232,
      "step": 4694
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.387843288331881,
      "learning_rate": 1.8701377257416762e-05,
      "loss": 1.0747,
      "step": 4695
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.175779082559378,
      "learning_rate": 1.8700734268231833e-05,
      "loss": 1.239,
      "step": 4696
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.973256239595401,
      "learning_rate": 1.870009113096342e-05,
      "loss": 1.1369,
      "step": 4697
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.311416686397317,
      "learning_rate": 1.8699447845622473e-05,
      "loss": 1.2644,
      "step": 4698
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.296643287538725,
      "learning_rate": 1.869880441221994e-05,
      "loss": 1.1792,
      "step": 4699
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.731476684695116,
      "learning_rate": 1.8698160830766774e-05,
      "loss": 1.346,
      "step": 4700
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.965188805574261,
      "learning_rate": 1.869751710127392e-05,
      "loss": 1.1328,
      "step": 4701
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.404243091560286,
      "learning_rate": 1.8696873223752345e-05,
      "loss": 1.3173,
      "step": 4702
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.2761638602828445,
      "learning_rate": 1.8696229198213003e-05,
      "loss": 1.3608,
      "step": 4703
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.533222335586309,
      "learning_rate": 1.8695585024666853e-05,
      "loss": 1.4188,
      "step": 4704
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.243686410348452,
      "learning_rate": 1.869494070312486e-05,
      "loss": 1.3473,
      "step": 4705
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.041016589164226,
      "learning_rate": 1.8694296233597996e-05,
      "loss": 1.1876,
      "step": 4706
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.428335480553839,
      "learning_rate": 1.869365161609722e-05,
      "loss": 1.5705,
      "step": 4707
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.40341832280282,
      "learning_rate": 1.8693006850633507e-05,
      "loss": 0.8938,
      "step": 4708
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.358502935952054,
      "learning_rate": 1.869236193721783e-05,
      "loss": 1.371,
      "step": 4709
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.693763709805102,
      "learning_rate": 1.8691716875861168e-05,
      "loss": 1.3635,
      "step": 4710
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.971331743006845,
      "learning_rate": 1.86910716665745e-05,
      "loss": 1.3244,
      "step": 4711
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.055594484060847,
      "learning_rate": 1.8690426309368807e-05,
      "loss": 1.108,
      "step": 4712
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.397840098317773,
      "learning_rate": 1.8689780804255063e-05,
      "loss": 1.0809,
      "step": 4713
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.727465670946058,
      "learning_rate": 1.8689135151244268e-05,
      "loss": 1.3539,
      "step": 4714
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.3017894307857425,
      "learning_rate": 1.86884893503474e-05,
      "loss": 1.08,
      "step": 4715
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.776647696842687,
      "learning_rate": 1.8687843401575466e-05,
      "loss": 1.5007,
      "step": 4716
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.270051005794018,
      "learning_rate": 1.868719730493944e-05,
      "loss": 1.49,
      "step": 4717
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.215906620078886,
      "learning_rate": 1.8686551060450333e-05,
      "loss": 1.3041,
      "step": 4718
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.0026977732707705,
      "learning_rate": 1.8685904668119136e-05,
      "loss": 1.2974,
      "step": 4719
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.888073913380022,
      "learning_rate": 1.868525812795685e-05,
      "loss": 1.1872,
      "step": 4720
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.679202962121157,
      "learning_rate": 1.8684611439974484e-05,
      "loss": 1.252,
      "step": 4721
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.36601338831469,
      "learning_rate": 1.8683964604183042e-05,
      "loss": 1.1688,
      "step": 4722
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.191039974000724,
      "learning_rate": 1.8683317620593537e-05,
      "loss": 1.0833,
      "step": 4723
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.349593375540642,
      "learning_rate": 1.868267048921697e-05,
      "loss": 1.5754,
      "step": 4724
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.798529148609207,
      "learning_rate": 1.8682023210064365e-05,
      "loss": 1.3122,
      "step": 4725
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.650846067785629,
      "learning_rate": 1.8681375783146736e-05,
      "loss": 1.4649,
      "step": 4726
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.532019510973847,
      "learning_rate": 1.86807282084751e-05,
      "loss": 1.2885,
      "step": 4727
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.58891008154229,
      "learning_rate": 1.8680080486060478e-05,
      "loss": 1.5137,
      "step": 4728
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.446722554253076,
      "learning_rate": 1.8679432615913896e-05,
      "loss": 1.3099,
      "step": 4729
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.64313021717344,
      "learning_rate": 1.8678784598046377e-05,
      "loss": 1.4818,
      "step": 4730
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.613170697404204,
      "learning_rate": 1.867813643246896e-05,
      "loss": 1.1787,
      "step": 4731
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.296348684861599,
      "learning_rate": 1.8677488119192663e-05,
      "loss": 1.4224,
      "step": 4732
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.47270479546295,
      "learning_rate": 1.8676839658228528e-05,
      "loss": 1.2251,
      "step": 4733
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.4365747261220365,
      "learning_rate": 1.867619104958759e-05,
      "loss": 1.2529,
      "step": 4734
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.415707330607988,
      "learning_rate": 1.867554229328089e-05,
      "loss": 1.3289,
      "step": 4735
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.4073800727616375,
      "learning_rate": 1.8674893389319467e-05,
      "loss": 1.3336,
      "step": 4736
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.657210902261495,
      "learning_rate": 1.867424433771437e-05,
      "loss": 1.3101,
      "step": 4737
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.131738628406179,
      "learning_rate": 1.8673595138476636e-05,
      "loss": 1.2415,
      "step": 4738
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.8460564948496945,
      "learning_rate": 1.8672945791617323e-05,
      "loss": 1.0397,
      "step": 4739
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.670050507897262,
      "learning_rate": 1.8672296297147476e-05,
      "loss": 1.0905,
      "step": 4740
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.850234097780103,
      "learning_rate": 1.8671646655078156e-05,
      "loss": 1.3881,
      "step": 4741
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.128882029633549,
      "learning_rate": 1.8670996865420413e-05,
      "loss": 1.2139,
      "step": 4742
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.093188129078133,
      "learning_rate": 1.8670346928185313e-05,
      "loss": 1.3208,
      "step": 4743
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.2276203921334226,
      "learning_rate": 1.866969684338391e-05,
      "loss": 1.1496,
      "step": 4744
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.605339787720142,
      "learning_rate": 1.8669046611027277e-05,
      "loss": 1.2104,
      "step": 4745
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.536707901802901,
      "learning_rate": 1.8668396231126473e-05,
      "loss": 1.5519,
      "step": 4746
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.814873219401962,
      "learning_rate": 1.866774570369257e-05,
      "loss": 1.4945,
      "step": 4747
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.066076103827806,
      "learning_rate": 1.8667095028736642e-05,
      "loss": 0.982,
      "step": 4748
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.250746764802006,
      "learning_rate": 1.866644420626976e-05,
      "loss": 1.1762,
      "step": 4749
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.199243605856641,
      "learning_rate": 1.8665793236303006e-05,
      "loss": 1.4328,
      "step": 4750
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.475364890457717,
      "learning_rate": 1.866514211884745e-05,
      "loss": 1.0401,
      "step": 4751
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.1999236409403,
      "learning_rate": 1.8664490853914182e-05,
      "loss": 1.1663,
      "step": 4752
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.018600678973663,
      "learning_rate": 1.8663839441514283e-05,
      "loss": 1.0869,
      "step": 4753
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.6438041189981245,
      "learning_rate": 1.866318788165884e-05,
      "loss": 1.0815,
      "step": 4754
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.579619104315967,
      "learning_rate": 1.8662536174358945e-05,
      "loss": 1.2675,
      "step": 4755
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.082688931156475,
      "learning_rate": 1.8661884319625685e-05,
      "loss": 1.212,
      "step": 4756
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.973808906955773,
      "learning_rate": 1.866123231747016e-05,
      "loss": 1.2132,
      "step": 4757
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.0846188113759565,
      "learning_rate": 1.866058016790346e-05,
      "loss": 1.4228,
      "step": 4758
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.823703889406248,
      "learning_rate": 1.8659927870936686e-05,
      "loss": 1.2,
      "step": 4759
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.121018575627162,
      "learning_rate": 1.865927542658095e-05,
      "loss": 1.2867,
      "step": 4760
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.892283770062672,
      "learning_rate": 1.865862283484734e-05,
      "loss": 1.2995,
      "step": 4761
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.967658062571588,
      "learning_rate": 1.8657970095746972e-05,
      "loss": 1.2397,
      "step": 4762
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.033827717678753,
      "learning_rate": 1.865731720929096e-05,
      "loss": 1.195,
      "step": 4763
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.079527170649566,
      "learning_rate": 1.8656664175490406e-05,
      "loss": 1.0242,
      "step": 4764
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.193608503157321,
      "learning_rate": 1.8656010994356432e-05,
      "loss": 1.2882,
      "step": 4765
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.593566102195246,
      "learning_rate": 1.865535766590015e-05,
      "loss": 1.2099,
      "step": 4766
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.40639838263613,
      "learning_rate": 1.865470419013268e-05,
      "loss": 1.1803,
      "step": 4767
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.541172416750155,
      "learning_rate": 1.8654050567065148e-05,
      "loss": 1.3586,
      "step": 4768
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.3537612936978425,
      "learning_rate": 1.8653396796708672e-05,
      "loss": 1.3302,
      "step": 4769
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.376515296397973,
      "learning_rate": 1.8652742879074384e-05,
      "loss": 1.4052,
      "step": 4770
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.835746066774642,
      "learning_rate": 1.8652088814173413e-05,
      "loss": 1.2548,
      "step": 4771
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.773991329597739,
      "learning_rate": 1.8651434602016892e-05,
      "loss": 0.8248,
      "step": 4772
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.663289873178941,
      "learning_rate": 1.865078024261595e-05,
      "loss": 1.3598,
      "step": 4773
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.507649548223041,
      "learning_rate": 1.865012573598173e-05,
      "loss": 1.3281,
      "step": 4774
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.913391200709237,
      "learning_rate": 1.8649471082125366e-05,
      "loss": 1.2519,
      "step": 4775
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.923231398853915,
      "learning_rate": 1.8648816281058004e-05,
      "loss": 1.3954,
      "step": 4776
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.1202108441893195,
      "learning_rate": 1.8648161332790787e-05,
      "loss": 1.0999,
      "step": 4777
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.891505809344293,
      "learning_rate": 1.864750623733486e-05,
      "loss": 1.3611,
      "step": 4778
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.676818872878414,
      "learning_rate": 1.864685099470138e-05,
      "loss": 1.1684,
      "step": 4779
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.808030408013159,
      "learning_rate": 1.8646195604901492e-05,
      "loss": 1.2939,
      "step": 4780
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.494419190993765,
      "learning_rate": 1.8645540067946353e-05,
      "loss": 1.0693,
      "step": 4781
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.263464120169202,
      "learning_rate": 1.8644884383847117e-05,
      "loss": 1.3863,
      "step": 4782
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.291149729998294,
      "learning_rate": 1.8644228552614944e-05,
      "loss": 1.0403,
      "step": 4783
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.062613785360267,
      "learning_rate": 1.8643572574261002e-05,
      "loss": 1.2038,
      "step": 4784
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.206786832677383,
      "learning_rate": 1.864291644879645e-05,
      "loss": 1.2475,
      "step": 4785
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.783698476757297,
      "learning_rate": 1.8642260176232455e-05,
      "loss": 1.1877,
      "step": 4786
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.331710295915455,
      "learning_rate": 1.8641603756580192e-05,
      "loss": 1.4113,
      "step": 4787
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.069246630342143,
      "learning_rate": 1.8640947189850826e-05,
      "loss": 1.1039,
      "step": 4788
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.554184198171185,
      "learning_rate": 1.8640290476055533e-05,
      "loss": 1.4079,
      "step": 4789
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.3529131806517505,
      "learning_rate": 1.8639633615205495e-05,
      "loss": 0.9639,
      "step": 4790
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.715335087719633,
      "learning_rate": 1.8638976607311886e-05,
      "loss": 1.196,
      "step": 4791
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.87440928256434,
      "learning_rate": 1.863831945238589e-05,
      "loss": 1.5349,
      "step": 4792
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.938206153473381,
      "learning_rate": 1.8637662150438695e-05,
      "loss": 1.5509,
      "step": 4793
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.15524131754018,
      "learning_rate": 1.863700470148148e-05,
      "loss": 1.2832,
      "step": 4794
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.148282200659404,
      "learning_rate": 1.8636347105525443e-05,
      "loss": 1.2743,
      "step": 4795
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.864419631367674,
      "learning_rate": 1.863568936258177e-05,
      "loss": 1.2384,
      "step": 4796
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.278371855908812,
      "learning_rate": 1.863503147266166e-05,
      "loss": 1.3792,
      "step": 4797
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.49037575877421,
      "learning_rate": 1.8634373435776305e-05,
      "loss": 1.4447,
      "step": 4798
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.904159189825487,
      "learning_rate": 1.863371525193691e-05,
      "loss": 1.4993,
      "step": 4799
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.776027675632459,
      "learning_rate": 1.8633056921154672e-05,
      "loss": 1.1887,
      "step": 4800
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.187091617505535,
      "learning_rate": 1.8632398443440802e-05,
      "loss": 1.0362,
      "step": 4801
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.588420278848564,
      "learning_rate": 1.86317398188065e-05,
      "loss": 1.0866,
      "step": 4802
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.6647837976013165,
      "learning_rate": 1.863108104726298e-05,
      "loss": 1.4587,
      "step": 4803
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.492153134442642,
      "learning_rate": 1.8630422128821454e-05,
      "loss": 1.1454,
      "step": 4804
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.802414360470013,
      "learning_rate": 1.8629763063493134e-05,
      "loss": 1.4279,
      "step": 4805
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.9288284096786965,
      "learning_rate": 1.862910385128924e-05,
      "loss": 1.3112,
      "step": 4806
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.704119453188851,
      "learning_rate": 1.8628444492220986e-05,
      "loss": 1.0877,
      "step": 4807
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.995887664956757,
      "learning_rate": 1.86277849862996e-05,
      "loss": 1.2255,
      "step": 4808
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.166099406459812,
      "learning_rate": 1.8627125333536308e-05,
      "loss": 1.1293,
      "step": 4809
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.343571635816863,
      "learning_rate": 1.862646553394233e-05,
      "loss": 1.1286,
      "step": 4810
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.985775864482364,
      "learning_rate": 1.8625805587528904e-05,
      "loss": 1.2365,
      "step": 4811
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.689721228105725,
      "learning_rate": 1.862514549430725e-05,
      "loss": 1.1879,
      "step": 4812
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.069639299293195,
      "learning_rate": 1.8624485254288615e-05,
      "loss": 1.3757,
      "step": 4813
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.199580740781717,
      "learning_rate": 1.862382486748423e-05,
      "loss": 1.1734,
      "step": 4814
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.120862349707271,
      "learning_rate": 1.8623164333905338e-05,
      "loss": 1.0758,
      "step": 4815
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.15592037595946,
      "learning_rate": 1.8622503653563173e-05,
      "loss": 1.2398,
      "step": 4816
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.881592619433803,
      "learning_rate": 1.862184282646899e-05,
      "loss": 1.4173,
      "step": 4817
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.796853809083868,
      "learning_rate": 1.8621181852634026e-05,
      "loss": 1.1753,
      "step": 4818
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.205943045916961,
      "learning_rate": 1.862052073206954e-05,
      "loss": 1.043,
      "step": 4819
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.100304994679119,
      "learning_rate": 1.861985946478678e-05,
      "loss": 1.3407,
      "step": 4820
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.755489154562663,
      "learning_rate": 1.8619198050796997e-05,
      "loss": 1.266,
      "step": 4821
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.493744874470877,
      "learning_rate": 1.8618536490111454e-05,
      "loss": 1.4557,
      "step": 4822
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.595080771834035,
      "learning_rate": 1.8617874782741407e-05,
      "loss": 1.0802,
      "step": 4823
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.388153791572185,
      "learning_rate": 1.861721292869812e-05,
      "loss": 1.3174,
      "step": 4824
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.434841545539097,
      "learning_rate": 1.8616550927992855e-05,
      "loss": 1.4697,
      "step": 4825
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.248928766233886,
      "learning_rate": 1.861588878063688e-05,
      "loss": 1.2175,
      "step": 4826
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.901263139385733,
      "learning_rate": 1.8615226486641464e-05,
      "loss": 1.2526,
      "step": 4827
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.006239180874968,
      "learning_rate": 1.861456404601788e-05,
      "loss": 1.2278,
      "step": 4828
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.847500450832106,
      "learning_rate": 1.861390145877741e-05,
      "loss": 1.1659,
      "step": 4829
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.80584992547266,
      "learning_rate": 1.8613238724931315e-05,
      "loss": 1.4008,
      "step": 4830
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.426613230165143,
      "learning_rate": 1.8612575844490886e-05,
      "loss": 0.9457,
      "step": 4831
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.1236759603202,
      "learning_rate": 1.86119128174674e-05,
      "loss": 1.2515,
      "step": 4832
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.7505908993379933,
      "learning_rate": 1.8611249643872145e-05,
      "loss": 1.1416,
      "step": 4833
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.466442378087132,
      "learning_rate": 1.861058632371641e-05,
      "loss": 1.2084,
      "step": 4834
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.154832628167044,
      "learning_rate": 1.8609922857011476e-05,
      "loss": 1.4096,
      "step": 4835
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.002153744121289,
      "learning_rate": 1.860925924376864e-05,
      "loss": 1.2256,
      "step": 4836
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.821181547160998,
      "learning_rate": 1.86085954839992e-05,
      "loss": 1.4562,
      "step": 4837
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.722170778633219,
      "learning_rate": 1.8607931577714448e-05,
      "loss": 1.2687,
      "step": 4838
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.759725447839159,
      "learning_rate": 1.8607267524925684e-05,
      "loss": 1.3116,
      "step": 4839
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.5230557663144975,
      "learning_rate": 1.860660332564421e-05,
      "loss": 1.3178,
      "step": 4840
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.757433690503789,
      "learning_rate": 1.8605938979881332e-05,
      "loss": 1.073,
      "step": 4841
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.598707789687225,
      "learning_rate": 1.8605274487648355e-05,
      "loss": 1.5515,
      "step": 4842
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.7984940617919145,
      "learning_rate": 1.8604609848956592e-05,
      "loss": 1.2201,
      "step": 4843
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.960996932815514,
      "learning_rate": 1.860394506381735e-05,
      "loss": 1.3589,
      "step": 4844
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9556005595880905,
      "learning_rate": 1.860328013224195e-05,
      "loss": 1.2378,
      "step": 4845
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.447257498384713,
      "learning_rate": 1.8602615054241698e-05,
      "loss": 1.1385,
      "step": 4846
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.734442285170665,
      "learning_rate": 1.8601949829827926e-05,
      "loss": 1.1862,
      "step": 4847
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.3463225657913815,
      "learning_rate": 1.860128445901195e-05,
      "loss": 1.4196,
      "step": 4848
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.735295704596837,
      "learning_rate": 1.8600618941805087e-05,
      "loss": 1.5105,
      "step": 4849
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.6766891958090975,
      "learning_rate": 1.8599953278218678e-05,
      "loss": 1.034,
      "step": 4850
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.845800929192445,
      "learning_rate": 1.8599287468264043e-05,
      "loss": 1.0754,
      "step": 4851
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.254493138577706,
      "learning_rate": 1.859862151195252e-05,
      "loss": 1.4979,
      "step": 4852
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.083320978406206,
      "learning_rate": 1.8597955409295435e-05,
      "loss": 1.3297,
      "step": 4853
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.47226686059179,
      "learning_rate": 1.8597289160304135e-05,
      "loss": 1.1258,
      "step": 4854
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.0556403894023045,
      "learning_rate": 1.859662276498995e-05,
      "loss": 0.9799,
      "step": 4855
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.466772433235342,
      "learning_rate": 1.8595956223364227e-05,
      "loss": 1.4965,
      "step": 4856
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.161583908604117,
      "learning_rate": 1.8595289535438313e-05,
      "loss": 1.2016,
      "step": 4857
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.874089943934934,
      "learning_rate": 1.8594622701223547e-05,
      "loss": 1.4652,
      "step": 4858
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.879669373744303,
      "learning_rate": 1.8593955720731284e-05,
      "loss": 1.2818,
      "step": 4859
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.590190458043105,
      "learning_rate": 1.859328859397287e-05,
      "loss": 0.9965,
      "step": 4860
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.737772243317405,
      "learning_rate": 1.859262132095967e-05,
      "loss": 1.2179,
      "step": 4861
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.742462165509475,
      "learning_rate": 1.8591953901703028e-05,
      "loss": 0.9621,
      "step": 4862
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.346595527850519,
      "learning_rate": 1.859128633621431e-05,
      "loss": 1.1401,
      "step": 4863
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.508671049098717,
      "learning_rate": 1.859061862450488e-05,
      "loss": 1.19,
      "step": 4864
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.684982391315792,
      "learning_rate": 1.8589950766586098e-05,
      "loss": 0.9082,
      "step": 4865
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.430807184769413,
      "learning_rate": 1.858928276246933e-05,
      "loss": 1.221,
      "step": 4866
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.075470382755096,
      "learning_rate": 1.858861461216595e-05,
      "loss": 1.2625,
      "step": 4867
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.807523917154261,
      "learning_rate": 1.8587946315687327e-05,
      "loss": 1.4001,
      "step": 4868
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.329045590205233,
      "learning_rate": 1.8587277873044832e-05,
      "loss": 1.248,
      "step": 4869
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.80783135384826,
      "learning_rate": 1.8586609284249845e-05,
      "loss": 1.1585,
      "step": 4870
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.717831259757224,
      "learning_rate": 1.858594054931375e-05,
      "loss": 1.4091,
      "step": 4871
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.02985272580209,
      "learning_rate": 1.8585271668247918e-05,
      "loss": 1.2069,
      "step": 4872
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.720288107516998,
      "learning_rate": 1.8584602641063744e-05,
      "loss": 1.0181,
      "step": 4873
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.399080979868415,
      "learning_rate": 1.8583933467772604e-05,
      "loss": 1.21,
      "step": 4874
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.9770937799087465,
      "learning_rate": 1.858326414838589e-05,
      "loss": 1.1564,
      "step": 4875
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.995687060218061,
      "learning_rate": 1.8582594682915002e-05,
      "loss": 1.0595,
      "step": 4876
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.450483826108329,
      "learning_rate": 1.8581925071371326e-05,
      "loss": 1.1849,
      "step": 4877
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.10970946674854,
      "learning_rate": 1.858125531376626e-05,
      "loss": 1.2444,
      "step": 4878
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.9021065889576505,
      "learning_rate": 1.8580585410111203e-05,
      "loss": 0.9461,
      "step": 4879
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.856278711207781,
      "learning_rate": 1.8579915360417556e-05,
      "loss": 0.9842,
      "step": 4880
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.986797212823269,
      "learning_rate": 1.857924516469673e-05,
      "loss": 1.2877,
      "step": 4881
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.588435111394594,
      "learning_rate": 1.857857482296012e-05,
      "loss": 1.3131,
      "step": 4882
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.343615242976199,
      "learning_rate": 1.857790433521914e-05,
      "loss": 1.3931,
      "step": 4883
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.502042508445991,
      "learning_rate": 1.8577233701485205e-05,
      "loss": 1.0414,
      "step": 4884
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.223020501939153,
      "learning_rate": 1.8576562921769727e-05,
      "loss": 1.2803,
      "step": 4885
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.481782721585964,
      "learning_rate": 1.857589199608412e-05,
      "loss": 1.287,
      "step": 4886
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.030920683006389,
      "learning_rate": 1.8575220924439805e-05,
      "loss": 1.2145,
      "step": 4887
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.16060356432405,
      "learning_rate": 1.85745497068482e-05,
      "loss": 1.3323,
      "step": 4888
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.6002218269859725,
      "learning_rate": 1.8573878343320734e-05,
      "loss": 1.4679,
      "step": 4889
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.651420465779763,
      "learning_rate": 1.8573206833868832e-05,
      "loss": 1.293,
      "step": 4890
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.7809131202827295,
      "learning_rate": 1.857253517850392e-05,
      "loss": 1.379,
      "step": 4891
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.971535161382314,
      "learning_rate": 1.8571863377237432e-05,
      "loss": 1.1716,
      "step": 4892
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.482091171353579,
      "learning_rate": 1.85711914300808e-05,
      "loss": 1.3958,
      "step": 4893
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.086356918943668,
      "learning_rate": 1.857051933704546e-05,
      "loss": 1.1768,
      "step": 4894
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.2387450614894,
      "learning_rate": 1.8569847098142856e-05,
      "loss": 1.2017,
      "step": 4895
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.591810515605802,
      "learning_rate": 1.8569174713384423e-05,
      "loss": 1.221,
      "step": 4896
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.681883975492516,
      "learning_rate": 1.8568502182781608e-05,
      "loss": 1.3842,
      "step": 4897
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.297753322349592,
      "learning_rate": 1.856782950634586e-05,
      "loss": 1.3155,
      "step": 4898
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.169940383647992,
      "learning_rate": 1.8567156684088616e-05,
      "loss": 1.3053,
      "step": 4899
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.091283678384803,
      "learning_rate": 1.856648371602134e-05,
      "loss": 1.4594,
      "step": 4900
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.354255528209761,
      "learning_rate": 1.8565810602155483e-05,
      "loss": 1.3114,
      "step": 4901
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.484746036595489,
      "learning_rate": 1.8565137342502496e-05,
      "loss": 1.5166,
      "step": 4902
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.554593337249125,
      "learning_rate": 1.856446393707384e-05,
      "loss": 1.1602,
      "step": 4903
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.634969094306412,
      "learning_rate": 1.8563790385880983e-05,
      "loss": 1.3846,
      "step": 4904
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.395913559727462,
      "learning_rate": 1.8563116688935375e-05,
      "loss": 1.0856,
      "step": 4905
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.828152307191678,
      "learning_rate": 1.8562442846248495e-05,
      "loss": 1.0387,
      "step": 4906
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.470281914151468,
      "learning_rate": 1.8561768857831802e-05,
      "loss": 1.2334,
      "step": 4907
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.438515434280546,
      "learning_rate": 1.8561094723696776e-05,
      "loss": 1.2373,
      "step": 4908
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.346184439167995,
      "learning_rate": 1.8560420443854884e-05,
      "loss": 1.3353,
      "step": 4909
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.2628086580650795,
      "learning_rate": 1.8559746018317603e-05,
      "loss": 1.2597,
      "step": 4910
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.507550559505195,
      "learning_rate": 1.855907144709641e-05,
      "loss": 1.3671,
      "step": 4911
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9946184783446776,
      "learning_rate": 1.855839673020279e-05,
      "loss": 1.2893,
      "step": 4912
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.553069372747117,
      "learning_rate": 1.855772186764823e-05,
      "loss": 0.8351,
      "step": 4913
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.264288976564596,
      "learning_rate": 1.8557046859444204e-05,
      "loss": 1.1057,
      "step": 4914
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.091333973259296,
      "learning_rate": 1.855637170560221e-05,
      "loss": 1.1846,
      "step": 4915
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.2464132178785885,
      "learning_rate": 1.8555696406133736e-05,
      "loss": 1.0988,
      "step": 4916
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.749325624671045,
      "learning_rate": 1.8555020961050273e-05,
      "loss": 1.3378,
      "step": 4917
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.529543426667807,
      "learning_rate": 1.8554345370363323e-05,
      "loss": 1.2738,
      "step": 4918
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.294208185826604,
      "learning_rate": 1.855366963408438e-05,
      "loss": 1.0611,
      "step": 4919
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.427671902762321,
      "learning_rate": 1.855299375222494e-05,
      "loss": 1.2723,
      "step": 4920
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.580751747904824,
      "learning_rate": 1.8552317724796513e-05,
      "loss": 0.952,
      "step": 4921
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.049578447339299,
      "learning_rate": 1.8551641551810607e-05,
      "loss": 1.2982,
      "step": 4922
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.6884778484588505,
      "learning_rate": 1.8550965233278725e-05,
      "loss": 1.2762,
      "step": 4923
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.424164564327418,
      "learning_rate": 1.855028876921238e-05,
      "loss": 1.2837,
      "step": 4924
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.111539229992815,
      "learning_rate": 1.854961215962308e-05,
      "loss": 1.4555,
      "step": 4925
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.8098181740773205,
      "learning_rate": 1.854893540452235e-05,
      "loss": 1.2716,
      "step": 4926
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.116227884846033,
      "learning_rate": 1.85482585039217e-05,
      "loss": 1.2532,
      "step": 4927
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.016747759695288,
      "learning_rate": 1.8547581457832658e-05,
      "loss": 1.1145,
      "step": 4928
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.796299870123977,
      "learning_rate": 1.854690426626674e-05,
      "loss": 1.1727,
      "step": 4929
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.601038854342326,
      "learning_rate": 1.8546226929235475e-05,
      "loss": 1.5806,
      "step": 4930
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.524929054731769,
      "learning_rate": 1.8545549446750392e-05,
      "loss": 1.1063,
      "step": 4931
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.006937327041565,
      "learning_rate": 1.8544871818823016e-05,
      "loss": 0.9364,
      "step": 4932
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.5852163600910325,
      "learning_rate": 1.8544194045464888e-05,
      "loss": 1.2487,
      "step": 4933
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.713908266356004,
      "learning_rate": 1.8543516126687536e-05,
      "loss": 1.3832,
      "step": 4934
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.419047460554571,
      "learning_rate": 1.8542838062502506e-05,
      "loss": 1.4038,
      "step": 4935
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.381601208807444,
      "learning_rate": 1.854215985292133e-05,
      "loss": 1.3771,
      "step": 4936
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.9280805930131075,
      "learning_rate": 1.8541481497955555e-05,
      "loss": 1.3752,
      "step": 4937
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.927455138636782,
      "learning_rate": 1.8540802997616726e-05,
      "loss": 1.4683,
      "step": 4938
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.83284083375383,
      "learning_rate": 1.8540124351916394e-05,
      "loss": 1.3374,
      "step": 4939
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8514139263067007,
      "learning_rate": 1.8539445560866106e-05,
      "loss": 1.1766,
      "step": 4940
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.348588983151623,
      "learning_rate": 1.8538766624477412e-05,
      "loss": 1.0839,
      "step": 4941
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.322861751543494,
      "learning_rate": 1.8538087542761873e-05,
      "loss": 1.3323,
      "step": 4942
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.914716813090794,
      "learning_rate": 1.853740831573104e-05,
      "loss": 1.1698,
      "step": 4943
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.231118437671922,
      "learning_rate": 1.8536728943396482e-05,
      "loss": 1.1192,
      "step": 4944
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.7958137513895105,
      "learning_rate": 1.8536049425769755e-05,
      "loss": 1.2501,
      "step": 4945
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.334595163464552,
      "learning_rate": 1.8535369762862427e-05,
      "loss": 1.1639,
      "step": 4946
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.451249255573226,
      "learning_rate": 1.853468995468606e-05,
      "loss": 1.1592,
      "step": 4947
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.7020273124957415,
      "learning_rate": 1.8534010001252233e-05,
      "loss": 1.1763,
      "step": 4948
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.859141311793975,
      "learning_rate": 1.8533329902572515e-05,
      "loss": 1.1168,
      "step": 4949
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.305421496794454,
      "learning_rate": 1.853264965865848e-05,
      "loss": 1.1514,
      "step": 4950
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.929952330165524,
      "learning_rate": 1.8531969269521706e-05,
      "loss": 1.1762,
      "step": 4951
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.086188822113843,
      "learning_rate": 1.853128873517377e-05,
      "loss": 1.0059,
      "step": 4952
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.276791128863552,
      "learning_rate": 1.8530608055626263e-05,
      "loss": 1.1431,
      "step": 4953
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.6961450974696115,
      "learning_rate": 1.8529927230890757e-05,
      "loss": 1.3135,
      "step": 4954
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.971810219513697,
      "learning_rate": 1.8529246260978855e-05,
      "loss": 1.3306,
      "step": 4955
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.813305709414453,
      "learning_rate": 1.852856514590213e-05,
      "loss": 1.1677,
      "step": 4956
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.865239177873526,
      "learning_rate": 1.8527883885672188e-05,
      "loss": 1.2409,
      "step": 4957
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.892825646524352,
      "learning_rate": 1.852720248030062e-05,
      "loss": 1.3161,
      "step": 4958
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.176479377748006,
      "learning_rate": 1.852652092979902e-05,
      "loss": 1.2345,
      "step": 4959
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.461495014834986,
      "learning_rate": 1.8525839234178987e-05,
      "loss": 1.371,
      "step": 4960
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.180044270748957,
      "learning_rate": 1.8525157393452133e-05,
      "loss": 1.2472,
      "step": 4961
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.8365271233441245,
      "learning_rate": 1.8524475407630052e-05,
      "loss": 1.2094,
      "step": 4962
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.547418145836118,
      "learning_rate": 1.8523793276724354e-05,
      "loss": 1.1385,
      "step": 4963
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.918281451372842,
      "learning_rate": 1.852311100074665e-05,
      "loss": 1.0587,
      "step": 4964
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.975016069866408,
      "learning_rate": 1.852242857970855e-05,
      "loss": 0.9834,
      "step": 4965
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.96099019548006,
      "learning_rate": 1.8521746013621674e-05,
      "loss": 1.1401,
      "step": 4966
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.548623644563767,
      "learning_rate": 1.8521063302497634e-05,
      "loss": 0.8945,
      "step": 4967
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.472231610639752,
      "learning_rate": 1.852038044634805e-05,
      "loss": 1.4251,
      "step": 4968
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.154301469831157,
      "learning_rate": 1.8519697445184542e-05,
      "loss": 1.0775,
      "step": 4969
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.4580070446352735,
      "learning_rate": 1.851901429901874e-05,
      "loss": 1.2789,
      "step": 4970
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.944024577010546,
      "learning_rate": 1.8518331007862268e-05,
      "loss": 1.0977,
      "step": 4971
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.764142644595848,
      "learning_rate": 1.8517647571726757e-05,
      "loss": 1.2337,
      "step": 4972
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.64092749714835,
      "learning_rate": 1.8516963990623834e-05,
      "loss": 1.5389,
      "step": 4973
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.774505039966708,
      "learning_rate": 1.851628026456514e-05,
      "loss": 1.2597,
      "step": 4974
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.394937601767726,
      "learning_rate": 1.8515596393562304e-05,
      "loss": 1.3122,
      "step": 4975
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.901728463538788,
      "learning_rate": 1.8514912377626974e-05,
      "loss": 1.5111,
      "step": 4976
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.213782703032213,
      "learning_rate": 1.8514228216770784e-05,
      "loss": 1.1308,
      "step": 4977
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.80899633128652,
      "learning_rate": 1.8513543911005384e-05,
      "loss": 1.2542,
      "step": 4978
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.93427102443219,
      "learning_rate": 1.8512859460342415e-05,
      "loss": 1.1726,
      "step": 4979
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.261431586668204,
      "learning_rate": 1.851217486479353e-05,
      "loss": 1.1313,
      "step": 4980
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.16267404226706,
      "learning_rate": 1.8511490124370383e-05,
      "loss": 1.2813,
      "step": 4981
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.771018765177063,
      "learning_rate": 1.851080523908462e-05,
      "loss": 1.474,
      "step": 4982
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.36394717980409,
      "learning_rate": 1.851012020894791e-05,
      "loss": 1.4049,
      "step": 4983
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.749617832085656,
      "learning_rate": 1.8509435033971898e-05,
      "loss": 1.2213,
      "step": 4984
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.225439203105179,
      "learning_rate": 1.8508749714168252e-05,
      "loss": 1.2758,
      "step": 4985
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.489371387065393,
      "learning_rate": 1.8508064249548637e-05,
      "loss": 1.1545,
      "step": 4986
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.661627381145513,
      "learning_rate": 1.850737864012472e-05,
      "loss": 1.4009,
      "step": 4987
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.135649099123507,
      "learning_rate": 1.8506692885908166e-05,
      "loss": 1.0118,
      "step": 4988
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.848416411233792,
      "learning_rate": 1.850600698691065e-05,
      "loss": 1.1515,
      "step": 4989
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.406624503002871,
      "learning_rate": 1.8505320943143843e-05,
      "loss": 1.265,
      "step": 4990
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.008468638102408,
      "learning_rate": 1.850463475461942e-05,
      "loss": 1.0288,
      "step": 4991
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.527073334637672,
      "learning_rate": 1.8503948421349064e-05,
      "loss": 1.2125,
      "step": 4992
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.097280638018664,
      "learning_rate": 1.8503261943344455e-05,
      "loss": 1.5152,
      "step": 4993
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.078347927440646,
      "learning_rate": 1.8502575320617274e-05,
      "loss": 1.1836,
      "step": 4994
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.783532659060466,
      "learning_rate": 1.8501888553179207e-05,
      "loss": 1.4794,
      "step": 4995
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.336082009639277,
      "learning_rate": 1.850120164104195e-05,
      "loss": 1.3371,
      "step": 4996
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.681780212040139,
      "learning_rate": 1.8500514584217182e-05,
      "loss": 1.1234,
      "step": 4997
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.023289979702203,
      "learning_rate": 1.8499827382716607e-05,
      "loss": 1.0907,
      "step": 4998
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.288865424461899,
      "learning_rate": 1.8499140036551914e-05,
      "loss": 1.2467,
      "step": 4999
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.959872191251502,
      "learning_rate": 1.8498452545734808e-05,
      "loss": 0.9999,
      "step": 5000
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.772532747816903,
      "learning_rate": 1.913888626681689e-05,
      "loss": 1.5018,
      "step": 5001
    },
    {
      "epoch": 0.2,
      "step": 5001,
      "total_flos": 705784014913536.0,
      "train_loss": 0.0003003027410036181,
      "train_runtime": 45.3711,
      "train_samples_per_second": 180.555,
      "train_steps_per_second": 5.642
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 256,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5000,
  "total_flos": 705784014913536.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}