{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.982278481012658,
  "eval_steps": 500,
  "global_step": 1230,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004050632911392405,
      "grad_norm": 6.493779741035002,
      "learning_rate": 3.2520325203252037e-07,
      "loss": 1.0369,
      "step": 1
    },
    {
      "epoch": 0.00810126582278481,
      "grad_norm": 6.515106229518725,
      "learning_rate": 6.504065040650407e-07,
      "loss": 1.0456,
      "step": 2
    },
    {
      "epoch": 0.012151898734177215,
      "grad_norm": 6.429995969071275,
      "learning_rate": 9.75609756097561e-07,
      "loss": 1.0189,
      "step": 3
    },
    {
      "epoch": 0.01620253164556962,
      "grad_norm": 6.387039354914473,
      "learning_rate": 1.3008130081300815e-06,
      "loss": 1.0409,
      "step": 4
    },
    {
      "epoch": 0.020253164556962026,
      "grad_norm": 6.116498044645081,
      "learning_rate": 1.6260162601626018e-06,
      "loss": 1.0311,
      "step": 5
    },
    {
      "epoch": 0.02430379746835443,
      "grad_norm": 5.941014474026115,
      "learning_rate": 1.951219512195122e-06,
      "loss": 1.0236,
      "step": 6
    },
    {
      "epoch": 0.028354430379746835,
      "grad_norm": 4.653955647394794,
      "learning_rate": 2.2764227642276426e-06,
      "loss": 0.9925,
      "step": 7
    },
    {
      "epoch": 0.03240506329113924,
      "grad_norm": 4.182108605052275,
      "learning_rate": 2.601626016260163e-06,
      "loss": 0.9875,
      "step": 8
    },
    {
      "epoch": 0.03645569620253165,
      "grad_norm": 2.7833095417146194,
      "learning_rate": 2.926829268292683e-06,
      "loss": 0.966,
      "step": 9
    },
    {
      "epoch": 0.04050632911392405,
      "grad_norm": 2.5982924224651462,
      "learning_rate": 3.2520325203252037e-06,
      "loss": 0.9588,
      "step": 10
    },
    {
      "epoch": 0.044556962025316456,
      "grad_norm": 2.496034039850331,
      "learning_rate": 3.577235772357724e-06,
      "loss": 0.9444,
      "step": 11
    },
    {
      "epoch": 0.04860759493670886,
      "grad_norm": 4.040707066018245,
      "learning_rate": 3.902439024390244e-06,
      "loss": 0.9377,
      "step": 12
    },
    {
      "epoch": 0.052658227848101265,
      "grad_norm": 4.292346111875974,
      "learning_rate": 4.227642276422765e-06,
      "loss": 0.9281,
      "step": 13
    },
    {
      "epoch": 0.05670886075949367,
      "grad_norm": 4.00842290769796,
      "learning_rate": 4.552845528455285e-06,
      "loss": 0.9214,
      "step": 14
    },
    {
      "epoch": 0.060759493670886074,
      "grad_norm": 3.7344061702213467,
      "learning_rate": 4.8780487804878055e-06,
      "loss": 0.9315,
      "step": 15
    },
    {
      "epoch": 0.06481012658227848,
      "grad_norm": 2.7848035744391844,
      "learning_rate": 5.203252032520326e-06,
      "loss": 0.8908,
      "step": 16
    },
    {
      "epoch": 0.06886075949367089,
      "grad_norm": 2.6387229811312456,
      "learning_rate": 5.528455284552846e-06,
      "loss": 0.8645,
      "step": 17
    },
    {
      "epoch": 0.0729113924050633,
      "grad_norm": 2.0405970359934886,
      "learning_rate": 5.853658536585366e-06,
      "loss": 0.842,
      "step": 18
    },
    {
      "epoch": 0.0769620253164557,
      "grad_norm": 1.5063531839872482,
      "learning_rate": 6.178861788617887e-06,
      "loss": 0.8444,
      "step": 19
    },
    {
      "epoch": 0.0810126582278481,
      "grad_norm": 1.2231838348418702,
      "learning_rate": 6.504065040650407e-06,
      "loss": 0.8399,
      "step": 20
    },
    {
      "epoch": 0.08506329113924051,
      "grad_norm": 1.1816006571265374,
      "learning_rate": 6.829268292682928e-06,
      "loss": 0.8338,
      "step": 21
    },
    {
      "epoch": 0.08911392405063291,
      "grad_norm": 1.187015584076262,
      "learning_rate": 7.154471544715448e-06,
      "loss": 0.7943,
      "step": 22
    },
    {
      "epoch": 0.09316455696202532,
      "grad_norm": 1.0826454198633744,
      "learning_rate": 7.4796747967479676e-06,
      "loss": 0.7991,
      "step": 23
    },
    {
      "epoch": 0.09721518987341772,
      "grad_norm": 0.9916375884291238,
      "learning_rate": 7.804878048780489e-06,
      "loss": 0.795,
      "step": 24
    },
    {
      "epoch": 0.10126582278481013,
      "grad_norm": 0.8768118030200569,
      "learning_rate": 8.130081300813009e-06,
      "loss": 0.7858,
      "step": 25
    },
    {
      "epoch": 0.10531645569620253,
      "grad_norm": 0.8843277046551754,
      "learning_rate": 8.45528455284553e-06,
      "loss": 0.7671,
      "step": 26
    },
    {
      "epoch": 0.10936708860759493,
      "grad_norm": 0.8869623782243337,
      "learning_rate": 8.78048780487805e-06,
      "loss": 0.7686,
      "step": 27
    },
    {
      "epoch": 0.11341772151898734,
      "grad_norm": 0.7637712328586501,
      "learning_rate": 9.10569105691057e-06,
      "loss": 0.7552,
      "step": 28
    },
    {
      "epoch": 0.11746835443037974,
      "grad_norm": 0.7726766110043194,
      "learning_rate": 9.43089430894309e-06,
      "loss": 0.7414,
      "step": 29
    },
    {
      "epoch": 0.12151898734177215,
      "grad_norm": 0.8359043319346965,
      "learning_rate": 9.756097560975611e-06,
      "loss": 0.7657,
      "step": 30
    },
    {
      "epoch": 0.12556962025316457,
      "grad_norm": 0.7278177337741392,
      "learning_rate": 1.008130081300813e-05,
      "loss": 0.737,
      "step": 31
    },
    {
      "epoch": 0.12962025316455697,
      "grad_norm": 0.7726847175656346,
      "learning_rate": 1.0406504065040652e-05,
      "loss": 0.7289,
      "step": 32
    },
    {
      "epoch": 0.13367088607594937,
      "grad_norm": 0.7146625232042311,
      "learning_rate": 1.0731707317073172e-05,
      "loss": 0.7401,
      "step": 33
    },
    {
      "epoch": 0.13772151898734178,
      "grad_norm": 0.769595424293189,
      "learning_rate": 1.1056910569105692e-05,
      "loss": 0.7332,
      "step": 34
    },
    {
      "epoch": 0.14177215189873418,
      "grad_norm": 0.7077250993009653,
      "learning_rate": 1.1382113821138213e-05,
      "loss": 0.7279,
      "step": 35
    },
    {
      "epoch": 0.1458227848101266,
      "grad_norm": 0.6259048383641109,
      "learning_rate": 1.1707317073170731e-05,
      "loss": 0.721,
      "step": 36
    },
    {
      "epoch": 0.149873417721519,
      "grad_norm": 0.7064145889175929,
      "learning_rate": 1.2032520325203254e-05,
      "loss": 0.7023,
      "step": 37
    },
    {
      "epoch": 0.1539240506329114,
      "grad_norm": 0.5739218033720924,
      "learning_rate": 1.2357723577235774e-05,
      "loss": 0.7075,
      "step": 38
    },
    {
      "epoch": 0.1579746835443038,
      "grad_norm": 0.5290977416891973,
      "learning_rate": 1.2682926829268294e-05,
      "loss": 0.709,
      "step": 39
    },
    {
      "epoch": 0.1620253164556962,
      "grad_norm": 0.6338445755753618,
      "learning_rate": 1.3008130081300815e-05,
      "loss": 0.7225,
      "step": 40
    },
    {
      "epoch": 0.1660759493670886,
      "grad_norm": 0.582008946784302,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.7013,
      "step": 41
    },
    {
      "epoch": 0.17012658227848101,
      "grad_norm": 0.46148416206026244,
      "learning_rate": 1.3658536585365855e-05,
      "loss": 0.6941,
      "step": 42
    },
    {
      "epoch": 0.17417721518987342,
      "grad_norm": 0.5518185411745675,
      "learning_rate": 1.3983739837398376e-05,
      "loss": 0.7036,
      "step": 43
    },
    {
      "epoch": 0.17822784810126582,
      "grad_norm": 0.4639122282449821,
      "learning_rate": 1.4308943089430896e-05,
      "loss": 0.6916,
      "step": 44
    },
    {
      "epoch": 0.18227848101265823,
      "grad_norm": 0.452346813462064,
      "learning_rate": 1.4634146341463415e-05,
      "loss": 0.6923,
      "step": 45
    },
    {
      "epoch": 0.18632911392405063,
      "grad_norm": 0.477960285448589,
      "learning_rate": 1.4959349593495935e-05,
      "loss": 0.694,
      "step": 46
    },
    {
      "epoch": 0.19037974683544304,
      "grad_norm": 0.48180303782475953,
      "learning_rate": 1.528455284552846e-05,
      "loss": 0.6917,
      "step": 47
    },
    {
      "epoch": 0.19443037974683544,
      "grad_norm": 0.5043342484339408,
      "learning_rate": 1.5609756097560978e-05,
      "loss": 0.6898,
      "step": 48
    },
    {
      "epoch": 0.19848101265822785,
      "grad_norm": 0.5563154851040636,
      "learning_rate": 1.5934959349593496e-05,
      "loss": 0.6752,
      "step": 49
    },
    {
      "epoch": 0.20253164556962025,
      "grad_norm": 0.5461413876421156,
      "learning_rate": 1.6260162601626018e-05,
      "loss": 0.6908,
      "step": 50
    },
    {
      "epoch": 0.20658227848101265,
      "grad_norm": 0.6520126658339338,
      "learning_rate": 1.6585365853658537e-05,
      "loss": 0.707,
      "step": 51
    },
    {
      "epoch": 0.21063291139240506,
      "grad_norm": 0.5813047978084712,
      "learning_rate": 1.691056910569106e-05,
      "loss": 0.7037,
      "step": 52
    },
    {
      "epoch": 0.21468354430379746,
      "grad_norm": 0.5577770947575851,
      "learning_rate": 1.7235772357723578e-05,
      "loss": 0.6686,
      "step": 53
    },
    {
      "epoch": 0.21873417721518987,
      "grad_norm": 0.5415169789148744,
      "learning_rate": 1.75609756097561e-05,
      "loss": 0.6789,
      "step": 54
    },
    {
      "epoch": 0.22278481012658227,
      "grad_norm": 0.6484662988339179,
      "learning_rate": 1.788617886178862e-05,
      "loss": 0.6889,
      "step": 55
    },
    {
      "epoch": 0.22683544303797468,
      "grad_norm": 0.5465724606158754,
      "learning_rate": 1.821138211382114e-05,
      "loss": 0.6959,
      "step": 56
    },
    {
      "epoch": 0.23088607594936708,
      "grad_norm": 0.5850420519700699,
      "learning_rate": 1.8536585365853663e-05,
      "loss": 0.6672,
      "step": 57
    },
    {
      "epoch": 0.23493670886075949,
      "grad_norm": 0.7922879583091695,
      "learning_rate": 1.886178861788618e-05,
      "loss": 0.6776,
      "step": 58
    },
    {
      "epoch": 0.2389873417721519,
      "grad_norm": 0.8134985296805135,
      "learning_rate": 1.91869918699187e-05,
      "loss": 0.6696,
      "step": 59
    },
    {
      "epoch": 0.2430379746835443,
      "grad_norm": 0.7031140151725898,
      "learning_rate": 1.9512195121951222e-05,
      "loss": 0.6731,
      "step": 60
    },
    {
      "epoch": 0.2470886075949367,
      "grad_norm": 0.4754470638622413,
      "learning_rate": 1.983739837398374e-05,
      "loss": 0.6822,
      "step": 61
    },
    {
      "epoch": 0.25113924050632913,
      "grad_norm": 0.6127230308312909,
      "learning_rate": 2.016260162601626e-05,
      "loss": 0.6727,
      "step": 62
    },
    {
      "epoch": 0.25518987341772154,
      "grad_norm": 0.655461466949352,
      "learning_rate": 2.048780487804878e-05,
      "loss": 0.6649,
      "step": 63
    },
    {
      "epoch": 0.25924050632911394,
      "grad_norm": 0.7142679376537177,
      "learning_rate": 2.0813008130081303e-05,
      "loss": 0.671,
      "step": 64
    },
    {
      "epoch": 0.26329113924050634,
      "grad_norm": 0.6355258786430444,
      "learning_rate": 2.1138211382113822e-05,
      "loss": 0.658,
      "step": 65
    },
    {
      "epoch": 0.26734177215189875,
      "grad_norm": 0.601739330859932,
      "learning_rate": 2.1463414634146344e-05,
      "loss": 0.6582,
      "step": 66
    },
    {
      "epoch": 0.27139240506329115,
      "grad_norm": 0.702154219398689,
      "learning_rate": 2.1788617886178863e-05,
      "loss": 0.6697,
      "step": 67
    },
    {
      "epoch": 0.27544303797468356,
      "grad_norm": 0.7634367175867832,
      "learning_rate": 2.2113821138211385e-05,
      "loss": 0.6636,
      "step": 68
    },
    {
      "epoch": 0.27949367088607596,
      "grad_norm": 0.9235175913292207,
      "learning_rate": 2.2439024390243907e-05,
      "loss": 0.6477,
      "step": 69
    },
    {
      "epoch": 0.28354430379746837,
      "grad_norm": 1.0254992259705693,
      "learning_rate": 2.2764227642276426e-05,
      "loss": 0.6615,
      "step": 70
    },
    {
      "epoch": 0.28759493670886077,
      "grad_norm": 0.7727646963955114,
      "learning_rate": 2.3089430894308948e-05,
      "loss": 0.6522,
      "step": 71
    },
    {
      "epoch": 0.2916455696202532,
      "grad_norm": 0.7790381555019846,
      "learning_rate": 2.3414634146341463e-05,
      "loss": 0.6526,
      "step": 72
    },
    {
      "epoch": 0.2956962025316456,
      "grad_norm": 0.6994731480588183,
      "learning_rate": 2.3739837398373985e-05,
      "loss": 0.6413,
      "step": 73
    },
    {
      "epoch": 0.299746835443038,
      "grad_norm": 0.9156293173170503,
      "learning_rate": 2.4065040650406507e-05,
      "loss": 0.6484,
      "step": 74
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 1.421362613775901,
      "learning_rate": 2.4390243902439026e-05,
      "loss": 0.6565,
      "step": 75
    },
    {
      "epoch": 0.3078481012658228,
      "grad_norm": 0.754699546883223,
      "learning_rate": 2.4715447154471548e-05,
      "loss": 0.6529,
      "step": 76
    },
    {
      "epoch": 0.3118987341772152,
      "grad_norm": 1.1756426852249031,
      "learning_rate": 2.5040650406504066e-05,
      "loss": 0.6616,
      "step": 77
    },
    {
      "epoch": 0.3159493670886076,
      "grad_norm": 0.9388260925549986,
      "learning_rate": 2.536585365853659e-05,
      "loss": 0.648,
      "step": 78
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9545136533609573,
      "learning_rate": 2.569105691056911e-05,
      "loss": 0.6694,
      "step": 79
    },
    {
      "epoch": 0.3240506329113924,
      "grad_norm": 1.302449385376475,
      "learning_rate": 2.601626016260163e-05,
      "loss": 0.6519,
      "step": 80
    },
    {
      "epoch": 0.3281012658227848,
      "grad_norm": 0.6623514031959914,
      "learning_rate": 2.634146341463415e-05,
      "loss": 0.6616,
      "step": 81
    },
    {
      "epoch": 0.3321518987341772,
      "grad_norm": 0.844931208384013,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.6592,
      "step": 82
    },
    {
      "epoch": 0.3362025316455696,
      "grad_norm": 1.4191222191333228,
      "learning_rate": 2.699186991869919e-05,
      "loss": 0.6638,
      "step": 83
    },
    {
      "epoch": 0.34025316455696203,
      "grad_norm": 0.6883700383015497,
      "learning_rate": 2.731707317073171e-05,
      "loss": 0.6445,
      "step": 84
    },
    {
      "epoch": 0.34430379746835443,
      "grad_norm": 1.0459717782117925,
      "learning_rate": 2.764227642276423e-05,
      "loss": 0.6405,
      "step": 85
    },
    {
      "epoch": 0.34835443037974684,
      "grad_norm": 1.3014935120548765,
      "learning_rate": 2.796747967479675e-05,
      "loss": 0.6408,
      "step": 86
    },
    {
      "epoch": 0.35240506329113924,
      "grad_norm": 0.7902832428651478,
      "learning_rate": 2.829268292682927e-05,
      "loss": 0.644,
      "step": 87
    },
    {
      "epoch": 0.35645569620253165,
      "grad_norm": 1.1790507455720673,
      "learning_rate": 2.8617886178861792e-05,
      "loss": 0.6557,
      "step": 88
    },
    {
      "epoch": 0.36050632911392405,
      "grad_norm": 0.9308727298662697,
      "learning_rate": 2.8943089430894314e-05,
      "loss": 0.6421,
      "step": 89
    },
    {
      "epoch": 0.36455696202531646,
      "grad_norm": 1.3852444679779106,
      "learning_rate": 2.926829268292683e-05,
      "loss": 0.6568,
      "step": 90
    },
    {
      "epoch": 0.36860759493670886,
      "grad_norm": 1.083175881998161,
      "learning_rate": 2.959349593495935e-05,
      "loss": 0.6548,
      "step": 91
    },
    {
      "epoch": 0.37265822784810126,
      "grad_norm": 1.3023777120228035,
      "learning_rate": 2.991869918699187e-05,
      "loss": 0.6309,
      "step": 92
    },
    {
      "epoch": 0.37670886075949367,
      "grad_norm": 1.169229690621438,
      "learning_rate": 3.0243902439024392e-05,
      "loss": 0.6354,
      "step": 93
    },
    {
      "epoch": 0.3807594936708861,
      "grad_norm": 1.2343495144807062,
      "learning_rate": 3.056910569105692e-05,
      "loss": 0.6518,
      "step": 94
    },
    {
      "epoch": 0.3848101265822785,
      "grad_norm": 0.7880681493978746,
      "learning_rate": 3.089430894308943e-05,
      "loss": 0.6454,
      "step": 95
    },
    {
      "epoch": 0.3888607594936709,
      "grad_norm": 0.9845266495398576,
      "learning_rate": 3.1219512195121955e-05,
      "loss": 0.6402,
      "step": 96
    },
    {
      "epoch": 0.3929113924050633,
      "grad_norm": 1.2835785749186486,
      "learning_rate": 3.154471544715447e-05,
      "loss": 0.6342,
      "step": 97
    },
    {
      "epoch": 0.3969620253164557,
      "grad_norm": 0.8941179056871833,
      "learning_rate": 3.186991869918699e-05,
      "loss": 0.6437,
      "step": 98
    },
    {
      "epoch": 0.4010126582278481,
      "grad_norm": 1.1492862601767255,
      "learning_rate": 3.2195121951219514e-05,
      "loss": 0.6269,
      "step": 99
    },
    {
      "epoch": 0.4050632911392405,
      "grad_norm": 0.9435850054148426,
      "learning_rate": 3.2520325203252037e-05,
      "loss": 0.6375,
      "step": 100
    },
    {
      "epoch": 0.4091139240506329,
      "grad_norm": 1.1560716364398882,
      "learning_rate": 3.284552845528456e-05,
      "loss": 0.6455,
      "step": 101
    },
    {
      "epoch": 0.4131645569620253,
      "grad_norm": 1.3752429986711991,
      "learning_rate": 3.3170731707317074e-05,
      "loss": 0.6391,
      "step": 102
    },
    {
      "epoch": 0.4172151898734177,
      "grad_norm": 0.8529747886238581,
      "learning_rate": 3.3495934959349596e-05,
      "loss": 0.6402,
      "step": 103
    },
    {
      "epoch": 0.4212658227848101,
      "grad_norm": 1.2384909332740734,
      "learning_rate": 3.382113821138212e-05,
      "loss": 0.6496,
      "step": 104
    },
    {
      "epoch": 0.4253164556962025,
      "grad_norm": 0.782143387645592,
      "learning_rate": 3.414634146341463e-05,
      "loss": 0.6422,
      "step": 105
    },
    {
      "epoch": 0.4293670886075949,
      "grad_norm": 1.2921701082047912,
      "learning_rate": 3.4471544715447155e-05,
      "loss": 0.6419,
      "step": 106
    },
    {
      "epoch": 0.43341772151898733,
      "grad_norm": 1.0295850101630235,
      "learning_rate": 3.479674796747968e-05,
      "loss": 0.6392,
      "step": 107
    },
    {
      "epoch": 0.43746835443037974,
      "grad_norm": 1.2451766619193538,
      "learning_rate": 3.51219512195122e-05,
      "loss": 0.6317,
      "step": 108
    },
    {
      "epoch": 0.44151898734177214,
      "grad_norm": 0.759219963665096,
      "learning_rate": 3.544715447154472e-05,
      "loss": 0.6377,
      "step": 109
    },
    {
      "epoch": 0.44556962025316454,
      "grad_norm": 0.9160457930109868,
      "learning_rate": 3.577235772357724e-05,
      "loss": 0.6284,
      "step": 110
    },
    {
      "epoch": 0.44962025316455695,
      "grad_norm": 1.2762244100941547,
      "learning_rate": 3.609756097560976e-05,
      "loss": 0.6314,
      "step": 111
    },
    {
      "epoch": 0.45367088607594935,
      "grad_norm": 0.953814132845107,
      "learning_rate": 3.642276422764228e-05,
      "loss": 0.6298,
      "step": 112
    },
    {
      "epoch": 0.45772151898734176,
      "grad_norm": 1.227529178232297,
      "learning_rate": 3.67479674796748e-05,
      "loss": 0.6405,
      "step": 113
    },
    {
      "epoch": 0.46177215189873416,
      "grad_norm": 1.3445919689496657,
      "learning_rate": 3.7073170731707325e-05,
      "loss": 0.6189,
      "step": 114
    },
    {
      "epoch": 0.46582278481012657,
      "grad_norm": 0.6641227292112333,
      "learning_rate": 3.739837398373984e-05,
      "loss": 0.6354,
      "step": 115
    },
    {
      "epoch": 0.46987341772151897,
      "grad_norm": 0.7345865382530172,
      "learning_rate": 3.772357723577236e-05,
      "loss": 0.643,
      "step": 116
    },
    {
      "epoch": 0.4739240506329114,
      "grad_norm": 1.2673684968122978,
      "learning_rate": 3.804878048780488e-05,
      "loss": 0.6377,
      "step": 117
    },
    {
      "epoch": 0.4779746835443038,
      "grad_norm": 1.0220561501182392,
      "learning_rate": 3.83739837398374e-05,
      "loss": 0.6212,
      "step": 118
    },
    {
      "epoch": 0.4820253164556962,
      "grad_norm": 1.6078379451949094,
      "learning_rate": 3.869918699186992e-05,
      "loss": 0.6239,
      "step": 119
    },
    {
      "epoch": 0.4860759493670886,
      "grad_norm": 0.6652128409014539,
      "learning_rate": 3.9024390243902444e-05,
      "loss": 0.6441,
      "step": 120
    },
    {
      "epoch": 0.490126582278481,
      "grad_norm": 1.192603228608864,
      "learning_rate": 3.9349593495934966e-05,
      "loss": 0.6322,
      "step": 121
    },
    {
      "epoch": 0.4941772151898734,
      "grad_norm": 1.2787886383147804,
      "learning_rate": 3.967479674796748e-05,
      "loss": 0.6317,
      "step": 122
    },
    {
      "epoch": 0.4982278481012658,
      "grad_norm": 0.7229243690355931,
      "learning_rate": 4e-05,
      "loss": 0.6294,
      "step": 123
    },
    {
      "epoch": 0.5022784810126583,
      "grad_norm": 1.0248057535736221,
      "learning_rate": 3.999991946137476e-05,
      "loss": 0.6349,
      "step": 124
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 1.525586054482558,
      "learning_rate": 3.999967784614766e-05,
      "loss": 0.6275,
      "step": 125
    },
    {
      "epoch": 0.5103797468354431,
      "grad_norm": 0.8882333381653731,
      "learning_rate": 3.9999275156264656e-05,
      "loss": 0.6234,
      "step": 126
    },
    {
      "epoch": 0.5144303797468355,
      "grad_norm": 2.1241713787889,
      "learning_rate": 3.999871139496895e-05,
      "loss": 0.6555,
      "step": 127
    },
    {
      "epoch": 0.5184810126582279,
      "grad_norm": 1.3919506047718353,
      "learning_rate": 3.9997986566800995e-05,
      "loss": 0.6323,
      "step": 128
    },
    {
      "epoch": 0.5225316455696203,
      "grad_norm": 1.6732908651424134,
      "learning_rate": 3.999710067759846e-05,
      "loss": 0.6277,
      "step": 129
    },
    {
      "epoch": 0.5265822784810127,
      "grad_norm": 1.3101489734815353,
      "learning_rate": 3.999605373449617e-05,
      "loss": 0.638,
      "step": 130
    },
    {
      "epoch": 0.5306329113924051,
      "grad_norm": 1.6015574611565624,
      "learning_rate": 3.9994845745926075e-05,
      "loss": 0.6357,
      "step": 131
    },
    {
      "epoch": 0.5346835443037975,
      "grad_norm": 1.0127353201103801,
      "learning_rate": 3.999347672161713e-05,
      "loss": 0.6203,
      "step": 132
    },
    {
      "epoch": 0.5387341772151899,
      "grad_norm": 1.5568024472012802,
      "learning_rate": 3.999194667259528e-05,
      "loss": 0.6343,
      "step": 133
    },
    {
      "epoch": 0.5427848101265823,
      "grad_norm": 1.169633471584172,
      "learning_rate": 3.999025561118334e-05,
      "loss": 0.6388,
      "step": 134
    },
    {
      "epoch": 0.5468354430379747,
      "grad_norm": 1.5252329091515757,
      "learning_rate": 3.998840355100086e-05,
      "loss": 0.6345,
      "step": 135
    },
    {
      "epoch": 0.5508860759493671,
      "grad_norm": 1.3444595017666914,
      "learning_rate": 3.998639050696409e-05,
      "loss": 0.6298,
      "step": 136
    },
    {
      "epoch": 0.5549367088607595,
      "grad_norm": 1.1443605082950221,
      "learning_rate": 3.998421649528582e-05,
      "loss": 0.6272,
      "step": 137
    },
    {
      "epoch": 0.5589873417721519,
      "grad_norm": 1.0079361454341442,
      "learning_rate": 3.9981881533475234e-05,
      "loss": 0.6425,
      "step": 138
    },
    {
      "epoch": 0.5630379746835443,
      "grad_norm": 1.0029542686653323,
      "learning_rate": 3.997938564033779e-05,
      "loss": 0.6249,
      "step": 139
    },
    {
      "epoch": 0.5670886075949367,
      "grad_norm": 1.015564295707624,
      "learning_rate": 3.9976728835975064e-05,
      "loss": 0.6311,
      "step": 140
    },
    {
      "epoch": 0.5711392405063291,
      "grad_norm": 1.2557549600983018,
      "learning_rate": 3.9973911141784605e-05,
      "loss": 0.6287,
      "step": 141
    },
    {
      "epoch": 0.5751898734177215,
      "grad_norm": 0.7468464286390285,
      "learning_rate": 3.997093258045973e-05,
      "loss": 0.6159,
      "step": 142
    },
    {
      "epoch": 0.579240506329114,
      "grad_norm": 1.242760939558488,
      "learning_rate": 3.996779317598936e-05,
      "loss": 0.6276,
      "step": 143
    },
    {
      "epoch": 0.5832911392405064,
      "grad_norm": 0.8876839140869978,
      "learning_rate": 3.996449295365782e-05,
      "loss": 0.6253,
      "step": 144
    },
    {
      "epoch": 0.5873417721518988,
      "grad_norm": 1.142462004782762,
      "learning_rate": 3.996103194004467e-05,
      "loss": 0.609,
      "step": 145
    },
    {
      "epoch": 0.5913924050632912,
      "grad_norm": 0.956802960437125,
      "learning_rate": 3.995741016302441e-05,
      "loss": 0.6155,
      "step": 146
    },
    {
      "epoch": 0.5954430379746836,
      "grad_norm": 0.8216663909823639,
      "learning_rate": 3.9953627651766364e-05,
      "loss": 0.6099,
      "step": 147
    },
    {
      "epoch": 0.599493670886076,
      "grad_norm": 0.843324655269676,
      "learning_rate": 3.9949684436734325e-05,
      "loss": 0.6204,
      "step": 148
    },
    {
      "epoch": 0.6035443037974684,
      "grad_norm": 0.7837263915872599,
      "learning_rate": 3.994558054968643e-05,
      "loss": 0.6344,
      "step": 149
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 0.9211042586175315,
      "learning_rate": 3.994131602367481e-05,
      "loss": 0.61,
      "step": 150
    },
    {
      "epoch": 0.6116455696202532,
      "grad_norm": 0.852038263104514,
      "learning_rate": 3.9936890893045376e-05,
      "loss": 0.6291,
      "step": 151
    },
    {
      "epoch": 0.6156962025316456,
      "grad_norm": 0.6083963620395382,
      "learning_rate": 3.993230519343752e-05,
      "loss": 0.6354,
      "step": 152
    },
    {
      "epoch": 0.619746835443038,
      "grad_norm": 0.8495674708153879,
      "learning_rate": 3.992755896178383e-05,
      "loss": 0.6033,
      "step": 153
    },
    {
      "epoch": 0.6237974683544304,
      "grad_norm": 0.9448267933712936,
      "learning_rate": 3.992265223630981e-05,
      "loss": 0.6162,
      "step": 154
    },
    {
      "epoch": 0.6278481012658228,
      "grad_norm": 0.8429402834252556,
      "learning_rate": 3.991758505653355e-05,
      "loss": 0.6031,
      "step": 155
    },
    {
      "epoch": 0.6318987341772152,
      "grad_norm": 0.8990977661755324,
      "learning_rate": 3.991235746326543e-05,
      "loss": 0.6194,
      "step": 156
    },
    {
      "epoch": 0.6359493670886076,
      "grad_norm": 0.6268515530458345,
      "learning_rate": 3.9906969498607745e-05,
      "loss": 0.6103,
      "step": 157
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5594599301744696,
      "learning_rate": 3.990142120595444e-05,
      "loss": 0.6214,
      "step": 158
    },
    {
      "epoch": 0.6440506329113924,
      "grad_norm": 0.7037342719961769,
      "learning_rate": 3.98957126299907e-05,
      "loss": 0.6086,
      "step": 159
    },
    {
      "epoch": 0.6481012658227848,
      "grad_norm": 0.8074926517981739,
      "learning_rate": 3.9889843816692596e-05,
      "loss": 0.6026,
      "step": 160
    },
    {
      "epoch": 0.6521518987341772,
      "grad_norm": 0.6177679520555956,
      "learning_rate": 3.9883814813326766e-05,
      "loss": 0.6112,
      "step": 161
    },
    {
      "epoch": 0.6562025316455696,
      "grad_norm": 0.5657529954730145,
      "learning_rate": 3.9877625668449956e-05,
      "loss": 0.605,
      "step": 162
    },
    {
      "epoch": 0.660253164556962,
      "grad_norm": 0.6813629216776935,
      "learning_rate": 3.98712764319087e-05,
      "loss": 0.6292,
      "step": 163
    },
    {
      "epoch": 0.6643037974683544,
      "grad_norm": 0.8974269735246865,
      "learning_rate": 3.9864767154838864e-05,
      "loss": 0.6105,
      "step": 164
    },
    {
      "epoch": 0.6683544303797468,
      "grad_norm": 0.8403900958702467,
      "learning_rate": 3.9858097889665277e-05,
      "loss": 0.6221,
      "step": 165
    },
    {
      "epoch": 0.6724050632911392,
      "grad_norm": 0.6698290422618715,
      "learning_rate": 3.985126869010129e-05,
      "loss": 0.6307,
      "step": 166
    },
    {
      "epoch": 0.6764556962025317,
      "grad_norm": 0.5509832314339451,
      "learning_rate": 3.984427961114833e-05,
      "loss": 0.6274,
      "step": 167
    },
    {
      "epoch": 0.6805063291139241,
      "grad_norm": 0.5525040237617902,
      "learning_rate": 3.9837130709095475e-05,
      "loss": 0.6203,
      "step": 168
    },
    {
      "epoch": 0.6845569620253165,
      "grad_norm": 0.836421107896438,
      "learning_rate": 3.982982204151901e-05,
      "loss": 0.6263,
      "step": 169
    },
    {
      "epoch": 0.6886075949367089,
      "grad_norm": 0.9552015114391759,
      "learning_rate": 3.982235366728193e-05,
      "loss": 0.5933,
      "step": 170
    },
    {
      "epoch": 0.6926582278481013,
      "grad_norm": 0.5322688550453594,
      "learning_rate": 3.9814725646533505e-05,
      "loss": 0.6054,
      "step": 171
    },
    {
      "epoch": 0.6967088607594937,
      "grad_norm": 0.6299715565354982,
      "learning_rate": 3.9806938040708746e-05,
      "loss": 0.6022,
      "step": 172
    },
    {
      "epoch": 0.7007594936708861,
      "grad_norm": 0.7014072892057753,
      "learning_rate": 3.9798990912527976e-05,
      "loss": 0.6179,
      "step": 173
    },
    {
      "epoch": 0.7048101265822785,
      "grad_norm": 0.6313198783319353,
      "learning_rate": 3.979088432599627e-05,
      "loss": 0.6036,
      "step": 174
    },
    {
      "epoch": 0.7088607594936709,
      "grad_norm": 0.4828474126073063,
      "learning_rate": 3.9782618346402964e-05,
      "loss": 0.6213,
      "step": 175
    },
    {
      "epoch": 0.7129113924050633,
      "grad_norm": 0.5221173010104053,
      "learning_rate": 3.977419304032111e-05,
      "loss": 0.6137,
      "step": 176
    },
    {
      "epoch": 0.7169620253164557,
      "grad_norm": 0.5619097443174723,
      "learning_rate": 3.976560847560697e-05,
      "loss": 0.6057,
      "step": 177
    },
    {
      "epoch": 0.7210126582278481,
      "grad_norm": 0.48458250268714326,
      "learning_rate": 3.9756864721399456e-05,
      "loss": 0.6078,
      "step": 178
    },
    {
      "epoch": 0.7250632911392405,
      "grad_norm": 0.5938847632181161,
      "learning_rate": 3.974796184811956e-05,
      "loss": 0.6154,
      "step": 179
    },
    {
      "epoch": 0.7291139240506329,
      "grad_norm": 0.8573099668344029,
      "learning_rate": 3.973889992746979e-05,
      "loss": 0.6058,
      "step": 180
    },
    {
      "epoch": 0.7331645569620253,
      "grad_norm": 0.9887704327797934,
      "learning_rate": 3.972967903243361e-05,
      "loss": 0.6076,
      "step": 181
    },
    {
      "epoch": 0.7372151898734177,
      "grad_norm": 0.8686374962177015,
      "learning_rate": 3.972029923727486e-05,
      "loss": 0.6217,
      "step": 182
    },
    {
      "epoch": 0.7412658227848101,
      "grad_norm": 0.7109343949802196,
      "learning_rate": 3.971076061753709e-05,
      "loss": 0.6035,
      "step": 183
    },
    {
      "epoch": 0.7453164556962025,
      "grad_norm": 0.493792875999254,
      "learning_rate": 3.9701063250043066e-05,
      "loss": 0.6179,
      "step": 184
    },
    {
      "epoch": 0.7493670886075949,
      "grad_norm": 0.6295602586739065,
      "learning_rate": 3.969120721289402e-05,
      "loss": 0.6,
      "step": 185
    },
    {
      "epoch": 0.7534177215189873,
      "grad_norm": 0.5993221419342887,
      "learning_rate": 3.9681192585469146e-05,
      "loss": 0.6059,
      "step": 186
    },
    {
      "epoch": 0.7574683544303797,
      "grad_norm": 0.6640111641033765,
      "learning_rate": 3.9671019448424865e-05,
      "loss": 0.606,
      "step": 187
    },
    {
      "epoch": 0.7615189873417721,
      "grad_norm": 0.5340855976513834,
      "learning_rate": 3.966068788369422e-05,
      "loss": 0.61,
      "step": 188
    },
    {
      "epoch": 0.7655696202531646,
      "grad_norm": 0.5094307398305351,
      "learning_rate": 3.965019797448622e-05,
      "loss": 0.6088,
      "step": 189
    },
    {
      "epoch": 0.769620253164557,
      "grad_norm": 0.5222761909192823,
      "learning_rate": 3.963954980528515e-05,
      "loss": 0.6177,
      "step": 190
    },
    {
      "epoch": 0.7736708860759494,
      "grad_norm": 0.5192079744187618,
      "learning_rate": 3.9628743461849905e-05,
      "loss": 0.6134,
      "step": 191
    },
    {
      "epoch": 0.7777215189873418,
      "grad_norm": 0.528368948026275,
      "learning_rate": 3.961777903121329e-05,
      "loss": 0.5897,
      "step": 192
    },
    {
      "epoch": 0.7817721518987342,
      "grad_norm": 0.4875660748536816,
      "learning_rate": 3.960665660168131e-05,
      "loss": 0.6054,
      "step": 193
    },
    {
      "epoch": 0.7858227848101266,
      "grad_norm": 0.4405163656576639,
      "learning_rate": 3.9595376262832485e-05,
      "loss": 0.6059,
      "step": 194
    },
    {
      "epoch": 0.789873417721519,
      "grad_norm": 0.5839220175599222,
      "learning_rate": 3.9583938105517127e-05,
      "loss": 0.599,
      "step": 195
    },
    {
      "epoch": 0.7939240506329114,
      "grad_norm": 0.5312412693211577,
      "learning_rate": 3.957234222185657e-05,
      "loss": 0.6185,
      "step": 196
    },
    {
      "epoch": 0.7979746835443038,
      "grad_norm": 0.4262982798701605,
      "learning_rate": 3.9560588705242474e-05,
      "loss": 0.6013,
      "step": 197
    },
    {
      "epoch": 0.8020253164556962,
      "grad_norm": 0.6199389475787582,
      "learning_rate": 3.954867765033605e-05,
      "loss": 0.6,
      "step": 198
    },
    {
      "epoch": 0.8060759493670886,
      "grad_norm": 0.7139097282232219,
      "learning_rate": 3.953660915306728e-05,
      "loss": 0.6053,
      "step": 199
    },
    {
      "epoch": 0.810126582278481,
      "grad_norm": 0.5962012783773114,
      "learning_rate": 3.952438331063419e-05,
      "loss": 0.6258,
      "step": 200
    },
    {
      "epoch": 0.8141772151898734,
      "grad_norm": 0.588874409693904,
      "learning_rate": 3.951200022150205e-05,
      "loss": 0.601,
      "step": 201
    },
    {
      "epoch": 0.8182278481012658,
      "grad_norm": 0.5285767735454469,
      "learning_rate": 3.949945998540253e-05,
      "loss": 0.6069,
      "step": 202
    },
    {
      "epoch": 0.8222784810126582,
      "grad_norm": 0.4613117720774959,
      "learning_rate": 3.9486762703332993e-05,
      "loss": 0.6083,
      "step": 203
    },
    {
      "epoch": 0.8263291139240506,
      "grad_norm": 0.44568915286136745,
      "learning_rate": 3.947390847755559e-05,
      "loss": 0.5986,
      "step": 204
    },
    {
      "epoch": 0.830379746835443,
      "grad_norm": 0.49643346478004824,
      "learning_rate": 3.946089741159648e-05,
      "loss": 0.6118,
      "step": 205
    },
    {
      "epoch": 0.8344303797468354,
      "grad_norm": 0.48199145601359433,
      "learning_rate": 3.944772961024501e-05,
      "loss": 0.6055,
      "step": 206
    },
    {
      "epoch": 0.8384810126582278,
      "grad_norm": 0.4385394919190343,
      "learning_rate": 3.943440517955285e-05,
      "loss": 0.6029,
      "step": 207
    },
    {
      "epoch": 0.8425316455696202,
      "grad_norm": 0.5227377418078802,
      "learning_rate": 3.9420924226833126e-05,
      "loss": 0.6123,
      "step": 208
    },
    {
      "epoch": 0.8465822784810126,
      "grad_norm": 0.5557498184432195,
      "learning_rate": 3.9407286860659566e-05,
      "loss": 0.5893,
      "step": 209
    },
    {
      "epoch": 0.850632911392405,
      "grad_norm": 0.47087717307213717,
      "learning_rate": 3.9393493190865657e-05,
      "loss": 0.5982,
      "step": 210
    },
    {
      "epoch": 0.8546835443037974,
      "grad_norm": 0.6724579733407169,
      "learning_rate": 3.937954332854371e-05,
      "loss": 0.5992,
      "step": 211
    },
    {
      "epoch": 0.8587341772151899,
      "grad_norm": 0.6020203634243056,
      "learning_rate": 3.9365437386044016e-05,
      "loss": 0.5989,
      "step": 212
    },
    {
      "epoch": 0.8627848101265823,
      "grad_norm": 0.5707962035272884,
      "learning_rate": 3.935117547697387e-05,
      "loss": 0.6123,
      "step": 213
    },
    {
      "epoch": 0.8668354430379747,
      "grad_norm": 0.5821269609930904,
      "learning_rate": 3.933675771619675e-05,
      "loss": 0.5991,
      "step": 214
    },
    {
      "epoch": 0.8708860759493671,
      "grad_norm": 0.6422558154108303,
      "learning_rate": 3.932218421983131e-05,
      "loss": 0.5926,
      "step": 215
    },
    {
      "epoch": 0.8749367088607595,
      "grad_norm": 0.5564434387251845,
      "learning_rate": 3.9307455105250484e-05,
      "loss": 0.6186,
      "step": 216
    },
    {
      "epoch": 0.8789873417721519,
      "grad_norm": 0.49360865756263667,
      "learning_rate": 3.929257049108054e-05,
      "loss": 0.5951,
      "step": 217
    },
    {
      "epoch": 0.8830379746835443,
      "grad_norm": 0.6371197132961622,
      "learning_rate": 3.927753049720011e-05,
      "loss": 0.5998,
      "step": 218
    },
    {
      "epoch": 0.8870886075949367,
      "grad_norm": 0.6402339583511993,
      "learning_rate": 3.9262335244739234e-05,
      "loss": 0.5996,
      "step": 219
    },
    {
      "epoch": 0.8911392405063291,
      "grad_norm": 0.5484042182270151,
      "learning_rate": 3.92469848560784e-05,
      "loss": 0.597,
      "step": 220
    },
    {
      "epoch": 0.8951898734177215,
      "grad_norm": 0.6337938041969547,
      "learning_rate": 3.923147945484751e-05,
      "loss": 0.5958,
      "step": 221
    },
    {
      "epoch": 0.8992405063291139,
      "grad_norm": 0.5434746613682024,
      "learning_rate": 3.9215819165924956e-05,
      "loss": 0.5926,
      "step": 222
    },
    {
      "epoch": 0.9032911392405063,
      "grad_norm": 0.39940339370234795,
      "learning_rate": 3.920000411543654e-05,
      "loss": 0.594,
      "step": 223
    },
    {
      "epoch": 0.9073417721518987,
      "grad_norm": 0.46933456129664114,
      "learning_rate": 3.9184034430754495e-05,
      "loss": 0.5936,
      "step": 224
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 0.5432580453018409,
      "learning_rate": 3.916791024049648e-05,
      "loss": 0.5997,
      "step": 225
    },
    {
      "epoch": 0.9154430379746835,
      "grad_norm": 0.5039426677924892,
      "learning_rate": 3.91516316745245e-05,
      "loss": 0.5983,
      "step": 226
    },
    {
      "epoch": 0.9194936708860759,
      "grad_norm": 0.5026347462058494,
      "learning_rate": 3.913519886394389e-05,
      "loss": 0.6007,
      "step": 227
    },
    {
      "epoch": 0.9235443037974683,
      "grad_norm": 0.6239700304616167,
      "learning_rate": 3.911861194110225e-05,
      "loss": 0.5975,
      "step": 228
    },
    {
      "epoch": 0.9275949367088607,
      "grad_norm": 0.7157505625944839,
      "learning_rate": 3.910187103958837e-05,
      "loss": 0.6019,
      "step": 229
    },
    {
      "epoch": 0.9316455696202531,
      "grad_norm": 0.4652247278302937,
      "learning_rate": 3.908497629423117e-05,
      "loss": 0.5904,
      "step": 230
    },
    {
      "epoch": 0.9356962025316455,
      "grad_norm": 0.6552376372616936,
      "learning_rate": 3.9067927841098614e-05,
      "loss": 0.5943,
      "step": 231
    },
    {
      "epoch": 0.9397468354430379,
      "grad_norm": 0.7142902168970736,
      "learning_rate": 3.9050725817496594e-05,
      "loss": 0.6188,
      "step": 232
    },
    {
      "epoch": 0.9437974683544303,
      "grad_norm": 0.614359786592411,
      "learning_rate": 3.9033370361967844e-05,
      "loss": 0.5995,
      "step": 233
    },
    {
      "epoch": 0.9478481012658228,
      "grad_norm": 0.5890522049738569,
      "learning_rate": 3.901586161429081e-05,
      "loss": 0.6062,
      "step": 234
    },
    {
      "epoch": 0.9518987341772152,
      "grad_norm": 0.5689810380195013,
      "learning_rate": 3.8998199715478545e-05,
      "loss": 0.6001,
      "step": 235
    },
    {
      "epoch": 0.9559493670886076,
      "grad_norm": 0.4291509046067673,
      "learning_rate": 3.8980384807777564e-05,
      "loss": 0.5898,
      "step": 236
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5121008007374366,
      "learning_rate": 3.896241703466667e-05,
      "loss": 0.6036,
      "step": 237
    },
    {
      "epoch": 0.9640506329113924,
      "grad_norm": 0.4813346213177489,
      "learning_rate": 3.894429654085585e-05,
      "loss": 0.5997,
      "step": 238
    },
    {
      "epoch": 0.9681012658227848,
      "grad_norm": 0.4793607764038437,
      "learning_rate": 3.892602347228505e-05,
      "loss": 0.5959,
      "step": 239
    },
    {
      "epoch": 0.9721518987341772,
      "grad_norm": 0.4478139598266485,
      "learning_rate": 3.890759797612307e-05,
      "loss": 0.5923,
      "step": 240
    },
    {
      "epoch": 0.9762025316455696,
      "grad_norm": 0.433162071207074,
      "learning_rate": 3.888902020076632e-05,
      "loss": 0.5936,
      "step": 241
    },
    {
      "epoch": 0.980253164556962,
      "grad_norm": 0.5729835230052921,
      "learning_rate": 3.887029029583764e-05,
      "loss": 0.5946,
      "step": 242
    },
    {
      "epoch": 0.9843037974683544,
      "grad_norm": 0.4347336495590074,
      "learning_rate": 3.8851408412185125e-05,
      "loss": 0.5929,
      "step": 243
    },
    {
      "epoch": 0.9883544303797468,
      "grad_norm": 0.5976960746396264,
      "learning_rate": 3.8832374701880855e-05,
      "loss": 0.5933,
      "step": 244
    },
    {
      "epoch": 0.9924050632911392,
      "grad_norm": 0.5338444592333565,
      "learning_rate": 3.881318931821972e-05,
      "loss": 0.5885,
      "step": 245
    },
    {
      "epoch": 0.9964556962025316,
      "grad_norm": 0.5303643665077618,
      "learning_rate": 3.879385241571817e-05,
      "loss": 0.5995,
      "step": 246
    },
    {
      "epoch": 1.000506329113924,
      "grad_norm": 0.5500071608632071,
      "learning_rate": 3.8774364150112955e-05,
      "loss": 0.587,
      "step": 247
    },
    {
      "epoch": 1.0045569620253165,
      "grad_norm": 0.47554915240229956,
      "learning_rate": 3.8754724678359884e-05,
      "loss": 0.5514,
      "step": 248
    },
    {
      "epoch": 1.0086075949367088,
      "grad_norm": 0.4992773099624107,
      "learning_rate": 3.873493415863256e-05,
      "loss": 0.5638,
      "step": 249
    },
    {
      "epoch": 1.0126582278481013,
      "grad_norm": 0.6072353660091911,
      "learning_rate": 3.871499275032111e-05,
      "loss": 0.5534,
      "step": 250
    },
    {
      "epoch": 1.0167088607594936,
      "grad_norm": 0.5181358091193021,
      "learning_rate": 3.869490061403091e-05,
      "loss": 0.5529,
      "step": 251
    },
    {
      "epoch": 1.0207594936708861,
      "grad_norm": 0.48118018117331696,
      "learning_rate": 3.867465791158124e-05,
      "loss": 0.5509,
      "step": 252
    },
    {
      "epoch": 1.0248101265822784,
      "grad_norm": 0.5357875757629944,
      "learning_rate": 3.865426480600407e-05,
      "loss": 0.5602,
      "step": 253
    },
    {
      "epoch": 1.028860759493671,
      "grad_norm": 0.6024913819748591,
      "learning_rate": 3.863372146154264e-05,
      "loss": 0.5595,
      "step": 254
    },
    {
      "epoch": 1.0329113924050632,
      "grad_norm": 0.5122244261222243,
      "learning_rate": 3.861302804365024e-05,
      "loss": 0.5646,
      "step": 255
    },
    {
      "epoch": 1.0369620253164558,
      "grad_norm": 0.4746427479000092,
      "learning_rate": 3.85921847189888e-05,
      "loss": 0.5392,
      "step": 256
    },
    {
      "epoch": 1.041012658227848,
      "grad_norm": 0.4273815868751707,
      "learning_rate": 3.85711916554276e-05,
      "loss": 0.5562,
      "step": 257
    },
    {
      "epoch": 1.0450632911392406,
      "grad_norm": 0.4241982358340587,
      "learning_rate": 3.85500490220419e-05,
      "loss": 0.5483,
      "step": 258
    },
    {
      "epoch": 1.0491139240506329,
      "grad_norm": 0.42810644286613164,
      "learning_rate": 3.852875698911154e-05,
      "loss": 0.5572,
      "step": 259
    },
    {
      "epoch": 1.0531645569620254,
      "grad_norm": 0.4866573211882829,
      "learning_rate": 3.850731572811963e-05,
      "loss": 0.5514,
      "step": 260
    },
    {
      "epoch": 1.0572151898734177,
      "grad_norm": 0.7019981598437524,
      "learning_rate": 3.848572541175116e-05,
      "loss": 0.5434,
      "step": 261
    },
    {
      "epoch": 1.0612658227848102,
      "grad_norm": 0.6509258776076047,
      "learning_rate": 3.846398621389154e-05,
      "loss": 0.5269,
      "step": 262
    },
    {
      "epoch": 1.0653164556962025,
      "grad_norm": 0.47697440269419494,
      "learning_rate": 3.84420983096253e-05,
      "loss": 0.5577,
      "step": 263
    },
    {
      "epoch": 1.069367088607595,
      "grad_norm": 0.629370595703202,
      "learning_rate": 3.8420061875234606e-05,
      "loss": 0.5521,
      "step": 264
    },
    {
      "epoch": 1.0734177215189873,
      "grad_norm": 0.4709554820325284,
      "learning_rate": 3.839787708819787e-05,
      "loss": 0.5525,
      "step": 265
    },
    {
      "epoch": 1.0774683544303798,
      "grad_norm": 0.46024366360545865,
      "learning_rate": 3.8375544127188325e-05,
      "loss": 0.5513,
      "step": 266
    },
    {
      "epoch": 1.081518987341772,
      "grad_norm": 0.4829546452797468,
      "learning_rate": 3.8353063172072564e-05,
      "loss": 0.5455,
      "step": 267
    },
    {
      "epoch": 1.0855696202531646,
      "grad_norm": 0.49588600155937496,
      "learning_rate": 3.8330434403909105e-05,
      "loss": 0.5425,
      "step": 268
    },
    {
      "epoch": 1.089620253164557,
      "grad_norm": 0.45222460171716566,
      "learning_rate": 3.8307658004946934e-05,
      "loss": 0.5542,
      "step": 269
    },
    {
      "epoch": 1.0936708860759494,
      "grad_norm": 0.48141903592195834,
      "learning_rate": 3.8284734158624046e-05,
      "loss": 0.5569,
      "step": 270
    },
    {
      "epoch": 1.0977215189873417,
      "grad_norm": 0.45639846540679607,
      "learning_rate": 3.826166304956594e-05,
      "loss": 0.5518,
      "step": 271
    },
    {
      "epoch": 1.1017721518987342,
      "grad_norm": 0.5150877153095482,
      "learning_rate": 3.8238444863584164e-05,
      "loss": 0.5645,
      "step": 272
    },
    {
      "epoch": 1.1058227848101265,
      "grad_norm": 0.47269601228990676,
      "learning_rate": 3.821507978767479e-05,
      "loss": 0.5416,
      "step": 273
    },
    {
      "epoch": 1.109873417721519,
      "grad_norm": 0.45054706027822283,
      "learning_rate": 3.819156801001693e-05,
      "loss": 0.5452,
      "step": 274
    },
    {
      "epoch": 1.1139240506329113,
      "grad_norm": 0.4288542529602165,
      "learning_rate": 3.816790971997121e-05,
      "loss": 0.5576,
      "step": 275
    },
    {
      "epoch": 1.1179746835443038,
      "grad_norm": 0.47553750754751306,
      "learning_rate": 3.8144105108078246e-05,
      "loss": 0.5372,
      "step": 276
    },
    {
      "epoch": 1.1220253164556961,
      "grad_norm": 0.45912063598866276,
      "learning_rate": 3.81201543660571e-05,
      "loss": 0.5568,
      "step": 277
    },
    {
      "epoch": 1.1260759493670887,
      "grad_norm": 0.533218488391862,
      "learning_rate": 3.809605768680377e-05,
      "loss": 0.5468,
      "step": 278
    },
    {
      "epoch": 1.130126582278481,
      "grad_norm": 0.5340660442916709,
      "learning_rate": 3.807181526438958e-05,
      "loss": 0.5302,
      "step": 279
    },
    {
      "epoch": 1.1341772151898735,
      "grad_norm": 0.5205756484859887,
      "learning_rate": 3.8047427294059697e-05,
      "loss": 0.5394,
      "step": 280
    },
    {
      "epoch": 1.1382278481012658,
      "grad_norm": 0.6326794284713414,
      "learning_rate": 3.802289397223145e-05,
      "loss": 0.5516,
      "step": 281
    },
    {
      "epoch": 1.1422784810126583,
      "grad_norm": 0.5752748780382161,
      "learning_rate": 3.7998215496492854e-05,
      "loss": 0.5468,
      "step": 282
    },
    {
      "epoch": 1.1463291139240506,
      "grad_norm": 0.7100075144875339,
      "learning_rate": 3.797339206560096e-05,
      "loss": 0.5583,
      "step": 283
    },
    {
      "epoch": 1.150379746835443,
      "grad_norm": 0.5194528593048473,
      "learning_rate": 3.794842387948027e-05,
      "loss": 0.5551,
      "step": 284
    },
    {
      "epoch": 1.1544303797468354,
      "grad_norm": 0.4495828385334631,
      "learning_rate": 3.7923311139221114e-05,
      "loss": 0.5437,
      "step": 285
    },
    {
      "epoch": 1.158481012658228,
      "grad_norm": 0.5473781759804024,
      "learning_rate": 3.7898054047078054e-05,
      "loss": 0.5444,
      "step": 286
    },
    {
      "epoch": 1.1625316455696202,
      "grad_norm": 0.571803502668593,
      "learning_rate": 3.787265280646825e-05,
      "loss": 0.5612,
      "step": 287
    },
    {
      "epoch": 1.1665822784810127,
      "grad_norm": 0.6358882536684146,
      "learning_rate": 3.7847107621969786e-05,
      "loss": 0.5419,
      "step": 288
    },
    {
      "epoch": 1.170632911392405,
      "grad_norm": 0.5690157108797107,
      "learning_rate": 3.7821418699320064e-05,
      "loss": 0.547,
      "step": 289
    },
    {
      "epoch": 1.1746835443037975,
      "grad_norm": 0.49333548405763644,
      "learning_rate": 3.7795586245414145e-05,
      "loss": 0.5506,
      "step": 290
    },
    {
      "epoch": 1.1787341772151898,
      "grad_norm": 0.5824239573734334,
      "learning_rate": 3.776961046830306e-05,
      "loss": 0.5367,
      "step": 291
    },
    {
      "epoch": 1.1827848101265823,
      "grad_norm": 0.6874573712018205,
      "learning_rate": 3.774349157719215e-05,
      "loss": 0.5489,
      "step": 292
    },
    {
      "epoch": 1.1868354430379746,
      "grad_norm": 0.5331898343471256,
      "learning_rate": 3.7717229782439365e-05,
      "loss": 0.554,
      "step": 293
    },
    {
      "epoch": 1.1908860759493671,
      "grad_norm": 0.45159237925798407,
      "learning_rate": 3.769082529555359e-05,
      "loss": 0.5447,
      "step": 294
    },
    {
      "epoch": 1.1949367088607594,
      "grad_norm": 0.46653595802228726,
      "learning_rate": 3.766427832919294e-05,
      "loss": 0.5377,
      "step": 295
    },
    {
      "epoch": 1.198987341772152,
      "grad_norm": 0.6102207124766773,
      "learning_rate": 3.7637589097163024e-05,
      "loss": 0.566,
      "step": 296
    },
    {
      "epoch": 1.2030379746835442,
      "grad_norm": 0.559336763542953,
      "learning_rate": 3.761075781441526e-05,
      "loss": 0.5644,
      "step": 297
    },
    {
      "epoch": 1.2070886075949367,
      "grad_norm": 0.4237326266656369,
      "learning_rate": 3.75837846970451e-05,
      "loss": 0.5484,
      "step": 298
    },
    {
      "epoch": 1.211139240506329,
      "grad_norm": 0.4295195248427764,
      "learning_rate": 3.755666996229032e-05,
      "loss": 0.5331,
      "step": 299
    },
    {
      "epoch": 1.2151898734177216,
      "grad_norm": 0.5285807695162228,
      "learning_rate": 3.752941382852927e-05,
      "loss": 0.5554,
      "step": 300
    },
    {
      "epoch": 1.2192405063291138,
      "grad_norm": 0.5789596902987457,
      "learning_rate": 3.7502016515279115e-05,
      "loss": 0.5446,
      "step": 301
    },
    {
      "epoch": 1.2232911392405064,
      "grad_norm": 0.4545036984330519,
      "learning_rate": 3.7474478243194043e-05,
      "loss": 0.5576,
      "step": 302
    },
    {
      "epoch": 1.2273417721518987,
      "grad_norm": 0.5013885182345832,
      "learning_rate": 3.744679923406351e-05,
      "loss": 0.5588,
      "step": 303
    },
    {
      "epoch": 1.2313924050632912,
      "grad_norm": 0.6072046752643824,
      "learning_rate": 3.741897971081043e-05,
      "loss": 0.5639,
      "step": 304
    },
    {
      "epoch": 1.2354430379746835,
      "grad_norm": 0.5815730659431051,
      "learning_rate": 3.739101989748946e-05,
      "loss": 0.5526,
      "step": 305
    },
    {
      "epoch": 1.239493670886076,
      "grad_norm": 0.5305379835506069,
      "learning_rate": 3.7362920019285066e-05,
      "loss": 0.5411,
      "step": 306
    },
    {
      "epoch": 1.2435443037974683,
      "grad_norm": 0.48569343206796184,
      "learning_rate": 3.73346803025098e-05,
      "loss": 0.5389,
      "step": 307
    },
    {
      "epoch": 1.2475949367088608,
      "grad_norm": 0.584739034141005,
      "learning_rate": 3.730630097460247e-05,
      "loss": 0.5325,
      "step": 308
    },
    {
      "epoch": 1.251645569620253,
      "grad_norm": 0.4193077559961448,
      "learning_rate": 3.727778226412628e-05,
      "loss": 0.5504,
      "step": 309
    },
    {
      "epoch": 1.2556962025316456,
      "grad_norm": 0.5526901611302676,
      "learning_rate": 3.7249124400767006e-05,
      "loss": 0.5495,
      "step": 310
    },
    {
      "epoch": 1.259746835443038,
      "grad_norm": 0.7133154742039458,
      "learning_rate": 3.722032761533114e-05,
      "loss": 0.5524,
      "step": 311
    },
    {
      "epoch": 1.2637974683544304,
      "grad_norm": 0.48721207071453104,
      "learning_rate": 3.719139213974403e-05,
      "loss": 0.5457,
      "step": 312
    },
    {
      "epoch": 1.2678481012658227,
      "grad_norm": 0.45461213715281523,
      "learning_rate": 3.7162318207048006e-05,
      "loss": 0.5393,
      "step": 313
    },
    {
      "epoch": 1.2718987341772152,
      "grad_norm": 0.5106805479818406,
      "learning_rate": 3.713310605140055e-05,
      "loss": 0.5391,
      "step": 314
    },
    {
      "epoch": 1.2759493670886077,
      "grad_norm": 0.4140392259424703,
      "learning_rate": 3.710375590807233e-05,
      "loss": 0.5398,
      "step": 315
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.4164917853050876,
      "learning_rate": 3.7074268013445365e-05,
      "loss": 0.5656,
      "step": 316
    },
    {
      "epoch": 1.2840506329113923,
      "grad_norm": 0.4682037370853518,
      "learning_rate": 3.7044642605011114e-05,
      "loss": 0.5398,
      "step": 317
    },
    {
      "epoch": 1.2881012658227848,
      "grad_norm": 0.4602566288065545,
      "learning_rate": 3.701487992136854e-05,
      "loss": 0.5461,
      "step": 318
    },
    {
      "epoch": 1.2921518987341774,
      "grad_norm": 0.40355892984393266,
      "learning_rate": 3.69849802022222e-05,
      "loss": 0.5424,
      "step": 319
    },
    {
      "epoch": 1.2962025316455696,
      "grad_norm": 0.5030157465478555,
      "learning_rate": 3.6954943688380334e-05,
      "loss": 0.5527,
      "step": 320
    },
    {
      "epoch": 1.300253164556962,
      "grad_norm": 0.4318964729187145,
      "learning_rate": 3.692477062175289e-05,
      "loss": 0.5497,
      "step": 321
    },
    {
      "epoch": 1.3043037974683545,
      "grad_norm": 0.394028927160198,
      "learning_rate": 3.689446124534958e-05,
      "loss": 0.5325,
      "step": 322
    },
    {
      "epoch": 1.308354430379747,
      "grad_norm": 0.42954927562396006,
      "learning_rate": 3.686401580327799e-05,
      "loss": 0.5574,
      "step": 323
    },
    {
      "epoch": 1.3124050632911393,
      "grad_norm": 0.4526565447881357,
      "learning_rate": 3.683343454074149e-05,
      "loss": 0.5536,
      "step": 324
    },
    {
      "epoch": 1.3164556962025316,
      "grad_norm": 0.4779515164221075,
      "learning_rate": 3.6802717704037386e-05,
      "loss": 0.5328,
      "step": 325
    },
    {
      "epoch": 1.320506329113924,
      "grad_norm": 0.40347027199928304,
      "learning_rate": 3.6771865540554855e-05,
      "loss": 0.5393,
      "step": 326
    },
    {
      "epoch": 1.3245569620253166,
      "grad_norm": 0.5572561392412894,
      "learning_rate": 3.674087829877297e-05,
      "loss": 0.5522,
      "step": 327
    },
    {
      "epoch": 1.3286075949367089,
      "grad_norm": 0.40317352917455396,
      "learning_rate": 3.6709756228258735e-05,
      "loss": 0.5533,
      "step": 328
    },
    {
      "epoch": 1.3326582278481012,
      "grad_norm": 0.5678906492158503,
      "learning_rate": 3.667849957966501e-05,
      "loss": 0.5548,
      "step": 329
    },
    {
      "epoch": 1.3367088607594937,
      "grad_norm": 0.47921280991988013,
      "learning_rate": 3.6647108604728546e-05,
      "loss": 0.5458,
      "step": 330
    },
    {
      "epoch": 1.3407594936708862,
      "grad_norm": 0.5123511240169487,
      "learning_rate": 3.661558355626795e-05,
      "loss": 0.5514,
      "step": 331
    },
    {
      "epoch": 1.3448101265822785,
      "grad_norm": 0.4020620349725275,
      "learning_rate": 3.658392468818163e-05,
      "loss": 0.5523,
      "step": 332
    },
    {
      "epoch": 1.3488607594936708,
      "grad_norm": 0.47002920021294187,
      "learning_rate": 3.655213225544574e-05,
      "loss": 0.5496,
      "step": 333
    },
    {
      "epoch": 1.3529113924050633,
      "grad_norm": 0.5040830445834008,
      "learning_rate": 3.652020651411218e-05,
      "loss": 0.5469,
      "step": 334
    },
    {
      "epoch": 1.3569620253164558,
      "grad_norm": 0.4230989726236193,
      "learning_rate": 3.6488147721306474e-05,
      "loss": 0.5432,
      "step": 335
    },
    {
      "epoch": 1.3610126582278481,
      "grad_norm": 0.4285348583010353,
      "learning_rate": 3.645595613522574e-05,
      "loss": 0.5492,
      "step": 336
    },
    {
      "epoch": 1.3650632911392404,
      "grad_norm": 0.47567361635291877,
      "learning_rate": 3.642363201513657e-05,
      "loss": 0.5472,
      "step": 337
    },
    {
      "epoch": 1.369113924050633,
      "grad_norm": 0.5605263023128432,
      "learning_rate": 3.6391175621373006e-05,
      "loss": 0.5519,
      "step": 338
    },
    {
      "epoch": 1.3731645569620254,
      "grad_norm": 0.37552053212175884,
      "learning_rate": 3.6358587215334355e-05,
      "loss": 0.554,
      "step": 339
    },
    {
      "epoch": 1.3772151898734177,
      "grad_norm": 0.4997050340494049,
      "learning_rate": 3.632586705948318e-05,
      "loss": 0.5449,
      "step": 340
    },
    {
      "epoch": 1.38126582278481,
      "grad_norm": 0.5562556711374784,
      "learning_rate": 3.629301541734311e-05,
      "loss": 0.5362,
      "step": 341
    },
    {
      "epoch": 1.3853164556962025,
      "grad_norm": 0.48991140891773255,
      "learning_rate": 3.626003255349676e-05,
      "loss": 0.5538,
      "step": 342
    },
    {
      "epoch": 1.389367088607595,
      "grad_norm": 0.4561967022339903,
      "learning_rate": 3.622691873358357e-05,
      "loss": 0.5471,
      "step": 343
    },
    {
      "epoch": 1.3934177215189874,
      "grad_norm": 0.3954695649456603,
      "learning_rate": 3.61936742242977e-05,
      "loss": 0.5402,
      "step": 344
    },
    {
      "epoch": 1.3974683544303796,
      "grad_norm": 0.47077689118293536,
      "learning_rate": 3.6160299293385864e-05,
      "loss": 0.5476,
      "step": 345
    },
    {
      "epoch": 1.4015189873417722,
      "grad_norm": 0.36429771524009263,
      "learning_rate": 3.612679420964516e-05,
      "loss": 0.5545,
      "step": 346
    },
    {
      "epoch": 1.4055696202531647,
      "grad_norm": 0.4253212657113402,
      "learning_rate": 3.609315924292092e-05,
      "loss": 0.551,
      "step": 347
    },
    {
      "epoch": 1.409620253164557,
      "grad_norm": 0.39418758899046846,
      "learning_rate": 3.6059394664104554e-05,
      "loss": 0.544,
      "step": 348
    },
    {
      "epoch": 1.4136708860759493,
      "grad_norm": 0.441803094431973,
      "learning_rate": 3.602550074513133e-05,
      "loss": 0.5428,
      "step": 349
    },
    {
      "epoch": 1.4177215189873418,
      "grad_norm": 0.4399055634781674,
      "learning_rate": 3.599147775897822e-05,
      "loss": 0.5476,
      "step": 350
    },
    {
      "epoch": 1.4217721518987343,
      "grad_norm": 0.4458150877037388,
      "learning_rate": 3.595732597966167e-05,
      "loss": 0.5414,
      "step": 351
    },
    {
      "epoch": 1.4258227848101266,
      "grad_norm": 0.44899552827415684,
      "learning_rate": 3.592304568223542e-05,
      "loss": 0.5432,
      "step": 352
    },
    {
      "epoch": 1.4298734177215189,
      "grad_norm": 0.35170470372855966,
      "learning_rate": 3.588863714278826e-05,
      "loss": 0.5441,
      "step": 353
    },
    {
      "epoch": 1.4339240506329114,
      "grad_norm": 0.3987402435012038,
      "learning_rate": 3.585410063844186e-05,
      "loss": 0.5455,
      "step": 354
    },
    {
      "epoch": 1.437974683544304,
      "grad_norm": 0.4610527746655635,
      "learning_rate": 3.581943644734846e-05,
      "loss": 0.543,
      "step": 355
    },
    {
      "epoch": 1.4420253164556962,
      "grad_norm": 0.3820155227887942,
      "learning_rate": 3.578464484868869e-05,
      "loss": 0.5431,
      "step": 356
    },
    {
      "epoch": 1.4460759493670885,
      "grad_norm": 0.40538970543505604,
      "learning_rate": 3.5749726122669316e-05,
      "loss": 0.5483,
      "step": 357
    },
    {
      "epoch": 1.450126582278481,
      "grad_norm": 0.4941609680886557,
      "learning_rate": 3.5714680550520943e-05,
      "loss": 0.5434,
      "step": 358
    },
    {
      "epoch": 1.4541772151898735,
      "grad_norm": 0.541997276032994,
      "learning_rate": 3.5679508414495794e-05,
      "loss": 0.5477,
      "step": 359
    },
    {
      "epoch": 1.4582278481012658,
      "grad_norm": 0.5642857519590568,
      "learning_rate": 3.564420999786543e-05,
      "loss": 0.5453,
      "step": 360
    },
    {
      "epoch": 1.4622784810126581,
      "grad_norm": 0.5323601713654209,
      "learning_rate": 3.560878558491842e-05,
      "loss": 0.5545,
      "step": 361
    },
    {
      "epoch": 1.4663291139240506,
      "grad_norm": 0.5159977383844734,
      "learning_rate": 3.5573235460958145e-05,
      "loss": 0.5365,
      "step": 362
    },
    {
      "epoch": 1.4703797468354431,
      "grad_norm": 0.4427609450451662,
      "learning_rate": 3.553755991230039e-05,
      "loss": 0.5473,
      "step": 363
    },
    {
      "epoch": 1.4744303797468354,
      "grad_norm": 0.47963303821292386,
      "learning_rate": 3.5501759226271144e-05,
      "loss": 0.5437,
      "step": 364
    },
    {
      "epoch": 1.4784810126582277,
      "grad_norm": 0.3692107078743929,
      "learning_rate": 3.546583369120419e-05,
      "loss": 0.5412,
      "step": 365
    },
    {
      "epoch": 1.4825316455696202,
      "grad_norm": 0.4726027151812385,
      "learning_rate": 3.5429783596438864e-05,
      "loss": 0.5487,
      "step": 366
    },
    {
      "epoch": 1.4865822784810128,
      "grad_norm": 0.5477158914701177,
      "learning_rate": 3.539360923231766e-05,
      "loss": 0.5495,
      "step": 367
    },
    {
      "epoch": 1.490632911392405,
      "grad_norm": 0.5112201820786759,
      "learning_rate": 3.535731089018394e-05,
      "loss": 0.5325,
      "step": 368
    },
    {
      "epoch": 1.4946835443037974,
      "grad_norm": 0.41235035484151883,
      "learning_rate": 3.532088886237956e-05,
      "loss": 0.5264,
      "step": 369
    },
    {
      "epoch": 1.4987341772151899,
      "grad_norm": 0.5601143157981107,
      "learning_rate": 3.528434344224253e-05,
      "loss": 0.5418,
      "step": 370
    },
    {
      "epoch": 1.5027848101265824,
      "grad_norm": 0.46037913089000987,
      "learning_rate": 3.524767492410464e-05,
      "loss": 0.5426,
      "step": 371
    },
    {
      "epoch": 1.5068354430379747,
      "grad_norm": 0.42209428923513703,
      "learning_rate": 3.521088360328908e-05,
      "loss": 0.5526,
      "step": 372
    },
    {
      "epoch": 1.510886075949367,
      "grad_norm": 0.411234079680589,
      "learning_rate": 3.517396977610811e-05,
      "loss": 0.5401,
      "step": 373
    },
    {
      "epoch": 1.5149367088607595,
      "grad_norm": 0.47642025260768994,
      "learning_rate": 3.5136933739860595e-05,
      "loss": 0.5466,
      "step": 374
    },
    {
      "epoch": 1.518987341772152,
      "grad_norm": 0.5099760316987239,
      "learning_rate": 3.509977579282971e-05,
      "loss": 0.5442,
      "step": 375
    },
    {
      "epoch": 1.5230379746835443,
      "grad_norm": 0.40371040377670553,
      "learning_rate": 3.5062496234280424e-05,
      "loss": 0.5477,
      "step": 376
    },
    {
      "epoch": 1.5270886075949366,
      "grad_norm": 0.5062065127252333,
      "learning_rate": 3.502509536445719e-05,
      "loss": 0.5538,
      "step": 377
    },
    {
      "epoch": 1.531139240506329,
      "grad_norm": 0.4234150075360343,
      "learning_rate": 3.498757348458147e-05,
      "loss": 0.5421,
      "step": 378
    },
    {
      "epoch": 1.5351898734177216,
      "grad_norm": 0.38730022641140516,
      "learning_rate": 3.4949930896849324e-05,
      "loss": 0.5451,
      "step": 379
    },
    {
      "epoch": 1.539240506329114,
      "grad_norm": 0.33241443525402675,
      "learning_rate": 3.491216790442899e-05,
      "loss": 0.546,
      "step": 380
    },
    {
      "epoch": 1.5432911392405062,
      "grad_norm": 0.4284241369469206,
      "learning_rate": 3.487428481145839e-05,
      "loss": 0.5548,
      "step": 381
    },
    {
      "epoch": 1.5473417721518987,
      "grad_norm": 0.37376058916670324,
      "learning_rate": 3.483628192304278e-05,
      "loss": 0.5359,
      "step": 382
    },
    {
      "epoch": 1.5513924050632912,
      "grad_norm": 0.36837594114970984,
      "learning_rate": 3.479815954525219e-05,
      "loss": 0.5427,
      "step": 383
    },
    {
      "epoch": 1.5554430379746835,
      "grad_norm": 0.3879312011678323,
      "learning_rate": 3.475991798511899e-05,
      "loss": 0.5438,
      "step": 384
    },
    {
      "epoch": 1.5594936708860758,
      "grad_norm": 0.39649417439881224,
      "learning_rate": 3.4721557550635464e-05,
      "loss": 0.5539,
      "step": 385
    },
    {
      "epoch": 1.5635443037974683,
      "grad_norm": 0.35506664906645824,
      "learning_rate": 3.468307855075128e-05,
      "loss": 0.5501,
      "step": 386
    },
    {
      "epoch": 1.5675949367088609,
      "grad_norm": 0.4174460997170703,
      "learning_rate": 3.4644481295371005e-05,
      "loss": 0.5363,
      "step": 387
    },
    {
      "epoch": 1.5716455696202531,
      "grad_norm": 0.32226208469568524,
      "learning_rate": 3.460576609535163e-05,
      "loss": 0.5593,
      "step": 388
    },
    {
      "epoch": 1.5756962025316454,
      "grad_norm": 0.4045986822087714,
      "learning_rate": 3.456693326250006e-05,
      "loss": 0.5378,
      "step": 389
    },
    {
      "epoch": 1.579746835443038,
      "grad_norm": 0.32051097916176075,
      "learning_rate": 3.452798310957058e-05,
      "loss": 0.5536,
      "step": 390
    },
    {
      "epoch": 1.5837974683544305,
      "grad_norm": 0.33587806002092474,
      "learning_rate": 3.4488915950262386e-05,
      "loss": 0.5454,
      "step": 391
    },
    {
      "epoch": 1.5878481012658228,
      "grad_norm": 0.3283821046323556,
      "learning_rate": 3.4449732099216985e-05,
      "loss": 0.5382,
      "step": 392
    },
    {
      "epoch": 1.591898734177215,
      "grad_norm": 0.3775819330210603,
      "learning_rate": 3.441043187201574e-05,
      "loss": 0.5383,
      "step": 393
    },
    {
      "epoch": 1.5959493670886076,
      "grad_norm": 0.4462507037045206,
      "learning_rate": 3.437101558517728e-05,
      "loss": 0.5461,
      "step": 394
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.39146848476142787,
      "learning_rate": 3.433148355615496e-05,
      "loss": 0.539,
      "step": 395
    },
    {
      "epoch": 1.6040506329113924,
      "grad_norm": 0.44343126791856813,
      "learning_rate": 3.4291836103334294e-05,
      "loss": 0.5446,
      "step": 396
    },
    {
      "epoch": 1.6081012658227847,
      "grad_norm": 0.41580431921978134,
      "learning_rate": 3.425207354603043e-05,
      "loss": 0.5418,
      "step": 397
    },
    {
      "epoch": 1.6121518987341772,
      "grad_norm": 0.34714839307359463,
      "learning_rate": 3.421219620448553e-05,
      "loss": 0.5305,
      "step": 398
    },
    {
      "epoch": 1.6162025316455697,
      "grad_norm": 0.36359874379967816,
      "learning_rate": 3.417220439986623e-05,
      "loss": 0.5449,
      "step": 399
    },
    {
      "epoch": 1.620253164556962,
      "grad_norm": 0.358430464566614,
      "learning_rate": 3.4132098454261024e-05,
      "loss": 0.554,
      "step": 400
    },
    {
      "epoch": 1.6243037974683543,
      "grad_norm": 0.40237926786474026,
      "learning_rate": 3.4091878690677676e-05,
      "loss": 0.5484,
      "step": 401
    },
    {
      "epoch": 1.6283544303797468,
      "grad_norm": 0.3737579658803788,
      "learning_rate": 3.405154543304065e-05,
      "loss": 0.5404,
      "step": 402
    },
    {
      "epoch": 1.6324050632911393,
      "grad_norm": 0.3599967872924873,
      "learning_rate": 3.401109900618843e-05,
      "loss": 0.5455,
      "step": 403
    },
    {
      "epoch": 1.6364556962025316,
      "grad_norm": 0.3530018194588855,
      "learning_rate": 3.3970539735870996e-05,
      "loss": 0.545,
      "step": 404
    },
    {
      "epoch": 1.640506329113924,
      "grad_norm": 0.2814475653399262,
      "learning_rate": 3.392986794874714e-05,
      "loss": 0.5355,
      "step": 405
    },
    {
      "epoch": 1.6445569620253164,
      "grad_norm": 0.3190457686061442,
      "learning_rate": 3.388908397238184e-05,
      "loss": 0.5284,
      "step": 406
    },
    {
      "epoch": 1.648607594936709,
      "grad_norm": 0.30743908814888693,
      "learning_rate": 3.384818813524362e-05,
      "loss": 0.5409,
      "step": 407
    },
    {
      "epoch": 1.6526582278481012,
      "grad_norm": 0.34052422733021664,
      "learning_rate": 3.380718076670195e-05,
      "loss": 0.543,
      "step": 408
    },
    {
      "epoch": 1.6567088607594935,
      "grad_norm": 0.33979173186098643,
      "learning_rate": 3.376606219702454e-05,
      "loss": 0.5369,
      "step": 409
    },
    {
      "epoch": 1.660759493670886,
      "grad_norm": 0.3504167763147821,
      "learning_rate": 3.372483275737468e-05,
      "loss": 0.5496,
      "step": 410
    },
    {
      "epoch": 1.6648101265822786,
      "grad_norm": 0.3593326852304351,
      "learning_rate": 3.368349277980861e-05,
      "loss": 0.5359,
      "step": 411
    },
    {
      "epoch": 1.6688607594936709,
      "grad_norm": 0.36321133560412144,
      "learning_rate": 3.3642042597272844e-05,
      "loss": 0.5564,
      "step": 412
    },
    {
      "epoch": 1.6729113924050631,
      "grad_norm": 0.3459380973026731,
      "learning_rate": 3.360048254360144e-05,
      "loss": 0.5409,
      "step": 413
    },
    {
      "epoch": 1.6769620253164557,
      "grad_norm": 0.3814388950086124,
      "learning_rate": 3.355881295351336e-05,
      "loss": 0.5403,
      "step": 414
    },
    {
      "epoch": 1.6810126582278482,
      "grad_norm": 0.33422596461029963,
      "learning_rate": 3.351703416260975e-05,
      "loss": 0.5425,
      "step": 415
    },
    {
      "epoch": 1.6850632911392405,
      "grad_norm": 0.4005529329042578,
      "learning_rate": 3.347514650737126e-05,
      "loss": 0.5483,
      "step": 416
    },
    {
      "epoch": 1.6891139240506328,
      "grad_norm": 0.4689663913144769,
      "learning_rate": 3.3433150325155295e-05,
      "loss": 0.5394,
      "step": 417
    },
    {
      "epoch": 1.6931645569620253,
      "grad_norm": 0.4129597775171812,
      "learning_rate": 3.339104595419334e-05,
      "loss": 0.5474,
      "step": 418
    },
    {
      "epoch": 1.6972151898734178,
      "grad_norm": 0.359522282612482,
      "learning_rate": 3.3348833733588204e-05,
      "loss": 0.5394,
      "step": 419
    },
    {
      "epoch": 1.70126582278481,
      "grad_norm": 0.3911319108521518,
      "learning_rate": 3.3306514003311305e-05,
      "loss": 0.5374,
      "step": 420
    },
    {
      "epoch": 1.7053164556962024,
      "grad_norm": 0.39425084594829646,
      "learning_rate": 3.326408710419996e-05,
      "loss": 0.5374,
      "step": 421
    },
    {
      "epoch": 1.709367088607595,
      "grad_norm": 0.30569033685733393,
      "learning_rate": 3.322155337795454e-05,
      "loss": 0.532,
      "step": 422
    },
    {
      "epoch": 1.7134177215189874,
      "grad_norm": 0.4114689788440459,
      "learning_rate": 3.317891316713587e-05,
      "loss": 0.546,
      "step": 423
    },
    {
      "epoch": 1.7174683544303797,
      "grad_norm": 0.33406229895892525,
      "learning_rate": 3.313616681516231e-05,
      "loss": 0.5415,
      "step": 424
    },
    {
      "epoch": 1.721518987341772,
      "grad_norm": 0.41967888871029113,
      "learning_rate": 3.309331466630713e-05,
      "loss": 0.541,
      "step": 425
    },
    {
      "epoch": 1.7255696202531645,
      "grad_norm": 0.484167006840953,
      "learning_rate": 3.305035706569563e-05,
      "loss": 0.5305,
      "step": 426
    },
    {
      "epoch": 1.729620253164557,
      "grad_norm": 0.3139086327219068,
      "learning_rate": 3.3007294359302433e-05,
      "loss": 0.5394,
      "step": 427
    },
    {
      "epoch": 1.7336708860759493,
      "grad_norm": 0.46066214072594497,
      "learning_rate": 3.296412689394864e-05,
      "loss": 0.5546,
      "step": 428
    },
    {
      "epoch": 1.7377215189873416,
      "grad_norm": 0.4185469507503111,
      "learning_rate": 3.292085501729909e-05,
      "loss": 0.5345,
      "step": 429
    },
    {
      "epoch": 1.7417721518987341,
      "grad_norm": 0.41935690556386274,
      "learning_rate": 3.2877479077859534e-05,
      "loss": 0.5351,
      "step": 430
    },
    {
      "epoch": 1.7458227848101266,
      "grad_norm": 0.34296794400016556,
      "learning_rate": 3.283399942497381e-05,
      "loss": 0.5325,
      "step": 431
    },
    {
      "epoch": 1.749873417721519,
      "grad_norm": 0.4811290118980767,
      "learning_rate": 3.279041640882108e-05,
      "loss": 0.5492,
      "step": 432
    },
    {
      "epoch": 1.7539240506329112,
      "grad_norm": 0.43846278587864007,
      "learning_rate": 3.2746730380412964e-05,
      "loss": 0.5281,
      "step": 433
    },
    {
      "epoch": 1.7579746835443038,
      "grad_norm": 0.3796763443512012,
      "learning_rate": 3.2702941691590726e-05,
      "loss": 0.5361,
      "step": 434
    },
    {
      "epoch": 1.7620253164556963,
      "grad_norm": 0.385344437134616,
      "learning_rate": 3.265905069502244e-05,
      "loss": 0.5342,
      "step": 435
    },
    {
      "epoch": 1.7660759493670886,
      "grad_norm": 0.3254915092609221,
      "learning_rate": 3.261505774420016e-05,
      "loss": 0.5446,
      "step": 436
    },
    {
      "epoch": 1.7701265822784809,
      "grad_norm": 0.4029359908992895,
      "learning_rate": 3.257096319343707e-05,
      "loss": 0.5343,
      "step": 437
    },
    {
      "epoch": 1.7741772151898734,
      "grad_norm": 0.38983654386834293,
      "learning_rate": 3.2526767397864614e-05,
      "loss": 0.53,
      "step": 438
    },
    {
      "epoch": 1.7782278481012659,
      "grad_norm": 0.4021689513277025,
      "learning_rate": 3.248247071342966e-05,
      "loss": 0.5444,
      "step": 439
    },
    {
      "epoch": 1.7822784810126582,
      "grad_norm": 0.3607143499620443,
      "learning_rate": 3.243807349689161e-05,
      "loss": 0.5336,
      "step": 440
    },
    {
      "epoch": 1.7863291139240505,
      "grad_norm": 0.38248097317152213,
      "learning_rate": 3.2393576105819544e-05,
      "loss": 0.541,
      "step": 441
    },
    {
      "epoch": 1.790379746835443,
      "grad_norm": 0.4645642163819531,
      "learning_rate": 3.2348978898589333e-05,
      "loss": 0.544,
      "step": 442
    },
    {
      "epoch": 1.7944303797468355,
      "grad_norm": 0.36762229321673434,
      "learning_rate": 3.230428223438075e-05,
      "loss": 0.536,
      "step": 443
    },
    {
      "epoch": 1.7984810126582278,
      "grad_norm": 0.5031829811007891,
      "learning_rate": 3.225948647317459e-05,
      "loss": 0.547,
      "step": 444
    },
    {
      "epoch": 1.80253164556962,
      "grad_norm": 0.5046474170023305,
      "learning_rate": 3.2214591975749745e-05,
      "loss": 0.5303,
      "step": 445
    },
    {
      "epoch": 1.8065822784810126,
      "grad_norm": 0.36754667338138813,
      "learning_rate": 3.216959910368034e-05,
      "loss": 0.5421,
      "step": 446
    },
    {
      "epoch": 1.8106329113924051,
      "grad_norm": 0.5850658243554608,
      "learning_rate": 3.212450821933277e-05,
      "loss": 0.5495,
      "step": 447
    },
    {
      "epoch": 1.8146835443037974,
      "grad_norm": 0.4182716792735133,
      "learning_rate": 3.207931968586281e-05,
      "loss": 0.5494,
      "step": 448
    },
    {
      "epoch": 1.81873417721519,
      "grad_norm": 0.45043513670861124,
      "learning_rate": 3.203403386721272e-05,
      "loss": 0.5411,
      "step": 449
    },
    {
      "epoch": 1.8227848101265822,
      "grad_norm": 0.4038368882468168,
      "learning_rate": 3.1988651128108245e-05,
      "loss": 0.5372,
      "step": 450
    },
    {
      "epoch": 1.8268354430379747,
      "grad_norm": 0.4298717566855572,
      "learning_rate": 3.194317183405573e-05,
      "loss": 0.5315,
      "step": 451
    },
    {
      "epoch": 1.8308860759493673,
      "grad_norm": 0.4726151353762397,
      "learning_rate": 3.189759635133914e-05,
      "loss": 0.553,
      "step": 452
    },
    {
      "epoch": 1.8349367088607595,
      "grad_norm": 0.39174820163890167,
      "learning_rate": 3.185192504701718e-05,
      "loss": 0.5357,
      "step": 453
    },
    {
      "epoch": 1.8389873417721518,
      "grad_norm": 0.40237597754217974,
      "learning_rate": 3.1806158288920234e-05,
      "loss": 0.5423,
      "step": 454
    },
    {
      "epoch": 1.8430379746835444,
      "grad_norm": 0.4747406954262624,
      "learning_rate": 3.1760296445647477e-05,
      "loss": 0.547,
      "step": 455
    },
    {
      "epoch": 1.8470886075949369,
      "grad_norm": 0.3541130128504745,
      "learning_rate": 3.1714339886563896e-05,
      "loss": 0.5342,
      "step": 456
    },
    {
      "epoch": 1.8511392405063292,
      "grad_norm": 0.5084942759911607,
      "learning_rate": 3.166828898179731e-05,
      "loss": 0.5351,
      "step": 457
    },
    {
      "epoch": 1.8551898734177215,
      "grad_norm": 0.4189932373056671,
      "learning_rate": 3.162214410223536e-05,
      "loss": 0.5415,
      "step": 458
    },
    {
      "epoch": 1.859240506329114,
      "grad_norm": 0.40641601601318933,
      "learning_rate": 3.157590561952257e-05,
      "loss": 0.5298,
      "step": 459
    },
    {
      "epoch": 1.8632911392405065,
      "grad_norm": 0.39071157968080206,
      "learning_rate": 3.152957390605732e-05,
      "loss": 0.55,
      "step": 460
    },
    {
      "epoch": 1.8673417721518988,
      "grad_norm": 0.407840622797626,
      "learning_rate": 3.148314933498886e-05,
      "loss": 0.5363,
      "step": 461
    },
    {
      "epoch": 1.871392405063291,
      "grad_norm": 0.38872539171390313,
      "learning_rate": 3.143663228021431e-05,
      "loss": 0.5215,
      "step": 462
    },
    {
      "epoch": 1.8754430379746836,
      "grad_norm": 0.4294321426809109,
      "learning_rate": 3.1390023116375624e-05,
      "loss": 0.545,
      "step": 463
    },
    {
      "epoch": 1.879493670886076,
      "grad_norm": 0.4308652310896162,
      "learning_rate": 3.134332221885661e-05,
      "loss": 0.5375,
      "step": 464
    },
    {
      "epoch": 1.8835443037974684,
      "grad_norm": 0.3387274210920273,
      "learning_rate": 3.129652996377987e-05,
      "loss": 0.5283,
      "step": 465
    },
    {
      "epoch": 1.8875949367088607,
      "grad_norm": 0.40569876035698665,
      "learning_rate": 3.12496467280038e-05,
      "loss": 0.5432,
      "step": 466
    },
    {
      "epoch": 1.8916455696202532,
      "grad_norm": 0.35560049369197233,
      "learning_rate": 3.120267288911952e-05,
      "loss": 0.5279,
      "step": 467
    },
    {
      "epoch": 1.8956962025316457,
      "grad_norm": 0.4191037550397765,
      "learning_rate": 3.11556088254479e-05,
      "loss": 0.5321,
      "step": 468
    },
    {
      "epoch": 1.899746835443038,
      "grad_norm": 0.41871093330355297,
      "learning_rate": 3.11084549160364e-05,
      "loss": 0.5459,
      "step": 469
    },
    {
      "epoch": 1.9037974683544303,
      "grad_norm": 0.4319516986373976,
      "learning_rate": 3.106121154065615e-05,
      "loss": 0.5342,
      "step": 470
    },
    {
      "epoch": 1.9078481012658228,
      "grad_norm": 0.3370333795937814,
      "learning_rate": 3.1013879079798805e-05,
      "loss": 0.5403,
      "step": 471
    },
    {
      "epoch": 1.9118987341772153,
      "grad_norm": 0.3807589850618723,
      "learning_rate": 3.096645791467348e-05,
      "loss": 0.535,
      "step": 472
    },
    {
      "epoch": 1.9159493670886076,
      "grad_norm": 0.40311505539743026,
      "learning_rate": 3.091894842720373e-05,
      "loss": 0.5295,
      "step": 473
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.3922674265774569,
      "learning_rate": 3.0871351000024425e-05,
      "loss": 0.5396,
      "step": 474
    },
    {
      "epoch": 1.9240506329113924,
      "grad_norm": 0.3738344532460926,
      "learning_rate": 3.0823666016478716e-05,
      "loss": 0.5285,
      "step": 475
    },
    {
      "epoch": 1.928101265822785,
      "grad_norm": 0.44853215225367477,
      "learning_rate": 3.0775893860614896e-05,
      "loss": 0.5465,
      "step": 476
    },
    {
      "epoch": 1.9321518987341773,
      "grad_norm": 0.43311216786335766,
      "learning_rate": 3.0728034917183336e-05,
      "loss": 0.5457,
      "step": 477
    },
    {
      "epoch": 1.9362025316455695,
      "grad_norm": 0.34032032637744697,
      "learning_rate": 3.06800895716334e-05,
      "loss": 0.5449,
      "step": 478
    },
    {
      "epoch": 1.940253164556962,
      "grad_norm": 0.3931144547446437,
      "learning_rate": 3.063205821011029e-05,
      "loss": 0.5494,
      "step": 479
    },
    {
      "epoch": 1.9443037974683546,
      "grad_norm": 0.4070740566429497,
      "learning_rate": 3.0583941219452016e-05,
      "loss": 0.5342,
      "step": 480
    },
    {
      "epoch": 1.9483544303797469,
      "grad_norm": 0.33179746275159194,
      "learning_rate": 3.053573898718618e-05,
      "loss": 0.5416,
      "step": 481
    },
    {
      "epoch": 1.9524050632911392,
      "grad_norm": 0.34338178798351077,
      "learning_rate": 3.0487451901526956e-05,
      "loss": 0.5209,
      "step": 482
    },
    {
      "epoch": 1.9564556962025317,
      "grad_norm": 0.3647239671816951,
      "learning_rate": 3.0439080351371875e-05,
      "loss": 0.5394,
      "step": 483
    },
    {
      "epoch": 1.9605063291139242,
      "grad_norm": 0.31168158356926057,
      "learning_rate": 3.0390624726298764e-05,
      "loss": 0.5329,
      "step": 484
    },
    {
      "epoch": 1.9645569620253165,
      "grad_norm": 0.3648233022038886,
      "learning_rate": 3.034208541656255e-05,
      "loss": 0.5246,
      "step": 485
    },
    {
      "epoch": 1.9686075949367088,
      "grad_norm": 0.38633942162924395,
      "learning_rate": 3.029346281309218e-05,
      "loss": 0.5403,
      "step": 486
    },
    {
      "epoch": 1.9726582278481013,
      "grad_norm": 0.2922907360326553,
      "learning_rate": 3.0244757307487415e-05,
      "loss": 0.5397,
      "step": 487
    },
    {
      "epoch": 1.9767088607594938,
      "grad_norm": 0.4006882666755273,
      "learning_rate": 3.019596929201569e-05,
      "loss": 0.5247,
      "step": 488
    },
    {
      "epoch": 1.980759493670886,
      "grad_norm": 0.33899849711146635,
      "learning_rate": 3.0147099159608985e-05,
      "loss": 0.5345,
      "step": 489
    },
    {
      "epoch": 1.9848101265822784,
      "grad_norm": 0.37265611273226523,
      "learning_rate": 3.0098147303860616e-05,
      "loss": 0.5345,
      "step": 490
    },
    {
      "epoch": 1.988860759493671,
      "grad_norm": 0.38542510250047196,
      "learning_rate": 3.0049114119022117e-05,
      "loss": 0.5344,
      "step": 491
    },
    {
      "epoch": 1.9929113924050634,
      "grad_norm": 0.3201548157337632,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.536,
      "step": 492
    },
    {
      "epoch": 1.9969620253164557,
      "grad_norm": 0.4263237846259171,
      "learning_rate": 2.995080534235264e-05,
      "loss": 0.5381,
      "step": 493
    },
    {
      "epoch": 2.001012658227848,
      "grad_norm": 0.36218251181469985,
      "learning_rate": 2.9901530542287044e-05,
      "loss": 0.5266,
      "step": 494
    },
    {
      "epoch": 2.0050632911392405,
      "grad_norm": 0.43475503289619366,
      "learning_rate": 2.9852175996655676e-05,
      "loss": 0.4831,
      "step": 495
    },
    {
      "epoch": 2.009113924050633,
      "grad_norm": 0.5522471080433271,
      "learning_rate": 2.980274210295326e-05,
      "loss": 0.479,
      "step": 496
    },
    {
      "epoch": 2.013164556962025,
      "grad_norm": 0.3763731344340183,
      "learning_rate": 2.9753229259313578e-05,
      "loss": 0.4681,
      "step": 497
    },
    {
      "epoch": 2.0172151898734176,
      "grad_norm": 0.34598711850824376,
      "learning_rate": 2.9703637864506274e-05,
      "loss": 0.4842,
      "step": 498
    },
    {
      "epoch": 2.02126582278481,
      "grad_norm": 0.3782078547535528,
      "learning_rate": 2.965396831793362e-05,
      "loss": 0.4757,
      "step": 499
    },
    {
      "epoch": 2.0253164556962027,
      "grad_norm": 0.37458123967019824,
      "learning_rate": 2.9604221019627316e-05,
      "loss": 0.4819,
      "step": 500
    },
    {
      "epoch": 2.0293670886075947,
      "grad_norm": 0.33955315351754267,
      "learning_rate": 2.955439637024526e-05,
      "loss": 0.4839,
      "step": 501
    },
    {
      "epoch": 2.0334177215189873,
      "grad_norm": 0.40896110380817935,
      "learning_rate": 2.9504494771068334e-05,
      "loss": 0.4759,
      "step": 502
    },
    {
      "epoch": 2.0374683544303798,
      "grad_norm": 0.3605416805918467,
      "learning_rate": 2.9454516623997156e-05,
      "loss": 0.4939,
      "step": 503
    },
    {
      "epoch": 2.0415189873417723,
      "grad_norm": 0.4221400567759612,
      "learning_rate": 2.9404462331548847e-05,
      "loss": 0.4825,
      "step": 504
    },
    {
      "epoch": 2.0455696202531644,
      "grad_norm": 0.4190616759508317,
      "learning_rate": 2.93543322968538e-05,
      "loss": 0.4712,
      "step": 505
    },
    {
      "epoch": 2.049620253164557,
      "grad_norm": 0.3224446286046344,
      "learning_rate": 2.9304126923652428e-05,
      "loss": 0.4834,
      "step": 506
    },
    {
      "epoch": 2.0536708860759494,
      "grad_norm": 0.39847285228636997,
      "learning_rate": 2.9253846616291896e-05,
      "loss": 0.4668,
      "step": 507
    },
    {
      "epoch": 2.057721518987342,
      "grad_norm": 0.3564689682421963,
      "learning_rate": 2.9203491779722896e-05,
      "loss": 0.481,
      "step": 508
    },
    {
      "epoch": 2.061772151898734,
      "grad_norm": 0.3429867251651154,
      "learning_rate": 2.9153062819496357e-05,
      "loss": 0.4737,
      "step": 509
    },
    {
      "epoch": 2.0658227848101265,
      "grad_norm": 0.38063658223910224,
      "learning_rate": 2.9102560141760178e-05,
      "loss": 0.4709,
      "step": 510
    },
    {
      "epoch": 2.069873417721519,
      "grad_norm": 0.37830501811677425,
      "learning_rate": 2.9051984153256004e-05,
      "loss": 0.4726,
      "step": 511
    },
    {
      "epoch": 2.0739240506329115,
      "grad_norm": 0.38477889289713846,
      "learning_rate": 2.900133526131588e-05,
      "loss": 0.4828,
      "step": 512
    },
    {
      "epoch": 2.0779746835443036,
      "grad_norm": 0.4445786120257467,
      "learning_rate": 2.8950613873859025e-05,
      "loss": 0.468,
      "step": 513
    },
    {
      "epoch": 2.082025316455696,
      "grad_norm": 0.41793146648504415,
      "learning_rate": 2.8899820399388515e-05,
      "loss": 0.4721,
      "step": 514
    },
    {
      "epoch": 2.0860759493670886,
      "grad_norm": 0.3860261304768386,
      "learning_rate": 2.8848955246988012e-05,
      "loss": 0.4697,
      "step": 515
    },
    {
      "epoch": 2.090126582278481,
      "grad_norm": 0.3720994154655176,
      "learning_rate": 2.879801882631847e-05,
      "loss": 0.4825,
      "step": 516
    },
    {
      "epoch": 2.094177215189873,
      "grad_norm": 0.35385203146409083,
      "learning_rate": 2.8747011547614808e-05,
      "loss": 0.4633,
      "step": 517
    },
    {
      "epoch": 2.0982278481012657,
      "grad_norm": 0.3603771922977517,
      "learning_rate": 2.8695933821682635e-05,
      "loss": 0.4861,
      "step": 518
    },
    {
      "epoch": 2.1022784810126582,
      "grad_norm": 0.3778891319887707,
      "learning_rate": 2.864478605989494e-05,
      "loss": 0.4681,
      "step": 519
    },
    {
      "epoch": 2.1063291139240508,
      "grad_norm": 0.349902238703878,
      "learning_rate": 2.8593568674188765e-05,
      "loss": 0.4929,
      "step": 520
    },
    {
      "epoch": 2.110379746835443,
      "grad_norm": 0.3746456445697909,
      "learning_rate": 2.8542282077061892e-05,
      "loss": 0.4749,
      "step": 521
    },
    {
      "epoch": 2.1144303797468353,
      "grad_norm": 0.3639611067974695,
      "learning_rate": 2.8490926681569523e-05,
      "loss": 0.4653,
      "step": 522
    },
    {
      "epoch": 2.118481012658228,
      "grad_norm": 0.3916102310017982,
      "learning_rate": 2.8439502901320956e-05,
      "loss": 0.4885,
      "step": 523
    },
    {
      "epoch": 2.1225316455696204,
      "grad_norm": 0.4047614105274411,
      "learning_rate": 2.8388011150476237e-05,
      "loss": 0.4793,
      "step": 524
    },
    {
      "epoch": 2.1265822784810124,
      "grad_norm": 0.38664428878208346,
      "learning_rate": 2.8336451843742866e-05,
      "loss": 0.4726,
      "step": 525
    },
    {
      "epoch": 2.130632911392405,
      "grad_norm": 0.3976255417102989,
      "learning_rate": 2.8284825396372387e-05,
      "loss": 0.4809,
      "step": 526
    },
    {
      "epoch": 2.1346835443037975,
      "grad_norm": 0.4235761412797531,
      "learning_rate": 2.8233132224157132e-05,
      "loss": 0.4646,
      "step": 527
    },
    {
      "epoch": 2.13873417721519,
      "grad_norm": 0.4003990097045247,
      "learning_rate": 2.8181372743426805e-05,
      "loss": 0.4603,
      "step": 528
    },
    {
      "epoch": 2.1427848101265825,
      "grad_norm": 0.35581775666365445,
      "learning_rate": 2.8129547371045128e-05,
      "loss": 0.4776,
      "step": 529
    },
    {
      "epoch": 2.1468354430379746,
      "grad_norm": 0.41567461108049586,
      "learning_rate": 2.8077656524406534e-05,
      "loss": 0.4775,
      "step": 530
    },
    {
      "epoch": 2.150886075949367,
      "grad_norm": 0.35808509338973943,
      "learning_rate": 2.802570062143278e-05,
      "loss": 0.4668,
      "step": 531
    },
    {
      "epoch": 2.1549367088607596,
      "grad_norm": 0.3818576059096302,
      "learning_rate": 2.7973680080569555e-05,
      "loss": 0.4768,
      "step": 532
    },
    {
      "epoch": 2.1589873417721517,
      "grad_norm": 0.359203069366619,
      "learning_rate": 2.792159532078314e-05,
      "loss": 0.4732,
      "step": 533
    },
    {
      "epoch": 2.163037974683544,
      "grad_norm": 0.3927046311769122,
      "learning_rate": 2.7869446761557033e-05,
      "loss": 0.47,
      "step": 534
    },
    {
      "epoch": 2.1670886075949367,
      "grad_norm": 0.36032673902912415,
      "learning_rate": 2.781723482288857e-05,
      "loss": 0.4912,
      "step": 535
    },
    {
      "epoch": 2.1711392405063292,
      "grad_norm": 0.38689889007703165,
      "learning_rate": 2.7764959925285517e-05,
      "loss": 0.4814,
      "step": 536
    },
    {
      "epoch": 2.1751898734177217,
      "grad_norm": 0.3499205347793075,
      "learning_rate": 2.771262248976272e-05,
      "loss": 0.4775,
      "step": 537
    },
    {
      "epoch": 2.179240506329114,
      "grad_norm": 0.3660668046449536,
      "learning_rate": 2.7660222937838677e-05,
      "loss": 0.4737,
      "step": 538
    },
    {
      "epoch": 2.1832911392405063,
      "grad_norm": 0.38354677422363054,
      "learning_rate": 2.7607761691532186e-05,
      "loss": 0.4679,
      "step": 539
    },
    {
      "epoch": 2.187341772151899,
      "grad_norm": 0.4363583648431936,
      "learning_rate": 2.7555239173358916e-05,
      "loss": 0.4782,
      "step": 540
    },
    {
      "epoch": 2.191392405063291,
      "grad_norm": 0.3758152190318853,
      "learning_rate": 2.7502655806328e-05,
      "loss": 0.4839,
      "step": 541
    },
    {
      "epoch": 2.1954430379746834,
      "grad_norm": 0.38797637545358965,
      "learning_rate": 2.7450012013938648e-05,
      "loss": 0.4788,
      "step": 542
    },
    {
      "epoch": 2.199493670886076,
      "grad_norm": 0.43575466484649805,
      "learning_rate": 2.739730822017673e-05,
      "loss": 0.4692,
      "step": 543
    },
    {
      "epoch": 2.2035443037974685,
      "grad_norm": 0.3396448750393881,
      "learning_rate": 2.7344544849511355e-05,
      "loss": 0.4845,
      "step": 544
    },
    {
      "epoch": 2.207594936708861,
      "grad_norm": 0.4449590039824782,
      "learning_rate": 2.7291722326891456e-05,
      "loss": 0.4657,
      "step": 545
    },
    {
      "epoch": 2.211645569620253,
      "grad_norm": 0.3763570885174587,
      "learning_rate": 2.723884107774236e-05,
      "loss": 0.4808,
      "step": 546
    },
    {
      "epoch": 2.2156962025316456,
      "grad_norm": 0.33682135330786095,
      "learning_rate": 2.718590152796239e-05,
      "loss": 0.4662,
      "step": 547
    },
    {
      "epoch": 2.219746835443038,
      "grad_norm": 0.3765215512411809,
      "learning_rate": 2.71329041039194e-05,
      "loss": 0.4789,
      "step": 548
    },
    {
      "epoch": 2.22379746835443,
      "grad_norm": 0.33605421933389956,
      "learning_rate": 2.7079849232447357e-05,
      "loss": 0.4759,
      "step": 549
    },
    {
      "epoch": 2.2278481012658227,
      "grad_norm": 0.3030403177119961,
      "learning_rate": 2.7026737340842895e-05,
      "loss": 0.4918,
      "step": 550
    },
    {
      "epoch": 2.231898734177215,
      "grad_norm": 0.4002023860778905,
      "learning_rate": 2.697356885686189e-05,
      "loss": 0.4937,
      "step": 551
    },
    {
      "epoch": 2.2359493670886077,
      "grad_norm": 0.2915139390600507,
      "learning_rate": 2.6920344208716014e-05,
      "loss": 0.4761,
      "step": 552
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.3326407417483593,
      "learning_rate": 2.6867063825069252e-05,
      "loss": 0.4749,
      "step": 553
    },
    {
      "epoch": 2.2440506329113923,
      "grad_norm": 0.33196624723074347,
      "learning_rate": 2.6813728135034494e-05,
      "loss": 0.4769,
      "step": 554
    },
    {
      "epoch": 2.248101265822785,
      "grad_norm": 0.3218304755722093,
      "learning_rate": 2.6760337568170056e-05,
      "loss": 0.4821,
      "step": 555
    },
    {
      "epoch": 2.2521518987341773,
      "grad_norm": 0.34508694911009485,
      "learning_rate": 2.6706892554476226e-05,
      "loss": 0.4657,
      "step": 556
    },
    {
      "epoch": 2.2562025316455694,
      "grad_norm": 0.33474022696005784,
      "learning_rate": 2.6653393524391795e-05,
      "loss": 0.4832,
      "step": 557
    },
    {
      "epoch": 2.260253164556962,
      "grad_norm": 0.34516062097121053,
      "learning_rate": 2.6599840908790592e-05,
      "loss": 0.4684,
      "step": 558
    },
    {
      "epoch": 2.2643037974683544,
      "grad_norm": 0.4150549466606544,
      "learning_rate": 2.6546235138978028e-05,
      "loss": 0.4689,
      "step": 559
    },
    {
      "epoch": 2.268354430379747,
      "grad_norm": 0.28113382538262743,
      "learning_rate": 2.6492576646687597e-05,
      "loss": 0.4769,
      "step": 560
    },
    {
      "epoch": 2.2724050632911394,
      "grad_norm": 0.34650250414383393,
      "learning_rate": 2.6438865864077425e-05,
      "loss": 0.4731,
      "step": 561
    },
    {
      "epoch": 2.2764556962025315,
      "grad_norm": 0.3094632400348333,
      "learning_rate": 2.6385103223726766e-05,
      "loss": 0.4769,
      "step": 562
    },
    {
      "epoch": 2.280506329113924,
      "grad_norm": 0.31966921165855666,
      "learning_rate": 2.6331289158632537e-05,
      "loss": 0.4783,
      "step": 563
    },
    {
      "epoch": 2.2845569620253166,
      "grad_norm": 0.335100320633454,
      "learning_rate": 2.6277424102205817e-05,
      "loss": 0.4688,
      "step": 564
    },
    {
      "epoch": 2.2886075949367086,
      "grad_norm": 0.3272543767643052,
      "learning_rate": 2.6223508488268374e-05,
      "loss": 0.4825,
      "step": 565
    },
    {
      "epoch": 2.292658227848101,
      "grad_norm": 0.3222351905734499,
      "learning_rate": 2.6169542751049148e-05,
      "loss": 0.4743,
      "step": 566
    },
    {
      "epoch": 2.2967088607594937,
      "grad_norm": 0.29133982571615624,
      "learning_rate": 2.6115527325180754e-05,
      "loss": 0.4734,
      "step": 567
    },
    {
      "epoch": 2.300759493670886,
      "grad_norm": 0.3526421961594564,
      "learning_rate": 2.606146264569603e-05,
      "loss": 0.4789,
      "step": 568
    },
    {
      "epoch": 2.3048101265822787,
      "grad_norm": 0.2855632943793194,
      "learning_rate": 2.6007349148024447e-05,
      "loss": 0.4742,
      "step": 569
    },
    {
      "epoch": 2.3088607594936708,
      "grad_norm": 0.2758318762766398,
      "learning_rate": 2.5953187267988694e-05,
      "loss": 0.4769,
      "step": 570
    },
    {
      "epoch": 2.3129113924050633,
      "grad_norm": 0.28536802347088636,
      "learning_rate": 2.5898977441801097e-05,
      "loss": 0.4828,
      "step": 571
    },
    {
      "epoch": 2.316962025316456,
      "grad_norm": 0.2912184815269533,
      "learning_rate": 2.584472010606015e-05,
      "loss": 0.4832,
      "step": 572
    },
    {
      "epoch": 2.321012658227848,
      "grad_norm": 0.2926513723531926,
      "learning_rate": 2.5790415697746976e-05,
      "loss": 0.4846,
      "step": 573
    },
    {
      "epoch": 2.3250632911392404,
      "grad_norm": 0.35350058469195755,
      "learning_rate": 2.5736064654221808e-05,
      "loss": 0.4851,
      "step": 574
    },
    {
      "epoch": 2.329113924050633,
      "grad_norm": 0.3230309374903499,
      "learning_rate": 2.568166741322048e-05,
      "loss": 0.484,
      "step": 575
    },
    {
      "epoch": 2.3331645569620254,
      "grad_norm": 0.3314096805628127,
      "learning_rate": 2.56272244128509e-05,
      "loss": 0.4855,
      "step": 576
    },
    {
      "epoch": 2.337215189873418,
      "grad_norm": 0.3490015273323093,
      "learning_rate": 2.55727360915895e-05,
      "loss": 0.4746,
      "step": 577
    },
    {
      "epoch": 2.34126582278481,
      "grad_norm": 0.32486101490565855,
      "learning_rate": 2.5518202888277734e-05,
      "loss": 0.4667,
      "step": 578
    },
    {
      "epoch": 2.3453164556962025,
      "grad_norm": 0.3274095087103921,
      "learning_rate": 2.5463625242118523e-05,
      "loss": 0.4765,
      "step": 579
    },
    {
      "epoch": 2.349367088607595,
      "grad_norm": 0.3422790574046964,
      "learning_rate": 2.5409003592672723e-05,
      "loss": 0.4694,
      "step": 580
    },
    {
      "epoch": 2.353417721518987,
      "grad_norm": 0.29283828797123745,
      "learning_rate": 2.535433837985559e-05,
      "loss": 0.479,
      "step": 581
    },
    {
      "epoch": 2.3574683544303796,
      "grad_norm": 0.30676104799342374,
      "learning_rate": 2.529963004393324e-05,
      "loss": 0.4839,
      "step": 582
    },
    {
      "epoch": 2.361518987341772,
      "grad_norm": 0.2862687560334886,
      "learning_rate": 2.524487902551908e-05,
      "loss": 0.4761,
      "step": 583
    },
    {
      "epoch": 2.3655696202531646,
      "grad_norm": 0.2918802675586107,
      "learning_rate": 2.519008576557029e-05,
      "loss": 0.4769,
      "step": 584
    },
    {
      "epoch": 2.369620253164557,
      "grad_norm": 0.27279853932837567,
      "learning_rate": 2.5135250705384254e-05,
      "loss": 0.4725,
      "step": 585
    },
    {
      "epoch": 2.3736708860759492,
      "grad_norm": 0.31605762881008237,
      "learning_rate": 2.5080374286595007e-05,
      "loss": 0.4829,
      "step": 586
    },
    {
      "epoch": 2.3777215189873417,
      "grad_norm": 0.33091440623652313,
      "learning_rate": 2.5025456951169677e-05,
      "loss": 0.4817,
      "step": 587
    },
    {
      "epoch": 2.3817721518987343,
      "grad_norm": 0.3408396691439725,
      "learning_rate": 2.4970499141404942e-05,
      "loss": 0.487,
      "step": 588
    },
    {
      "epoch": 2.3858227848101268,
      "grad_norm": 0.3093916353671197,
      "learning_rate": 2.491550129992345e-05,
      "loss": 0.4651,
      "step": 589
    },
    {
      "epoch": 2.389873417721519,
      "grad_norm": 0.37549655522254166,
      "learning_rate": 2.486046386967024e-05,
      "loss": 0.491,
      "step": 590
    },
    {
      "epoch": 2.3939240506329114,
      "grad_norm": 0.33252863629927754,
      "learning_rate": 2.4805387293909214e-05,
      "loss": 0.4844,
      "step": 591
    },
    {
      "epoch": 2.397974683544304,
      "grad_norm": 0.34696471159228814,
      "learning_rate": 2.4750272016219552e-05,
      "loss": 0.4825,
      "step": 592
    },
    {
      "epoch": 2.4020253164556964,
      "grad_norm": 0.27874228653082794,
      "learning_rate": 2.4695118480492114e-05,
      "loss": 0.4767,
      "step": 593
    },
    {
      "epoch": 2.4060759493670885,
      "grad_norm": 0.3526802932142688,
      "learning_rate": 2.4639927130925898e-05,
      "loss": 0.4636,
      "step": 594
    },
    {
      "epoch": 2.410126582278481,
      "grad_norm": 0.3193929033201143,
      "learning_rate": 2.458469841202444e-05,
      "loss": 0.461,
      "step": 595
    },
    {
      "epoch": 2.4141772151898735,
      "grad_norm": 0.2822612975314097,
      "learning_rate": 2.452943276859226e-05,
      "loss": 0.471,
      "step": 596
    },
    {
      "epoch": 2.418227848101266,
      "grad_norm": 0.3263694612527554,
      "learning_rate": 2.447413064573125e-05,
      "loss": 0.471,
      "step": 597
    },
    {
      "epoch": 2.422278481012658,
      "grad_norm": 0.2691493980927656,
      "learning_rate": 2.4418792488837095e-05,
      "loss": 0.4793,
      "step": 598
    },
    {
      "epoch": 2.4263291139240506,
      "grad_norm": 0.3592824488420059,
      "learning_rate": 2.4363418743595713e-05,
      "loss": 0.4814,
      "step": 599
    },
    {
      "epoch": 2.430379746835443,
      "grad_norm": 0.3272004210136757,
      "learning_rate": 2.430800985597963e-05,
      "loss": 0.4826,
      "step": 600
    },
    {
      "epoch": 2.4344303797468356,
      "grad_norm": 0.3475348254171806,
      "learning_rate": 2.4252566272244415e-05,
      "loss": 0.475,
      "step": 601
    },
    {
      "epoch": 2.4384810126582277,
      "grad_norm": 0.34155619397368925,
      "learning_rate": 2.4197088438925063e-05,
      "loss": 0.4719,
      "step": 602
    },
    {
      "epoch": 2.44253164556962,
      "grad_norm": 0.36272403804820563,
      "learning_rate": 2.4141576802832417e-05,
      "loss": 0.4758,
      "step": 603
    },
    {
      "epoch": 2.4465822784810127,
      "grad_norm": 0.35029028480560104,
      "learning_rate": 2.408603181104957e-05,
      "loss": 0.4723,
      "step": 604
    },
    {
      "epoch": 2.4506329113924052,
      "grad_norm": 0.30681816427051056,
      "learning_rate": 2.4030453910928245e-05,
      "loss": 0.4711,
      "step": 605
    },
    {
      "epoch": 2.4546835443037973,
      "grad_norm": 0.3116692192646477,
      "learning_rate": 2.397484355008521e-05,
      "loss": 0.4684,
      "step": 606
    },
    {
      "epoch": 2.45873417721519,
      "grad_norm": 0.2902988121144744,
      "learning_rate": 2.3919201176398662e-05,
      "loss": 0.4689,
      "step": 607
    },
    {
      "epoch": 2.4627848101265823,
      "grad_norm": 0.32033334456934104,
      "learning_rate": 2.3863527238004633e-05,
      "loss": 0.4744,
      "step": 608
    },
    {
      "epoch": 2.466835443037975,
      "grad_norm": 0.30502349143264085,
      "learning_rate": 2.380782218329337e-05,
      "loss": 0.4736,
      "step": 609
    },
    {
      "epoch": 2.470886075949367,
      "grad_norm": 0.30995443251241084,
      "learning_rate": 2.3752086460905725e-05,
      "loss": 0.4771,
      "step": 610
    },
    {
      "epoch": 2.4749367088607594,
      "grad_norm": 0.3248501480122993,
      "learning_rate": 2.3696320519729544e-05,
      "loss": 0.4771,
      "step": 611
    },
    {
      "epoch": 2.478987341772152,
      "grad_norm": 0.25236006669814004,
      "learning_rate": 2.3640524808896045e-05,
      "loss": 0.4791,
      "step": 612
    },
    {
      "epoch": 2.4830379746835445,
      "grad_norm": 0.3114865461571946,
      "learning_rate": 2.3584699777776222e-05,
      "loss": 0.4791,
      "step": 613
    },
    {
      "epoch": 2.4870886075949366,
      "grad_norm": 0.31236338864705204,
      "learning_rate": 2.3528845875977195e-05,
      "loss": 0.4714,
      "step": 614
    },
    {
      "epoch": 2.491139240506329,
      "grad_norm": 0.28424314151270896,
      "learning_rate": 2.3472963553338614e-05,
      "loss": 0.474,
      "step": 615
    },
    {
      "epoch": 2.4951898734177216,
      "grad_norm": 0.3336640402538505,
      "learning_rate": 2.341705325992901e-05,
      "loss": 0.4725,
      "step": 616
    },
    {
      "epoch": 2.499240506329114,
      "grad_norm": 0.2656189488110121,
      "learning_rate": 2.336111544604222e-05,
      "loss": 0.4763,
      "step": 617
    },
    {
      "epoch": 2.503291139240506,
      "grad_norm": 0.36847590302993144,
      "learning_rate": 2.33051505621937e-05,
      "loss": 0.475,
      "step": 618
    },
    {
      "epoch": 2.5073417721518987,
      "grad_norm": 0.35753646224280883,
      "learning_rate": 2.324915905911693e-05,
      "loss": 0.4733,
      "step": 619
    },
    {
      "epoch": 2.511392405063291,
      "grad_norm": 0.3494215093098695,
      "learning_rate": 2.319314138775977e-05,
      "loss": 0.4859,
      "step": 620
    },
    {
      "epoch": 2.5154430379746833,
      "grad_norm": 0.38249763770266965,
      "learning_rate": 2.3137097999280856e-05,
      "loss": 0.4807,
      "step": 621
    },
    {
      "epoch": 2.519493670886076,
      "grad_norm": 0.3186718222404792,
      "learning_rate": 2.308102934504593e-05,
      "loss": 0.4574,
      "step": 622
    },
    {
      "epoch": 2.5235443037974683,
      "grad_norm": 0.35887336926079594,
      "learning_rate": 2.3024935876624222e-05,
      "loss": 0.4753,
      "step": 623
    },
    {
      "epoch": 2.527594936708861,
      "grad_norm": 0.34259229369636574,
      "learning_rate": 2.2968818045784813e-05,
      "loss": 0.4812,
      "step": 624
    },
    {
      "epoch": 2.5316455696202533,
      "grad_norm": 0.3979949591322303,
      "learning_rate": 2.2912676304493006e-05,
      "loss": 0.4801,
      "step": 625
    },
    {
      "epoch": 2.5356962025316454,
      "grad_norm": 0.2892341488647758,
      "learning_rate": 2.2856511104906668e-05,
      "loss": 0.4802,
      "step": 626
    },
    {
      "epoch": 2.539746835443038,
      "grad_norm": 0.33317174265454685,
      "learning_rate": 2.2800322899372586e-05,
      "loss": 0.4872,
      "step": 627
    },
    {
      "epoch": 2.5437974683544304,
      "grad_norm": 0.3132697338781742,
      "learning_rate": 2.2744112140422844e-05,
      "loss": 0.4889,
      "step": 628
    },
    {
      "epoch": 2.547848101265823,
      "grad_norm": 0.36276410992858144,
      "learning_rate": 2.2687879280771177e-05,
      "loss": 0.4817,
      "step": 629
    },
    {
      "epoch": 2.5518987341772155,
      "grad_norm": 0.3515364607483676,
      "learning_rate": 2.26316247733093e-05,
      "loss": 0.4821,
      "step": 630
    },
    {
      "epoch": 2.5559493670886075,
      "grad_norm": 0.3604397519094138,
      "learning_rate": 2.257534907110328e-05,
      "loss": 0.4867,
      "step": 631
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.30478968859405936,
      "learning_rate": 2.2519052627389882e-05,
      "loss": 0.4845,
      "step": 632
    },
    {
      "epoch": 2.5640506329113926,
      "grad_norm": 0.3292016380262924,
      "learning_rate": 2.246273589557294e-05,
      "loss": 0.4732,
      "step": 633
    },
    {
      "epoch": 2.5681012658227846,
      "grad_norm": 0.3515522413548847,
      "learning_rate": 2.240639932921966e-05,
      "loss": 0.4854,
      "step": 634
    },
    {
      "epoch": 2.572151898734177,
      "grad_norm": 0.31006476976564123,
      "learning_rate": 2.2350043382056995e-05,
      "loss": 0.4853,
      "step": 635
    },
    {
      "epoch": 2.5762025316455697,
      "grad_norm": 0.3556850594205431,
      "learning_rate": 2.2293668507968015e-05,
      "loss": 0.4758,
      "step": 636
    },
    {
      "epoch": 2.580253164556962,
      "grad_norm": 0.2797881880751504,
      "learning_rate": 2.2237275160988186e-05,
      "loss": 0.473,
      "step": 637
    },
    {
      "epoch": 2.5843037974683547,
      "grad_norm": 0.3434045556527874,
      "learning_rate": 2.2180863795301787e-05,
      "loss": 0.4716,
      "step": 638
    },
    {
      "epoch": 2.5883544303797468,
      "grad_norm": 0.31737375629911485,
      "learning_rate": 2.212443486523819e-05,
      "loss": 0.4794,
      "step": 639
    },
    {
      "epoch": 2.5924050632911393,
      "grad_norm": 0.2900116209765602,
      "learning_rate": 2.2067988825268243e-05,
      "loss": 0.4823,
      "step": 640
    },
    {
      "epoch": 2.596455696202532,
      "grad_norm": 0.34729503449377835,
      "learning_rate": 2.2011526130000596e-05,
      "loss": 0.4795,
      "step": 641
    },
    {
      "epoch": 2.600506329113924,
      "grad_norm": 0.2867838883203389,
      "learning_rate": 2.1955047234178038e-05,
      "loss": 0.4767,
      "step": 642
    },
    {
      "epoch": 2.6045569620253164,
      "grad_norm": 0.41088219198172937,
      "learning_rate": 2.1898552592673825e-05,
      "loss": 0.4809,
      "step": 643
    },
    {
      "epoch": 2.608607594936709,
      "grad_norm": 0.28244752841301873,
      "learning_rate": 2.184204266048803e-05,
      "loss": 0.4719,
      "step": 644
    },
    {
      "epoch": 2.6126582278481014,
      "grad_norm": 0.38570926123915483,
      "learning_rate": 2.1785517892743887e-05,
      "loss": 0.4833,
      "step": 645
    },
    {
      "epoch": 2.616708860759494,
      "grad_norm": 0.2854265371746082,
      "learning_rate": 2.17289787446841e-05,
      "loss": 0.4841,
      "step": 646
    },
    {
      "epoch": 2.620759493670886,
      "grad_norm": 0.3071722363054009,
      "learning_rate": 2.1672425671667198e-05,
      "loss": 0.4778,
      "step": 647
    },
    {
      "epoch": 2.6248101265822785,
      "grad_norm": 0.28661409653577496,
      "learning_rate": 2.161585912916385e-05,
      "loss": 0.4757,
      "step": 648
    },
    {
      "epoch": 2.628860759493671,
      "grad_norm": 0.2627374037283905,
      "learning_rate": 2.1559279572753214e-05,
      "loss": 0.4819,
      "step": 649
    },
    {
      "epoch": 2.632911392405063,
      "grad_norm": 0.30707730798794175,
      "learning_rate": 2.1502687458119268e-05,
      "loss": 0.4826,
      "step": 650
    },
    {
      "epoch": 2.6369620253164556,
      "grad_norm": 0.25852415290575975,
      "learning_rate": 2.1446083241047116e-05,
      "loss": 0.4789,
      "step": 651
    },
    {
      "epoch": 2.641012658227848,
      "grad_norm": 0.2903138910648452,
      "learning_rate": 2.1389467377419333e-05,
      "loss": 0.4847,
      "step": 652
    },
    {
      "epoch": 2.6450632911392407,
      "grad_norm": 0.2571669475389089,
      "learning_rate": 2.133284032321232e-05,
      "loss": 0.4797,
      "step": 653
    },
    {
      "epoch": 2.649113924050633,
      "grad_norm": 0.2773078827892671,
      "learning_rate": 2.1276202534492566e-05,
      "loss": 0.4771,
      "step": 654
    },
    {
      "epoch": 2.6531645569620252,
      "grad_norm": 0.27219386243791915,
      "learning_rate": 2.121955446741306e-05,
      "loss": 0.4807,
      "step": 655
    },
    {
      "epoch": 2.6572151898734178,
      "grad_norm": 0.2968779714208418,
      "learning_rate": 2.1162896578209517e-05,
      "loss": 0.48,
      "step": 656
    },
    {
      "epoch": 2.6612658227848103,
      "grad_norm": 0.3062583497382708,
      "learning_rate": 2.1106229323196813e-05,
      "loss": 0.4847,
      "step": 657
    },
    {
      "epoch": 2.6653164556962023,
      "grad_norm": 0.2796401366125329,
      "learning_rate": 2.1049553158765214e-05,
      "loss": 0.4743,
      "step": 658
    },
    {
      "epoch": 2.669367088607595,
      "grad_norm": 0.2883572189424844,
      "learning_rate": 2.0992868541376764e-05,
      "loss": 0.477,
      "step": 659
    },
    {
      "epoch": 2.6734177215189874,
      "grad_norm": 0.27801694602980254,
      "learning_rate": 2.093617592756158e-05,
      "loss": 0.4798,
      "step": 660
    },
    {
      "epoch": 2.67746835443038,
      "grad_norm": 0.3193627113211239,
      "learning_rate": 2.0879475773914167e-05,
      "loss": 0.4761,
      "step": 661
    },
    {
      "epoch": 2.6815189873417724,
      "grad_norm": 0.26885974629122417,
      "learning_rate": 2.082276853708978e-05,
      "loss": 0.4717,
      "step": 662
    },
    {
      "epoch": 2.6855696202531645,
      "grad_norm": 0.2867671259884346,
      "learning_rate": 2.076605467380071e-05,
      "loss": 0.465,
      "step": 663
    },
    {
      "epoch": 2.689620253164557,
      "grad_norm": 0.30369190188821454,
      "learning_rate": 2.0709334640812613e-05,
      "loss": 0.4848,
      "step": 664
    },
    {
      "epoch": 2.6936708860759495,
      "grad_norm": 0.32604623804109556,
      "learning_rate": 2.0652608894940824e-05,
      "loss": 0.4783,
      "step": 665
    },
    {
      "epoch": 2.6977215189873416,
      "grad_norm": 0.32735817951139423,
      "learning_rate": 2.0595877893046722e-05,
      "loss": 0.4817,
      "step": 666
    },
    {
      "epoch": 2.701772151898734,
      "grad_norm": 0.3087098683795272,
      "learning_rate": 2.0539142092033985e-05,
      "loss": 0.4737,
      "step": 667
    },
    {
      "epoch": 2.7058227848101266,
      "grad_norm": 0.29584430334651424,
      "learning_rate": 2.048240194884496e-05,
      "loss": 0.4825,
      "step": 668
    },
    {
      "epoch": 2.709873417721519,
      "grad_norm": 0.250267747047268,
      "learning_rate": 2.042565792045695e-05,
      "loss": 0.4693,
      "step": 669
    },
    {
      "epoch": 2.7139240506329116,
      "grad_norm": 0.32890306174242084,
      "learning_rate": 2.036891046387857e-05,
      "loss": 0.4898,
      "step": 670
    },
    {
      "epoch": 2.7179746835443037,
      "grad_norm": 0.26997334534174616,
      "learning_rate": 2.0312160036146036e-05,
      "loss": 0.4745,
      "step": 671
    },
    {
      "epoch": 2.7220253164556962,
      "grad_norm": 0.2717094895407256,
      "learning_rate": 2.025540709431948e-05,
      "loss": 0.4826,
      "step": 672
    },
    {
      "epoch": 2.7260759493670887,
      "grad_norm": 0.2806731459090754,
      "learning_rate": 2.0198652095479298e-05,
      "loss": 0.4712,
      "step": 673
    },
    {
      "epoch": 2.730126582278481,
      "grad_norm": 0.2635982610716554,
      "learning_rate": 2.014189549672245e-05,
      "loss": 0.4881,
      "step": 674
    },
    {
      "epoch": 2.7341772151898733,
      "grad_norm": 0.29174167625157627,
      "learning_rate": 2.0085137755158776e-05,
      "loss": 0.4787,
      "step": 675
    },
    {
      "epoch": 2.738227848101266,
      "grad_norm": 0.28356391401337294,
      "learning_rate": 2.0028379327907327e-05,
      "loss": 0.4836,
      "step": 676
    },
    {
      "epoch": 2.7422784810126584,
      "grad_norm": 0.3051622905610548,
      "learning_rate": 1.9971620672092676e-05,
      "loss": 0.4773,
      "step": 677
    },
    {
      "epoch": 2.746329113924051,
      "grad_norm": 0.3150607098143032,
      "learning_rate": 1.991486224484123e-05,
      "loss": 0.4706,
      "step": 678
    },
    {
      "epoch": 2.750379746835443,
      "grad_norm": 0.2833550130785345,
      "learning_rate": 1.985810450327756e-05,
      "loss": 0.4784,
      "step": 679
    },
    {
      "epoch": 2.7544303797468355,
      "grad_norm": 0.29800274057004567,
      "learning_rate": 1.9801347904520706e-05,
      "loss": 0.4865,
      "step": 680
    },
    {
      "epoch": 2.758481012658228,
      "grad_norm": 0.3659637134691945,
      "learning_rate": 1.974459290568053e-05,
      "loss": 0.4743,
      "step": 681
    },
    {
      "epoch": 2.76253164556962,
      "grad_norm": 0.28047729144792405,
      "learning_rate": 1.968783996385397e-05,
      "loss": 0.4855,
      "step": 682
    },
    {
      "epoch": 2.7665822784810126,
      "grad_norm": 0.39218986424799895,
      "learning_rate": 1.963108953612143e-05,
      "loss": 0.4763,
      "step": 683
    },
    {
      "epoch": 2.770632911392405,
      "grad_norm": 0.2723243967544914,
      "learning_rate": 1.9574342079543056e-05,
      "loss": 0.4719,
      "step": 684
    },
    {
      "epoch": 2.7746835443037976,
      "grad_norm": 0.37145420446621163,
      "learning_rate": 1.9517598051155046e-05,
      "loss": 0.4812,
      "step": 685
    },
    {
      "epoch": 2.77873417721519,
      "grad_norm": 0.3061463333237582,
      "learning_rate": 1.9460857907966025e-05,
      "loss": 0.4714,
      "step": 686
    },
    {
      "epoch": 2.782784810126582,
      "grad_norm": 0.25483731288591094,
      "learning_rate": 1.9404122106953285e-05,
      "loss": 0.4687,
      "step": 687
    },
    {
      "epoch": 2.7868354430379747,
      "grad_norm": 0.27962319481160547,
      "learning_rate": 1.9347391105059176e-05,
      "loss": 0.477,
      "step": 688
    },
    {
      "epoch": 2.790886075949367,
      "grad_norm": 0.2851455439563356,
      "learning_rate": 1.92906653591874e-05,
      "loss": 0.4772,
      "step": 689
    },
    {
      "epoch": 2.7949367088607593,
      "grad_norm": 0.279462694472255,
      "learning_rate": 1.9233945326199295e-05,
      "loss": 0.487,
      "step": 690
    },
    {
      "epoch": 2.798987341772152,
      "grad_norm": 0.24248303724865086,
      "learning_rate": 1.917723146291022e-05,
      "loss": 0.4646,
      "step": 691
    },
    {
      "epoch": 2.8030379746835443,
      "grad_norm": 0.26246364236862585,
      "learning_rate": 1.912052422608584e-05,
      "loss": 0.4873,
      "step": 692
    },
    {
      "epoch": 2.807088607594937,
      "grad_norm": 0.2762944044046232,
      "learning_rate": 1.9063824072438428e-05,
      "loss": 0.4738,
      "step": 693
    },
    {
      "epoch": 2.8111392405063294,
      "grad_norm": 0.2812736100561901,
      "learning_rate": 1.9007131458623246e-05,
      "loss": 0.4777,
      "step": 694
    },
    {
      "epoch": 2.8151898734177214,
      "grad_norm": 0.26215713489178744,
      "learning_rate": 1.895044684123479e-05,
      "loss": 0.477,
      "step": 695
    },
    {
      "epoch": 2.819240506329114,
      "grad_norm": 0.27664925570323157,
      "learning_rate": 1.8893770676803194e-05,
      "loss": 0.4708,
      "step": 696
    },
    {
      "epoch": 2.8232911392405065,
      "grad_norm": 0.29264477980840436,
      "learning_rate": 1.8837103421790486e-05,
      "loss": 0.4825,
      "step": 697
    },
    {
      "epoch": 2.8273417721518985,
      "grad_norm": 0.2780685787822027,
      "learning_rate": 1.8780445532586952e-05,
      "loss": 0.4891,
      "step": 698
    },
    {
      "epoch": 2.831392405063291,
      "grad_norm": 0.28561716925670644,
      "learning_rate": 1.872379746550743e-05,
      "loss": 0.4678,
      "step": 699
    },
    {
      "epoch": 2.8354430379746836,
      "grad_norm": 0.28336347582050636,
      "learning_rate": 1.866715967678769e-05,
      "loss": 0.4893,
      "step": 700
    },
    {
      "epoch": 2.839493670886076,
      "grad_norm": 0.2783766954935952,
      "learning_rate": 1.861053262258067e-05,
      "loss": 0.4907,
      "step": 701
    },
    {
      "epoch": 2.8435443037974686,
      "grad_norm": 0.2849008998529939,
      "learning_rate": 1.8553916758952897e-05,
      "loss": 0.4627,
      "step": 702
    },
    {
      "epoch": 2.8475949367088607,
      "grad_norm": 0.2915410596288912,
      "learning_rate": 1.8497312541880735e-05,
      "loss": 0.4801,
      "step": 703
    },
    {
      "epoch": 2.851645569620253,
      "grad_norm": 0.28530563148503196,
      "learning_rate": 1.8440720427246786e-05,
      "loss": 0.4821,
      "step": 704
    },
    {
      "epoch": 2.8556962025316457,
      "grad_norm": 0.2818309014268536,
      "learning_rate": 1.8384140870836157e-05,
      "loss": 0.4825,
      "step": 705
    },
    {
      "epoch": 2.8597468354430378,
      "grad_norm": 0.29390373688934235,
      "learning_rate": 1.8327574328332806e-05,
      "loss": 0.4717,
      "step": 706
    },
    {
      "epoch": 2.8637974683544303,
      "grad_norm": 0.24495911510773882,
      "learning_rate": 1.8271021255315906e-05,
      "loss": 0.4646,
      "step": 707
    },
    {
      "epoch": 2.867848101265823,
      "grad_norm": 0.3026569601410154,
      "learning_rate": 1.8214482107256117e-05,
      "loss": 0.4848,
      "step": 708
    },
    {
      "epoch": 2.8718987341772153,
      "grad_norm": 0.2823724263776704,
      "learning_rate": 1.8157957339511968e-05,
      "loss": 0.4772,
      "step": 709
    },
    {
      "epoch": 2.875949367088608,
      "grad_norm": 0.31224754823779277,
      "learning_rate": 1.8101447407326182e-05,
      "loss": 0.4866,
      "step": 710
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.3105167232519128,
      "learning_rate": 1.8044952765821966e-05,
      "loss": 0.4913,
      "step": 711
    },
    {
      "epoch": 2.8840506329113924,
      "grad_norm": 0.29260198358429834,
      "learning_rate": 1.7988473869999407e-05,
      "loss": 0.4763,
      "step": 712
    },
    {
      "epoch": 2.888101265822785,
      "grad_norm": 0.2864280785116009,
      "learning_rate": 1.7932011174731764e-05,
      "loss": 0.474,
      "step": 713
    },
    {
      "epoch": 2.892151898734177,
      "grad_norm": 0.27861503618550526,
      "learning_rate": 1.7875565134761817e-05,
      "loss": 0.4811,
      "step": 714
    },
    {
      "epoch": 2.8962025316455695,
      "grad_norm": 0.30879463826854525,
      "learning_rate": 1.7819136204698226e-05,
      "loss": 0.4635,
      "step": 715
    },
    {
      "epoch": 2.900253164556962,
      "grad_norm": 0.27861226754598467,
      "learning_rate": 1.776272483901182e-05,
      "loss": 0.4763,
      "step": 716
    },
    {
      "epoch": 2.9043037974683545,
      "grad_norm": 0.3160370177740227,
      "learning_rate": 1.7706331492031995e-05,
      "loss": 0.4834,
      "step": 717
    },
    {
      "epoch": 2.908354430379747,
      "grad_norm": 0.28104722121537434,
      "learning_rate": 1.764995661794301e-05,
      "loss": 0.4857,
      "step": 718
    },
    {
      "epoch": 2.912405063291139,
      "grad_norm": 0.29967480423503984,
      "learning_rate": 1.759360067078035e-05,
      "loss": 0.4874,
      "step": 719
    },
    {
      "epoch": 2.9164556962025316,
      "grad_norm": 0.2763044671677636,
      "learning_rate": 1.7537264104427064e-05,
      "loss": 0.4656,
      "step": 720
    },
    {
      "epoch": 2.920506329113924,
      "grad_norm": 0.29848930441575344,
      "learning_rate": 1.748094737261012e-05,
      "loss": 0.4754,
      "step": 721
    },
    {
      "epoch": 2.9245569620253162,
      "grad_norm": 0.2813936957891289,
      "learning_rate": 1.7424650928896726e-05,
      "loss": 0.4702,
      "step": 722
    },
    {
      "epoch": 2.9286075949367087,
      "grad_norm": 0.28861730672302316,
      "learning_rate": 1.7368375226690712e-05,
      "loss": 0.4707,
      "step": 723
    },
    {
      "epoch": 2.9326582278481013,
      "grad_norm": 0.24693368164976032,
      "learning_rate": 1.731212071922883e-05,
      "loss": 0.4738,
      "step": 724
    },
    {
      "epoch": 2.9367088607594938,
      "grad_norm": 0.2710623013366391,
      "learning_rate": 1.7255887859577156e-05,
      "loss": 0.4744,
      "step": 725
    },
    {
      "epoch": 2.9407594936708863,
      "grad_norm": 0.27899506630638915,
      "learning_rate": 1.7199677100627427e-05,
      "loss": 0.4805,
      "step": 726
    },
    {
      "epoch": 2.9448101265822784,
      "grad_norm": 0.24794190806912542,
      "learning_rate": 1.7143488895093343e-05,
      "loss": 0.4827,
      "step": 727
    },
    {
      "epoch": 2.948860759493671,
      "grad_norm": 0.2738379651196646,
      "learning_rate": 1.7087323695506994e-05,
      "loss": 0.4758,
      "step": 728
    },
    {
      "epoch": 2.9529113924050634,
      "grad_norm": 0.2574243838368711,
      "learning_rate": 1.7031181954215194e-05,
      "loss": 0.4817,
      "step": 729
    },
    {
      "epoch": 2.9569620253164555,
      "grad_norm": 0.32093151182428625,
      "learning_rate": 1.6975064123375788e-05,
      "loss": 0.4818,
      "step": 730
    },
    {
      "epoch": 2.961012658227848,
      "grad_norm": 0.27264824183723835,
      "learning_rate": 1.6918970654954084e-05,
      "loss": 0.4689,
      "step": 731
    },
    {
      "epoch": 2.9650632911392405,
      "grad_norm": 0.3826800453853356,
      "learning_rate": 1.686290200071915e-05,
      "loss": 0.4682,
      "step": 732
    },
    {
      "epoch": 2.969113924050633,
      "grad_norm": 0.25243349645292534,
      "learning_rate": 1.6806858612240234e-05,
      "loss": 0.472,
      "step": 733
    },
    {
      "epoch": 2.9731645569620255,
      "grad_norm": 0.3221055195117261,
      "learning_rate": 1.6750840940883078e-05,
      "loss": 0.4811,
      "step": 734
    },
    {
      "epoch": 2.9772151898734176,
      "grad_norm": 0.3115920830162225,
      "learning_rate": 1.6694849437806305e-05,
      "loss": 0.4789,
      "step": 735
    },
    {
      "epoch": 2.98126582278481,
      "grad_norm": 0.30484074672963696,
      "learning_rate": 1.663888455395778e-05,
      "loss": 0.4764,
      "step": 736
    },
    {
      "epoch": 2.9853164556962026,
      "grad_norm": 0.2919317308934776,
      "learning_rate": 1.6582946740070995e-05,
      "loss": 0.4866,
      "step": 737
    },
    {
      "epoch": 2.9893670886075947,
      "grad_norm": 0.27078260294352235,
      "learning_rate": 1.6527036446661396e-05,
      "loss": 0.4717,
      "step": 738
    },
    {
      "epoch": 2.993417721518987,
      "grad_norm": 0.2439449818749629,
      "learning_rate": 1.6471154124022818e-05,
      "loss": 0.4842,
      "step": 739
    },
    {
      "epoch": 2.9974683544303797,
      "grad_norm": 0.2893038867282254,
      "learning_rate": 1.6415300222223788e-05,
      "loss": 0.4833,
      "step": 740
    },
    {
      "epoch": 3.0015189873417722,
      "grad_norm": 0.292183359435827,
      "learning_rate": 1.6359475191103958e-05,
      "loss": 0.4558,
      "step": 741
    },
    {
      "epoch": 3.0055696202531648,
      "grad_norm": 0.4287388818809184,
      "learning_rate": 1.6303679480270466e-05,
      "loss": 0.4017,
      "step": 742
    },
    {
      "epoch": 3.009620253164557,
      "grad_norm": 0.6405146802296977,
      "learning_rate": 1.624791353909428e-05,
      "loss": 0.4152,
      "step": 743
    },
    {
      "epoch": 3.0136708860759494,
      "grad_norm": 0.39000129626059893,
      "learning_rate": 1.619217781670663e-05,
      "loss": 0.4079,
      "step": 744
    },
    {
      "epoch": 3.017721518987342,
      "grad_norm": 0.4819800653144911,
      "learning_rate": 1.6136472761995373e-05,
      "loss": 0.3991,
      "step": 745
    },
    {
      "epoch": 3.0217721518987344,
      "grad_norm": 0.3557365651832404,
      "learning_rate": 1.608079882360134e-05,
      "loss": 0.4057,
      "step": 746
    },
    {
      "epoch": 3.0258227848101265,
      "grad_norm": 0.3928838264136881,
      "learning_rate": 1.60251564499148e-05,
      "loss": 0.4098,
      "step": 747
    },
    {
      "epoch": 3.029873417721519,
      "grad_norm": 0.40434179498614453,
      "learning_rate": 1.596954608907176e-05,
      "loss": 0.4084,
      "step": 748
    },
    {
      "epoch": 3.0339240506329115,
      "grad_norm": 0.30780651453264884,
      "learning_rate": 1.591396818895043e-05,
      "loss": 0.4185,
      "step": 749
    },
    {
      "epoch": 3.037974683544304,
      "grad_norm": 0.3931479240202925,
      "learning_rate": 1.585842319716759e-05,
      "loss": 0.4159,
      "step": 750
    },
    {
      "epoch": 3.042025316455696,
      "grad_norm": 0.33303071423313646,
      "learning_rate": 1.5802911561074944e-05,
      "loss": 0.4226,
      "step": 751
    },
    {
      "epoch": 3.0460759493670886,
      "grad_norm": 0.33189024750192686,
      "learning_rate": 1.5747433727755595e-05,
      "loss": 0.4166,
      "step": 752
    },
    {
      "epoch": 3.050126582278481,
      "grad_norm": 0.35020159759839736,
      "learning_rate": 1.5691990144020376e-05,
      "loss": 0.4066,
      "step": 753
    },
    {
      "epoch": 3.0541772151898736,
      "grad_norm": 0.29372359714540297,
      "learning_rate": 1.5636581256404297e-05,
      "loss": 0.4221,
      "step": 754
    },
    {
      "epoch": 3.0582278481012657,
      "grad_norm": 0.357782655397944,
      "learning_rate": 1.558120751116291e-05,
      "loss": 0.4064,
      "step": 755
    },
    {
      "epoch": 3.062278481012658,
      "grad_norm": 0.3181027274281555,
      "learning_rate": 1.552586935426876e-05,
      "loss": 0.4181,
      "step": 756
    },
    {
      "epoch": 3.0663291139240507,
      "grad_norm": 0.325968737360748,
      "learning_rate": 1.547056723140774e-05,
      "loss": 0.418,
      "step": 757
    },
    {
      "epoch": 3.0703797468354432,
      "grad_norm": 0.32482107280557443,
      "learning_rate": 1.5415301587975565e-05,
      "loss": 0.4176,
      "step": 758
    },
    {
      "epoch": 3.0744303797468353,
      "grad_norm": 0.3087307699702407,
      "learning_rate": 1.536007286907411e-05,
      "loss": 0.4076,
      "step": 759
    },
    {
      "epoch": 3.078481012658228,
      "grad_norm": 0.27648293336701857,
      "learning_rate": 1.5304881519507896e-05,
      "loss": 0.4015,
      "step": 760
    },
    {
      "epoch": 3.0825316455696203,
      "grad_norm": 0.28823527616677114,
      "learning_rate": 1.5249727983780453e-05,
      "loss": 0.4032,
      "step": 761
    },
    {
      "epoch": 3.086582278481013,
      "grad_norm": 0.2753793130881606,
      "learning_rate": 1.5194612706090786e-05,
      "loss": 0.4065,
      "step": 762
    },
    {
      "epoch": 3.090632911392405,
      "grad_norm": 0.2842583349395612,
      "learning_rate": 1.5139536130329771e-05,
      "loss": 0.4183,
      "step": 763
    },
    {
      "epoch": 3.0946835443037974,
      "grad_norm": 0.3139698593956553,
      "learning_rate": 1.508449870007656e-05,
      "loss": 0.4086,
      "step": 764
    },
    {
      "epoch": 3.09873417721519,
      "grad_norm": 0.26890735341676075,
      "learning_rate": 1.5029500858595056e-05,
      "loss": 0.4209,
      "step": 765
    },
    {
      "epoch": 3.1027848101265825,
      "grad_norm": 0.34493076298877834,
      "learning_rate": 1.4974543048830328e-05,
      "loss": 0.4278,
      "step": 766
    },
    {
      "epoch": 3.1068354430379745,
      "grad_norm": 0.25067460468583025,
      "learning_rate": 1.4919625713405e-05,
      "loss": 0.4316,
      "step": 767
    },
    {
      "epoch": 3.110886075949367,
      "grad_norm": 0.31800266759966306,
      "learning_rate": 1.4864749294615756e-05,
      "loss": 0.4249,
      "step": 768
    },
    {
      "epoch": 3.1149367088607596,
      "grad_norm": 0.2796030532912686,
      "learning_rate": 1.4809914234429716e-05,
      "loss": 0.4316,
      "step": 769
    },
    {
      "epoch": 3.118987341772152,
      "grad_norm": 0.28097556614399005,
      "learning_rate": 1.4755120974480923e-05,
      "loss": 0.4238,
      "step": 770
    },
    {
      "epoch": 3.123037974683544,
      "grad_norm": 0.276711665638413,
      "learning_rate": 1.4700369956066771e-05,
      "loss": 0.4091,
      "step": 771
    },
    {
      "epoch": 3.1270886075949367,
      "grad_norm": 0.2891319331223242,
      "learning_rate": 1.4645661620144413e-05,
      "loss": 0.4004,
      "step": 772
    },
    {
      "epoch": 3.131139240506329,
      "grad_norm": 0.27271292664078356,
      "learning_rate": 1.4590996407327284e-05,
      "loss": 0.4085,
      "step": 773
    },
    {
      "epoch": 3.1351898734177217,
      "grad_norm": 0.2708752039695535,
      "learning_rate": 1.4536374757881487e-05,
      "loss": 0.4163,
      "step": 774
    },
    {
      "epoch": 3.1392405063291138,
      "grad_norm": 0.2625803779346398,
      "learning_rate": 1.4481797111722271e-05,
      "loss": 0.4046,
      "step": 775
    },
    {
      "epoch": 3.1432911392405063,
      "grad_norm": 0.2770914398005877,
      "learning_rate": 1.4427263908410507e-05,
      "loss": 0.4153,
      "step": 776
    },
    {
      "epoch": 3.147341772151899,
      "grad_norm": 0.24729893699837818,
      "learning_rate": 1.4372775587149108e-05,
      "loss": 0.4156,
      "step": 777
    },
    {
      "epoch": 3.1513924050632913,
      "grad_norm": 0.26274937712690305,
      "learning_rate": 1.4318332586779522e-05,
      "loss": 0.4226,
      "step": 778
    },
    {
      "epoch": 3.1554430379746834,
      "grad_norm": 0.23027884807607785,
      "learning_rate": 1.4263935345778202e-05,
      "loss": 0.406,
      "step": 779
    },
    {
      "epoch": 3.159493670886076,
      "grad_norm": 0.2566954956603609,
      "learning_rate": 1.420958430225303e-05,
      "loss": 0.405,
      "step": 780
    },
    {
      "epoch": 3.1635443037974684,
      "grad_norm": 0.23870188842733345,
      "learning_rate": 1.415527989393985e-05,
      "loss": 0.4268,
      "step": 781
    },
    {
      "epoch": 3.167594936708861,
      "grad_norm": 0.2542217701973114,
      "learning_rate": 1.410102255819891e-05,
      "loss": 0.4101,
      "step": 782
    },
    {
      "epoch": 3.171645569620253,
      "grad_norm": 0.2297688802305136,
      "learning_rate": 1.404681273201131e-05,
      "loss": 0.4169,
      "step": 783
    },
    {
      "epoch": 3.1756962025316455,
      "grad_norm": 0.23208207745358939,
      "learning_rate": 1.399265085197556e-05,
      "loss": 0.4178,
      "step": 784
    },
    {
      "epoch": 3.179746835443038,
      "grad_norm": 0.24198725359335324,
      "learning_rate": 1.393853735430398e-05,
      "loss": 0.4107,
      "step": 785
    },
    {
      "epoch": 3.1837974683544306,
      "grad_norm": 0.23566692449585475,
      "learning_rate": 1.3884472674819246e-05,
      "loss": 0.4122,
      "step": 786
    },
    {
      "epoch": 3.1878481012658226,
      "grad_norm": 0.23009057752998352,
      "learning_rate": 1.3830457248950864e-05,
      "loss": 0.4175,
      "step": 787
    },
    {
      "epoch": 3.191898734177215,
      "grad_norm": 0.27165907041772225,
      "learning_rate": 1.377649151173163e-05,
      "loss": 0.4175,
      "step": 788
    },
    {
      "epoch": 3.1959493670886077,
      "grad_norm": 0.23219126326594666,
      "learning_rate": 1.3722575897794181e-05,
      "loss": 0.4167,
      "step": 789
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.308267007995278,
      "learning_rate": 1.3668710841367472e-05,
      "loss": 0.4146,
      "step": 790
    },
    {
      "epoch": 3.2040506329113922,
      "grad_norm": 0.24362350765015298,
      "learning_rate": 1.361489677627324e-05,
      "loss": 0.4268,
      "step": 791
    },
    {
      "epoch": 3.2081012658227848,
      "grad_norm": 0.25649431305766557,
      "learning_rate": 1.3561134135922585e-05,
      "loss": 0.4077,
      "step": 792
    },
    {
      "epoch": 3.2121518987341773,
      "grad_norm": 0.23974743222644912,
      "learning_rate": 1.350742335331241e-05,
      "loss": 0.4233,
      "step": 793
    },
    {
      "epoch": 3.21620253164557,
      "grad_norm": 0.24322276663098819,
      "learning_rate": 1.345376486102198e-05,
      "loss": 0.4246,
      "step": 794
    },
    {
      "epoch": 3.220253164556962,
      "grad_norm": 0.2524623307729589,
      "learning_rate": 1.3400159091209414e-05,
      "loss": 0.4248,
      "step": 795
    },
    {
      "epoch": 3.2243037974683544,
      "grad_norm": 0.2544289605549649,
      "learning_rate": 1.3346606475608216e-05,
      "loss": 0.4173,
      "step": 796
    },
    {
      "epoch": 3.228354430379747,
      "grad_norm": 0.2517436273066714,
      "learning_rate": 1.3293107445523781e-05,
      "loss": 0.4302,
      "step": 797
    },
    {
      "epoch": 3.2324050632911394,
      "grad_norm": 0.24437546900755755,
      "learning_rate": 1.3239662431829949e-05,
      "loss": 0.4149,
      "step": 798
    },
    {
      "epoch": 3.2364556962025315,
      "grad_norm": 0.24606508495132434,
      "learning_rate": 1.3186271864965509e-05,
      "loss": 0.4172,
      "step": 799
    },
    {
      "epoch": 3.240506329113924,
      "grad_norm": 0.26034544477073057,
      "learning_rate": 1.3132936174930756e-05,
      "loss": 0.4196,
      "step": 800
    },
    {
      "epoch": 3.2445569620253165,
      "grad_norm": 0.24988231205698694,
      "learning_rate": 1.3079655791283995e-05,
      "loss": 0.4086,
      "step": 801
    },
    {
      "epoch": 3.248607594936709,
      "grad_norm": 0.26776621437793613,
      "learning_rate": 1.3026431143138108e-05,
      "loss": 0.4193,
      "step": 802
    },
    {
      "epoch": 3.252658227848101,
      "grad_norm": 0.2375731637015717,
      "learning_rate": 1.2973262659157114e-05,
      "loss": 0.4146,
      "step": 803
    },
    {
      "epoch": 3.2567088607594936,
      "grad_norm": 0.244978924810613,
      "learning_rate": 1.2920150767552651e-05,
      "loss": 0.4199,
      "step": 804
    },
    {
      "epoch": 3.260759493670886,
      "grad_norm": 0.25074854815687186,
      "learning_rate": 1.2867095896080607e-05,
      "loss": 0.4117,
      "step": 805
    },
    {
      "epoch": 3.2648101265822786,
      "grad_norm": 0.2681728464688326,
      "learning_rate": 1.2814098472037612e-05,
      "loss": 0.4202,
      "step": 806
    },
    {
      "epoch": 3.2688607594936707,
      "grad_norm": 0.2382962649747543,
      "learning_rate": 1.276115892225764e-05,
      "loss": 0.4216,
      "step": 807
    },
    {
      "epoch": 3.2729113924050632,
      "grad_norm": 0.27795968521183745,
      "learning_rate": 1.2708277673108555e-05,
      "loss": 0.4275,
      "step": 808
    },
    {
      "epoch": 3.2769620253164558,
      "grad_norm": 0.2357551035755549,
      "learning_rate": 1.2655455150488649e-05,
      "loss": 0.4091,
      "step": 809
    },
    {
      "epoch": 3.2810126582278483,
      "grad_norm": 0.2589408902208313,
      "learning_rate": 1.2602691779823272e-05,
      "loss": 0.4243,
      "step": 810
    },
    {
      "epoch": 3.2850632911392403,
      "grad_norm": 0.2290011345122997,
      "learning_rate": 1.2549987986061355e-05,
      "loss": 0.4176,
      "step": 811
    },
    {
      "epoch": 3.289113924050633,
      "grad_norm": 0.2607625839963637,
      "learning_rate": 1.2497344193672005e-05,
      "loss": 0.403,
      "step": 812
    },
    {
      "epoch": 3.2931645569620254,
      "grad_norm": 0.24882122318002547,
      "learning_rate": 1.2444760826641092e-05,
      "loss": 0.414,
      "step": 813
    },
    {
      "epoch": 3.297215189873418,
      "grad_norm": 0.2651345018666523,
      "learning_rate": 1.2392238308467817e-05,
      "loss": 0.4173,
      "step": 814
    },
    {
      "epoch": 3.30126582278481,
      "grad_norm": 0.23418704458099618,
      "learning_rate": 1.2339777062161326e-05,
      "loss": 0.4246,
      "step": 815
    },
    {
      "epoch": 3.3053164556962025,
      "grad_norm": 0.2620879973047422,
      "learning_rate": 1.2287377510237293e-05,
      "loss": 0.4204,
      "step": 816
    },
    {
      "epoch": 3.309367088607595,
      "grad_norm": 0.2256133547326787,
      "learning_rate": 1.2235040074714488e-05,
      "loss": 0.4191,
      "step": 817
    },
    {
      "epoch": 3.3134177215189875,
      "grad_norm": 0.2488387438125186,
      "learning_rate": 1.2182765177111434e-05,
      "loss": 0.4208,
      "step": 818
    },
    {
      "epoch": 3.3174683544303796,
      "grad_norm": 0.22881522051173686,
      "learning_rate": 1.213055323844297e-05,
      "loss": 0.4089,
      "step": 819
    },
    {
      "epoch": 3.321518987341772,
      "grad_norm": 0.2492282772486322,
      "learning_rate": 1.2078404679216864e-05,
      "loss": 0.4232,
      "step": 820
    },
    {
      "epoch": 3.3255696202531646,
      "grad_norm": 0.26899723993922114,
      "learning_rate": 1.2026319919430458e-05,
      "loss": 0.4112,
      "step": 821
    },
    {
      "epoch": 3.329620253164557,
      "grad_norm": 0.2553620436787898,
      "learning_rate": 1.1974299378567227e-05,
      "loss": 0.416,
      "step": 822
    },
    {
      "epoch": 3.333670886075949,
      "grad_norm": 0.24862664853191566,
      "learning_rate": 1.1922343475593462e-05,
      "loss": 0.4149,
      "step": 823
    },
    {
      "epoch": 3.3377215189873417,
      "grad_norm": 0.2567726127178886,
      "learning_rate": 1.187045262895488e-05,
      "loss": 0.4174,
      "step": 824
    },
    {
      "epoch": 3.3417721518987342,
      "grad_norm": 0.30776014500523113,
      "learning_rate": 1.1818627256573203e-05,
      "loss": 0.4024,
      "step": 825
    },
    {
      "epoch": 3.3458227848101267,
      "grad_norm": 0.24246952412399594,
      "learning_rate": 1.1766867775842864e-05,
      "loss": 0.4135,
      "step": 826
    },
    {
      "epoch": 3.349873417721519,
      "grad_norm": 0.28921455644413563,
      "learning_rate": 1.1715174603627615e-05,
      "loss": 0.4168,
      "step": 827
    },
    {
      "epoch": 3.3539240506329113,
      "grad_norm": 0.27341121915144334,
      "learning_rate": 1.1663548156257147e-05,
      "loss": 0.4119,
      "step": 828
    },
    {
      "epoch": 3.357974683544304,
      "grad_norm": 0.24157935925064114,
      "learning_rate": 1.161198884952377e-05,
      "loss": 0.412,
      "step": 829
    },
    {
      "epoch": 3.3620253164556964,
      "grad_norm": 0.2682560402716377,
      "learning_rate": 1.1560497098679056e-05,
      "loss": 0.4239,
      "step": 830
    },
    {
      "epoch": 3.3660759493670884,
      "grad_norm": 0.24965206450469038,
      "learning_rate": 1.1509073318430479e-05,
      "loss": 0.4132,
      "step": 831
    },
    {
      "epoch": 3.370126582278481,
      "grad_norm": 0.25351031909033794,
      "learning_rate": 1.1457717922938116e-05,
      "loss": 0.4295,
      "step": 832
    },
    {
      "epoch": 3.3741772151898735,
      "grad_norm": 0.26113421007043197,
      "learning_rate": 1.1406431325811233e-05,
      "loss": 0.4174,
      "step": 833
    },
    {
      "epoch": 3.378227848101266,
      "grad_norm": 0.27981607823239624,
      "learning_rate": 1.135521394010506e-05,
      "loss": 0.4167,
      "step": 834
    },
    {
      "epoch": 3.382278481012658,
      "grad_norm": 0.23846038254912133,
      "learning_rate": 1.1304066178317367e-05,
      "loss": 0.4214,
      "step": 835
    },
    {
      "epoch": 3.3863291139240506,
      "grad_norm": 0.27233418305862495,
      "learning_rate": 1.1252988452385199e-05,
      "loss": 0.4083,
      "step": 836
    },
    {
      "epoch": 3.390379746835443,
      "grad_norm": 0.2588421707282755,
      "learning_rate": 1.1201981173681536e-05,
      "loss": 0.4204,
      "step": 837
    },
    {
      "epoch": 3.3944303797468356,
      "grad_norm": 0.2570394300812571,
      "learning_rate": 1.1151044753011991e-05,
      "loss": 0.4201,
      "step": 838
    },
    {
      "epoch": 3.3984810126582277,
      "grad_norm": 0.24761949566787725,
      "learning_rate": 1.1100179600611491e-05,
      "loss": 0.4185,
      "step": 839
    },
    {
      "epoch": 3.40253164556962,
      "grad_norm": 0.2486705406466546,
      "learning_rate": 1.1049386126140985e-05,
      "loss": 0.4127,
      "step": 840
    },
    {
      "epoch": 3.4065822784810127,
      "grad_norm": 0.2644145764037628,
      "learning_rate": 1.0998664738684128e-05,
      "loss": 0.4075,
      "step": 841
    },
    {
      "epoch": 3.410632911392405,
      "grad_norm": 0.2408390578931503,
      "learning_rate": 1.0948015846744e-05,
      "loss": 0.4164,
      "step": 842
    },
    {
      "epoch": 3.4146835443037973,
      "grad_norm": 0.26164206829670067,
      "learning_rate": 1.0897439858239832e-05,
      "loss": 0.4188,
      "step": 843
    },
    {
      "epoch": 3.41873417721519,
      "grad_norm": 0.24724214061227429,
      "learning_rate": 1.0846937180503652e-05,
      "loss": 0.4115,
      "step": 844
    },
    {
      "epoch": 3.4227848101265823,
      "grad_norm": 0.25431948698305656,
      "learning_rate": 1.0796508220277117e-05,
      "loss": 0.4093,
      "step": 845
    },
    {
      "epoch": 3.426835443037975,
      "grad_norm": 0.24655570889215117,
      "learning_rate": 1.0746153383708107e-05,
      "loss": 0.4084,
      "step": 846
    },
    {
      "epoch": 3.430886075949367,
      "grad_norm": 0.22637941004332657,
      "learning_rate": 1.0695873076347579e-05,
      "loss": 0.4156,
      "step": 847
    },
    {
      "epoch": 3.4349367088607594,
      "grad_norm": 0.25855325916706956,
      "learning_rate": 1.0645667703146205e-05,
      "loss": 0.411,
      "step": 848
    },
    {
      "epoch": 3.438987341772152,
      "grad_norm": 0.23680441314171682,
      "learning_rate": 1.0595537668451161e-05,
      "loss": 0.4144,
      "step": 849
    },
    {
      "epoch": 3.4430379746835444,
      "grad_norm": 0.288873723647931,
      "learning_rate": 1.0545483376002854e-05,
      "loss": 0.4198,
      "step": 850
    },
    {
      "epoch": 3.4470886075949365,
      "grad_norm": 0.23915771006144626,
      "learning_rate": 1.0495505228931676e-05,
      "loss": 0.4098,
      "step": 851
    },
    {
      "epoch": 3.451139240506329,
      "grad_norm": 0.2737761284337262,
      "learning_rate": 1.044560362975474e-05,
      "loss": 0.4105,
      "step": 852
    },
    {
      "epoch": 3.4551898734177215,
      "grad_norm": 0.26020823797270015,
      "learning_rate": 1.0395778980372695e-05,
      "loss": 0.4182,
      "step": 853
    },
    {
      "epoch": 3.459240506329114,
      "grad_norm": 0.22579604999796674,
      "learning_rate": 1.0346031682066381e-05,
      "loss": 0.4078,
      "step": 854
    },
    {
      "epoch": 3.463291139240506,
      "grad_norm": 0.23397779978133962,
      "learning_rate": 1.0296362135493724e-05,
      "loss": 0.4219,
      "step": 855
    },
    {
      "epoch": 3.4673417721518986,
      "grad_norm": 0.22284483884897047,
      "learning_rate": 1.0246770740686422e-05,
      "loss": 0.4122,
      "step": 856
    },
    {
      "epoch": 3.471392405063291,
      "grad_norm": 0.23308368128194587,
      "learning_rate": 1.0197257897046743e-05,
      "loss": 0.4144,
      "step": 857
    },
    {
      "epoch": 3.4754430379746837,
      "grad_norm": 0.22348651516165416,
      "learning_rate": 1.014782400334433e-05,
      "loss": 0.4064,
      "step": 858
    },
    {
      "epoch": 3.479493670886076,
      "grad_norm": 0.22448152832942533,
      "learning_rate": 1.009846945771296e-05,
      "loss": 0.4107,
      "step": 859
    },
    {
      "epoch": 3.4835443037974683,
      "grad_norm": 0.23592451253463717,
      "learning_rate": 1.0049194657647363e-05,
      "loss": 0.4198,
      "step": 860
    },
    {
      "epoch": 3.487594936708861,
      "grad_norm": 0.2837071401232295,
      "learning_rate": 1.0000000000000006e-05,
      "loss": 0.4213,
      "step": 861
    },
    {
      "epoch": 3.4916455696202533,
      "grad_norm": 0.22658549406727618,
      "learning_rate": 9.950885880977891e-06,
      "loss": 0.4186,
      "step": 862
    },
    {
      "epoch": 3.4956962025316454,
      "grad_norm": 0.22800204156713488,
      "learning_rate": 9.901852696139382e-06,
      "loss": 0.431,
      "step": 863
    },
    {
      "epoch": 3.499746835443038,
      "grad_norm": 0.24901185418217667,
      "learning_rate": 9.852900840391027e-06,
      "loss": 0.4262,
      "step": 864
    },
    {
      "epoch": 3.5037974683544304,
      "grad_norm": 0.23894269704674706,
      "learning_rate": 9.804030707984313e-06,
      "loss": 0.4086,
      "step": 865
    },
    {
      "epoch": 3.507848101265823,
      "grad_norm": 0.23708588509530604,
      "learning_rate": 9.755242692512599e-06,
      "loss": 0.4116,
      "step": 866
    },
    {
      "epoch": 3.5118987341772154,
      "grad_norm": 0.24528497674057603,
      "learning_rate": 9.70653718690782e-06,
      "loss": 0.4087,
      "step": 867
    },
    {
      "epoch": 3.5159493670886075,
      "grad_norm": 0.22894842494049852,
      "learning_rate": 9.657914583437454e-06,
      "loss": 0.4115,
      "step": 868
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.25380579728332203,
      "learning_rate": 9.609375273701246e-06,
      "loss": 0.4268,
      "step": 869
    },
    {
      "epoch": 3.5240506329113925,
      "grad_norm": 0.23316294992730854,
      "learning_rate": 9.560919648628133e-06,
      "loss": 0.4128,
      "step": 870
    },
    {
      "epoch": 3.5281012658227846,
      "grad_norm": 0.23580209724821408,
      "learning_rate": 9.512548098473047e-06,
      "loss": 0.4155,
      "step": 871
    },
    {
      "epoch": 3.532151898734177,
      "grad_norm": 0.2540105571637628,
      "learning_rate": 9.464261012813825e-06,
      "loss": 0.4158,
      "step": 872
    },
    {
      "epoch": 3.5362025316455696,
      "grad_norm": 0.22524760223632734,
      "learning_rate": 9.416058780547987e-06,
      "loss": 0.4146,
      "step": 873
    },
    {
      "epoch": 3.540253164556962,
      "grad_norm": 0.2277901771775277,
      "learning_rate": 9.367941789889714e-06,
      "loss": 0.415,
      "step": 874
    },
    {
      "epoch": 3.5443037974683547,
      "grad_norm": 0.22577518782390343,
      "learning_rate": 9.319910428366607e-06,
      "loss": 0.4217,
      "step": 875
    },
    {
      "epoch": 3.5483544303797467,
      "grad_norm": 0.2175416082327203,
      "learning_rate": 9.271965082816667e-06,
      "loss": 0.4273,
      "step": 876
    },
    {
      "epoch": 3.5524050632911393,
      "grad_norm": 0.23317608598560186,
      "learning_rate": 9.224106139385111e-06,
      "loss": 0.4221,
      "step": 877
    },
    {
      "epoch": 3.5564556962025318,
      "grad_norm": 0.24601282797386567,
      "learning_rate": 9.176333983521291e-06,
      "loss": 0.4206,
      "step": 878
    },
    {
      "epoch": 3.560506329113924,
      "grad_norm": 0.24308692475906507,
      "learning_rate": 9.12864899997558e-06,
      "loss": 0.4222,
      "step": 879
    },
    {
      "epoch": 3.5645569620253164,
      "grad_norm": 0.22926730360327544,
      "learning_rate": 9.08105157279628e-06,
      "loss": 0.4159,
      "step": 880
    },
    {
      "epoch": 3.568607594936709,
      "grad_norm": 0.26225859555362585,
      "learning_rate": 9.03354208532653e-06,
      "loss": 0.4199,
      "step": 881
    },
    {
      "epoch": 3.5726582278481014,
      "grad_norm": 0.23241415479050753,
      "learning_rate": 8.986120920201205e-06,
      "loss": 0.4228,
      "step": 882
    },
    {
      "epoch": 3.576708860759494,
      "grad_norm": 0.23202425826349732,
      "learning_rate": 8.938788459343852e-06,
      "loss": 0.4156,
      "step": 883
    },
    {
      "epoch": 3.580759493670886,
      "grad_norm": 0.24328469880457745,
      "learning_rate": 8.8915450839636e-06,
      "loss": 0.4075,
      "step": 884
    },
    {
      "epoch": 3.5848101265822785,
      "grad_norm": 0.2349804396314518,
      "learning_rate": 8.844391174552116e-06,
      "loss": 0.4237,
      "step": 885
    },
    {
      "epoch": 3.588860759493671,
      "grad_norm": 0.2527982014761089,
      "learning_rate": 8.797327110880479e-06,
      "loss": 0.4126,
      "step": 886
    },
    {
      "epoch": 3.592911392405063,
      "grad_norm": 0.2471290002240711,
      "learning_rate": 8.750353271996206e-06,
      "loss": 0.4024,
      "step": 887
    },
    {
      "epoch": 3.5969620253164556,
      "grad_norm": 0.23516849230653505,
      "learning_rate": 8.703470036220132e-06,
      "loss": 0.4199,
      "step": 888
    },
    {
      "epoch": 3.601012658227848,
      "grad_norm": 0.22477549350550297,
      "learning_rate": 8.656677781143394e-06,
      "loss": 0.4151,
      "step": 889
    },
    {
      "epoch": 3.6050632911392406,
      "grad_norm": 0.2715290853641857,
      "learning_rate": 8.609976883624377e-06,
      "loss": 0.4217,
      "step": 890
    },
    {
      "epoch": 3.609113924050633,
      "grad_norm": 0.2439671739231745,
      "learning_rate": 8.563367719785698e-06,
      "loss": 0.4179,
      "step": 891
    },
    {
      "epoch": 3.613164556962025,
      "grad_norm": 0.22435655523306888,
      "learning_rate": 8.516850665011138e-06,
      "loss": 0.4014,
      "step": 892
    },
    {
      "epoch": 3.6172151898734177,
      "grad_norm": 0.2649896697483519,
      "learning_rate": 8.47042609394269e-06,
      "loss": 0.4264,
      "step": 893
    },
    {
      "epoch": 3.6212658227848102,
      "grad_norm": 0.2520517710281064,
      "learning_rate": 8.424094380477432e-06,
      "loss": 0.4207,
      "step": 894
    },
    {
      "epoch": 3.6253164556962023,
      "grad_norm": 0.23995721739748685,
      "learning_rate": 8.37785589776465e-06,
      "loss": 0.4331,
      "step": 895
    },
    {
      "epoch": 3.629367088607595,
      "grad_norm": 0.22714763458819415,
      "learning_rate": 8.331711018202694e-06,
      "loss": 0.4191,
      "step": 896
    },
    {
      "epoch": 3.6334177215189873,
      "grad_norm": 0.2213990563435402,
      "learning_rate": 8.285660113436104e-06,
      "loss": 0.4148,
      "step": 897
    },
    {
      "epoch": 3.63746835443038,
      "grad_norm": 0.25045323986475415,
      "learning_rate": 8.239703554352527e-06,
      "loss": 0.4128,
      "step": 898
    },
    {
      "epoch": 3.6415189873417724,
      "grad_norm": 0.209656478652168,
      "learning_rate": 8.193841711079775e-06,
      "loss": 0.4149,
      "step": 899
    },
    {
      "epoch": 3.6455696202531644,
      "grad_norm": 0.2163158110944908,
      "learning_rate": 8.148074952982828e-06,
      "loss": 0.4013,
      "step": 900
    },
    {
      "epoch": 3.649620253164557,
      "grad_norm": 0.23295745873847007,
      "learning_rate": 8.102403648660859e-06,
      "loss": 0.4189,
      "step": 901
    },
    {
      "epoch": 3.6536708860759495,
      "grad_norm": 0.23137395377970035,
      "learning_rate": 8.056828165944282e-06,
      "loss": 0.4149,
      "step": 902
    },
    {
      "epoch": 3.6577215189873415,
      "grad_norm": 0.22401234293706565,
      "learning_rate": 8.011348871891762e-06,
      "loss": 0.4211,
      "step": 903
    },
    {
      "epoch": 3.661772151898734,
      "grad_norm": 0.23794233918064692,
      "learning_rate": 7.965966132787287e-06,
      "loss": 0.4218,
      "step": 904
    },
    {
      "epoch": 3.6658227848101266,
      "grad_norm": 0.21841298535534123,
      "learning_rate": 7.920680314137189e-06,
      "loss": 0.4144,
      "step": 905
    },
    {
      "epoch": 3.669873417721519,
      "grad_norm": 0.23031236227950844,
      "learning_rate": 7.875491780667246e-06,
      "loss": 0.423,
      "step": 906
    },
    {
      "epoch": 3.6739240506329116,
      "grad_norm": 0.21997196834727709,
      "learning_rate": 7.830400896319667e-06,
      "loss": 0.4189,
      "step": 907
    },
    {
      "epoch": 3.6779746835443037,
      "grad_norm": 0.2193877823817885,
      "learning_rate": 7.785408024250259e-06,
      "loss": 0.4216,
      "step": 908
    },
    {
      "epoch": 3.682025316455696,
      "grad_norm": 0.22784323033308543,
      "learning_rate": 7.74051352682542e-06,
      "loss": 0.41,
      "step": 909
    },
    {
      "epoch": 3.6860759493670887,
      "grad_norm": 0.2110364447805807,
      "learning_rate": 7.695717765619257e-06,
      "loss": 0.4191,
      "step": 910
    },
    {
      "epoch": 3.690126582278481,
      "grad_norm": 0.21202241707789757,
      "learning_rate": 7.651021101410673e-06,
      "loss": 0.4238,
      "step": 911
    },
    {
      "epoch": 3.6941772151898733,
      "grad_norm": 0.22357752887348697,
      "learning_rate": 7.606423894180464e-06,
      "loss": 0.4124,
      "step": 912
    },
    {
      "epoch": 3.698227848101266,
      "grad_norm": 0.21063077489491633,
      "learning_rate": 7.56192650310839e-06,
      "loss": 0.4179,
      "step": 913
    },
    {
      "epoch": 3.7022784810126583,
      "grad_norm": 0.23903542018274826,
      "learning_rate": 7.517529286570349e-06,
      "loss": 0.412,
      "step": 914
    },
    {
      "epoch": 3.706329113924051,
      "grad_norm": 0.21127505329479815,
      "learning_rate": 7.473232602135387e-06,
      "loss": 0.4161,
      "step": 915
    },
    {
      "epoch": 3.710379746835443,
      "grad_norm": 0.2215793412064054,
      "learning_rate": 7.429036806562935e-06,
      "loss": 0.4059,
      "step": 916
    },
    {
      "epoch": 3.7144303797468354,
      "grad_norm": 0.21759408966787905,
      "learning_rate": 7.3849422557998455e-06,
      "loss": 0.4137,
      "step": 917
    },
    {
      "epoch": 3.718481012658228,
      "grad_norm": 0.20943276647208486,
      "learning_rate": 7.340949304977567e-06,
      "loss": 0.4161,
      "step": 918
    },
    {
      "epoch": 3.72253164556962,
      "grad_norm": 0.22884365488661843,
      "learning_rate": 7.297058308409282e-06,
      "loss": 0.4289,
      "step": 919
    },
    {
      "epoch": 3.7265822784810125,
      "grad_norm": 0.23885052454790517,
      "learning_rate": 7.25326961958704e-06,
      "loss": 0.4103,
      "step": 920
    },
    {
      "epoch": 3.730632911392405,
      "grad_norm": 0.21266592073714152,
      "learning_rate": 7.209583591178921e-06,
      "loss": 0.4114,
      "step": 921
    },
    {
      "epoch": 3.7346835443037976,
      "grad_norm": 0.23470405797765628,
      "learning_rate": 7.1660005750261925e-06,
      "loss": 0.4324,
      "step": 922
    },
    {
      "epoch": 3.73873417721519,
      "grad_norm": 0.25058079538140254,
      "learning_rate": 7.1225209221404765e-06,
      "loss": 0.4111,
      "step": 923
    },
    {
      "epoch": 3.742784810126582,
      "grad_norm": 0.222732028345954,
      "learning_rate": 7.079144982700909e-06,
      "loss": 0.4148,
      "step": 924
    },
    {
      "epoch": 3.7468354430379747,
      "grad_norm": 0.27007402715285894,
      "learning_rate": 7.0358731060513695e-06,
      "loss": 0.3989,
      "step": 925
    },
    {
      "epoch": 3.750886075949367,
      "grad_norm": 0.23353294458660478,
      "learning_rate": 6.99270564069757e-06,
      "loss": 0.4173,
      "step": 926
    },
    {
      "epoch": 3.7549367088607593,
      "grad_norm": 0.22159547255301118,
      "learning_rate": 6.949642934304375e-06,
      "loss": 0.4028,
      "step": 927
    },
    {
      "epoch": 3.7589873417721518,
      "grad_norm": 0.22828919328155972,
      "learning_rate": 6.906685333692871e-06,
      "loss": 0.4247,
      "step": 928
    },
    {
      "epoch": 3.7630379746835443,
      "grad_norm": 0.23118277604580306,
      "learning_rate": 6.86383318483769e-06,
      "loss": 0.4149,
      "step": 929
    },
    {
      "epoch": 3.767088607594937,
      "grad_norm": 0.21273100254565397,
      "learning_rate": 6.821086832864139e-06,
      "loss": 0.4224,
      "step": 930
    },
    {
      "epoch": 3.7711392405063293,
      "grad_norm": 0.23219951763411456,
      "learning_rate": 6.77844662204546e-06,
      "loss": 0.4084,
      "step": 931
    },
    {
      "epoch": 3.7751898734177214,
      "grad_norm": 0.226925716981963,
      "learning_rate": 6.7359128958000455e-06,
      "loss": 0.4027,
      "step": 932
    },
    {
      "epoch": 3.779240506329114,
      "grad_norm": 0.21473361282129558,
      "learning_rate": 6.693485996688695e-06,
      "loss": 0.4106,
      "step": 933
    },
    {
      "epoch": 3.7832911392405064,
      "grad_norm": 0.23236362423744558,
      "learning_rate": 6.651166266411801e-06,
      "loss": 0.3986,
      "step": 934
    },
    {
      "epoch": 3.7873417721518985,
      "grad_norm": 0.23269266573674577,
      "learning_rate": 6.6089540458066725e-06,
      "loss": 0.4042,
      "step": 935
    },
    {
      "epoch": 3.791392405063291,
      "grad_norm": 0.2315704181853091,
      "learning_rate": 6.566849674844711e-06,
      "loss": 0.414,
      "step": 936
    },
    {
      "epoch": 3.7954430379746835,
      "grad_norm": 0.2214571438442625,
      "learning_rate": 6.524853492628747e-06,
      "loss": 0.4102,
      "step": 937
    },
    {
      "epoch": 3.799493670886076,
      "grad_norm": 0.24337763709699473,
      "learning_rate": 6.4829658373902536e-06,
      "loss": 0.4144,
      "step": 938
    },
    {
      "epoch": 3.8035443037974686,
      "grad_norm": 0.2396319997263314,
      "learning_rate": 6.441187046486648e-06,
      "loss": 0.4173,
      "step": 939
    },
    {
      "epoch": 3.8075949367088606,
      "grad_norm": 0.23684991729507926,
      "learning_rate": 6.399517456398567e-06,
      "loss": 0.4049,
      "step": 940
    },
    {
      "epoch": 3.811645569620253,
      "grad_norm": 0.23968732498055503,
      "learning_rate": 6.357957402727164e-06,
      "loss": 0.4203,
      "step": 941
    },
    {
      "epoch": 3.8156962025316457,
      "grad_norm": 0.24061892599113532,
      "learning_rate": 6.316507220191395e-06,
      "loss": 0.4185,
      "step": 942
    },
    {
      "epoch": 3.8197468354430377,
      "grad_norm": 0.2251851377384505,
      "learning_rate": 6.275167242625331e-06,
      "loss": 0.4184,
      "step": 943
    },
    {
      "epoch": 3.8237974683544302,
      "grad_norm": 0.20821558885180502,
      "learning_rate": 6.233937802975471e-06,
      "loss": 0.4165,
      "step": 944
    },
    {
      "epoch": 3.8278481012658228,
      "grad_norm": 0.23944763525235213,
      "learning_rate": 6.192819233298046e-06,
      "loss": 0.4164,
      "step": 945
    },
    {
      "epoch": 3.8318987341772153,
      "grad_norm": 0.2245392447971014,
      "learning_rate": 6.151811864756383e-06,
      "loss": 0.4101,
      "step": 946
    },
    {
      "epoch": 3.835949367088608,
      "grad_norm": 0.23192260049069344,
      "learning_rate": 6.1109160276181655e-06,
      "loss": 0.4197,
      "step": 947
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.21809494368236756,
      "learning_rate": 6.070132051252868e-06,
      "loss": 0.4218,
      "step": 948
    },
    {
      "epoch": 3.8440506329113924,
      "grad_norm": 0.22848808394465095,
      "learning_rate": 6.0294602641290034e-06,
      "loss": 0.4052,
      "step": 949
    },
    {
      "epoch": 3.848101265822785,
      "grad_norm": 0.20138747488615916,
      "learning_rate": 5.988900993811575e-06,
      "loss": 0.4271,
      "step": 950
    },
    {
      "epoch": 3.852151898734177,
      "grad_norm": 0.22893895094104608,
      "learning_rate": 5.948454566959363e-06,
      "loss": 0.4012,
      "step": 951
    },
    {
      "epoch": 3.8562025316455695,
      "grad_norm": 0.23675805788204565,
      "learning_rate": 5.908121309322328e-06,
      "loss": 0.4234,
      "step": 952
    },
    {
      "epoch": 3.860253164556962,
      "grad_norm": 0.2175216033208925,
      "learning_rate": 5.867901545738976e-06,
      "loss": 0.406,
      "step": 953
    },
    {
      "epoch": 3.8643037974683545,
      "grad_norm": 0.22309328743752352,
      "learning_rate": 5.827795600133774e-06,
      "loss": 0.4112,
      "step": 954
    },
    {
      "epoch": 3.868354430379747,
      "grad_norm": 0.2325710304434826,
      "learning_rate": 5.787803795514466e-06,
      "loss": 0.4157,
      "step": 955
    },
    {
      "epoch": 3.872405063291139,
      "grad_norm": 0.22639122172184742,
      "learning_rate": 5.747926453969576e-06,
      "loss": 0.4011,
      "step": 956
    },
    {
      "epoch": 3.8764556962025316,
      "grad_norm": 0.207762678314419,
      "learning_rate": 5.708163896665708e-06,
      "loss": 0.4177,
      "step": 957
    },
    {
      "epoch": 3.880506329113924,
      "grad_norm": 0.20941446635391753,
      "learning_rate": 5.668516443845047e-06,
      "loss": 0.4089,
      "step": 958
    },
    {
      "epoch": 3.884556962025316,
      "grad_norm": 0.2016024119168464,
      "learning_rate": 5.6289844148227225e-06,
      "loss": 0.4032,
      "step": 959
    },
    {
      "epoch": 3.8886075949367087,
      "grad_norm": 0.21106573967197506,
      "learning_rate": 5.5895681279842615e-06,
      "loss": 0.4229,
      "step": 960
    },
    {
      "epoch": 3.8926582278481012,
      "grad_norm": 0.20938293844352165,
      "learning_rate": 5.550267900783019e-06,
      "loss": 0.4102,
      "step": 961
    },
    {
      "epoch": 3.8967088607594937,
      "grad_norm": 0.2054727514311623,
      "learning_rate": 5.511084049737623e-06,
      "loss": 0.4083,
      "step": 962
    },
    {
      "epoch": 3.9007594936708863,
      "grad_norm": 0.21082234034973055,
      "learning_rate": 5.4720168904294215e-06,
      "loss": 0.4025,
      "step": 963
    },
    {
      "epoch": 3.9048101265822783,
      "grad_norm": 0.21828389716217125,
      "learning_rate": 5.433066737499948e-06,
      "loss": 0.4163,
      "step": 964
    },
    {
      "epoch": 3.908860759493671,
      "grad_norm": 0.21741859542928438,
      "learning_rate": 5.394233904648376e-06,
      "loss": 0.4097,
      "step": 965
    },
    {
      "epoch": 3.9129113924050634,
      "grad_norm": 0.2036628056226865,
      "learning_rate": 5.355518704628997e-06,
      "loss": 0.4149,
      "step": 966
    },
    {
      "epoch": 3.9169620253164554,
      "grad_norm": 0.22609446431050897,
      "learning_rate": 5.316921449248731e-06,
      "loss": 0.416,
      "step": 967
    },
    {
      "epoch": 3.921012658227848,
      "grad_norm": 0.19809165048987415,
      "learning_rate": 5.278442449364538e-06,
      "loss": 0.4258,
      "step": 968
    },
    {
      "epoch": 3.9250632911392405,
      "grad_norm": 0.20903757304729806,
      "learning_rate": 5.240082014881016e-06,
      "loss": 0.4089,
      "step": 969
    },
    {
      "epoch": 3.929113924050633,
      "grad_norm": 0.2177478018514132,
      "learning_rate": 5.201840454747822e-06,
      "loss": 0.3946,
      "step": 970
    },
    {
      "epoch": 3.9331645569620255,
      "grad_norm": 0.2063122063537382,
      "learning_rate": 5.163718076957223e-06,
      "loss": 0.4179,
      "step": 971
    },
    {
      "epoch": 3.9372151898734176,
      "grad_norm": 0.22167598918014525,
      "learning_rate": 5.125715188541609e-06,
      "loss": 0.4272,
      "step": 972
    },
    {
      "epoch": 3.94126582278481,
      "grad_norm": 0.2059865007268975,
      "learning_rate": 5.087832095571021e-06,
      "loss": 0.4082,
      "step": 973
    },
    {
      "epoch": 3.9453164556962026,
      "grad_norm": 0.2024651577652389,
      "learning_rate": 5.0500691031506766e-06,
      "loss": 0.4342,
      "step": 974
    },
    {
      "epoch": 3.9493670886075947,
      "grad_norm": 0.2253317018411249,
      "learning_rate": 5.01242651541854e-06,
      "loss": 0.4169,
      "step": 975
    },
    {
      "epoch": 3.953417721518987,
      "grad_norm": 0.21989367340190785,
      "learning_rate": 4.974904635542815e-06,
      "loss": 0.4051,
      "step": 976
    },
    {
      "epoch": 3.9574683544303797,
      "grad_norm": 0.2099645783074133,
      "learning_rate": 4.937503765719582e-06,
      "loss": 0.4068,
      "step": 977
    },
    {
      "epoch": 3.961518987341772,
      "grad_norm": 0.21267115678868861,
      "learning_rate": 4.900224207170299e-06,
      "loss": 0.4001,
      "step": 978
    },
    {
      "epoch": 3.9655696202531647,
      "grad_norm": 0.22541040516362024,
      "learning_rate": 4.8630662601394065e-06,
      "loss": 0.4188,
      "step": 979
    },
    {
      "epoch": 3.969620253164557,
      "grad_norm": 0.21418806896324039,
      "learning_rate": 4.8260302238918995e-06,
      "loss": 0.4287,
      "step": 980
    },
    {
      "epoch": 3.9736708860759493,
      "grad_norm": 0.20741973897264557,
      "learning_rate": 4.789116396710924e-06,
      "loss": 0.4227,
      "step": 981
    },
    {
      "epoch": 3.977721518987342,
      "grad_norm": 0.2093470804808695,
      "learning_rate": 4.752325075895368e-06,
      "loss": 0.4155,
      "step": 982
    },
    {
      "epoch": 3.981772151898734,
      "grad_norm": 0.22312326190880916,
      "learning_rate": 4.715656557757473e-06,
      "loss": 0.3976,
      "step": 983
    },
    {
      "epoch": 3.9858227848101264,
      "grad_norm": 0.2199180688840271,
      "learning_rate": 4.679111137620442e-06,
      "loss": 0.4216,
      "step": 984
    },
    {
      "epoch": 3.989873417721519,
      "grad_norm": 0.1995684398388577,
      "learning_rate": 4.6426891098160585e-06,
      "loss": 0.4327,
      "step": 985
    },
    {
      "epoch": 3.9939240506329114,
      "grad_norm": 0.2064085472983062,
      "learning_rate": 4.6063907676823474e-06,
      "loss": 0.4051,
      "step": 986
    },
    {
      "epoch": 3.997974683544304,
      "grad_norm": 0.22533187015405287,
      "learning_rate": 4.570216403561141e-06,
      "loss": 0.4035,
      "step": 987
    },
    {
      "epoch": 4.002025316455696,
      "grad_norm": 0.23383029925732315,
      "learning_rate": 4.534166308795815e-06,
      "loss": 0.3954,
      "step": 988
    },
    {
      "epoch": 4.006075949367089,
      "grad_norm": 0.38174957203950993,
      "learning_rate": 4.498240773728859e-06,
      "loss": 0.3755,
      "step": 989
    },
    {
      "epoch": 4.010126582278481,
      "grad_norm": 0.25485253160493926,
      "learning_rate": 4.462440087699609e-06,
      "loss": 0.3655,
      "step": 990
    },
    {
      "epoch": 4.014177215189873,
      "grad_norm": 0.3354595309853449,
      "learning_rate": 4.426764539041861e-06,
      "loss": 0.378,
      "step": 991
    },
    {
      "epoch": 4.018227848101266,
      "grad_norm": 0.386395627276107,
      "learning_rate": 4.391214415081582e-06,
      "loss": 0.3667,
      "step": 992
    },
    {
      "epoch": 4.022278481012658,
      "grad_norm": 0.30723388314085925,
      "learning_rate": 4.355790002134579e-06,
      "loss": 0.3866,
      "step": 993
    },
    {
      "epoch": 4.02632911392405,
      "grad_norm": 0.3085180578178756,
      "learning_rate": 4.320491585504207e-06,
      "loss": 0.3674,
      "step": 994
    },
    {
      "epoch": 4.030379746835443,
      "grad_norm": 0.33714406822492904,
      "learning_rate": 4.2853194494790615e-06,
      "loss": 0.3583,
      "step": 995
    },
    {
      "epoch": 4.034430379746835,
      "grad_norm": 0.299137326578124,
      "learning_rate": 4.250273877330691e-06,
      "loss": 0.3783,
      "step": 996
    },
    {
      "epoch": 4.038481012658228,
      "grad_norm": 0.22329026458726728,
      "learning_rate": 4.215355151311313e-06,
      "loss": 0.3769,
      "step": 997
    },
    {
      "epoch": 4.04253164556962,
      "grad_norm": 0.26793845532691984,
      "learning_rate": 4.180563552651542e-06,
      "loss": 0.3473,
      "step": 998
    },
    {
      "epoch": 4.046582278481012,
      "grad_norm": 0.33626885054521666,
      "learning_rate": 4.145899361558147e-06,
      "loss": 0.3757,
      "step": 999
    },
    {
      "epoch": 4.050632911392405,
      "grad_norm": 0.28426618636167433,
      "learning_rate": 4.111362857211738e-06,
      "loss": 0.3695,
      "step": 1000
    },
    {
      "epoch": 4.054683544303797,
      "grad_norm": 0.22195107257668395,
      "learning_rate": 4.076954317764592e-06,
      "loss": 0.3649,
      "step": 1001
    },
    {
      "epoch": 4.0587341772151895,
      "grad_norm": 0.26785142882777535,
      "learning_rate": 4.042674020338335e-06,
      "loss": 0.3654,
      "step": 1002
    },
    {
      "epoch": 4.062784810126582,
      "grad_norm": 0.2625421227461147,
      "learning_rate": 4.0085222410217835e-06,
      "loss": 0.3709,
      "step": 1003
    },
    {
      "epoch": 4.0668354430379745,
      "grad_norm": 0.2631886694870951,
      "learning_rate": 3.974499254868674e-06,
      "loss": 0.3656,
      "step": 1004
    },
    {
      "epoch": 4.0708860759493675,
      "grad_norm": 0.20826551800005616,
      "learning_rate": 3.940605335895451e-06,
      "loss": 0.3828,
      "step": 1005
    },
    {
      "epoch": 4.0749367088607595,
      "grad_norm": 0.2684905782015678,
      "learning_rate": 3.90684075707908e-06,
      "loss": 0.3801,
      "step": 1006
    },
    {
      "epoch": 4.078987341772152,
      "grad_norm": 0.2543442983997565,
      "learning_rate": 3.8732057903548505e-06,
      "loss": 0.365,
      "step": 1007
    },
    {
      "epoch": 4.083037974683545,
      "grad_norm": 0.2520058582774448,
      "learning_rate": 3.8397007066141375e-06,
      "loss": 0.3665,
      "step": 1008
    },
    {
      "epoch": 4.087088607594937,
      "grad_norm": 0.21152238316758754,
      "learning_rate": 3.806325775702304e-06,
      "loss": 0.3846,
      "step": 1009
    },
    {
      "epoch": 4.091139240506329,
      "grad_norm": 0.24916557840095122,
      "learning_rate": 3.773081266416434e-06,
      "loss": 0.3545,
      "step": 1010
    },
    {
      "epoch": 4.095189873417722,
      "grad_norm": 0.24467551009361926,
      "learning_rate": 3.739967446503245e-06,
      "loss": 0.3688,
      "step": 1011
    },
    {
      "epoch": 4.099240506329114,
      "grad_norm": 0.22433643051439828,
      "learning_rate": 3.706984582656894e-06,
      "loss": 0.3727,
      "step": 1012
    },
    {
      "epoch": 4.103291139240507,
      "grad_norm": 0.22135742282350201,
      "learning_rate": 3.6741329405168237e-06,
      "loss": 0.3546,
      "step": 1013
    },
    {
      "epoch": 4.107341772151899,
      "grad_norm": 0.23302536731241344,
      "learning_rate": 3.641412784665648e-06,
      "loss": 0.3627,
      "step": 1014
    },
    {
      "epoch": 4.111392405063291,
      "grad_norm": 0.23825820469848602,
      "learning_rate": 3.608824378627005e-06,
      "loss": 0.3908,
      "step": 1015
    },
    {
      "epoch": 4.115443037974684,
      "grad_norm": 0.23520368025775154,
      "learning_rate": 3.5763679848634337e-06,
      "loss": 0.3715,
      "step": 1016
    },
    {
      "epoch": 4.119493670886076,
      "grad_norm": 0.21767111507708836,
      "learning_rate": 3.544043864774269e-06,
      "loss": 0.3869,
      "step": 1017
    },
    {
      "epoch": 4.123544303797468,
      "grad_norm": 0.21089851143911936,
      "learning_rate": 3.5118522786935282e-06,
      "loss": 0.3792,
      "step": 1018
    },
    {
      "epoch": 4.127594936708861,
      "grad_norm": 0.251215748387612,
      "learning_rate": 3.479793485887819e-06,
      "loss": 0.3734,
      "step": 1019
    },
    {
      "epoch": 4.131645569620253,
      "grad_norm": 0.21834712244561058,
      "learning_rate": 3.4478677445542653e-06,
      "loss": 0.3703,
      "step": 1020
    },
    {
      "epoch": 4.135696202531646,
      "grad_norm": 0.2279142889503118,
      "learning_rate": 3.4160753118183767e-06,
      "loss": 0.3771,
      "step": 1021
    },
    {
      "epoch": 4.139746835443038,
      "grad_norm": 0.21345342388834926,
      "learning_rate": 3.3844164437320527e-06,
      "loss": 0.3659,
      "step": 1022
    },
    {
      "epoch": 4.14379746835443,
      "grad_norm": 0.2309406083196365,
      "learning_rate": 3.3528913952714558e-06,
      "loss": 0.3594,
      "step": 1023
    },
    {
      "epoch": 4.147848101265823,
      "grad_norm": 0.23063299884765595,
      "learning_rate": 3.321500420335e-06,
      "loss": 0.3664,
      "step": 1024
    },
    {
      "epoch": 4.151898734177215,
      "grad_norm": 0.21742164750394125,
      "learning_rate": 3.290243771741275e-06,
      "loss": 0.3824,
      "step": 1025
    },
    {
      "epoch": 4.155949367088607,
      "grad_norm": 0.2223866988319986,
      "learning_rate": 3.2591217012270325e-06,
      "loss": 0.3582,
      "step": 1026
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.21280159557629047,
      "learning_rate": 3.228134459445149e-06,
      "loss": 0.3643,
      "step": 1027
    },
    {
      "epoch": 4.164050632911392,
      "grad_norm": 0.20637562932794526,
      "learning_rate": 3.1972822959626205e-06,
      "loss": 0.3793,
      "step": 1028
    },
    {
      "epoch": 4.168101265822785,
      "grad_norm": 0.21897310073601947,
      "learning_rate": 3.166565459258513e-06,
      "loss": 0.3779,
      "step": 1029
    },
    {
      "epoch": 4.172151898734177,
      "grad_norm": 0.21302999454045682,
      "learning_rate": 3.1359841967220193e-06,
      "loss": 0.3869,
      "step": 1030
    },
    {
      "epoch": 4.176202531645569,
      "grad_norm": 0.21171192395807253,
      "learning_rate": 3.105538754650419e-06,
      "loss": 0.3765,
      "step": 1031
    },
    {
      "epoch": 4.180253164556962,
      "grad_norm": 0.20050435107605363,
      "learning_rate": 3.07522937824712e-06,
      "loss": 0.3782,
      "step": 1032
    },
    {
      "epoch": 4.184303797468354,
      "grad_norm": 0.21614543714119977,
      "learning_rate": 3.0450563116196697e-06,
      "loss": 0.3645,
      "step": 1033
    },
    {
      "epoch": 4.188354430379746,
      "grad_norm": 0.21162064917406376,
      "learning_rate": 3.0150197977778008e-06,
      "loss": 0.3764,
      "step": 1034
    },
    {
      "epoch": 4.192405063291139,
      "grad_norm": 0.23302613588742882,
      "learning_rate": 2.985120078631465e-06,
      "loss": 0.3782,
      "step": 1035
    },
    {
      "epoch": 4.1964556962025314,
      "grad_norm": 0.20512339617170444,
      "learning_rate": 2.9553573949888893e-06,
      "loss": 0.3724,
      "step": 1036
    },
    {
      "epoch": 4.200506329113924,
      "grad_norm": 0.2059723565213932,
      "learning_rate": 2.9257319865546384e-06,
      "loss": 0.3688,
      "step": 1037
    },
    {
      "epoch": 4.2045569620253165,
      "grad_norm": 0.21614949937773495,
      "learning_rate": 2.896244091927678e-06,
      "loss": 0.378,
      "step": 1038
    },
    {
      "epoch": 4.2086075949367086,
      "grad_norm": 0.209332276856673,
      "learning_rate": 2.8668939485994584e-06,
      "loss": 0.3659,
      "step": 1039
    },
    {
      "epoch": 4.2126582278481015,
      "grad_norm": 0.21178681908852773,
      "learning_rate": 2.837681792951994e-06,
      "loss": 0.3534,
      "step": 1040
    },
    {
      "epoch": 4.216708860759494,
      "grad_norm": 0.2070364856152208,
      "learning_rate": 2.808607860255981e-06,
      "loss": 0.3796,
      "step": 1041
    },
    {
      "epoch": 4.220759493670886,
      "grad_norm": 0.20009534086523942,
      "learning_rate": 2.7796723846688634e-06,
      "loss": 0.3741,
      "step": 1042
    },
    {
      "epoch": 4.224810126582279,
      "grad_norm": 0.21253796738678016,
      "learning_rate": 2.7508755992329937e-06,
      "loss": 0.3694,
      "step": 1043
    },
    {
      "epoch": 4.228860759493671,
      "grad_norm": 0.21097440012940472,
      "learning_rate": 2.722217735873718e-06,
      "loss": 0.3655,
      "step": 1044
    },
    {
      "epoch": 4.232911392405064,
      "grad_norm": 0.20788182132024463,
      "learning_rate": 2.6936990253975315e-06,
      "loss": 0.3645,
      "step": 1045
    },
    {
      "epoch": 4.236962025316456,
      "grad_norm": 0.2047893388293901,
      "learning_rate": 2.665319697490205e-06,
      "loss": 0.3523,
      "step": 1046
    },
    {
      "epoch": 4.241012658227848,
      "grad_norm": 0.20799250426352872,
      "learning_rate": 2.637079980714945e-06,
      "loss": 0.3669,
      "step": 1047
    },
    {
      "epoch": 4.245063291139241,
      "grad_norm": 0.21707979113687223,
      "learning_rate": 2.6089801025105453e-06,
      "loss": 0.3706,
      "step": 1048
    },
    {
      "epoch": 4.249113924050633,
      "grad_norm": 0.214981195790325,
      "learning_rate": 2.581020289189571e-06,
      "loss": 0.3607,
      "step": 1049
    },
    {
      "epoch": 4.253164556962025,
      "grad_norm": 0.1990070403781358,
      "learning_rate": 2.553200765936501e-06,
      "loss": 0.3687,
      "step": 1050
    },
    {
      "epoch": 4.257215189873418,
      "grad_norm": 0.20461338858838066,
      "learning_rate": 2.525521756805962e-06,
      "loss": 0.3728,
      "step": 1051
    },
    {
      "epoch": 4.26126582278481,
      "grad_norm": 0.20534396049047576,
      "learning_rate": 2.497983484720885e-06,
      "loss": 0.3668,
      "step": 1052
    },
    {
      "epoch": 4.265316455696203,
      "grad_norm": 0.2018613229566384,
      "learning_rate": 2.470586171470728e-06,
      "loss": 0.3753,
      "step": 1053
    },
    {
      "epoch": 4.269367088607595,
      "grad_norm": 0.20552579834127008,
      "learning_rate": 2.4433300377096836e-06,
      "loss": 0.3704,
      "step": 1054
    },
    {
      "epoch": 4.273417721518987,
      "grad_norm": 0.19970645706925913,
      "learning_rate": 2.4162153029549073e-06,
      "loss": 0.3718,
      "step": 1055
    },
    {
      "epoch": 4.27746835443038,
      "grad_norm": 0.20026505318398374,
      "learning_rate": 2.3892421855847458e-06,
      "loss": 0.3836,
      "step": 1056
    },
    {
      "epoch": 4.281518987341772,
      "grad_norm": 0.21027219020513316,
      "learning_rate": 2.362410902836978e-06,
      "loss": 0.3596,
      "step": 1057
    },
    {
      "epoch": 4.285569620253165,
      "grad_norm": 0.2118451262746592,
      "learning_rate": 2.3357216708070653e-06,
      "loss": 0.3697,
      "step": 1058
    },
    {
      "epoch": 4.289620253164557,
      "grad_norm": 0.20233883361564772,
      "learning_rate": 2.309174704446411e-06,
      "loss": 0.3667,
      "step": 1059
    },
    {
      "epoch": 4.293670886075949,
      "grad_norm": 0.20270637402517822,
      "learning_rate": 2.2827702175606437e-06,
      "loss": 0.3597,
      "step": 1060
    },
    {
      "epoch": 4.297721518987342,
      "grad_norm": 0.20298746828581632,
      "learning_rate": 2.256508422807855e-06,
      "loss": 0.3667,
      "step": 1061
    },
    {
      "epoch": 4.301772151898734,
      "grad_norm": 0.20458200992114323,
      "learning_rate": 2.230389531696946e-06,
      "loss": 0.3651,
      "step": 1062
    },
    {
      "epoch": 4.305822784810126,
      "grad_norm": 0.20959424136592025,
      "learning_rate": 2.204413754585857e-06,
      "loss": 0.3699,
      "step": 1063
    },
    {
      "epoch": 4.309873417721519,
      "grad_norm": 0.2128210058431168,
      "learning_rate": 2.1785813006799406e-06,
      "loss": 0.3706,
      "step": 1064
    },
    {
      "epoch": 4.313924050632911,
      "grad_norm": 0.23248604856307095,
      "learning_rate": 2.1528923780302224e-06,
      "loss": 0.3705,
      "step": 1065
    },
    {
      "epoch": 4.317974683544303,
      "grad_norm": 0.20632093360575463,
      "learning_rate": 2.127347193531757e-06,
      "loss": 0.3735,
      "step": 1066
    },
    {
      "epoch": 4.322025316455696,
      "grad_norm": 0.20662097888350422,
      "learning_rate": 2.101945952921942e-06,
      "loss": 0.3685,
      "step": 1067
    },
    {
      "epoch": 4.326075949367088,
      "grad_norm": 0.20833808710608634,
      "learning_rate": 2.0766888607788906e-06,
      "loss": 0.3731,
      "step": 1068
    },
    {
      "epoch": 4.330126582278481,
      "grad_norm": 0.20278302919591829,
      "learning_rate": 2.0515761205197337e-06,
      "loss": 0.3577,
      "step": 1069
    },
    {
      "epoch": 4.334177215189873,
      "grad_norm": 0.21837350899250932,
      "learning_rate": 2.0266079343990453e-06,
      "loss": 0.3641,
      "step": 1070
    },
    {
      "epoch": 4.3382278481012655,
      "grad_norm": 0.21626643769958315,
      "learning_rate": 2.0017845035071494e-06,
      "loss": 0.3594,
      "step": 1071
    },
    {
      "epoch": 4.3422784810126585,
      "grad_norm": 0.20503880082099307,
      "learning_rate": 1.9771060277685537e-06,
      "loss": 0.3718,
      "step": 1072
    },
    {
      "epoch": 4.3463291139240505,
      "grad_norm": 0.20874805569291383,
      "learning_rate": 1.95257270594031e-06,
      "loss": 0.3778,
      "step": 1073
    },
    {
      "epoch": 4.3503797468354435,
      "grad_norm": 0.21206102902433097,
      "learning_rate": 1.9281847356104188e-06,
      "loss": 0.3828,
      "step": 1074
    },
    {
      "epoch": 4.3544303797468356,
      "grad_norm": 0.2166065958360577,
      "learning_rate": 1.9039423131962365e-06,
      "loss": 0.3697,
      "step": 1075
    },
    {
      "epoch": 4.358481012658228,
      "grad_norm": 0.21790940853039328,
      "learning_rate": 1.8798456339429027e-06,
      "loss": 0.3716,
      "step": 1076
    },
    {
      "epoch": 4.362531645569621,
      "grad_norm": 0.20606713748473623,
      "learning_rate": 1.8558948919217612e-06,
      "loss": 0.3639,
      "step": 1077
    },
    {
      "epoch": 4.366582278481013,
      "grad_norm": 0.1947323325298249,
      "learning_rate": 1.8320902800287954e-06,
      "loss": 0.3743,
      "step": 1078
    },
    {
      "epoch": 4.370632911392405,
      "grad_norm": 0.19914611357063197,
      "learning_rate": 1.8084319899830726e-06,
      "loss": 0.3849,
      "step": 1079
    },
    {
      "epoch": 4.374683544303798,
      "grad_norm": 0.20817045308452728,
      "learning_rate": 1.7849202123252097e-06,
      "loss": 0.368,
      "step": 1080
    },
    {
      "epoch": 4.37873417721519,
      "grad_norm": 0.20178263149439804,
      "learning_rate": 1.7615551364158401e-06,
      "loss": 0.3704,
      "step": 1081
    },
    {
      "epoch": 4.382784810126582,
      "grad_norm": 0.19809386859714087,
      "learning_rate": 1.738336950434061e-06,
      "loss": 0.3655,
      "step": 1082
    },
    {
      "epoch": 4.386835443037975,
      "grad_norm": 0.2037435484156601,
      "learning_rate": 1.715265841375957e-06,
      "loss": 0.3651,
      "step": 1083
    },
    {
      "epoch": 4.390886075949367,
      "grad_norm": 0.2116806539312197,
      "learning_rate": 1.6923419950530684e-06,
      "loss": 0.3709,
      "step": 1084
    },
    {
      "epoch": 4.39493670886076,
      "grad_norm": 0.20006758641049305,
      "learning_rate": 1.6695655960909008e-06,
      "loss": 0.3855,
      "step": 1085
    },
    {
      "epoch": 4.398987341772152,
      "grad_norm": 0.2052280215216722,
      "learning_rate": 1.646936827927441e-06,
      "loss": 0.3592,
      "step": 1086
    },
    {
      "epoch": 4.403037974683544,
      "grad_norm": 0.19971320977576953,
      "learning_rate": 1.6244558728116766e-06,
      "loss": 0.3703,
      "step": 1087
    },
    {
      "epoch": 4.407088607594937,
      "grad_norm": 0.20483161955919033,
      "learning_rate": 1.6021229118021265e-06,
      "loss": 0.3569,
      "step": 1088
    },
    {
      "epoch": 4.411139240506329,
      "grad_norm": 0.20350198618111182,
      "learning_rate": 1.5799381247653967e-06,
      "loss": 0.3597,
      "step": 1089
    },
    {
      "epoch": 4.415189873417722,
      "grad_norm": 0.20251067852038496,
      "learning_rate": 1.5579016903747013e-06,
      "loss": 0.3782,
      "step": 1090
    },
    {
      "epoch": 4.419240506329114,
      "grad_norm": 0.20496444710265416,
      "learning_rate": 1.5360137861084656e-06,
      "loss": 0.3624,
      "step": 1091
    },
    {
      "epoch": 4.423291139240506,
      "grad_norm": 0.19899724491931184,
      "learning_rate": 1.5142745882488475e-06,
      "loss": 0.3771,
      "step": 1092
    },
    {
      "epoch": 4.427341772151899,
      "grad_norm": 0.20081948661757276,
      "learning_rate": 1.4926842718803691e-06,
      "loss": 0.3671,
      "step": 1093
    },
    {
      "epoch": 4.431392405063291,
      "grad_norm": 0.19799720295608475,
      "learning_rate": 1.4712430108884657e-06,
      "loss": 0.3631,
      "step": 1094
    },
    {
      "epoch": 4.435443037974683,
      "grad_norm": 0.20402443350828575,
      "learning_rate": 1.4499509779581078e-06,
      "loss": 0.369,
      "step": 1095
    },
    {
      "epoch": 4.439493670886076,
      "grad_norm": 0.20203686152029354,
      "learning_rate": 1.4288083445723988e-06,
      "loss": 0.385,
      "step": 1096
    },
    {
      "epoch": 4.443544303797468,
      "grad_norm": 0.19693128366630258,
      "learning_rate": 1.4078152810112045e-06,
      "loss": 0.3603,
      "step": 1097
    },
    {
      "epoch": 4.44759493670886,
      "grad_norm": 0.20235386080871376,
      "learning_rate": 1.3869719563497697e-06,
      "loss": 0.3781,
      "step": 1098
    },
    {
      "epoch": 4.451645569620253,
      "grad_norm": 0.195405500740495,
      "learning_rate": 1.3662785384573663e-06,
      "loss": 0.3629,
      "step": 1099
    },
    {
      "epoch": 4.455696202531645,
      "grad_norm": 0.19168854165298244,
      "learning_rate": 1.3457351939959383e-06,
      "loss": 0.3717,
      "step": 1100
    },
    {
      "epoch": 4.459746835443038,
      "grad_norm": 0.1989151166679931,
      "learning_rate": 1.3253420884187551e-06,
      "loss": 0.3636,
      "step": 1101
    },
    {
      "epoch": 4.46379746835443,
      "grad_norm": 0.2031461991236245,
      "learning_rate": 1.3050993859690953e-06,
      "loss": 0.3672,
      "step": 1102
    },
    {
      "epoch": 4.467848101265822,
      "grad_norm": 0.19301451873678271,
      "learning_rate": 1.2850072496788869e-06,
      "loss": 0.3685,
      "step": 1103
    },
    {
      "epoch": 4.471898734177215,
      "grad_norm": 0.19147885094725345,
      "learning_rate": 1.2650658413674434e-06,
      "loss": 0.3712,
      "step": 1104
    },
    {
      "epoch": 4.4759493670886075,
      "grad_norm": 0.19426906670979383,
      "learning_rate": 1.2452753216401226e-06,
      "loss": 0.3658,
      "step": 1105
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.19365285387095088,
      "learning_rate": 1.2256358498870503e-06,
      "loss": 0.3749,
      "step": 1106
    },
    {
      "epoch": 4.4840506329113925,
      "grad_norm": 0.19970098931469718,
      "learning_rate": 1.2061475842818337e-06,
      "loss": 0.3703,
      "step": 1107
    },
    {
      "epoch": 4.488101265822785,
      "grad_norm": 0.19665228000280865,
      "learning_rate": 1.1868106817802816e-06,
      "loss": 0.3736,
      "step": 1108
    },
    {
      "epoch": 4.4921518987341775,
      "grad_norm": 0.19549894514662816,
      "learning_rate": 1.1676252981191482e-06,
      "loss": 0.3638,
      "step": 1109
    },
    {
      "epoch": 4.49620253164557,
      "grad_norm": 0.19829355520960645,
      "learning_rate": 1.1485915878148823e-06,
      "loss": 0.3702,
      "step": 1110
    },
    {
      "epoch": 4.500253164556962,
      "grad_norm": 0.20024815716156083,
      "learning_rate": 1.1297097041623584e-06,
      "loss": 0.3657,
      "step": 1111
    },
    {
      "epoch": 4.504303797468355,
      "grad_norm": 0.19699001112848946,
      "learning_rate": 1.1109797992336847e-06,
      "loss": 0.3648,
      "step": 1112
    },
    {
      "epoch": 4.508354430379747,
      "grad_norm": 0.20185222140879966,
      "learning_rate": 1.092402023876933e-06,
      "loss": 0.3778,
      "step": 1113
    },
    {
      "epoch": 4.512405063291139,
      "grad_norm": 0.20579367848989755,
      "learning_rate": 1.0739765277149527e-06,
      "loss": 0.3699,
      "step": 1114
    },
    {
      "epoch": 4.516455696202532,
      "grad_norm": 0.20217953410887549,
      "learning_rate": 1.0557034591441596e-06,
      "loss": 0.3654,
      "step": 1115
    },
    {
      "epoch": 4.520506329113924,
      "grad_norm": 0.20025906754426578,
      "learning_rate": 1.0375829653333324e-06,
      "loss": 0.3684,
      "step": 1116
    },
    {
      "epoch": 4.524556962025317,
      "grad_norm": 0.2013741644452221,
      "learning_rate": 1.0196151922224385e-06,
      "loss": 0.3583,
      "step": 1117
    },
    {
      "epoch": 4.528607594936709,
      "grad_norm": 0.1931035924325039,
      "learning_rate": 1.0018002845214526e-06,
      "loss": 0.3647,
      "step": 1118
    },
    {
      "epoch": 4.532658227848101,
      "grad_norm": 0.2056383904410794,
      "learning_rate": 9.841383857091947e-07,
      "loss": 0.3652,
      "step": 1119
    },
    {
      "epoch": 4.536708860759494,
      "grad_norm": 0.2167731280067832,
      "learning_rate": 9.666296380321616e-07,
      "loss": 0.3514,
      "step": 1120
    },
    {
      "epoch": 4.540759493670886,
      "grad_norm": 0.18949598719854865,
      "learning_rate": 9.492741825034124e-07,
      "loss": 0.378,
      "step": 1121
    },
    {
      "epoch": 4.544810126582279,
      "grad_norm": 0.19729747504624592,
      "learning_rate": 9.320721589013892e-07,
      "loss": 0.3736,
      "step": 1122
    },
    {
      "epoch": 4.548860759493671,
      "grad_norm": 0.19773888349845925,
      "learning_rate": 9.150237057688339e-07,
      "loss": 0.362,
      "step": 1123
    },
    {
      "epoch": 4.552911392405063,
      "grad_norm": 0.19775726765053958,
      "learning_rate": 8.981289604116328e-07,
      "loss": 0.3684,
      "step": 1124
    },
    {
      "epoch": 4.556962025316456,
      "grad_norm": 0.19450990915648061,
      "learning_rate": 8.813880588977542e-07,
      "loss": 0.3778,
      "step": 1125
    },
    {
      "epoch": 4.561012658227848,
      "grad_norm": 0.19979736708473808,
      "learning_rate": 8.648011360561126e-07,
      "loss": 0.3561,
      "step": 1126
    },
    {
      "epoch": 4.56506329113924,
      "grad_norm": 0.2002945429319717,
      "learning_rate": 8.483683254755037e-07,
      "loss": 0.3541,
      "step": 1127
    },
    {
      "epoch": 4.569113924050633,
      "grad_norm": 0.2009398397073559,
      "learning_rate": 8.320897595035227e-07,
      "loss": 0.3663,
      "step": 1128
    },
    {
      "epoch": 4.573164556962025,
      "grad_norm": 0.19938474291938896,
      "learning_rate": 8.159655692455093e-07,
      "loss": 0.3564,
      "step": 1129
    },
    {
      "epoch": 4.577215189873417,
      "grad_norm": 0.2053831500990717,
      "learning_rate": 7.999958845634648e-07,
      "loss": 0.3689,
      "step": 1130
    },
    {
      "epoch": 4.58126582278481,
      "grad_norm": 0.2002666205430569,
      "learning_rate": 7.841808340750478e-07,
      "loss": 0.3908,
      "step": 1131
    },
    {
      "epoch": 4.585316455696202,
      "grad_norm": 0.19722575218962698,
      "learning_rate": 7.685205451524869e-07,
      "loss": 0.3801,
      "step": 1132
    },
    {
      "epoch": 4.589367088607595,
      "grad_norm": 0.1982693395304635,
      "learning_rate": 7.530151439216027e-07,
      "loss": 0.3732,
      "step": 1133
    },
    {
      "epoch": 4.593417721518987,
      "grad_norm": 0.1970493793762997,
      "learning_rate": 7.376647552607675e-07,
      "loss": 0.3705,
      "step": 1134
    },
    {
      "epoch": 4.597468354430379,
      "grad_norm": 0.1977876043852225,
      "learning_rate": 7.224695027998963e-07,
      "loss": 0.3713,
      "step": 1135
    },
    {
      "epoch": 4.601518987341772,
      "grad_norm": 0.19782936150519176,
      "learning_rate": 7.07429508919466e-07,
      "loss": 0.3652,
      "step": 1136
    },
    {
      "epoch": 4.605569620253164,
      "grad_norm": 0.19172441693328024,
      "learning_rate": 6.925448947495206e-07,
      "loss": 0.3741,
      "step": 1137
    },
    {
      "epoch": 4.609620253164557,
      "grad_norm": 0.1902556406650782,
      "learning_rate": 6.778157801686936e-07,
      "loss": 0.3807,
      "step": 1138
    },
    {
      "epoch": 4.613670886075949,
      "grad_norm": 0.19361001895133612,
      "learning_rate": 6.632422838032515e-07,
      "loss": 0.3702,
      "step": 1139
    },
    {
      "epoch": 4.6177215189873415,
      "grad_norm": 0.20171357861933498,
      "learning_rate": 6.488245230261281e-07,
      "loss": 0.3623,
      "step": 1140
    },
    {
      "epoch": 4.6217721518987345,
      "grad_norm": 0.19450110326742778,
      "learning_rate": 6.345626139559868e-07,
      "loss": 0.365,
      "step": 1141
    },
    {
      "epoch": 4.6258227848101265,
      "grad_norm": 0.19279589447192502,
      "learning_rate": 6.204566714562866e-07,
      "loss": 0.3657,
      "step": 1142
    },
    {
      "epoch": 4.629873417721519,
      "grad_norm": 0.1891792082212798,
      "learning_rate": 6.06506809134344e-07,
      "loss": 0.3744,
      "step": 1143
    },
    {
      "epoch": 4.633924050632912,
      "grad_norm": 0.20770590369979192,
      "learning_rate": 5.927131393404373e-07,
      "loss": 0.3669,
      "step": 1144
    },
    {
      "epoch": 4.637974683544304,
      "grad_norm": 0.1973042212462415,
      "learning_rate": 5.790757731668817e-07,
      "loss": 0.3782,
      "step": 1145
    },
    {
      "epoch": 4.642025316455696,
      "grad_norm": 0.19665404411387785,
      "learning_rate": 5.655948204471507e-07,
      "loss": 0.3777,
      "step": 1146
    },
    {
      "epoch": 4.646075949367089,
      "grad_norm": 0.19861713550745552,
      "learning_rate": 5.522703897549875e-07,
      "loss": 0.3898,
      "step": 1147
    },
    {
      "epoch": 4.650126582278481,
      "grad_norm": 0.1982323502976002,
      "learning_rate": 5.391025884035239e-07,
      "loss": 0.3731,
      "step": 1148
    },
    {
      "epoch": 4.654177215189874,
      "grad_norm": 0.19478016162515197,
      "learning_rate": 5.260915224444207e-07,
      "loss": 0.389,
      "step": 1149
    },
    {
      "epoch": 4.658227848101266,
      "grad_norm": 0.19719066950952938,
      "learning_rate": 5.132372966670129e-07,
      "loss": 0.3687,
      "step": 1150
    },
    {
      "epoch": 4.662278481012658,
      "grad_norm": 0.20381126769759292,
      "learning_rate": 5.005400145974704e-07,
      "loss": 0.3516,
      "step": 1151
    },
    {
      "epoch": 4.666329113924051,
      "grad_norm": 0.21065490548614954,
      "learning_rate": 4.879997784979562e-07,
      "loss": 0.3621,
      "step": 1152
    },
    {
      "epoch": 4.670379746835443,
      "grad_norm": 0.1909279592487378,
      "learning_rate": 4.7561668936580984e-07,
      "loss": 0.3789,
      "step": 1153
    },
    {
      "epoch": 4.674430379746836,
      "grad_norm": 0.19744198060802876,
      "learning_rate": 4.6339084693272306e-07,
      "loss": 0.3807,
      "step": 1154
    },
    {
      "epoch": 4.678481012658228,
      "grad_norm": 0.20286663873428457,
      "learning_rate": 4.5132234966395847e-07,
      "loss": 0.3624,
      "step": 1155
    },
    {
      "epoch": 4.68253164556962,
      "grad_norm": 0.2038201677001349,
      "learning_rate": 4.3941129475752795e-07,
      "loss": 0.3633,
      "step": 1156
    },
    {
      "epoch": 4.686582278481013,
      "grad_norm": 0.19378805976242464,
      "learning_rate": 4.27657778143431e-07,
      "loss": 0.3717,
      "step": 1157
    },
    {
      "epoch": 4.690632911392405,
      "grad_norm": 0.19836069228993353,
      "learning_rate": 4.1606189448287757e-07,
      "loss": 0.3696,
      "step": 1158
    },
    {
      "epoch": 4.694683544303797,
      "grad_norm": 0.2003501897772341,
      "learning_rate": 4.046237371675177e-07,
      "loss": 0.3693,
      "step": 1159
    },
    {
      "epoch": 4.69873417721519,
      "grad_norm": 0.2002909191004781,
      "learning_rate": 3.9334339831869963e-07,
      "loss": 0.3696,
      "step": 1160
    },
    {
      "epoch": 4.702784810126582,
      "grad_norm": 0.19302104583763816,
      "learning_rate": 3.8222096878671955e-07,
      "loss": 0.371,
      "step": 1161
    },
    {
      "epoch": 4.706835443037974,
      "grad_norm": 0.19936465127981864,
      "learning_rate": 3.7125653815009545e-07,
      "loss": 0.3711,
      "step": 1162
    },
    {
      "epoch": 4.710886075949367,
      "grad_norm": 0.19608820295892057,
      "learning_rate": 3.6045019471484974e-07,
      "loss": 0.3763,
      "step": 1163
    },
    {
      "epoch": 4.714936708860759,
      "grad_norm": 0.19528544524784472,
      "learning_rate": 3.498020255137813e-07,
      "loss": 0.3759,
      "step": 1164
    },
    {
      "epoch": 4.718987341772152,
      "grad_norm": 0.19027108814032834,
      "learning_rate": 3.393121163057811e-07,
      "loss": 0.3787,
      "step": 1165
    },
    {
      "epoch": 4.723037974683544,
      "grad_norm": 0.190751056390856,
      "learning_rate": 3.289805515751399e-07,
      "loss": 0.3789,
      "step": 1166
    },
    {
      "epoch": 4.727088607594936,
      "grad_norm": 0.19018446397745253,
      "learning_rate": 3.188074145308573e-07,
      "loss": 0.3704,
      "step": 1167
    },
    {
      "epoch": 4.731139240506329,
      "grad_norm": 0.19280650352498363,
      "learning_rate": 3.087927871059804e-07,
      "loss": 0.3755,
      "step": 1168
    },
    {
      "epoch": 4.735189873417721,
      "grad_norm": 0.1993488817814719,
      "learning_rate": 2.989367499569418e-07,
      "loss": 0.3723,
      "step": 1169
    },
    {
      "epoch": 4.739240506329114,
      "grad_norm": 0.1964693866269329,
      "learning_rate": 2.8923938246290917e-07,
      "loss": 0.3641,
      "step": 1170
    },
    {
      "epoch": 4.743291139240506,
      "grad_norm": 0.19814854748267932,
      "learning_rate": 2.7970076272514804e-07,
      "loss": 0.362,
      "step": 1171
    },
    {
      "epoch": 4.7473417721518985,
      "grad_norm": 0.1872480286124407,
      "learning_rate": 2.703209675663887e-07,
      "loss": 0.383,
      "step": 1172
    },
    {
      "epoch": 4.751392405063291,
      "grad_norm": 0.19567977488750696,
      "learning_rate": 2.6110007253021374e-07,
      "loss": 0.3668,
      "step": 1173
    },
    {
      "epoch": 4.7554430379746835,
      "grad_norm": 0.19785047840077458,
      "learning_rate": 2.520381518804471e-07,
      "loss": 0.3754,
      "step": 1174
    },
    {
      "epoch": 4.759493670886076,
      "grad_norm": 0.19390834188199224,
      "learning_rate": 2.4313527860054585e-07,
      "loss": 0.3688,
      "step": 1175
    },
    {
      "epoch": 4.7635443037974685,
      "grad_norm": 0.18736341473743742,
      "learning_rate": 2.343915243930317e-07,
      "loss": 0.3692,
      "step": 1176
    },
    {
      "epoch": 4.767594936708861,
      "grad_norm": 0.19656104506659142,
      "learning_rate": 2.2580695967889367e-07,
      "loss": 0.3687,
      "step": 1177
    },
    {
      "epoch": 4.7716455696202535,
      "grad_norm": 0.19271883447822538,
      "learning_rate": 2.1738165359704189e-07,
      "loss": 0.3712,
      "step": 1178
    },
    {
      "epoch": 4.775696202531646,
      "grad_norm": 0.1973038938289027,
      "learning_rate": 2.0911567400373257e-07,
      "loss": 0.3685,
      "step": 1179
    },
    {
      "epoch": 4.779746835443038,
      "grad_norm": 0.19463660747370465,
      "learning_rate": 2.0100908747202607e-07,
      "loss": 0.3703,
      "step": 1180
    },
    {
      "epoch": 4.783797468354431,
      "grad_norm": 0.19456683151812587,
      "learning_rate": 1.9306195929125638e-07,
      "loss": 0.3704,
      "step": 1181
    },
    {
      "epoch": 4.787848101265823,
      "grad_norm": 0.19431278369990157,
      "learning_rate": 1.8527435346650247e-07,
      "loss": 0.3647,
      "step": 1182
    },
    {
      "epoch": 4.791898734177215,
      "grad_norm": 0.19307066710229734,
      "learning_rate": 1.7764633271807108e-07,
      "loss": 0.3648,
      "step": 1183
    },
    {
      "epoch": 4.795949367088608,
      "grad_norm": 0.18947942350530061,
      "learning_rate": 1.7017795848099262e-07,
      "loss": 0.3676,
      "step": 1184
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.20041649621852745,
      "learning_rate": 1.6286929090452596e-07,
      "loss": 0.3589,
      "step": 1185
    },
    {
      "epoch": 4.804050632911393,
      "grad_norm": 0.20017240714629844,
      "learning_rate": 1.557203888516745e-07,
      "loss": 0.3578,
      "step": 1186
    },
    {
      "epoch": 4.808101265822785,
      "grad_norm": 0.19570389918229877,
      "learning_rate": 1.487313098987131e-07,
      "loss": 0.3685,
      "step": 1187
    },
    {
      "epoch": 4.812151898734177,
      "grad_norm": 0.19082862238865853,
      "learning_rate": 1.4190211033472402e-07,
      "loss": 0.384,
      "step": 1188
    },
    {
      "epoch": 4.81620253164557,
      "grad_norm": 0.18679813542766394,
      "learning_rate": 1.3523284516113955e-07,
      "loss": 0.3783,
      "step": 1189
    },
    {
      "epoch": 4.820253164556962,
      "grad_norm": 0.19661130566995297,
      "learning_rate": 1.2872356809130682e-07,
      "loss": 0.3752,
      "step": 1190
    },
    {
      "epoch": 4.824303797468355,
      "grad_norm": 0.19391957230969104,
      "learning_rate": 1.2237433155004807e-07,
      "loss": 0.3706,
      "step": 1191
    },
    {
      "epoch": 4.828354430379747,
      "grad_norm": 0.20058116435785212,
      "learning_rate": 1.1618518667323886e-07,
      "loss": 0.3675,
      "step": 1192
    },
    {
      "epoch": 4.832405063291139,
      "grad_norm": 0.19198612751526883,
      "learning_rate": 1.1015618330740385e-07,
      "loss": 0.3743,
      "step": 1193
    },
    {
      "epoch": 4.836455696202532,
      "grad_norm": 0.19297074492098343,
      "learning_rate": 1.042873700093061e-07,
      "loss": 0.3841,
      "step": 1194
    },
    {
      "epoch": 4.840506329113924,
      "grad_norm": 0.19608792911552783,
      "learning_rate": 9.857879404556291e-08,
      "loss": 0.3756,
      "step": 1195
    },
    {
      "epoch": 4.844556962025316,
      "grad_norm": 0.19101916300205224,
      "learning_rate": 9.303050139225722e-08,
      "loss": 0.3729,
      "step": 1196
    },
    {
      "epoch": 4.848607594936709,
      "grad_norm": 0.1949415869902579,
      "learning_rate": 8.76425367345779e-08,
      "loss": 0.3756,
      "step": 1197
    },
    {
      "epoch": 4.852658227848101,
      "grad_norm": 0.19249833424262044,
      "learning_rate": 8.241494346644897e-08,
      "loss": 0.37,
      "step": 1198
    },
    {
      "epoch": 4.856708860759493,
      "grad_norm": 0.23286209050406617,
      "learning_rate": 7.734776369019204e-08,
      "loss": 0.3698,
      "step": 1199
    },
    {
      "epoch": 4.860759493670886,
      "grad_norm": 0.19340348074693778,
      "learning_rate": 7.244103821617332e-08,
      "loss": 0.3622,
      "step": 1200
    },
    {
      "epoch": 4.864810126582278,
      "grad_norm": 0.1909225852090869,
      "learning_rate": 6.769480656248606e-08,
      "loss": 0.3764,
      "step": 1201
    },
    {
      "epoch": 4.868860759493671,
      "grad_norm": 0.19385442214560197,
      "learning_rate": 6.310910695462635e-08,
      "loss": 0.3781,
      "step": 1202
    },
    {
      "epoch": 4.872911392405063,
      "grad_norm": 0.1911492486362271,
      "learning_rate": 5.8683976325191185e-08,
      "loss": 0.3604,
      "step": 1203
    },
    {
      "epoch": 4.876962025316455,
      "grad_norm": 0.18974172573682738,
      "learning_rate": 5.4419450313571984e-08,
      "loss": 0.3754,
      "step": 1204
    },
    {
      "epoch": 4.881012658227848,
      "grad_norm": 0.19609372990000615,
      "learning_rate": 5.031556326567488e-08,
      "loss": 0.3737,
      "step": 1205
    },
    {
      "epoch": 4.88506329113924,
      "grad_norm": 0.20612469170085448,
      "learning_rate": 4.637234823364312e-08,
      "loss": 0.3804,
      "step": 1206
    },
    {
      "epoch": 4.889113924050633,
      "grad_norm": 0.18577331687414014,
      "learning_rate": 4.258983697558838e-08,
      "loss": 0.3728,
      "step": 1207
    },
    {
      "epoch": 4.8931645569620255,
      "grad_norm": 0.19330057506595444,
      "learning_rate": 3.896805995533548e-08,
      "loss": 0.3681,
      "step": 1208
    },
    {
      "epoch": 4.8972151898734175,
      "grad_norm": 0.1980700084393809,
      "learning_rate": 3.550704634218028e-08,
      "loss": 0.3682,
      "step": 1209
    },
    {
      "epoch": 4.9012658227848105,
      "grad_norm": 0.19784576796836684,
      "learning_rate": 3.2206824010647676e-08,
      "loss": 0.355,
      "step": 1210
    },
    {
      "epoch": 4.905316455696203,
      "grad_norm": 0.1956268676122362,
      "learning_rate": 2.9067419540278476e-08,
      "loss": 0.3561,
      "step": 1211
    },
    {
      "epoch": 4.909367088607595,
      "grad_norm": 0.19159434473455017,
      "learning_rate": 2.6088858215400638e-08,
      "loss": 0.3729,
      "step": 1212
    },
    {
      "epoch": 4.913417721518988,
      "grad_norm": 0.19780025649020805,
      "learning_rate": 2.3271164024940564e-08,
      "loss": 0.377,
      "step": 1213
    },
    {
      "epoch": 4.91746835443038,
      "grad_norm": 0.19339086733272057,
      "learning_rate": 2.061435966221881e-08,
      "loss": 0.3716,
      "step": 1214
    },
    {
      "epoch": 4.921518987341772,
      "grad_norm": 0.1907790419174612,
      "learning_rate": 1.811846652477245e-08,
      "loss": 0.3861,
      "step": 1215
    },
    {
      "epoch": 4.925569620253165,
      "grad_norm": 0.19657974787570867,
      "learning_rate": 1.5783504714184106e-08,
      "loss": 0.3749,
      "step": 1216
    },
    {
      "epoch": 4.929620253164557,
      "grad_norm": 0.1964104046217562,
      "learning_rate": 1.360949303591097e-08,
      "loss": 0.376,
      "step": 1217
    },
    {
      "epoch": 4.93367088607595,
      "grad_norm": 0.19755700041732416,
      "learning_rate": 1.1596448999144916e-08,
      "loss": 0.3558,
      "step": 1218
    },
    {
      "epoch": 4.937721518987342,
      "grad_norm": 0.1917928696270574,
      "learning_rate": 9.744388816668172e-09,
      "loss": 0.3731,
      "step": 1219
    },
    {
      "epoch": 4.941772151898734,
      "grad_norm": 0.18988638296207275,
      "learning_rate": 8.05332740472009e-09,
      "loss": 0.3724,
      "step": 1220
    },
    {
      "epoch": 4.945822784810127,
      "grad_norm": 0.20178301101296467,
      "learning_rate": 6.523278382872811e-09,
      "loss": 0.3537,
      "step": 1221
    },
    {
      "epoch": 4.949873417721519,
      "grad_norm": 0.1992005662267341,
      "learning_rate": 5.15425407393133e-09,
      "loss": 0.3816,
      "step": 1222
    },
    {
      "epoch": 4.953924050632912,
      "grad_norm": 0.19050865077285728,
      "learning_rate": 3.94626550383137e-09,
      "loss": 0.3742,
      "step": 1223
    },
    {
      "epoch": 4.957974683544304,
      "grad_norm": 0.1872926460109918,
      "learning_rate": 2.899322401546112e-09,
      "loss": 0.3895,
      "step": 1224
    },
    {
      "epoch": 4.962025316455696,
      "grad_norm": 0.20099451486761116,
      "learning_rate": 2.013433199010706e-09,
      "loss": 0.3608,
      "step": 1225
    },
    {
      "epoch": 4.966075949367089,
      "grad_norm": 0.19890470296681306,
      "learning_rate": 1.2886050310556563e-09,
      "loss": 0.3743,
      "step": 1226
    },
    {
      "epoch": 4.970126582278481,
      "grad_norm": 0.1966593391401625,
      "learning_rate": 7.248437353468695e-10,
      "loss": 0.3623,
      "step": 1227
    },
    {
      "epoch": 4.974177215189873,
      "grad_norm": 0.1991880070762382,
      "learning_rate": 3.221538523412449e-10,
      "loss": 0.3808,
      "step": 1228
    },
    {
      "epoch": 4.978227848101266,
      "grad_norm": 0.19001535140018364,
      "learning_rate": 8.053862524670663e-11,
      "loss": 0.375,
      "step": 1229
    },
    {
      "epoch": 4.982278481012658,
      "grad_norm": 0.18580033503179685,
      "learning_rate": 0.0,
      "loss": 0.3718,
      "step": 1230
    },
    {
      "epoch": 4.982278481012658,
      "step": 1230,
      "total_flos": 5.917335126752231e+18,
      "train_loss": 0.49487486435630457,
      "train_runtime": 63553.2393,
      "train_samples_per_second": 2.486,
      "train_steps_per_second": 0.019
    }
  ],
  "logging_steps": 1,
  "max_steps": 1230,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.917335126752231e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}